Identification of Lung and Blood Microbiota Implicated in COVID-19

cells

Article Identiﬁcation of Lung and Blood Microbiota Implicated in COVID-19 Prognosis

Kypros Dereschuk 1,2 , Lauren Apostol 1,2, Ishan Ranjan 1,2, Jaideep Chakladar 1,2, Wei Tse Li 1,2, Mahadevan Rajasekaran 3,4, Eric Y. Chang 5,6 and Weg M. Ongkeko 1,2,*

1 Division of Otolaryngology-Head and Neck Surgery, Department of Surgery, University of California, San Diego, CA 92093, USA; [email protected] (K.D.); [email protected] (L.A.); [email protected] (I.R.); [email protected] (J.C.); [email protected] (W.T.L.) 2 Research Service, VA San Diego Healthcare System, San Diego, CA 92161, USA 3 Department of Urology, University of California San Diego, La Jolla, CA 92093, USA; [email protected] 4 Urology Service, VA San Diego Healthcare System, San Diego, CA 92161, USA 5 Department of Radiology, University of California, San Diego, CA 92093, USA; [email protected] 6 Radiology Service, VA San Diego Healthcare System, San Diego, CA 92161, USA * Correspondence: [email protected]; Tel.: +1-(858)-552-8585-7165

Abstract: The implications of the microbiome on Coronavirus disease 2019 (COVID-19) prognosis has not been thoroughly studied. In this study we aimed to characterize the lung and blood microbiome and their implication on COVID-19 prognosis through analysis of peripheral blood mononuclear cell (PBMC) samples, lung biopsy samples, and bronchoalveolar lavage fluid (BALF) samples. In all three tissue types, we found panels of microbes differentially abundant between COVID-19 and Citation: Dereschuk, K.; Apostol, L.; normal samples correlated to immune dysregulation and upregulation of inflammatory pathways, Ranjan, I.; Chakladar, J.; Li, W.T.; including key cytokine pathways such as interleukin (IL)-2, 3, 5-10 and 23 signaling pathways and Rajasekaran, M.; Chang, E.Y.; downregulation of anti-inflammatory pathways including IL-4 signaling. In the PBMC samples, Ongkeko, W.M. Identification of Lung six microbes were correlated with worse COVID-19 severity, and one microbe was correlated with and Blood Microbiota Implicated in improved COVID-19 severity. Collectively, our findings contribute to the understanding of the human COVID-19 Prognosis. Cells 2021, 10, microbiome and suggest interplay between our identified microbes and key inflammatory pathways 1452. https://doi.org/10.3390/cells which may be leveraged in the development of immune therapies for treating COVID-19 patients. 10061452

Keywords: microbiome; lung microbiota; blood microbiota; SARS-CoV-2; COVID-19; coronavirus; in- Academic Editor: Maria Cristina ﬂammation Curia

Received: 11 May 2021 Accepted: 8 June 2021 1. Introduction Published: 10 June 2021 As of 19 March 2021, 122,044,376 people have been infected and 2,695,014 have Publisher’s Note: MDPI stays neutral died from coronavirus disease 2019 (COVID-19), caused by the severe acute respiratory with regard to jurisdictional claims in syndrome coronavirus 2 (SARS-CoV-2) [1]. Symptoms and severity of COVID-19 vary published maps and institutional afﬁl- drastically. The most common symptoms are fever, cough, fatigue, headache, myalgias, iations. and diarrhea. The severity of COVID-19 symptoms can range from very mild to severe [2,3]. Around one in six infected individuals present with no symptoms at all [4]. Approximately 10% of people infected with COVID-19 experience symptoms’ that persist beyond three weeks in what is known as “long haul COVID-19” [5]. The cause of the large variance

Copyright: © 2021 by the authors. in COVID-19 severity and length is not fully understood, and research into this topic is Licensee MDPI, Basel, Switzerland. of great importance to learn how to prevent long haul COVID-19 [6,7]. In more severe This article is an open access article cases of COVID-19, the innate immune system fails to stop viral replication of SARS-CoV-2, distributed under the terms and leading to a substantial immune response from immune effector cells which has previously conditions of the Creative Commons been characterized by a large amount of cytokines in the body. This hyperinﬂammatory Attribution (CC BY) license (https:// condition manifested as a ‘cytokine storm’ is called COVID-19 ARDS, and this is one of the creativecommons.org/licenses/by/ most dangerous and potentially life-threatening events related to COVID-19 [8]. 4.0/).

Cells 2021, 10, 1452. https://doi.org/10.3390/cells10061452 https://www.mdpi.com/journal/cells Cells 2021, 10, 1452 2 of 18

The gut microbiome and lung microbiome have both been found previously to be altered in COVID-19 patients, in addition to likely playing a role in explaining the variation in critical COVID-19 response [9–11]. The lung microbiome has been shown to play an important role in maintaining lung homeostasis and plays a role in prompting an adequate immune response to pathogens, preventing a hyper-inflammatory response via crosstalk between microbes and receptors on immune cells within the lungs, while some microbes/microbiota compositions appear to prompt more inflammatory responses to pathogens [11–14]. Previous study has found that severity of COVID-19 disease is correlated with the predominance of opportunistic pathogens in the gut [10]. Building upon this prior research, we wished to investigate the microbiome further by correlating microbe abundance, both fungi and bacteria, to COVID-19 severity, immune cell abundances, and inflammatory and immune-associated gene pathways, and investigate, in addition to the lung microbiome, the effect the blood microbiome may have in modulating the immune response to COVID-19. Thus, in this study we aimed to characterize the lung and blood microbiome and their implication on COVID-19 prognosis through analysis of peripheral blood mononuclear cell (PBMC) samples, lung biopsy samples, and bronchoalveolar lavage fluid (BALF) samples.

2. Materials and Methods An overview of our methods is presented in Figure1.

2.1. Data Acquisition RNA-seq data of 17 PBMC normal samples and 17 PBMC COVID-19 samples were downloaded from GEO (accession code GSE152418 https://www.ncbi.nlm.nih.gov/geo/ accessed 15 July 2020), 8 lung biopsy normal samples and 8 lung biopsy COVID-19 samples were downloaded from GEO (accession code GSE147507 https://www.ncbi.nlm.nih. gov/geo/ accessed 15 July 2020), and 20 BALF normal samples and 8 BALF COVID-19 samples were downloaded from GSA (accession code HRA000143 https://bigd.big.ac. cn/gsa-human/ accessed 15 July 2020) and 4 additional BALF COVID-19 samples were downloaded from GSA (accession code CRA002390 https://bigd.big.ac.cn/gsa-human/ accessed 15 July 2020). Arunachalam et al. determined disease severity for PBMC samples and classiﬁed patients as convalescent, moderate, severe, or ICU [15]. The dataset included 1 convalescent sample, 4 moderate samples, 8 severe samples, and 4 ICU samples.

2.2. Extraction of Microbial Reads Pathoscope 2.0 (Boston University School of Medicine, Boston, MA, USA) was used to separate the microbe-speciﬁc reads incorporated in the human reads of high-throughput RNA-seq and align it to the reads in a target library, producing levels of microbe abundance and individual taxonomic lineage [16]. This was done two times, once with a target library containing reads of bacteria and once with a target library containing reads of fungus. Microbes with total abundance less than the number of COVID-19 patients per tissue type were excluded from analysis. The batch correction method ComBat was used to adjust for batch effects when combining BALF data from the HRA000143 and CRA002390 datasets [17].

2.3. Differential Microbial Abundance between COVID19 and Normal Patients The Kruskal-Wallis statistical test was used to determine differential abundance between COVID-19 samples and normal samples and correlations between microbe abundance and disease severity (p < 0.05). For correlation between microbe abundance and disease severity, ICU and Severe patients were grouped together into one group called Severe.

2.4. Correlation between Microbial Abundance and IA Gene Expression All RNA-seq samples were aligned to human genome version GRCh38.p13 and its an- notation from NCBI using STAR (Spliced Transcript Alignment to a Reference) (Cold Spring Cells 2021, 10, 1452 3 of 18

Harbor Laboratory, Cold Spring Harbor, New York, NY, USA) version 2.7.4 with sjdbOver- hang set to optimize value of 99. Other parameters were left to default values [18]. We used feature counts function from package Rsubread v2.0.1 (Subread Sequence Alignment and Counting for R) (Olivia Newton-John Cancer Research Institute, Melbourne, Australia) to obtain raw read count and the result was subsequently passed to EdgeR v3.28.1 (Walter and Eliza Hall Institute of Medical Research, Parkville, Australia) for normalization using TMM metrics. TMM normalization enables us to compare gene expression across samples. The Kruskal-Wallis statistical test was used to correlate gene expression with COVID-19 status and to correlate microbe abundance to dysregulated immune associated (IA) genes.

