Myeloid-derived Suppressor Cell and Non-classical Monocyte Frequencies Are Directly Related to the Immunological Status of Patients With Active Tuberculosis

Alterations of myeloid cell populations have been reported in patients with tuberculosis (TB). Since myeloid-derived suppressor cells (MDSC) and non-classical monocytes (CD14 + CD16 + ) seem to play an important role in TB, we studied the relationship between these cells and the immunological status of patients. TB patients were classied as high responders (HR-TB) or low responders (LR-TB) according to their T cell responses against a whole cell lysate of Mycobacterium tuberculosis (Mtb-Ag). Thus, LR-TB, individuals with severe disease, display a weaker immune response to Mtb compare to HR-TB, subjects with strong immunity against the bacteria. We observed that LR-TB presented higher percentages of CD14 + CD16 + monocytes as compared to HR-TB and healthy donors (HD). Moreover, monocyte-like (M-MDSC) and polymorphonuclear-like MDSC (PMN-MDSC) were increased in TB patients as compared to HD. Furthermore, the proportion of M-MDSC of TB patients inversely correlated with the levels of IFN-γ released after Mtb-Ag stimulation. We also found that LR-TB displayed the highest levels of circulating M-MDSC. Interestingly, in LR-TB, the frequencies of non-classical monocytes and M-MDSC were restored after only three weeks of anti-TB treatment. Together, our ndings show a direct relationship between the immunological status of TB patients and the levels of different circulating myeloid cell populations.

After infection, Mtb interacts with different cells of both, innate and adaptive immune compartments.
These cells play an important role in the modulation and the development ofthe pathology 4 . An e cient host protection against Mtb infection is associated with the induction, activation and proliferation of Th1 and Th17 cells [5][6][7] , whom promote the release of cytokines such as IL-2, TNF-α, IFN-γ, and IL-17; and the activation of effector monocytes [5][6][7] . Moreover, it has been previously demonstrated that reduced IFN-γ production is a marker of severe disease 8 and de ciencies in the IL-12-IFN-γ-STAT1 signaling pathway lead to the dissemination of mycobacterial infections 3,9 . Nevertheless, how Mtb is able to evade host immune surveillance and persist, particularly inside myeloid cells, is not yet fully elucidated. In ammatory myeloid cells are key players in the pathophysiology of tuberculosis 10 . In fact, it has been proposed that the phenotype of the populations of myeloid cells involved in early granuloma formation may in uence substantially the progression of TB and the outcome of the infection 7,11,12 . Many studies have shown that mycobacterial infections can also affect the differentiation of progenitors and immature myeloid cells to mature and polarized phagocytes 11 . Mtb can alter the activation of recruitedmacrophages and shape thecytokines and chemokinesthat they produce 13 . Furthermore, we have previously demonstrated that IFN-γ and IL17A differentially regulate the autophagy process in Mtb-infected monocytes derived from TB patients in correlation with the severity of the disease 5,14 . Phenotypically and functionally different subsets of monocytes were identi ed based on the relative expression of CD14 (co-receptor for toll-like receptor 4) and CD16 (Fc gamma receptor IIIa) 15 . Besides, it has been described that the ability of circulating non-classical monocytes (CD14 + CD16 + )to differentiate into dendritic cells and induce T-cell activation is decreased in TB patients 16  Although no speci c markers have been described for MDSC identi cation 17 , human M-MDSC were shown to express CD14, CD33, and CD11b with a lack of CD15 and low or no HLA-DR expression [20][21][22][23] .
The accumulation of MDSC during mycobacterial infections was rst reported in mice models 24,25 . Furthermore, the expansion of MDSCs was observed bothin blood of patients with pulmonary TB, and in pleural uid of extra-pulmonary pleural TB 7 . Besides, it was also described a clear decrease of circulating MDSC frequencies at the end of successful anti-TB treatments 7 . However, further characterization of these cells considering the immunological status of the individuals is currently needed.
Due to the important role of monocytes and MDSC during Mtb infection, in this study, we evaluated the levels of these circulating myeloid cells in TB patients.
