University of Southern California Dissertations and Theses

Wastewater-based epidemiology for emerging biological contaminants

(USC Thesis Other)

Wastewater-based epidemiology for emerging biological contaminants

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Phillip Wang

Wastewater-based Epidemiology for Emerging Biological Contaminants

By

Phillip Wang

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(ENVIRONMENTAL ENGINEERING)
December 2022
ii

Acknowledgments
I would like to thank my advisor Dr. Adam Smith, whose patience, support, and
insight have made this success possible. I would also like to express thanks to my
thesis committee members Dr. Amy Childress, Dr. Daniel McCurry, and Dr. Cameron
Thrash for their thoughtful comments and recommendations on this dissertation.
Furthermore, I would like to thank Drs. Moustapha Harb, Ali Zarei-Baygi, Lauren
Stadler, and Jeseth Delgado-Vela for their invaluable assistance and contributions
during my Ph.D. journey. A special thanks to my family for their unconditional support.
Lastly, I would also like to acknowledge the following sources of financial support for
this research: Viterbi Graduate School Fellowship, Water for Agriculture Grant from the
USDA National Institute of Food and Agriculture, and US-Egypt Science and
Technology Joint Fund.

iii

Table of Contents
ACKNOWLEDGMENTS ................................................................................................................................ ii
LIST OF TABLES ......................................................................................................................................... vi
LIST OF FIGURES ....................................................................................................................................... vii
ABSTRACT ................................................................................................................................................... x
1. INVESTIGATING BIOLOGICAL CONTAMINANTS WITHIN OUR WASTEWATER MANAGEMENT
SYSTEM........................................................................................................................................................ 1
1.1 INTRODUCTION ...................................................................................................................................... 1
1.2 DISSERTATION OVERVIEW ...................................................................................................................... 3
REFERENCES .............................................................................................................................................. 5
2. LONG-TERM SURVEILLANCE OF WASTEWATER SARS-COV-2 IN LOS ANGELES COUNTY ......... 6
ABSTRACT .................................................................................................................................................. 6
2.1 INTRODUCTION ...................................................................................................................................... 7
2.2 MATERIALS AND METHODS ................................................................................................................... 10
2.2.1 Sample Collection and Enveloped Virus Concentration ............................................................ 10
2.2.2 RNA Extraction ........................................................................................................................... 12
2.2.3 SARS-CoV-2 Quantification ....................................................................................................... 13
2.2.4 Process flow and inhibition control assessment ........................................................................ 14
2.2.5 Variant Analysis with RT-ddPCR ............................................................................................... 15
2.2.6 Contributing Cases by Sewershed ............................................................................................. 16
2.2.7 System-Specific Factor Analysis ................................................................................................ 17
2.2.8 Monte Carlo Simulation .............................................................................................................. 17
2.3 RESULTS AND DISCUSSION .................................................................................................................. 19
2.3.1 SARS-CoV-2 Detected in sampled WWTPs .............................................................................. 19
2.3.2 Wastewater SARS-CoV-2 levels show strong sensitivity and correlation to reported new
cases of Covid-19 in Los Angeles County .......................................................................................... 20
2.3.3 Wastewater SARS-CoV-2 levels strongly correlates to the average new cases by
contributing sewershed ....................................................................................................................... 22
2.3.4 Comparison of daily viral load to 20-day case count could help identify under testing
communities ........................................................................................................................................ 24
2.3.5 Clinical cases normalized by influent flow rate, TSS, and BOD5 are potential factors that
may influence Wastewater SARS-CoV-2 Correlation to clinical data ................................................. 28
2.3.6 Daily Wastewater sample could lead clinical data by up to five days ........................................ 30
2.3.7 Variant Analysis ......................................................................................................................... 32
2.3.8 Estimated Infected Population from Monte Carlo Simulations Exceed the Reported Clinical
Cases by more than 200 Percent ....................................................................................................... 33
2.4 CONCLUSION ....................................................................................................................................... 36
REFERENCES ............................................................................................................................................ 38
3. ASSESSMENT OF WASTEWATER SARS-COV-2 MUTATION PROFILE IN LOS ANGELES
COUNTY ..................................................................................................................................................... 42
ABSTRACT ................................................................................................................................................ 42
3.1 INTRODUCTION .................................................................................................................................... 43
3.2 MATERIAL AND METHODS ..................................................................................................................... 46
3.2.1 Wastewater Samples Selection and RNA Extraction ................................................................ 46
3.2.2 Sequencing and Data processing .............................................................................................. 48
3.2.3 Clinical Data ............................................................................................................................... 48
iv

3.3 RESULTS ............................................................................................................................................. 48
3.3.1 Overview of wastewater sequencing quality .............................................................................. 48
3.3.2 Mutation Type and Frequency for JW and Hyperion (WW and Biosolid) ................................ 49
3.3.3 Comparison between Influent and Biosolid Samples ................................................................ 52
3.3.4 Top mutations in JW and Hyperion samples overtime .............................................................. 54
3.3.5 Comparison of WW mutations to clinical data ........................................................................... 56
3.3.6 Mutations relating to variants of concern and variants of interest.............................................. 57
3.4 CONCLUSION ....................................................................................................................................... 59
REFERENCES ............................................................................................................................................ 61
4. METAGENOMIC ANALYSIS OF THE ANTIBIOTIC RESISTANCE RISK BETWEEN AN AEROBIC
AND ANAEROBIC MEMBRANE BIOREACTOR........................................................................................ 65
ABSTRACT ................................................................................................................................................ 65
4.1 INTRODUCTION .................................................................................................................................... 66
4.2 MATERIAL AND METHODS ..................................................................................................................... 68
4.2.1 Bioreactor operating conditions ................................................................................................. 68
4.2.2 Sample collection and DNA extraction ....................................................................................... 69
4.2.3 ARG quantification via qPCR ..................................................................................................... 69
4.2.4 Illumina sequencing and sequenced data assembly ................................................................. 70
4.2.5 Microbial community assessment .............................................................................................. 71
4.3 RESULTS AND DISCUSSION ................................................................................................................... 71
4.3.1 Robust performance of parallel bioreactors ............................................................................... 71
4.3.2 Antibiotics loading displayed stronger effects on the alpha and beta diversity of the AeMBR
compared to the AnMBR ..................................................................................................................... 72
4.3.3 Microbial community members were distinct across sample types ........................................... 75
4.3.4 Contribution of shared ARGs across sample types were inversely impacted by the addition
of antibiotics in the AeMBR and AnMBR ............................................................................................ 79
4.3.5 Anitbiotics addition enriched resistance classes in the AeMBR biomass, AeMBR iDNA, and
AnMBR exDNA.................................................................................................................................... 81
4.3.6 AeMBR biomass and effluent iDNA showed greater increase in ARGs/rpoB compared to its
AnMBR counterpart ............................................................................................................................. 84
REFERENCES ............................................................................................................................................ 88
5. CHARACTERIZING MOBILE COLISTIN RESISTANCE GENES WITHIN LOS ANGELES COUNTY
WASTEWATER .......................................................................................................................................... 92
ABSTRACT ................................................................................................................................................ 92
5.1 INTRODUCTION .................................................................................................................................... 93
5.2 MATERIAL AND METHODS ..................................................................................................................... 94
5.2.1 Sample processing ..................................................................................................................... 94
5.2.2 Genomic DNA extraction ............................................................................................................ 95
5.2.3 Plasmid extraction and processing ............................................................................................ 95
5.2.4 PCR-based screening of mcr-1 through mcr-9 .......................................................................... 96
5.2.5 qPCR analysis of mcr-4 and mcr-5 ............................................................................................ 96
5.2.6 Culture-based screening of colistin-resistant bacteria ............................................................... 96
5.2.7 Illumina and Nanopore Sequencing and Assembly ................................................................... 97
5.3 RESULTS ............................................................................................................................................. 98
5.3.1 PCR and qPCR detection of mcr genes in wastewater samples ............................................... 98
5.3.2 Characterization of cultured colistin-resistant bacterial isolates. ............................................... 99
5.3.3 Culture-independent metagenomic analysis of wastewater plasmids ..................................... 103
mcr contigs ........................................................................................................................................ 103
5.3.4 Complete circular plasmids containing mcr-5 .......................................................................... 103
REFERENCES .......................................................................................................................................... 106
v

6. CONCLUSION ...................................................................................................................................... 109
6.1 OVERVIEW ........................................................................................................................................ 109
6.2 WBE IS AN EFFECTIVE COMPLEMENTARY TOOL FOR TRACKING SARS-COV-2 VIRAL LOAD ACROSS
SEWERSHEDS .......................................................................................................................................... 110
6.3 WASTEWATER SARS-COV-2 GENOMES ARE HIGHLY HETEROGENOUS WITH KEY MUTATIONS SHOWING
DOMINANCE ACROSS VARIANTS ................................................................................................................ 111
6.4 ANMBR EXHIBIT GREATER RESILIENCE TOWARD ANTIBIOTIC RESISTANCE PROPAGATION COMPARED TO
AEMBR .................................................................................................................................................. 111
6.5 WBE IS AN EFFECTIVE TOOL TO INVESTIGATE ARGS OF CLINICAL IMPORTANCE .................................... 112
6.6 FUTURE RESEARCH ........................................................................................................................... 113
APPENDIX A: SUPPLEMENTAL INFORMATION FOR CHAPTER 2 ..................................................... 117
APPENDIX B: SUPPORTING INFORMATION FOR CHAPTER 4 .......................................................... 126
B.1 AEMBR AND ANMBR OPERATION ...................................................................................................... 126
B.2 ANALYSIS METHODS .......................................................................................................................... 126
B.3 SYNTHETIC FEED FOR BIOREACTOR ................................................................................................... 127
REFERENCES .......................................................................................................................................... 130
APPENDIX C: SUPPORTING INFORMATION FOR CHAPTER 5 .......................................................... 131
REFERENCES .......................................................................................................................................... 133

vi

List of Tables
Table 3. 1: Overview of the average flow rate, population serviced, and number of samples collected
for Hyperion and Joint Water ...................................................................................................................... 47
Table 3. 2: Table showing the distribution of mutation types within the influent and biosodatasetsaset. .. 50
Table 3. 3: Summary table of the VOC related mutations detected in our analysis and their relative
abundance within wastewater clinical dataset. The last column contains the earliest detection data of
the respective mutation within the wastewater dataset. ............................................................................. 59

Table 5. 1: Genome properties and antibiotic resistance profile of the A. hydrophilia and H. alvei
isolates harboring mcr, isolated from Los Angeles County wastewater. .................................................. 102

Table S2. 1: Primer and probe names, sequences, and final concentrations used in this study. ............ 118
Table S2. 2: Raw data from the Monte Carlo Simulations for the compiled and individual sewersheds . 119

Table S4. 1: Composition of the synthetic feed used in this study. .......................................................... 127
Table S4. 2: Forward and reverse primers and qPCR thermocycling conditions of all targeted genes... 128

Table S5. 1: Primers and thermocycling conditions used in this study
1,2,8
............................................... 131
Table S5. 2: Results for the NCBI Conserved Domain search for (A) Hyp p1 and (B) Hyp p2 ............... 132

vii

List of Figures
Figure 2. 1: (A) Map of LA county and sampled sewersheds. Stars represent approximate location of
each sampled WWTP. Orange = Hyperion, Blue = Joint Water Pollution Control Plant, Purple= Long
Beach Water Reclamation Plant, Red = San Jose Creek Water Reclamation Plant, and Yellow =
Whittier Narrows Water Reclamation Plant. (B) Diagram of the general workflow used in this study to
concentrate and measure wastewater SARS-CoV-2. ................................................................................. 11
Figure 2. 2: (A) Time series analysis of the averaged viral copies/L for all sampled WWTPs and
averaged new cases rate for Los Angeles County over the duration of this study. (B-F) Time series
analysis of SARS-CoV-2 viral copies/L for each sampled WWTP and the averaged new positive tests
for the respective sewershed. The dark red line represents quantified N1 data, the pink line represents
quantified N2 data the and blue bar chart represents the moving seven-day average of daily new
cases A= Total, B=HYP, C=JW, D=SJ, E=LB, and F= WN. Error bars represent the standard deviation
of the measured N1 or N2 value. Pearson coefficient r represents the correlation between the N1 or
N2 gene copies/L and regional clinical data. .............................................................................................. 24
Figure 2. 3: Box plot analysis of daily viral load to the moving 20-day case count for each sampled
sewershed. (A) The duration of the study May 2020-March 2021 (B) May-August 2020 (C) September-
October 2020 (D) November-March 2021. Box represents the median, 25
th
, and 75
th
percentile. The
whiskers represent the largest and smallest values and outliers are shown as circles. Error bars
represent standard deviation. ..................................................................................................................... 27
Figure 2. 4: Comparison of daily wastewater SARS-CoV-2 virus copies/L (N1=red and N2= grey) from
JW to its respective daily new cases (blue). (A) Represents same day comparison. (B) Represents
measured SARS-CoV-2 virus copies/L compared to the daily new cases five days later. Error bars
represent standard deviation. ..................................................................................................................... 32
Figure 2. 5: Comparison of the reported adjusted cases per 100,000 people (light blue) vs. simulated
NIF per 100,000 people (dark blue). Error bars represent the 95% CI. ...................................................... 36

Figure 3. 1: (A) Map of Los Angeles County with the serviced sewershed area shaded in green
(Hyperion) and blue (Joint Water). The star icon represents the approximat location of the wastewater
treatment plant. (B and C) Overlayed line and bar chart depicting the SARS-CoV-2 viral load/L and the
respective 7-day average case count for each sewershed. ReThe d line erepresents the N1 gene
concentratithe on, pink line represents the N2 gene concentration, and the blue bars represent the 7-
day average of new case counts for Hyp or JW. Dark blue arrows signify the dates of the sequenced
samples. ...................................................................................................................................................... 47
Figure 3. 2: Distribution of mutations within the SARS-CoV-2 genes (A). Normalized distribution of
mutation frequency among the SARS-CoV-2 genes (B). Blue bars represent influent data and orange
bars represent biosolid data. ....................................................................................................................... 50
Figure 3. 3: Distribution of nucleotide substitution class within the synonymous mutations (A) and non-
synonymous mutations (B). Light blue bars represent influent data and blue bars represent biosolid
data. ............................................................................................................................................................ 52
Figure 3. 4: Heatmap comparison of SARS-CoV-2 variants between composite influent and biosolid
samples for (A) Hyperion Water Reclamation Plant (Hyp) and (B) Joint Water Pollution Control Plant
(JW). ............................................................................................................................................................ 54
Figure 3. 5: Heatmap with dendrogram of most frequently detected allelic variants within composite
influent samples. (A) Hyperion Water Reclamation Plant (Hyp) and (B) Joint Water Pollution Control
Plant (JW).................................................................................................................................................... 56

Figure 4. 1: Species richness displayed as the anti-log of the Shannon indices for the AeMBR and
AnMBR biomass and effluent iDNA microbial diversity at increasing antibiotic concentrations. ................ 73
Figure 4. 2: Multi-dimensional scaling analyses between the microbial communities collected from the
(A) AeMBR system and the (B) AnMBR system at increasing antibiotic concentrations. Increasing
color darkness represents increasing antibiotic concentrations at 0 µg/L, 10 µg/L, and 250 µg/L.
viii

Dashed loops represent distinct clusters and the arrows signify the directional movement of the
samples from the no antibiotic time point. ................................................................................................... 75
Figure 4. 3: Community structure of the AeMBR and AnMBR biomass and iDNA, at the phylum level
(A and B) and genus level (C and D). The top 15 phylum or genera from each sample type and date
are shown. The remaining phylum and genera are grouped under the label “Others’. Samples
represents antibiotics loading conditions at 0 µg/L, 10 µg/L, and 250 µg/L. ............................................... 77
Figure 4. 4: Box plot analyses of the ARG distribution among the biomass, iDNA, and exDNA sample
types for the (A) AeMBR and (B) AnMBR at increasing antibiotic concentrations. The boxes represent
the 25th and 75th percentile. The whiskers represent the largest and smallest values and outliers are
shown as closed circles. Each shared ARG is plotted as the relative abundance among the three
sample types. .............................................................................................................................................. 80
Figure 4. 5: Heat map analysis of the metagenomic ARGs grouped by resistant classes. Data is
depicted as the log (fold change) of examined resistant classes relative to the 0 µg/L antibiotic
loading period. Results for the biomass, iDNA, and exDNA are grouped as (A) AeMBR (B) and
AnMBR. ....................................................................................................................................................... 83
Figure 4. 6: Bar chart analysis of resistant classes for beta-lactam, mls and sulfonamide, multi-
biocide, multi-drug, and multi-metal. (A) AeMBR biomass and iDNA. (B) AnMBR biomass, iDNA.
Antibiotic concentrations are color coded red (0 µg/L), green (10 µg/L), and yellow (250 µg/L). ............... 85

Figure 5. 1: qPCR analysis of mcr-4 and mcr-5 normalized to rpoB. Blue bars represent samples
collected from Joint Water Pollution Control Plant (JW) and red bars represent samples collected from
the Hyperion Water Reclamation Plant (Hyp). Error bars represent the standard deviation from the
mean for each qPCR run (n=3). .................................................................................................................. 99
Figure 5. 2: Multiple sequence alignment with EasyFig v2.2.5
29
showing (A) A. hydrophilia
scaffold_46 aligned to A. hydrophilia strain ZYAH75, AH1805, AH2359, AH3924, and AH3019 (B) H.
Alvei, scaffold_26 aligned to L. adecarboxylata strain L21, plasmid pAV50177-a, plasmid pEL-ars2,
and E. kobei strain IB2020. Sequence identity is marked by the intensity of the color gray. Genes are
color coded by protein function. ................................................................................................................ 102
Figure 5. 3: Pairwise alignment between Hyp p1 and Hyp p2 using Clinker v0.0.5
30
. Sequence
identity is shown by increasing intensity in the color black. Hypothetical protein coding sequences
have not been confirmed. Genes are color coded by protein function. .................................................... 104

Figure S2. 1: Comparison of measured N1 Copies/L versus N2 Copies/L over the duration of this
study. Linear fit is represented by R
2
and the Pearson correlation between N1 and N2 Copies/L is
represented by r. ....................................................................................................................................... 117
Figure S2. 2: (A) Time series analysis of the moving 20-day Covid-19 tests given in Los Angeles
County. (B) Heat map of the cumulative adjusted rate for persons tested in Los Angeles County.
Adjusted rates are per 100,000 people. Data grouped by service planning area. Image is taken from
the LA County COVID-19 Surveillance Dasboard
http://dashboard.publichealth.lacounty.gov/covid19_surveillance_dashboard/ on Oct 11, 2021. ............ 125
Figure S4. 1: A violin box plot showing the log ratio of the measured genes in the AnMBR and
AeMBR effluent exDNA (copies/L). The box area represents the 25%-75% range of the data.
The colored region of the violin depicts the distribution of the data. The white circle represents the
median. The vertical lines above and below the box are the 1.5 interquartile range ……………………..128
Figure S4. 2: qPCR results of measured rpoB gene copies/L of the effluent iDNA through the study
period. The box represents the 25%-75% range of the measured data. The median is represented by
a horizontal line inside the boxes, the mean is represented by a square inside the boxes, and the 1.5
interquartile range is represented by the bars above and below the colored boxes. ............................... 129
Figure S4. 3: Venn diagram analyses of the ARG type distribution among the biomass, iDNA, and
exDNA of the AeMBR (A) and AnMBR (B) at increasing antibiotic concentrations. ................................. 129

Figure S5. 1: Multiplex PCR detection of mcr genes. Genomic DNA extracts from Joint Water
Pollution Control Plant and Hyperion Water Reclamation plant were used as templates. Numbers 1
ix

through 4 represent collection dates of Nov 2021, Jan 2021, Feb 2021, and Jan 2022, respectively.
Agarose gel electrophoresis (1.5% w/v) was used to separate multiplex PCR products. L indicates the
molecular ladder (Quick-Load 100bp DNA Ladder, Ipswich, MA). The size of the molecular weight
markers are listed on the left side of the gel and the size of the amplicons are indicated on the right
side. NTC = no template control with nuclease-free water. ...................................................................... 131

x

Abstract
Wastewater treatment plants (WWTPs) receive a wide diversity of human-
associated viruses and microorganisms, which can be used to assess the disease
burden within serviced communities. This thesis aims to demonstrate several
applications of how wastewater-based epidemiology (WBE) can be leveraged to track
and characterize biological contaminants in wastewater to better understand how global
health issues such as viral diseases and antibiotic resistant bacteria (ARB) spread,
evolve, and persist within the natural and built environment. First, we tracked
wastewater SARS-CoV-2 levels using stool-associated SARS-CoV-2 viral particles of
the infected population in Los Angeles County. Measured wastewater SARS-CoV-2
levels were strongly correlated to in-person testing data and could be used to model the
infected population within each sewershed. Next, we sequenced select wastewater
SARS-CoV-2 samples from our previous study to assess the SARS-CoV-2 variant
profile over nine months. Analysis of our sequencing data revealed a wide array of
mutations compared to the first sequenced Wuhan SARS-CoV-2 reference genome.
Further, around 67.6% of the detected mutations from our analysis were not found in
sequenced clinical samples from Los Angeles. Turning our focus toward antibiotic
resistance, we used metagenomic sequencing to assess the antibiotic resistance
propagation risk between an aerobic membrane bioreactor (AeMBR) and an anaerobic
membrane bioreactor (AnMBR) under identical operational parameters and antibiotic
stress. Our results showed differential antibiotic resistance selection patterns between
the AeMBR and AnMBR. Under antibiotic stress, the AeMBR enriched for ARB while
the AnMBR enriched for extracellular antibiotic resistant genes. Moreover, antibiotics
xi

added to the shared influent feed showed a greater impact on the microbial diversity
within the AeMBR compared to the AnMBR. Lastly, we report the first detection and
characterization of the mobile-colistin resistant gene (mcr) and its variants in the urban
wastewater of Los Angeles County. We assembled draft genomes of mcr-positive
bacterial hosts and obtained two complete mcr-positive plasmid sequences. This thesis
provides gainful information on the potential application of WBE to explore a broad
spectrum of biological contaminants within our sewer network.

1

Chapter 1
1. Investigating biological contaminants within our
wastewater management system
1.1 Introduction
Wastewater-based epidemiology (WBE) serves as a cost-effective and non-
invasive tool for assessing communal health. The continuous emergence of new
infectious diseases and pathogens creates a significant strain on the public health
authorities’ ability to monitor the various diseases and pathogens among the human
population. Since the 1970s, more than 1,500 new pathogens and approximately 40
novel infectious diseases have been reported
1
. However, due to limited financial and
labor resources for in-person testing, the prevalence of most diseases and pathogens
circulating within the built environment remains a significant knowledge gap. WBE
incorporates the use of molecular biology assays that can help monitor and quantify a
diverse range of fecal-associated viruses and diseases in serviced communities. Data
generated through WBE can be used to trace disease carriers
2,3
, monitor disease
presence
4
, assess the efficacy of public health intervention
5
, and provide early warnings
for potential disease outbreaks
5
.
In addition to infectious viral diseases, the WHO and the Center of Disease
Control has named antibiotic resistance as one of the greatest threats to human health,
with over 700,000 reported deaths annually worldwide
7,8
. Moreover, rising awareness of
the link between human, animal, and environmental health has led to the adoption of
2

the One Health approach to address systemic health issues such as rising antibiotic
resistance, vector-borne diseases, and water contamination
9
. WWTPs are logical
settings for WBE as it serves as a central point to monitor, manage, and characterize
biological contaminants from large human populations. Further, WWTPs are the last
barriers for anthropogenic biological contaminants before release into the natural
environment. Several metagenomic studies of untreated wastewater samples collected
across the United States reported around 15% of their sequenced reads can be traced
back to human fecal bacteria and captures up to 97% of the microbial diversity in
human fecal samples
10–12
. Moreover, influent ARG profiles can be used to assess
regional life attributes such as antibiotic use patterns and socioeconomic factors
13
.
While the relative abundance of ARGs in treated wastewater is significantly reduced
compared to the influent, approximately 90% of the effluent ARGs are also present in
the influent ARG profile
14
. Currently, the fate and interaction of the secreted human
fecal bacteria and its genomic content with the microbial community of non-fecal origin
is a substantial knowledge gap and could contain prominent dissemination pathways for
ARGs between distinct ecologies.
Traditional WBE have relied on the detection of pre-determined targets and
offers limited insight into the evolution and mobility of biological targets (e.g., plasmids,
integrons, transposons, and genomic islands). However, with the declining cost of
sequencing assays
15
and the maturity of analytical tools for sequenced data, modern
WBE can be broadened to encompass the exploration, characterization, and
documentation of wastewater viruses and microorganisms relating to human health.
With the increasing adoption of high-throughput sequencing platforms (e.g., Illumina,
3

PacBio, and Oxford Nanopore), researchers are tapping into the power of nucleic acid
sequencing and uncovering the microbial community, antibiotic resistance, and plasmid
profiles in wastewater. In addition, routine sequencing of human viral diseases in
wastewater such as influenza A, polio, and SARS-CoV-2 offers temporal data of the
mutation profile and evolutionary shift between viral strains
16
. A comprehensive global
database of viral pathogens and antibiotic resistance in our wastewater network would
be integral to our understanding of how biological contaminants spread, evolve, and
persist.
1.2 Dissertation Overview
This dissertation focuses on wastewater surveillance for SARS-CoV-2 and
antibiotic resistance and the assessment of antibiotic-resistance proliferation during
wastewater treatment. We first assessed the utility of WBE to track wastewater SARS-
CoV-2 levels in Los Angeles County as a complementary tool to in-person testing
(Chapter 2). Quantification of wastewater SARS-CoV-2 concentration was done using
RT-qPCR targeting the N1 and N2 genes as recommended by the Centers for Disease
Control. Next, we used amplicon tiling and Illumina sequencing to examine the evolution
of the SARS-CoV-2 genome throughout our previous study (Chapter 3). We then shifted
our focus from viruses to ARGs and ARB. We examined the antibiotic-resistance
propagation risk between an aerobic membrane bioreactor (AeMBR) and an anaerobic
membrane bioreactor (AnMBR) operated in parallel, treating identical synthetic feed
(Chapter 4). Finally, we characterized the emergence of mobile-colistin resistance
4

genes (mcr) in Los Angeles County urban wastewater through sequencing of cultured
colistin-resistant bacteria and wastewater plasmid metagenome.

