Abstract
Immune response dysregulation plays a key role in severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pathogenesis. In this study, we evaluated immune and endothelial blood cell profiles of patients with coronavirus disease 2019 (COVID-19) to determine critical differences between those with mild, moderate, or severe COVID-19 using spectral flow cytometry. We examined a suite of immune phenotypes, including monocytes, T cells, NK cells, B cells, endothelial cells, and neutrophils, alongside surface and intracellular markers of activation. Our results showed progressive lymphopenia and depletion of T cell subsets (CD3+, CD4+, and CD8+) in patients with severe disease and a significant increase in the CD56+CD14+Ki67+IFN-γ+ monocyte population in patients with moderate and severe COVID-19 that has not been previously described. Enhanced circulating endothelial cells (CD45−CD31+CD34+CD146+), circulating endothelial progenitors (CD45−CD31+CD34+/−CD146−), and neutrophils (CD11b+CD66b+) were coevaluated for COVID-19 severity. Spearman correlation analysis demonstrated the synergism among age, obesity, and hypertension with upregulated CD56+ monocytes, endothelial cells, and decreased T cells that lead to severe outcomes of SARS-CoV-2 infection. Circulating monocytes and endothelial cells may represent important cellular markers for monitoring postacute sequelae and impacts of SARS-CoV-2 infection during convalescence and for their role in immune host defense in high-risk adults after vaccination.
Introduction
Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the causative virus of coronavirus disease 2019 (COVID-19), has devastated global public health and led to millions of deaths (1, 2). In-hospital mortality rates during the first 6 months of the pandemic reached >15–19% in some institutions (3, 4); however, disease presentation is varied and dependent on several previously identified risk factors (5–16). Some adults suffer only mild illness, yet mount and maintain an immune response to SARS-CoV-2 (17, 18), while others, primarily those with risk factors such as advanced age and obesity, develop severe disease and respiratory failure (6, 11–13, 15, 19). The initial days after COVID-19 symptom onset are typically experienced as an outpatient (20), but the immune dysregulation that follows can lead to severe respiratory failure requiring hospitalization and sometimes resulting in death (21). As with other RNA viruses, the immune system must function precisely to achieve virucidal activity without overwhelming inflammation, which is often detrimental to the host (22–24).
Immune response dysregulation plays a crucial role in SARS-CoV-2 pathogenesis (25–28) and is likely central to the progression from a mild upper respiratory tract illness to fulminant respiratory failure and fatal coagulopathy (21, 29). The driving immune factors that precipitate such a devastating response to SARS-CoV-2 are unknown. A thorough understanding of the immunopathology caused by SARS-CoV-2 is of paramount importance as more infectious and virulent variants emerge (30).
Severe COVID-19 (31–34) has been associated with changes to several immune cell populations, including monocytes, basophils, plasmacytoid dendritic cells, NK cells, and T cells (25, 31, 32). Depletion of CD8+ T cells has been associated with hospital mortality (35, 36), and Ag-specific T cell circulation is delayed (25). In contrast, some patients with severe COVID-19 have demonstrated high CCR7+CD8+ T cell populations (31). Many studies have shown an association with disease severity and cytokines such as IL-6, IL-8, and TNF-α (32, 37, 38). Hyperelevated cytokines cause endovascular permeability, leading to acute respiratory distress syndrome, as seen with COVID-19 and other disease processes (38, 39). Direct endothelial cell injury also has been associated with severe disease (40, 41) and may be associated with thrombotic complications in patients with severe COVID-19 (42, 43).
Whole blood profiling with immune and vascular elements is essential to delineate underlying mechanisms and markers of severe immunopathology. In this study, we evaluated immune and endothelial blood cell profiles of patients with COVID-19 to determine critical differences and identify novel immune marker signatures between those with mild, moderate, or severe COVID-19. CD56 is one phenotypic marker of NK cells but can also be expressed by other immune cells, including monocytes (44), that have immune-stimulatory effector functions and possess efficient cytotoxic properties (45, 46). CD56+ monocytes have been reported in patients with various infectious, autoimmune, or malignant diseases (47–49). During our initial analysis of the PBMCs in patients, we identified a monocyte population expressing CD56 in patients with COVID-19. Therefore, we aimed to investigate CD56+ monocytes in patients with mild/asymptomatic, moderate, and severe disease.
Materials and Methods
Human subjects and study design
Enrollment of participants was a joint effort between Colorado State University (CSU) and the University of Colorado Health System (UCH), Medical Center of the Rockies Medical Research. Adults >18 y of age with a SARS-CoV-2 PCR+ test result and adults with SARS-CoV-2–negative tests (uninfected) were enrolled into a biorepository (ClinicalTrials.gov Identifier: NCT04603677) between July 2020 and February 2021. Participants were recruited from the community via the health department screening, local medical clinics, e-mails, Web-based announcements, and directly through northern Colorado hospitals, including Poudre Valley Hospital (Fort Collins, CO), Medical Center of the Rockies (Loveland, CO), and Greeley Hospital (Greeley, CO). The UCH Trauma Research Department identified eligible hospitalized patients through Epic, the EMR platform. UCH staff directly approached eligible, hospitalized patients on the ward for consent and enrollment. The UCH investigators also conducted frequent EMR searches to recruit recently discharged/diagnosed COVID-19 patients. In addition, enrolled participants also assisted via word of mouth and personal networks. SARS-CoV-2+ patients were graded according to the severity of disease based on the oxygen requirement and Yale score of severity: mild/asymptomatic COVID-19 = no oxygen requirement; moderate COVID-19 = oxygen requirement of 1–5 l/min; and severe COVID-19 = oxygen requirement >5 l/min and/or required mechanical ventilation (50).