2.5. Correlation between Microbial Abundance and Immune Cell Abundances The CIBERSORTx software (Stanford University, Stanford, CA, USA) was used to de- convolute RNA-sequencing data to estimate the inﬁltration levels of 22 immune cell types. These immune cell types include the following: CD8 T-cells, CD4 naïve T-cells, CD4 memory resting T-cells, CD4 memory activated T-cells, follicular helper T-cells, regulatory T-cells, gamma-delta T-cells, naïve B-cells, memory B-cells, plasma cells, M0-M2 macrophages, resting dendritic cells, activated dendritic cells, resting NK cells, activated NK cells, monocytes, resting mast cells, activated mast cells, eosinophils, and neutrophils [19]. We then correlated microbe abundance with expression levels of the different immune cells using the Kruskal-Wallis statistical test. (p < 0.05). Patients with lower or higher microbial read counts than the median microbial read count of a particular microbe across all patients were deﬁned as “LOW” or “HIGH,” respectively.

2.6. GSEA (Correlation of Microbial Abundance and Covid Status to Canonical Pathways and Immune-Associated Signatures) We used Gene Set Enrichment Analysis (GSEA) version 4.1.0 (UCSD and Broad Institute, San Diego and Cambridge, United States) to find gene enrichment related to microbe abundance and covid status. Final best hit read numbers, scores that are used by Pathoscope to represent numbers of microbe reads in RNA-seq numbers, are used as numerical phenotype input. In separate runs, covid status was used as categorical phenotype input. We manually selected immune related pathways from CP (canonical pathways) and pooled the gene sets with C7 (immunologic signature gene sets) for gene set input. We set the number of permutations to be 1000 and no collapse gene symbols. Metric for ranking genes was set to Pearson. Everything else was left with default parameters. Significantly enriched signatures were identified by a nominal p-value < 0.05 and ranked by normalized enrichment score [20]. Genes involved in the three immunological signatures correlated to the greatest number of differentially abundant microbes were input into Cytoscape Reactome FIViz software version 3.7.2 to visualize the gene networks [21].

2.7. Contamination Correction Using Spearman’s Correlation The abundance of individual microbes in each patient were plotted against total microbe reads in the same patient separated by tissue type to determine if any microbe is a likely contaminant. Best hit results from Pathoscope were used. If a scatterplot shows a positive slope, it suggests that the microbe was biologically relevant. If there is a vertical or near vertical slope and the counts of all the microbes are substantially above zero, then the microbe is likely a contaminant. If there is a slope close to zero or less than zero, the test is inconclusive. This reasoning follows from the assumption that similar amounts of microbes will be present regardless of how many microbes are present in the tissue sample if the microbe is an environmental contaminant. Cells 2021, 10, 1452 4 of 18 Cells 2021, 10, x FOR PEER REVIEW 4 of 18

FigureFigure 1. 1.Schematic Schematic of of analysis analysis and and workﬂow. workflow.

3.3. Results Results 3.1.3.1. Differential Differential Microbial Microbial Abundance Abundance in in COVID-19 COVID-19 Tissue Tissue and and Normal Normal Tissue Tissue UsingUsing Pathoscope Pathoscope 2.0, 2.0, we we found found 91 91 bacteria bacteria and and 14 14 fungus fungus differentially differentially abundant abundant in in lunglung biopsy biopsy samples, samples, 13 13 bacteria bacteria and and 9 9 fungus fungus differentially differentially abundant abundant in in PBMC PBMC samples, samples, andand 12 12 bacteria bacteria and and 57 57 fungus fungus differentially differentially abundant abundant in in BALF BALF samples samples (Figure (Figure2 A–C).2A–C). Bacteroides fragilis Thermoanaerobacterium thermosaccharolyticum DSM 571 Escherichia Bacteroides fragilis, , Thermoanaerobacterium thermosaccharolyticum DSM 571, and, and Escherichia (E.) coli were the only bacteria and Tremella fuciformis and Aspergillus oryzae were the only (E.) coli were the only bacteria and Tremella fuciformis and Aspergillus oryzae were the only fungus found differentially abundant between COVID-19 samples and normal samples in fungus found differentially abundant between COVID-19 samples and normal samples in both BALF and PBMC tissues. There was no overlap in microbes found to be differentially both BALF and PBMC tissues. There was no overlap in microbes found to be differentially abundant in lung biopsy samples and BALF or PBMC samples. abundant in lung biopsy samples and BALF or PBMC samples.

Cells 2021, 10, 1452 5 of 18

Lachancea thermotolerans Penicillium rubens Wisconsin 54-1255 uncultured Morganella Candida orthopsilosis Cladosporium sp. A04 Moraxella bovoculi A C Kocuria rhizophila DC2201 Kytococcus sedentarius DSM 20547 Deinococcus geothermalis DSM 11300 Bacillus subtilis subsp. subtilis str. 168 Mycoplasma hyorhinis HUB-1 Pseudocercosporella sp. CPC 4008 Azospirillum sp. B510 Conidiobolus coronatus Veillonella parvula DSM 2008 Rhopalomyces elegans uncultured planctomycete Beauveria bassiana Clostridium perfringens Aspergillus oryzae Deinococcus radiodurans Bacteroides fragilis YCH46 Rothia mucilaginosa DY-18 Micrococcus luteus NCTC 2665 Streptomyces ambofaciens Campylobacter hominis ATCC BAA-381 Psychrobacter lutiphocae Thermoanaerobacterium thermosaccharolyticum DSM 571 Rhodobacter capsulatus Sugiyamaella castrensis Moraxella catarrhalis Thermoanaerobacter pseudethanolicus ATCC 33223 Acidiphilium symbioticum Staphylococcus epidermidis Streptomyces roseoverticillatus Delftia acidovorans SPH-1 Micrococcus luteus NCTC 2665 Staphylococcus capitis Lactobacillus vermiforme Pseudomonas sp. I-09 Haemophilus parain uenzae T3T1 fungal sp. JF52 Paracoccus denitricans Tremella fuciformis uncultured bacterium HF0010_04H24 Bacillus sp. PL-12 Stenotrophomonas rhizophila Escherichia coli Actinomadura citrea Bacillus subtilis Oerskovia turbata NBRC 15015 Vibrio vulnicus PBMC COVID-19 (n=17) Normal (n=17) Pseudomonas mendocina ymp Finegoldia magna ATCC 29328 Corynebacterium propinquum DSM 44285 Burkholderia cenocepacia MC0-3 Streptococcus oralis Achromobacter xylosoxidans A8 Mycoplasma hyopneumoniae Gardnerella vaginalis Ralstonia solanacearum Cardiobacterium hominis Arthrobacter sp. FB24 Komagataeibacter intermedius Klebsiella aerogenes Bordetella bronchiseptica uncultured beta proteobacterium CBNPD1 BAC clone 578 Pantoea endophytica B Paenibacillus popilliae uncultured bacterium 270 fungal sp. JF28 Methylobacterium sp. 4-46 Postia sp. CLZ-2014b uncultured bacterium zdt-44a23 Mesophellia glauca uncultured delta proteobacterium Agaricus bisporus Mycoplasma hyorhinis fungal sp. JF8 Leptothrix cholodnii SP-6 fungal sp. JF46 Streptomyces alboniger fungal sp. JF42 Helicobacter canis Tremella fuciformis Cupriavidus pinatubonensis JMP134 Kluyveromyces marxianus Salmonella enterica subsp. enterica serovar Schwarzengrund str. CVM19633 fungal sp. JF33 Anaerostipes hadrus Saccharomycopsis buligera Clavibacter michiganensis subsp. tessellarius Taiwanofungus camphoratus Xanthomonas fragariae fungal sp. JF39 Bidobacterium mongoliense Funneliformis mosseae Bidobacterium sp. GC61 fungal sp. JF65 Gardnerella vaginalis ATCC 14019 fungal sp. JF38 uncultured Friedmanniella sp. Botrytis cinerea Atopobium parvulum DSM 20469 fungal sp. JF2 Varibaculum cambriense Phakopsora pachyrhizi Paenibacillus lentimorbus Aspergillus oryzae uncultured bacterium 59 fungal sp. JF10 Ochrobactrum anthropi fungal sp. JF57 Rhizobium etli CFN 42 Thermothielavioides terrestris NRRL 8126 Mycolicibacterium smegmatis MC2 155 Histophilus somni 129PT Terrisporobacter glycolicus ATCC 14880 = DSM 1288 Haemophilus in uenzae Rd KW20 Cellulomonas avigena DSM 20109 Haemophilus in uenzae R2846 Burkholderia vietnamiensis Haemophilus in uenzae R2866 Comamonas testosteroni Haemophilus in uenzae PittGG Corynebacterium glutamicum R Moraxella catarrhalis BBH18 Staphylococcus carnosus subsp. carnosus TM300 Mycoplasma conjunctivae Acinetobacter baumannii ATCC 17978 Chlorobium phaeobacteroides BS1 Corynebacterium amycolatum Anabaena sp. K119 Roseburia intestinalis XB6B4 Bacteroides fragilis YCH46 Roseburia intestinalis M50/1 Escherichia coli Mycobacterium kansasii ATCC 12478 Thermoanaerobacterium thermosaccharolyticum DSM 571 Sinorhizobium meliloti Naumovozyma castellii CBS 4309 Streptococcus gordonii Trichophyton equinum Escherichia coli str. K-12 substr. W3110 Aspergillus scheri NRRL 181 Burkholderia pseudomallei 1106a Saccharomyces cerevisiae YJM1341 Trichormus variabilis ATCC 29413 Monilinia laxa Burkholderia multivorans ATCC BAA-247 Cryphonectria parasitica uncultured proteobacterium EBAC39D12 Colletotrichum capsici Mycobacterium malmoense Histoplasma capsulatum Pediococcus cellicola Saccharomyces cerevisiae YJM450 Streptomyces netropsis Saccharomyces cerevisiae YJM1592 uncultured beta proteobacterium fungal endophyte Agrobacterium vitis Saccharomyces cerevisiae YJM1307 Campylobacter ureolyticus Saccharomyces cerevisiae YJM1389 Meiothermus silvanus DSM 9946 Saccharomyces cerevisiae YJM1574 Megamonas hypermegale ART12/1 Monilinia fructicola Leuconostoc kimchii IMSNU 11154 Suillus pictus Anoxybacillus avithermus WK1 Seiridium cardinale Caulobacter segnis ATCC 21756 Ampelomyces quisqualis Deinococcus radiophilus Heterobasidion annosum Klebsiella pneumoniae 342 Saccharomyces cerevisiae YJM1078 Psychrobacter cryohalolentis K5 Aureobasidium subglaciale EXF-2481 Faecalibacterium prausnitzii L2-6 Lichtheimia ramosa Citrobacter koseri ATCC BAA-895 Candida parapsilosis Pseudomonas syringae pv. tomato str. DC3000 Tremella mesenterica DSM 1558 Novosphingobium aromaticivorans DSM 12444 Mucor racemosus Sphingomonas sp. KT0216 Chaetomium thermophilum var. thermophilum DSM 1495 Faecalibacterium prausnitzii SL3/3 Moesziomyces antarcticus cyanobiont of Histioneis sp. Pseudocercospora musae Bacteroides eggerthii DSM 20697 Syncephalastrum monosporum var. pluriproliferum Herbaspirillum autotrophicum Saccharomyces cerevisiae YJM1444 Streptococcus sanguinis SK1 = NCTC 7863 Lentinula edodes Pseudomonas baetica Yarrowia lipolytica CLIB122 Myroides odoratus Pseudocercospora eumusae Staphylococcus epidermidis RP62A Aspergillus niger Corynebacterium diphtheriae BALF COVID-19 (n=12) Normal (n=20) Pseudomonas aeruginosa Pseudomonas syringae pv. syringae B728a Flavobacterium johnsoniae UW101 Streptococcus thermophilus LMD-9 Leuconostoc gelidum subsp. gelidum KCTC 3527 Pseudomonas uorescens SBW25 Streptococcus thermophilus uncultured bacterium HF0500_12O04 fungal sp. JF54 fungal sp. JF53 1 0 uncultured Tuber fungal sp. JF3 Rozella sp. JEL347 fungal sp. SNB-GTC2112 Normalized Peziza vesiculosa uncultured Leucocortinarius Microbial Abundance uncultured fungus Lung Biopsy Normal (n=8) COVID-19 (n=8)

