Complement Alternative and Mannose-Binding Lectin Pathway Activation Is Associated With COVID-19 Mortality

Background The SARS-CoV-2 infection triggers excessive immune response resulting in increased levels of pro-inflammatory cytokines, endothelial injury, and intravascular coagulopathy. The complement system (CS) activation participates to this hyperinflammatory response. However, it is still unclear which activation pathways (classical, alternative, or lectin pathway) pilots the effector mechanisms that contribute to critical illness. To better understand the immune correlates of disease severity, we performed an analysis of CS activation pathways and components in samples collected from COVID-19 patients hospitalized in Grenoble Alpes University Hospital between 1 and 30 April 2020 and of their relationship with the clinical outcomes. Methods We conducted a retrospective, single-center study cohort in 74 hospitalized patients with RT-PCR-proven COVID-19. The functional activities of classical, alternative, and mannose-binding lectin (MBL) pathways and the antigenic levels of the individual components C1q, C4, C3, C5, Factor B, and MBL were measured in patients’ samples during hospital admission. Hierarchical clustering with the Ward method was performed in order to identify clusters of patients with similar characteristics of complement markers. Age was included in the model. Then, the clusters were compared with the patient clinical features: rate of intensive care unit (ICU) admission, corticoid treatment, oxygen requirement, and mortality. Results Four clusters were identified according to complement parameters. Among them, two clusters revealed remarkable profiles: in one cluster (n = 15), patients exhibited activation of alternative and lectin pathways and low antigenic levels of MBL, C4, C3, Factor B, and C5 compared to all the other clusters; this cluster had the higher proportion of patients who died (27%) and required oxygen support (80%) or ICU care (53%). In contrast, the second cluster (n = 19) presented inflammatory profile with high classical pathway activity and antigenic levels of complement components; a low proportion of patients required ICU care (26%) and no patient died in this group. Conclusion These findings argue in favor of prominent activation of the alternative and MBL complement pathways in severe COVID-19, but the spectrum of complement involvement seems to be heterogeneous requiring larger studies.


INTRODUCTION
The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection drives sustained inflammatory response considered to be a major cause of disease severity and death in patients with COVID-19 (1). Growing evidence suggests that the complement system (CS) activation instigates this dysregulated inflammatory reaction in COVID-19: elevated levels of the anaphylatoxin C5a have been reported to be proportional to disease severity (2); the membrane attack complex concentration has been linked with respiratory failure and systemic inflammation in infected patients (3); and deposits of mannose-binding lectin (MBL) and MBL-associated protease MASP-2 have been found in the microvasculature of critical patients with SARS CoV-2 infection (4). On the other hand, patients treated with complement blockers (anti-C5a mAb [eculizumab] or C3-inhibitor) exhibited a drop in inflammatory markers and significant clinical improvement (5)(6)(7)(8)(9). However, it is still not fully understood which of the three complement activation pathways (classical, alternative, or lectin pathway) drives the effector mechanisms that contribute to the tissue injury. To address these questions, we performed extensive analysis of CS activation pathways and components in samples collected from hospitalized COVID-19 patients and their relationship with the clinical outcomes.

Study Participants
This retrospective, single center study included 74 patients with RT-PCR-proven COVID-19 admitted to Grenoble Alpes University Hospital from April 1 to 30, 2020. Samples were collected during hospitalization in infectious/pneumology/ internal medicine department or intensive care unit (ICU) of our hospital. The study was performed in accordance with the Declaration of Helsinki, Good Clinical Practice guidelines, and CNIL (Commission Nationale de l'Informatique et des Liberteś) methodology reference. Patients were informed and consent was obtained, according to French law. Demographic, clinical characteristics (oxygen requirements, ICU admission, mortality, steroid treatment) and laboratory data were collected from electronic clinical records and included in an anonymized database. Patients were classified as severe on the basis of oxygen requirement (>2 L O 2 /min), ICU admission, limitation of therapeutic effort (LTE), and mortality, according to (10).

Complement Testing
Peripheral blood samples were collected in citrate anticoagulated or without anticoagulant tubes for hemolytic and functional assays or antigenic level measurement, respectively. Total hemolytic assays for classical pathway (CP, TH50c) and alternative pathway (AP, TH50a) were assessed as previously published (11); 100% lysis is defined by the TH50c/TH50a of the control sample. Reference values (TH50c: 86-126%; TH50a: 84-150%) were established by testing samples from 50 blood donors.
Determination of MBL protein concentration and function was realized by an enzyme-linked immunosorbent assay (ELISA) as described previously (12). The characterization of MBL protein expression deficiency was established by the combination of three assays: ELISA for antigen and functional MBL and hemolytic activity of C4 (C4H) (13) normal values of C1q confirmed the absence of CP activation. Low values of antigenic (<100 µg/L) (14) and functional MBL associated with a normal value of C4H defined patients with MBL protein expression deficiency. Low levels of antigenic and functional MBL associated with decreased value of C4H identified patients with MBL pathway activation. Reference values for MBL protein concentration and function and C4H were determined from 50 blood donors (MBL antigen: 30-3000 µg/L; MBL function: 35-115%; C4H: 70-130%).

