Diverse macrophage populations contribute to distinct manifestations of human cutaneous graft-versus-host disease

Abstract Background Graft-versus-host disease (GvHD) is a major life-threatening complication of allogeneic haematopoietic stem cell transplantation (HSCT), limiting the broad application of HSCT for haematological malignancies. Cutaneous GvHD is described as a post-transplant inflammatory reaction by skin-infiltrating donor T cells and remaining recipient tissue-resident memory T cells. Despite the major influence of lymphocytes on GvHD pathogenesis, the complex role of mononuclear phagocytes (MNPs) in tissues affected by GvHD is increasingly appreciated. Objectives To characterize the identity, origin and functions of MNPs in patients with acute cutaneous GvHD. Methods Using single-cell RNA sequencing and multiplex tissue immunofluorescence, we identified an increased abundance of MNPs in skin and blood from 36 patients with acute cutaneous GvHD. In cases of sex-mismatched transplantation, we used expression of X-linked genes to detect rapid tissue adaptation of newly recruited donor MNPs resulting in similar transcriptional states of host- and donor-derived macrophages within GvHD skin lesions. Results We showed that cutaneous GvHD lesions harbour expanded CD163+ tissue-resident macrophage populations with anti-inflammatory and tissue-remodelling properties including interleukin-10 cytokine production. Cell–cell interaction analyses revealed putative signalling to strengthen regulatory T-cell responses. Notably, macrophage polarization in chronic cutaneous GvHD types was proinflammatory and drastically differed from acute GvHD, supporting the notion of distinct cellular players in different clinical GvHD subtypes. Conclusions Overall, our data reveal a surprisingly dynamic role of MNPs after HSCT. Specific and time-resolved targeting to repolarize this cell subset may present a promising therapeutic strategy in combatting GvHD skin inflammation.


Graphical Abstract
What is already known about this topic?
• Cutaneous graft-versus-host disease (GvHD) is a life-threatening complication limiting the broad application of haematopoietic stem cell transplantation (HSCT).• GvHD can manifest as acute-and chronic-type skin inflammation with distinct clinical presentations and lymphocyte cytokine profiles.• Skin-infiltrating donor T cells significantly propagate GvHD, but little is known about donor and host macrophages (MΦs) as potential drivers of skin inflammation.

What is the translational message?
• Both host-and donor-derived MΦs were present in skin lesions and showed few transcriptomic differences, indicating rapid differentiation and polarization of donor mononuclear phagocytes to tissue MΦs as they enter the skin in acute GvHD.• Furthermore, diverse MΦ responses in clinically distinct GvHD subtypes reflect the high plasticity of skin MΦs and illustrate the complex setting that is present in the cutaneous immune system after HSCT.• Our results demonstrate a major role of MΦs as cellular players in GvHD-affected skin, with possibilities for therapeutic targeting.

What does this study add?
• We detected donor-and host-derived MΦs as major cellular players in cutaneous GvHD clinical subtypes.
Cutaneous graft-versus-host disease (GvHD) is a major complication of allogeneic haematopoietic stem cell trans plantation (HSCT), occurring in a large proportion of patients. 1][5][6] Tissue-resident macrophages (MΦs) are a major antigen-presenting cell subset in adult human barrier organs with a wide spectrum of pro-and anti-inflammatory properties, depending on their polarization state.MΦs have the ability to rapidly change their gene expression in response to environmental stimuli along a continuum of what has formerly been simplified as M1 and M2 polarization.
Skin-resident MΦs are induced by a T helper (Th)-2 cytokine environment 7 and express surface receptors including CD11b, CD68 and CD206. 8Their detrimental role in inflammatory skin diseases has been shown for atopic dermatitis, psoriasis and discoid lupus skin lesions, where MΦs were characterized by the expression of F13A1 or CD163 and genes related to chemotaxis and transforming growth factor beta (TGF-β) signalling. 9,10In a cytokine-rich environment, proinflammatory MΦs may produce interferon gamma (IFN-γ), which is a potent autocrine mediator inducing the MΦ phenotype found in psoriatic skin. 11owever, skin-resident MΦs have also been appreciated for their anti-inflammatory properties and can promote tissue repair by local production of interleukin (IL)-10. 12Recently, a role for MΦs in GvHD pathology has been implicated in studies of tissues obtained from human HSCT recipients.Aasebo et al. detected a five-fold increase in donor-derived proinflammatory MΦs in colon biopsies of patients with gastrointestinal aGvHD. 13In cutaneous aGvHD, Jardine et al. found a population of monocyte-derived MΦs that mediated keratinocyte cytopathicity ex vivo. 6owever, the functions and polarization profiles of donorand host-derived MΦs in acute and chronic cutaneous GvHD remain unclear.Here, we used single-cell RNA sequencing (scRNA-seq) and tissue immunofluorescence (IF) in GvHD skin lesions from adult HSCT recipients to investigate MΦ subsets in acute and chronic cutaneous GvHD.We detected expanded CD163 + host and donor MΦs with low IFN-γ production and a tissue-remodelling cytokine signature (TGF-β, IL-10) in cutaneous aGvHD, and proinflammatory MΦs with decreased IL-10 production in cutaneous chronic GvHD.

