Evaluation of altered miRNA expression pattern to predict COVID-19 severity.

Swati srivastava, Ph.D.; Iti Garg: Conceived and designed the experiments; Performed the experiments; Analyzed and interpreted the data; Wrote the paper. Nilanjana Ghosh; Babita Kumari; Vinay Kumar: Contributed reagents, materials, analysis tools or data; Performed the experiments. Yamini Singh; Ramesh Meena; Malleswara Rao Eslavat; Sayar Singh; Vikas Dogra; Mona Bargotya; Sonali Bhattar; Utkarsh Gupta; Shruti Jain; Javid Hussain:Performed the experiments. Rajeev Varshney; Lilly Ganju: Analyzed and interpreted the data.

10.13039/501100001849 Defence Research and Development Organisation , DIPAS-DRDO, 2021.

An average of 19,182,338.5 high quality and non-redundant reads were retained for analysis. After trimming of the raw reads an average of 13,113,032.2 reads. Homo sapiens (GRCh38) genome downloaded from the Ensembl database and mo sapiens cDNA sequences from Ensembl was used as reference genome. An average of 1048 known miRNAs and 244 novel miRNAs were identified in all samples. Average total number of miRNA (known and novel) identified across each group along with average total number of miRNA detected with read count≥50 and read count≥10 is depicted in Table 2 . Table 2 Sequencing data report. Table 2 Subjects Dead Severe Moderate Control Avg. Total Reads 19797381 19453872.5 19054232 18423869.7 Avg. Trimmed Reads 16928692 17361656.8 16618799 1542981.6 % of reads align to Genome 87.43 87.92 89.14 88.68 % of reads align to ncRNA 7.33 6.58 7.81 8.81 Known miRNA Avg. total miRNA detected 988 1060 1054 1022 Avg. Total miRNA detected with read count ≥500 307 312 308 277 Avg. Total miRNA detected with read count >10 530 536 525 486 Novel miRNA Avg. total miRNA detected 238 251 249 159 Avg. Total miRNA detected with read count ≥500 17 17 18 14 Avg. Total miRNA detected with read count >10 99 109 107 68 Sequencing data report. Variation in the reads were normalized by library normalization method opted from DESeq library. DESeq calculated size factor, each read count is normalized by dividing with size factor. Mean normalized read counts of the samples in a given condition were used for DGE calculation. To understand the regulation of expression between the samples, log2fold of 2 fold change (FC) and p value ≤ 0.05 was used as cutoff. MiRNAs ≥2 FC were considered as “UP” regulated, miRNAs ≤ -2 FC were considered as “DOWN” and those between 2 and 2 were flagged as “NEUTRAL”. Fold changes were calculated based of “Expression of treated sample/Expression of control sample. A total of 103 known miRNAs were up-regulated, 44 in dead, 46 in severe and 13 in moderate patients group. Similarly, 118 known miRNAs were down-regulated, 39 in dead patients, 34 in severe patients and 45 in moderate patients group. Venn diagram (plotted using webtool ) identified the common and unique DE miRNAs which were further categorized into those above >2, >3 and > 4 FC ( Fig. 2 C). The details of both common and unique miRNAs in the study groups as obtained from venn diagram has been detailed in Table 3 . Table 3 List of differentially expressed (DE) miRNAs in all study groups (Severe, Moderate and Mild COVID-19 Patients). Table 3 Study Groups Expression Number MiRNAs identified Common miRNAs amongst Study Groups Dead, Moderate & Severe Up-Regulated 4 hsa-miR-320a-5p, hsa-miR-377-5p,hsa-miR-4800-3p, hsa-miR-4484 Down-Regulated 15 hsa-miR-20b-5p, hsa-miR-559, hsa-miR-4999-5p, hsa-miR-624-5p, hsa-miR-496, hsa-miR-1255b-2-3p, hsa-miR-15a-5p, hsa-miR-624-3p, hsa-miR-3613-5p, hsa-miR-20a-5p, hsa-miR-106a-5p, hsa-miR-126-3p, hsa-miR-590-3p, hsa-miR-31-5p, hsa-miR-32-5p. Dead & Severe Up-Regulated 18 hsa-miR-6502-5p,hsa-miR-34c-5p, hsa-miR-6868-3p, hsa-miR-5585-3p, hsa-miR-212-5p, hsa-miR-3614-3p, hsa-miR-1262, hsa-miR-2115-5p, hsa-miR-3162-3p, hsa-miR-4772-5p, hsa-miR-96-5p, hsa-miR-4668-3p, hsa-miR-618, hsa-miR-2355-3p, hsa-miR-143-3p, hsa-miR-203a-3p, hsa-miR-1179, hsa-miR-671-5p, Down-Regulated 8 hsa-miR-6859-3p, hsa-miR-1248, hsa-miR-374b-3p, hsa-miR-449a, hsa-miR-3184-5p, hsa-miR-1911-5p, hsa-miR-411-5p, hsa-miR-153-3p. Dead & Moderate Up-Regulated 1 hsa-miR-203b-5p Down-Regulated 4 hsa-miR-548am-5p, hsa-miR-1277-5p, hsa-miR-548l,hsa-miR-95-3p. Moderate & Severe Up-Regulated 2 hsa-miR-6793-5p, hsa-miR-885-5p. Down-Regulated 5 hsa-miR-548b-5p, hsa-miR-5698, hsa-miR-486-3p, hsa-miR-16-5p, hsa-miR-374a-5p. Unique to study groups Dead Up-Regulated 21 hsa-miR-582-3p, hsa-miR-1908-5p, hsa-miR-125b-1-3p, hsa-miR-490-3p, hsa-miR-490-5p, hsa-miR-518c-3p, hsa-miR-6876-3p, hsa-miR-193b-5p, hsa-miR-1299, hsa-miR-3074-5p, hsa-miR-3158-3p, hsa-miR-320b, hsa-miR-3614-5p, hsa-miR-27a-5p, hsa-miR-125a-3p, hsa-miR-5189-5p, hsa-miR-760, hsa-miR-504-5p, hsa-miR-6787-3p, hsa-miR-3177-3p, hsa-miR-1256. Down-Regulated 12 hsa-miR-551b-3p, hsa-miR-6751-5p, hsa-miR-147b-5p, hsa-miR-3683, hsa-miR-449b-5p, hsa-miR-548p, hsa-miR-218-5p, hsa-miR-369-3p, hsa-miR-3136-5p, hsa-miR-20a-3p, hsa-miR-107, hsa-miR-548aq-5p. Severe Up-Regulated 22 