We found a differential expansion pro le of monocytes and MDSC accordingto the immunological status of TB patients. We could observe that low responder (LR-TB) TB patients presented higher percentages of circulating CD14 + CD16 + monocytes and M-MDSC as compared tohigh responder (HR-TB) patientsand healthy donors (HD). Furthermore, levels of CD14 + CD16 + monocytes and M-MDSC were restored to normal after the rst three weeks of anti-TB treatment. Therefore, wecould show a direct relationship between the immunological status of TB patients and the frequencies of circulating CD14 + CD16 + monocytes and M-MDSC. Taking into account the crucial role of cellular immunity during TB, these ndings could be important pieces to better understand Mtb infection and disease.

Results
Healthy BCG-vaccinated donors (HD) and pulmonaryTBpatientswere analyzed. Patients with active disease were classi ed ashigh responders(HR-TB) or low responders (LR-TB) based on their in vitro lymphocyte responses against Mtb-Ag.HR-TB patients display signi cant proliferative responses, IFN-γ production and increased percentage of signaling lymphocyte activation molecule (SLAMF1) positive cells after the exposure to the antigen. On the other hand, LR-TBexhibit low proliferative responses, IFN-γ production and SLAMF1-positive cells 26 . Therefore, LR-TB present a more severe TB disease. Clinical and immunological parameters of participating TB patientsare shown inTable 1.

Analysis of circulating monocyte subsets frequencies in TB patients and healthy donors
Previous reports have shown that CD16 + monocytes are expanded in TB infection in association with the severity of the disease 27,28 . Then, we initially decided to analyze the subsets of circulating blood monocytes in our study population based on the expression of CD14 and CD16 molecules. TB patients showed higher percentage of circulating CD14 + CD16 + cells than HD (%CD14 + CD16 + cells (TB) = 15.4± 3.5; %CD14 + CD16 + cells (HD) = 8.1 ± 0.9. Mean ± SEM. **P < 0.01, Mann-Whitney U test). Furthermore, the percentage of CD14 + CD16 + monocytes was also analyzed in the two groups of TB patients classi ed as previously described. As we described, this classi cation is based on the immunological status of patients. No perturbation of classical monocytes (CD14 + CD16 -) was observed when comparing HR-TB, LR-TB and HD (Fig. 1a), but, surprisingly, we found that LR-TB patients presented the highest percentages of CD14 + CD16 + circulating cells (Fig. 1b). On the other hand, HR-TB patients showed signi cantly lower levels of these non-classical monocytes than LR-TB and similar frequencies as compared to those observed in HD (Fig. 1b). Nevertheless, the analysis of the clinical parameters showed that there were no differences in monocytes count between HR-TB and LR-TB. No differences were also observed in the number of circulating monocytes between patients classi ed as mild, moderate or severe according to radiological lesions, nor between patients classi ed according to acid-fast bacilli in sputum smear (see Supplementary Fig. S1 online). In view of our results, and to evaluate the potential use of CD14 + CD16 + cells frequencies to discriminate individuals LR-TB from HR-TB, we performed a ROC analysis (Fig. 1e). From this study, signi cantresults were obtained (AUC = 0.8750; P<0.01; 95% CI: 0.68 -1.07), demonstrating that the percentages of circulating non-classical monocytes allow differentiating between these patients with distinct immunological status (Cut-off = 8.31%; Sensitivity = 81.8%; Speci city = 87.5%).

Analysis of circulating MDSC in TB patients and healthy donors
Then, we decided to compare the frequencies of circulating MDSC in TB patients and HD. PBMC were obtained by Ficoll-Hypaque gradient and then MDSC percentages were analyzed by ow cytometry.
As previous studies have suggested that M-MDSCare more potent immune suppressors than PMN-MDSC 29 , we then investigated whether M-MDSC and PMN-MDSC subsets presented any correlation with IFN-γ levels produced by T cells after Mtb-Ag stimulation. For this, we stimulated PBMC with the whole cell lysateof the bacteria for 48h and then, IFN-γ levels were measured in culture supernatants. The performed analysis showed that the percentages of M-MDSC were inversely correlated with IFN-γ levels in TB patients. However, no correlation between the percentages of PMN-MDSC and the levels of IFN-γ were detected ( Fig. 2e and 2f). These results suggested an immunosuppressive role of M-MDSC on TB patients´ lymphocytes.
