University of Southern California Dissertations and Theses

Characterization of senescent cell heterogeneity using cell culture models

(USC Thesis Other)

Characterization of senescent cell heterogeneity using cell culture models

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content CHARACTERIZATION OF SENESCENT CELL HETEROGENEITY
USING CELL CULTURE MODELS
by
Francesco Neri
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
(BIOLOGY OF AGING)
August 2024
Copyright 2024 Francesco Neri

ii
Table of Contents
List of Figures..................................................................................................................iii
Abstract........................................................................................................................... v
Chapter 1: Quantitative Proteomic Analysis of the Senescence-Associated Secretory
Phenotype by Data-Independent Acquisition................................................................... 1
1.01. Introduction ...................................................................................................... 1
1.02. Strategic planning ............................................................................................ 4
1.03. Basic protocol 1 ............................................................................................... 5
1.04. Alternate protocol 1A........................................................................................ 8
1.05. Alternate protocol 1B........................................................................................ 9
1.06. Alternate protocol 1C ..................................................................................... 11
1.07. Alternate protocol 1D ..................................................................................... 13
1.08. Support protocol 1.......................................................................................... 15
1.09. Basic protocol 2 ............................................................................................. 24
1.10. Alternate protocol 2 ........................................................................................ 26
1.11. Basic protocol 3.............................................................................................. 28
1.12. Reagents and solutions ................................................................................. 37
1.13. Commentary .................................................................................................. 37
1.14. Conclusion ..................................................................................................... 44
Chapter 2: A Fully-Automated Senescence Test (FAST) for the High-Throughput
Quantification of Senescence-Associated Markers ....................................................... 45
2.1. Introduction ...................................................................................................... 45
2.2. Methods ........................................................................................................... 47
2.3. Results............................................................................................................. 54
2.4. Discussion........................................................................................................ 77
Chapter 3: High-Content Image Analysis Identifies DNA Content As A Key Contributor
To Senescent Cell Heterogeneity And A Determinant Of Senolytic Response .............. 83
3.1. Introduction ...................................................................................................... 83
3.2. Methods ........................................................................................................... 85
3.3. Results............................................................................................................. 90
3.4. Discussion...................................................................................................... 100
Bibliography ................................................................................................................ 104

iii
List of Figures
Figure 1. SASP analysis workflow.................................................................................. 3
Figure 2. EdU incorporation. ........................................................................................ 19
Figure 3. SA-β-Gal staining.......................................................................................... 21
Figure 4. Gene expression analysis using qPCR. ........................................................ 22
Figure 5. Propidium iodide (PI) staining for cell death.................................................. 24
Figure 6. SASP proteomic analysis figures generated by the Spectronaut software.... 36
Figure 7. FAST workflow. ............................................................................................. 56
Figure 8. Standardization of detection.......................................................................... 59
Figure 9. Additional graphs generated with FAST ........................................................ 60
Figure 10. FAST is compatible with different microscope setups. ................................ 61
Figure 11. FAST distinguishes senescent cell populations in experimental settings
with a high rate of false positives. ................................................................................. 64
Figure 12. Comparison of colorimetric X-Gal and fluorescent C12FDG live cell
staining with FAST......................................................................................................... 65
Figure 13. C12FDG SA-β-Gal redistributes inside and in between cells upon fixation
and permeabilization. .................................................................................................... 68
Figure 14. Benchmarking FAST with senescence induction......................................... 69
Figure 15. Benchmarking FAST with a senolytic compound. ....................................... 71
Figure 16. ABT263 treatment does not affect SA-β-Gal and EdU staining in
senescent microvascular endothelial cells. ................................................................... 73
Figure 17. Combination of markers using machine learning improves repeatability of
senescence detection.................................................................................................... 76

iv
Figure 18. High-content imaging workflow.................................................................... 91
Figure 19. Validation of senescence induction and population-level heterogeneity...... 94
Figure 20. G1 and G2 senescent cells express different levels of senescence
markers. ........................................................................................................................ 96
Figure 21. G1 and G2 enrichment protocol for senescent endothelial cells. ................ 97
Figure 22. G1 and G2 senescent endothelial cells show different levels of IL-6
secretion and ABT263 susceptibility............................................................................ 100
Figure 23. Senescent heterogeneity model based on DNA content. .......................... 102

v
Abstract
Cellular senescence is a complex stress response that induces an essentially
permanent cell cycle arrest, and a complex secretory phenotype termed the
senescence-associated secretory phenotype (SASP). Over the past few decades, a
substantial body of literature has established a potentially causal link between the
accumulation of senescent cells and several age-related diseases. Consequently,
significant efforts have been made to develop interventions that target cellular
senescence. However, more work is needed to create senotherapeutics that can
substantially improve human healthspan. Challenges in developing such therapies
include the difficulty of detecting and quantifying senescent cells and their significant
heterogeneity. Novel methods for assessing senescence burden and a deeper
understanding of senescent cell heterogeneity are necessary to develop the next
generation of senotherapeutics. Therefore, my thesis work focused on developing and
optimizing workflows and tools for studying senescent cells and their heterogeneity by
using cell culture models.
In Chapter 1, a mass spectrometry (MS)-compatible protocol is described to i)
generate senescent cells using different stimuli, ii) collect conditioned media containing
proteins secreted by senescent cells (SASP), and iii) prepare the SASP for quantitative
proteomic analysis using data-independent acquisition (DIA) MS. This workflow can
help elucidate senescent cell heterogeneity by characterizing the SASP of different
senescent cell populations, as well as provide insights into aging and disease
mechanisms related to senescence.

vi
In Chapter 2, the Fully-Automated Senescence Test (FAST) is described, an
image-based method for the high-throughput, single-cell assessment of senescence in
cultured cells. FAST quantifies three of the most widely adopted senescenceassociated markers for each cell imaged: senescence-associated β-galactosidase
activity (SA-β-Gal) using X-Gal, proliferation arrest via lack of 5-ethynyl-2’-deoxyuridine
(EdU) incorporation, and enlarged morphology via increased nuclear area. Additionally,
proof of concept is provided that FAST is suitable for screening compounds that modify
senescence burden. This novel method enables rapid, unbiased, and user-friendly
quantification of senescence burden in culture, as well as facilitating large-scale
experiments that were previously impractical.
Finally, Chapter 3 describes a study where high-content image analysis was
employed to identify functionally distinct senescent subpopulations. It was found that
G2-arrested senescent cells have higher senescence marker expression than G1-
arrested cells. Additionally, it was found that G2-arrested senescent cells secrete more
IL-6 – a pro-inflammatory cytokine part of the SASP – and are more sensitive to the
senolytic ABT263 than G1-arrested cells. Thus, this study demonstrates the existence
of functionally distinct senescent subpopulations in culture and points to DNA content as
a key contributor to the heterogeneity within senescent populations.
Overall, my thesis work provided new workflows and tools for studying senescent
cells and their heterogeneity, both between as well as within senescent cell populations.
Such tools generated insights into the factors that contribute to senescent cell
heterogeneity and will be broadly useful to the aging field by enabling further
characterization of cellular senescence and its heterogeneity.

1
Chapter 1: Quantitative Proteomic Analysis of the SenescenceAssociated Secretory Phenotype by Data-Independent Acquisition
1.01. Introduction
Cellular senescence is a complex stress response that leads to an exit from the
cell cycle that is essentially permanent, although it can in some cases be overcome by
genetic or pharmacological manipulation in culture1,2
. It is not known whether
senescence reversal occurs in vivo. Most dividing cell types can become senescent,
and subsequently influence both physiological and pathological processes3,4{Basisty,
2019, A Proteomic Atlas of Senescence-Associated Secretomes for Aging Biomarker
Development}. Mouse models and human cell and tissue experiments show that
senescent cells can contribute to frailty and many age-related conditions, including type
II diabetes, pulmonary fibrosis, and cancer, making cellular senescence a key driver of
aging5,6
. Many biological effects (both beneficial and detrimental) that senescent cells
exert on organisms are triggered by their secretome, termed Senescence-Associated
Secretory Phenotype (SASP)7
.
The SASP is a complex mixture of biologically active molecules, including lipids,
metabolites, and proteins7–9
. Among the SASP factors are inflammatory cytokines,
growth factors, hormones, proteases, hemostasis factors10
, and many other still
unknown components. Interestingly, senescent cells release not only soluble
metabolites, lipids, and proteins, but also extracellular vesicles, including
exosomes8,11,12
. Overall, the SASP is extremely heterogeneous and varies depending
on cell type and the senescence-inducing stimulus8,13
.

2
To add even more complexity, the SASP is very dynamic and changes
significantly with time after the initial senescence induction8,14
. With newer technological
developments we are now able to use proteomic unbiased discovery workflows to gain
deeper insights into highly complex SASP profiles15 and are only beginning to
appreciate their true heterogeneity. A better characterization of the SASP is of
importance, as it will allow us to better understand the biological effects of cellular
senescence, as well as offer new potential biomarkers and therapeutic targets that
could be key tools in the fight against age-related diseases.
Given the vast number and heterogeneity of protein components in the SASP,
quantitative and comprehensive mass spectrometry-based proteomic approaches, such
as data-independent acquisition (DIA) mass spectrometry (MS), are excellent
techniques for unbiased profiling of the SASP. This protocol will guide researchers in
the generation and validation of cultured senescent cells, collection, and processing of
conditioned medium containing the SASP, and discovery and quantification of SASP
proteins using a comprehensive DIA-MS-based acquisition workflow16–18
. These SASP
isolation protocols are also compatible with other downstream quantitative mass
spectrometric methods, such as data-dependent acquisition (DDA) MS.
See Figure 1 for a schematic of the protocols. Basic Protocol 1 and its Alternate
Protocols 1A-D describe in detail the preparation of senescent cells and appropriate
control cells using some of the most common senescence inducers: ionizing radiation
(Basic Protocol 1), the chemotherapeutic drug doxorubicin (Alternate Protocol 1A),
oncogenic RAS overexpression (Alternate Protocol 1B), mitochondrial disruption

3
(Alternate Protocol 1C) and the HIV protease inhibitor atazanavir (Alternate Protocol
1D).
Figure 1. SASP analysis workflow.
Protocol schematics for senescence induction (A, B), collection of conditioned media (C),
and proteomics analysis (D).

4
Support Protocol 1 describes assays to check for the presence of senescenceassociated markers and confirm successful senescence induction.
Basic Protocol 2 and Alternate Protocol 2 describe the procedure for generating
and collecting conditioned media (CM) from senescent and quiescent control cells.
Basic Protocol 3 describes the proteomic workflow that we recommend for
identifying and quantifying the SASP proteins secreted by senescent cells in culture,
including sample processing, mass spectrometric acquisition by DIA-MS, and analysis
of raw mass spectrometric data files for protein identification and quantification.
1.02. Strategic planning
Each researcher will choose a specific senescence inducer and version of
Protocol 1 to use based on the best models for the biological system and context under
investigation. For example, if the objective is to identify SASP factors involved in the
side-effects of chemotherapy, then the doxorubicin-induced senescence protocol below
(Alternate Protocol 1A) may be the most appropriate approach. Another important
factor in deciding which senescence inducer to use is the availability of
materials/instrumentation to perform senescence induction. For example, senescence
induction using oncogenic RAS overexpression requires having cells harboring an
inducible vector such as pLVX-RASV12, while induction of senescence by ionizing
radiation requires access to an x-ray generator. The same rationale applies to the
decision about which cell type should be investigated. The selection of cell type will be
based on what is most relevant to the tissue or disease of interest.
It should be noted that the Protocols outlined here are optimized for IMR-90
human fetal lung fibroblasts cultured at physiological O2 concentration (3% O2, 10%

5
CO2 at 37° C), and the inducer doses and culture conditions should be individually
tailored and optimized for other cell types. It should also be pointed out that the
approaches described in Basic Protocol 1 and Alternate Protocols 1A-D are not the only
means of senescence induction.
To generate SASP-containing and quiescent control conditioned media (CM),
Basic Protocol 2 should be followed. If after senescence induction the cell type
investigated does not remain viable in low-serum, then Alternate Protocol 2 should be
used. Also, when using any combination of cell type and senescence induction for the
first time, successful induction of senescence should be verified. Support Protocol 1
describes several assays that are commonly used to confirm senescence induction.
Please note that other assays, such as assessing the lack of proliferation through Ki67
staining or detecting increased lysosomal activity by staining for lipofuscin granules,
could be used to confirm the presence of senescent cells19
. However, the assays listed
in Support Protocol 1 are widely used, robust, and easy to perform.
1.03. Basic protocol 1
Generating ionizing radiation-induced senescent and control cells
The method described here allows for the generation of senescent cells by Xrays (ionizing radiation; IR) and appropriate control cells. From these two cell
populations, the researcher will later generate the SASP-containing and control CM
(see Basic Protocol 2 / Alternate Protocol 2). Cells are seeded at the desired density,
cultured overnight, and irradiated the following day. After IR, media are replaced with
fresh complete media. Then, media are switched to fresh complete media every two
days. Three days before the irradiated cells reach a full senescent state, control cells

6
are seeded and cultured overnight. These control cells are derived from the same stock
culture used to seed the irradiated cells and have been cultured in parallel up to this
point. The next day, control cells are mock irradiated. This protocol has been optimized
for IMR-90 primary human fetal lung fibroblasts and can serve as a guide for the use of
other adherent human cells.
Materials
• Desired cell type
o Such as human primary fibroblasts strain IMR-90 (ATCC, cat. no. CCL186), WI-38 (ATCC, cat. no. CCL-75), or BJ (ATCC, cat. no. CRL-2522)
• Complete medium appropriate for the cell type used
o When using IMR-90 fibroblasts, the complete medium composition is the
following: DMEM (Gibco, cat. no. 12430-054) supplemented with 100
U/mL penicillin-streptomycin (Gibco, cat. no. 15070063) and 10% fetal
bovine serum (Gibco, cat. no. 2614079)
• Serum-free, phenol red-free media
o When culturing IMR-90 fibroblasts, use phenol red-free DMEM (Gibco,
cat. no. 21063-029) supplemented with 100 U/mL penicillin-streptomycin
(Gibco, cat. no. 15070063) to avoid interferences for subsequent protein
concentration measurements, such as BCA assays.
• Dulbecco’s phosphate-buffered saline (PBS) (Gibco, cat. no. 21600-010)
• Trypsin-EDTA (Corning, cat. no. 25-051-CI)
• Incubator set at the optimal conditions of temperature and partial pressures of
air gases to mimic the physiological requirement of the cell type used
o IMR-90 cells are cultured at 37° C, 3% O2, 10% CO2 in HERA CELL 240i
CO2 incubators (ThermoFisher, cat. no. 51026331)
• T175 Tissue culture flasks (Genesee Scientific, cat. no. 25-211)
• X-ray generator (X-Ray associates LLC, Polaris 320 kV X-ray Generator set)
Generation of senescent cells

7
1. Seed 3 T175 flasks with 2 x 106
fibroblasts each (or another cell type of
interest), using complete DMEM medium (see “Materials”); culture overnight.
These cells will be used to generate SASP-containing CM. Seed (at least) 3
replicates per condition to ensure that changes observed in the MS data can
be statistically analyzed.
o The number of cells needed to obtain enough SASP protein amount for
MS analysis may vary depending on the cell type and MS instrumentation
available. Ideally, the total amount of secreted proteins for each replicate
should be above 50 μg. If less protein is available, the protocol should be
optimized for smaller protein amounts and tested. To estimate the level of
protein secretion, conduct a pilot experiment by seeding the desired
number of cells, allowing them to recover overnight, then wash them twice
with PBS, add serum-free, phenol red-free media, collect CM 24 h later,
and perform a Bicinchoninic Acid (BCA) assay. This quantification will
allow an estimation of the number of cells needed to obtain enough SASP
proteins for MS analysis. Culture media used during this step must be
free of serum and, as much as possible, of protein
components/contaminants. This way, the protein concentration measured
will reflect only the proteins secreted by the cells. Also, the CM needs to
be phenol red-free because this compound interferes with the
quantification of proteins using BCA.
2. Treat fibroblasts the next day with 15 Gy IR to induce senescence and then
replace the media with fresh complete media.
o Fibroblasts will develop a full senescent phenotype approximately 10 days
later. By then, almost all fibroblasts should develop the enlarged, flattened
morphology associated with senescence. The replacement of media
should not wash off cells as senescent fibroblasts tend to stick to plates
more firmly than their control counterpart. The radiation dose and time
needed may vary depending on the cell type.
3. Replace media with fresh complete media every 2 days until day 6 (including
on day 6) after IR treatment. On day 8, proceed with the “Generation of CM
samples” step in Basic Protocol 2 (or Alternate Protocol 2) to prepare the cells
for later CM collection.
Generation of control cells
4. Seed flasks with control fibroblasts 7 days after IR treatment of experimental
cells (that is, 3 days before the IR-treated fibroblasts reach a full senescent
phenotype) and allow them to recover overnight. These cells will be used to
generate control CM.
5. On the next day, mock irradiate the control cells by placing them into the Xray chamber without irradiating them. Leave control cells in the X-ray

8
chamber for as long as it took to give the irradiated cells their full dose of
radiation.
6. After mock irradiation, proceed to Basic Protocol 2 (or Alternate Protocol 2).
1.04. Alternate protocol 1A
Generating doxorubicin-induced senescent and control cells
This method allows for the generation of senescent cells upon treatment with the
chemotherapeutic drug doxorubicin and appropriate control cells. From these two cell
populations, the researcher will later generate the SASP-containing and control CM
(see Basic Protocol 2 / Alternate Protocol 2). Cells are seeded at the desired density,
cultured overnight, and treated with doxorubicin for 24 h. Thereafter, cells are washed
with PBS, fresh media is added, and cells are cultured until the development of a full
senescent phenotype. Four days before the doxorubicin-treated cells reach full
senescence, the control cells are seeded and cultured overnight. These control cells
are derived from the same stock culture used to seed the doxorubicin-treated cells and
have been cultured in parallel up to this point. The next day, control cells are treated
with vehicle (DMSO) for 24 h. This protocol has been optimized for IMR-90 primary
human fetal lung fibroblasts and can serve as a guide for the use of other adherent
human cells.
Materials (see also Basic Protocol 1)
• Doxorubicin stock solution (2.5 mM in DMSO) (see recipe details in the
“Reagent and Solutions” section)
• Dimethyl sulfoxide (DMSO) (MilliporeSigma, cat. no. 67-68-5)
Generation of senescent cells
7. Seed 3 T175 flasks with 2 x 106
fibroblasts each (or another cell type) using
the appropriate complete medium (see “Materials”), and culture overnight.

9
These cells will be used to generate SASP-containing CM. Seed (at least) 3
replicates to ensure changes observed in the MS data can be statistically
analyzed.
o The number of cells needed to obtain enough SASP proteins for MS
analysis may vary depending on the cell type and MS instrumentation
available. To verify that protein concentrations are sufficient for
downstream MS analyses, see step 1 comment in Basic Protocol 1.
8. On the next day, dilute the doxorubicin stock solution (2.5 mM doxorubicin in
DMSO) in complete media for a final concentration of 250 nM doxorubicin
(dilution 1:10,000).
9. Remove culture media from the flasks and then switch to doxorubicincontaining media. Incubate for 24 h to induce senescence.
o Fibroblasts will develop a full senescent phenotype 10 days later. By then,
almost all fibroblasts should develop the enlarged, flattened morphology
associated with senescence. The dose of doxorubicin and time needed
may vary depending on the cell type.
10.After doxorubicin treatment, wash cells twice with 25 mL PBS (same volume
as doxorubicin-containing media fed to the cells) and add fresh complete
media. Then replace media every 2 days until day 6 (included) after the
beginning of doxorubicin treatment. On day 8, proceed with the “Generation
of CM samples” step in Basic Protocol 2 (or Alternate Protocol 2) to prepare
the cells for subsequent CM collection.
Generation of control cells
11.Seed new flasks with fibroblasts 6 days after doxorubicin treatment (that is, 4
days before the doxorubicin-treated cells reach senescence) and allow the
cells to recover overnight. These cells will be used to generate quiescent
control CM.
12.On the next day, treat the control cells with vehicle (DMSO) for 24 h.
13.After treatment with vehicle DMSO, proceed to Basic Protocol 2 (or Alternate
Protocol 2) to prepare the cells for subsequent CM collection.
1.05. Alternate protocol 1B
Generating oncogenic RAS-induced senescent cells and control cells
The method described here allows for the generation of senescent cells via
oncogenic RAS overexpression and appropriate control cells. From these cell

10
populations, the researcher will later generate SASP-containing and control CM (see
Basic Protocol 2 / Alternate Protocol 2). Cells transduced with a pLVX-RASV12 vector or
GFP control vector are seeded at the desired density, allowed to recover overnight, and
then cultured with doxycycline to induce RAS or GFP overexpression. Media are
changed every 2 days with fresh complete media containing doxycycline. Three days
before the RAS-overexpressing cells reach a full senescent phenotype, GFPoverexpressing control cells are re-seeded and cultured overnight. Until then, it is
important to passage the control cells when appropriate (after reaching 80-90%
confluency) since these cells keep proliferating. This protocol has been optimized for
IMR-90 primary human fetal lung fibroblasts and can serve as a guide for the use of
other adherent human cells.
Materials (see also Basic Protocol 1)
• Desired cell type transduced with the inducible vector pLVX (Lenti-X™ Tet
On®Advanced Inducible Expression System; Takara Bio, cat. no. 632162)
expressing either a constitutively active RASV12 oncogene or GFP8
.
o For example, IMR-90 primary human fetal lung fibroblasts (ATCC, cat. no.
CCL¬186) transduced with pLVX-RASV12 or pLVX-GFP8
• Tetracycline-free FBS (Tet-free FBS) (Takara Bio, cat. no. 631105)
• Doxycycline hyclate (MilliporeSigma, cat. no. D9891-5G)
Generation of senescent and control cells
14.Seed 3 T175 flasks with 1.5 x 106
fibroblasts transduced with pLVX-RASV12
(RAS cells), and 3 T175 flasks with 1.5 x 106 fibroblasts transduced with
pLVX-GFP (GFP cells). Use the appropriate complete medium (see
“Materials”) and allow them to recover overnight. Use media containing Tetfree FBS (see “Materials”) to ensure that RASV12 or GFP is not expressed
until the addition of doxycycline. These cell populations will be used to
generate SASP-containing or control CM, respectively. Seed (at least) 3
replicates for each condition to ensure changes observed in the MS data can
be statistically analyzed.

