Identification of Critical Transcriptomic Signaling Pathways in Patients with H Syndrome and Rosai-Dorfman Disease

Patient blood samples used for this study were obtained with ethics approval (REC 10/H1306/88, National Research Ethics Committee Yorkshire and Humber–Leeds East), and all the studies involving human samples from healthy control (HC) volunteers were approved by the Health Research Authority (REC reference 17/YH/0084). Patients were recruited from the Department of Clinical Immunology and Allergy, St James’s University Hospitals, Leeds, UK. Informed written consent was obtained from all participants at the time of the sample collection. Age- and sex-matched HC were recruited from the St James’s University Hospitals, Leeds, UK. Details for the tissue biopsies can be found in Supplementary Table 1.

Blood samples were collected in EDTA precoated tubes and processed within the next 4 h. Peripheral blood mononuclear cells (PBMCs) were isolated from whole blood using a standard density gradient centrifugation method. Blood was mixed with an equal volume of DPBS (without Ca²⁺ and Mg²⁺, containing 2% heat-inactivated fetal bovine serum (FBS)) and carefully layered onto Lymphoprep (StemCell) and centrifuged at 1100×g for 20 min without brakes. The white buffy layer was collected and washed twice in DPBS (2% FBS) by centrifuging at 150×g for 10 min without brakes, to remove platelets. After PBMCs were obtained, CD14⁺ monocytes were isolated by immunomagnetic negative selection, using the EasySep Human Monocyte Isolation Kit (StemCell). Isolated monocytes were immediately lysed with TRIzol or differentiated to macrophages as described in the next sections. Table 1

Isolated human monocytes were cultured in complete RPMI medium (10% FBS, 50 U/ml penicillin, 50 μg/ml streptomycin and 1% L-glutamine) (Merck) supplemented with 20 ng/mL human GM-CSF (PeproTech) for macrophage differentiation and incubated for 6 days, adding fresh media on day 3. On day 6, M0 macrophages were activated with 100 ng/mL human IFN-γ (PeproTech) and 50 ng/mL LPS, for M1 macrophage polarization, or 20 ng/mL IL-13 (PeproTech) and IL-4 (PeproTech), for M2 macrophage polarization, and incubated for 24 h. Monocytes were initially seeded at a density of 0.5 × 10⁶ and cultured in tissue culture-treated 12 well plates.

Characterization of the macrophages was done through flow cytometry. On day 7, cells were washed twice with DBPS and detached using DPBS with EDTA 10 mM. Cells were washed with DPBS and resuspended in brilliant stain buffer (BSB) with human and mouse serum for 20 min on ice. Cells were stained with the surface markers for M1 (CD64⁺, CD80⁺, and CD86⁺) and M2 (CD64⁺, CD206^+, and CD209⁺) for 30 min on ice. Finally, cells were resuspended in BSB in FACS collection tubes, and samples were run in the CytoFLEX-LS (Beckman Coulter). All antibodies used are listed in detail in the table of regents.

Total RNA was obtained by using TRIzol and Phasemaker Tubes (Thermo Fisher Scientific) according to the manufacturers’ protocol. RNA quality and quantity were further determined by NanoDrop spectrophotometer and with the Agilent TapeStation (Agilent Technologies).

Library preparation and RNA sequencing (RNA-seq) were performed on an Illumina Novaseq and generating 150 bp paired-ended reads (Novogene Bioinformatics Technology Co., LTD). An average of 55.0 million raw reads were generated per sample (effective rate average 97.6% and Q30 average of 94.5%). RNA-seq raw reads were trimmed using TrimGalore [12] to remove library adaptors sequences and low-quality reads (QV ≤ 30). High-quality reads were mapped to the reference genome (GRCh38/hg38) using STAR [13], and then the Cufflinks-Cuffdiff (v2.2.1) pipeline [14] was employed to perform transcriptome assembly, normalization, and differential expression (DE) analyses. Monocytes and macrophages showed an average of 89.3% uniquely mapped reads, whereas tissue samples showed an average of 48.5%. Processed and raw RNA-seq data have been deposited to the NCBI Gene Expression Omnibus (GEO) database (accession number: GSE155697). Gene-ontology (GO) enrichment analysis of the DEGs was performed using the DAVID (v6.8) functional annotation analysis tool [15].

