Differential MicroRNA Expression Levels in Cutaneous Acute Graft-Versus-Host Disease

Allogeneic hematopoietic stem cell transplantation is a curative treatment for numerous hematological malignancies. However, acute graft-versus-host disease (aGvHD) is a major complication affecting 40–70% of all transplant patients, whereby the earliest and most frequent presentation is in the skin. MicroRNAs play a role in varied biological process and have been reported as potential biomarkers for aGvHD. More recently, microRNAs have received added attention as circulatory biomarkers that can be detected in biofluids. In this study, we performed global microRNA expression profiling using a discovery cohort of diagnostic cutaneous aGvHD biopsies (n = 5, stages 1–3) and healthy volunteers (n = 4), in order to identify a signature list of microRNAs that could be used as diagnostic biomarkers for cutaneous aGvHD. Candidate microRNAs (n = 8) were then further investigated in a validation cohort of post-HSCT skin biopsies (n = 17), pre-HSCT skin biopsies (n = 6) and normal controls (n = 6) for their association with aGvHD. Expression of let-7c (p = 0.014), miR-503-5p (p = 0.003), miR-365a-3p (p = 0.02), miR-34a-5p (p < 0.001) and miR-34a-3p (p = 0.006) were significantly differentially expressed between groups and significantly associated with survival outcome in post-HSCT patients (miR-503-5p ROC AUC = 0.83 p = 0.021, Log Rank p = 0.003; miR-34a-3p ROC AUC = 0.93, p = 0.003, Log Rank p = 0.004). There was no association with relapse. A statistical interaction between miR-34a-3p and miR-503-5p (p = 0.016) was diagnostic for aGvHD. Expression levels of the miR-34a-5p protein target p53 were assessed in the epidermis of the skin, and an inverse correlation was identified (r2 = 0.44, p = 0.039). Expression of the validated candidate microRNAs was also assessed at day 28 post-HSCT in the sera of transplant recipients, in order to investigate their potential as circulatory microRNA biomarkers. Expression of miR-503-5p (p = 0.001), miR-34a-5p (p = 0.005), and miR-34a-3p (p = 0.004) was significantly elevated in the sera of patients who developed aGvHD versus no-aGvHD (n = 30) and miR-503-5p was associated with overall survival (OS) (ROC AUC = 0.80, p = 0.04, Log Rank p = 0.041). In conclusion, this investigation reports that microRNA expression levels in clinical skin biopsies, obtained at the time of cutaneous aGvHD onset, show potential as diagnostic biomarkers for aGvHD and as predictive biomarkers for OS. In addition, the same microRNAs can be detected in the circulation and show predictive association with post-HSCT outcomes.

Allogeneic hematopoietic stem cell transplantation is a curative treatment for numerous hematological malignancies. However, acute graft-versus-host disease (aGvHD) is a major complication affecting 40-70% of all transplant patients, whereby the earliest and most frequent presentation is in the skin. MicroRNAs play a role in varied biological process and have been reported as potential biomarkers for aGvHD. More recently, microRNAs have received added attention as circulatory biomarkers that can be detected in biofluids. In this study, we performed global microRNA expression profiling using a discovery cohort of diagnostic cutaneous aGvHD biopsies (n = 5, stages 1-3) and healthy volunteers (n = 4), in order to identify a signature list of microRNAs that could be used as diagnostic biomarkers for cutaneous aGvHD. Candidate microRNAs (n = 8) were then further investigated in a validation cohort of post-HSCT skin biopsies (n = 17), pre-HSCT skin biopsies (n = 6) and normal controls (n = 6) for their association with aGvHD. Expression of let-7c (p = 0.014), miR-503-5p (p = 0.003), miR-365a-3p (p = 0.02), miR-34a-5p (p < 0.001) and miR-34a-3p (p = 0.006) were significantly differentially expressed between groups and significantly associated with survival outcome in post-HSCT patients (miR-503-5p ROC AUC = 0.83 p = 0.021, Log Rank p = 0.003; miR-34a-3p ROC AUC = 0.93, p = 0.003, Log Rank p = 0.004). There was no association with relapse. A statistical interaction between miR-34a-3p and miR-503-5p (p = 0.016) was diagnostic for aGvHD. Expression levels of the miR-34a-5p protein target p53 were assessed in the epidermis of the skin, and an inverse correlation was identified (r 2 = 0.44, p = 0.039). Expression of the validated candidate microRNAs was also assessed at day 28 post-HSCT in the sera of transplant recipients, in order to investigate their potential as circulatory microRNA biomarkers. Expression of miR-503-5p (p = 0.001), miR-34a-5p (p = 0.005), and miR-34a-3p (p = 0.004) was significantly elevated in the sera of patients who developed aGvHD versus no-aGvHD (n = 30) and miR-503-5p was associated with overall survival (OS) (ROC AUC = 0.80, p = 0.04, Log Rank p = 0.041). In conclusion, this investigation reports that microRNA expression levels in clinical skin biopsies, obtained at the time of cutaneous aGvHD onset, show potential as diagnostic biomarkers for aGvHD and as predictive biomarkers for OS. In addition, the same microRNAs can be detected in the circulation and show predictive association with post-HSCT outcomes.
