University of Southern California Dissertations and Theses

Mechanisms of long-term survival of Vibrio harveyi and Escherichia coli: linking damage and senescence

(USC Thesis Other)

Mechanisms of long-term survival of Vibrio harveyi and Escherichia coli: linking damage and senescence

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content
Mechanisms of Long-Term Survival of Vibrio harveyi and Escherichia coli:
Linking Damage and Senescence

By

Calista Allen

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
(MOLECULAR BIOLOGY)

August 2022

Copyright 2022 Calista Allen

ii
Acknowledgments
I have many people, both professionally and personally, to thank for the creation of this
dissertation. First, to my advisor, Steve Finkel, thank you for consistently supporting my
endeavors, allowing me to freely explore this project in my own way while being a great mentor
throughout the process. I would like to thank my committee members, Moh El-Naggar, Kenneth
Nealson, Ian Ehrenreich, and Matt Dean, as well for their contributions towards this project in
addition to being caring individuals.
Ino Chang, thank you for loving and supporting me throughout the latter half of my PhD,
taking care of my needs while I’ve focused on completing this document. I appreciate your
consistent support lifting me up and your desire to help ease my burdens. To my family members,
Steve Allen, Melanie Peterson, and John Peterson, I thank you for providing for me financially
when needed and reminding me of my drive so I could pursue my dreams of higher-level education
and becoming a scientist. Also, to my cat, Conan, for always being a part of the process when I
have worked from home during the COVID-19 shut down.
Next, I’d like to thank lab and cohort members that provided guidance and friendship
throughout this process. Thank you to Dr. Namita Shroff, Dr. Nicole Ratib, Dr. Lacey Westphal,
and Dr. Karin Kram for serving as exceptional female scientist role models; Hans Sebastian for
being my lab buddy; Dr. Alberto Robador and Dr. Christopher Corzett for the insightful comments
on my project; and my cohort members Katie Orban, Justin Cayford, Celja Uebel, Gary Molano,
and James Nhan for the comradery during our PhD journey.
Lastly, I would like to thank the organizations I was a part of while at USC – the Molecular
Biology Graduate Student Association, Women in Molecular Biology, and especially the USC

iii
Gymnastics team for providing me with a sense of self-accomplishment outside of my academic
endeavors. Thank you for always cheering me on.

iv
Table of Contents

Acknowledgments .......................................................................................................................... ii
List of Tables .................................................................................................................................. vi
List of Figures ............................................................................................................................... vii
Abstract ........................................................................................................................................ viii
Chapter 1: Linkage of Survival and Senescence of Bacterial Populations ...................................... 1
1.1 Introduction to Bacterial Survival ....................................................................................................... 1
1.2 Stresses Experienced during Long-Term Batch Culturing ................................................................. 3
1.3 Generation of Mutations within Bacterial Populations ....................................................................... 4
1.4 Individual Bacterial Aging .................................................................................................................. 5
1.5 Implications of Aging at the Population Level ................................................................................... 6
Chapter 2: Vibrio harveyi exhibits the Growth Advantage in Stationary Phase Phenotype during
Long-Term Incubation ..................................................................................................................... 8
2.1 Abstract ............................................................................................................................................... 8
2.2 Importance ........................................................................................................................................... 9
2.3 Introduction ....................................................................................................................................... 10
2.4 Results ............................................................................................................................................... 14
V. harveyi survives into Long-Term Stationary Phase when initially incubated in rich culture medium. ......... 14
V. harveyi exhibits the GASP phenotype. .......................................................................................................... 17
The V. harveyi GASP phenotype strengthens over time. ................................................................................... 21
2.5 Discussion ......................................................................................................................................... 24
2.6 Materials and Methods ...................................................................................................................... 30
Bacterial strains and growth medium. ................................................................................................................ 30
Initiation of cultures. ........................................................................................................................................... 30
GASP competition experiments. ........................................................................................................................ 31
Monitoring survival dynamics. ........................................................................................................................... 32
Chapter 3: Inequality of Daughter Cells and the Benefits of Asymmetry during Escherichia coli
Cell Division .................................................................................................................................. 36
3.1 Abstract ............................................................................................................................................. 36
3.2 Introduction ....................................................................................................................................... 38
3.3 E. coli Assembly of Daughter Peptidoglycan and Membrane at New Poles Creates Asymmetric
Aging. ...................................................................................................................................................... 40
3.4 Individual Pole Age Relates to Individual Fitness. ........................................................................... 43
3.5 Protein Aggregates Partially Responsible for Effects on Fitness Observed during Aging. .............. 44
3.6 E. coli Populations Benefit from the Asymmetric Aging Phenomenon through Enhanced Long-
Term Survival. ......................................................................................................................................... 47
3.7 Conclusion ......................................................................................................................................... 53
Chapter 4: Escherichia coli Subpopulations Exhibit Differential Fitness Over Time When
Separated by Buoyant Density ...................................................................................................... 57
4.1 Abstract ............................................................................................................................................. 57
4.2 Introduction ....................................................................................................................................... 59
4.3 Results ............................................................................................................................................... 64
E. coli viable cell distribution in Percoll gradients changes throughout long-term incubation. ......................... 64
Subpopulations can survive into Long Term Stationary Phase. ......................................................................... 73

v
Buoyant density separated subpopulations exhibit differential competitive phenotypes. .................................. 79
Stress assays reveal subpopulation phenotypes spread throughout the culture as cultures age. ........................ 81
4.4 Discussion ......................................................................................................................................... 84
4.5 Materials and Methods ...................................................................................................................... 89
Percoll gradient centrifugation and fractionation of subpopulations. ................................................................. 89
Monitoring survival dynamics in monoculture, outgrowth, and competition experiments. ............................... 90
Monitoring survival dynamics. ........................................................................................................................... 91
Stress assay. ........................................................................................................................................................ 92
Chapter 5: Future Directions for Clarifying the Linkage between Survival and Senescence of
Bacterial Populations ..................................................................................................................... 99
5.1 Linking Pole Age and Damage Levels with Percoll Gradient Subpopulations ................................ 99
5.2 Further Assessment of Competitive Phenotypes ............................................................................. 103
5.3 RNA Sequencing of Bacterial Subpopulations ............................................................................... 104
5.4 Assessment of DNA Damage and Mutation Frequencies of Subpopulations ................................. 104
5.5 Closing Remarks ............................................................................................................................. 105
References ................................................................................................................................... 106

vi
List of Tables
Supplemental Table 2.1. Average CFU/ml and standard error of monoculture Vibrio harveyi
survival curves (Fig. 2.1) ………………………………………………………………………..34
Supplemental Table 2.2. Ratios of aged CFU/mL to unaged CFU/mL and standard error at the
conclusion of a GASP assay. ……………………………………………………………………34
Table 3.1 Culminated list of fitness implication between new and old-pole cell lineages. ….....52
Supplemental Table 4.1. Average CFU Values from Fig. 4.4. …………………….…...……..98

vii
List of Figures
Figure 2.1. Vibrio harveyi long-term population dynamics in SWC Media…………………….16
Figure 2.2. V. harveyi SFP118 populations aged in conditions for 10, 20, and 30 days competed
against unaged V. harveyi SFP119……………………………………………………………….20
Figure 2.3. Competitive advantage of aged populations increases with previous exposure time to
conditions………………………………………………………………………………………...23
Supplemental Figure 2.1. Experimental design for GASP competition assays. …………….…35
Figure 3.1. Asymmetric aging of daughter cells through E. coli division……………………….42
Figure 3.2. Our model for asymmetric segregation of damage leading to a more fit population
through introduction of genetic mutations. ………………………………………………………49
Figure 3.3. Schematic of Percoll gradient centrifugation of cell cultures to respective damage
levels using buoyant density. …………………………………………………………………….56
Figure 4.1. Visual banding patterns for 1, 4, 7, and 10-Day Percoll gradients. …………………65
Figure 4.2. Percoll gradient profiles of viable cell counts for aged E. coli populations. …….67/68
Figure 4.3. Percoll gradient profiles for the percentage of cells recovered in each fraction for aged
E. coli cultures. ……………………………………………………………………………….71/72
Figure 4.4. Long-term population dynamics of subpopulations isolated from Percoll gradients..74
Figure 4.5. Outgrowth of subpopulations isolated from Percoll gradients. ……………………..78
Figure 4.6. Competitive phenotypes for Percoll gradient subpopulations……………………….80
Figure 4.7. Minimum Inhibitory Concentrations for Percoll gradient subpopulations to exogenous
stressors. …………………………………………………………………………………………83
Supplemental Figure 4.1. All replicates for fraction monocultures from Fig. 4.4. ……………..93
Supplemental Figure 4.2. All replicates of 1- and 7-day fraction outgrowths examining lag time
and logarithmic doubling times from Fig. 4.5. …………………………………………………..94
Supplemental Figure 4.3. Competitive fitness of 10-Day subpopulations (A), and 7-Day
subpopulations (B) vs unaged parental strain when inoculated at 1:1000 (vol:vol). ………….…95
Supplemental Figure 4.4. Minimum Inhibitory Concentration of Percoll gradient subpopulations
to exogenous stressors – secondary trial. ………………………………………………………..96
Supplemental Figure 4.5. Percoll gradient of diluted 1-Day culture. ………………………….97
Figure 5.1 Flow cytometry counts from mixed cultures of NADA-labeled cells and unlabeled
cells. …………………………………………………………………………………………… 101
Figure 5.2 Flow cytometry counts of dividing NADA-labeled cells inoculated in fresh medium.
…………………………………………………………………………………………………..102

viii
Abstract
The only things certain about life are aging and death. While this is typically a principle
thought about for people, it is also true for bacterial species. Bacteria can exist within many
environments and populations can persist for many years under conditions of famine, but there is
frequently a constant flux of new cells that arise and cells that die, particularly within long-term
cultures. As individual cells age within a population, they accumulate damage, experience slowed
growth rates, eventually stop dividing, and die off. Despite the negative implications associated
with aging, the damage accumulation within individual cells may allow for the generation of
additional mutations, some of which might be beneficial. Therefore, when a cell divides and two
daughter cells are produced: one old-pole receiving cell obtains the maternal damage and one
“new-pole” cell is cleansed of the maternal damage and experiences a rejuvenated growth rate.
While both the old- and the new-pole cell receive the mutations present within the maternal cell,
the new-pole daughter can express any beneficial mutations and, along with its faster growth rate
and doubling time, pass any benefits on to its progeny. This allows the population to persist while
the older individuals die off. Therefore, while aging is an inevitable fact of life with cells marching
towards death, the aging process may also contribute to the genetic diversity within the population.
Here we explore this linkage between aging and survival at the population level.

1
Chapter 1: Linkage of Survival and Senescence of Bacterial Populations
1.1 INTRODUCTION TO BACTERIAL SURVIVAL
Bacteria are found within many different environments and experience times of “feast and
famine” (Finkel, 2006), including in batch culture where periods of nutrient excess as well as
nutrient scarcity exists at different times within the culture environment. Further, many external
stresses exist within natural and laboratory environments. Bacterial populations must possess the
ability to navigate these periods of nutrient shifts, nutrient scarcity, and other external stressors.
The ability of populations to survive, whether bacterial or eukaryotic, is related to the number of
diverse strategies available to individuals within the population. Darwinian principles suggests that
the greater the amount of diversity, the more strategies the populations may possess to acclimate
to new external stresses – whether surviving nutrient scarcity or a buildup of toxins. The population
may suffer more deleterious mutations at the price of getting a few beneficial mutations, but if a
population remains genetically identical, it may not survive significant environmental shifts of any
kind. Genetic diversity is crucial for population survival. Understanding the mechanisms of
survival of bacteria within natural and laboratory environments, along with an understanding of
the mechanisms that generate genetic diversity, is an important topic to explore as it may have
direct bearing on fitness of both bacterial species and higher-level organisms.
A bacterium highlighted within Chapter 2 of this dissertation, Vibrio harveyi, experiences
these described periods of “feast and famine” within their natural marine environment. V. harveyi
exists both within a host organism (Stabili et al., 2006; Guerrero-Ferreira et al., 2013) experiencing
a feast of nutrients provided by the host, as well as planktonically in the open seawater (Nealson
and Hastings, 1979; Boedicker and Nealson, 2016) when one of their host organisms, a squid
species, expels ~95% of the bacteria each dawn into the open ocean seawater (Nyholm and McFall-

2
Ngai, 2004). The model laboratory bacterial organism, Escherichia coli, the other bacterium
highlighted within Chapters 3 and 4 of this dissertation, possesses the ability to survive long
periods of incubation within an initially rich medium, LB, without any addition of nutrients for
many years (Finkel and Kolter, 1999; Zinser and Kolter, 1999; Finkel, 2006; Ratib et al., 2021).
The batch culture environment can mimic the periods of time in which bacterial populations
experience famine, such as in within their natural environments and allows us to study the
generation of diverse strategies to deal with the various changes of environment that come with
the periods of famine or stress. After inoculation of E. coli within an initially rich medium, the
culture experiences five distinct phases: lag, exponential/logarithmic, stationary, death, and long-
term stationary phase. Lag phase is a period defined by the acclimation of the bacteria to their new,
nutrient rich environment, where no growth in population numbers is observed. Once the
population acclimates to their new environment and starts dividing this signifies the entry into
exponential phase or the logarithmic growth phase with a constant rotation of cell doubling.
Stationary phase is defined as the period in which population growth slows, and peak population
density is observed. After stationary phase, depending on specific strains and culture conditions,
99% of the population experiences cell death and the surviving fraction of the population has
entered long-term stationary phase. During long-term stationary phase the population numbers
reach a steady state, dynamic equilibrium where the “birth” and “death” rates are roughly equal,
therefore little increase or decrease in overall cell yield is observed. Long-term stationary phase
best characterizes the periods of famine that a population may experience within their natural
environment. This dissertation utilizes batch culture methodology to examine the dynamics of V.
harveyi and E. coli long-term populations.

3
1.2 STRESSES EXPERIENCED DURING LONG-TERM BATCH CULTURING
During long-term culturing of a population, many endogenous stresses accumulate within
the culture. Both oxidative species and Advanced Glycation End (AGE) products build-up within
bacterial cultures as they age throughout batch cultures or within their natural environments
(Dimitrova et al., 2004). One common form of AGEs found within aging batch cultures is
carboxymethyl lysine (CML). CML accumulates over time within a population of E. coli aging
within LB and other culture media (Pepper et al., 2010; Kram and Finkel, 2015). Reactive oxygen
species (ROS) similarly accumulate over time within LB cultures as well, as determined by
monitoring expression levels of oxyR, a regulatory gene involved in response systems to hydrogen
peroxide generated due to an increase in ROS within the cell (Kram and Finkel, 2015). Bacterial
populations have many ways of coping with the buildup of metabolic stresses during stationary
and long-term stationary phase. During long-term culturing, the Sigma factor associated with gene
regulation in response to stationary phase stress, RpoS (Sigma-S), is upregulated. RpoS alters
expression levels for ~23% of E. coli’s genome (Wong et al., 2017). For example, oxyR as
discussed above, is within the RpoS regulon. ROS detoxification enzymes such as superoxide
dismutases, catalases, and peroxidases are important defenses against oxidative species and
carbonylation (Nyström, 2005). Protein aggregation increases within mutants deficient in the
oxidative stress response system, while overexpression of a superoxide dismutase, sodA, reduced
the levels of protein aggregates in aging cultures (Maisonneuve et al., 2008). Other genes within
the heat shock, RpoH, regulon include genes that are involved in disaggregation of damaged
materials within the cell. These include proteases such as clpB and chaperone dnaK which work
to solubilize aggregates in an ATP-dependent process (Rokney et al., 2009). For all cells, a balance

4
between replication and repair exists, and after a certain point of damage the cells will no longer
be able to replicate due to shunting energy into repair, thus dying off within the culture.
1.3 GENERATION OF MUTATIONS WITHIN BACTERIAL POPULATIONS
Incubating E. coli under long-term batch culture conditions revealed an increase in fitness
when aged populations are co-cultured with unaged, parental populations, termed the Growth
Advantage in Stationary Phase (GASP) phenotype (Zambrano et al., 1993; Finkel and Kolter,
1999; Finkel, 2006; Ratib et al., 2021). This increase in fitness is due to the generation of beneficial
mutations that are selected within the population. As E. coli experiences long-term batch culturing,
the cells are constantly replaced by other novel genotypes (Ratib et al., 2021). As bacterial cultures
age throughout long-term batch culturing or throughout famine periods within their natural
environments, they need to scavenge resources when nutrients are scarce. In some cases, this
means utilizing the amino acid detritus, as well as other compounds, from their aged siblings. Some
of the mutations that arise early during LTSP culturing include enhanced ability to scavenge
nutrients and/or improved ability to resist stress (Zinser and Kolter, 1999). rpoS is frequently
mutated within the GASP populations perhaps adjusting the stress response.
Mutations within the populations of bacteria occur in many ways. During replication,
bacteria have a basal spontaneous mutation rate which is determined by the error rates of DNA
polymerases, the proofreading capabilities of the DNA polymerases, and the ability of the
mismatch repair (MMR) system, base excision repair (BER), and nucleotide excision repair (NER)
to correct the errors made by DNA polymerases. The mutational errors made by DNA polymerases
can be deleterious, neutral, or advantageous.
Mutations can also be generated in bacterial populations through damage. Reactive Oxygen
Species can interact with DNA material and introduce mutations through oxidation of guanine to

5
form 8-oxo-guanine. Then during replication, DNA polymerase incorrectly compliments the 8-
oxo-G with adenine instead of the correct pairing with cytosine increasing G-C to T-A transversion
mutations. Another mechanism of mutation introduction by damage is through AGEs.
Accumulated AGEs can form cross-links between the glycated sugars and the DNA, creating a
lesion that may stall replication.
During times of stress or in the presence of DNA damage, bacterial populations can activate
another set of “error-prone” polymerases under the SOS response system to circumvent lesions
rather than utilizing the high-fidelity DNA pol III or pol I typically utilized in “housekeeping”
DNA synthesis. These include pol IV, pol V, and pol II. Pol II is involved in error-free replication
restart, pol IV rescues the replication fork when stalled over a lesion, and pol V is involved in
translesion synthesis particularly with UV damage. Without these error-prone polymerases, E. coli
populations are less fit when competed against a parental population with these low-fidelity
polymerases (Yeiser et al., 2002; Corzett et al., 2013).
1.4 INDIVIDUAL BACTERIAL AGING
Within a long-term culture of bacteria, or within a natural population, there are individual
cells that are “older” and cells that are “younger” as defined by the generational age of their older
pole. During cell division, even though it might appear that binary fission should produce two
daughter cells that are exactly equal, one daughter inherits the older, more-damaged maternal
material (See Chapter 3 for more information). The synthesis of new peptidoglycan, cell wall, is
inserted along the mid-line of the maternal cell creating two daughter cells each with one new pole,
but the other pole is inherited from the maternal cell (De Pedro et al., 1997). One pole from the
maternal cell is older than the other due to one of her poles possessing the new peptidoglycan
produced during the generation of that maternal cell. Protein aggregates, AGEs, and ROS

6
segregate asymmetrically between the two daughter cells coinciding with the older maternal pole
(Aguilaniu et al., 2003; Lindner et al., 2008; Koleva and Hellweger, 2015; Proenca et al., 2019).
Visual protein aggregates are inherited by the old-pole-receiving daughter ~80% of the time and,
through modeling approaches, it is calculated that the “older” daughter inherits ~63% of the
maternal damage (Proenca et al., 2019). Within the older daughter, these damaged materials
continually accumulate until that cell is no longer able to divide (Boehm et al., 2016; Proenca et
al., 2019) and experiences cell death due to protein aggregate levels negatively correlating with
growth rates and doubling times (Lindner et al., 2008). Prior to cell death, the damage within the
cells could introduce DNA damage and thus introduce new spontaneous mutations, as described
above, within the population during replication and division.
1.5 IMPLICATIONS OF AGING AT THE POPULATION LEVEL
Despite the aging of individual cells eventually leading to cell death, this phenomenon
could also be an evolutionarily advantageous strategy and important for generating and
manifesting beneficial mutations within the population. If the damage to cells were equally
segregated, each daughter cell would experience negative implications from aging at the same rate,
thus reducing fitness and the overall population growth rate (Koleva and Hellweger, 2015). If any
mutations were to arise that are beneficial, both daughters would experience those benefits equally,
but with the caveat of being “weighed down” by their damage. Since the damage is asymmetrically
segregated, one daughter cell would be burdened by the damaged components, but the other would
be substantially cleansed of this damage and fully express the beneficial mutations that may have
arisen during replication of the maternal DNA material. The new-pole daughters have a
rejuvenated growth rate, thus populating more progenitors compared to their old-pole sibling
(Stewart et al., 2005; Lindner et al., 2008). Further, there is a difference between gene expression

7
activity levels between the old- and new-poles in individual cells, as well as between new- and
old-pole receiving daughters (Shi et al., 2020). Higher gene expression levels are observed near
the new-pole end of cells and this difference between new-pole and old-pole expression becomes
greater as the old-pole becomes older. The new-pole daughters may be able to express certain
genes more efficiently than old-pole daughters, allowing for shifting strategies depending upon the
particular stresses present. For example, the new-pole cells accumulated glucose for use more
quickly than old-pole siblings (Lapinska et al., 2019). While we typically think of survival and
senescence in opposition with each other, within this dissertation I will evaluate how the two are
linked and how forms of aging provide benefits to the population.

