Heightened Circulating Interferon-Inducible Chemokines, and Activated Pro-Cytolytic Th1-Cell Phenotype Features Covid-19 Aggravation in the Second Week of Illness

Covid-19 features a delayed onset of critical illness occurring approximately one week from the beginning of symptoms, which corresponds to the bridging of innate and adaptive immunity. We reasoned that the immune events occurring at the turning point of disease might mark the direction toward pathogenic versus protective inflammatory responses. Subjects with either severe (s; PaO2/FiO2 ratio <200) or mild (m; PaO2/FiO2 ratio>300) Covid-19 were enrolled. A range of chemokines and cytokines as well as reactive oxygen species (ROS) were measured in plasma. Dendritic and NK cell frequency, monocyte and B-/T-cell phenotype and SARS-CoV-2-specific T-cell responses were assessed in PBMC. Twenty mCovid-19 and 20 sCovid-19 individuals were studied. sCovid-19 patients displayed higher non-classical monocytes, plasma chemokines (CXCL8, CXCL9, CXCL10), cytokines (IL-6, IL-10), and ROS versus mCovid-19. sCovid-19 also showed significantly increased activated CD38+HLA-DR+ T-lymphocyte, and granzyme-B+/perforin+ pro-cytolytic T-cells. All Covid-19 patients showed SARS-CoV-2 specific-T-cell response with a predominance of Th1 bi- or trifunctional IFN-γ/IL-2/TNF-α-expressing CD4+, while no difference according to disease severity was observed. Severe Covid-19 features heightened circulating IFN-inducible chemokines and activated pro-cytolytic Th1 cell phenotype in the second week of illness, yet SARS-CoV-2-specific responses are similar to that of mild illness. Altogether, our observations suggest Th1 polarization coupled to higher cytolytic profile in sCovid-19 as correlate of disease pathogenesis and as potential targets to be investigated in the roadmap to therapy and vaccine development.

INTRODUCTION SARS-CoV-2 is the etiologic agent of Coronavirus disease 2019  which may feature interstitial pneumonia leading to severe respiratory distress and death (1). While literature findings point to the presence of older age, neutrophilia, organ, and coagulation dysfunction in subjects with critical disease (2)(3)(4), scant data exist on the immune correlates of Covid-19 progression.
Literature evidence suggests that individuals with severe SARS-CoV-2 infection may have a "cytokine storm syndrome", characterized by increased levels of cytokines and chemokines (5)(6)(7), crucial mediators of the adaptive immune response.
Lymphopenia also appears to feature SARS-Cov-2 infection (1,7,8) with low CD4+, CD8+, B and NK counts (7,9). In line with these findings, functional exhaustion of cytotoxic lymphocytes in adult individuals with Covid-19, presenting lower intracellular cytokine expression compared to healthy controls, has also been described (3). Of note, significantly lower T-and B-cell counts as well as skewing of T-cell maturation were detected in subjects with severe Covid-19 compared to those with mild disease (6,9), suggesting that diverse adaptive immunity phenotypes may feature Covid-19 severity. In contrast, a limited number of studies report conflicting results on the functional profile of T-lymphocytes in SARS-CoV-2-infected subjects with different clinical presentation (6,7).
Clinical aggravation of Covid-19 occurs approximately one week from the onset of illness (5,10,11), which corresponds to the temporal bridging of the innate and adaptive immune responses. In the animal model, CoV-specific T-cells are not only necessary and sufficient for virus clearance but also account for protection from clinical disease (12), as they dampen overactive innate immune responses (13,14), thus limiting further damage to the host (12,15). This observation has led us to hypothesize that the immune characteristics at this specific time-point in the course of SARS-CoV-2 infection may represent a watershed for the clinical outcome of the disease.
In the attempt to outline an immune signature of severe and mild Covid-19, we conducted an extensive analysis of innate and adaptive immune parameters of SARS-CoV-2-infected subjects at the defining moment of illness.

