1. Introduction
In recent years, proteomics has become an attractive tool for uncovering senescence biology, biomarkers, and therapeutic targets. Researchers anticipate the application of mass spectrometry (MS)-based proteomics will accelerate the translation therapies targeting senescent cells. MS-based proteomics is a gold-standard technique for large-scale protein identification and quantification in complex samples. To overcome the challenges associated with analyzing complex protein samples with a large dynamic range of protein concentrations, sophisticated separation techniques and advanced MS instrumentation have been developed to extend coverage and sensitivity to accommodate the detection of biologically relevant and low-abundance proteins.
2. Emerging MS-Based Approaches for Biomarker Discovery
2.1. Data Acquisition and MS Instrumentation
MS data acquisition methods and analysis pipelines have significantly improved data reproducibility, quantification, and completeness in mass spectrometry-based workflows. The advancement of data-independent acquisition (DIA or SWATH) methods is a major step toward increased reproducibility and quantitative accuracy for biomarker discovery compared with traditional data-dependent acquisition (DDA) methods applied in untargeted proteomics studies
[1][2][3][4,119,120]. DIA MS acquires MS/MS spectra for all peptides in a sample without requiring precursor information, which overcomes several challenges associated with DDA, including reduced bias toward abundant peptide selection that results in data missingness and decreased quantitative interference at the MS1 level
[2][4][119,121]. DIA-MS also offers a reliable and consistent analysis of complex proteomic samples
[5][6][122,123] that have been validated in consortium-level biomarker studies
[3][120]. DIA analysis also lends itself nicely to PTM profiling and quantification, as quantification at the MS2 level allows one to localize PTM sites and distinguish peptide isomers via site-localized fragment ions
[7][8][124,125]. DIA-MS acquisition generally uses wider precursor isolation windows to activate all ions for collision in each mass-to-charge (
m/
z) range, repeated across the full mass range in a single duty cycle. Innovations to the size and placement of precursor isolation windows, such as variable-window DIA
[9][126] and overlapped (staggered)-window DIA
[10][11][127,128], were developed to improve peptide selectivity. The generation of the first comprehensive SASP profiles in SASP Atlas were generated utilizing variable-window DIA to identify candidate senescence biomarkers such as GDF15, STC1, and SERPINs that are associated with aging in human plasma
[12][2]. Another newly developed novel acquisition method called ‘parallel accumulation–serial fragmentation’ (PASEF)
[13][129], combined with DIA (diaPASEF)
[14][130], improves sensitivity and scanning speed. Unlike typical DIA approaches based predominantly on chromatographic separation, diaPASEF increases the sensitivity and coverage of proteomic analysis with the inclusion of an additional ‘ion mobility’ peptide separation dimension.
2.2. Identification of ‘Proteoform’-Level Biomarkers with Top-Down Proteomics
Hundreds of thousands of proteins are produced by approximately 20,000 human coding genes
[15][16][17][18,131,132]. Following protein biosynthesis, post-translational modification (PTM) and proteolytic processing events can modify protein structures and interactions
[18][19][133,134]. PTMs include chemical groups such as phosphorylation and acetylation and more complex glycan structures, among many others
[20][21][135,136]. Any protein variation (protein products of a single gene) due to PTMs, alternative RNA splicing, and genetic mutation is defined by a term named ‘proteoform’
[22][137], which can have distinct functions and change dynamically in response to diverse stimuli
[23][24][138,139]. Proteoforms are challenging to characterize because of their similar sequences and masses, which make proteomic analysis more complex
[15][18]. Most ‘discovery’ proteomic approaches rely on peptide-level measurements or antibody binding at a specific location on a protein, and therefore are unable to distinguish proteoforms. However, proteoform-level measurements will potentially improve the sensitivity and specificity of proteomic biomarkers.