To further investigate the clinical signi cance of MDSC in TB disease, we also studied whether LR-TB, HR-TB and HD presented different levels of circulating M-MDSC and PMN-MDSC. The FACS analysis showed signi cant higher levels of M-MDSC in LR-TB in comparison with HR-TB and HD individuals (Fig. 3a). On the contrary, HR-TB patients presented the highest levels of PMN-MDSC compared to TB-LR and HD (Fig.   3b). Furthermore, signi cant differences in the production of INF-γ and the proliferation of lymphocytes of LR-TB and HR-TB patients were detected after Mtb-Ag stimulation ( Fig. 3c and 3d). Correlation analysis showed that the percentage of M-MDSC from HR-TB patients negatively correlated with IFN-γ levels from Mtb-Ag stimulated PBMC. However, there was no correlation between M-MDSC frequencies and IFN-γ production by cells from LR-TB patients (Fig. 3e).
In order to evaluate the potential use of circulating M-MDSC frequencies in discriminating LR-TB vs. HR-TB, we next also performed a ROC analysis (Fig. 3f)  Thus, since due to the fact that LR-TB patients present severe TB disease 6,26 , we decided to study whether few weeks of anti-TB treatment affected the frequencies of these circulating cell subsets in these patients. We then compared percentages of blood circulating PMN-MDSC, M-MDSC and classical and non-classical monocytes at the beginning of the anti-TB treatment (blood samples taken during the rst weekof chemotherapy) and after three weeks (blood samples taken between 14 -21 days later). No differences in the percentage of classical monocytes (CD14 + CD16 -) were observed after 14-21 days of anti-TB treatment. However, we observed a signi cant decrease in non-classical monocytes(CD14 + CD16 + ) levels after this short period of treatment ( Fig. 4a y 4b). In addition, we analyzed whether the percentages of MDSC subsets were also modi ed during the rst days of chemotherapy. Interestingly, we observed a signi cant reduction of the frequency of M-MDSC in LR-TB patients (Fig. 4c). In contrast, the percentage of PMN-MDSC was not affected by the anti-TB treatment during this period of time (Fig. 4d). Importantly, IFN-γ production and the proliferation index signi cantly augmented during these three weeks of anti-TB treatment (Fig. 4e and 4f).
Together, our present ndings indicate that non-classical monocytes and M-MDSC would play an important role in TB patients with severe disease. Moreover, the accumulation of these subsets of cells is rapidly reversed after a few days of anti-TB treatment, suggesting that non-classical monocytes and M-MDSC could serve as a marker of TB progression or even treatment monitoring.

Discussion
A spectrum of stages caused by Mtb infection leads to identify that some patients can control the bacterial infection and others cannot. The IFN-γ producing Th1 cells are essential to control mycobacterial replication 31,32 . Indeed, reduced IFN-γ production is a well-known marker of disease severity 33 . In our study population, HR-TB and LR-TB patients were identi ed based on their T cell responses to Mtb-Ag 26 . In spite of that, Th1 and Th17 cells alone do not explain the resistance/susceptibility to infection and disease 6,34 , suggesting that other actors might be required during the immune regulation of TB. Myeloid cells are a heterogeneous group of cells that plays a major role in the regulation of immune responses in many pathological conditions. However, the imbalance of myeloid cells during human TB has been poorly studied. Therefore, we aimed to investigate the populations of monocyte and MDSC cells during human active TB disease. In the present report, we identify an increase in different circulating myeloid cell populations in severe TB patients. Both, immature myeloid suppressor cells M-MDSC and PMN-MDSC were augmented in blood from TB patients. Furthermore, we found that patients that showing weak or no IFN-γ responses to Mtb-Ag presented higher percentages of non-classical CD14 + CD16 + monocytes as compared to HR-TB and HD (Fig. 1). Moreover, LR-TB individuals presented the highest percentages of circulating M-MDSC.On the contrary, HR-TB individuals displayed the highest percentages of PMN-MDSC (Fig. 3). We further observed that the frequencies of CD14 + CD16 + monocytes and M-MDSC return to healthy control levels after a short time of anti-TB treatment (Fig. 4). Therefore, we showed that these cell subsets contribute to a better characterization of the immunological pro le of TB patients and could be new targets for the development of host-directed therapies.