Study Population
We included in the study 40 individuals hospitalized with ascertained acute SARS-CoV-2 infection (positive nasopharyngeal swab) presenting with either severe (n = 20, PaO2/ FiO2 ratio< 200) or mild (n = 20, PaO2/FiO2 ratio>300) Covid-19. At hospitalization, demographic, and clinical characteristics were recorded from electronic clinical charts. All enrolled patients provided written informed consent according to the Ethical Committee of our institution (no. 2020/ST/049). Following informed consent, peripheral blood samples were collected from all study participants for plasma and peripheral blood mononuclear cell (PBMC) separation which were stored for laboratory analyses.
Immune parameters were compared to those of 10 healthy controls (HC) in archived laboratory samples.
Briefly, the samples were diluted 1 to 4 with assay diluent and incubated for 3 h with the relative capture beads and with the human chemokine detection reagent. Samples were washed in a wash buffer resuspended in 0.3 ml of the same buffer and acquired by flow cytometry. Data were acquired on (BD) FACSCanto II flow-cytometer and analyzed by FCAP v3 array software.

Soluble ROS Quantification
Plasma levels of ROS were measured by an enzyme-linked immunosorbent assay (ELISA; LSBio), according to the manufacturer's instructions.

Statistics
Twenty individuals with severe and 20 with mild Covid-19 (sCovid-19 and mCovid-19, respectively) were enrolled. Aside for a non-significant trend to older age in sCovid patients, no significant differences were registered between groups in terms of sex, comorbidities, type, and duration of Covid-19-related symptoms ( Table 1). The two groups were comparable in terms of pulmonary radiologic findings and medical therapy, except for a higher proportion of subjects with sCovid-19 on supplemental oxygen upon hospital admittance and more frequent treatment with Continuous Positive Airway Pressure (CPAP) and mechanical ventilation during hospitalization ( Table 1).
Samples for immune investigations were collected at a median (IQR) of 7 (3-8) days from symptoms onset, with no differences between mCovid and sCovid patients ( Table 1).
We next investigated T-cell maturation, finding significant differences according to disease severity, with sCovid-19 patients displaying lower EM cells (p = 0.03; Figure 3C and p = 0.02; Figure  3D) and higher granzyme B-and perforin-producing CD4+ ( Figure  3E). A similar trend was observed in CD8+ compartment, albeit not reaching statistical significance ( Figure 3F).
Covid-19 subjects as a whole presented higher CD4+ and CD8+ IL-17A, IL-4, IFN-g, IL-2, TNF-a cytokine production in response to SARS-CoV-2-specific stimulation compared to HC from archived materials who were not exposed to SARS-CoV-2 infection ( Figure  4). This suggests a specific T-cell response in Covid-19 patients, while we could not detect differences in intracellular cytokine production between mCovid-19 and sCovid-19 subjects (Figure 4).
Although about 80% of Covid-19 patients display a benign clinical phenotype, up to 20% of the patients can develop rapidly progressing respiratory failure. While several clinical and epidemiological factors have been associated with poor outcome, specific immunologic aspects featuring the worst clinical outcome are still elusive. Clinical experience with Covid-19 has demonstrated that the second week of illness seemingly represents a turning point in disease, suggesting that the immune events occurring at this phase of the infection might mark the direction toward pathogenic versus protective inflammatory responses.
With this idea in mind we comparatively assessed innate and adaptive immunity measured at approximately one week from the onset of symptoms in a cohort of Covid-19 patients featuring severe versus milder illness.
In patients developing severe Covid-19, we demonstrate: i) elevated inflammatory chemokines, cytokines and ROS positively associating with neutrophilia and pro-inflammatory monocytes; ii) T-cell immune phenotype characterized by increased activated, granzyme/perforin-producing T-lymphocytes, and reduced effector-memory cells; iii) evidence of SARS-CoV-2-specific intracellular cytokine production, with a predominance of Th1 CD4+ T-cells, similar to patients with milder disease.
The direct comparison of patients with severe versus moderate disease revealed highest inflammatory chemokines including MCP-1/CCL2, IP-10/CXCL10, IL-8/CXCL8, as well as cytokines in subjects with critical illness, correlating with both neutrophilia and increased circulating pro-inflammatory CD14+ CD16++ monocytes with reduced HLA-DR surface expression.
In particular, while finding elevated circulating IL-6 and decreased HLA-DR expression on circulating monocytes in sCovid-19, we describe less relevant changes in other proinflammatory cytokines, in line with the predominant role of IL-6 as driver of Covid-19 hyperinflammatory response, immune dysregulation and respiratory failure (19)(20)(21).