hsa-miR-193a-3p, hsa-miR-4755-3p, hsa-miR-4707-3p, hsa-miR-4743-3p, hsa-miR-145-3p, hsa-miR-495-3p, hsa-miR-202-5p, hsa-miR-221-3p, hsa-miR-4642, hsa-miR-181a-5p, hsa-miR-3679-3p, hsa-miR-582-5p, hsa-miR-335-3p, hsa-miR-4659b-3p, hsa-miR-605-5p, hsa-miR-337-3p, hsa-miR-7107-3p, hsa-miR-377-3p, hsa-miR-1178-3p, hsa-miR-130a-5p, hsa-miR-433-3p, hsa-miR-6834-3p. Down-Regulated 6 hsa-miR-224-5p, hsa-miR-101-2-5p, hsa-miR-6825-5p, hsa-miR-6736-5p, hsa-miR-1278, hsa-miR-3912-3p. Moderate Up-Regulated 6 hsa-miR-6507-5p, hsa-miR-3675-3p, hsa-miR-4700-5p, hsa-miR-205-5p, hsa-miR-3944-3p, hsa-miR-3611. Down-Regulated 21 hsa-miR-196b-5p, hsa-miR-199a-3p,hsa-miR-32-3p, hsa-miR-548am-3p, hsa-let-7f-5p, hsa-miR-651-5p, hsa-miR-144-3p, hsa-miR-17-5p, hsa-miR-141-3p, hsa-miR-98-5p, hsa-miR-374c-3p, hsa-miR-493-5p, hsa-miR-26b-5p, hsa-let-7g-5p, hsa-miR-5583-3p, hsa-let-7a-5p, hsa-miR-548k, hsa-miR-21-5p, hsa-miR-545-3p, hsa-miR-454-3p, hsa-miR-1289. List of differentially expressed (DE) miRNAs in all study groups (Severe, Moderate and Mild COVID-19 Patients). Volcano plots were also generated to identify DE miRNAs in the three study groups (as depicted in Fig. 2 D), using webtool VolcaNoseR (doi: 10.1101/2020.05.07.082263 ). Also, 25 miRNAs were commonly up-regulated and 32 were commonly down-regulated in all three study groups in comparison to compared to healthy controls. Heat map in Fig. 3 depicts these miRNAs. The differential colour pattern shows the change in expression amongst the patient groups. Group wise principal component analysis (PCA) of common up-regulated and down-regulated miRNAs shows that the three study groups lying in different quadrants showing distinct pattern of expression. Fig. 3 Heat map and PCA analysis: ( A) Heat map of differentially expressed (DE) miRNAs common in the three study groups. Each column represents a group of COVID-19 patient (Dead, Severe, Moderate), and each row represents a single miRNA. The fold change of miRNAs is represented in blue (negative), red (positive) and intermediate colours (white colour represents no significant change or absence of data). (B) Group wise prinicipal Component Analysis (PCA) of Up-regulated and Down-regulated miRNAs of Dead, Severe and Moderate COVID-19 patients. Fig. 3 Heat map and PCA analysis: ( A) Heat map of differentially expressed (DE) miRNAs common in the three study groups. Each column represents a group of COVID-19 patient (Dead, Severe, Moderate), and each row represents a single miRNA. The fold change of miRNAs is represented in blue (negative), red (positive) and intermediate colours (white colour represents no significant change or absence of data). (B) Group wise prinicipal Component Analysis (PCA) of Up-regulated and Down-regulated miRNAs of Dead, Severe and Moderate COVID-19 patients. Besides known miRNAs, DE was also observed in novel miRNA sequences. A total of 19 novel miRNAs were observed (log2 FC±2, p value ≤ 0.05), out of which 8 were up-regulated miRNAs (2 in Dead, 5 in severe and 1 in moderate group) and 11 were down-regulated miRNAs (2 in Dead, 2 in Severe and 1 in Moderate group and 6 in All patients group) ( Table 4 ). A novel miRNA sequence was commonly up-regulated in Dead and Severe group and another was commonly down-regulated in Moderate and All patients group (highlighted in Table 3 ). MiRNA target prediction and downstream bioinformatic analysis for these novel miRNAs was not carried out. Table 4 Novel miRNA sequences (log2 FC±2, p value ≤ 0.05). Sequences shaded in grey colour are common amongst the two or more study groups. Table 4 miRNA log2FoldChange p-Value Dead vs Control ATTCTACAGTCCTACGAGCA −4.807 0.037 AGACTGACCCAGAGTCTCAAGC 6.016 0.044 TCAGCGTTGGTTCCCGTCTTGGCC −2.948 0.044 CCTCTCCGCCACCTCCACCGCGGC 4.940 0.049 Severe vs Control AGACTGACCCAGAGTCTCAAGC 5.619 0.009 TATCCCACCACTGCCACCATTATT −3.280 0.014 ACTGCCTTTTGATGACCGGGACGA 4.259 0.014 CACTGACCCTCTTCTCTGCACAGC 4.134 0.028 TACCTTCACACCTCTGACTCTGA 3.745 0.047 CTTGAGACTCTGGGTCAGTCTATT 3.104 0.049 TGCCCTGAGACTTTTGCTCTAAT −4.188 0.050 Moderate vs Control GCGGCCGGGGAGAACTGCGCCTGC 3.623 0.020 TGAGGTAGTAGGTTGTATAGTTT −4.205 0.035 All patients vs control CGTCCAGTCCTGGCCCCCTAGGAG −4.625 0.007 TGAGGTAGTAGGTTGTATAGTTT −3.381 0.014 ACTGACCCTCTTCTCTGCACAGCT −2.972 0.030 CTCGTACCGTGAGTAATAATGCG −3.053 0.030 CAACGGAATCCCAAAAGCAGCTG −4.528 0.035 TTTAGTGGCTCCCTCTGCCTGC −5.099 0.042 Novel miRNA sequences (log2 FC±2, p value ≤ 0.05). Sequences shaded in grey colour are common amongst the two or more study groups. Sample wise expression data for each DE miRNA was subjected to receiver-operating characteristic (ROC) analysis using online webtool EPITOOLS ( https://epitools.ausvet.com.au/roccurves ) ( Fig. 4 ). This probability curve that plots true positive rate against false positive rate essentially separated the low quality data and further refined the list of DE miRNAs. All DE miRNAs were subjected to ROC analysis. Cut-off value of area under the ROC curve (AUC) was taken as 0.8 to select miRNAs for further analysis. MiRNAs having AUC values lower than 0.8 were not included for further analysis. 