key: cord-0728157-p5poj5tk
authors: Gurshaney, Sanjeev; Alvarez, Anamaria Morales; Ezhakunnel, Kevin; Manalo, Andrew; Huynh, Thien-Huong; Le, Nhat-Tu; Lupu, Daniel S.; Gardell, Stephen J.; Nguyen, Hung
title: Metabolic dysregulation induces impaired lymphocyte memory formation during severe SARS-CoV-2 infection
date: 2021-12-08
journal: bioRxiv
DOI: 10.1101/2021.12.06.471421
sha: 6da865e62420050a05bfb39bc44b15e4f65da8c1
doc_id: 728157
cord_uid: p5poj5tk

Cellular metabolic dysregulation is a consequence of COVID-19 infection that is a key determinant of disease severity. To understand the mechanisms underlying these cellular changes, we performed high-dimensional immune cell profiling of PBMCs from COVID-19-infected patients, in combination with single cell transcriptomic analysis of COVID-19 BALFs. Hypoxia, a hallmark of COVID-19 ARDS, was found to elicit a global metabolic reprogramming in effector lymphocytes. In response to oxygen and nutrient-deprived microenvironments, these cells shift from aerobic respiration to increase their dependence on anaerobic processes including glycolysis, mitophagy, and glutaminolysis to fulfill their bioenergetic demands. We also demonstrate metabolic dysregulation of ciliated lung epithelial cells is linked to significant increase of proinflammatory cytokine secretion and upregulation of HLA class 1 machinery. Augmented HLA class-1 antigen stimulation by epithelial cells leads to cellular exhaustion of metabolically dysregulated CD8 and NK cells, impairing their memory cell differentiation. Unsupervised clustering techniques revealed multiple distinct, differentially abundant CD8 and NK memory cell states that are marked by high glycolytic flux, mitochondrial dysfunction, and cellular exhaustion, further highlighting the connection between disrupted metabolism and impaired memory cell function in COVID-19. Our findings provide novel insight on how SARS-CoV-2 infection affects host immunometabolism and anti-viral response during COVID-19. Graphical Abstract Highlights Hypoxia and anaerobic glycolysis drive CD8, NK, NKT dysfunction Hypoxia and anaerobic glycolysis impair memory differentiation in CD8 and NK cells Hypoxia and anaerobic glycolysis cause mitochondrial dysfunction in CD8, NK, NKT cells

CoV-2 vaccines have been launched, herd immunity may not be reached in many countries until 79 late 2021 or early 2022 2 . Additionally, mutant COVID-19 strains including the novel delta variant, 80 which have the potential to partially evade immunity induced by currently available vaccines, as 81 well as display significantly increased rates of infection are rapidly increasing in prevalence 3 . 82

Therefore, novel therapeutics to combat SARS-CoV-2 are urgently needed. 83

Metabolic syndrome and its cluster of conditions pose risk factors for severe COVID-19 84 pathogenesis 4,5 . The growing body of evidence suggests that individuals with pre-existing 85 metabolic comorbidities are at far higher risk of suffering severe complications from COVID-19 86 COVID-19 patients, the primary target for virus infection, using publicly available single cell RNA-149 sequencing data from BALF samples 18 . Louvain optimization and UMAP dimensionality reduction 150 were applied to generate unsupervised clusters (Fig. 1G ) which were then annotated by 151 expression of canonical markers ( Supplementary Fig. 1A , B) to define highly resolved populations 152 (Fig. 1H) . Canonical markers used to annotate the unsupervised clusters and define cell subsets Fig. 1A-B) . 159

After initial cell type identification, T cells were subsetted and unsupervised clustering was 160 performed again to achieve higher resolution into distinct T cell lineages (Fig. 1I, J and  161 Supplementary Fig. 2A ). Abundance of CD8 memory and NKT cells was found reduced and 162 positively correlated to disease severity (Fig. 1K) . Taken together, these results suggest that 163 memory differentiation of CD8 + T and NK cells is impaired in both the lungs and in circulation of 164 COVID-19 patients. 165