11
o The number of cells needed to obtain enough SASP proteins for MS
analysis may vary depending on the cell type and MS instrumentation
available. To verify that protein secretion is sufficient for downstream MS
analyses, see step 1 comment in Basic Protocol 1.
15.The next day, treat the two fibroblast populations with 1 μg/mL doxycycline,
which will trigger senescence in RAS cells while serving as a control
treatment for GFP cells.
o Fibroblasts will develop a full senescent phenotype approximately 7 days
later. By then, almost all fibroblasts should develop the enlarged, flattened
morphology associated with senescence. The time needed may vary
depending on the cell type used.
16.Replace media for both cell populations with fresh culture media containing
doxycycline every 2 days until day 4 (included) after the beginning of
treatment. On day 5, proceed with the “Generation of CM samples” step in
Alternate Protocol 2 (or in Basic Protocol 2) to prepare the cells for
subsequent CM collection. Note that RAS cells will undergo a hyperproliferative phase before undergoing growth arrest. Importantly, IMR-90
fibroblasts induced to senescence by oncogenic RAS overexpression are not
viable in low-serum. Therefore, after completing Alternate Protocol 1B, we
suggest using Alternate Protocol 2 (rather than Basic Protocol 2) when using
this method of senescence induction in fibroblasts.
o Note that when instructed to feed cells with fresh media in Alternate
Protocol 2 (or Basic Protocol 2), such media must be supplemented with
doxycycline. Only serum-free media given to both cell populations 24 h
before CM collection (“Collection of CM” section) should be devoid of
doxycycline.
1.06. Alternate protocol 1C
Generating mitochondrial dysfunction-induced senescent and control
cells
The method described here allows for generating senescent cells by
mitochondrial dysfunction (mitochondrial dysfunction-associated senescence or MiDAS)
and appropriate control cells. From these cell populations, the researcher will later
generate the SASP-containing and control CM (see Basic Protocol 2 / Alternate Protocol
2). Cells are seeded at the desired density, cultured overnight, and treated with either

12
the mitochondrial complex III inhibitor antimycin A or vehicle DMSO. Media are
replaced every 2 days with fresh complete media containing either antimycin A or
DMSO. Three days before the antimycin A-treated cells reach a full senescent
phenotype, DMSO-treated cells are re-seeded and cultured overnight. Until then, it is
important to passage the DMSO-treated cells when appropriate (after reaching 80-90%
confluency) since these cells keep proliferating. This protocol has been optimized for
IMR-90 primary human fetal lung fibroblasts and can serve as a guide for other types of
adherent human cells.
Materials (see also Basic Protocol 1)
• Antimycin A from Streptomyces species. (Sigma Aldrich, cat. no. A8674-
25MG)
• Dimethyl sulfoxide (DMSO) (MilliporeSigma, cat. no. 67-68-5)
Generation of senescent and control cells
17.Seed 6 T175 flasks (3 for antimycin A treatment, 3 for DMSO treatment) with
2 x 106
fibroblasts each using an appropriate complete medium (see
“Materials”), and culture overnight. These cells will be used to generate
SASP-containing or control CM. Seed (at least) 3 replicates for each
condition to ensure that changes observed in the MS data can be statistically
analyzed.
o The number of cells needed to obtain enough SASP proteins for MS
analysis may vary depending on the cell type and MS instrumentation
available. To verify that protein secretion is sufficient for downstream MS
analyses, see step 1 comment in Basic Protocol 1.
18.The next day, treat fibroblasts with 250 nM antimycin A or vehicle DMSO.
Cells treated with antimycin A will become senescent and generate the SASPcontaining CM. Cells treated with DMSO will be used as quiescent controls to
generate the control CM.
o Fibroblasts will develop a full senescent phenotype approximately 10 days
later. By then, almost all fibroblasts should develop the enlarged, flattened
morphology associated with senescence. The concentration of antimycin
A and time needed may vary depending on the cell type used.

13
19.Replace media for both cell populations with fresh culture media containing
antimycin A or DMSO every 2 days until day 6 (included) after the beginning
of treatment. On day 8, proceed to the “Generation of CM samples” step in
Basic Protocol 2 (or in Alternate Protocol 2) to prepare the cells for
subsequent CM collection. Note that DMSO-treated cells will keep
proliferating, so they must be passaged when appropriate (after reaching
80%- 90% confluency).
20.To generate control cells, re-seed proliferating DMSO-treated control cells 7
days after the beginning of vehicle treatment (that is, 3 days before antimycin
A-treated cells reach a full senescent phenotype). By doing so, it will be
possible to obtain similar numbers of senescent and control cells when
collecting their CM.
21.Allow the control DMSO-treated cells to recover overnight and then, on the
next day, proceed to Basic Protocol 2 (or Alternate Protocol 2) to prepare the
cells for subsequent CM collection.
o Note that when instructed to feed cells with fresh media in Basic Protocol
2 (or Alternate Protocol 2), media must be supplemented with 250 nM
antimycin A or vehicle DMSO. Only serum-free media given to both cell
populations 24 h before CM collection (“Collection of CM” section) should
be devoid of antimycin A or DMSO.
1.07. Alternate protocol 1D
Generating atazanavir/ritonavir-induced senescent and control cells
The method described here allows for the generation of senescent cells upon
treatment with the anti-HIV drugs atazanavir and ritonavir (ATV/r), as well as appropriate
control cells. From these cell populations, the researcher can generate SASPcontaining and control CM (see Basic Protocol 2 / Alternate Protocol 2). Cells are
seeded at the desired density, cultured overnight, and treated with either ATV/r or
vehicle (DMSO). Media are changed every 2 days with fresh complete media
containing either ATV/r or DMSO. Three days before the ATV/r-treated cells reach a full
senescent phenotype, DMSO-treated cells are re-seeded and cultured overnight. Until
then, it is important to passage the DMSO-treated cells when appropriate (after

14
reaching 80-90% confluency) since these cells keep proliferating. This protocol has
been optimized for IMR-90 primary human fetal lung fibroblasts and can serve as a
guide for the use of other adherent human cells.
Materials (see also Basic Protocol 1)
• Atazanavir/ritonavir solution (ATV/r) (see recipe details in the “Reagent and
Solutions” section)
• Dimethyl sulfoxide (DMSO) (MilliporeSigma, cat. no. 67-68-5)
Generation of senescent and control cells
22.Seed 6 T175 flasks (3 for ATV/r treatment, 3 for DMSO treatment) with 2 x 106
fibroblasts each using the appropriate complete medium (see “Materials”),
and culture them overnight. These cells will be used to generate SASPcontaining or control CM. Seed (at least) 3 replicates for each condition to
ensure that changes observed in the MS data can be statistically analyzed.
o The number of cells needed to obtain enough SASP proteins for MS
analysis may vary depending on the cell type and MS instrumentation
available. To verify that protein secretion is sufficient for downstream MS
analyses, see step 1 comment in Basic Protocol 1
23.On the next day, treat fibroblasts with 25 μM ATV/r or vehicle DMSO. Cells
treated with ATV/r will become senescent and generate SASP-containing CM.
Cells treated with DMSO will be used as quiescent controls to generate the
control CM.
o Fibroblasts will develop a full senescent phenotype approximately 14 days
later. By then, almost all fibroblasts should develop the enlarged, flattened
morphology associated with senescence. The concentration of ATV/r and
time needed may vary depending on the cell type.
24.Replace media for both cell populations with fresh complete media containing
either ATV/r or DMSO every 2 days until day 10 (included), and on day 12
proceed with the “Generation of CM samples” step in Basic Protocol 2 (or in
Alternate Protocol 2) to prepare the cells for subsequent CM collection. Note
that DMSO-treated cells will keep proliferating, so they will have to be
passaged when appropriate (after reaching 80%-90% confluency).
25.Re-seed the proliferating DMSO-treated control cells 11 days after the
beginning of vehicle treatment (that is, 3 days before ATV/r-treated cells reach
full senescence). By doing so, it will be possible to obtain similar numbers of
senescent and control cells when collecting CM.

15
26.Allow the control DMSO-treated cells to recover overnight and then proceed
to Basic Protocol 2 (or Alternate Protocol 2) to prepare the cells for
subsequent CM collection.
o Note that when instructed to feed cells with fresh media in Basic Protocol
2 (or Alternate Protocol 2), such media must be supplemented with ATV/r
or DMSO for the culture of senescent or control cells, respectively. Only
serum-free media given to both cell populations 24 h before CM collection
(“Collection of CM” section) will be devoid of ATV/r and DMSO.
1.08. Support protocol 1
A multiple-assays approach to confirm the phenotype of senescent
cells
This protocol describes multiple assays to verify that the senescence induction
performed in Basic Protocol 1 (or Alternate Protocol 1 versions) has been successful.
Supporting Protocol 1 is used to prove that the secreted proteins to be analyzed by MS
are produced by senescent cells and therefore part of a SASP. This supporting protocol
should be performed when the researcher uses any combination of a senescence
inducer (Basic Protocol 1 or Alternate Protocol 1 versions) and a desired cell type.
These assays are not compatible with the generation of CM described in Basic Protocol
2 (or Alternate Protocol 2) because they require fewer cells compared to the number
needed to analyze secreted proteins by MS. Therefore, cells should be seeded in the
appropriate tissue culture dishes (see corresponding assay protocol below for details) to
perform the assays described, and then the chosen senescence induction method
(Basic Protocol 1 or Alternate Protocol 1 versions) is performed. Once senescent and
control cells are generated, the steps described below must be performed for each
assay. The assays we recommend are used to verify reduced cell proliferation (see
“EdU incorporation”), upregulation of senescence-associated β-galactosidase (SA-β-

16
Gal) activity (see “Senescence-associated β-galactosidase”), and a gene transcription
signature associated with senescence (see “Senescence-associated gene expression
analysis by qPCR”). We also recommend performing a viability assay (see “Propidium
iodide inclusion”) to verify that both senescent and control cells are healthy before
generating CM for MS analysis.
Materials (see also Basic Protocol 1 and its Alternate Protocol 1 versions)
• Reverse Transcriptase (Thermo Fisher Scientific, cat. no. 43-112-35)
• Serum-free media
o For IMR-90 fibroblasts, use DMEM (Gibco, cat. no. 12430-054)
supplemented with 100 U/mL penicillin-streptomycin (Gibco, cat. no.
15070063)
• Click-iT EdU Kit Alexa Fluor 488 HCS Assay (Thermo Fisher Scientific, cat.
no. C10351)
• Senescence Detection Kit (BioVision, cat. no. K320-250)
• X-Gal (Life Technologies, cat. no. 15520-018)
• ISOLATE II RNA Micro Kit (Bioline, cat. no. BIO-52075)
• Senescence-associated human genes-specific primers (Eurofins)
o CDKN1A (p21)
▪ Forward: TCACTGTCTTGTACCCTTGTGC
▪ Reverse: GGCGTTTGGAGTGGTAGAAA
o CDKN2A (p16)
▪ Forward: GAGCAGCATGGAGCCTTC
▪ Reverse: CGTAACTATTCGGTGCGTTG
o CDKN2B (p15)
▪ Forward: CTCCCGAAACGGTTGACTC
▪ Reverse: GCGGGGACTAGTGGAGAAG
o CXCL1

17
▪ Forward: GCTGAACAGTGACAAATCCAAC
▪ Reverse: CTTCAGGAACAGCCACCAGT
o CXCL10
▪ Forward: GAAAGCAGTTAGCAAGGAAAGGT
▪ Reverse: GACATATACTCCATGTAGGGAAGTGA
o IL1B
▪ Forward: CTGTCCTGCGTGTTGAAAGA
▪ Reverse: TTGGGTAATTTTTGGGATCTACA
o IL6
▪ Forward: GCCCAGCTATGAACTCCTTCT
▪ Reverse: GAAGGCAGCAGGCAACAC
o MMP3
▪ Forward: CAAAACATATTTCTTTGTAGAGGACAA
▪ Reverse: TTCAGCTATTTGCTTGGGAAA
o SERPINE1
▪ Forward: CCAGCTGACAACAGGAGGAG
▪ Reverse: CCCATGAGCTCCTTGTACAGAT
o LAMNB1
▪ Forward: TTGGATGCTCTTGGGGTTC
▪ Reverse: AAGCAGCTGGAGTGGTTGTT
• Propidium iodide staining solution (eBioscience, cat. no. 00-6990-50)
• 8-well chamber slides Lab-Tek II (Sigma-Aldrich, cat. no. C7057-1CS)
• Inverted fluorescence microscope (Olympus, IX70 Fluorescence Microscope)
• 6-well cell culture plates (Genesee Scientific, 25-105)
• T75 Tissue culture flasks (Genesee Scientific, cat. no. 25-211)
• LightCycler® 480 Instrument II (Roche, cat. no. 05015278001)
• Flow cytometer (BD™ LSR II with BD FACS Diva v8.0.2 acquisition software)
EdU incorporation

18
27.Follow the steps listed in Basic protocol 1 (or in one of Alternate Protocol 1
versions) to generate senescent and control cells of the chosen cell type. For
this immunostaining assay, seed 10,000 fibroblasts per well in 8-well chamber
slides for each replicate.
28.Wash senescent and control cells twice with PBS, then switch them to media
containing 0.2% FBS (instead of 10% FBS) and culture for 48 h to induce
quiescence by serum starvation in control cells while maintaining viability for a
few days for both cell populations.
29.The following day (24 h after switching to low serum-containing media), seed
cells in new wells using complete media and allow recovery overnight. These
cells will be used as a positive control for EdU staining (proliferating control
cells).
o Use complete media (containing 10% FBS when using fibroblasts) to keep
cells in a proliferative state.
30.The next day, remove half the volume of media in each well and replace it
with fresh media containing EdU (Click-iT EdU kit) as instructed by the
manufacturer’s protocol. Use serum-free media for the senescent and
quiescent cells. Use complete media (10% FBS) for the proliferating control
cells.
o For example, if each well of the 8-well chamber slide contains 250 μL of
media, remove 125 μL and add 125 μL of the appropriate media (serumfree or complete media) containing EdU (see manufacturer’s protocol for
the EdU concentration).
31.Incubate for 24 h.
32.24 h later, follow the instructions (manufacturer’s protocol) to fix,
permeabilize, detect EdU incorporation, and stain cellular DNA with Hoechst
33342.
o EdU staining is also compatible with DAPI DNA staining.
33.Remove all liquid from the wells, remove the well chambers, add mounting
media, and then apply coverslips onto the slides (avoid air bubbles).
34.Compare DNA staining with EdU staining to confirm that senescent and
quiescent control populations have low numbers of EdU-positive cells
(whereas proliferating control cells should have high numbers of EdU-positive
cells; see Figure 2).

19
Figure 2. EdU incorporation.
A) Representative images of quiescent control (top), ionizing radiation (IR)-induced
senescent (center), and proliferating control IMR-90 cells (bottom). DAPI staining is
shown on the left and EdU staining in the middle. Merged images are shown on the right.
B) Quantification of EdU-positive cells for quiescent control, IR-induced senescent, and
proliferating control cells. Data shown are means of 3 replicates ± SD.
Senescence-associated β-galactosidase
35.Follow the steps listed in Basic protocol 1 (or one of the Alternate Protocol 1
versions) to generate senescent and control cells of the chosen cell type. For
each replicate, seed 80,000 fibroblasts in a well of a 6-well plate.
36.Wash senescent and control cells twice with PBS, then switch to media
containing 0.2% FBS (instead of 10% FBS) and culture for 48 h. This will
induce quiescence by serum starvation in control cells while maintaining
viability for a few days in both cell populations.
37.After 48 h, wash the cells twice with PBS, then switch to serum-free media
and culture for 24 h.
38.Fix and incubate the cells with the X-Gal staining solution at 37° C overnight
(as instructed in the manufacturer’s protocol from the Senescence Detection
Kit).
o When preparing the staining solution as instructed by the manufacturer’s
protocol, we recommend using X-Gal from Life Technologies (see
“Materials”) rather than X-Gal provided with the Senescence Detection Kit

20
to optimize staining. Use the suggested solvent and concentration of XGal described in the Senescence Detection Kit manufacturer’s manual.
39.On the next day, remove the staining solution and leave the cells in PBS.
o The staining reaction might require incubations longer than the overnight
suggested in the manufacturer’s protocol. Cells should be monitored
under a microscope before removing the staining solution. If senescent
cells appear to be poorly stained, they can remain in the staining solution
at 37° C for longer times. Observe the cells under a microscope about
every 2 h until the senescent cells appear clearly stained, while quiescent
control cells should remain unstained. However, avoid extended
incubation times as the quiescent cells might turn positive. We
recommend acquiring images right after removing the staining solution
and adding PBS, even though cells can be stored at 4° C for 1-2 days
before imaging.
40.Acquire bright-field images using an inverted microscope and confirm that
senescent cells have high SA-β-Gal staining compared to quiescent control
cells (see Figure 3).

21
Figure 3. SA-β-Gal staining.
A) Representative images of quiescent control (top) and IR senescent IMR90 cells
(bottom). B) Quantification of SA-β-Gal positive cells for quiescent control and senescent
cells. Data shown are means of 3 replicates ± SD.
Senescence-associated gene expression analysis by qPCR
41.Follow the steps listed in Basic protocol 1 (or in one of the Alternate Protocol
1 versions) to generate senescent and control cells of the chosen cell type.
For each replicate, seed 200,000 fibroblasts in a well of a 6-well plate.
42.Wash both senescent and control cells twice with PBS, then switch them to
media containing 0.2% FBS (instead of 10% FBS) and culture for 48 h to
induce quiescence by serum starvation in control cells while maintaining
viability for a few days in both cell populations.
43.After 48 h, wash the cells twice with PBS, then switch them to serum-free
media and culture for 24 h.

22
44.24 h later, remove media from the wells and extract total RNA using the
ISOLATE II RNA Micro Kit. We recommend adding Lysis Buffer RLY directly
to the wells after removing the media, making sure to detach all cells with the
help of a cell scraper.
45.After RNA extraction, convert the RNA samples into cDNA using the Applied
Biosystems MultiScribe Reverse Transcriptase kit.
46.Use the cDNA to analyze by qPCR mRNA levels of the senescenceassociated genes.
o See “Materials” for a list of genes that are significantly upregulated (all
genes except LAMNB1) or downregulated (LAMNB1) upon senescence
induction20
. The primer sequences provided have been tested with the
Universal Probe Library System. If using another probe system, the
primer sequences listed are not guaranteed to work, and primer sequence
optimization might be needed.
47.Analyze qPCR data to compare the relative expression of senescenceassociated genes between senescent and quiescent control cells. Confirm
that most genes have the same upregulation or downregulation trends shown
in Figure 4.
o Different cell types may express different complements of SASP factors.
The senescence-associated genes listed here are optimized for
fibroblasts.
Figure 4. Gene expression analysis using qPCR.
Relative mRNA levels of the indicated genes in senescent cells normalized to quiescent
IMR-90 cells. Data shown are means of 3 replicates ± SD.

