Multi-Omics Reveals Mechanisms of Partial Modulation of COVID-19 Dysregulation by Glucocorticoid Treatment

Treatments for COVID-19 infections have improved dramatically since the beginning of the pandemic, and glucocorticoids have been a key tool in improving mortality rates. The UK’s National Institute for Health and Care Excellence guidance is for treatment to be targeted only at those requiring oxygen supplementation, however, and the interactions between glucocorticoids and COVID-19 are not completely understood. In this work, a multi-omic analysis of 98 inpatient-recruited participants was performed by quantitative metabolomics (using targeted liquid chromatography-mass spectrometry) and data-independent acquisition proteomics. Both ‘omics datasets were analysed for statistically significant features and pathways differentiating participants whose treatment regimens did or did not include glucocorticoids. Metabolomic differences in glucocorticoid-treated patients included the modulation of cortisol and bile acid concentrations in serum, but no alleviation of serum dyslipidemia or increased amino acid concentrations (including tyrosine and arginine) in the glucocorticoid-treated cohort relative to the untreated cohort. Proteomic pathway analysis indicated neutrophil and platelet degranulation as influenced by glucocorticoid treatment. These results are in keeping with the key role of platelet-associated pathways and neutrophils in COVID-19 pathogenesis and provide opportunity for further understanding of glucocorticoid action. The findings also, however, highlight that glucocorticoids are not fully effective across the wide range of ‘omics dysregulation caused by COVID-19 infections.


Introduction
Whilst great strides have been made in testing for and vaccinating against COVID-19, due to the rapid spread of new variants, the disease is progressing towards a globally endemic status. More positively, treatment of the illness has also improved, with mortality risk declining since the initial wave of COVID-19 infections in early 2020 [1]. A key treatment for COVID-19 infections is glucocorticoid medication, for example, dexamethasone. Synthetic glucocorticoids are a class of immunosuppressive drugs and additionally have powerful anti-inflammatory effects, an important benefit given that severe cases of COVID-19 are characterised by a hyperinflammatory state. The clinical action of glucocorticoids such as dexamethasone is well described for COVID-19 treatment, with the drug's antiinflammatory properties limiting lung injury, reducing both the likelihood of respiratory complications and also death [2]. It should be noted, however, that the UK's National bile acid dysregulation. The data also, however, illustrate that in many cases, glucocorticoid treatment is not sufficient to reverse the observed phenomena of the impact of COVID-19, for example in compensating for dyslipidemia or elevated amino acid levels in serum.

Population Metadata Overview
The study population analysed in this work totalled 98 hospital inpatients recruited between May 2020 and March 2021. These included 37 participants with a positive COVID-19 RT-PCR test and treated with glucocorticoids up to 48 h prior to sampling, and 36 participants with a positive RT-PCR test but not treated with glucocorticoids. A control group (n = 25) with a negative COVID-19 RT-PCR test but with symptoms consistent with a suspected COVID-19 infection was also recruited, in a shorter timeframe between May 2020 and July 2020. Of the 37 participants treated with synthetic glucocorticoids, 30 were treated with dexamethasone, 6 with prednisolone, 1 with methylprednisolone and none with betamethasone or hydrocortisone. A summary of the population characteristics is shown in Table 1. Age distributions for the glucocorticoid-treated COVID-19-positive cohort and the untreated cohort were similar, but the treated cohort included proportionately more males. Comorbidities are associated with severity, representing both a causative and confounding factor. Due to hospital recruitment, however, comorbidities including type 2 diabetes mellitus, hypertension, high cholesterol and ischaemic heart disease were present in both the treated and untreated groups; former smokers were somewhat more represented in the glucocorticoid-treated group. Levels of C-Reactive Protein (CRP) and eosinophils were reduced by glucocorticoid treatment (p-values of 0.03 and 0.02, respectively), but levels of lymphocytes were not. Within the treated cohort, participants were more likely to present with bilateral chest X-ray changes (p-value of 0.001), more likely to require continuous positive airways pressure (p-value of 0.008) and were also escalated to the hospital Medical Acute Dependency Unit more frequently (p-value of 0.001), indicating that those treated with glucocorticoids are more severely affected by COVID-19 than those not treated.