This study of whole blood profiling of patients with COVID-19 was approved by CSU’s Research Integrity and Compliance Review Office Institutional Review Board #2105 (protocol ID 20-10063H), as well as UCH IRB (Colorado Multiple IRB 20-6043). All enrolled participants provided written informed consent. This study followed the ethical principles outlined in the Declaration of Helsinki. All participants received $25 cash compensation for participation. All clinical data were deidentified and password protected for analysis. Clinical data obtained from the EMR were stored in Research Electronic Data Capture, and each record was assigned a unique identifier (51).
Blood was collected from study participants within 16 d of SARS-CoV-2 PCR+ in cell preparation tubes (CPTs), and PBMCs were isolated. PBMCs were stained with cell surface and intracellular activation markers to define cell populations (Supplemental Table I) using high-dimensional spectral flow cytometry (52).
Isolation of human PBMCs
Approximately 40 mL of whole blood from each study participant was collected in five 8-mL CPTs. CPTs were centrifuged at 1500 × g for 30 min at room temperature with the brake turned off. The buffy coat was aspirated from each CPT with a 5-mL serological pipette, transferred into a 500-mL conical tube, and washed twice with PBS. The pellet was resuspended in 5 mL of PBS and counted, and 1 × 106 cells were added to a V-bottom 96-well plate in triplicate for flow cytometric staining.
Flow cytometry staining of PBMCs
PBMCs isolated earlier were centrifuged at 500 × g for 5 min at 4°C, supernatant was decanted, and the cells were incubated in 100 µL of 1× brefeldin A solution (5 µg/ml final concentration) at 37°C and 5% CO2. After 2 h, brefeldin was removed, and the cells were washed with 150 µL FACS buffer (PBS, 2% heat-inactivated FBS, and 0.05% sodium azide) by centrifugation at 500 × g for 5 min at 4°C. Fc receptors were blocked by incubating cells with 50 µl of 1:200 dilution of human Fc block (BD Biosciences) at 4°C for 15 min. Cells were stained with 50 µl surface staining Ab mixture with different concentrations of Abs (Supplemental Table I) and incubated at 4°C for 30 min in the dark. After surface staining, cells were washed and fixed using 100 µL 1× Foxp3 Fix/Perm Staining Buffer (eBioscience) and incubated for 1 h at room temperature. Fix/perm buffer was removed, and cells were washed with 1× permeabilization buffer (eBioscience) and incubated with intracellular cytokines and Ki67 Ab mixture overnight at 4°C in the dark. Supernatant was removed the following day, and the cells were washed with 1× permeabilization buffer and resuspended in 200 µL FACS buffer. Samples were read using a Cytek 4 Laser Aurora spectral flow cytometer, and 50,000 events in the leukocyte gate were recorded.
Flow cytometry staining of whole blood
Before isolating PBMCs, 200 µL whole blood from each patient was transferred into a 1.5-mL microcentrifuge tube. Erythrocytes were lysed using Gey’s RBC lysis buffer, cells were suspended in PBS, and 50 µL of cells was added to a V-bottom 96-well plate in triplicate. The plate was centrifuged at 500 × g for 5 min, the supernatant was carefully removed, and 100 µL of Zombie NIR amine reactive viability dye was added in a 1:2000 dilution (BioLegend) to the samples and incubated for 15 min in the dark. Cells were washed using FACS buffer, Fc receptors were blocked using human Fc block (BD Biosciences), and cells were incubated with surface Ab mixture (Supplemental Table II) for 30 min at 4°C. Cells were fixed using 4% paraformaldehyde, incubated for 15 min at room temperature, washed twice with FACS buffer, and resuspended in 300 µl FACS buffer. Samples were read using a Cytek 4L Aurora spectral flow cytometer, and 500,000 events were recorded.
Indirect ELISA for anti–receptor-binding domain IgG detection
An indirect ELISA was performed to evaluate Ab binding to the SARS-CoV-2 spike protein receptor-binding domain (RBD; catalog numbers 40592-V08H; Sino Biological US, Wayne, PA). The protocol for ELISA was adapted from Yonemura et al. (53). In brief, high-binding 96-half-well microplates (Corning Life Sciences, Tewksbury, MA, USA) were coated with 50 ng RBD protein prepared in PBS and incubated overnight at 4°C. Plates were washed the next day five times with wash solution (PBS + 0.05% Tween 20) and incubated with blocking buffer (PBS + 0.05% Tween 20 + 2% BSA + 2% normal goat serum; Jackson Immunoresearch, West Grove, PA) for 2 h. Plates were washed, and different dilutions of human plasma prepared in blocking buffer were added to the wells and incubated for 1 h. Plates were then washed five times and incubated for 1 h with HRP-conjugated anti-human IgG Abs (Jackson Immunoresearch) prepared in blocking buffer (1:10,000 dilution). A total of 100 µl 3,3′,5,5′-tetramethylbenzidine was added to develop colorimetric product (Thermo Fisher Scientific, Rockford, IL), and the reaction was stopped by adding 50 µl of 1 M sulfuric acid. OD was measured at 450 nm using a BioTek Synergy 2 plate reader (BioTek Instruments, Winooski, VT).
