Proteomic Profiling of Protease-Primed Virus-Permissive Caco-2 Cells Display Abortive-Interferon Pathway and Deregulated Thromboinflammatory SERPINS

Emerging paradigms in interferon (IFN) biology suggest a dynamic INF induced interactome that extends through broader Interferon Stimulated Gene (ISG)- induction, which implicates interferon- ISG coordinated cross-talk with mRNA processing, post-translational modification and metabolic processes that underlie pathological (viral, autoimmune and tumor biology) and physiological (stem cell regenerative pathways) processes. INF immune responses can also be triggered by endogenous host-derived molecules that are generated in response to cellular stress or hemostasis imbalance to establish tissue repair and regeneration in first place, however, overactivation or lack of countermeasures can result in host tissue damage. The proteases are integral to viral and tumor pathology, and importantly serine proteases TMPRSS2 and trypsin have been identified as important molecular determinants underlying COVID-19 pathology, and emergence of coronaviruses cultured in vitro, respectively. We propose that pathogen associated proteases can act as novel stress-inducers to facilitate viral- competent immunomodulation. We term it as Protease Induced Transcriptomic/ epi-Transcriptomic Reshaping (PITTR) of host cells to counter cellular stress. We present a novel experimental model and our preliminary findings of trypsin- primed Caco-2 cells (CPT) that result in translational halt comparable to cells grown under serum-starvation conditions (CSS). CPT at escalating trypsin concentration (CPT- EC) induce upregulation of selective proteins that majorly map to ribosomal, RNA transport, and spliceosome ribonucleoproteins (RNPs). The inclusion of proinflammatory IL1-b to CPT (CPT- IL) resulted in global overexpression of proteins comparable to Caco-2 cells cultured in growth-factor rich serum conditions (CFBS), indicating a likely de-repression of trypsin- induced translational halt. Caco-2 cells display abortive interferon proteome under differential trypsin conditions (CPT, CPT-EC and CPT-IL), which is marked by complete lack of INF generation despite induction of intermediate ISGs, suggestive of protease (trypsin)- dependent regulation of INF response. Viruses regulate the proteome of stress granules (SGs) that are induced to cope transient translational halt as a central adaptive response to pathogen induced cellular stress. The integral components of SGs include non-translating mRNA, ribonucleoproteins (RNPs) and RNA binding proteins (RBPs), which together form biological condensates through a biophysical process involving weak electrostatic interactions through intrinsically disordered regions in RBPs resulting in liquid- liquid phase separation. We compared the CPT- EC proteome to the Mammalian Stress Granules Proteome (MSGP) database to explore potential RBPs that could possibly regulate INF response (and could act as potential anti-viral targets). Notably, differentially upregulated RNPs and potential RBPs from ISG family including ADAR and PRKRA, and RNA helicases implicated in viral pathogenesis were found to be upregulated in the CPT- EC proteome further strengthening the role of proteases (trypsin) in regulating INF pathways independent of the pathogen. We propose that the supplementation of viable SARS-CoV-2 viral loads to trypsin- primed host cells could recapitulate an infectious disease model, which may closely phenocopy pathogen- driven inflammation and signaling events. Based on the global downregulation of seven SERPINS (serine protease inhibitors) linked to thromboinflammation in our LCMS profiling data, we support the candidature of serine protease inhibitors for protease mediated viral pathologies. COVID-19 is increasingly linked to coagulopathy and resemblance to Neutrophil Extracellular Trap (NET) related thromboinflammatory features; SERPIN A1AT (alpha 1 antitrypsin) being a potent neutrophil- elastase inhibitor and a negative regulator of coagulation complement pathway may be a promising candidate for establishing hemostasis rebalancing in COVID-19 pathology.


Introduction
A novel betacoronavirus, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is currently spreading across many countries globally Sun et al., 2020;Li et al. 2020;Lai et al., 2020). Coronavirus disease 2019  is a pandemic with no specific therapeutic agents and hence substantial mortality. The potential drugs in preclinical and clinical studies include viral protease targeting inhibitors-Lopinavir-Ritonavir , lysmotrophic molecule hydroxychloroquine (Gautret et al., 2020), ribosome and RNA metabolism modulators -remdesivir and ribavirin (Beigel et al., 2020;Kmietowicz, 2020;Alexander et al., 2020;Khalili et al., 2020), anticoagulant heparin (Negri et al., 2020), IL-6 neutralizing antibody , triple combination of interferon-1b, lopinavir-ritonavir, and ribavirin (Hung et al., 2020), and ribavirin in combination with anti-viral interferons a/ b (Khalili et al., 2020). Other suggested therapeutic strategies are directed toward the viral entry pathway that involves host cell receptors ACE2 and serine protease TMPRSS2. TMPRSS2 serine protease inhibitors Camostat and Nafamostat have been implicated in COVID-19 treatment (Bittmann et al., 2020). ACE2 (virus-binding receptor) blockers, antimalarials; antivirals; immunoenhancers, monoclonal antibodies, and vaccines  are proposed for COVID-19 treatment. In silico approaches (Mercurio et al., 2020) driven potential drug candidates (Balkrishna, 2020), non-human primate model for COVID-19 pathogenesis (Lu et al., 2020), and nutritional supplements (lactoferrin) (Chang et al., 2020), are other research guided strategies that are being actively investigated. However, exploring COVID-19 treatment still remains a global challenge (Mohta et al., 2020), and accelerating drug development through repurposed FDA approved drugs is encouraged for COVID-19 treatment, which involves employing pathogen-or hostdirected approaches that may potentiate anti-viral immune response by interfering with host cell factors or dampening pro-inflammatory cytokine storm (Gordy et al, 2020). Further, the emerging studies indicate that COVID-19 is commonly complicated with coagulopathy (Gris J-Christophe et al., 2020; Connors and Levy, 2020), marked by higher platelet count, and elevated D-dimer (Yin et al., 2020;. Continuous efforts to find potential treatments and understand the molecular determinants of disease mechanism are underway (Wang et. al COVID-19 (van der Heide, 2020). Neutrophil infiltration has also been reported in pathological findings from autopsied COVID-19 patients (Fox et al., 2020;Yao et al., 2020) implicating neutrophilia as a potential source of excess Neutrophil Extracellular Traps (NETs) in case of COVID-19 Mozzini and Girelli, 2020;, Bonow et al., 2020Thierry and Roch, 2020). The clinical presentations of severe COVID-19 patients (including ARDS and microthrombosis) strikingly overlap with the established NETopathies . Most notably, COVID-19 is characterized by a unique immunological profile marked by a profoundly impaired IFN type I response, which features a low interferon production and downregulation of interferon-stimulated genes (ISGs). Improved identification of therapeutic targets requires better understanding of the differential immunological responses underlying SARS-CoV-2 pathogenicity in COVID-19 disease. Intriguingly, sustained refractory interferon response (resulting in reduced viral clearance) in parallel to massive proinflammatory responses (leading to hosttissue destruction) in COVID-19 disease (Hadjadj et al., 2020), suggests existence of yet unexplored pathways supportive of persistent low level of type I interferon induction. Identification of the precise mechanisms that contribute to reduced type I interferon activity are proposed to be critical in the development of targeted immunomodulatory strategies in patients with COVID-19 (Merad and Martin, 2020).
Several coronavirus proteins including MERS-CoV encoded ORF 8b (Lee et al., 2019), SARS-CoV proteins 8b and 8ab , and NS7b and NS8, in 2019-nCoV (Fahmi et al., 2020), are implicated in immune modulation. The coronavirus S glycoprotein binds host cell receptors and promotes fusion of the viral and cellular membranes (Walls et al., 2016;. The extensively N-glycosylated S protein trimers are the main target of neutralizing antibodies and the focus of therapeutic and vaccine design efforts (Tortorici and Veesler, 2019), the recent comprehensive structural and functional insights into the SARS-CoV-2 spike S protein and its interaction with host protein ACE2 and TMPRSS2 (Walls et  Herpes Simplex Virus (HSV-1) serine protease UL24 is known to impair NF-κB activation and modulate host antiviral immune pathways (Xu et al., 2017). Emerging studies indicate that the viral proteases possess side-functions extending beyond the polyprotein clipping, which are linked to host immune evasion by these viruses (Kikkert, 2020).
SARS-CoV-2 enters the host cells by using its spike protein receptor binding domain, which binds to the host cell receptor ACE2 (carboxypeptidase) Letko et al., 2020), and a nearby serine protease TMPRSS2 later cleaves it off to permit the cell-cell fusion and viral entry (Hoffmann et al., 2020). TMPRSS2 is also likely key protease for SARS-CoV-2 replication (Matsuyama et al., 2020). Expression of two mucosa-specific serine proteases, TMPRSS2 and TMPRSS4 have recently been shown to facilitate SARS-CoV-2 spike fusogenic activity highlighting the intestine as a potential site of SARS-CoV-2 replication (Zang et al., 2020). Importantly, serine proteases TMPRSS2 and trypsin have been identified as important molecular players underlying pathology and emergence of coronaviruses, respectively. Particularly trypsin-treatment (proteolytic processing) of viral spike protein has come up as potential species barrier for emergence of zoonotic coronaviruses, hence posing a potential threat for future outbreaks driven by cross-species transmission of coronaviruses Hoffmann et al., 2020;Matsuyama et al., 2020Menachery et al., 2020Letko et al., 2020). It was recently proposed that the small intestine additional proteases such as trypsin might enhance SARS-CoV-2 viral infection and pathogenesis by triggering more robust cell fusion (Zang et al., 2020).