46 down-regulated and 37 up-regulated miRNAs showed AUC values > 0.8. These miRNAs were then further subjected to target prediction followed by pathway and network analysis. Fig. 4 [(A), (B), (C), & (D)] and Fig. 5 [(A), (B), (C), & (D)] respectively, shows ROC curves and sample wise box plots of DE common miRNAs and DE miRNAs in individual study groups. These 31 miRNAs (both up-regulated and Down-regulated) were finally selected as candidate miRNAs for COVID-19 infection severity and disease prognosis. Fig. 4 Receiver operating characteristic (ROC) curve of 31 candidate microRNAs. Area under the ROC curve (AUC); SEN, sensitivity; SPE, specificity. (A) ROC curves of down-regulated miRNAs in each study group; (B) ROC curves of down-regulated miRNAs common to all study groups; (C) ROC curves of up-regulated miRNAs in each study group; (D) ROC curves of up-regulated miRNAs common to all study groups. Fig. 4 Fig. 5 Box plot representation of 31 candidate microRNAs. Each plot represents Patient miRNA expression profile (Dead, Severe or Moderate) in comparison to healthy controls. (A) Box plot of down-regulated miRNAs in each study group; (B) Box plot of down-regulated miRNAs common to all study groups; (C) Box plot of up-regulated miRNAs in each study group; (D) Box plot of up-regulated miRNAs common to all study groups. Fig. 5 Receiver operating characteristic (ROC) curve of 31 candidate microRNAs. Area under the ROC curve (AUC); SEN, sensitivity; SPE, specificity. (A) ROC curves of down-regulated miRNAs in each study group; (B) ROC curves of down-regulated miRNAs common to all study groups; (C) ROC curves of up-regulated miRNAs in each study group; (D) ROC curves of up-regulated miRNAs common to all study groups. Box plot representation of 31 candidate microRNAs. Each plot represents Patient miRNA expression profile (Dead, Severe or Moderate) in comparison to healthy controls. (A) Box plot of down-regulated miRNAs in each study group; (B) Box plot of down-regulated miRNAs common to all study groups; (C) Box plot of up-regulated miRNAs in each study group; (D) Box plot of up-regulated miRNAs common to all study groups. Target gene prediction was done online using miRDB database [ 29 , 30 ]. Only those target genes were selected for further analysis that showed target score match of 80%. 3 miRNAs did not show any target genes above 80% match and hence were not included in downstream analysis. Pathway analysis was done using PANTHER database [ 31 ]. Each miRNA target gene list was sought for genes involved in eighteen biological pathways putatively associated in COVID-19 pathogenesis ( Fig. 6 A). These included inflammation, B-cell and T-cell activation, interleukin signaling, blood coagulation, oxidative stress response etc. Those miRNAs showing over 45 target genes involved in these pathways were selected as candidate miRNAs significantly linked to COVID-19 prognosis ( Fig. 6 A). Network analysis for the candidate miRNAs (up-regulated and down-regulated) was performed using online database, MiRnet to create MiRNA-mRNA target networks ( Fig. 6 B). Fig. 6 ( A) Pathway analysis of differentially expressed Down-regulated (n = 24) and Up-regulated (n = 7) miRNAs predicted by PANTHER online tool. (B). Network analysis of differentially expressed miRNAs; Down-regulated (n = 24) and Up-regulated (n = 7) miRNAs of Dead, Severe and Moderate COVID-19 patients. Fig. 6 ( A) Pathway analysis of differentially expressed Down-regulated (n = 24) and Up-regulated (n = 7) miRNAs predicted by PANTHER online tool. (B). Network analysis of differentially expressed miRNAs; Down-regulated (n = 24) and Up-regulated (n = 7) miRNAs of Dead, Severe and Moderate COVID-19 patients. Present analysis resulted in a panel of 31 miRNAs ( Fig. 7 ). These were 24 down-regulated and 7 up-regulated miRNAs, including both unique and common elements in the study groups. Fig. 7 Statistical derivation of miRNA panel for COVID-19 severity. Fig. 7 Statistical derivation of miRNA panel for COVID-19 severity. Hsa-miR-147b-5p and hsa-miR-107 was uniquely down-regulated whereas hsa-miR-1299 was uniquely up-regulated in Dead patients. In the Severe patients group, hsa-miR-224-5p was uniquely down-regulated and three miRNAs, hsa-miR-4659b-3p, hsa-miR-495-3p and hsa-miR-335-3p were uniquely up-regulated. Eleven miRNAs were uniquely down-regulated in Moderate patients group, including, hsa-miR-32-3p, hsa-miR-144-3p, hsa-miR-5583-3p, hsa-miR-26b-5p, hsa-miR-199a-3p, hsa-miR-98-5p, hsa-miR-548k, hsa-miR-493-5p, hsa-miR-454-3p, hsa-miR-141-3p and hsa-miR17-5p. Only one miRNA, hsa-miR-4700-5p was uniquely up-regulated in moderate patients. Ten miRNAs common to all three study groups were significantly down-regulated. These were hsa-miR, 31-5p, hsa-miR-20a-5p, hsa-miR-20b-5p, hsa-miR-590-3p, hsa-miR-15a-5p, hsa-miR-374b-3p, hsa-miR-16-5p, hsa-miR-1277-5p, hsa-miR-32-5p, hsa-miR-548I. Two miRNAs were commonly up-regulated, hsa-miR-4668-3p and hsa-miR-320a-5p.