Statistics
Nonparametric tests of association are used throughout the study for integrated flow cytometric analysis and clinical parameters. Tests of association were performed by unpaired Wilcoxon test (for n = 2 categories) or Kruskal–Wallis test (for n > 2 categories). Spearman rank correlation coefficients were assessed by the corresponding nonparametric methods for significance. All tests were performed in a two-sided manner, and p < 0.05 was considered significant. False discovery rate correction was performed (when required) using the Benjamini–Hochberg method at the false discovery rate < 0.05 significance threshold. All statistical tests were performed in RStudio using stats package and also in GraphPad Prism.
Results
Participant characteristics and demographics
Whole blood was collected from 56 patients with confirmed SARS-CoV-2 infection, including 14 participants with mild, 23 with moderate, and 19 with severe COVID-19 (Supplemental Fig. 1A, Table I). The mean age of participants who suffered mild disease was 27.5 y, while those with moderate and severe disease were older (mean age 56 and 61 y, respectively). Those who had mild disease were more likely to be female (79%) than those who had moderate (52%) or severe disease (26%). In addition, those with moderate or severe disease had more comorbidities and significantly higher mean body mass index (BMI; mild = 24.5, moderate = 34.8, severe = 33.8) than those with mild disease. No participants enrolled with mild disease were Hispanic or Latinx, in contrast with 26% in the moderate and 47% in the severe group. Fourteen SARS-CoV-2 PCR-negative participants were enrolled as uninfected adults for group comparisons.
Table I.
Patient characteristics and demographics (total N = 70)
| Symptom/Disease Category Characteristics | Mild/Asymptomatic (n = 14) | Moderate (n = 23) | Severe (n = 19) | Uninfected (n = 14) |
|---|---|---|---|---|
| Age, mean (SD), y | 27.5 (18.7) | 56 (13.4) | 61 (13.5) | 47.8 (8.5) |
| Sex, n (%) | ||||
| Female | 11 (79) | 12 (52) | 5 (26) | 11 (78) |
| Male | 3 (21) | 11 (48) | 14 (74) | 3 (22) |
| BMI, mean (SD) | 24.5 (6.4) | 34.8 (7.3) | 33.8 (11.2) | 24.6 (5) |
| Ethnicity, n (%) | ||||
| Non-Hispanic/Latinx | 15 (100) | 17 (74) | 10 (53) | 13 (93) |
| Hispanic/Latinx | 0 (0) | 6 (26) | 9 (47) | 1 (7) |
| Coexisting conditions, n (%) | ||||
| CVA | 0 (0) | 2 (8.6) | 1 (5.2) | 0 (0) |
| COPD | 1 (6) | 4 (17.4) | 2 (10.5) | 0 (0) |
| DM | 1 (6) | 8 (35) | 10 (53) | 0 (0) |
| HTN | 1 (6) | 11 (48) | 15 (79) | 0 (0) |
| CAD | 1 (6) | 3 (13) | 1 (5.2) | 0 (0) |
| Active cancer | 0 (0) | 2 (8.6) | 0 (0) | 0 (0) |
| Days after PCR+ test, mean (SD) | 9 (3.4) | 7.6 (4.1) | 7 (5.2) | n/a |
| COVID-19 therapeutics received, n (%) | ||||
| Steroids | 1 (6) | 21 (91) | 19 (100) | n/a |
| Convalescent plasma | 1 (6) | 13 (59) | 12 (63) | n/a |
| Remdesivir | 1 (6) | 18 (78) | 17 (89) | n/a |
| Tocilizumab | 0 (0) | 0 (0) | 0 (0) | n/a |
CAD, coronary artery disease; COPD, chronic obstructive pulmonary disease; CVA, cerebrovascular accident; DM, diabetes mellitus; HTN, hypertension; n/a, not applicable.
Data were analyzed using the cyto-feature engineering pipeline (54) for unbiased analysis generating different populations (Supplemental Fig. 1B, 1C). Cyto-feature engineering is a novel R-based flow cytometry analysis pipeline to identify a plethora of cell populations efficiently. This pipeline uses Fluorescence Minus One controls or distinct population differences to characterize cells and different populations. Because this pipeline gates cells in an unbiased manner, one can find populations that can be missed by manual gating. The identified populations were further renamed based on marker expression and divided into T cell, B cell, monocyte, or NK cell populations and were compared among uninfected, mild/asymptomatic, moderate, and severe COVID-19 participants.