Feature Identification
We identified 472 serum metabolites and lipids that were reliably quantified in samples (out of a theoretical maximum of 630). We then identified those metabolites that were differentiated between those treated with glucocorticoids and those not treated at a p-value of 0.05 or less by the non-parametric Wilcoxon rank sum test. This resulted in a list of 53 metabolites and lipids significantly altered by glucocorticoid treatment.
For proteomics, SWATH-MS identified 754 proteins. This initial dataset was then filtered by the same method as for metabolites in order to identify proteins differentially expressed between glucocorticoids-treated and -untreated participants. A total of 68 proteins were found to be modulated.
The top 15 altered metabolites (Table 2) and proteins (Table 3) ranked by p-value are summarised below. Complete lists of differentiated proteins and metabolites are shown in Supplementary Materials Tables S1 and S2, respectively; the full lists of proteins and metabolites identified are shown in Supplementary Materials Tables S3 and S4.

Metabolomic Analysis: Key Metabolites Altered by Glucocorticoid Treatment
The initial data frame of metabolites measured in participants was subjected to pathway analysis. The most statistically significant pathway when controlled for false discovery was steroid hormone biosynthesis (Table 4). Several amino acid pathways also showed statistically significant differences, including lysine, phenylalanine, tyrosine and tryptophan. Univariate analyses by boxplot for cortisol, bile acids and amino acids (selected for their relevance to the statistically significant pathways in Table 4 or their prior literature identification relevant to COVID-19 infections) are also shown in Figure 1. Fewer lipids were given categorical identifiers in the pathway analysis described above due to assay and database limitations. Statistically significant triglycerides are also shown in Figure 1.

Proteomic Analysis: Investigation of Pathway Changes Due to Glucocorticoid Treatment
Next, differentially expressed proteins were used for pathway analysis. The results of the pathway analysis as performed in ClueGO are summarised in Table 5. The most

Proteomic Analysis: Investigation of Pathway Changes Due to Glucocorticoid Treatment
Next, differentially expressed proteins were used for pathway analysis. The results of the pathway analysis as performed in ClueGO are summarised in Table 5. The most statistically significant set of pathways when controlled for false discovery were neutrophil degranulation and the innate immune system, followed by exocytosis of platelets, platelet activation, signalling and degranulation. A summary of the affected pathways is also presented in network form in Figure 2.  Univariate analyses by boxplot for significantly altered proteins are also shown in Figure 3, showing the treated and untreated groups, as well as the control group (COVID-19 negative, no glucocorticoid treatment) for comparison.

Relevance Network Analysis of Glucocorticoid Treatment
Finally, a relevance network analysis base was performed on the proteomic and metabolomic datasets, using a bipartite implementation applied to identify related features. Figure 4 shows these related features as nodes (proteins and metabolites) together with the links between them (similarity scores of greater than 0.65). Eight proteins and 20 metabolites were identified in this relevance network.
Lipids incorporating the fatty acid arachidonic acid (denoted 20.4 equalling 20:4n6) were over-represented in the relevance network, comprising 8 of the 20 metabolites identified. In the overall feature set, lipids incorporating arachidonic acid comprised 13 of the 483 metabolites measured by the targeted LC-MS/MS method used in this work.

Relevance Network Analysis of Glucocorticoid Treatment
Finally, a relevance network analysis base was performed on the proteomic and metabolomic datasets, using a bipartite implementation applied to identify related features. Figure 4 shows these related features as nodes (proteins and metabolites) together with the links between them (similarity scores of greater than 0.65). Eight proteins and 20 metabolites were identified in this relevance network.
tures. Figure 4 shows these related features as nodes (proteins and metabolites) toge with the links between them (similarity scores of greater than 0.65). Eight proteins an metabolites were identified in this relevance network.
Lipids incorporating the fatty acid arachidonic acid (denoted 20.4 equalling 20: were over-represented in the relevance network, comprising 8 of the 20 metabolites i tified. In the overall feature set, lipids incorporating arachidonic acid comprised 13 o 483 metabolites measured by the targeted LC-MS/MS method used in this work.