8
Chapter 2: Vibrio harveyi exhibits the Growth Advantage in Stationary Phase Phenotype
during Long-Term Incubation
The content of this chapter is essentially as published in January 2022 in Microbiology Spectrum,
10(1):e02144-21: Allen C., Finkel SE., “Vibrio harveyi exhibits the Growth Advantage in
Stationary Phase Phenotype during Long-Term Incubation.”
2.1 ABSTRACT
The bioluminescent marine bacterium Vibrio harveyi can exist within a host, acting as a mutualist
or a parasitic microbe, and as planktonic cells in open seawater. This study demonstrates the ability
of V. harveyi populations to survive and adapt under nutrient stress conditions in the laboratory,
starting in an initially rich medium. V. harveyi populations remain viable into long-term stationary
phase, for at least one month, without the addition of nutrients. To determine whether these
communities are dynamic, populations were sampled after 10, 20, and 30 days of incubation and
examined for their competitive ability when co-cultured with an unaged, parental population.
While populations incubated for 10 or 20 days showed some fitness advantage over parental
populations, only after 30 days of incubation do all populations examined outcompete parental
populations in co-culture, fully expressing the Growth Advantage in Stationary Phase (GASP)
phenotype. The ability to express GASP, in the absence of additional nutrients after inoculation,
verifies the dynamism of long-term stationary phase V. harveyi populations, implies the ability to
generate genetic diversity, and demonstrates the plasticity of the V. harveyi genome, allowing for
rapid adaptation for survival in changing culture environments. Despite the dynamism, the
adaptation to the changing culture environment occurs less rapidly than in Escherichia coli,
possibly due to Vibrio harveyi’s lower mutation frequency.

9
2.2 IMPORTANCE
Vibrio harveyi populations exist in many different niches within the ocean environment – as free-
living cells, symbionts with particular squid and fish species, and parasites to other marine
organisms. It is important to understand V. harveyi’s ability to survive and evolve within each of
these niches. This study focuses on V. harveyi’s lifestyle outside of the host environment,
demonstrating this microbe’s ability to survive long-term culturing after inoculation in an initially
rich medium, and revealing increased competitive fitness correlated with incubation time when
aged V. harveyi populations are co-cultured with unaged, parental cultures. Thus, this study
highlights the development of the Growth Advantage in Stationary Phase (GASP) phenotype in V.
harveyi populations suggesting a dynamic population with fluctuating genotype frequencies
throughout long-term, host-independent, incubation.

10
2.3 INTRODUCTION
Vibrio harveyi, a bioluminescent, ocean-dwelling microbe, occupies a number of different
environments, with numerous ecological roles. V. harveyi can exist as planktonic cells in seawater,
a mutualistic microbe, or as a parasite. As a parasitic microbe, V. harveyi causes mass mortality in
shrimp populations by disrupting stomach function upon attaching to the stomach’s chitinous
lining (Guerrero-Ferreira et al., 2013; Soonthornchai et al., 2015). Other marine organisms such
as pearl oysters, seahorses, and lobsters are similarly plagued by V. harveyi colonization of the
stomach lining (Guerrero-Ferreira et al., 2013). In mutualistic relationships, V. harveyi provides
bacterial bioluminescence or specialized metabolic capabilities in order to occupy a nutrient-rich
habitat within its host organism. Mutualistic relationships between V. harveyi populations and the
hydrozoan Aglaophenia octadonta have been reported throughout areas of the Mediterranean Sea
(Stabili et al., 2006). It is proposed that A. octadonta provides nutrients for the Vibrio species
within the chitinous structures forming the hydroid, and it is feasible that the bacteria may degrade
the chitinous material into more usable forms for the host organism. Much of our knowledge on
mutualistic relationships between bioluminescent bacteria and host organisms focuses on the
relationship between marine Vibrio and squid. This relationship is established within the squid’s
light organ where nutrients from the host are provided to the bacterium in exchange for the
bacterium’s ability to produce bioluminescence for a variety of purposes, including intraspecific
host communication, camouflage against moonlight, and attracting prey (Ruby, 1996; Guerrero-
Ferreira et al., 2013; Boedicker and Nealson, 2016; Bongrand et al., 2016).
While the specific establishment of V. harveyi symbionts within squid species is only
recently characterized (Guerrero-Ferreira et al., 2013)—in another Vibrio species, Vibrio fischeri,
colonization is better understood. Euprymna scolopes, a squid species that is solely colonized by

11
V. fischeri, hatches with no bacteria present in its two light organs and bacterial cells in the
surrounding seawater must overcome several physical and chemical obstacles and respond to other
chemical cues to colonize the juvenile squids’ light organ crypts. This initial colonization by
surrounding bacteria establishes a lifelong mutualism (Nyholm and McFall-Ngai, 2004;
Thompson et al., 2017). Once the bacteria successfully colonize the light organ, the bacteria divide
and increase in density within the light organ, resulting in the induction of bioluminescence.
While V. harveyi exists within host environments, it is frequently found as planktonic cells
in seawater in contrast to other closely related, bioluminescent bacterial species primarily existing
within host organisms (Hastings and Nealson, 1977). Many studies confirm V. harveyi’s existence
in the planktonic form, maintaining relatively high cell density, in both coastal and surface
seawaters (Nealson and Hastings, 1979; Boedicker and Nealson, 2016). The numerous possible
niches for V. harveyi likely explains its ability to use a broad range of substrates for growth within
laboratory environments (Hastings and Nealson, 1977; Nealson and Hastings, 1979; Nealson and
Hastings, 2006). Much of the research on Vibrio species tends to focus on the bacterium’s role in
relation to a host organism; the planktonic lifestyle of Vibrio harveyi remains less well understood.
Further, Vibrio harveyi has not previously been examined for evidence of adaptive evolution
throughout long-term incubation and is the focus of this study.
It is well established that Escherichia coli populations possess the ability to survive long-
term batch culture for many years after inoculation into a rich medium such as LB, without any
further addition of nutrients (Finkel and Kolter, 1999; Zinser and Kolter, 1999; Finkel, 2006; Ratib
et al., 2021). The inoculated culture initially experiences four phases (lag phase,
exponential/logarithmic phase, stationary phase, and death phase) before transitioning into the
final phase, long-term stationary phase (LTSP). During LTSP, subpopulations with beneficial

12
mutations including, for example, an enhanced ability to scavenge nutrients and/or improved
ability to resist stress, increase in frequency taking over the population (Zambrano et al., 1993;
Finkel, 2006; Kram and Finkel, 2014; Kram and Finkel, 2015; Ratib et al., 2021). The benefit to
cultures aging into long-term stationary phase was first observed by introducing a sample of a 10-
day-old, aged E. coli population as a minority into a majority, 1-day-old parental population to
examine the dynamics between the two populations in competition, revealing the enhanced relative
fitness of cells from the aged population. When in competition with the parental population, the
10-day subpopulation overtakes the unaged cells within a few days—termed the Growth
Advantage in Stationary Phase (GASP) phenotype; the aged population is ultimately able to drive
the parental lineage to extinction (Zambrano et al., 1993; Finkel and Kolter, 1999; Finkel, 2006;
Ratib et al., 2021).
The GASP phenotypes observed within a competition experiment can be categorized into
four different classes depending upon the final cell densities of the two populations (Finkel, 2006).
Class I populations distinctly outcompete the ancestral lineage at the conclusion of the
experiment—typically exhibiting at least a 100-fold fitness advantage and frequently
demonstrating the ability to drive the ancestral population to extinction. Class II GASP populations
increase in frequency to achieve similar cell counts to the ancestral population, but do not drive
the parental population to extinction during the transition into long-term stationary phase; instead
both populations can co-exist for long periods of time. Evolved populations that show an initial
increase in cell density, but do not succeed in surpassing or reaching cell densities equivalent to
the ancestral population fall into class III. Finally, aged populations that never show any increase
in frequency after introduction into the ancestral lineage and eventually die out, do not possess the
GASP phenotype, and fall into class IV.

13
The ability to survive in multiple different environments would suggest that V. harveyi
populations possess a level of genomic plasticity sufficient to provide enough genetic diversity to
allow for adaptation and survival in these various niches. V. harveyi in planktonic form, living in
the open water—where the availability of nutrients may remain low before encountering another
host—suggests that the organism can survive under conditions akin to long-term stationary phase.
After prolonged incubation in LTSP, populations may adapt to those conditions, potentially
express the GASP phenotype after evolving under LTSP conditions, and thus could specialize
more in an open seawater type environment without a host. V. fischeri, a bacterium occurring more
strictly within a host, has been examined for the GASP phenotype and shows evidence for type II
GASP with populations aged 7 (Petrun and Lostroh, 2013), 10 (Maher, Bansal, and S. Finkel—
unpublished results), or 22 days (Soto et al., 2019). While varying in experimental design, in each
experiment cells achieve similar population densities as the unaged cells during co-culture. V.
harveyi, a free-living planktonic bacterium, existing within multiple environments, could have
more targets for adaptive mutations within its genome to manifest the GASP phenotype through
long-term culturing in the laboratory. Potentially, costly mutualistic traits, such as
bioluminescence which consumes ~12% of Vibrio harveyi’s usable oxygen, may be lost or affected
by long-term culturing as selection for these traits may be relaxed within the free-living
environment (Makemson, 1986). This study aims to understand V. harveyi’s ability to survive into
long-term stationary phase and to determine whether populations incubated for prolonged periods
are able to adapt to LTSP conditions and manifest the GASP phenotype. The data presented
demonstrates the expression of the GASP phenotype in aged V. harveyi populations; although, the
mode and tempo of acquiring the phenotype may take longer to manifest under laboratory
conditions, compared to Escherichia coli.

14
2.4 RESULTS
V. harveyi survives into Long-Term Stationary Phase when initially incubated in rich culture
medium. After a month-long, batch monoculture incubation in the rich medium SWC, both the
parental Vibrio harveyi and its nalidixic acid-resistant derivative (SFP119) experience all five
phases usually observed in the laboratory, with populations reaching cell densities of ~2 x 10
8

CFU/mL at the end of log phase (Fig. 2.1 and Supplemental Table 2.1); both strains display a
stationary phase lasting for one day. Similar to E. coli, following the peak stationary phase density,
the population enters death phase where the cell density decreases 10-fold to ~2 x 10
7
CFU/mL.
The cell density of the nalidixic acid resistant strain decreases further during death phase compared
to the parental population. After death phase the cell densities for both strains increase, reaching
concentrations of ~10
8
CFU/mL by day 4 of incubation, but only briefly before dropping again to
~10
7
CFU/mL by day 5, where the cell densities remained for the duration of the experiment. Upon
continued incubation, the viable cell counts of both strains fluctuate with various “dips” in
population densities, with both strains most frequently experiencing the dips simultaneously (Fig.
2.1 – Days 2-3, 10-12, 16-17). All replicates of the nalidixic acid-resistant strain experience a dip
on day 22, while the replicates of the unmarked strain do not. Despite these minor differences, the
population densities of the two strains generally track closely with each other. The results shown
in Fig. 2.1 highlight V. harveyi’s ability to survive into long-term stationary phase for at least one
month without the addition of nutrients. Further, no significant differences in the long-term
survivability between the parental and nalidixic acid-resistant strain are observed (Fig. 2.1), which
is important for conducting competition experiments between the strains where at least one drug-
resistant isogenic strain is necessary to differentiate strains for tracking the differently aged sub-
populations during competition experiments. In E. coli GASP competitions, typically two

15
antibiotic-resistant strains are utilized allowing for separate tracking of each sub-population;
however, in V. harveyi, nalidixic acid was the only antibiotic tested which yielded a stable,
spontaneously-resistant strain despite examining a wide range of antibiotics; (the strain is naturally
resistant to to streptomycin, spectinomycin, and ampicillin; and no spontaneous drug-resistant
mutants were identified for kanamycin, chloramphenicol, or rifampin). V. harveyi’s ability to
survive in long-period batch cultures under conditions of nutrient deprivation in LTSP lead us to
ask whether cells in aged cultures evolve to adapt to LTSP conditions and thus will exhibit a fitness
advantage over the parental strain, akin to the GASP phenotype (Zambrano et al., 1993; Finkel
and Kolter, 1999; Finkel, 2006).

16

Figure 2.1. Vibrio harveyi long-term population dynamics in SWC Media. Average viable cell
counts (with standard error) throughout a month-long period of V. harveyi SFP118 (black) and
nalidixic acid resistant V. harveyi B-392 SFP119 (gray) monocultures aerated in test tubes at 30
0
C
demonstrates the ability of V. harveyi cultures to persist into long-term stationary phase without
an addition of nutrients (n=3).
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ● ●
●
●
●
●
●
●
●
5
6
7
8
9
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
Day
Log
10
CFU/mL

17
V. harveyi exhibits the GASP phenotype. To test for the expression of the GASP phenotype, the
three cultures incubated into LTSP in Fig. 2.1 were sampled after 10, 20, and 30 days; samples
were frozen at the appropriate time and utilized to initiate aged starter cultures for experiments
(Supplemental Fig. 2.1). A sample of each aged starter culture was used to inoculate three separate
high density, unaged, parental cultures to test for the expression of the GASP phenotype (Fig. 2.2).
In total, competitive phenotypes were determined for nine replicates for each LTSP incubation
timepoint (Fig. 2.3A). In addition to the co-culture competitions between aged population samples
and the unaged strain, co-culture competitions between two unaged populations, one strain
possessing nalidixic acid resistance and one strain unmarked, were performed by inoculating a
small sample (5 µl) of each unaged population into one test tube to serve as an unaged control
(Fig. 2.3A – Unaged); the unaged co-cultures reached similar population densities for each
population at the conclusion of the competition experiment (Fig. 2.3B – Unaged), revealing no
consistent advantage for either unaged parental strain. The unaged competitions provide a baseline
for comparison when examining the results from co-cultures with aged sub-populations. The
competitions performed between the aged populations and unaged V. harveyi revealed that
populations incubated into LTSP can express the GASP phenotype. The relative competitive
fitness of the aged populations compared to the parental population increases with incubation time
in LTSP, with the GASP phenotype shifting from class II after 10 days of prior incubation to class
I after 20 to 30 days (Fig. 2.2 and 2.4).
Each panel in Fig. 2.2 represents average population densities of competitions performed,
in triplicate, from each of the three originally aged samples against the unaged parent. 10-day V.
harveyi populations (Fig. 2.2 – A, D, G) in competition with the unaged parent (Fig. 2.2) grow in
the presence of the parental cells throughout the early time points of the experiment, outnumbering

18
parental cells by ~100-fold by Day 8 of co-incubation; however, with continued incubation
(between days 10 and sixteen), the parental population increases in population density ultimately
reaching similar densities as the “evolved” population. This result can be classified as a class II
GASP phenotype (Finkel, 2006).
Populations sampled further into long-term stationary phase incubation at 20 days (Fig. 2.2
– B, E, H), allowing additional time for cells to evolve under LTSP conditions, show a more robust
GASP phenotype. Similar to the populations aged for 10 days, the 20-day populations exhibit
higher relative fitness than the unaged population throughout early long-term stationary phase, but
here the 20-day populations outnumber the parental majority by ~100-fold by day 5 rather than on
day 8 as observed for 10-day-old populations. As co-culture incubation continues, the phenotypes
observed for each of the 20-day aged samples start to diverge. One 20-day sample (Fig. 2.2B and
Fig. 2.3 – 20-B) ultimately converges with the parental population by the end of the experiment,
similar to what is observed with 10-day populations. However, the two other samples of 20-day
populations clearly outcompete the unaged population by more than 100-fold by the end of the
experiment (Fig. 2.2 – E, and H; and Fig. 2.3 – 20-A and 20-C).
The populations incubated for 30 days prior to competition show the strongest GASP
phenotypes (Fig. 2.2 – C, F, and I). Unlike cultures aged for 10 or 20 days, cells from the 30-day-
old populations first show enhanced relative fitness on day 1 and the enhanced fitness continues
into early long-term stationary phase, where replicates of all three aged population samples
examined outnumber the parental population. While the parental populations rebound briefly—
ultimately the 30-day populations dominate the culture by the end of the experiment. 30-day
populations possess an ~100-fold relative fitness advantage for two of the three examined samples
(Fig. 2.2 – C and I), with the third sample showing an even stronger phenotype, consistently driving

19
the parental population below the limit of detection by the conclusion of the experiment (Fig. 2.2
– F).

20

Figure 2.2. V. harveyi SFP118 populations aged in conditions for 10, 20, and 30 days
competed against unaged V. harveyi SFP119. Each panel shows the average viable cell counts
(with standard error, n=3) for co-cultures of previously aged cultures (black) for 10 days (left
panel- squares), 20 days (middle panel- triangles), or 30 days (right panel- diamonds) when
introduced as a minority (5 microliters) into a dense culture (5 mL) that has not been exposed to
culture conditions (unaged in all panels- gray circles).

21
The V. harveyi GASP phenotype strengthens over time. Our results show the strength of the
GASP phenotype is correlated with the length of long-term stationary phase incubation time. To
provide a summary of each competition and to determine whether culture age correlates with
competitive ability, the ratio of the unmarked aged population to the nalidixic acid resistant unaged
population at the conclusion of each co-culture experiment was calculated and plotted in Fig. 2.3A
(and Supp. Table 2.2). Since three independent cultures were originally aged and each of these
were examined for the GASP phenotype in triplicate, a total of nine replicates are shown for each
timepoint. In addition to the GASP competitions, the ending ratios of the co-culture competitions
between the two unaged populations are included in this analysis as a baseline for relative fitness
between the strains (Fig. 2.3A – Unaged). Fig. 2.3A and Table S2 shows the ratios of the final
population yields at the end of each individual competition performed. 10-day populations sampled
from replicates 10-A and 10-B outnumbered the parental populations by approximately 10-fold
while competitions performed with samples from replicate 10-C reached similar cell densities as
the parent. Samples from replicate 20-A have lower relative competitive fitness and reach similar
population densities as the parental population, while competitions with populations 20-B and 20-
C successfully outcompete the parental population. Each of the replicates aged to 30 days
outcompeted the parental populations by at least 1000-fold.
The relative fitness ratios were then averaged for each incubation time (unaged, 10 days,
20 days, and 30 days) prior to co-culture competition to determine the correlation between relative
competitive fitness and incubation time in LTSP. On average, cells aged for 10 days can express
the GASP phenotype and outnumber unaged cells by at least 7-fold. When aged for 20 days, aged
cells overtake the population, outnumbering the unaged cells by ~350-fold on average. Cells aged
for 30 days exhibit the strongest GASP phenotype and have final cell ratios that are significantly

22
different compared to all other competitions (ANOVA: p = 1.92 x 10
-2
for Unaged, 7.46 x 10
-4
for
10-Day, and 3.76 x 10
-3
for 20-Day). On average, the 30-day populations outcompete the unaged
cells by ~2400-fold at the conclusion of the co-culture competition.

23
Figure 2.3. Competitive advantage of aged populations increases with previous exposure
time to conditions. (A) Final ratios of the previously aged population to the unaged population at
the conclusion (day 28) of each competition tested compared to the final ratio of competitions
between two unaged populations (Unaged). Each replicate was examined three times and are
grouped by having the same letter (A, B, or C—Corresponding with Figure 2.2) while the number
corresponds to the incubation time prior to competition (B) The final ratios for all competitions
(n=9) averaged (with standard error) by initial age with significant differences depicted from
ANOVA (* = p<0.05, ** = p<0.01, *** = p<0.001).