No additional information is available for this paper.

The miRNA signatures identified in the present study could be used as novel tool for early prediction of vital status of COVID-19 patients during disease prognosis. Specific miRNA expression data may be useful in prediction of severity of COVID-19 infection in an individual. Validation of these miRNA targets with larger cohort could also be helpful in using them as therapeutic targets.

Present study generated a panel of 31 candidate miRNAs (24 Down-Regulated and 7 up-Regulated) which could be potentially involved in COVID-19 disease severity. These circulating candidate miRNAs were identified in Dead, Severe and Moderate COVID-19 patients with a distinct miRNA signature pattern in each study group. The criteria for selecting candidate DE miRNA after sequencing were extremely stringent in the current study. The selection was based on miRNA expression fold change ±≥2, p ≤ 0.05, AUC≥0.8 and number of target genes associated with selected pathways ≥45. Panel of biomarkers generated in the present study could be more Robust than depending upon a single biomarker. Present study is an unbiased high throughput screening of miRNA biomarkers for COVID-19 infection. It can be considered as discovery phase study, which includes a smaller sample size. However, it cannot be considered as an underpowered study, as statistical analysis has been conducted with stringent criteria to exclude non significant leads at each step [ 32 ]. Each candidate miRNA identified in the present discovery cohort shall be validated by RT-qPCR in a larger validation cohort. Validation of potentiality of 31 miRNAs as putative markers for COVID-19 shall circumvent the chances of failure of potential marker, which could be further taken up for clinical trials. Pubmed search revealed that all the DE miRNAs identified in the present study, have established role in cancer, coronary artery disease, chronic heart failure, diabetic retinopathy, stroke, hypertension, thrombosis etc. except for up-regulated hsa-miR-4659b-3p ( Table 5 ). Interestingly, 15 out of 24 down-regulated miRNAs and 2 out of 7 up-regulated have been associated with COVID-19 in the previous reports ( Table 5 ). On the other hand, 9 down-regulated and 5 up-regulated miRNAs have been empanelled for COVID-19 association for the first time in this study. Table 5 Disease association of 31 DE miRNAs identified in the present study. Table 5 S. No. MiRNA Association with COVID-19 Association with other Diseases 1 miR-147b-5p Unknown Tuberculosis [ 44 ], Cancer [ 45 ] 2 miR-107 Unknown Cancer [ 46 ], Alzheimer [ 47 ] 3 miR-224-5p Unknown Cancer [ 48 ], Cystic Fibrosis [ 49 ] 4 miR-32-3p Down-regulated in COVID-19 [ 50 ] Myocardial Injury [ 51 ], Ischemia [ 52 ] 5 miR-144-3p Reduced by more than 1.3-fold [ 53 ] Myocardial Infarction [ 54 ], Cancer [ 55 ] 6 miR-5583-3p Unknown Cancer [ 56 ], Arthritis [ 57 ] 7 miR-26b-5p Regulator of ACE2 network [ 58 ] Cardiac hypertrophy [ 53 ]; Cancer [ 59 ] 8 miR-199a-3p Anti-viral miRNA, down-regulated [ 60 ] Cancer [ 61 , 62 ] 9 miR-98-5p Regulator of TMPRSS2 transcription [ 63 ] Asthma [ 64 ] 10 miR-548k Unknown Cancer [ 65 ], Myasthenia gravis [ 66 ] 11 miR-493-5p Unknown Stroke [ 67 ], Cancer [ 68 ] 12 miR-454-3p Down-regulated in COVID-19 [ 25 ] Cancer [ 69 ] Alzheimer's Disease [ 70 ] 13 miR-141-3p Indirectly affect ACE2/TMPRSS2 expression [ 71 ] Stroke [ 72 ], Atherosclerosis [ 73 ] 14 miR-17-5p Down-regulated in COVID-19 [ 39 ] Myocardial infarction [ 59 ], Chronic kidney disease [ [74] , [75] ] 15 miR-31-5p Up-regulated in COVID-19 [ 33 ] Diabetes [ 51 ], Osteoarthritis [ 76 ] 16 miR-20b-5p Unknown Coronary Heart disease [ 77 ], Cancer [ 78 ] 17 miR-20a-5p Anti-viral miRNA, down regulated [ 60 ], Severity marker for COVID-19 [ 79 ] Diabetic cardiacmyopthy [ 80 ], Non alcoholic fatty liver disease [ 81 ] 18 miR-590-3p SARS-CoV-2 Spike transfected cells release a significant amount of exosomes loaded with miRNAs [ 82 ] Chronic Heart Failure [ 83 ], Cardiomyocyte injury [ 84 ] 19 miR-15a-5p Up-regulated in COVID-19 [ 85 ] Acute myocardial infarction [ 86 ], Endometriosis [ 87 ] 20 miR-374b-3p Host immune response [ 88 ] Systemic lupus erythematosus [ 67 ], Osteoporosis [ 89 ] 21 miR-16-5p Regulator of ACE2 network [ 58 ] Cardiomyocyte injury [ 90 ], Atherosclerosis [ 91 ] 22 miR-1277-5p Unknown Osteoarthritis [ 92 ], Parkinsons [ 93 ] 23 miR-32-5p Silencing of Tmprss2 with maximum gene suppression [ 50 ] Coronary artery disease [ 94 ], Acute myocardial infarction [ 95 ] 24 miR-548l Unknown Cancer [ 96 , 97 ] Up-regulated 25 miR-1299 Unknown Cancer [ 98 ], Diabetes [ 99 ] 26 miR-4659b-3p Unknown Unknown 27 miR-495-3p Unknown Rheumatoid Arthritis [ 100 ], Chronic Kidney disease [ 101 ] 28 miR-335-3p Unknown Hypertension [ 102 ], Cancer [ 103 ] 29 miR-4700-5p Unknown Cancer [ 104 ] 30 miR-4688-3p Unknown Aneurysm [ 105 ], Cancer [ 106 ] 31 miR-320a-5p Down-Regulated in COVID-19 [ 23 ] Thrombosis [ 107 ], Chronic heart failure [ 108 ] Disease association of 31 DE miRNAs identified in the present study. Several recent studies have listed various miRNAs significantly linked to COVID-19 pathology and severity. In a recently published study consisting of 10 COVID-19 patients and 10 