23
Propidium iodide inclusion
48.Follow the steps listed in Basic protocol 1 (or Alternate Protocol 1 versions) to
generate senescent and control cells of the chosen cell type. Seed 1 x 106
fibroblasts in a T75 flask for each replicate.
49.Wash senescent and control cells twice with PBS, then switch to media
containing 0.2% FBS (instead of 10% FBS) and culture for 48 h. This will
induce quiescence by serum starvation in control cells while maintaining
viability for a few days in both cell populations
50.After 48 h, wash the cells twice with PBS, then switch to serum-free media
and culture them for 24 h.
51.Wash the cells with PBS and detach them from the substratum by incubation
with Trypsin/EDTA at 37°C for 5 min.
52.Resuspend the cells in complete media to inhibit the trypsin, then wash the
cell suspensions twice with PBS.
o To perform a PBS wash, pellet the cells by centrifugation at 250 x g for 5
min, remove the supernatant, resuspend the pellet in PBS, centrifuge the
cells again at 250 x g for 5 min, and remove the PBS.
53.Resuspend the cell pellets in PBS to a final concentration of 106
-107 cells/μl.
54.Aliquot 100 μl of each cell suspension in a flow cytometry tube and add 5 μl
PI staining solution.
55.Mix the solution by pipetting and incubate for 1 min at room temperature.
56.Analyze the samples for PI staining by flow cytometry. Confirm that the
number of PI-positive cells is low in both senescent and control cells (see
Figure 5).
o PI dye is impermeable to cells, so will not stain cells unless their
membrane integrity is compromised (that is, the cells are dead or dying).

24
Figure 5. Propidium iodide (PI) staining for cell death.
Percentage of PI-positive cells in quiescent control and IR-induced senescent IMR-90 cell
populations. Data shown are representative of 3 replicates.
1.09. Basic protocol 2
Generating conditioned media from senescent cells cultured in low
serum and quiescent control cells
With this method, SASP-containing and control CM are generated from
senescent and control cells, prepared following Basic Protocol 1 (or an Alternate
Protocol 1 version). Both senescent and control cells are cultured in low-serum media
(0.2% FBS) for 48 h, which allows control cells to become quiescent. Then, cells are
cultured in phenol red-free, serum-free media for 24 h. CM from senescent and control
quiescent cells are collected, and the cells are counted, which will allow normalization
for proteomic analysis. Normalization to cell count is more meaningful as it normalizes
to ‘proteins secreted per cell’. This protocol has been optimized for IMR-90 primary

25
human fetal lung fibroblasts and can serve as a guide for the use of other adherent
human cells.
Materials (also see Basic Protocol 1)
• Senescent and control cells generated using Basic Protocol 1 or an Alternate
Protocol 1 version
• Low-serum media
o When using IMR-90 fibroblasts, the low-serum medium composition is the
following: DMEM (Gibco, cat. no. 12430-054) supplemented with 100
U/mL penicillin-streptomycin (Gibco, cat. no. 15070063) and 0.2% fetal
bovine serum (Gibco, cat. no. 2614079)
• Phenol red-free, serum-free media
o When using IMR90 fibroblasts, use phenol-red free DMEM (Gibco, cat. no.
21063-029) supplemented only with 100 U/mL penicillin-streptomycin
(Gibco, cat. no. 15070063)
• Beckman Coulter Z1 Particle Counter (Beckman, cat. no. 6605698)
• Swing Rotor Centrifuge 5804 (Eppendorf, cat. no. 022622501)
Generation of CM samples
57.Wash senescent and control cells twice with PBS, then add low-serum media
(see Materials) and culture for 48 h.
o Using low-serum media induces quiescence in control cells by serum
starvation, while maintaining viability for a few days in both cell
populations.
58.After 48 h, wash senescent and quiescent cells twice with PBS, then switch to
phenol red-free and serum-free media, and incubate for 24 h.
o The collected CM needs to be phenol red-free because this compound
interferes with the quantification of protein using BCA. Also, the culture
media used during this step must be free of serum and, as much as
possible, of protein components/contaminants. Abundant exogenous
protein contamination can limit the identification and quantification of
secreted proteins. High concentrations of proteins present in serum and
other cell culture supplements interfere with and suppress the ionization of
secreted proteins during MS analysis. If for some reason the culture

26
media contain protein components, these proteins must be excluded from
later proteomic analysis.
Collection of CM
59.After the 24-hour incubation, collect CM from each replicate and perform cell
counts.
o Right after collection, keep the CM on ice to prevent protein degradation
during cell counting. Also, cell counts are important to normalize the
levels of secreted proteins during proteomic analysis (see Basic Protocol
3). Senescent fibroblasts tend to attach more firmly to plates, so a slightly
longer incubation (1-2 min) with trypsin-EDTA might be required for the
senescent cells to detach.
60.Centrifuge the collected CM at 10,000 x g for 15 min to pellet cell debris and
transfer the supernatant into a new tube.
61.Proceed to Basic Protocol 3 (“Conditioned media processing and proteomic
analysis”) or store the samples at -80°C for later processing.
1.10. Alternate protocol 2
Generating CM from senescent cells cultured in complete media and
quiescent control cells
If after senescence induction the cell type investigated does not remain viable in
low-serum, then Alternate Protocol 2 should be used to generate CM for MS analysis.
Some indicators of poor viability are loss of cell number and cell detachment. Loss of
viability can be quantitatively confirmed by increasing cell death, as measured by cell
viability assays (for example, see “Propidium iodide inclusion” in Support Protocol 1). In
this protocol, CM containing the SASP are collected from senescent cells that are
cultured in complete media up to 24 h before CM collection. Control cells are still
cultured in low serum to induce quiescence. However, when comparing these two
conditions it is difficult to determine whether differences are due to a senescent vs nonsenescence state or due to culturing in complete media vs low-serum media. This

27
protocol is optimized for primary lung fibroblasts, which remain viable in low serum
under control conditions. If the cell type under investigation is not viable under control
conditions, we recommend optimization of culture conditions appropriate to the cell type
under investigation. A possible alternative approach may be to compare CM collected
from senescent cells cultured in complete media versus CM collected from nonsenescent cells cultured in complete media. However, under these conditions, one
cannot distinguish if changes in protein secretion are the result of comparing
proliferating cells versus non-proliferating cells, differences in cell density between
control and senescent conditions, or differences between senescent and non-senescent
cells. Also, see the section “Cell cultures in low-serum media before CM collection”
under “Critical Parameters and Troubleshooting”.
Materials
• See Basic Protocol 2 Materials section.
Generation of CM samples
62.Wash senescent and control cells twice with PBS. Then switch control cells
to low serum-containing media and add complete media to senescent cells.
Culture both cell populations for 48 h.
o Using low-serum media induces quiescence in control cells by serum
starvation, while maintaining viability for a few days.
63.After 48 h, remove culture media by aspiration, then wash senescent and
quiescent control cells twice by adding PBS and subsequently aspirate the
media/PBS for removal. After the washes, add serum-free and phenol-redfree media, and incubate for 24 h.
o The CM needs to be phenol red-free because this compound interferes
with the quantification of protein using BCA. Also, the culture media used
during this step must be free of serum and, as much as possible, of
protein components/contaminants. Abundant exogenous protein
contamination can limit the identification and quantification of secreted
proteins. High concentrations of proteins present in the serum and other

28
cell culture supplements interfere with and suppress the ionization of
secreted proteins during MS analysis. If for some reason the culture
media contain protein components, these proteins must be excluded from
later MS data analysis.
Collection of CM
64.For the collection of CM, follow steps 59 through 61 from Basic Protocol 2.
1.11. Basic protocol 3
Quantitative proteomic analysis of the SASP
This protocol describes a comprehensive unbiased mass spectrometry-based
approach to identify and quantify the secreted proteins of cultured cells. CM prepared
in Basic protocol 2 (or Alternate Protocol 2) are concentrated, digested, and desalted.
The processed samples are analyzed using a liquid chromatography-mass
spectrometry (LC-MS) method, specifically a data-independent acquisition (DIA)
workflow. The protocol illustrates in detail how to use an LC-MS system composed of a
nano-LC 2D HPLC coupled to a TripleTOF 6600 high-resolution mass spectrometer.
The DIA quantitative proteomics analysis software Spectronaut (Biognosys) is used to
perform relative quantification of protein levels and create reports of protein abundance,
fold changes, and statistics.
Materials
• SASP CM and control CM (CTL CM) generated in Basic Protocol 2 (or
Alternate Protocol 2)
• Ammonium bicarbonate (NH4HCO3) (SigmaAldrich, cat. no. A 6141)
• Dithiothreitol (DTT) (Sigma, cat. no. D9779)
• Iodoacetamide (IAA) (Sigma, cat. no. I1149)
• Sequencing grade trypsin (Promega, cat. no. V5113)

29
• Acetic acid (Sigma-Aldrich, cat. no. 695092-100ML)
• Formic acid (Sigma, cat. no. 94318)
• Acetonitrile (Burdick & Jackson, cat. no. AH015)
• Indexed retention time iRT peptide standards (Biognosys, cat. no. Ki-3002)
• Sequencing grade trypsin (Promega, cat. no. V5113)
• Amicon® Ultra centrifugal filters with 3 kDa molecular weight cutoff
(MilliporeSigma, cat. no. UFC9003)
• Bicinchoninic Acid (BCA) kit (Thermo Fisher Scientific, cat. no. PI23225)
• Litmus strips (VWR International, cat. no. EM1095350007)
• Heating Shaking Drybath (Thomas Scientific, cat. no. 1199A66)
• Benchtop microcentrifuge (ThermoFisher, cat. no. 75002401)
• Oasis HLB Solid-Phase Extraction (SPE) cartridges, 10 mg Sorbent per
Cartridge, 30 µm (Waters, cat. no. 186006339)
• SpeedVac concentrator (Thermo Scientific, Savant™ SPD131DDA)
• Ultrasonic Bath Sonicator (ThomasScientific)
• Autosampler vials for HPLC (Agilent Technologies, cat. no. 5190-3155)
• Nano-LC 2D HPLC system (Eksigent Ultra Plus, Eksigent)
• cHiPLC system (Eksigent)
• 200 µm x 0.4 mm ChromXP C18-CL chip, 3 µm, 120 Å (SCIEX)
• 75 µm x 15 cm ChromXP C18-CL chip, 3 µm, 120 Å (SCIEX)
• High-resolution mass spectrometer TripleTOF 6600 System (SCIEX) – or
other high-resolution mass spectrometer
• SpectronautTM software (Biognosys, Schlieren)
Concentrating the CM
65.Transfer CM to Amicon ultrafilter tubes, without exceeding the maximum
volume capacity (see manufacturer instructions).

30
o It is likely that the volumes of CM samples are higher than the ultrafilter
tubes’ capacity. If so, use multiple centrifugations until the volume of
every CM sample has been reduced to <0.5 mL (see next step).
66.Following manufacturer instructions, buffer exchange the protein samples into
50 mM NH4HCO3 buffer. After initially concentrating the protein samples in
the columns, we recommend performing one buffer exchange with a
maximum volume of 50 mM NH4HCO3, followed by a buffer exchange with a
maximum volume of 8 M urea in 50 mM NH4HCO3. After the addition of 8 M
urea in 50 mM NH4HCO3, centrifuge the samples until reaching a final volume
<0.5 mL in NH4HCO3 buffer.
67.Buffer exchanged samples can be stored long-term at -80°C.
In-solution proteolytic digestion
68.Quantify secreted protein concentrations by BCA assay.
o Use at least a 1:3 dilution of the sample in milliQ water for the BCA assay.
The BCA assay is compatible with urea concentrations up to 3M.
69.Aliquot 50 µg of each sample into new 1.5 mL tubes.
o If sample yields are lower than 50 µg, we recommend aliquoting at least
20 µg of secreted proteins due to sample losses during subsequent steps.
Still, sample amounts as low as 10 µg can be used.
70.Bring all samples to equal volumes with 8M urea in 50 mM NH4HCO3.
71.Vortex to mix.
72.Add DTT (1 M in 50 mM NH4HCO3 buffer stock solution) to a final
concentration of 20 mM (in 50 mM NH4HCO3 buffer) to each sample to
reduce disulfide bridges.
73.Incubate samples for 30 min at 37° C with shaking/agitation.
74.Allow samples to cool to room temperature.
75.Add IAA (200 mM in 50 mM NH4HCO3 buffer stock solution) to a final
concentration of 40 mM IAA (in 50 mM NH4HCO3 buffer) to each sample to
alkylate reduced thiols.
o It is important to use an IAA concentration at least double that of DTT to
ensure that all the reduced thiols are alkylated.
76.Incubate samples in the dark at room temperature for 30 min.

31
77.Dilute all samples 1:6 in 50 mM NH4HCO3 to reduce the concentration of urea
for trypsin digestion.
78.Verify that the pH of samples is between 7.0-8.5 by pipetting small volumes
(<1 µL) onto litmus strips. Adjust the pH as needed.
79.Resuspend lyophilized trypsin in 50 mM acetic acid to a final concentration of
0.1 mg/mL.
o For example, 20 μg of lyophilized trypsin are resuspended in 200 µL of
solution at pH 8, for a final concentration of 0.1 μg/μl trypsin.
80.Digest protein samples by adding trypsin at a protease-to-substrate protein
ratio of 1:50 (wt:wt). Incubate samples overnight at 37° C on a shaking dry
bath.
81.Quench the protein digestion by adding formic acid to a final concentration of
1% by volume from a 10% formic acid (in water) stock.
o Stock concentrations of formic acid less than or equal to 10% can be
pipetted safely with plastic pipette tips. Higher concentrations may
dissolve plasticware and should be handled with glass to avoid polymer
contamination of samples.
82.Spin sample at 5,000 x g for 15 min at room temperature to pellet insoluble
material. The supernatants, which contain peptides, are desalted in the next
steps.
Desalting samples
83.Samples are desalted using commercially available Solid-Phase Extraction
(SPE) cartridges. The next steps describe a desalting process using Oasis
HLB SPE cartridges.
84.Wet each HLB SPE cartridge twice with 800 µL 50% acetonitrile (ACN) in
0.2% formic acid in water.
85.Equilibrate each cartridge 3 times with 800 µL 0.2% formic acid in water.
86.Load peptide samples onto HLB SPE cartridges.
87.Wash each cartridge 3 times with 800 µL 0.2% formic acid in water.
88.Elute peptides once with 800 µL 50% ACN in 0.2% formic acid in water and
once with 400 µL 50% ACN in 0.2% formic acid in water in the same tube.
89.Dry samples completely in a SpeedVac.

32
o If desired, one can pause sample processing; dried peptide pellets can be
safely stored long-term at -80° C.
90.Resuspend dry pellets in 0.2% formic acid to a final concentration of 1 µg/µL.
o Calculate the final peptide concentration based on the initial protein
sample mass aliquoted for digestion. For example, if the initial aliquot
contained 50 μg of protein, the dried pellet will be resuspended in 50 µL of
0.2% formic acid.
91.Sonicate samples in a water bath sonicator for 5 min.
92.Vortex samples at 4° C for 10 min.
93.Centrifuge samples at 15,000 x g for 15 min.
94.Transfer supernatants to autosampler vials.
95.Add retention time standard peptides to each autosampler vial.
o For example, use iRT peptide standards at a 1:20 dilution.
96.Centrifuge samples in autosampler vials at 5,000 x g in a clinical centrifuge
for 1 min to remove bubbles.
Mass spectrometry acquisition
o The following steps describe in detail the use of a nano-LC 2D HPLC
coupled to a TripleTOF 6600. The protocol can be adjusted depending on
the available MS instruments or preferences.
97.Transfer the autosampler vials into the autosampler tray with cooling set at 4°
C.
98.Create loading, injection, and analytical gradient methods with the following
settings:
o Load a total of 1 µg sample into the autosampler loop.
o After injection, transfer the peptide mixtures onto a C18 pre-column chip
(200 µm x 0.4 mm ChromXP C18-CL chip, 3 µm, 120 Å) (or column) and
desalt by washing with aqueous mobile phase A at 2 µL/min for 10 min.
Then, transfer the peptides to an analytical chip (75 µm x 15 cm ChromXP
C18-CL chip, 3 µm, 120 Å) (or column) and elute at a flow rate of 300
nL/min with a gradient method using mobile phases A (aqueous) and B
(organic). Apply a linear gradient from 5% mobile phase B to 35% mobile
phase B over 120 min.
o Subsequently, ramp the mobile phase B to 80% over 5 min, then hold at
80% B for 8 min before returning to 5% B for a 25 min re-equilibration.

33
99.Build a DIA MS instrument method and define the following instrument scan
experiments:
o Experiment 1: perform MS1 precursor ion scan from m/z 400-1,250
(accumulation time of 250 ms).
o Experiment 2-65: perform MS/MS product ion scans for 64 variable
window segments with an MS2 scan range from m/z 100-1,500
(accumulation time of 45 ms per each of the 64-product ion scans per
cycle). Set the collision energy spread to CES = 10, then select the "high
sensitivity product ion scan mode".
o Use the 64-variable window DIA acquisition strategy as described by
Schilling et al.18 for a total cycle time of ~3.2 s. In this acquisition, a series
of variable window widths (5 -90 m/z) is stepped over the full mass range
(m/z 400 -1,250 over 64 SWATH segments, each with a 45 ms
accumulation time, yielding a cycle time of 3.2 s, which includes one MS1
scan with a 250 ms accumulation time). NOTE: The variable window
width is adjusted according to the complexity of the typical MS1 ion
current observed within a certain m/z range using a variable window
calculator algorithm (more narrow windows are chosen in "busy" m/z
ranges, wide windows in m/z ranges with few eluting precursor ions). On
other MS instrument platforms, other DIA window strategies (isolation
schemes) may be implemented.
100. Create a sample queue/batch for all biological replicate samples.
o To obtain consistent and reliable data, regular LC-MS system suitability
assessments need to be performed before and during the entire SWATH
study. Initially, use pre-digested Beta Galactosidase standards and/or
more complex human HeLa cell digests and perform QC acquisitions, or
mass calibration acquisitions typically used in your laboratory or
proteomics core. Also, randomize study samples to avoid systematic
errors; block-randomization of biologically different samples is often
applied in proteomics studies.
o Assign file names for each sample.
o Set injection volumes to the corresponding 1μg of sample (for example, 1
μL of a 1 μg/μL sample).
o Assign the acquisition method generated above.
101. Submit the samples for MS acquisition.
Mass spectrometry data analysis
o Here we describe a workflow using the commercially available software
Spectronaut (alternative software packages can also be used). The
procedure for entering settings and processing data will vary depending
on the analysis software.

34
102. To analyze and quantify protein levels, open the DIA analysis software
Spectronaut.
103. To start the Quantification Analysis, select the "Pipeline" tab, click "Set up
a DIA Analysis from File", and open the MS DIA raw files of interest for
relative quantification.
104. Select "Assign Spectral Library", select the ‘Pan Human Library’, and click
"load" | "next".
o The Pan Human Spectral Library21
should be used only for human
samples (however other Spectral Library approaches can also be applied).
105. Select the "BGS Factory Settings" analysis schema. Verify that the
peptide FDR settings are set to Q-value < 0.01 with sparse identifications and
click "next".
106. Select the appropriate human database FASTA file (the default UniProt
FASTA file assigned to the Pan Human Library) and click "next".
107. Define the condition set-up (e.g., ‘senescent’ and ‘non-senescent’).
108. In the condition set-up form, to each sample assign a correction factor that
is equal to the inverse of its cell count (1 divided by the cell count) and click
"next".
o It is critical to assign correction factors based on cell counts to account for
differences in protein secretion levels due to differences in the number of
cells in each cell culture flask.
109. Select "goa_human" as the gene annotation (ontology) file and click
"next".
110. Review the analysis overview (summary of the experiment set-up) and
select "output directory" to assign an output directory. Click "finish".
111. Finally, click "Run Pipeline" to perform the label-free quantitative analysis.
112. Review the results in the “output directory”. The Spectronaut DIA
Quantitative Analysis Software automatically performs FDR analysis,
generates heat maps and volcano plots (Figure 6), generates lists of identified
and quantified peptides and proteins, and provides Q-values along with
relative fold changes comparing different conditions in a “candidates.tsv” file.
o Expected results: The elevation of several SASP factors is expected in a
successful proteomic analysis, including GDF15, MMP1, STC1, and
CXCL1. The number of proteins identified in an experiment with 5
replicates per condition (5 senescent and 5 controls) is typically

35
approximately 1000 proteins identified in the secretome. The number of
significantly changed proteins (SASP proteins) will typically vary
depending on the senescence inducer or cell type used. For example, we
have previously reported 548 proteins significantly increased in the SASP
of irradiated senescent fibroblasts, but 332 proteins in the SASP of ATVtreated senescent fibroblasts, and 180 proteins in the SASP of irradiated
epithelial cells8
. Significant SASP factors change heterogeneously with
senescence induction (and cell type). In senescent fibroblasts, it is
typically expected that the majority of significant protein changes are
increases in protein secretion.