Expression of cellular markers for metabolism and exhaustion was evaluated in CD8 T cells from Fig. 5D ). Decreased NADH oxidation and a concomitant increased NAD + level is 215 associated with impaired cytokine secretion, cell proliferation, and exhaustion 24,25 . Indeed, the 216 expression of CD38, an NAD + hydrolase linked to T cell exhaustion 25 , was increased in COVID-217

against viral re-infection 26 . To better understand the kinetics and dynamics of CD8 memory cell 226 differentiation during SARS-CoV-2 infection, we performed trajectory inference and 227 pseudotemporal modeling analysis on BALF CD8 T cells ( Fig. 2A) . Differential analysis revealed 228 decreased pseudotime values for CD8 memory in S compared to M-COVID-19 and healthy 229 control ( Figs. 2A, B) . The frequencies of proliferating CD8 and GNLY + effector CD8 T cells were 230 significantly increased in severe COVID-19 compared to healthy controls (Fig. 2C) . Moreover, 231 CD8 cells were found highly proliferating in M-than S-COVID-19 patients (Fig. 2C) . In contrast, 232 a lower abundance of GNLY + effector CD8 cells was found in S-COVID-19. These findings 233

suggest that CD8 T cells are stalled along the memory differentiation trajectory and are unable to 234 reach the terminal state during S-COVID-19 infection. 235

During viral infection, circulating effector memory cells migrate to the infected tissue and 236 differentiate into tissue-resident memory (TRM) cells to provide the first response defense against 237 reencounter of the pathogen 27 . Consistently, CD8 effector memory (CD8EM) cells expressed 238 higher tissue residence phenotype (i.e., increased expression of ITGA1 and ZNF683) in M-as 239 compared to S-COVID-19 patients (Fig. 2D) Fig. 3A ). GSEA analysis revealed that CD8M cells from S-or M-COVID-19 patients were highly dependent on glycolysis for their energetic demands (Figs. 3A, B) . Close 249 clustering and shared upregulation of glycolytic enzyme coding genes GALM, GAPDH, GPI, and 250 ALDOA with HIF1A, and transcripts regulating exhaustion (TIGIT and LAG3) (Fig. 3B) Pearson correlation analysis revealed a negative correlation (R = -0.73, p = 0.011) between 259 module scores for glycolysis and FAO (Fig. 3D) , which further showed a potential association 260 between prolonged anaerobic glycolysis and reduced mitochondrial fitness. Moreover, a strong 261 positive correlation between module scores for glycolysis and exhaustion (R = 0.85, p = 0.00026) 262 ( Fig. 3D) validates that excessive glycolytic dependence is likely tightly linked to CD8M 263 exhaustion. Genes involved in cellular senescence and mitophagy were also upregulated in S-264 COVID-19 CD8M cells (Fig. 3A) , implying that CD8M cells metabolically switch to these pathways 265 to satisfy bioenergetics demands in response to impaired mitochondrial metabolism. Likewise, 266 glutaminolysis is also used as an alternative bioenergetic pathway, evident by upregulation of 267 glutamate oxidation regulating genes GLUD1 and DGLUCY in CD8M cells during COVID-19, likely 268 in response to reduced lipid uptake and FAO (Fig. 3A) . 269 PCA analysis performed on 30 differentially expressed metabolic genes (Table S1) showed 270 distinct clustering of CD8M cells across different groups, further highlighting the potential of these 271 pathways to be used as predictive markers for disease severity (Fig. 3E) . In this line, Pearson 272 correlation analysis showed a strong positive correlation between expression of genes regulating of ROS + NKT cells in COVID-19 PBMCs ( Supplementary Fig. 8B ). Interestingly, these cells 299 displayed elevated HIF-1α expression ( Supplementary Fig. 8B ), which further implied that hypoxia 300 also triggers oxidative stress in NKT cells during COVID-19 infection. To further establish the 301 relationship between this altered metabolism and cellular function, a secondary unsupervised 302 clustering was performed. We identified a population of GLUT1 + mitochondrially exhausted NKT 303 cells, representative of combined augmented glycolytic phenotype, impaired mitochondrial 304 function, and cellular exhaustion, that were significantly increased in COVID-19 patients 305 ( Supplementary Fig. 8C ). Higher HIF-1α expression in this cell subset further validated that 306 hypoxia-induced glycolysis is a key mechanism underlying NKT mitochondrial dysfunction 307 ( Supplementary Fig. 8C ). In summary, these results suggest that NKT cells in circulation acquire 308 increased in COVID-19 PBMCs (Fig. 5B) . Importantly, GLUT1 + CD62L + NK frequency was 334 positively correlated with serum glucose levels (R = 0.76, p = 0.0042) (Fig. 5C) , which unveils a 335 pivotal relationship between blood glucose levels and memory cell glucose uptake. GLUT1 + 336 CD62L + NK also expresses significantly higher levels of HIF-1α (Fig. 5D) and ROS (Fig. 5E) , 337