Keywords: microrna, gvhD, biomarker, molecular profiling, cutaneous inTrODUcTiOn Skin is the largest organ of the human body, and it is more frequently affected by acute graft-versus-host disease (aGvHD) (1). Cutaneous aGvHD manifests as a sequence of events that can be divided into three phases. Phase I of cutaneous aGvHD involves the activation of resting Langerhans cells, due to the release of cytokines from damaged host tissues during the chemotherapy regimen and/ or irradiation therapy. Phase II of cutaneous aGvHD occurs post allogeneic hematopoietic stem cell transplantation (allo-HSCT) when donor T-cells recognize the antigens presented by activated host dendritic cells (DCs). During phase III, immature keratinocytes undergo apoptosis due to the "cytokine storm, " which exacerbates histological skin damage (2). Cutaneous aGvHD severity is determined by trained independent histopathologists (3).
Currently, biomarkers that can be reliably used in the clinic to stratify patients for improved treatment response, diagnose aGvHD, or predict patient survival outcome are absent. Recently, the roles of several microRNAs have been investigated in aGvHD. miR-146a has been shown to be downregulated in severe-aGvHD murine models, and its target Traf6 elevated, leading to activation of the NFκB pathway (4). Likewise, elevated levels of miR-155 were expressed in the gut of severe-aGvHD patients when compared with normal biopsies (5). miR-34a has been shown to be expressed at higher levels post-transplant in the gut of patients with aGvHD grades II-IV compared with patients with grades 0-I (6). miR-100 expression was shown to be increased in the gut of mice without aGvHD, therefore suggesting a protective function (7). A predictive signature list of microRNAs (miR-423, miR-199a-3p, miR-93*, and miR-377) for aGvHD has previously been detected and validated in the plasma/sera of allo-HSCT patients (8,9). These microRNAs were able to discriminate patients who had developed aGvHD from those who had not (non-aGvHD group). The microRNAs were associated with reduced overall survival (OS) and aGvHD severity (8). Circulatory microRNAs have also been globally profiled in the context of aGvHD (10) and tissue-specific microRNA expression patterns have been examined in a rat model of aGvHD (11). Circulatory microRNAs show great promise as potential biomarkers, due to the ease and non-invasive nature of sample collection.
In this study, in the discovery cohort, skin biopsies were collected from normal controls, as well as pre-transplant and post-transplant patients at the time of aGvHD disease onset. The samples were analyzed to determine the diagnostic potential of microRNAs in cutaneous aGvHD and their prognostic potential for predicting patient survival. Global microRNA profiling using RT-qPCR was performed to identify a signature microRNA list, which was then further validated in a larger patient validation cohort. Expression of miR-503-5p, miR-34a-5p, miR-34a-3p, and let-7c-5p was associated with aGvHD incidence and survival. Results showed that miR-34a-5p was the most promising micro-RNA for determining aGvHD severity, based on its significant differential expression levels across all the comparative groups. The protein targets (p53 and c-Myc) of miR-34a-5p were investigated by immunohistochemistry and showed association with expression of miR-34a-5p and skin aGvHD stage. Studies have shown the existence of a positive feedback loop between p53 and miR-34a (12). Likewise, c-Myc expression is suppressed by miR-34a (13). Interestingly, c-Myc has numerous functions in hematopoiesis such as regulating hematopoietic stem cell regeneration and differentiation (14). c-Myc protein is a transcription factor that is constitutively expressed in many tumors and impacts microRNA biogenesis steps in the nucleus (15). Moreover, c-Myc, p53, and miR-34a are all involved in the p53 pathway and the epithelial-mesenchymal transition network (16). In the context of allo-HSCT, p53-deficient mice that develop aGvHD have a better OS when compared with mice with at least one copy of the gene (17). Finally, analysis of the expression of candidate micro-RNAs was assessed in non-invasive serum samples, in order to evaluate their circulatory biomarker capacity. This would be of great potential clinical benefit, as biofluid samples are preferable to using invasive diagnostic skin biopsies.