Top-down proteomics is a powerful technology for analysis of the intact proteins/peptides via tandem mass spectrometry to characterize unique proteoforms and localize PTMs
[23][24][25][138,139,140] to reveal structural and functional information on each proteoform
[26][27][141,142]. Based on recent studies, protein modifications and diseases can be closely associated with each other
[28][143]. For instance, the relationship between the occurrence and development of cancer and phosphorylation sites has been described
[29][30][31][144,145,146]. The Kelleher Research Group at Northwestern University has reported 29,620 unique proteoforms which are expressed from human blood and bone marrow, accessible in the Blood Proteoform Atlas (BPA) (
https://blood-proteoform-atlas.org (accessed on 24 August 2023)). The potential applications of BPA in clinical validation, such as the prostate-specific antigen isoform (IsoPSA) test in prostate cancer, have been noted
[32][147] and there are many more proteoform associations yet to be explored.
Proteoforms are potential biomarkers that can be directly connected to multifaceted phenotypes
[27][142], and thus the separation of intact proteins is necessary and one of the most significant challenges of top-down proteomics. Intact proteins are typically separated using reversed-phase liquid chromatography (RPLC) coupled with tandem mass spectrometry (MS/MS) analysis. Separation techniques have been developed for multidimensional separation platforms such as capillary zone electrophoresis (CZE), two-dimensional (2D) gel electrophoresis, and 2D-LC platform using high-pH RPLC and low-pH RPLC separation
[33][34][148,149]. In polyacrylamide gel electrophoresis (PAGE), Passively Eluting Proteins from Polyacrylamide gels as Intact species for MS (‘PEPPI-MS’) is a recent advancement for clean separation and extraction of intact proteins
[35][150]. Future advancements in MS instrumentation and data analysis tools are needed for pervasive application of top-down proteomics in biomedical research
[31][146]. such as streamling and upgrading of database search tools. For example, the presence of unexpected modifications complicates database matching. Fortunately, there is a rapid growth of top-down proteomics
[36][151], propelled by community groups such as the Consortium of Top-Down Proteomics (
www.topdownproteomics.org (accessed on 24 August 2023)) and initiatives such as the Human Proteoform Project, which strive to map the human proteoform at proteoform resolution
[26][141].
2.3. Sample Preparation Workflows to Overcome Challenges in Blood Biomarker Discovery
Circulating biomarkers are promising indicators to quantify age-related declines and senescent cell burden
[37][38][1,152]. The translation of therapies, like senolytics and senomorphics, will require biomarkers of senescence burden to identify and stratify individuals with elevated senescent cell burden and evaluate the efficacy of senolytic interventions. Interrogation of the blood proteome is key to establishing senescence biomarkers and subsequent senotherapeutics. However, the high dynamic range of protein abundance in plasma/serum, which spans over 10 orders of magnitude, poses a challenge in the detection of lower abundance components using mass spectrometry
[39][153]. The 22 most abundant proteins in blood, including albumin, immunoglobulins, myoglobin, transferrin, and haptoglobin, together constitute 99% of the total protein content, whereas albumin itself comprises nearly 60% of the total serum proteome
[40][154]. Low-abundant proteins having biomarker potential may be obscured from detection by high-abundance proteins. Thus, sample preparation protocols are critical for minimizing sample complexity. Various strategies therefore play a crucial role in reducing sample complexity before LC-MS/MS analysis, including depletion of high-abundance proteins, affinity enrichment of low-abundant proteins, and a variety of chromatographic
[41][155] and electrophoretic fractionation techniques
[42][156]. The depletion of highly abundant proteins, reviewed in detail elsewhere
[43][157], is a viable strategy for reducing the dynamic range at the expense of increasing the number of samples to be analyzed. Immunodepleting highly abundant proteins is a routinely applied method with high specificity and reproducibility
[43][157]. Antibody-based depletion of the top 7 or top 14 high-abundance plasma proteins showed 25% gains in proteome coverage
[44][158]. However, these gains come at the expense of added processing steps, additional variability, and the cost of the depletion columns. Another method based on bead-bound hexapeptides known as combinatorial peptide ligand library (CPLL, commercially named Proteominer) reduces the dynamic range of protein concentrations
[45][159]. CPLL enriches low-abundant proteins in a reproducible manner
[46][160], and with greater performance compared to the other available depletion/enrichment approaches, yielding 68.6% higher protein identification compared to the untreated serum
[47][161]. The impact of CPLL in serum/plasma biomarker discovery is excellently reviewed elsewhere
[48][162]. In one notable study, CPLL increased plasma protein identifications by 22% in breast cancer patients, of which 23 were differentially expressed and thus potential biomarker candidates
[49][163]. Another study applying CPLL identified 26 differentially expressed proteins, of which ficolin-2 was reported as a potential diagnostic marker for rheumatoid arthritis
[50][164].
