Elevated frequencies of CD14+HLA-DRlo/neg MDSCs in COVID-19 patients

Background: The immune responses, hyper-inflammation or immunosuppression, may be closely related to COVID-19 progression. We aimed to evaluate the changes of frequency of CD14+HLA-DRlo/neg MDSCs, a population of cells with potent immunosuppressive capacity, in COVID-19 patients. Methods: The levels of CD14+HLA-DRlo/neg MDSCs were determined by flow cytometry in 27 COVID-19 patients, and their association with clinical characteristics and laboratory data were analyzed. Results: The frequency of CD14+HLA-DRlo/neg MDSCs was elevated in COVID-19 patients, particularly severe patients. A follow-up comparison revealed a decline of CD14+HLA-DRlo/neg MDSCs percentages in most patients 1 day after testing negative for SARS-CoV-2 nucleic acid, but the levels of CD14+HLA-DRlo/neg MDSCs were still greater than 50.0% in 3 ICU patients 4-10 days after negative SARS-CoV-2 results. Elevated frequency of CD14+HLA-DRlo/neg MDSCs was positively correlated with oropharyngeal viral loads and length of hospital stay, while negatively correlated with lymphocyte counts and serum albumin. Moreover, strong correlations were observed between the frequency of CD14+HLA-DRlo/neg MDSCs and T cell subsets, NK cell counts, and B cell percentages. The frequency of CD14+HLA-DRlo/neg MDSCs could be used as a predictor of COVID-19 severity. Conclusions: A high frequency of CD14+HLA-DRlo/neg MDSCs, especially in severe patients, may indicate an immunoparalysis status and could be a predictor of disease severity and prognosis.


INTRODUCTION
SARS-CoV-2 infection now has a global resonance, with the disease termed COVID-19. As of December 29, 2020, the most recent data confirm more than eighty-one million cases in 191 countries/regions, leading to over 1,770,000 deaths [1]. The illness severity of COVID-19 can range from mild to critical. Although most patients exhibit mild or moderate symptoms, it is expected that approximately 19% of patients will develop severe or critical disease, constituting a group with a high risk of death [2]. Some patients with mild symptoms in early stage will suddenly deteriorate to severe respiratory failure, requiring intubation and mechanical ventilation. These patients have a high risk of death, even reaching 60% mortality rate [3]. Hence, management of the disease progression of severe patients is essential for COVID-19 prevention and control. A large number of studies are currently concerned about AGING the "cytokine storm" induced by COVID-19 [4]; however, over the course of severe illness, patients may show moderate to low fever, or even no obvious fever [5], and have dramatically reduced numbers of circulating T lymphocytes and NK cells [6], suggesting these patients may have developed immunosuppression similar to that seen in sepsis caused by bacteria [7]. When transitioning to an immunosuppressive state, monocytes are highly sensitive and become inactive, leading to a phase called "immunoparalysis" [8]. This is defined as a decrease in the level of HLA-DR expression on monocytes, measured by the frequency of one kind of myeloid-derived suppressor cells (MDSCs), defined as CD14 + HLA-DR lo/neg MDSCs [9]. MDSCs are a group of immature cells that can mediate suppression of the immune response. In mice, MDSCs are characterized as CD11b + Gr-1 + Ly6G + granulocyte MDSCs (G-MDSCs) and CD11b + Gr-1 + Ly6C + monocyte MDSCs (M-MDSCs) [10]. Nevertheless, the definition and role of G-MDSCs and M-MDSCs in human remain elusive due to the lack of specific phenotypic markers [11,12]. Currently, G-MDSCs are mainly described as CD11b + CD33 + CD15 + HLA-DR + , whereas M-MDSCs are mainly defined as CD11b + CD33 + CD14 + HLA-DR lo/neg or CD14 + HLA-DR lo/neg in human. Filipazzi P et al. identified and characterized for first time the new subset of MDSCs, CD14 + HLA-DR lo/neg , in peripheral blood of melanoma patients with high suppressive of lymphocyte functions [13]. In clinical practice, we noticed that many COVID-19 patients admitted to ICU showed a significant decrease in T lymphocyte subsets, suggesting that they were in an immunosuppressive state. Therefore, we are interested in the difference between CD14 + HLA-DR lo/neg MDSC frequency in ICU patients and non-ICU patients, and the changes before and after negative SARS-CoV-2 nucleic acid tests.