5

References
1. World Health Organization. The World Health Report 2007: A Safer Future. Glob.
Public Health (2007).
2. MOORE, B. The detection of enteric carriers in towns by means of sewage
examination. J. R. Sanit. Inst. 71, (1951).
3. Moore, B., Perry, C. E. L. & Chard, S. T. A Survey by the sewage swab method of
latent enteric infection in an urban area. J. Hyg. (Lond). 50, (1952).
4. Trask, J. D., Paul, J. R. & Riordan, J. T. Periodic examination of sewage for the
virus of poliomyelitis. J. Exp. Med. 75, (1942).
5. Ahmed, W. et al. SARS-CoV-2 RNA monitoring in wastewater as a potential early
warning system for COVID-19 transmission in the community: A temporal case
study. Sci. Total Environ. 761, (2021).
6. World Health Organization. New report calls for urgent action to avert
antimicrobial resistance crisis. Joint News Release vol. 29 (2019).
7. Thompson, T. The staggering death toll of drug-resistant bacteria. Nature (2022)
doi:10.1038/d41586-022-00228-x.
8. Mackenzie, J. S. & Jeggo, M. The one health approach-why is it so important?
Tropical Medicine and Infectious Disease vol. 4 (2019).
9. Newton, R. J. et al. Sewage reflects the microbiomes of human populations. MBio
6, (2015).
10. Shanks, O. C. et al. Comparison of the microbial community structures of
untreated wastewaters from different geographic locales. Appl. Environ. Microbiol.
79, (2013).
11. Vandewalle, J. L. et al. Acinetobacter, Aeromonas and Trichococcus populations
dominate the microbial community within urban sewer infrastructure. Environ.
Microbiol. 14, (2012).
12. Hendriksen, R. S. et al. Global monitoring of antimicrobial resistance based on
metagenomics analyses of urban sewage. Nat. Commun. 10, (2019).
13. Majeed, H. J. et al. Evaluation of Metagenomic-Enabled Antibiotic Resistance
Surveillance at a Conventional Wastewater Treatment Plant. Front. Microbiol. 12,
(2021).
14. Wetterstrand, K. A. The Cost of Sequencing a Human Genome. National Human
Genome Research Institute (2020).
15. Agrawal, S. et al. Prevalence and circulation patterns of SARS-CoV-2 variants in
European sewage mirror clinical data of 54 European cities. Water Res. 214,
(2022).
6

Chapter 2
2. Long-Term Surveillance of Wastewater SARS-
CoV-2 in Los Angeles County
Phillip Wang
a
, Ali Zarei-Baygi
a
, Connor Sauceda
a
, Syeed Md Iskander
b
, Adam L. Smith
a
*
a
Sonny Astani Department of Civil and Environmental Engineering, University of Southern California,
3620 South Vermont Avenue, Los Angeles, California 90089, United States
b
Department of Civil, Construction and Environmental Engineering, North Dakota State University, Fargo,
ND 58102, USA
*Corresponding author: smithada@usc.edu
Abstract
Wastewater-based epidemiology (WBE) is an effective and versatile tool for
monitoring communal viral load. In addition, WBE can enhance clinical surveillance by
identifying potential under-testing communities. Here we report the results of WBE
surveillance of Los Angeles County, CA, one of the largest and most populated
metropolises in the United States. We collected weekly samples of 24-hour flow-
weighted composite influent from five wastewater treatment plants for 44 weeks.
Wastewater SARS-CoV-2 levels were quantified using RT-qPCR targeting the CDC
recommended nucleocapsid genes N1 and N2. During our study, wastewater SARS-
CoV-2 levels in Los Angeles County experienced two large spikes, once during July-
August 2020 and a second during December 2020-January 2021. Wastewater SARS-
CoV-2 levels peaked at 3.85E+05 N1 gene copies/L and 3.79E+05 N2 gene copies/L
during the first spike and 2.55E+06 N1 gene copies/L and 2.15E+06 N2 gene copies/L
during the second spike. Pearson correlation analysis of wastewater SARS-CoV-2
7

levels with clinical data showed strong correlations of r = 0.94, p << 0.01 for N1 and N2.
Further, wastewater SARS-CoV-2 levels from samples collected once a day, over the
course of a week, led clinical data by up to 5 days, which suggests WBE could be used
as an early warning system for rising community infections. Monte Carlo simulations,
using our measured wastewater SARS-CoV-2 dataset, estimated the number of
infected individuals peaked on January 19
th
, 2021 with about 1.25 million active cases.
The estimated total number of infected individuals for the duration of this study was 3.42
million people, which represents 34.2% of the population residing in Los Angeles
County. Interestingly, our estimated number exceeds the cumulative clinical case count
by almost 2 million people. This study demonstrates the utility of WBE to track infection
dynamics within large communities. Further, WBE data can be used in Monte Carlo
simulations to estimate the size of the infected population and complement clinical data-
based models to better understand the disease impact on different communities.
2.1 Introduction
With the rapid and extensive spread of coronavirus disease 2019 (Covid-19)
across the United States, large-scale monitoring tools, such as wastewater-based
epidemiology (WBE), are gaining interest as a potential solution to help identify and
track the spread of the virus, severe acute respiratory syndrome coronavirus 2 (SARS-
CoV-2). WBE is an attractive candidate for communal monitoring of SARS-CoV-2 as it
offers public health systems an economical, non-invasive, and readily deployable tool
that complements in-person clinical nasal/saliva testing
1,2
. While in-person testing
provides invaluable diagnostic power to understand the size and demographic of the
8

infected population, issues with potential sampling bias, reporting delays, and costs are
exacerbated when the capacity to provide in-person testing is limited. WBE offers many
advantages that mitigate the shortcomings of in-person testing. For example,
wastewater samples are taken from communal waste streams that draws from all
contributing individuals with equal probability and therefore avoids the potential
sampling bias and inconvenience stemming from the opt-in nature of in-person testing.
Further, WBE requires a relatively short turn-around time, with a number of workflows
offering samples to results in under six hours, regardless of community size, versus one
to three days for clinical nasal/saliva tests
3
. With faster time to results, WBE can identify
outbreaks or emerging hotspots in near real-time. Wastewater samples can be used in
a variety of assays such as RT-qPCR to quantify biomarkers for SARS-CoV-2 (e.g.,
nucleocapsid genes N1 and N2) and metagenomic sequencing to assess variant
composition. In the wake of evolving SARS-CoV-2 variants, RT-qPCR and
metagenomic sequencing may be used on stored or fresh wastewater samples to
assess the SARS-CoV-2 variant composition in specific communities over time
4,5
.
To date, SARS-CoV-2 has been detected in wastewater from countries around
the world including, but not limited to, Turkey
6
, Germany
7
, Netherlands
8
, Australia
9
,
Japan
10
, and the United States
11
. While previous WBE studies have demonstrated that
trends in regional clinical cases of Covid-19 are reflected in wastewater
12
, attempts to
back-calculate the infected population size from wastewater SARS-CoV-2 data face
multi-faceted and complex uncertainties stemming from inconsistent viral loading and
system-specific factors of each wastewater collection system. For instance, while
studies estimate around 48% of the SARS-CoV-2 infected population shed detectable
9

levels of the virus in their stool
13
, reports of SARS-CoV-2 in stool samples range
between 10
2
-10
8
virus copies/gram of feces. Further, estimates place the viral shedding
period range between 1-33 days, and in rare cases up to 47 days
12,13
. Apart from viral
loading, the impact of system-specific factors ( e.g., combined or separate collection
systems, wastewater strength, sewer travel time, and temperature) on the detection and
interpretation of the data remains an area needing further refinement
14
. For a detailed
assessment of the various variables and uncertainties in back-calculating infected
population size with measured wastewater SARS-CoV-2 data please refer to the review
on this topic by Li et al., 2021
15
.
In this study, we showcase the utility of WBE by monitoring wastewater SARS-
CoV-2 load from five wastewater treatment plants (WWTPs) in Los Angeles County, CA
over the course of 44 weeks. The five WWTPs sampled in this study vary in size, with
an average influent flow rate between 9.87 to 941 million liters per day (MLD) and
service populations between 150,000 to 4,000,000 residents each. Collectively, the
sampled WWTPs serve over 9 million people, which accounts for more than 90% of the
population in Los Angeles County. More notably, Los Angeles County experienced the
highest prevalence of new Covid-19 cases of all US cities during the winter surge
(November-January), with 27,906 new cases per day at its peak. Wastewater SARS-
CoV-2 load was quantified using the CDC recommended N1 and N2 gene targets within
the SARS-CoV-2 genome via RT-qPCR. Wastewater SARS-CoV-2 data along with
parameters for stool load, viral load, and percentage of infected population for viral
shedding were used to estimate the number of infected individuals via Monte Carlo
simulations. System-specific factors (SSF) for each of the sampled WWTPs were
10

explored for possible influencing variables that may improve the implementation of
large-scale WBE efforts.
2.2 Materials and Methods
2.2.1 Sample Collection and Enveloped Virus Concentration
Twenty four-hour flow-weighted composite influent samples were collected from
each WWTP on a weekly basis starting from May 12
th
, 2020 to March 10th, 2021
(Figure 1, Table 1), except for Joint Water Pollution Control Plant which was collected
twice per week. All samples were kept on ice during transport from the collection site to
the lab and immediately heat treated at 60°C for 90 min to inactivate the SARS-CoV-2
virus. Samples were concentrated in duplicates, where one replicate underwent RNA
extraction, and the second replicate was stored at -80°C. The second replicate was
processed when reverse transcription or qPCR inhibition was detected during RT-qPCR
runs. A total sample volume of 50 mL was transferred from each sample into respective
sterile 50 mL falcon tubes (Thermo Fisher Scientific, PA). An adsorption and elution
method was used to concentrate the SARS-CoV-2 virus in each 50 mL sample
16,17
.
Briefly, each 50 mL sample was conditioned to an approximate 25 mM MgCl2 final
concentration. Conditioned samples were filtered through a 0.45 um HA membrane filter
(Whatman) using a 250 mL vacuum filtration setup (Sterlitech, WA). HA filters trap
enveloped viruses based on charge repulsion.

11

Figure 2. 1: (A) Map of LA county and sampled sewersheds. Stars represent approximate location of
each sampled WWTP. Orange = Hyperion, Blue = Joint Water Pollution Control Plant, Purple= Long
Beach Water Reclamation Plant, Red = San Jose Creek Water Reclamation Plant, and Yellow = Whittier
Narrows Water Reclamation Plant. (B) Diagram of the general workflow used in this study to concentrate
and measure wastewater SARS-CoV-2.

A
B
12

Table 2. 1: Collections Sites, Flow Rate, People Served, and Sample Frequency. Composite influent
samples

2.2.2 RNA Extraction
Total RNA was extracted from processed HA membrane filters using zirconium
bead beating and the Maxwell 16 LEV simply RNA, blood purification kit (Promega,
Madison WI) according to the manufacturer’s instructions. HA filtration was chosen as
the virus concentration method based on its ease of use, low cost, and consistent high
recovery of gene targets compared to other widely used virus concentration methods
18
.
Preliminary assessment of our process workflow efficiency, using spiked bovine
coronavirus (BCoV) seeded in heat-inactivated wastewater samples, achieved an
Utility Average
Flow Rate
(MLD)
Population
Serviced
Frequency
(per week)
Hyperion
(HYP)
941 4,000,000 1
San Jose
Creek
(SJ)
36.8 992,000 1
Joint
Water
(JW)
305 3,500,000 2
Whittier
Narrows
(WN)
9.87 150,000 1
Long
Beach
(LB)
16.5 250,000 1
13

extraction efficiency of 64.7% ± 1.86% (n=3), which is comparable to the 65.7% ± 23%
recovery efficiency reported by Ahmed et al. using murine hepatitis virus
17
. Similarly, our
results are comparable to the recovery efficiency reported for HA filtration studies using
BCoV as the proxy virus which range between 27.3-60.5% ± 22.2%
18,19
. Although many
studies use a proxy virus of known titer in each sample to adjust the measured SARS-
CoV-2 data, we feared adjusting our measured wastewater SARS-CoV-2 data with the
recovery efficiency of a proxy virus may introduce more biases than it corrects for
20
.
Therefore, we here report all SARS-CoV-2 data in its unadjusted form
2.2.3 SARS-CoV-2 Quantification
RNA extracts were analyzed using the SARS-CoV-2 RT-qPCR detection kit
(Promega, Madison, WI). Each reaction consisted of 2.5 µL of RNA extract, 5 µL of 2x
Go Taq Wastewater Master Mix, 0.5 µL of 20X Prime/Probe/Internal Amplification
Control Mix, 0.2 µL, GoScript Reverse Transcriptase (50X), and 1.8 µL of nuclease free
water, for a total final reaction volume of 10 µL. Each RT-qPCR reaction is designed to
be a triplex assay targeting either the N1 or N2 gene (Hex), Pepper Mild Mottle Virus
(reverse transcription inhibition control, FAM), and an internal DNA template (qPCR
inhibition control, Cy5). All RT-qPCR reactions were done in triplicates and carried out
using the LightCycler 96 instrument (Roche). All RT-qPCR runs included a no template
control, where 3 µL nuclease free water were used in place of the RNA extract and an
RNA positive control containing the N and E genes (Promega, Madison WI). Only RT-
qPCR runs with non-detects in the no template control were used for downstream
analysis. Standard curves were generated by analyzing five ten-fold serial dilutions of
the manufacturer provided linear dsDNA template (1x10
5
–1x10
1
gene copies/reaction),
14

which contains partial fragments of the N1, N2, and E gene. Please refer to SI Table 2
for a complete list of names and concentrations of the primes and probes used for the
RT-qPCR assay. The thermocycling condition for all RT-qPCR runs were 1 cycle at
42°C for 15 min and 95°C for 2 min, followed by 40 cycles at 95 °C for 3 sec and 62°C
for 30 seconds. Cq values above 40 were considered invalid and not used for
downstream analysis. Only data within the acceptable Cq range for each inhibition
control and a qPCR reaction efficiency value greater than or equal to 90% were used for
downstream analysis. RT-qPCR reaction efficiencies were generated by the software
provided with the LightCycler instrument (Roche)The SARS-CoV-2 copy number in
gene copies per L for each wastewater sample was calculated using the RT-qPCR data
for N1 and N2 multiplied by the concentration factor used in this study. The
concentration factor was determined based on equation (1).
(𝑬𝒍𝒖𝒕𝒊𝒐𝒏 𝑽𝒐𝒍𝒖𝒎𝒆 )
(𝑨𝒏𝒂 𝒍𝒚𝒛𝒆𝒅 𝑽𝒐𝒍𝒖𝒎𝒆 )
×
(𝟏 , 𝟎𝟎𝟎 𝒎𝑳 )
(𝑺𝒂𝒎𝒑𝒍𝒆 𝑽𝒐𝒍𝒖𝒎𝒆 )
= 𝑪𝒐𝒏𝒄𝒆𝒏𝒕𝒓𝒂𝒕𝒊𝒐𝒏 𝑭𝒂𝒄𝒕𝒐𝒓 (𝟏 )
Limit of detection for our RT-qPCR assay was assessed using the manufacturer
provided RNA template, which encodes the N and E genes. A set of ten-fold serial
dilutions were prepared from 1x10
5
–1x10
0
gene copies/reaction and mixed with the
reaction mix as previously described. The limit of detection was called when only 60%
(3 out of 5) of the prepared concentration successfully amplified.
2.2.4 Process flow and inhibition control assessment
Process flow and amplification control were simultaneously assessed along with
each sample through a triplex assay design provided by the SARS-CoV-2 RT-qPCR
15

detection kit (Promega, WI). Process flow control was assessed through targeting the
PMMoV RNA using the Cy5 channel of the RT-qPCR instrument. While some studies
advocate using PMMoV as a normalizing gene, our study quantified PMMoV to
establish a Cq value range to serve as a benchmark for subsequent reactions. Our
preliminary assessment of PMMoV in our wastewater samples from each sampled
WWTP established a baseline Cq range of 16-19. Cq shifts of more than 2 or negative
PMMoV results signify signs of potential reverse transcription inhibition, qPCR inhibition,
or lab processing error according to the manufacturer’s instructions. Amplification
control was assessed through the HEX channel by quantifying a 435bp linear DNA
template that is pre-mixed in the 20X Primer/Probe/Internal Amplification Control tube.
Amplification inhibition is defined by a Cq shift of greater than 3 compared to the no
template control according to the manufacturer’s instructions. Sample dates flagged for
potential inhibition were processed a second time using the replicate samples stored in
the -80°C. Replicate RNA extracts were prepared in three concentrations, undiluted,
diluted 1:2, and diluted 1:5 before performing the RT-qPCR assay as previously
described. Dilutions were carried out using nuclease free water.
2.2.5 Variant Analysis with RT-ddPCR
SARS-CoV-2 variant analysis was carried out on a QX200 AutoDG Droplet
Digital PCR (ddPCR) instrument (Bio-Rad, CA). Primers and probes used to detect and
discriminate SARS-CoV-2 Alpha variant from the parental Wuhan strain were obtained
from the GT dPCR
TM
Mutation Detection Assays, Validated Kit (GT Molecular, CO).
Positive controls for Alpha and Wuhan strains were provided by GT dPCR
TM
Mutation
Detection Assays, Validated kit. Design specifics for the primer and probes from the GT
16

dPCR
TM
Mutation Detection Assays, Validated Kits are not available and therefore not
included in this study. Reaction mixtures consisted of 5µL RNA extract, 5.5µL of 2x
Super Mix, 1µL of GT-Primer-Probe Solution 4-Plex, 2.2µL Reverse Transcriptase,
1.1µL of DTT, and 7.2µL of nuclease free water, in a final volume of 22 µL. Each
sample was analyzed in duplicates. Non template controls were included in each RT-
ddPCR run, where nuclease free water was used in place of the RNA extract.
Thermocylcing conditions for all RT-ddPCR reactions were 1 cycle at 50°C for 60, 95°C
for 10 min, followed by 45 cycles at 94°C for 30 sec and 60°C for 60 seconds, then 1
cycle at 98°C for 10 min, and 4°C for 30 min. RT-ddPCR results were analyzed using
QuantaSoft Analysis Pro software v.1.0 (Bio-Rad, CA).
2.2.6 Contributing Cases by Sewershed
GIS shapefiles of the WWTP sewersheds were obtained from the Los Angeles
County Sanitation District (LACSD) and Los Angeles Sanitation and Environment
(LASAN). The GIS shapefiles were overlaid onto the shapefile for Countywide Statistical
Areas (CSAs), acquired from the Los Angeles County GIS Hub. Using QGIS, the
distribution of these CSAs within their corresponding WWTP sewershed was
determined. Using this distribution, the proportion of a CSA lying within a given WWTP
sewershed was used as a proxy for the portion of cases that CSA contributed to the
viral loading of the WWTP. Using data made available by the Los Angeles County
Department of Public Health, the new cases per day per CSA were distributed to each
of the five WWTPs sampled. New cases per day represents the number of positive tests
for the samples collected on each specific date. Only cases within these five
sewersheds were counted.
17

2.2.7 System-Specific Factor Analysis
Influent wastewater quality was obtained from LACSD and Hyperion Water
Reclamation Plant. Sampled WWTPs were ranked in respect to each SSF between 1-5,
where 1 = highest value and 5 = lowest value. The SSF assessed in this study were
total suspended solids (TSS), biochemical oxygen demand5 (BOD5), population
serviced, influent flow rate, new cases, and new cases per influent flow rate for the
duration of the study. Our set of SSFs were chosen based on the completeness of their
dataset and availability across all sampled WWTPs. Spearman ranked correlation was
used to assess the relationship between each SSF-ranked list to a separate ranked list
in decreasing Pearson coefficient strength between wastewater SARS-CoV-2 levels and
sewershed specific clinical data.
2.2.8 Monte Carlo Simulation
The number of SARS-CoV-2 infected individuals was estimated via Monte Carlo
simulations using Oracle Crystal Ball (Version Number 11.1.2.4.850, Redwood City CA).
The equation used for our model is presented below
9,21
:
NIF = Rq*Q/(F*Rf*P)
NIF = Estimated number of infected people, Rq = Viral load in wastewater (virus
copies/L), Q = Wastewater flow rate (L/day), Rf = Viral load in stool (virus copies/g
stool), F = Daily production of stool per capita (g stool/capita-day), and P = % of SARS-
CoV-2 infected individual who shed the virus in their stool. While previous studies
suggest wastewater SARS-CoV-2 RNA follows a first order decay rate in the sewer
lines, a variable parameter for SARS-CoV-2 RNA loss was not included in the model
18

due to the limited number of studies on the decay constant for the sampled sewersheds.
Further, the rate loss parameter of wastewater SARS-CoV-2 is further confounded by
the report of relatively high wastewater SARS-CoV-2 RNA signal persistence in
untreated wastewater
13
. In one study, wastewater SARS-CoV-2 RNA signal achieved a
persistence of T90 = 3.3 to 33 days in untreated wastewater, depending on the
wastewater SARS-CoV-2 concentration
13
, where T90 is the time it takes to lose 90% of
the maximum signal. Our omission of the parameter for viral RNA loss simplifies real
world conditions and shifts our estimation toward the conversative side. Our estimation
can be refined as more representative data emerges.
Data for the 24-hour averaged wastewater flow rate (Q) was provided to us by
LACSD and the Hyperion Water Reclamation Plant. Stool viral load (Rf) (virus copies/g)
had a log-normal distribution with a mean of 7.18 and standard deviation of 0.67 in
log10
22
. Daily stool production per capita (F) (g/capita-day) had a log-normal distribution
with a mean of 149 and standard deviation of 95 according to reports for high income
earning countries
23
. The percentage of SARS-CoV-2 infected individuals who shed the
virus in their stool (P) was simulated as a uniform distribution from 0.29 to 0.55
24–26
.
Estimates for each sample point are based on 50,000 simulations.
The median value was used to represent each estimate due to the right skewed
probability distribution of the two input variables, Rf and F. The median value is more
robust toward extreme values drawn from the input variables compared to the mean
value. Further, the 95% confidence interval (CI) for our estimates were determined
through bootstrapping the model using the parameter of 200 experiments and 1,000
simulations each.
19

2.3 Results and Discussion
2.3.1 SARS-CoV-2 Detected in sampled WWTPs
A total of 250 composite influent samples were collected during the period of this
study (Table 1). Overall, samples from San Jose Creek Water Reclamation Plant (SJ),
Hyperion Water Reclamation Plant (HYP), Joint Water Pollution Control Plant (JW),
Whittier Narrows Water Reclamation Plant (WN), and Long Beach Water Reclamation
Plant (LB) contained a positivity rate of over 80% for SARS-CoV-2 (202 positive
detections/250 samples). As most of the non-detects occurred during the early stage of
the pandemic, non-detects were simply omitted from downstream analysis to prevent
the possibility of overcorrection. Samples from WN and LB contained more frequent
non-detects for SARS-CoV-2 in the early phase of the study compared to the remaining
WWTPs. A potential explanation for our observation could be smaller WWTPs require a
greater degree of SARS-CoV-2 penetrance within their serviced population for the
excreted viral load to rise above the limit of detection of our RT-qPCR assay (1,200
copies/L). In agreement, samples from SJ, HYP, and JW consistently contained higher
levels of SARS-CoV-2 than WN and LB due to the larger population serviced by the
former three WWTPs. Consistent with previous reports, Pearson correlation analysis of
the quantified N1 and N2 genes within each WWTP was strongly correlated to each
other r = 0.90-0.99, p < 0.05
27
(SI Figure 1).
20

2.3.2 Wastewater SARS-CoV-2 levels show strong sensitivity and
correlation to reported new cases of Covid-19 in Los Angeles County
Wastewater SARS-CoV-2 levels (virus copies/L) for Los Angeles County were
obtained by using the mean value from all sampled WWTPs for each date. In addition,
all datasets were smoothed using a non-parametric regression to reduce background
noise. Smoothing was done in XLSTAT (Addinsoft) using the built-in Brown’s linear
exponential smoothing function, with 500 iterations and self-optimized alpha value.
Since measurable wastewater SARS-CoV-2 levels are known to vary due to several
external factors such as variable viral load, wastewater flows, wastewater quality,
collection protocol, and lab processing method, statistical smoothing was used to
denoise the imperfect and variable datasets and highlight general patterns. Smoothing
of the datasets improved the correlation coefficient between wastewater SARS-CoV-2
levels and new cases from rraw=0.87 N1 and 0.88 N2, p << 0.01 to rsmooth =0.94 N1 and
0.94 N2, p << 0.01. Consistent with previous studies, our measured wastewater SARS-
CoV-2 levels reflect reported new cases of Covid-19 to its corresponding regions
8,28
.
The two major surges of new Covid-19 cases in Los Angeles County at the beginning of
June and November 2020 coincide with elevated levels of wastewater SARS-CoV-2
during the same period (Figure 2). Interestingly, during both summer (June 2020) and
winter (November 2020) Covid-19 outbreaks, wastewater SARS-CoV-2 levels showed
high sensitivity toward the accumulation and decline of reported average daily new
cases. From June 2
nd
to July 28
th
, 2020, wastewater SARS-CoV-2 levels increased by
roughly 64,000 virus copies/L (530%) from a corresponding increase of 1,722 average
daily new cases (120%). Similarly, from November 3
rd
to January 18, 2021, wastewater
SARS-CoV-2 levels increased by more than 2 million virus copies/L (>1,500%) from an
21

increase of 9,943 averaged daily new cases (500%). As the infection rate fell following
the summer peak, July 28
th
to August 11
th
, 2020, wastewater SARS-CoV-2 levels
decreased by 210,000 virus copies/L (54.6%) and then 85,000 virus copies/L (48.7%) in
the first and second week, respectively, which correspond to a decline in the average
daily new cases by 434 (13.8%) and then 363 average daily new cases (13.3%) over
the same period. The steep increase in wastewater SARS-CoV-2 levels in response to
rising daily new cases is likely due to the fecal SARS-CoV-2 pattern, where fecal viral
titers peak in the first 1-2 weeks after the onset of symptoms followed by a steady
decline in the following weeks
29
. Therefore, wastewater SARS-CoV-2 levels are
sensitive to rising and falling community infections rates, which makes WBE suitable for
community-level surveillance.
Interestingly, a recent WBE study in Southern California overlapped with our
sampling period by approximately 30 weeks and allows a unique opportunity to
compare WBE data from two independent labs sampling from the same WWTPs (JW
and SJ)
28
. Despite significantly different sample processing methods between the two
labs (direct extraction, Qiamp Viral RNA kit, and recovery adjusted data versus HA
filtration, Maxwell 16 LEV simply RNA blood purification kit, and unadjusted data),
reported wastewater SARS-CoV-2 levels from both labs captured the summer and
winter surge of Covid-19 cases in Los Angeles County. Moreover, the overall similar
wastewater SARS-CoV-2 trend in both studies suggest temporal measured wastewater
SARS-CoV-2 data can be presented in its unadjusted form if proper process and
inhibition controls are done.
22

2.3.3 Wastewater SARS-CoV-2 levels strongly correlates to the average
new cases by contributing sewershed
Data from the Covid-19 Dashboard for Los Angeles County was separated by
CSA to obtain a representative dataset for the sewershed corresponding to each
sampled WWTP. Again, all datasets were smoothed using a non-parametric regression
analysis to reduce background noise. Pearson correlation analysis for both raw and
smoothed datasets were strongly correlated to the averaged new cases of each
respective sewershed. In every case, smoothing increased the correlation coefficient
(Pearson rraw = 0.45-0.85, p << 0.01 and Pearson rsmooth = 0.76-0.95, p <<0.01, Figure
2). Interestingly, the average daily new cases across all sampled sewersheds rose and
fell around similar dates during the summer and winter peaks. The near-synchronous
wave of averaged daily new cases in the sampled communities could likely be
explained by the highly infectious nature of SARS-CoV-2
30
, the extensive traveling
between neighboring communities within Los Angeles County, and large centralized
WWTPs that service multiple zip codes. While previous WBE studies conducted in
areas with smaller decentralized WWTPs could provide greater geographic resolution
due to their smaller sewershed size, here we demonstrate large centralized WWTPs
can provide insights for communal SARS-CoV-2 load in sewersheds serving up to 4
million people without compromising the sensitivity to reflect corresponding clinical case
counts.
We acknowledge a level of uncertainty in splitting LA County Covid-19
Dashboard data by CSA for comparison between sewershed-specific case counts and
its respective wastewater SARS-CoV-2 level. Our approximation of the sewershed
23

boundaries is confounded by the interconnected sewer lines and broadly defined
sewershed borders within LACSD. For instance, LACSD WWTPs that are upstream of
JW (Whittier Narrows, San Jose Creek, and Long Beach Water Reclamation Plant) are
designed to divert excess wastewater to JW as part of their overflow management
practice and could dilute or enrich the wastewater SARS-CoV-2 level in JW. However,
from a size perspective, JW treats an averaged influent flow rate of 305 MLD, which is
one or two orders of magnitude greater than the average influent flow rate of SJ (36.8
MLD), LB (16.5 MLD), and WN (9.87 MLD), which could make measured wastewater
SARS-CoV-2 levels in JW robust toward dilution or enrichment effects from upstream
LACSD WWTPs, under non-extreme conditions. Further, wastewater in LACSD is
designed to flow southwest which adds confidence that wastewater SARS-CoV-2 levels
for WN, SJ, and LB, which are upstream of JW, should be largely unaffected by non-
regional wastewater.
24

