Glucocorticoid treatment in patients with newly diagnosed immune thrombocytopenia switches CD14++CD16+ intermediate monocytes from a pro‐inflammatory to an anti‐inflammatory phenotype

Immune thrombocytopenia (ITP) is thought to result from an aberrant adaptive autoimmune response, involving autoantibodies, B and T lymphocytes, directed at platelets and megakaryocytes. Previous reports have demonstrated skewed CD4+ T‐helper subset distribution and enhanced production of pro‐inflammatory cytokines such as interleukin 17A and interferon gamma. The role of monocytes (MCs) in ITP is less widely described, but innate immune cells have a role in shaping CD4+ T‐cell phenotypes. Glucocorticoids (GCs) are commonly used for first‐line ITP treatment and modulate a broad range of immune cells including T cells and MCs. Using multiparameter flow cytometry analysis, we demonstrate the expansion of intermediate MCs (CD14++CD16+) in untreated patients with newly diagnosed ITP, with these cells displaying a pro‐inflammatory phenotype, characterised by enhanced expression of CD64 and CD80. After 2 weeks of prednisolone treatment (1 mg/kg daily), the proportion of intermediate MCs reduced, with enhanced expression of the anti‐inflammatory markers CD206 and CD163. Healthy control MCs were distinctly different than MCs from patients with ITP before and after GC treatment. Furthermore, the GC‐induced phenotype was not observed in patients with chronic ITP receiving thrombopoietin receptor agonists. These data suggest a role of MCs in ITP pathogenesis and clinical response to GC therapy.


Introduction
Immune thrombocytopenia (ITP) is a rare (2-3/100 000) autoimmune condition characterised by a low platelet count (<100 9 10 9 /l) with an increased risk of bleeding and fatigue. 1,2 It can be categorised clinically as newly diagnosed (<3 months from diagnosis), persistent (3-12 months) or chronic ITP (>12 months). 3,4 High-dose glucocorticoids (GCs) are the recommended first-line treatment for ITP. Patients who fail GC treatment, either due to lack of response, intolerance or relapse after GC is discontinued, receive second-line treatments, such as thrombopoietin receptor agonists (TPO-RA), mycophenolate mofetil (MMF) or rituximab. ITP pathogenesis is driven by B and T cells through antibody-and cell-mediated mechanisms, targeting platelets and megakaryocytes, resulting in increased consumption and decreased production of platelets. [5][6][7] Patients with ITP have elevated interleukin 17A (IL-17A) and interferon gamma (IFN-c) producing CD4 + T cells, as well as lower numbers of circulating forkhead box P3 (FOXP3) + regulatory T cells (Tregs). [8][9][10] Monocytes (MCs) play a key role in shaping the T-cell response through processes of antigen presentation and cytokine production. 11 The contribution of MCs to the development of ITP is less well understood than classical antibody and T cell-mediated autoimmunity. It is nonetheless likely that MCs shape these adaptive immune drivers of ITP, as well as the T-cell response to treatment.
Human MCs are classified into three subsets based on their cell surface expression of CD14 [lipopolysaccharide (LPS) coreceptor] and CD16 [activatory Fc gamma receptor III (FccRIII)] into classical (C-MCs; CD14 ++ CD16 À ), intermediate (I-MCs; CD14 ++ CD16 + ) and non-classical (NC-MCs; CD14 + CD16 ++ ) MCs. 12 In many autoimmune conditions, I-MCs are enriched 13 and in the context of inflammation drive enhanced memory T-cell proliferation and inflammatory cytokine expression. 14,15 CD16 + MCs from patients with ITP have been shown to specifically promote the expansion of IFNc + CD4 + T cells and concomitantly inhibit the proliferation of Tregs, 16 providing evidence that MCs shape T-cell responses in ITP. However, a detailed MC phenotype in untreated, newly diagnosed patients with active ITP has not yet been described. Different MC subsets have unique abilities to shape T cells with I-MCs from healthy controls (HCs) demonstrating reduced priming of na€ ıve T cells and reduced pro-inflammatory CD4 + T-cell polarisation, while also promoting secretion of IL-10 by Tregs, implying that I-MCs under homeostatic conditions attenuate CD4 + T-cell activity. 17 GCs, such as prednisolone or dexamethasone (dex), aim to suppress aberrant immune responses, 3,4 and while the T cell effect of these drugs has been described, 18 the effect of GCs on MC subsets has been less well researched. GC treatment of MCs in vitro induces an expansion of CD16 + MCs and induction of an anti-inflammatory monocytic phenotype, characterised by increased CD163 and CD206 expression and increased IL-10 production. [19][20][21][22] However, these studies did not investigate the in vivo consequences of GC treatment on MC subsets. Much of the published data regarding GC effect on immune cellular populations in patients with autoimmune diseases (including ITP) is limited by the heterogeneity of patients, including treatment history, disease severity and chronicity. Study of the in vivo effects of GC treatment is particularly confounded by these variables.
Given previous reports in other autoimmune diseases and alongside observations in ITP, we hypothesised that the circulating CD14 ++ CD16 + I-MC subset would have pro-inflammatory characteristics in untreated patients presenting acutely with active ITP, and that this would be reversed following successful GC treatment. In the present study, to further elucidate the potential role of MCs in disease pathogenesis and resolution following GC treatment, we examined the phenotype of the MC subsets before and after in vivo GC treatment in untreated patients with newly diagnosed ITP.