CD56+ monocytes showed inflammatory behavior and were elevated in patients with moderate and severe COVID-19
Unbiased computational analysis and confirmatory FlowJo v10 analysis (Supplemental Figs. 1B, 2A) showed a CD56+ monocyte population associated with severe COVID-19. The CD14+CD56+ monocyte population (Fig. 1A, 1B) was significantly higher in moderate (4.71 ± 3.7; p < 0.01) and severe participants (5.58 ± 2.6; p < 0.005) compared with mild COVID-19 participants (2.07 ± 1.02) and uninfected participants (>1%). We also evaluated CD14+CD56− monocytes and total CD14+ monocytes among COVID-19 severity groups and observed a significant increase in both populations in moderate and severe patients compared with mild/asymptomatic patients (Fig. 1B, 1C). CD56+ and CD56− monocyte proliferation was also investigated using Ki67 and showed that both populations were highly proliferative in moderate and severe patients (∼95% of total CD56+ and CD56− populations) compared with mild/asymptomatic patients (Fig. 1D). It was interesting to note that CD56+ monocytes expressed different subsets of markers and intracellular cytokines than CD56− monocytes (Fig. 1D). CD56+, but not CD56−, monocytes expressed IFN-γ and were significantly higher in patients with moderate (1.24 ± 1.12; p < 0.05) and severe (1.48 ± 0.98; p < 0.005) COVID-19 compared with patients with mild disease (0.16 ± 0.25) (Fig. 1Di). CD56− monocytes expressed HLA-DR, CD38, and PD1 and were further categorized based on these markers. We found that HLA-DR−CD38+ and HLA-DR−CD38+PD1+ populations increased significantly with disease severity and were highest in patients with severe disease (Fig. 1Dii). To further evaluate the response of CD56+, CD56−, and total monocytes during the course of infection, patients were stratified based on days after PCR+ diagnostic test results (1–5, 6–10, or 11–16 d) after SARS-CoV-2 PCR+ (Fig. 1C, 1E). Total monocyte population is consistently higher in patients with moderate and severe disease compared with mild/asymptomatic patients at all time points; however, this significant difference between mild and moderate/severe was highest (p < 0.001) at 11–16 d after SARS-CoV-2 PCR+ test results (Fig. 1C, right panel). CD56+ monocytes were significantly higher in patients with severe and moderate disease compared with mild participants in all three time points; however, their IFN-γ–releasing capacity declined with time. In contrast, CD56− monocytes increased over time from 20.53 ± 5.8 (1–5 d) to 35.84 ± 9.2 (11–16 d) in patients with severe disease; however, in patients with moderate disease, CD56− monocytes fluctuated between days 1 and 10 and became similar to patients with severe disease at 11–16 d postinfection. These CD56− populations were still proliferating and presented CD38 activation marker at 11–16 d postinfection.
FIGURE 1.
Evaluation of monocyte populations in PBMCs in SARS-CoV-2–positive and uninfected adults. (A) Flow cytometry bivariate dot plot showing CD14 on the x-axis and CD56 on the y-axis. Q1 represents CD14−CD56+ cells, Q2 represents CD14+CD56+ cells (highlighted quadrant), Q3 shows CD14+CD56− cells, and Q4 shows CD14−CD56− cells. (B) Comparison of total CD14+ monocytes between uninfected, mild, moderate, and severe COVID-19 participants. y-Axis represents percentage of cells out of total leukocytes, gated after singlets, live, and CD3− cells; x-axis represents disease severity (left panel). CD14+ monocytes in mild, moderate, and severe COVID-19 participants were grouped based on days from SARS-CoV-2 PCR+ test. Data are represented as 1–5, 6–10, and 11–16 days since SARS-CoV-2 PCR+ test on x-axis and percentage of leukocytes on the y-axis (left panel). (C) Comparison of CD14+CD56+ monocytes and CD14+CD56− monocytes between uninfected, mild, moderate, and severe COVID-19 participants. y-Axis represents percentage of cells out of total leukocytes, gated after singlets and live cells; x-axis represents disease severity. Each dot represents a separate participant within that group. Color key is provided on top of the graph. (D) Ki67 (proliferation marker) monocytes and monocytes releasing cytotoxic molecules (IFN-γ and Granzyme B). Data are presented as percentage of leukocytes on y-axis and uninfected, mild, moderate, and severe COVID-19 participants on the x-axis. (E) CD14+CD56+ and CD14+CD56− cells and their subtypes in participants with mild, moderate, and severe COVID-19 were grouped based on days from SARS-CoV-2 PCR+ test. Data are represented as 1–5, 6–10, and 11–16 days since SARS-CoV-2 PCR+ test on x-axis and percentage of leukocytes on the y-axis. All graphs were created using ggplot2 in RStudio, and significance was calculated using R package stats. Mild/asymptomatic (n = 14), moderate (n = 23), and severe disease (n = 19) and SARS-CoV-2 uninfected (n = 14). Each dot in the graph represents mean value (n = 3) of each patient. Significance was determined by unpaired Wilcoxon test: *p < 0.05, **p < 0.005, ***p < 0.0005, ****p < 0.0001.
Lymphopenia and T cell activation significantly increased in severe COVID-19
T cell subset differences in SARS-CoV-2 PCR+ patients were evaluated by disease severity using cyto-feature engineering (Supplemental Figs. 1B, 2B). We found that CD3+ T cells were significantly diminished in patients with moderate (35.74 ± 10.5; p < 0.05) and severe (30.29 ± 8.3; p < 0.005) COVID-19 compared with participants with mild disease (51.67 ± 11.6) (Fig. 2Ai). This decrease in T cells was true for both CD4+ and CD8+ T cells (Fig. 2A). Interestingly, cytotoxic T cells (CD8+CD56−), but not NKT cells (CD8+CD56+), were significantly reduced in moderate (p < 0.005) and severe patients (p < 0.005) compared with mild participants. We further evaluated the percentage of T cells by the number of days from SARS-CoV-2 PCR+ and found that in mild participants, the CD3+, CD4+, and CD8+CD56− T cell populations were highest at 5–10 d after PCR+ test results, the expected time frame of adaptive immune response action; however, in the same phase, these populations were decreased in patients with moderate and severe disease. CD3+, CD4+, and CD8+CD56− T cell populations were lowest in participants with severe disease at 11–16 d after PCR+ results (Fig. 2Aii).
FIGURE 2.