Figure 2. 2: (A) Time series analysis of the averaged viral copies/L for all sampled WWTPs and averaged
new cases rate for Los Angeles County over the duration of this study. (B-F) Time series analysis of
SARS-CoV-2 viral copies/L for each sampled WWTP and the averaged new positive tests for the
respective sewershed. Dark red line represents quantified N1 data, pink line represents quantified N2
data and blue bar chart represents the moving seven-day average of daily new cases A= Total, B=HYP,
C=JW, D=SJ, E=LB, and F= WN. Error bars represent the standard deviation of the measured N1 or N2
value. Pearson coefficient r represents the correlation between the N1 or N2 gene copies/L and regional
clinical data.
2.3.4 Comparison of daily viral load to 20-day case count could help
identify under testing communities
We compared the daily viral load to a moving 20-day case count in each
sewershed to assess the variability between measured wastewater SARS-CoV-2 levels
and clinical data over the course of this study. Our motivation for this comparison stems
from the length of our WBE work, which allowed us to compare the relationship between

25

our WBE data to different stages of the pandemic as public perception and participation
toward in-person testing evolved over the course of this 44-week study. Compared to
WBE, in-person testing is likely more susceptible to public perception and preparedness
such as education, testing availability, and fatigue. Although the reported SARS-CoV-2
shedding period in stool ranges from 1-47 days, the general consensus places the
mean shedding period to be around 12-20 days
31,32
. Therefore, we used the number of
reported cases within a 20-day window to represent the viral shedding population. Viral
load for each WWTP was calculated by multiplying measured virus copies/L by
averaged influent flowrate (MLD) on each date. In general, larger sewersheds (HYP,
JW, and SJ) exhibited a lower ratio and variability compared to smaller sewersheds (LB
and WN, Figure 3A). The observed pattern could be the result of large sewersheds
having a higher testing capacity and participation than smaller sewersheds. Limited
testing capacity or participation poses a risk of under reporting or overlooking disease
outbreaks.
For greater temporal resolution, we divided our analysis into three segments,
June-August 2020, September-October 2020, and November 2020-March 2021 (Figure
3B-D). Segments were divided based on timeframes that allow comparisons of
individual outbreaks. During June-August 2020, the ratio and variability of viral load to
the moving 20-day case count for each sampled sewershed was inversely related to the
size of the serviced population (Figure 3B), which is consistent with our previous
explanation. For HYP, JW, SJ, LB, and WN, the median ratios were 0.8, 8.3, 9.3, 10.3
and 21.8, respectively. During September-October, the median ratio of daily viral load to
the moving 20-day case count increased by 80-320% for HYP, JW, SJ, and WN and
26

2,700% for LB compared to the ratios during June-August 2020. While the increased
ratios for HYP, JW, SJ, and WN during September-October is likely attributed to the
decline in administered Covid-19 tests in September compared to August (SI Figure 2),
the stark increase of 2,700% in LB suggests potential under reporting of a resurgence in
LB following the summer peak. On closer examination, the moving 20-day case count in
LB steadily declined in the first half of September and reached a low of 141 cases on
September 17
th
. Afterwards, the moving 20-day case count reached a peak of 156
cases on October 1
st
, which represents an increase of 15 new cases from September
17
th
and could likely be considered insignificant. However, over a similar period the viral
load for LB increased from < 9,000 virus copies/day to 1.2 million virus copies/day
(September 15
th
-October 6
th
, 2020). Based on the discrepancy between the 20-day
case count and the daily viral load seen in LB, future WBE implementations could be
used to validate clinical data for sewersheds where testing capacity and participation
are likely strained. From November-March 2021, the median ratio of daily viral load to
the 20-day case count decreased by 83.3%, 81.2% and 37.4% for JW, LB, and WN,
respectively, compared to ratios from September-October. The decreased median ratios
for JW, LB, and WN is likely due to the record number of Covid-19 tests administered
during November-March 2021. Surprisingly, the median ratio of daily viral load to the
moving 20-day case count increased by 39.3% and 18.2% for SJ and HYP,
respectively, which could be due to the rate of new infections exceeding the increased
Covid-19 tests performed in SJ and HYP. While the relationship between viral load and
infected individuals is highly variable, monitoring the daily viral load to the moving 20-
day case count in each sewershed over time could help identify under testing
27

communities. We acknowledge that our analysis would require additional studies for
validation. For instance, sudden shifts to high fecal-viral load SARS-CoV-2 variants in
the infected population my cause an inflation to the ratio of wastewater SARS-CoV-2 to
20-day case count. Ongoing data for community-specific SARS-CoV-2 variant profile
would help refine our analytical approach. Further, our assumption for a fixed 20-day
period to estimate the number of active cases is subject to change as additional high-
quality data for the recovery window recovery continues to surface.
Figure 2. 3: Box plot analysis of daily viral load to the moving 20-day case count for each sampled
sewershed. (A) The duration of the study May 2020-March 2021 (B) May- August 2020 (C) September-
October 2020 (D) November-March 2021. Box represents the median, 25
th
, and 75
th
percentile. The
whiskers represent the largest and smallest values and outliers are shown as circles. Error bars represent
standard deviation.
28

2.3.5 Clinical cases normalized by influent flow rate, TSS, and BOD5 are
potential factors that may influence Wastewater SARS-CoV-2 Correlation
to clinical data
Interestingly, the Pearson correlation strengths between the wastewater SARS-
CoV-2 level of each sampled WWTP and its respective regional new cases did not
follow any obvious trend. While HYP and JW are the two largest WWTPs sampled (941
and 305 MLD, respectively), the corresponding Pearson coefficient (rsmooth = 0.90 N1
/0.87 N2 and 0.82 N1/ 0.80 N2, p << 0.01) ranks 2
nd
and 4
th
in our dataset. Whereas SJ
being the third largest WWTP sampled (36.8 MLD) displayed the strongest Pearson
coefficient (rsmooth= 0.94 N1/ 0.91 N2, p << 0.01). Further, WN (9.83 MLD) treats
significantly less wastewater than HYP and JW, but its Pearson coefficient (rsmooth= 0.85
N1/ 0.88 N2 p << 0.01) is comparable to HYP and stronger than JW. To assess SSFs
that may influence the correlative strength of wastewater SARS-CoV-2 levels to clinical
data, we ranked our sampled WWTPs, in decreasing Pearson coefficient, and
compared it to individual lists of sampled WWTPs, each ranked in respect to one SSF.
The SSF examined in this study were TSS, BOD5, serviced population size, averaged
influent flowrate, new cases, and new cases per averaged influent flowrate. In total,
seven ranked lists were created, one list ranked by decreasing Pearson coefficient and
six lists each ranked by one SSF in decreasing value. Spearman ranked correlations
were used to assess the relationship between the Pearson coefficient ranked list to the
SSF ranked lists. Based on our assessment, new cases normalized by average influent
flow rate showed a very strong relationship to the Pearson coefficient ranking between
wastewater SARS-CoV-2 levels and regional new cases (r = 0.9 and p < 0.05, Table
2B). While TSS and BOD5 showed strong relationship (r = 0.06, p > 0.05, Table 2B) to
29

the Pearson coefficient ranking between wastewater SARS-CoV-2 levels and regional
new cases, our dataset of fived sampled WWTPs did not reach statistical significance of
p-value < 0.05. However, given previous reports of higher wastewater SARS-CoV-2
detection in biosolids compared to primary influent
33
and a positive association of
coronavirus survival rate with biosolids and organic matter concentrations
34
, we
encourage future studies to conduct a similar analysis with a greater number of
WWTPs. Overall, correlative strength between wastewater SARS-CoV-2 and clinical
data is most strongly influenced by the ratio of new cases per averaged influent
flowrate, whereas TSS and BOD5 levels showed potential strong relationship. While we
acknowledge that the p values for the TSS and BOD5 analysis presented here are
greater than the commonly accepted p = 0.05, we believe increasing levels of TSS and
BOD5 would facilitate the adsorption of wastewater SARS-CoV-2 to particulates. An
increased fraction of adsorbed SARS-CoV-2 could increase the measurable wastewater
SARS-CoV-2 fraction and improve the correlative strength between wastewater SARS-
CoV-2 levels to regional new cases. Further, concentrating solids in primary influent
may be a viable alternative for WWTPs that serve areas with a low number of Covid-19
cases or a low ratio of Covid-19 cases per averaged influent flow rate.

30

Table 2. 2: (A) Summary table representing the ranking of the Pearson Correlation coefficient of all
sampled WWTPs to regional clinical cases. 1 = highest Spearman correlation coefficient and 5 = lowest
Spearman correlation coefficient (B) Summary table of the Spearman rank correlation coefficient of each
SSF to the Pearson Correlation ranking of each sampled WWTPs.

Utility
Clinical and
Wastewater
Correlation Ranking
Correlation of SSF to
Clinical and Wastewater
Correlation Ranking

SJ

1
New Cases/MLD
r = 0.9, p < 0.05

JW

2
BOD5
r = 0.6, p > 0.05

WN

3
TSS
r = 0.6, p > 0.05

HYP

4
New Cases
r = 0.3, p > 0.05

LB

5
Population
r = 0.1, p > 0.05
Average MLD
r = 0.1, p > 0.05

2.3.6 Daily Wastewater sample could lead clinical data by up to five days
Although previous studies report wastewater and primary sludge SARS-CoV-2
levels lead clinical and hospitalization cases by 0-6 day
27,35
, our dataset showed no
significant signs of lead time when offsetting measured wastewater SARS-CoV-2 levels
to the daily new cases for JW. The initial assessment was done using the wastewater
SARS-CoV-2 data for JW for each month and by complete dataset. We hypothesized
that the short 0-2 day lead time in wastewater SARS-CoV-2 may have been lost when
A)
B)
31

sampling once or twice a week. To examine the potential lead time of wastewater
SARS-CoV-2 to reported daily cases, we collected daily composite influent samples
from JW from August 16
th
, 2020 to August 22
nd
, 2020 and compared our results to the
daily new COVID-19 cases corresponding to the JW sewershed. The comparison was
done by aligning measured SARS-CoV-2 levels to the reported daily new cases on the
same day of sample collection or by offsetting the two datasets from 1-5 days. Pearson
correlation analysis of daily wastewater SARS-CoV-2 levels to same day daily new
cases showed a correlation coefficient of rsmooth=0.81 N1/0.69 N2, p < 0.05 N1 and p >
0.05 N2. However, offsetting the reported daily new cases by 5 days improved the
correlation coefficient to rsmooth = 0.96 N1/0.96 N2, p < 0.005 (Figure 4A and 4B).
Interestingly, offsetting the reported daily new cases by 2 days also improved the
correlation coefficient to rsmooth = 0.91 N1/ 0.92 N2, p < 0.05, whereas offsetting the
reported daily new cases by 1, 2, and 3 days showed decreased or mixed improvement
to the correlation coefficient (rsmooth = 0.78 N1/ 0.77 N2 p < 0.05, rsmooth = 0.78 N1/ 0.82
N2 p < 0.05, rsmooth = 0.72 N1/ 0.71 N2 p > 0.05, respectively. In agreement with our
hypothesis, future WBE implementation using daily sampling could offer improved
sensitivity in detecting early rises in community infections compared to weekly samples.
While the increased correlation coefficient through temporal offsetting the daily reported
new cases by 2 and 5 days highlights the susceptibility of our comparison to spurious
correlations, future studies can limit this shortcoming by increasing the number of
samples collected over a greater number of WTTPs.
32

Figure 2. 4: Comparison of daily wastewater SARS-CoV-2 virus copies/L (N1=red and N2= grey) from
JW to its respective daily new cases (blue). (A) Represents same day comparison. (B) Represents
measured SARS-CoV-2 virus copies/L compared to the daily new cases five days later. Error bars
represent standard deviation.
2.3.7 Variant Analysis
At the time of our study, reports of SARS-CoV-2 variants began to emerge
around the United States
36
. Specifically, the novel B.1.1.7 variant or UK variant was
reported to be more infectious than the original Wuhan strain
36–38
. The highly infectious
nature of the UK variant caused some experts to estimate its dominant circulation in the
United States by March 2020
36
. To showcase the utility of WBE to assess communal
variant composition, select replicate extracts from HYP and JW were analyzed for the
UK variant by quantifying the presence of the point mutation 501Y and del 69-70
37
.
Each sample was benchmarked to the presence of the Wuhan strain by measuring the
targets N501 and HV69-70. Variant analysis was performed using reverse transcription
droplet digital PCR (RT-ddPCR) due to its reported lower limit of detection than RT-
qPCR. Despite reports of increasing prevalence of the UK variant in sequenced clinical
samples in the United States (https://www.gisaid.org/hcov19-variants/), we did not
detect the UK variant in any of our samples. Although we did not detect the UK variant
33

in our select samples, we cannot conclude the absence of any UK variant infections in
Los Angeles County during the time of our study. While the relative abundance of the
UK variant among sequenced SARS-CoV-2 strains reached a high of 14% around
March 1, 2021, the averaged new cases of Covid-19 in Los Angeles County on March
1, 2020 fell to 1.16E+03 cases, which marks a 90% reduction from its all-time high. The
viral load from the UK infections within the declining case count in March 2021 were
likely further diluted in communal waste streams to below the limit of detection of our
RT-ddPCR assay (10 copies/reaction). Quantified Wuhan variant concentrations were
comparable to our qPCR results (data not shown), which suggests both RT-qPCR and
RT-ddPCR are suitable assays to measure wastewater SARS-CoV-2. For future
studies, we recommend metagenomic sequencing of wastewater samples as a
preliminary step to assess all possible variant strains and their relative abundance to
better customize the selection of variant targets for each geographical region.
2.3.8 Estimated Infected Population from Monte Carlo Simulations Exceed
the Reported Clinical Cases by more than 200 Percent
Monte Carlo simulations were used to estimate the median number of infected
individuals (NIF) for Los Angeles County and for the sewershed of each sampled
WWTP. A total of 50,000 simulations were performed for each data point. Here we
report a summary of the median NIF obtained from the simulations. Summary results
from the simulation are presented in Table 3. Full results from the simulation can be
found in SI Table 2. We used a conservative 20-day window to estimate the cumulative
NIF for Los Angeles County. Our simulation estimates the peak NIF for Los Angeles
County to be 1.25 million (95% CI: 4.91E+05 - 3.55E+06), which occurred on January
34

19
th
, 2021 (Table 3). In contrast, the cumulative reported case count for the 20-days
leading up to January 19
th
was 2.08E+05 people, which falls short of the simulated NIF
by a factor of 6. As expected, the simulated NIF are well above the reported case
counts and demonstrates the potential utility of WBE to consistently sample a far
greater population than in-person testing.
Our estimate raises the prevalence of Covid-19 from 14.5% to 34.2% for Los
Angeles County. Using a 20-day window, we estimate the cumulative NIF in Los
Angeles County to be around 3.42 million people (95% CI: 7.91E+05 - 9.12E+06) from
the period between May 2020 to March 2021. Our estimate exceeds the 1.45 million
reported cases over the same duration by more than a factor of 2. While we
acknowledge there are multiple uncertainties in our prediction, we believe our estimate
to be on the conservative side due to the inevitable viral loss through the sewer
networks and sample processing that were not factored into the simulation. For
instance, the HA filtration method used to concentrate wastewater samples in this study
has a reported extraction efficiency between 27.3-60.5% ± 22.2%
18,19
. Adjusting our
estimate by a fixed extraction efficiency of 60.5% would increase the cumulative NIF
estimate to 5.7 million people. However, since we did not track extraction efficiency
during this study our estimate remains 3.42 million people (95% CI: 7.91E+05 -
9.12E+06). Interestingly, our Covid-19 prevalence estimate of 34.2% closely resembles
the 37.5% prevalence estimate reported by Los Angeles Health Services and the 30-
50% prevalence estimate by previous reports
39,40
.

35

Table 2. 3: Summary values of the peak median NIF, 95% CI, and peak date from the Monte Carlo
simulations using 50,000 simulations per datapoint.
Sewershed Peak Median
NIF
95% Confidence
Interval
Peak Date
SJ 5.17E+04 (1.89E+04 -1.41E+05) December 22
nd
, 2020
HYP 1.17E+06 (4.33E+05 - 3.23E+06 January 19
th
, 2021
JW 9.04 E+04 (3.54E+04 - 2.39E+05) December 22
nd
, 2020
LB 8.45E+03 (3.20E+03 - 2.46E+04) January 5
th
, 2021
WN 1.43E+03 (5.27E+02 - 3.69E+03) December_29
th
, 2020

Aggregate 1.25E+06 (4.91E+05 - 3.55E+06) January 19
th
, 2021

Using the same 20-day window, we estimate the total number of infected
individuals for SJ, HYP, JW, LB, and WN to be 1.21E+04, 4.4E+05, 1.53E+04,
5.94E+03, and 3.57E+03 per 100,000 people, respectively. Based on our simulation,
the highest NIF per 100,000 people occurred in the catchment area belonging to HYP,
followed by JW, SJ, LB, and WN (Figure 5). In contrast to our simulation, the reported
public health data (separated by sewershed) ranked SJ to have the highest rate of
infection, followed by HYP, WN, JW, and LB. We believe the disagreement in our
simulated ranking versus the public health data stems from the uncertainty surrounding
the true infected population. If the true infected population lies closer to 1.45 million
people or 14.4% of the population in Los Angeles County, then the adjusted case rate
should have hot spots and follow the public health data. However, if the true infected
population is far more prevalent and closer to 3.42–5.7 million people or 34.2- 57% of
the population in Los Angeles County, then the infections are likely less concentrated in
one community and the adjusted rate would likely mirror the ranking for serviced
population size.
36

Figure 2. 5: Comparison of the reported adjusted cases per 100,000 people (light blue) vs. simulated NIF
per 100,000 people (dark blue). Error bars represent the 95% CI.
We acknowledge that the model used for our estimate requires further calibration
and should not be taken as an absolute calculation for the infected population. Further,
variations in lab processes, sample handling experience, and SARS-CoV-2 variant
profile could significantly alter the input parameters and the estimated output. However,
despite the uncertainties listed, we believe Monte Carlo simulations using measured
wastewater SARS-CoV-2 can be beneficial to the improvement of future models as
more specific and representative data emerges. Monte Carlo simulations can be
another option in the suite WBE tools that can be used to complement clinical data.
2.4 Conclusion
In this comprehensive study, we demonstrate the utility of WBE for SARS-CoV-2
using various approaches ranging from RT-qPCR to statistical estimation. We first
quantified wastewater SARS-CoV-2 levels in Los Angeles County and showcased the
effectiveness of WBE to track regional SARS-CoV-2 load for communities ranging from
37

150,000 to 4,000,000 people. While measured wastewater SARS-CoV-2 concentrations
varied from sample-to-sample, smoothing the dataset was effective in denoising
background variability to reveal the general trend of wastewater SARS-CoV-2 levels
over time. Further, wastewater SARS-CoV-2 trends measured from daily samples may
lead reported daily new cases by 2 or 5 days. SSF such as dilution factors should be
considered for future WBE implementation as the ratio of new cases to averaged
influent flowrate is a better indicator of correlation strength than serviced population
size, averaged influent flowrate, TSS, and BOD5. Measured wastewater SARS-CoV-2
data were used to estimate the median NIF via Monte Carlo simulations. Our simulation
estimates the largest active infection population peaked on January 19
th
, 2021 with 1.25
million NIF (95% CI: 4.91E+05 - 3.55E+06) and a total of 3.42 million NIF (95% CI:
7.91E+05 - 9.12E+06) between the period of May 2020-March 2021 (34.2% of Los
Angeles County’s population). In comparison to the reported case count, our simulated
infected population exceeds the reported number of cases by almost 2 million people.
Conflicts of interest
We do not have any conflicts of interest to declare.
Acknowledgements
The figure for the table of contents entry was adapted from “Quantifying SARS-CoV-2
Virions in City Wastewater”, by BioRender.com (2021). Retrieved from
https://app.biorender.com/biorender-templates. Figure 2B was created with
Biorender.com.
38

References
1 S. Arora, A. Nag, J. Sethi, J. Rajvanshi, S. Saxena, S. K. Shrivastava and A. B.
Gupta, Sewage surveillance for the presence of SARS-CoV-2 genome as a useful
wastewater based epidemiology (WBE) tracking tool in India, Water Sci. Technol.,
, DOI:10.2166/wst.2020.540.
2 D. Polo, M. Quintela-Baluja, A. Corbishley, D. L. Jones, A. C. Singer, D. W.
Graham and J. L. Romalde, Making waves: Wastewater-based epidemiology for
COVID-19 – approaches and challenges for surveillance and prediction, Water
Res., , DOI:10.1016/j.watres.2020.116404.
3 Overview of Testing for SARS-CoV-2 (COVID-19) | CDC,
https://www.cdc.gov/coronavirus/2019-ncov/hcp/testing-overview.html, (accessed
23 June 2021).
4 A. Crits-Christoph, R. S. Kantor, M. R. Olm, O. N. Whitney, B. Al-Shayeb, Y. C.
Lou, A. Flamholz, L. C. Kennedy, H. Greenwald, A. Hinkle, J. Hetzel, S. Spitzer, J.
Koble, A. Tan, F. Hyde, G. Schroth, S. Kuersten, J. F. Banfield and K. L. Nelson,
Genome sequencing of sewage detects regionally prevalent SARS-CoV-2
variants, MBio, , DOI:10.1128/mBio.02703-20.
5 B. S. Beresford-Jones and Y. Shao, Nat. Rev. Microbiol., 2021, 19.
6 B. A. Kocamemi, H. Kurt, S. Hacıoglu, C. Yaralı, A. M. Saatci and B. Pakdemirli,
medRxiv, 2020.
7 M. Döhla, G. Wilbring, B. Schulte, B. M. Kümmerer, C. Diegmann, E. Sib, E.
Richter, A. Haag, S. Engelhart, A. M. Eis-Hübinger, M. Exner, H. Streeck and R.
M. Schmithausen, medRxiv, 2020.
8 G. Medema, L. Heijnen, G. Elsinga, R. Italiaander and A. Brouwer, Presence of
SARS-Coronavirus-2 RNA in Sewage and Correlation with Reported COVID-19
Prevalence in the Early Stage of the Epidemic in the Netherlands, Environ. Sci.
Technol. Lett., 2020, 7, 511–516.
9 W. Ahmed, N. Angel, J. Edson, K. Bibby, A. Bivins, J. W. O’Brien, P. M. Choi, M.
Kitajima, S. L. Simpson, J. Li, B. Tscharke, R. Verhagen, W. J. M. Smith, J.
Zaugg, L. Dierens, P. Hugenholtz, K. V. Thomas and J. F. Mueller, First confirmed
detection of SARS-CoV-2 in untreated wastewater in Australia: A proof of concept
for the wastewater surveillance of COVID-19 in the community, Sci. Total
Environ., 2020, 728, 138764.
10 A. Hata, R. Honda, H. Hara-Yamamura and Y. Meuchi, medRxiv, 2020.
11 F. Wu, A. Xiao, J. Zhang, K. Moniz, N. Endo, F. Armas, R. Bonneau, M. A. Brown,
M. Bushman, P. R. Chai, C. Duvallet, T. B. Erickson, K. Foppe, N. Ghaeli, X. Gu,
W. P. Hanage, K. H. Huang, W. L. Lee, M. Matus, K. A. McElroy, J. Nagler, S. F.
Rhode, M. Santillana, J. A. Tucker, S. Wuertz, S. Zhao, J. Thompson and E. J.
Alm, medRxiv, 2020.
39

12 J. Weidhaas, Z. T. Aanderud, D. K. Roper, J. VanDerslice, E. B. Gaddis, J.
Ostermiller, K. Hoffman, R. Jamal, P. Heck, Y. Zhang, K. Torgersen, J. Vander
Laan and N. LaCross, Correlation of SARS-CoV-2 RNA in wastewater with
COVID-19 disease burden in sewersheds, Sci. Total Environ., ,
DOI:10.1016/j.scitotenv.2021.145790.
13 S. Gupta, J. Parker, S. Smits, J. Underwood and S. Dolwani, Color. Dis., 2020,
22.
14 P. Foladori, F. Cutrupi, N. Segata, S. Manara, F. Pinto, F. Malpei, L. Bruni and G.
La Rosa, SARS-CoV-2 from faeces to wastewater treatment: What do we know?
A review, Sci. Total Environ., , DOI:10.1016/j.scitotenv.2020.140444.
15 X. Li, S. Zhang, J. Shi, S. P. Luby and G. Jiang, Chem. Eng. J., 2021, 415.
16 E. Haramoto, M. Kitajima, H. Katayama, T. Ito and S. Ohgaki, Development of
virus concentration methods for detection of koi herpesvirus in water, J. Fish Dis.,
, DOI:10.1111/j.1365-2761.2008.00977.x.
17 W. Ahmed, V. J. Harwood, P. Gyawali, J. P. S. Sidhu and S. Toze, Comparison of
Concentration Methods for Quantitative Detection of Sewage-Associated Viral
Markers in Environmental Waters, Appl. Environ. Microbiol., 2015, 81, 2042 LP –
2049.
18 Z. W. LaTurner, D. M. Zong, P. Kalvapalle, K. R. Gamas, A. Terwilliger, T.
Crosby, P. Ali, V. Avadhanula, H. H. Santos, K. Weesner, L. Hopkins, P. A.
Piedra, A. W. Maresso and L. B. Stadler, Evaluating recovery, cost, and
throughput of different concentration methods for SARS-CoV-2 wastewater-based
epidemiology, Water Res., 2021, 197, 117043.
19 W. Ahmed, P. M. Bertsch, A. Bivins, K. Bibby, K. Farkas, A. Gathercole, E.
Haramoto, P. Gyawali, A. Korajkic, B. R. McMinn, J. F. Mueller, S. L. Simpson, W.
J. M. Smith, E. M. Symonds, K. V Thomas, R. Verhagen and M. Kitajima,
Comparison of virus concentration methods for the RT-qPCR-based recovery of
murine hepatitis virus, a surrogate for SARS-CoV-2 from untreated wastewater,
Sci. Total Environ., 2020, 739, 139960.
20 R. S. Kantor, K. L. Nelson, H. D. Greenwald and L. C. Kennedy, Environ. Sci.
Technol., 2021, 55.
21 S. W. Hasan, Y. Ibrahim, M. Daou, H. Kannout, N. Jan, A. Lopes, H. Alsafar and
A. F. Yousef, Detection and quantification of SARS-CoV-2 RNA in wastewater
and treated effluents: Surveillance of COVID-19 epidemic in the United Arab
Emirates, Sci. Total Environ., 2021, 764, 142929.
22 F. X. Lescure, L. Bouadma, D. Nguyen, M. Parisey, P. H. Wicky, S. Behillil, A.
Gaymard, M. Bouscambert-Duchamp, F. Donati, Q. Le Hingrat, V. Enouf, N.
Houhou-Fidouh, M. Valette, A. Mailles, J. C. Lucet, F. Mentre, X. Duval, D.
Descamps, D. Malvy, J. F. Timsit, B. Lina, S. van-der-Werf and Y. Yazdanpanah,
Clinical and virological data of the first cases of COVID-19 in Europe: a case
series, Lancet Infect. Dis., , DOI:10.1016/S1473-3099(20)30200-0.
40