36
Figure 6. SASP proteomic analysis figures generated by the Spectronaut software.
A) Heatmap depicting the abundance of all proteins identified in the conditioned medium
of senescent (n=6) and control (n=6) lung fibroblasts. B) Volcano plot of the log2 fold
changes in abundance of proteins secreted by senescent versus control cells. Red dots
represent fold changes greater than 1.5-fold, and p-values less than 0.05. C) Line plot
showing the abundance of a classical SASP protein, MMP1, in the secretomes of
senescent and non-senescent cells.

37
1.12. Reagents and solutions
Doxorubicin stock solution
Prepare a stock solution of doxorubicin by dissolving doxorubicin hydrochloride
(Tocris, cat. no. 2252) in DMSO (MilliporeSigma, cat. no. 67-68-5) to obtain a final
concentration of 2.5 mM. This solution can be aliquoted and stored at -20° C shortterm, and at -80° C for long-term storage.
Atazanavir/ritonavir stock solution (ATV/r)
Prepare a stock solution of ATV/r by mixing atazanavir (sulfate) (MedChem
Express, cat. no. HY-17367A) with ritonavir (MedChem Express, cat. no. HY-90001) (wt
ratio atazanavir:ritonavir of 4:1) and dissolving in DMSO (MilliporeSigma, cat. no. 67-68-
5) to obtain a concentration of 8 mg/mL and 2 mg/mL respectively. For example, mix 32
mg of atazanavir with 8 mg of ritonavir and dissolve in 4 mL DMSO. In a sterile
environment, filter the solution using a sterile 0.2 μm filter compatible with DMSO
(Fisherbrand Syringe Filters Sterile, Fisher Scientific, 09-719C). After filtration, keep the
solution sterile. The stock solution can be stored at -20° C short-term, and at -80° C for
long-term storage. The working concentration of ATV/r described in Alternate Protocol
1D (25 μM) is calculated considering only the atazanavir sulfate concentration in the
stock solution (that is 8 mg/mL, approximately 10 mM). Therefore, to prepare media
with a final concentration of 25 μM ATV/r, dilute the stock solution by a factor of 400 in
media.
1.13. Commentary
Background Information

38
Although senescent cells participate in several physiological functions, they are
known to contribute to frailty as well as many age-related diseases and therefore are
considered a major driver of aging and age-related diseases3,4
The secretome of
senescent cells, termed the senescence-associated secretory phenotype (SASP), is
known to cause many of these effects (both beneficial and detrimental)5,6
. For example,
the SASP is linked to optimal wound healing22, but also to pathological states such as
osteoarthritis and tumor growth and metastasis23,24
. The SASP is extremely
heterogeneous and dynamic, varying with the cell type and senescence-inducing
stimulus, as well as how much time has passed since senescence was induced20
.
Characterization of the SASP on a molecular level may provide novel mechanistic
insights into how cellular senescence drives aging, potentially leading to the discovery
of aging biomarkers and targets for preventing or counteracting the deleterious effects
of senescent cells8
.
Previously, antibody arrays were used to study the SASP protein composition7
.
However, antibody arrays can detect only a pre-selected number of proteins. This
method is therefore both biased and limited. Another approach to studying the SASP is
unbiased gene expression analysis of senescent cells20
. This approach enables
researchers to analyze a much larger number of genes compared to a small number of
proteins using antibody arrays. Nevertheless, gene expression analysis cannot directly
provide information about proteins secreted by cells. Recently, more comprehensive
and unbiased mass spectrometric studies have been used to characterize and quantify
SASP protein profiles directly8–10
. A study by Basisty et al.8
found the SASP to be much
larger than previously thought and extremely heterogeneous, depending on the cell type

39
and senescence inducer. Interestingly, many of the SASP proteins overlapped with the
plasma protein signature of aging identified in the Baltimore Longitudinal Study of Aging
(BLSA). This overlap demonstrates the extent to which this approach not only expands
our understanding of the SASP but also permits the identification of promising candidate
protein biomarkers of organismal senescent cell burden in a biofluid that is easily
accessible. The step-by-step procedure described here is meant as a guide to i)
generate senescent cells, ii) collect SASP-containing conditioned media, and iii) perform
a quantitative, unbiased analysis of the SASP using MS with a DIA workflow.
Critical Parameters and Troubleshooting
The methods for senescence induction described in this protocol collection have
been developed and optimized for human IMR90 fibroblasts. Therefore, some
adjustments might be required when using other cell strains. Nevertheless, these
protocols have been successfully used in our laboratory to induce senescence in
multiple human primary cell strains with little to no modification. Hence, we are
confident that they can be applied as they are, or with little optimization, to many human
primary cell strains.
For compatibility with downstream mass spectrometry analysis, there are several
important general considerations. First, the conditioned medium in which the final
collection of secreted proteins takes place should be free of highly abundant protein
additives and free of serum because abundant exogenously added proteins will interfere
with the detection of secreted proteins. Secondly, all buffers and reagents used during
the sample processing for mass spectrometry should be mass spectrometry grade.

40
Thirdly, during all steps of sample processing, the sample should be completely free of
detergents and chemicals that are incompatible with mass spectrometry analysis.
Senescence Induction
• Observation: cells do not develop a senescent phenotype.
o Possible causes and solutions: the dose of senescence inducer is too low.
The cell strain or line used might be resistant to the senescence inducer.
Increase the dose and perform a dose titration checking for senescence
induction using only a couple of easily performed assays, such as SA-βGal activity and cell counts as a proxy for growth arrest. Once a promising
dose is identified, perform Support Protocol 1 to more thoroughly confirm
that the cells developed a senescent phenotype.
• Observation: significant cell death upon treatment with the senescence
inducer.
o Possible causes and solutions: the senescence inducer chosen is toxic to
the cell type used. Reduce the dose (perform a dose titration) and check
for senescence induction using only a couple of easily performed assays,
as suggested above. Alternatively, use another senescence inducer.
Once a promising inducer and dose are identified, perform Support
Protocol 1 to more thoroughly confirm that the cells developed a
senescent phenotype.
Cell cultures in low-serum media before CM collection
• Observation: significant death of senescent cells (viability ≤ 80%, as measured
by PI inclusion assay) when cultured in low-serum media for a prolonged
period.
o Possible causes and solutions: the cell strain or line is highly sensitive to
low-serum media upon senescence induction. Increase the percentage of
serum in the media by performing a titration. Check that the concentration
at which senescent cells remain viable still induces quiescence in the
control cells. Otherwise, do not culture the senescent cells in low-serum
media (except for the last 24 h before CM collection) as described in
Alternate Protocol 2.
• Observation: significant death in control cells when cultured in low-serum
media for a prolonged period.
o Possible causes and solutions: the cell strain or line is highly sensitive to
low-serum media. Increase the percentage of serum in the media,
performing a titration. Check that the concentration at which control cells

41
remain viable still induces quiescence. Otherwise, try to induce
quiescence by other means, such as contact inhibition. If quiescence
cannot be achieved in the control cells, then culture both control and
senescent cells in complete media (except for the last 24 h before CM
collection) and compare the SASP-containing CM to CM generated from
these proliferating control cells. However, under these conditions, one
cannot distinguish if changes in protein secretion are the result of
comparing proliferating cells versus non-proliferating cells, major
differences in cell density between control and senescent conditions, or
differences between senescent and non-senescent cells.
Sample processing and MS analysis
• Observation: High background signal or poor standard curve in BCA results.
o Possible causes and solutions: residual phenol red or incompletely bufferexchanged samples. Ensure phenol red-free media is used for the
collection of CM containing secreted proteins. Alternatively, the Buffer
exchange was not complete. Ensure the buffer exchange is followed
according to the manufacturer protocol, making sure to follow multiple
rounds of exchanging buffer with 50 mM NH4HCO3.
• Observation: Albumin is the only protein detected.
o Possible causes and solutions: the final CM is contaminated with residual
serum. Take care to wash cells 2x or more with PBS before the addition of
serum-free medium for collection of CM.
• Observation: Unexpected direction of protein changes between senescent
and control cells, or unexpectedly high biological variation among samples.
o Possible causes and solutions: Mass spectrometry analysis is not
corrected for cell count. Verify that correction factors are properly
assigned in the MS software before sample acquisition. To ensure protein
levels reflect changes in secretion levels, rather than differences in cell
numbers, it is critical to correct for cell counts.
Understanding Results
It is important to confirm that the desired cell line develops a senescent
phenotype following senescence induction. To this end, multiple senescenceassociated markers should be tested.
EdU incorporation

42
Senescent cells are characterized by an essentially irreversible growth arrest.
Since the senescent cells do not divide, they should not incorporate nucleotides in DNA.
The Click-iT EdU kit detects nucleotide incorporation by supplementing cells with a
nucleotide analog (EdU) that can be later tagged by a selective chemical reaction. The
tag contains a fluorophore and thus incorporation into DNA can be detected with a
fluorescence microscope. Senescent cells, as well as quiescent control cells, should
have low EdU incorporation (Figure 2). Conversely, non-senescent, proliferating cells
should have high EdU staining.
SA-β-Gal activity
Senescent cells increase lysosomes and lysosomal β-galactosidase activity (βGal). Therefore, the β-Gal substrate X-Gal is broken down at a higher rate in senescent
cells and at a suboptimal pH compared to control cells. The breakdown of X-Gal
causes the formation of blue indole crystals, visible by light microscopy. Senescent
cells should have high SA-β-Gal staining, whereas the opposite should be true for
control cells (Figure 3). Note that if cells are close to confluency, staining can increase
significantly, leading to false SA-β-Gal positivity in control cells.
Senescence-associated gene expression
The transcriptome of senescent cells changes significantly compared to that of
non-senescent cells. While the transcriptome is very heterogeneous – dependent on
cell strain and senescence inducer – there is a transcription signature associated with
the senescent phenotype. This signature includes upregulated (e.g. p16INK4a) and
down-regulated (e.g. LaminB1) genes. These changes in transcription can be detected
by qPCR. Figure 4 shows a panel of selected genes and their expected relative mRNA

43
levels in senescent cells compared to quiescent control cells. Depending on the cell
type and senescence inducer, the expression of some of these genes might remain
unchanged or trend in the opposite direction. However, most of the genes in Figure 4
should show similar results, regardless of the cell type and inducer.
Propidium iodide inclusion
Cells used to generate SASP and control CM should be viable and healthy when
producing their secretome, otherwise, proteins from dying cells might significantly
contaminate the CM and falsely be assigned to the SASP. To verify that the cells are
viable and healthy when producing CM, staining for dead or unhealthy cells, for
example using propidium iodide, can be used with flow cytometry. Both senescent and
control cells should have a low percentage of positive cells.
Time Considerations
Characterizing the SASP of a desired cell type using a chosen senescence
inducer will require about 2 months of dedicated effort. The procedure can be divided
into sequential and independent stages.
Validation of senescence induction: generating senescent and control cells and
checking for senescence markers takes about 2 weeks. It might be necessary to invest
more time if the senescence induction protocol requires optimization.
Production of CMs: the generation of senescent and control cells and collection
of CMs should take 10-14 days.
Processing of CMs for MS analysis: proteins in CM samples can be
concentrated, digested, and desalted in 2 to 3 days.

44
LC-MS analysis: running one sample takes approximately 1.5 h. When planning
experiments, keep in mind that the LC-MS in your laboratory/MS core might not be able
to process your samples immediately. Data analysis and identification of SASP proteins
can be done in as little as a few hours to a few days, depending on the complexity of the
MS data.
1.14. Conclusion
This protocol provides a workflow to generate and validate senescence cultures,
as well as to analyze their SASP. Thus, this workflow can be a valuable tool for
studying the heterogeneity and dynamics of the SASP.
An important quality-control step in this process is the validation of senescence
induction. While the imaging assays described to validate senescence induction (EdU
incorporation, SA-β-Gal) have been extensively used to assess senescence burden,
their quantification remains suboptimal. This is because they typically entail manual
scoring by eye, making such assessments low-throughput and potentially subject to
user bias. Indeed, the quantification of senescent cells – whether for validation of
senescence induction or any other cellular senescence study – remains one of the
biggest challenges in the senescence field. To address this challenge, a new workflow
was developed to robustly and unbiasedly quantify senescent cells, which is described
in Chapter 2.

45
Chapter 2: A Fully-Automated Senescence Test (FAST) for the
High-Throughput Quantification of Senescence-Associated
Markers
2.1. Introduction
Cellular senescence is a complex stress response typically characterized by
essentially irreversible cell cycle arrest, altered morphology, increased lysosomal
activity, and profound changes in gene expression, including the acquisition of a
senescence-associated secretory phenotype (SASP)4
. This cellular response can be
triggered by many different types of stressors, such as telomere dysfunction25,26, direct
DNA damage27, oncogenic signaling28–30, and mitochondrial dysfunction31
. Cellular
senescence phenotypes are highly heterogeneous and dependent on tissue and cell
type8,20
. Therefore, detection technologies need to allow for robust adaptability to
different specimens.
Cellular senescence has important physiological roles. For instance, the
transient presence of senescent cells is beneficial for cancer prevention29,30, embryonic
development32,33, and wound healing22
. In contrast, senescent cells accumulate in
aging tissues34,35, which promotes chronic inflammation and increases the risk of agerelated diseases4
. Preclinical studies have demonstrated that targeting senescent cells
can mitigate age-related diseases and increase median lifespan36
. Hence, drugs that
selectively eliminate senescent cells (“senolytics”) or dampen their SASP
(“senomorphics”) have the potential to improve the treatment and prevention of agerelated diseases4,37,38
. Indeed, several human clinical trials are currently underway38,39
,

46
some of which have shown promising outcomes40
. However, a deeper understanding of
this Janus-faced stress response is needed to develop safe and effective senescencetargeting therapies that can combat age-related dysfunction and disease39
.
One major hurdle in studying cellular senescence is the detection and
quantification of senescent cells, primarily because there are no senescence-specific
markers4
. Instead, detection relies on using one or more senescence-associated
marker(s) – which are not unique to senescent cells. Some of the most widely adopted
senescence-associated markers include senescence-associated β-galactosidase
activity (SA-β-Gal) and proliferation arrest measurements, such as lack of EdU
incorporation41
. Because SA-β-Gal is a colorogenic stain, not fluorescent, its
quantitative analysis is uncommon. Current methods often rely on manual scoring of
microscopy images41 or use semi-automated, low throughput image analysis workflows
that either do not assess multiple markers42,43 or have limited sample processing
capabilities44
. Thus, quantification of cellular senesce is often subjective and timeconsuming, lacking standardization, altogether precluding its use in high-content / highthroughput settings.
To overcome these challenges, the Fully-Automated Senescence Test (FAST)
was developed. This method produces unbiased assessments of SA-β-Gal and EdU
staining by 1) calculation of colorimetric SA-β-Gal optical density (OD), which makes the
quantification independent of microscope model and settings; 2) leveraging internal
background controls, which allow unbiased staining thresholding and scoring with no
assumptions on the senescence phenotype; alternatively, 3) using machine learning
(ML, hence ML-FAST) with biological controls for scoring based on the combination of

47
SA-β-Gal, EdU and nucleus size; 4) automating image acquisition, image processing,
and data analysis, which enable high-throughput workflows. FAST was implemented in
the commercial image analysis software Image Analyst MKII to provide microplatebased automation and in R to provide custom data analysis and graphing, and here a
protocol and all required pipelines and R scripts are provided to implement this assay.
FAST is agnostic to the microscopy system used, and examples are provided using a
Nikon Eclipse Ti-PFS wide field setup and a Zeiss LSM 980 laser scanning confocal
microscope. Moreover, FAST simultaneously allows to evaluate cell counts and
morphological alterations – a third senescence hallmark – via nuclear area
measurements. Hence, FAST serves as a comprehensive, unbiased tool to rapidly
assess senescence burden by measuring three key senescence-associated markers.
2.2. Methods
Cell culture
Primary human lung microvascular endothelial cells (HMVEC-L) were purchased
from Lonza (CC-2527). HMVEC-L were cultured in EGMTM-2MV Microvascular
Endothelial Cell Growth Medium-2 BulletKitTM (Lonza, CC-3202) at 37°C, 14% O2, 5%
CO2. Human lung fibroblasts IMR-90 were purchased from Coriell Institute (I90). IMR90 cells were cultured in DMEM (Corning, 01-017-CV) supplemented with 10% FBS
(R&D Systems, S11550H), 100 units/mL penicillin, and 100 µg/mL streptomycin (R&D
Systems, B21210) at 37°C, 3% O2, 10% CO2. For all experiments performed, both
HMVEC-L and IMR-90 were cultured in 96-well microplates appropriate for microscopy
imaging (Corning, 3904), with media changes every 2-3 days.

48
To achieve serum starvation in HMVEC-L, cells were washed twice in DPBS
containing Ca2+ and Mg2+ (Gibco, 14040-117) and then cultured for 72 h in low-serum
EGMTM-2MV medium (0.5% FBS instead of 5.0% FBS). To achieve high confluency
conditions, cells were seeded at high density (25,000 cells/cm2
) and further cultured for
7 days before irradiation; non-senescent control cells were also seeded at the same
density and cultured for 7 days before analysis.
Senescence induction
Senescence was induced as previously described45
. For ionizing-radiationinduced senescence, the cells were irradiated with 15 Gy, and medium change was
performed immediately after treatment. Cells were considered senescent after at least
7 days since irradiation, during which medium was regularly changed (every 2-3 days).
For doxorubicin-induced senescence, cells were treated with different dilutions of the
drug (Millipore Sigma, D1515-10MG), while non-senescent control cells were treated
with vehicle only (DMSO, ThermoFisher Scientific, BP231-100). Cells were cultured in
doxorubicin/vehicle-containing medium for 24 h, after which two washes were
performed with DPBS containing Ca2+ and Mg2+ (Gibco, 14040-117) before adding
regular medium. Cells were further cultured for 6 days before analysis (i.e. 7 days
since the beginning of doxorubicin treatment).
Senolytic Treatment
On the last day before analysis, cells were treated with the senolytic
ABT263/Navitoclax (Selleck Chemicals, S1001) at different concentrations for 24 h,
while only vehicle (DMSO) was given as mock treatment.
SA-β-Gal and EdU Staining

49
A detailed, step-by-step protocol is provided at protocol.io
(dx.doi.org/10.17504/protocols.io.kxygx3ypwg8j/v1). Commercially available kits were
used to perform SA-β-Gal (Abcam, ab65351) and EdU staining (ThermoFisher
Scientific, C10351). Cells in “Staining wells” were given medium containing 2.5 µM EdU
for 24 h before fixation. Cells in “Background wells” were instead given medium
containing vehicle (DMSO). After 24h, cells were fixed by adding 8% PFA in PBS prewarmed to 37°C directly to the medium up to a final concentration of 4% PFA and
incubated for 15 min at RT. Subsequently, cells were washed twice with PBS, and SAβ-Gal staining was performed.
To stain for SA-β-Gal, fixed cells were given the staining solution mix as
recommended by the manufacturer (Abcam, ab65351). However, the X-Gal powder
used was separately purchased (Life Technologies, 15520-018). “Staining wells” were
given the complete staining solution mix, whereas “Background” wells were given a
solution that did not contain X-Gal, but only vehicle (DMSO). Staining was performed
overnight at 37°C in an incubator without CO2 control. To prevent nonspecific indole
crystal formation, empty spaces in between wells of the microplates were filled with
PBS, and parafilm was used to seal the microplates before the overnight incubation.
After the overnight incubation, cells were washed twice with PBS to stop the staining.
After SA-β-Gal staining, EdU detection was performed. Briefly, cells were
permeabilized at room temperature for 15 min with 0.5% Triton X-100 (Millipore Sigma,
T9284-500ML) in PBS. After permeabilization, the Click-iT Reaction Cocktail was
added as per the user manual, and cells were incubated for 30 min in the dark. After
the incubation, cells were washed once with PBS, counterstained with 0.5 μg/ml DAPI in

50
MilliQ water for 30 min at room temperature in the dark, then washed once with MilliQ
water. Finally, cells and cell-free wells were covered with PBS and imaged.
Image Acquisition
Wide-field microscopy was performed on a Nikon Eclipse Ti-PFS fully motorized
microscope controlled by NIS Elements AR 5.21 (Nikon, Melville, NY). The setup
comprised of a Lambda 10-3 emission filter wheel, a SmartShutter in the brightfield light
path, and a 7-channel Lambda 821 (Sutter Instruments, Novato, CA) LED
epifluorescence light source with excitation filters on the LEDs, controlled by a PXI 6723
DAQ (NIDAQ; National Instruments) board. Images were acquired by an Andor iXon
Life 888 EMCCD camera (Oxford Instruments) using 10 ms exposure times, with a
Nikon S Fluor 10× DIC NA=0.5 lens, using the following filter sets (Semrock; excitation
– dichroic mirror – emission given as center/bandwidth in nm): for DAPI: using the 385
nm LED 390/40 – 409 – 460/80; for EdU 480 nm LED with 480/17 – 495 – 542/27. For
SA-β-GAL an incandescent Koehler illumination was used and a 692/40 “emission”
filter. Using the ND-Acquisition feature of Elements, 3×3 tiled images were recorded
without overlap or registration, using the full 1024×1024 resolution of the camera (1.3
µm/pixel), and the above-defined 3 channels. The Kohler condenser was carefully
focused for each experiment in the center of a well, with the aperture diaphragm semiopen. For autofocusing, the Nikon’s Perfect Focus System was used. For each
microplate, two acquisitions were run: one to image the wells containing samples
(“Staining” and “Background wells”; typically center 60 wells), and another to image
wells that did not contain any cells (“Blank wells”; 36 edge wells). In addition to the
above data, the average pixel intensity measured with no illumination (dark current) was

51
determined for precise OD calculation below. Data were saved and analyzed as native
*.nd2 files.
Image Acquisition on an Alternative Setup
Confocal microscopic image acquisition was performed on a Zeiss LSM 980
laser scanning confocal microscope. Standard (Smart Setup) settings were used for
DAPI and EdU, and SA-β-GAL was recorded using the transmitted light detector and
the 639 nm laser. Koehler illumination and tiling were set up as described above for
wide-field microscopy. Here a Plan-Apochromat 10× NA=0.45 lens was used and 3.78
s frame time. Optical zoom of 1.0 resulted in 0.83 µm/pixel resolution in 1024×1024 tile
frames. For autofocusing the Zeiss Definite Focus 3 was used. The microplate-based
acquisition was set up using the AI sample finder feature of the controller software Zen
2.3, placing one tile region into the center of each well, and data were saved and
analyzed as native *.czi files.
Image Processing
Native format image files were opened in Image Analyst MKII (Image Analyst
Software, Novato, CA) as a Multi-Dimensional Open Dialog, representing one
microplate (or its blank recordings) at a time. Analysis was performed using modified
standard and custom pipelines (https://github.com/gerencserlab/FAST). A detailed,
step-by-step protocol is provided at dx.doi.org/10.17504/protocols.io.kxygx3ypwg8j/v1.
Briefly, “Blank well” reference images were created by the median projection of replicate
wells using the “Create BLANK reference image for multiwell plate using median”
pipeline. These reference images in conjunction with pixel intensity related to detector
dark current were used then by the main pipeline for SA-β-GAL OD calculation below.