Patient inclusion, sampling and tissue processing
Thirty-six adult individuals presenting with acute, previously untreated cutaneous GvHD following allogeneic HSCT, at the University Hospital, Medical University of Vienna, Austria, were included in the study after obtaining appropriate fully informed written consent.5][16][17] Of the 36 patients, 5 with previously untreated cutaneous aGvHD grade 2-3 were included for scRNA-seq, and 31 with previously untreated aGvHD, chronic lichenoid (cl)GvHD or chronic sclerotic (cs)GvHD were included for IF work-up.We performed 6-mm punch biopsies of lesional skin under local anaesthesia at disease onset before start of treatment with corticosteroids.Tissue was either deep frozen for IF work-up or immediately processed by enzymatic digestion for scRNA-seq.From patients presenting with aGvHD, blood was sampled in heparinized sample collection tubes and immediately processed for scRNA-seq.The study was approved by the local ethics committee, Medical University of Vienna (ECS 1087/2016).

Cell sorting and library preparation
Single leucocyte suspensions were prepared from skin via enzymatic digestion and from whole blood using a Ficoll-Paque gradient, as previously described in detail. 18For sc-RNA-seq using the 10X Genomics platform (10X Genomics, Pleasanton, CA, USA), CD45 + live cells were sorted into phosphate-buffered saline (PBS)/10% fetal calf serum for library preparation with the Chromium™ Single Cell 5′ Library & Gel Bead Kit v2 (10X Genomics) according to the manufacturer's instructions.Briefly, sorted cells were partitioned into single-cell GEMs (Gel beads in EMulsion droplets) for cDNA synthesis and subsequently amplified and prepared for sequencing.The samples were sequenced at the Biomedical Sequencing Facility on the Illumina NovaSeq 6000 SP platform (Illumina, San Diego, CA, USA) in the 50-bp paired-end configuration.Raw sequencing data were processed as described below.

Analysis of single-cell RNA-sequencing data
Raw sequencing data were quantified using the Cell Ranger single-cell gene expression software version 3.1.0(10X Genomics) and aligned to the GRCh38 reference genome.Cells were filtered for detected gene number (< 200) and excessive mitochondrial gene expression (> 20%).Genes were filtered for expression in < 3 cells.After preprocessing, datasets were integrated and batch correction was performed using a BBKNN (batch balanced k nearest neighbours) graph-based data integration algorithm. 19Differential gene expression analysis was performed with the SCANPY toolkit 20 using a Wilcoxon rank-sum test and Student's t-test.

Cell-cell interaction analysis
Intercellular communication between mononuclear phagocytes (MNPs) and T cells was inferred from single-cell transcriptome data using the R toolkit CellChat version 1.5.0. 21ased on cell clusters identified by the Leiden clustering algorithm, normalized count data were used to identify overexpressed genes and overexpressed interactions, and compute communication probability with the default package functions.Receptors/ligands were considered if expressed in > 15 cells per group.