Each strategy employed for reducing the dynamic range of protein samples has its own merits and drawbacks
[43][51][157,165] that impose trade-offs between sample throughput and proteome coverage. Extensive offline fractionation increases the number of samples and MS analysis time, whereas depletion/enrichment strategies add processing steps that introduce technical variability and cost. Further depth can be achieved by utilizing a combination of approaches
[52][53][166,167]. A recent study assessed the plasma proteome by comparing depletion via Human 6 (Hu6) and Human 14 (Hu14) columns and the ProteoMiner Kit, each in combination with protein level fractionation via SDS PAGE or via offline C18 chromatography before trypsin digestion
[54][168]. This study identified a total of 4385 proteins, 3064 (70%) of which were common between all methods. Another study reported 3400 proteins from pooled plasma samples from AD patients and healthy controls
[55][169] by using a combination of immune depletion, HPLC fractionation, and 2D gel electrophoresis before LC-MS/MS analysis. Throughput loss resulting from extensive fractionation can be mitigated by the utilization of isobaric tags for relative and absolute quantification (iTRAQ) or tandem mass tags (TMT) for multiplexing analysis
[56][170]. Keshishian et al. utilized abundant protein depletion, iTRAQ isobaric labeling, and offline basic PH reversed-phase fractionation to identify >4500 proteins in plasma with high reproducibility
[57][171].
A recently developed nanoparticle-based method has reported promising results for the analysis of blood samples
[58][59][172,173]. Over the last decade, there has been an increased understanding of the mechanisms of generation of ‘protein corona’, or the protein mixture bound at the interface between the surface of nanoparticles and a biological fluid, and which factors influence its composition
[60][61][174,175]. Nanoparticles can reduce the proportion of highly abundant proteins and enrich low-abundance proteins in serum and plasma. Notably, the composition of protein coronas can be tuned for a swath of plasma proteins in a reproducible manner, enabling the use of nanoparticle-based automation workflows in combination with MS analysis, enabling protein coverage exceeding 2000 proteins
[62][176]. Although initial studies utilizing nanoparticles suggest great promise for proteomic analysis of blood samples, these methods need to be further evaluated and improve reproducibility across different cores and institutions
[63][177]. Improvements in standardization and automation will aid in these efforts
[58][172].
3. MS-Based Approaches for the Discovery of Novel Therapeutic Targets
Proteomic methodologies can be leveraged for the discovery of novel senotherapeutic approaches in multiple ways. First, the identification of senescence-specific cell-surface proteins will unveil senotherapeutic targets. Second, proteomic profiling of protein–drug interactions associated with senolysis can aid in the identification of new pathways and proteins for the development of senolytics. As the protein targets of several existing senolytics are not yet known, the identification of protein interactors from existing senolytics presents an intriguing area for the discovery of novel senolytic pathways and the development of new drugs. Identifying specific drug/protein binding interactions may also enable the development of scientific interventions to address issues such as drug delivery, bioavailability, and off-target binding. Numerous methods for proteomic target discovery have surfaced over the last few decades, many of them relying on the quantitative unbiased measurements given by mass spectrometry.