Discussion
We have employed a quantitative multiomics approach combined with use of clinical data to investigate glucocorticoid action in patients with COVID-19 across metabolomic and proteomic features. Metabolomic pathway analysis shows that the largest difference between the glucocorticoid-treated and -untreated groups was in steroid hormone biosynthesis, as would be expected. Cortisol levels in the treated group were below those seen in either the untreated group, or the control group. The remaining pathways influenced were predominantly related to amino acids. Amino acid dysregulation was greater in glucocorticoid-treated participants than in -untreated participants. Consequently, these data show no evidence of glucocorticoids modulating COVID-19-driven amino acid dysregulation. Furthermore, glucocorticoid treatment did not alleviate increased serum concentrations of triglycerides. This result is consistent with dexamethasone previously being identified as potentially causing hypertriglyceridemia in non-COVID-19 settings [7], and suggests that in part, the effect of COVID-19 and glucocorticoid treatment on dyslipidemia may be additive. These data illustrate the complexity of drug/disease interactions.
For the proteomic dataset, the most statistically significant pathway change caused by glucocorticoid treatment was that for neutrophil degranulation. This pathway is related to platelet activation and degranulation in an inflammatory response. The importance of neutrophil-and platelet-associated pathways described here is consistent with the COVID-19 proteomics literature, especially Sinha et al. and Panda et al. [19,21], and reveals the molecular mechanisms for synthetic glucocorticoids alleviating COVID-19 symptoms. Of interest is the fact that within these protein pathway alterations, the impact of glucocorticoids was not consistent, with some proteins moderated towards the levels seen in the control group whilst others were not, or indeed showed amplified changes.
The relevance network presented in this work (Figure 4) also shows relationships that have previously been related to COVID-19 infections, albeit care is required in interpreting relevance networks as they cannot show causation. For example, P04196 has a role in immune complex and pathogen clearance, cell chemotaxis, cell adhesion, angiogenesis, coagulation and fibrinolysis. Its modulated expression and relationship to cortisol fits with a dampening of the innate and acquired immune system. P35542 or Serum Amyloid A4 Protein is critical in the acute phase response to infection and/or injury and, as such, it is not a surprise to see it is modulated by dexamethasone. Q96IY4 inhibits thrombolytic response via specific proteolytic catalysis. Again, this is in keeping with glucocorticoids moderating the innate and acquired response to infection. Furthermore neutrophil activation can be caused by elevated free fatty acids and triglycerides [30], as well as GM-CSF (a cytokine elevated in COVID-19). GM-CSF expression is abrogated by dexamethasone [31,32]. Additionally, triglycerides incorporating arachidonic acid were related to the proteins P25311, Q96IY4, P22532 and P10909. Arachidonic acid has been identified as a potential marker of severe COVID-19 infection, acting as a substrate for the lipoxygenase and cyclooxygenase pathways, which contribute to increased levels of eicosanoids [33,34]. Whilst the disproportionate presence of arachidonic acid in differentiating lipids cannot be causatively linked to leukotriene or prostaglandin expression, arachidonic acid-derived oxylipins are generally viewed as pro-inflammatory [35], and have been associated with the immune response to COVID-19 [36]. Furthermore, patients with hypertriglyceridemia also demonstrate increased platelet activation [37]. Thus, there may well be a relationship between increased triglycerides and platelet activation in COVID-19 which is not being alleviated by treatment with glucocorticoids.
It should further be noted that the World Health Organization recommends the use of dexamethasone only in severe/critical patients, and not routinely, as do the UK's NICE guidelines. More recently, clinical data have shown that treatment with dexamethasone in patients that do not require intensive respiratory support is not associated with any improvement in outcomes, and that there is evidence of potential harm in such cases [38]. We now provide 'omics-driven evidence of the potential pleiotropic actions of synthetic glucocorticoids.
This study does include clear limitations. As a retrospective observational study, results will naturally be less robust than those that would be obtained from a prospective randomized and controlled trial. Furthermore, dexamethasone became standard of care in the UK in June 2020, albeit the implementation was not uniform [39], which meant that the cohort not treated by glucocorticoids had a time difference to the treated group. As discussed previously, whilst all participants were recruited in an in-patient hospital setting and there were no differences in admittance to ICU, there was a difference in treatment with CPAP and MADU admittance, strongly suggestive of the glucocorticoid-treated cohort having more severe symptoms. Other standards of care (ventilation, anti-clotting drugs) were also changing over the recruitment period. In addition, participants were partially recruited at a time when new variants were beginning to emerge, specifically the Alpha variant, which was present in the UK from January to April 2021, and the Delta variant, which was present in the UK from April to December 2021 [40]. This may have led to differences in underlying severity or clinical presentation. Furthermore, samples in this work were not sequenced so cannot be directly mapped to their variants. It should be noted that multi-omics can include the full range of genomics, transcriptomics, proteomics and metabolomics, but in this work, proteomics and metabolomics were employed. Finally, these analyses were conducted in a pandemic setting, where the circulation of competing respiratory viruses was limited. The specificity of the findings here to COVID-19 (as opposed to other respiratory diseases) has not been tested.
In conclusion, these results provide the first multi-omics investigation into the action of synthetic glucocorticoids in COVID-19 treatment, and are concordant with previous findings that glucocorticoids modulate the neutrophil response [19]. The findings are also consistent with clinical observations of reduced hyperinflammatory reactions [41], potentially limiting immune system-related harm to the respiratory system. The data do, however, illustrate that in many cases, glucocorticoid treatment will not fully alleviate the impact of COVID-19. In some instances, it is possible that the previously described use of dexamethasone causing hypertriglyceridemia may also play a role in this observation, albeit it is not possible in this work to separate the impact of glucocorticoids from the higher severity of COVID-19 in the treated cohort. Whilst glucocorticoids have been demonstrated to be effective in the treatment of severe cases of COVID-19, as with any class of drugs, caution is needed in deeming any treatment regimen as suitable for universal use.