23 C. Rose, A. Parker, B. Jefferson and E. Cartmell, The Characterization of Feces
and Urine: A Review of the Literature to Inform Advanced Treatment Technology,
Crit. Rev. Environ. Sci. Technol., 2015, 45, 1827–1879.
24 K. S. Cheung, I. F. N. Hung, P. P. Y. Chan, K. C. Lung, E. Tso, R. Liu, Y. Y. Ng,
M. Y. Chu, T. W. H. Chung, A. R. Tam, C. C. Y. Yip, K. H. Leung, A. Y. F. Fung,
R. R. Zhang, Y. Lin, H. M. Cheng, A. J. X. Zhang, K. K. W. To, K. H. Chan, K. Y.
Yuen and W. K. Leung, Gastrointestinal Manifestations of SARS-CoV-2 Infection
and Virus Load in Fecal Samples From a Hong Kong Cohort: Systematic Review
and Meta-analysis, Gastroenterology, , DOI:10.1053/j.gastro.2020.03.065.
25 Y. Wu, C. Guo, L. Tang, Z. Hong, J. Zhou, X. Dong, H. Yin, Q. Xiao, Y. Tang, X.
Qu, L. Kuang, X. Fang, N. Mishra, J. Lu, H. Shan, G. Jiang and X. Huang,
Prolonged presence of SARS-CoV-2 viral RNA in faecal samples, Lancet
Gastroenterol. Hepatol., 2020, 5, 434–435.
26 W. Wang, Y. Xu, R. Gao, R. Lu, K. Han, G. Wu and W. Tan, Detection of SARS-
CoV-2 in Different Types of Clinical Specimens, JAMA, 2020, 323, 1843–1844.
27 J. Peccia, A. Zulli, D. E. Brackney, N. D. Grubaugh, E. H. Kaplan, A. Casanovas-
Massana, A. I. Ko, A. A. Malik, D. Wang, M. Wang, J. L. Warren, D. M.
Weinberger, W. Arnold and S. B. Omer, Measurement of SARS-CoV-2 RNA in
wastewater tracks community infection dynamics, Nat. Biotechnol., ,
DOI:10.1038/s41587-020-0684-z.
28 Z. Song, R. Reinke, M. Hoxsey, J. Jackson, E. Krikorian, N. Melitas, D. Rosso
and S. Jiang, Detection of SARS-CoV-2 in Wastewater: Community Variability,
Temporal Dynamics, and Genotype Diversity, ACS ES&T Water, ,
DOI:10.1021/acsestwater.1c00119.
29 N. Sethuraman, S. S. Jeremiah and A. Ryo, JAMA - J. Am. Med. Assoc., 2020,
323.
30 F. Elrashdy, E. M. Redwan and V. N. Uversky, Biomolecules, 2020, 10.
31 A. S. van Doorn, B. Meijer, C. M. A. Frampton, M. L. Barclay and N. K. H. de
Boer, Aliment. Pharmacol. Ther., 2020, 52.
32 M. Cevik, M. Tate, O. Lloyd, A. E. Maraolo, J. Schafers and A. Ho, SARS-CoV-2,
SARS-CoV, and MERS-CoV viral load dynamics, duration of viral shedding, and
infectiousness: a systematic review and meta-analysis, The Lancet Microbe, ,
DOI:10.1016/S2666-5247(20)30172-5.
33 K. E. Graham, S. K. Loeb, M. K. Wolfe, D. Catoe, N. Sinnott-Armstrong, S. Kim,
K. M. Yamahara, L. M. Sassoubre, L. M. Mendoza Grijalva, L. Roldan-Hernandez,
K. Langenfeld, K. R. Wigginton and A. B. Boehm, SARS-CoV-2 RNA in
Wastewater Settled Solids Is Associated with COVID-19 Cases in a Large Urban
Sewershed, Environ. Sci. Technol., , DOI:10.1021/acs.est.0c06191.
34 P. M. Gundy, C. P. Gerba and I. L. Pepper, Survival of Coronaviruses in Water
and Wastewater, Food Environ. Virol., , DOI:10.1007/s12560-008-9001-6.
41

35 S. Karthikeyan, N. Ronquillo, P. Belda-Ferre, D. Alvarado, T. Javidi, C. A.
Longhurst and R. Knight, High-Throughput Wastewater SARS-CoV-2 Detection
Enables Forecasting of Community Infection Dynamics in San Diego County,
mSystems, , DOI:10.1128/msystems.00045-21.
36 S. E. Galloway, P. Paul, D. R. MacCannell, M. A. Johansson, J. T. Brooks, A.
MacNeil, R. B. Slayton, S. Tong, B. J. Silk, G. L. Armstrong, M. Biggerstaff and V.
G. Dugan, Emergence of SARS-CoV-2 B.1.1.7 Lineage — United States,
December 29, 2020–January 12, 2021, MMWR. Morb. Mortal. Wkly. Rep., ,
DOI:10.15585/mmwr.mm7003e2.
37 P. Horby, C. Huntley, N. Davies, J. Edmunds, N. Ferguson, G. Medley, A.
Hayward, M. Cevik and C. Semple, NERVTAG note on B.1.1.7 severity, Sage.
38 N. L. Washington, S. White, K. M. Schiabor Barrett, E. T. Cirulli, A. Bolze and J. T.
Lu, medRxiv, 2020.
39 S. Banks, Los Angeles Times, 2021.
40 T. Belin, A. Bertozzi, N. Chaudhary, T. Graves, J. Guterman, C. Jarashow, R. J.
Lewis, J. Marion, F. Schoenberg, M. Shah, J. Tolles, E. Traub, K. Viele and F.
Wu, Projections of Hospital-based Healthcare Demand due to COVID-19 in Los
Angeles County, Los Angeles, 2021.

42

Chapter 3
3. Assessment of Wastewater SARS-CoV-2
Mutation Profile in Los Angeles County
Phillip Wang
a
, Ali Zarei-Baygi
a
, Syeed Md Iskander
b
, Adam L. Smith
a
*
a
Sonny Astani Department of Civil and Environmental Engineering, University of Southern California,
3620 South Vermont Avenue, Los Angeles, California 90089, United States
b
Department of Civil, Construction and Environmental Engineering, North Dakota State University, Fargo,
ND 58102, USA
*Corresponding author: smithada@usc.edu
Abstract
Wastewater-based epidemiology (WBE) shows promise as a tool for inferring
circulating SARS-CoV-2 mutation profile. While sequencing clinical SARS-CoV-2 strains
provide high quality genomic data, performing patient-level sequencing at scale would
not be feasible due to cost and labor constraints. WBE provides a cost-effective and
labor-saving alternative to monitor the circulating SARS-CoV-2 mutation profile. In this
study, we showcased WBE as a surveillance tool for SARS-CoV-2 mutations through
sampling of two WWTPs between July 2020 to March 2021, representing approximately
75% of the population in Los Angeles County. Our wastewater dataset revealed 351
non-redundant mutations during our sample collection period. Of the 351 non-redundant
mutations, only 138 of the 351 non-redundant mutations were found in the sequenced
clinical SARS-CoV-2 genomes collected in Los Angeles County. While our results did
not reveal any variants of concerns, we detected an incomplete and disjointed set of
43

characteristic mutations for B.1.1.7 (Alpha) and B.1.429 (Beta) between the two
WWTPs more than six months before the designation Alpha and Beta.
3.1 Introduction
The coronavirus disease 2019 (COVID-19) pandemic has extended pass its third
year with over 480 million reported cases worldwide
1
. A major driver for the prolonged
pandemic stems from the continued evolution of the causative agent, severe respiratory
syndrome coronavirus-2 (SARS-CoV-2)
2
. SARS-CoV-2 is an enveloped virus with a
positive sense, single stranded RNA genome that is approximately 30 kbp in length
3
.
From the analysis of sequenced SARS-CoV-2 genomes, the SARS-CoV-2 genomic
mutation rate is estimated to be similar to SARS-CoV-1 (0.8-2.38 x 10
-3

nucleotide/site/year)
4–6
, influenza A (2.3x10
-3
nucleotide/site/year), and Middle East
respiratory syndrome coronavirus (1.12 x10
-3
nucleotide/site/year)
7
. Since the start of
the pandemic, the original SARS-CoV-2 genome has spawned numerous variant
strains, with some strains harboring mutations that increase infectivity, immune evasion,
or disease severity compared to the parental viral strain. SARS-CoV-2 variants are
labeled as variants of concern (VOC) when the hallmark mutations have been shown to
increase transmission and/or disease severity
8
.
Wastewater surveillance (WWS) of SARS-CoV-2 variants provides a high-
throughput and low-cost method to assess the relative abundance of known and novel
mutations and strains within sewersheds
9
. Although high quality genome coverage can
be obtained from sequencing clinical samples, attempts to sequence samples from a
significant portion of the infected population would be hindered by high manual and
44

financial resources
10
. Furthermore, sequencing of clinical viral samples requires a
COVID-19 confirmation in the patient, which biases the sequencing results to represent
predominately hospitalized patients and not patients with mild or no symptoms. While
over seven million SARS-CoV-2 genomes from clinical samples have been sequenced
around the world and made publicly available on GISAID
11
, the global number of
sequenced SARS-CoV-2 genomes from clinical settings represent less than 3% of the
480 million reported Covid-19 cases to date. As a complementary surveillance tool,
WWS can be used for near real-time monitoring of the dynamic SARS-CoV-2 variant
profile excreted in stools. Assessment of wastewater SARS-CoV-2 variants may offer a
more complete picture of the circulating variant diversity compared to sequencing
clinical samples. Moreover, sample collection at WWTPs can represent thousands to
millions of people per sample, with unbiased collection from asymptomatic and
symptomatic individuals. With 15-24% of the American population lacking broadband
connection, WWS may play a vital role in monitoring infection dynamics in communities
with low income or health literacy
12

To date, wastewater SARS-CoV-2 variant studies have employed either
amplicon-based tiling of the SARS-CoV-2 genome
13,14
or untargeted metatranscriptomic
sequencing followed by viral enrichment
15
to obtain reads for the downstream
bioinformatics analysis. While both methods have been used to identify wastewater
SARS-CoV-2 mutations, wastewater settings likely contain fragmented SARS-CoV-2
genomes and enzyme inhibitors which could lead to under-reporting of circulating
SARS-CoV-2 mutations. In addition, wastewater SARS-CoV-2 genomes are a collection
of viruses from many humans and potentially animal contributors which makes viral
45

haplotyping and lineage calling beyond dominant strains challenging. Despite difficult
sample matrices and potential gaps in the wastewater SARS-CoV-2 genomes,
previously reported wastewater SARS-CoV-2 variant profiles have identified several
mutations corresponding to clinical samples in addition to novel mutations and variant
strains missing from clinical databases
13,15,16
. The ability to detect novel mutations in
wastewater settings provides valuable insight for understanding the circulating variant
diversity which could help monitor rising VOC and aid in genomic epidemiology efforts
to better understand the spread and evolution of SARS-CoV-2
17
.
Here we demonstrate the utility of WWS to monitor wastewater SARS-CoV-2
mutations using samples collected from two wastewater treatment plants (WWTPs) in
Los Angeles between July 2020 to March 2021. The two WWTPs collectively serve over
7.5 million people (75% of Los Angeles County population). Our analysis detected 351
non-redundant mutations from 24 samples of which 138 out of 351 mutations were
present in sequenced clinical samples from Los Angeles County. Further, we
showcased the viability of sampling biosolids for SARS-CoV-2 variant analysis as a
potential alternative to composite-influent samples. Overall, our result showed a
relatively stable variant profile in Los Angeles County during our sample collection
period, which could point to the slow evolutionary driving force pre-dating the vaccine-
induced selective pressure.
46

3.2 Material and Methods
3.2.1 Wastewater Samples Selection and RNA Extraction
A total of 18 twenty four-hour flow-weighted composite influent samples and 6
primary biosolid grab samples were selected for sequencing from a previously collected
sample set
18
. Samples were collected from Hyperion Water Reclamation Plant (Hyp)
and Joint Water Pollution Control Plant (JW) from July 7
th
, 2020 to February 16
th
2021
(Figure1, Table 1). Selected samples were chosen to represent the circulating SARS-
CoV-2 virus during and in-between the two major peaks of COVID-19 cases in Los
Angeles County (Figure 1B and 1C). All samples were concentrated using an
adsorption and elution method
19
. Briefly, 50 mL (influent) or 5 mL (biosolid) of each
sample was transferred into respective sterile 50 mL falcon tubes (Thermo Fisher
Scientific, PA). Samples were then conditioned to an approximate 25 mM MgCl2 final
concentration. Conditioned samples were filtered through a 0.45 µm HA membrane filter
(Whatman) using a 250 mL vacuum filtration setup (Sterlitech, WA). Total RNA was
extracted from processed HA membrane filters using zirconium bead beating and the
Maxwell 16 LEV simply RNA, blood purification kit (Promega, Madison WI) according to
the manufacturer’s instructions and eluted in 50 µL of nuclease free water.
47

Figure 3. 1: (A) Map of Los Angeles County with the serviced sewershed area shaded in green
(Hyperion) and blue (Joint Water). The star icon represents the approximated location of the wastewater
treatment plant. (B and C) Overlayed line and bar chart depicting the SARS-CoV-2 viral load/L and the
respective 7-day average case count for each sewershed. Red line represents the N1 gene
concentration, pink line represents the N2 gene concentration, and the blue bars represent the 7-day
average of new case counts for Hyp or JW. Dark blue arrows signify the dates of the sequenced samples.

Table 3. 1: Overview of the average flow rate, population serviced, and number of samples collected for
Hyperion and Joint Water

Utility
Average
flow rate
(MLD)
Population
serviced
Number of
composite
influent samples
Number of
biosolid
samples
Hyperion Water
Reclamation Plant
(Hyp)
941 4,000,000 12 3
Joint Water Pollution
Control Plant
(JW)
305 3,500,000 12 3

A

48

3.2.2 Sequencing and Data processing
RNA extracts were sent to the University of Minnesota Genomics Center for amplicon
tiling using ARTIC primers version 3 and Illumina MiSeq (PE 2x300 bp). Sequenced
reads in FASTW formatted files were uploaded to usegalaxy.org and processed using
the SARS-CoV-2 variant analysis protocol
20
. Briefly, raw reads were filtered using
Fastp
21
to remove adaptors, ambiguous bases, low quality reads (Phred score <30),
and short fragments (<50 bp). Filtered reads were mapped against the complete
genome of the Wuhan SARS-CoV-2 reference strain NCBI NC_045512.2, using the
default settings of BWA-MEM
22
. Alignment statistics were generated using SAMtools
23
.
Consensus sequences and variants were generated using iVar
24
. Variants were called
using lofreq
25
. Called variants with an allele frequency of less than 0.05% and 10
supporting reads were removed from the final dataset of reported variants.
3.2.3 Clinical Data
The mutation profile of 5,592 clinical samples collected from Los Angeles between Dec
19
th
, 2019 to Feb 16
th
, 2021 were downloaded from covidcg.org. All analyses of the
mutational profile were done in XLstat (Addinsoft).
3.3 Results
3.3.1 Overview of wastewater sequencing quality
A total of 24 wastewater samples were sent to the University of Minnesota
Genomics Center for sequencing (18 twenty-four composite influent samples and 6
biosolid samples). SARS-CoV-2 concentrations were assessed before library
49

preparation using RT-qPCR with the CDC recommended N1 and N2 primers. The
median CT value for the samples before sequencing was 30.6 (min =27.2, max = 39.7,
and standard deviation = 3.46). Although 12 out of 24 samples were above the
recommended CT value of 30 for amplicon tiling (cite), wastewater samples are
complex matrices containing a mixture of unknown inhibitors that may inflate reported
CT values. Consistent with our explanation, all sequencing runs yielded more than 3
billion bases with an average read quality between Q34.9 to Q35.58. The average
reference genome coverage with at least 10X coverage was 69.48% ± 3.99%.
3.3.2 Mutation Type and Frequency for JW and Hyperion (WW and
Biosolid)
In total, we detected 351 non-redundant nucleotide mutations from 24
wastewater samples (18 influent and 6 biosolid samples). Detected nucleotide
mutations with an allele frequency (AF) less than 5% and less than 10 supporting reads
were filtered out from the final dataset. The filtered mutations were unevenly distributed
between five categories; nonsynonymous (63.05%), synonymous (26.98%), frameshift
(6.74%), stop gained (2.64%), and codon deletion (0.59%) (Table 2). The higher impact
mutations (frameshift, codon-deletions, and stop gained) represented less than 10% of
the total mutations identified, which was likely due to the increased risk of lethality to the
virus stemming from large-scale amino acid changes. The relative abundance of each
mutation type was consistent across all samples and in previous studies conducted with
wastewater and clinical samples
13,15
.
50

Table 3. 2: Table showing the distribution of mutation types within influent and biosolid dataset.

Figure 3. 2: Distribution of mutations within the SARS-CoV-2 genes (A). Normalized distribution of
mutation frequency among the SARS-CoV-2 genes (B). Blue bars represent influent data and orange
bars represent biosolid data.
Next, we examined the distribution of non-redundant mutations within each gene.
Overall, the distribution of non-redundant mutations was nearly identical between
influent and biosolid samples (Figure 2A). The three genes with the highest non-
redundant mutation counts were ORF1ab (211 influent and 140 biosolid mutations),
followed by N (42 influent and 22 biosolid mutations), and S (31 influent and 16 biosolid
mutations). While ORF1ab contained the highest number of non-redundant mutations,
the gene length of ORF1ab is 21,290 bp, which represents more than 70% of the
SARS-CoV-2 genome. However, the N gene (1,260 bp) contained the second highest
0
1
2
3
4
5
6
ln( Mutation count)
Influent
Bio-solid
0
5
10
15
20
25
30
35
40
Mutations/ 1 kbp
Influent
Bio-solid
B)
A)
51

number of non-redundant mutations despite being close to three times shorter than the
S gene (3,822 bp). To examine mutation rate, we normalized our dataset to obtain
mutation rate per kbp gene length (Figure 2B). Interestingly, genes N and ORF10 had
the highest mutation rate in influent and biosolid samples, respectively. Despite the high
non-redundant mutation count within genes ORF1ab and S, the mutation rate for both
genes were less than 10 mutations per kbp and toward the bottom half among the
SARS-CoV-2 genes. Our data suggest the N gene may be more prone to mutations
compared to genes ORF1ab and S during our collection period.
A deeper assessment of the nucleotide substitution class revealed a strong bias
for C →T and G →T mutations (Figure 3). Here, thymine (T) is used to represent uracil
(U) in the original SARS-CoV-2 RNA genome since the RNA virus was reversed
transcribed to cDNA before mapped to the reference genome, NCBI NC_045512.2.
Mutation frequency for each nucleotide class was normalized by the abundance of A, T,
G, or C within the original SARS-CoV-2 genome to obtain a normalized mutation
frequency per kbp. Interestingly, C →T transitions were strongly favored within
synonymous mutations, whereas G →T transversions were most dominant within non-
synonymous mutations. Along with the bias for C →T and G →T mutations, we also
detected a strong asymmetrical mutation between C →T versus T →C and G →T versus
T →C. Further, the asymmetry between C →T versus T →C and G →T versus T →G was
amplified in non-synonymous mutations compared to synonymous mutations. While
previous studies suggest the abnormal rate of C →T transitions are driven by
apolipoprotein B mRNA editing enzyme, a catalytic polypeptide-like (APOBEC) family of
proteins within the host
26
, no characterized RNA-editing enzymes are known to drive
52

G →T transversions. Our data suggests the mutational bias for C →T transitions and
G →T transversions would lead to circulating SARS-CoV-2 genomes with lower GC
content overtime. However, the fitness advantage and mutational consequence of the
observed biases for the evolution of SARS-CoV-2 is unclear and warrants further
study
27
.

Figure 3. 3: Distribution of nucleotide substitution class within the synonymous mutations (A) and non-
synonymous mutations (B). Light blue bars represent influent data and blue bars represent biosolid data.
3.3.3 Comparison between Influent and Biosolid Samples
To assess the sensitivity and consistency to detect wastewater SARS-CoV-2
variants between influent and biosolid sample types, we compared the dataset between
3 influent and 3 biosolid samples from each WWTP collected during the month of Nov
and Dec, 2020 (Figure 4A-B). Influent samples contained 296 mutations (147 Hyp and
139 JW) and biosolid samples contained 272 mutations (162 Hyp and 110 JW). While
our influent dataset contained more unique mutations compared to our biosolid dataset,
meaningful conclusions are difficult to draw given the difference in processed volume. In
total, influent and biosolid samples shared 98 unique mutations for Hyp and 77 unique
mutations for JW. The mean AF of the shared mutations were 42% and 48% for Hyp
B A
53

and JW influent samples, respectively and 0.45 and 52% for Hyp and JW biosolids
samples, respectively. For both WWTPs, the mean AF of the shared mutations were 3-
4% higher in biosolid samples compared to influent samples. Interestingly, the shared
mutations were more frequently detected in biosolid samples compared to influent
samples for both Hyp and JW. In influent samples, 11 mutations for Hyp and 3
mutations for JW were detected in all sampling dates, whereas in biosolids samples, 33
mutations for Hyp and 23 mutations for JW were detected in all sampling dates. The
higher detection frequency of the shared mutations in biosolid samples compared to
influent is likely due to the adsorption of SARS-CoV-2 to particulates in the biosolid
28,29
.
Communities with lower COVID-19 case counts or SARS-CoV-2 gene copies/L could
consider assessing wastewater SARS-CoV-2 variant profile using biosolids instead of
composite-influent samples. Further, the high solids retention time (SRT) of 15-30 days
in the primary clarifier could allow grab samples to be viewed as composite samples,
which would reduce sampling time and cost compared to twenty-four composite influent
samples. To our knowledge, this is the first report to compare the consistency and mean
AF for wastewater SARS-CoV-2 variants between influent and biosolid samples. Our
data suggest biosolids may provide more consistent detection of prevailing mutations
compared to composite-influent samples. We acknowledge the limited sampling points
in our comparison and suggest future studies to examine the consistency and mean AF
of shared mutations between influent and biosolid samples collected across a greater
number of dates and WWTPs.

54

Figure 3. 4: Heatmap comparison of SARS-CoV-2 variants between composite influent and biosolid
samples for (A) Hyperion Water Reclamation Plant (Hyp) and (B) Joint Water Pollution Control Plant
(JW).
3.3.4 Top mutations in JW and Hyperion samples overtime
To assess the wastewater SARS-CoV-2 variant profile over time, we examined
the AF of recurring mutations across our sampling timeline. Variant profile analysis was
carried out separately for influent and biosolid samples for each respective WWTP. We
defined recuring mutations as a minimum of 7 out of 9 samples for influent samples and
a more stringent 3 out of 3 samples for biosolid samples. A total of 100 mutations (75
Hyp and 25 JW) in the influent samples and 110 mutations (73 Hyp and 37 JW) in the
biosolid samples met our cutoff criteria (Figure 5A-B). Overall, the variant profile of the
recuring mutations showed relative stability throughout our sampling period with no
55

clear trend toward enrichment or purging of specific mutations. We hypothesize the
relative stability of the wastewater SARS-CoV-2 variant profile could be due to the low
vaccinated population in Los Angeles during our sample collection period, July 2020 to
March 2021
30
. Previous studies have suggested infectivity-strengthening mutations as
the primary driver for SARS-CoV-2 evolution during the early stages of the pandemic
before the vaccination campaign in December 2020. Mutations such as S:D614G which
increases the infectivity of SARS-CoV-2
31,32
provides an evolutionary advantage and
could be the reason behind its dominance in nearly all sequenced genomes to date.
However, given the mutation rate of 0.8 to 2.3 x 10
-3
nucleotide/site/year
4–6
, the random
probability of novel beneficial mutation over-taking existing variants is low without
additional selective pressure
33
. Consistent with this hypothesis, we detected just 12
unique mutations in the S gene from all 24 wastewater samples over the course of 8
months. The strong correlation between vaccination rate and vaccine-resistant
mutations
34
highlights the risk of low diversity in vaccine targets and rate of vaccination
among the general population. Interestingly, the increasing vaccination rate among
people ages 12+ in March, 2021
35
coincided with the increased presence of the VOC
B.1.1.7 within sequenced clinical samples. The combination of vaccine-induced
selective pressure and high-transmission rate could have primed the population for
novel variants with vaccine-resistant mutations to establish dominance within the
infected population.
56

Figure 3. 5: Heatmap with dendrogram of most frequently detected allelic variants within composite
influent samples. (A) Hyperion Water Reclamation Plant (Hyp) and (B) Joint Water Pollution Control Plant
(JW).
3.3.5 Comparison of WW mutations to clinical data
To assess the consistency of variant surveillance between wastewater and clinical
data, we compared our dataset to 5,592 sequenced clinical samples from Los Angeles
County between the period Dec 15
th
, 2019, to February 16
th
, 2021. Of the 351 unique
mutations in our wastewater dataset, 138 out of 351 mutations (39.3%) were found in
sequenced clinical samples. The relative abundance of the 138 unique mutations ranged
from 0.017% to 99.28% within the sequenced clinical samples. While a total of 8,674
unique mutations were present in the sequenced clinical samples, only 312 unique
mutations occurred at or above 0.05% relative abundance. Overall, we did not find a
correlation between clinical mutation counts and wastewater supporting reads for the
57

shared mutations (r= 0.012, p=0.89). The lack of correlation between the two datasets
could be due to the lack of sample size within clinical settings or variable viral load and
degradation of the SARS-CoV-2 genome in WWTPs. Further research into the
relationship between the two datasets could increase our understanding of SARS-CoV-2
genome evolution across the world.
3.3.6 Mutations relating to variants of concern and variants of interest
A key interest for wastewater surveillance is the ability to assess the relative
abundance of characteristic mutations pertaining to VOC and variants of interest (VOI).
Of the previously identified VOC/VOI, the emergence of variant B.1.1.7 (Alpha) and
B.1.429 (Cali.20C) within Los Angeles County clinical samples lies closest to our
sample collection period. Variant B.1.1.7 contains 23 characteristic mutations, of which
7 out of the 23 mutations were detected in our dataset, S: A570D, S: D614G, S: P681H,
S: T716I, ORF8:Y73C, N: R203K, and N: G204R (Table 4). Mutations S: D614G, and
ORF8:Y73C were among the most frequently detected mutations, appearing in 22 out of
24 samples. Mutation S: A70D appeared in 18 out of 24 samples. Interestingly,
mutations N: R203K and N: G204R appeared in 5 out of 12 samples in just Hyp
samples and mutations S:P 681H and S: T716I occurred in 2 out of 9 and 4 out of 9
samples, respectively, in only JW samples. The disjointed detection of the B.1.1.7
related mutations between Hyp and JW samples adds evidence to the sequential
acquisition of mutations within circulating SARS-CoV-2 genomes that may have later
recombined to become a VOC/VOI
36
. Similarly, we detected mutation ORF3: Q57H and
N: T205I associated with B.1.351 variant (Beta), but not the remaining characteristic
mutations.
58

Although B.1.1.7 was documented in the U.S. by Jan 15
th
, 2020
37
, we did not
detect the hallmark S: H69del mutation for B.1.1.7. While Hyp and JW effectively serve
more than 75% of the population in Los Angeles County, we cannot rule out the
possibility of detectable B.1.1.7 mutations in the remaining sewersheds. However, our
data suggests very low levels of B.1.1.7 in Los Angeles County during our sample
collection period. Consistent with our hypothesis, the frequency of S: del 69/70 within
Los Angeles clinical samples was 18 out of 701 (0.64%) sequenced samples between
Jan 1
st
, 2021, to Feb 16
th
, 2021 (covidgc.org). Despite several reports of rising B.1.429
detection within clinical samples in California in late 2020, our analysis only detected
one mutation, S: L452R (AF = 0.14), associated with B.1.429. The inconsistency
between wastewater and clinical data could be partly due to the biases from limited
sample size and infection clusters affecting clinical surveillance and the higher detection
limit troubling WWS. While clinical settings have a theoretical lower limit of detection for
specific mutations than WWS, attempting to routinely collect, sequence, and analyze a
representative portion of the infected population would be highly labor intensive and
cost prohibitive. Further, clinical samples are biased toward symptomatic patients and
fail to capture the relative abundance of the true SARS-CoV-2 diversity. Therefore, a
complementary wastewater and clinical surveillance and cross referencing between the
two datasets would enhance attempts to monitor mutations of interest among the
general population.

59

Table 3. 3: Summary table of the VOC related mutations detected in our analysis and their relative
abundance within wastewater clinical dataset. The last column contains the earliest detection data of the
respective mutation within the wastewater dataset.