Participants
Regulatory approval was granted in accordance with the NHS Health Research Authority (HRA) at the University Hospitals Bristol NHS Foundation Trust, UK. Untreated patients with newly diagnosed ITP, aged ≥16 years, had peripheral blood samples taken at the point of diagnosis before treatment and after an average of 2 weeks prednisolone (1 mg/kg daily) treatment, in accordance with international consensus guidelines. 3,4 Additional patients with varying levels of disease chronicity who had received TPO-RA therapy, were also recruited (HRA ref: 15

MC and platelet count
Peripheral blood was collected from patients by sterile venepuncture into ethylenediaminetetraacetic (EDTA) containing tubes. MC and platelet counts were derived from the full blood count (FBC) processed on the Sysmex XN-20 analyser. Samples were not analysed for anti-platelet autoantibodies.

Human cell isolation
Peripheral blood mononuclear cells (PBMCs) were isolated from peripheral blood by density gradient centrifugation using Ficoll-Paque PLUS (GE Healthcare, Chicago, IL, USA). TM Human Monocyte Enrichment Cocktail (Stem Cell Technologies UK, Cambridge, UK) according to the manufacturer's instructions.

Phenotyping by flow cytometry
PBMCs were assessed using a panel of fluorescent-conjugated antibodies outlined in Table SI. Cultured MCs were harvested on ice and stained with CD14, CD16, CD64, CD80, CD163 and CD206. Analysis was performed on a BD Fortessa X20 flow cytometer (BD Biosciences, San Jose, CA, USA). FMO (fluorescence minus one) controls were used for each fluorochrome. Analysis for t-distributed stochastic neighbour embedding (t-SNE) and sequential cluster analysis was performed using FlowJo 10.6.1. The number of MCs [identified by live, singlets expressing CD4, CD14 and human leucocyte antigen-DR isotype (HLA-DR)] were equalised, at random, from each donor and all samples were concatenated prior to t-SNE analysis. The analysis included 11 markers outlined in Table SI and excluded CD4, side scatter-area (SSC-A), forward scatter (FSC)-A, FSC-height (H) and the live/dead discriminator.

Statistical analysis
Normality of grouped data was determined, and the statistical significance was assessed by ANOVA (parametric data), Kruskal-Wallis (non-parametric data) or Friedman (paired, non-parametric data) using GraphPad PRISM software 8.2.1 (GraphPad Software Inc., La Jolla, CA, USA). Comparison between individual data sets was determined by either unpaired t-test (parametric data) or Mann-Whitney test (nonparametric data) using GraphPad PRISM software 8.2.1.

Intermediate MCs are expanded in untreated patients with newly diagnosed ITP
The patients with ITP demographics are outlined in Table I. PBMCs from 11 untreated patients with newly diagnosed ITP, presenting with a platelet count of <30 9 10 9 /l, were examined by multiparameter flow cytometry to determine the MC subset phenotype before and after GC treatment [median (range) of 13 (8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26) days]. MCs were gated from isolated PBMCs based on CD4 int , CD14 and HLA-DR hi as described by Abeles et al. 23 and shown in Figure S1. Based on CD14 and CD16 expression (Fig 1A), untreated patients with newly diagnosed ITP had an expansion of I-MCs compared with HC [median (range) age 32 (23-52) years, all female], with a concurrent reduction of the C-MC. The I-MC and NC-MC populations were reduced after GC therapy, with a concurrent increase in C-MC (Fig 1A,B). The proportional change in MC subsets was not reflective of an overall change in total MC count before or after GC therapy ( Fig 1C). There was a significant increase in platelet count after GC therapy ( Fig 1D).