Gene expression is a tightly regulated process and integrally linked to regulation of RNA homeostasis. Cellular mRNA transcription, decay, transport, localization, translation and degradation is further regulated by specific proteins, which majorly belong to the family of ribonucleoproteins (RNPs) or RNA binding proteins (RBPs). Currently, these RNA-protein complexes are called membraneless organelles (MOs), and are focus of intensive research to understand precise regulation of gene expression under physiological and pathological conditions. RNAs and RBPs enriched MOs can form and localize to nucleus as speckles, paraspeckles or Cajal bodies, and in the cytoplasm as P-bodies and stress granules (FU, 2020). MOs are preferred targets of invading viruses as they exploit them for faithful replication (hiding from host immune sentinels), and in case MOs are associated with gene repression, viruses tend to evolve mechanisms to counteract their assembly and formation (Gaete-Argel et al., 2019) Different types of MO ribonucleoprotein granules (speckles, paraspeckles, SGs or P-bodies) have overlapping constituents however they involve distinct RBPs scaffolds to form. Interestingly recent studies report the G3BP protein homologues (G3BP1 and G3BP2) and other intrinsically disordered region (IDR) containing proteins as integral to MLO biogenesis via liquid-liquid phase separation (LLPS) process (Deniz, 2020; Guillen-Boixet, 2020), which is an adaptive response to environmental stresses required for cellular function (reviewed in Fay and Anderson, 2018). Viral infections are a major trigger of cellular stress and many classes of viruses are known to interact with host ribonucleoproteins, RNA transport, and RNA binding proteins  with implication in MO formation, regulation or counter formation.
Based on authors' previous findings (Sharma et al., 2013;Sharma et al., 2017;Sharma et al., 2019) and recent emerging studies, we envisaged a novel role for proteases in viral pathogenesis, which suggests Pathogen (protease) Induced Transcriptomic/ epiTranscriptomic Reshaping (PITTR) of host cells for pathogencompetent immunomodulation. We tested the following reasonable hypothesis -"Trypsin-priming of virus-permissive host cells reshapes host metabolic circuitry, which drives transcriptional and epitranscriptomic remodeling conducive to viral-competent immunomodulation. We present our preliminary findings that offer a new framework of protease induced interferon deregulation of host cells and implicate SERPINS as potential therapeutic molecules that may help trigger anti-viral immune response. Our high-throughput proteomics data links partial loss of A1AT in protease-primed host cells (SARS-CoV-2 permissive Caco-2) to induction of viral-competent immunomodulatory pathways.

Materials and Methodology:
Culture of human intestinal cells derived from Caco-2 cells: Caco-2 cell line was obtained from National Repository for cell lines at National Centre for Cell Science (NCCS), Pune, India. The cells were maintained in Dulbecco's Modified Eagle Medium with 10% fetal bovine serum and grown to confluency before using them for the desired treatments for proteomic studies. The study was approved by the Institute Ethics Committee of Postgraduate Institute of Medical Education and Research, (PGIMER) Chandigarh, India.

LC-MS and protein data analysis:
For LC-MS studies, total cellular proteins were extracted from Caco-2 cells grown under 5 different conditions (Fig.1)-i) DMEM serum free medium-Control 1, ii) DMEM medium and 10% FBS, iii) DMEM serum free medium supplemented with varying doses of trypsin (Catalog #: 17-161E, Lonza). T1 is the maximum concentration (500µg /ml) and T2 is 1:10 of T1 (50 µg /ml). T3 is a unique treatment, which includes the well characterized conditioned media of ARPE19 (adult retinal pigment epithelial) cell line cultured in presence of residual trypsin (50 µg /ml). Briefly, confluent Caco-2 cells were trypsinized using 500µg /ml of trypsin solution, the trypsinized cells were spun and 1/10 th of residual trypsin was retained, diluted 10 times in serum free media and replated at a high cell density of 1 million cells/ml for 24-48h. This conditioned media is marked by secretion of biologically active bFGF (basic fibroblast growth factor) and interleukin IL-1b as determined by using multiplex assays previously reported by the author (Sharma et al., 2013;Sharma et al., 2017;Sharma et al., 2019). Therefore, T3 is ARPE19 conditioned media containing bFGF, IL-1b and active trypsin (as qualitatively determined by gelatin zymography-data not shown), however, absolute concentrations of trypsin were not determined using a commercial trypsin substrate. The Caco-2 cells cultured under different conditions as aforementioned, were incubated for 48 h in standard CO2 incubator at 37°C and 5% CO2.
Sample preparation for LCMS proteomic profiling-at the end of experiment, the cells were washed in 1X TBS to remove media or any other residual buffer. 90 μl of hot 6M Guanidine Hydrochloride (GnHCl)/ 0.1M Tris (pH 8.5) was added. The mixture was thoroughly mixed and vortexed and immediately kept at 90°C for 10 minutes to achieve lysis of the cells. Brief sonication (5-10 second single pulse) of the samples was carried out followed by returning the samples to 90°C for further 5 minutes. The cell lysates were spun at maximum speed (14000rpm/20 minutes at room temperature) to allow the debris to settle completely. The clear supernatants were secured sparing the residual dirty glue of the pellet to prevent any contaminating artifact (lysis buffer 6M GnHCl). 25 μg of each sample was taken and subjected to reduction and alkylation as follows-10 mM DTT (2 μl of 200mM stock) was added to each tube containing 20 μl protein sample. Tubes were incubated for 45 minutes in dark at 37°C, this was followed by addition of IAA (Idoacetamide) to each tube to achieve final concentration of 50mM (2 μl of 500mM stock was added to each tube) and incubated for 45 minutes in dark at 37°C.
Trypsinization-after reduction and alkylation, the protein samples were diluted 10X (180 μl of LCMS grade H2O was added to 20 μl of protein sample) to bring down the GnHCl concentration from 6M to 0.6M. The pH for trypsin activity is crucial, therefore, pH for all tubes was checked using precise pH strips. If it was not found to be in range of 8.0 -8.5, it was reset with 0.1M tris (pH 8.0). Trypsin (2 μl) was added to each tube at the concentration of 0.5 μg/ μl, making final amount to 1 μg. The tubes were incubated overnight (16h) at 37 °C.
Sample Clean up-following trypsinization, the pH was readjusted to around 2.0 using 2-3 μl of 10% TFA. The C18 clean-up columns were washed with 400 μl of methanol, spun at 2000 rpm for 1 minute (this step was repeated 2 times). It was followed by washing with 400 μl of Solution A. Composition of Solution A was as follows 80% ACN, 20% H20, 0.1% formic acid. The column was the equilibrated with 400 μl of Solution B-2% ACN, 98% H20, 0.1% formic acid. This step was repeated twice. The samples were now loaded onto the respective columns, which were spun at 1500rpm for 1 minute (the flow through was made to pass the columns thrice). The columns were washed with Solution B five times. The bound proteins were eluted with 150 μl Solution C-50% ACN, 50% H20, 0.1% formic acid. Bound proteins were twice more eluted using Solution A (150 μl and 100 μl). At this step the total volume of eluted protein was 400 μl.
Protein identification and data analyses-the samples were concentrated by speed vacuuming that was kept on heater-off mode to prevent sample evaporation. After drying up, the samples were resuspended in 30-50ul of 0.1% formic acid and mixed well. The protein concentrations were measured using nanodrop (ThermoFisher Scientific). The samples were then bath sonicated briefly (1 minute) to break down the air bubbles if any and transferred to auto sampler vials to be loaded onto the LC-MS machine (Thermo Scientific™ Orbitrap Fusion™ Tribrid™ Mass Spectrometer, ThermoFischer Scientific) at the Central Sophisticated Instrument Cell (CSIC) at Postgraduate Institute of Medical Education and Research, Chandigarh, India.
The experiment was performed using EASY-nLC 1000 system (Thermo Fisher Scientific, Waltham, MA, USA) coupled to Thermo Fisher-Orbitrap Fusion Trybrid mass spectrometer equipped with nano-electrospray ion source. 1.0 µg of the peptide mixture was separated using a 25 cm thermos Easy-spray PepMap C18 column. Elution buffer was a 0-40% gradient of (95% acetonitrile, 0.1% formic acid) buffer at a flow rate of 300 nl/ min for 180 min. Mass Spectrometric (MS) data was acquired using a data-dependent top10 method dynamically choosing the most abundant precursor ions from the survey scan. The data was further processed for protein database search.
The system recommends loading protein samples with 1-2 μg /μl concentration, the injection volume was adjusted accordingly. The run time for sample was 180 minutes, following which the raw files were generated and were analysed in Proteome Discoverer 2.2 analysis software. Identification and quantification of proteins in complex biological samples was carried out using Thermo Scientific™ Proteome Discoverer™ software. This software simplifies a wide range of proteomics workflows, from protein and peptide identification to posttranslational modification analysis in label-free quantitation. The analyses were carried out against a standard homo sapiens database from UniprotKB. The samples were divided into desired groups/conditions/replicates and compared on intra and inter basis (to evaluate replicability and assess differentially expressed proteins, respectively). The results thus obtained reflected the PEP scores, unique peptides and more importantly the abundances of detected proteins in each sample.
The RAW files generated were analysed with Proteome Discoverer (v2.2) against the Uniprot HUMAN reference proteome database. For Sequest search, the precursor and fragment mass tolerances were set at 10 ppm and 0.5 Dalton, respectively. The protease used to generate peptides, i.e. enzyme specificity was set for trypsin/P (cleavage at the C terminus of "K/R: unless followed by "P") along with maximum missed cleavages value of 2. Carbamidomethyl on cysteine amino acid was considered as fixed modification and oxidation of methionine & N-terminal acetylation were considered as variable modifications for database search. Both peptide spectrum match and protein false discovery rate (FDR) were set to 0.01.