24
2.5 DISCUSSION
This study examines the long-term survivability of Vibrio harveyi and whether the GASP
phenotype is manifested in this bacterial species, as it is for Vibrio fischeri and Escherichia coli.
By inoculating V. harveyi into a rich medium and assessing the viable population density over
time, it was determined that V. harveyi maintains viability throughout long-term stationary phase
culture for at least 30 days (Fig. 2.1). Previously it has been shown that V. harveyi populations
maintained viable counts up to 14 days with a slight reduction in viability, but in a larger volume
of marine broth, a similar rich medium (Kaberdin et al., 2015). Furthermore, our study provides
evidence that V. harveyi cultures acquire the GASP phenotype throughout LTSP incubation,
adding another bacterial organism to the pantheon of GASP. The correlation between the relative
strengthening of the GASP phenotype with LTSP incubation time, and the similarity of GASP
phenotypes between replicate aged cultures (Fig. 2.2 and Fig. 2.3), provides evidence for cell
turnover and altered genotype frequencies throughout long-term incubation, suggesting adaptive
evolution is occurring. Similar to E. coli, the aging V. harveyi parental population is likely being
replaced with subpopulations selected for their enhanced LTSP survival by possessing beneficial
mutations, allowing for GASP phenotype progression (Finkel, 2006; Ratib et al., 2021).
Though aged V. harveyi populations manifest the GASP phenotype, there are some notable
differences from the phenomenon originally observed in E. coli. While V. harveyi cells aged for
10 days can initially overtake unaged populations, this does not occur in all replicates (Fig. 2.3A).
This is unlike populations of E. coli aged for 10 days in rich medium where virtually all aged
populations drive the unaged, parental population below the limit of detection (Finkel, 2006),
possibly due differences in mutation frequency and population yield after exponential phase,
discussed below. Further, when cells from 10-day-old V. harveyi cultures do initially overtake the

25
majority population, the aged and unaged populations are at equal cell densities by the end of the
experiment (Fig. 2.3 – A and B), exhibiting a class II GASP phenotype (Finkel, 2006). This is
consistent with previous studies that found the closely related organism, V. fischeri, displays a
class II GASP phenotype after aging for 7 days or 22 days in LTSP (Petrun and Lostroh, 2013;
Soto et al., 2019).
When using the final cell ratio of the aged to unaged populations in the co-culture
competitions as a measure of relative fitness, a significant strengthening of the GASP phenotype
is observed with the length of LTSP incubation time prior to competition. When the populations
are aged for 20 days, two of the three aged populations examined for GASP demonstrated a class
I phenotype (Fig. 2.2 – E and H), compared to the 10-day samples which all exhibited a class II
GASP phenotype (Fig. 2.2 – D and G); at 30 days, all populations exhibited a class I GASP
phenotype (Fig. 2.2 – C, F and I). A majority of the competitions between 30-day populations and
unaged populations resulted in the detection of only the aged populations at the conclusion of the
experiment, with the unaged population frequently below the limit of detection, highlighting the
increase in relative fitness throughout LTSP incubation (Fig. 2.3).
As a population ages, the observed strengthening of the GASP phenotype is likely due to
an accumulation or replacement of novel mutations providing enhanced survival in LTSP. This
replacement of mutations and accumulation throughout time has been confirmed for E. coli, where
populations aged for 10 days possess numerous genotypes. In one experiment, sequencing 10-day-
old clones from 4 different aging populations revealed 27 novel mutations with no detection of the
parental genotype (Ratib et al., 2021). When examining the final sub-population ratios in V.
harveyi competitions, it is clear that 30-day populations exhibit the strongest GASP phenotype;
however, it is worth examining the co-culture dynamics over the course of all competitions. While

26
V. harveyi populations aged for only 10 days do not exhibit a strong GASP phenotype relative to
20- and 30-day populations, the 10-day populations are the most competitive against the parent at
day 10 of co-incubation, showing the highest relative cell density for that timepoint (Fig. 2.2 – A,
D, G compared to all other panels on day 10). After 10 days of aging in monoculture, cells likely
possess mutations allowing for better survival through day 10 compared to the parent, but the
benefit received from the mutations at day 10 does not continue further into LTSP or under
conditions of co-incubation, thus resulting in a weaker GASP phenotype at the conclusion of the
experiment. Further aging in monoculture prior to competition is necessary for long-term
competitiveness against the parent in prolonged co-culture experiments. However, despite the
stronger GASP phenotype, 20- or 30-day samples show reduced fitness in co-culture on day 10
suggesting a form of antagonistic pleiotropy; mutations beneficial in extended LTSP may not be
beneficial at earlier time points and are thus not selected for early on. The data supports a model
where the genetic composition of the aging population is highly dynamic and is changing
throughout LTSP incubation. These genetic differences may cause the GASP phenotype to shift
from class II at 10 days of aging to class I upon further incubation where the aged population
drives the unaged population below the limit of detection, similar to the phenomenon observed in
E. coli at 10 days of aging. The different competitive phenotypes and the overall strengthening of
the GASP phenotype correlated with culture age provides evidence for dynamic V. harveyi
populations during LTSP.
While each of the aged populations achieves a strong GASP phenotype, likely due to
beneficial mutations for enhanced LTSP survival, the evolutionary paths within the aged
populations almost assuredly differ from each other and differ from what is observed for E. coli.
Population A (Fig. 2.2 – A, B, C and Fig. 2.3A: 10-A, 20-A, and 30-A) does not acquire the class

27
I GASP phenotype until aged for 30 days compared to populations B and C that start to exhibit a
stronger GASP phenotype by day 20. It is probable that each population is accruing different
mutations, some of which provide a stronger advantage during LTSP. This provides evidence for
multiple paths of evolution all ultimately resulting in improved relative fitness during LTSP. The
dynamics of nutrient availability within the aging cultures throughout LTSP incubation are
unknown and may significantly differ as cultures progress from 10 to 30 days. Further, changes in
nutrient availability likely will differ from culture to culture depending on what genotypes are
present within a given population. As novel mutations appear, the shift in nutrient availability, and
thus selection for different mutations within each culture likely creates the differences observed in
the GASP phenotypes, yet all mutants improve fitness over time. Both the environments and the
specific paths of evolution are complex and the “beneficial” mutations for each aging Vibrio
population may differ from each other and from those observed in E. coli populations.
Why the GASP phenotype is not fully manifested by day 10 in Vibrio populations, both as
observed here, and in the studies of V. fischeri (Petrun and Lostroh, 2013; Soto et al., 2019),
compared to 10-day E. coli populations, is not understood, however, we can speculate. One
important difference between the population dynamics of E. coli evolving within LB cultures and
V. harveyi evolving within SWC cultures is the maximum population density achieved by each
species in stationary phase. E. coli populations reach a maximum population density (Finkel, 2006)
at least one order-of-magnitude greater than that observed in V. harveyi (~5 x 10
9
compared to ~2
x 10
8
CFU/mL). During exponential phase growth, mutations are introduced within the population
primarily through random errors made during DNA replication. Each of these mutations can either
be beneficial, neutral, or deleterious to the individual bacterium. The greater numbers of cells in
E. coli cultures almost assuredly leads to additional genetic diversity within the larger population.

28
Therefore E. coli populations have a greater chance of possessing strong beneficial mutations.
After death phase, similar population densities are observed for E. coli and V. harveyi populations
with both species entering into long-term stationary phase at ~2 x 10
7
CFU/mL. After death phase,
cells possessing beneficial mutations for LTSP survival increase in frequency within the
population. E. coli, generating enough genetic diversity during exponential phase, possesses
genotypes with a strong GASP phenotype by day 10; while V. harveyi’s lower cell yield throughout
exponential phase reduces the chance for a strong beneficial mutant present early on, and thus
necessitates further incubation with cell turnover for a class I GASP population to appear.
This model is further supported when examining the differences in spontaneous mutation
frequency between E. coli and Vibrio species. Previous studies report that the Vibrio genus has a
lower base-substitution mutation rate compared to other bacteria. V. fischeri has a rate of 2.07 x
10
-10
mutations/base pair/generation and Vibrio cholerae a rate of 1.07 x 10
-10
(Dillon et al., 2017).
Per genome, this means that V. fischeri experiences 8.85 x 10
-4
base-pair substitutions and V.
cholerae experiences 4.38 x 10
-4
base-pair substitutions when factoring in genome sizes (~4.23 x
10
6
and ~4.09 x 10
6
, respectively). Estimates for E. coli show a mutation rate of 2.2 x 10
-10
per
nucleotide but when factoring in genome size (~4.6 x 10
6
bp) this results in 1 x 10
-3
mutations per
genome, much greater than the frequency observed for Vibrio strains (Lee et al., 2012) The
generation of fewer mutations in the Vibrio, along with the decreased cell yield at stationary phase,
may contribute to the delay in the manifestation of a strong GASP phenotype within the Vibrio
genus simply because the chance of getting a beneficial mutation within the Vibrio population is
lower when compared to E. coli. However, after more time to generate (by chance) and accumulate
beneficial mutations during the dynamic LTSP, a strong GASP phenotype can occur in V. harveyi.
While the specific mutations present within the aged populations of V. harveyi and the mutational

29
spectrum of V. harveyi populations throughout time will need to be determined, this study has
revealed the expression of the GASP phenotype in Vibrio harveyi and the shift in phenotype
strength through long-term stationary phase culturing, reflecting a level of genotypic adaptation
not previously observed.

30
2.6 MATERIALS AND METHODS
Bacterial strains and growth medium. SWC, a rich culture medium, was utilized for both liquid
culturing and plating of V. harveyi. SWC is an artificial seawater medium containing sodium
chloride (2.33%), magnesium sulfate (1.85%), calcium chloride (0.22%), and potassium chloride
(0.11%); with peptone (0.50%), glycerol (0.30%), and yeast extract (0.05%) as carbon sources
(Nealson, 1978; Wimpee et al., 1991; Setiawan et al., 2015). V. harveyi strain SFP118, derived
from B392 (MAV/Photobacterium fischeri/Benecka harveyi) (Reichelt and Baumann, 1973;
Baumann et al., 1980), and a spontaneous nalidixic acid resistant mutant (SFP119) were utilized
for this study. SFP119, the spontaneous nalidixic acid resistant strain, was isolated from SFP118
using a previously published protocol (Yeiser et al., 2002; Petrun and Lostroh, 2013; Kram et al.,
2017). Briefly, antibiotic-sensitive cells of SFP118 were plated onto SWC plates containing 20
µg/mL nalidixic acid to select for spontaneously resistant mutants able to form colonies in the
presence of the antibiotic (Zambrano et al., 1993). The survival dynamics of populations initiated
from these resistant colonies were screened throughout LTSP to identify a mutant strain
phenotypically identical to the parental strain in the absence of the drug selection, to be utilized
for tracking throughout co-culture incubation between aged and unaged strains. (Other antibiotics,
streptomycin, spectinomycin, ampicillin, kanamycin, chloramphenicol, and rifampin, were tested
at various concentrations in a similar manner to nalidixic acid, but we were unable to select for a
reliable spontaneously resistant mutant population).
Initiation of cultures. For each experiment, cultures were inoculated from a frozen stock (stored
in SWC plus 20% glycerol at -80°C) into 18 x 150 mm borosilicate test tubes containing 5 mL of
SWC medium. Cultures were incubated with aeration for two days at 30°C in a rolling drum (TC-
7; New Brunswick Scientific, Edison, NJ) to achieve maximal population density. These starter

31
cultures were used to initiate experiments by inoculating fresh medium at a 1:1000 (vol:vol)
dilution. At least three biological replicates were generated in this manner and used for all
experimental conditions. For the experiments depicted in Fig. 2.1, this transfer represents day 0.
After 10 days, 20 days and 30 days of incubation, 100 µl of each of the biological replicates were
sampled and frozen in SWC plus 20% glycerol at -80°C for utilization in GASP experiments
(Supplemental Fig. 2.1).
GASP competition experiments. For the GASP competitions, “aged” starter cultures are initiated
from the frozen populations described above (three for each age) and incubated for two days to
achieve high cell density. After reaching high cell density, a sample of each aged starter culture is
used to inoculate three separate high density, unaged, SFP119 parental cultures possessing
resistance to nalidixic acid at a 1:1000 (vol:vol) ratio to establish day 0 of the GASP competition
(Supplemental Fig. 2.1). Each of the unaged, parental cultures inoculated in the GASP
competitions were started from the same original parental starter culture in the manner described
above. Each of the unaged, parental cultures were incubated for two days, prior to initiation of the
GASP experiment, to achieve high cell density. This positions the parental culture at high
population density and cells from the aged culture at low population density at the start of the
GASP competition. Each aged starter culture is examined in triplicate, thus generating nine total
replicate GASP competitions for each age. After GASP competitions were initiated, cell densities
of the two populations throughout the co-incubation period were measured over 28 days (Fig. 2.2).
To serve as a control for comparison to the aged GASP competitions, three replicate co-culture
competitions between two unaged strains (an unaged, unmarked population and an unaged,
nalidixic acid-resistant population) were performed with both strains inoculated at similar
densities. This sets each population at a similar low cell density at the start of the competition.

32
Monitoring survival dynamics. Population densities for the monoculture, control, and GASP
competitions were determined by quantifying viable cell counts by titering appropriate dilutions
of cells, and plating in the presence or absence of nalidixic acid as appropriate. The limit of
detection for this method is <1000 CFU/mL (Kraigsley and Finkel, 2009). For monoculture
experiments, dilutions were plated on SWC agar to determine viable cell counts of the total
population, in this case, the one strain present in the culture. For co-cultures, total cell counts,
including both the unaged and aged populations, were determined by plating on SWC, while the
unaged population (nalidixic acid resistant strain) density was determined independently by plating
onto SWC plates containing 20 µg/mL nalidixic acid. The aged population density was then
calculated by subtracting the cell counts observed on the nalidixic-acid containing SWC plates
from the total observed counts on the SWC plates. This method is similarly applied for control
competitions between the two parental strains (SFP118 and SFP119). At day 0 in GASP
experiments, the unmarked, aged population is undetected because it comprises so little of the total
population that viable counts from the nalidixic acid resistant strain and total counts are similar.
In this case, the aged population is assigned a value 1000-fold lower than that of the detected
nalidixic acid resistant strain since it was inoculated at a 1:1000 (vol:vol) ratio. Due to the limit of
detection for the aged strain, after day 0, if the aged population is undetected at a time point but is
at or above the unaged population counts in the previous or subsequent time point, it is assigned
equal counts to the unaged, nalidixic acid-resistant strain. This correction was only necessary for
one or two timepoints within the first four days of co-culture incubation (Fig. 2.2, ~Day 1-4; 3A:
Day 2, 3B: Day 2, 3C: Day 2-4, 3E: Day 2, 3F: Day 2-3, 3G: Day 1, 3H: Day 2-3, 3I: Day 2). For
each experiment, viable cell counts were averaged between biological replicates for each
strain/population with the standard error indicated (Fig. 2.1 and Fig. 2.2). Standard error is

33
calculated using the standard deviation in R of the viable cell counts divided by the square root of
the length of the data.
***

34
Supplemental Table 2.1. Average CFU/ml and standard error of monoculture Vibrio harveyi
survival curves (Fig. 2.1)
Parental Nalidixic Acid Resistant
Day CFU Standard Error CFU Standard Error
1 2.8 x 10
8
3.3 x 10
7
2.1 x 10
8
1.8 x 10
6
2 2.6 x 10
7
4.4 x 10
6
1.2 x 10
7
9.1 x 10
5

3 3.3 x 10
7
2.4 x 10
6
2.3 x 10
7
2.0 x 10
6
4 1.4 x 10
8
1.3 x 10
7
1.4 x 10
8
1.5 x 10
7
5 4.4 x 10
7
1.9 x 10
6
2.4 x 10
7
1.3 x 10
6
6 3.8 x 10
7
2.2 x 10
6
1.5 x 10
7
1.4 x 10
6
7 1.7 x 10
7
2.1 x 10
6
1.7 x 10
7
3.4 x 10
6
8 2.8 x 10
7
1.2 x 10
6
3.4 x 10
7
1.6 x 10
6
9 2.6 x 10
7
1.3 x 10
6
2.0 x 10
7
4.1 x 10
6
10 1.9 x 10
7
9.7 x 10
5
1.8 x 10
7
1.8 x 10
5
11 1.1 x 10
7
2.5 x 10
6
1.9 x 10
7
4.8 x 10
5
12 4.0 x 10
6
1.6 x 10
6
7.9 x 10
6
1.5 x 10
6
14 2.4 x 10
7
8.0 x 10
5
2.1 x 10
7
6.3 x 10
5
15 1.5 x 10
7
1.8 x 10
6
2.1 x 10
7
1.9 x 10
6
16 1.7 x 10
7
7.3 x 10
5
2.0 x 10
7
2.3 x 10
6
17 1.2 x 10
7
1.0 x 10
6
1.0 x 10
7
1.8 x 10
5
18 1.4 x 10
7
1.7 x 10
6
1.4 x 10
7
1.3 x 10
6
19 1.7 x 10
7
8.4 x 10
5
1.3 x 10
7
6.6 x 10
5
20 1.5 x 10
7
2.0 x 10
6
1.4 x 10
7
1.1 x 10
6
22 1.4 x 10
7
8.0 x 10
5
4.3 x 10
6
2.4 x 10
5
24 1.0 x 10
7
0.0 x 10
0
1.2 x 10
7
4.8 x 10
5
30 4.3 x 10
6
1.6 x 10
5
5.7 x 10
6
1.3 x 10
6

Supplemental Table 2.2. Ratios of aged CFU/mL to unaged CFU/mL and standard error at
the conclusion of a GASP assay
Unaged Control 10 Day 20 Day 30 Day
All Populations 0.72 ± 0.46 7.01 ± 3.65 347.95 ± 140.52 2416.83 ±
673.08
Population A 10.54 ± 9.25 0.65 ± 0.30 1643.27 ±
1428.74
Population B 9.64 ± 6.72 263.63 ± 94.63 4100.00 ±
115.47
Population C 0.84 ± 0.36 779.56 ± 266.42 1507.21 ±
1119.07

35

Supplemental Figure 2.1. Experimental design for GASP competition assays. Three parental
cultures are incubated into long-term stationary phase and are sampled after 10, 20, and 30 days.
The frozen samples are used to generate starter cultures and incubated to high cell density. A
sample of the aged starter cultures is inoculated as a minority into three replicate parental cultures
that have been incubated for 2 days to achieve high cell density. Each of the three aged cultures
are tested in triplicate generating nine total replicates for each age.

36
Chapter 3: Inequality of Daughter Cells and the Benefits of Asymmetry during Escherichia
coli Cell Division

3.1 ABSTRACT
While it may appear that Escherichia coli binary fission creates identical daughter cells,
there is an inherit asymmetry during cell division of the segregation of older maternal components,
generating daughter cells with differential fitness levels. During E. coli cell division, new
peptidoglycan is integrated at the midline of the cell, resulting in each daughter cell having one
new pole of contemporaneously synthesized peptidoglycan and one pre-existing pole, containing
maternal cell wall and membrane. Consequently, while both daughters each inherit one new pole,
one daughter cell inherits the younger maternal pole created during the previous generation of the
maternal cell, while the other daughter inherits the older maternal pole which can be many
generations old. Damaged maternal components, such as protein aggregates, tend to localize
towards and are inherited along with the older pole each generation. Receiving older maternal
components has negative fitness implications for the individual cell receiving the damaged
material, but allows the other daughter cell to receive less damage and enter a rejuvenated state to
propagate the population forward. Despite the negative consequences for old-pole receiving cells,
the asymmetric segregation of damage has larger population benefits. This review examines the
asymmetry that occurs upon E. coli cell division, which cellular components are affected by this
asymmetry, and the fitness implications of the asymmetric segregation at both the cellular and
population levels. Asymmetric aging may directly provide evolutionary benefits to E. coli, and
other microbes that reproduced via binary fission, by creating variation in growth rates and death

37
dynamics, as well as by allowing the accelerated introduction of beneficial mutations within a
population.

38
3.2 INTRODUCTION
During Escherichia coli cell division, the insertion of newly synthesized peptidoglycan is
at the midcell septation site (De Pedro et al., 1997) and creates a generational age gap between the
two poles once cell division and daughter cell formation is complete (Stewart et al., 2005). In each
daughter cell, one pole is newly synthesized, and one pole is inherited from the maternal cell. The
generation of new peptidoglycan at one cell pole means that one daughter inherits the younger,
maternal cell pole, previously synthesized upon the creation of the maternal cell, and the other
daughter inherits the older maternal cell pole, passed down from previous generation(s). The
asymmetry between the poles, and thus the daughter cells, creates a landscape of various pole ages
within a population of cells from the same generation (Figure 3.1). This differential aging between
the two daughters extends further to other components of the cell. Along with the older maternal
pole, the daughter cell may also inherit damaged protein aggregates (Lindner et al., 2008; Proenca
et al., 2019). The “older” daughter cell suffers consequences from inheriting the older components.
The growth rate is reduced causing a subsequent reduction in offspring biomass of the older pole-
receiving lineages. The cells receiving the younger maternal pole, the “new-pole cells”, experience
a rejuvenated growth rate (Stewart et al., 2005; Lindner et al., 2008).
The asymmetric aging among the poles and sequestration of protein aggregates within one
of the daughters is thought to be beneficial for the population. The asymmetric distribution of the
older pole and damaged components generates two separate growth rates for each daughter cell,
instead of similar longer generation times than the original maternal cell, with those growth rate
differences becoming larger with each generation. The new-pole cells, cleansed of the older
maternal components and the majority of the protein aggregates, grow at a faster rate and propagate
a population forward (Table 3.1). If cell division separated the maternal components

39
symmetrically, then the overall population growth rate would be lowered (Rang et al., 2011). The
older-pole lineage serves as a reservoir for the damage within the population. We hypothesize that
the damage within the older pole cells may additionally serve as a source of mutations within an
aging population. The protein aggregates and older maternal components could potentially
contribute to a mechanism for the creation of new genetic variants within the older pole daughters
without the risk of excess DNA damage within the overall population. This is due to the
observation that the segregation of damaged protein to old poles generates redox active
compounds, such as reactive oxidative species (ROS), capable of damaging protein, lipids and
DNA, causing mutations disproportionately within the older pole daughters as a form of bet
hedging. In this model, if bacteria symmetrically distributed their aging peptidoglycan, membrane,
and other cellular damage fewer mutations would occur in the older pole population, but more
mutations would be generated overall in all cells, potentially leading to a general reduction in
fitness and cell death. Asymmetric segregation generates a more fit population capable of coping
with stress, along with an increased growth rate (Chao, 2010; Chao et al., 2016; Proenca et al.,
2019). This review focuses on how the unavoidable aging of one pole during each generation
results in asymmetric damage segregation during cell division between the daughter cells could be
an evolutionarily advantageous strategy. Population level advantages include both reducing the
overall growth rate and possibly the controlled generation of mutations within the population
allowing for bet hedging when encountered with various stresses.

40
3.3 E. COLI ASSEMBLY OF DAUGHTER PEPTIDOGLYCAN AND MEMBRANE AT
NEW POLES CREATES ASYMMETRIC AGING.
Escherichia coli reproduces through binary fission. As the maternal cell replicates the
chromosome and synthesizes other cellular machinery, the cell elongates and undergoes cell
division through cytokinesis at the midline, creating two daughter cells of equal size, each with
one copy of the chromosome. During the elongation phase, newly synthesized peptidoglycan is
inserted at the midline of the maternal cell creating the asymmetry between the previously existing,
“old,” and newly synthesized, “new,” cell poles. Peptidoglycan contains alternating glycan strands,
N-acetylglucosamine (GlcNAc) and N-acetylmuramic acid (MurNAc), which are cross-linked by
beta-1-4 linkages between the carboxyl group of D-Ala at position 4 on one glycan and the diamino
acid at position 3 on the other glycan strand (Vollmer et al., 2008; Kysela et al., 2013). The
insertion location for the newly synthesized peptidoglycan was discovered by replacing the
terminal D-Ala with D-Cys, as this doesn’t affect peptidoglycan formation, visualizing the foreign
D-Cys in the peptidoglycan through biotinylation, and immunodetection with an anti-biotin
antibody (De Pedro et al., 1997). Incubating E. coli with the label revealed mostly equal
distribution of labeled peptidoglycan throughout the cell. Once E. coli is removed from the labeled
medium and incubated in unlabeled medium, the labeled peptidoglycan is conserved at the cell
poles and the septum is fully unlabeled. After two cell doublings, cells with no labeling were
observed suggesting one pole of each daughter cell is always comprised of newly formed
peptidoglycan upon the creation of that cell. After 5 doublings, 4.5% of the cells conserved the
label at their poles, in good agreement with a model where, if the peptidoglycan was perfectly
conserved, the expected percentage of cells with labeled poles would be 6.2% after 5 doublings
(De Pedro et al., 1997).