3.4 Conclusion
Our analysis of 24 wastewater samples (18 composite-influent and 6 biosolid
samples) collected from two WWTPs in Los Angeles County during July 7
th
, 2020 to
February 16
th
, 2021 revealed 351 unique mutations. Of the 351 unique mutations, 213
mutations were not present in sequenced clinical samples for Los Angeles. Our dataset
contained an incomplete and disjointed set of characteristic mutations for variant B.1.1.7
and B.1.429 that were detected months before the designation of the variant strains,
which suggests the possibility of recombination among circulating SARS-CoV-2 variants
leading to the emergence one or more VOC. Our wastewater dataset contained
significant discrepancies compared to the clinical data, which could be due to the limited
60

size of the sequenced clinical samples or higher detection limit in wastewater settings.
We believe there is a need for increased wastewater variant surveillance to complement
clinical surveillance as new evolutionary pressures (such as new types of vaccines) are
introduced to the general population.

61

References
1 WHO Coronavirus (COVID-19) Dashboard | WHO Coronavirus (COVID-19)
Dashboard With Vaccination Data, https://covid19.who.int/info, (accessed 2 April
2022).
2 N. D. Rochman, Y. I. Wolf, G. Faure, P. Mutz, F. Zhang and E. V. Koonin,
Ongoing global and regional adaptive evolution of SARS-CoV-2, Proc. Natl. Acad.
Sci. U. S. A., , DOI:10.1073/pnas.2104241118.
3 C. Cao, Z. Cai, X. Xiao, J. Rao, J. Chen, N. Hu, M. Yang, X. Xing, Y. Wang, M. Li,
B. Zhou, X. Wang, J. Wang and Y. Xue, The architecture of the SARS-CoV-2
RNA genome inside virion, Nat. Commun., , DOI:10.1038/s41467-021-22785-x.
4 T. Koyama, D. Platt and L. Parida, Variant analysis of SARS-cov-2 genomes, Bull.
World Health Organ., , DOI:10.2471/BLT.20.253591.
5 R. Sender, Y. M. Bar-On, S. Gleizer, B. Bernshtein, A. Flamholz, R. Phillips and
R. Milo, The total number and mass of SARS-CoV-2 virions, Proc. Natl. Acad. Sci.
U. S. A., , DOI:10.1073/pnas.2024815118.
6 S. Duchene, L. Featherstone, M. Haritopoulou-Sinanidou, A. Rambaut, P. Lemey
and G. Baele, Temporal signal and the phylodynamic threshold of SARS-CoV-2,
Virus Evol., , DOI:10.1093/ve/veaa061.
7 M. Cotten, S. J. Watson, A. I. Zumla, H. Q. Makhdoom, A. L. Palser, S. H. Ong, A.
A. Al Rabeeah, R. F. Alhakeem, A. Assiri, J. A. Al-Tawfiq, A. Albarrak, M. Barry,
A. Shibl, F. A. Alrabiah, S. Hajjar, H. H. Balkhy, H. Flemban, A. Rambaut, P.
Kellam and Z. A. Memish, Spread, circulation, and evolution of the Middle East
respiratory syndrome coronavirus, MBio, , DOI:10.1128/mBio.01062-13.
8 Tracking SARS-CoV-2 variants, https://www.who.int/en/activities/tracking-SARS-
CoV-2-variants, (accessed 2 April 2022).
9 J. A. Baaijens, A. Zulli, I. M. Ott, M. E. Petrone, T. Alpert, J. R. Fauver, C. C.
Kalinich, C. B. F. Vogels, M. I. Breban, C. Duvallet, K. McElroy, N. Ghaeli, M.
Imakaev, M. Mckenzie-Bennett, K. Robison, A. Plocik, R. Schilling, M. Pierson, R.
Littlefield, M. Spencer, B. B. Simen, Y. S.-C.-2 G. S. Initiative, W. P. Hanage, N.
D. Grubaugh, J. Peccia and M. Baym, Variant abundance estimation for SARS-
CoV-2 in wastewater using RNA-Seq quantification, medRxiv, 2021,
2021.08.31.21262938.
10 S. Knyazev, K. Chhugani, V. Sarwal, R. Ayyala, H. Singh, S. Karthikeyan, D.
Deshpande, Z. Comarova, A. Lu, Y. Porozov, A. Wu, M. Abedalthagafi, S.
Nagaraj, A. Smith, P. Skums, J. Ladner, T. T.-Y. Lam, N. Wu, A. Zelikovsky, R.
Knight, K. Crandall and S. Mangul, Unlocking capacities of viral genomics for the
COVID-19 pandemic response.
11 GISAID - Initiative, https://www.gisaid.org/, (accessed 2 April 2022).
12 V. Yee, S. S. Bajaj and F. C. Stanford, Building a pandemic supply chain — equity
62

over equality, Nat. Med. 2022, 2022, 1–2.
13 M. Herold, A. F. D’hérouël, P. May, F. Delogu, A. Wienecke-Baldacchino, J. Tapp,
C. Walczak, P. Wilmes, H. M. Cauchie, G. Fournier and L. Ogorzaly, Genome
sequencing of sars-cov-2 allows monitoring of variants of concern through
wastewater, Water (Switzerland), , DOI:10.3390/w13213018.
14 R. Izquierdo-Lara, G. Elsinga, L. Heijnen, B. B. Oude Munnink, C. M. E.
Schapendonk, D. Nieuwenhuijse, M. Kon, L. Lu, F. M. Aarestrup, S. Lycett, G.
Medema, M. P. G. Koopmans and M. De Graaf, Monitoring SARS-CoV-2
circulation and diversity through community wastewater sequencing, the
netherlands and belgium, Emerg. Infect. Dis., , DOI:10.3201/eid2705.204410.
15 A. Crits-Christoph, R. S. Kantor, M. R. Olm, O. N. Whitney, B. Al-Shayeb, Y. C.
Lou, A. Flamholz, L. C. Kennedy, H. Greenwald, A. Hinkle, J. Hetzel, S. Spitzer, J.
Koble, A. Tan, F. Hyde, G. Schroth, S. Kuersten, J. F. Banfield and K. L. Nelson,
Genome sequencing of sewage detects regionally prevalent SARS-CoV-2
variants, MBio, , DOI:10.1128/mBio.02703-20.
16 S. Agrawal, L. Orschler and S. Lackner, Metatranscriptomic Analysis Reveals
SARS-CoV-2 Mutations in Wastewater of the Frankfurt Metropolitan Area in
Southern Germany, Microbiol. Resour. Announc., , DOI:10.1128/mra.00280-21.
17 D. S. Smyth, M. Trujillo, D. A. Gregory, K. Cheung, A. Gao, M. Graham, Y. Guan,
C. Guldenpfennig, I. Hoxie, S. Kannoly, N. Kubota, T. D. Lyddon, M. Markman, C.
Rushford, K. M. San, G. Sompanya, F. Spagnolo, R. Suarez, E. Teixeiro, M.
Daniels, M. C. Johnson and J. J. Dennehy, Tracking cryptic SARS-CoV-2
lineages detected in NYC wastewater, Nat. Commun. 2022 131, 2022, 13, 1–9.
18 P. Wang, A. Zarei-Baygi, C. Sauceda, S. M. Iskander and A. L. Smith, Long-term
surveillance of wastewater SARS-CoV-2 in Los Angeles County, Environ. Sci.
Water Res. Technol., 2021, 7, 2282–2294.
19 E. Haramoto, M. Kitajima, H. Katayama, T. Ito and S. Ohgaki, Development of
virus concentration methods for detection of koi herpesvirus in water, J. Fish Dis.,
, DOI:10.1111/j.1365-2761.2008.00977.x.
20 E. Afgan, D. Baker, B. Batut, M. Van Den Beek, D. Bouvier, M. Ech, J. Chilton, D.
Clements, N. Coraor, B. A. Grüning, A. Guerler, J. Hillman-Jackson, S. Hiltemann,
V. Jalili, H. Rasche, N. Soranzo, J. Goecks, J. Taylor, A. Nekrutenko and D.
Blankenberg, The Galaxy platform for accessible, reproducible and collaborative
biomedical analyses: 2018 update, Nucleic Acids Res., 2018, 46, W537–W544.
21 S. Chen, Y. Zhou, Y. Chen and J. Gu, in Bioinformatics, 2018, vol. 34.
22 H. Li, [Heng Li - Compares BWA to other long read aligners like CUSHAW2]
Aligning sequence reads, clone sequences and assembly contigs with BWA-
MEM, arXiv Prepr. arXiv.
23 H. Li, B. Handsaker, A. Wysoker, T. Fennell, J. Ruan, N. Homer, G. Marth, G.
Abecasis and R. Durbin, The Sequence Alignment/Map format and SAMtools,
63

Bioinformatics, , DOI:10.1093/bioinformatics/btp352.
24 S. Castellano, F. Cestari, G. Faglioni, E. Tenedini, M. Marino, L. Artuso, R.
Manfredini, M. Luppi, T. Trenti and E. Tagliafico, Ivar, an interpretation‐oriented
tool to manage the update and revision of variant annotation and classification,
Genes (Basel)., , DOI:10.3390/genes12030384.
25 A. Wilm, P. P. K. Aw, D. Bertrand, G. H. T. Yeo, S. H. Ong, C. H. Wong, C. C.
Khor, R. Petric, M. L. Hibberd and N. Nagarajan, LoFreq: A sequence-quality
aware, ultra-sensitive variant caller for uncovering cell-population heterogeneity
from high-throughput sequencing datasets, Nucleic Acids Res., ,
DOI:10.1093/nar/gks918.
26 J. Ratcliff and P. Simmonds, Potential APOBEC-mediated RNA editing of the
genomes of SARS-CoV-2 and other coronaviruses and its impact on their longer
term evolution, Virology, , DOI:10.1016/j.virol.2020.12.018.
27 K. Yi, S. Y. Kim, T. Bleazard, T. Kim, J. Youk and Y. S. Ju, Mutational spectrum of
SARS-CoV-2 during the global pandemic, Exp. Mol. Med., , DOI:10.1038/s12276-
021-00658-z.
28 Y. Ye, R. M. Ellenberg, K. E. Graham and K. R. Wigginton, Survivability,
Partitioning, and Recovery of Enveloped Viruses in Untreated Municipal
Wastewater, Environ. Sci. Technol., , DOI:10.1021/acs.est.6b00876.
29 J. Carrillo-Reyes, M. Barragán-Trinidad and G. Buitrón, Surveillance of SARS-
CoV-2 in sewage and wastewater treatment plants in Mexico, J. Water Process
Eng., , DOI:10.1016/j.jwpe.2020.101815.
30 A. A. Capoferri, W. Shao, J. Spindler, J. M. Coffin, J. W. Rausch and M. F.
Kearney, A Pre-Vaccination Baseline of SARS-CoV-2 Genetic Surveillance and
Diversity in the United States., Viruses, , DOI:10.3390/v14010104.
31 J. A. Plante, Y. Liu, J. Liu, H. Xia, B. A. Johnson, K. G. Lokugamage, X. Zhang, A.
E. Muruato, J. Zou, C. R. Fontes-Garfias, D. Mirchandani, D. Scharton, J. P.
Bilello, Z. Ku, Z. An, B. Kalveram, A. N. Freiberg, V. D. Menachery, X. Xie, K. S.
Plante, S. C. Weaver and P. Y. Shi, Spike mutation D614G alters SARS-CoV-2
fitness, Nature, , DOI:10.1038/s41586-020-2895-3.
32 L. Zhang, C. B. Jackson, H. Mou, A. Ojha, H. Peng, B. D. Quinlan, E. S.
Rangarajan, A. Pan, A. Vanderheiden, M. S. Suthar, W. Li, T. Izard, C. Rader, M.
Farzan and H. Choe, SARS-CoV-2 spike-protein D614G mutation increases virion
spike density and infectivity, Nat. Commun., , DOI:10.1038/s41467-020-19808-4.
33 S. A. Rella, Y. A. Kulikova, E. T. Dermitzakis and F. A. Kondrashov, Rates of
SARS-CoV-2 transmission and vaccination impact the fate of vaccine-resistant
strains, Sci. Rep., , DOI:10.1038/s41598-021-95025-3.
34 R. Wang, J. Chen and G.-W. Wei, The evolution of the mechanisms of SARS-
CoV-2 evolution revealing vaccine-resistant mutations in Europe and America.,
ArXiv.
64

35 LA County COVID-19 Vaccine Dashboard - LA County Department of Public
Health, http://publichealth.lacounty.gov/media/coronavirus/vaccine/vaccine-
dashboard.htm, (accessed 4 April 2022).
36 B. Jackson, M. F. Boni, M. J. Bull, A. Colleran, R. M. Colquhoun, A. C. Darby, S.
Haldenby, V. Hill, A. Lucaci, J. T. McCrone, S. M. Nicholls, Á. O’Toole, N.
Pacchiarini, R. Poplawski, E. Scher, F. Todd, H. J. Webster, M. Whitehead, C.
Wierzbicki, N. J. Loman, T. R. Connor, D. L. Robertson, O. G. Pybus and A.
Rambaut, Generation and transmission of interlineage recombinants in the SARS-
CoV-2 pandemic, Cell, , DOI:10.1016/j.cell.2021.08.014.
37 R. Lee, B.1.1.7: What we know about the novel SARS-CoV-2 variant.

65

Chapter 4
4. Metagenomic analysis of the antibiotic resistance
risk between an aerobic and anaerobic membrane
bioreactor
Phillip Wang
a
, Ali Zarei-Baygi
a
, Jeseth Delgado Vela
b
, Adam .L. Smith
a*

a
Sonny Astani Department of Civil and Environmental Engineering, University of Southern California
b
Department of Civil and Environmental Engineering, Howard University
*Corresponding author: smithada@usc.edu

Abstract
Membrane bioreactors are advanced treatment technologies that, compared to
conventional treatment processes, are expected to reduce antibiotic resistance spread
in the environment due to their superior biomass retention. In this study, metagenomic
sequencing was used to assess the response of the microbial community and antibiotic
resistance profile of an aerobic membrane bioreactor (AeMBR) and an anaerobic
membrane bioreactor (AnMBR) operated in parallel, treating identical synthetic
wastewater supplemented with select antibiotics. Our results revealed a greater
antibiotic impact on the microbial community and antibiotic resistance gene (ARG)
proliferation within the AeMBR compared to the AnMBR. Moreover, antibiotics loading
elicited a more pronounced disruption of the microbial diversity of the AeMBR biomass
and effluent intracellular DNA compared to its AnMBR counterpart. Under antibiotics
loading, the AeMBR effluent enriched ARGs in the form of antibiotic resistant bacteria
66

while the AnMBR effluent enriched a wide set extracellular ARGs across a broad
spectrum of resistance classes. These results provide a theoretical basis for process
selection and controlling the spread of ARGs during wastewater management.
4.1 Introduction
With the continual rise of antibiotic resistance worldwide, wastewater treatment
plants (WWTPs) are under increased scrutiny for their potential role in propagating
antibiotic resistance genes (ARGs) and antibiotic resistant bacteria (ARB) in
downstream environments
1,2
. Indeed, conventional WWTPs are not designed or
operated to remove ARGs and ARB, leaving the discharged effluent and biosolids as
potential transmission routes for ARGs and ARB in receiving environments
1,3
. WWTPs
also inevitably receive a constant influx of antibiotics at sub-lethal concentrations
resulting from antibiotic usage that could increase the abundance of ARGs and ARB
during treatment
4,5
, necessitating an evaluation of treatment systems to consider the
management of emerging contaminants such as antibiotic resistance. To that end,
membrane-based treatment technologies (e.g., aerobic membrane bioreactors
(AeMBR) and anaerobic membrane bioreactors (AnMBR)), can achieve superior log
removal of ARGs and ARB compared to conventional activated sludge processes,
thereby significantly reducing the antibiotic resistance load discharged into receiving
environments
6–8
.
While removal of ARGs and ARB in AeMBR and AnMBR systems are primarily
attributed to membrane separation, the effect of redox conditions on microbial
community characteristics could also aid in the removal of ARGs and ARB during
67

wastewater treatment. For instance, biofouling and wastewater-colloids have been
shown to adsorb ARGs and ARB and increase their retention within bioreactors
6,7,9
. As
such, the distinct biofouling mechanism and wastewater-colloid concentrations between
AeMBR and AnMBR systems could provide differential removal capacities for ARGs
and ARB. Moreover, microbial growth rate, which is significantly lower in anaerobic
communities
10
, could also affect the proliferation rate of ARGs and ARB during
wastewater treatment. While numerous studies have examined the effects of antibiotics
loading on the microbial and antibiotic resistance profile in AeMBR or AnMBR
systems
7,11–13
, no study to our knowledge has examined the antibiotic resistance
propagation risk between an AeMBR and an AnMBR in a side-by-side comparison.
Here, we operated two bench-scale AeMBR and AnMBR systems in parallel, treating
identical synthetic low-strength wastewater. We assessed the microbial community and
antibiotic resistance profiles of both systems under increasing antibiotic concentrations,
including sulfamethoxazole (SMX), erythromycin (ERY), and ampicillin (AMP) in weekly
stepwise increases at 0 µg/L, 10 µg/L, 50 µg/L, and 250 µg/L. The microbial community
and antibiotic resistance profile were assessed using metagenomic sequencing of the
bioreactor-biomass DNA, effluent-intracellular DNA (iDNA), and effluent-extracellular
DNA (exDNA). Further, qPCR was used to determine the absolute abundance of a
subset of 8 ARGs, intI1 (a proxy for mobile genetic element), and rpoB (a biomarker for
cell count) within the biomass and effluent samples throughout the study.
68

4.2 Material and Methods
4.2.1 Bioreactor operating conditions
Two bench-scale membrane bioreactors with an effective volume of 5 L each
were operated in parallel at 25°C. Both reactors were fed identical synthetic feed,
prepared twice per week. The synthetic feed was formulated to represent U.S. domestic
low-strength wastewater
14
(SI Table 1). Both reactors were seeded with sludge from a
mesophilic anaerobic digester collected from the Joint Water Pollution Control Plant.
The use of sludge sourced from an anaerobic digester for inoculation of both the
AeMBR and AnMBR was intended to reduce the initial variability in the microbial and
antibiotic resistance profile. The AeMBR was continuously aerated to a dissolved
oxygen content of 2.5 mg/L while the AnMBR was kept strictly anaerobic. The
bioreactors were continuously stirred-tank reactors (Chemglass Life Science, Vineland,
NJ) with three separate submerged-microfiltration silicon carbide membrane modules
(Cembrane, Denmark). The effective membrane area of each module was
approximately 0.015 m
2
and the membrane pore size was 0.1 μm. Both bioreactors
were operated until steady-state conditions, defined by consistent chemical oxygen
demand (COD) removal of >85% and a methane content >60% of produced biogas (for
AnMBR only), before starting the experiment. After seven days of steady-state
operation, sulfamethoxazole (SMX, a sulfonamide), erythromycin (ERY, a macrolide),
and ampicillin (AMP, a beta-lactam) were simultaneously added to the shared influent
feed. Antibiotic concentrations were held at 0 μg/L, 10 μg/L, 50 μg/L and 250 μg/L for a
period of seven days each. Antibiotic concentrations were chosen to represent typical
antibiotic levels found in U.S. WWTPs
15,16
. The hydraulic retention time (HRT) of both
69

reactors were maintained at 16 hrs and the solids retention time (SRT) for the AeMBR
and AnMBR were 15 days and 300 days, respectively.
4.2.2 Sample collection and DNA extraction
Three sample types (biomass, iDNA, and exDNA) were collected from each
bioreactor at the start and end of each antibiotic concentration period. A total of 2 mL of
biomass was taken from each bioreactor and centrifuged at 10,000 g @ 4°C for 1 min
before discarding the supernatant. iDNA was obtained by filtering 150 mL of effluent
through a 0.2 µm filter. exDNA was collected from the filtrate using isopropanol
precipitation and resuspended in 100 µL of elution buffer (Promega, Madison, WI)
17
.
DNA extraction for the biomass and iDNA was carried out using Maxwell 16 LEV Blood
DNA Purification Kit (Promega, Madison, WI) according to manufacturer’s instructions.
Extracted DNA were resuspended in 60 µL of elution buffer (Promega, Madison, WI). All
DNA extracts were split into two equal volumes for either DNA sequencing or qPCR
analysis.
4.2.3 ARG quantification via qPCR
qPCR was performed using a LightCycler 96 (Roche, Basel, Switzerland)
targeting 8 ARGs commonly found in domestic wastewater including genes conferring
resistance to sulfonamides (sul1 and sul2), macrolides (ermB and ermF), β-lactams
(ampC and oxa-1), and tetracycline (tetO and tetW), as well as a class 1 integron gene
(intI1), and a single gene copy per cell biomarker (rpoB)
18–20
. All qPCR reactions were
carried out in 20 μL reactions containing 10 μL qPCR master mix (Forget-Me-Not
EvaGreen, Biotium, Fermont, CA), 1 μL each of forward and reverse primers (0.25 μM
70

final concentration), 1 μL of DNA template, and 7 μL nuclease free water. Each reaction
was performed in triplicates. Thermocycling conditions for all gene targets are provided
in SI Table 2.The qPCR efficiency, slope, and R

squared values for all qPCR runs were
between 86% to 103.5%, -3.71 to -3.17, and >0.98, respectively. A detailed description
of these parameters for each primer assay can be found in SI Table 3. qPCR results
with Cq values greater than 40 or above the no template control (NTC) were discarded.
4.2.4 Illumina sequencing and sequenced data assembly
One DNA sample from each sample type during the 0 μg/L ,10 μg/L, and 250
μg/L antibiotic concentration periods were selected for metagenomic sequencing.
Samples were chosen to represent no, low, and high antibiotic concentrations, such as
those observed in hospital wastewaters
16,21
. DNA extracts were shotgun sequenced via
Illumina NextSeq platform with PE 150 bp reads (USC Genome Core, Los Angeles,
CA). Raw reads were dereplicated, and QC filtered and trimmed using Trimmomatic.
Metagenomic assemblies were completed using MegaHit (mink=21; maxk=255; step
size=6) and Prodigal was used to develop protein-coding genes and amino acid
sequences. Mapping was completed using bwa and contigs were binned using
CONCOCT. To obtain ARG and rpoB annotations, amino acid sequences were
compared with Megares 2.0 database (for ARGs) or a custom database of rpoB
sequences obtained from NCBI using blastp (align length > 50 amino acids, bitscore >
100 and percent ID greater than 60). Coverage of ARGs were normalized to coverage
of rpoB and reads per kilobase per million mapped reads (rpkm). The rpoB custom
database was retrieved from the RefSeq NCBI database using the "Identical Protein
71

Groups" function (retrieved on June 22, 2021). The custom rpoB database FASTA file is
presented as SI File 1.
4.2.5 Microbial community assessment
Sequenced DNA reads in FASTQ format were uploaded to MG-RAST for
taxonomic classification
22
. Phylum and genus level data were downloaded and analyzed
using Xlstat (Addinsoft). The alpha diversity for each sample was calculated using the
Shannon index equation, H = -Σpi * ln(pi), where pi is the proportion of the community
made up of species i. The alpha diversity in this study is represented as the anti-log of
H. The beta diversity between samples were visualized using a multi-dimensional plot
from a Pearson correlation similarity matrix. All sequenced reads are publicly available
via MG-RAST, sample IDs and its corresponding sample type can be found in SI Table
4.
4.3 Results and discussion
4.3.1 Robust performance of parallel bioreactors
The AeMBR and AnMBR displayed robust performance throughout the length of
the study, achieving average chemical oxygen demand (COD) removal efficiencies of
97.8 ± 1.9% (mean and standard deviation) and 92.2 ± 3.2% (mean and standard
deviation), respectively. Average AeMBR and AnMBR effluent COD values were 15.8 ±
5.2 mg/L and 38.0 ± 16.9 mg/L (mean and standard deviation), respectively. Throughout
operation, mixed liquor suspended solids (MLSS) and mixed liquor volatile suspended
solids (MLVSS) were 4.42 ± 0.30 g/L and 4.01 ± 0.37 g/L for the AeMBR and 8.95 ±
72

0.70 g/L and 8.11 ± 0.62 g/L for the AnMBR, respectively. The AnMBR produced 702 ±
24 mL/d biogas with a methane content of 69.6 ± 4.4%. The higher COD removal and
lower MLSS and MLVSS values in the AeMBR compared to the AnMBR indicates
higher microbial activity per cell under aerobic conditions compared to anaerobic
conditions. Although higher microbial activity and growth rate tend to improve COD
removal, it may increase antibiotic susceptibility and ARG selection potential
23
. Details
of sample preparation procedures can be found in the SI.
4.3.2 Antibiotics loading displayed stronger effects on the alpha and beta
diversity of the AeMBR compared to the AnMBR
Shannon indices were used to assess the effect of antibiotics loading on the
microbial diversity within the AeMBR and AnMBR systems. Overall, the microbial
diversity in the AnMBR displayed greater resilience against increasing antibiotics
concentration compared to the AeMBR (Figure 1). While increasing antibiotics loading
reduced the microbial diversity of the AeMBR biomass and effluent iDNA from 426 to
327 and 410 to 145, respectively, the addition of antibiotics stimulated the microbial
diversity of the AnMBR biomass and effluent iDNA from 473 to 485 and 252 to 285,
respectively. The greater antibiotic tolerance within the AnMBR compared to the
AeMBR could be driven by several biotic (e.g., microbial growth rate)
24,25
and abiotic
factors (e.g., physiochemical properties of the sludge)
26
. Many bactericidal antibiotics,
such as beta-lactam, are known to exert maximal effect on rapidly growing bacteria
compared to slow or non-dividing cells
23,25,27,28
. As such, the high cell density and low
microbial growth rate within AnMBRs could confer non-inheritable antimicrobial
73

resistance distinct from the classical method of ARGs
29
. In addition, AnMBRs are known
to contain higher soluble microbial products (SMP), colloids, and organic matter
compared to AeMBRs
30,31
. The higher SMP, colloids, and organics within AnMBRs
could lead to greater gel and cake layer growth within the AnMBR compared to the
AeMBR, which could help shelter antibiotic-sensitive microbial cells from antibiotic
exposure
32,33
.
Figure 4. 1: Species richness displayed as the anti-log of the Shannon indices for the AeMBR and
AnMBR biomass and effluent iDNA microbial diversity at increasing antibiotic concentrations.
Multidimensional scaling plots were used to assess the similarity between the
microbial communities across samples from the AeMBR and AnMBR (Figure 2A-B).
Consistent with our previous publication
12
, the microbial communities within the AeMBR
and AnMBR biomass were distinct from their respective effluent counterparts, which
highlights the effectiveness of membrane separation in reducing overall ARB and ARG
load discharged into receiving environments. Interestingly, the AnMBR effluent iDNA
and effluent exDNA microbial communities displayed greater similarity toward each
other compared to the microbial communities between the AeMBR effluent iDNA and
effluent exDNA. Moreover, antibiotics loading exhibited less impact on the microbial
74

communities within all AnMBR sample types compared to the antibiotics effect on the
AeMBR samples. Consistent with our previous explanation, we hypothesize that the
greater similarity and antibiotic resilience of the microbial communities in the AnMBR
samples compared to the AeMBR samples stem from a combination of biotic and
physiochemical factors. Primarily, the lower microbial growth rate and greater biofouling
within AnMBRs compared to AeMBRs could greatly reduce the effect of antibiotics on
its microbial targets. The greater similarity of the microbial communities between the
AnMBR effluent iDNA and effluent exDNA compared to the similarity between the
AeMBR effluent iDNA and effluent exDNA could also be attributed to higher biofouling in
the AnMBR. In our previous publication, we reported a strong positive correlation
between membrane fouling and exARG retention
13
. Therefore, biofouling on the
membrane and biofilms within the effluent line likely serve as a physical barrier against
the antibiotic selective pressure and as a reservoir for exDNA
6,13
. Consistent with our
theory, the median abundance of the measured AnMBR effluent exDNA genes,
quantified through qPCR, were 2.45 ± 8.9 folds greater than their aerobic counterparts
(SI Figure 1). Moreover, measured rpoB copies/L for AnMBR effluent iDNA were
approximately 1.3 log higher compared to the AeMBR effluent iDNA, which could be
due to the greater degree of attached growth within the AnMBR effluent line compared
to the AeMBR effluent line (SI Figure 2).
75

Figure 4. 2: Multi-dimensional scaling analyses between the microbial communities collected from the (A)
AeMBR system and the (B) AnMBR system at increasing antibiotic concentrations. Increasing color
darkness represents increasing antibiotic concentrations at 0 µg/L, 10 µg/L, and 250 µg/L. Dashed loops
represents distinct clusters and the arrows signify the directional movement of the samples from the no
antibiotic time point.
4.3.3 Microbial community members were distinct across sample types
A total of 71 and 59 phyla were identified from all samples collected from the
AeMBR and AnMBR, respectively. The top 10 phyla (relative abundance >0.005) from
the biomass and effluent iDNA samples are shown in Figure 3A-B. The minor phyla are
grouped together and labeled as ‘Others’. All groups are shown with its minimum and
maximum relative abundance across this study. The most relative abundant phylum
was Proteobacteria (43.0 - 84.7%) in all sample types in the AeMBR. Interestingly, the
following most relative abundant phyla varied between the biomass and effluent iDNA
samples. In the AeMBR biomass, the most relative abundant phyla were Bacteroidetes
(22.5 - 37.6%), Chloroflexi (14.7 - 16.1%), Nitrosperiae (1.52 - 2.81%), and
Actinobacteria (2.28 - 2.55%). In the AeMBR effluent iDNA, the most relative abundant
phyla were Bacteroidetes (7.50 - 20.6%), Actinobacteria (1.45 - 3.72%),
Verrucomicrobia (1.72 - 2.39%), and Eukarota unclassified (0.84 - 1.81%). The variation
76

in the dominant phyla between the biomass and effluent iDNA samples is likely due to
the size differences between microbial species which would affect the microbial
community after membrane separation.
In the AnMBR, variation in the phyla composition was more distinct across
sample types compared to the AeMBR. In the AnMBR biomass, the most relative
abundant phyla were Bacteroidetes (44.2 - 46.1%), Proteobacteria (17.2 - 18.2%),
Chloroflexi (13.3 - 15.6%), Firmicutes (6.04 - 6.18%), and Euryarchaeota (3.15 -
3.17%). In the AnMBR effluent iDNA, the most dominant phyla were Proteobacteria
(57.9 - 93.9%), Firmicutes (12.7 – 36.0%), Bacteroidetes (0.92 - 3.13%), Actinobacteria
(0.36 - 0.63%), and Lentisphaerae (0.07 - 0.27%). In addition to membrane separation,
the potential diffusion of oxygen through the walls of the effluent line may have created
a micro-aerobic environment, which would select against strictly anaerobic microbial
community members and support aerobic and facultative aerobic microorganisms. To
estimate the dissolved oxygen levels within the effluent at the time of collection, we
used equation (1) and (2).
𝐽 = 𝐾 ∗
(𝑃𝑜 −𝑃𝑖 )
𝑇 ℎ
(1)
𝑃𝑉 = 𝑛𝑅𝑇 (2)
In equation (1), J = flux of oxygen gas into the effluent, K = permeability constant
of the effluent line material, Po = partial pressure of oxygen in the atmosphere, Pi =
partial pressure of oxygen inside the tube, Th= thickness of the effluent wall. In equation
(2), P = partial pressure of oxygen, V = volume, n = number of moles, R = gas constant,
and T = room temperature in Kelvin. To simplify the calculation, we assumed negligible
77

oxygen consumption by the effluent microbial community and uniform oxygen
permeability throughout the effluent line. Based on our calculations, we estimate a
dissolved oxygen content within the effluent at the time of collection would reach 0.11
mg/L, which could inhibit a number of anaerobic microbial species
34
.