Overall, the aims of this investigation were (1) to identify a list of microRNAs with significant potential to diagnose cutaneous aGvHD and to explore their impact on their predicted protein targets, (2) to predict OS and risk of relapse, (3) to further explore the circulatory biomarker potential of candidate microRNAs.

Patient criteria
Research was granted ethical approval by the Newcastle and North Tyneside Research Ethics Committee (REC Ref: 14/NE/1136 and 07/H0906/131). Participants gave full informed written consent for their samples to be used for research purposes. Skin bio psies were obtained from allo-HSCT patients pre-(n = 6) and posttransplantation (n = 17) as well as from healthy volunteers (n = 6) with informed written consent. Skin histopathological aGvHD stage was evaluated by two independent histopathologists who were blinded to the origin of the sample, using the Glucksburg (3) criteria. All skin biopsies were collected pre-transplantation and at the time of aGvHD onset. Blood samples were collected from HSCT patients (n = 30) at day 28 (D28) post-HSCT in 7-ml vacutainers containing no anti-coagulant from patients undergoing allo-HSCT (years 2009-2013). Clinical details of the cohorts are shown in Tables 1-3. To verify that differential microRNA expression levels were only due to either aGvHD or a result of transplantation, stringent criteria were set that excluded any patient who had received donor-lymphocyte infusion, had cyclosporine withdrawal, had received steroids, or had developed late onset aGvHD (post-100 days from the date of transplantation). Thus, only patients with classic aGvHD (allo-HSCT to onset ≤100 days) were included in all analyses.

Total rna extraction
Skin biopsies (1-4 mm) collected from patients and healthy volunteers were stored in RNA later buffer at −20°C until RNA extraction. Total RNA was extracted as per the manufacturer's protocol from clinical skin biopsies using the mirVana microRNA Isolation kit (Life Technologies, USA). Blood samples were left to clot, the supernatant centrifuged at 500 × g for 5 min and stored at −80°C. Serum aliquots were centrifuged at 4,500 × g for 15 min to remove platelets before use. Total RNA was extracted using the Cycler) conditions comprised of 42°C for 60 min, 95°C for 5 min, and then samples were stored at −4°C until use. The RT-PCR reaction step consisted of microRNA-specific forward and reverse primers (see Table S1 in Supplementary Material) to amplify individual microRNA of interest. ExiLENT SYBR ® Green (Product No: 203402) was used for the detection of amplified microRNA products, which was followed by melting curve analysis upon standard qPCR completion as per the manufacturer's protocol. Normfinder was used to determine the most suitable microRNA for normalization of the validation data (18). miR-103-3p and SNORD48 were used as endogenous controls, with analysis based on the geometric mean for normalization of the expression results. For serum samples, microRNA-specific cDNA was generated using the TaqMan ® MicroRNA Reverse Transcription Kit (Applied Biosystems ® ) and TaqMan ® MicroRNA Assays (Applied Biosystems ® ), according to the supplier's instructions. Each reverse transcription contained 5 µl of total RNA and heating was performed according to supplier recommended cycling conditions. MicroRNA-specific qRT-PCR reactions comprised of TaqMan ® MicroRNA Assays (Applied Biosystems ® ) and SensiFAST™ Probe Hi-Rox Reagent (Bioline). Each 10 µl reaction was carried out in triplicate. MicroRNA expression was normalized to two previously identified endogenous controls (U6 and HY3) (19). Reactions were cycled using the 7900 Real-Time PCR System (Applied Biosystems ® ), using manufacturer's recommended cycling conditions.

Quickscore Method for assessment of immunohistopathological samples
The "quick score method" is a semi-quantitative method (20), used for assessing the percentage of positively stained cells for both p53 and c-Myc proteins. A consensus score was calculated as each section was assessed by two independent researchers, who were blinded to the origin of the samples. The score for each biopsy was calculated by considering both the intensity and the proportion of cells (see Table S2 in Supplementary Material) within that intensity [Quickscore formula = (Intensity 0 × Proportion) + (Intensity 1 × Proportion) + (Intensity 2 × Proportion) + (Intensity 3 × Proportion)]. The whole section was used for calculating the percentage positivity scores. Since miR-34a expression was quantified in both the epidermis and dermis, quick scores were calculated (1) for the entire section regardless of the skin layer and (2) in each skin layer individually.