Model of inducing protease-mediated cellular stress in Caco-2 cells:
We hypothesized that the COVID-19 associated serine proteases could act as stressinducers and display stress-granule (SG)-like proteome, marked by compromised translational output with suppressed anti-viral interferon pathway. However, the protease-primed cellular proteome would retain the translational expression of proteins involved in the formation of biological substructures like SGs and negative regulation of interferon pathway. We established a novel experimental model of inducing cellular stress using serine protease trypsin as a stress-inducer. Caco-2 cell cells were maintained in Dulbecco's Modified Eagle Medium with 10% fetal bovine serum and grown to confluency before using them for the desired treatments for proteomic studies. The different treatments and the cellular morphology under respective treatment has been shown by bright field microscopy ( Fig. 1A) Bioinformatics analysis of identified proteins Differential analysis-Abundance values of Treated (T) and Control (C) samples were taken and differential analysis was performed to see the degree of fold change among the conditions. Abundance values were Log2 transformed and then filtered protein wise for whole samples on the basis of valid values by taking at least present in all conditions. Total 3088 proteins were filtered out from 4860 proteins. These values were then Z-score standardized. Categorical grouping was also for both the conditions: Control and Treatment-Control: (Control1:F2, Control2:F4), Treatment: (Treatment 1: F1, Treatment 2: F3, Treatment 3: F5), as detailed (Fig S1 A-F). Student t-test was used as the numbers of groups needed to be compared were 2 and statistical significance was considered for P value < = 0.05 and fold change > = 2.

Plots and histograms
Correlation Plot-The correlation plot shows the spearman correlation coefficient value that lies between -1 (negative correlation) to +1 (positive correlation). The colour code follows the indicated values of correlation coefficient. PCA-PCA is a widely used mathematical technique designed to extract, display and rank the variance within a data set. The overall goal of PCA is to reduce the dimensionality of a data set, simultaneously retaining the information present in the data. The PCA-biplot contains a lot of information and can be helpful in interpreting relationships between experimental samples. In this biplot, we can see that runs from each sample as representing different experimental conditions are far from each other. Screeplot: The contribution of all ten principal components to the profiling classification in plot of percentage explained variance of PC (Figs. 2 A-C, Fig. S2 A-F). Heatmaps-Clustering was performed on the LFQ abundance of total 23 statistically Significant proteins (Fig. 2D) and on all the 3088 proteins ( Fig. 1B) that were filtered as identified in all samples. Red and Blue colours, respectively, represent low and high expression of proteins. whereas yellow (median) blocks indicate no difference in expression. Heatmap plot of the samples produced after abundance values were log2 transformed and z-normalized. Hierarchical clustering was performed using Euclidian distance and average linkage using In-House R script (Package: ComplexHeatmap).
Pathway Enrichment analysis was performed on both group of Proteins filtered (3088 and 23). 3,077 out of 3,088 Proteins were successfully mapped to 3,123 Entrez Gene ID and used for pathway analysis. Out of 202 KEGG pathways gene sets were statistical significantly enriched over following pathways and highlighted in "Red" colour as identified in the KEGG pathway diagram-hsa03010 Ribosome, hsa03040 Spliceosome, hsa03013 RNA transport (Figure 5), hsa03050 proteosome and hsa04141 protein processing in endoplasmic reticulum (data not shown). 23 Proteins were found to be T-test statistically significant. These 3088 Proteins and 23 statistically significant proteins values were then visualized using In-house R Programming Scripts. 23 out of 23 proteins were successfully mapped to 23 Entrez Gene IDs. Out of 202 KEGG pathways gene sets were statistical significantly enriched over 1 pathway and genes were highlighted in "Red" colour as Identified in the KEGG pathway diagram-hsa00190 oxidative phosphorylation (Figure 3 A and B).

Mitochondrial proteins in trypsin-treated Caco-2 cells and implication in viral pathogenesis-
We compared the upregulated proteins in oxidative phosphorylation pathway (Figure 3) under various trypsin treatments (T1>T2>T3) compared to control 2. Interestingly, 5 out of 6 significant proteins were upregulated in T1 and T2, however these proteins were either down regulated or modestly upregulated in T3 (Fig. 3B, logarithmic fold changes). The upregulated proteins (NDUFB1, UQCRB, COX5A, ATP5O and ATP5J) are associated with assembly of the five oxidative phosphorylation system complexes (OXPHOS) in the inner mitochondrial membrane.

KEGG pathway analyses for biological pathways implicated in viral pathogenesis by employing colour coded expression tags
For the current set of LCMS experiments, the triplicates for each condition were pooled and processed prior to LCMS profiling, therefore, we could not apply ANOVA statistical significance between the groups since we had only one representative treatment (intra group-containing 3 pooled independent replicates) in each group. We, therefore, tried to leverage the 3 independent escalating trypsin concentration (T3<T2<T1) data (intergroup) for some meaningful interpretations. We attempted to carry out a trypsin dose response pathway analyses using simple colour coded expression tags for 26 pathways that are implicated in viral pathogenesis using unique expression tags (Figures 5, 11, 12, 14). Following modifications were made for further analysis -for KEGG pathway analysis and mappings, all the identified proteins were included in the analysis as there are many possibilities of missed identification mainly due to uncharged peptides; masking by other peptides etc. (these are being overcome by using multiple biological replicates in our in-process experiments). For this study, missing abundance values were imputed on the basis of normal distribution and used for the pathway analysis. Protein abundances (Log2 transformed and z-score standardized values) for Treatment 1, Treatment 2 and Treatment 3 with respect to Control 2 were shown by using in-house R scripts for pathway based data integration and visualization (Package: "Pathview"). Red and green colours were used to show the up-regulation and down-regulation respectively. The pathways were slightly modified for comparative analyses of trypsin concentrations with copyright permission ( We comprehensively analysed 26 biological pathways ( Figure 4A), which are linked to viral pathogenesis and generated modified respective KEGG pathways highlighting the expression of a particular pathway intermediate protein/gene using color-coded gradations (log2FC -1 to +1, green to red colour gradients), for the most relevant pathways (Figures 5B, 11, 12, 14). This helped us to assess the dose dependent expression (3 independent concentrations of trypsin) of the "concerned" pathway protein in color-coded read outs. However, owing to the limitation of missing independent replicates and inclusion of all the proteins for pathway analyses (as discussed above), a fraction of proteins did not accurately map to the color coding on the pathways (the missing upregulated proteins in T1 group are highlighted in blue arrows). Therefore, to get an accurate quantitative fold change value for the identified proteins we tabulated all the pathway proteins (above and below 1, on Log2 scale) including upregulated (+1 to +7) and downregulated (-1 to -6) proteins on log2 FC scale (supplementary files 1 and 2). Interestingly, there were many proteins present (up or own regulated) in trypsin treated Caco-2 cells as can been seen from colored boxes in the respective pathways ( Figures 5, 11, 12, 14). Further, we observed a consistent trend in protein upregulation (red color) that was negatively correlated with trypsin concentration (T) for most of the pathway proteins. Major fraction of mapped proteins was upregulated (red color) in T3 (lowest concentration of trypsin used) and hence most of the "red-color" is restricted to right 1/3 rd of each box (for most of the pathways), which is the position assigned to T3. The mid part of each box is T2 and left 1/3 rd is T1 (maximum concentration of trypsin) (expression tags, Figure 4B).

Modulation of ribosome and RNA pathways in trypsin-treated Caco-2 cells
The ribosome, spliceosome and RNA transport pathways emerged as most significantly abundant pathways in our data set ( Figure 5 A-C, upper panel). To explore their relative abundance in T1, T2 and T3 over control 2, we did the KEGG pathways analysis employing colour coded expression tags as described above ( Figure 5 A-C, bottom panel). We hypothesized that serine-proteases linked to COVID-19 pathology might act as stress-inducers to trigger biological condensate associated translational halt, involving global proteome suppression that compromises induction of anti-viral proteins, primarily the interferon pathway. We asked if trypsin treatment of Caco-2 cells was associated with the integral components of stress associated biological condensates, which include i) non-translating mRNA pool, ii) ribonucleoproteins, and iii) RNA binding proteins. In support of our hypothesis, we found the ribosome, spliceosome and RNA pathways had 68 proteins upregulated in their T1 group (Figure 6A), implicated in RNA transport, metabolism, transcription and translation. We also analysed the other (non-protease) stress control (control1), where we grew Caco-2 cells in presence of DMEM under serum starvation (SS) for extended 48h. We did not observe obvious differences at the level of ribosomal proteins across T1 and T5 (SS) ( Figure 6B). However, interestingly we observed appreciable heterogenous and small ribonucleoproteins (Figure 7 A and B) upregulated in the trypsin treatment groups (T1 and T3) compared to stress control C1 (SS), indicating limiting ribonucleoprotein availability in SS conditions, which may discourage faithful RNP-concentration dependent formation of biological condensates. We further explored the potential of RNA binding proteins present in our dataset, which could offer potential anti-viral therapeutic value. We took leverage of the open access electronic resource The Mammalian Stress Granule Proteome (MSGP) containing all known SG recruited proteins reported to date (Nunes et al., 2019) available at https://msgp.pt/. We particularly compared the reported RBPs in MSGP database to our proteomic data set, and it was found that nearly three-fourth (171) of the total RBPs were present in our database out of which nearly one-third were upregulated (~ 1.4-fold, 0.5 log2) in T1 ( Figure 8A and B). Most notably, we identified 2 interferon stimulatory genes ADAR and PRKRA that were upregulated in T1 and in T1/T5 respectively ( Figure 9A and B). We observed appreciable number of RNA-binding RNA helicase proteins, which are implicated in viral pathogenesis. It is noteworthy that we identified several helicases in our dataset (Figure 10 A and B) that could be potential G4 interactors.