Baseline characteristics
27 COVID-19 patients were included in our study. The median age was 58 years (IQR, . Of these, 17 (62.96%) were men and 10 (37.04%) were women (Table 1). Diabetes (14.81%) and hypertension (25.93%) were the two most common comorbidities in the patient group. Of the 27 cases, 8 patients (29.62%) were admitted to the ICU, and 19 patients were managed outside of the ICU (non-ICU). The levels of C-reactive protein, prothrombin time, lactate dehydrogenase and D-dimer were higher in ICU patients than non-ICU patients. Compared with non-ICU cases, the level of albumin was reduced in ICU patients (P=0.003). Moreover, duration of hospital stay was significantly longer for the ICU group compared to the non-ICU group (P=0.003).

Changes in CD14 + HLA-DR lo/neg MDSCs before and after negative SARS-CoV-2 nucleic acid testing
To further investigate the clinical significance of CD14 + HLA-DR lo/neg MDSCs in COVID-19, we performed an evaluation of 15 COVID-19 patients (7 ICU patients and 8 non-ICU patients) before and 1 day after negative SARS-CoV-2 nucleic acid results, and in 7 patients (5 ICU patients and 2 non-ICU patients) before, 1 day after, and 4-10 days after testing negative for the virus. The data showed that there was a reduction in the levels of CD14 + HLA-DR lo/neg MDSCs 1 day after testing negative for SARS-CoV-2 when compared to the levels during active disease (P=0.0002) ( Figure 2). Although the MDSC levels decreased 4-10 days after negative nucleic acid testing, the difference was not statistically significant (P>0.05). For 3 patients admitted to ICU, represented by the data points inside the dotted frame in Figure 3, high frequencies (>50.0%) of CD14 + HLA-DR lo/neg MDSCs were still observed 4-10 days after testing negative for viral nucleic acid.

Elevated frequencies of CD14 + HLA-DR lo/neg MDSCs were correlated with neutrophil, lymphocyte, albumin, D-dimer, CRP, and length of hospitalization
The correlations between CD14 + HLA-DR lo/neg MDSCs frequencies and baseline clinical and laboratory indexes were determined and analyzed for COVID-19 patients. As shown in Figure 3, MDSCs frequencies were positively correlated with neutrophil count (weak), D-dimer (weak), C-reactive protein, and hospital stay, while inversely associated with lymphocyte count and serum albumin (weak). No obvious correlation was observed between MDSCs and other indicators.

Elevated frequency of CD14 + HLA-DR lo/neg MDSCs is correlated with SARS-CoV-2 viral RNA load
We analyzed the relationship between frequencies of CD14 + HLA-DR lo/neg MDSCs and viral RNA loads, and the results indicated patients with COVID-19 with more severe disease requiring ICU admission had high viral RNA loads when compared to non-ICU patients. And a negative correlation (weak, r=-0.4723) was observed between CD14 + HLA-DR lo/neg MDSCs percentages and viral loads ( Figure 5).