The supernatant from cultured macrophages and patients’ serum were frozen and stored at − 80 °C. TNF, IL-12, IL-6, IL-1β, IFNγ, and IL-2 (R+D Systems Fluorokinemap) cytokines were measured by using the multiplex Luminex analyzer (Bio-Plex, Bio-Rad, UK). Serum was collected immediately after the sample collection, and supernatants from macrophages were collected before processing the cells.

Flow cytometry and cytokine analysis were analyzed by multiple independent t tests, and statistical significance determined by the Holm-Sidak method. Each parameter was analyzed individually without assuming a consistent SD. GraphPad Prism 8 software was used to do all analyses. P values of < 0.05 were statistically significant. Nonparametric tests were used to compare the medians between groups.

Details of all reagents can be found in the Supplementary Table 2.

To study the transcriptomic profile of monocytes, M0, M1, and M2 macrophages, we obtained CD14⁺ monocytes from 2 HS, 2 SAID patients, and 5 HC and differentiated the cells according to the protocol summarized in Fig. 1a. To determine whether the macrophage differentiation was successful, the global transcriptomic differences between the different cell types, in each group, were assessed and visualized using multidimensional scaling (MDS) plots (Fig. 1b–d). Monocytes, M0, and M1 macrophages were clearly separated, by at least three units on the MDS plot, for each sample (Fig. 1b). Although the M2 and M0 macrophages did not show a distinct global transcriptomic difference by MDS, we found this to be consistent with another published dataset available of M2 differentiated from M0 macrophages [18]. Moreover, expression of the surface markers CD80/CD86 and CD206/CD209 showed high expression in M1 and M2 macrophages, respectively. To further corroborate the differences between the cell types, pairwise comparisons were made between the different cell types in each group (Fig. 1e–g). As shown in the Venn diagrams, a large number of transcripts were found to be differentially expressed in M0, M1, and M2, when compared to monocytes; for example, HC (n = 5) and HS1 showed 2210 and 2045 DEGs in M1 macrophages, respectively, and M1 macrophages from patient HS2 presented 905 DEGs (Fig. 1e–g top). When compared to M0 macrophages, the transcriptomic differences of monocytes were more evident, with 2714 (HC), 2337 (HS1), and 1273 (HS2) DEGs (Fig. 1e–g bottom). Similarly, when comparing to monocytes, M2 macrophages showed 344 (HC), 440 (HS1), and 128 (HS2) DEGs in each respective group (Fig. 1e–g top).

The number of DEGs was reduced in the different cell types from the uSAID and TRAPS patients. We observed 80 and 149 unique DEGs in M2 and M1 macrophages from uSAID patient and 102 and 93 in M2 and M1 macrophages from the patient with TRAPS (Supplemental Fig. 1C-D). The remaining comparisons of all the DEGs in each cell type from each sample group are shown in Supplemental Fig. 1E-G.

To determine which cellular processes might be responsible for the differences in the inflammatory phenotypes observed between the two HS patients, we first compared the global transcriptomic profile of cells derived from HS patients. While heterogeneity is an important factor that influences the transcriptomic profiles of cell subset in different individuals [19], we still observed a clear degree of separation in the monocytes from the two HS patients when compared to the HC samples (Fig. 1c) and to a lesser degree in the other cell subsets (Fig. 1d; Supplemental Fig. 1A-B).

We next conducted pairwise comparisons of the DEGs between both HS patients, as well as independently against the HC samples, for each cell type. Monocytes from HS patients shared a cluster of upregulated genes when compared to the HC controls (Fig. 2a), and a total of 73 and 70 DEGs were shown in patients HS1 and HS2, respectively (Supplement Table 4). The shared upregulated DEGs in both HS patients included SERPINA1, TNFAIP6, and SLC39A8 (Fig. 2b). When monocytes from patients HS1 and HS2 were compared, we found 77 DEGs with several HLA genes being overexpressed in patient HS2, as well as B2M, which encodes β2-Microglobulin, a component of MHC class I molecules (Fig. 2b).