Tissue processing for multiplex immunofluorescence
For validation experiments, we collected lesional skin biopsies of 31 patients presenting with previously untreated cutaneous GvHD in different disease stages and pathologies: aGvHD (n = 15), clGvHD (n = 9) and csGvHD (n = 7).In addition, we collected excised skin from plastic surgery procedures of healthy individuals (n = 10).Tissue samples were embedded in Tissue-Tek ® optimum cutting temperature compound (O.C.T.™ Compound; Sakura Finetek, Torrance, CA, USA), deep frozen and cryosectioned.MΦs, cytokines and surface receptors were visualized with multistep IF staining.In brief, slides were defrosted, blocked with mouse serum (3%) and bovine serum albumin (2%) in PBS and incubated with directly conjugated primary antibodies overnight.CD68 + CD11b + MΦs were costained either with anti-CCR7 or with anti-CD163.For cytokine labelling, cells contained in skin sections were previously permeabilized using Tween-20 solution.Primary anti-IFN-γ or anti-IL-10 antibody staining was performed by incubation for 2 h at room temperature, followed by secondary antibody labelling to increase signal strength.To reveal the cellular source of cytokines, MΦ labelling was performed on the same sections using incubation with anti-CD68 antibody overnight, followed by 4′,6-diamidino-2-phenylindole (DAPI) counterstaining.For visualization of MΦ-T-cell contact, IF was performed in aGvHD skin sections from three patients by incubation with directly conjugated anti-CD3, anti-CD68, anti-CD45 and anti-CD206 antibodies for 120 min at room temperature, and DAPI counterstaining after blocking steps as described above.Labelled sections were acquired using a ZEISS Axio Observer Z1 microscope (Zeiss Group, Oberkocken, Germany) with a TissueFAXS imaging system (TissueGnostics, Vienna, Austria) and processed using image analysis software (TissueGnostics).

Results
The mononuclear phagocyte fraction is expanded in graft-versus-host disease aGvHD of the skin may occur as early as post-transplant week 2, and is classically described as an adverse inflammatory reaction by skin-infiltrating donor T cells.However, it is well established that recovery of MNPs precedes lymphocyte engraftment after HSCT, 22,23 prompting speculation on the contribution of MNP subsets to disease pathogenesis.To investigate MNP subsets in aGvHD, we performed scRNA-seq of blood and skin leucocytes from five HSCT recipients presenting with aGvHD of the skin (> grade 2) (Figure 1a, Table 1) and compared them with our previously published datasets from skin and blood of healthy individuals. 24ollowing quality control filtering and doublet exclusion, we performed analysis on 24 346 sequenced skin-and 21 616 blood leucocytes [Figure 1b; Figure S1a (see Supporting Information)].Following batch correction, Leiden clustering identified nine distinct cell states (Figure 1c).Differential expression of known cell-type marker genes allowed annotation of broad leucocyte clusters as T cells, cytotoxic T-and NK cells (T/NK), B cells, haematopoietic stem cells/immature leucocytes (HSCs) and MNPs (Figure 1d, e; Figure S1b).To identify the major leucocyte subsets contributing to inflammation in GvHD, we analysed the relative cell fraction for each cell type.We identified T/NK and MNP subsets to be relatively expanded in aGvHD compared with skin and blood from healthy donors (Figure 1f).Furthermore, immature leucocytes contained in the HSC cluster were derived almost exclusively from aGvHD datasets.When we mapped the relative fraction of cell types according to individual samples (Figure S1c), we noticed differential ratios of cells according to time of sampling after HSCT.Samples early after HSCT (< 3 weeks, GvHD3, GvHD5) contributed to MNP-and T-cell clusters, while B cells were mostly derived from healthy controls and samples > 40 days after HSCT (GvHD2, GvHD4).

