Immunological imprint of COVID‐19 on human peripheral blood leukocyte populations

Abstract Background SARS‐CoV‐2 has triggered a pandemic that is now claiming many lives. Several studies have investigated cellular immune responses in COVID‐19‐infected patients during disease but little is known regarding a possible protracted impact of COVID‐19 on the adaptive and innate immune system in COVID‐19 convalescent patients. Methods We used multiparametric flow cytometry to analyze whole peripheral blood samples and determined SARS‐CoV‐2‐specific antibody levels against the S‐protein, its RBD‐subunit, and viral nucleocapsid in a cohort of COVID‐19 convalescent patients who had mild disease ~10 weeks after infection (n = 109) and healthy control subjects (n = 98). Furthermore, we correlated immunological changes with clinical and demographic parameters. Results Even ten weeks after disease COVID‐19 convalescent patients had fewer neutrophils, while their cytotoxic CD8+ T cells were activated, reflected as higher HLA‐DR and CD38 expression. Multiparametric regression analyses showed that in COVID‐19‐infected patients both CD3+CD4+ and CD3+CD8+ effector memory cells were higher, while CD25+Foxp3+ T regulatory cells were lower. In addition, both transitional B cell and plasmablast levels were significantly elevated in COVID‐19‐infected patients. Fever (duration, level) correlated with numbers of central memory CD4+ T cells and anti‐S and anti‐RBD, but not anti‐NC antibody levels. Moreover, a “young immunological age” as determined by numbers of CD3+CD45RA+CD62L+CD31+ recent thymic emigrants was associated with a loss of sense of taste and/or smell. Conclusion Acute SARS‐CoV‐2 infection leaves protracted beneficial (ie, activation of T cells) and potentially harmful (ie, reduction of neutrophils) imprints in the cellular immune system in addition to induction of specific antibody responses.

Current data indicate a long and relatively mild initial course of the disease after SARS-CoV-2 infection, which may be followed by a severe course after a median time of 8 days from the first symptoms.
The mild period may be followed by a period of breathlessness (approximately on day 8), which may turn into acute respiratory distress syndrome (ARDS) on day 9, followed by admission to intensive care on day 10.5. [15][16][17] Acute infection with SARS-CoV-2 can be proven by detecting nucleic acids from the virus in nasopharyngeal swabs or lavages. 18 After the acute infection, SARS-CoV-2-specific antibodies 9,16 are usually formed, which coincides with the clearance of the virus from the body and a delayed response might be associated with mortality. 19 However, SARS-CoV-2-specific antibodies have varying specificities and titers for different antigens/epitopes and it has been shown that antibodies from convalescents after a mild COVID-19 infection cannot always prevent the virus from binding to ACE2 and may therefore not have a protective effect. 9 Similarly, several studies have investigated the effect(s) of a primary SARS-CoV-2 infection on leukocyte populations in the body in general, and on T and B lymphocyte subsets in particular. Also, the SARS-CoV-2-specific T cell response has been investigated by several research groups. 17,[20][21][22][23][24][25][26][27][28][29] Notably, memory CD4 + T cells have been suggested to mediate protective immunity against the previously circulating CoV-strains SARS-CoV 30 and MERS, 31 respectively, and CD8 + memory T cell responses have been observed in SARS-CoV convalescent patients years after primary infection. 32 However, regarding infection with the novel coronavirus SARS-CoV-2 recent reports have revealed conflicting results with one study (analyzing PB lymphocyte subsets in 44 patients) pointing toward reduced T memory and T regulatory cell populations in severe COVID-19-infected patients, suggesting that immune dysregulation potentially leads to aggravated inflammatory responses in patients. 28 In contrast, another study provided evidence for segregation between memory T cells from patients with acute severe or acute moderate COVID-19 disease and the development of memory T cells from recently convalescent individuals (42-58 days after infection) and healthy controls. 33 The latter study found only very moderate changes in a group of 40 COVID-19 convalescent patients who underwent a mild COVID-19 disease course (49-64 days after disease onset).
However, the available studies on cellular immune responses were mainly conducted in patients suffering from acute COVID-19 but little is known about the protracted effects of COVID-19 on cellular immune responses. Consequently, there is a need for systematic studies analyzing COVID-19 effects on cellular immune responses in cohorts of well documented COVID-19 convalescent patients, in particular, for those who had experienced mild disease constituting the majority of cases, to understand if and how the infection continues to stimulate and/or perturb the immune system of the convalescent host. It is also unclear how possible changes in peripheral leukocyte/lymphocyte compartments correlate with SARS-CoV-2-specific antibody levels and/or clinical symptoms.
Accordingly, the aims of this study were to identify the immunological imprint of primary COVID-19 infection on peripheral blood (PB) cell populations and antibody levels. We were especially interested whether distinct changes in blood cell or antibody parameters would separate COVID-19 convalescent patients from healthy subjects and whether we could define parameters which may be associated with the disease course, disease duration, pre-existing medical conditions, regular medications, or demographic parameters. Thus, we performed here a systematic study to investigate the distribution of leukocyte subsets in peripheral blood with a special interest in T and B lymphocyte subpopulations of COVID-19 convalescent patients 10 weeks after mild SARS-CoV-2 infection and compared results with a large, age-matched, healthy control group, which was recruited in parallel and was negative for SARS-CoV-2 antibodies and had a negative SARS-CoV-2 rtPCR at the time of venipuncture. Therefore, and in order to unequivocally determine leukocyte, T and B lymphocyte populations we performed whole-blood multiparametric flow cytometry on COVID-19-infected patient and control peripheral blood (PB) samples. Furthermore, we determined SARS-CoV-2-specific antibody levels against S protein, its subunit the RBD protein, and the viral nucleocapsid protein and analyzed whether thymic and/or bone marrow output were affected in COVID-19-infected patients who underwent a mild disease course.