suggesting that hypoxia-mediating anaerobic glycolysis induces oxidative stress in CD62L + NK 338 cells during COVID-19 infection. This interpretation was supported by upregulation of HIF-1α in 339 ROS + CD62L + NK cells (Fig. 5F ). Given that elevated GLUT1 expression is associated with 340 lymphocyte exhaustion and mitochondrial dysfunction, these results may suggest that 341 hyperglycemic COVID-19 infected patients will likely exhibit exacerbated CD62L + NK cell 342 dysfunction because of hypoxia-driven glycolytic metabolic reprogramming. 343 Induction of a pro-inflammatory cascade including type 1 IFN, toll-like receptor, NF-kB, and 351 chemokine signaling and PI3K/AKT pathway was observed in COVID-19 ECs (Figs. 6C,G) . 352

Glucose metabolism mediates type I IFN secretion through enhancing transcriptional expression 353 and epigenetic acetylation 34 . Indeed, we found a positive correlation between module scores for 354 glycolysis and type 1 IFN signaling, as well as for glycolysis and NF-kB signaling (Figs. 6D, E) . 355

We observed that genes encoding HLA class 1 (HLA-E, PSMA-6, TAP1, IFI30) were enriched in 357 COVID-19 ECs (Fig. 6A) . GSEA analysis further confirmed the upregulation of HLA class 1 358 antigen presentation in bulk ECs. In contrast, downregulation of genes encoding HLA class 2 359 (HLA-DRA, HLA-DPA1, HLA-DMA, DYNLL1) was found in COVID-19 ECs (Fig. 6A ) which was 360 further confirmed by GSEA analysis (Fig. 6F) . Glycolysis was reported to repress functional 361 response of antigen presenting cells during infection 36 . We indeed observed a negative correlation 362 of glycolysis and genes encoding HLA class 2 machinery (Fig. 6H) . These results revealed 363 potential links between dysregulated EC metabolism with cytokine release syndrome and 364 adaptive immune dysfunction in COVID-19. 365 BALF ECs were next identified and subsetted for downstream analysis (Fig. 7A) . Differential 366 expression analysis revealed key differences in the expression of transcripts governing key 367 metabolic pathways (Fig. 7B) . Additionally, UMAP performed solely on differentially expressed 368 metabolic genes revealed distinct clustering of bulk epithelial cells along disease severity (Fig.  369   7C) . Pearson correlation analysis performed on ECs revealed a strong positive correlation 370 between HIF1A and key glycolytic transcripts, suggesting a hypoxia-induced glycolytic metabolic subtypes based on the expression of canonical genes associated with cilia production (CFAP126, 373

and DNAAF) (Figs. 7E) . The ratio of pseudostratified ciliated ECs to nonciliated epithelial cells 374 was inversely correlated with COVID-19 disease severity (Fig. 7F) . This finding suggested that 375 SARS-CoV-2 infection produced direct injury to the ciliated EC compartment. Overexpression of 376 glycolytic transcripts (ENO1, ADH1A3, GAPDH, ALDOA, PCK2) was noted in both ciliated and 377 nonciliated EC subsets from COVID-19 infected patients (Fig. 7G) . These results were validated 378 by GSEA analysis, which demonstrated enrichment of glycolysis genes (Fig. 7G) . We also 379 observed decreased expression of FAO regulating genes to different extent in ciliated and 380 nonciliated ECs from S-COVID-19 compared to HC (Fig. 7G) . HIF-1A and anaerobic glycolysis 381 gene expression was strongly correlated with reduced expression of the OXPHOS and TCA cycle 382 genes in these EC subsets from S-COVID-19 (Fig. 7H) . GSEA analysis demonstrated enrichment 383 of glycolysis, as well as a large downregulation of OXPHOS and TCA cycle regulating genes in 384 S-COVID-19 ciliated and nonciliated ECs (Fig. 7H) . Collectively, these results suggested that 385 oxygen deprived conditions in the COVID-19 lung mediates a metabolic switch from aerobic FAO 386 and OXPHOS towards anaerobic glycolysis in ECs, which is strongly linked to mitochondrial 387 dysfunction. Warburg like metabolic adaptation to rely on aerobic glycolysis; thus, a metabolic switch from 439 aerobic to HIF-1a mediated anaerobic glycolysis will not impair initial T cell activation into effector 440 subsets, which is consistent with reports demonstrating hyperactivation of CD8 T cells 42 . 441