41
The cell membrane, anchored with the peptidoglycan, undergoes a similar phenomenon
during synthesis and results in an asymmetry during cell division as well. Tracking cell divisions
after a pulse chase with a cell-membrane stain, TRSE, similar to the experiment described above,
revealed a similar asymmetric inheritance as the peptidoglycan. After 2.5 generations the old poles
showed ~65% of the remaining label, while diffusion of the signal would predict 16% of the signal
to be located at the pole (De Pedro et al., 2004). Further, strong interactions between the
peptidoglycan and outer membrane proteins result in proteins near the polar cap being nearly
immobile and, thus, similarly segregate asymmetrically. Examples of polar proteins include those
such as Braun’s Lipoprotein (Lpp) binding covalently to the outer membrane (Vollmer et al.,
2004), peptidoglycan association lipoprotein (PAL), and OmpA binding non-covalently between
the outer membrane and murein sacculus (De Pedro et al., 2004), among others preferentially
located at one or both poles (Janakiraman and Goldberg, 2004).
Despite the fact that cell division occurs symmetrically, midway across the cell length
(Gupta et al., 2014), the phenomenon of conserving peptidoglycan at the cell poles creates an
asymmetry in pole ages across every daughter cell, a further asymmetry between the two daughter
cells from the same mother, and thus is the basis for the term asymmetric aging as used throughout
this review.

42

Figure 3.1. Asymmetric aging of daughter cells through E. coli division. As E. coli cells divide
one pole is always 0 generations in age upon creation of that cell while the other is inherited from
the maternal cell. This creates a variety of pole ages within the same generation, but half of the
daughters are always new-pole cells.

43
3.4 INDIVIDUAL POLE AGE RELATES TO INDIVIDUAL FITNESS.
The asymmetric aging observed across daughter cells in E. coli coincides with
physiological consequences (Table 3.1). The new-pole cells are slightly larger than older-pole cells
due to having a faster growth rate when compared to the old-pole-receiving daughters (Stewart et
al., 2005). Automated fluorescence microscopy tracking of E. coli cells growing on an agarose
pad, allowing for the ability to track old-pole vs new-pole lineages, revealed old-pole lineage cells
had a growth rate ~2.2% slower than new-pole cells on average. Inheriting the older maternal pole
for subsequent generations further reduces the old-pole lineages growth rate, while the new-pole
cell from the same maternal cell experiences a rejuvenated, faster growth rate. The reduced growth
rate in old-pole-receiving cells results in dividing slower compared to new-pole cells, and
consequently a decrease in biomass production. The reduced growth rate of the old-pole lineages
has been observed many times under various stresses (Rang et al., 2012; Clark et al., 2015), as
well as a wide range of nutrient sources (Lapinska et al., 2019; Proenca et al., 2019). The new-
pole lineage and old-pole lineage stabilize to two separate growth rate equilibriums (Proenca et
al., 2018; Lapinska et al., 2019; Proenca et al., 2019). Stresses, such as light excitation, which
induces ROS, decrease elongation rates with increased light exposure causing old-pole daughters
to stop dividing, with doubling times approaching infinity (Proenca et al., 2019).
Reduced elongation rates in older-pole cells correspond to an increased death probability
for these older-pole cells compared to the new-pole siblings (Stewart et al., 2005; Boehm et al.,
2016; Proenca et al., 2019; Yang et al., 2019). In Stewart et al.’s study of dividing cells, cells
observed to stop growing were more likely to have inherited the older, maternal pole in previous
generations (Stewart et al., 2005). In a study of an increased glycogen production mutant (csrA),
at pole age 3.76 there is a 50% probability the population will stop dividing, with the probability

44
increasing with pole age (Boehm et al., 2016). In the case of exposure to UV, heat, or streptomycin,
the new-pole lineage displayed lower probability of death compared to the lineages inheriting the
older, maternal pole (Proenca et al., 2019).
3.5 PROTEIN AGGREGATES PARTIALLY RESPONSIBLE FOR EFFECTS ON
FITNESS OBSERVED DURING AGING.
The age-associated phenotypes observed in the daughters with older poles can partially be
attributed to protein aggregates inherited along with the older maternal pole. Time-lapsed
microscopy studies using a YFP fusion to an inclusion binding protein (IbpA) that colocalizes with
protein aggregates showed protein aggregates first appear throughout the cell, but are rapidly re-
distributed toward the older pole within a couple cell divisions due to being relatively immobile
(Lindner et al., 2008; Proenca et al., 2019). Thus, in a population of cells, most of the protein
aggregates accumulate towards the older pole. Models predicting the location of these damaged
proteins based on nucleoid exclusion, random movement, and melting capabilities with other
misfolded proteins also confirm the bias towards the older pole (Koleva and Hellweger, 2015).
Misfolded proteins are consequently inherited with the older maternal pole in ~80% of cell division
events (Proenca et al., 2019). Furthermore, the aggregate protein levels increase with cell culture
incubation time (Proenca et al., 2019) and also with pole age (Lindner et al., 2008). Closely
tracking cell populations growing within a monolayer, through time-lapse images to determine
relative pole age and fluorescence of the protein aggregate marker, showed new-pole daughters
are lacking the fluorescent marker, indicating they are “cleansed” of the protein aggregates upon
cell division. Cell division times, expressed as growth rates, indicated a negative correlation with
fluorescence intensity for the protein aggregate levels along with pole age (Lindner et al., 2008).

45
Protein aggregates are insoluble forms of damaged proteins that clump together in the cell.
One class of damage, Advanced Glycation End products (AGEs), are formed through a
condensation reaction (Maillard reaction) occurring between reducing sugars and primary amino
groups forming aldimines/Schiff bases which then transfigure into more stable Amadori products
(Hodge, 1953). The free amino groups of proteins (primarily the N-terminus of proline, arginine,
lysine, and threonine) (reviewed in Nyström, 2005) and even DNA (Bucala et al., 1984) are then
glycated to form AGEs. This irreversible process of carbonylation/glycation increases with levels
of oxidative damage (Dalle-Donne et al., 2003), in fact levels of oxidative damage increase with
culture age (Dukan and Nyström, 1998; Dukan and Nyström, 1999; Pepper et al., 2010). Under
conditions of oxidative stress, carbonyl groups are produced when there is a direct metal catalyzed
oxidative attack on the amino acid side chains of lysine, arginine, and threonine. Carbonylation
also occurs in starving E. coli cultures due to reduced translational accuracy and increased
available substrates creating more aberrant proteins (Ballesteros et al., 2001). These carbonylated
proteins can either be in a soluble state or in an aggregated state. The increased levels of
carbonylated proteins leads to the formation of the protein aggregates (reviewed in Maisonneuve,
Fraysse, et al., 2008). Most of the carbonylated proteins in an exponentially growing E. coli culture
are in an aggregated state and do not degrade over time, further increasing in stationary phase
(Maisonneuve, Fraysse, et al., 2008). Cellular defenses to prevent protein aggregation include
chaperone systems such as DnaJ/K (Fredriksson and Nyström, 2006) and ROS detoxification
systems such as sodA and hpx (Maisonneuve, Ezraty, et al., 2008), where overexpression reduces
the pool of aberrant proteins and prevents protein aggregation. Mutants of these systems show
increased carbonylated protein levels with increasing culture age (Maisonneuve, Ezraty, et al.,
2008). The growth rate of dnaK mutants within a microfluidics device increased, along with the

46
ratio between the old and new-pole daughters (control – new: 21.79 min vs old: 23.23 min; dnaK
mutant - new: 28.06 min vs old: 30.20 min) (Proenca et al., 2018). Population survival studies,
modeled from the microfluidics growth rate data, showed increased mortality in the older-pole
inheriting lineages (Proenca et al., 2018; Proenca et al., 2019). The negative correlation between
growth rate and protein aggregate levels provide evidence for their role in bacterial aging, but the
inheritance of the protein aggregate is not the only determining factor for the reduction in growth
rate of the older-pole cells. Aging effects are still observed without inheritance of the protein
aggregate in older-pole cells and the aging effects are not as severe when new-pole daughters
inherit the protein aggregate (Lindner et al., 2008; Koleva and Hellweger, 2015). The inheritance
of a protein aggregate by a new-pole daughter does come with a price, however, as those cells do
not experience the expected rejuvenated growth rates.
Potentially, one of the reasons for the continued aging effect in older-pole cells is due to
differences between gene expression levels across the older and younger pole in individual cells.
Using GFP as a proxy of expressed proteins, the level of expression differed across the cell length.
Near the newly synthesized pole expression of the GFP was higher than near the maternally
inherited old pole (Shi et al., 2020). The asymmetry in gene expression likewise follows the
asymmetry in pole age after cell division occurs. New-pole daughters have 8% higher GFP
fluorescence levels than older-pole receiving daughter cells, on average. After the next round of
cell division, the discrepancy becomes even larger between the new-pole and old-pole receiving
daughters when from an older maternal cell with a 12% difference in GFP fluorescence, compared
to a 4% difference between new-pole and old-pole receiving daughters when from a new-pole
mother (Table 3.1). The expression ratios between new-pole and old-pole daughters correlates with

47
growth rate ratios, with larger discrepancies observed when the daughters are from a mother with
an older pole.
3.6 E. COLI POPULATIONS BENEFIT FROM THE ASYMMETRIC AGING
PHENOMENON THROUGH ENHANCED LONG-TERM SURVIVAL.
The asymmetric segregation of irreparably damaged protein, as well as the peptidoglycan
itself, may permit the survival of the new-pole lineage at the expense of the “older” cell. Modeling
growth rates based on the asymmetry levels in damage segregation between daughter cells revealed
higher predicted evolutionary fitness with asymmetric damage segregation by generating variation
in the growth rates (Rang et al., 2012). This asymmetry lowers the overall mean population
doubling time by increasing the variation in the population (Vedel et al., 2016). The two daughter
cell lineages segregate into two growth rate equilibrium attractor states, new-pole daughters are
cleansed of the protein aggregates and experience a faster growth rate, while older daughters are
burdened by the damaged proteins and nucleic acid along with older cellular material experience
a slower growth rate (Rang et al., 2011; Proenca et al., 2018). If the segregation of damage were
symmetrical, the entire population would experience a slower, average growth rate between the
two lineages, continually getting slower with each division (Rang et al., 2011). The difference
between the two lineages growth rates becomes more pronounced in the presence of environmental
stress (Vedel et al., 2016). The increased growth rate experienced by the new-pole daughters
propagate a population forward with greater success than if the damage is partitioned
symmetrically (Chao, 2010; Chao et al., 2016).
It is possible the accumulated damage at the older poles may provide an additional benefit
to the population outside of serving as the “dumpster” for the population by providing beneficial
mutations. Non-enzymatic glycation reactions can occur between sugars and amino groups of

48
DNA (Bucala et al., 1984), in addition to the direct effect of ROS on DNA; thus, it is possible that
protein aggregates may interact directly with the chromosome; protein aggregates and DNA have
both been observed at the cell pole (Rokney et al., 2009). Glycation within aging cultures of E.
coli increases and correlates with mutation frequency (Pepper et al., 2010; Kram and Finkel, 2014).
It is possible that the glycation of DNA increases damage of bases in DNA and thus new mutations
upon cell division. The new-pole daughter, spared of the protein aggregate damage, experiences
rejuvenated growth, but inherits the genetic background of the damaged mother, this would include
beneficial mutations, as well as those that are deleterious. A study examining protein aggregate-
bearing cells and protein aggregate-free cells revealed that protein aggregate-bearing cells had
higher survival rates and resume growth more quickly after experiencing various stresses (Govers
et al., 2018). This has been referred to as a form of epigenetic memory for bacteria (Mortier et al.,
2019). The older-pole cells can both sequester cellular damage and also generate genetic variation,
but both roles need further investigation to understand bacterial senescence.

49

Figure 3.2. Our model for asymmetric segregation of damage leading to a more fit population
through introduction of genetic mutations. Created with Biorender.com, adapted from “Binary
Fission (Layout)”, 2022.

50
The generation of genetic variation is a critical topic to explore within populations of
microbes. E. coli cultures initially introduced into a rich medium, such as LB, possess the ability
to survive within this environment for many years without any addition of nutrients (Finkel, 2006).
E. coli when cultured in this manner experiences five phases: lag phase, exponential or logarithmic
phase, stationary phase, death phase, and long-term stationary phase (Finkel, 2006). During long-
term stationary phase survival, the bacterial population is not dormant, but evidence shows
continual turnover of novel genotypes. Despite the apparent stability in population density, there
are subpopulations increasing in frequency within the population, while other subpopulations
decrease or go extinct. Genotyping of clones from aging populations throughout time confirms
that genotypes are replaced constantly revealing very few, if any, of the parental genotype present
as early as after ten days of incubation (Ratib et al., 2021). During the course of aging, due to this
replacement of genotypes, the population becomes more fit where 10-day populations outcompete
1-day populations, 20-day populations outcompete 10-day populations, and 30-day populations
outcompete 20-day population when in co-culture – termed the Growth Advantage in Stationary
Phase phenotype (GASP) (Finkel and Kolter, 1999; Finkel, 2006). Typically, mutations arise in a
population through errors by DNA polymerases during replication; however, the cell turnover is
slower than during exponential phase, so the number of mutations is greater than expected during
long-term stationary phase. One model to explain this is that asymmetric aging plays a role in
generating the mutations necessary within a population for long term survival. The protein
aggregates within the older cells damage DNA at a higher frequency generating additional
mutations. As populations age, the damage increases, thus generating more mutations. While
genetic damage is initially more present within older daughter cells, as this cell undergoes cell
division, it gives rise to a rejuvenated young, new-pole daughter. This new-pole daughter receives

51
the mutation from the damaged maternal cell but is rid of the cellular damage that results from
aging, so it experiences a faster growth rate, expressing any mutational benefit, and passing those
benefits to further generations (Figure 3.2). This model may account for the mutations observed
within aging populations and how long-term populations are dynamic despite cell division
occurring less frequently. This model may also result in less mutational damage throughout the
entirety of the population if damage were symmetrically segregated. The asymmetry of damage
segregation, and thus mutation generation, creates a balance allowing the population to explore
additional genetic strategies without the risk of reducing the overall fitness or even killing off the
entire population. Thus, the asymmetric segregation of damage to the old pole-receiving daughters
may provide further positive evolutionary benefits, far more significant than the sequestration of
damage; however, the direct effects on mutation rates needs to be further studied.

52
Trait New-Pole Cells Old-Pole Cells Reference
Cell Size

Larger Smaller Stewart et al., 2005
Growth Rate

Faster Slower (2.2%) Stewart et al., 2005
Doubling Times

Faster (21.79 min)
Faster
Slower (23.23 min)
Slower
Proenca et al., 2018
Chapter 4 – Fig. 4.5
First appearance of
Protein Aggregates
within a cell

30% at new-pole 31% at old-pole Lindner et al., 2008
Protein Aggregate
Location after Two
Divisions

<10% at new-pole >60% around old-
pole
Lindner et al., 2008
Damage Inherited

~37% ~63% Proenca et al., 2019
Inherited Protein
Aggregate

19.6% of division
events
80.4% of division
events
Proenca et al., 2019
Glucose
Accumulation

Faster Slower Lapinska et al., 2019
Expression Levels

Higher (8%) Lower Shi et al., 2020
Expression Ratios
between new:old
inherited pole within
the cell

Smaller Differences
(5.2% higher for new
pole)
Larger Differences
(9.1% higher in new
pole)
Shi et al., 2020
Expression Ratios
between new and old
daughters
(second generation)
Smaller Differences
(2% higher in new-
pole daughters)
Larger Differences
(5.7% higher in new-
pole daughters)
Shi et al., 2020
Lag Times Shorter Longer Chapter 4 – Fig. 4.5
Minimum Inhibitory
Concentrations to
Stressors
Lower Higher Chapter 4 – Fig. 4.7
Competitive Fitness Less Competitive More Competitive Chapter 4 – Fig. 4.6

Table 3.1 Culminated list of fitness implication between new and old-pole cell lineages.

53
3.7 CONCLUSION
Despite the negative connotation with aging in humans and other eukaryotic cells, we have
highlighted the potential benefits bacterial populations experience while aging. While the
accumulation of damage occurs within the older pole-receiving E. coli cells which slows down the
growth rate, doubling time, and offspring biomass production (Table 3.1); we have described how
this could result in beneficial attributes to the progeny and thus the overall population. The oldest
cells within a population, too ridden with damage, may stop dividing; but prior to this point of no
return the damage within the cell may occasionally generate beneficial mutations as described
above. The beneficial mutations are distributed to the new-pole daughter which can fully express
the benefits with the rejuvenated growth rate. The introduction of mutations potentially through
accumulated damage in older-pole containing cells allows the population to utilize “bet hedging”
as a strategy (Grimbergen et al., 2015) for various environmental stresses a population may
encounter within their natural environment. The overall doubling time of the population is lowered
through this asymmetric aging between the daughter cells allowing a population to propagate
forward at a faster rate than if the damage were segregated symmetrically, more quickly passing
on the beneficial mutations. The asymmetric aging phenomenon may be critical to how bacterial
populations, especially E. coli, survive into long-term stationary phase and increase in fitness over
time. If all cells segregated their damage symmetrically, the entire population could crash due to
consistent accumulation of damage in all cells reaching the point of senescence faster. The
asymmetric aging allows half the population at any time to start as new, but with the “genetic
memory” of the older generations.
Studying the asymmetric aging phenomenon has its challenges due to the unidentified, if
any, regulatory factors involved in damage segregation. It is unknown if individual E. coli cells

54
have any control over the degree of asymmetry for segregating maternal damage. Thus far the
damage accumulation at the pole and the asymmetric segregation of damage is believed to be a
relatively passive system since the protein aggregates are relatively immobile within the cell. It is
possible that regulatory elements are involved in controlling the level of asymmetric segregation
of the damaged maternal components, but none have been discovered thus far. The only
characterized regulation is the insertion of peptidoglycan along the midline creating the divide
between the poles that continues into further generations, but this cannot be altered because the
cells will not divide properly. One approach to study the protein aggregates’ role in generating
mutations and modulating fitness would be to increase the stress on the population to heighten
damage accumulation within the cells and compare to an ideal environment lacking additional
stressors. Although too much additional stress may promote senescence of individual cells at a
faster rate and not allow for expression of the benefits.
Our approach to study the asymmetric aging phenomenon in bacteria is to utilize Percoll
gradient separation techniques to divide the population into subpopulations based on relative
age/damage levels (Figure 3.3) (Makinoshima et al., 2002; Maisonneuve, Ezraty, et al., 2008;
Pandey et al., 2013). Induced inclusion body formation in E. coli increases density within Percoll
gradients (Pandey et al., 2013). Increased glycogen levels are another component which increases
buoyant density. Glycogen increases in cultures throughout stationary phase and as cultures
transition from exponential phase to stationary phase cultures become denser within Percoll
gradients (Makinoshima et al., 2003). The sub-fractionation of populations within a Percoll
gradients also show some correlation with cell phase state. Less dense subpopulations appear to
be “exponential phase-like cells” while denser subpopulations are more “stationary phase-like”
when examining various subpopulations’ intracellular protein concentration profiles using phase

55
indicators such as Dps, RpoS, IHF, Hfq, GroE, H-NS, RpoD, and OxyR protein levels
(Makinoshima et al., 2003). The denser fractions show higher levels of Dps, RpoS, IHF, Hfq, and
GroE, which are strong indicators of stationary phase. This separation system could be useful for
studying the older vs younger cell populations to further the understanding of bacterial aging and
the asymmetry in the division of maternal components between daughter cells. However, even
with this system the new-pole cells found within the gradient originated from older-pole cells in
the population. Aging is a fundamental process to eukaryotic life and bacteria was formerly
thought to escape this phenomenon, but further research is necessary to examine how aging occurs
and how it benefits bacterial populations for long-term population fitness.

56

Figure 3.3. Schematic of Percoll gradient centrifugation of cell cultures to respective damage
levels using buoyant density. Left: Created with Biorender.com. Right: a 4-day E. coli culture
separated using Percoll gradient centrifugation with visual bands at different buoyant density
levels compared to density marker beads (purple: 1.06 g/ml, red: 1.09 g/ml) within a Percoll
gradient.