| Patients, control subjects, and trial conduct
Between May 11, 2020, and July 2, 2020, 109 patients diagnosed with COVID-19 disease 10 weeks previously (71.2 ± 16.5 days) were enrolled into this study. The 109 patients had rtPCR-confirmed (n = 92, 84.4%) and/or SARS-CoV-2 antibody-confirmed (Elecsys ® Anti-SARS-CoV-2 assay Roche) (n = 108 tested, n = 107 positive, 99.1%) COVID-19 disease. 34 In parallel, 98 healthy control subjects, who were reportedly asymptomatic for the last 10 weeks and who were SARS-CoV-2 negative by certified SARS-CoV-2 antibody test (Elecsys ® Anti-SARS-CoV-2 assay Roche) and had a negative rtPCR test for SARS-CoV-2 at the time of venipuncture, were enrolled into the study. Similar to the COVID-19-infected patients also the HC collective had different co-morbidities as specified in Table 1. All patients gave their written informed consent in accordance with the Declaration of Helsinki.
The study was approved by the Ethics Committee of the Medical University of Vienna (EK No.: 1302/2020). Venous blood was drawn from all subjects and was either EDTA-anticoagulated (for flow cytometric analyses and determination of lymphocyte emigration rates from thymus using TREC and bone marrow using KREC technology), heparin-anticoagulated (for cryopreservation of PBMC), or silicon dioxide coagulated (to obtain serum for determining specific antibodies).

| Immunophenotyping by multiparametric flow cytometry
Immunophenotyping was performed by using optimal concentrations of directly conjugated monoclonal antibodies (Table S1)   Additional methods can be found in the Appendix 1 in the online repository to this article.

| Clinical characterization of COVID-19 convalescent patients
Between May 11, 2020, and July 2, 2020, 109 COVID-19 convalescent patients and 98 healthy control subjects were enrolled into this retrospective study. The group of COVID-19 convalescent patients and the group of healthy control subjects were well balanced regarding gender, age, height, and weight (

| Primary SARS-CoV-2 infection induces protracted reduction of neutrophil counts and reveals increased numbers of activated cytotoxic T cells in the circulation
First, we determined overall peripheral blood (PB) leukocyte subpopulations in the COVID-19-infected patient collective and in healthy control subjects. Remarkably, the COVID-19 convalescent patients showed significantly less leukocytes compared with the healthy control subjects (6384 ± 1677 × 10 6 /L vs 7006 ± 1961 × 10 6 /L; P = .0156), which was due to highly significantly lower neutrophil counts (4111 ± 1477 × 10 6 /L vs 4693 ± 1588 × 10 6 /L; P = .0033) (Figure 1; Figure S1 and Table S3). Neutropenia is frequently associated with  Days of hospitalization 5.9 ± 5.8 March-May, that is, when the first wave of the COVID-19 pandemic was at its peak in Austria, no capacity for additional PCR tests was available, which represents a limitation of this study.