However, because of prolonged anaerobic glycolysis, the lung microenvironment becomes 442 increasingly hostile, resulting in nutrient depleted, hyperlactatemic, and hypoxic conditions. 443

Herein, we show that this dysregulated metabolism is heavily tied with impaired memory 444 lymphocyte formation. Although memory lymphocytes are traditionally associated with a heavy reliance on OXPHOS and fatty acid oxidation, we have detected multiple clusters of cellular state 446 associated with high glucose uptake, ROS production, hypoxia mediated transcriptional 447 response, and cellular exhaustion in CD8 and NK memory cell populations that were specifically 448 enriched in hospitalized COVID-19 patients. These populations were negatively associated with 449 memory lymphocyte frequency, suggesting that this metabolic switch may halt the transition of 450 activated lymphocytes into memory cells. Pseudotime analysis and trajectory interference with 451 transcriptomic data further demonstrate differentiation of tissue-resident memory cells that was 452 tied to altered lymphocyte metabolism. In addition to anaerobic glycolysis, upregulation of 453 glutaminolysis and mitophagy was also seen in CD8 memory cells in order to sustain bioenergetic 454 demands. Clustering solely on metabolic phenotype revealed clear distinctions between healthy, 455 moderate, and severe patients in CD8 memory cells, suggesting the potential use of metabolic 456 markers in predicting memory cell response. 457

Interestingly, we found that SARA-CoV-2 derived EC damage creates oxygen-deprived 458 conditions in the lungs that not only induce metabolic reprogramming of various immune cell 459

subsets, but also themselves. We found that during COVID-19 infection, differential metabolism 460 drives lung ECs towards senescence and towards acquiring a significant SASP phenotype, 461 leading to secretion of proinflammatory cytokines, reduced HLA class 2 mediated 462 immunosurveillance, and increased HLA class 1 machinery. Chronic stimulation of exhausted 463 lymphocytes, that demonstrate attenuated effector function and cytokine secretion in nutrient-464 depleted microenvironments, by antigen presenting cells via HLA class 1 leads to significantly 465 increased cellular exhaustion, which further impairs the capacity of cells to differentiate into 466 memory phenotypes. Our results therefore show that the immunometabolic rewiring of ECs in the 467 BALF can be a potential mechanism for organ-specific lymphocyte exhaustion and memory cell immunity has yet to be established 57,58 . Furthermore, there has yet to be an attempt to understand 533 the longevity of convalescence-induced protective immune responses in COVID-19 patients with 534 metabolic disorders. 535

Despite claims of biomarkers to predict COVID-19 severity 59 , no specific markers for COVID-19 536 patients with metabolic co-morbidities have been yet discovered. In the current study, using high 537 dimensional analyses, we provide a number of lowly abundant cell populations in the blood of 538 severe COVID-19 patients that can potentially predict disease severity including GLUT1 + 539 mitochondrially exhausted CTL, CD8CM, NKT and NK cells. Noticeably, a clear correlation 540 between the serum glucose level, recently identified as risk factor COVID-19 severity in patients 541 with pneumonia 60 , and GLUT-1 high CD62L + NK cells was observed, suggesting that the use of 542 metabolic biomarkers in combination can be strong prognostic indicator for COVID-19 disease 543 severity.