52
Next, the file containing all other assay wells, including “Stained wells” and “Background
wells” images was opened, and the pipeline “FAST Analysis Pipeline - Basic” was
executed in all wells. The output Excel file containing single-cell measurements for the
whole microplate was saved, and further data analysis was triggered by executing the
“Run FAST.R Shiny App” pipeline.
Data Analysis
Single-cell measurements were analyzed using a web browser-based Rapplication, FAST.R (https://github.com/f-neri/FASTR), which we developed using the R
Shiny package46 (version 1.7.4, using R 4.3.1) and is designed to run locally. No
command line or scripting knowledge is required for its installation or use (for a detailed,
step-by-step protocol, see protocol.io:
dx.doi.org/10.17504/protocols.io.kxygx3ypwg8j/v1). The inputs of FAST.R are the
above Excel output file containing the single-cell measurements, and a plate map in
*.csv format. The output consists of i) a single-cell data table, containing single-cell
measurements integrated with the user-defined metadata; ii) an analysis report table,
which contains staining scoring and nuclear area data for each sample imaged; iii)
automatically generated graphs, as shown in the figures below, which help users
visualize and understand their data. Briefly, the app associates image analysis results
with well-condition labels and determines the positive staining thresholds, independently
for each condition based on the labels. Then, these thresholds are used for generating
positive cell counts and percents for each well and condition. The app also provides
additional summary information for each well, such as cell counts, quartile values for
each staining, and fold changes. The machine learning-based senescence

53
classification was implemented with the R package caret47 using random forest
classifiers. Briefly, single-cell measurements of all senescence markers (SA-β-Gal,
EdU, and nuclear area) were pre-processed by centering and scaling, and then the
classifier model was trained using a random forest algorithm and fine-tuned by repeated
k-fold cross-validation method. The trained classifier was used to calculate the
percentage of predicted senescent cells in each well not used for training.
X-Gal and C12FDG co-staining
IMR-90 cells were cultured as described above, except they were seeded on
Matrigel-coated glass bottom plates (Greiner SensoPlate #655892). For senescence
staining, the culture media was removed from each well containing cells (irradiated and
non-irradiated controls) and replaced with custom low sodium bicarbonate, clear
imaging media (Image Analyst Software, Novato CA) comprised of DMEM, 1% FBS, LGln 4 mM, sodium pyruvate 1 mM, and 25 mM glucose, at 37°C. The media was
supplemented with Bafilomycin A (1 µM) and the cultures were incubated for 40 min at
37°C in an air incubator (no CO2 control). Cells were finally stained with 30 µM C12FDG
(ThermoFisher Scientific, D2893) prepared in imaging media (with 1 µM Bafilomycin A
and 1 µg/mL Hoechst 33342) and incubated for 1.5 h. Live cell imaging was performed
on the above-specified Nikon Eclipse Ti-PFS microscope. After imaging C12FDG and
Hoechst fluorescence, cells were immediately fixed for 10 min in 2% PFA and stained
for 20 hours with X-Gal as described above. The microplate was subsequently
reimaged for both fluorescence (C12FDG and Hoechst) and chromogenic X-Gal staining.
Analysis was performed using the “FAST Analysis Pipeline - Basic - Modified for Live
plus Fixed Merging” pipeline, where an image registration step matching live- and fixed-

54
cell imaged Hoechst fluorescence was added in front of the basic pipeline, and a
customized version of the FAST.R Shiny app.
Statistics
Statistical tests employed are exhaustively described in each figure legend.
Such statistical tests were either performed using GraphPad Prism or the R package
“stats” (v 4.4.0).
2.3. Results
FAST Workflow
The FAST workflow is comprised of four parts: sample preparation, image
acquisition, image analysis, and data analysis (Figure 7) to assess senescence burden
in a fully automated process. To demonstrate these, we used senescence-induced and
non-senescent primary human lung endothelial cells (HMVEC-L) and fibroblasts (IMR90) cultured in optical quality 96-well plates. Regardless of the assayed biology, each
microplate included wells containing cells stained with DAPI, EdU, and SA-β-Gal
(“Staining wells”), or DAPI only (“Background wells”). Moreover, selected wells
contained no cells and no stain but PBS only (“Blank wells”) (Figure 7A).

55

56
Figure 7. FAST workflow.
A) For each condition (e.g. Control and Senescent), cells are given substrates for SA-βGal and EdU staining (Staining) or vehicle (Background). All cell wells are DAPI-stained.
Some wells do not contain any cells (Blank). B) Automated image acquisition is
performed to capture nuclear counterstain DAPI (blue channel), EdU staining (green
channel), and SA-β-Gal (bright field, BF). The use of a red wavelength emission filter
(690 nm) for BF imaging results in SA-β-Gal crystals appearing as dark pixels. In
combination with the acquisition of Blank images, this modified BF imaging enables
optical density (OD) measurements of SA-β-Gal staining. C) Image analysis in Image
Analyst MKII. DAPI images are segmented, resulting in nuclear and perinuclear labels,
and these are used to measure integrated intensities of EdU and nuclear area, or
integrated ODs of SA-β-Gal, respectively. D) Single-cell measurements are analyzed and
graphed with FAST.R, a custom R Shiny application.
The entire microplate was imaged, covering most of the bottom of each well
using tiling. Fluorescent images were captured for both DAPI (blue channel) and EdU
staining (green channel). SA-β-Gal, on the other hand, was imaged as a monochrome
bright field channel using a red filter in the light path (or using a red laser to illuminate)
matching the peak absorption of the SA-β-Gal staining (Figure 7B). Image data was
saved in a 3-channel *.nd2 or *.czi single file for each microplate, containing a single
stitched large view field for each imaged well, and well labels as metadata.
Single-cell EdU intensities, SA-β-Gal optical densities, and nucleus size were
determined using automated execution of a single image analysis pipeline for each well
in the native microscopy format image data in Image Analyst MKII (Figure 7C). As a
preparation, first, a blank reference image was created from the pixel-wise median of all
Blank wells. By calculating the OD using blank images internal to each experiment,
optical effects, such as experiment-to-experiment variations in condenser or lamp
settings, and vignetting from tiling were canceled. Furthermore, precise OD calculation
was obtained by a low-pass spatial filter48 to remove non-specific signals in images
originating from cellular processes. To generate single-cell data, DAPI images were
used to segment nuclei. Here we used a watershed and flood-fill-based morphological

57
segmentation. We chose this method over neural-network-based segmentation, such
as Cellpose49,50 (also available in our pipeline repository), because we occasionally
observed biases in cell detection due to changing nucleus shape (data not shown). SAβ-Gal absorbance was measured over perinuclear ring-shaped areas and EdU staining
intensities over the nuclei. Because in each case the total amount of the marker was
relevant to the biology, optical densities or fluorescence intensities were integrated in
these areas. The above analysis was performed as a single image analysis pipeline
and results were saved as one tabular data (Excel) file per microplate.
For data analysis and visualization, an open-source R-based application,
FAST.R, was developed using R Shiny46 and was integrated into the above workflow
(Figure 7D). FAST.R allows the association of plate maps with the single-cell data and
calculates thresholds to be applied to all “Staining wells” based on the “Background
wells”. The application outputs consist of an analysis report, which details staining
quartiles, percentages of SA-B-Gal and EdU positive cells, and nuclear area data for
each well, as well as graphs presented below. For enhanced senescence detection,
senescence scoring can alternatively be performed using a machine learning approach
that combines the three staining readouts.
Automation of the Senescence Marker Scoring
A critical component of automated analysis is the definition of marker positivity
using objective criteria. The above workflow provided a standardized input for data
analysis by OD calculation for SA-β-Gal and recording of “Background wells” containing
unstained cells for both SA-β-Gal and EdU.

58

59
Figure 8. Standardization of detection.
A) Representative images of primary human microvasculature endothelial cells. Top
panels show background cells, while bottom panels show stained cells; left panels show
non-senescent cells (CTL), right panels show ionizing radiation-induced senescent cells
(IR). B) Standardized thresholding for percentage staining calculation. 1) Signal
thresholds are generated based on the 95th percentile in SA-β-Gal and EdU staining
measurements of background cells. 2) Signal thresholds are then used to establish SAβ-Gal and EdU positivity in stained cells. 3) The percentage of EdU- and SA-β-Galpositive cells is calculated for each well. 1-2: each dot corresponds to one cell in a
representative microplate (n cells: CTL = 6359, CTL background = 5012, IR = 1183, IR
background = 884). 3: each dot corresponds to one well (n = 9) from the same plate. CE) Boxplots with median well values for SA-β-Gal staining (C), EdU staining (D), and
nuclear area (E). Each dot corresponds to one well in the same microplate (n = 9). F)
3D scatterplot showing all 3 measured variables for each well (n = 9). ****, p<0.0001 by
Mann-Whitney test.
We first tested the FAST workflow using primary human lung microvascular
endothelial (HMVEC-L) cells induced to senesce through ionizing radiation (IR; Figure
8)
27,45
. Proliferating cells that were mock irradiated served as the non-senescent control
(CTL). The FAST.R app generated detection thresholds for SA-β-Gal and EdU by
calculating OD and intensity values at the 95th percentile of cells in a pool of
“Background wells” (Figure 8B.1). Then, these values were used to determine SA-βGal and EdU positivity in the “Staining wells” (Figure 8B.2). Importantly, different
conditions, such as IR and CTL, may absorb or scatter light differently, therefore each
condition had its own “Background well” control, and FAST.R automatically matched
these to the “Staining wells”. Finally, the application computed the percentage of SA-βGal- and EdU-positive cells in each well (Figure 8B.3). As expected, the analysis
showed a statistically significant increase in the proportion of SA-β-Gal-positive cells
and a concomitant reduction of EdU-positive cells in the IR senescent well compared to
non-senescent CTL wells. In addition, FAST.R provided statistics on signal intensities
and nuclear area (Figure 8C-F). Figure 8 and Figure 9 show graphs automatically
generated by FAST.R.

60
Figure 9. Additional graphs generated with FAST
A) Scatterplot showing SA-β-Gal and EdU signal intensity of all cells. Each dot is a cell
(n cells: CTL = 6359, IR = 1183). B) Bar graph showing the percentage of all cells
belonging to one of the four possible staining categories: EdU+/-, SA-β-Gal +/-. C) 2D
boxplot showing SA-β-Gal and EdU signal intensity of cells grouped by well. Dots indicate
median (50th percentile) values, solid lines show interquartile (25th to 75th percentile)
range, dashed lines show min to max range. Each data point is a well (n = 9) from the
same plate. D) Violin plot showing nuclear area distribution.
We also confirmed that the FAST workflow is compatible with different
microscope setups (Figure 10).

61
Figure 10. FAST is compatible with different microscope setups.
The same microplate containing cells treated with different concentrations of doxorubicin
or DMSO vehicle was imaged with a Nikon Eclipse Ti-PFS wide-field microscope (A, C)
and a Zeiss LSM 980 laser scanning confocal microscope (B, D). A, B) Percentage of
SA-β-Gal- and EdU-positive cells per well for each condition. Each dot is a well (n = 4).
C, D) Boxplot plot showing median nuclear area values for each condition. Each dot is a
well (n = 4).

62
FAST tracks senescent cell populations in experimental settings beset with
a high rate of false positive staining
The markers SA-β-Gal and lack of EdU incorporation are not specific to cellular
senescence. Culturing conditions can result in a significant number of non-senescent
cells with false-positive senescence staining, i.e. high SA-β-Gal and low EdU
incorporation.
To test the sensitivity of FAST to this intrinsic limitation of senescence-associated
markers, we evaluated SA-β-Gal and EdU staining of both senescent and nonsenescent cells under conditions known to produce false-positive senescence staining
(Figure 11). Specifically, we analyzed the SA-β-Gal and EdU staining in senescent and
non-senescent HMVEC-L cells subjected to serum starvation (Figure 11A-B), prolonged
(48 h) SA-β-Gal staining (Figure 11C-D), or cultured at a high cell density (Figure 11EF). In every experimental condition tested, FAST allowed to detect a statistically
significant increase in the proportion of SA-β-Gal positive and EdU-negative cells
between the senescent samples and the highly false-positive non-senescent samples
(Figure 11B, D, F).

63

64
Figure 11. FAST distinguishes senescent cell populations in experimental settings with a
high rate of false positives.
A) Representative images of serum-starved (SS) non-senescent (CTL) and ionizing
radiation-induced senescent (IR) endothelial cells. B) Quantification of (A) with
percentages of SA-β-Gal- and EdU-positive cells grouped by well. Both SS and fullserum (standard) samples are shown. Each dot is a well (n = 9) in a representative
microplate of 3 independent experiments. C) Representative images of CTL and IR
endothelial cells with prolonged SA-β-Gal staining (48 h). D) Quantification of (C) with
percentages of SA-β-Gal- and EdU-positive cells grouped by well. Both samples with
prolonged staining (48h stain) and standard overnight staining (standard) are shown.
Each dot is a well (n wells: CTL & IR 48h stain = 4, CTL & IR standard = 9). E)
Representative images of CTL and IR endothelial cells at high confluency. F)
Quantification of (E) with percentages of SA-β-Gal- and EdU-positive cells grouped by
well. Both samples at high confluency (confluent) and low confluency (standard) are
shown. Each dot is a well (n wells: CTL & IR confluent = 4, CTL & IR standard = 9). *,
p<0.05; ****, p<0.0001 by Mann-Whitney test.
Comparison of colorimetric X-Gal staining with fluorescent C12FDG staining
While FAST relies on the classical SA-β-Gal staining, which uses chromogenic XGal as substrate34, fluorogenic β-galactosidase substrates have been increasingly
employed for cellular senescence detection51–53
. Despite this, quantitative comparisons
of SA-β-Gal staining between X-Gal and fluorogenic dyes are scarce and predominantly
restricted to population-level analyses51
. The principle of the SA-β-Gal staining is the
detection of an enzyme activity that has been constrained by assay conditions, i.e.
suboptimal pH and PFA fixation. Thus, it is unclear how replacing one substrate with
another or staining live vs fixed cells affects the detection, and therefore whether
fluorescence stains are direct substitutes for X-Gal. Utilizing FAST, here we directly
compared the colorimetric and fluorescent SA-β-Gal staining on the single cell level
(Figure 12).

65
Figure 12. Comparison of colorimetric X-Gal and fluorescent C12FDG live cell staining
with FAST.
A) Representative images of non-senescent (CTL) and ionizing radiation-induced
senescent (IR) IMR-90 fibroblasts. Live cells were stained with fluorescent C12FDG and
imaged, followed by fixation, staining with colorimetric X-Gal, and subsequent re-imaging
of the same view fields. B) Scatterplot with median signal of X-Gal and C12FDG for each
well (n = 3). C) Boxplot showing the fold change in median signal intensity of IR wells
relative to CTL (n wells = 3). The p-value was calculated by Mann-Whitney test. D)
Scatterplots of single-cell staining intensities for X-Gal versus C12FDG in CTL (left) and
IR (right) conditions. Fitted linear regression models are indicated by solid black lines; n
cells: CTL = 22601, IR = 5198. Representative of 3 experiments.
To do this, CTL and IR IMR-90 live fibroblasts were incubated with the cellpermeable fluorogenic substrate C12FDG (in the presence of 1 µM Bafilomycin A to
increase lysosomal pH), followed by live imaging, fixation, staining with X-Gal, and reimaging the identical view-fields (Figure 12A). As anticipated, both X-Gal and C12FDG
staining intensity were elevated in IR-induced senescent cells compared to controls
(Figure 12B). Notably, X-Gal staining exhibited a trending larger relative increase than
C12FDG (Figure 12C). Despite both stains being strongly induced by IR, linear

66
regression analysis revealed only a modest correlation at the single-cell level between
X-Gal and C12FDG staining in IR-induced cultures (R² = 0.142), with no observable
correlation in CTL (R² = 0.002) (Figure 12D). These data question whether the same
molecular entity is reported by the two stains. Furthermore, while we successfully
combined X-Gal with fluorescence (DAPI, Alexa488), C12FDG exhibited redistribution
inside and in between cells upon fixation and permeabilization (Figure 13), precluding
its use in multiplexed immunocytochemistry paradigms.

67

68
Figure 13. C12FDG SA-β-Gal redistributes inside and in between cells upon fixation and
permeabilization.
A-C) Live (A), fixed (B), and merged images (C) of senescent IMR-90 fibroblasts stained
with C12FDG. For ease of distinction, fluorescence from live images is shown in orange,
while fluorescence from fixed cells is shown in green. Orange arrows indicate example
cells with bright staining during live imaging subsequently lost after fixation and
permeabilization. Green arrow shows an example cell with low staining during live
imaging which subsequently becomes highly fluorescent after fixation and
permeabilization.
Benchmarking FAST: Senescence Inducers
The sensitivity of FAST was tested by analyzing dose-responses and calculating
Z-factors, in different aspects of cellular senescence. As a proof of concept, first, we
tested senescence induction in lung fibroblasts IMR-90 after treatment with increasing
concentrations of doxorubicin (Doxo), a chemotherapeutic drug known to induce
senescence45,54,55 (Figure 14A-B). FAST quantifies single cells and it sensitively tracked
the expected reduction in cell counts seven days after the commencement of treatment
compared to the vehicle-treated (DMSO) cells (Figure 14C). FAST resolved a
concentration-dependent increase in the fraction of SA-β-Gal positive and EdU negative
cells (Figure 14D), consistent with a senescent phenotype, which plateaued at 250 nM
and 500 nM for SA-β-Gal and EdU respectively.