statistical analysis
Unsupervised hierarchical clustering analysis was performed using RStudio™ Version 0.98.501 (RStudio, Inc., USA). The global profiling results were analyzed using one-way ANOVA on SPSS  aGvHD severity, the Jonckheere-Terpstra test (SPSS version 21) was also performed for the microRNAs in the signature list. Generalized linear models (ordinal logistic regression) (GLMs) were performed to find the significance and statistical interactions between microRNAs in the signature list using SPSS version 21 (IBM, SPSS Inc., USA). The main effect is statistically defined as an effect of an independent variable on the outcome, while an interaction is a measure of two or more independent variables on the dependent outcome (21). In this scenario, the main effect was the effect of the individual miRNA expressions or significant clinical risk factors on aGvHD outcome and the interaction was a measure of the effect of one independent miRNA expression level as it differed at every condition in comparison to the other independent miRNA expression levels (21). In order to predict aGvHD outcome from a list of covariates, GLMs were built using the binary and/or ordinal logistic regressions. The equation for the binary or ordinal regression models that was used for determining aGvHD grade (outcome) was aGvHD grade (cumulative logit function) = β0 + β1 (miRNA expression 1) + β2 (miRNA expression 2) + β3 (miRNA expression 1 × miRNA expression 2). β0 = intercept coefficient and β1-3 = the coefficient of each miRNA expression. Rotational 3D scatterplots were generated using JMP ® In the discovery cohort, skin samples with similar microRNA expression values were grouped by performing unsupervised hierarchical clustering. Only microRNAs that were expressed in all skin biopsies, including the healthy control biopsies (n = 186), were included in the final analysis. MicroRNA expression in the skin biopsies was used to create a dendrogram and heatmap (Figure 1). The microRNA expression profiles showed considerable patient-to-patient variation, as depicted by the height of the branches, but the most pronounced partition on the dendrogram led to segregation of the skin biopsies into two distinct clusters: the healthy control group in one branch and the allo-HSCT group in the other (Figure 1). The heatmap showed three distinct expression regions, whereby highly upregulated microRNAs clustered in the top branch, and the downregulated microRNAs clustered in the lower section of the heat map. miR-720 was highly expressed in all nine skin biopsies (Mean ΔCp = 9.10) (Figure 1) and has been reported to be an important regulator of T cell proliferation (22). miR-34a-5p expression was significantly different in the healthy controls compared with the allo-HSCT group [control mean ΔCp = 1.97 (range = 1.74-2.25), allo-HSCT mean ΔCp = 3.30 (range = 2.57-4.07); p < 0.001] (Figure 1). miR-451-5p is highly expressed in erythrocytes (23) and its levels were highly variable in all the skin biopsies (range = −1.33 to 4.90, SD = 2.28) (Figure 1). This could be attributed to the differences in the vascular network present in each skin biopsy.

lower cutaneous mir-503-5p and mir-34a-3p expression is associated With improved Os
Receiver operating characteristic analysis of the validation cohort, based on individual microRNA expression, showed that miR-503-5p and miR-34a-3p predicted for OS (miR-503-5p AUC = 0.83, p = 0.021, miR-34a-3p AUC = 0.93, p = 0.003) (Figures 3A,B). MicroRNA expression was dichotomized (low or high expression), based on the optimal thresholds evaluated by ROC analysis. The K.M method was used to generate survival plots. Low expression of miR-503-5p and miR-34a-3p FigUre 3 | Association between cutaneous miR-503-5p and miR-34a-3p expression with overall survival (OS) and skin acute graft-versus-host disease (aGVHD) severity in the validation cohort. (a-D) Expression of miR-503-5p and miR-34a-3p in the skin validation cohort (n = 17) was analyzed according to OS. (a,b) miR-503-5p and miR-34a-3p expression was associated with OS according to receiver operator characteristic (ROC) analysis. ROC curves detailing area under the curve (AUC) and p-value significance are shown. (c,D) miR-503-5p and miR-34a-3p expression was dichotomized (low or high expression), based on the optimal thresholds evaluated by ROC analysis. Kaplan-Meier survival curves for OS were then generated, based on dichotomized microRNA expression. p-Values were calculated using the Log-Rank test. For both microRNAs, high expression was significantly associated with poor OS. (e) In patients with high levels of miR-34a-3p, miR-503-5p expression was associated with skin histopathological aGvHD. (F) However, in patients with low levels of miR-34a-3p, there was no association between miR-503-5p expression and skin aGvHD severity. (e,F) Histopathological grades are shown on the y-axis. The x-axis denotes microRNA expression. miR-503-5p expression levels were dichotomized based on miR-34a-3p expression.  Figures 3C,D).