Figure 4. 3: Community structure of the AeMBR and AnMBR biomass and iDNA, at the phylum level (A
and B) and genus level (C and D). The top 15 phylum or genera from each sample type and date are
shown. The remaining phylum and genera are grouped under the label “Others’. Samples represents
antibiotics loading conditions at 0 µg/L, 10 µg/L, and 250 µg/L.
At the genus level, a total of 1,063 and 881 genera were identified in all samples
collected from the AeMBR and AnMBR, respectively. The top 10 genera (relative
abundance > 0.005) from all biomass and effluent iDNA samples are shown in Figure
3C-D and the minor phyla are merged and labeled as ‘Other’. Each group is shown with
78

the beginning and final range in parenthesis. In the AeMBR biomass, Roseiflexus
(11.2%) was the most relative abundant genus during the no antibiotic period, but
gradually decreased to 10.5%, and 9.45% at antibiotic concentrations of 10 µg/L and
250 µg/L, respectively. Under antibiotic loading, a selection occurred for the genus
Chitinophaga in the AeMBR biomass, which rose from 9.24% to 11.0% and 14.9% with
increasing antibiotic concentrations. Despite the high relative abundance of Roseiflexus
in the AeMBR biomass, we did not detect the presence of Roseiflexus in the effluent
iDNA fraction. In the effluent iDNA, there was a clear selection for Acidovorax (3.46% to
17.5%), Alicycliphilus (0.71% to 4.23%), Delftia (ND to 2.57%), and Polaromonas
(0.63% to 1.83%) with increase antibiotic concentrations. Of concern is the incomplete
removal of the pathogenic genus, Mycobacterium (1.17% to 0.75%) and Legionella
(3.4% to 2.32%) in the iDNA samples. The discharge of pathogens and ARGs into
receiving environments are potential promoters for the development of antibiotic
resistant pathogens, however, further studies would be required to evaluate this risk.
In the AnMBR biomass, the most relative abundant genera throughout the study
were Anaerolinea (16.2 - 18.3%), Bacteroides (5.77 - 5.89%), Paludibacter (3.01 -
3.1%), and Syntrophus (2.07 - 2.37%). Most of the detected genera in the AnMBR
biomass changed by less than 1% throughout the study. In contrast, the most relative
abundant genera in the AnMBR effluent iDNA decreased at elevated antibiotic
concentrations. The change in the most relative abundant genera in the AnMBR effluent
iDNA were Decholormonas (23.4% to 17.2%), Acinetobacter (11.8% to 1.46%),
Azoarcus (5.0% to 1.0%), Thauera (3.74% to 0.62%), and Aromatoleum (3.32% to
0.47%). The greater antibiotic impact on the AnMBR effluent microbial community
79

compared to the biomass could be due to the reduced cell density and increased
microbial growth following membrane separation, which could re-vert the antimicrobial
resistant phenotype as discussed previously. In addition, the diffusion of oxygen through
the effluent line, as calculated previously, likely provides low levels of oxygen to support
micro-aerobic to aerobic metabolisms which correlates to increased microbial activity
and antibiotic sensitivity. The observed physiological shift in this study would likely occur
at full-scale treatment plants as the microbial community transitions from an anaerobic
environment during treatment to an aerobic receiving environment.
4.3.4 Contribution of shared ARGs across sample types were inversely
impacted by the addition of antibiotics in the AeMBR and AnMBR
Venn diagrams were used to assess the distribution of ARG types among the
sample types collected from the AeMBR and AnMBR (SI Figure 3). Overall, the AeMBR
and AnMBR systems contained comparable amounts of ARG subtypes with similar
distribution among the sample types. In total, 486 and 412 unique ARG subtypes were
identified in the AeMBR and AnMBR, respectively. In the AeMBR, 79, 100, and 77 ARG
subtypes were shared between all sample types at 0, 10, and 250 µg/L antibiotic
concentrations, respectively. Similarly, in the AnMBR, 79, 93, and 51 ARG subtypes
were shared between all sample types at 0, 10, and 250 µg/L, respectively. Surprisingly,
a significantly greater number of ARG subtypes were detected exclusively in the
AeMBR and AnMBR effluent (243-249 and 164-309 ARGs, respectively) compared to
the AeMBR and AnMBR biomass (18-26 and 8-25 ARGs, respectively). The exclusive
effluent ARGs could be due to the limit of detection of metagenomic-based analyses
80

and the distinct microbial community shift from the AeMBR and AnMBR biomass to their
respective effluent iDNA. Of note, the high number of exARG subtypes exclusively
found in the effluent exDNA fraction of both the AeMBR (56-131 ARGs) and AnMBR
(50-82 ARGs) highlights the importance of including exDNA in future antibiotic
resistance risk analyses. The continuous release of exARGs from WWTPs and its
persistence in the downstream environments could contribute to the global increase of
antibiotic resistance
35,36
.
Figure 4. 4: Box plot analyses of the ARG distribution among the biomass, iDNA, and exDNA sample
types for the (A) AeMBR and (B) AnMBR at increasing antibiotic concentrations. The boxes represent the
25th and 75th percentile. The whiskers represent the largest and smallest values and outliers are shown
as closed circles. Each shared ARG is plotted as the relative abundance among the three sample types.
To examine the relative contribution of each sample type toward the shared ARG
pool, the abundance of each ARG, in reads per kilobase per million mapped reads
(rpkm), was divided by the total summed abundance in the biomass, iDNA, and exDNA.
The relative contribution of the shared ARGs from each sample type was visualized in
box plot format across all antibiotic concentrations (Figure 4A-B). Under the
experimental conditions tested, AeMBR and AnMBR biomass contributed the least
81

toward the shared ARG pool, with a median relative abundance between 9% and 19%.
Surprisingly, under increasing antibiotic loading concentrations, the AeMBR and
AnMBR systems displayed an inverse pattern for the partitioning of the shared ARGs.
During the no antibiotic loading period, AeMBR exDNA had a relative abundance
median of 52% toward the shared ARGs whereas iDNA contributed a median of 30%.
Under increasing antibiotics loading, the median contribution of shared ARGs from the
AeMBR exDNA fell to 28% and the median contribution of the AeMBR iDNA rose to
50%. Conversely, the AnMBR iDNA median contribution during the no antibiotic loading
period was 43% and the median contribution of the AnMBR exDNA was 30%. During
the antibiotic loading period, the AnMBR iDNA median contribution fell to 19% while the
AnMBR exDNA median contribution increased to 68%. Our data suggest that under
increasing antibiotic loading, AeMBR enrich ARGs within ARB while AnMBR enrich
exARGs. While ARGs in the form of ARB or exDNA could potentially be propagated in
downstream environments, ARGs in ARB are the active and more immediate risk
toward human health. Moreover, reported abundance of exARGs are likely an
overestimate of viable/intact genes since qPCR and shot-gun sequencing would
incorporate both intact and degraded gene fragments.
4.3.5 Anitbiotics addition enriched resistance classes in the AeMBR
biomass, AeMBR iDNA, and AnMBR exDNA
The relative fold change across various anitbiotic, mutli-metal, multi-biocide, and
multi-drug resistnace classes under increasing antibiotic loading conditions compared to
the no antibiotic loading period were visualized using heatmaps (Figure 5A-B).
82

Additional resistance classes such as multi-metal, multi-biocide, and multi-drug were
included in our analyses due to the potential of cross-resistance through broad-
spectrum efflux pumps. All detected resistance genes from the AeMBR and AnMBR
were grouped into 46 and 48 resistance classes, respetively. Overall, our data indicated
both AeMBR and AnMBR biomass displayed greater resilience toward resistance
enrichment compared to their respective effluent counterpart. Interestingly, the AeMBR
biomass and effluent iDNA showed increased abundance for the total number of multi-
metal and antibiotic resistance classes at antibiotic concentrations of 10 µg/L and 250
µg/L, whereas the total number of resistance classes for the AnMBR biomass and
effluent iDNA increased at 10 µg/L but sharply decreased at 250 µg/L (Table 1). In the
AnMBR effluent exDNA, 32 out of 46 extracellular resistance classes showed increased
abundance during the antibiotic loading period. Consistent with our previous
explanation, the high prevalence of extracellular multi-metal, multi-biocide, multi-drug
and antibiotic resistance classes is likely due to the higher degree of adsorption of
exDNA to particulates such as SMP, biofilm, and colloidal compounds in the AnMBR
effluent versus AeMBR effluent
9
. While total EPS and SMP concentrations from the
membranes or effluent lines were not measured in this study, we observed a more rapid
temporal increase in the transmembrane pressure (TMP) within the AnMBR compared
to the AeMBR. TMP for the AeMBR and AnMBR were both maintained between 13.79
kPa and 20.68 kPa, however, the faster temporal rise in the AnMBR TMP compared to
the AeMBR suggests a greater degree of membrane fouling occurring in the AnMBR
versus AeMBR.
83

Despite numerous studies suggesting biofilms as hotspots for HGT, the majority
of these studies involve simplified biofilm setups with a limited and defined set of donor
and recipient bacterial species
37
. Moreover, several studies have shown HGT within
biofilms are readily inhibited, which highlights the presence of regulatory factors in
natural environments
38–43
. In fact, our data does not support the paradigm of rampant
HGT within biofilms as the AnMBR effluent, with greater degree of fouling compared to
the AeMBR, is not associated with greater selection of ARB. To date, natural barriers
toward HGT remain understudied and the true extent to which HGT events occur within
complex, natural biofilms remain largely unknown.

Figure 4. 5: Heat map analysis of the metagenomic ARGs grouped by resistant classes. Data is depicted
as the log (fold change) of examined resistant classes relative to the 0 µg/L antibiotic loading period.
Results for the biomass, iDNA, and exDNA are grouped as (A) AeMBR (B) and AnMBR.
84

4.3.6 AeMBR biomass and effluent iDNA showed greater increase in
ARGs/rpoB compared to its AnMBR counterpart
Bar chart comparison between the AeMBR and AnMBR biomass and effluent
iDNA were used to assess the antibiotic resistant proliferation risk for select resistance
classes. The select resistance classes (beta-lactam, mls, sulfonamide, multi-biocide,
multi-drug, and multi-metal) were chosen based on their ability to confer resistance to
the administered antibiotics (beta-lactam, mls, and sulfonamide, Figure 6A-B). In
general, the abundance of beta-lactam, mls, sulfonamide, multi-biocide, multi-drug, and
multi-metal ARGs/rpoB showed greater increase in the AeMBR system compared to the
AnMBR. In AeMBR biomass, beta-lactam, mls, and sulfonamide ARGs/rpoB decreased
during the antibiotic loading period. However, multi-biocide, multi-drug, and multi-metal
ARGs/rpoB increased between 0.1-0.4 log in the AeMBR biomass during the same
period, which suggests antibiotic resistance within AeMBR biomass could stem from the
enrichment of cross-resistant genes such as broad-spectrum efflux pumps
44,45
. Within
the AnMBR biomass, the select resistance classes decreased between 0.37-1.27 log
ARGs/rpoB during the antibiotic loading period, suggesting antibiotic resistance in the
AnMBR biomass could be from unknown ARGs or through an ARG-independent
mechanism (e.g., low microbial growth and activity). In the AeMBR iDNA, select
resistant classes, beside beta-lactam, increased between 0.14-1.72 log ARGs/rpoB
during the antibiotics loading period. In contrast, all select resistant classes, beside mls,
decreased between 0.18-0.75 log ARGs/rpoB in the AnMBR iDNA. The greater
increase of select resistance classes in the AeMBR iDNA versus AnMBR iDNA could be
85

due to the greater ecological similarity between the AeMBR biomass and effluent
compared to the AnMBR biomass and effluent.
The transition from anaerobic to micro-aerobic/aerobic metabolism likely imposes
a strong metabolic barrier for obligate anaerobes to proliferate. As a result, microbial
regrowth downstream of the membrane would be more phylogenetically distant to the
biomass in an AnMBR system compared to an AeMBR system. While horizontal gene
transfer of ARGs through mobile genetic elements such as plasmids could propagate
antibiotic resistant determinants across various redox conditions, preliminary
assessment of large-scale plasmid metagenomic data suggest plasmid host-ranges are
largely governed by the ecological and phylogenetic distance of the bacterial hosts
46
.
Our results suggest that AeMBRs are more sensitive to antibiotic resistant proliferation
than AnMBRs under the conditions tested in this study. However, to elucidate the
mechanism of antibiotic resistance proliferation and to distinguish between vertical and
horizontal gene transfer would require higher-resolution molecular biology tools
47
.
Figure 4. 6: Bar chart analysis of resistant classes or beta-lactam, mls and sulfonamide, multi-biocide,
multi-drug, and multi-metal. (A) AeMBR biomass and iDNA. (B) AnMBR biomass, iDNA. Antibiotic
concentrations are color coded red (0 µg/L), green (10 µg/L), and yellow (250 µg/L).
86

In this study we assessed the antibiotic resistance risk between an AeMBR and
AnMBR treating identical synthetic wastewater supplemented with select antibiotics
using a metagenomic-based approach. To our knowledge, this is the first study to
examine the impact of antibiotics loading on a wide range of ARGs and the microbial
community within two operationally identical AeMBR and AnMBR systems. Our results
showed greater antibiotic impact on the microbial community and ARG proliferation
within the AeMBR compared to the AnMBR. Antibiotics loading elicited more
pronounced impact on the microbial diversity of the AeMBR biomass and effluent iDNA
compared to its AnMBR counterpart. Under antibiotics loading, the AeMBR effluent
enriched ARGs in the form of ARB while the AnMBR effluent enriched a wide set
exARGs across a broad spectrum of resistance classes. Further, the slow microbial
growth and high biofouling within AnMBRs may confer non-inheritable antibacterial
resistance phenotypes within the biomass and effluent iDNA, which may help reduce
the risk of ARG propagation during wastewater treatment. Future studies examining
specific ARG host-range (e.g., EPIC PCR and chromosomal cross-linking) with a range
of antibiotic classes and operational parameters would help broaden our understanding
of the antibiotic resistance propagation risks for AeMBR and AnMBR systems.
Supporting Information
• Additional experimental details and methods including bioreactor setup, qPCR
targets, qPCR assay performance, and metagenomic dataset sample IDs.
• Amino acid sequences of our reference rpoB database.

87

Acknowledgement
This work was funded from Water for Agriculture grant no. 2016-68007-25044
from the USDA National Institute of Food and Agriculture and partially supported by the
US-Egypt Science and Technology Joint Fund (NAS Subaward No. 2000012477). This
publication is derived from Subject Data funded in whole or part by USAID and NAS
through Subaward [20000012477]. Any opinions, findings, conclusions, or
recommendations expressed in this publication are those of the authors alone, and do
not necessarily reflect the views of USAID or NAS. This work used the Extreme Science
and Engineering Discovery Environment (XSEDE) Bridges-2 resource at the Pittsburgh
Supercomputer Center through allocation EVB210001, which is supported by National
Science Foundation grant number ACI-1548562.

88

References
1. Berendonk, T. U. et al. Tackling antibiotic resistance: The environmental
framework. Nature Reviews Microbiology vol. 13 (2015).
2. Makowska, N., Koczura, R. & Mokracka, J. Class 1 integrase, sulfonamide and
tetracycline resistance genes in wastewater treatment plant and surface water.
Chemosphere 144, (2016).
3. Fahrenfeld, N., Ma, Y., O’Brien, M. & Pruden, A. Reclaimed water as a reservoir
of antibiotic resistance genes: Distribution system and irrigation implications.
Front. Microbiol. 4, (2013).
4. Guo, J., Li, J., Chen, H., Bond, P. L. & Yuan, Z. Metagenomic analysis reveals
wastewater treatment plants as hotspots of antibiotic resistance genes and mobile
genetic elements. Water Res. (2017) doi:10.1016/j.watres.2017.07.002.
5. Ju, F. et al. Antibiotic resistance genes and human bacterial pathogens: Co-
occurrence, removal, and enrichment in municipal sewage sludge digesters.
Water Res. 91, (2016).
6. Cheng, H. & Hong, P. Y. Removal of antibiotic-resistant bacteria and antibiotic
resistance genes affected by varying degrees of fouling on anaerobic
microfiltration membranes. Environ. Sci. Technol. (2017)
doi:10.1021/acs.est.7b03798.
7. Zhu, Y. et al. Robust performance of a membrane bioreactor for removing
antibiotic resistance genes exposed to antibiotics: Role of membrane foulants.
Water Res. 130, 139–150 (2018).
8. Li, B., Qiu, Y., Li, J., Liang, P. & Huang, X. Removal of antibiotic resistance genes
in four full-scale membrane bioreactors. Sci. Total Environ. 653, (2019).
9. Riquelme Breazeal, M. V., Novak, J. T., Vikesland, P. J. & Pruden, A. Effect of
wastewater colloids on membrane removal of antibiotic resistance genes. Water
Res. (2013) doi:10.1016/j.watres.2012.09.044.
10. Ghangrekar, M. M. & Behera, M. 3.5 - Suspended Growth Treatment Processes.
in (ed. Ahuja, S. B. T.-C. W. Q. and P.) 74–89 (Elsevier, 2014).
doi:https://doi.org/10.1016/B978-0-12-382182-9.00087-6.
11. Zarei-Baygi, A., Harb, M., Wang, P., Stadler, L. B. L. B. & Smith, A. L. A. L.
Evaluating Antibiotic Resistance Gene Correlations with Antibiotic Exposure
Conditions in Anaerobic Membrane Bioreactors. (2019)
doi:10.1021/acs.est.9b00798.
12. Zarei-Baygi, A., Harb, M., Wang, P., Stadler, L. B. & Smith, A. L. Microbial
community and antibiotic resistance profiles of biomass and effluent are distinctly
affected by antibiotic addition to an anaerobic membrane bioreactor. Environ. Sci.
Water Res. Technol. (2020) doi:10.1039/C9EW00913B.
89

13. Zarei-Baygi, A., Wang, P., Harb, M., Stadler, L. B. & Smith, A. L. Membrane
Fouling Inversely Impacts Intracellular and Extracellular Antibiotic Resistance
Gene Abundances in the Effluent of an Anaerobic Membrane Bioreactor. Environ.
Sci. & Technol. 54, (2020).
14. Smith, A. L., Skerlos, S. J. & Raskin, L. Psychrophilic anaerobic membrane
bioreactor treatment of domestic wastewater. Water Res. 47, (2013).
15. Karthikeyan, K. G. & Meyer, M. T. Occurrence of antibiotics in wastewater
treatment facilities in Wisconsin, USA. Sci. Total Environ. 361, (2006).
16. Kulkarni, P. et al. Antibiotic concentrations decrease during wastewater treatment
but persist at low levels in reclaimed water. Int. J. Environ. Res. Public Health 14,
(2017).
17. Green, M. R. & Sambrook, J. Precipitation of DNA with isopropanol. Cold Spring
Harb. Protoc. 2017, (2017).
18. Peng, Y. P. et al. Fluorescence Nanobiosensor for Simultaneous Detection of
Multiple Veterinary Drugs in Chicken Samples. J. Anal. Test. (2021)
doi:10.1007/s41664-021-00199-4.
19. Munir, M., Wong, K. & Xagoraraki, I. Release of antibiotic resistant bacteria and
genes in the effluent and biosolids of five wastewater utilities in Michigan. Water
Res. 45, 681–693 (2011).
20. Pruden, A., Pei, R., Storteboom, H. & Carlson, K. H. Antibiotic resistance genes
as emerging contaminants: Studies in northern Colorado. Environ. Sci. Technol.
40, (2006).
21. McKinney, C. W., Dungan, R. S., Moore, A. & Leytem, A. B. Occurrence and
abundance of antibiotic resistance genes in agricultural soil receiving dairy
manure. FEMS Microbiol. Ecol. 94, 1–10 (2018).
22. Keegan, K. P., Glass, E. M. & Meyer, F. MG-RAST, a metagenomics service for
analysis of microbial community structure and function. in Methods in Molecular
Biology vol. 1399 (2016).
23. Lee, A. J. et al. Robust, linear correlations between growth rates and β-lactam-
mediated lysis rates. Proc. Natl. Acad. Sci. U. S. A. 115, (2018).
24. Eng, R. H. K., Padberg, F. T., Smith, S. M., Tan, E. N. & Cherubin4, C. E.
Bactericidal Effects of Antibiotics on Slowly Growing and Nongrowing Bacteria.
Antimicrob. Agents Chemother. 35, 1824–1828 (1991).
25. Gilbert, P., Collier, P. J. & Brown, M. R. W. Influence of growth rate on
susceptibility to antimicrobial agents: Biofilms, cell cycle, dormancy, and strigent
response. Antimicrob. Agents Chemother. 34, 1865–1868 (1990).
26. Deowan, S. A., Bouhadjar, S. I. & Hoinkis, J. Membrane bioreactors for water
treatment. in Advances in Membrane Technologies for Water Treatment:
Materials, Processes and Applications (2015). doi:10.1016/B978-1-78242-121-
90

4.00005-8.
27. Claudi, B. et al. Phenotypic Variation of Salmonella in Host Tissues Delays
Eradication by Antimicrobial Chemotherapy. Cell 158, 722–733 (2014).
28. Tuomanen, E., Cozens, R., Tosch, W., Zak, O. & Tomasz, A. The rate of killing of
Escherichia coli by β-lactam antibiotics is strictly proportional to the rate of
bacterial growth. J. Gen. Microbiol. 132, (1986).
29. Lewis, K. Persister cells, dormancy and infectious disease. Nature Reviews
Microbiology vol. 5 (2007).
30. Wang, K. M., Garcia, N. M., Soares, A., Jefferson, B. & McAdam, E. J.
Comparison of fouling between aerobic and anaerobic MBR treating municipal
wastewater. H2Open Journal vol. 1 (2018).
31. Spagni, A., Casu, S., Crispino, N. A., Farina, R. & Mattioli, D. Filterability in a
submerged anaerobic membrane bioreactor. Desalination 250, (2010).
32. Flemming, H. C. Biofouling in water systems - Cases, causes and
countermeasures. Applied Microbiology and Biotechnology vol. 59 (2002).
33. Wang, L. et al. Extracellular polymeric substances affect the responses of multi-
species biofilms in the presence of sulfamethizole. Environ. Pollut. 235, (2018).
34. Lu, Z. & Imlay, J. A. When anaerobes encounter oxygen: mechanisms of oxygen
toxicity, tolerance and defence. Nature Reviews Microbiology vol. 19 (2021).
35. Zarei-Baygi, A. & Smith, A. L. Intracellular versus extracellular antibiotic
resistance genes in the environment: Prevalence, horizontal transfer, and
mitigation strategies. Bioresource Technology vol. 319 (2021).
36. Sivalingam, P., Poté, J. & Prabakar, K. Extracellular DNA (eDNA): Neglected and
potential sources of antibiotic resistant genes (ARGs) in the aquatic
environments. Pathogens vol. 9 (2020).
37. Abe, K., Nomura, N. & Suzuki, S. Biofilms: Hot spots of horizontal gene transfer
(HGT) in aquatic environments, with a focus on a new HGT mechanism. FEMS
Microbiology Ecology vol. 96 (2021).
38. Stalder, T. & Top, E. Plasmid transfer in biofilms: A perspective on limitations and
opportunities. npj Biofilms and Microbiomes vol. 2 (2016).
39. Christensen, B. B., Sternberg, C. & Molin, S. Bacterial plasmid conjugation on
semi-solid surfaces monitored with the green fluorescent protein (GFP) from
Aequorea victoria as a marker. in Gene vol. 173 (1996).
40. Seoane, J. et al. An individual-based approach to explain plasmid invasion in
bacterial populations. FEMS Microbiol. Ecol. 75, (2011).
41. Krone, S. M., Lu, R., Fox, R., Suzuki, H. & Top, E. M. Modelling the spatial
dynamics of plasmid transfer and persistence. Microbiology 153, (2007).
91

42. Reisner, A., Wolinski, H. & Zechner, E. L. In situ monitoring of IncF plasmid
transfer on semi-solid agar surfaces reveals a limited invasion of plasmids in
recipient colonies. Plasmid 67, (2012).
43. Headd, B. & Bradford, S. A. The conjugation window in an Escherichia coli K-12
strain with an IncFII plasmid. Appl. Environ. Microbiol. 86, (2020).
44. Chapman, J. S. Disinfectant resistance mechanisms, cross-resistance, and co-
resistance. in International Biodeterioration and Biodegradation vol. 51 (2003).
45. Pal, C. et al. Metal Resistance and Its Association With Antibiotic Resistance. in
Advances in Microbial Physiology vol. 70 (2017).
46. Redondo-Salvo, S. et al. Pathways for horizontal gene transfer in bacteria
revealed by a global map of their plasmids. Nat. Commun. 11, (2020).
47. Rice, E. W., Wang, P., Smith, A. L. & Stadler, L. B. Determining Hosts of
Antibiotic Resistance Genes: A Review of Methodological Advances. Environ. Sci.
Technol. Lett. (2020) doi:10.1021/acs.estlett.0c00202.