Overall, our current study sheds important new light on the molecular and cellular mechanisms 545 by which immune cell metabolism regulates COVID-19 pathobiology. Shortcomings of our study 546 include a limited sample size for analysis of BALF transcriptomic data, along with a lack of 547 proteomic data for our study of lung ECs. Additionally, our assessment of cellular metabolism is 548 limited to analysis surface and intracellular metabolic marker expression, coupled with 549 transcriptomic data for key metabolic pathways. Future studies should aim to validate our results 550 using approaches that shed more light into functional metabolism such as Seahorse Metabolic 551

Flux profiling or metabolomics. Given the large diversity in the range of responses towards SARS-552

CoV-2 infection, stratification for factors such as gender, BMI, preexisting conditions, and prior 553 treatments would help to further validate our analysis. Nonetheless, this study provides crucial 554 information about the pivotal relationship between cellular metabolism and the memory 555 lymphocyte response during severe COVID-19. Our results shed light on novel biomarkers, 556 therapeutic targets, and strategies for the COVID-19 therapy. We also provide a novel data 557 analysis pipeline for understanding single cell metabolism in an organ-specific manner. We are 558 confident that this comprehensive single-cell transcriptomic and proteomic portrait of immune cell 559 

PBMCs were isolated by density-gradient centrifugation using Ficoll. Briefly, blood specimens 577 were centrifuged at 700G for 7 min at RT for serum collection. The pellets were resuspended in 578 phosphate buffer saline (PBS). Cell suspension were carefully overlay on the top of 4 mL Ficoll in 579 15 mL conical tube, followed by centrifugation at 700G at RT for 25 min without break. PBMCs 580 were collected from interphase between plasma and Ficoll layers. Cells were wash twice with 581 PBS to remove Ficoll residue. All the procedures were approved at BSL2 + level by UCF 582 Environmental Health and Safety. 583

About 5x10 5 cells from each sample were used for flow cytometry staining. See Table S3 for 585 antibody information. PBMCs were first stained with live/dead in PBS for 15 min, washed with 586 FACS buffer, and stained with surface markers in FACS buffer at 4 O C for 30min. Following identify unsupervised clusters (Fig. 1H and Fig. S3A ) which were assigned to specific populations 612 based on canonical marker expression (Fig. 1I and Fig. S3B ).

healthy donors were used for analysis 66 . This study defined moderate and severe COVID-19 616 patients as those with pneumonia experiencing respiratory distress and hypoxia and with critical 617 condition, requiring ICU care, and having been placed under mechanical ventilation, respectively. 618

Prefiltered expression matrices with UMI counts were downloaded from the GEO Database with 619 accession number GSE145926. Additionally, as suggested by the original study, data from an 620 additional BALF sample derived from a healthy donor from a separate study was used as a 621 reference 67 . Prefiltered expression matrices with UMI counts were downloaded from the GEO 622

Database with accession number GSE128033 and sample number GSM3660650. 623

Quality control and data preprocessing was done using Seurat 68,69 . First, cells for which more 625 than 10% of reads were mitochondrial transcripts were discarded. Next, we removed cells that 626 had less than 1000 detected transcripts. Cells with less than 200 and greater than 6000 unique 627 genes were also filtered. Filtered data from different 14 patient samples were integrated in Seurat. 628

Individually, data from each sample was log 2 normalized and the top 2000 variable genes were 629 identified using the "vst" method in Seurat. Data from each sample was next scaled and PCA was 630 run with percentage of mitochondrial DNA and number of detected unique genes regressed out. 631

Alignment and batch effect correction was done using reciprocal PCA and canonical correlation 632 analysis (CCA) (in accordance to standard Seurat integrated analysis workflow) on the first 30 633 dimensions of the data. Next, a shared nearest neighbor graph was constructed and Louvain-634 based optimization was run to perform unsupervised clustering. UMAP was next run on the first 635 

Monocle 3 was used to construct a trajectory upon UMAP embeddings and order cells in 652 pseudotime 70 . Analysis was done on both CD8 and CD4 T cells. Seurat wrapper function 653 "asMonocle" was used to create Monocle CellDataSet object from an existing Seurat object. 654 "learn_graph" function was used to construct trajectory mappings onto transferred UMAP 655 embeddings. "order_cells" was used to estimate and order cells in pseudotime. All samples for 656 CD8 and CD4 populations were ordered together and were split by disease state after ordering 657 for differential comparison of pseudotime. 658

To investigate whether metabolic phenotypes of certain cell populations alone could be used 660 alone as predictive indicators of disease severity, dimensionality reduction at both a single cell 661 and sample-wide resolution was done only on key identified differentially expressed metabolic analysis, principal component analysis was conducted and the first three principal components 664 were visualized. For analysis at single cell resolution, UMAP was done and the first two 665 components were visualized. 666