To obtain a set of predictors in a multivariate model for OS, variable selection was performed using both the clinical risk factors and microRNAs. From the preliminary univariate variable selection (which included patient age, donor-patient relationship, patient gender, CMV status, graft regimen, skin aGvHD stage, mismatches in HLA Class I, and underlying disease), clinical risk factors with p < 0.2 (24) were chosen as candidates in a "Forward Likelihood Ratio" stepwise Cox regression procedure alongside the microRNAs [i.e., patient CMV status (p = 0.146), mismatches in HLA Class I (p = 0.187), skin histopathological aGvHD stage (p = 0.115), miR-503-5p (p = 0.014), and miR-34a-3p (p = 0.026) expression] ( statistical interaction between cutaneous mir-503-5p and mir-34a-3p is associated With skin agvhD stage Statistical interactions between all microRNAs were also assessed in the validation cohort. Ordinal logistic regression analysis of skin histopathological aGvHD stage was performed incorporating both the main effect and interaction of all the validated microRNAs in a stepwise manner. Results showed that there was a statistically significant interaction between miR-503-5p and miR-34a-3p expression (p = 0.016), with miR-503-5p having the most significant effect (p = 0.011) and miR-34a-3p as the moderator (p = 0.020) in this interaction. Clinical covariates (as for the survival analysis) were considered in the stepwise procedure, but none were significant in the final model.
A conditional relationship was also identified between miR-503-5p and miR-34a-3p, such that when high levels of miR-34a-3p were expressed, higher levels of miR-503-5p were more clearly associated with severity of skin histopathological aGvHD stage ( Figure 3E). However, the relationship between miR-503-5p and skin histopathological aGvHD severity was less apparent when levels of miR-34a-3p expression were low (Figure 3F). This finding requires further investigation to elucidate whether a clinical or biological factor was the reason for this trend. To further understand the conditional interaction between miR-503-5p and miR-34a-3p, rotational 3D scatterplots were generated ( Figure S3 in Supplementary Material). Since miR-503-5p and miR-34a-3p were significantly associated with OS, non-parametric density contours based on the patient survival status were included, to aid visualization. The plots demonstrated the interaction between miR-503-5p and miR-34a-3p expression. None of the other microRNAs were associated with skin aGvHD stage, OS, or relapse.

c-Myc and p53-Positive cells are Present in cutaneous biopsies of allo-hscT Patients
Immunohistochemical analysis was performed to test whether c-Myc and p53 proteins, both of which are miR-34a-5p targets, were expressed in cutaneous biopsies from allo-HSCT patients pre-and post-transplantation. The majority of the activated c-Myc-positive cells, particularly those with high c-Myc intensity, were present in the basement membrane of skin histopathological stage 0 and stage 1 aGvHD biopsies (n = 9) (Figure 4). In stage 2 and stage 3 aGvHD biopsies (n = 7), the complete epidermis stained positive for c-Myc and very few negatively staining cells were present. A limited number of cells demonstrated cytoplasmic-only staining for c-Myc, while the majority showed both cytoplasmic and nuclear staining.