When comparing transcriptomic profiles of M0 macrophages, patient HS1 showed a profile resembling the HC samples rather than HS2 (Fig. 2c). The number of DEGs in M0 macrophages was 58 in HS1 and 154 in HS2, when compared to HC samples. Similarly, when the two HS patients were compared to each other, we observed 71 DEGs, including B2M and HLA genes, upregulated and lysosomal-associated membrane protein 3 (LAMP3) downregulated in patient HS2 (Fig. 2d).

A cluster of overexpressed genes was found in M1 macrophages from the two HS patients when compared to HC samples (Fig. 2e). In M1 macrophages, several inflammatory genes such as IFN-responsive genes, TNF, IL-1α and IL-1β were upregulated in both HS patients (Fig. 2f). In M2 macrophages, we still observed a different transcriptomic profile in the two HS patients (Fig. 2g). A total of 653 and 422 DEGs were found in M2 macrophages from HS1 and HS2 patients when compared to HC samples, several of the genes shown in Fig. 2h. The full list of DEGs can be found in Supplemental Table 4.

To classify genes with the most significant differences in the patients, the top 15 upregulated and 15 downregulated DEGs (ranked by fold change) in each cell type were identified (Fig. 3a–d). Additionally, to determine whether the gene expression profile of H syndrome patients resembles that of patients with SAID, we included the data from two SAID patients, which were also characterized by transcriptomic profiling alongside the HS patients (Fig. 1b–d; Supplemental Fig. 1A-D).

The SERPINA1 gene, which encodes alpha1-antitrypsin (AAT), was found to be the most or second most upregulated gene in all cell types from both HS and SAID patients compared to HC (Fig. 3a–d). M0 macrophages from HS and SAID patients exhibited a comparable gene expression profile with similar levels of SERPINA1, LOC105376781, F13A1, CD1A, ASGR2, CCL17, MMP12, and CLDN1 when compared to controls (Fig. 3b). Moreover, in M1 macrophages, 12 genes were found to be similarly expressed among the two HS and SAID patients, with SERPINA1 being the top upregulated gene (Fig. 3c).

To further elucidate the unique DEGs, in monocytes and macrophages, which may be responsible for the autoinflammatory phenotype presented by HS2 patient, we compared the transcriptomic profile HS2 against HS1 and ranked the DEGs by fold change (Fig. 3e–h). Corroborating our previous findings, SERPINA1 was not found in the list of DEGs, when comparing the two HS patients, suggesting that this gene is a shared marker in SAIDs. Remarkably, B2M was the top upregulated gene in all cell types from HS2 patient (Fig. 3e–h). M1 macrophages from HS2 patient showed increased levels of NINL and SNHG5, which have been associated with Hodgkin’s lymphoma lymphocytic-histiocytic predominance (Fig. 3g). In the M0 macrophages from the HS2 patient, we observed downregulation of LAMP3, which is related to lysosomal function.

We next performed gene ontology (GO) analysis of the DEGs in HS and SAIDs, when compared to HC. Macrophages from HS patients revealed a significant enrichment of GO terms related to immune responses, IFN signaling, and MHC class I (Fig. 4) (Supplemental Fig. 2A). For instance, the biological process (BP) GO term “Interferon-gamma-mediated signaling pathway” (GO:0060333) displayed a fold enrichment of 27.4 in monocytes, 14.1 in M0, 15.2 in M1, and 5.6 in M2, in both HS patients with FDR < 0.001 (Supplemental Table 5). Several of the top 15 GO terms in HS patients were also shown in uSAID and TRAPS patients (Fig. 4). For example, the GO term “Immune response” (GO:0006955) was in the top 4 terms from all cell types in HS and SAID patients (Fig. 4) (Supplemental Fig. 2A). Although the GO term “Inflammatory response” (GO:0006954) was significantly enriched in all the patients, this term was more enriched in M2 macrophages from all the patients, fold enrichment of 3.0 (HS) and 2.5 (uSAID and TRAPS) with a FDR < 0.0001 (Fig. 4d–e) (Supplemental Table 5).