92

Chapter 5
5. Characterizing mobile colistin resistance genes
within Los Angeles County wastewater
Phillip Wang
a
and Adam L. Smith
a*

a
Sonny Astani Department of Civil and Environmental Engineering, University of Southern California
*Corresponding author: smithada@usc.edu

Abstract
Wastewater-based epidemiology offers insight into the circulating antibiotic
resistant profile of its serviced population. Continual surveillance and characterization of
wastewater antibiotic resistant genes (ARGs) could help identify emerging ARGs of
clinical importance and its dissemination pathway across bacterial species. In this study,
we detected four mobile colistin-resistant (mcr) gene variants (mcr-3.12, mcr-4.3, mcr-
5.1, and mcr-9.1) in Los Angeles County wastewater between the period of Sep 2020 to
Feb 2022. Through sequencing of cultured colistin-resistant bacterial isolates and
wastewater plasmids, we characterized the genetic context of the detected mcr gene
variants and revealed conserved mcr gene cassettes between our dataset and
published data. Moreover, we discovered two novel small plasmids (approximately 8.29
-8.67 kbp) with near identical mcr-5 gene cassettes flanked by a non-homologous
segment. From the cultured colistin-resistant bacterial isolates, we detected mcr-3.12 in
Aeromonas hydrophilia and mcr-9.1 in Hafnia alvei. Our findings underscore the utility of
wastewater-based epidemiology to detect and characterize circulating ARGs and help
uncover their dissemination pathways.
93

5.1 Introduction
The global dissemination of plasmid-mediated mobile colistin resistance (mcr)
poses a severe risk to human health
1–3
. Colistin is a last resort antibiotic typically
reserved for the treatment of multi-drug resistant Gram-negative pathogens
4
. Colistin
resistance can arise through mutations in chromosomal genes (e.g., eptA, pmrABCD,
phoPQ, mgrB, crrAB, yciM, lpxM, ramA, and ompW) or through horizontal gene transfer
(HGT) of mcr within foreign DNA (e.g., plasmids, transposons, and integrons)
4
. The
rising prevalence of mcr-positive bacterial isolates in diverse pathogenic species such
as Escherichia coli, Klebsiella spp., Moraxella spp., Aeromonas spp., Shewanella spp.,
and Salmonella spp. is concerning due to the parallel spread of antibiotic resistance,
particularly carbapenem resistance
5,6
. The acquisition of mcr within carbapenem-
resistant Enterobacteriaceae would confer resistance to all known antibiotic treatment
options
7
.
Since the initial report of mobile colistin resistance 1 (mcr-1) in China, nine
additional variants (mcr-2 through mcr-10) have been reported across the world
8–17
. All
mcr variants encode for a lipid A phosphoethanolamine transferase which catalyzes the
addition of a phosphoethanolamine to the lipid A moiety of the lipopolysaccharide
(LPS)
18
. The addition of phosphoethanolamine to the LPS mitigates its ionic interaction
with colistin and confers colistin resistance to the bacterium. Characterization studies of
known mcr variants have often identified mobile genetic elements (MGE) such as
insertion sequences and transposons in close vicinity to the sequenced mcr gene and
could be a major contributor to the mobility of mcr between plasmids and bacterial
hosts
19
.
94

Here, we report the first detection of mcr-3.12, mcr-4.3, mcr-5.1, and mcr-9.1 in
the urban wastewater of Los Angeles County. Select wastewater samples between the
period of Sep 2020-Feb 2022 were screened for the presence of mcr and its gene
variants using standard multiplex PCR. Multiplex PCR results were used to inform our
qPCR targets. qPCR was used to measure the relative abundance of mcr-4 and mcr-5
(normalized to rpoB) over the same sample period. The surrounding genetic
environment of the detected mcr variants in this report were further characterized using
whole-genome sequencing of cultured colistin-resistant bacterial isolates and culture-
independent metagenomic sequencing of the extracted wastewater plasmids. Parallel
short (Illumina) and long (Oxford Nanopore) read sequencing of the extracted
wastewater plasmids were conducted to obtain mcr-positive contigs and complete
plasmid assemblies.
5.2 Material and Methods
5.2.1 Sample processing
Archived 24-hour composite influent samples from the Joint Water Pollution
Control Plant (JW) and Hyperion Water Reclamation Plant (Hyp) between the dates of
September 2020 to February 2021 were selected for multiplex PCR screening (3
samples per location). In addition to the archived influent samples, two fresh composite
influent samples from JW and Hyp, collected on February 12
th
, 2022, were included in
the multiplex PCR screening and used for isolation of colistin-resistant bacteria and
wastewater plasmid extraction. Archived samples were stored on processed filters and
600 µL of lysis buffer at -80ºC. The multiplex PCR assays used in this study target
95

genes mcr-1 through mcr-9. Primers for mcr-10 were not included in our multiplex PCR
assay due to its relatively recent discovery and published primers were incompatible
with the mcr-1 through mcr-9 primer set.
5.2.2 Genomic DNA extraction
For wastewater genomic DNA extraction, sample volumes of 50 mL were
concentrated using vacuum filtration (Sterlitech, Auburn, WA) and a 0.45 µm Whatman
filter (Fischer Scientific, Pittsburgh, PA). DNA extraction from the processed
membranes was carried out using zirconium bead beating (Bio-spec, Bartlesville, OK)
and the Maxwell 16 LEV DNA kit (Promega, Madison, WI) according to the
manufacturer’s instructions. Extracted DNA was eluted in 50 µL of nuclease-free water.
5.2.3 Plasmid extraction and processing
Wastewater sample volumes of 50 mL from JW and Hyp samples collected on
Jan 12
th
, 2022, were centrifuged at 10,000 g for 5 min at 4°C, decanted, and
resuspended in 1 mL nuclease-free water before plasmid extraction. Plasmid
extractions were performed using a Plasmid Midiprep kit (Qiagen, Germantown, MD)
according to the manufacturer’s instructions and resuspended in a final volume of 100
µL of nuclease-free water. To reduce chromosomal DNA contamination, plasmid
extracts were digested with Plasmid-Safe DNAse (Lucigen, Radnor, PA) according to
the manufacturer’s protocol.
96

5.2.4 PCR-based screening of mcr-1 through mcr-9
Wastewater genomic DNA and plasmid extracts were screened for mcr-1 through
mcr-9 using 9 sets of primers and two separate thermocycling conditions
1,2
. All primers
and thermocycling conditions used in this study are listed in Supporting Information (SI)
Table S1. Forward and reverse primer sets for mcr-1 through mcr-5 (set 1) and mcr-6
through mcr-9 (set 2) were mixed into four working multiplex stock solutions containing
a final concentration of 10µM of each primer. Each PCR reaction consisted of 5 µL of
Q5 2X PCR master mix (NEB, Ipswich, MA), 3.5 µL of nuclease-free water, 0.25 µL of
each forward and reverse multiplex primer solution, and 1 µL of DNA extract. PCR
products were visualized using agarose gel electrophoresis with 1.5% agarose gel (w/v)
operated at 100V for 30 min. DNA bands were purified using Agarose Gel DNA
Extraction Kit (Roche, St. Louis, MO) and sent for Sanger sequencing (Azenta, South
Plainfield, NJ).
5.2.5 qPCR analysis of mcr-4 and mcr-5
Using our multiplex PCR-screening results, all wastewater genomic DNA extracts
were used for qPCR analysis targeting mcr-4, mcr-5, and rpoB (a biomarker for cell
count). Primers and thermocycling conditions used in this study are listed in Table S5.1.
5.2.6 Culture-based screening of colistin-resistant bacteria
To isolate colistin-resistant bacteria, 50 µL of wastewater sample (JW and Hyp)
was transferred onto McConkey Agar plates supplemented with 2 mg/L of colistin
sulfate (Thermo Fisher, USA) and incubated at 36°C for 16 hours. Twenty-four
overnight colonies (12 from JW and 12 from Hyp) were picked and inoculated into 24x1
97

mL nutrient broth supplemented with 2 mg/L of colistin sulfate. Liquid cultures were
stirred at 180 rpm for 16 hours at 36°C. Grown liquid cultures underwent genomic DNA
extraction via zirconium bead beating and Maxwell LEV DNA extraction as described
above. The 24 genomic DNA extracts were screened for mcr-1 through mcr-9 using the
multiplex PCR protocol described previously. PCR products were visualized through
agarose gel electrophoreses using 1.5% agarose (w/v) operated at 100V for 30 min.
5.2.7 Illumina and Nanopore Sequencing and Assembly
Select genomic DNA extracts were sent to Snpsaurus (Eugene, Oregon) for
sequencing using Illumina Next-seq (PE 2x150 bp). All samples were sequenced to
approximately 60X coverage. Paired-end sequences were trimmed by bbduk with
parameters: ktrim=r, k=17, hdist=1, mink=8, ref=/bbmap/resources/nextera.fa.gz,
minlen=100, ow=t, qtrim=r, trimq=10 pig, z=t, unpigz=t and then assembled with
SPAdes
3
-3.13.0 using -k 77,99. Draft genome assemblies were imported into Geneious
Prime (version 2022.1.1) for antibiotic resistance gene analysis against the CARD
database
4
.
Wastewater plasmid extracts from JW and Hyp were amplified through rolling
circle amplification using 4BB TruePrime (4basebio, Madrid, Spain) according to the
manufacturer’s protocol. Plasmid amplicons were split into two equal volumes, where
one half was sent for Illumina sequencing (Azenta, South Plainfield, NJ) on the Nova-
seq platform (PE 2x150 bp) and the other half was sent for Nanopore on the
Promethion platform sequencing (UC Davis DNA Sequencing Facility, Davis, CA).
Illumina paired-end sequences were trimmed and filtered using Geneious Prime
98

(version 2022.1.1) with parameters: trimq= 30 and minlen= 150. Trimmed and filtered
Illumina reads were assembled into contigs using Megahit
5
. Assembled contigs were
searched for mcr variants using a curated set of 39 mcr variant sequences obtained
from the CARD database and one mcr-10 sequence from Wang et al
6
.
Nanopore reads were imported into Geneious Prime (version 2022.1.1) and
filtered based on the following parameters: minlen = 1,000 bp and tirmq=10. Filtered
Nanopore reads were mapped to the same curated mcr database as above using
minimap-2
7
. Nanopore reads with mcr hits were manually inspected and confirmed
using NCBI Blastn. Mapped Nanopore reads were searched with primer sequences for
mcr-1 through mcr-9 (allowing up to 2 mismatches). Nanopore reads with less than 2
repeats were discarded. Nanopore reads with linear mcr repeats were split at the start
of the mcr gene. The split fragments from the same parental read were aligned to form
a circular consensus sequence.
5.3 Results
5.3.1 PCR and qPCR detection of mcr genes in wastewater samples
Genes for mcr-4 and mcr-5 were detected in both Joint Water Pollution Control
Plant (HW) and Hyperion Water Reclamation Plant (Hyp) samples collected on Jan
2021, Feb 2021, and Feb 2022 (SI Figure 1). For samples collected in Nov 2020, mcr-4
and mcr-5 were detected only in the Hyp sample. PCR amplicons were purified and
sent for Sanger sequencing (Azenta, South Plainfield, NJ) to confirm the identity of the
PCR products. Using NCBI BLASTn, the sequenced amplicons were confirmed to be
99

mcr-4 and mcr-5 with >98% query coverage and sequence identity. Details on sample
processing and data analysis are presented in the SI.
To quantify the relative abundance of mcr-4 and mcr-5 over our sampling
timeline, wastewater genomic DNA extracts were used for qPCR analysis targeting mcr-
4, mcr-5, and rpoB (Figure 1). In general, mcr-4 and mcr-5 gene copies/rpoB increased
from 2020 to 2022. The increased abundance of mcr gene copies/rpoB is consistent
with the CDC report of marked increases in antimicrobial resistant infections during the
pandemic
22
.

Figure 5. 1: qPCR analysis of mcr-4 and mcr-5 normalized to rpoB. Blue bars represent samples
collected from Joint Water Pollution Control Plant (JW) and red bars represent samples collected from the
Hyperion Water Reclamation Plant (Hyp). Error bars represent the standard deviation from the mean for
each qPCR run (n=3).
5.3.2 Characterization of cultured colistin-resistant bacterial isolates.
Colistin-resistant bacterial isolates were obtained from raw wastewater samples
(JW and Hyp) collected on Feb 12
th
, 2022, using McConkey agar supplemented with 2
mg/L of colistin sulfate. From overnight plates, a total of twenty-four bacterial colonies
100

were selected for multiplex PCR screening targeting mcr-1 through mcr-9 as described
previously. Multiplex PCR results showed that 10 out of 24 colistin-resistant bacterial
colonies contained DNA amplicons of expected lengths. Genomic DNA extracts of the
10 PCR-positive bacterial colonies were sent for whole genome sequencing to confirm
and characterize the potential mcr genes and their surrounding genetic environment.
Interestingly, analysis of the draft genomes revealed only 2 out of the 10 draft genomes
contained mcr genes (mcr-3.12 and mcr-9.1, Table 1). The remaining 8 draft genome
assemblies contain chromosomal genes, eptA, pmrAB, or phoPQ, which confer intrinsic
colistin resistance. While the reason behind the false positive PCR results for the 8
bacterial isolates remain unclear, we were able to confirm the presence or absence of
mcr genes through Illumina sequencing of the extracted bacterial genomes.
The draft genome containing mcr-3.12 was classified as Aeromonas hydrophilia.
NCBI BLASTn analysis of the scaffold containing mcr-3.12 (scaffold_46, 9.81 kbp)
matched with A. hydrophilia strain ZYAH75 (accession No. CP016990.1) as the top hit
with 98% query coverage and 99.02% sequence identity. In addition, scaffold_46
shares similar surrounding genetic sequences to several mcr-3-positive A. hydrophilia
strains identified in U.S. food-producing animals particularly strains AH1805, AH2359,
AH3019, and AH3924
23
. Multiple sequence alignment of these six A. hydrophilia strains
show a hypothetical protein and an MFS transporter (major facilitator superfamily)
directly upstream and an mprF (multiple peptide resistance factor) downstream of the
mcr-3/3.12 gene (Figure 2A). A. hydrophilia is an opportunistic gastrointestinal pathogen
documented in several cases of human diarrheal diseases
24
. The characterization of A.
hydrophilia strains with highly similar mcr-3 genetic environments within food-producing
101

animals and urban wastewater (this study) highlights the risk for the potential
transmission of mcr-positive bacterial hosts from food to humans. In addition to mcr-
3.12 within the draft genome of A. hydrophilia, we also identified imiH-1, qacJ, blaCMY-
8b-1, oxa-724, and rsmA. The presence of carbapenem resistant genes in addition to
mcr-3 raises concern as it potentially confers carbapenem and colistin resistance.
The draft genome harboring mcr-9.1 was classified as Hafnia alvei. While mcr-
9.1 has been reported in a related bacterial species within the Hafnia genus, Hafnia
paralvei
25
, this work marks the first detection of mcr-9.1 within the bacterial species H.
alvei. NCBI BLASTn analysis of the mcr-9.1-positive scaffold (scaffold_26, 8.96 kbp),
matched with four entries within the NCBI databank, Enterobacter ludwigii strain Sb-9
plasmid pEL-ars2 (accession no CP094843.1), Hafnia paralvei plasmid pAVS0177-a
(accession no CP083738.1), Leclercia adecarboxylata strain L21 chromosome
(accession no CP043397.1), and Enterobacter kobei strain IB2020 chromosome
(accession no CP059481.1) with 100% query coverage and sequence identity (Figure
2B). The mcr-9.1 gene within all five alignments shares an IS110, IS6, and IS481 family
transposase downstream of the detected mcr-9.1 gene, which suggests the mcr-9.1
cassette in this study is mobile between plasmids and chromosomes. Although plasmid
prediction analysis with PlasmidFinder
26
failed to assign scaffold_26 to any known
plasmid replicon type, a subsequent analysis with PlasFlow
27
predicted scaffold_26 to
be of plasmid origin with a 64.6% probability. Moreover, a similar analysis of all
scaffolds for the H. alvei draft genome revealed scaffold_9 (202 kbp) belonging to the
plasmid type IncHI2, which is the same plasmid type as the mcr-9.1-positive pEL-ars2
and several mcr-9.1-positive IncHI2 plasmids identified from Salmonella Typhimurium
28
.
102

This work marks the second report of mcr-9.1 within the intrinsically colistin-resistant
genus Hafnia, which adds to the growing list of mcr acquisition within intrinsically
colistin-resistant bacterial species. In addition to mcr-9.1, we also identified acc-3,
ampH-1, blaCMY-8b-1, oxa-724, and rsmA, which encodes for carbapenem, beta-lactam,
and aminoglycoside resistance.

Figure 5. 2: Multiple sequence alignment with EasyFig v2.2.5
29
showing (A) A. hydrophilia scaffold_46
aligned to A. hydrophilia strain ZYAH75, AH1805, AH2359, AH3924, and AH3019 (B) H. Alvei,
scaffold_26 aligned to L. adecarboxylata strain L21, plasmid pAV50177-a, plasmid pEL-ars2, and E.
kobei strain IB2020. Sequence identity is marked by the intensity of the color gray. Genes are color coded
by protein function.

Table 5. 1: Genome properties and antibiotic resistance profile of the A. hydrophilia and H. alvei isolates
harboring mcr, isolated from Los Angeles County wastewater.

103

5.3.3 Culture-independent metagenomic analysis of wastewater plasmids
mcr contigs
Metagenomic sequencing of the wastewater plasmid amplicons revealed plasmid
contigs containing complete mcr-4 or mcr-5 along with its surrounding genetic
environment. NCBI BLASTn analysis of the mcr-4 containing plasmid contig matched
with an unnamed plasmid (accession no. CP077333.1) with 100% query coverage and
99.97% sequence identity and the mcr-5 containing plasmid contig matched with
plasmid pSGMCR103 (accession no. MK731877.1) with 96% query coverage and
99.96% sequence identity. The mcr-4 plasmid contig also contains an IS6 family
transpose and the mcr-5 plasmid contig contains a Tn3 superfamily.
5.3.4 Complete circular plasmids containing mcr-5
In addition to the mcr-4 and mcr-5 plasmid contigs, we identified two complete circular
plasmids containing mcr-5, denoted as Hyp p1 and Hyp p2 (Figure 3, SI File 1).
Interestingly, both plasmids are similar in size, Hyp p1= 8.67 kbp and Hyp p2 = 8.29
kbp. Alignment of the two plasmids revealed a near identical 4.15 kbp segment (>99%
sequence identity) encoding for IR148-chrB-mcr-5-mfs-IRL. Although the remaining
segments of Hyp p1 and Hyp p2 were not homologous to each other, they contained
several cryptic open reading frames identified as mobilization genes through NCBI
conserved domain search (E value cut off < 0.005) (SI Table 2A and 2B). While most
reports of mcr-5 plasmids have been on self-transmissible plasmids greater than 30 kbp
and contain multiple antibiotic-resistant genes, here we report two novel small plasmids
encoding for a single antibiotic resistance gene, mcr-5.1. Moreover, the shared
104

functionality of the non-identical segment between Hyp p1 and Hyp p2 suggests
independent recombination events between an mcr-5.1 donor and a small mobilizable
plasmid recipient could be a common mechanism for the dissemination of mcr genes.

Figure 5. 3: Pairwise alignment between Hyp p1 and Hyp p2 using Clinker v0.0.5
30
. Sequence identity is
shown by increasing intensity in the color black. Hypothetical protein coding sequences have not been
confirmed. Genes are color coded by protein function.
This report marks the first detection and characterization of mcr-3.12, mcr-4.3,
mcr-5.1, and mcr-9.1 in Los Angeles County wastewater. Moreover, qPCR analysis
revealed an increase in mcr-4 and mcr-5 gene copies per rpoB from Sep 2020 to Feb
2022. Whole genome analysis of colistin-resistant bacterial isolates and metagenomic
sequencing of wastewater plasmid extracts revealed a wide variety of MGE surrounding
the characterized mcr genes and the detection of two small mcr-5 plasmids. The
detection of a diverse set of mcr variants suggests there is an established mcr-positive
microbial community within Los Angeles County. Additional characterization studies of
the wastewater mcr profile could help identify novel genetic carriers and intermediary
steps as mcr genes are mobilized between plasmids, chromosomes, and bacterial
hosts.

105

Accession number
Draft genome sequences for Aeromonas hydrophila and Hafnia alvei have been
uploaded to GeneBank under BioProject ID PRJNA887832 and PRJNA887846,
respectively.
Acknowledgements
This work was funded by the US-Egypt Science and Technology Joint Fund (NAS
Subaward No. 2000012477).

106

References
1. Ling, Z. et al. Epidemiology of mobile colistin resistance genes mcr-1 to mcr-9.
Journal of Antimicrobial Chemotherapy vol. 75 (2020).
2. Cuadrat, R. R. C., Sorokina, M., Andrade, B. G., Goris, T. & Dávila, A. M. R.
Global ocean resistome revealed: Exploring antibiotic resistance gene abundance
and distribution in TARA Oceans samples. Gigascience 9, (2020).
3. Wang, R. et al. The global distribution and spread of the mobilized colistin
resistance gene mcr-1. Nat. Commun. (2018) doi:10.1038/s41467-018-03205-z.
4. El-Sayed Ahmed, M. A. E. G. et al. Colistin and its role in the Era of antibiotic
resistance: an extended review (2000–2019). Emerging Microbes and Infections
vol. 9 (2020).
5. Marchaim, D. et al. Outbreak of colistin-resistant, carbapenem-resistant Klebsiella
pneumoniae in Metropolitan Detroit, Michigan. Antimicrob. Agents Chemother. 55,
(2011).
6. Huang, H. et al. Colistin-resistance gene mcr in clinical carbapenem-resistant
Enterobacteriaceae strains in China, 2014–2019. Emerg. Microbes Infect. 9,
(2020).
7. Jiang, Y. et al. Clinical relevance and plasmid dynamics of mcr-1-positive
Escherichia coli in China: a multicentre case-control and molecular
epidemiological study. The Lancet Microbe 1, (2020).
8. Liu, Y. Y. et al. Emergence of plasmid-mediated colistin resistance mechanism
MCR-1 in animals and human beings in China: A microbiological and molecular
biological study. Lancet Infect. Dis. (2016) doi:10.1016/S1473-3099(15)00424-7.
9. Wang, C. et al. Identification of novel mobile colistin resistance gene mcr-10.
Emerg. Microbes Infect. 9, (2020).
10. Erratum to: Identification of a novel plasmid-mediated colistin-resistance gene,
mcr-2, in Escherichia coli, Belgium, June 2016 (Eurosurveillance, (2016) 21:27,
10.2807/1560-7917.ES.2016.21.27.30280). Eurosurveillance vol. 21 (2016).
11. Yin, W. et al. Novel plasmid-mediated colistin resistance gene mcr-3 in
Escherichia coli. MBio 8, (2017).
12. Carattoli, A. et al. Novel plasmid-mediated colistin resistance mcr-4 gene in
Salmonella and Escherichia coli, Italy 2013, Spain and Belgium, 2015 to 2016.
Eurosurveillance 22, (2017).
13. Borowiak, M. et al. Identification of a novel transposon-associated
phosphoethanolamine transferase gene, mcr-5, conferring colistin resistance in d-
tartrate fermenting Salmonella enterica subsp. enterica serovar Paratyphi B. J.
Antimicrob. Chemother. 72, (2017).
14. Yang, Y. Q., Li, Y. X., Lei, C. W., Zhang, A. Y. & Wang, H. N. Novel plasmid-
107

mediated colistin resistance gene mcr-7.1 in Klebsiella pneumoniae. J.
Antimicrob. Chemother. 73, (2018).
15. Wang, X. et al. Emergence of a novel mobile colistin resistance gene, mcr-8, in
NDM-producing Klebsiella pneumoniae article. Emerg. Microbes Infect. 7, (2018).
16. Carroll, L. M. et al. Identification of novel mobilized colistin resistance gene mcr-9
in a multidrug-resistant, colistin-susceptible salmonella enterica serotype
typhimurium isolate. MBio 10, (2019).
17. AbuOun, M. et al. mcr-1 and mcr-2 variant genes identified in Moraxella species
isolated from pigs in Great Britain from 2014 to 2015. J. Antimicrob. Chemother.
72, (2017).
18. Osei Sekyere, J. Mcr colistin resistance gene: a systematic review of current
diagnostics and detection methods. Microbiologyopen 8, (2019).
19. Partridge, S. R., Kwong, S. M., Firth, N. & Jensen, S. O. Mobile Genetic Elements
Associated with Antimicrobial Resistance. Clin. Microbiol. Rev. 31, e00088-17
(2018).
20. Strepis, N. et al. Genetic Analysis of mcr-1-Carrying Plasmids From Gram-
Negative Bacteria in a Dutch Tertiary Care Hospital: Evidence for Intrapatient and
Interspecies Transmission Events. Front. Microbiol. 12, (2021).
21. Li, Y. et al. Characterization of the global distribution and diversified plasmid
reservoirs of the colistin resistance gene mcr-9. Sci. Rep. 10, (2020).
22. COVID-19: U.S. Impact on Antimicrobial Resistance, Special Report 2022.
(2022).
23. Wang, Y., Hou, N., Rasooly, R., Gu, Y. & He, X. Prevalence and Genetic Analysis
of Chromosomal mcr-3/7 in Aeromonas From U.S. Animal-Derived Samples.
Front. Microbiol. 12, (2021).
24. Daskalov, H. The importance of Aeromonas hydrophila in food safety. Food
Control 17, (2006).
25. Biggel, M. et al. Complete Genome Sequence of Hafnia paralvei Isolate
AVS0177, Harboring mcr-9 on a Plasmid . Microbiol. Resour. Announc. 11,
(2022).
26. Carattoli, A. & Hasman, H. PlasmidFinder and In Silico pMLST: Identification and
Typing of Plasmid Replicons in Whole-Genome Sequencing (WGS). in Methods
in Molecular Biology vol. 2075 (2020).
27. Krawczyk, P. S., Lipinski, L. & Dziembowski, A. PlasFlow: predicting plasmid
sequences in metagenomic data using genome signatures. Nucleic Acids Res.
46, (2018).
28. Diaconu, E. L. et al. Emergence of IncHI2 Plasmids With Mobilized Colistin
Resistance (mcr)-9 Gene in ESBL-Producing, Multidrug-Resistant Salmonella
108

Typhimurium and Its Monophasic Variant ST34 From Food-Producing Animals in
Italy. Front. Microbiol. 12, (2021).
29. Sullivan, M. J., Petty, N. K. & Beatson, S. A. Easyfig: A genome comparison
visualizer. Bioinformatics 27, (2011).
30. Gilchrist, C. L. M. & Chooi, Y. H. Clinker & clustermap.js: Automatic generation of
gene cluster comparison figures. Bioinformatics 37, (2021).