Participant Recruitment and Ethics
All participants for this study were recruited as part of an observational cohort study. Ethical approval (IRAS project ID 155921) was obtained via the NHS Health Research Authority (REC reference: 14/LO/1221). The participants were recruited consecutively at Frimley Park NHS Trust, UK, between May 2020 and March 2021. Participants were identified by clinical staff to ensure that they had the capacity to consent to the study and were asked to sign an Informed Consent Form based on the International Severe Acute Respiratory and emerging Infection Consortium/World Health Organisation (ISARIC/WHO) Clinical Characterisation Protocol for Severe Emerging Infections. Those patients that did not have this capacity were not sampled. Signatures were witnessed by University of Surrey researchers. At the time of recruitment, participants were categorised by the hospital as either "query COVID" (meaning there was clinical suspicion of COVID-19 infection, but a negative positive RT-PCR SARS-CoV-2 test result had been recorded during their admission) or "COVID positive" (meaning that a positive test result had been recorded). All participants were provided with a Patient Information Sheet explaining the goals of the study. All methods part of this study were performed in accordance with the relevant guidelines and regulations.

Sample Collection and Extraction
Collection of the samples was performed by researchers from the University of Surrey at Frimley Park NHS Foundation Trust hospitals; collection took place on admission or in some cases shortly afterwards. Alongside the collection of blood samples, metadata for all participants were also collected, covering the inter alia medication regime (specifically including dexamethasone treatment or other glucocorticoids), sex, age, comorbidities (based on whether the participant was receiving treatment), the results and dates of COVID-19 PCR (polymerase chain reaction) tests, bilateral chest X-Ray changes, smoking status and whether the participant presented with clinical symptoms of COVID-19. Values for lymphocytes, CRP and eosinophils were also taken-here, values within five days of biofluid sampling were recorded. Serum collection and extraction followed the protocols set out by the COVID-19 Coalition for metabolomics [42] and proteomics [43]. In brief, venous blood was collected in 3 mL serum tubes, transported to the University of Surrey by courier whilst stored on ice, and centrifuged on arrival at 1600× g for 10 min at 4 • C. All samples with a sampling time interval greater than four hours were rejected. Serum was then decanted into 100 µL aliquots and stored at −80 • C until processing. Prior to analysis, the serum was sterilised using 200 µL of ethanol into 100 µL of serum (2:1 v/v solvent/sample ratio).