Figure 2. Differential abundance summary. Heatmaps showing normalized microbe abundance for COVID-19 samples vs. normal samples for (A) PBMC, (B) BALF, and (C) lung biopsy. Cells 2021, 10, 1452 6 of 18

3.2. Immune Landscape of COVID-19 Patients Most immunologic gene signatures correlated to COVID-19 status are of cells in both the innate and adaptive immune systems such as macrophages, dendritic cells, mast cells, T-cells, B-cells, and monocytes. Other results of significance include the following: across PBMC samples, T cell receptor (TCR) signaling was downregulated in COVID-19 patients, immunological signatures characteristic of increased levels of pro-inflammatory interleukins, such as IL-6, IL-8, and IL-12 were upregulated in COVID-19 patients and anti-inflammatory interleukins, including IL-4 and IL-10, were downregulated in COVID- 19 patients (Figure3C). We found plasma cells and memory B cells to be significantly increased in COVID-19 patients and naive B cells, resting memory CD4 T cells, resting natural killer (NK) cells, and Eosinophils significantly decreased in COVID-19 patients (Figure3E). Across BALF samples, retinoic acid-inducible gene-I (RIG-I)-like receptor signaling, transforming growth factor (TGF) β signaling, and TCR signaling were upregulated. Similar interleukin dysregulation found in PBMC samples was found in BALF samples (Figure3A). We found activated mast cells, M1 macrophages, eosinophils, neutrophils abundances were significantly increased in COVID-19 patients and T regulatory cells (Tregs) M0 macrophages and activated dendritic cells abundance was significantly decreased in COVID-19 patients (Figure3D). Across lung biopsy samples, IL-4 and IL-10 signaling, TCR signaling, and the complement system pathway of the innate immune system were down regulated (Figure3B). We found follicular helper T cells, gamma delta T cells, naive CD4 T cells, activated NK cells, and memory B cells significantly increased in COVID-19 patients and resting NK cells, plasma cells, M0, M1, and M2 macrophages, memory CD4 T cells, CD8 T cells, naive B cells, activated dendritic cells, and eosinophils significantly decreased in COVID-19 patients (Figure3D).

3.3. Microbes in Blood Are Correlated to Disease Severity E. coli abundance, Bacillus sp. PL-12 abundance, Campylobacter hominis ATCC BAA-381 abundance, Pseudomonas sp. I-09 abundance, Thermoanaerobacter pseudethanolicus ATCC 33223 abundance, Thermoanaerobacterium thermosaccharolyticum DSM 571 abundance, and Staphylococcus epidermis abundance were correlated with COVID-19 severity. Bacillus subtilis subsp. subtilis str. 168 abundance was inversely correlated with COVID-19 severity (Figure4). Staphylococcus capitis abundance correlation to COVID-19 severity was not signiﬁcant. No signiﬁcant correlation to age was found with these microbes. Cells 2021, 10, 1452 7 of 18

A BALF Genes up-regulated in CD8 T cells: ID3 [GeneID=3399] knockout versus wildtype Genes down-regulated in B lymphocytes treated by anti IgM for 3h: wildtype versus MAP3K7 [GeneID=6885] knockout Genes down-regulated in bone marrow-derived macrophages treated with IL4 [GeneID=3565]: wildtype versus PPARG [GeneID=5468] knockout Genes down-regulated in bone marrow-derived macrophages with STAT6 [GeneID=6778] knockout treated with rosiglitazone [PubChem=77999]: control versus IL4 [GeneID=3565] Genes up-regulated in comparison of dendritic cells (DC) stimulated with CpG DNA (TLR9 agonist) at 0.5 h versus those stimulated with CpG DNA (TLR9 agonist) at 24 h Genes up-regulated in comparison of unstimulated CD8 T cells at 72 h versus CD8 T cells at 72 h after stimulation with IL12 Genes down-regulated in comparison of dendritic cells (DC) exposed to L. donovani versus DCs exposed to M. tuberculosis Genes up-regulated in macrophages dierentiated for 5 days in the presence of: IL4 [GeneID=3565] versus IL4 [GeneID=3565] and dexamethasone [PubChem=5743] Genes down-regulated in comparison of polysome bound (translated) mRNA before and 16 h after LPS (TLR4 agonist) stimulation Genes down-regulated in comparison of unstimulated macrophage cells versus macrophage cells stimulated with LPS (TLR4 agonist) for 240 min Genes up-regulated in comparison of adult DN2 thymocytes versus fetal DN2 thymocytes Genes down-regulated in the immature neuron cell line: control versus infected with western equine encephalitis viruss KEGG RIG-I-like receptor signaling pathway REACTOME Inuenza Infection Genes up-regulated in the activated CD4 [GeneID=920] T cells (48h): control versus IL-12 Genes down-regulated in B lymphocytes: naïve versus B1 REACTOME MAPK6/MAPK4 signaling Genes down-regulated in untreated double positive thymocytes: wildtype versus ELK1 and ELK4 [GeneID=2002;2005] knockout Genes up-regulated in thymic: dendritic cells versus macrophages Genes down-regulated in thymocytes: cortical versus medullary sources KEGG TGF-beta signaling pathway Genes down-regulated in macrophages with IL10 [GeneID=3586] knockout treated by LPS and IL10 [GeneID=3586]: 10 min versus 30 min Genes down-regulated in the activated CD4 [GeneID=920] T cells (48h): IL-12 versus interferon alpha Genes down-regulated in comparison of control dendritic cells (DC) at 24 h versus those stimulated with Gardiquimod (TLR7 agonist) at 24 h Genes up-regulated in comparison of IgD+ B cells versus IgD- B cells