healthy subjects, researchers observed alteration of 55 miRNAs in COVID-19 patients. They stated that miR-31-5p, associated with inflammation, is most up-regulated among other 55miRNAs. Along with this, they revealed by supervised machine learning that miR-423-5p, miR-23a-3p and miR-195-5p are independently associated with COVID-19 patients [ 33 ]. On the contrary, we report significant down-regulation of hsa-miR-31-5p in all COVID-19 patients in the present study. In another recent study, transcriptome analysis was done in COVID-19 patients and observed that miR-2392, a circulatory miRNA, has strong association COVID-19 infection. They designed and tested miRNA -based antiviral therapeutics that targeted miR-2392 by using in vitro human, in vivo hamster models and potential findings were noted [ 34 ]. It is also well known that miRNA are potential regulator of endothelial homeostasis which are in turn associated with vascular diseases. Predominance of endothelial dysfunction and thrombus formation has been reported in various clinical studies in COVID-19 patients. Centa and co-workers, observed the levels of miRNAs, miR-26a-5p, miR-29b-3p and miR-34a-5p, which are associated with regulation of endothelial and inflammatory signaling pathways as well as viral diseases in lung biopsies of COVID-19 patients and healthy subjects [ 35 ]. Present study also identified hsa-miR-26b-5p significantly down-regulated in Moderate patients group. Another interesting finding was that hsa-miR-320a-5p, which has been previously linked to thrombosis [ 36 ] was up-regulated in all patient groups indicating that COVID-19 pathology leads to thrombus formation. However, a contradicting finding by Deuker and co-workers suggest that hsa-miR-320 family is significantly down-Regulated in COVID-19 [ 37 ]. A small cohort study was done in serum of 40 patients diagnosed with COVID-19 along with healthy subjects and expression pattern of 20 miRNAs was evaluated using Reverse transcription quantitative real-time PCR (RT-qPCR). Researchers observed that out of twenty miRNAs, only nine miRNAs were deferentially expressed in study groups. Expression levels of hsa-miR-190a and hsa-miR-203 were significantly elevated in COVID-19 patients with respect to healthy subjects. On the other side seven miRNAs (hsa-let-7d, hsa-miR-17, hsa-miR-34b, hsa-miR-93, hsa-miR-200b, hsa-miR-200c and hsa-miR-223) showed reduced level of expression in COVID-19 patients in comparison to healthy subjects [ 38 ]. However, hsa-miR-17 is significantly down-regulated in Moderate patients group in the current study, which is in concurrence with the findings of Li and co-workers [ 39 ] who stated that COVID-19 patients and four healthy subjects showed up-regulation of 35miRNAs as well as down regulation of 38 miRNAs in patient group. One observational and multicentric study was conducted in Spain from March to June 2020 consisting of 84 COVID-19 patients. They observed that differential expression of two miRNAs (hsa-miR-192-5p and hsa-miR-323a-3p) in ICU non survivors from survivors. They also identified three miRNAs (hsa-miR-148a-3p, hsa-miR-451a, hsa-miR-486-5p) that distinguishes between ICU and ward patients [ 40 ]. Present study also identified two down-regulated (hsa-miR-147b-5p and hsa-miR-107) and one up-regulated (hsa-miR-1299) miRNAs uniquely expressed in Dead patients group (non-survivors). These miRNAs have not been linked to COVID-19 so far! Also, four uniquely expressed miRNAs (1 down-regulated: hsa-miR-224-5p and 3 up-regulated: hsa-miR-4659b-3p, hsa-miR-495-3p and hsa-miR-335-3p) were expressed in Severe patients group. Interestingly, none of these four miRNAs have been directly linked to COVID-19 so far, however hsa-miR-335-5p has been associated with COVID-19 in a previous report in which bioinformatic analysis predicted that hsa-miR-335-5p is modulated by Spike and ACE together with histone deacetylate (HDAC) pathway [ 41 ]. There are few other clinical studies on miRNAs and COVID-19. In one of them, elevated levels of expression of miR-499, miR-21, miR-155 and miR-208-a were determined in serum of COVID-19 infected patients in comparison to healthy subject [ 42 ]. Another clinical study was done in 150 COVID-19 patients and 50 controls along with evaluation of plasma miR-155 expression level using RT-qPCR. They demonstrated that there is increased miR-155 expression level in COVID-19 patients, severe and non-survivors in comparison to controls, moderate, survivors COVID-19 patients respectively [ 43 ]. Of note, we identified three families of miRNA significantly down-regulated in COVID-19 patients. Hsa-miR-20a-5p and hsa-miR-20b-5p were commonly down-regulated in all patient groups. Also, hsa-miR-32-5p and hsa-miR-548I, both down-regulated in all patient groups and their family members, hsa-miR-32-3p and hsa-miR-548K both were down-regulated in moderate patients group. There is a complex interplay of host induced miRNAs and virus encoded miRNAs that determine the severity of infection and host defence response. Recent reports indicate that miRNAs could be promising tools for early diagnosis and treatment of COVID-19 infection. This alternative novel therapy with miRNA regulation could prevent disease progression and severe condition [ 22 ]. The miRNAs identified in the present study could be used as potential biomarkers for identification of individuals who could be more prone to develop severe symptoms of COVID-19 as well as could be used as putative targets for developing treatment strategies for COVID-19 in future.