Several existing methods, Thermal Proteome Profiling (TPP), Limited Proteolysis (LIP), and Stability of Proteins from Rates of Oxidation (SPROX), have proven to be effective in the successful identification and quantification of specific intracellular drug/target binding interactions. New approaches, in addition to these previously well-established methods, have also been recently introduced such as pulse proteolysis (PP) and Proteome Integral Solubility Alteration (PISA).
3.1. Limited Proteolysis Coupled to Mass Spectrometry (LiP-MS)
Lip-MS is a technique that compares proteolytic fragmentation patterns between a drug-treated group and a non-treated control
[64][178]. Drug treatment induces conformational changes in drug-bound protein targets. Then, when a broad-spectrum protease is added to the samples for a short time, these conformational changes cause the protease to have altered accessibility to its cleavage sites at the site of drug–protein interactions. This results in distinct peptide fragment signatures in the drug-treated group compared to the control
[64][178]. After disrupting the protein–drug interaction, specific digestion with trypsin is then used to further break down the remaining large peptide fragments, leaving tryptic peptides enriched at the sites of protein–drug interaction to be detected via MS. Lip-MS is useful in complex mixtures, such as cell lysates, for understanding the structure of proteins and any changes that occur due to drug binding
[65][179].
3.2. Pulse Proteolysis (PP)
Pulse proteolysis is another proteomic technique used for identifying potential drug targets, as well as protein–protein interactions. In this method, a cell lysate is incubated with a drug and urea, while a control sample is incubated under identical conditions without the drug. The drug will theoretically stabilize its target proteins and prevent their denaturation by urea. Then, a protease is introduced for a short period, referred to as a ‘pulse’. PP causes complete digestion of unfolded proteins, but drug-stabilized proteins remain intact
[66][180]. After the pulse period, the reaction is quenched by rapidly lowering the temperature or by adding protease inhibitors. The protein is run on a gel, and the intact protein bands (higher molecular weight) are quantitatively compared. Drug-bound targets are indicated by an increased amount of intact protein in the drug-treated group
[66][180]. A later improvement in this method incorporated mass spectrometry for the untargeted discovery of drug-binding interactions. In this variation, the portion of the gel containing intact protein is excised, digested in gel with trypsin, and analyzed via MS to compare the fraction of intact proteins between drug-treated and untreated groups
[67][181].
3.3. Stability of Proteins from Rates of Oxidation (SPROX)
Stability of Proteins from Rates of Oxidation (SPROX) employs a hydrogen peroxide buffer with increasing levels of a denaturant to generate chemical denaturation curves
[68][182]. After an equal amount of time for the oxidation rate, the samples are quenched. Chemical denaturation curves are generated based on the oxidation of methionine residues on peptides detected and quantified with mass spectrometry
[68][182]. SPROX reveals proteins that are stabilized or destabilized by a bound target molecule, indicated by either a right or left shift in the chemical denaturation curve
[69][183]. It is especially beneficial in evaluating the disassociation constants of protein–ligand complexes as oxidation decreases protein stability. Since its first use, SPROX has been updated to allow for higher throughput detection with isobaric tags
[69][183].