MCs from patients with newly diagnosed ITP before and after GC treatment have distinct phenotypes
The phenotypic differences between whole MCs from HCs and patients with ITP before (ITP pre-GC) and after GC treatment (ITP post-GC) based on multiparameter flow cytometry were visualised by generating a t-SNE analysis and summarised in Table SII. This analysis revealed that MCs from the three cohorts were markedly different from each other (Fig 2A). Clustering analysis identified that while some phenotypes were shared between these cohorts, some cellular clusters were unique. For example, cluster 12 is only present in untreated patients with newly diagnosed ITP, whereas cluster 6 is only observed in patients with ITP after GC therapy (Fig 2B,C). Detailed examination of these phenotypes in the MC subsets from untreated patients with newly diagnosed ITP demonstrates enhanced expression of pro-inflammatory markers CD64 on all MC subsets and CD80 on C-MCs and I-MCs, but not NC-MCs compared with HCs. These pro-inflammatory markers were reduced after GC therapy on the C-MCs and I-MCs, but not on NC-MCs ( Fig 3A,B respectively). CD206 and CD163 have previously been shown to be upregulated in vitro and in vivo after GC treatment. 21 In the ITP cohort, CD206 was upregulated on all MC subsets (Fig 3C), with CD163 being specifically upregulated on both I-MCs and NC-MCs, but not C-MCs ( Fig 3D) following in vivo GC treatment. These data suggest that MCs from newly diagnosed ITP have a pro-inflammatory phenotype and that this switches to an anti-inflammatory phenotype after treatment, with the nature of this shift varying between MC subsets.
Changes in CD163 and CD206 associated with GC therapy are not seen with TPO-RA treatment To demonstrate that the observed phenotypic changes were specifically due to GC therapy and not due to disease resolution, the PBMC phenotype of patients with chronic ITP receiving TPO-RA treatment (n = 9) was investigated. The platelet counts for the chronic ITP patient cohort was below the normal range, although higher than the counts from the untreated newly diagnosed cohort (Fig 1D; Figure S2A). However, MC counts were within normal parameters ( Fig S2B). Consistent with the untreated patients with newly diagnosed ITP, there was an increase in the proportion of I-MCs and NC-MCs in patients with ITP treated with TPO-RA with a decrease in C-MCs compared with the concurrently analysed HC cohort [median (range) age 27  years, all female] (Fig 4A). These post-TPO-RA I-MCs did not express increased CD163 or CD206 compared with the HCs (Fig 4B,C respectively Fig 5A). Therefore, further studies were focussed on only these two MC subsets. Despite previous reports that GC treatment increases MC viability, 20,21 in the present study there was good viability amongst all groups and therefore GC treatment did not augment MC survival ( Figure S3). In vitro treatment with dex leads to an enrichment of the I-MCs compared with both non-treated (NT) and LPS-treated MCs (Fig 5A,B). This is in contrast with the ex vivo observations after in vivo GC treatment outlined in Fig 1, but is consistent with our previously reported in vitro data in HCs. 17 There was no significant increase in CD64 mean fluorescent intensity (MFI) after LPS stimulation compared with untreated MCs (Fig 5C). However, there was a significant increase in expression of CD80 on both C-MCs and I-MCs after treatment with the inflammatory stimulus LPS compared with both NT-and dex-treated MCs (Fig 5D). I-MCs significantly upregulate CD80 to a greater extent than C-MCs under the same treatment conditions (Fig 5D). In contrast to ex vivo observations, CD206 was not significantly upregulated after dex treatment (Fig 5E). However, after dex treatment, I-MCs did have greater CD206 expression than the C-MCs in the same culture (Fig 5E). After dex treatment, the I-MCs had significantly upregulated CD163 compared with NT-and LPS-treated MCs (Fig 5F). In addition, I-MCs upregulated CD163 to a greater extent after dex treatment than C-MCs, the latter of which showed a non-significant increase in CD163 compared with untreated C-MCs or those stimulated with LPS ( Fig 5F). Therefore, these data (summarised in Table SIV) show that while both subsets partly mimic ex vivo phenotypic changes observed in the untreated ITP cohort, the I-MCs are most changed after incubation with either dex or LPS treatment. Furthermore, MCs isolated from patients with newly diagnosed and chronic ITP and treated for 24 h with dex, showed similar expansion of CD16 + I-MCs (Fig 5G). I-MCs significantly upregulated CD163 compared with C-MCs ( Fig 5H); neither MC population significantly upregulated CD206 (Fig 5I). This suggests that MCs from patients with ITP respond similarly to HC MCs in response to dex in vitro.