Previously, we observed that monocytes from LR-TB individuals were unable to activate autophagy process through IL17A at least in part because of a defect in the MAPK1/3 signaling pathway. In contrast, both IFN-γ and IL17A increased the levels of autophagy in patients with strong immunity to Mtb 5 . Our present results extended those ndings, demonstrating that LR-TB patients, individuals whose severity clinical and immunological parameters have been already described 6,26 , showed higher proportions of non-classical monocytes and M-MDSC in peripheral blood as compared to HR-TB patients. In fact, we founded an inverse correlation between IFN-γ released in response to Mtb-Ag stimulation and the levels of circulating M-MDSC. Thus, the proportion of circulating CD14 + CD16 + monocytes and M-MDSC in patients with active TB might be re ecting the severity of the disease.
It was suggested that MDSC are closely correlated with the progression of TB infection in humans 7,30 . In accordance with these reports, we found that MDSC are expanded in peripheral blood of active TB patients (Fig. 2). Moreover, we show for the rst time that there is a characteristic expansion pro le of M-MDSC and PMN-MDSC in TB patients related to their immunological status. Our results in PBMC con rmed that the majority of HR-TB patients present augmented percentages of PMN-MDSC. Furthermore, patients with weak immunity to Mtb presented the highest levels of circulating M-MDSC. However, an inverse correlation between the percentages of circulating M-MDSC and IFN-γ levels was only observed in HR-TB patients (Fig. 3). These would suggest that M-MDSC from HR-TB present more important suppressive effects than M-MDSC from LR-TB. These results are in agreement with those reported previously where a functional correlation between MDSC and TB infection was also Additionally, it has been previously shown that pharmacological therapy reduce the accumulationof myeloid cell population 7,28,30 . Sanchez et al. demonstrated that the expression of CD14, HLA-DR and CD36 wasdecreased in monocytes of TB patients 28 ; nevertheless, normal expression of these molecules was restored during anti-TB treatment. Furthermore, in TB patients, MDSC population was reduced at the end of anti-TB therapy 7,30 . However, it is important to mention that in our study we could observe for the rst time to our knowledge, a restoration in the circulating levels of CD14 + CD16 + monocytes and M-MDSC after a short term (three weeks) of anti-TB treatment. In this short period of time we did not observe differences in percentages ofcirculating classical monocytes (CD14 + CD16 − ) or in PMN-MDSC levels. Therefore, considering previous published data and our present results we hypothesize that some myeloid cell populations could serve as a treatment-response marker during TB. In addition, by providing information about the immunological status of the patient, the circulating levels of CD14 + CD16 + monocytesand M-MDSC would allow to know about the severity of the active disease in infected individuals.
In summary, our ndingsshow that accumulation of circulating non-classical monocytes and monocyticlike MDSC during pulmonary tuberculosis infection reveals the host immunological status. Furthermore, we could describe that the anti-TB treatment rapidly restore the normal levels of these cells. The recruit of these expanded myeloid populations in individuals with a weakened immune response could exacerbate the detrimental pro-in ammatory responses in infected lung tissue and the prognosis of tuberculosis.

Methods
Study Subjects. Patients with TB were diagnosed at Dr. F. Muñiz or P. Piñero Hospitals (Buenos Aires, Argentina) based on clinical and radiological data together with the identi cation of acid-fast bacilli (AFB) in sputum. First peripheral blood samples were collected before one week of anti-tuberculosis therapy administration and second samples were obtained between 14 and 21days of anti-TB treatment. All patients had received anti-tuberculosis (anti-TB) regular therapy for drug sensitive Mtb strains. Bacillus Calmette-Guerin (BCG) vaccinated healthy control individuals lacking a history of TB (HD) participated in this study.
Peripheral blood was collected in heparinized tubes from each participant after obtaining a written informed consent for the collection of samples and the subsequent analysis. All methods were carried out in accordance with relevant guidelines and regulations.
The protocols conducted in this work were approved by the Ethics Committees of Muñiz and Piñero Hospitals.