3.4. Thermal Denaturation-Based Target Discovery: TPP, CETSA, and PISA
The Cellular Thermal Shift Assay (CETSA) compares thermal denaturation curves between drug-treated and control groups to identify potential drug targets. Drug-bound proteins exhibit stabilization that protects the drug–protein complex from thermal denaturation, allowing for the interaction to be identified by identifying proteins with a right-shifted curve
[70][184]. Thermal Proteome Profiling (TPP) combines CETSA with mass spectrometry analysis to identify novel protein–drug interactions in untargeted mass spectrometry analysis
[71][185]. Both techniques can be used in cell extracts and intact cells, enabling the identification of target engagement within the intact cellular environment. Notably, TPP can also be used to identify drug targets in tissues from animal studies. For example, one study demonstrated the dose-dependent engagement of methionine aminopeptidase-2 (MetAP2) with its inhibitor, TNP-470, in the livers of mice treated with the drug
[70][184]. These techniques are beneficial for target discovery following treatment with a ligand, and TPP can also be applied to analyze downstream effects and other factors that affect protein stability such as post-translational modifications (PTMs)
[71][185]. Finally, Proteome Integral Solubility Alteration (PISA) pairs thermal denaturation with the use of isobaric labeling to enable sample multiplexing during MS analysis
[72][186]. By combining the soluble protein fractions obtained after a thermal gradient, targets can be identified by looking at the area under the curve across plexes within each sample
[72][186]. PISA also allows for comparative quantification at a 10-or-more-fold increased throughput when compared to CETSA/TPP.
3.5. Size-Exclusion Chromatography and Affinity Selection Mass Spectrometry (SEC-ASMS)
SEC-ASMS is a hybrid analytical technique that combines the size separation capability of SEC with further separation via affinity selection which is then analyzed with MS
[73][74][187,188]. SEC is a liquid chromatography method that separates molecules based on size and shape. Affinity selection allows for the separation of ligand-bound proteins in a complex mixture. After the SEC separation, the eluted protein fractions are subjected to affinity selection, adding the ligand of interest, to selectively capture and enrich their target molecules from the complex protein mixture
[73][74][187,188]. After affinity selection, captured protein–ligand complexes are analyzed using mass spectrometry to characterize the interaction and identify targets
[73][74][187,188]. SEC-ASMS is also helpful for examining protein complexes, protein–protein interactions, and mapping post-translational modifications to study their function.
3.6. Targeting and Quantifying Senescent Cells through the ‘Surfaceome’
For decades there have been efforts to identify and quantity proteins on cell surfaces because of their value as markers of cell identity and as therapeutic targets. A prime example of this is immunophenotyping, a method that essentially exploits the presence or absence of specific cell-surface markers to count and discriminate cell populations using antibody detection. Immunophenotyping is now a tool of choice to probe the phenotype and function of the immune system
[75][189]. This method can be used to count and discriminate specific lymphocyte populations, including progenitor cells and various hematopoietic lineages, with ever-increasing granularity based on the presence or absence of cell-surface markers
[75][76][189,190]. Knowing the identity and quantity of proteins on the cell surface has enormous potential benefits for translational research and the development of disease diagnostics and therapeutics, not only for its established benefit in immunophenotyping, but for the potential to identify therapeutic drugs, immunotherapy targets, and cell-surface targets for the isolation of cell populations.
Mass spectrometry-based proteomics has been applied to a variety of senescence models to identify surface markers that can be used to target these cells for senotherapy (
Table 1). For example, unbiased mass spectrometry-based studies have identified dipeptidyl peptidase 4 (DPP4) to be enriched on the surface of senescent versus proliferating fibroblasts
[77][191]. Importantly, DPP4 can be exploited for both sorting senescent cells via flow cytometry and for targeting them in vitro using antibody-dependent cell-mediated cytotoxicity. A subsequent study demonstrated that senescence-associated DPP4 activity can be pharmacologically targeted to improve hemostasis and plaque stability in old mice
[78][192]. Another promising example of a therapeutically relevant surface marker is the urokinase-type plasminogen activator receptor (uPAR), originally identified as a preferentially expressed marker on the surface of senescent human lung adenocarcinoma cells and melanocytes
[79][193]. In the same study, the authors demonstrated that chimeric antigen receptor (CAR) T cells that target uPAR efficiently eliminate senescent cells in vivo and extend the survival of mice with lung adenocarcinoma following treatment with MEK and CDK4/6 inhibitors, and ameliorated liver fibrosis in two independent models
[79][193]. These examples highlight the utility of cell-surface proteins as potential senotherapeutic targets.