Acute graft-versus-host disease skin lesions harbour increased numbers of macrophages
To further determine the role of MNPs in aGvHD, we next performed differential gene expression analysis of all aGvHD vs. healthy control datasets.Unsupervised analysis showed upregulation of MNP marker gene CST3 and downregulation of T-cell-marker genes CD3D and CD3E in aGvHD (Figure S2a; see Supporting Information).Reclustering of cells identified as MNPs (Figure 2a) using the Leiden algorithm revealed eight distinct MNP transcriptional states (Figure 2a).MNPs clustered according to tissue origin in skin and blood MNPs (Figure 2b).We annotated distinct MNP clusters according to expression of published marker genes 9 (Figure 2c; Figure S2b).Leiden clusters 1, 3 and 5 were almost exclusively derived from peripheral blood, and corresponded to classical and nonclassical monocytes by expression of CD14 and FCGR3A (which encodes CD16).Cluster 6 contained circulating cells expressing neutrophil-related genes LTF and CD177, and was thus annotated as granulocyte cluster.Clusters 4 and 7 displayed a gene signature associated with dendritic cells (DCs) and were annotated as CLEC4C-expressing plasmacytoid DCs (pDCs, cluster 7) and skin DCs, including Langerhans cells, expressing FCER1A and CD1C (cluster 4).Clusters 0 and 2 corresponded to tissue MΦs, characterized by the expression of CD68, F13A1, C1QC and IL1B.
To gauge the relevance of respective MNP subsets to GvHD inflammation, we compared the expression of proand anti-inflammatory marker genes in MΦ, monocyte and DC subclusters from GvHD samples (Figure S2c, top panels).Monocytes expressed high levels of STAT1, while TGFB1 and IL10RA were expressed by DC and pDC subsets.Interestingly, cells of the MΦ cluster displayed the highest expression of TNF, VEGFA and IL10 (Figure S2c, bottom panels), suggesting a tissue-remodelling cytokine signature.
We next mapped cell clusters back to their dataset origin and found the major fraction of MΦs, neutrophils, nonclassical monocytes and pDCs to derive from aGvHD datasets, while DCs were largely from healthy controls (Figure 2d).In GvHD dataset-derived MΦs, marker genes for cell proliferation (MKI67, MYBL2) and cell-cycle-regulated genes (CCND1) were upregulated (Figure S2d), suggesting in situ proliferation of cutaneous MΦ.To validate that the MΦ fraction was expanded in aGvHD skin, we investigated CD68 and CD11b protein expression in skin cryosections of individuals presenting with aGvHD > grade 2 (Figure 2e, f).We detected a mean number of 200 MΦs (CD68 + CD11b + cells) mm -2 in aGvHD skin, which was significantly increased compared with healthy control skin.Taken together, our data show a clear expansion of tissue MΦs in aGvHD skin lesions.

Activated conventional dendritic cells are present in acute graft-versus-host disease skin
Merely a small fraction of the DC cluster was derived from GvHD datasets (Figure 2d).While this finding is in line with delayed DC reconstitution after HSCT, we were interested in the phenotype of DC subsets present in aGvHD skin.To define DC subsets involved in GvHD, we subclustered cluster 4 (Figure 2a) using the Leiden algorithm.Within this DC cluster, we detected five distinct cell states (Figure S3a; see Supporting Information).Interestingly, only one subcluster (Leiden cluster 2) was mainly derived from GvHD skin datasets, the rest from healthy skin (Figure S3b).The GvHD DC cluster contained activated skin CD1C + conventional DCs (CD1C, FCER1A, CELC10A expression) with high antigen-presenting function, indicated by increased expression of NFKB1, CXCL8, BCL3 and LGALS3 (Figure S3c).Functionally, Leiden cluster 2 resembled healthy skinderived DC subclusters with a similar expression of pro-and anti-inflammatory cytokines (Figure S3d).

CD163 + tissue-remodelling macrophages are increased in acute graft-versus-host disease skin lesions
To gain insights into the functional role of the numerically expanded MΦs in aGvHD skin lesions, we performed Leiden subclustering of MNPs derived from aGvHD skin lesions and retrieved four distinct cell states (Figure S2e).Annotation to known marker genes identified LC, DC and two MΦ clusters (Figure 3a; Figure S2f).We next investigated transcriptional profiles within the MΦ clusters isolated from aGvHD skin and detected high expression of genes described in skin-resident, anti-inflammatory, tissue-remodelling and tumour-associated MΦ, including MRC1 (encoding CD206), C1QC and CD163 (Figure 3b). 25,26Healthy skin MΦs displayed a similar transcriptome with low expression of proinflammatory genes and migration marker CCR7 [Figure S4a (see Supporting Information)].Interestingly, top differentially expressed genes between aGvHD and healthy skin MNPs included MT2A (encoding metallothionein-2), a gene upregulated in tissue repair MΦ, 27 TIMP1, which regulates the angiogenetic capacity of MΦs 28 and PLIN2, a gene involved in lipid storage, which is expressed by immunosuppressive tumour-associated MΦs (Figure S4b). 29o validate our findings, we investigated protein expression of scavenger receptor CD163 and the migration/ lymph node homing receptor CCR7 on MΦs in aGvHD skin sections.In aGvHD, we detected significantly higher percentages of CD163 + MΦs compared with CCR7 + MΦs (Figure 3c, d).Accordingly, MΦs derived from aGvHD skin expressed tissue-remodelling cytokines TGFB and IL10 on the RNA level (Figure 3e).Using intracellular cytokine staining on skin sections, we detected a significantly higher proportion of IL-10 + MΦs compared with IFN-γ + MΦs in aGvHD (Figure 3f), which was less pronounced in healthy skin (Figure S4c-e).This resulted in an inverse ratio of proinflammatory vs. tissue-remodelling/anti-inflammatory MΦs in aGvHD (Figure 3 g).