57
Chapter 4: Escherichia coli Subpopulations Exhibit Differential Fitness Over Time When
Separated by Buoyant Density

4.1 ABSTRACT
During Escherichia coli cell division, the segregation of maternal damage is asymmetrically
divided between the two daughter cells. As cells replicate, new peptidoglycan and cell membrane
is inserted along the midline of the mother cell producing newly synthesized material around one
pole of each daughter cell. The other pole is inherited from the maternal cell intact, with one
daughter inheriting the most recently synthesized pole of the maternal cell, termed the “new-pole”
daughter cell, while the other daughter cell inherits the maternal pole that has passed through the
generations. This creates differential pole ages within individual cells and between the daughter
cells. With every cell division, one new-pole cell is produced, meaning that every generation
produces half new-pole daughter cells. The other half of the cells disproportionately inherit
maternal damage, including accumulated protein aggregates, each generation. The new-pole cells
thus experience a rejuvenated growth rate compared to the original maternal cell and the other
daughter. Using Percoll gradient centrifugation, we separate cells in an aging population by their
buoyant density, which corresponds to the relative damage levels accumulated and examine
potential fitness implications for those aging individual cells, as well as the overall population
based on the asymmetric aging phenomenon. We explore how damage levels may correlate to
survival in the overall population using monoculture survival assays, competitive assays between
differentially aged/damaged cells, as well as survival to various environmental stressors. We show
that subpopulations separated based on damage levels through buoyant density exhibit differential
fitness. Subpopulations with more accumulated damage eventually die out within aging

58
subpopulations, but first provide benefits, including novel mutant alleles, that pass on to new-pole
cells prior to death, allowing for an overall increase in population fitness.

59
4.2 INTRODUCTION
Escherichia coli populations possess the ability to survive batch culture conditions for long
periods of time after initial inoculation without providing any additional nutrients (Finkel and
Kolter, 1999; Finkel, 2006; Ratib et al., 2021). After inoculation, E. coli K-12 populations
experience five distinct phases: lag phase, exponential/log phase, stationary phase, death phase,
and long-term stationary phase (LTSP) (Finkel, 2006). The initial lag phase is a relatively short
period of time where the population is acclimating to their new environment and retooling cellular
metabolism; population numbers remain constant with no increase in viable cell counts. After the
acclimation period, the population counts increase throughout log phase in an exponential fashion
as cells undergo rapid division. Exponential growth continues until the population reaches a peak
population density and the total population counts remain at that peak of cell counts as the culture
enters stationary phase. During stationary phase, population growth ceases and population density
remains constant for one or two days, depending on strain and culture conditions. After stationary
phase, approximately 99% of the cells lose viability as cells transition into death phase. Despite
the extreme loss of viable cells during death phase, a small fraction of the population can survive
into LTSP. Populations of E. coli maintain roughly stable population densities during LTSP, and
this phase most closely models naturally occurring populations as they endure nutrient-limited
environments for long periods of time.
During LTSP, the population is highly dynamic with a constant turnover of different
genotypes despite the observance of relatively stable viable cell counts (Finkel, 2006; Ratib et al.,
2021). The selection of beneficial genotypes within the population as E. coli progress through
LTSP results in an overall increase in competitive fitness to LTSP conditions over time. The
increase in competitive fitness of LTSP populations is observed through co-culturing of a

60
population aged through LTSP with the initial parental population at a 1:1000 ratio (vol:vol). In
this competition assay, cells from 10-day populations outcompete parental populations, cells from
20-day populations outcompete 10-day populations, and cells from 30-day populations
outcompete 20-day populations (Finkel and Kolter, 1999). The increase of competitive fitness as
populations are incubated further into LTSP is referred to as the Growth Advantage in Stationary
Phase (GASP) phenotype.
When initially studied, the aged populations’ fitness advantage within the competitive
environment was attributed to beneficial mutations for scavenging amino acids (Zambrano et al.,
1993; Zinser and Kolter, 1999; Finkel, 2006). However, subsequent experiments where many
individual clones from a 10-day LTSP population have been sequenced revealed that virtually
every clone possessed a mutation compared to the parental genotype – with a number of different
genotypes present (Ratib et al., 2021); not all mutations can be associated with amino acid
scavenging. The amount of genetic diversity observed on day 10, including identifying genotypes
with 4 or more mutations suggests diversity is present even earlier within monoculture growth.
These GASP populations may initially be present as early as day 1, once a culture arrives in
stationary phase, prior to death phase, albeit at low frequency. The genetic diversity present early
on in these cultures could play a role in the survival during LTSP by providing diverse strategies
for natural selection to act upon.
Despite the overall aging of the population being seemingly beneficial with the
manifestation of the GASP phenotype, individual E. coli cells do experience senescence. The
senescence of individual cells or subpopulations of cells may contribute to some of the cycling
observed among different genotypes present within the long-term culture, in part due to the
differences in the relative ages of the two existing poles of separating daughter cells. When an E.

61
coli cell undergoes cell division, it elongates from the mid-cell plane, producing new
peptidoglycan at the septum, while the existing peptidoglycan remains in the distal poles of the
cell (De Pedro et al., 1997). Since the septum is entirely composed of newly synthesized
peptidoglycan, each daughter cell has one new pole (zero generations in age) and the other with
the inherited, maternal peptidoglycan. Thus, with each generation, half of the daughter cells
produced have the youngest possible poles (0 and 1 generation in age) inheriting the maternal pole
synthesized just one generation prior to the creation of that maternal cell. The other half of the
population possesses poles of various ages depending upon the number of previous divisions
experienced along the maternal lineage (Stewart et al., 2005). Thus, within the overall population
a variety of pole ages exist, with ~50% of the cells 0-1, ~25% 0-2, ~12% 0-3, etc.; reflecting a
form of aging in these bacteria. This asymmetry during division is observed among other cellular
components near the cell poles, such as outer membrane proteins and accumulations of damage in
the form of inclusion bodies (Janakiraman and Goldberg, 2004; De Pedro et al., 2004; Nyström,
2007). The amount of aggregated proteins accumulated within a cell correlates with pole age with
new pole cells devoid of their maternal protein aggregate levels (Lindner et al., 2008). This
asymmetry in the division of damaged maternal components results in negative fitness
implications for the older-maternal pole-receiving cell compared to the new-pole cell produced
every generation. The older-pole-inheriting cells shows a slower growth rate, on average,
compared to their new-pole siblings which experience a similar or faster growth rate compared to
the maternal cell. The decrease in growth rate of the older-pole cells and longer time to cell division
results in lower offspring biomass in relation to their siblings (Stewart et al., 2005). The inheritance
of the maternal protein aggregates and other forms of damage within the older-pole daughters can
partially explain the slower growth rate, but this is not the only factor decreasing the generation

62
time within older-pole cells (Lindner et al., 2008). Older-pole cells receiving protein aggregates
show a larger decrease compared to new-pole cells that may receive the maternal protein aggregate
(Koleva and Hellweger, 2015). Despite the senescence observed by older-pole receiving cells, the
asymmetric segregation of the older-pole and damaged maternal components is thought to be an
evolutionarily advantageous strategy for the population by increasing variation in growth rates and
thus offspring performance discovered through tuning the level of asymmetry in damage and
subsequent growth rates of daughter cells within population survival models (Ackermann et al.,
2007; Chao, 2010; Chao et al., 2016).
The ability for E. coli to survive into LTSP and express the GASP phenotype may be
related to this asymmetric aging phenomenon, but this correlation has not been examined in an
experimental setting. Perhaps asymmetric aging is beneficial to sustain long-term populations
because older-pole cells generate more genetic diversity in a population, while the younger-pole
cells manifest these novel beneficial alleles more quickly. The accumulation of damage in the
older-pole cells could damage the chromosome through glycation of the free amino group of DNA
bases to form advanced glycation end products (AGEs) or by oxidative damage converting guanine
to 8-oxo-G causing replication errors during division inducing mutations. Cells with older poles
could potentially be reservoirs for damage, allowing the younger-pole cells to have maximum
reproductive possibilities (Stephens, 2005). If the rejuvenated, new-pole cells inherit beneficial
mutations, natural selection selects for those genotypes, and the younger-pole cells have increased
offspring production compared to the older-pole cells with the same beneficial mutations. Thus,
beneficial genotypes then increase in frequency at a faster rate than if damage was segregated
symmetrically with less variation in growth rates (Chao, 2010; Chao et al., 2016). The beneficial
mutations could first arise in older-pole cells from the accumulated damage, induce mutations,

63
then the new-pole cell generated from that cell’s division is able to fully manifest the benefits from
that mutation and generate the GASP populations observed in LTSP. Asymmetric aging could
provide a model for how bacterial populations are able to generate versatile populations that can
ultimately survive harsh conditions.

64
4.3 RESULTS
E. coli viable cell distribution in Percoll gradients changes throughout long-term incubation.
To address the role asymmetric aging plays in LTSP survival and the manifestation of the
GASP phenotype, we separated E. coli populations by relative damage levels using Percoll density
gradient centrifugation and examined subpopulation fitness in LTSP and GASP assays based on
their inferred relative damage levels. Percoll gradient centrifugation has previously shown that E.
coli with inclusion bodies and a greater accumulation of damaged proteins are located further along
the gradient within zones of higher density (Maisonneuve et al., 2008; Pandey et al., 2013). As the
density in the gradient increases, the cells are thought to be older and more damaged compared to
less dense zones higher up in the gradient.
Differentially aged E. coli cultures show different buoyant density patterns within the
Percoll gradient. After incubating E. coli for 1, 4, 7, or 10 days and performing Percoll gradient
centrifugation, there is a clear shift in the visual band into higher up in the gradient, towards the
lower density positions, within the Percoll gradient as cultures age. Cultures that have entered
stationary phase after 1 day of incubation show a distinctive band at ~1.09 g/ml compared with the
known density marker beads (Figure 4.1 – red). After 4 days of incubation the visual band splits
into two separate bands, at both the stationary phase banding location and a novel, less-dense band
closer to the ~1.06 g/ml marker bead location (Figure 4.1 – purple). Further aging the culture for
7 and then 10 days, into long-term stationary phase, showed a shift from the denser stationary
phase band to the lower density banding position first observed at Day 4. Despite the changes in
banding patterns observed, it must be noted that the observable biomass does not represent the
total cell population. In fact, cells are distributed throughout the entire Percoll gradient, not simply
where the visual band appears. However, the position of the band typically corresponds to where

65
the bulk of the population resides. Cell counts determined for 0.5 mL fractions removed starting
from the top of the gradient revealed viable cells distributed throughout the entirety of the gradient
with clear patterns observed for each culture age (Figure 4.2).

Figure 4.1. Visual banding patterns for 1, 4, 7, and 10-Day Percoll gradients. Density marker
beads and E. coli populations aged for 1, 4, 7, and 10 days after centrifugation within Percoll
gradients revealed a shift in visual band position from ~1.09 g/mL (tube 1 – red density marker
bead and tube 2 visual band of cells) to ~1.06 g/mL (tube 1 – purple density marker bead and tube
3-5 visual band of cells).

66
Gradients from differentially aged E. coli populations at numerous times revealed
consistent viable cell count distribution patterns within the gradients for each incubation age
(Figure 4.2). In stationary phase, Day 1, E. coli cultures, the bulk of the viable cells are observed
near the visual band peaking at Fraction 9 with 4.6 x 10
8
CFU/mL ± 7.1 x 10
7
standard error. The
viable cell counts in fractions prior to Fraction 9 steadily increase and viable cell counts in lower
density fractions steadily decrease. The shape of the distribution curve is consistent between
replicates, which each show a steady increase in viable cell counts until a peak around Fractions
9-11 (Figure 4.2). At the end of death phase, after 4 days of incubation, the distribution pattern for
viable cell counts throughout the gradient flattens out, reflective of the two visual bands observed
within the 4-day gradient. Instead of a continual increase in viable cell counts throughout the
gradient until Fraction 9, instead we observe consistent levels of viable cell counts from Fractions
6 to11. The lower density Fractions 4-6 of the 4-day gradients show similar cell numbers to the 1-
day gradients, but after Fraction 7 level out instead of continuing to increase. Cell counts decrease
throughout the lower density fractions as observed for 1-day cultures. Once in long-term stationary
phase, both the 7- and 10-day culture gradient profiles are similar. The viable cell numbers found
in each fraction similarly increase up to Fraction 7/8, but instead of leveling off as observed for
the 4-day culture gradients, the cell counts continue to decrease after the peak fraction counts. This
is consistent with observing the visual band at a lower density in the gradient. Despite further aging
from 7 to 10 days, the distribution within the gradient does not change. While there is variation in
individual distribution patterns from each gradient, as depicted in Figure 4.2B-D, the overall
pattern remains consistent.

67

68
Figure 4.2. Percoll gradient profiles of viable cell counts for aged E. coli populations. (A)
Average viable cell counts per mL (with standard error) for fractionated subpopulations from
Fractions 4 (lower density) to Fraction 16 (higher density) for each overall culture age (1 day –
black, 4 day – pink, 7 day – green, and 10 day – blue). (B-E) Viable cell counts for each individual
gradient of that culture age included within (A). (n = 12, 7, 9, 8 for 1-, 4-, 7-, and 10-day culture
gradients respectively).

69
Due to the progressively lower cell counts within aging E. coli cultures, I examined the
same data not simply looking at total numbers of cells, but with cell counts in each fraction
represented as a percentage of the total cells recovered to better understand the distribution of
viable cells throughout the gradient (Figure 4.3). With the viable cell count data represented as a
percentage of the total, a shift in the fraction possessing the largest total percentage number of
living cells within each population is clearly observed. The bulk of the 1-day population resides at
Fraction 9 with ~24% of the population at that density within the Percoll gradient. The percentage
of the viable cells within the gradient for the other fractions ramp up towards Fraction 9 and ramp
back down after. Fraction 7 possesses ~5%, Fraction 8 ~10%, then Fraction 10 ~18% and Fraction
11 ~14%. About 70% of the population resides at the buoyant densities within the middle of the
gradient, while less than 3% exist on the extreme lower density Fractions 1-6 and less than 25%
exist on the higher density Fractions 12-20 of the gradient. Gradients from 4-day cultures show a
much wider range of fractions/densities where viable cells reside. This is consistent with the two
bands visually observed in 4-day cultures along with the leveling out in CFU/mL across 4-day
gradients. The majority of the viable cells are found within Fractions 8-11 accounting for ~60% of
the total cell count (15.67%, 16.28%, 14.42%, 13.27% respectively). The peak viable cell counts
stretched across many more fractions for the 4-day gradients. The 7-day and 10-day gradients
percentage yields are similarly distributed as observed on Day 1, but shifted towards lesser density
fractions. For example, 7-day gradients peak at Fraction 7 with ~24% of the viable cell population
housed within this portion of the gradient, while the peak population in 10-day gradients fall
primarily within Fraction 8 with ~24% of the population residing there. The percentage of viable
cells in the 7-day gradients begin decreasing after Fraction 7 with Fraction 8 having 16.83%,
Fraction 9 having 12.34%, Fraction 10 5.90% and the percentage of viable cells becoming

70
undetectable past Fraction 11. 10-day gradients percentage of viable cells shows a similar decrease
past Fraction 8 with Fraction 9-11 having 17.64%, 10.20%, and 6.83% respectively. Past Fraction
12 the number of viable cells within those relative densities becomes very small. While each
individual gradient shown in Figure 4.3B-E has its own signature of viable cell distribution, the
bulk of the cell population is within range of the average peak CFU fractions. and the shift between
the bulk of the cell population residing at higher to lower density positions within Percoll gradients
is still maintained. 1-Day populations show a range from Fraction 9-11 having the peak
recoverable CFU within that portion of the gradient. While 7- and 10-day gradients have a range
of Fraction 5-9 and Fraction 7-9 respectively possessing the greatest CFU/mL within those
gradient fractions.

71

72
Figure 4.3. Percoll gradient profiles for the percentage of cells recovered in each fraction for
aged E. coli cultures. (A) Average percentage of recovered viable cell counts per mL (with
standard error) for fractionated subpopulations from Fractions 4 (lower density) to Fraction 16
(higher density) for each overall culture age (1 day – black, 4 day – pink, 7 day – green, and 10
day – blue). (B-E) Percentage of viable cell counts recovered for each individual gradient of that
culture age included within (A). (n = 12, 7, 9, 8 for 1-, 4-, 7-, and 10-day culture gradients
respectively).

73
Subpopulations can survive into Long Term Stationary Phase. Samples of fractionated
subpopulations from multiple positions along the density gradient were examined for long-term
survival in rich medium. Typically, E. coli populations can survive for months in rich medium
without any addition of nutrients, with a relatively stable population density (Finkel, 2006; Ratib
et al., 2021). Cells sampled from Fractions 4, 8, 12, and 16 from 1-, 4-, 7-, and 10-day culture
gradients were inoculated into fresh LB cultures and viable cell counts were determined daily for
ten days (Figure 4.4, Supplemental Figure. 4.1). Cells from all fractions tested survived into long-
term stationary phase with final yields of 10
7
-10
8
CFU/mL at the end of the ten-day period.
However, variations in the growth and survival patterns between the fractions and between the
different days were observed. After initial inoculation all fractions and total populations reached
~4x10
9
CFU/mL at stationary phase after overnight growth. Then all fractions experience death
phase from day 1 to day 2 dropping to ~3x10
8
CFU/mL cell counts. The viable cell counts remain
consistent between the fractions at 10
8
CFU/mL until day 5 where some differences start to be
observed between the different days and fractions.
For Day 1 fractions, all experience a dip in viable cell counts either at day 7 for Fraction
12, 16, and the total culture prior to centrifugation, or at day 8 for Fractions 4 and 8. For the most
part all Day 1 fractions show little variation between their respective long-term survival curves.
Day 4 fractions show a bifurcation between the lower density and higher density fraction
phenotypes. Fraction 4, 8, and the total 4-day population experience a dip at day 8 similarly to the
Day 1 lower density fractions, but the dip shifts up to day 6 for 4 Day Fractions 12 and 16.

74
Figure 4.4. Long-term population dynamics of subpopulations isolated from Percoll
gradients. Average viable cell counts (with 95% CI) for monocultures aerated in test tubes at 37
0
C
for a ten-day period of aged E. coli total populations (grey) and Fraction 4 (red), 8 (green), 12
(blue), and 16 (purple) subpopulations from previously aged cultures within Percoll gradients (A:
1-day gradient, B: 4-day gradient, C: 7-day gradient, and D: 10-day gradient) demonstrated the
ability for all subpopulations to survive long-term culturing from all initial culture ages (n = 3).

75
The long-term stationary phase phenotypes of Day 7 fractions begin merging showing
more similarities than the Day 4 gradient subpopulations. All fractions along with the total
population dip into 10
7
CFU/mL viable cell counts on day 6, similar to the Day 4 higher density
Fractions 12 and 16. After day 7 the fraction phenotypes show more variation with their paths
leading to the end of the survival curve at day 10 with each fraction subpopulation showing their
own phenotypes. 10 Day fractions start to reconverge with less variations observed throughout the
10-day survival curve showing a more gradual decrease towards day 10 cell yields. Overall, all
fractions survive into long-term stationary phase with some variations that generally track with
their density levels with Fraction 4 and 8 showing more similarities with each other and Fractions
12 and 16 being more alike. Despite the higher density fractions, Fraction 12 and Fraction 16,
having less concentration of cells as cultures age, cells from the higher density fractions survive
into long-term stationary phase.
While all the subpopulation fractions examined achieved similar cell numbers at stationary
phase on day 1, we wanted to examine whether fractions showed differences in the length of lag
phase and the doubling times (Figure 4.5, Supplemental Figure 4.2). Inoculating pairs of fractions
into fresh medium and taking time points every 20-30 minutes for 5.5 hours, at 12 hours, and 24
hours allowed us to estimate relative lag times and log phase generation times. 1- and 7-day
cultures were studied because of the clear shift observed in the Percoll gradient characteristics
from 1 to 7 days. Again, in this experiment all subpopulations and the total population,
representative of the population prior to Percoll gradient centrifugation, achieve 10
9
CFU/mL by
12 hours (Figure 4.5 – A and D: 720 min). Each of the cultures inoculated from different
subpopulations take a slightly different path to arrive at their stationary phase densities, with
variations in lag phase times and doubling times during exponential phase. The time of entry into

76
log phase is, on average, ~187.5 min for cultures inoculated from Day 1 subpopulations, but
slightly shorter at ~150.6 min with 7-day subpopulations as inocula (p-value = 0.0015). For the 1-
day subpopulations, the cultures started from Fractions 2 and 3 samples started doubling in 180
minutes. Cultures inoculated from higher density fractions, Fractions 10 and above, took longer to
exit lag phase and enter logarithmic growth, with lag times increasing with fractions from higher
densities (Figure 4.5 – G). For 7-day subpopulations, the transition from lag phase into exponential
phase is more consistent with the majority of the cultures exiting lag phase by 150 minutes except
for Fraction 6+7 and one replicate from Fraction 2+3 which take longer to exit lag phase (Figure
4.5 – G).
Generation times were calculated for each culture from appropriate time points between
210-330 minutes of incubation once the cultures had entered exponential growth phase (Figure 4.5
– C, F, and H). Cultures inoculated from both 1- and 7-day subpopulations had a doubling time
ranging from 14.4-30.7 min with the average doubling time significantly different between the two
ages (19.6 and 24.3 min for Day 1 and Day 7 respectively, p-value = 0.0002). Cells from both aged
cultures showed a general increase in doubling times correlated with increasing density for the
seed population for the culture (Figure 4.5 – H). Cultures started from 1-day subpopulations
showed a larger increase in doubling times from the lesser density to higher density starting
populations. Cultures started from Fractions 2+3 showed an average doubling time of 18.7 min ±
0.22 95% CI, but cultures started from Fractions 14+15 showed an average doubling time of 24.8
min ± 0.94 95% CI. The increase in doubling time from Fractions 2+3 cultures to 14+15 is
consistent through the other fractions. Populations started from 7-day subpopulations had longer
doubling times than 1-day subpopulation cultures for all fractions. A general increase is still
observed for the cultures started from 7-day subpopulations with increasing seed population

77
density. The increase is less pronounced than what is observed for the 1-day subpopulation
cultures. Fraction 2+3 subpopulation cultures showed a doubling time of 21.1 min ± 1.5 95% CI
and the doubling time increased to 26.2 ± 1.6 95% CI. Only ~5 min increase in doubling times
were observed for 7-day subpopulation cultures while there was a difference of ~6 minutes
between Fraction 2+3 subpopulation cultures and 14+15 subpopulation cultures. Looking at the
highest and the lowest doubling times for each age group the variation is much larger for 1-day
starting subpopulations, ranging from 14.6 min for the total population to 24.75 min for Fraction
14+15 populations, than for 7-day starting subpopulations, ranging from 21.1 min for Fraction 2+3
to 26.8 min for Fraction 12+13.