Participant Selection
From the initial population of 115 recruited participants, a number of exclusions were made to remove participants where incomplete metadata were provided, where the gap between the initial positive COVID-19 test and sampling exceeded 14 days or where the participants were already being medicated with glucocorticoids prior to COVID-19 treatment. The remaining population of 98 was then segmented into a COVID-19-negative control group for comparison purposes (n = 25), a group of COVID-19-positive patients treated with a glucocorticoid in the 48 h prior to sampling (n = 37), and a group of COVID-19-positive patients not treated with a glucocorticoid in in the 48 h prior to sampling (n = 36).

Serum Instrumentation and Analysis: Metabolomics
Serum samples were analysed using the Biocrates MxP Quant 500 system using a Xevo TQ-S Triple Quadrupole Mass Spectrometer coupled to an Acquity UPLC system (Waters Corporation, Milford, MA, USA). The MxP Quant 500 system provides targeted quantification of metabolites including amino acids and derivatives, bile acids, biogenic amines, acylcarnitines, carbohydrates and other small molecule metabolites, plus a wide array of lipids. Analysis took place via a single assay, and two analytical procedures. The first of these procedures operated by liquid chromatography (operated in both positive and negative ion mode) and the second by flow injection analysis (positive ion mode), both coupled to tandem mass spectrometry with isotopically labelled internal standards for quantification. The sample order was randomised across 96-well plates, and 3 levels of quality controls (QC) were run on each plate. Blank PBS (phosphate-buffered saline) samples (three technical replicates) were used for the calculation of the limits of detection (LOD). Biogenic amines and amino acids were quantified for each plate using a seven-point calibration curve, with other analytes semi-quantitated with a single point standard (i.e., assuming concentration linearity in the range measured). The levels of metabolites present in each QC were compared to the expected values and the CV% calculated. Data were normalised between the three batches using the results of quality control level 2 (QC2) repeats across the plate (n = 5) and between plates (n = 3) using Biocrates METIDQ software (QC2 correction). Metabolites where >25% concentrations were at or below the limit of detection ( LOD), above the limit of quantification (>LOQ), or where the blank was out of range were excluded (total n excluded in serum = 150). The remaining 474 quantified metabolites comprised of 8 acylcarnitines, 20 amino acids, 26 biogenic amines, 11 bile acids, 53 ceramides, 15 cholesteryl esters, 1 cresol, 9 diglycerides, 4 carboxylic and fatty acids, 85 phosphatidyl cholines, 14 sphingolipids, 222 triglycerides, 2 hormones, 2 indoles, 1 nucleobase and 1 vitamin.

Plasma Instrumentation and Analysis: Proteomics
Mass spectrometry of clinical specimens using the SWATH-MS technique implements a Data-Independent Acquisition (DIA) approach for precision identification and accurate quantification of proteins. Samples were quality-checked, assigned a unique in-ternal study ID and groups were randomized. Immunodepletion was performed by using Top14 Abundant Protein Depletion Resin (ThermoFisher Scientific, Macclesfield, UK) in a 96-well format. Samples were reduced, alkylated and digested with trypsin (Promega, Macclesfield, UK) using S-trap columns (Protifi, New York, NY, USA) prior to lyophylisation. Digitized proteomic maps were generated by using 100 variable window SWATH-MS (68 min) with a micro-flow LC-MS system. SWATH-MS analysis was performed on a 6600 TripleTOF mass spectrometer with DuoSpray Ion Source (AB Sciex Limited, Alderley Park, Macclesfield, UK) coupled to an Eksigent 425 LC (AB Sciex Limited) with specific mass spectrometric conditions (including isolation window size and overlap and total cycle time), as previously described. All study samples were run in duplicate and CV percentages were calculated across the two injections. Only samples with median CV's across all proteins quantified with less than 20% were taken forward for analysis, with the best injection for each sample included in the final protein quantification. To ensure instrument performance remained consistent throughout the study and to control for batch-related effects, both commercially available plasma (Human K2EDTA gender pooled plasma, Sera Laboratories International Ltd. trading as BioIVT, Burgess Hill, UK) and a total pool of all study samples (TOTs) were processed, digested and run alongside each batch of samples.