The collection of samples and data was according to the guidelines laid by Indian Council of Medical Research (ICMR) and was duly approved by ethical committee (Ref code: IEC RGSSH/01/2020/03). High throughput miRNA sequencing and bioinformatics analysis was employed to study differential expression pattern of miRNAs in three study groups (i) Dead COVID-19 patients, (ii) Severe COVID-19 patients and (iii) Moderate COVID-19 patients in comparison to healthy controls ( Fig. 1 ). Fig. 1 Schematic workflow representation of the steps involved in miRNA sequencing and data analysis. Fig. 1 Schematic workflow representation of the steps involved in miRNA sequencing and data analysis. Moderate patients were having symptoms like pneumonia with fever, cough, sore throat, nasal congestion. There saturation of peripheral oxygen (SpO 2 ) was maintained from 90 to 93% and these patients were treated in dedicated COVID health centers. Severe patients had higher respiratory rate along with severe pneumonia and respiratory distress, SpO 2 <90%. In some severe patients who could not survive, sepsis and signs of organ dysfunction were observed. Severe patients were admitted to Intensive care unit (ICU) and were also given supplemental oxygen for varying duration, based on their clinical conditions. Nasopharyngeal swabs were collected from individuals with symptoms of COVID-19 and the diagnosis of infection was confirmed by clinicians using RT-PCR (real time reverse transcription polymerase chain reaction) at Rajiv Gandhi Super Specialty Hospital (RGSSH), Delhi, India. Patients with pneumonia and respiratory distress were hospitalized at RGSSH, Delhi, India and were grouped under ‘Moderate’ and ‘Severe’ category. This categorization was based on the severity of infection and symptoms, according to the guidelines laid by Ministry of health and family welfare, Government of India ( https://www.mohfw.gov.in/pdf/ ) in accordance with World Health Organization (WHO). Peripheral blood samples were collected in PAXgene blood RNA tubes (Qiagen, Germany) from a total of 33 COVID-19 infected patients. For the present study, the patients have been grouped under dead (n = 16), severe (n = 7) and moderate (n = 10) category ( Fig. 2 A). The medical reports so obtained, were analyzed at Defence Institute of Physiology and Allied Sciences (DIPAS), DRDO, Delhi. Patient categorization with their basic details (age, sex), and co-morbidity status along with details of healthy controls has been elaborated in Table 1 . Blood was also collected in PAXgene blood RNA tubes (Qiagen, Germany) from ten healthy individuals (controls) who did not succumb to COVID-19 infection. Fig. 2 Diagrammatic representation of miRNA sequencing data analysis (A) Study groups (B) Library preparation and sequencing (C) Venn diagram shows common and unique differentially expressed miRNA in each study group i.e. Dead, Sever, Moderate with respect to Healthy subjects. Histograms shows distribution of differentially expressed miRNAs in each study group on the basis of Fold Change >±2. (D)Volcano plot of DE miRNAs (Fold Change >±2) with p value ≤ 0.05 in Dead, Severe and Moderate COVID-19 patients. Fig. 2 Table 1 Basic characteristics and clinical profile of COVID-19 patients included in the study. Table 1 Variables All Patients N = 33 Patients with co-morbidity Patients without co-morbidity Controls (With no co-morbidity) Age Group (years) <40 8 4 4 5 40–49 5 3 2 4 50–59 9 8 1 1 ≥ 60 11 10 1 – Sex MALE 22 16 6 5 FEMALE 11 9 2 5 Severity of Infection Moderate 7 6 1 – Severe 10 7 3 – Dead 16 12 4 – Co-morbidity Type 2 Diabetes Mellitus Hypertension Cardiovascular Disorder Bronchial Asthma Moderate 1 1 1 – 3 Patients suffering from both Diabetes and Hypertension *Severe 3 1 – – **Dead 1 1 2 (These two patients also suffered from hypertension) – 7 Patients suffering from both Diabetes and Hypertension *1 Severe Patient suffered from right nepherectomy **1 Dead patient suffered from hypothyroidism. Diagrammatic representation of miRNA sequencing data analysis (A) Study groups (B) Library preparation and sequencing (C) Venn diagram shows common and unique differentially expressed miRNA in each study group i.e. Dead, Sever, Moderate with respect to Healthy subjects. Histograms shows distribution of differentially expressed miRNAs in each study group on the basis of Fold Change >±2. (D)Volcano plot of DE miRNAs (Fold Change >±2) with p value ≤ 0.05 in Dead, Severe and Moderate COVID-19 patients. Basic characteristics and clinical profile of COVID-19 patients included in the study. *1 Severe Patient suffered from right nepherectomy **1 Dead patient suffered from hypothyroidism. Isolation of total RNA including miRNA from human whole blood was done using PAXgene Blood miRNA kit (Cat No. 763134, Qiagen) following manufacturer's protocol. Briefly, tubes were incubated at room temperature for 2 h before starting the isolation. The tubes were centrifuged using swing out rotor at 3000-5000× g for 