| Levels of transitional B cells and plasmablasts are higher in the blood of COVID-19 convalescent patients while B memory compartments are unaffected
Next, we determined B-cell subset distribution and found that major  Table S5). This finding was interesting, given the observed and above-described gross distortions of T memory cell populations in COVID-19 convalescent patients. However, CD19 + CD38 high IgM + transitional B cells were significantly higher in COVID-19 convalescent patients when compared to healthy control subjects (5.00 ± 3.07% vs 3.57 ± 2.30%; P = .0020) (Figure 3; Figures S6, S7 and Table S5). Similarly, CD19 + CD38 + IgMplasmablasts were both absolutely (2.5 ± 1.8 × 10 6 /L vs 1.6 ± 1.3 × 10 6 /L; P = .0015) and relatively (1.56 ± 0.97% vs 1.04 ± 0.95%; P < .0001 higher in COVID-19-infected patients as compared to healthy control subjects (Figure 3; Figures S6, S7 and Table S5). That the current bone marrow output of B-cell precursors was not depressed in COVID-19 convalescent patients could be demonstrated by determining kappa recombination excision circles (KRECs), which were not different from the healthy control population (Figure 3).

| Associations between clinical parameters and changes in blood cells and serum parameters
Next, we performed univariate analyses in order to investigate whether blood cell parameters and/or SARS-CoV-2-specific antibody levels (S-, RBD-, and NC-protein) would correlate with clinical symptoms ( Figure 4A), co-morbidities, and premedications of COVID-19-infected patients. Most notably, we found that convalescent patients who had more severe fever during their disease had significantly higher relative numbers of CD3 + CD4 + CD45RO + CCR7 + central memory helper T cells ( Figure 4B) in their blood (P = .0003). In addition, relative numbers of CD3 + CD4 + CD45RO + CCR7 + central memory helper T cells were significantly associated with chills (P = .0030; data not shown), and with the number of fever days (P = .0045; Figure 4C). These findings were corroborated by significant inverse correlations observed between highest fever and relative and absolute numbers of naïve CD3 + CD4 + CD45RO -CCR7 + T cells (P = .0025 and P = .0034, respectively; not shown).
Notably, patients who had fever during their disease also presented with significantly more activated CD19 + CD21 low CD38 low B cells 41 ( Figure 4D; P = .0045).
Moreover, the number of fever days and the presence of fever in general (not shown), correlated well with both the levels of anti-S-protein IgG (P = .0002) and anti-RBD IgG (P = .0166) as determined by ELISA ( Figure 4E, F). In contrast, anti-NC antibodies only moderately correlated with fever/highest temperature (not shown) but not duration of fever ( Figure 4G).
Another very interesting finding was the fact that patients suffering from loss of taste/smell had significantly more CD3 + CD45RA + CD62L + CD31 + RTEs in their blood (P = 0.0172) ( Figure 4H). While CD3 + CD45RA + CD62L + CD31 + recent thymic emigrant cells positively correlate with TREC numbers (P < .0001) and negatively with biological age (P < .0001), this association was only evident when the age was determined "immunologically," that is, by multicolor flow cytometry but not upon determination of birth age.
This finding may indicate that individuals with a "young(er) immune system" are more likely to suffer from loss of taste/smell when infected by SARS-CoV-2. That loss of taste/smell represents indeed a separate clinical factor in COVID-19-infected patients, which is clearly different from, for example, the fever/coughing/fatigue-or the arthralgia/myalgia-factor was confirmed by principal component analysis (PCA), reducing clinical symptoms to factors best explaining the variance of variables, followed by rotated component matrix relation determination (Varimax-Rotation) using the collected clinical symptom categories (Table 2). These analyses revealed seven factors defined by certain symptoms (factor 1: fever, chills, rigor, fatigue, cough; factor 2: running nose, nasal congestion, sore throat, sneeze; factor 3: arthralgia, myalgia; factor 4: conjunctivitis, other GI problems; factor 5: pneumonia, shortness of breath, wheezing; factor 6: vomiting, nausea, headache, diarrhea; convalescent patients, however, surprisingly not with disease duration and severity (fever days, fever height) although age is frequently associated with a higher mortality risk. 42 Notably, a total of 45% of COVID-19 convalescent patients took regular medications (Table S2) with 16 out of the 109 patients reporting intake of ACE-or angiotensin receptor (AT) blockers (candesartan, lisinopril, enalapril, ramipril, irbesartan, valsartan), that means, substances which have been shown to lead to increased ACE2 receptor expression in cardiac tissue in preclinical models previously. 43 While ACE/AT inhibitor intake positively correlated with the age of patients (P < .0001), it apparently did not correlate with the disease course, for example, duration of fever (P = .7316) or of clinical symptoms (P = .6131). Of significance, also 11 subjects in the control group took ACE/AT inhibitors as regular medication; thus, there was no significant difference between the patient and control group regarding ACE/AT inhibitor consumption.  53 Moreover, cytotoxic T cells are also proficient to express the canonical co-stimulatory molecules CD86 and CD80, which are essential for the priming and activation of effector T cells. 53 In addition, a previous report has shown that T cell-T-cell synapses have a major role in the generation of protective CD8 + T-cell memory since conjugated cells critically polarize each other toward the secretion of interferon-gamma (IFN-gamma). 54 We here also found increased CD38 expression on both CD4 + and CD8 + T cells of COVID-19 convalescent patients. In fact, CD38