Comprehensive analysis of T cell populations in participants with mild, moderate, and severe COVID-19. (A) Frequencies of CD3+ (all T cells), CD4+ (helper T cells), CD8+CD56− (cytotoxic T cells), and CD8+CD56+ (NK T cells) out of total leukocytes (Ai). Percentage of T cells in mild, moderate, and severe disease represented as days (1–5, 6–10, and 11–16 days) from SARS-CoV-2 PCR+ test (Aii). (B) CD8+CD56− and CD8+CD56+ cells were plotted for the expression of PD1. Percentage represents PD1+ cells out of CD8+CD56− and CD8+CD56+ cells. (C) T cell subpopulations: CD25+ regulatory T cells, Ki67+ proliferating T cells, and PD1+ T cells. Data are represented as percentage of cells out of CD8+CD56+ T cells (Ci), CD8+CD56− T cells (Cii), and CD4+ T cells (Ciii). (D) T cell subpopulations are divided by days after SARS-CoV-2 PCR+ test and plotted as percentage (Di–iii). Graphs were created using ggplot2 in RStudio, and statistical significance was calculated using R package stats. Each dot in the graph represents mean value (n = 3) of each patient. Significance was determined by unpaired Wilcoxon test: *p < 0.05, **p < 0.005, ***p < 0.0005.
To identify the proliferating, exhausted, and cytokine-positive T cell subpopulations in each COVID-19 severity group, we plotted the frequency of CD8+CD56+, CD8+CD56−, or CD4+ T cells for different cytokines, proliferation, and exhaustion markers. We found that CD8+CD56−, but not CD8+CD56+, T cells positive for PD1 were significantly higher in patients with severe disease compared with all other severity groups (Fig. 2B). Although, as noted earlier, T cell percentages were significantly lower in the moderate and severe groups, Granzyme B+CD56+ and Granzyme B+CD56− CD8+ T cells were significantly higher in these groups (Fig. 2Ci, ii). Furthermore, Granzyme B+CD56+CD8+ T cells were significantly higher in moderate and severe patients than in mild participants at 1–5, 6–10, and 11–16 d from the SARS-CoV-2 PCR+ test (Fig. 2Di). In contrast, CD8+CD56− T cell numbers were similar among patients with mild, moderate, and severe disease during the early stage (1–5 d after PCR+) of infection but decreased in patients with mild disease over time. Notably, CD8+CD56− T cells from patients with moderate and severe disease continued to produce IFN-γ and Granzyme B until 16 d after PCR+ test results, consistent with hyperinflammation (Fig. 2Dii). We also found a significant increase in CD4+CD25+ regulatory T cells in patients with moderate (5.94 ± 2.9; p < 0.05) and severe (8.23 ± 4.3; p < 0.05) disease compared with patients with mild disease (3.37 ± 1.9), whereas uninfected participants had <2% CD4+CD25+ cells (Fig. 2Ciii). Overall, these T cell findings demonstrate that severe COVID-19 is dually marked by lymphopenia and persistent T cell activation.
Elevated frequency of activated B cells and humoral responses in patients with severe COVID-19
Comparing B cell population dynamics between disease severity groups (Supplemental Figs. 1B, 2C) showed that CD19+ B cells were significantly higher in patients with moderate (26.99 ± 4.4; p < 0.05) and severe disease (29.14 ± 6.5; p < 0.05) compared with patients with mild disease (20.55 ± 4.5), with variable levels observed for the uninfected population (Fig. 3Ai). We further analyzed B cell responses by days after SARS-CoV-2 PCR+ tests and found that CD19+ B cells were higher in patients with moderate and severe disease than uninfected participants at all three time points: 1–5, 6–10, and 11–16 d after PCR+ tests (Fig. 3Aii). In mild disease, CD19+ cells were lower than moderate and severe groups in the first 5 d, increased significantly at 6–10 d, and declined at 11–16 d after PCR+ tests (Fig. 3Aii).
FIGURE 3.
B cells and Ab responses in COVID-19 progression. (A) Frequency of CD19+ B cells in uninfected adults alongside patients with mild, moderate, and severe COVID-19 (Ai). B cell responses over time from days after SARS-CoV-2 PCR+ (Aii). Activated B cells (HLA-DR−CD38+, HLA-DR−CD38+Ki67+, and HLA-DR+CD38+) among three disease severity groups and uninfected participants (Aiii) and their responses over time (i.e., days after SARS-CoV-2 PCR+ test) (Aiv). (B) Data showing plasma B cells (CD27−CD138+ and CD27+CD138+) and memory B cells (CD27+CD138−) out of total CD19+ B cells. (C) Memory B cells and plasma cells are divided based on days since SARS-CoV-2 PCR+ test and plotted as the frequency of CD19+ cells. (D) IgG antibody reactivity to SARS-CoV-2 RBD in plasma samples. Data are represented as OD at 450 nm on the y-axis for the corresponding plasma dilutions (1:250, 1:1250, and 1:6250). (E) Plot representing AUC was calculated using all dilutions of anti-RBD IgG in plasma of patients with mild, moderate, and severe COVID-19 and uninfected adults. Closed triangles represent patients who received convalescent plasma treatment. All graphs were created using ggplot2 in RStudio, and significance was calculated using R package stats. Each dot in the graph represents mean value (n = 3) of each patient. Significance was determined by unpaired Wilcoxon test with Benjamini–Hochberg correction: *p < 0.05, **p < 0.005, ***p < 0.0005.