Normal Enrichment Score B Lung Biopsy BIOCARTA T Cell Receptor Signaling Pathway Genes up-regulated in macrophages dierentiated for 5 days in the presence of: IL4 [GeneID=3565] versus IL4 [GeneID=3565] and dexamethasone [PubChem=5743] Genes down-regulated in comparison of memory CD8 T cells versus eector CD8 T cells KLRG1 intermediate [GeneID=10219] enes up-regulated in comparison of dendritic cells (DC) versus DCs exposed to M.tuberculosis Genes up-regulated in comparison of dendritic cells (DC) stimulated with poly(I:C) (TLR3 agonist) at 16 h versus DC cells stimulated with Gardiquimod (TLR7 agonist) at 16 h Genes up-regulated in B lymphocytes: B2 versus B1 Genes down-regulated in comparison of control dendritic cells (DC) at 0 h versus those stimulated with Gardiquimod (TLR7 agonist) at 0.5 h Genes down-regulated in B lymphocytes treated by anti IgM for 24h: wildtype versus MAP3K7 [GeneID=6885] knockout Genes up-regulated in CD4 [GeneID=920] T helper cells Th0: 10h versus 60h Genes down-regulated in comparison of unstimulated NK cells versus those stimulated with IL2 [GeneID=3558] at 16 h Genes down-regulated in comparison of neutrophils versus eector memory CD4 [GeneID=920] T cells Genes down-regulated in comparison of naive B cell versus pre-germinal tonsil B cells KEGG Systemic lupus erythematosus Genes up-regulated in comparison of dendritic cells (DC) versus monocytes Genes down-regulated in IKZF1 [GeneID=10320] knockout: lymphoid-primed multipotent progenitors versus granulo-monocyte progenitors Genes up-regulated in comparison of Ig isotype switched memory B cells versus plasma cells Genes down-regulated in cortical thymic epithelial cells (cTEC) versus thymic macrophages Genes down-regulated in comparison of monocytes treated with anti-TREM1 [GeneID=54210] and 5000 ng/ml LPS (TLR4 agonist) versus untreated monocytes PID IL2-mediated signaling events Genes up-regulated in CD4 [GeneID=920]: naïve versus FOXP3+ [GeneID=50943] T reg Genes up-regulated in comparison of Th2 cells versus Th17 cells Genes down-regulated in monocytes (3h): untreated versus M. tuberculosis 19 kDa lipopeptide Genes down-regulated in mature neuron cell line: control versus interferon alpha (6h) Genes up-regulated in dendritic cells: CIITA [GeneID=4261] knockout versus I ab-/- mice Genes up-regulated in memory CD8 T cells: ITGAE- [GeneID=3682] from spleen versus ITGAE+ [GeneID=3682] from brain Genes up-regulated in Vd1 gamma delta T cells: LPS versus phorbol myristate acetate and ionomycin [PubChem=4792;3733] Genes down-regulated in comparison of dendritic cells (DC) stimulated with Gardiquimod (TLR7 agonist) at 4 h versus those stimulated with Gardiquimod (TLR7 agonist) at 24 h Genes up-regulated in bone marrow-derived macrophagesat 180 min stimulation by LPS: IL6 [GeneID=3469] knockout versus IL10 [GeneID=3486] knockout Genes up-regulated in pro-B cells versus RAG2 [GeneID=5897] knockout NK cells Genes down-regulated in liver: untreated versus IL6 [GeneID=3569] injection Genes down-regulated in comparison of dendritic cells (DC) stimulated with LPS (TLR4 agonist) at 0.5 h versus those stimulated at 24 h Genes down-regulated in comparison of unstimulated macrophage cells versus macrophage cells stimulated with LPS (TLR4 agonist) for 20 min BIOCARTA Complement Pathway Genes down-regulated in double positive thymocytes from OT-2 transgenic mice: control versus injected with agonist peptide Genes down-regulated in CD4 [GeneID=920] Th1 cells with EGR2 [GeneID=1959] knockout: control versus anergic Genes down-regulated in comparison of control dendritic cells (DC) at 2 h versus those stimulated with Gardiquimod (TLR7 agonist) at 2 h Genes up-regulated during primary acute viral infection in dendritic cells with IFNAR1 [GeneID=3454] knockout: CD8A [GeneID=925] versus ITGAM+ [GeneID=3684] Genes down-regulated in eector CD8 T cells at the peak expansion phase (day 8 after LCMV-Armstrong infection) compared to memory CD8 T cells (day 40+ after LCMV-Armstrong infection) Genes up-regulated in Th1 CD4 [GeneID=920] SMARTA T cells: eector during acute infection of LCMV versus memory

Normal Enrichment Score

PBMC Genes up-regulated in CD8 T cells: ID3 [GeneID=3399] knockout versus wildtype C Genes down-regulated in B lymphocytes treated by anti IgM for 3h: wildtype versus MAP3K7 [GeneID=6885] knockout Genes down-regulated in bone marrow-derived macrophages treated with IL4 [GeneID=3565]: wildtype versus PPARG [GeneID=5468] knockout Genes down-regulated in bone marrow-derived macrophages with STAT6 [GeneID=6778] knockout treated with rosiglitazone [PubChem=77999]: control versus IL4 [GeneID=3565] Genes up-regulated in comparison of dendritic cells (DC) stimulated with CpG DNA (TLR9 agonist) at 0.5 h versus those stimulated with CpG DNA (TLR9 agonist) at 24 h Genes up-regulated in liver from SOCS3 [GeneID=9021] knockout: untreated versus IL6 [GeneID=3569] injection Genes up-regulated in comparison of unstimulated CD8 T cells at 72 h versus CD8 T cells at 72 h after stimulation with IL12 Genes down-regulated in comparison of dendritic cells (DC) exposed to L. donovani versus DCs exposed to M. tuberculosis Genes up-regulated in macrophages dierentiated for 5 days in the presence of: IL4 [GeneID=3565] versus IL4 [GeneID=3565] and dexamethasone [PubChem=5743] Genes down-regulated in comparison of polysome bound (translated) mRNA before and 16 h after LPS (TLR4 agonist) stimulation Genes down-regulated in comparison of unstimulated macrophage cells versus macrophage cells stimulated with LPS (TLR4 agonist) for 240 min Genes up-regulated in comparison of adult DN2 thymocytes versus fetal DN2 thymocytes Genes down-regulated in the immature neuron cell line: control versus infected with western equine encephalitis viruss Genes up-regulated in the activated CD4 [GeneID=920] T cells (48h): control versus IL-12 Genes down-regulated in B lymphocytes: naïve versus B1 REACTOME FCGR activation BIOCARTA Transcription factor CREB and its extracellular signals REACTOME ERK/MAPK targets REACTOME CREB phosphorylation BIOCARTA T Cell Receptor Signaling Pathway REACTOME TCR signaling Genes down-regulated in untreated double positive thymocytes: wildtype versus ELK1 and ELK4 [GeneID=2002;2005] knockout PID IL12-mediated signaling events Genes up-regulated in thymic: dendritic cells versus macrophages PID TCR signaling in naïve CD4+ T cells Genes down-regulated in thymocytes: cortical versus medullary sources Genes down-regulated in macrophages with IL10 [GeneID=3586] knockout treated by LPS and IL10 [GeneID=3586]: 10 min versus 30 min Genes down-regulated in the activated CD4 [GeneID=920] T cells (48h): IL-12 versus interferon alpha Genes down-regulated in comparison of control dendritic cells (DC) at 24 h versus those stimulated with Gardiquimod (TLR7 agonist) at 24 h Genes up-regulated in comparison of IgD+ B cells versus IgD- B cells

Normal Enrichment Score D 100% 90%

80% Neutrophils Eosinophils 70% Mast cells activated Mast cells resting Dendritic cells activated 60% Dendritic cells resting Macrophages M2 50% Macrophages M1 Macrophages M0 Monocytes 40% NK cells activated NK cells resting T cells gamma delta Relative Percentage Relative 30% T cells regulatory (Tregs) T cells follicular helper T cells CD4 memory activated 20% T cells CD4 memory resting T cells CD4 naive 10% T cells CD8 Plasma cells B cells memory 0% B cells naive HRR057173 HRR057174 HRR057175 HRR057176 HRR057177 HRR057178 HRR057179 HRR057180 HRR057181 HRR057182 HRR057183 HRR057184 HRR057185 HRR057186 HRR057187 HRR057188 HRR057189 HRR057190 HRR057191 HRR057192 CRR11894 CRR11895 CRR11896 CRR11897 HRR057165 HRR057166 HRR057167 HRR057168 HRR057169 HRR057170 HRR057171 HRR057172 SRR11517733 SRR11517734 SRR11517735 SRR11517736 SRR11517737 SRR11517738 SRR11517739 SRR11517740 SRR11517725 SRR11517726 SRR11517727 SRR11517728 SRR11517729 SRR11517730 SRR11517731 SRR11517732 Normal COVID-19 COVID-19 Normal E Lung BALF 100%

90%

80% Neutrophils Eosinophils 70% Mast cells activated Mast cells resting Dendritic cells activated 60% Dendritic cells resting Macrophages M2 Macrophages M1 50% Macrophages M0 Monocytes 40% NK cells activated NK cells resting