Discussion
Studying an untreated newly diagnosed ITP patient cohort (all patients with initial platelet count of <30 9 10 9 /l) has avoided confounding factors such as heterogeneous ITP treatments (current and previous), variable disease severity and chronicity. Consistent with previous reports, we observed an expansion of the I-MC population in patients with active ITP, which returns to a normal proportion of the peripheral MCs following successful treatment. 24 In addition, we demonstrate that MCs differ in their cell surface phenotype in untreated patients with newly diagnosed ITP before and after GC treatment, and in both cases, compared to HC samples. However, this does contrast with other autoimmune diseases, where I-MC expansion is seen after in vivo GC treatment, 17 and suggests that other factors such as different disease-specific autoimmune context may influence MC phenotype before and after exposure to GCs. Variable disease duration and severity, or co-administration of other therapies may also play a role. Recently, in a cohort of untreated patients with ITP, Monz on Manzano et al. 25 found an increased proportion of C-MCs and decreased proportion of NC-MCs, but in contrast to our present results there was no change in the percentage of I-MCs. However, the untreated patients in the Monz on Manzano et al. 25 study had ITP for a minimum of 6 months, some for >12 months, with average platelet counts of~100 9 10 9 /l (hence not requiring treatment). This contrasts with our present patients with newly diagnosed (acute) untreated ITP with platelet counts of <30 9 10 9 /l and a clinical need for prompt treatment. Therefore, it is possible that the differences between these cohorts in terms of disease chronicity, stability and severity of thrombocytopenia, explains the differing observations in the MCsubset proportions. Concordant with our observations, previous work by Liu et al. 26 demonstrated a reduction of CD64 (FccRI) and consequentially a reduction of opsonised bead phagocytosis, following high-dose in vivo dex treatment. As expression of Fc receptors contributes to ITP pathogenesis, 27 this suggests that reduction of FccRs after GC therapy may help treat ITP by reducing antibody-mediated platelet phagocytosis.
MCs are the precursors of macrophages, the vast diversity of which has been described as the 'M1/M2 paradigm'. M1like macrophages are characterised by, but not limited to, expression of CD64 and CD80, with these M1 myeloid cells being pro-inflammatory and driving type I inflammatory responses in adaptive immune effector cells. In contrast, M2 macrophages, often expressing CD206 and CD163, are wound-healing or anti-inflammatory cells. 28 Our present study examined circulating MCs, which can share characteristics with these M1-like or pro-inflammatory cells and conversely M2-like or anti-inflammatory cells. Specifically, we showed that prior to treatment, all MC subsets showed a pro-inflammatory, activated phenotype, and this switched to an anti-inflammatory (or more M2-like) phenotype after in vivo GC therapy. This suggests that systemic administration of GCs, rather than MCs returning to a homeostatic phenotype, skews MCs to be anti-inflammatory. The present study clarifies that all MCs, in particular I-MCs, have the ability to be either pro-or anti-inflammatory depending on the context in which they are activated, shedding light on the dichotomous role of MCs in both health and disease.
As the ex vivo anti-inflammatory phenotype (namely CD206 and CD163 upregulation) is not observed with other therapeutic interventions such as TPO-RAs, we can conclude that the upregulation of these molecules is due to GC-specific effects on MCs. This is further supported by previous reports of CD163 and CD206 upregulation in MCs treated with GCs in vitro, 21,29 and as the in vitro upregulation of CD163 has been shown in the present study in both patients with newly diagnosed and chronic ITP, this phenotypic response to GCs is apparently retained regardless of disease stage. With regard to the upregulation of CD206 expression after in vitro dex treatment, our present findings were on the margins of significance (Fig 5E; P = 0Á0515 compared with HCs), and given that CD206 regulation by GCs has previously been examined in murine models or at the genomic level, 21,29,30 further investigation into the kinetics of CD206 protein expression after in vitro GC treatment in human MCs is warranted. Consistent with our present findings, Monz on Manzano et al. 25 also demonstrated an increase in I-MCs after TPO-RA treatment, but in contrast to our present findings they did not observe an increase in the proportion of NC-MCs or decrease in C-MCs compared with HCs (this may again simply reflect differences in our patient cohorts and treatment criteria). Nonetheless, there is a consistent increase in I-MCs across these studies, which may reflect ongoing ITP disease activity, as following successful treatment of ITP, MC subsets have previously been found to return to normal proportions. 