Host cell immunomodulation in trypsin-treated Caco-2 cells
We mapped the total proteins to the known viral pathways in the KEGG data base including-05168-Herpes simplex virus 1 infection, 05203-viral carcinogenesis, 05416viral myocarditis and 04061-Viral protein interaction with cytokine and cytokine receptor (Figure 11 A-D). With most proteins downregulated in these pathways under T1 conditions (green color restricted to left), it was intriguing to observe upregulation of few proteins in T1 (red -shit toward left side of box) including the host factors Oct-1 and HCF (in 05168-Herpes simplex virus 1 infection pathway), which are known to promote viral infection in host cells (detailed in discussion). The T1 treatment appeared to be conducive to productive viral replication/ infection and, therefore, we focused on upregulated proteins in T1 in remaining pathways (Figures 12,13) to establish a correlation (if any) between pro-viral activity in T1 (indicated by Oct-1 overexpression and interferon-downregulation) ( Figure 11A) with other potential host cell determinants as indicated in our data set to speculate a possible hypothesis underlying protease-mediated downmodulation of immune cell pathways, and identify potential host target proteins for viral drug development. Further, the viral nucleotide sensing cytosolic proteins (RIG-1 and cGAS, see discussion for details) eventually converge at generating antiviral interferon response (INF-a and INF-b) and proinflammatory cytokines. INF-g induced activation of INFGR (interferon gamma receptor) (upregulated in T3), which results in transcription of antiviral proteins showed upregulation of OAS interferon-stimulatory gene (ISGs) in T3, however, no IFN gene transcription was seen for any of the antiviral pathways at any of the trypsin concentration tested (Figures 11A, 12).

Induction of abortive-interferon anti-viral immune pathways in trypsin-treated Caco-2 cells
As shown (Figure 12 A-D) the major innate immune (TLR, RIG-I, cGAS, NF-B) pathway proteins were downregulated in T1 (green color restriction to left part of the box) as opposed to T3 (red color restriction to right part of the same box). Notably, the central (RIG-I RNA sensing pathway) protein ISG-ISG15 was appreciably downregulated in T1, modestly upregulated in T2 and over 2 fold upregulated in T3, thereby, indicating a likely "trypsin-dependent" suppression of interferon pathway intermediate/ISG15, which however fail to trigger the production of terminal effector antiviral interferons (at all the trypsin concentrations-T1, T2, T3), thereby resulting in an "abortive-interferon" pathway (Figure 12 and 13). This is in contrast to cyclic GAMP-AMP synthase (cGAS), which was downregulated in all the treatments (T1, T2, T3) (Figure 13). Intriguingly, we found upregulation of interferon protein ADAR (adenosine deaminase acting on RNA), RNA-editing enzyme in T1 (Figure 12C, 13C).
ISGs execute their inhibitory function by degrading viral RNA and/or blocking translation of viral mRNAs including 2′,5′-oligoadenylate synthetase (OAS) and latent ribonuclease L (RNase L), protein kinase R (PKR), Moloney leukemia virus 10 homolog (MOV10), and zinc-finger antiviral protein (ZAP), many of which were downregulated in our data set (figure 13). ISG15 is the most highly induced interferon stimulated gene, which can inhibit viral translation, replication, or egress. It is a ubiquitin-like protein with an ability to covalently attach to target proteins (ISGylation). Plethora of proteins are involved in ISGylation and deISGylation including UBE2L6, HERC5, HERC6, UBE1LA, TRIM25, and USP18. TRIM25 was downregulated in T1, T2, and T3 (Figure 13), and HERC4 was highly downregulated in T1 (log2 -4.0) (supplementary file 2). Since ubiquitination and phosphorylation are key regulatory post translation modification in innate immune RIG-I -ISG15-dependent signaling pathway, we tabulated all the proteins with keyword "ubiquitin" and "phosphatase" (Figure S3 and B), respectively. To our surprise we found many ubiquitinating/deubiquitinating proteins that may have potential implication in interferon response pathway including Lys-63-specific deubiquitinase BRCC36 (downregulated in T1-log2 -4.9), which is known to have role in interferon signaling by deubiquitination of the type 1 interferon receptor alpha (IFNAR1), resulting in stabilization and cell surface expression of INFRA1 ( Figure 13B).
We further did a "theme" based search with key word "Interferon" and tabulated 11 entries that were present in at least 2 groups ( Figure 13B). 7/11 proteins were down regulated in T1, and 2 proteins were not found. The 3 most downregulated proteins included Interferon-induced, double-stranded RNA-activated protein kinase (EIF2AK2), Interferon gamma receptor 1 (IFNGR1) and Interferon-induced 35 kDa protein (IFI35) that showed highest downregulation (log2 -4.0). The study by Kerr et al., observed an increase in ribosome RPL28, which was selectively incorporated into ribosomes/polysomes during the IFN response, they further reported an increase in ISG abundance upon RPL28 knockdown (siRPL28) in cells, indicating it as a negative regulator of ISG synthesis. We tabulated the proteins present in the heatmap (Kerr et al., 2020; Fig. 6H) showing abundance of select well-studied ISGs in siRPL28-treated and IFN-stimulated cells compared to controls. To our surprise, out of 12 proteins in their heatmap, we were able to find 10 proteins in our proteomics data including RPL28, which was over 2 fold upregulated in T1 (log2, 1.2), unchanged in T2 and modestly upregulated in T3. Further, we could correlate the downregulation of ISGs to the expression levels of RPL28 and interestingly, increased levels of RPL28 appreciably correlated with more downregulation of ISGs in T1 compared to T2 and T3 wrt Control 2 ( Figure 13D). The most highly downregulated ISG was CD55, a complement decay-accelerating factor that showed a remarkable downregulation (logF2 -4.9), other ISGs that were highly downregulated included anti-viral SAMHD1 (logF2 -2.49), proteasome subunit PSDM1(logF2 -1.9) and ISG15 (logF2 -1.8) in T1.
Trim21, a E3 ubiquitin ligase was reported as a negative regulator of anti-viral Nmi-IFI35 complex with implications for various autoimmune diseases associated uncontrolled antiviral signaling (Das et al., 2015). We did not find Trim21 in our data, however, Trim21 regulated ISG protein-Sterile alpha motif and histidine-aspartic acid domain-containing protein 1 (SAMHD1) was downregulated in T1 (log2 -2.5) ( Figure  13D). Further, we also tabulated the INF responsive proteins as reported in the mRNA expression profiles of proteins that were significantly upregulated at the proteomic level at 24 h of IFN stimulation . We could find 17 proteins in our data set that were present in at least 2 conditions out of 3 (T1, T2, T3), with top downregulated proteins being IFI35 (logF2 -4.0), STAT2 (logF2 -3.09), E3 ubiquitinprotein ligase RNF213 (logF2 -3.02), and E3 ubiquitin/ISG15 ligase TRIM25, which ubiquitinates RIG-I and activates it ( Figure 13E).
We further explored the spectra of ISGs present in our proteomic profiling data set by tabulating shared ISGs ( Figure 13C) as reported in (Lamers et al., 2020). Interestingly, we could find total of 15 proteins that were present in at least 2 treatments out of 3 (T1, T2, T3 wrt control2) and 1 protein (TRIM56) was present only in T3. Most of the proteins were downregulated in T1 as (as opposed Lamers et al., results) and interestingly, modestly upregulated in T3. We observed another interesting shared result with Lamers et al., as the expression of 3 genes (downregulated in the expansion organoids/ intestinal progenitor populations) compared to more differentiated organoids (Lamers et al., 2020) were found to be downregulated in our cell culture model of trypsin treated Caco-2 cells. These included Transthyretin/TTR (logF2 -2.92), Apolipoprotein A-1/ APOA1 (logF2 -4.3) and Galectin-2/ LGALS2 (logF2 -2.76). The level of downregulation for APOA-1 and LGALS2 was higher in T1, however, interestingly, TTR was more downregulated in T3 (log2 -4.30) (supplementary file 2).

Extending the Interferon-Ribosomal-RNA pathway axis to Coagulation-Complement Networking via thromboinflammatory SERPINS
We observed downregulation of a coagulation-complement pathway intermediate serine protease inhibitor A1AT (alpha-1 antitrypsin)/ SERPINA1 in all three trypsin treatments (T1, T2, and T3) (Figure 14A-C). Encouraged by the presence of SERPIN (serine protease inhibitor) A1AT in our system, we explored other potential SERPINS (if any) present in our data set (that may have potential thromboinflammatory regulating functions) using a keyword "serine protease inhibitor and/or SERPIN". We could find 7 serine protease inhibitors with implications in thromboinflammatory diseases. 7/7 serine protease inhibitors were downregulated in T1, and interestingly, 6/7 serine protease inhibitors except SPINT2 were downregulated in T2 and T3 ( Figure 14B). SERPINS downregulation at all trypsin concentrations indicated a likely similar pattern of downmodulation as seen for abortive-interferon response in T1, T2, and T3. We further searched for serine proteases present in our data set to assess their expression pattern and draw a correlation (if any) with protease/antiprotease balance and interferon response. Total of 3 serine proteases were identified in our dataset as opposed to 7 SERPINs ( Figure 14C). The serine proteases present in the data set included Prostasin (PRSS8), Suppressor of tumorigenicity 14 protein (ST14) encoding Matriptase, and Serine protease HTRA2, mitochondrial (HTRA2), which showed the following expression levels in T1 (log2 -4.25, log2 -0.79 and log2 1.46) respectively, indicating a more complex regulation of proteases compared to serine protease inhibitors, which were all downregulated in T1. Prostasin was recently shown to have an anti-inflammatory effect via downregulation of TLR4 expression in colonic epithelial cells (Sugitani et al., 2020). SPINT1 was the most downregulated serine protease inhibitor in T1 (log2 -4.0), SERPINB6 in T2 (log2 -4.0), and SERPINA1 or A1AT in T3 (log2 -2.76). Interestingly, SERPINB6 and A1AT were appreciably downregulated in all 3 treatments of trypsin (T1, T2, and T3) ( Figure 14C . Dissecting pathophysiology of COVID-19 in terms of interplaying mechanisms that underlie SARS-CoV-2 virulence, human immune response, overactive inflammatory reactions and majorly the coagulation pathways (Kowalik et al., 2020) is urgently required. Identifying and targeting possible molecular determinants driving coagulation-thromboinflammation axis/ loop in COVID-19 to achieve hemostasis rebalancing, is a potential novel area to design better therapeutic options for COVID19 treatment.