Exclusion criteria and classi cation of patients. The exclusion criteria were carried out as previously described 36 . Brie y, individuals participating of the study were 18 -60 years old and had no history of diseases affecting the immune system, such as HIV infection, treatment with immunosuppressive drugs,a recent diagnosis of cancer, hepatic or renal disease, pregnancy, or positive serology for other viral (e.g., hepatitis A, B or C), or bacterial (e.g., leprosy, syphilis) infections. Subjects with anticoagulant medication orbleeding disorders that might be at an increased risk of bleeding during the procedure of obtaining the sample were excluded from the study.
Individuals with latent infection were excluded from the present study by using the QuantiFERON-TB Gold Plus kit (Qiagen, Germany, USA).
TB Patients were classi ed as high responders (HR-TB) or low responders (LR-TB), based on their in vitro lymphocyte responses to a whole cell lysate of M. tuberculosis (Mtb-Ag) as previously described 26 . Brie y, HR-TB patients are individuals displaying signi cant proliferative responses, IFN-γ production and an increased percentage of SLAMF1 positive cells after Mtb-Ag stimulation; whereas LR-TB patients exhibit low proliferative responses, IFN-γ release and SLAMF1 positive cells. LR-TB patients had more severe pulmonary disease compared with HR individuals.
In the studied population, no differences regarding age distribution or sex were found ( Table 2).
Cell preparation and reagents. PBMC were isolated by centrifugation on Ficoll-Hypaque (Amersham Biosciences, NJ, USA). Flow cytometry assays were performed after PBMC were washed with PBS plus 1% BSA and 0.1% NaN 3 and resuspended in staining buffer (PBS plus 1% Fetal Bovine Serum (FBS)).
Viability was determined by exclusion of trypan blue (≥ 98%). PBMC were also cultured (1 × 10 6 cells/mL), with or without Mtb-Ag (10 μg/mL) with RPMI 1640 medium (Gibco, MD, USA) supplemented with 1% L-glutamine, 1% penicillin/streptomycin, and 10% FBS (Gibco, MD, USA) during 48 h. Then, levels of IFN-γ were measured using a commercial ELISA kit (Human IFN-γ ELISA MAX Standard Kit, BioLegend, USA) following the manufacturer´s instructions. Statistical analysis. Analysis of variance (ANOVA) and post hoc Tukey´s multiple comparisons test were used as indicated in gure legends. The Mann-Whitney U test was used to analyze differences between groups. For categorical variables, the Chi-square (and Fisher´s exact) test for homogeneity was performed to compare proportions of subjects between groups. In the graphs each symbol represents an individual and the horizontal lines indicate the mean ± standard error of the mean (SEM). Correlations were calculated using the non-parametric Spearman correlation test. Receiver operating characteristic (ROC) curve analysis was performed to analyze the predictive value of the frequencies of CD14 + CD16 + and M-MDSC cells populations, calculating the area under the curve (AUC) and the 95% con dence interval (CI). Analyses were performed using GraphPad Prism 8.0.2 software. P < 0.05 was considered statistically signi cant.  Tables   Table 1. Immunological and clinical parameters of TB patients. LR-TB and HR-TB were classi ed according to IFN-γ production [fold-stimulation: (pg/mL after Mtb-Ag stimulation)/ (pg/mL after culturing with medium)]; proliferation [proliferation index: (c.p.m. after Mtb-Ag stimulation)/ (c.p.m. after culturing with medium)]; and increase in the percentage of SLAM-positive cells in response to Mtb-Ag stimulation. Proportions of lymphocytes, monocytes and neutrophils are shown as percentages of total white blood cells. Time of disease evolution (days) was established by analyzing the following clinical symptoms in patients previous to hospital admission: weight loss, night sweats, symptoms of malaise or weakness, persistent fever, presence of cough, history of shortness of breath, and/or hemoptysis. Mean ± SEM are shown for continuous data. Categorical data (AFB in sputum smear) are expressed as percentages. a P values were calculated by the Mann-Whitney U test. b P value was calculated by Chi-Square (and Fisher´s exact) test for categorical variables. P values < 0.05 were considered signi cant. LR TB: Low responder tuberculosis patients; HR TB: High responder tuberculosis patients. AFB: Acid-Fast Baclli.