Relative Percentage Relative T cells gamma delta 30% T cells regulatory (Tregs) T cells follicular helper T cells CD4 memory activated 20% T cells CD4 memory resting T cells CD4 naive T cells CD8 10% Plasma cells B cells memory 0% B cells naive GSM4615003 GSM4615006 GSM4615008 GSM4615011 GSM4615014 GSM4615016 GSM4615019 GSM4615022 GSM4615025 GSM4615027 GSM4615030 GSM4615032 GSM4615033 GSM4615034 GSM4615035 GSM4615036 GSM4615037 GSM4614985 GSM4614986 GSM4614987 GSM4614988 GSM4614989 GSM4614990 GSM4614991 GSM4614992 GSM4614993 GSM4614994 GSM4614995 GSM4614996 GSM4614997 GSM4614998 GSM4614999 GSM4615000 GSM4615001

COVID-19 PBMC Normal

Figure 3. Immune landscape background. Bar plots showing nominal enrichment score for GSEA pathways signiﬁcantly correlated to COVID-19 status for (A) BALF, (B) lung biopsy, and (C) PBMC samples. Stacked bar plots showing relative immune cell abundances for (D) BALF, lung biopsy, and (E) PBMC samples. Cells 2021, 10, x FOR PEER REVIEW 8 of 18 Cells 2021, 10, 1452 8 of 18

Figure 4. Correlation to disease severity. Boxplots of microbes signiﬁcantly correlated to disease severity in PBMC samples. AllFigure boxplots 4. Correlation were produced to disease using severity. the Kruskal-Wallis Boxplots of test.microbes significantly correlated to disease severity in PBMC samples. All boxplots were produced using the Kruskal-Wallis test.

Cells 2021, 10, 1452 9 of 18

3.4. Microbes from PBMC Samples Correlate to Immune Infiltration and to Dysregulation of Immunological Signatures and Canonical Pathways Across PBMC samples, all microbes correlated to disease severity also correlated to immune dysregulation and pathways involved in inflammation. Bacillus subtilis subsp. subtilis str. 168 was correlated to upregulation of TCR signaling, the ERK pathway, and the antigen dependent B cell activation pathway; Bacillus sp. PL-12, Campylobacter hominis ATCC BAA-381, Thermoanaerobacterium thermosaccharolyticum DSM 571, and Thermoaner- obacter pseudethanolicus ATCC 33223 all correlated to upregulation of pathways involved in increasing inflammation. Campylobacter hominis ATCC BAA-381 and Thermoanaerobac- terium thermosaccharolyticum DSM 571 both correlated to upregulation of retinol metabolism and retinoic acid biosynthesis, which both play key roles in regulating antiviral immune response [22,23]. Staphylococcus epidermis correlated to dysregulation of various immunological signatures but interestingly not to any interleukin signaling pathways nor any immune cell abundances. Gene pathways for immunological signatures each correlated to the greatest number of dysregulated microbes in PBMC samples are visualized (Figure5). Bacillus sp. PL-12 and Thermoanerobacter pseudethanolicus ATCC 33223 both correlated to upregulation of the aurora A pathway and Thermoanerobacter pseudethanolicus ATCC 33223 correlated to upregulation of the KEGG Asthma pathway (Figure6A,B).

3.5. Microbes from BALF Samples Correlate to Immune Inﬁltration and to Dysregulation of Immunological Signatures and Canonical Pathways Across BALF samples, panels of GSEA and immune cell abundance correlated microbes were discovered. The most notable results include that Saccharomyces cerevisiae YJM1444, Syncephalastrum monosporum var. Pluriproliferum, Candida parapsilosis, and Histo- plasma capsulatum correlated to upregulation of TCR signaling, inﬂammatory interleukin signaling, TGFβ signaling, interferon IFN α and IFNβ signaling, RIG-I-like receptor signaling, and to dysregulation of immunological signatures (Figure6A,B).

3.6. Microbes from Lung Biopsy Samples Correlate to Immune Infiltration and to Dysregulation of Immunological Signatures and Canonical Pathways Across lung biopsy samples, panels of GSEA and immune cell abundance correlated microbes were identified. The most notable results include that Campylobacter ureolyticus, fungal sp. JF54, Ochrobactrum anthropi, and uncultured beta proteobacterium correlated to upregulation of TCR activation, IFNα signaling, activation of the complement system, tumor necrosis factor signaling, the P38MAPK pathway, the Toll like receptor (TLR) TLR1:TLR2 cascade, and to upregulation of inflammatory interleukin signaling pathways. Addition- ally, Ochrobactrum anthropi and uncultured beta proteobacterium were both found inversely correlated to gamma-delta T-cell abundance. Streptococcus sanguinis SK1 = NCTC 7863 also correlated to upregulation of inflammatory interleukin signaling, T cell receptor signaling, and TLR signaling, and it correlated down regulation of IL-4 signaling. Other Streptococcus species did not correlate to interleukin signaling or inflammatory pathways (Figure6A,B).

3.7. Negligible Contaminants Found in Differentially Abundant Microbes Scatter plots using Spearman’s correlation to correlate individual microbe abundance to total microbe reads in each patient. Normal and COVID-19 samples were correlated separately. Scatter plots that display a vertical or near vertical regression line deemed the microbe a contaminant, and no GSEA and immune cell correlated microbes were found to be contaminants (Figure7). Scatter plots for BALF COVID-19 samples (Figure S1A), BALF normal samples (Figure S1B), lung biopsy COVID-19 samples (Figure S1C), and lung biopsy normal samples (Figure S1D). Cells 2021, 10, 1452 10 of 18 Cells 2021, 10, x FOR PEER REVIEW 9 of 18

FigureFigure 5. Select 5. Select gene gene networks. networks. Gene Gene networks networks of GSEA of GSEA pathways pathways each correlated each correlated to three microbes to three with microbes boxplots with of box- microbe abundance vs. IA gene expression for microbes significantly correlated to IA gene expression for genes found to plots ofbe enriched microbe within abundance the respective vs. IA gene gene network. expression The gene for networks microbes were signiﬁcantly produced usin correlatedg the Cytoscape to IA Reactome gene expression FI for genes found to be enriched within the respective gene network. The gene networks were produced using the Cytoscape Reactome FI software. All boxplots were produced using the Kruskal-Wallis test. The networks are A B for the immunological signatures ( ) GSE17974_CTRL_VS_ACT_IL4_AND_ANTI_IL12_72H_CD4_TCELL_DN, ( ) GSE4590_SMALL_VS_LARGE_PRE_BCELL_UP, and (C) GSE3565_DUSP1_VS_WT_SPLENOCYTES_UP. Cells 2021, 10, 1452 11 of 18 Cells 2021, 10, x FOR PEER REVIEW 11 of 18

Figure 6. Microbe correlation to GSEA pathways and immune cells. Volcano plots of (A) microbe abundance correlated to Figure 6. Microbe correlation to GSEA pathways and immune cells. Volcano plots of (A) microbe abundance correlated to GSEA pathways with the most significant pathways indicated and of (B) microbe abundance correlated to immune cell GSEA pathways with the most signiﬁcant pathways indicated and of (B) microbe abundance correlated to immune cell abundances with the most signiﬁcant microbe immune cell pairs indicated. Labeled data points correspond to those with cyan outlines. Cells 2021, 10, x FOR PEER REVIEW 12 of 18

Cells 2021, 10, 1452 12 of 18 abundances with the most significant microbe immune cell pairs indicated. Labeled data points correspond to those with cyan outlines.

Figure 7. Contamination correction. Spearman’s correlation for select disease severity associated microbes. Scatterplots showing microbe abundance vs. total microbial reads for the most significantly differentially abundant microbes from (A) showing microbe abundance vs. total microbial reads for the most signiﬁcantly differentially abundant microbes from (A) PBMC COVID-19 samples and (B) PBMC normal samples. PBMC COVID-19 samples and (B) PBMC normal samples.