The current study correlates the partial loss of A1AT, a coagulation-complement pathway protein (potent neutrophil elastase inhibitor) to immunomodulation of protease-treated Caco-2 cells. Based on our preliminary findings (discussed below), we propose the following hypothesis-"down-regulation of A1AT enhances proteasemediated (virus-competent) immunometabolic reprograming of Caco-2 cells, rescuing this loss by exogenous A1AT may be a promising candidate in treating viral pathologies like COVID-19. The unique immunological profile marked by a profoundly impaired interferon (IFN) type I response in parallel to massive proinflammatory responses (host-tissue destruction) in COVID-19 disease (Hadjadj et  Based on the emerging studies and other drug candidates implicated in COVID-19; it emerges that the potential targets for COVID-19 treatment include-1) viral proteases, 2) host serine protease TMPRSS2, 3) ribosomal/ RNA pathways, 4) host proinflammatory cytokines, 5) anti-viral interferon pathways, and 6) coagulation pathways. We had a "virus-free" cell culture model with complete absence of proinflammatory cytokines (as opposed to cytokine storm seen in COVID-19). Therefore, we excluded 2 druggable targets (viral proteases and proinflammatory cytokines) owing to their absence in our in vitro culture system that lacks infective viral loads. We tried to gather insights from our high throughput proteomics data and recent literature on COVID-19 pathology, to establish a functional/mechanistic network with an effort to understand the cross-talk between innate immune interferon response and ribosomal/ RNA pathways, which may help to develop improved strategies/ repurpose drugs to treat COVID-19 disease.
The proteases are integral to viral and tumor pathology and emerging studies are establishing parallels in metabolic reprogramming between oncogenesis and human oncogenic viruses, which hijack host metabolite and lipid machinery to leverage progeny virus formation (Purdy and Luftig, 2019). Further, the interplay of cellular metabolism (immunometabolism) with innate immune response has been linked to respiratory viral infections (Zhang et al., 2018; Cheng et al., 2014), pointing to a (possible) viral infection-associated metabolic remodeling for achieving host immunomodulation. Interferon signalling induced alterations in ribosome composition that selectively downregulated ISG synthesis associated with IFN response modulation, offers a differential network map of the IFN-induced interactome to interpret IFN response in the context of viral infection and autoimmune diseases (Kerr et al., 2019). Notably, the IFN stimulation response extends beyond antiviral program and can facilitate ISG induction more broadly by activating mTOR-AKT-S6K signalling pathways of the MAPK signaling cascade (Ivashkiv and Donlin, 2012). The evolving interferon biology landscape implicates interferon/ISG signaling in regulation of mRNA processing, post-translational modification, metabolism, cellular trafficking, chromatin organization, and the cytoskeleton pathways (Schneider et al., 2014), which underlie pathological (viral, autoimmune and tumor biology) and physiological (stem cell regenerative pathways) processes.
Recent emerging paradigms point towards an increased trend towards "trypsinreliance" for zoonotic coronavirus emergence that might predict cross-species transmission of coronaviruses possibly employing alternative route(s) like digestive tract/gut, which is proposed to be a potential site for future coronavirus emergence events in humans and perhaps alternative yet unidentified ACE2-independent hostcell receptors (Shulla et et al., 2020). These paradigms encouraged us to explore the potential mechanisms that might be active at the trypsin (protease)host interface, which may extend beyond the layer of "viral-spike processing" and could possibly be linked to "host-receptor alterations" (Letko et al., 2020).  , 2019), we speculated that the pathogen/protease induced (metabolic) reprogramming events may underlie SARS-CoV-2 mediated infection of host cells to trigger viral-competent immunomodulatory pathways. We were interested to investigate the proposed hypothesis based on the recent findings from diverse groups reporting the "co-emergence/ -existence of SARS-CoV-2 and reprogramming events" indicating extended SARS-CoV-2 associated reprogramming trajectories including the possible haematopoiesis, (Blanco-melo et al., 2020) , preferred enterocyte progenitor (ACE2 low) permissive replication (Lamers et al., 2020), and transit-amplifying cells or intestinal stem cells supporting SARS-CoV-2 infections (Zhang et al., 2020). We were intrigued to explore if CoV associated serine protease(s) facilitated host/Caco-2 cell remodeling to usurp its cellular machinery resulting in dampened immune response.
Our preliminary proteomics data with trypsin-treated Caco-2 (SARS-CoV-2 permissive gut/intestinal) cells suggests that the enhanced bioenergetics in trypsin treated Caco-2 cells might drive metabolic resetting, which can facilitate viral competent immunomodulation. Assembly of the OXPHOS system (upregulated in trypsin treatment, Fig 3) is an intricate and vaguely defined process (Signes and Fernandez-Vizarra, 2018), however, the mitochondrial F type -ATP synthase, is established to be critical for mitochondrial functions. The deregulation of this enzyme was recently shown to be associated with dampened mitochondrial oxidative phosphorylation (OXPHOS) and activated mitochondrial permeability transition via OSCP (ATP5O) (Beck et al., 2016). Moreover, defects in COX5A overexpression were reported to attenuate mitochondrial respiratory dysfunction (Zhang et al., 2020). PB1-52, a mitochondrial inner membrane interacting viral protein has emerged as an innate immune modulator and virulence factor that regulated Type 1 interferon response (Cheung et al., 2020), thereby reflecting a possible host cell-metabolism rewiring that may regulate interferon responses for pathogen benefitting. Emerging studies are increasingly implicating RNA metabolism to viral immune evasion (Oyarzun-Arrau et al., 2020), and SARS-CoV-2 is also proposed to camouflage its RNA from innate immune recognition by modifying RNA capping machinery (Encinar and Menendez, 2020). Importantly, certain interferon-stimulated genes (ISGs) including PKR, ADAR1, RIG-I, RNase L, and OAS also belong to the family of RNA binding proteins, therefore, anti-viral interferon response is interlinked with regulation of MO biogenesis/disassembly. SARS-CoV-2 transmits rapidly and its infection has been associated with poor anti-viral interferon response. Therefore, we further investigated if SARS-CoV-2 associated protease trypsin can induce stressgranule like proteome marked by suppressed interferon response. Interestingly, we observed that trypsin-primed Caco-2 cells displayed abortive-interferon proteome under differential stress conditions that was marked by complete lack of INF generation despite induction of intermediate ISGs under certain treatments. We asked if the lack of interferon response was a generalizable phenomenon related to the translational halt triggered by cellular stress, or if we had some success in simulating the SG-like proteome phenotype, which is linked to regulation of interferon pathway as many interferon proteins like ADAR and PRKRA (RNA binding proteins) can interact with the SG components. We briefly discuss our findings in the subsequent section.
We included protease-primed stress conditions in our experimental set up with a goal to develop protease-mediated induction of host -dependent (viral regulatory) biological substructures that may offer investigation of interferon-regulatory networks with implications in viral pathology. The cellular-stress inducers included-a) Caco-2 cells grown under serum-free conditions in DMEM medium with no added growth supplement (SS/ C1)-T5; b) Caco-2 cells grown under trypsin (50 µg /ml) conditions-T2 (CPT); c) Caco-2 cells grown under ARPE19 conditioned media as established by the author (Sharma et al., 2013; Sharma et al., 2017; Sharma et al., 2019). This conditioned media is marked by presence of activated trypsin (50 µg /ml), IL-1b, and biologically active bFGF (basic fibroblast growth factor). IL-1b is strongly implicated in COVID-19 pathology, therefore, we asked if we could more closely approximate the cellular stress underlying SARS-CoV-2 infection by combining trypsin and proinflammatory IL-1b (CPT-IL). We envisaged it to be a more potent stress inducer than CPT, and d) Caco-2 cells grown under escalating trypsin dose (500 µg /ml)-T1 (CPT-EC). Notably, trypsin (Catalog # 17-161E, Lonza) used in this study was procured from Lonza, which comes at a working concentration of 500 µg /ml with 5mM dextrose, and physiologic ionic concentration of 5mM KCl and 137mM NaCl. However, the concentration of sodium bicarbonate is around 7mM in trypsin solution as opposed to around 44mM bicarbonate in DMEM media. Since bicarbonate exists as a conjugate base (the dissolved bicarbonate ions in the medium) to control a stable physiological pH through the bicarbonate buffering system under artificial cell culture conditions in vitro, we consistently observed a slight drop in pH (it was around 6.9) over 48h priming of Caco-2 cells in trypsin T1 conditions. Further, cell culture media of a standard composition contains source of energy and compounds which regulate the cell cycle and supplementation with amino acids, vitamins, inorganic salts, glucose, and serum as a source of growth factors, hormones, and attachment factors. In addition to nutrients, the medium also helps maintain pH and osmolality. The T1 condition with no added amino acids, vitamins or growth factors could support the survival of cells (Figure 1 A) until 48h post trypsin-inclusion. Apparently dextrose (1000mg/ml or 5mM) was sufficient to keep cells viable under stress. We believe that the trypsin activation of its receptor PAR-2 on Caco-2 cells resulted in upregulation of various ion channels and transporters (Figure S3 C) to further regulate the osmolality of the system. Glutamine and pyruvate have also been excluded from T1 condition (as there is no basal media), yet the cells could cope up with stress and survived for 48h indicating yet unidentified pathways to be driving survival under stress. Importantly, recent studies implicate glutamine independence as a selectable feature of pluripotent stem cells (Vardhana et al., 2019). Presence of anaplerotic molecules, pentose phosphate pathway intermediates and purine-pyrimidine pathway intermediates are also indicated in our data, however, detailed investigation would provide better insights to the metabolic pathways governing T1 cell survival under extreme stress. We attempted to test this culture condition as we were aiming to establish proteasemediated LLPS (liquid-liquid phase separation) driven stress granule formation in vitro. It is reported that LLPS formation is facilitated by temperature, pH, salt, and crowding agent dependent manner (Fu, 2020), therefore, we hypothesized that physiological salts and slight pH alterations (and perhaps perturbed osmolality) might promote SGlike features in vitro under CPT-EC conditions.