Likewise, cells that positively stained for p53 protein (Figure 4) were also analyzed for differences between pre-and postallo-HSCT skin biopsies. Visualization of the sections showed positive cytoplasmic-nuclear staining for p53 under all histopathological conditions (pre-transplantation to stage 3 aGvHD). p53-positive cells were localized in the epidermis. In healthy volunteers, p53-positive cells are confined to the basement membrane (25). Only occasional cells in the dermis were p53-positive. The majority of p53-positive cells stained with strong to moderate intensity. The dermis had less than 4% p53-positive cells and was therefore considered as p53-negative. Thus, for p53, the quick score method was calculated based only on the epidermis.   severity. (a,b) The first row shows the two pre-transplant skin sections that were taken from the individuals with cutaneous histopathological aGvHD stage 0 and stage 1 (second row), respectively. Stage 3 skin sections showed severe cutaneous aGvHD, with separation between the epidermis and dermis. For both c-Myc and p53, the number of cells that infiltrated the dermis gradually increased from no GVHD (stage 0) to severe skin aGvHD (stage 3). Breast carcinoma tissues were used as a negative and positive control for c-Myc and p53. The images were taken at 20× magnification using AxioCamMR3. miR-34a-5p expression alone positively correlated with the percentage of p53-positive cells in the epidermis (rs = 0.44, p = 0.039) ( Figure S4 in Supplementary Material). In postallo-HSCT skin biopsies only (n = 17), there was no significant correlation between p53-positive cells and microRNA expression. There was a significant difference between the percentage p53-positive cells pre-transplantation and skin aGvHD stages 2 and 3 (p = 0.045) ( Figure S4 in Supplementary Material). Jonckheere's trend test confirmed the significant trend toward higher p53-positive cells with increasing skin aGvHD stage post allo-HSCT (p = 0.009).
When microRNAs were assessed in relation to individual skin aGvHD stage or clinical aGvHD grade, there was no significant difference in expression between disease stages or grades (p > 0.05, data not shown), indicating the microRNAs to be associated with skin and overall aGvHD incidence, but not severity (data not shown).

DiscUssiOn
The critical role of microRNA in disease pathology and development has been highlighted in numerous studies to-date (26). MicroRNAs are cell-specific, stable, and therefore ideal biomarkers (27). Thus, this study aimed to identify a signature list of microRNAs that could discriminate mild (stage 1) from moderate to severe (stages 2 and 3) cutaneous aGvHD. Global microRNA profiling by RT-qPCR was carried out on normal and post-HSCT biopsies taken from patients with cutaneous aGvHD. Eight microRNAs that were differentially expressed across the groups (miR-142-3p, miR-34a-5p, miR-503-5p, let-7c-5p, miR-21-3p, miR-365a-3p, miR-23b-3p, and miR-24-3p) were selected for individual validation by RT-qPCR in healthy controls and pre-transplant and post-transplant skin biopsies from patients with and without aGvHD. Four microRNAs (miR-34a-5p, miR-34a-3p, let-7c-5p, and miR-503-5p) were further investigated in serum samples for their circulatory biomarker potential for skin and overall aGvHD.
Unsupervised hierarchical clustering of the global microRNA expression levels showed that there were two distinct clusters, supporting the hypothesis that healthy controls and allo-HSCT patients have different cutaneous microRNA expression patterns. The findings suggested that microRNAs could be deregulated as a result of both the allo-HSCT procedure and aGvHD disease onset. This is in agreement with previous studies that have reported differential expression of circulatory microRNAs post-cytotoxic therapy prior to HSCT, as well as in relation to aGvHD onset (10,28). In the subsequent validation study, five microRNAs of the eight observed in the discovery cohort retained their significance (let-7c-5p, miR-503-5p, miR-365a-3p, and miR-34a-5p/-3p). Interestingly, only let-7c was significantly differentially expressed between normal controls and pre-HSCT patients (lower in pre-HSCT), suggesting expression of this microRNA to be altered as a results of underlying disease. The results showed that let-7c-5p expression increased after transplantation, but it was surprising to find that levels in skin biopsies with skin aGvHD histopathological stages 2-3 were approximately the same as that of healthy controls. A previous study demonstrated that let-7c-5p is highly expressed in CD4 + T-cells in healthy individuals (29). Results in the healthy controls were, therefore, in concordance with this study. Swaminathan et al. (29) have shown that the expression of IL-10 protein could be modulated by let-7c-5p. IL-10 is an anti-inflammatory cytokine that has been extensively studied in GvHD (30). At this stage, it could be postulated that let-7c-5p might play a role in immune-regulation rather than cutaneous aGvHD. As the skin is involved in ~80% of GvHD cases, further understanding the repertoire of microRNA expression in a cutaneous aGvHD setting is important in order to better understand the complex nature of this disease. Although the data presented here strengthens the potential of miR-503-5p, miR-34a-5p, miR-34a-3p, and let-7c-5p for use as prospective cutaneous aGvHD biomarkers, the necessity for invasive biopsies is not advantageous. Therefore, studies were extended to assess their expression in serum samples pre-and post-transplant. Results showed that miR-503-5p, miR-34a-5p, and miR-34a-3p retained their biomarker potential with respect to skin and overall clinical aGvHD incidence, as demonstrated by high expression in aGvHD patients at D28 post-HSCT and subsequent ROC analysis.