Table 1. Surface markers from senescent cells identified using mass spectrometry-based proteomics. RS = replicative senescence; OIS = oncogene-induced senescence; IR = irradiation; PM = plasma membrane; ADCs = antibody–drug conjugates; ADCC = antibody-dependent cell-mediated toxicity.
Surface Protein |
Cell/Tissue Type |
Senescence Inducer |
Strategy |
Senolytic Validation |
Reference |
DEP1/PTPRJ/CD148 |
Bladder cancer (EJp21 and EJp16); human lung fibroblast (IMR-90); human fibrosarcoma (HT1080p21); mouse lung adenomas (V600EBRAF); human melanocytes |
p21/p16 overexpression; RS, RAS OIS - - - |
PM isolation, in-gel digestion, and LC-MS/MS |
|
[80][195] |
B2MG |
Bladder cancer (EJp21 and EJp16); mouse lung adenomas (V600EBRAF) |
p21/p16 overexpression - |
PM isolation, in-gel digestion, and LC-MS/MS |
Gold nanoparticles, ADCs |
[80][81][82][13,195,196] |
NOTCH1 |
Human lung fibroblast (IMR-90); mouse pancreatic neoplasm (p48-cre) |
HRASG12V, OIS, Etoposide KrasG12D OIS |
SILAC |
|
[83][197] |
DPP4/CD26 |
Human lung fibroblast (WI-38, IMR-90); human aortic endothelial cells (HAEC); human umbilical vein endothelial cells (HUVEC); mouse embryonic fibroblasts (MEF) |
RS, IR, Doxorubicin IR IR
HRASG12V OIS |
PM isolation, in-gel digestion, and LC-MS/MS |
ADCC |
[77][191] |
SCAMP4 |
Human lung fibroblast (WI-38, IMR-90); human aortic endothelial cells (HAEC); human umbilical vein endothelial cells (HUVEC) |
RS, IR, Doxorubicin, HRASG12V OIS IR IR |
PM isolation, in-gel digestion, and LC-MS/MS |
|
[84][198] |
CD24 |
Bone marrow mesenchymal cells from INK-ATTAC mice |
- |
CyTOF |
|
[85][199] |
Despite its huge benefits in diagnostics and cell phenotyping, immunophenotyping is a limited tool for the discovery of new cell-surface markers. First, it is limited to the set of cell-surface proteins that are already known and have high-quality antibodies. Secondly, immunophenotyping has limited multiplexing capability. Thus, unbiased approaches are critically needed for the discovery of new surfaceome targets that will aid in the discrimination of the highly diverse and heterogeneous cell types in the human body. Technological advancements in mass spectrometry approaches now enable comprehensive and unbiased identification and quantification of surfaceome proteins. Over the past decade, an increasing number of sample preparation workflows, computational tools, and curated surfaceomes have emerged to aid in the discovery and validation of cell-surface proteins using mass spectrometry, and are reviewed more extensively elsewhere
[86][87][200,201]. The most rigorous methods for identification of the cell-surface proteome and the surface interactome utilize biorthogonal chemistry to biotinylate cell-surface proteins for subsequent affinity enrichment. For example, the cell-surface capture (CSC)
[88][202] method utilizes non-cell-permeable reagents to biotinylate cell-surface glycoproteins, which comprise 90% of cell-surface proteins. Following the labeling of live cells in culture, labeled glycoproteins can be enriched either at the protein or peptide levels before mass spectrometry analysis. Modifications of CSC include Cys-Glyco-CSC, which allows the co-enrichment of peptides that are linked to the glycopeptides via disulfide bonds, and Lys-CSC, which allows labeling and enrichment of all surface-exposed lysine-containing proteins
[89][203]. Modern variations of CSC have enhanced its utility for smaller sample inputs
[90][204] and implemented automation, such as autoCSC
[91][92][205,206] and µCSC
[93][207], resulting in vastly improved sensitivity, reproducibility, and throughput and making surfaceome characterizations of many samples and cell types more feasible. Another recent innovation termed LUX-MS enables a precise spatiotemporal elucidation of the surface interactome of ligands or antibodies using a light-mediated reaction, thus enabling the precise mapping of cell-surface nanoscale organization with targeted therapies
[94][208].