Macrophages interact with T cells in acute graftversus-host disease skin
To address the consequence of anti-inflammatory MΦ polarization in aGvHD skin on T cells, we inferred putative cell-cell interactions from T-cell and MΦ scRNA states (Figure 4a) using the R toolkit CellChat. 21Unsupervised analysis yielded several significant receptor-ligand partners expressed by the investigated cell types (Figure 4b).Interestingly, the majority of signals predicted by CellChat were reported to induce regulatory T cells: (i) CD206 (encoded by MRC1) induces T-cell tolerance via inhibition of CD45 (PTPRC); 30 (ii) the interaction of galectin-9 (LGALS9) with CD44 enhances the stability and function of adaptive regulatory T cells; 31 and (iii) CD55 costimulation via CD97 (ADGRE5) induces IL-10 production in T cells. 32,33In addition, we detected potential instruction of tissue-regenerative T cells by MΦs via TNF/ TNF receptor 2 (TNFRSF1B, alias TNFR2) signalling 34 and interaction via PILRA, which has recently been reported to maintain CD8 T-cell quiescence in mice. 35To confirm a spatial relationship between MΦs and T cells for possible interaction, we performed immunofluorescence costaining in aGvHD skin and detected many instances of cell-cell surface contact (Figure 4c).In addition, we found CD206 + MΦs and CD45 + T cells in direct contact, indicating interaction via these receptors (Figure 4c, d).Overall, the polarization and functional profiles of cutaneous MΦs detected in our cohort suggest that MΦs induce a regulatory T-cell phenotype in the skin to limit GvHD inflammation.

Remaining host skin-resident-and donor macrophages join forces in acute graft-versushost disease skin
Although most tissue MΦ populations develop during embryogenesis from yolk sac progenitors, it is generally accepted that tissue-resident MΦs derived from HSC transplants repopulate the niche left by their naturally occurring counterparts 36 and may even engraft as brain-resident MΦs. 37owever, for skin-resident memory T cells we recently showed that host cells resist conditioning therapies preceding HSCT and remain in the skin for several years, 2 where they may contribute to local GvHD and mediate Th2-driven inflammation at distant body sites. 38o investigate the potential role of host tissue-resident MΦs in aGvHD skin inflammation, we analysed scRNA-seq datasets of two male individuals after sex-mismatched  transplantation (Figure 5a).Notably, sex-mismatched HSCT is a well-established individual risk factor for the development of GvHD. 39,40Using the X-chromosome-linked gene XIST to distinguish donor from host cells, 41 we identified 675 host-and 784 donor-derived cells within the MΦ skin cluster (Figure 5b).Unlike T cells, 2 MΦs did not cluster according to genotype, indicating similar transcriptomic profiles of donor-and host-derived MΦs (Figure 5c, left panel).Interestingly, cell abundance analysis revealed relatively large fractions of remaining host MΦs in both aGvHD samples (Figure 5c, right panel).Lastly, gene expression analysis showed expression of tissue-remodelling genes by both donor-and host-derived MΦs (Figure 5d), and we found few genes with differential expression between XISTnegative host and XIST-expressing donor MΦs [Figure 4a-d; Figure S5a-d