DISCUSSION
The treatment of severe COVID-19 patients has been a focus in the prevention and management of SARS-CoV-2 infection due to the higher mortality [14]. However, little is known about the detailed factors governing the development or remission of severe disease. To date, the most important predictors of COVID-19 severity were related to the activation or suppression of the host's AGING immune response [7]. Immune system dysregulation has been observed and is believed to be related to the severity of COVID-19 [15]. Zheng et al. recently discovered that increased levels of T cell exhaustion and reduced diversity of T cell function may be indicators of disease progression in COVID-19 patients [16]. Similarly, our previous study found severe COVID-19 patients had significantly reduced lymphocyte subset counts compared to mild-to-moderate patients [17]. However, the mechanism underlying T cell lymphopenia remains unknown. High levels of antiinflammatory mediators, such as IL-10 and IL-4, have also been reported in COVID-19 patients, particularly ICU patients [18]. These results suggest that dysregulation of the immune response may present in the form of immuno-suppression as well as a cytokine storm. Indeed, it has been reported that COVID-19 is associated with states of both immunodeficiency and hyperinflammation [19]. Furthermore, even without significant hypotension, many severe or critically ill COVID-19 patients develop typical clinical manifestations of shock. Severe metabolic acidosis and impaired liver and kidney function caused by SARS-CoV-2 infection indicates that viral sepsis may occur in severe patients [20]. The low expression of HLA-DR on monocytes in critically ill patients is the most widely recognized sign of innate immune suppression, and the most studied biomarker of sepsis-induced immune suppression, termed "immunoparalysis" [21]. Monocytes with low or no HLA-DR expression are often referred to   AGING as CD14 + HLA-DR lo/neg MDSCs. HLA-DR is an MHC class II glycoprotein expressed on antigen presenting cells, which presents peptides derived from antigens to T cell receptors, resulting in T cell activation [9,22]. Therefore, the ability of CD14 + HLA-DR lo/neg MDSCs to present antigens to T cells is diminished, and many previous studies have demonstrated that MDSCs have a strong immunosuppressive effect [23]. Volk et al. defined immunoparalysis in sepsis as CD14 + HLA-DR lo/neg MDSCs being >70% of the monocyte pool [8]. Studies have shown that the decrease of HLA-DR expression on monocytes is a vital predictor of forthcoming sepsis [24]. In this study, we found that COVID-19 patients showed significant increases in the frequency of CD14 + HLA-DR lo/neg MDSCs when compared to HCs, especially in patients admitted to ICU. This may signify immunosuppression in severe COVID-19 patients, and also suggests that while focusing on the treatment of cytokine storms clinically, we should also pay attention to possible immunosuppression in patients to prevent secondary infection. Using single-cell RNA sequencing and single-cell proteomics of whole blood and peripheral blood mononuclear cells, Schulte-Schrepping J et al. identified a high proportion of HLA-DR lo monocytes clusters in severe COVID-19 patients, and low expression of HLA-DR is an established marker of monocyte dysfunction as well as immunosuppression in sepsis [25]. Furthermore, monocytes isolated from COVID-19 patients exhibited impaired cytokine secretion upon LPS stimulation, particularly severe cases, indicating an immunosuppressive phenotype of monocytes in COVID-19 patients. This is consistent with the results of the present study. In addition, they also found that HLA-DR hi monocytes were predominant in mild patients, while HLA-DR lo monocytes were present mainly in severe patients, with HLA-DR lo CD163 hi monocytes in the early stage and HLA-DR lo S100A hi monocytes at the later stage. The present study found frequencies of HLA-DR hi monocytes were still relatively high after testing negative for SARS-CoV-2 nucleic acid. Unfortunately, what is not clear is which kind of HLA-DR lo monocytes is the main form at this phase. In addition, elucidating the underlying mechanism of the increased frequency of MDSCs in COVID-19 patients requires further study. High CD14 + HLA-DR lo/neg MDSC frequencies have also been reported in other patients with viral infection, such as hepatitis B [26] and hepatitis C [27]. In patients with H7N9 infection, Diao et al. found that the expression of HLA-DR on CD14 + cells was significantly lower in patients with severe than mild disease and, importantly, was related to the severity of H7N9 infection [28].