Statistical Analysis
Statistical analysis was performed using hierarchical ascendant clustering (HAC) in order to identify groups of COVID-19 patients with similar characteristics ("clusters") in terms of complement variables: TH50c, TH50a, C1q, C4, C3, C5, Factor B, and MBL antigen. Age was included in the model (Supplementary Methods).
The biological significance of the clusters was analyzed by comparing the values of every complement parameter between the clusters. The ANOVA F-test was performed for variables with a Gaussian distribution, and the Kruskal-Wallis test for the variables with other distribution. Mean (standard deviation [sd]) or median (interquartile range [IQR]) were presented for the parametric or non-parametric variables, respectively. For the markers with a significant difference between the clusters, specific cluster by cluster tests were performed using Student or Wilcoxon tests to identify the cluster significantly different from the others. Post-hoc analysis with the Fisher's exact test was used to test specific difference between the clusters. Table 1 details the main demographic and clinical characteristics of our cohort. The median age of patients was 72 years (IQR: 62;82; range: 32-96); more than half of patients were men (58%). Of the 74 patients, 43 (58%) were severe, 8 (11%) died, 23 (31%) were admitted in the ICU, 3 (4%) were with LTE, and 23 (31%) required treatment by corticoids.

RESULTS
The results of complement proteins and activation pathways analysis performed in samples collected during hospitalization are summarized in Table 2.
Four clusters of individuals were identified in the studied cohort by HAC analysis according to the complement data (Supplementary Figure S1). The comparison between the clusters (cluster 2 against others and/or cluster 4 against others) is presented in Figure 1 and Table 3. There was no statistically significant difference among the four clusters for sex, age, CRP, severity, ICU admission, O 2 requirement, or corticoid treatment ( Table 4).

DISCUSSION
SARS-CoV-2 infection triggers an innate immune response including CS activation which is a key weapon both implicated in disease resolution and organ damage depending on the time of infection (15).
Recent studies addressing the role of complement in the pathogenesis of COVID-19 have shown a relationship between respiratory failure, intravascular coagulopathy, and complement overactivation in COVID-19 patients (3,16). There are several lines of evidence for local deposition of complement proteins and activation products in lung, skin, and other tissues showing activation of the three pathways, CP, LP, and AP (17)(18)(19)(20)(21). Furthermore, systemic complement activation and consumption were related to severe COVID-19 and predictive of in-hospital mortality (22). However, despite in vitro lines of evidence suggesting that the SARS-CoV-2 spike proteins activate the AP (23), it remains incompletely understood which complement activation pathways contribute to critical illness in COVID-19.
Concerning the possible involvement of MBL in coronavirus infection, it has been described so far for SARS-CoV in several   (24,25). More recently, Ali and al showed binding of LP recognition molecules to S-and N-proteins of SARS-CoV-2, as also robust LP activation on the surface of HEK 293 cells expressing SARS-CoV-2 S protein (26).
One of the limitation of the study is that it reports data about the original variant of the virus, while the delta variant is now a majority among the infected patients. The crucial mutations leading to delta variant concern the S1 subunit (intimately involved in the initiation of infection) for which we have no data concerning its molecular interaction with complement components. However, the replication rate of delta strain is much higher. It would therefore be interesting to compare our results with samples collected from patients infected with the delta variant of the virus.
Using an original, unsupervised statistical approach by hierarchical clustering on complement and clinical parameters, this study reveals for the first time the association between AP and LP activation, and the mortality in COVID-19 patients (cluster 4). Our data are in line with Ma et al. showing increased AP components in COVID-19 patients with worse prognosis (27) and with Sinkovits et al. relating significant association between AP activity and COVID-19 severity (22). Furthermore, our results provide additional evidence for an association between MBL pathway activation and mortality, supported by the data of Eriksson et al. (4). In contrast, Sinkovits et al. reported that the LP activity showed no difference between severity groups (22). Our data are consistent with recent reports describing deposits of MBL and MASP2 in affected tissues of COVID-19 patients and in vitro MBL pathway activation by recombinant SARS-CoV2 proteins (17,28). Collectively, these findings are consistent with the CS implication in the pathogenesis of severe COVID-19 infection. As consequences of unrestrained complement activation, the strong pro-inflammatory C5a-C5aR axis promotes neutrophil/monocyte infiltration and the "cytokine storm" driving lung inflammation and injury, responsible for complications in hospitalized COVID-19 patients (2).
If complement activation is evident from our results and associated with the severity of the disease, the spectrum of involvement of the complement cascade in COVID-19 seems to be heterogeneous and depending on patients: notably, complement activation could be deleterious in some ones and expression of the severity of the disease in other ones. A recent review summarizes the current knowledge about modulating complement cascade as therapeutic approach in COVID-19 patients (29). In brief, despite nonconclusive studies, the available data suggested favorable outcomes in a small number of patients with severe COVID-19 treated either with C1 inhibitor, MASP-2 monoclonal antibodies, compstatin-based complement C3 inhibitor, anti-C5 drugs, or C5a-C5aR1 antagonists. Clinical trials of complement inhibitors in COVID-19 are ongoing (NCT04288713, NCT04414631, NCT04395456). However, awaiting final results from the clinical trials, the potential benefits from complement inhibition in COVID-19 remain to be elucidated.   Interestingly, we found high incidence of MBL expression deficiency in patients of our cohort (16%, established as described in Methods), all clusters combined. Further studies based on MBL genotyping would be of interest to support biochemical data. MBL deficiency is fairly common, affecting approximately 5-10% of individuals and usually associated with increased susceptibility to bacterial infections of the upper respiratory in young children (30).
Our results highlight the dichotomous nature of the complement MBL pathway: on one hand, MBL appears to contribute to the pathogenesis of disease because it mediates complement activation that is related to clinical deterioration of patients; on the other hand, MBL could have a protective role against SARS-CoV-2 infection by promoting phagocytosis and virus lysis or neutralization. Further in vitro studies using pseudoviral particles or the SARS-CoV-2 virus are necessary to check the latter hypothesis.
In summary, our study suggests that alternative and lectin pathways assessment might be useful as biomarker of disease severity. Extensive investigations of complement pathways have to be performed on a larger cohort of patients with SARS-CoV-2 infection to help rationalize therapeutic choices.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding author.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by Commission Nationale de l'Informatique et des Libertés (CNIL). Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.