Diverse macrophage populations contribute to distinct manifestations of cutaneous graft-versushost disease
Using disease onset and clinical presentation, cutaneous GvHD can be classified into an acute form and two chronic forms: aGvHD, clGvHD and csGvHD, respectively.These three most common clinical presentations of skin GvHD differ by histopathological features and distinct molecular signatures.Type-2 and Th-22 T-cell responses characterize aGvHD, while Th1/Th17 cytokines are upregulated in clG-vHD, and increased transcripts of the type-1 cytokine IFN-γ were found in csGvHD. 42o investigate whether distinct MΦs may help shape these immunophenotypes, we investigated MΦ polarization profiles in a-, cl-and csGvHD skin lesions (Figure 6a).CD68 + CD11b + MΦ numbers were increased in aGvHD and clGvHD, and decreased in csGvHD (Figure 6b).Interestingly, in contrast to the expanded CD163 + tissue-resident MΦs with tissue-remodelling and anti-inflammatory properties, we detected that aGvHD, clGvHD and csGvHD lesions were populated by CCR7 + MΦs (Figure 6c, d).Furthermore, chronic GvHD MΦs were characterized by normal IFN-γ and decreased IL-10 expression compared with aGvHD and healthy skin (Figure 6e-g).Overall, our findings support the notion that acute and chronic cutaneous GvHD subtypes indeed represent distinct adverse immunological reactions after HSCT.

Discussion
In the present study we used scRNA-seq and tissue IF to resolve the cellular heterogeneity of MNPs in cutaneous GvHD.Our data show a significant expansion of both hostand donor-derived MΦs in aGvHD skin and a clear distinction from chronic GvHD MΦ polarization, indicated by transcriptional profile, CD163 surface protein expression, TGFB transcription and IL-10 cytokine production.
Overall, the importance of myeloid cells in the initiation and propagation of GvHD has been widely acknowledged. 4 histopathological study investigating aGvHD skin lesions found that increased CD163 + cell infiltration was a significant predictive factor for steroid-refractory GvHD as well as a negative prognostic factor for survival. 43However, more recent studies suggest a protective role of this type of MΦ in murine GvHD and radiation injury, where mesenchymal stem cell-educated CD163 + MΦs improved survival. 44In our study, we corroborate the presence of increased numbers of CD163 + cells in aGvHD skin and characterized them as tissue-remodelling CD11b + CD68 + MΦs with increased production of the anti-inflammatory cytokine IL-10 and potential instruction of regulatory T cells.
This type of MΦ polarization has previously been found in other, more common inflammatory skin diseases including atopic dermatitis and psoriasis, which both share some clinical and molecular features with aGvHD. 9Although skin-resident MΦs may produce high levels of IL-10 and TGF-β, providing anti-inflammatory functions in tissue repair, they are highly plastic, and promote tissue inflammation in Th-2-mediated conditions including contact dermatitis and asthma, 45,46 rather than dampening the inflammatory response.Interestingly, TGF-β is a well-characterized factor promoting establishment of tissue-resident memory T cells with cytotoxic functions in human skin. 47Therefore, while our data generally suggest an anti-inflammatory role of MΦs in previously untreated aGvHD, changes in the skin environment could lead to rapid repolarization in the later stages of disease, including steroid-refractory GvHD.Notably, therapeutic repolarization of MΦs -from anti-to proinflammatory -has been implicated in preclinical models as a potential treatment of solid tumours. 48Conversely, specific targeting to increase anti-inflammatory functions of the numerous MΦs in GvHD skin may present a novel therapeutic strategy to induce regulatory T cells after transplantation.
After HSCT, immunosuppressive therapy dampens inflammatory responses induced by loss of epithelial integrity, microbial dysbiosis and host-cell antigen presentation. 49,50Both host and donor antigen-presenting cells were shown to play a prominent role in GvHD tissue inflammation. 6,51,52In patients after sex-mismatched transplantation, we found host-and donor-derived MΦs to exhibit highly similar transcriptional profiles in GvHD-affected tissues.This is surprising, as tissue-resident MΦs differ from monocyte-derived macrophages by their bone marrow origin, and fulfil niche-specific homeostatic functions. 8However, MΦs are remarkably plastic immune cells 53 and the cytokine milieu present in GvHD-affected tissues may prompt incoming donor monocyte-derived MΦs to rapidly adapt and polarize towards a skin-resident phenotype.
In line with the diverse T-cell responses present in cutaneous GvHD subtypes, 42 we found MΦ responses in aGvHD to be distinct from chronic GvHD lesions, where we detected MΦs with low IL-10 production, high CCR7 expression and relatively higher IFN-γ production.Our data are compatible with findings of a recent study investigating chronic GvHD skin using bulk RNA sequencing, which described a cytokine signature dominated by IFN-γ in lichenoid and sclerotic skin lesions. 54Furthermore, clGvHD was associated with increased expression of TREM1, a molecule that characterizes proinflammatory MΦs in inflammatory bowel disease. 55In an IL-17-dependent murine model of scleroderma-like GvHD, donor MΦs expressed CD206 and TGF-β but not iNOS, identifying them as profibrotic skin-resident macrophages, 52 which suggests central differences between murine cutaneous GvHD models and human skin.
Overall, contrasting MNP-and T-cell phenotypes in acute and chronic GvHD potentially reflects differences in disease pathogenesis and pathophysiology: aGvHD develops upon barrier dysfunction and dysbiosis after conditioning therapy, 56 and the subsequent IL-4-containing cytokine milieu 57 may induce skin homing of donor monocyte-derived MΦs with tissue-remodelling properties.IL-1 and IL-6rich environments in damaged skin later after HSCT may propagate the proinflammatory MΦ bias in chronic GvHD.Finally, the lack of anti-inflammatory MΦs -for example as a long-lasting effect of total body irradiation 58 -may contribute to sclerosis in csGvHD.
In summary, our results demonstrate a major role of MΦs as cellular players in GvHD-affected skin, with possibilities for therapeutic targeting.In our study, both hostand donor-derived MΦs were present in skin lesions and showed few transcriptomic differences, indicating rapid differentiation and polarization of donor MNPs to tissue MΦs as they enter the skin in aGvHD.Furthermore, diverse MΦ responses in clinically distinct GvHD subtypes reflect the high plasticity of skin MΦs and illustrate the complex setting that is present in the cutaneous immune system after HSCT.