4. Discussion Few studies have investigated the lung microbiome in COVID-19 patients, and no studies to date have investigated the blood microbiome in COVID-19 patients. In this study, we identified individual bacterial and fungal sequences via high-throughput RNA sequencing differentially abundant between COVID-19 and normal patients. Despite no explicit contamination correction for all microbes identified, we are confident that our findings of immune cell abundance and GSEA pathway correlated dysregulated microbes are authentic due to no contaminants found through our Spearman’s Correlation results. We identified 91 bacteria and 14 fungus differentially abundant in lung biopsy, 13 bacteria and 9 fungus differentially abundant in PBMC, and 12 bacteria and 57 fungus differen-

Cells 2021, 10, 1452 13 of 18

tially abundant in BALF. We demonstrate significant associations between many bacterial and fungal species from the lungs and blood and COVID-19, suggesting that the lung and blood microbiota could play a role in modulating host immune response and potentially influence disease severity and outcomes. Specifically, the depletion or enrichment of above listed key bacterial and fungal species in COVID-19 patient samples were associated with downregulation of the anti-inflammatory IL-4 signaling and associated with upregulation of IL-2, IL-3, IL-5, IL-6, IL-7, IL-10, IL-20, IL-22 and IL-23 signaling, IFNα signaling pathways, TGFβ signaling pathways, inflammatory pathways, and immune dysregulation. As shown in a previous study, “shifts in cytokine profiles mediated by changes in the microbiota may also promote epithelial injury and fibrotic outcomes” in chronic lung disease and in COVID-19, and similar associations were observed in our study [24,25]. Across lung biopsy samples, Campylobacter ureolyticus, fungal sp. JF54, Ochrobactrum anthropi, and uncultured beta proteobacterium correlated to upregulation of TCR activation, IFNα signaling, activation of the complement system, tumor necrosis factor signaling, and several inflammatory interleukin signaling pathways. Burgos-Portugal et al. demonstrated that IFNα causes cells to produce significantly greater amounts of IL-8 and thus inflammation, and cells infected with Campylobacter ureolyticus produced even higher levels of IL-8 and thus more inflammation [26]. Ochrobactrum anthropi is emerging as an opportunistic pathogen that causes infections in severely ill or immunocompromised patients [27] and has been reported as the cause of diverse inflammatory ailments including infective endocarditis [28,29], endophthalmitis [30], meningitis [31], and osteomyelitis [32]. This previous research lends support to our findings that these microbes may be implicated in modulating the immune response to COVID-19 producing a pro-inflammatory environment and thus worse outcomes for COVID-19 patients. Fungal sp. JF54 and uncultured beta proteobacterium, while not thoroughly studied before, had similar associations as Ochrobactrum anthropi and Campylobacter ureolyticus. Ochrobactrum anthropi and uncultured beta proteobacterium additionally correlated to decreased gamma delta T cell abundance. Gamma delta T cells play an important role in immunosurveillance in mucosal and epithelial barriers in the lungs and have been demonstrated to play a critical protective role in response to both SARS-CoV-1 and SARS-CoV-2 [33]. This may be another potential mechanism by which opportunistically pathogenic microbiota modulate the response to COVID-19, by disrupting the balance between a sufficient antiviral and hyper-inflammatory response by decreasing the appropriate antiviral function of gamma delta T cells. Furthermore, Streptococcus sanguinis has previously been found to be opportunistically pathogenic when the host already suffers certain inflammatory ailments [34]. This lends support to our findings that Strepto- coccus sanguinis SK1 = NCTC 7863 may influence the immune response to COVID-19 by upregulating inflammatory IL-6 signaling and downregulating IL-4 signaling. Interestingly, all other differentially abundant Streptococcus species did not have significant correlations to inflammatory pathways, which does not support the findings of other studies that have demonstrated Streptococcus species correlating to COVID-19 severity [35]. Across BALF samples, Saccharomyces cerevisiae YJM1444, Syncephalastrum monosporum var. Pluriproliferum, Candida parapsilosis, and Histoplasma capsulatum were correlated to upregulation of TGFβ signaling pathways and inflammatory interleukin signaling pathways. These four microbes also correlated to increased M1 macrophage abundance. Inciden- tally, other microbes had similar associations but have been previously shown to be only pathogenic to plants. It has previously been shown that in severe COVID-19, SARS-CoV-2 triggers a chronic immune reaction instructed by TGFβ, which contributes to chronic upregulation of IL-6 signaling and B cell activation leading to worse COVID-19 outcomes [36]. By modulating inflammatory interleukin pathways and TGFβ signaling pathways, the above microbes may promote a pro-inflammatory response to COVID-19 worsening patient outcomes. Syncephalastrum monosporum var. Pluriproliferum also correlated to the local acute inflammatory response pathway. These particular species of Syncephalastrum and Saccharomyces have not been thoroughly studied, other species have been found implicated in other pulmonary ailments and inflammation in immunocompromised patients and are Cells 2021, 10, 1452 14 of 18

emerging as opportunistic pathogens. Histoplasma capsulatum and Candida parapsilosis are also emerging as opportunistic pathogens, as demonstrated in past studies [37–40]. The above four microbes in addition to Tremella mesenterica DSM 1558 correlated to upregulation of RIG-I-like receptors, and RIG-I-like receptors detect viral RNA and mediate the host antiviral response by inducing IFN 1 transcription [41]. Additionally, these microbes may contribute to a hyperinflammatory response by overactivation of the inflammatory M1 macrophages and fewer macrophages differentiating to the anti-inflammatory M2 macrophage mediated by upregulation of inflammatory cytokine signaling and downregulation of IL-4 signaling, as excessive M1-polarized immune responses has been shown to lead to tissue damage in inflammatory diseases [42]. Interestingly, Fungal sp. 57 was found inversely correlated with local acute inflammatory pathways and inflammatory interleukin signaling and may play a beneficial role in immunomodulation in COVID-19. Previous study has found that SARS-CoV-2 allows anaerobic bacteria to colonize the lungs and consequently disrupt lung homeostasis [43]. We found four Haemophilus influenzae species, two Bacteroides species, and Chlorobium phaeobacteroides BS1, all being anaerobic bacteria, to have greater abundance in COVID-19 samples in BALF tissue; however, the majority of our data does not corroborate the previous finding that anaerobic bacteria colonize the COVID-19 effected lung, as we found that the majority of bacterial abundance in both lung biopsy tissues and BALF tissues are facultative anaerobic bacteria in both normal samples and COVID-19 samples, and the anaerobic species Veillonella parvula DSM 2008 and Haemophilus parainfluenzae T3T1 abundance was greater in normal samples in lung biopsy tissue. Other studies characterizing the COVID-19 respiratory tract microbiomes have found that Streptococcus and Veillonella dominate the upper respiratory tract and Streptococcus and Haemophilus abundance are the most important features for segregating COVID-19 clinical outcomes [35,44]. Our findings do not corroborate that Streptococcus and Haemophilus are most associated with COVID-19 severity nor do Streptococcus or Haemophilus dominate the lung microbiome in our study. This may provide evidence of some difference between the upper respiratory tract microbiome and the lung microbiome or that the particular identity of commensal microbes populating the lung microbiome may be less important than the severity of differential abundance of commensal microbes. This conclusion may be supported by other studies finding decreased microbiota alpha-diversity in the lung of COVID-19 patients [45]. Our study does corroborate the finding of dysbiosis of commensal microbes and increased abundance of opportunistic pathogens. Across PBMC samples, Campylobacter hominis ATCC BAA-381 and Thermoanaerobac- terium thermosaccharolyticum DSM 571 both correlated to upregulation of retinol metabolism and retinoic acid biosynthesis, which both play key roles in regulating antiviral immune response [22,23]. Dysregulation in metabolism and transport of retinol may contribute to inflammatory responses to infection. Furthermore, this dysregulation of retinol metabolism and retinoic acid biosynthesis may play a role in the upregulation of RIG-I-like receptors that was observed in the BALF samples, as retinoids induce RIG-I expression in concert with IFN signaling to lead to the secretion of various pro-inflammatory cytokines [46]. Bacillus sp. PL-12 abundance correlated to decreased plasma cell abundance and correlated to upregulation of IFNα signaling, inflammatory interleukin signaling, and TGFβ signaling, whereas Bacillus subtilis subsp. subtilis str. 168 inversely correlated to IFNα signaling and inflammatory interleukin signaling, which may explain their correlations to disease severity. Additionally, Bacillus subtilis subsp. subtilis str. 168 correlated to increased plasma cell abundance and memory B cell abundance and microbes correlated with COVID-19 severity, namely E. coli, Pseudomonas sp. I-09, and Bacillus sp. PL-12 correlated to decreased plasma cell abundance and memory B cell abundance. Previous study has found an association between plasma cells in the blood in patients with severe COVID-19 and improved survival [47]. In corroboration with all of our methods and correlations involving Bacillus subtilis subsp. subtilis str. 168, in addition to the strength and linearity of correlation in our Spearman’s Correlation results make us believe future study of this microbe in particular may potentially be useful in immunotherapy for treating COVID-19. Campylobacter hominis Cells 2021, 10, 1452 15 of 18

ATCC BAA-381 and Pseudomonas sp. I-09 have not been previously thoroughly studied; however other species of Campylobacter hominis and other Pseudomonas have been found to induce inflammation in other diseases such as gastroenteritis lending support to our findings that they may modulate the host immune response to COVID-19 [48–51]. How- ever, it remains unknown if the above inflammatory-associated lung and blood bacterial and fungal species enriched in COVID-19 do in fact play an active role in the prognosis of COVID-19 or simply flourish opportunistically due to a depletion of other lung and blood microbes, but our correlations to immune dysregulation suggest they do in fact play a role in the prognosis of COVID-19. Our study has several limitations. First, our study used a fairly small sample size for each sample type with 17 PBMC COVID-19 samples and 17 PBMC normal samples, 12 BALF COVID-19 samples and 20 BALF normal samples, and 8 lung biopsy COVID-19 samples and 8 lung normal samples. While enough samples were used to identify nu- merous dysregulated microbes, further studies with greater sample sizes could be used to increase statistical power and confirm our results. Secondly, metadata of age and disease severity were only available for the PBMC samples used. Further studies correlating microbe abundance to COVID-19 severity using BALF samples and/or lung biopsy samples to identify microbes of the lung microbiome directly correlated with COVID-19 severity should be performed. Thirdly, our study used samples sourced from three different hospi- tals, and as the human microbiome is highly impacted by geography, our findings may not necessarily reflect the lung and blood microbiome of COVID-19 patients from different geographies. Regardless of these limitations, our study demonstrating associations between microbes found in the lung and blood to immune dysregulation pathways and our associations between microbes in the blood to COVID-19 severity suggests that the lung and blood microbiome are likely implicated in modulating host inflammatory responses to COVID-19 and thus COVID-19 severity.

Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/cells10061452/s1, Figure S1: The non-contaminant microbes’ plots in lung biopsy and BALF samples. Scatterplots showing microbe abundance vs. total microbial reads for GSEA and immune cell abundance correlated microbes from (A) BALF COVID-19 samples, (B) BALF normal samples, (C) lung biopsy COVID-19 samples, and (D) lung biopsy normal samples. Author Contributions: Conceptualization, W.M.O.; methodology, K.D., J.C., W.T.L. and W.M.O.; software, N/A; validation, N/A; formal analysis, K.D., L.A. and I.R.; investigation, W.M.O.; resources, W.M.O.; data curation, N/A; writing—original draft preparation, K.D.; writing—review and editing, K.D., J.C., W.T.L., M.R., E.Y.C. and W.M.O.; visualization, K.D., L.A. and I.R.; supervision, W.M.O.; project administration, W.M.O.; funding acquisition, W.M.O. All authors have read and agreed to the published version of the manuscript. Funding: This research was funded by the University of California, Office of the President/Tobacco- Related Disease Research Program Emergency COVID-19 Research Seed Funding Grant (Grant number: R00RG2369) to W.M.O. and the University of California, San Diego Academic Senate Grant (Grant number: RG096651) to W.M.O. Institutional Review Board Statement: Not applicable. Informed Consent Statement: Not applicable. Data Availability Statement: RNA sequencing data for PBMC samples are available at Gene Ex- pression Omnibus (accession code GSE152418). RNA sequencing data for lung biopsy samples are available at Gene Expression Omnibus (accession code GSE147507). RNA sequencing data for BALF samples are available at Genome Sequence Archive (accession code HRA000143 and CRA002390). Conflicts of Interest: The authors declare no conflict of interest. Cells 2021, 10, 1452 16 of 18

References 1. COVID-19 Dashboard by the Center for Systems Science and Engineering (CSSE) at Johns Hopkins University. Available online: https://gisanddata.maps.arcgis.com/apps/opsdashboard/index.html#/bda7594740fd40299423467b48e9ecf6 (accessed on 19 March 2021). 2. Berlin, D.A.; Gulick, R.M.; Martinez, F.J. Severe Covid-19. N. Engl. J. Med. 2020, 383, 2451–2460. [CrossRef][PubMed] 3. Interim Clinical Guidance for Management of Patients with Confirmed Coronavirus Disease (COVID-19). Available online: https://www.cdc.gov/coronavirus/2019-ncov/hcp/clinical-guidance-management-patients.html (accessed on 19 March 2021). 4. Byambasuren, O.; Cardona, M.; Bell, K.; Clark, J.; McLaws, M.L.; Glasziou, P. Estimating the extent of asymptomatic COVID-19 and its potential for community transmission: Systematic review and meta-analysis. J. Assoc. Med. Microbiol. Infect. Dis. Can. 2020, 5, 223–234. [CrossRef] 5. Greenhalgh, T.; Knight, M.; Buxton, M.; Husain, L. Management of post-acute covid-19 in primary care. BMJ 2020, 370, 3026. [CrossRef] 6. Siordia, J.A., Jr. Epidemiology and clinical features of COVID-19: A review of current literature. J. Clin. Virol. 2020, 127, 104357. [CrossRef][PubMed] 7. Ruan, Q.; Yang, K.; Wang, W.; Jiang, L.; Song, J. Clinical predictors of mortality due to COVID-19 based on an analysis of data of 150 patients from Wuhan, China. Intensive Care Med. 2020, 46, 846–848. [CrossRef] 8. Coperchini, F.; Chiovato, L.; Croce, L.; Magri, F.; Rotondi, M. The cytokine storm in COVID-19: An overview of the involvement of the chemokine/chemokine-receptor system. Cytokine Growth Factor Rev. 2020, 53, 25–32. [CrossRef][PubMed] 9. Khatiwada, S.; Subedi, A. Lung microbiome and coronavirus disease 2019 (COVID-19): Possible link and implications. Hum. Microb. J. 2020, 17, 100073. [CrossRef] 10. Yeoh, Y.K.; Zuo, T.; Lui, G.C.; Zhang, F.; Liu, Q.; Li, A.Y.; Chung, A.C.; Cheung, C.P.; Tso, E.Y.; Fung, K.S.; et al. Gut microbiota composition reflects disease severity and dysfunctional immune responses in patients with COVID-19. Gut 2021, 70, 698–706. [CrossRef] 11. Dickson, R.P.; Schultz, M.J.; van der Poll, T.; Schouten, L.R.; Falkowski, N.R.; Luth, J.E.; Sjoding, M.W.; Brown, C.A.; Chanderraj, R.; Huffnagle, G.B.; et al. Biomarker Analysis in septic ICU patients (BASIC) consortium. Lung microbiota predict clinical outcomes in critically ill patients. Am. J. Respir. Crit. Care Med. 2020, 201, 555–563. [CrossRef] 12. Sommariva, M.; Le Noci, V.; Bianchi, F.; Camelliti, S.; Balsari, A.; Tagliabue, E.; Sfondrini, L. The lung microbiota: Role in maintaining pulmonary immune homeostasis and its implications in cancer development and therapy. Cell. Mol. Life Sci. 2020, 77, 2739–2749. [CrossRef] 13. Fan, J.; Li, X.; Gao, Y.; Zhou, J.; Wang, S.; Huang, B.; Wu, J.; Cao, Q.; Chen, Y.; Wang, Z.; et al. The lung tissue microbiota features of 20 deceased patients with COVID-19. J. Infect. 2020, 81, e64–e67. [CrossRef] 14. Yuki, K.; Fujiogi, M.; Koutsogiannaki, S. COVID-19 pathophysiology: A review. Clin. Immunol. 2020, 215, 108427. [CrossRef] 15. Arunachalam, P.S.; Wimmers, F.; Mok, C.K.P.; Perera, R.A.; Scott, M.; Hagan, T.; Sigal, N.; Feng, Y.; Bristow, L.; Tsang, O.T.Y.; et al. Systems biological assessment of immunity to mild versus severe COVID-19 infection in humans. Science 2020, 369, 1210–1220. [CrossRef][PubMed] 16. Hong, C.; Manimaran, S.; Shen, Y.; Perez-Rogers, J.F.; Byrd, A.L.; Castro-Nallar, E.; Crandall, K.A.; Johnson, W.E. PathoScope 2.0: A complete computational framework for strain identification in environmental or clinical sequencing samples. Microbiome 2014, 2, 33. [CrossRef] 17. Johnson, W.E.; Li, C.; Rabinovic, A. Adjusting batch effects in microarray expression data using empirical Bayes methods. Biostatistics 2007, 8, 118–127. [CrossRef] 18. Dobin, A.; Davis, C.A.; Schlesinger, F.; Drenkow, J.; Zaleski, C.; Jha, S.; Batut, P.; Chaisson, M.; Gingeras, T.R. STAR: Ultrafast universal RNA-seq aligner. Bioinformatics 2013, 29, 15–21. [CrossRef] 19. Newman, A.M.; Steen, C.B.; Liu, C.L.; Gentles, A.J.; Chaudhuri, A.A.; Scherer, F.; Khodadoust, M.S.; Esfahani, M.S.; Luca, B.A.; Steiner, D.; et al. Determining cell type abundance and expression from bulk tissues with digital cytometry. Nat. Biotechnol. 2019, 37, 773–782. [CrossRef] 20. Subramanian, A.; Tamayo, P.; Mootha, V.K.; Mukherjee, S.; Ebert, B.L.; Gillette, M.A.; Paulovich, A.; Pomeroy, S.L.; Golub, T.R.; Lander, E.S.; et al. Gene set enrichment analysis: A knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl. Acad. Sci. USA 2005, 102, 15545–15550. [CrossRef] 21. Shannon, P.; Markiel, A.; Ozier, O.; Baliga, N.S.; Wang, J.T.; Ramage, D.; Amin, N.; Schwikowski, B.; Ideker, T. Cytoscape: A software environment for integrated models of biomolecular interaction networks. Genome Res. 2003, 13, 2498–2504. [CrossRef] 22. Ross, A.C.; Stephensen, C.B. Vitamin A and retinoids in antiviral responses. FASEB J. 1996, 10, 979–985. [CrossRef][PubMed] 23. Pino-Lagos, K.; Guo, Y.; Noelle, R.J. Retinoic acid: A key player in immunity. BioFactors 2010, 36, 430–436. [CrossRef] 24. O’Dwyer, D.N.; Dickson, R.P.; Moore, B.B. The lung microbiome, immunity, and the pathogenesis of chronic lung disease. J. Immunol. 2016, 196, 4839–4847. [CrossRef][PubMed] 25. Leng, L.; Cao, R.; Ma, J.; Mou, D.; Zhu, Y.; Li, W.; Lv, L.; Gao, D.; Zhang, S.; Gong, F.; et al. Pathological features of COVID- 19-associated lung injury: A preliminary proteomics report based on clinical samples. Sig. Transduct. Target. Ther. 2020, 5, 240. [CrossRef][PubMed] Cells 2021, 10, 1452 17 of 18