69
Figure 14. Benchmarking FAST with senescence induction.
A) Experimental design to test senescence induction with a given compound, in this case,
the known senescence inducer doxorubicin (Doxo). B) Representative images of primary
human IMR-90 fibroblasts treated with different concentrations of Doxo. C) Cell counts
per well for each Doxo concentration (n wells = 6). D) Percentage of SA-β-Gal- and EdUpositive cells per well for each Doxo concentration. Each dot is a well (n = 6). E) Z-factor
calculations across different metrics: percentage of SA-β-Gal-positive cells, percentage
of EdU-positive cells, and cell counts per well. SA-β-Gal and EdU percentages for each
Doxo concentration are compared to the DMSO condition. Cell counts for each Doxo
concentration are compared to the (8-days cultured) Doxo 0 nM condition. ns, adjustedp>0.05; ****, adjusted-p<0.0001 by Tukey’s test after significant (p<0.05) one-way
ANOVA.
To evaluate the sensitivity of FAST, the Z-factor was calculated (Figure 14E).
The Z-factor is a statistical data quality indicator often used to evaluate the performance
and signal robustness of high-throughput screening bioassays 56
. Assays with Z-factors

70
between 0.5 and 1 are considered to be of good quality, and suitable for highthroughput screenings. Across all the metrics (i.e., SA-β-Gal, EdU, and cell counts), the
Z-factor exceeded 0.5 for all concentrations tested. Given that Doxo is a recognized
senescence inducer, this data suggests that FAST could serve as a high-quality
bioassay for high-throughput screenings of senescence inducers.
Benchmarking FAST: Senolytics
As a proof of concept, the ability of FAST to measure senolytic activity was tested
by treating IR-induced HMVEC-L cells with increasing concentrations of ABT263
(Navitoclax), a chemotherapeutic compound recognized for its senolytic properties57
(Figure 15A).

71
Figure 15. Benchmarking FAST with a senolytic compound.
A) Representative images of senescent HMVEC-L cells treated with different
concentrations of the senolytic compound ABT263 for 24 h. CTL: non-senescent cells,
IR: ionizing radiation-induced senescent cells. B) Percentage of SA-β-Gal- and EdUpositive cells per well for all ABT263 concentrations (n = 4 wells). C) Percentage of viable
cells based on cell counts per well normalized to vehicle (0 µM ABT263) condition (n = 4
wells). Comparison of viability between CTL and IR cells for each ABT263 concentration:
***, adjusted-p <0.001; ****, adjusted-p<0.0001 by Bonferroni-Dunn test. D) Z-factor
calculations for viability measurements at each ABT263 concentration. For each ABT263
concentration, the viability of IR cells was compared to the viability of CTL cells treated
with the same senolytic concentration.
As expected, compared to the non-senescent CTL group (Figure 15B), the IRtreated HMVEC-L showed an increase in the proportion of SA-β-Gal positive and EdU
negative cells. All ABT263 concentrations tested resulted in a significant reduction of
viability (based on cell count) in IR cells compared to CTL cells (Figure 15C).

72
Interestingly, the proportions of SA-β-Gal positive and EdU negative IR cells that
survived the senolytic treatment remained the same as that of the vehicle-treated IR
cells at all concentrations of ABT263 tested (Figure 15B and Figure 16).
To evaluate the sensitivity of FAST at detecting changes in viability upon drug
treatment, the Z-factor was calculated (Figure 15D). One of the ABT263 concentrations
tested, 1 µM, had a Z-factor that exceeded 0.5. Considering that ABT263 is a known
senolytic, this data suggests that FAST could be used for high-throughput screening of
senolytics.

73
Figure 16. ABT263 treatment does not affect SA-β-Gal and EdU staining in senescent
microvascular endothelial cells.
A-C) Median SA-β-Gal (A), EdU (B), and nuclear area values (C) per well at different
ABT263 concentrations in non-senescent control (CTL) and IR-induced senescent (IR)
cell populations (n = 4). Non-significant ANOVA p-values (p>0.05) are shown (A, B). ns,
adjusted-p>0.05; *, adjusted-p<0.05, **, adjusted-p<0.01 by Tukey’s test after significant
(p<0.05) one-way ANOVA. D) 3D scatterplots with all 3 variables for CTL (left) and IR
(right) wells at different ABT263 concentrations.
Analysis of senescence-associated markers using machine learning
The above-presented method scores a cell as senescent or non-senescent
based on unbiased thresholds for two markers (SA-β-Gal and EdU), determined in
unstained samples, thus using no prior knowledge of the senescence phenotype. While

74
nuclear area was informative (Figure 8E-F), the above threshold generation method of
using unstained controls is not applicable for geometric parameters. Therefore, in order
to combine all three markers (SA-β-Gal, EdU, and nuclear area), here we describe a
trainable, ML approach (ML-FAST) using the unscored, raw data the FAST workflow
provides (Figure 17).

75

76
Figure 17. Combination of markers using machine learning improves repeatability of
senescence detection.
A) Conceptual scheme for machine learning classifiers. In each replicate microplate, 6
non-senescent (CTL) and 6 IR senescent wells (SEN) including both serum conditions
were randomly selected for model training, and the remaining wells were used for the
data shown. For each training dataset, random forest classifiers were trained using the
indicated combinations of the measured senescence markers (SA-β-Gal, EdU, and
nuclear area). B) Percentage of cells classified as senescent in CTL (green) and SEN
(purple) test wells by each model in HMVEC-L endothelial cells in either full-serum (FS)
or serum-starved conditions (SS). Representative of 3 experiments, dots are technical
replicate wells (n = 6). C) Z-factors comparing SEN and CTL in (B) for both FS (grey) and
SS conditions (white). Dots are independent experiments (n=3). D, E) Senescence
classification in IMR-90 fibroblasts as described (B) and (C), using n = 6 well replicates
in D) and n=2 experimental replicates in E). *, p<0.05 using two-way repeated-measures
ANOVA with Dunnett’s multiple comparison test comparing to the triple marker classifier.
For this all data from C) and E) were pooled and accounted for repeated use of the same
data with multiple classifiers. While cell type and serum condition did not have a
significant effect, the z-factors significantly varied between the used ML models
Classifier models distinguishing senescent from non-senescent cells were based
on the random forest algorithm and trained on positive (IR) and negative (mock) sample
wells (ignoring impurity) using one, two, or all three markers (Figure 17A). The training
was performed on single cells (1461- 6042 cells) pooled from 6 positive and 6 negative
control wells. Because the random forest model requires relatively little training data
and time, we performed the training and testing independently for each cell type in each
replicate microplate. Figure 17B and D show model predictions in test wells for
HMVEC-L and IMR-90 cells, respectively. The model trained on all three markers was
the only one providing consistently greater than 0.5 z-factors in all trials, and for this
model in all cell conditions, mean z-factors were significantly or trending greater than
with all the other models (Figure 17C, E). The combination of nuclear area with SA-βGal performed similarly well as the combination of EdU with SA-β-Gal, showing a small
difference compared to the triple marker model. This latter finding suggests a low-cost
variant of FAST that uses only SA-β-Gal and DAPI staining, but no EdU. Altogether,

77
ML-FAST with two or three markers showed high performance across different cell
types (HMVEC-L and IMR-90) and culturing conditions (full-serum and serum-starved
conditions) as measured by Z-factor (Figure 17C, E). Thus, the ML-FAST can support
sensitive detection of modulators of cellular senescence at high throughput.
2.4. Discussion
FAST represents a significant advancement over the current manual41 and semiautomatic scoring42–44,58–60 methods used to quantify colorimetric SA-β-Gal and EdU
staining, which remains low-throughput and not standardized. Key innovations
presented here are: 1) calculating SA-β-Gal optical density (OD), which makes the
quantification independent of microscope model and settings; 2) implementing internal
negative controls (i.e. ”Background wells”) to standardize staining thresholding and
scoring; 3) automating image acquisition, image processing and data analysis and
visualization; 4) combining multiple senescence-associated markers (i.e. SA-β-Gal,
proliferation arrest, and enlarged morphology). The pipelines and data analysis app for
FAST are readily available, open-source, and support a GUI-based, user-friendly
installation and operation for a general biologist audience. A step-by-step protocol is
available at protocol.io: dx.doi.org/10.17504/protocols.io.kxygx3ypwg8j/v1).
We implemented image analysis for FAST in Image Analyst MKII because of the
microplate-level pipeline-based automation and the fast desktop parallel computing
offered by this solution. This software also acts as a GUI to launch all aspects of the
analysis. Image Analyst MKII allows manual exploration and well-by-well visualization
of cells scored as positive or negative, thus supporting explorative research work.
Sequential, unsupervised analysis of multiple microplates is supported by batch-based

78
analysis allowing high throughput applications. Pipelines used here are publicly
available and are in human-readable format. Hence, FAST may be also implemented
on open-source pipeline-based or scriptable platforms, such as Cell Profiler or
FIJI/ImageJ.
Published semi-automated methods42,44,58–60 fall short of FAST in several of its
innovative elements. For example, none of these methods standardized the SA-β-Gal
quantification by calculating OD, and, as far as we know, no method employed internal
controls to unbiasedly establish staining thresholds, which are instead arbitrarily
assigned. Furthermore, the automation component of these workflows was limited. In
fact, a few of these relied on manual segmentation58,60, and none of them described and
provided protocols for automated sample acquisition, data analysis, or visualization.
Finally, often only a single senescence-associated marker (SA-β-Gal) was measured
42,58,60, and methods measuring more than one marker had limited sample processing
capacity44,59
.
One limitation of our current work on FAST is that it was only optimized and
tested to analyze adherent cell culture samples, unlike other semi-automated, flow
cytometry-based workflows that can also analyze ex-vivo samples44,59
. However, this
limitation could be addressed in future versions of FAST, after adapting it for the
analysis of ex-vivo immobilized cell suspensions or tissue cryosections. In fact, the key
elements employed by FAST (OD SA-β-Gal measurement, background staining
controls, automation, and use of multiple senescence markers combined via ML) remain
applicable. In tissue sections, color deconvolution approaches61 can be used to convert
a color image into OD values specific to stains. A confounder of accurate determination

79
of SA-β-Gal positivity using FAST is that cellular processes and thicker cell bodies can
cause a non-specific increase in measured cellular OD not due to light scatter. We
mitigated these effects by 1) calculating staining threshold values for SA-β-Gal (and
EdU) scoring in a condition-specific manner, thus in control wells with similar cell
morphology; 2) applying a low-pass spatial filter during image processing to suppress
signal from thin cellular processes. An additional way of suppressing this non-specific
signal is increasing the refraction index of the medium during imaging with media such
as Optiprep, which has already been used for analyzing formazan OD for succinate
dehydrogenase activity cytochemistry62
.
The ML-FAST approach relies on a classifier trained for the particular experiment
using on-plate positive and negative controls, and therefore it is expected to adapt to a
broad range of biology. The random forest classifier can be trivially extended using
additional markers. The overall cell size is also enlarged in cellular senescence4
, along
with other morphological changes63–65
, and therefore these can contribute to the
identification of senescence phenotypes. However, the visible cell area is also highly
dependent on the growth surface available for each cell, therefore on cell density, and
this may inadvertently lead to biases. Altogether, ML-FAST can serve as a platform for
building more complex cellular senescence assays.
We showed that FAST is capable of distinguishing senescent cells in
experimental conditions that cause false-positive senescence staining, due to its
graded, quantitative response to staining intensity, thus reducing the intrinsic limitations
of the senescence-associated markers used here. We employed FAST across multiple
types of primary human cells – that is, endothelial cells, fibroblasts, and astrocytes (data

80
not shown) – and two different senescence-inducing stimuli – specifically, IR and Doxo.
This suggests that FAST can likely be used regardless of the cell type or senescenceinducing stimulus used. Moreover, we also show that the FAST workflow is compatible
with different automated microscopes.
To the best of our knowledge, we showed for the first time a single-cell-level
direct comparison of colorimetric (X-Gal) and fluorescent (C12FDG) SA-β-Gal staining.
These data confirm the utility of both X-Gal and C12FDG in detecting SA-β-Gal activity at
the population level. However, they also quantitatively demonstrate that these methods
are not interchangeable on the cellular, and perhaps molecular level due to the weak
correlation observed. Because C12FDG fluorescence was captured before X-Gal
staining, this latter could not optically interfere with measuring fluorescence intensity.
While X-Gal staining is commonly thought to be incompatible with fluorescence (except
for a few demonstrated applications41,44) due to its absorption and different microscopy
modality, here we showed that X-Gal is better suitable for multiplexed fluorescence
assays than C12FDG because of the redistribution of the latter during subsequent
staining. The X-Gal staining OD values observed at 692nm, near its absorption peak,
were in the range of 0.1-0.2. Below 550nm, its absorption is less than ~1/5th of the
peak66 and this equates to absorption of up to 4-8% green fluorescence signal, thus
causing little interference. FAST quantifies Alexa488-tagged EdU over the nuclei that
typically lack X-Gal staining, further diminishing the possibility of optical interference
between probes in this paradigm. For this reason, FAST is especially suitable for
combination with other fluorescence stains, especially over the nucleus, such as
fluorescence in situ hybridization (FISH) or immunocytochemistry (ICC). The staining

81
protocol for X-Gal is compatible with common FISH and ICC protocols, and analysis of
such stains, including spot counting, can be added to the image analysis pipeline and
FAST.R application presented here.
Interestingly, we observed that the IR endothelial cells that survived the ABT263
senolytic treatment showed the same fraction of senescence-associated marker
positivity as the untreated IR cells (Figure 15B and Figure 16). We know this is not due
to a lack of senolytic activity, as we confirmed that ABT263 preferentially eliminated IR
cells compared to control mock-IR cells (Figure 15C). Thus, ABT263 does not seem to
be preferentially targeting IR senescent endothelial cells with canonical senescence
staining (SA-β-Gal +, EDU -) compared to senescent cells with non-canonical staining.
This could be due to a lack of correlation in IR senescent endothelial cells between
senescence-associated staining and expression of ABT263 targets (anti-apoptotic Bcl-2
family proteins). Alternatively, ABT263 might be preferentially targeting the IR
senescent cells with canonical senescence staining (SA-β-Gal +, EDU -), but their death
might cause a cell non-autonomous cytotoxic effect on the other senescent cells as
well. This observation exemplifies how quantitative assaying could propel future studies
on mechanisms of cellular senescence and senolysis.
Benchmarks indicate that FAST is suitable for high-throughput screening.
Possible applications include the identification of environmental pollutants that might
exacerbate senescence burden42,67, chemical compound screens for senolytics or
compounds that prevent senescence induction, or validation and optimization of
senescence induction methods. Because FAST directly measures cell viability while
assessing multiple senescence-associated markers, this single assay provides

82
important controls and counter-screen data that can help the identification of
compounds with a specific action.
In summary, FAST has the potential to substantially advance senescence
research by offering a rapid, unbiased, and robust means to assess senescence burden
at the single-cell level. These qualities make FAST ideal for validating senescence
induction in cell culture models and aiding the study of senescence heterogeneity.
Chapter 3, which describes the use of a larger high-content imaging workflow to study
senescent cell heterogeneity, exemplifies how FAST can be used to this end.

83
Chapter 3: High-Content Image Analysis Identifies DNA Content
As A Key Contributor To Senescent Cell Heterogeneity And A
Determinant Of Senolytic Response
3.1. Introduction
Cellular senescence is a complex stress response, generally characterized by an
essentially irreversible cell cycle arrest, altered morphology, increased lysosomal
activity, macromolecular damage, and profound changes in gene expression, such as
the acquisition of a senescence-associated secretory phenotype (SASP)4
. Genetic and
pharmacological interventions that promote the elimination of senescent cells, a process
termed senolysis, have been shown to benefit the healthspan and lifespan of
mice36,57,68
. Because of their promising results in animal models, senolytic therapies
have now entered clinical trials38,39
. Notably, a senolytic drug tested in patients with
diabetic macular degeneration, UBX132569
, was recently shown to improve vision for up
to 48 weeks after a single drug injection in a Phase 2 clinical trial40
. While current
treatments hold promises to improve some age-related pathologies, much more work is
needed to develop safe and effective senolytics that can significantly improve human
healthspan. An important challenge in the development of such therapeutics is the
heterogeneity of senescent cells, which is still poorly understood70
.
Senescent cells are highly heterogeneous, as many different cell types can
become senescent due to a variety of stressors capable of inducing this cell state.
Indeed, it was shown that senescent cell culture models based on diverse cell types and
senescence inducers resulted in senescent cells with different SASP7,8
. These different

84
secretomes are the direct consequence of distinct gene expression profiles across
these heterogeneous senescent cell populations20
. However, such studies involved the
use of bulk techniques, and therefore could not analyze the heterogeneity within a given
senescent population. Recently, single-cell RNA-sequencing (scRNA-Seq) has been
used to analyze the diversity within senescent cell populations in culture71–73
. Despite
the highly controlled experimental conditions, these studies have highlighted significant
heterogeneity, suggesting the existence of senescent subpopulations in different cell
states with distinct biology. However, these experiments were limited to the description
of these senescent subpopulations. Further research is therefore needed to
characterize the functional differences between senescent subpopulations.
Previous studies that employed bulk techniques have shown that senolytics have
different efficacies across senescent cell culture models (obtained by using different cell
types and senescence inducers)
57,74
. However, it is unknown whether there is
heterogeneity in senolytic responses within senescent cell populations. While this
seems likely because of the existence of heterogeneous subpopulations highlighted by
previous single-cell studies, no evidence of this is yet available. Understanding whether
heterogeneous subpopulations of senescent cells indeed respond differently to
senolytics is critical for the development of the next generation of senolytic therapies.
Unlike scRNA-Seq, high-content imaging enables rapid and cost-effective
measurements of several senescence markers at a single-cell level. Even though
imaging can limit the number of markers assessed at once, it allows measurements at
the protein level – one step closer to function than RNA. Additionally, imaging can
readily assess cell viability and hence response to senolytics. Thus, we set out to

85
identify functionally distinct senescent subpopulations that might respond differently to
senolytics using a high-content imaging workflow.
3.2. Methods
Cell Culture
Primary human lung microvascular endothelial cells (HMVEC-L) were purchased
from Lonza (CC-2527). HMVEC-L were cultured in EGMTM-2MV Microvascular
Endothelial Cell Growth Medium-2 BulletKitTM (Lonza, CC-3202) at 37°C, 14% O2, 5%
CO2. Human lung fibroblasts IMR-90 were purchased from Coriell Institute (I90). IMR90 cells were cultured in DMEM (Corning, 01-017-CV) supplemented with 10% FBS
(R&D Systems, S11550H), 100 units/mL penicillin, and 100 µg/mL streptomycin (R&D
Systems, B21210) at 37°C, 3% O2, 10% CO2. For all experiments, both HMVEC-L and
IMR-90 were cultured in 96-well microplates appropriate for microscopy imaging
(Corning, 3904), with media changes every 2-3 days.
To achieve serum starvation in HMVEC-L, cells were washed twice in DPBS
containing Ca2+ and Mg2+ (Gibco, 14040-117) and then cultured for 72 h in low-serum
EGMTM-2MV medium (0.5% FBS instead of 5% FBS). To achieve serum starvation in
IMR-90, cells were washed twice in DPBS containing Ca2+ and Mg2+ (Gibco, 14040-117)
and then cultured for 72 h in low-serum DMEM medium (0.2% FBS instead of 10%
FBS).
To achieve G1-enrichment in HMVEC-L, cells were washed twice in DPBS
containing Ca2+ and Mg2+ (Gibco, 14040-117) and then cultured for 48 h in low-serum
EGMTM-2MV medium (0.5% FBS instead of 5% FBS) followed by 24 h in serum-free

86
EGMTM-2MV medium. To achieve G2 enrichment in HMVEC-L, cells were seeded 40 h
before irradiation was carried out.
Senescence Induction
Senescence was induced as previously described75
. Briefly, cells were irradiated
with 15 Gy, and medium change was performed immediately after treatment. Cells
were considered senescent after at least 7 days since irradiation, during which medium
was regularly changed (every 2-3 days).
SA-β-Gal and EdU Staining
SA-β-Gal and EdU staining was performed following the Fully-Automated
Senescence Test (FAST) workflow76
. Briefly, commercially available kits were used to
perform SA-β-Gal (Abcam, ab65351) and EdU staining (ThermoFisher Scientific,
C10351). For EdU, cells were given medium containing 2.5 µM EdU 24 h before
fixation. After 24 h, cells were fixed by adding 8% PFA in PBS pre-warmed to 37°C
directly to the medium up to a final concentration of 4% PFA and incubated for 15 min at
RT. Subsequently, cells were washed twice with PBS, and SA-β-Gal staining was
performed.
To stain for SA-β-Gal, fixed cells were treated with the staining solution mix as
recommended by the manufacturer (Abcam, ab65351). However, the X-Gal powder
used was separately purchased (Life Technologies, 15520-018). Staining was
performed overnight at 37°C in an incubator at atmospheric CO2 conditions. To prevent
nonspecific indole crystal formation, empty spaces in between wells of the microplates
were filled with PBS, and parafilm was used to seal the microplates before the overnight

87
incubation. After the overnight incubation, cells were washed twice with PBS to stop the
staining.
After SA-β-Gal staining, EdU detection was performed. Briefly, cells were
permeabilized at room temperature for 15 min with 0.5% Triton X-100 (Millipore Sigma,
T9284-500ML) in PBS. After permeabilization, the Click-iT Reaction Cocktail was
added as per the user manual, and cells were incubated for 30 min in the dark. After
the incubation, cells were washed once with PBS, counterstained with 0.5 μg/mL DAPI
in MilliQ water for 30 min at room temperature in the dark, and then washed once with
MilliQ water. Finally, wells were covered with PBS and imaged.
Immunocytochemistry Staining
Immunocytochemistry staining was performed using standard protocols. Briefly,
cells were first fixed and permeabilized as described in the SA-β-Gal and EdU staining
section above. Then, cells were incubated with 10% goat serum for 1 h for blocking,
incubated with primary antibodies (see Table 1) overnight at 4°C, washed 3 times with
PBS, and incubated with secondary antibodies (see Table 1) for 1 h at room
temperature in the dark. Afterward, samples were washed once with PBS,
counterstained with 0.5 μg/mL DAPI in MilliQ water for 30 min at room temperature in
the dark, and then washed once with MilliQ water. Finally, wells were covered with PBS
and imaged.