78
Figure 4.5. Outgrowth of subpopulations isolated from Percoll gradients. (A-F) Average
viable cell counts (with 95% CI) of 1-day (top row) and 7-day (middle row) E. coli total cultures
(grey) and subpopulation pairs (Fraction 2+3: pink, 4+5: red, 6+7: orange, 8+9: green, 10+11: light
blue, 12+13: dark blue, and 14+15: purple) isolated from Percoll gradients monocultures aerated
in test tubes at 37
0
C for 12 hours. (G) Average timepoint (with 95% CI) total cultures or
subpopulation pairs from 1- (black) or 7-day (grey) gradients first showed doubling of viable cell
counts to signify entry into the exponential growth phase. (H) Average calculated doubling time
(with 95% CI) for each total culture or subpopulation pairs from the 1- (black) and 7-day (grey)
gradients. (n = 2).

79
Buoyant density separated subpopulations exhibit differential competitive phenotypes. Since
cultures from all subpopulations examined were able to survive into long-term stationary phase,
but differences in doubling times were observed in the outgrowth experiment, we decided to
examine the cultures competitive fitness compared to each other. We first examined
subpopulations from 10-day and 7-day gradients versus the parent population in a GASP
competition setting, but observed no differences between the relative fitness for different starting
subpopulations (Supplemental Figure 4.3). All fractions from 10-day populations exhibited a full
GASP phenotype, while all fractions from 7-day populations did not fully manifest the GASP
phenotype when co-cultured against an unaged, parental culture. To examine the competitive
phenotype of the fractions more directly, we competed a lower density population versus a higher
density population in a 1:1000 (vol:vol) starting ratio (Figure 4.6). Since the 4-day gradient is a
transitionary period between the 1-day and 10-day Percoll phenotype we chose to examine
subpopulations from 4-day populations in the competitive assay. To examine lower density
fractions fitness compared to higher density fractions, I competed cultures started from 4-day
Fraction 6 and 4-day Fraction 8 vs 4-day Fraction 12 populations. When introducing a 4-day total
population as a minority vs a 4-day total population as a majority, both total cultures conclude the
competition assay with similar population numbers within 10-fold of each other. When introducing
a population from 4-day Fraction 8 vs 4-day Fraction 12 as a majority, the competition results are
like the control competition where Fraction 8 and Fraction 12 end the 10-day competition within
10-fold of each other for viable cell counts within the culture. However, when I examined Fraction
6 as a minority population vs Fraction 12, the Fraction 6 population does not increase in cell density
and falls below the limit of detection while Fraction 12 viable cell numbers remain consistent with
what is observed when Fraction 12 is against fraction 8 and like the total population. Despite

80
Fraction 6 and 8 coming from the same gradient and original population, the subpopulations
separated through the density gradient show widely different competitive phenotypes. This
confirms the variation and diversity of phenotypes for various subpopulations within a total
initially isogenic population all aged together. When Fraction 6 and Fraction 8 are competed in a
1:1000 ratio with 4-day Fraction 10 cultures, similar phenotypes are observed. Fraction 6 is a poor
competitor vs the higher density fractions, while Fraction 8 has similar relative fitness.

Figure 4.6. Competitive phenotypes for Percoll gradient subpopulations. (A) Average viable
cell counts (with standard error, n = 5) for each population within 10-day co-culture experiments
between 4-day total populations (A, B: black), Fraction 6 (A, C: solid green) vs Fraction 12 (A, C:
dashed green) subpopulations, and Fraction 8 (A, D: solid blue) vs Fraction 12 (A, D: dashed blue)
subpopulations with individual replicates shown as well (B-D).

81
Stress assays reveal subpopulation phenotypes spread throughout the culture as cultures age.
Since fraction vs fraction competitions revealed differential fitness between subpopulations that
are close together in the gradient, next we examined the ability of cells from these fractions to
survive in the presence of exogenous stressors – saline, and the antibiotics nalidixic acid, and
streptomycin (Figure 4.7 and Supplemental Figure 4.4). Fractionated subpopulations from the 1-,
4-, 7-, and 10-day culture gradients were inoculated into LB medium supplemented along with
either saline or an antibiotic. Fractions 4-14 were examined for growth, survival, and/or death after
two days of incubation with the stressor. The minimum inhibitory concentration (MIC) is then
noted for each of the stressors examined. This assay revealed differential fitness depending on the
subpopulation examined from the same gradient, but the results change as the culture ages. For the
saline stress, 1-day Fraction 4, 5, and 6 subpopulations have a MIC of 1M while all other fraction
subpopulations have a lower MIC (900 mM for Fractions 7-13 and 750 mM for Fraction 14
subpopulations). For 4-day subpopulations the higher density Fractions 12, 13, and 14 all show
growth at the highest concentration tested, 1 M, and thus have an increased MIC. Fraction 4 and 5
subpopulations exhibited survival, but not growth, with saline concentrations 750 mM-1M.
Fractions 6 through 11 all showed lower MICs at 900 mM or below. After 7 days of incubation,
subpopulations become more similar in their MICs where all showed inhibited growth at 1 M. 10-
day subpopulations are also more consistent with each other all surviving 750 mM to 1 M saline,
but without growth observed.
For antibiotic stresses, the fraction subpopulations revealed similar patterns as with saline.
1-day gradient Fraction 5 subpopulations exhibited a higher MIC surviving 5-100 µg/mL with no
growth and death with 200 µg/mL nalidixic acid. All other fraction subpopulations died at 5
µg/mL. 4-day fraction subpopulations exhibited similar phenotypes with nalidixic acid as with

82
saline. Again, Fraction 12, 13, and 14 showed a higher MIC than other subpopulations. The same
MIC phenotype was observed in these subpopulations as for Fraction 5 from the 1-day gradient
with survival observed from 5-100 µg/mL and death at 200 µg/mL. Fraction 5 from the 4-day
gradient showed slightly lower MIC with survival observed from 5-20 µg/mL and death occurred
at 50 µg/mL nalidixic acid. Once cultures are incubated to 7 and 10 days, fewer differences are
observed in the phenotypes between the subpopulations, such as the saline MICs. For 7-day
gradient subpopulations, Fraction 10 showed the highest MIC surviving 10-100 µg/mL and dying
at 200 µg/mL while Fraction 4 showed the lowest MIC with values the same as 1- and 4-day
fraction 4 subpopulations. Fractions 6 through 9 and Fractions 11-14 all showed the same MIC
which showed survival at 5-100 µg/mL and death at 200 µg/mL. For 10-day gradients there is even
more consistency between the fraction subpopulation MICs. All fractions examined show the same
MIC phenotype as most 7-day gradient subpopulations except for Fraction 7 subpopulations which
have an MIC phenotype of death at 5 µg/mL.
Streptomycin again showed a similar pattern as the other stressors with most 1- and 4-day
fraction subpopulations having lower MIC phenotypes than 7- and 10-day gradient fractions. The
increase in MIC is again observed within the higher density Fractions 12, 13, and 14 where the
subpopulations exhibited survival with no growth at 6.25-250 µg/mL for Fractions 12 and 14 and
Fraction 13 showed growth at 6.25 µg/mL and shifted into survival without growth at 12.5-250
µg/mL. All other fraction subpopulations showed lesser MICs with growth last observed at 6.25
µg/mL, survival in 12.5-65.5 µg/mL and death at 125 µg/mL. 7- and 10-day fraction
subpopulations all show similar or higher MIC to the 4-day Fraction 12 and 14 subpopulations
except for 7-day Fraction 13, 10-day fraction 4 and 10-day Fraction 14 which all have a slightly

83
lower MIC phenotype with no growth observed even at 6.25 µg/mL but survival observed at all
concentrations examined up to 250 µg/mL.
Similar to the competition assays, subpopulations from different fractions or different ages
shows a range of phenotypes when it comes to responding to the various stressors that a population
encounters. The differential phenotypes of different subpopulations from the same initial culture
repeated within other gradients as well (Supplemental Figure 4.4). There is a general adoption of
the higher MIC concentration that begins within the higher density fractions and then spreads to
the rest of the population.

Figure 4.7. Minimum Inhibitory Concentrations for Percoll gradient subpopulations to
exogenous stressors. MIC for Fraction 4-14 subpopulations introduced to saline stress (blue),
nalidixic acid (green), and streptomycin (yellow). The darker the color the higher the MIC of those
subpopulations. Listed concentrations within parenthesis indicate survival (no growth) at those
concentrations while other listed concentrations without parenthesis indicate death. The
concentrations for each color block are indicated only once with all similar color blocks matching
those concentration values.

84
4.4 DISCUSSION
This work shows that as populations progress through different growth and survival phases,
buoyant density patterns of cells within the cultures shift with culture age. Despite the observable
bands concentrated at specific locations in the gradients, viable cells are found throughout all
gradients. As LB cultures age from 1 to 10 days into long-term stationary phase, 10 to100-fold cell
loss is observed in the overall population and this observed loss in total viable cell count is mostly
observed within the higher density fractions of the gradients. The shift in the position of observable
bands and the patterns of viable cell counts found throughout the gradients of aging cultures is not
just due to changes in initial total cell number. I examined 1-day cultures diluted 10-fold and 100-
fold to similar cell numbers as 7- and 10-day cultures and the observable band decreased in
visibility, but the location remained the same and showed a similar cell distribution to 1-day
gradients compared to the 7- or 10-day gradients (Supplemental Figure 4.5). The banding position
is related more to the aging of cells in the culture than the total numbers of cells. The number of
viable cells at higher density positions within the gradient decrease with culture age. Cells that are
becoming more damaged in the population are thought to progress further in the gradient as this
has been observed for inclusion body producing cells compared to non-inclusion body containing
cells (Pandey et al., 2013). Cells with increased buoyant density have previously been found to
have increased glycogen (Makinoshima et al., 2003) and cultures increasing in incubation age
show an increase in glycogen concentration within the cultures (Kram and Finkel, 2015). It is
feasible to posit that at a certain point the cells found at higher densities collect too much damage
and are not able to maintain viability as the cultures age. Past observations of individual cells aging
and dividing revealed that the only cells that stop dividing were those inheriting the older maternal
pole upon subsequent cell divisions (Stewart et al., 2005). Furthermore, populations with increased

85
glycogen production because of a csrA deletion show 50% probability of cell division stopping for
cells with a pole age of >3.76 and this probability increases with increasing pole age (Boehm et
al., 2016). Cells found within the higher buoyant density zones within the gradients likely show
increased levels of glycogen, have higher amounts of damage, are older in pole age and thus
possibly stop dividing and eventually rupture leaving fewer viable cells within those gradients as
the cultures age from 1 to 10 days. The viable cell counts found within the lower density fractions,
that change less as cultures age, are likely the new-pole cells produced with every division and are
constantly replaced in the population. The balance between life, aging, and death stabilizes after
7-days of culture incubation when the gradients begin to look similar as cultures are further aged.
Up to 10-day gradients are shown here, but 20- and 30-day gradients revealed similar patterns, so
further aging doesn’t alter the dynamics of the Percoll gradient (data not shown).
Despite the decrease in higher buoyant density subpopulation cell numbers as E. coli
populations enters long-term stationary phase, all subpopulations examined from each gradient
age survived into long-term stationary phase for 10 days (Figure 4.4). Even with potentially more
damaged subpopulations initiating the cultures, populations reach high cell density at stationary
phase and survive the entire assay. This could be due to an older-cell dividing and producing a
new-pole cell that isn’t plagued by the damage that would lead to a termination of cell division.
While certain cells within the gradient have likely reached the point of senescence where they are
too damaged to divide, others within the same subpopulation haven’t reached that point yet. As
soon as one new-pole cell is produced that cell generates the divisions necessary for achieving
high cell counts in stationary phase and survival into long-term stationary phase. It is interesting
that there is some inheritance of phenotypes observed, passing between the subpopulations
throughout aging. For example, populations started from 4-day Fraction 12 and Fraction 16

86
subpopulations show a dip in viable cell counts at day 6, which is not observed for the other 4-day
fraction populations. This dip in viable cell counts at day 6 is then inherited by all the populations
started from 7-day fraction subpopulations. It is possible that older-pole containing, more damaged
cells found at higher buoyant densities within the 4-day gradient divide and produce new-pole cells
with their phenotype that are spread throughout the gradient at 7 days. As populations age from 4
to 7 days the higher density fractions within the Day 4 gradients such as Fractions 12 and 16 would
divide generating new-pole daughters that would fall into lower density zones within the earlier
fractions. The phenotypes become more similar for 7- and 10-Day gradient fractions due to this.
This phenomenon of inheriting the phenotype from previous subpopulations is further
observed with the saline and antibiotic stress assay (Figure 4.7). The initial increase in minimum
inhibitory concentration is first observed within the higher density fractions from 4-day gradients.
The lower density fractions such as Fraction 5 also typically show an increased MIC from the rest
of the subpopulations. The higher density fractions increase in MIC to various stresses is caused
by damaged, older-pole cells generating a more active stress response. This stress response is
triggered due to the damage within their cells. The damage within the older cells found at the
higher densities in the gradient may have further implications on the chromosome. Increased
reactive oxygen species (ROS) and damaged aggregates could interact with the chromosome,
damaging the DNA bases, and generate mutations upon replication. The heightened stress response
and the potential mutations allowing for stress resistance could spread through the population after
division. The new-pole cells, found in the lower density fractions such as Fraction 5, then also
possess increased stress resistance. As these cells become older, they will fall within the gradient
to higher density positions and the new-pole daughters will take the place of these cells in the

87
lower density positions. This generates the 7- and 10-day subpopulation phenotypes where nearly
all fractions have increased stress resistance.
The generation of mutations through the damaged material has implications for the GASP
phenotype observed within the 4-day fraction competitions. Competing a lower density fraction
from a 4-day gradient, which hasn’t gained the increased stress resistance programming, against
one of the higher density fractions which does experience increased stress resistance will result in
the higher density fraction winning (Figure 4.6). It could also be due to the damage DNA receives
through ROS and advanced glycation end products (AGEs) build up within the higher density
fractions generating mutations, providing more possibilities for a potential beneficial mutation.
The lower density fractions are not as competitive at first within 4-day gradients because the
generation of mutations or an adoption of a more stress resistant programming has not provided
the population with as much of an advantage as the higher density, Fraction 12, population. When
we move further along the gradient, Fraction 8 has more similar fitness as Fraction 12 due to those
cells starting slightly older, slightly more damaged, and thus having more opportunities for
mutations upon division than the Fraction 6 subpopulation starting cultures. After 7-days of
incubation, the subpopulations of the culture begin to look more similar with lower density
fractions competing more similarly to their higher density counterparts (Supplemental Figure 4.3).
Likely because these cells have originated from those that have beneficial programming in the 4-
day gradients and beyond.
The asymmetric aging may be a beneficial strategy for the population as it is a form of “bet
hedging.” The populations’ ability to handle external stresses it may encounter within the natural
environment because the older-pole cells provide a stress resistance programming or beneficial
mutations to their new-pole daughters. The asymmetric aging generates cells of various ages and

88
damage levels compared to a symmetrically damage-segregating population. The new-pole
daughters, with the genetic make-up or physiological memory of their older-pole mothers can fully
manifest any benefits received since they experience a rejuvenated growth rate faster than their
damaged mother. This allows for populations to thrive within various conditions and continue to
propagate forward in less-than-ideal conditions. The ability for older, more damaged cells to
generate mutations or a stress response memory still needs to be examined, but this data along with
the theoretical approaches suggests the asymmetric aging phenomenon is a beneficial and
unavoidable strategy for a population and may not have arisen by chance, but by evolutionary
modes. Symmetrically dividing populations would not be selected as they wouldn’t generate the
mutations necessary for a diverse population, resulting in possible extinction. While having
damage possibly generates mutations that new-pole daughters can manifest if the mutations are
beneficial; however, there is a point of no return for the damaged, older-pole cells. Eventually
these cells will no longer divide due to their increasing damage. If the level of stress overloads this
system, the population will also experience extinction due to the lack of division creating the new-
pole daughters. There is a fine balance between life, senescence, and death at both the individual
cell level as well as the population level. Aging is fundamental process of life and bacteria do not
escape this phenomenon, but it appears that aging is not all negative as it is perceived in humans.
Research on bacterial aging could provide insight into the possible benefits aging may have.

89
4.5 MATERIALS AND METHODS
Percoll gradient centrifugation and fractionation of subpopulations. Percoll gradients were
formed by first generating a stock isotonic solution to the LB used for culture incubation by mixing
1 part 0.946M saline with 10 parts Percoll, then further diluted to the desired ~1.100 g/mL density
with a 0.086M saline solution (Biosciences, 2001). 10 mL polycarbonate tubes (Beckman) were
filled with 9.5 mL of the diluted Percoll solution and centrifuged in a Beckman Avanti J-E high-
speed centrifuge in a JA-25.50 rotor with the appropriate adaptors for 1 hour at 15,500 rpm at 2
degrees Celsius.
Cultures were inoculated from frozen stock (40% LB-glycerol at -80°C) into 18 x 150 mm
borosilicate test tubes containing 5 mL of LB medium. Both a kanamycin resistant E. coli strain
(SF2504, K-12 ZK126 DtnaA::KanR-YFP) and a chloramphenicol resistant E. coli strain (SF2503,
K-12 ZK126 DtnaA::CamR-CFP) was utilized for the presented studies. The use of two markers
allows for studying two populations within the same culture environment as is presented in the
competition experiments in Figure 4.6. Cultures were incubated overnight with aeration at 37°C
in a rolling drum (TC-7; New Brunswick Scientific, Edison, NJ) to achieve maximal population
density. These started cultures were started 1, 4, 7, and 10 days prior to the Percoll centrifugation
day in order to fractionate differentially aged populations within the same experiment day as much
as possible. Multiple Percoll gradients were performed for each age (1 day - 12, 4 day - 7, 7 day -
9, and 10 day - 8 replicates) at different time points and combined in a larger analysis. Each culture
is pelleted at 5000 rpm and resuspended in 500 µl of saline isotonic to the LB (0.086M) used for
incubation of the cultures prior to layering 0.5 mL onto the Percoll gradients. The Percoll gradient
of the cell suspension is then performed by centrifuging at 15,500 rpm for 45 minutes at 2 degrees
Celsius. Each time cell gradients are performed, a control gradient containing marker beads of

90
known densities (1.06 g/mL, 1.09 g/mL, 1.13 g/mL) is similarly performed. Gradients performed
at different times show consistent results according to the density marker bead placements (data
not shown) and they give a basis for the cell population densities observed in the gradient. Visual
bands are measured in mm from the bottom of the gradient (top of the gradient tube is 80 mm).
The gradients are then extracted using a pipette from the top in 0.5 mL increments and labeled
Fraction 1 for the top of the gradient to Fraction 20 for the bottom of the gradient. After extraction,
the cell fractions are pelleted, the supernatant is discarded, and resuspended in 0.5 mL 0.086M
saline. The viable cell counts per mL for each fraction were quantified by tittering appropriate
dilutions of cells and plating onto LB plates. The limit of detection for this method is <1000
CFU/mL (Kraigsley and Finkel, 2009).
Monitoring survival dynamics in monoculture, outgrowth, and competition experiments. For
the long-term monoculture survival assay, 5 µl of the Fraction 4, 8, 12, and 16 resuspensions from
1, 4, 7, and 10-day gradients were used to independently inoculate cultures containing 5 mL of
fresh LB to initiate a population from various subpopulations. At least three biological replicates
were generated in this manner for each fraction. For the experiments depicted in Figure 4.4, this
transfer represents day 0. Viable cell counts were determined as described above daily for ten days.
For the outgrowth of the 1- and 7-day fraction subpopulations to assess lag phase time and
exponential doubling rate, the fractions were diluted to ~10
6
CFU/mL with isotonic saline and then
25 µl of each fraction in the fraction pair was inoculated into the test tubes with 5 mL of LB. Two
biological replicates were generated for each fraction pair. Every 20 minutes for the first 2 hours
and every 30 minutes for 4 hours after that till 6 hours viable cell counts were determined by titer
as described above. Timepoints were taken at 12 and 24 hours to assess stationary phase population
counts from the starting subpopulation fraction pairs. The exit from lag phase and entrance into

91
exponential phase is marked by the first timepoint the viable cell counts were observed to double
from the previous timepoint. Doubling times for each individual culture were calculated for
appropriate time points, 240-330 minutes for all fractions except 1-day Fraction 6+7 which was
calculated from 210-270 minutes, by taking the log2 of the difference in time divided by the
difference in the logarithmic viable cell counts from those same timepoints.
For GASP Competition experiments (Figure 4.6, and Supplemental Figure 4.3), the 500 µl
fractions were saved in ~500 µl 40% glycerol-LB. at -80°C. The Fraction 12 populations starting
at higher initial density in the GASP competitions are initiated from each respective frozen stocks
for each age as described above and incubated overnight to achieve high cell density before
transferring 5 µl to test tubes with 5 mL of LB. Fraction 6 and Fraction 8 populations are started
from frozen and incubated overnight to achieve high cell density to inoculate 5 µl at a 1:1000
(vol:vol) ratio into the Fraction 12 population incubated to stationary phase the day prior. The
inoculation of the lower density population into the higher density population establishes day 0 of
the competition. Viable cell counts of the two populations throughout the co-incubation period
were measured over 10 days. To provide a baseline, total populations for each age sampled prior
to Percoll gradient centrifugation and taken through the same processing as the fractions are
examined in competition with the initially high viable cell count stationary phase total population
of the other marker.
Monitoring survival dynamics. Population densities for the monoculture, outgrowth, and GASP
competitions were determined by quantifying viable cell counts by titering appropriate dilutions
of cells, and plating in the presence or absence of chloramphenicol or kanamycin as appropriate.
The limit of detection for this method is <1000 CFU/mL (Kraigsley and Finkel, 2009). For
monoculture experiments, dilutions were plated on LB agar to determine viable cell counts of the

92
total population, in this case, the one strain present in the culture. For co-cultures, Fraction 6 and
Fraction 8 population cell counts were determined through plating on LB containing 30 µg/mL
chloramphenicol while the Fraction 12 population counts were determined by plating onto LB
plates containing 50 µg/mL kanamycin. For each experiment, viable cell counts were averaged
between biological replicates for each strain/population with the standard error or 95% confidence
interval shown as indicated. Standard error was calculated using the standard deviation in R of the
viable cell counts divided by the square root of the length of the data. 95% CI is expressed as two
times the standard error.
Stress assay. The stress assay plates are set up with 100 µl of 2X LB and 100 µl of a saline solution
for saline stress or nalidixic acid or streptomycin for antibiotic stress to the desired concentration
(saline: 250 mM-1 M, antibiotics: 3.125-250 µg/mL). Fractions are diluted to ~10
6
CFU/mL prior
to inoculating 10 µl in each stress concentration well along with a control well containing just LB
and sterile water used to dilute the stresses. Plates are incubated for two days overnight at 37
degrees Celsius and then scored for no growth, slight growth, or growth by eye.