Furthermore, to compare gene expression profiles between HS patients and other defined SAID patients, we obtained and analyzed published transcriptome data from individuals diagnosed with either neonatal onset multisystem inflammatory disease (NOMID), NLRC4 macrophage activation syndrome (NLRC4-MAS) or A20 haploinsufficiency (due to mutation of TNFAIP3) [20, 21] (GSE57253 and GSE95078). DEGs were identified for each disorder, compared to the matched controls present in the same dataset and then compared to the DEGs identified in HS patients (Fig. 5). Consistent with our previous findings, the upregulation of SERPINA1 was found to be shared in all SAID and HS patients’ samples (Fig. 5a–d). Moreover, the inflammatory markers CXCL8, TNFAIP6, and IL-1β were also shared between the conditions (Fig. 5a–d).

GO analysis of shared DEGs between HS and SAID patients revealed significant enrichment of GO terms, such as “type I interferon signaling pathway” (GO:0060337), “MHC class I protein complex” (GO:0042612) and “Immune response” (GO:0006955) (Fig. 5e–h) (Supplementary Table 3). Altogether, these data suggest a similar transcriptomic profile in HS and SAID patients.

To determine if similar biological processes are involved in the pathogenesis of RDD related to H syndrome and in idiopathic cases, we compared transcriptomic profiles of tissue lymph node biopsies obtained from five sporadic cases of RDD and the HS2 patient. Pairwise comparison of biopsies from the five sporadic RDD cases against the HS2 failed to detect any DEGs suggesting that irrespective of the cause, there is a common pathological mechanism underpinning the development of RDD. To identify signaling pathways that might be associated with the pathogenesis of RDD, we filtered all the genes with fold change less than two and greater than − 2 and took the top 3000 genes with the highest expression. Panther pathway analysis revealed 27 significantly enriched pathways including “Inflammation mediated by chemokine and cytokine signaling pathway,” “JAK/STAT signaling pathway,” and “p38 MAPK pathway” (please see supplementary Tables 7 and 8 for full details).

Cellular plasticity confers macrophages the capacity to elicit different immune responses, pro-inflammatory (M1), or anti-inflammatory (M2), according to the environmental circumstances [22]. Abnormal proportions of M1 and M2 macrophages have also been reported in several immune disorders and are associated with disease progression [23]. We evaluated macrophage polarization in HS, uSAID, and TRAPS patients, where macrophages were classified as M1 (CD64⁺, CD80⁺, and CD86⁺) or M2 (CD64⁺, CD206⁺, and CD209⁺) according to their surface marker expression (Fig. 6). Prior to initiating differentiation, the TRAPS patient showed an increased number of M1 macrophages (27.3%); in contrast, the proportion of M2 macrophages was significantly increased in the uSAID, TRAPS, and HS1 patients compared to HC (Fig. 6a). Although proportionally no differences were observed in CD64 expression, the CD64 mean fluorescent intensity (MFI) was raised in all patients (Fig. 6a–b). When macrophages were activated, as described in Fig. 1a, HS patients showed a decrease in the proportion of M1 macrophages, which was more pronounced in HS2, and the proportion of M2 was significantly increased in the uSAID, TRAPS, and HS1 patients (Fig. 6c). CD80 MFI was slightly decreased in the two HS patients, while CD209 MFI was increased considerably in the two SAID and HS1 patients (Fig. 6d). The M1 and M2 gating strategy can be found in Supplemental Fig. 3. Cytokine profiles of patients presented some anomalies, showing high TNF in M0 macrophages from the TRAPS patient and significantly increased levels of TNF in M1 macrophages from all patients (Fig. 6e–f). Although M2 macrophages from the uSAID, TRAPS, and HS1 patients presented odd polarization ratios, no significant differences were observed in their cytokine levels (Fig. 6g). Moreover, increased serum levels of IFNγ, TNF, IL-6, and IL-2 were shown by HS2 patient, consistent with the patient’s inflammatory clinical features (Fig. 6h). Interestingly, both HS patients showed increased levels of IFNγ and IL-2 (Fig. 6h).