88
Table 1: Antibody list
Antibody Company Catalog # Dilution
HMGB1 Abcam ab18256 1000
LaminB1 Abcam ab229025 1000
p21_Abcam Abcam ab109520 100
p21_CellSign Cell Signaling Technology 2946 850
yH2AX_Abcam Abcam ab81299 250
yH2AX_CellSign Cell Signaling Technology 80312 200
Goat anti-Mouse
AlexaFluor488
Invitrogen A11001 2000
Goat anti-Rabbit
AlexaFluor555
Invitrogen A27039 2000
ELISA
2 days before analysis, cells were given fresh medium. Conditioned medium
(CM) was collected 48 h later, after which cells were fixed in 4 % PFA for 15 min at room
temperature, washed twice with PBS, covered with fresh PBS, and microplates were
stored at 4°C. Each sample for downstream ELISA analysis was obtained by pooling
CM from at least three 96-well microplate wells of the same group (CTL, IR-G1-E, or IRG2-E). To remove potential cell debris, CM was then centrifuged at 10,000 g for 1 min
and the supernatant was moved to a new tube. A human IL-6 ELISA kit (Thermo Fisher
Scientific, KHC0061) was then used as instructed by the manufacturer to measure IL-6
concentrations. To normalize IL-6 secretion, fixed cells were washed twice with PBS,
permeabilized with 0.5% Triton X-100 at room temperature for 15 min, washed twice
with PBS, counterstained with 0.5 μg/mL DAPI in MilliQ water for 30 min at room
temperature in the dark, then washed once with MilliQ water. Finally, wells were

89
covered with PBS and imaged. DAPI staining was used to obtain cell counts per well,
which were then used to normalize IL-6 concentrations of all samples.
Senolytic Treatment
On the last day before analysis, cells were treated with the senolytic
ABT263/Navitoclax (Selleck Chemicals, S1001) at different concentrations for 24 h,
while only vehicle (DMSO) was given as mock treatment.
Image Acquisition
Wide-field microscopy was performed on a Nikon Eclipse Ti-PFS fully motorized
microscope controlled by NIS Elements AR 5.21 (Nikon, Melville, NY). The setup
comprised a Lambda 10-3 emission filter wheel, a SmartShutter in the brightfield light
path, and a 7-channel Lambda 821 (Sutter Instruments, Novato, CA) LED
epifluorescence light source with excitation filters on the LEDs, controlled by a PXI 6723
DAQ (NIDAQ; National Instruments) board. Images were acquired by an Andor iXon
Life 888 EMCCD camera (Oxford Instruments) using 10-100 ms exposure times, with a
Nikon S Fluor 20× DIC NA=0.5 lens. To image SA-β-Gal staining, an incandescent
Koehler illumination was used, and a 692/40 “emission” filter. The Kohler condenser
was carefully focused for each experiment in the center of a well, with the aperture
diaphragm semi-open. 5×5 tiled images were recorded without overlap or registration,
using the full 1024×1024 resolution of the camera (1.3 µm/pixel). For autofocusing, the
Nikon’s Perfect Focus System was used. Data were saved as native *.nd2 files for SAβ-Gal and EdU staining, or as *.tiff files for immunocytochemistry staining.
Image Processing

90
For SA-β-Gal and EdU images, native format *.nd2 image files were opened in
Image Analyst MKII (Image Analyst Software, Novato, CA). Analysis was performed
using modified standard and custom pipelines described in the FAST workflow76
. The
output Excel file containing single-cell measurements for each whole microplate was
saved, and further data analysis was performed in R. For immunocytochemistry
images, *.tiff image files were analyzed in MATLAB.
Data Analysis
Data were analyzed either with custom R pipelines, available on GitHub
(https://github.com/f-neri/Wirtz-collaboration), or MATLAB pipelines. For SA-β-Gal and
EdU data, raw single-cell measurements were first pre-processed using an R Shinybased application, FAST.R76
.
Statistics
Statistical tests employed are exhaustively described in each figure legend.
Such statistical tests were either performed in R (v4.3.2) or MATLAB (R2023b).
3.3. Results
Validation of senescence and population-level heterogeneity
To study senescence heterogeneity, the expression of several senescenceassociated markers was measured at the single-cell level by using a high-content image
analysis workflow (Figure 18). As a senescence model, primary human endothelial cells
(HMVEC-L) and fibroblasts (IMR-90) were used and made senescent using ionizing
radiation (IR) (Figure 18A). Both senescent (SEN) and their control samples (CTL)
were either cultured in full-serum medium for the entire experiment (FS) or were

91
switched to low-serum medium for the last 3 days of culture to induce quiescence by
serum-starvation in CTL cells (SS). SEN and CTL samples were subsequently costained either for SA-β-Gal and proliferation (EdU incorporation) or for other
senescence markers (yH2AX, LaminB1, HMGB1, p21) via immunocytochemistry (ICC)
(Figure 18B). Then, an automated microscope was employed to acquire thousands of
images, which were further processed using image analysis software to generate tens
of thousands of single-cell measurements (Figure 18C).
Figure 18. High-content imaging workflow.
A) Sample preparation. Human lung primary microvascular endothelial cells (HMVEC-L)
and fibroblasts (IMR-90) were induced to senescence using ionizing radiation (IR). IR
and mock-irradiated cells (CTL) were either cultured in full-serum medium the entire time
(FS) or switched to low-serum medium for the last 3 days of culture to induce quiescence
in CTL cells (SS). B) Staining for senescence markers. Prepared samples were either
co-stained for senescence-associated beta-galactosidase activity (SA-β-Gal) and
proliferation via EdU incorporation (EdU); or for other senescence markers (γH2AX,
LaminB1, HMGB1, p21) using immunocytochemistry (ICC). C) High-content image
analysis was performed to identify senescent subpopulations.
First, this dataset was used to assess senescence-associated markers at the
well or population level (Figure 19). As expected, SA-β-Gal was higher in IR cells
compared to their respective CTL cells and EdU incorporation was lower (Figure 19B).
Notably, the EdU signal was not higher in CTL samples in SS conditions compared to IR
cells, indicating the SS CTL cells were indeed quiescent. An increased number of
nuclear yH2AX foci and higher p21 nuclear levels were observed using ICC staining,
while both LaminB1 and HMGB1 nuclear staining were lower in IR compared to CTL

92
cells. This was true both when comparing IR cells to proliferating CTL cells (FS
conditions) and IR cells to quiescent CTL cells (SS conditions). Taken together, these
data confirm the senescence induction in IR samples. Additionally, differences in
marker expression were observed between irradiated HMVEC-L endothelial cells and
IMR-90 fibroblasts, i.e., SA-β-Gal was much higher in IR HMVEC-L compared to IR
IMR-90, while p21 levels were lower. This observation demonstrates the existence of
heterogeneity between senescent populations that differ on the cell type investigated.

93

94
Figure 19. Validation of senescence induction and population-level heterogeneity.
A) Representative images of senescence marker staining from full-serum (FS) samples.
Top: mock-irradiated cells (CTL); bottom: ionizing-radiation-induced senescent cells (IR).
For each co-staining, endothelial cells (HMVEC-L) are shown on the left; fibroblasts (IMR90) are shown on the right. B-D) Image quantification of HMVEC-L and IMR-90 cells for
both FS and serum-starved (SS) conditions. CTL samples are in green, while IR samples
are in purple. Data shown are from 2 independent experiments. B) SA-β-Gal (left) and
EdU (right) staining quantification. Each data point corresponds to one well (n = 18); bars
indicate mean values. C) Immunocytochemistry staining quantification. Top-left: γH2AX;
top-right: p21; bottom-left: LaminB1; bottom-right: HMGB1. Each data point corresponds
to one well (γH2AX n = 27; p21, LaminB1, and HMGB1 n = 9); bars indicate mean values.
D) Nuclear morphology feature quantification. Left: nuclear area; right: shape factor.
Each data point corresponds to one well (n = 27); bars indicate mean values. ***: p-value
< 10-3
; ****: p-value < 10-4
; non-significant values (p-value > 0.05) are shown.
G2-arrested senescent cells express higher levels of senescence markers
than G1-arrested senescent cells
After comparing population-level data, an exploratory analysis was performed
within senescent populations using single-cell measurements of the markers assayed
(Figure 20). For several of the co-stains performed, two subpopulations of senescent
endothelial cells were observed with distinct senescence marker expression (Figure
20A). We hypothesized that these differences might be related to the phase of the cell
cycle at which senescent cells were arrested. To test this hypothesis, senescence
marker staining was analyzed in relation to DNA content (measured via DAPI staining),
separating senescent cells into G1- and G2-arrested subpopulations (Figure 20B).
Indeed G1 and G2 senescent cells expressed different senescence marker levels, with
G2-arrested senescent cells showing higher marker staining compared to G1-arrested
cells (Figure 20C). Additionally, plotting G1 and G2 senescent cells separately resulted
in uniformly stained subpopulations (Figure 20D), suggesting DNA content could be the
main driver of the observed heterogeneity at the population level.

95
a
p21 Intensity
FS S S
CTL IR
H2AX Intensity
LaminB1 Intensity
FS S S
CTL IR
p21 Intensity
p21 Intensity
CTL IR
G1 G2
H2AX Intensity
LaminB1 Intensity
CTL IR
G1 G2
H2AX Intensity
HMGB1 Intensity
CTL IR
G1 G2
H2AX Intensity
LaminB1 Intensity
CTL IR
G1 G2
p21 Intensity
HMGB1 Intensity
CTL IR
G1 G2
p21 Intensity
LaminB1 Intensity
FS S S
CTL IR
H2AX Intensity
HMGB1 Intensity
FS S S
CTL IR
H2AX Intensity
HMGB1 Intensity
FS S S
CTL IR
p21 Intensity
b
c
d
H2AX Intensity p21 Intensity LaminB1 Intensity HMGB1 Intensity
Figure 3 TOTAL INTENSITY
****
****
****
****
****
****
**** ****
CTL IR
G1 G2 G1 G2

96
Figure 20. G1 and G2 senescent cells express different levels of senescence markers.
A) Single-cell scatterplots showing co-staining data of immunocytochemistry markers
measured. FS: full serum; SS: serum starved; CTL: non-senescent control cells; IR:
irradiated senescent cells. B) Histogram showing G1 and G2 peaks, distinguishable via
DAPI staining. C) Violin plots with mean marker intensity per well; only FS data is shown.
For both non-senescent (Mock, left) and senescent cells (IR, right), mean marker intensity
was calculated for all (Total, blue), G1-only (red), or G2-only cells (dark grey). Data shown
is from 2 independent experiments (γH2AX n = 27; p21, LaminB1, and HMGB1 n = 9).
D) Single-cell scatterplots showing co-staining data of cells separated into G1 and G2
subpopulations; only FS data is shown. ****: p-value < 10-4 by Mann-Whitney nonparametric test.
G1- and G2-arrested senescent endothelial cells have different IL-6
secretion and ABT263 susceptibility
Based on the differences in senescence marker expression, we hypothesized
that G1- and G2-arrested senescent cells might also be functionally distinct. To test this
hypothesis, we focused on senescent endothelial cells, where differences between G1-
and G2-arrested cells were more prominent.
First, protocols were developed to enrich senescent endothelial cell populations
in G1 or G2 (Figure 21). We hypothesized that enriching cells in G2 or G1 just before
irradiation would result in senescent populations enriched in G2 or G1 respectively.
Thus, G2- and G1-enriched populations were generated by either precisely timing cell
seeding to obtain cells in their exponential growth phase (IR-G2-E) or using serumstarved culturing conditions (IR-G1-E), respectively (Figure 21A). To compare the DNA
content of senescent cells with that of cells at the time of irradiation, additional samples
were prepared in parallel, which were fixed instead of being irradiated (CTL-G2-E, CTLG1-E). As expected, the G2-E samples had a higher percentage of G2 cells than the
G1-E samples (Figure 21B). This was the case both at the time of irradiation (CTL
samples) and, more importantly, in senescent populations (IR samples). Interestingly,

97
the relative difference in the percentage of G2 cells between G2-E and G1-E samples
was about half for both CTL and IR samples (Figure 21C).
Figure 21. G1 and G2 enrichment protocol for senescent endothelial cells.
A) Workflow to compare DNA content in cells just before irradiation (CTL) and senescent
cells 10 days after irradiation (IR) when enriched for either G1 (G1-E) or G2 (G2-E) cells.
B) Percentage of G1 and G2 cells per well. Each data point is a well (n = 30); bars indicate
mean values. C) Fold change of G2 percentages between G2-E vs G1-E groups. Each
data point is a well (n = 30); bars indicate mean values.
Upon establishing that DNA content enrichment at the time of senescence
induction was maintained after cells became fully senescent, we proceeded to compare
G2- and G1-enriched senescent populations (Figure 22). First, their IL-6 secretion
levels were compared (Figure 22A). For this purpose, IR-G2-E and IR-G1-E samples
were generated. Conditioned medium (CM) was collected over the last two days of
culture, and then the samples were fixed and imaged. Using the imaging data, we
validated the expected DNA content enrichment (Figure 22B-C). IL-6 secretion in the

98
CM was measured by ELISA and normalized to cell counts obtained using imaging. IL6 levels were higher in IR-G2-E compared to IR-G1-E samples (Figure 22D). Then, the
sensitivity to ABT263 senolytic treatment was compared in IR-G2-E and IR-G1-E by
measuring cell counts obtained via imaging (Figure 22E). DNA content enrichment was
validated (Figure 22F), and viability was compared across different ABT263
concentrations (Figure 22G). Differences in ABT263 sensitivity were observed between
IR-G2-E and IR-G1-E populations in two out of the three concentrations tested (0.33
and 1.00 µM). The same data were further analyzed to measure differences in viability
within IR-G2-E and IR-G1-E populations by comparing their G1 and G2 subpopulations
(Figure 22H). Differences in ABT263 sensitivity were observed between G1 and G2
subpopulations at all three concentrations tested (including 0.11 µM) both in the IR-G2-
E and IR-G1-E populations. Thus, G2-arrested senescent endothelial cells secreted
higher levels of IL-6 and were more sensitive to ABT263 senolytic treatment than G1-
arrested cells. This suggests the existence of functionally distinct senescent cell
subpopulations, which underscores the importance of considering senescence
heterogeneity during the development of senotherapeutics.

99

100
Figure 22. G1 and G2 senescent endothelial cells show different levels of IL-6 secretion
and ABT263 susceptibility.
A) Workflow for the comparison of IL-6 secretion in G1 vs G2 senescent endothelial cells.
Three conditions were prepared: non-senescent mock-irradiated cells (CTL, green);
ionizing radiation-induced senescent cells enriched in G2 (IR-G2-E, purple); ionizing
radiation-induced senescent cells enriched in G1 (IR-G1-E, orange). Condition media
(CM) were collected from the last 2 days of culture, after which cells were fixed,
counterstained with DAPI, and imaged. IL-6 concentration was quantified by ELISA and
normalized to cell counts. B) DNA content distribution histogram, showing G1 (light grey)
and G2 (dark grey) peaks in IR-G2-E and IR-G1-E senescent populations. The plot
shows all IR-G2-E and IR-G1-E cells from a single representative experiment. C)
Quantification of (B), showing the sample percentages of senescent cells arrested in G1
and G2. Data shown is from 3 independent experiments; each data point is a sample
(CTL n = 12; IR-G1-E and IR-G2-E n = 16). D) IL-6 secretion across CTL, IR-G2-E, and
IR-G1-E groups normalized to cell counts. Data shown are from 3 independent
experiments; each data point is a sample (CTL n = 12; IR-G1-E and IR-G2-E n = 16). ***:
p-value < 10-3
; ***: p-value < 10-4
; by one-way ANOVA followed by post-hoc Tukey’s test.
E) Workflow for the comparison of ABT263 susceptibility in G1 vs G2 senescent
endothelial cells. CTL, IR-G2-E, and IR-G1-E were prepared as described in (A), but cells
were treated with ABT263 for the last 24 h of culture. After treatment, cells were fixed,
counterstained with DAPI, and imaged. F) Percentages of senescent cells per well
arrested in G1 and G2. Data shown are from 3 independent experiments; each data point
is a well (n = 30). G) Cell viability comparison after ABT263 treatment between IR-G2-E
and IR-G1-E senescent populations, measured by cell counts normalized to vehicle (0.00
µM ABT263). Data shown are mean ± SEM for each ABT263 concentration from 3
independent experiments. Viability was compared between IR-G2-E and IR-G1-E
populations across all ABT263 concentrations (n = 30). ns: p-value > 0.05; *: p-value <
0.05; **: p-value < 0.01; by non-parametric Mann-Whitney test corrected for multiple
comparisons by FDR method. H) Cell viability comparison of G1 and G2 subpopulations
within IR-G2-E and IR-G1-E populations from (G). Data shown are mean ± SEM of G1
(light grey) and G2 (dark grey) subpopulations for each ABT263 concentration from 3
independent experiments. Viability was compared between G1 and G2 cells across all
ABT263 concentrations (n = 30). **: p-value < 10-2
; ***: p-value < 10-3
; ****: p-value < 10-
4
; by non-parametric Mann-Whitney test corrected for multiple comparisons by FDR
method.
3.4. Discussion
In this study, we aimed to identify functionally distinct subpopulations of
senescent cells by using high-content imaging workflows. Specifically, our goal was to
establish whether we could identify subpopulations of senescent cells with
heterogeneous sensitivity to senolytic treatment.

101
By leveraging cell culture senescence models and analyzing single-cell
measurements of several senescence-associated markers, we found a relationship
between marker expression and DNA content. Specifically, we observed that G2-
arrested senescent cells had higher levels of senescence markers than G1-arrested
cells. Additionally, we found that G2-arrested senescent cells secreted higher levels of
IL-6 and were more sensitive to ABT263 senolytic treatment. This suggests that DNA
content can differentiate senescent subpopulations with distinct functions, as highlighted
by senescence marker expression, SASP factor secretion, and drug response. While
our study focused on ABT263, we speculate that the cytotoxic effect of other senolytic
compounds might be heterogeneous and depend on the DNA content of the treated
cells. Interestingly, previous studies in cancer cells have highlighted that the cytotoxic
profile of several drugs is influenced by the DNA content and cell cycle phase at the
time of treatment, with some drugs preferentially targeting cells in G1 and others in
G277
.
It is important to note that this study focused solely on the analysis of two cell
culture models (primary human endothelial cells and fibroblasts) of DNA-damageinduced senescence. Future studies should determine whether our findings are
generalizable to other cell culture models (with different cell types and senescence
inducers) and – most importantly – to senescent cells in vivo. While our study focused
on senolytics, other types of senotherapeutics are being developed, such as
senomorphics. Senomorphics can alleviate senescence-related tissue dysfunction by
attenuating the SASP, rather than eliminating senescent cells39
. Future studies should
address whether senomorphics display heterogeneity between senescent

102
subpopulations too, and whether such heterogeneity is tied to the same factors that
influence the response to senolytics (i.e. DNA content).
Overall, we demonstrated the existence of functionally distinct senescent
subpopulations in culture, which can be differentiated based on G1 and G2 DNA content
(Figure 23). To the best of our knowledge, our data also constitute the first evidence of
heterogenous senolytic response between subpopulations of senescent cells. These
findings highlight the importance of studying senescent cell heterogeneity and that their
diversity should be considered when developing senolytic treatments.
Figure 23. Senescent heterogeneity model based on DNA content.
Compared to G1-arrested cells, G2-arrested senescent cells express higher levels of
senescence-associated markers, secrete more IL-6, and are more sensitive to senolytic
treatment.
In summary, the research described in my thesis provides novel protocols and
tools to assess senescence burden and study senescence heterogeneity, both between
and within senescent cell populations. Using these tools, I started elucidating which
factors contribute to senescent cell heterogeneity and showed that such heterogeneity

103
might also affect response to senotherapeutics. I speculate the methods here
described will be broadly useful to the aging field by enabling further characterization of
cellular senescence and its heterogeneity.