The behavior of other pro-inflammatory cytokines in Covid-19 has proven more erratic across the literature, with discordant findings being published (6,22,23). In particular, the failure to detect differences in IL-1b and TNF-a might reflect receptor antagonism or differential cytokine concentrations in diseased tissue and peripheral blood (24).
Having ascertained a pattern of "cytokine storm" (6,22,(25)(26)(27), more evident in patients with severe disease, and given the crucial interplay between innate and adaptive immunity, we next investigated T-cell responses, aiming to shed light on the contribution of adaptive immunity on Covid-19 course.
To date, little is known about the antiviral T-cell responses in Covid-19. In animal models of coronavirus infection, adaptive T-cell immunity has proven essential in tempering the innate immune response, in turn mitigating immunopathology. Indeed, upon acute coronavirus infection, T-cell deficient mice mounted an exaggerated innate immune response with high levels of circulating pro-inflammatory cytokines, resulting in rapid lethality, proving that an unleashed innate response together with the lack of antiviral-specific responses can be a direct cause of death (12,13,28).
Less is known about T-cell response in human coronavirus, whether it contributes to disease progression or recovery (25,29). In MERS, SARS, and Covid-19, patients with severe/fatal outcomes present progressive lymphopenia and neutrophilia peaking at approximately days 7-10 from symptom onset, while healing patients efficiently recover physiologic lymphocyte/neutrophil counts, suggesting a crossover between innate and adaptive immunity in dictating disease outcome, with adaptive immunity on the one hand controlling the infection and on the other hand slowing immune pathology (6).
Accordingly, in our patient cohort, despite similar demographic and pre-existing comorbidities, as well as analogous lung damage, we show reduced neutrophil and increased lymphocyte counts in moderately versus severely ill individuals at an average of 7 days from symptom onset. Because lymphocyte activation/exhaustion has been suggested in Covid-19 adult patients (3,22), we next characterized T-cell immunophenotype and function according to disease severity, with particular focus on SARS-Cov-2-specific response. As compared to milder disease, sCovid patients show highly activated CD4+/CD8+ T-cells, with reduced effector-memory and raised pro-cytolytic phenotypes. Given the paradigm of T-cell differentiation described in humans featuring naïve!centralmemory!effector-memory!terminally-differentiated (30), our findings of higher proportion of activated T-cells, with lower effector-memory cells and higher cytolytic potential in sversus mCovid patients would altogether suggest higher T-lymphocyte activation upon acute SARS-CoV-2 infection in individuals with a severe disease course, resulting in the continuous T-cell differentiation, overall resulting in a net reduction of the effector-memory at the advantage of cytolytic T-cell phenotype.
Having shown that despite their bad prognosis, sCovid patients display more vigorous T-lymphocyte engagement, activation and function, we next sought to investigate SARS-CoV-2-specific response according to disease severity.
Interestingly, we demonstrate the presence of T-cell reactivity to SARS-CoV-2 S-, M-, N-overlapping antigen pool, at higher magnitude within the CD4+ compartment, confirming the pivotal role of CD4+ in the control over SARS-CoV infection, confirming data on co-dominant M, spike and N-specific CD4+ response in 100% of Covid-19 convalescent patients (16), as well as both humans and animal data correlating disease severity and CD4+ responses in the course of SARS (25,31).
Further detailing T-cell response revealed virus-specific IL-2producing CD4+ in the whole Covid-19 cohort, with up to 75% of the patients displaying bi-or trifunctional INF-g/IL-2/TNF-aexpressing CD4+ and similar virus-specific CD8+ functionality despite lower frequency and magnitude. Because polyfunctional T-cell response has been associated with better immunity versus several infections (32)(33)(34), our data indicate an ongoing Th1polarized response to SARS-Cov-2, in agreement with recent data showing predominant Th1 responses in convalescent (16) as well as in ARDS ICU patient cohort (20).
Unexpectedly however, despite a more activated/pro-cytolytic T-lymphocyte asset, we failed to detect any difference in virusspecific intracellular cytokine response according to disease severity. In particular, despite data suggesting Th2 polarization as correlate of immunopathology (25), about half of our patients presented IL-4-expressing CD4+ irrespective of disease outcome, with a reasonable proportion of IL-4 and Th1 cytokine-coexpressing T-cells to possibly suggest a Th0 profile in this stage of the disease. Given recent data showing a predominant Th1 response in a small but well-defined cohort of convalescent uncomplicated non-hospitalized Covid-19 cases (16), it will be interesting to longitudinally follow up the fate of Th1/Th2 ratio in later disease phases as well as in patients recovering from complicated versus uncomplicated disease.
Likewise, our finding of a relevant proportion of IL-2/TNF-a/ IL-17 co-expressing T-cells suggest the activation of IL-17mediated immune pathway whose role in neutrophil recruitment and immune regulation in Covid-19 will need to be further investigated.