When microRNA expression was assessed in relation to post-HSCT survival, lower miR-503-5p expression levels were associated with improved OS when assessed in both the skin and the serum. miR-503 has been shown to negatively regulate CD40 gene expression when malignant human monocyte-derived cells (U937 cells) were irradiated (31). The function of the CD40-CD40 ligand (CD40L) pathway has been investigated in GvHD models (32). T and NK cells are activated when CD40 and CD40L ligate, which results in the release of established pro-inflammatory cytokines such as IFN-γ, IL-2, and IL-12 (33). Studies have shown that anti-CD40L antibodies can decrease GvHD severity (32). Activated CD4 + T-cells express CD40L that can be targeted by anti-CD40L antibodies. These antibodies block CD4 + T-cells, thus making them tolerant to host antigens, resulting in lower GvHD (34). Moreover, CD40 is also expressed in human keratinocytes in the skin (35). Since miR-503 and miR-424 are polycistronic, they have very similar seed sequences and have been shown to target the same genes and also be involved in monocyte to macrophage differentiation in an AML cell line study (36). In a GvHD-related study, investigators have shown that miR-424 levels are higher in cord blood CD4 + and CD8 + cells when compared to adult peripheral blood cells (37). It was also demonstrated that miR-424 is overexpressed in adult blood CD14 + monocytes, when compared with CD4 + and CD8 + T-cells (37). It is known that during inflammation monocytes are recruited to the affected site and differentiate to either macrophages and/or DCs [as reviewed in Ref. (38)]. Therefore, miR-503-5p could be overexpressed to target CD40 gene expression and thus, inhibit T-cell activation. Its expression levels may also be upregulated as more monocytes are recruited and migrate from the blood to the host skin tissue.
Likewise, miR-365a-3p was under-expressed post-transplantation. Xu et al. (39) have shown that miR-365a-3p is a negative regulator of IL-6, which is a cytokine involved in the control of immune responses (40). Studies have demonstrated that IL-6 levels are elevated at the time of aGvHD onset in the sera of patients (41). Skin fibroblasts have also been shown to express IL-6 under inflammatory conditions (42). It may be possible that the decrease in miR-365a-3p expression post-transplant causes IL-6 level to increase in aGvHD patients. miR-34a-5p was significantly differentially expressed between pre-transplant and skin stages 0-1 as well as stages 2-3 aGvHD biopsies. Similarly, it was differentially expressed between pre-transplant and grades 0-I and grades II-IV aGvHD when assessed according to overall clinical aGvHD. Expression was also significantly higher in D28 serum samples in patients who developed overall aGvHD or skin aGvHD. However, no significant difference was observed between mild and moderate aGvHD in either the skin or the serum. Interestingly, miR-34a was not expressed at all when tested in a small cohort of scleroderma skin biopsies (n = 5, data not shown). Thus, the findings suggest that miR-34a-5p may be upregulated as a result of the allo-HSCT, as well as inflammation due to GvHD. miR-34a has previously been associated with aGvHD in Fanconi anemia (FA) (6). Patients with FA are particularly susceptible to developing high-grade aGvHD, thought to be mediated through impaired DNA repair processes leading to increased apoptosis (43). It was demonstrated that miR-34a expression was higher in aGvHD compared with no-aGvHD, in both the skin and gut and that expression directly correlated with the number of apoptotic cells present in the gut (6). This supports the findings of this study in which miR-34a levels were elevated in aGvHD. Thus, serum miR-34a levels may be indicative of apoptosis occurring in target GvHD organs, thus serving as a potential serum biomarker of aGvHD. In allo-HSCT and GvHD, T-cell receptors (TCRs) are stimulated during the GvH cycle when T-cells recognize HLA mismatches between the patient and donor. This results in partial activation of the T-cells (44). Recently, it has been shown that overexpression of miR-34a is related to higher TCR T-cell activation. The positive correlation between miR-34a expression and TCR stimulation has been linked to diacylglycerol kinase-zeta (DAGK-ζ), which is an enzyme responsible for metabolizing diacylglycerol (DAG) (45). Lower DAGK-ζ results in higher T-cell activation as the cells become sensitive to TCR triggers, and miR-34a directly targets DAGK-ζ mRNA, leading to more un-metabolized DAG, which stimulates TCRs and enhanced T-cell activation (45).