HLA-DR, CD38, and Ki67 were used to assess B cell activation and proliferation. We found that CD19+CD38+HLA-DR−Ki67+ B cells significantly increased with disease severity, yet without significant variation in B cell HLA-DR expression according to disease severity groups (Fig. 3Aiii). In addition, HLA-DR−CD38+ and HLA-DR−CD38+Ki67+ B cells were similar in patients with mild, moderate, and severe disease at 1–5 d after PCR+ test results. These cell types continued to proliferate and increased at 11–16 d after PCR+ tests in patients with moderate and severe disease, while in mild participants, HLA-DR−CD38+Ki67+ B cells decreased over time (Fig. 3Aiv).
(Fig. 3B and Supplemental Figs. 1C and 2D show the whole blood analysis (n = 48) for the presence of CD138+ plasma B cells and CD27+ memory B cells. The CD27+, CD138+CD27−, and CD138+CD27+ B cells were plotted as the frequency of CD19+ cells and demonstrated that CD138+CD27− B cells increased significantly in participants with mild (p < 0.05), moderate (p < 0.0005), and severe (p < 0.0005) disease as compared with uninfected participants. No difference was detected in CD19+CD27+ B cells across disease severity groups and uninfected adults. Further dividing B cell populations by the number of days from the SARS-CoV-2 PCR+ test result, we found that CD138+CD27− B cells were highest in severe disease at 11–16 d after PCR+. In moderate disease, CD138+CD27− B cells were consistent through all time points; however, they decreased significantly at 11–16 d after PCR+ in mild participants (Fig. 3C).
Using ELISA, we evaluated plasma for differences in SARS-CoV-2 spike protein RBD IgG Ab among the three severity groups. Participants with severe disease had the highest levels of anti-RBD IgG (area under the curve [AUC] = 11768 ± 8154) when compared with participants with moderate (AUC = 6064 ± 4142) and mild (AUC = 1237 ± 1100) disease (Fig. 3D, 3E). Sixty percent of the severe and 40% of moderate patients received convalescent plasma therapy, which could explain the high IgG titers in these patients (Fig. 3E). As expected, uninfected participants did not show the presence of anti-RBD IgG (Fig. 3D, 3E).
Participants with severe COVID-19 displayed dysregulation of endothelial cells and neutrophils
Thromboembolic disease and vascular endothelial cell damage have been detected in patients with severe COVID-19 (55), so we next evaluated whole blood for the presence of circulating endothelial cells (CECs; CD45−CD31+CD34+CD146+), circulating endothelial progenitors (CEPs) 1 (CEP-1s: CD45−CD31+CD34−CD146−), and CEP-2s (CD45−CD31+CD34+CD146−) (Supplemental Fig. 2E). CEP-1s were significantly increased in moderate (2.98 ± 1.8; p < 0.05) and severe (3.48 ± 2.8; p < 0.05) disease; however, in uninfected and mild participants, CEP-1s were <1% of the total live cells. CECs were also significantly higher in patients with moderate (p < 0.05) and severe (p < 0.05) disease compared with uninfected participants, suggesting the presence of vascular endothelial cell damage in patients with COVID-19 (Fig. 4Ai). Evaluation of the presence of blood CECs and CEPs according to days since SARS-CoV-2 PCR+ showed augmentation of CEP-1 on 1–5 d in patients with moderate and severe disease, which persisted at 11–16 d after PCR+ test results. CECs and CEP-2 increased at 11–16 d after PCR+ test results in patients with moderate and severe disease. Mild patients and uninfected adults did not display shedding of CECs and CEPs (Fig. 4Aii).
FIGURE 4.
Increased endothelial cells and neutrophils in whole blood of patients with severe COVID-19. (A) Whole blood from each patient was stained with CD45+ antibody, and CD45− live cells were gated to evaluate endothelial cell markers. Graphs show percentage of CD31+CD146−CD34− (CEP-1s), CD31+CD146+CD34+ (CECs), and CD31+CD34+CD146− (CEP-2) of CD45− live cells (Ai). Endothelial cell distribution after 1–5, 6–10, and 11–16 d after SARS-CoV-2 PCR+ test (Aii). (B) Live CD45+ cells were gated for neutrophils from whole blood as CD16+CD66b+CD11bint and CD16+CD66b+CD11bhi. Data are represented as percentage of CD45+ cells on y-axis and disease severity groups on x-axis (Bi). Neutrophil data are also represented as days after SARS-CoV-2 PCR+ test (Bii). All graphs were created using ggplot2 in RStudio, and significance was calculated using R package stats. Significance was determined by unpaired Wilcoxon test: *p < 0.05, **p < 0.005, ***p < 0.0005.
Neutrophils play a crucial role in virus elimination (56) but are also involved in hyperinflammation and cytokine storm (57, 58). We investigated the percentages of neutrophils in whole blood (Supplemental Fig. 2E) as a requency of CD45+ cells (Fig. 4Bi). CD16+CD66b+CD11bint and CD16+CD66b+CD11bhi neutrophils were elevated in patients with moderate (69.13 ± 7.8 and 0.82 ± 0.3, respectively) and severe COVID-19 (78.6 ± 11.6 and 1.11 ± 0.8, respectively) compared with uninfected (48.75 ± 8.1 and 0.21 ± 0.1, respectively) and mild participants (50.17 ± 10.4 and 0.44 ± 0.28, respectively). CD16+CD66b+CD11bint neutrophils were present in whole blood at 1–5 d after SARS-CoV-2 PCR+ test and continue to increase up to 11–16 d, suggesting dysregulated neutrophil response in patients with severe COVID-19 (Fig. 4Bii).