Our preliminary findings show near-global downregulated protein expression with lower trypsin concentrations (T2/CPT) closer to the proteome patterns of stressed, nutrient-deprived serum starved (T5/ SS) cells (Figure 1B), indicating low translational activity. Surprisingly, distinct and moderately stretched tracts of overexpressed proteins appeared in proteome of Caco-2 cells primed with trypsin at escalating concentrations, (CPT-EC), which could be majorly mapped to ribosomal, RNA transport, and spliceosome ribonucleoproteins (Figures 5-10). Intriguingly, the proteome of T1 and T3 contained upregulated SG-forming components, including ribonucleoproteins and RNA-binding proteins (Figures 7-10) that were absent or downregulated in T2 and T5 , indicating propensity of T1 and T3 towards formation of SGs. However, T3 exhibited overwhelming expression of protein tracts as seen for Caco-2 cells cultured under growth-factor rich serum conditions (C2) indicating a likely de-repression of trypsin-induced translational halt, and diminishing possibility of formation of SGs owing to global translation upregulation as seen for serum treated Caco-2 cells (Figure 1B). It appears that the inclusion of growth factor bFGF in the T3 conditions released the apparent translational halt in T3 treated cells. Intriguingly, the inclusion of proinflammatory IL-1b to trypsin treated Caco-2 cells (CPT-IL) failed to trigger INF and proinflammatory cytokines, despite the presence of IL-1b in the cell culture milieu suggestive of protease/ trypsin-dependent regulation of INF response, this has a close resemblance to refractory interferon response amidst proinflammatory cytokine storm as seen in COVID-19 disease, suggesting a yet underappreciated role of serine proteases in regulating anti-viral interferon response. We further investigated if interferon stimulated genes were a part of the SG-like proteome in T1 conditions. It is established that following viral infection the cells happen to sense dsRNA (viral) detection, and alternate between periods of translation inhibition and process of active protein synthesis. The cytokines mRNA translation is hampered by translation inhibition (as appear under T1 conditions), which is mediated by dsRNA-induced PKR activation and consecutive translation initiation factor eIF2a phosphorylation indicating translational halt. PKR-dependent eIF2a phosphorylation is key for translation shut-off and stress granules formation (Dalet et al., 2015). Interferon induced proteins like ADAR and PRKRA are RNA binding proteins that can interact with the SG components and are negative regulators of interferon response. ADAR deficiency is shown to cause chronic type I interferon production and inflammatory pathology in humans and mice (Hartner et al., 2009;Rice et al., 2012). A recent study linked the loss of ADAR1 in tumours to MDA5-dependent IFN-I production and inflammation (Ishizuka et al.,  2019).
The proposed mechanisms that cells may utilize to restrict the formation or disassembly SGs/ MOs involve regulation of RNA synthesis rates and RNA modifications underlying RNA-RNA interactions, ribosome association and regulation of RNA concentrations through RNA decay pathway (Burke et al., 2019, Tauber et  al., 2020). RNA sensing is central to viral infections, therefore, RNA unwinding enzymes called RNA helicases (DEAD-box proteins) can perturb RNA-RNA interactions by invoking their inherent ATPase activity. This can block the formation of RNP condensates like stress granule, hence RNA helicases can counteract biological RNA-RNP condensate formation as has been recently shown for RNA helicase eIF4 (Tauber et al., 2020). Four-stranded G-quadruplex (G4) secondary structures are key structural features in viral mRNA and they tend to be highly conserved despite the high rate of mutations in viruses, indicating the importance of G4 targeting approach for broader antiviral therapy. Viral G-quadruplexes are new frontiers in virus pathogenesis and antiviral therapy (Ruggiero and Richter, 2020). Notably, G4s are enriched in functionally important regions in mRNA including 5 ′ -and 3 ′ -untranslated regions (UTRs), and recognition of mRNA secondary structures by RNA binding proteins (RBPs) is essential for post-transcriptional modifications. Therefore, recognition and unwinding of G4s are thought to be important for post-transcriptional processes such as mRNA translation, transport or stability. Interestingly, the major group of G4 interacting proteins consist of proteins involved in RNA splicing and processing, followed by proteins involved in translation. A dynamic balance between formation and resolution of G4 structures in host cells and presence of conserved G4 structures in viruses, offer a promising window of opportunity for targeting viral pathologies. Our data presents appreciable expression of RNA helicases in T1 and T3 (Figure 10 A-B) indicating ensuing secondary structural alterations in mRNA, which may underlie the formation of SG-like condensates under T1 conditions.
Other candidates that can regulate translational outputs under T1 primed conditions include PUS7 and HTATSF1. It was recently reported that PUS7 bound to distinct tRNAs to control biogenesis of tRNA derived fragments and hence regulated translation in stem cells (Guzzi et al., 2018). HTATSF1 was recently shown to regulate ribosomal RNA transcription and processing and subsequent efficient protein synthesis by specifically controlling splicing and intron retention (Corsini et al., 2018). Nuclear retention of incompletely spliced or mature mRNAs is emerging as a novel layer of gene regulation, which sensitizes the cell to instantly respond to stress, viral infection, differentiation cues or changing environmental conditions (Wegener et al., 2017). Notably, the host RNA modification inventory is dynamic, and emerging modulators of RNA virus gene expression have been proposed (Netzband and Pager, 2019). These RNA regulatory events can significantly impact virus gene expression, regulation, cytopathology and pathogenesis. Therefore, RNA viruses use, usurp and/or avoid the associated RNA machinery to impact the outcome of infection (Cross et al., 2019). RNA splicing is a new paradigm in host-pathogen interactions (Chuahan et al., 2019), and it was recently proposed that the alternative splicing landscape of host-cells determined the outcome of infection (productive vs abortive) (Boudreault et al., 2019). Recent studies report exciting convergence of the epitranscriptomic and virology research areas involving the N7-methylguanosine, ribose 2' -O-methylation, pseudouridine, inosine, N6-methyladenosine, and 5methylcytosine epitranscriptomic marks on gene expression of RNA viruses. Interestingly, viral interaction and subversion of host cellular RNA transcription, splicing, translation, storage, and decay pathways underlie faithful viral infection cycles (Cross, Michalski . We believe that our data set might offer novel host cell candidates/determinants that may be potential targets for viral drug discovery. The functional characterization of such candidates will provide viable drug discovery platforms. The predominant candidates in our data that support the display of SG-like proteome signatures belong to the family of snRNPs, and many splicing factors, including the hnRNP proteins and the serine/arginine (SR)-rich protein family, which are reported to identify primary cis elements in the pre-mRNA (Sebbag-Sznajder et al., 2020). Interestingly, recent data implicate LLPS as a pre-requisite for transcription activation (Kantidze and Razin, 2020). Nuclear speckles (subnuclear compartments) are hubs of active genes/ pre-mRNA metabolism and are driven by ribonucleoproteins dependent large-scale condensates (Smith et al., 2020), reinforcing the emerging role of pre-mRNA in mediating euchromatin regulation (Li and Fu, 2019).
The POU transcription factor Oct-1 is known to represses virus-induced interferon gene expression (Mesplede et al., 2005), and promote disruption of viral latency (Robinson et al., 2011). The mammalian transcriptional coactivator host cell factor-1 (HCF-1) is known to function in concert with Oct-1 and VP16 to assemble the HSV immediate early (IE) transcription enhancer core complexes, which mediated the enhanced transcription of these genes upon infection (Narayanan et al., 2005), and triggered herpesvirus reactivation (Carroll et al., 2007). Both Oct-1 and HCF were upregulated in T1 conditions in our data ( Figure 11A, supplementary file 1). Further, type I interferons can act in an autocrine/ paracrine manner by binding to their receptor (interferon-α/β receptor-IFNAR), which activates the Janus kinases (JAKs). JAKs subsequently activate the transcription factors signal transducer and activator of transcription 1 (STAT1) and STAT2, resulting in expression of major anti-viral interferon-stimulated genes (ISGs) (reviewed in Rehwinkel and Gack, 2020 and cited references). Multiple signaling pathways that are targeted by viruses tend to converge on preventing the phosphorylation, dimerization, and nuclear translocation of the transcription factors IRF-3/7 (Soham et al., 2020). The coordinated ISG-ISGylation network has been recently shown to play a central role in providing anti-tumor immunity (Fan et al., 2020). IFNs activate the JAK-STAT pathway, which triggers induction of many IFN-stimulated genes (ISGs), and not all of these have been completely characterized (Stark et al., 2012; Borden et al., 2019). Since ISG inventory is dynamic and many ISGs are yet to be reported/ characterized, we consulted recent rigorous papers (Kerr et al., 2019; Lamers et al., 2020) and tabulated shared ISGs and Interferon Response Genes (IRGs) from these studies (comparing them to our data proteins) to extract meaningful ISGs/ IRGs (Figure 13 C-E), which may have a potential role in viral pathogenesis.