Feature Identification
Serum metabolites were identified and quantified using isotopically labelled internal standards, retention times and multiple reaction monitoring. In this study, identifications were made in accordance with the Metabolomics Standards Initiative for metabolite identification [44], using both accurate m/z values referenced to library values as well as orthogonal information in the form of MS/MS fragmentation spectra and retention time matching against isotopically labelled internal standards. The mass spectrometry conditions were used as optimised and provided by Biocrates.
For the proteomic workflow, extraction and processing followed the method described in Salie et al. [45]. Briefly, SWATH-MS data files were searched using openSWATH (version 2.0.0) against a published plasma spectral library [46]. Spectral library matches were scored and filtered for quality using pyProphet (version 0.18.1) then aligned across runs using the TRIC alignment algorithm from MSproteomicstools. The transition-level quantification contained within the aligned data was then normalised and summarised to protein-level quantification in R (version 3.4.1) using the Bioconductor (version 3.5) packages SWATH2stats and MSstats, with the equalized medians and top 3 features per peptide parameters. Proteins quantified in at least 30% of samples were retained in the following biomarker analysis.
The two approaches described generated two data frames, each comprising a matrix of n participants by p features, with matched n of 98 participants. These data frames are provided in Tables S3 (metabolomics) and S4 (proteomics) within Supplementary Materials.

Statistical Analysis
Both the metabolomic and proteomic data frames were log2 transformed to stabilize variance and reduce heteroscedasticity. In order to identify proteins and metabolites differentially expressed between the two conditions (COVID-19-positive participants treated with glucocorticoids, and not treated), the difference between the two populations for each feature was measured by p-value. Specifically, a p-value threshold of 0.05 was used for significance by Wilcoxon rank-sum test.
Pathway analysis for the proteomic differences between the two classes of participants was conducted using the ClueGo application (version 2.5.9) within Cytoscape (version 3.9.1) [47,48], matching Uniprot identifiers for differentiating proteins to Reactome pathway databases. Pathway analysis for the metabolomic differences between the two classes of participants was conducted using MetaboAnalyst (version 5.0) [49], using the Pathway Analysis module, matching HMDB identifiers to the MetaboAnalyst database. Participant metadata characteristics for the two main cohorts (COVID-19 positive split between glucocorticoid-treated and -untreated patients) were assessed by two-tailed Student t-tests for continuous variables, and by two-tailed z-score tests for population proportions.
The relevance network was constructed using the R package MixOmics [50]. This approach uses the two data frames (metabolomic and proteomic) to construct a multiblock partial least squares discriminant analysis (PLS-DA) model. This model was then used to generate a relevance network for the variables selected by the multiblock PLS-DA, to highlight differential relationships between the treated cohort and the untreated (by glucocorticoid) cohort. Rather than a matched-pairs correlation network, a bipartite model was constructed, showing protein-to-metabolite relationships and excluding protein-toprotein/metabolite-to-metabolite relationships, using a similarity matrix to capture the relationships between the features and components of the predictive multiblock PLS-DA model [51]. The relevance network output file was then processed in Cytoscape to produce the visual summary.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/ijms232012079/s1, Table S1: Full list of proteins showing statistical significance in differentiating the glucocorticoid-treated and -untreated cohort; Table S2: Full list of metabolites showing statistical significance in differentiating the glucocorticoid-treated anduntreated cohort; Table S3: Feature by participant data frame for proteomics; Table S4: Feature by participant data frame for metabolomics. Funding: The authors would like to acknowledge funding from the EPSRC Impact Acceleration Account for sample collection, as well as EPSRC Fellowship Funding EP/R031118/1. Mass spectrometry at the University of Surrey was funded under EP/P001440/1; Equipment used in the Stoller Biomarker Discovery Centre was funded by a donation received from the Stoller Charitable Trust and a research grant was awarded by the Medical Research Council (MR/M008959/1). Sample collection and processing was funded by the University of Surrey and the BBSRC BB/T002212/1.

Institutional Review Board Statement:
The study was reviewed and approved by UK London REC 14/LO/1221 for COVID-19 patient samples.
Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.
Data Availability Statement: Data frames for proteomics and metabolomics relating to this work are included within Supplementary Materials.