Collectively, our data provide some hints to better understand Covid-19 pathogenesis. Together, the findings of similar plasma IFN-g, higher circulating IFN-stimulated chemokines CXCL9 and CXCL10 (35) in sCovid, and an overall heightened Th1 virus-specific ex vivo T-cell response, suggest Th1 polarization. The lack of difference in IFN-g plasmatic levels might reflect a more vigorous response in lymphoid organs and diseased tissues (24) that might therefore fail to be captured by peripheral blood cytokine assessment, in accordance with post-mortem data showing mononuclear cell accumulation in the lungs (36).
Interesting speculations derive from the investigation of Tlymphocyte phenotype and function. While the prevalent activated/cytolytic T-cell phenotype would indicate vigorous Tcell activation and function to neutralize the infection, the lack of a difference in virus-specific T-cell response in sCovid versus mCovid was somehow unexpected and contrasts previous data in SARS (25). Higher T-cell activation and differentiation in the face of non-efficacious virus-specific response have been described in other models of viral infections such as HIV (37,38), where ongoing non-virus-specific immune activation has been long proven a major driver of disease progression even after viral abatement by antiretroviral therapy (39,40).
As a caveat in the interpretation, it must be noted that all but two Covid-19 patients in our cohort developed pneumonia: it will be interesting to assess SARS-Cov-2 specific T-cell responses in pauci-symptomatic patients without pneumonia. However, 14/15 patients who died were within the sCovid groups, so we can assume that the immune picture that we describe realistically captures the immune events contributing to the most severe immune pathology and clinical prognosis.
Limitations of this study include the patient's size and the lack of a longitudinal assessment that was not possible given the high mortality within the sCovid group. In analogy to what was described in other models of infectious diseases where disease progression has been associated with different profiles of virusspecific T-cell functionality (34,41), a detailed longitudinal profiling of virus-specific intracellular cytokine asset in larger patients cohorts with different disease phenotype will further inform on the immune features and tempo of disease progression and severity.
While the hectic run to anti-SARS-Cov2 therapy and vaccine is ongoing, our study adds to the body of literature aimed at broadening the knowledge about T-cell responses (42).
While no current targeted treatment is available thus far, combined antiviral, antimalarials, corticosteroids, anti-inflammatory molecules, convalescent plasma, and anticoagulant approaches are being used and investigated for the treatment of Covid-19 (43). Among immunomodulants, biologics interfering with the cytokine storm, mainly the IL-6/IL-6R axis, as well as JAK-STAT signaling inhibitors (i.e. bariticitinib, ruxolitinib, fedtratinib) have raised expectations, prompting pilot studies and clinical trials (44,45). By showing elevated IFN-inducible chemokines as well as IL-6 at the end of the first week of disease in patients developing severe versus milder Covid-19, our findings are informative on the rationale to the therapeutic exploitation of JAK/STAT inhibitors or for cytokine targeting antibodies in patients who develop severe Covid-19. Because JAK/STAT activation is triggered by a wide range of cytokines, our data support its broader inhibition as valid therapeutic candidate to finely modulate the pro-inflammatory cascade downstream of single cytokine signaling, possibly redirecting the disastrous inflammatory response toward disease containment. Likewise, the most thorough understanding adaptive immunity fingerprints of protective immunity versus immune-mediated enhancement of SARS-CoV-2 pathology will be essential to the evaluation and design of a vaccine. By demonstrating similar virusspecific T-cell response, our findings comfort on the presence of M, S and N T-cell response as correlate of acute Covid-19 irrespective of disease severity that will need to be further profiled in the course of disease and convalescence to further inform the requisites of candidate Covid-19 vaccine (16,20).

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by ASST Santi Paolo and Carlo ethics committee. Patients/participants provided their written informed consent to participate in this study.

AUTHOR CONTRIBUTIONS
CT designed the study, analyzed, and interpreted the data, designed the figures, and wrote the manuscript. EC designed the study, performed experiments, analyzed, and interpreted the data designed the figures, and wrote the manuscript. MG performed the experiments. RB interpreted the data and wrote the manuscript. Ad'A helped with interpreting the results and edited the manuscript. GM conceived and designed the study, interpreted the data and wrote the manuscript. All authors contributed to the article and approved the submitted version.