104
Bibliography
1. Beausejour, C. M. et al. Reversal of human cellular senescence: roles of the p53
and p16 pathways. EMBO J 22, 4212–22 (2003).
2. Lee, S. & Schmitt, C. A. The dynamic nature of senescence in cancer. Nat Cell Biol
21, 94–101 (2019).
3. Campisi, J. & d’Adda di Fagagna, F. Cellular senescence: when bad things happen
to good cells. Nat Rev Mol Cell Biol 8, 729–40 (2007).
4. Gorgoulis, V. et al. Cellular Senescence: Defining a Path Forward. Cell 179, 813–
827 (2019).
5. Campisi, J. Aging, cellular senescence, and cancer. Annu Rev Physiol 75, 685–705
(2013).
6. Munoz-Espin, D. & Serrano, M. Cellular senescence: from physiology to pathology.
Nat Rev Mol Cell Biol 15, 482–96 (2014).
7. Coppé, J.-P. et al. Senescence-Associated Secretory Phenotypes Reveal CellNonautonomous Functions of Oncogenic RAS and the p53 Tumor Suppressor.
PLoS Biol 6, e301 (2008).
8. Basisty, N. et al. A proteomic atlas of senescence-associated secretomes for aging
biomarker development. PLOS Biology 18, e3000599 (2020).
9. Schafer, M. J. et al. The senescence-associated secretome as an indicator of age
and medical risk. JCI Insight 5, (2020).
10. Wiley, C. D. et al. SILAC Analysis Reveals Increased Secretion of HemostasisRelated Factors by Senescent Cells. 28, 3329–3337 (2019).
11. Borghesan, M. et al. Small Extracellular Vesicles Are Key Regulators of Non-cell
Autonomous Intercellular Communication in Senescence via the Interferon Protein
IFITM3. Cell Rep 27, 3956-3971 e6 (2019).
12. Jeon, O. H. et al. Senescence cell-associated extracellular vesicles serve as
osteoarthritis disease and therapeutic markers. JCI Insight 4, (2019).
13. Ozcan, S. et al. Unbiased analysis of senescence associated secretory phenotype
(SASP) to identify common components following different genotoxic stresses.
Aging (Albany NY) 8, 1316–29 (2016).
14. James, E. N. L., Bennett, M. H. & Parkinson, E. K. The induction of the fibroblast
extracellular senescence metabolome is a dynamic process. Sci Rep 8, 12148–
12148 (2018).

105
15. Basisty, N., Kale, A., Patel, S., Campisi, J. & Schilling, B. The power of proteomics to
monitor senescence-associated secretory phenotypes and beyond: toward clinical
applications. Expert Rev Proteomics 17, 297–308 (2020).
16. Collins, B. C. et al. Multi-laboratory assessment of reproducibility, qualitative and
quantitative performance of SWATH-mass spectrometry. Nat Commun 8, 291
(2017).
17. Gillet, L. C. et al. Targeted data extraction of the MS/MS spectra generated by dataindependent acquisition: a new concept for consistent and accurate proteome
analysis. Mol Cell Proteomics 11, O111 016717 (2012).
18. Schilling, B., Gibson, B. W. & Hunter, C. L. Generation of High-Quality SWATH((R))
Acquisition Data for Label-free Quantitative Proteomics Studies Using
TripleTOF((R)) Mass Spectrometers. Methods Mol Biol 1550, 223–233 (2017).
19. Gonzalez-Gualda, E., Baker, A. G., Fruk, L. & Munoz-Espin, D. A guide to assessing
cellular senescence in vitro and in vivo. FEBS J (2020) doi:10.1111/febs.15570.
20. Hernandez-Segura, A. et al. Unmasking Transcriptional Heterogeneity in Senescent
Cells. Curr Biol 27, 2652-2660 e4 (2017).
21. Rosenberger, G. et al. A repository of assays to quantify 10,000 human proteins by
SWATH-MS. Sci Data 1, 140031 (2014).
22. Demaria, M. et al. An Essential Role for Senescent Cells in Optimal Wound Healing
through Secretion of PDGF-AA. Developmental Cell 31, 722–733 (2014).
23. Coppe, J. P., Desprez, P. Y., Krtolica, A. & Campisi, J. The senescence-associated
secretory phenotype: the dark side of tumor suppression. Annu Rev Pathol 5, 99–
118 (2010).
24. Jeon, O. H., David, N., Campisi, J. & Elisseeff, J. H. Senescent cells and
osteoarthritis: a painful connection. J Clin Invest 128, 1229–1237 (2018).
25. Harley, C. B., Futcher, A. B. & Greider, C. W. Telomeres shorten during ageing of
human fibroblasts. Nature 345, 458–460 (1990).
26. Bodnar, A. G. et al. Extension of Life-Span by Introduction of Telomerase into
Normal Human Cells. Science 279, 349–352 (1998).
27. Leonardo, A. D., Linke, S. P., Clarkin, K. & Wahl, G. M. DNA damage triggers a
prolonged p53-dependent G1 arrest and long-term induction of Cip1 in normal
human fibroblasts. Genes Dev. 8, 2540–2551 (1994).
28. Serrano, M., Lin, A. W., McCurrach, M. E., Beach, D. & Lowe, S. W. Oncogenic ras
Provokes Premature Cell Senescence Associated with Accumulation of p53 and
p16INK4a. Cell 88, 593–602 (1997).

106
29. Di Micco, R. et al. Oncogene-induced senescence is a DNA damage response
triggered by DNA hyper-replication. Nature 444, 638–642 (2006).
30. Bartkova, J. et al. Oncogene-induced senescence is part of the tumorigenesis
barrier imposed by DNA damage checkpoints. Nature 444, 633–637 (2006).
31. Wiley, C. D. et al. Mitochondrial Dysfunction Induces Senescence with a Distinct
Secretory Phenotype. Cell Metabolism 23, 303–314 (2016).
32. Muñoz-Espín, D. et al. Programmed Cell Senescence during Mammalian Embryonic
Development. Cell 155, 1104–1118 (2013).
33. Storer, M. et al. Senescence Is a Developmental Mechanism that Contributes to
Embryonic Growth and Patterning. Cell 155, 1119–1130 (2013).
34. Dimri, G. P. et al. A biomarker that identifies senescent human cells in culture and in
aging skin in vivo. 92, 9363–9367 (1995).
35. Idda, M. L. et al. Survey of senescent cell markers with age in human tissues. Aging
12, 4052–4066 (2020).
36. Xu, M. et al. Senolytics Improve Physical Function and Increase Lifespan in Old
Age. Nature medicine 24, 1246 (2018).
37. Muñoz-espín, D. & Serrano, M. Cellular senescence : from physiology to pathology.
Nature Publishing Group 15, 482–496 (2014).
38. Di Micco, R., Krizhanovsky, V., Baker, D. & d’Adda di Fagagna, F. Cellular
senescence in ageing: from mechanisms to therapeutic opportunities. Nature
Reviews Molecular Cell Biology 2020 22:2 22, 75–95 (2020).
39. Chaib, S., Tchkonia, T. & Kirkland, J. L. Cellular senescence and senolytics: the path
to the clinic. Nat Med 28, 1556–1568 (2022).
40. UNITY Biotechnology Announces Positive 48-Week Results from Phase 2 BEHOLD
Study of UBX1325 in Patients with Diabetic Macular Edema | Unity Biotechnology.
https://ir.unitybiotechnology.com/news-releases/news-release-details/unitybiotechnology-announces-positive-48-week-results-phase-2/.
41. Kohli, J. et al. Algorithmic assessment of cellular senescence in experimental and
clinical specimens. Nat Protoc 16, 2471–2498 (2021).
42. Woods, G. & Andersen, J. K. Screening Method for Identifying Toxicants Capable of
Inducing Astrocyte Senescence. Toxicological Sciences 166, 16–24 (2018).
43. Krzystyniak, A., Gluchowska, A., Mosieniak, G. & Sikora, E. Fiji-Based Tool for Rapid
and Unbiased Analysis of SA-β-Gal Activity in Cultured Cells. Biomolecules 13, 362
(2023).

107
44. Biran, A. et al. Quantitative identification of senescent cells in aging and disease.
Aging Cell 16, 661–671 (2017).
45. Francesco Neri, Nathan Basisty, Pierre-Yves Desprez, Judith Campisi, and B. S.
Quantitative Proteomic Analysis of the Senescence-Associated Secretory
Phenotype by Data-Independent Acquisition.
46. Chang, W. et al. shiny: Web Application Framework for R. (2023).
47. Kuhn, M. Building Predictive Models in R Using the caret Package. Journal of
Statistical Software 28, 1–26 (2008).
48. Gerencser, A. A., Doczi, J., Töröcsik, B., Bossy-Wetzel, E. & Adam-Vizi, V.
Mitochondrial Swelling Measurement In Situ by Optimized Spatial Filtering:
Astrocyte-Neuron Differences. Biophysical Journal 95, 2583–2598 (2008).
49. Stringer, C., Wang, T., Michaelos, M. & Pachitariu, M. Cellpose: a generalist
algorithm for cellular segmentation. Nat Methods 18, 100–106 (2021).
50. Pachitariu, M. & Stringer, C. Cellpose 2.0: how to train your own model. Nat
Methods 19, 1634–1641 (2022).
51. Kurz, D. J., Decary, S., Hong, Y. & Erusalimsky, J. D. Senescence-associated (beta)-
galactosidase reflects an increase in lysosomal mass during replicative ageing of
human endothelial cells. Journal of Cell Science 113, 3613–3622 (2000).
52. Martínez-Zamudio, R. I. et al. Senescence-associated β-galactosidase reveals the
abundance of senescent CD8+ T cells in aging humans. Aging Cell 20, e13344
(2021).
53. Flor, A. C., Doshi, A. P. & Kron, S. J. Modulation of therapy-induced senescence by
reactive lipid aldehydes. Cell Death Discov 2, 16045 (2016).
54. Maejima, Y., Adachi, S., Ito, H., Hirao, K. & Isobe, M. Induction of premature
senescence in cardiomyocytes by doxorubicin as a novel mechanism of myocardial
damage. Aging Cell 7, 125–136 (2008).
55. Piegari, E. et al. Doxorubicin induces senescence and impairs function of human
cardiac progenitor cells. Basic Res Cardiol 108, 334 (2013).
56. Zhang, J.-H. & Oldenburg, K. R. Z-Factor. in Encyclopedia of Cancer (ed. Schwab,
M.) 4885–4887 (Springer, Berlin, Heidelberg, 2017). doi:10.1007/978-3-662-46875-
3_6298.
57. Chang, J. et al. Clearance of senescent cells by ABT263 rejuvenates aged
hematopoietic stem cells in mice. Nat Med 22, 78–83 (2016).

108
58. Shlush, L. I. et al. Quantitative digital in situ senescence-associated β-galactosidase
assay. BMC Cell Biology 12, 16 (2011).
59. Flor, A., Pagacz, J., Thompson, D. & Kron, S. Far-Red Fluorescent SenescenceAssociated β-Galactosidase Probe for Identification and Enrichment of Senescent
Tumor Cells by Flow Cytometry. JoVE (Journal of Visualized Experiments) e64176
(2022) doi:10.3791/64176.
60. Fuhrmann-Stroissnigg, H. et al. SA-β-Galactosidase-Based Screening Assay for the
Identification of Senotherapeutic Drugs. JoVE (Journal of Visualized Experiments)
e58133 (2019) doi:10.3791/58133.
61. Marini, N. et al. Data-driven color augmentation for H&E stained images in
computational pathology. Journal of Pathology Informatics 14, 100183 (2023).
62. Brand, M. D. et al. Suppressors of Superoxide-H2O2 Production at Site IQ of
Mitochondrial Complex I Protect against Stem Cell Hyperplasia and IschemiaReperfusion Injury. Cell Metab 24, 582–592 (2016).
63. Smer-Barreto, V. et al. Discovery of senolytics using machine learning. Nat Commun
14, 3445 (2023).
64. Heckenbach, I. et al. Nuclear morphology is a deep learning biomarker of cellular
senescence. Nat Aging 1–14 (2022) doi:10.1038/s43587-022-00263-3.
65. Duran, I. et al. Detection of senescence using machine learning algorithms based on
nuclear features. Nat Commun 15, 1041 (2024).
66. Levitsky, K. L., Toledo-Aral, J. J., López-Barneo, J. & Villadiego, J. Direct confocal
acquisition of fluorescence from X-gal staining on thick tissue sections. Sci Rep 3,
2937 (2013).
67. Liu, Y. et al. Environmental pollutants exposure: A potential contributor for aging and
age-related diseases. Environmental Toxicology and Pharmacology 83, 103575
(2021).
68. Baker, D. J. et al. cells shorten healthy lifespan. Nature (2016)
doi:10.1038/nature16932.
69. Tsuruda, P. et al. UBX1325, a small molecule inhibitor of Bcl-xL, attenuates vascular
dysfunction in two animal models of retinopathy.
70. Cohn, R. L., Gasek, N. S., Kuchel, G. A. & Xu, M. The heterogeneity of cellular
senescence: insights at the single-cell level. Trends in Cell Biology (2022)
doi:10.1016/j.tcb.2022.04.011.
71. Tang, H. et al. Single senescent cell sequencing reveals heterogeneity in senescent
cells induced by telomere erosion. Protein Cell 10, 370–375 (2019).

109
72. Fard, A. T. et al. Deconstructing replicative senescence heterogeneity of human
mesenchymal stem cells at single cell resolution reveals therapeutically targetable
senescent cell sub-populations. 2022.01.24.476823 Preprint at
https://doi.org/10.1101/2022.01.24.476823 (2022).
73. Ciotlos, S., Wimer, L., Campisi, J. & Melov, S. A Single Cell Time Course of
Senescence Uncovers Discrete Cell Trajectories and Transcriptional Heterogeneity.
2023.02.17.529001 Preprint at https://doi.org/10.1101/2023.02.17.529001 (2023).
74. Zhu, Y. et al. The Achilles’ heel of senescent cells: from transcriptome to senolytic
drugs. Aging Cell 14, 644–658 (2015).
75. Neri, F., Basisty, N., Desprez, P., Campisi, J. & Schilling, B. Quantitative Proteomic
Analysis of the Senescence‐Associated Secretory Phenotype by Data‐Independent
Acquisition. Curr Protoc 1, e32 (2021).
76. Neri, F. et al. A Fully-Automated Senescence Test (FAST) for the high-throughput
quantification of senescence-associated markers. bioRxiv 2023.12.22.573123
(2024) doi:10.1101/2023.12.22.573123.
77. Johnson, T. I. et al. Quantifying cell cycle-dependent drug sensitivities in cancer
using a high throughput synchronisation and screening approach. eBioMedicine 68,
(2021).

Abstract (if available)

Abstract Cellular senescence is a complex stress response that induces an essentially permanent cell cycle arrest, and a complex secretory phenotype termed the senescence-associated secretory phenotype (SASP). Over the past few decades, a substantial body of literature has established a potentially causal link between the accumulation of senescent cells and several age-related diseases. Consequently, significant efforts have been made to develop interventions that target cellular senescence. However, more work is needed to create senotherapeutics that can substantially improve human healthspan. Challenges in developing such therapies include the difficulty of detecting and quantifying senescent cells and their significant heterogeneity. Novel methods for assessing senescence burden and a deeper understanding of senescent cell heterogeneity are necessary to develop the next generation of senotherapeutics. Therefore, my thesis work focused on developing and optimizing workflows and tools for studying senescent cells and their heterogeneity by using cell culture models.

In Chapter 1, a mass spectrometry (MS)-compatible protocol is described to i) generate senescent cells using different stimuli, ii) collect conditioned media containing proteins secreted by senescent cells (SASP), and iii) prepare the SASP for quantitative proteomic analysis using data-independent acquisition (DIA) MS. This workflow can help elucidate senescent cell heterogeneity by characterizing the SASP of different senescent cell populations, as well as provide insights into aging and disease mechanisms related to senescence.

In Chapter 2, the Fully-Automated Senescence Test (FAST) is described, an image-based method for the high-throughput, single-cell assessment of senescence in cultured cells. FAST quantifies three of the most widely adopted senescence-associated markers for each cell imaged: senescence-associated β-galactosidase activity (SA-β-Gal) using X-Gal, proliferation arrest via lack of 5-ethynyl-2’-deoxyuridine (EdU) incorporation, and enlarged morphology via increased nuclear area. Additionally, proof of concept is provided that FAST is suitable for screening compounds that modify senescence burden. This novel method enables rapid, unbiased, and user-friendly quantification of senescence burden in culture, as well as facilitating large-scale experiments that were previously impractical.

Finally, Chapter 3 describes a study where high-content image analysis was employed to identify functionally distinct senescent subpopulations. It was found that G2-arrested senescent cells have higher senescence marker expression than G1-arrested cells. Additionally, it was found that G2-arrested senescent cells secrete more IL-6 – a pro-inflammatory cytokine part of the SASP – and are more sensitive to the senolytic ABT263 than G1-arrested cells. Thus, this study demonstrates the existence of functionally distinct senescent subpopulations in culture and points to DNA content as a key contributor to the heterogeneity within senescent populations.

Overall, my thesis work provided new workflows and tools for studying senescent cells and their heterogeneity, both between as well as within senescent cell populations. Such tools generated insights into the factors that contribute to senescent cell heterogeneity and will be broadly useful to the aging field by enabling further characterization of cellular senescence and its heterogeneity.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

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

PDF

The overlap between mTOR signaling, rapamycin and cellular senescence

PDF

A small molecule protease inhibitor induces senescence phenotypes that are reversible upon drug removal

PDF

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

PDF

Identifying and measuring cell-intrinsic and cell-extrinsic factors influencing aging

PDF

Longitudinal assessment of neural stem-cell aging

PDF

Defining the functional roles of neurotransmitters and neuropeptides in neural circuits

PDF

Investigating brain aging and neurodegenerative diseases through omics data

PDF

Phenotypic and multi-omic characterization of novel C. elegans models of Alzheimer's disease

PDF

Genomic and phenotypic novelties in the Southeast Asian house mouse

PDF

The effects of a fasting mimicking diet (FMD) on mouse models of Alzheimer's and Parkinson's disease

PDF

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

PDF

Computational approaches to identify genetic regulators of aging and late-life mortality

PDF

Alpha-ketoglutarate, an endogenous metabolite, extends lifespan and compresses morbidity in aging mice

PDF

Metabolomic and proteomic approaches to understanding senescence and aging in mammary epithelial cells

PDF

Coenzyme A binding sites induce proximal acylation

PDF

Modeling neurodegenerative diseases using induced pluripotent stem cells and identifying therapeutic targets

PDF

Swimming exercise reduces native ⍺-synuclein protein species in a transgenic C. elegans model of Parkinson’s disease and confers downstream protective effects independent of food restriction…

PDF

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

PDF

Synaptic transmission, nutrient sensors, and aging in Drosophila melanogaster

Asset Metadata

Creator Neri, Francesco (author)

Core Title Characterization of senescent cell heterogeneity using cell culture models

School Leonard Davis School of Gerontology

Degree Doctor of Philosophy

Degree Program Biology of Aging

Degree Conferral Date 2024-08

Publication Date 12/28/2024

Defense Date 06/24/2024

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

Tag cellular senescence,heterogeneity,high-content image analysis,OAI-PMH Harvest,SASP,senolytics

Format theses (aat)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Andersen, Julie (committee chair), Desprez, Pierre-Yves (committee member), Garrison, Jennifer (committee member), Schilling, Birgit (committee member), Tower, John (committee member)

Creator Email f.francesconeri@gmail.com,fneri@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC113997AIY

Unique identifier UC113997AIY

Identifier etd-NeriFrance-13171.pdf (filename)

Legacy Identifier etd-NeriFrance-13171

Document Type Dissertation

Format theses (aat)

Rights Neri, Francesco

Internet Media Type application/pdf

Type texts

Source 20240701-usctheses-batch-1176 (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.

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