26. Burgos-Portugal, J.A.; Kaakoush, N.O.; Raftery, M.J.; Mitchell, H.M. Pathogenic potential of Campylobacter ureolyticus. Infect. Immun. 2012, 80, 883–890. [CrossRef] 27. Aguilera-Arreola, M.G.; Ostria-Hernández, M.L.; Albarrán-Fernández, E.; Juárez-Enriquez, S.R.; Majalca-Martínez, C.; Rico- Verdín, B.; Ruiz, E.A.; Ruiz-Palma, M.; Morales-García, M.R.; Contreras-Rodríguez, A. Correct identification of ochrobactrum anthropi from blood culture using 16rRNA sequencing: A first case report in an immunocompromised patient in Mexico. Front. Med. 2018, 5, 205. [CrossRef] 28. Romero Gómez, M.P.; Peinado Esteban, A.M.; Sobrino Daza, J.A.; Sáez Nieto, J.A.; Alvarez, D.; Peña García, P. Prosthetic mitral valve endocarditis due to Ochrobactrum anthropi: Case report. J. Clin. Microbiol. 2004, 42, 3371–3373. [CrossRef] 29. Mahmood, M.S.; Sarwari, A.R.; Khan, M.A.; Sophie, Z.; Khan, E.; Sami, S. Infective endocarditis and septic embolization with Ochrobactrum anthropi: Case report and review of literature. J. Infect. 2000, 40, 287–290. [CrossRef] 30. Braun, M.; Jonas, J.B.; Schönherr, U.; Naumann, G.O. Ochrobactrum anthropi endophthalmitis after uncomplicated cataract surgery. Am. J. Ophthalmol. 1996, 122, 272–273. [CrossRef] 31. Chang, H.J.; Christenson, J.C.; Pavia, A.T.; Bobrin, B.D.; Bland, L.A.; Carson, L.A.; Arduino, M.J.; Verma, P.; Aguero, S.M.; Carroll, K.; et al. Ochrobactrum anthropi meningitis in pediatric pericardial allograft transplant recipients. J. Infect. Dis. 1996, 173, 656–660. [CrossRef] 32. Jelveh, N.; Cunha, B.A. Ochrobactrum anthropi bacteremia. Heart Lung 1999, 28, 145–146. [CrossRef] 33. Yazdanifar, M.; Mashkour, N.; Bertaina, A. Making a case for using γδ T cells against SARS-CoV-2. Crit. Rev. Microbiol. 2020, 46, 689–702. [CrossRef] 34. Martini, A.M.; Moricz, B.S.; Ripperger, A.K.; Tran, P.M.; Sharp, M.E.; Forsythe, A.N.; Kulhankova, K.; Salgado-Pabón, W.; Jones, B.D. Association of novel Streptococcus sanguinis virulence factors with pathogenesis in a native valve infective endocarditis model. Front. Microbiol. 2020, 11, 10. [CrossRef] 35. De Castilhos, J.; Zamir, E.; Hippchen, T.; Rohrbach, R.; Schmidt, S.; Hengler, S.; Schumacher, H.; Neubauer, M.; Kunz, S.; Müller-Esch, T.; et al. COVID-19 severity and complications associated with low diversity, dysbiosis and predictive metagenome features of the oropharyngeal microbiome. Res. Sq. 2021.[CrossRef] 36. Ferreira-Gomes, M.; Kruglov, A.; Durek, P.; Heinrich, F.; Tizian, C.; Heinz, G.A.; Pascual-Reguant, A.; Du, W.; Mothes, R.; Fan, C.; et al. In severe COVID-19, SARS-CoV-2 induces a chronic, TGF-β-dominated adaptive immune response. medRxiv 2020. [CrossRef] 37. Irshad, M.; Nasir, N.; Hashmi, U.H.; Farooqi, J.; Mahmood, S.F. Invasive pulmonary infection by Syncephalastrum species: Two case reports and review of literature. IDCases 2020, 21, e00913. [CrossRef][PubMed] 38. Pérez-Torrado, R.; Querol, A. Opportunistic strains of saccharomyces cerevisiae: A potential risk sold in food products. Front. Microbiol. 2016, 6, 1522. [CrossRef] 39. Trofa, D.; Gácser, A.; Nosanchuk, J.D. Candida parapsilosis, an emerging fungal pathogen. Clin. Microbiol. Rev. 2008, 21, 606–625. [CrossRef] 40. Horwath, M.C.; Fecher, R.A.; Deepe, G.S., Jr. Histoplasma capsulatum, lung infection and immunity. Future. Microbiol. 2015, 10, 967–975. Available online: https://www.nature.com/articles/s41577-020-0288-3 (accessed on 1 January 2021). [CrossRef] [PubMed] 41. Rehwinkel, J.; Gack, M.U. RIG-I-like receptors: Their regulation and roles in RNA sensing. Nat. Rev. Immunol. 2020, 20, 537–551. [CrossRef] 42. Orekhov, A.N.; Orekhova, V.A.; Nikiforov, N.G.; Myasoedova, V.A.; Grechko, A.V.; Romanenko, E.B.; Zhang, D.; Chistiakov, D.A. Monocyte differentiation and macrophage polarization. Vessel Plus 2019, 3, 10. [CrossRef] 43. Chackraborty, S. SARS-Cov2 enables anaerobic bacteria (Prevotella, et al) to colonize the lungs disrupting homeostasis, causing long-drawn chronic symptoms, and acute severe symptoms (ARDS, septic shock, clots, arterial stroke) which finds resonance, with key differences, in the ‘forgotten disease’ Lemierre Syndrome, enabled by Epstein Barr Virus. OSF Prepr. 2020.[CrossRef] 44. Ren, L.; Wang, Y.; Zhong, J.; Zhang, D.; Xiao, Y.; Yang, J.; Fan, G.; Guo, L.; Shen, Z.; Liu, W.; et al. Dynamics of the upper respiratory tract microbiota and its association with fatality in COVID-19 patients. Res. Sq. 2020.[CrossRef] 45. Xu, R.; Lu, R.; Zhang, T.; Zhang, T.; Wu, Q.; Cai, W.; Han, X.; Wan, Z.; Jin, X.; Zhang, Z.; et al. Temporal association between human upper respiratory and gut bacterial microbiomes during the course of COVID-19 in adults. Commun. Biol. 2021, 4, 240. [CrossRef] 46. Soye, K.J.; Trottier, C.; Richardson, C.D.; Ward, B.J.; Miller, W.H., Jr. RIG-I is required for the inhibition of measles virus by retinoids. PLoS ONE 2011, 6, 22323. [CrossRef][PubMed] 47. Boulanger, M.; Molina, E.; Wang, K.; Kickler, T.; Xu, Y.; Garibaldi, B.T. Peripheral plasma cells associated with mortality benefit in severe COVID-19: A marker of disease resolution. Am. J. Med. 2021, 1, 40. [CrossRef] 48. Man, S.M. The clinical importance of emerging Campylobacter species. Nat. Rev. Gastroenterol. Hepatol. 2011, 8, 669. [CrossRef] 49. Markazi, A.; Bracci, P.M.; McGrath, M.; Gao, S.J. Pseudomonas aeruginosa stimulates inflammation and enhances Kaposi’s sarcoma herpesvirus-induced cell proliferation and cellular transformation through both lipopolysaccharide and flagellin. mBio 2020, 11, e02843-20. [CrossRef] Cells 2021, 10, 1452 18 of 18

50. Rajkumari, N.; John, N.V.; Mathur, P.; Misra, M.C. Antimicrobial resistance in Pseudomonas sp. causing infections in trauma patients: A 6 year experience from a South Asian country. J. Glob. Infect. Dis. 2014, 6, 182–185. [CrossRef][PubMed] 51. Lin, C.K.; Kazmierczak, B.I. Inﬂammation: A double-edged sword in the response to Pseudomonas aeruginosa infection. J. Innate Immun. 2017, 9, 250–261. [CrossRef]