A blood-based comprehensive and systems-level analysis of disease stages, immune 1 regulation and symptoms in COVID-19 patients

50 COVID-19 patients show significant clinical heterogeneity in presentation and outcomes that 51 makes pandemic control and strategy difficult; optimising management requires a systems 52 biology approach of understanding the disease. Here we sought to understand and infer 53 complex system-wide changes in patients infected with coronaviruses (SARS-CoV and SARS- 54 CoV-2; n=38 and 57 samples) at two different disease stages compared with healthy 55 individuals (n=16) and patients with other infections (n=144). We applied inferential 56 statistics/machine-learning approaches (the COVID-engine platform) to RNA profiles derived 57 from peripheral blood mononuclear cells (PBMCs). Compared to healthy individuals, an 58 integrated blood-based gene signatures distinguished acute-like (mimicking coronavirus- 59 infected patients with prolonged hospitalisation) from recovering-like patients. These 60 signatures also hierarchically represented systems-level parameters associated with PBMC 61 including dysregulated cytokines, genes, pathways, networks of pathways/concepts, immune 62 status, and cell types. Proof-of-principle confirmatory observations included PBMC-associated 63 increases in ACE2 , cytokine storm-associated IL6 , enhanced innate immunity (macrophages 64 and neutrophils), and lower adaptive T and B cell immunity in patients with acute-like disease 65 compared to those with recovery-like disease. Patients in the recovery-like stage had 66 significantly enhanced TNF , IFN- g , anti-viral, HLA-DQA1 , and HLA-F gene expression and 67 cytolytic activity, and reduced pro-viral gene expression compared to those in the acute-like 68 stage in PBMC. Besides, PBMC-derived surrogate-based approach revealed overlapping 69 genes associated with comorbidities (associated diabetes), and disease-like symptoms 70 (associated with thromboembolism, pneumonia, lung disease and septicaemia). Overall, our 71 study involving PBMC-based RNA profiling may further help understand complex and variable 72 systems-wide responses displayed by coronavirus-infected patients.


79
The spread of COVID-19, a disease caused by Severe Acute Respiratory Syndrome 80 CoronaVirus 2 (SARS-CoV-2), has led to the current global pandemic with already more than 81 4 million people with confirmed infection and nearly 300,000 deaths within a few months(1).

82
According to the World Health Organization (WHO), the mode of infection for COVID-19 is 83 predominantly through respiratory droplets, aerosol transmission due to pathogen-laden viral 84 particles in the air, or close contact with infected people with increased viral loads, especially 85 in the early stages of disease(2). The mechanism of human pathogenesis, to a great extent, there are now reports suggesting heterogeneous manifestation of the disease affecting 91 multiple organs, including kidney, liver, and brain (7). Although age and compromised health 92 history are considered critical prognostic signatures, certain patients of younger age and good 93 health have shown severe progression of the disease(8).

95
Patients with COVID-19 may be asymptomatic(9, 10), but can still transmit infection(11). Viral 96 shedding from an infected person may occur, although resolution of symptoms (12), and 97 relapse has been reported despite consecutive negative testing(13). Currently, there is no 98 standard of care to treat COVID-19 respiratory symptoms (14,15), screen for potential organ 99 failures, disease aggression, and systemic changes in patients.

101
In this study, we sought to understand and infer changes in coronavirus-infected patients at

128
Nearest Template Prediction(22) was used to derive distance between two signatures -acute-129 like vs. recovering-like patients. Different gene set databases were from EnrichR(23) and 130 MSigDB(20). Immune gene sets were from Rooney et al(24). The references to the used gene

204
there was no significant differences between acute-like and recovering-like patients ( Figure   205 2A-B). Interestingly, TNF was highly expressed in acute-like patients compared to recovering-206 like patients and healthy individuals ( Figure 2C). Besides, analysis of subunits of lactate 207 dehydrogenase (LDHA and LDHB; associated with hypoxia) genes showed LDHB was highly 208 expressed in healthy individuals compared to coronavirus-infected patients, and inverse 209 trends were observed for LDHA ( Figure 2D-E). Among these five genes, TNF was the only 210 gene that showed differential expression between acute-like and recovering-like patients.