According to "Notice on printing and distributing the diagnosis and treatment plan of pneumonia with new coronavirus infection (trial version 7)", two consecutive negative RT-PCR test results at least 24 h apart is one of the criteria for patient discharge, which to some extent indicates recovery from infection. We found that the frequency of  [29]. This is consistent with the trend of M-MDSCs in the present study, suggesting that both MDSCs may be involved in the immune abnormal process in COVID-19 patients. In addition, for patients with severe disease, G-MDSCs decreased in the later stage, yet were still remained higher level compared with those in healthy population. The current dynamic follow-up analysis also revealed that the proportion of M-MDSCs was higher than in healthy subjects even 4-10 days post-infection. These results suggested that persistent abnormal MDSCs, including G-MDSCs and M-MDSCs, may be accompanied by a persistent state of "immunoparalysis". Chen et al. reported that in severe H7N9 patients, the total lymphocyte count and percentages of different lymphocyte subtypes returned to normal levels 1 month post-infection, while 3 months were needed for the normalization of CD14 + HLA-DR lo/neg MDSCs [30]. Unfortunately, our study did not record when the CD14 + HLA-DR lo/neg MDSC frequency of COVID-19 patients returned to normal levels.
Lymphopenia is a common sign of COVID-19 that has been observed in more than 70% of patients, particularly severe patients [31,32]; however the mechanism underlying this observation remains unclear. Interestingly, the frequency of CD14 + HLA-DR lo/neg MDSCs showed negative correlation to lymphocyte counts, suggesting that MDSCs may be involved in the lymphopenia detected in COVID-19. The relationship between malnutrition and infections is a vicious cycle, whereby not only are the risk and severity of infections increased by malnutrition, but malnutrition may also result from infection [33]. During infection inducedhyperinflammation, serum albumin is reduced. Previous studies have shown that hypoalbuminemia can indicate AGING a high risk of death for patients with sepsis [34]. We also found a notable decrease of albumin and increase of CRP in COVID-19 patients, especially in ICU patients, which is consistent with a recent study [35]. The correlations between MDSCs and albumin and CRP imply that a high MDSC percentage may be an indicator of the systematic inflammatory and nutritional state, and the significant correlation between CD14 + HLA-DR lo/neg MDSC percentage and length of hospital stay implies that this may be a prognostic predictor for COVID-19 patients.
Studies have reported that the number of total T cells, and CD4 + and CD8 + T cells, were dramatically reduced in COVID-19 patients, especially in patients requiring ICU care, and this was negatively correlated with patient survival [36]. It is well established that CD14 + HLA-DR lo/neg MDSCs mediate immunosuppression in tumors by suppressing the proliferation of T cells [37,38]. Thus, what role does the expanded MDSC population play in COVID-19? We found that MDSC frequency showed a strong inverse correlation with counts of T cell subsets and NK cells. These results indicate that the decrease in lymphocyte subsets in COVID-19 may be partly due to the expansion of MDSCs in response to hyperinflammation. Further research needs to be carried out to explore the underlying mechanism of this observation. Moreover, a positive correlation was found between CD14 + HLA-DR lo/neg MDSCs frequencies and SARS-CoV-2 viral RNA loads, indicating SARS-CoV-2 infection may cause immunosuppression through inducing expansion of CD14 + HLA-DR lo/neg MDSCs. Indeed, high levels of MDSCs were found in patients experiencing viral infections, such as dengue, and correlated with virus viral load and severity [39].

Data collection
Demographic information and comorbidities were collected. Blood cells count, biochemical indexes, coagulation parameters and inflammatory marker were also collected. The thresholds selected for laboratory measurements were based on the normal range set by each hospital. The viral loads from nasopharyngeal were estimated by the ΔCt method (Ctsample-Ctref).

Statistical analysis
Statistical analyses were performed using SPSS 23.0 and GraphPad Prism 8.0. Categorical variables were described using percentages, and continuous variables were described as mean±SD or median (interquartile range, IQR). When the data were normally distributed, the means of continuous variables were compared using the independent t-test; otherwise, the Mann-Whitney U test was used. Categorical variables were compared using the χ 2 test. Likewise, the Pearson method was used for correlation analysis. A value of P <0.05 was considered statistically significant.

ACKNOWLEDGMENTS
We would like to thank all the patients involved in the study, and to thank Editage (https://www. editage.cn) for English language editing.

CONFLICTS OF INTEREST
All the authors state that there are no conflicts of interest related to this study.