109

Chapter 6
6. Conclusion
6.1 Overview
The primary objective of this dissertation was to investigate WBE for tracking and
characterizing SARS-CoV-2 and antibiotic resistance. WBE is a relatively low cost and
non-invasive method for assessing multiple aspects of communal health, which can be
used to inform intervention policies such as lockdown measures and antibiotic
stewardship. With the increasing affordability and ease of use of nucleic acid
sequencing assays, the role of traditional WBE can be expanded to incorporate routine
characterization of viral and microbial genomes to assess the temporal and
geographical variation of various biological contaminants across sewersheds. In
addition, WBE can be used to assess antibiotic resistance management within WWTPs
and help determine the risk of antibiotic resistant propagation between wastewater
treatment technologies. This dissertation particularly focused on wastewater SARS-
CoV-2, mobile-colistin resistance genes (mcr), and the propagation risk of antibiotic
resistance between an AeMBR and an AnMBR. First, we used RT-qPCR to track
wastewater SARS-CoV-2 levels in Los Angeles County over the course of nine months
and compared our results to in-person testing data (Chapter 2). Next, we performed
amplicon tiling and second-generation sequencing of select wastewater samples to
evaluate the evolution of wastewater SARS-CoV-2 genomes over the collection period
of our previous study (Chapter 3). We then investigated the antibiotic resistance
110

propagation response between an AeMBR and an AnMBR under identical antibiotic
stress conditions (Chapter 4). Last, we surveyed the urban wastewater of Los Angeles
County and characterized four mcr variants (Chapter 5).
6.2 WBE is an effective complementary tool for tracking SARS-
CoV-2 viral load across sewersheds
The widespread outbreak of COVID-19 across the United States placed a
significant strain on in-person testing facilities across the country. WBE offers an
effective and versatile alternative for monitoring communal viral load and enhances
clinical surveillance by identifying potential under-testing communities. In this study, we
assessed weekly samples of 24-hour flow-weighted composite influent from five
WWTPs for 44 weeks. Wastewater SARS-CoV-2 levels were quantified using RT-qPCR
targeting the CDC recommended nucleocapsid genes N1 and N2. Pearson correlation
analysis of wastewater SARS-CoV-2 levels with clinical data showed strong correlations
of r = 0.94, and p << 0.01 for N1 and N2. Moreover, comparison of daily wastewater
SARS-CoV-2 levels to daily clinical data over a one-week period showed the strongest
correlation of r = 0.96 N1/N2, p <0.005 when wastewater data was off-set with the
clinical data by a 5-day lead period. The improved correlation of the daily wastewater
SARS-CoV-2 levels to daily clinical data using a 5-day lead period suggests WBE could
be used as an early warning system for rising community infections. We used Monte
Carlo simulations to estimate the number of infected individuals in our sampled
sewershed. Our results estimate the number of infected individuals peaked on January
19
th
, 2021 with about 1.25 million active cases. The estimated total number of infected
individuals for the duration of this study was 3.42 million people, which represents
111

34.2% of the population residing in Los Angeles County. Our estimated number
exceeded the cumulative clinical case count by almost 2 million people.
6.3 Wastewater SARS-CoV-2 genomes are highly heterogenous
with key mutations showing dominance across variants
The evolution of the SARS-CoV-2 genome has prolonged the course of the
COVID-19 pandemic. To understand the shifting mutation profile in circulation, we
investigated the genetic sequence of the wastewater SARS-CoV-2 genomes using
amplicon tiling and second-generation sequencing of select wastewater samples
representing distinct periods in the first year of the pandemic. We demonstrated the
utility of WWS to monitor wastewater SARS-CoV-2 mutations using 24 samples
collected from two WWTPs in Los Angeles between July 2020 to March 2021. We
assessed samples from two WWTPs collectively serving over 7.5 million people (75% of
the Los Angeles County population). Our analysis detected 351 non-redundant
mutations from 24 samples of which 138 out of 351 mutations were present in
sequenced clinical samples from Los Angeles County. Further, we showcased the
viability of sampling biosolids for SARS-CoV-2 variant analysis as a potential alternative
to composite-influent samples.
6.4 AnMBR exhibit greater resilience toward antibiotic resistance
propagation compared to AeMBR
Process selection for wastewater treatment plays an important role in shaping
the antibiotic resistance profile and propagation risk within WWTPs. While membrane
bioreactors can achieve greater biosolid retention and antibiotic resistance removal
112

compared to conventional treatment trains, a direct comparison between an AeMBR
and an AnMBR on antibiotic resistant management remained a knowledge gap in the
literature. In this sense, we assessed the antibiotic resistance propagation risk between
an AeMBR and AnMBR operated in parallel treating identical synthetic feed and
antibiotics addition. The synthetic feed of the AeMBR and AnMBR was challenged with
weekly increasing concentrations of three antibiotics (SMX, AMP, and TET) at 0 µg/L,
10 µg/L, 50 µg/L, and 250 µg/L. Our results revealed a lesser antibiotic impact on the
microbial community and ARG proliferation within the AnMBR compared to the AeMBR.
Moreover, antibiotics loading elicited a less pronounced disruption of the microbial
diversity of the AnMBR biomass and effluent intracellular DNA compared to its AeMBR
counterpart. Under antibiotics loading, the AeMBR effluent enriched ARGs in the form of
ARB while the AnMBR effluent enriched a wide set exARGs across a broad spectrum of
resistance classes.
6.5 WBE is an effective tool to investigate ARGs of clinical
importance
The increasing prevalence of colistin resistance in previously colistin-sensitive
bacteria is a serious threat to human health. Mobile-colistin resistant genes (mcr) are
believed to be the causative agent behind the rapid spread of colistin resistance across
bacterial hosts. Here, we surveyed the urban wastewater of Los Angeles County and
report the first detection of mcr-3.12, mcr-4.3, mcr-5.1, and mcr-9.1. Moreover, select
wastewater samples showed increasing relative abundance of mcr-4 and mcr-5 per
rpoB between the time period of Sep 2020-Feb 2022. The detected mcr variants in this
report were isolated for further characterization of their surrounding genetic environment
113

through whole-genome sequencing of cultured colistin-resistant bacterial isolates and
culture-independent metagenomic sequencing of the extracted wastewater plasmids. In
summary, within colistin-resistant bacterial isolates, we detected a chromosome-
associated mcr-3.12 in Aeromonas hydrophilia and a plasmid-associated mcr-9.1 in
Hafnia alvei. Within the wastewater metagenomic plasmid dataset, we identified plasmid
contigs containing complete mcr-4.3 or mcr-5.1 as well as their immediate surrounding
sequences, which offers insight into their associated mobile genetic elements (e.g.,
Tn3). Further, we recovered two complete novel plasmids harboring mcr-5.1. While
several reports have identified mcr-5 on large self-transmissible plasmids (>30 kbp),
here we report two small novel plasmids around 8.24 kbp and 8.69 kbp in length. The
two novel plasmids share a 4.15 kbp region (>99% sequence identity) encoding for
IR148-chrB-mcr-5-mfs-IRL. The remaining segment of the two plasmids contain a non-
homologous set of plasmid mobilization genes, which suggests independent
recombination events.
6.6 Future Research
While WBE has been used since the 1940s, the COVID-19 pandemic has drawn
renewed interest in WBE for broad public health surveillance. This dissertation
investigated the utility of WBE for monitoring the spread and evolution of SARS-CoV-2
and antibiotic resistance in serviced populations. The findings and challenges garnered
from monitoring wastewater SARS-CoV-2 could be readily applied to a range of viral
diseases such as monkeypox, influenza A, and polio. In addition, WBE for antibiotic
resistance shows promise as a routine surveillance tool to assess the antibiotic
114

resistance profile from human excreta and its fate in WWTPs. The findings of this work
can guide future WBE strategies for monitoring and assessing biological contaminants.
In Chapter 2, we showed wastewater SARS-CoV-2 levels correlate strongly to in-
person testing data across five sewersheds in Los Angeles County. Our work supports
WBE as an effective tool to complement in-person testing by confirming community
outbreaks and identifying areas needing additional testing. While we used Monte Carlo
simulations to estimate the number of active cases in Los Angeles County, the accuracy
of these calculations remains unverified. Several factors such as viral load in human
stool, RNA virus degradation rate in the sewer networks, extraction efficiency, and
quantification efficiency all contribute considerable variability to the measurable SARS-
CoV-2 quantity. While several studies have attempted to quantify the uncertainty factors
along the sample collection and assessment process, there remains no method or
benchmark dataset to verify the estimated case count from wastewater data. Future
research direction should investigate the confidence range of current WBE practices. In
this view, universities may serve as an ideal location to conduct this research direction
as many universities have invested in equipment for campus-level wastewater
surveillance during the COVID-19 pandemic and have the potential to validate campus
wastewater data with their student health department.
Monitoring HGT events and ARG-host relationships remains an elusive goal for
WBE. While genome assemblies from metagenomic datasets often contain ARGs in
association with known phylogenetic markers (e.g., 16S rRNA and rpoB), a significant
fraction of ARGs reside within mobile genetic elements (e.g., plasmids, integrons,
transposons, and genomic islands) which can be shared across bacterial genus or
115

higher taxonomic groups. Metagenomic sequencing is a powerful tool for providing a
snapshot of the genetic content within a sample, however, it falls short in providing
dynamic or temporal information such as HGT events and ARG-host relationships.
Understanding the dissemination pathways of HGT and ARG-host relationships are
fundamental steps toward mitigating the spread of antibiotic resistance and the
development of multi-antibiotic resistant pathogens, while helping extend the lifetime of
our current antibiotic treatment options. Novel molecular assays such as single-cell
fusion PCR and genomic cross-linking are promising approaches to identifying HGT and
ARG-host relationships and would be powerful tools to incorporate into WBE to
elucidate the ARG-host relationships within the natural and built environment.
In Chapter 4, we investigated the antibiotic resistance response between an
AnMBR and an AeMBR under identical antibiotic stress conditions. While our data
showed AnMBRs exhibit markedly lower antibiotic resistance proliferation risks
compared to an AeMBR under the conditions tested, significant ARGs in the form of
iARGs and exARGs remain detectable in both AnMBR and AeMBR effluents. While
ARGs are natural to microbial life and complete removal of ARGs during treatment is
unrealistic and unwarranted, future research should investigate the antibiotic resistance
propagation risk in receiving environments downstream of WWTPs. Currently, the
antibiotic resistance propagation risk from WWTPs to the receiving microbial
communities remains an area of debate. While several studies have shown
environments receiving human fecal pollution to be strongly correlated with increased
ARG abundance, direct evidence of HGT remains scarce. The use of single fusion PCR
or genome cross-linking, as mentioned above, could be used to assess the HGT of
116

ARGs originating from WWTPs to receiving microbial communities. Results from this
antibiotic propagation risk assessment would help inform regulatory guidelines for the
ARG and ARB discharge limit.

117

Appendix A: Supplemental Information for Chapter 2

Figure S2. 1: Comparison of measured N1 Copies/L versus N2 Copies/L over the duration of this study.
Linear fit is represented by R
2
and the Pearson correlation between N1 and N2 Copies/L is represented
by r.

r
r=0.98
118

Table S2. 1: Primer and probe names, sequences, and final concentrations used in this study.

Primer/Oligo

Sequence

Final
Concentration
2019- nCoV_N1-Fwd 5’-GAC CCC AAA ATC AGC GAA AT-3’ 500 nM
2019- nCoV_N1-Rev 5’-TCTGGTTACTGCCAGTTGAATCTG-3’ 500 nM
2019- nCoV_N1-P 5’-FAM-ACCCCGCATTACGTTTGGTGGACC-
BHQ1-3’
125 nM
2019- nCoV_N2-Fwd 5’-TTACAAACATTGGCCGCAAA-3’ 500 nM
2019- nCoV_N2-Rev 5’-GCGCGA CAT TCC GAA GAA-3’ 500 nM
2019- nCoV_N2-P 5’-FAM-ACA ATT TGC CCC CAG CGC TTC AG-
BHQ1-3’
125 nM
PMMV-Fwd 5’-GAGTGGTTTGACCTTAACGTTGA-3’ 500 nM
PMMV-Fwd 5’-TTGTCGGTTGCAATGCAAGT-3’ 500 nM
PMMV-P
5’-Quasar 670-CCTACCGAAGCAAATG-BHQ1-3’
125 nM
CC169-Fwd 5’-CATTCCGGATACTGCGATTTTAAGTG-3’ 500 nM
NF002-Rev 5'-GCTTCCCCGACTTCTTTCGA-3' 500 nM
DS121-Org-560 5’-CAL Fluor Orange 560-
CGCCCCCAGAAGCAATTTCGTGTAAA-BHQ1-3’
125 nM
119

Table S2. 2: Raw data from the Monte Carlo Simulations for the compiled and individual sewersheds

Total
120

San Jose Creek (SJ)
121

Hyperion (HYP)
122

Joint Water (JW)
123

Long Beach (LB)
124

Whittier Narrows (WN)
125

Figure S2. 2: (A) Time series analysis of the moving 20-day Covid-19 tests given in Los Angeles County.
(B) Heat map of the cumulative adjusted rate for persons tested in Los Angeles County. Adjusted rates
are per 100,000 people. Data grouped by service planning area. Image taken from the LA County
COVID-19 Surveillance Dasboard
http://dashboard.publichealth.lacounty.gov/covid19_surveillance_dashboard/ on Oct 11, 2021.

0.0E+00
5.0E+05
1.0E+06
1.5E+06
2.0E+06
2.5E+06
20-Day Test Count
126

Appendix B: Supporting Information for Chapter 4
B.1 AeMBR and AnMBR operation
Bioreactor headspace was recirculated with biogas through sparging tubes below
the membrane modules at a rate of 30 mL/min (for each membrane module) to help
scour the surface of the membranes and control membrane fouling. Effluent permeate
flow was controlled using a peristaltic pump (BT100-1L Multi-channel Peristaltic Pump,
Longer, China) at a rate of 8 min filtration and 2 min backwashing. Silicon carbide
membrane modules were purchased from Cembrane (Denmark). The average effective
area for each module was 150 cm
2
, and the nominal pore size was 0.1 µm. The
maximum temperature and pressure tolerance of the membrane modules were 60 C
and 6 bar, respectively. AnMBR operational parameters were monitored and recorded
using LabVIEW 2014 (Student Edition)
B.2 Analysis methods
Total biogas produced was measured using a flowmeter (GFM17 Flow Meter,
Aalborg, 21 Orangeburg, NY).Chemical oxygen demand (COD) was measured in
accordance with USEPA Method 410.4 using a HI801 Spectrophotometer (Hanna
Instruments, Woonsocket, RI, USA). Mixed liquor suspended solid (MLSS) and mixed
liquor volatile suspended solid (MLVSS) were determined according to standard
methods described previously
1
. Biogas samples from the reactor headspace and
effluent were analyzed using a Trace 1310 GC system (Thermo Scientific, NY) with
127

flame ionization detection (FID) as described previously
1
. Bioreactor COD, MLSS, and
MLVSS were measured once per week.
B.3 Synthetic Feed for Bioreactor
The synthetic wastewater was prepared twice a week and fed to both the AeMBR
and AnMBR to prevent degradation. To prevent degradation of the feed, the synthetic
wastewater was prepared as a concentrate, acidified to a pH of 3.5, and refrigerated.
The concentrated synthetic wastewater was blended with a basic dilution of water
before feeding to the AeMBR and AnMBR. SI Table 1 shows the composition of these
two solutions and their final concentration in the synthetic wastewater.
Table S4. 1: Composition of the synthetic feed used in this study.

128

Table S4. 2: Forward and reverse primers and qPCR thermocycling conditions of all targeted genes.

Figure S4. 1: A violin box plot showing the log ratio of the measured genes in the AnMBR and AeMBR
effluent exDNA (copies/L). The box area represents the 25%-75% range of the data. The colored region
of the violin depicts the distribution of the data. The white circle represents the median. The vertical lines
above and below the box are the 1.5 interquartile range.
129

Figure S4. 2: qPCR results of measured rpoB gene copies/L of the effluent iDNA through the study
period. The box represents the 25%-75% range of the measured data. The median is represented by a
horizontal line inside the boxes, the mean is represented by a square inside the boxes, and the 1.5
interquartile range is represented by the bars above and below the colored boxes.

Figure S4. 3: Venn diagram analyses of the ARG type distribution among the biomass, iDNA, and exDNA
of the AeMBR (A) and AnMBR (B) at increasing antibiotic concentrations.

130

References
1. Zarei-Baygi A, Harb M, Wang P, Stadler LB, Smith AL. Evaluating Antibiotic
Resistance Gene Correlations with Antibiotic Exposure Conditions in Anaerobic
Membrane Bioreactors. Environmental science & technology. 2019 Feb 27;53(7):3599-
609.
2. Chen, J.; Yu, Z. T.; Michel, F. C.; Wittum, T.; Morrison, M., Development and
application of real-time PCR assays for quantification of erm genes conferring
resistance to macrolides-lincosamides-streptogramin B in livestock manure and manure
management systems. Appl. Environ. Microbiol. 2007, 73, (14), 4407-4416.
3. Pei, R.; Kim, S.-C.; Carlson, K. H.; Pruden, A., Effect of river landscape on the
sediment concentrations of antibiotics and corresponding antibiotic resistance genes
(ARG). Water Res. 2006, 40, (12), 2427-2435.
4. Barlow, R. S.; Pemberton, J. M.; Desmarchelier, P. M.; Gobius, K. S., Isolation and
characterization of integron-containing bacteria without antibiotic selection. Antimicrob.
Agents Chemother. 2004, 48, (3), 838-842.
5. Feria, C.; Ferreira, E.; Correia, J. D.; Goncalves, J.; Canica, M., Patterns and
mechanisms of resistance to beta-lactams and beta-lactamase inhibitors in
uropathogenic Escherichia coli isolated from dogs in Portugal. J Antimicrob. Chemother.
2002, 49, (1), 77- 85.
6. Szczepanowski, R.; Linke, B.; Krahn, I.; Gartemann, K. H.; Gutzkow, T.; Eichler, W.;
Puhler, A.; Schluter, A., Detection of 140 clinically relevant antibiotic-resistance genes in
the plasmid metagenome of wastewater treatment plant bacteria showing reduced
susceptibility to selected antibiotics. Microbiology 2009, 155, (Pt 7), 2306-19.
7. Aminov, R. I.; Garrigues-Jeanjean, N.; Mackie, R. I., Molecular ecology of tetracycline
resistance: development and validation of primers for detection of tetracycline
resistance genes encoding ribosomal protection proteins. Appl. Environ. Microbiol.
2001, 67, (1), 22-32.
8. Dahllöf, I.; Baillie, H.; Kjelleberg, S., rpoB-based microbial community analysis avoids
limitations inherent in 16S rRNA gene intraspecies heterogeneity. Appl. Environ.
Microbiol. 2000, 66, (8), 3376-3380.

131

Appendix C: Supporting information for Chapter 5
Table S5. 1: Primers and thermocycling conditions used in this study
1,2,8

Figure S5. 1: Multiplex PCR detection of mcr genes. Genomic DNA extracts from Joint Water Pollution
Control Plant and Hyperion Water Reclamation plant were used as templates. Numbers 1 through 4
represent collection dates of Nov 2021, Jan 2021, Feb 2021, and Jan 2022, respectively. Agarose gel
electrophoresis (1.5% w/v) was used to separate multiplex PCR products. L indicates the molecular
ladder (Quick-Load 100bp DNA Ladder, Ipswich, MA). The size of the molecular weight markers are listed
on the left side of the gel and the size of the amplicons are indicated on the right side. NTC = no template
control with nuclease-free water.

132

Table S5. 2: Results for the NCBI Conserved Domain search for (A) Hyp p1 and (B) Hyp p2
A)
Name Accession Description Nucleotide Position E-value
MFS_MJ1317_like cd17370 MJ1317 and similar transporters of
the Major Facilitator Superfamily
2516-3685 5.54E-88
PRK11598 super
family
cl32711
putative metal dependent hydrolase
843-2456 1.02E-147
CcdB pfam01845
CcdB protein
3869-4159 9.46E-36
MbeB_N pfam04837
MbeB-like, N-term conserved region
5219-5374 5.04E-12
Replicase pfam03090
Replicase family
7555-7938 6.87E-46
CcdA pfam07362
Post-segregation antitoxin CcdA
4165-4377 5.12E-19
Relaxase super
family
cl21589 Relaxase/Mobilisation nuclease
domain
5542-6027 6.27E-11
HTH super family cl21459
Helix-turn-helix domains
7033-7125 2.74E-05
PriCT_1 super
family
cl07362
Primase C terminal 1 (PriCT-1)
7276-7485 3.50E-03

B)
Name Accession Description
Nucleotide
Position
E-value
CcdA
pfam07362
Post-segregation antitoxin CcdA
7756-7968 2.00E-19
Relaxase super
family
cl21589 Relaxase/Mobilisation nuclease domain
6103-6738 5.46E-15
MFS_MJ1317_like
cd17370
MJ1317 and similar transporters of the Major
Facilitator Superfamily
2516-3685 1.16E-87
ALP_like super
family
cl23718 alkaline phosphatases and sulfatases
1496-1957 3.95E-43
MbeB_N
pfam04837
MbeB-like, N-term conserved region
6725-6880 6.66E-12
MbeD_MobD
super family
cl04831 MbeD/MobD like
7163-7366 1.14E-08
MobC
pfam05713
Bacterial mobilisation protein (MobC);
5906-6028 3.85E-07
PRK11598 super
family
cl32711 putative metal dependent hydrolase;
864-2456 2.09E-85
Replicase super
family
cl03886 Replicase family...
4080-4463 2.13E-40
PemK_toxin super
family
cl00995
PemK-like, MazF-like toxin of type II toxin-antitoxin
system
7974-8171 1.14E-24

133

References
1. Rebelo, A. R. et al. Multiplex PCR for detection of plasmid-mediated colistin
resistance determinants, mcr-1, mcr-2, mcr-3, mcr-4 and mcr-5 for surveillance
purposes. Eurosurveillance 23, 1 (2018).
2. Borowiak, M. et al. Development of a Novel mcr-6 to mcr-9 Multiplex PCR and
Assessment of mcr-1 to mcr-9 Occurrence in Colistin-Resistant Salmonella
enterica Isolates From Environment, Feed, Animals and Food (2011–2018) in
Germany. Front. Microbiol. 11, 80 (2020).
3. Bankevich, A. et al. SPAdes: A new genome assembly algorithm and its
applications to single-cell sequencing. J. Comput. Biol. 19, (2012).
4. Alcock, B. P. et al. CARD 2020: Antibiotic resistome surveillance with the
comprehensive antibiotic resistance database. Nucleic Acids Res. 48, (2020).
5. Li, D., Liu, C. M., Luo, R., Sadakane, K. & Lam, T. W. MEGAHIT: An ultra-fast
single-node solution for large and complex metagenomics assembly via succinct
de Bruijn graph. Bioinformatics 31, (2015).
6. Wang, C. et al. Identification of novel mobile colistin resistance gene mcr-10.
Emerg. Microbes Infect. 9, (2020).
7. Li, H. Minimap2: Pairwise alignment for nucleotide sequences. Bioinformatics 34,
(2018).
8. Tolosi, R. et al. Rapid detection and quantification of plasmid-mediated colistin
resistance genes (mcr-1 to mcr-5) by real-time PCR in bacterial and
environmental samples. J. Appl. Microbiol. 129, (2020).

Abstract (if available)

Abstract Wastewater treatment plants (WWTPs) receive a wide diversity of human-associated viruses and microorganisms, which can be used to assess the disease burden within serviced communities. This thesis aims to demonstrate several applications of how wastewater-based epidemiology (WBE) can be leveraged to track and characterize biological contaminants in wastewater to better understand how global health issues such as viral diseases and antibiotic resistant bacteria (ARB) spread, evolve, and persist within the natural and built environment. First, we tracked wastewater SARS-CoV-2 levels using stool-associated SARS-CoV-2 viral particles of the infected population in Los Angeles County. Measured wastewater SARS-CoV-2 levels were strongly correlated to in-person testing data and could be used to model the infected population within each sewershed. Next, we sequenced select wastewater SARS-CoV-2 samples from our previous study to assess the SARS-CoV-2 variant profile over nine months. Analysis of our sequencing data revealed a wide array of mutations compared to the first sequenced Wuhan SARS-CoV-2 reference genome. Further, around 67.6% of the detected mutations from our analysis were not found in sequenced clinical samples from Los Angeles. Turning our focus toward antibiotic resistance, we used metagenomic sequencing to assess the antibiotic resistance propagation risk between an aerobic membrane bioreactor (AeMBR) and an anaerobic membrane bioreactor (AnMBR) under identical operational parameters and antibiotic stress. Our results showed differential antibiotic resistance selection patterns between the AeMBR and AnMBR. Under antibiotic stress, the AeMBR enriched for ARB while the AnMBR enriched for extracellular antibiotic resistant genes. Moreover, antibiotics added to the shared influent feed showed a greater impact on the microbial diversity within the AeMBR compared to the AnMBR. Lastly, we report the first detection and characterization of the mobile-colistin resistant gene (mcr) and its variants in the urban wastewater of Los Angeles County. We assembled draft genomes of mcr-positive bacterial hosts and obtained two complete mcr-positive plasmid sequences. This thesis provides gainful information on the potential application of WBE to explore a broad spectrum of biological contaminants within our sewer network.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Fate of antibiotic resistance in anaerobic membrane bioreactors

PDF

Advancing energy recovery from food waste using anaerobic biotechnologies: performance and microbial ecology

PDF

A rigorous benchmarking of methods for SARS-CoV-2 lineage abundance estimation in wastewater

PDF

Wastewater reclamation and potable reuse with novel processes: membrane performance and system integration

PDF

The interaction of SARS-CoV-2 with CD1d/NKT antigen presentation pathway

PDF

Modulation of the renin-angiotensin system (RAS) for the treatment of COVID-19

PDF

Identification of molecular mechanism for cell-fate decision in liver; &, SARS-CoV replicon inhibitor high throughput drug screening

PDF

Exploration of latent warheads for covalent inhibition of therapeutic proteins

PDF

SARS-CoV-2 suppression of CD1d expression and NKT cell function

PDF

SARS-CoV-2 interaction with the CD1d/NKT cell antigen presentation system

PDF

Integrated technologies, blending schemes, and reuse practices to address contaminant and energy challenges in water reclamation

PDF

The analysis and modeling of signaling pathways induced by the interactions of the SARS-CoV-2 spike protein with cellular receptors

PDF

Pseudotyped viral vectors: HIV gene therapy applications and basic studies of SARS-COV-2

PDF

In2O3 COVID-19 biosensors and two-dimensional materials: synthesis and applications

PDF

Bioelectrochemical treatment of anaerobic process effluents: mitigation of dissolved methane and sulfide

PDF

Analyzing disease progression using cross-sectional data

PDF

Development of selective covalent probes to identify and modulate protein targets

PDF

Signal processing for channel sounding: parameter estimation and calibration

PDF

A low-cost and open-source automation platform for HIV surveillance

PDF

Investigating a physiological pathway for the effect of guided imagery on insulin resistance

Asset Metadata

Creator Wang, Phillip (author)

Core Title Wastewater-based epidemiology for emerging biological contaminants

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Environmental Engineering

Degree Conferral Date 2022-12

Publication Date 12/14/2022

Defense Date 09/19/2022

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag AnMBR,antibiotic resistance,COVID-19,OAI-PMH Harvest,SARS-CoV-2,wastewater surveillance,wastewater-based epidemiology

Format theses (aat)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Smith, Adam (committee chair), Childress, Amy (committee member), Mccurry, Daniel (committee member), Thrash, Cameron (committee member)

Creator Email pxwang2012@gmail.com,wang305@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC112620981

Unique identifier UC112620981

Identifier etd-WangPhilli-11364.pdf (filename)

Legacy Identifier etd-WangPhilli-11364

Document Type Dissertation

Format theses (aat)

Rights Wang, Phillip

Internet Media Type application/pdf

Type texts

Source 20221216-usctheses-batch-998 (batch), University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Repository Email cisadmin@lib.usc.edu