| D ISCUSS I ON
(adenosine diphosphate (ADP) ribosyl cyclase) marks T cells with increased cytotoxic capability 38,39 and enhanced ability to produce cytokines. 55 Furthermore, CD38 endows T cells with improved interaction with endothelial cells via CD31, 56 which has been shown to lead to improved adhesion of lymphocytes to endothelium. 57 Whether this also leads to enhanced extravasation of T cells is an open question, but it is tempting to speculate that CD38 + T cells might have an improved capability to target sites of inflammation.
Antibodies and T cells are the two main antigen-specific effector systems for resolving viral infections. Notably, both T cell subsets contribute to viral clearance. 58,59 In fact, anti-viral CD4 + T cell responses are important for optimal antibody and CD8 + T cell responses. Thus, it is not entirely unexpected that we found significantly elevated numbers of effector memory CD4 + CD45RA + CD127 + T cells in COVID-19 convalescent patients. What is, however, surprising is the level and degree of maintenance of these cells in patients with a mild COVID-19 disease course, because we could easily identify their high numbers even without applying antigen-specific detection methods such as tetramer staining. One mechanism for the sustained expansion of these effector cells seems to be the mitigation of T regulatory responses, which is reflected by significantly lower CD25 + Foxp3 + CD127 -T cells in convalescents as compared to healthy controls ( Figure 2 and Table S4).
Whether the CD4 + effector cells have mere helper function for the clonal expansion of the SARS-CoV-2-specific B cells or may also exert cytotoxic cellular programs remains to be shown in future analyses. Nevertheless, besides the expanded CD4 + T cells, we also detected a significant expansion of the CD8 + effector memory pool. Moreover, two patients clearly reacted in the NC-test (36.0 and 74.1, respectively), but no anti-S-protein or anti-RBD protein antibodies were detected. Together, these findings indicate that the spectrum and magnitude of antibody specificities in COVID-19 convalescent patients may considerably vary. In this context, it will be interesting to perform a comprehensive analysis of the SARS-CoV-2-specific antibody response using a broad spectrum of virus-derived antigens and epitopes and testing of several isotypes and IgG subclasses by multiplex tests such as chips containing micro-arrayed components similar as developed for other viral infections. 64,65 The fact that more than 25% of the convalescent patients did not mount a relevant anti-RBD antibody response fits to our earlier finding that up to 50% of convalescent patients do not mount antibodies which can inhibit the docking of the virus via RBD to its receptor, ACE2. 9 Also in this study, we found that ~50%-60%  Our study thus not only reports definitive evidence for a protracted immunological imprint of COVID-19 on human peripheral leukocyte populations but also raises several new aspects of COVID-19 which require further studies.

ACK N OWLED G EM ENTS
This study was supported by grants from the "Medizinisch- Guentcheva for their help regarding recruitment and administration of study subjects. We are indebted to all individuals who participated in our study.