10 min at 15 °C. The supernatant was decanted, the pellet dissolved in RNase free water and centrifuged. Again the supernatant was discarded and the pellet was washed with BM1. The samples were transferred to new microcentrifuge tubes, 300 μl of buffer BM2 and 40 μl of Proteinase K was added, vortex mixed briefly and incubated for 10 min at 55 °C shaker incubator at 400–1400 rpm. The samples are loaded into PAXgene shredder spin column placed in 2 ml processing tube and centrifuged at maximum speed for 3 min. The supernatant obtained was carefully transferred to new microcentrifuge tube and 700 μl of isopropanol was added and mixed. The sample was transferred into PAXgene RNA spin column, placed in 2 ml processing tube, and centrifuged for 1 min at 8000–20,000× g . The column wash was performed according to manufacturer's protocol. On column Dnase treatment was done using 80 μl of DNase. The concentration and purity of the RNA extracted was evaluated using Qubitfluorometer (Thermo Fisher Scientific, MA, USA). The samples were run on Bioanalyzer (Agilent) and % miRNA was obtained ( Fig. 2 B). Small RNA sequencing (smRNA) libraries were prepared with QIAseq® miRNA Library Kit (Cat: 331,502) protocol (Qiagen, Maryland, U.S.A.) at Genotypic Technology Pvt. Ltd., Bangalore, India. Briefly, 10 ng–63 ng of Qubit quantified total RNA was used as starting material. 3′ adapters were ligated to the specific 3′OH group of micro RNAs followed by ligation of 5′ adapter. Adapter ligated fragment was reverse transcribed with Unique Molecular Index (UMI) assignment by priming with reverse transcription primers. cDNA thus formed was enriched and barcoded by PCR amplification (17–19 cycles). The 3′ and 5′ adapters used in the prep are: 5′ GTTCAGAGTTCTACAGTCCGACGATC; 5′ AACTGTAGGCACCATCAAT. The Illumina-compatible sequencing libraries were quantified by Qubit fluorometer. The fragment size distribution of the libraries was analyzed on Agilent 2200 TapeStation. The sequencing was carried out for 50 cycles on Illumina NovaSeq 6000 sequencing platform following manufacturer's instructions at Genotypic Technology Pvt. Ltd., Bangalore, India ( Fig. 1 ). The raw reads were processed for filtering of specific length of 16–40 bases and were mapped to genome. The raw data was processed by srna-workbenchV3.0_ALPHA 1 which was used to trim 3′ adapter and performed length filtering. The criteria for removing contaminated reads were elimination of low quality reads (<q30), elimination of 3′ adapters and reads 40bp as well as elimination of reads matching to reads matching to other ncRNAs (r, t, sn, and snoRNAs). These reads were further mapped to non-coding RNA database. The final clean reads were made unique and hence read count profile was generated. This information was used for understanding the expression pattern of miRNAs. Differential Expression (DE) analysis was carried out using DESeq 6 tool. The unmapped reads were further used for classification of known and novel miRNAs. Conserved miRNAs were identified by a homology approach against matured human miRNAs from miRBase. Homology search was performed using miRNA sequences retrieved from miRbase-22 using ncbi-blast-2.2.30 with e-value of e −4 and non-gapped alignment. Novel miRNAs were identified based on the secondary structure prediction and target prediction. Sequences which have no homology with known miRNAs were extracted and considered for prediction of potential novel miRNAs. Firstly, the sequences were aligned to the reference genome using bowtie 2. The aligned sequences were used for novel miRNA prediction using Mireap_0.22b 5. The prediction is based on the identification of stem loop structure and miRNA with such secondary structure are reported as potential novel miRNAs ( Fig. 1 ). DE of miRNAs was compared across three study groups, Dead, Severe and Moderate to identify unique and common miRNAs expressed in each group. Selection criteria for miRNAs in each group were on the basis of fold change and p-value. Receiver operating characteristic (ROC) curve for each DE miRNA was plotted using webtool, EPITOOLS ( https://epitools.ausvet.com.au/roccurves ). Target prediction for all shortlisted miRNAs was done using online tool, miRDB [ 29 , 30 ]. Target genes of DE miRNAs were subjected to pathway and network analysis. Candidate miRNAs were selected on the basis of their significant expression and target genes associated with biological pathways closely involved in COVID-19 pathogenesis. The qualitative and quantitative assessment of samples was done to obtain optimal yield for Illumina library preparation. The ratio of absorbance at 260 nm and 280 nm was used to assess the purity RNA, which lied in between 2.0 and 2.1. The Illumina libraries showed average fragment size range of 220 bp with sufficient concentration to get the desired amount of sequencing data. The data obtained from the sequencing run was demultiplexed using Bcl2fastq software v2.20 and FastQ files were generated based on the unique dual barcode sequences. The sequencing quality was assessed using FastQC v0.11.8 software. The adapter sequences were trimmed and bases above Q30 were considered and low quality bases were filtered off during read pre-processing and used for downstream analysis.