A critical aspect of mass spectrometry-based identification of the surfaceome is the correct annotation of surface proteins from the huge proteomic dataset, which requires a stringent bioinformatic analysis based on specific characteristics of surface proteins. A widely used approach is gene ontology (GO) annotations to identify which proteins are reported to be associated with the cell surface or plasma membrane. More rigorous filtering criteria for the identification of surface glycoproteins after CSC rely on the detection of a mass shift of 0.984 Da that corresponds to the deamidation of asparagine residues through enzymatic deglycosylation during the enrichment process. The selected peptides can be further filtered for the presence of the conserved NXS/T motif that is present in more than 90% of the N-glycosylation sites, allowing the exclusion of the peptides that might have co-eluted during the enrichment process
[91][205]. Similarly, Lys-CSC requires the selection of all peptides that contain a mass shift of 145.019 Da corresponding to the 3-(carbamidomethylthio) propanoyl modification on the lysine residues
[89][203].
Numerous software tools and databases can be employed to aid in the identification, filtering, and prioritization of candidate cell-surface proteins identified from mass spectrometry-based approaches. A suite of tools in the CellSurfer portal
[93][207] (
https://www.cellsurfer.net/ (accessed on 29 September 2023)), developed by the Gundry laboratory, is one such resource that aids in annotating and scoring candidate cell-surface markers from mass spectrometry data. The site includes several tools
[93][95][96][207,209,210] that enable the rigorous bioinformatic validation, prioritization, ranking, and visualization of candidate surface proteins for future validation and clinical studies. Protter (
https://wlab.ethz.ch/protter/start/ (accessed on 29 September 2023)) is another online open-source tool that allows visualization of peptides in the protein topology and provides evidence to validate whether the identified peptides are exposed on the cell surface
[97][211].
Increased interest in exploring the cell surfaceome as a source of potential biomarkers has led to the development of online databases that can serve as a reference for future surfaceome studies. The Cell Surface Protein Atlas
[98][212] (
https://wlab.ethz.ch/cspa/ (accessed on 29 September 2023)) is a repository of mass spectrometry-derived cell-surface proteins from 41 human and 31 mouse cell types, providing a snapshot of the surface proteome from different cell types, including cancer cells. Similarly, the in silico human surfaceome (
http://wlab.ethz.ch/surfaceome/ accessed on 29 September 2023)) is an online database containing 2886 proteins predicted using a machine-learning-based predictor, SURFY, developed by the Wollschied laboratory
[99][213]. Despite these available databases, the unique cell surfaceome signature has not yet been described for many cell types and disease states, and thus continued efforts and innovation in surface protein discovery and annotation are needed.
There are also potential limitations to utilizing the surfaceome for targeting. First, just as in other markers, cell-surface markers are heterogeneous on senescent cells. Even in presumably homogenous cell culture conditions, not all induced senescent cells express surface markers such as DPP4
[77][191]. This is particularly true for markers of senescence, since senescent cell populations are inherently heterogenous and markers identified via bulk proteomic approaches may overestimate the ability of a protein to be a true marker of senescence. Strategies coupling single-cell technologies with surface labeling can provide a better approach to identifying markers from such heterogenous populations
[86][87][200,201]. Secondly, it is not yet known how well surface markers distinguish different populations of senescent cells based on their phenotypes. Another technical challenge is that the validation of surface markers discovered via MS is limited by the availability of highly sensitive and specific antibodies to cell-surface proteins. This may limit the validation of surfaceome candidates using orthogonal approaches such as flow cytometry.