For construction of gene pathway enrichment network, networkanalyst.ca was used 71 . All 668 statistically significant genes were inputted along with log fold change values to construct 669 enrichment network. Transcription factor -gene interaction network was also constructed using 670 networkanalyst.ca 71 . Statistically significant genes along with log fold changes vales were 671 inputted. The "degree" filter was first set to 100 and then the "betweenness" filter was set to 170. 672

For heatmap visualizations, scaled SCTransformed values were used and the Complexheatmap 674 package was used to generate visualization 71 . Hierarchical clustering and dendrogram 675 generation were done using default settings of the package. Outliers with extremely high scaled 676 expression values ( >> 2) were set to a maximum value of 2 to not distort the rest of the Fig. . For  677 dotplot visualizations, first a euclidean distance matrix was generated for which hierarchical 678 clustering was then applied. Ggtree was next used for dendrogram construction 72 . ReactomePA 679 package was used for functional GSEA 73 . All unique detected genes in the cell subset were 680 sorted by log fold change values to create ranked list that was inputted for GSEA analysis. enrichR 681 was used to determine over and under expressed pathways from differential expression analysis 682 ECs, adjacent is a GSEA enrichment plot for "Glycolysis" comparing ciliated ECs from severe vs 805 healthy control patients; H. Dot plots demonstrating expression and hierarchical clustering of 806 select key mitochondrial metabolism genes for ciliated ECs, adjacent is a GSEA enrichment plot 807

for "TCA and Respiratory Electron Transport" comparing ciliated ECs from severe vs healthy 808 control patients; I. Dot plots demonstrating expression and hierarchical clustering of select key comparing nonciliated ECs from severe vs healthy control patients; J. Dot plots demonstrating 811 expression and hierarchical clustering of select key mitochondrial metabolism genes for 812 nonciliated ECs, adjacent is a GSEA enrichment plot for "TCA and Respiratory Electron 813

Transport"" comparing nonciliated ECs from severe vs healthy control patients 814 Panel for metabolic phenotype-based clustering  CTL  APOE, APOC1, FABP4, MARCO, NFUFB8, NDUFC2, NDUFA11, ATP5MC2,  COX5A, ATP6V0C, CD38, LAG3, HIF1A, CFLAR, GABARAP, HSPA8, TOMM5,  OPTN, USP15, NFATC2, PPP3CC, CCND3, GAPDH, GALM, ALDOA  CD8  Memory   APOE, APOC, OLR1, MARCO, FABP4, GABARAP, CFLAR, HSPA8, CCND3,  PPP3CC, NFATC2, DGLUCY, GLUD1, CD38, TIGIT, LAG3, GAPDH, GALM,  GPI, ALDOA, HIF1A, TOMM5, USP15, OPTN, NDUFB8, OSP15, OPTN 

A Novel Coronavirus from Patients with Pneumonia in China

Modeling COVID-19 scenarios for the United States

Efficacy of the ChAdOx1 nCoV-19 Covid-19 Vaccine against the B

Omics-Driven Systems Interrogation of Metabolic Dysregulation in COVID-19

Elevated Glucose Levels Favor SARS-CoV-2 Infection and Monocyte 971 Response through a HIF-1alpha/Glycolysis-Dependent Axis

Metabolic Syndrome and COVID-19 Mortality Among Adult Black Patients in New 974

Plasma metabolomic and lipidomic alterations associated with COVID-19

Metabolic stress and disease-stage specific basigin expression of peripheral 978 blood immune cell subsets in COVID-19 patients. medRxiv, 2020

The tumor microenvironment as a metabolic barrier 981 to effector T cells and immunotherapy

A crucial role of angiotensin converting enzyme 2 (ACE2) in SARS coronavirus-983 induced lung injury

Risk factors for severity and mortality in adult COVID-19 inpatients in Wuhan

Lactate dehydrogenase, an independent risk factor of severe COVID-19 patients: a 987 retrospective and observational study

Clinical features of patients infected with 2019 novel coronavirus in Wuhan

Lactate Is a Natural Suppressor of RLR Signaling by Targeting MAVS

Lactate Regulates Metabolic and Pro-inflammatory Circuits in Control of T Cell 994