Coronavirus disease 2019 (COVID-19) caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has rapidly evolved and spread as pandemic. This highly transmissible virus, primarily affecting lungs and causing severe pneumonia, belongs to genus Betacoronavirus and was first identified in China in late 2019 [ 1 ]. Most common symptoms of this novel coronavirus include fever, cough, cold, shortness of breath, nausea, fatigue, muscle soreness, sputum production, dyspnea, diarrhea and severe pneumonia [ 2 , 3 ] and start appearing after an incubation period of 1–14 days. SARS-CoV-2 genome has a positive single stranded RNA with size of around 30 kbs in length having major structural proteins like spike glycoprotein (S), envelope (E), membrane (M) and nucleocapsid (N). S protein has two major subunits, S1 and S2, which binds to host angiotensin-converting enzyme 2 (ACE-2) and facilitates entry of virus into target proteins [ 4 ]. The symptoms of COVID-19 may range from mild, moderate to severe depending upon age, immune response and co-morbidity in a person. Various clinical studies have been reported that co-morbidities like cancer, diabetes, obesity and genetic deficiencies like inborn error of immunity (EI) complicate the pathophysiology and severity of COVID-19. Patients with co-morbidities need special medical attention and treatment to overcome the risk and mortality [ [5] , [6] , [7] ]. Under treatment perspective, trials are being conducted to examine anti-PD1 (programmed death-1) antibody efficacy for cancer patients suffering with COVID-19 infection [ 8 ]. Major determinants of infection severity include high levels of circulating cytokines and chemokines [importantly interleukin (IL)-6, IL-8, and tumor necrosis factor (TNF)-α], lymphopenia and immune cell infiltration in infected organ, which are responsible for varied immune response [ 9 , 10 ]. Thus progression and severity of COVID-19 disease, caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infections largely depends on host cell response. Apart from reverse transcription polymerase chain reaction (RT-PCR) method, several other diagnostic strategies for detection of COVID-19 infection are in process of approval from competent authority. These include isothermal nucleic acid amplification, cluster of regularly interspaced short pallidromic repeats/cas based approaches and digital PCR [ 11 ]. Biomarkers currently being used for COVID-19 prognosis and diagnosis may be categorized as inflammatory, biochemical, cardiac, immunological, haematological and coagulatory markers etc [ 12 ]. Several biomarkers have been experimentally evaluated for estimating risk of COVID-19 severity such as D-dimer, cardiac troponin, lactate dehydrogenase (LDH), C-reactive protein (CRP), Vitamin D etc. [ 13 , 14 ]. However, efficacy of these biomarkers for clinical use requires necessary confirmatory studies. Till now, COVID-19 treatment modalities include anti-biotic and anti-viral drugs, glucocorticoids/steroids treatment, invasive and non-invasive ventilation and renal replacement therapy [ 2 , 15 , 16 ]. However, none of these strategies are very specific, completely safe and definite for COVID-19 treatment and have varying response on individuals [ 17 ]. MicroRNAs (miRNAs) are small, non-coding post transcriptional repressors of gene expression which inhibit gene activity either by binding to a specific region of gene in 3′-untranslated region (3′-UTR) thereby degrading mRNA or by blocking the translation process [ 18 ]. Extensive research in the last decade established that small RNAs play a central role in pathogenesis and progression of many diseases. Previous studies have demonstrated that the binding of miRNAs on viral RNA has adverse effects on the virus genome [ 18 ]. Host miRNA expression can be important anti-viral tool [ 19 ] as they stimulate immune response [ 20 ] by mediating T cells and antiviral effector functions [ 21 ]. Thus miRNA biomarkers could be an alternate treatment strategy for COVID-19 [ 22 ]. In addition, pathogeneis of several human diseases are associated with dysregulation of miRNAs and biological processes controlled by these miRNAs. In this regard, previous reports have shown that severity of COVID-19 infection is associated with changes in miRNA levels especially in patients with co-morbidities such as respiratory diseases, diabetes, heart failure and kidney problems [ 23 ]. In case of COVID-19 infection, miRNAs can inhibit the viral translation by binding to 3′-UTR of the viral genome or target the receptors, structural or non-structural proteins of virus without affecting the expression of human genes [ 24 ]. MiRNAs associated with breast cancer, diet, physical activity, plant food bioactive (polyphenols) etc. may also play a critical role in epigenetic regulation of COVID-19 pathophysiology [ 25 , 26 ]. Differential expression of miRNAs may increase complications among COVID-19 patients by modulating expression of several genes like ACE-2 and other associated pathways [ 27 ]. Thus, miRNAs have potential to be used as preventive and therapeutic tool to minimize the risk and complications of COVID-19 infection [ 28 ]. Mechanism of epigenetic interplay of SARS-CoV-2 and host miRNAs is yet to be elucidated. However, understanding miRNA expression pattern in COVID-19 patients could help in explaining the basis of pathogenesis of novel SARS-CoV-2. Present study was undertaken to examine the miRNA expression pattern in peripheral blood of COVID-19 infected individuals with different levels of infection severity in comparison to healthy controls in order to identify potential miRNA biomarkers that can predict the severity of COVID-19 infection.

The authors declare no competing interests.

Data included in article/supp. material/referenced in article.