Age, obesity, and hypertension were associated with CD56+ monocytes, endothelial cells, and neutrophils for patients with severe COVID-19
The integrated flow cytometric and clinical metadata analysis completed for participants using the Spearman rank correlation analysis indicated key associations between clinical features and immune cell populations. The analysis shows multiple significant correlations between clinical parameters, immune marker activation status, and disease severity. CD56+ monocytes were significantly increased in participants with moderate and severe disease and were significantly positively correlated with age and BMI (Fig. 5A, 5B). We found that CD56+ monocytes were also correlated to hypertension, which could be another comorbidity in COVID-19 (Fig. 5A). In contrast, T cells were negatively correlated with age, obesity, and hypertension in our cohort (Fig. 5A, 5B).
FIGURE 5.
Integrated analysis of immune cells and clinical parameters. (A) Spearman correlation of indicated features for participants with COVID-19. Data include both PBMCs and whole blood analysis and were correlated with age, disease severity, BMI, anti-RBD IgG titer, and convalescent plasma therapy received. Plot was created using R package corrplot. n = 3 for each patient. Significance was calculated using Wilcoxon test with Benjamini–Hochberg correction. *p < 0.05, **p < 0.01, ***p < 0.001. (B) Correlation analysis using linear regression of significant R value is shown. Data show correlation of CD56+ monocytes with BMI and age, CD4+ and CD8+ T cells with age, and CEP and neutrophils with age. Plots are created in ggplot2 using package ggpubr, method used = “lm”. n = 3 for each patient. Statistical significance is calculated using t test. Correlation value (R value) and p values are listed on each graph. (C) Heatmap showing different PBMC populations of each mild, moderate, and severe participant with their respective parameters in column annotations. Row annotation shows cell type. Graph was generated using pheatmap function.
CECs and CEPs were also directly correlated with disease severity, age, and BMI (Fig. 5A, 5B). Neutrophils were divided into two categories based on CD11b expression, and we found that CD11bint neutrophils were higher in patients with COVID-19 with hypertension and were directly correlated with disease severity, age, and BMI (Fig. 5A). However, CD11bhi neutrophils were neither positively nor negatively correlated with hypertension or with severe COVID-19. The heatmap in (Fig. 5C shows that patients with higher BMI and age had moderate and severe COVID-19, and 80% of them were hospitalized. These results suggest that age, BMI, and hypertension can dysregulate the immune response by upregulating the number of CD56+ monocytes, CD11bint neutrophils, and CECs and by reducing T cell populations, leading to severe COVID-19.
Discussion
COVID-19 severity varies widely among infected individuals. Obesity (6) and older age (59) are among the comorbidities most associated with mortality. SARS-CoV-2 is increasing in populations with low vaccination rates, and future variant strains may continue to threaten immunized individuals (60). Although strides have been made in SARS-CoV-2 immunopathology research, a need remains to qualify additional measurable immune parameters that correlate with severe disease. We studied a cohort of 56 SARS-CoV-2–infected adults with mild, moderate, and severe COVID-19 and 14 uninfected adults to further investigate differences in immune profiles in these groups by using high-dimensional spectral flow cytometry and unbiased cyto-feature engineering analysis. We performed comprehensive immune profiling of monocytes, T cells, B cells, neutrophils, and NK cells and combined this immune analysis with clinical data to understand the relationships between immune cell phenotypes and COVID-19 severity.
This analysis had several key findings. The first key observation is the presence of a unique activated and proliferating CD56+Ki67+IFN-γ+ monocyte population in patients with moderate and severe COVID-19. CD56 is a neural cell adhesion molecule expressed in various immune cell populations and has immune-stimulatory functions (44). CD56+ monocytes are elevated in cancer (47), Crohn’s disease (48), obesity (61), and rheumatoid arthritis (49) and have direct cytolytic activity toward malignant or infected cells. We found that in moderate and severe SARS-CoV-2 infections, CD56+ monocytes were activated and produced IFN-γ and Granzyme B. In moderate and severe COVID-19, CD56+ monocytes were elevated at 1–5 d after PCR+ and continued to increase at 11–16 d after PCR+ test. These increasingly activated monocytes are hypothesized to be contributing to a dysregulated immune response and cytokine release that leads to direct tissue damage, capillary leak, acute respiratory distress syndrome, and organ dysfunction. Activated CD56+ monocytes were directly correlated with older age and higher BMI, contributing to increased risk for development of severe COVID-19. The chronic hyperinflammation from adipose tissue secreted cytokines may put the obese at higher risk for uncontrolled immune response (62–66). Despite the development of immunosenescence, myeloid cells do not decline with age, and monocyte activation and dysregulation in the elderly are feasible (67).
The second key finding in this study was the diminished CD4+ and CD8+ T cells in patients with moderate and severe disease. Lymphopenia has previously been associated with severe SARS-CoV-2 infection, and our observation is consistent with other studies (32). Several hypotheses for causes of T cell lymphopenia have been proposed. First, T cell destruction could be incited by cytokine release and has been associated with increased IL-6 levels (68). Dysregulated inflammatory cytokines could lead to lymphocyte apoptosis and exhaustion (69). In this study, continued release of IFN-γ and Granzyme B by cytotoxic T cells was noted even after 11–16 d of infection. Second, enhanced vascular sequestration of lymphocytes could be associated with increased soluble VCAM-1 levels (70). Third, SARS-CoV-2 could directly infect T cells, leading to lymphocyte death (71). Finally, the use of corticosteroids could also affect T cell count (72). These mechanisms merit detailed evaluation to better understand the underlying cause of lymphopenia in patients with severe COVID-19 so that targeted immune-based therapeutics can be developed and applied.