IFI35 and CD55 emerged as the highest downregulated (log2Fc -4.0) ISGs in our data set (see results section). IPF35/ IFI35 was reported as a potential cellular factor that attenuated the IFN response by negatively regulating RIG-I antiviral signaling leading to vesicular stomatitis virus replication (Das et al., 2014), increasing susceptibility to highly pathogenic H5N1-and pandemic H1N1-influenza A infection (Gounder et al., 2018;, and promoting sustained virological response to pegylated-interferon plus ribavirin treatment in chronic HCV infection (Estrabaud et al., 2015). Interestingly, intracellular IFP/ IFI35 and its partner NMI (N-myc interacting protein) are reported to act as negative regulators of innate immune signaling pathways, targeting RIG-I, nuclear factor IFN regulatory factor 7 and NF-κB (Wang et al., 2013;Hou et al., 2016). However, recent reports assign them exciting (extracellular) role as proinflammatory DAMPs (danger associated molecular patterns), which activate the NF-κB signaling pathway through TLR4 and facilitate the release of TNF (Xiahou et al., 2017). Therefore, IFI35 appears to have a context dependent role during inflammatory pathologies. A reduced intracellular IFI35 in the our study is suggestive of a suppressive interferon response, however, our study "did not evaluate the secreted extracellular proteome signatures". Therefore, it would be interesting to assess the extracellular levels of the IFI35 as enhanced trypsin concentrations might induce certain forms of cell death/ cell debris, we expect (some extent) of collateral detrimental effects to contribute to likely generation of DAMPs in our cell culture model. DAMPs can be produced or released by damaged and dying cells facilitating sterile inflammation, which underlies tissue repair and regeneration, however, overactivation of DAMP signaling or lack of counter measures can result in development of numerous inflammatory diseases including metabolic disorders, autoimmune diseases, and tumours. Emerging studies clearly demonstrate that PAMPs and DAMPs can initiate immune responses by activating PRRs-including TLRs, RIG-I-like receptors (RLRs), C-type lectin receptors (CLRs) and multiple intracellular DNA sensors. Alternatively, DAMPs can also be sensed by non-PRRs that include receptor for advanced glycation end products (RAGE), triggering receptors expressed on myeloid cells (TREMs), Gprotein-coupled receptors (GPCRs) and ion channels (Gong et al.,2020). The predominant DAMP present in our data set include Histone deacetylase complex subunit SAP130 (log2 3.4) and expression (up/down) of other histones ( Figure S4). A future work up to identify and annotate functional roles to the possible DAMPs in our current experimental model, and upgrading the cell-culture model (by adding viable SARS-CoV-2 viral load) to recapitulate infectious disease would phenocopy PAMPdriven 'pathogenic inflammation' and signaling events, as opposed to likely DAMPdriven 'sterile -inflammation' in the current model.
Interestingly, SAMHD1 (other potential ISG downregulated in our dataset) was initially identified as an interferon-g-induced gene, and an anti-HIV-1 restriction factor (Hrecka et al., 2011;Laguette et al., 2011). Its anti-viral activity is attributed to depletion of the intracellular pool of deoxynucleotide triphosphates (SAMHD1 is a deoxynucleoside triphosphate triphosphohydrolase) thereby limiting viral DNA synthesis. HIV-2 and SIV evolved the Vpx protein to degrade SAMHD1 via the proteasome pathway by hijacking the Cul4A/DDB1/DCAF1 E3 ligase, and a recent study demonstrates Trim21 dependent degradation of SAMHD1 in modulation of its anti-viral activity Coggins et al., 2020). DDB1-and CUL4-associated factor 7 (DCAF7) (log2Fc -1.7), Cullin-4A (CUL4A) (log2 Fc -2.3) and DNA damage-binding protein 1 (DDB1) (log2Fc -2.4) were downregulated in T1 (supplementary file 2). Another recent study implicated O-GlcNAcylation of SAMHD1 in promoting anti-HBV immunity as opposed to its anti-viral activity towards HIV (Hu et al., 2020), indicating another complex and crucial role of metabolic reprogramming underlying viral pathogenesis. The ubiquitin deconjugases encoded in the N-terminal domain of Epstein-Barr virus (EBV), Kaposi Sarcoma herpesvirus (KSHV) and human cytomegalovirus (HCMV) proteins were reported to target an early step of the IFN signaling cascade characterized by trimolecular complex formation with the ubiquitin ligase TRIM25 and the 14-3-3 molecular scaffold (Gupta et al., 2020). TRIM25 is suppressed in our study upon trypsin treatment of Caco-2 cells. These studies again support the absence activated interferon immune pathways in trypsin-treated Caco-2 cells.
Further, ISGs upregulated in intestinal organoids upon SARS-CoV-2 infection were recently reported (Lamers et al., 2020). Since our source/host cells were Caco-2 derived intestinal epithelial cells, we found it a reasonable reference data set to evaluate if we had any similar ISGs. A comparable data would strengthen our hypothesis that implicates the role of protease treatment of Caco-2 cells in partially recapitulating the viral infection. A unifying observation in Lamers et al., and our study was the lack of robust interferon response. Despite the induction of ISGs, the authors (Lamers et al., 2020) did not observe a robust Type I and III interferon response. Therefore, with our current preliminary findings, we propose that our trypsin treated Caco-2 cells proteomics data partially and appreciably recapitulates the features of SARS-CoV-2 infected intestinal organoids. Nevertheless, the lack of appreciable antiviral interferon response despite the robust (Lamers et al., 2020) and modest (in our T3) upregulation of ISGs, point to a yet unidentified layer of complexity that regulates interferon production under in vitro conditions. We do not have additional data (at this point) to justify this observation, however, we could speculate a higher propensity of virus infection in Caco-2 cells that are treated with trypsin as opposed to serum (rich in trypsin-inhibitors). The "trypsin-reliance" for enhanced infection under in vitro conditions in presence of viable virus loads (missing in our experimental design) would recapitulate the current emerging studies, which demonstrate indispensable trypsin requirement of coronaviruses under in vitro conditions for productive infection underlying zoonotic emergence (Letko et al., 2020, Menachery et al., 2020. We are currently attempting organoid generation using diverse trypsin formulations employing ARPE19 and Caco-2 cells (Sharma et al. unpublished  Encouraged by the coagulopathy manifestation in COVID-19 patients, we considered re-analysing the complement-coagulation KEGG pathway to see if we had some potential hit(s) that could help us establish mechanistic networks with interferon and ribosome/RNA pathways. Uncontrolled enzyme activity is known to have pathological consequences as in thrombosis marked by excessive activity of serine proteases of the coagulation system and kallikrein-kinin system. Serine proteases are regulated by SERPINs (serine protease inhibitors), which are widely implicated in regulating thromboinflammatory processes and hemostatic maintenance/ rebalancing (Crocetti et al., 2019;Emmerich, 2009). We explored the potential serine protease and SERPINS present in our data set with implication in thromboinflammatory disorder like COVID-19. Our data identified seven serine protease inhibitors, which were all downregulated in T1, and interestingly, 6/7 serine protease inhibitors except SPINT2 were downregulated in T2 and T3 (see results for details). We briefly discuss their established roles in inflammatory disorders. The regulation of protease/antiprotease balance during pathologic insults (marked by up or down regulation of serine protease/ SERPIN), is very complex. Type II transmembrane serine proteases such as matriptase and kallikreins, are known to activate protease activated receptor-2 (PAR-2), and regulated by SERPIN SPINT1/ HAI-1 (Kawaguchi et al., 2020). Matriptase is regulated by hepatocyte activator inhibitors (HAIs), HAI-1 and HAI-2, encoded by the genes serine peptidase inhibitor Kunitz type -1 and -2 (SPINT1 and SPINT2), and a recent study demonstrated intestinal regulation of suppression of tumorigenicity 14 (ST14 encodes matriptase) and serine peptidase inhibitor SPINT1 by intestinal transcription factor CDX2 (Danielsen et al., 2018); further SPINT2 was found to be an essential cellular inhibitor for maintaining intestinal epithelium architecture (Kawaguchi et al., 2019). Overactivation of HGF (hepatocyte growth factor) receptor MET by upregulation of matriptase and hepsin driven by downregulation of SPINT1 was recently reported in urological cancer (Mukai et al., 2020), indicating the crucial role of protease/antiprotease balance in inflammation, tumor biology and tissue integrity.