211
These results show increased expression of these key COVID-19/SARS genes in patient 212 PBMC samples.

214
We further examined multiple other candidate genes that act as chemo-attractants to 215 monocytes and macrophages, specifically those that interfere with innate and adaptive 216 immunity and viral replication (29). Among those genes, we observed CXCL8 (IL8) and 217 CCL13, which are associated with chemoattraction of neutrophils/macrophages (innate 218 immunity), to be highly expressed in acute-like patients, compared to recovering-like patients 219 and healthy individuals ( Figure 3A). On the other hand, OAS2 and IL16 associated with T cells 220 (adaptive immunity) and inhibition of viral replication were highly expressed in recovering-like 221 patients and healthy individuals ( Figure 3B). These results suggest that PBMC from acute-like patients may be associated with the activity of innate immunity, whereas PBMC from Given that different pathways were enriched in infected patients, next, we interrogated how 249 these pathways are linked together to convey a network of processes or changes at the 250 cellular level. Hence, we used the REACTOME pathway database(34) to connect different but 251 related pathways that were enriched in infected patients using network analysis. Two evident 252 and distinct networks were: a) interleukins and cytokine signaling (potentially representing 253 cytokine storm), and b) neutrophils and innate immunity showed significant enrichment using

254
CoV-Up-gene signatures ( Figure 4A). Nevertheless, we observed an increased enrichment of 255 a unique network linking granulopoiesis, megakaryocyte differentiation, and platelet activation 256 ( Figure 4A). This may be linked to coagulation system that can activate the innate immune

290
Nonetheless, the host-specific subcellular changes in PBMC are also evident from this 291 analysis. An increased replication and proliferation of potential host cells, mainly involving the 292 innate immune system, may be evident based on the enrichment of genes associated with 293 DNA polymerase processivity factor and proliferating cell nuclear antigen (PCNA) complex.

294
Also, the production of immunoglobin (IgG) complexes along with NFkB complex is higher in 295 these patient gene signatures, again, representing increased immune responses. At the same time, the host's potential responses to death signals associated with BCL2 complex are also 297 enriched in this analysis. Again, this potentially represents lymphocyte apoptosis in connection with an enriched apoptotic pathway in Figure 3C. Neutrophil specific S100A8/A9 complexes 299 are also enriched ( Figure 4C). Overall, these results suggest PBMC-based subcellular level 300 changes associated with the viral integration in immune cells and associated pathophysiology.

302
Recovery from coronavirus infection is associated with increased cytolytic activity and 303 IFN-g but not increased B-cell levels

305
In order to gain insight into the cellular dynamics of the SARS-CoV-2 immune response, we 306 calculated immune signature scores from Rooney et al (24)  and compared these with healthy control samples. As expected, we observed that the innate 309 immune system involving macrophages and neutrophils was highly active in the acute-like 310 patients, suggesting that they were the first to encounter the coronavirus. with these 311 decreasing in recovering patients ( Figure 5A).

313
Perhaps most interestingly, there was a significant increase in NK cells, cytolytic activity, and 314 plasmacytoid dendritic cells (pDCs) in recovering-like patients compared to acute-like patients 315 ( Figure 5A). It is noteworthy that the absolute levels of CD8 + T cells and co-stimulating helper

316
T cells are not different between recovering-like and acute-like patients ( Figure 5A). This result 317 suggests that the CD8 + T cells are potentially activated (cytolytic) in the recovering patients.

318
These results from SARS patients were confirmed using PBMC from COVID-19 samples 319 ( Figure 5B). There were no differences in B cells between both types of patients and healthy

326
Next, we examined the differential expression of major histocompatibility complex (MHC) 327 class-I and class-II HLA that may reflect antigen presentation to and/or activation of 328 CD4 + /CD8 + T cells, and whose levels are increased by IFN-g ( Figure 5C). Among the MHC

357
We found that CoV-Up-gene signatures was enriched for various diseases, including 358 septicaemia, pneumonia, lung disease, arthritis, cystic fibrosis, thalassemia, pre-eclampsia, 359 bacterial infections, asthma, acute coronary syndrome and others ( Figure 7A). The overlap of

360
CoV-Up-gene signatures and those genes from selected diseases -septicaemia, pneumonia, 361 lung diseases, arthritis, and cystic fibrosis are shown in Figure 7B. These results suggest that 362 the disease symptoms due to coronavirus may be complex and highly variable and may affect 363 patients with pre-existing disease conditions as recently reported (7). Further refinement of

387
In this study, we performed a comprehensive analysis using publicly available blood cell RNA

405
While there is no change in overall CD8 + T cell population between patients in their acute-like as represented in multiple reports(45). Congruently, this is associated with lower innate 409 immunity in recovering-like patients than acute-like patients and associated with increased 410 expression of anti-viral genes OAS2 and IL16.

412
Similarly, the high HLA-F gene, which is known to be associated with the interaction between

428
There are a number of questions that arose from this study that could be of relevance in

449
In conclusion, PBMC has information related to infection status, immune states, disease 450 aggression, severity, and disease symptoms that are likely going to be manifested due to

479
There is no funding to declare.