In this study, low expression levels of miR-503-5p and miR-34a-3p were indicative of improved OS. Interestingly, when considering skin histopathological aGvHD stage as the outcome, a significant statistical interaction was present between miR-503-5p and miR-34a-3p and this was distinctly evident when survival status was added onto 3D plots for visualization. The biological mechanism underlying the statistical interaction needs further investigation, but our results are consistent with a "gate" mechanism, whereby at high levels of miR-34a-3p, miR-503-5p is the dominant effector and most patients have a lower chance of survival, while higher levels of miR-503-5p accompany more severe aGvHD. By contrast, as miR-34a-3p expression decreased, the effect of miR-503-5p was negligible, most patients survived, and the severity of aGvHD increased. It would be interesting to assess whether cutaneous microRNA expression patterns directly reflect aGvHD severity. However, patients who develop severe aGvHD are rare in the Newcastle Transplantation facility due to the controlled conditioning and prophylaxis treatments. Almost every patient in this center is administered with alemtuzumab (Campath). In this study, only one patient from the cutaneous cohort was diagnosed with skin aGvHD stage 3. Therefore, we could only analyze stage 2 and 1 versus pre-transplant biopsies. None of the microRNAs were associated with risk of relapse and it was not possible to perform competing risk analysis, due to the few cases present per group. It will therefore be important to validate these results in larger, multi-center studies that incorporate a higher number of high stage 3-4 aGvHD patients. Furthermore, although it is encouraging that the potential biomarkers identified in this study can be confirmed in an independent cohort, additional expanded cohorts will allow for sub-stratification in order to investigate the impact of heterogeneous conditioning regimens and pre-transplant prophylaxis, as well as improve the strength of multivariate analyses.
In this study, whole skin biopsies were used for microRNA expression analysis. However, aGvHD mainly attacks the epidermal layer of the skin (2). Thus, the microRNA expression profiles observed are collective and may decrease the discriminatory specificity of the results. In addition, the proportion and types of cells in the epidermis and dermis are different. Thus, in order to fully understand the biology of the microRNAs, it may be essential to assess their expression levels separately in each layer of the skin, separated by laser capture microdissection.
The protein expression of microRNA targets was investigated in the same cohort. Results showed that c-Myc and p53-positive cells were present in cutaneous biopsies of allo-HSCT patients. The proto-oncogene c-Myc is involved in fundamental processes such as differentiation, apoptosis, and cellular proliferation (46,47). In this study, no significant association was determined between miR-34a expression and c-Myc-positivity; however, significant positive correlation between p53-positive cells and miR-34a-5p expression was observed. This result is in support of the p53-positive feedback loop with miR-34a (12). p53-positive cells were observed even in pre-transplantation skin biopsies, and overall the number of p53-positive cells increased with aGvHD severity (p = 0.045). Taken together, these results may suggest that pre-transplant activation of p53 is due to the conditioning regimen, which is not sufficient to trigger miR-34a-5p overexpression. In the post allo-HSCT skin, p53 is overexpressed due to the conditioning regimen as well as the GvH reaction, which led to higher miR-34a-5p expression and therefore positive correlation in their expression.
In summary, this research has shown that miR-34a, miR-503-5p, let-7c-5p, and miR-365a-3p may be implicated in both the transplant procedure and the pathogenesis of aGvHD. Expression of miR-503-5p, miR-34a-5p, and miR-34a-3p in serum demonstrated potential non-invasive biomarker potential. Further investigation into the experimentally validated targets of miR-34a, miR-503-5p, let-7c-5p, and miR-365a-3p may shed light on their impact at the protein level and their roles in maintaining skin homeostasis. These results need to be validated in a larger, multicentre cohort.

eThics sTaTeMenT
Research was granted ethical approval by the Newcastle and North Tyneside Research Ethics Committee. Participants gave full informed written consent for their samples to be used for research purposes. aUThOr cOnTribUTiOns SA designed and performed the experiments, evaluated the data, and helped prepare the manuscript. JN assisted with designing and performing experiments and drafting of the manuscript. LB performed experiments in serum samples. AJ, PR, and X-NW graded the patients GvHD and non-GvHD skin sections. CL performed clinical data collection and advised on statistical interactions analysis. KP advised and performed statistical analysis. SO and JL provided scleroderma samples. MC contributed to clinical samples and data collection. RC and AD developed the overall concept, supervised the research, and prepared the manuscript. All the authors approved the final manuscript.