Recent studies implicate coagulation proteases like TF: FVIIa and FXa in activation of PAR-2, which is linked to pro-viral responses via TLR3 ( A recent study proposed that targeting FXa and thrombin by using specific anticoagulant drugs can inhibit PARs, which may have beneficial role in human inflammatory diseases (Posma et al., 2019). PAR-2 signaling can also be triggered trypsin, tryptase, neutrophil elastase, GPI anchored serine protease (PRSS8) prostasin, and membrane anchored hepsin and TMPRSS2 (a potential COVID-19 associated serine protease). Notably, PRSS8/ prostasin is downregulated in T1, and upregulated in T2 and T3 (Figure 14C), and we did not detect PAR-2 and TMPRSS2 expression in our data; however, beta-arrestin (PAR-2 regulator) was found to be modestly downregulated in T1 and T2 and modestly upregulated in T3. A recent study demonstrated increased activation of PAR-2 in beta-arrestin depleted cells (Thibeault and Ramachandran, 2020), indicating possibility of functional trypsin-mediated-PAR-2 signalling in PAR-2 replete Caco-2 cells. Nevertheless, a detailed mechanistic insight of PAR-2 signaling outcomes in Caco-2 cells treated with varying concentrations of trypsin will determine its candidature for regulating proteasedriven immunomodulatory pathways and evaluate if PAR-2 targeting can have beneficial effects in context of infectious pathology, notably PAR-2 mediated NETosis was recently linked to pathogen benefits (Bryzek et al., 2019). We suggest that the activation of PAR-2 by trypsin, coagulation serine proteases and neutrophil elastase, makes it a tempting target of A1AT (neutrophil elastase inhibitor), which may exert its anticoagulant effects (in part) by inhibiting coagulation serine protease-PAR-2 signaling node, besides its potential role in binding complement C3 component and inhibiting the downstream coagulation-complement pathways as has been recently reported (O'Brien et al., 2020).
As discussed before, the interferon treated cells preferably incorporated RPL28 ribosomal subunit to regulate the ISG expression. We found remarkable similarity in the overexpression of expression of RPL28 in T1 and concomitant downregulation of ISG CD55 (Figure 13C), which is a complement pathway protein. Decay-accelerating factor (DAF or CD55) is a known inhibitor of the complement system and CD55 deficiency results in complement hyperactivation, and angiopathic thrombosis. In addition, CD55 has been reported to be an essential host receptor for infection by the malaria parasite (Dho et al., 2018), however, CD55 was found to be the most potent complement regulatory protein in counteracting complement-mediated virus inactivation (Rangaswamy et al., 2016). Parainfluenza virus incorporated CD55 and CD46 to delay complement-mediated neutralization and exhibited an increased potency of inhibiting complement-mediated neutralization with CD55 (Li and Parks, 2018). Interestingly, a recent study demonstrated a correlation between complement component 3 (C3) and alpha-1 antitrypsin deficiency (AATD), displaying airways disease and pulmonary emphysema (O'Brien et al., 2020); implicating A1AT in thromboinflammatory diseases.
A1AT is at the intersection node of coagulation and innate immune system. A1AT variants are proposed as potential SERPINs to block contact system-driven thromboinflammation (de Maat et al., 2019). Interestingly, a spontaneous mutant of A1AT, A1AT-P (M358R)-acquires antithrombin as opposed anti-neutrophil elastase activity, thereby linking A1AT and its variants to "serine protease/ anti-protease stoichiometric deregulation" underlying coagulopathies (such as COVID-19). Further, A1AT-P scaffold has been adapted to acquire functionality to inhibit furin and complement factor C1s to yield a Serp1 (viral serpin), inhibiting furin pro-convertases mediated viral pathogenicity (Tardif et al., 2010). Serpin PC (also called 1 -AT KRK) and 1 -AT-Fc fusion proteins are other designer SERPINs in early clinical trials (NCT04073498; NCT03815396, 2019). The crystallographic datasets for wild type and mutant AAT proteins point to considerable variability in the surface clefts, indicative of AAT protein as a potential target for structure-based drug designing (Janciauskiene     . Total of 26 pathways that are implicated in viral pathogenesis were analyzed using KEGG pathway analysis (A). Expression tag that were used to highlight the protein expression of Treatment1, 2, 3 with respect to Control2. The left one-third of each box is position assigned to T1 and right one-third is position is for T3, mid-part is for T2. The red color denotes upregulation and green color is down regulation on, the colored scale denotes log2 fold change. Figure 5. KEGG pathway analysis of total 3088 proteins showing significant pathways (p-value cutoff 0.05 calculated by hypergeometric distribution) identified in all treatments (upper panel); the relative abundance in T1, T2 and T3 with respect to control 2 (Caco-2 cells grown in DMEM and 10% serum) (bottom panel). Ribosome Pathway (A), Spliceosome pathway (B), RNA transport pathway (C). The bottom panel could not map a fraction of proteins accurately to the assigned color codes. These proteins are marked by blue arrow for T1 proteins. The link to these pathways is provided in figure 4. Figure 6. The proteins upregulated in T1 from figure 5 (A, B, and C-bottom panel), were tabulated and compared to relative fold changes in T2 and T3 and T5 (C1, positive stress control-it represents the proteome of serum starved Caco-2 cells) over Caco-2 cells grown in serum (control 2). The red color in table denotes all proteins with log2 fold change of 0.5 (~1.4 fold) and above. Bottom of each box is divided in 4 quadrants (q) (q1-T1, q2-T2, q3-T3, q4-T5, from left to right) and represent log2 FC of T1, T2, T3 and T5 over Control 2 (A). Heat map of the proteins searched with key word "ribosom/al " proteins depicting the log2 FC of T1, T2, T3 and T5 over C2 (B).

Figure 7.
The tabulated proteins represent all the proteins searched with key word "ribo/nucleo/protein" in the total data set and compared to relative fold changes in T1, T2, T3 and T5 (C1, positive stress control-it represents the proteome of serum starved Caco-2 cells) over Caco-2 cells grown in serum (control 2). The red color in table denotes all proteins with log2 fold change of 0.5 (~1.4 fold) and above (A). Heat map of the proteins tabulated in (7A), depicting the relative log2 FC of T1, T2, T3 and T5 over C2 (B). , and available at https://msgp.pt/ . The RBPs are tabulated for relative log2 FC in T1, T2, and T3 over Caco-2 cells grown in serum (control 2). The red color denotes all proteins with log2 fold change of 0.5 (~1.4 fold) and above (A). Heat map of the proteins tabulated in (8 A) and T5, depicting the relative log2 FC of T1, T2, T3 and T5 over C2 (B).  Figure 10. The tabulated proteins represent all the proteins searched with key word RNA helicase / DDx/ DHx in the total data set and compared to relative fold changes in T1, T2, T3 and T5 (C1, positive stress control-it represents the proteome of serum starved Caco-2 cells) over Caco-2 cells grown in serum (control 2). The red color in table denotes all proteins with log2 fold change of 0.5 (~1.4 fold) and above (A). Heat map of the proteins tabulated in (8 A), depicting the relative log2 FC of T1, T2, T3 and T5 over C2 (B). Figure 11. KEGG pathway analysis showing relative log2 FC expression of proteins in T1, T2 and T3 with respect to control C2. Blue arrows denote proteins not mapped accurately to color gradations. Herpes Simplex Virus pathway (A), viral carcinogenesis pathway (B), viral myocarditis pathway (C), and viral protein interaction with protein and cytokine receptor (D). The pathway were modified with copyright permission.   Figure 14. KEGG pathway analysis showing relative log2 FC expression of proteins in T1, T2 and T3 with respect to control 2 (A), the identified proteins as mapped in 13 A, tabulated and the red color depicting all proteins with log2 fold change of 0.5 (~1.4 fold) and above (B). SERPINS and serine proteases identified in our data set, tabulated to show relative log2 FC expression of proteins in T1, T2 and T3 with respect to control 2 (C). Figure 1. Venn diagrams and histogram comparison plots depicting the total proteins (A and D), treatment proteins (B and E) and control proteins (C and F). The unique proteins can be seen for each group in the Venn diagrams and histogram plots.

Supplementary
Supplementary Figure 3. It shows the normalized sample abundance plot (A), correlation plots (B and C), screeplot -PCs vs variance (D), PC-correlation plot (E), and PCA-biplot (F). Abundance values of Treated (T) and Control (C) samples were taken and differential analysis was performed to see the degree of fold change among the conditions. Abundance values were Log2 transformed and then filtered protein wise for whole samples on the basis of valid values by taking at least present in all conditions. Total 3088 Proteins were filtered out from 4860 Proteins. These values were then Z-score standardized. Categorical grouping was also done for both the conditions: Control and Treatment-Control (Control1:F2, Control2:F4), Treatment (Treatment1:F1, Treatment2: F3, Treatment3: F5), as detailed above. Student t-test was used as the number of groups needed to be compared were 2 and statistical significance was considered for P value < = 0.05 and fold change > = 2. Correlation Plot: The correlation plot shows the spearman correlation coefficient value that lies between -1 (negative correlation) to +1 (positive correlation). The colour code follows the indicated values of correlation coefficient. PCA: PCA is a widely used mathematical technique designed to extract, display and rank the variance within a data set. The overall goal of PCA is to reduce the dimensionality of a data set, simultaneously retaining the information present in the data. The PCA-biplot contains a lot of information and can be helpful in interpreting relationships between experimental samples. In this biplot, we can see that runs from each sample as representing different experimental conditions are far from each other. Screeplot: The contribution of all ten principal components to the profiling classification in plot of percentage explained variance of PC. Figure S3. The tabulated proteins represent all the proteins searched with key word "ubiquitin" (A), "phosphatase" (B), "Transporter/ SLC" (C), and "chaperon" (D) in the total data set, and are presented as relative fold changes in T1, T2, T3 over Caco-2 cells grown in serum (control 2). The red color in table denotes all proteins with log2 fold change of 0.5 (~1.4 fold) and above Figure S4. The tabulated proteins represent all the proteins searched with key word "histone" in the total data set, and are presented as relative fold changes in T1, T2, T3 over Caco-2 cells grown in serum (control 2). The red color in table denotes all proteins with log2 fold change of 0.5 (~1.4 fold) and above.