Migration and Effector Functions

Natural killer cell immunotypes related to COVID-19 disease severity

Impaired Cytotoxic CD8(+) T Cell Response in Elderly COVID-19 Patients

Single-cell landscape of immunological responses in patients with COVID-19

The glucose transporter Glut1 is selectively essential for CD4 T cell 1003 activation and effector function

Inhibiting glycolytic metabolism enhances CD8+ T cell memory and antitumor function

Induction of HIF-1alpha by HIV-1 Infection in CD4(+) T Cells Promotes Viral

VDAC oligomers form mitochondrial pores to release mtDNA fragments and 26

Metabolic plasticity of HIV-specific CD8(+) T cells is associated with enhanced 1018 antiviral potential and natural control of HIV-1 infection

Tissue-resident memory T cells

Metabolic switching and fuel choice during T-cell 1023 differentiation and memory development

A metabolic handbook for the COVID-19 pandemic

Role of type 1 natural killer T cells in pulmonary immunity

TLR-driven early glycolytic reprogramming via the kinases TBK1-IKKvarepsilon 1030 supports the anabolic demands of dendritic cell activation

CD62L expression identifies a unique subset of polyfunctional CD56dim NK cells

Viral and host factors related to the clinical outcome of COVID-19

Lactate modulation of immune 1037 responses in inflammatory versus tumour microenvironments

Cell Exhaustion During Chronic Viral 1040 Infection and Cancer

Increased Tumor Glycolysis Characterizes Immune Resistance to Adoptive T 1043

Metformin use is associated with reduced mortality rate from 1045 coronavirus disease 2019 (COVID-19) infection

Identification of a novel coronavirus in patients with severe acute respiratory 1048 syndrome

SARS-CoV-2 infection paralyzes cytotoxic and metabolic functions of immune 1050 cells. bioRxiv

Marked T cell activation, senescence, exhaustion and skewing towards TH17 in 1052 patients with COVID-19 pneumonia

Metabolic programs define dysfunctional immune responses in severe 1055

COVID-19 patients. medRxiv

Aberrant hyperactivation of cytotoxic T-cell as a potential determinant of 1058

COVID-19 severity

Type 1 Interferons Induce Changes in Core Metabolism that Are Critical for Immune 1073

Type I and Type III Interferons -Induction, Signaling, Evasion, and 1075 Application to Combat COVID-19

Disturbed mitochondrial dynamics in CD8(+) TILs reinforce T cell exhaustion

Association Between Hypoxemia and Mortality in Patients With COVID-19

Genomic evidence for reinfection with SARS-CoV-2: a case study

Regulation of metabolic supply and demand during B cell activation 1084 and subsequent differentiation

Loss of Bcl-6-Expressing T Follicular Helper Cells and Germinal Centers in 1087 COVID-19

Extrafollicular B cell responses correlate with neutralizing antibodies and 1089 morbidity in COVID-19

Immunological memory to SARS-CoV-2 assessed for up to 8 months after 1091 infection

Evolution of Antibody Immunity to SARS-CoV-2. bioRxiv

Longitudinal analyses reveal immunological misfiring in severe COVID-19

Metabolic biomarker profiling for identification of susceptibility to severe pneumonia and COVID-1098 19 in the general population

flowCore: a Bioconductor package for high throughput flow cytometry

OpenCyto: an open source infrastructure for scalable, robust, reproducible, and 1102 automated, end-to-end flow cytometry data analysis

Compensation of Signal Spillover in Suspension and Imaging Mass Cytometry

Normalization Algorithm for Cytometry Data

FlowSOM: Using self-organizing maps for visualization and interpretation of 1110 cytometry data

Single-cell landscape of bronchoalveolar immune cells in patients with COVID-19

Using ggtree to Visualize Data on Tree-Like Structures

Corrigendum: Gut microbiome-derived metabolites modulate intestinal 1128 epithelial cell damage and mitigate graft-versus-host disease

Enrichr: a comprehensive gene set enrichment analysis web server 2016 1131 update

The current study is conducted with the support from the University of Central Florida start-up 900 funding and College of Medicine, University of Central Florida COVID-19 seed funding support to

The authors declare no conflicts of interest.