Figure 1
Figure 1 Cell populations in skin and blood of patients with acute graft-versus-host disease (aGvHD) and healthy donors.(a) Graphical overview of patient sampling.(b) UMAP clustering of aGvHD (n = 5) and healthy (n = 4) single-cell RNA-sequencing (scRNa-seq) datasets (left panel) and blood and skin cells according to cell barcode distribution (right panel).(c) UMAP showing Leiden clustering according to neighbourhood analysis in cells from (b).(d) Dot plot displaying cell type marker gene expression in Leiden clusters from (c).Data are shown as mean expression of respective genes in each group.(e) Leucocyte clusters identified by marker gene expression in (d).(f) Relative fraction of cell clusters identified in (e) derived from healthy (n = 4) and aGvHD (n = 5) datasets.Data are shown as mean cell fraction per cell type.HSC, haematopoietic stem cells; MNP, mononuclear phagocytes; NK, natural killer; UMAP, Uniform Manifold Approximation and Projection.

Figure 4
Figure 4 Interaction of macrophages (MΦ) and T cells in acute graft-versus-host disease (aGvHD) skin.(a) UMAP clustering of aGvHD skin-derived cells used for cell-cell interaction predictions.(b) Dot plot displaying putative receptor-ligand signalling of mononuclear phagocytes (MNP) to T cells (left) and T cells to MNP (right).Interactions were considered for P-values < 0.01 and significant gene upregulation in > 15 cells of each cluster.Data are shown as relative communication probability.(c) Image showing immunofluorescence labelling of CD68 + CD206 + and CD3 + CD45 + cells in aGvHD skin.Left panel is an overview image: arrows show cell-cell contact of MΦ and T cells, and line indicates dermoepidermal border (scale bar = 100 µm).Right panel shows magnification of MΦ-T-cell contact (scale bar = 20 µm).(d) Graphical representation of putative cell-cell interaction proteins between MΦ and T cells in GvHD skin.IFN-γ, interferon gamma; TNF-α, tumour necrosis factor alpha; TNFR, TNF receptor; UMAP, Uniform Manifold Approximation and Projection.

Figure 5 X
Figure 5 X-linked gene expression reveals similar transcriptional profiles of host and donor macrophages (MΦ).(a) Graphical representation of mismatched haematopoietic stem cell transplantation (HSCT) and sampling.(b) UMAP showing XIST gene expression in acute graft-versus-host disease (aGvHD) lesional skin MΦ of patients after mismatched HSCT (n = 2).(c) UMAP clustering (left panel) and relative cell fraction (right panel) of host and donor skin MΦ identified by XIST expression (b).(d) Dot plot showing proinflammatory and tissue-remodelling gene expression in cells from (c).IFNG, interferon gamma; IL, interleukin; scRNA-seq, single-cell RNA sequencing; TGFB, transforming growth factor beta; UMAP, Uniform Manifold Approximation and Projection.