24,25 Alternatively, increase in I-MCs may be an indirect TPO-RA effect, as TPO-RAs have been associated with other immune changes in B , T and Treg cells, and altered production of anti-platelet autoantibodies. [31][32][33] Under steady state and during acute inflammation, C-MCs are released from the bone marrow and differentiate first into I-MCs and secondly into NC-MCs. 34 C-MCs are relatively short-lived (~1 day) and differentiate into I-MCs and then the NC-MCs in vivo in humans and in humanised murine models during homeostasis and following acute inflammatory stimulus. 34 In our present study, the observed expansion of the I-MC populations expressing pro-inflammatory markers is likely a persistent phenotype caused by enhanced bone marrow egressing C-MCs and the sequential differentiation into I-MCs due to sustained systemic inflammation. It is not clear whether, after GC treatment, the phenotypic shift in the I-MCs and NC-MCs is due to suppression and phenotype change of circulating MCs or replacement of the circulating MC populations. Given that I- MCs and NC-MCs, have been shown to persist in the periphery longer than C-MCs (~4Á3 and~7Á4 days respectively) it is most likely that over the 2-week treatment period the pro-inflammatory MCs (observed before treatment) are being replaced by fresh bone marrow egressing C-MCs which, due to the systemic GC therapy, differentiate into the anti-inflammatory MCs seen post-GC therapy. Previous studies support this mechanism, demonstrating expansion of a CD14, CD16 and CD163 population in vitro after GC treatment, with CD163 expression being linked exclusively to GC treatment. 17,21,35,36 However, our present observed reduction in the percentage of circulating CD16 + MCs after GC treatment in vivo has also been reported in multiple sclerosis 37 and in HCs. 38 This suggests that there are distinct differences with regard to CD16 expression after GC treatment in vitro and in vivo. In addition, the ex vivo MC phenotype of untreated newly diagnosed ITP shows an increase in CD64 on all MC subsets, but this is not observed in vitro after LPS stimulus. LPS treatment in vitro is a single stimulus used to drive inflammatory myeloid cells but does not reflect the complex, multifaceted pro-inflammatory stimuli myeloid cells are exposed to in vivo. Therefore, while in vitro models are ideal for studying an individual stimulus, they do not always reflect the complexities of in vivo systems. It has been widely reported that both patients with newly diagnosed and chronic ITP have elevated numbers of IFN-cand IL-17A-producing T cells. 8 In rheumatoid arthritis, MCs from inflamed joints enhance IL-17A production from autologous CD4 + T cells. 39 In ITP, CD16 + MCs from patients with ITP promote the expansion of IFN-c + CD4 + T cells 16 and separately, IL-17A production has been shown to correlate with increased CD68 + myeloid cells present in bone marrow of patients with ITP. 40 Alongside this, it has been shown that M1-like MC-derived macrophages (MDMs) generated from circulating MCs from patients with ITP augment cytokine production from CD4 + T cells compared with M2-like MDMs. Taken together, these data demonstrate that inflammatory myeloid cells, such as those observed in the present study upon diagnosis, could contribute to enhanced CD4 + T-cell activity. In contrast, M2-MDMs from patients with ITP differentiated in the presence of dex for 7 days, similar to the post-GC MC phenotype observed in our present study, had decreased CD80 and increased CD163 expression and were able to suppress CD4 + T-cell proliferation and cytokine production, 41 suggesting that GC-derived MCs are able to suppress CD4 + T-cell responses. Combined with this, patients with newly diagnosed ITP treated with high-dose dex have enhanced circulating myeloid-derived suppressor cells (CD11b + CD33 + HLA-DR À ) and M2-like macrophages (CD68 + CD163 + ). 42 These data suggest that not only are the circulating peripheral MCs in patients with ITP shifted to an anti-inflammatory phenotype post-GC treatment, but as they also give rise to M2-like macrophages these circulating GC-MCs may contribute to a reduction in overall immune activation and consequently support disease resolution.

Supporting Information
Additional supporting information may be found online in the Supporting Information section at the end of the article. Fig S1. Identification of human peripheral blood monocytes by flow cytometry. Fig S2. Peripheral blood monocyte and platelet counts in patients with ITP treated with TPO-RA. Fig S3. Viability of in vitro monocytes following 24-h culture. Table SI. Ex vivo monocyte phenotyping panel. Table SII. Mean fluorescence intensity of monocyte subset cell surface markers as determined by ex vivo flow cytometry of patients with ITP before and after glucocorticoid treatment in vivo. Table SIII. Mean fluorescence intensity of monocyte subset cell surface markers as determined by ex vivo flow cytometry of patients with ITP receiving TPO-RA treatment in vivo. Table SIV. Monocyte subset cell surface marker expression following 24-h in vitro treatment as determined by flow cytometry.