93

Supplemental Figure 4.1. All replicates for fraction monocultures from Fig 4.4.
10
6
10
7
10
8
10
9
10
10
0 1 2 3 4 5 6 7 8 9 10
Day
CFU per mL
1 Day A
10
6
10
7
10
8
10
9
10
10
0 1 2 3 4 5 6 7 8 9 10
Day
CFU per mL
4 Day B
10
6
10
7
10
8
10
9
10
10
0 1 2 3 4 5 6 7 8 9 10
Day
CFU per mL
7 Day C
10
6
10
7
10
8
10
9
10
10
0 1 2 3 4 5 6 7 8 9 10
Day
CFU per mL
10 Day D

94

Supplemental Figure 4.2. All replicates of 1- and 7-day fraction outgrowths examining lag
time and logarithmic doubling times from Fig. 4.5.

95

Supplemental Figure 4.3. Competitive fitness of 10-Day subpopulations (A), and 7-Day
subpopulations (B) vs unaged parental strain when inoculated at 1:1000 (vol:vol). 10-day
subpopulations all exhibit a full GASP phenotype when competed against the unaged parental
strain, while 7-Day subpopulations do not fully express the GASP phenotype.

96

Supplemental Figure 4.4. Minimum Inhibitory Concentration of Percoll gradient
subpopulations to exogenous stressors – secondary trial.

97

Supplemental Figure 4.5. Percoll gradient of diluted 1-Day culture. 1-Day cultures diluted 10-
fold (tube 3) and 100-fold (tube 4) reveal similar gradient banding positions as a full yield 1-Day
culture gradient (tube 2). Loss of cell numbers does not change position of the visual band.

98
Day Total Fraction 4 Fraction 8 Fraction 12 Fraction 16
0 6.93x10
6

6.73x10
5
3.57x10
5
4.55x10
5

1.00x10
3
1.00x10
3
5.67x10
3
1.00x10
3

2.60x10
4
2.37x10
5

2.40x10
4
2.07x10
4

2.63x10
5

2.23x10
4

1.00x10
3

1.33x10
3

1.27x10
5

1.00x10
3

1.00x10
3

1.00x10
3

1 Day
4 Day
7 Day
10 Day
1 4.40x10
9

3.80x10
9
5.60x10
9
3.25x10
9

4.70x10
9
5.07x10
9
3.67x10
9
3.60x10
9

4.73x10
9

5.83x10
9

4.27x10
9

4.73x10
9

3.70x10
9
6.0x10
9

4.20x10
9

3.23x10
9

3.53x10
9
4.43x10
9

3.23x10
9

5.03x10
9

2 5.17x10
8

3.60x10
8
3.40x10
8
4.00x10
8

5.60x10
8
2.43x10
8
4.87x10
8
4.03x10
8

3.83x10
8

3.83x10
8

5.93x10
8

4.40x10
8

4.10x10
8

2.90x10
8
4.65x10
8

3.83x10
8

4.37x10
8
3.60x10
8
4.33x10
8
4.00x10
8

3 4.57x10
8

4.07x10
8
4.07x10
8
5.05x10
8

4.17x10
8
4.57x10
8
3.93x10
8

4.00x10
8

3.67x10
8

4.87x10
8

4.70x10
8

5.10x10
8

4.03x10
8

5.77x10
8
4.85x10
8

4.13x10
8

4.37x10
8

4.03x10
8

4.17x10
8
4.20x10
8

4 4.43x10
8
3.53x10
8
2.13x10
8
3.20x10
8

3.80x10
8
3.00x10
8
3.10x10
8
2.43x10
8

2.57x10
8

2.93x10
8

2.90x10
8

3.63x10
8

2.90x10
8
3.83x10
8

3.35x10
8

3.87x10
8

3.70x10
8

4.50x10
8
3.20x10
8

3.57x10
8

5 4.20x10
8
2.37x10
8
1.67x10
8
1.00x10
8

1.97x10
8
1.43x10
8
2.20x10
8
1.07x10
8

1.87x10
8

2.93x10
8

1.20x10
8

5.93x10
7

2.80x10
8

2.80x10
8

1.55x10
8

1.33x10
8

3.13x10
8
3.23x10
8
1.43x10
8
6.27x10
7

6 2.20x10
8
1.00x10
8
3.37x10
7
1.40x10
8

2.37x10
8
1.57x10
8
3.03x10
7
7.00x10
7

2.03x10
8

1.23x10
8

1.90x10
7

4.47x10
7

3.10x10
8

3.97x10
7

3.40x10
7
4.27x10
7

2.47x10
8

2.63x10
7
2.23x10
7

1.27x10
8

7 3.30x10
7
5.83x10
7
1.27x10
8
1.10x10
8

1.57x10
8
4.77x10
7
8.93x10
7
5.80x10
7

1.77x10
8

1.03x10
8

1.87x10
8

1.27x10
8

9.87x10
7

6.67x10
7

1.15x10
8
3.80x10
7

6.77x10
7
1.50x10
8

4.73x10
7
1.20x10
8

8 2.87x10
7
3.37x10
7
4.87x10
7
2.30x10
7

3.53x10
7
2.03x10
7
1.80x10
8
2.73x10
7

3.50x10
7

4.27x10
7

1.50x10
8
2.40x10
7

3.27x10
7

6.20x10
7

4.70x10
7
2.57x10
7

2.93x10
7

1.46x10
8
1.57x10
8

2.77x10
7

9 2.80x10
7
2.23x10
8
1.57x10
8
3.50x10
7

2.30x10
7
2.17x10
8
3.73x10
7

3.13x10
7

4.13x10
7
2.10x10
8

2.53x10
7
2.13x10
7

4.30x10
7
2.27x10
8

2.00x10
8

3.43x10
7

2.50x10
7

1.24x10
8

1.20x10
8

3.07x10
7

10 1.97x10
8
3.20x10
7
5.37x10
7
2.75x10
7

1.70x10
8
5.07x10
7
3.73x10
7
3.23x10
7

1.63x10
8

4.03x10
7

3.23x10
7

2.73x10
7

1.77x10
8

3.43x10
7

4.15x10
7

2.90x10
7

1.50x10
8

4.80x10
7

2.57x10
7

2.87x10
7

Supplemental Table 4.1. Average CFU values from Fig. 4.4.

99
Chapter 5: Future Directions for Clarifying the Linkage between Survival and Senescence
of Bacterial Populations

5.1 LINKING POLE AGE AND DAMAGE LEVELS WITH PERCOLL GRADIENT
SUBPOPULATIONS
While I have shown that subpopulations separated with Percoll density gradient
centrifugation have differential fitness patterns when in competition and stress assays, the link
between specific damage levels and fitness hasn’t been fully elucidated. Previous studies suggest
that higher density subpopulations are more like stationary-phase cells and lower density
subpopulations are more exponential phase-like (Makinoshima et al., 2002) and that inclusion
body or damaged biomaterial aggregate levels have some impact on where cells fall within the
gradient (Pandey et al., 2013). However, a direct correlation with pole age has not been directly
proven. Here I propose a few approaches to further characterize the subpopulations based on
damage levels and relative pole age.
One method to assess damage levels of the subpopulations is through quantifying the level
of carbonylation occurring within the subpopulations through carboxymethyl lysine detection
(CML), a common AGE found within bacterial cultures (Pepper et al., 2010). A direct ELISA
approach using an antibody against CML can be used for this and has been used to quantify CML
levels in previous studies (Kram and Finkel, 2015). An ELISA approach could be useful in
quantifying other forms of damage as well if there exists a targeted antibody.
Another approach to assess damage levels of subpopulation fractions and relative pole age
would be to quantify the amount of fluorescence in each subpopulation separated by Percoll
gradient centrifugation by using an inclusion binding protein (IbpA) (Allen et al., 1992; Lindner

100
et al., 2008) tagged with a fluorescent molecule (GFP/YFP/etc) within a flow cytometer. This
would allow for quantification of the protein aggregate levels within each subpopulation. One
could use a population with the IbpA-YFP label for separation in the Percoll gradients, then
quantify the levels of fluorescence of each subpopulation fraction in a flow cytometer to gauge the
amount of damage each subpopulation possesses. Even further, one could use a FACS cell sorter
to separate the population based on fluorescence levels, then utilize the separated populations for
further fitness studies.
Flow cytometry and FACS cell sorting methods could be useful for quantifying relative
pole age as well through peptidoglycan or membrane markers. I have previously tried to
characterize pole age through the use of NADA, a fluorescently tagged D-amino acid that is
incorporated into newly synthesized peptidoglycan as cells undergo division (Kuru et al., 2013;
Kuru et al., 2015; Hsu et al., 2017). I used a pulse-chase method of incorporating the NADA into
exponentially dividing cells and then reinoculated fully labeled cells into fresh NADA-free, LB
medium to determine whether we could titrate out the NADA with each cell division, giving some
measure of generation time at least in exponentially dividing cultures. Labeled cells could be
distinguished from unlabeled cells in directly mixed cultures with values very similar to the
expected ratios (Figure 5.1). Further, cells labeled with NADA and recently birthed cells within
the reinoculation experiment could be observed separately within the flow cytometer, but without
variation within the NADA fluorescence signal. The labeled cells are washed out of the culture by
about 6 hours post inoculation falling below 5% of the cells counted within the flow cytometer
(Figure 5.2). The NADA marker was therefore deemed not feasible for tracking generation times.
Perhaps a different peptidoglycan or membrane marker that has a stronger signal could be used to
track the older generations vs new-pole cells from the pole age perspective.

101

Figure 5.1 Flow cytometry counts from mixed cultures of NADA-labeled cells and unlabeled
cells. Cell counts (y-axis, left and top right) of fluorescence intensities (x-axis, left plot) within the
mixed cultures with various ratios of fully labeled cells with NADA to unlabeled cells within the
population show separation between the two groups and percentages close to expected (bottom
right plot).

102

Figure 5.2 Flow cytometry counts of dividing NADA-labeled cells inoculated in fresh
medium. Cell counts (y-axis, left and top right) of fluorescence intensities (x-axis, left plot) of
NADA-labeled cells compared to unlabeled cells during outgrowth of a fully NADA-labeled
starting population shows the labeled group falling below 5% after 6 hours (bottom right).

103
5.2 FURTHER ASSESSMENT OF COMPETITIVE PHENOTYPES
Despite not fully elucidating the subpopulations pole ages and direct damage levels, we
have demonstrated that subpopulations fractionated from 4-day cultures exhibit differential fitness
when competed against a higher-density, Fraction 12 subpopulation (Chapter 4). Fraction 6 is less
fit when competed against Fraction 12 subpopulations, while moving further down in the gradient
to Fraction 8 subpopulations have similar fitness to Fraction 12 subpopulations. The specific
competitive fitness of each subpopulation for each age still needs to be determined. It is unclear if
what we observe is the case in all Fraction 6 subpopulations regardless of gradient and regardless
of age or whether this only happens from cells harvested from 4-day gradients. The MICs to
various stresses, lag times, and doubling times of each subpopulation begin to look more similar
using cells from 7-day gradients compared to 4-day gradient subpopulation phenotypes. For
cultures that are aged further, into 10-days and beyond, the gradient distributions and the stress
assay phenotypes appear more like the 7-day gradients, so it is possible that there may be less
subpopulation variation at this time point of culturing.
Another more stringent way to test subpopulation fitness in periods of famine is to first
condition the media with an aging culture and then examine subpopulation survival within that
conditioned media. The culture may be more depleted of nutrients and have increased levels of
toxins, providing a less than ideal environment for the culture, allowing further assessment of the
diversity present that may allow acclimation to these environments. Despite not observing many
differences between the monocultures within LB cultures started from different subpopulations
(see Figure 4.4), additional differences may be observed when examining the subpopulations of
cells incubated within conditioned media. Factors that might vary could include long-term survival
ability or changes in outgrowth parameters.

104
5.3 RNA SEQUENCING OF BACTERIAL SUBPOPULATIONS
Gene expression differences between the subpopulations would also be interesting to
examine along with determining whether these patterns shift in cells over an individual gradient
and/or over time from 1 to 10 days. Perhaps the programming for subpopulations is different within
the same gradient. This would be consistent with the differential fitness observed within the stress
assay in Figure 4.7. Expression levels for similar fractions between different ages could also tell
us whether those cells are more “stationary phase-like” or more “exponential phase-like” and
possibly also identify differences between “newer” vs “older” cells. Cells that are older may
upregulate certain repair genes as discussed in Chapter 1, while cells that are younger may be more
focused on maintenance and replication. Performing RNA-sequencing on the subpopulations
isolated from the Percoll gradients might answer this directly by showing the change in expression
patterns of all possible genes; a more targeted RT-qPCR approach could also be used on candidate
genes if there is too little cellular material to perform RNA-sequencing due to the much smaller
viable cell counts within each fraction compared to most cultures at peak viable cell counts used
for RNA-sequencing. The more targeted RT-qPCR approach could detect expression patterns at
much lower density, but with less scope of the genome.
5.4 ASSESSMENT OF DNA DAMAGE AND MUTATION FREQUENCIES OF
SUBPOPULATIONS
Since differential phenotypes were observed within the competitive fitness assays shown
in Figure 4.6 and within the stress assays shown within Figure 4.7, characterizing the mutations
and/or mutation frequencies within the subpopulations could reveal pertinent information about
how mutations arise and spread within the greater population. If our hypothesis is that older-pole
cells generate more mutations is correct, then it is possible that differences in the number of

105
mutations would be observed between populations started from different subpopulations.
Populations started recently from higher density subpopulations could show higher amounts of
diversity present due to the error prone polymerases discussed in Chapter 1 copying through
damaged DNA lesions adding non-complimentary bases compared to the rejuvenated new-pole
mothers. Characterizing the specific SNPs within the subpopulation cultures and then further doing
Percoll gradient fractionation with those populations to examine where those genotypes fall could
also show how a mutation from an older-pole cell within a higher density fraction could be passed
to a lower density new-pole cell. This could explain the adoption of phenotypes within the
monoculture survival curves in Figure 4.4 and the adoption of the MICs as culture age within
Figure 4.7.
5.5 CLOSING REMARKS
Overall, studying the asymmetric aging phenomenon, the segregation of damage, and the
relationship of these both to genetic diversity is complex. Despite the complications, aging in the
context of survival is an important topic to explore. Aging occurs in every species across the earth
and the benefit of aging to populations could be why it occurs in the first place. The advantages
aging provides to bacterial survival and evolution could have been present prior to the
endosymbiosis theory and thus is passed on to higher level organisms. Perhaps aging isn’t
something that should be complained about, but is a benefit to all species’ populations in some
way or another.

106
References
Ackermann, M., Chao, L., Bergstrom, C., and Doebeli, M. (2007) On the Evolutionary Origin of
Aging. Aging Cell, 6: 235–244. https://doi.org/10.1111/j.1474-9726.2007.00281.x
Aguilaniu, H., Gustafsson, L., Rigoulet, M., and Nyström, T. (2003) Asymmetric Inheritance of
Oxidatively Damaged Proteins During Cytokinesis. Science (80- ) 299: 1751–1754.
https://doi.org/10.1126/science.1080418
Allen, S.P., Polazzi, J.O., Gierse, J.K., and Easton, A.M. (1992) Two novel heat shock genes
encoding proteins produced in response to heterologous protein expression in Escherichia coli. J
Bacteriol, 174: 6938–6947. https://doi.org/10.1128/jb.174.21.6938-6947.1992
Ballesteros, M., Fredriksson, Å., Henriksson, J., and Nyström, T. (2001) Bacterial senescence:
Protein oxidation in non-proliferating cells is dictated by the accuracy of the ribosomes. EMBO
J, 20: 5280–5289. https://doi.org/10.1093/emboj/20.18.5280
Baumann, P., Baumann, L., Bang, S.S., and Woolkalis, M.J. (1980) Reevaluation of the
taxonomy of Vibrio, beneckea, and Photobacterium: Abolition of the genus Beneckea. Current
Microbiology, 4: 127–132. https://doi.org/10.1007/BF02602814
Biosciences, A. (2001) Percoll Methodology and Applications. Handbooks from Amersham
Biosci: 1–84.
Boedicker, J., and Nealson, K. (2016) Microbial Communication via Quorum Sensing. IEEE
Trans Mol Biol Multi-Scale Commun, 1(4): 310-320.
https://doi.org/10.1109/TMBMC.2016.2587629
Boehm, A., Arnoldini, M., Bergmiller, T., Röösli, T., Bigosch, C., and Ackermann, M. (2016)
Genetic Manipulation of Glycogen Allocation Affects Replicative Lifespan in E. coli. PLoS
Genet, 12. https://doi.org/10.1371/journal.pgen.1005974
Bongrand, C., Koch, E. J., Moriano-Gutierrez, S., Cordero, O. X., McFall-Ngai, M., Polz, M. F.,
& Ruby, E. G. (2016). A genomic comparison of 13 symbiotic Vibrio fischeri isolates from the
perspective of their host source and colonization behavior. The ISME journal, 10(12): 2907–
2917. https://doi.org/10.1038/ismej.2016.69
Bucala, R., Modelt, P., and Cerami, A. (1984) Modification of DNA by reducing sugars : A
possible mechanism for nucleic acid aging and age-related dysfunction in gene expression. Proc
Natl Acad Sci USA, 81: 105–109. https://doi.org/10.1073/pnas.81.1.105
Chao, L. (2010) A model for damage load and its implications for the evolution of bacterial
aging. PLoS Genet, 6. https://doi.org/10.1371/journal.pgen.1001076
Chao, L., Rang, C.U., Proenca, A.M., and Chao, J.U. (2016) Asymmetrical Damage Partitioning
in Bacteria: A Model for the Evolution of Stochasticity, Determinism, and Genetic Assimilation.
PLoS Comput Biol, 12. https://doi.org/10.1371/journal.pcbi.1004700

107
Clark, M.W., Yie, A.M., Eder, E.K., Dennis, R.G., Basting, P.J., Martinez, K.A., et al. (2015)
Periplasmic acid stress increases cell division asymmetry (Polar Aging) of Escherichia coli.
PLoS One, 10. https://doi.org/10.1371/journal.pone.0144650
Corzett, C.H., Goodman, M.F., and Finkel, S.E. (2013) Competitive fitness during feast and
famine: How SOS DNA polymerases influence physiology and evolution in Escherichia coli.
Genetics, 194: 409–420. https://doi.org/10.1534/genetics.113.151837
Dalle-Donne, I., Rossi, R., Giustarini, D., Milzani, A., and Colombo, R. (2003) Protein carbonyl
groups as biomarkers of oxidative stress. Clin Chim Acta, 329: 23–38.
https://doi.org/10.1016/S0009-8981(03)00003-2
Dillon, M. M., Sung, W., Sebra, R., Lynch, M., & Cooper, V. S. (2017). Genome-Wide Biases in
the Rate and Molecular Spectrum of Spontaneous Mutations in Vibrio cholerae and Vibrio
fischeri. Molecular Biology and Evolution, 34(1): 93–109.
https://doi.org/10.1093/molbev/msw224
Dimitrova, R., Mironova, R., and Ivanov, I. (2004) Glycation of Proteins in Escherichia coli:
Effect of nutreint broth ingredients on glycation. Biotechnol Biotechnol Equip, 18: 99–103.
https://doi.org/10.1080/13102818.2004.10817094
Dukan, S., and Nyström, T. (1998) Bacterial senescence: Stasis results in increased and
differential oxidation of cytoplasmic proteins leading to developmental induction of the heat
shock regulon. Genes Dev, 12: 3431–3441. https://doi.org/10.1101/gad.12.21.3431
Dukan, S., and Nyström, T. (1999) Oxidative stress defense and deterioration of growth-arrested
Escherichia coli cells. J Biol Chem, 274: 26027–26032. https://doi.org/10.1074/jbc.274.37.26027
Finkel S. E. (2006). Long-term survival during stationary phase: evolution and the GASP
phenotype. Nature reviews. Microbiology, 4(2): 113–120. https://doi.org/10.1038/nrmicro1340
Finkel, S. E., & Kolter, R. (1999). Evolution of microbial diversity during prolonged
starvation. Proceedings of the National Academy of Sciences of the United States of
America, 96(7): 4023–4027. https://doi.org/10.1073/pnas.96.7.4023
Fredriksson, A., and Nyström, T. (2006) Conditional and replicative senescence in Escherichia
coli. Curr Opin Microbiol, 9: 612–618. https://doi.org/10.1016/j.mib.2006.10.010
Govers, S.K., Mortier, J., Adam, A., and Aertsen, A. (2018) Protein aggregates encode
epigenetic memory of stressful encounters in individual Escherichia coli cells. PLOS Biol, 16:
e2003853. https://doi.org/10.1371/journal.pbio.2003853
Grimbergen, A.J., Siebring, J., Solopova, A., and Kuipers, O.P. (2015) Microbial bet-hedging:
The power of being different. Curr Opin Microbiol, 25: 67–72.
https://doi.org/10.1016/j.mib.2015.04.008