Emerging evidence provides support that cardiovascular complications play a significant role in COVID-19 pathology. Given that SARS-CoV-2 infects cells via the angiotensin-converting enzyme 2 receptor and endothelial cells express high levels of angiotensin-converting enzyme 2 receptors (73), the virus can directly infect endothelial cells (74) and cause endothelial inflammation in patients with COVID-19. It also has been shown by Qin et al. (75) that SARS-CoV-2 infection causes endotheliitis and infection-mediated immune activation that contribute to the pathogenesis of severe COVID-19. In this study, we found increased CECs in the whole blood of patients with moderate and severe disease. This finding suggests that SARS-CoV-2 infection can alter the integrity of the vessel barrier and induce endothelial inflammation that can lead to a precoagulative state. We also have noticed increased endothelial progenitor cells in patients with severe COVID-19. Because of increased shedding of endothelial cells in blood, more endothelial progenitor cells are expected to be in circulation to replenish the endothelial integrity. Increased CECs were correlated to increased age, obesity, and diabetes. A better understanding of endothelial activation and dysfunction mechanisms in COVID-19 is needed, which will help in the early identification of individuals who are at risk for suffering from severe complications.
The outcome and disease severity of SARS-CoV-2 infection could be a result of dysregulated immune profiles, i.e., both too weak and too strong of an immune response can lead to severe COVID-19. Our data suggest that the combined effect of upregulated CD56+ monocytes, endothelial cells, neutrophils, decreased but hyperactivated T cells, and older age, high BMI, and hypertension can lead to severe outcomes of SARS-CoV-2 infection. With the presence of new variants of SARS-CoV-2 and insufficient vaccinations in developing countries, it is essential that we continue to work toward a better understanding of productive immune responses against SARS-CoV-2 and the immunopathological mechanisms underlying severe disease.
Limitations of the study
The study limitations included a small sample size for the study groups designated according to number of days after the SARS-CoV-2 PCR+ test result. The whole blood staining of CECs and neutrophils was not completed on all participants. Notably, the whole blood endothelial markers in the flow cytometric panel were introduced in October 2020, which is after patient enrollment began in July 2020. Another limitation of the study was the limited number of participants enrolled who were not infected with SARS-CoV-2 and the lack of a control group. Enrolling adults into a non–SARS-CoV-2–infected study arm that had similar comorbidities is a promising future direction that was challenging to accomplish before vaccination availability. Therefore, we have not included a traditional “control” group but rather designated the adults without SARS-CoV-2 infection as an “uninfected” group. Comparing our SARS-CoV-2–infected participants with matched participants with other respiratory viral infections, such as influenza, would have been more informative. Given the mask mandates and sanitization protocols throughout Colorado while this cohort was assembled, it was difficult to enroll patients with other infections. Therefore, the key objective from this study is to illustrate immune phenotype comparisons between mild/asymptomatic, moderate, and severe adult patients from northern Colorado.
This study was limited by the imbalance of pre-existing comorbidities in our asymptomatic/mild, moderate, and severe disease groups. The presence of comorbidities was not part of the eligibility criteria when enrolling participants with COVID-19, and we showed that participants with severe disease had more underlying conditions than those in the moderate and mild/asymptomatic groups. Our findings are consistent with established links between severe disease and the presence of pre-existing comorbidities. Further investigation is warranted to establish whether CD31+CD56+ cellular markers are implicated in severe disease that is irrespective of underlying conditions, and this would require a larger study with more study participants.
Acknowledgements
We have gratitude and appreciation for all patients/participants enrolled into the longitudinal Northern Colorado COVID-19 research biorepository. We thank Emma McGinnis and Sarah Mast from the CSU Human Performance Clinical Research Laboratory for phlebotomy assistance and the late Dr. Chris Allen, former co-director of the CSU Flow Cytometry Facility, for technical assistance alongside facility and instrumentation access during the pandemic. We also thank CSU BSL-3 core facilities with infrastructure to safely conduct this research.
Footnotes
This work was supported by Colorado State University (CSU) and Poudre Valley Hospital Foundation. Staff, blood collection, processing, materials, reagents, and flow cytometry were supported by E.P.R. in the CSU Department of Environmental and Radiological Health Sciences in collaboration with the Medical Center of the Rockies Trauma Research Department (J.D.). Additional support was granted from the University of Colorado Health Northern Foundation for the clinical recruitment in the local hospitals.
Conceptualization: E.P.R., M.H.-T., S.M.L., T.L.W., J.D., and T.S.D.; methodology: E.P.R., M.H.-T., T.S.D., S.M.L., T.L.W., B.A.B., S.S., and K.B.; data curation: T.S.D., T.L.W., S.M.L., B.A.B., S.S., K.B., and M.T.; data analysis and interpretation: T.S.D., S.M.L., T.L.W., K.M., M.H.-T., and E.P.R.; consulting statistician: K.M.; samples and patient data collection and processing: B.A.B., L.Z., O.A., J.D., K.M., S.S., K.B., S.M.L., T.S.D., G.E., and T.L.W.; manuscript writing – original draft and figures: T.S.D., S.M.L., M.H.-T., and E.P.R.; review and editing: T.L.W., S.M.L., K.M., B.A.B., M.H.-T., and E.P.R. The final draft of the manuscript has been read and accepted by each author.