108
Guerrero-Ferreira, R., Gorman, C., Chavez, A. A., Willie, S., & Nishiguchi, M. K. (2013).
Characterization of the bacterial diversity in Indo-West Pacific loliginid and sepiolid squid light
organs. Microbial Ecology, 65(1): 214–226. https://doi.org/10.1007/s00248-012-0099-6
Gupta, A., Lloyd-Price, J., Oliveira, S.M.D., Yli-Harja, O., Muthukrishnan, A.B., and Ribeiro,
A.S. (2014) Robustness of the division symmetry in Escherichia coli and functional
consequences of symmetry breaking. Phys Biol, 11. https://doi.org/10.1088/1478-
3975/11/6/066005
Hastings, J.W., and Nealson, K.H. (1977) Bacterial Bioluminescence. Annual Review
Microbiology, 31: 549–595. https://doi.org/10.1146/annurev.mi.31.100177.003001
Hodge, J.E. (1953) Browning Reaction Theories Integrated in Review Chemistry of Browning
Reactions in Model Systems. Agric Food Chem, 1: 928–943.
Hsu, Y., Rittichier, J., Kuru, E., Yablonowski, J., Pasciak, E., Tekkam, S., et al. (2017) Full color
palette of fluorescent D -amino acids for in situ labeling of bacterial cell walls. Chem Sci, 8:
6313–6321. https://doi.org/10.1039/C7SC01800B
Janakiraman, A., and Goldberg, M. (2004) Recent advances on the development of bacterial
poles. Trends in Microbiol, 121356: 518–525. https://doi.org/10.1016/j.tim.2004.09.003
Kaberdin, V. R., Montánchez, I., Parada, C., Orruño, M., Arana, I., & Barcina, I. (2015).
Unveiling the Metabolic Pathways Associated with the Adaptive Reduction of Cell Size During
Vibrio harveyi Persistence in Seawater Microcosms. Microbial Ecology, 70(3): 689–700.
https://doi.org/10.1007/s00248-015-0614-7
Koleva, K.Z., and Hellweger, F.L. (2015) From protein damage to cell aging to poulation fitness
in E. coli: Insights from a multi-level agent-based model. Ecol Modell, 301: 62–71.
https://doi.org/10.1016/j.ecolmodel.2015.01.024
Kraigsley, A. M., & Finkel, S. E. (2009). Adaptive evolution in single species bacterial
biofilms. FEMS microbiology letters, 293(1): 135–140. https://doi.org/10.1111/j.1574-
6968.2009.01526.x
Kram, K. E., & Finkel, S. E. (2014). Culture volume and vessel affect long-term survival,
mutation frequency, and oxidative stress of Escherichia coli. Applied and environmental
microbiology, 80(5): 1732–1738. https://doi.org/10.1128/AEM.03150-13
Kram, K. E., & Finkel, S. E. (2015). Rich Medium Composition Affects Escherichia coli
Survival, Glycation, and Mutation Frequency during Long-Term Batch Culture. Applied and
environmental microbiology, 81(13): 4442–4450. https://doi.org/10.1128/AEM.00722-15
Kram, K. E., Geiger, C., Ismail, W. M., Lee, H., Tang, H., Foster, P. L., & Finkel, S. E. (2017).
Adaptation of Escherichia coli to Long-Term Serial Passage in Complex Medium: Evidence of
Parallel Evolution. mSystems, 2(2): e00192-16. https://doi.org/10.1128/mSystems.00192-16

109
Kuru, E., Hughes, H.V., Brown, P.J., Hall, E., Tekkam, S., Cava, F., et al. (2013) In situ Probing
of Newly Synthesized Peptidoglycan in Live Bacteria with Fluorescent D-Amino Acids. Angew
Chem Int Ed Engl, 51: 12519–12523. https://doi.org/10.1002/anie.201206749
Kuru, E., Tekkam, S., Hall, E., Brun, Y. V, and Vannieuwenhze, M.S. (2015) Synthesis of
fluorescent D-amino acids (FDAAs) and their use for probing peptidoglycan synthesis and
bacterial growth in situ. Nat Protoc, 10: 33–52. https://doi.org/10.1038/nprot.2014.197
Kysela, D.T., Brown, P.J.B., Huang, K.C., and Brun, Y. V. (2013) Biological Consequences and
Advantages of Asymmetric Bacterial Growth. Annu Rev Microbiol, 67: 417–435. https://doi.org/
10.1146/annurev-micro-092412-155622
Lapinska, U., Glover, G., Capilla-Lasheras, P., Young, A.J., and Pagliara, S. (2019) Bacterial
Ageing in the Absence of External Stressors. Phil Trans R Soc B, 374: 1–13.
https://doi.org/10.1098/rstb.2018.0442
Lee, H., Popodi, E., Tang, H., & Foster, P. L. (2012). Rate and molecular spectrum of
spontaneous mutations in the bacterium Escherichia coli as determined by whole-genome
sequencing. Proceedings of the National Academy of Sciences of the United States of
America, 109(41): E2774–E2783. https://doi.org/10.1073/pnas.1210309109
Lindner, A.B., Madden, R., Demarez, A., Stewart, E.J., and Taddei, F. (2008) Asymmetric
segregation of protein aggregates is associated with cellular aging and rejuvenation. Proc Natl
Acad Sci, 105: 3076–3081. https://doi.org/10.1073/pnas.0708931105
Maisonneuve, E., Ezraty, B., and Dukan, S. (2008) Protein aggregates: An aging factor involved
in cell death. J Bacteriol, 190: 6070–6075. https://doi.org/10.1128/JB.00736-08
Maisonneuve, E., Fraysse, L., Lignon, S., Capron, L., and Dukan, S. (2008) Carbonylated
proteins are detectable only in a degradation-resistant aggregate state in Escherichia coli. J
Bacteriol, 190: 6609–6614. https://doi.org/10.1128/JB.00588-08
Makemson J. C. (1986). Luciferase-dependent oxygen consumption by bioluminescent
Vibrios. Journal of Bacteriology, 165(2): 461–466. https://doi.org/10.1128/jb.165.2.461-
466.1986
Makinoshima, H., Aizawa, S.I., Hayashi, H., Miki, T., Nishimura, A., and Ishihama, A. (2003)
Growth phase-coupled alterations in cell structure and function of Escherichia coli. J Bacteriol,
185: 1338–1345. https://doi.org/10.1128/JB.185.4.1338-1345.2003
Makinoshima, H., Makinoshima, H., Nishimura, A., Nishimura, A., Ishihama, A., and Ishihama,
A. (2002) Fractionation of Escherichia coli cell populations at different stages during growth
transition to stationary phase. Mol Microbiol, 43: 269–79. https://doi.org/10.1046/j.1365-
2958.2002.02746.x
Mortier, J., Tadesse, W., Govers, S.K., and Aertsen, A. (2019) Stress-induced protein aggregates
shape population heterogeneity in bacteria. Curr Genet: 1–5. https://doi.org/10.1007/s00294-
019-00947-1

110
Nealson, B.K.H. (1978) Isolation, Identification, and Manipulation of Luminous Bacteria.
Methods Enzymology LVI: 153–166. https://doi.org/10.1016/0076-6879(78)57017-1
Nealson, K.H., and Hastings, J.W. (1979) Bacterial Bioluminescence: Its Control and Ecological
Significance. Microbiology Reviews 43(4): 496–518. https://doi.org/10.1128/mr.43.4.496-
518.1979
Nealson, K. H., & Hastings, J. W. (2006). Quorum sensing on a global scale: massive numbers
of bioluminescent bacteria make milky seas. Applied and Environmental Microbiology, 72(4):
2295–2297. https://doi.org/10.1128/AEM.72.4.2295-2297.2006
Nyholm, S. V., & McFall-Ngai, M. J. (2004). The winnowing: establishing the squid-Vibrio
symbiosis. Nature reviews. Microbiology, 2(8): 632–642. https://doi.org/10.1038/nrmicro957
Nyström, T. (2005) Role of oxidative carbonylation in protein quality control and senescence.
EMBO J, 24: 1311–1317. https://doi.org/10.1038/sj.emboj.7600599
Pandey, N., Sachan, A., Chen, Q., Ruebling-Jass, K., Bhalla, R., Panguluri, K.K., et al. (2013)
Screening and identification of genetic loci involved in producing more/denser inclusion bodies
in Escherichia coli. Microb Cell Fact, 12. https://doi.org/10.1186/1475-2859-12-43
Pedro, M.A. De, Grünfelder, C.G., and Schwarz, H. (2004) Restricted Mobility of Cell Surface
Proteins in the Polar Regions of Escherichia coli. J Bacteriol, 186: 2594–2602.
https://doi.org/10.1128/JB.186.9.2594-2602.2004
Pedro, M.A. De, Quintela, J.C., Höltje, J.V., and Schwarz, H. (1997) Murein segregation in
Escherichia coli. J Bacteriol, 179: 2823–2834. https://doi.org/10.1128/jb.179.9.2823-2834.1997
Pepper, E.D., Farrell, M.J., Nord, G., and Finkel, S.E. (2010) Antiglycation effects of carnosine
and other compounds on the long-term survival of Escherichia coli. Appl Environ Microbiol, 76:
7925–7930. https://doi.org/10.1128/AEM.01369-10
Petrun, B., and Lostroh, C.P. (2013) Vibrio fischeri exhibit the growth advantage in stationary-
phase phenotype. Can J Microbiol, 59(2): 130–135. https://doi.org/10.1139/cjm-2012-0439
Proenca, A.M., Rang, C.U., Buetz, C., Shi, C., and Chao, L. (2018) Age structure landscapes
emerge from the equilibrium between aging and rejuvenation in bacterial populations. Nat
Commun, 9. https://doi.org/10.1038/s41467-018-06154-9
Proenca, A.M., Rang, C.U., Id, A.Q., Shi, C., and Chao, L. (2019) Cell aging preserves cellular
immortality in the presence of lethal levels of damage. PLoS Biol: 1–21.
https://doi.org/10.1371/journal.pbio.3000266
Rang, C.U., Peng, A.Y., and Chao, L. (2011) Temporal dynamics of bacterial aging and
rejuvenation. Curr Biol, 21: 1813–1816. https://doi.org/10.1016/j.cub.2011.09.018

111
Rang, C.U., Peng, A.Y., Poon, A.F., and Chao, L. (2012) Ageing in Escherichia coli requires
damage by an extrinsic agent. Microbiol (United Kingdom), 158: 1553–1559.
https://doi.org/10.1099/mic.0.057240-0
Ratib, N. R., Seidl, F., Ehrenreich, I. M., and Finkel, S. E. (2021) Evolution in Long-Term
Stationary-Phase Batch Culture: Emergence of Divergent Escherichia coli Lineages over 1,200
Days. mBio, 12(1): 1–18. https://doi.org/10.1128/mBio.03337-20
Reichelt, J., and Baumann, P. (1973) Taxonomy of Marine, Luminous Bacteria. Archiv
Mikrobiol,, 94: 283–330. https://doi.org/10.1007/BF00769027
Rokney, A., Shagan, M., Kessel, M., Smith, Y., Rosenshine, I., and Oppenheim, A.B. (2009) E.
coli Transports Aggregated Proteins to the Poles by a Specific and Energy-Dependent Process. J
Mol Biol, 392(3): 589–601. https://doi.org/10.1016/j.jmb.2009.07.009
Ruby, E.G. (1996) Lessons from a Cooperative, Bacterial-Animal Association: The Vibrio
fischeri–Euprymna scolopes Light Organ Symbiosis. Ann Rev Microbiol, 50: 591–624.
https://doi.org/10.1146/annurev.micro.50.1.591
Setiawan, W.A., Widyastuti, U., and Yuhana, M. (2015) Detection of Luminous Vibrio harveyi
in Penaeid Shrimp Through Nested PCR Using Haemolysin Gene Primer. HAYATI Journal of
Biosciences, 22(2): 60–66 https://doi.org/10.4308/hjb.22.2.60
Shi, C., Chao, L., Proenca, A.M., Qiu, A., Chao, J., and Rang, C.U. (2020) Allocation of gene
products to daughter cells is determined by the age of the mother in single Escherichia coli cells.
Proc R Soc B Biol Sci, 287: 1–9. https://doi.org/10.1098/rspb.2020.0569
Soonthornchai, W., Chaiyapechara, S., Jarayabhand, P., Söderhäll, K., & Jiravanichpaisal, P.
(2015). Interaction of Vibrio spp. with the Inner Surface of the Digestive Tract of Penaeus
monodon. PloS one, 10(8): e0135783. https://doi.org/10.1371/journal.pone.0135783
Soto, W., Travisano, M., Tolleson, A. R., & Nishiguchi, M. K. (2019). Symbiont evolution
during the free-living phase can improve host colonization. Microbiology (Reading,
England), 165(2): 174–187. https://doi.org/10.1099/mic.0.000756
Stabili, L., Gravili, C., Piraino, S., Boero, F., & Alifano, P. (2006). Vibrio harveyi associated
with Aglaophenia octodonta (Hydrozoa, Cnidaria). Microbial Ecology, 52(4): 603–608.
https://doi.org/10.1007/s00248-006-9010-7
Stephens, C. (2005) Senescence: Even bacteria get old. Curr Biol 15(8).
https://doi.org/10.1016/j.cub.2005.04.006
Stewart, E.J., Madden, R., Paul, G., and Taddei, F. (2005) Aging and death in an organism that
reproduces by morphologically symmetric division. PLoS Biol 3: 0295–0300.
https://doi.org/10.1371/journal.pbio.0030045

112
Thompson, L. R., Nikolakakis, K., Pan, S., Reed, J., Knight, R., & Ruby, E. G. (2017).
Transcriptional characterization of Vibrio fischeri during colonization of juvenile Euprymna
scolopes. Environmental Microbiology, 19(5): 1845–1856. https://doi.org/10.1111/1462-
2920.13684
Vedel, S., Nunns, H., Semsey, S., Semsey, S., and Trusina, A. (2016) Asymmetric Damage
Segregation Constitutes an Emergent Population-Level Stress Response. Cell Systems, 3(2):
187–198. https://doi.org/10.1016/j.cels.2016.06.008
Vollmer, W., Blanot, D., and Pedro, M.A. De (2008) Peptidoglycan structure and architecture.
FEMS Microbiol Rev, 32(2): 149–167. https://doi.org/10.1111/j.1574-6976.2007.00094.x
Vollmer, W., Höltje, J., and Ho, J. (2004) The Architecture of the Murein(Peptidoglycan) in
Gram-Negative Bacteria : Vertical Scaffold or Horizontal Layer(s). J Bacteriol, 186(18): 5978–
5987. https://doi.org/10.1128/JB.186.18.5978-5987.2004
Wimpee, C. F., Nadeau, T. L., & Nealson, K. H. (1991). Development of species-specific
hybridization probes for marine luminous bacteria by using in vitro DNA amplification. Applied
and environmental microbiology, 57(5): 1319–1324. https://doi.org/10.1128/aem.57.5.1319-
1324.1991
Wong, G.T., Bonocora, R.P., Schep, A.N., Beeler, S.M., Fong, A.J.L., Shull, L.M., et al. (2017)
Genome-Wide Transcriptional Response to Varying RpoS Levels in Escherichia coli K-12. J
Bacteriol, 199(7): 1–17. https://doi.org/10.1128/JB.00755-16
Yang, Y., Santos, A.L., Xu, L., Lotton, C., Taddei, F., and Lindner, A.B. (2019) Temporal
scaling of aging as an adaptive strategy of Escherichia coli. Sci Adv, 5(5): 1–9.
https://doi.org/10.1126/sciadv.aaw2069
Yeiser, B., Pepper, E. D., Goodman, M. F., & Finkel, S. E. (2002). SOS-induced DNA
polymerases enhance long-term survival and evolutionary fitness. Proceedings of the National
Academy of Sciences of the United States of America, 99(13): 8737–8741.
https://doi.org/10.1073/pnas.092269199
Zambrano, M. M., Siegele, D. A., Almirón, M., Tormo, A., & Kolter, R. (1993). Microbial
competition: Escherichia coli mutants that take over stationary phase cultures. Science (New
York, N.Y.), 259(5102): 1757–1760. https://doi.org/10.1126/science.7681219
Zinser, E.R., and Kolter, R. (1999) Mutations Enhancing Amino Acid Catabolism Confer a
Growth Advantage in Stationary Phase. Journal of Bacteriology, 181(18): 5800–5807.
https://doi.org/10.1128/JB.181.18.5800-5807.1999

Abstract (if available)

Abstract The only things certain about life are aging and death. While this is typically a principle thought about for people, it is also true for bacterial species. Bacteria can exist within many environments and populations can persist for many years under conditions of famine, but there is frequently a constant flux of new cells that arise and cells that die, particularly within long-term cultures. As individual cells age within a population, they accumulate damage, experience slowed growth rates, eventually stop dividing, and die off. Despite the negative implications associated with aging, the damage accumulation within individual cells may allow for the generation of additional mutations, some of which might be beneficial. Therefore, when a cell divides and two daughter cells are produced: one old-pole receiving cell obtains the maternal damage and one “new-pole” cell is cleansed of the maternal damage and experiences a rejuvenated growth rate. While both the old- and the new-pole cell receive the mutations present within the maternal cell, the new-pole daughter can express any beneficial mutations and, along with its faster growth rate and doubling time, pass any benefits on to its progeny. This allows the population to persist while the older individuals die off. Therefore, while aging is an inevitable fact of life with cells marching towards death, the aging process may also contribute to the genetic diversity within the population. Here we explore this linkage between aging and survival at the population level.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Genomic and physiological characterization of Escherichia coli evolving in long-term batch culture

PDF

Bacterial survival under low nutrient conditions in the laboratory and the deep biosphere

PDF

Mechanisms of long-term survival in Escherichia coli

PDF

Evolution of E. coli in unstructured and structured environments

PDF

Genome-wide GATC methylation in long-term cultures of Escherichia coli K-12

PDF

Survival and evolution of Shewanella oneidensis MR-1: applications for microbial fuel cells

PDF

Genetic diversity and bacterial death in the context of adaptive evolution

PDF

Dps contributes to typical growth, survival, and genome organization in E. coli

PDF

Microbial interaction networks: from single cells to collective behavior

PDF

Physiological roles and evolutionary implications of alternative DNA polymerases in Escherichia coli

PDF

The molecular adaptation of Trichodesmium to long-term CO₂-selection under multiple nutrient limitation regimes

PDF

A single cell time course of senescence uncovers discrete cell trajectories and transcriptional heterogeneity

PDF

Natural divergence of traits in species of mice reveals novel molecular mechanisms of cellular senescence

PDF

Characterization of senescent cell heterogeneity using cell culture models

PDF

From single molecules to bacterial nanowires: functional and dynamic imaging of the extracellular electron transfer network in Shewanella oneidensis MR-1

PDF

Ancestral inference and cancer stem cell dynamics in colorectal tumors

PDF

Mediating the homeostatic response during light & cytotoxic induced retinal damage

PDF

Incidence and survival rates of the three major histologies of renal cell carcinoma

PDF

Sex differences in aging and the effects of mitochondria

PDF

Biomarkers of age-related health changes: associations with health outcomes and disparities

Asset Metadata

Creator Allen, Calista (author)

Core Title Mechanisms of long-term survival of Vibrio harveyi and Escherichia coli: linking damage and senescence

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Molecular Biology

Degree Conferral Date 2022-08

Publication Date 07/20/2022

Defense Date 05/04/2022

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

Tag asymmetric aging,bacterial population fitness,damage accumulation,damage and senescence,E. coli,Escherichia coli,GASP,growth advantage in stationary phase,long-term survival,OAI-PMH Harvest,pole age,V. harveyi,Vibrio harveyi

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Finkel, Steve (committee chair), Dean, Matthew (committee member), Ehrenreich, Ian (committee member), El-Naggar, Moh (committee member), Nealson, Kenneth (committee member)

Creator Email calistaa@usc.edu,calistaallen@outlook.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC111373668

Unique identifier UC111373668

Legacy Identifier etd-AllenCalis-10881

Document Type Dissertation

Format application/pdf (imt)

Rights Allen, Calista

Type texts

Source 20220721-usctheses-batch-958 (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