Patient characteristics. Microdissected lung and lymph node granulomas from 18 sarcoidosis subjects, four TB subjects, three CM subjects and six healthy controls were assessed. There were no significant differences in age (p = 0.1) in the three granulomatous disease groups (Table 1). The majority of sarcoidosis subjects were female and non-Hispanic whites while the demographic most frequent reported in the CM and TB groups was Hispanic. Clinical data related to organ involvement and radiographic Scadding stage was available for 70% of sarcoidosis subjects as detailed in Supplemental Table 2.
After excluding probes with multiple hits or partial hits, our gene expression data in granulomatous tissues included 2,430 probes that uniquely mapped to single genes. We averaged gene expression levels after normalization and compared diseased groups versus healthy tissue controls from the same tissue origin, the fold change (FC) was obtained as the ratio between the averages. We identified 250 significantly dysregulated genes between all granulomatous-diseased tissues and healthy control tissues. Figure 1 represents the differentially expressed genes (DEGs) volcano plots, in each category; granulomas from CM exhibited the fewest number of differentiated transcripts, 34 in total (Fig. 1A), sarcoidosis granulomas from the lung (Fig. 1B) had more DEGs than those granulomas from lymph nodes (Fig. 1C), 88 vs 60 dysregulated transcripts respectively. Notably, in both sarcoidosis-affected tissues, the number of downregulated genes (FC<-2) outpaced the number of upregulated genes (FC > 2), with downregulation more marked in lung granulomas (93% of the transcripts) than in lymph node granulomas (56% of the transcripts). Tuberculosis granulomas (Fig. 1D), showed the highest number of the differentiated transcripts, 140 in total. The heat maps comparing the gene expression in each group against healthy tissues are presented in Fig. 2. In addition, we screened the total of 250 DEGs among those four sets performing a Venn diagram analysis (33). Figure 2E shows the overlapping and unique transcripts in the different granulomas. Sarcoidosis granulomas exhibited 138 dysregulated transcripts in both lung (Fig. 2A) and lymph node (Fig. 2B), 87 transcripts were exclusive present in sarcoidosis, notoriously, 30% of the DEGs in lung granulomas were unique to sarcoidosis, and the overlapping transcripts dysregulated in both sarcoidosis tissues was only 4%. TB had 89 DEG that were exclusively dysregulated in TB granulomas. When compared TB to sarcoidosis 17% of the transcripts were dysregulated in both diseases, while only 4% of the transcripts were common between CM and lung sarcoidosis.
Comparisons of expression profiles in sarcoidosis granulomas.
We identified that sarcoidosis granulomas from lung and lymph nodes share only ten DEGs. NR1H3 and CXCL9 were upregulated whereas the remainder (ADAMTS1, CXCL2, HSPB6, ITGA9, NPR1, NR4A1, CCL14, FABP4) showed downregulation in both lung and lymph nodes. Interrogation of the expression levels of these 10 sarcoidosis genes to TB and CM DEGs revealed CCL14, CXCL9, FABP4, NR1H3 were also significantly dysregulated in TB. Only six genes (ADAMTS1, HSPB6, NR4A1, CXCL2, NPR1, ITGA9) were exclusively dysregulated in both lung and lymph node in sarcoidosis granulomas but not in CM or TB (Fig. 3). These results are very consistent with the notion that granuloma gene expression is highly tissue- and disease-specific.
Comparison of DEGs in sarcoidosis vs CM and TB granulomas. Comparison of TB lymph node granulomas expression profiles to lymph node healthy controls identified 140 DEGs with a surprising number of DEGs common to sarcoidosis, 43 transcripts, 39 present in lymph granulomas. Unlike sarcoidosis where DEGs were predominantly downregulated, TB DEGs showed a balanced proportion of expression, 64 down-regulated and 76 upregulated. Notably, we observed the same direction in the transcript regulation, in all the 43-shared transcripts, except for one, OLR1, a low density lipoprotein receptor involved in Fas-induced apoptosis, this gene was upregulated in TB and down regulated in sarcoidosis, over-expression of OLR1 result in upregulation of NF-κB and its target pro-oncogenes (34).
Gene expression profiles from CM lung granulomas exhibited the fewest number of dysregulated genes, 34 DEGs with a balanced pattern of dysregulation, 18 DEGs downregulated, 16 DEGs upregulated. A total of 15 DEGs were exclusively dysregulated in CM granulomas. Eight upregulated transcripts were common to sarcoidosis lung granulomas (CSF3, SERTAD1, MMP9, CCL19, BCL3, AREG, PTGS, F3). Only three DEGs, SLAMF7, DC27, and CXCL13, were dysregulated in all three diseases (sarcoidosis, TB, CM). Expression of CXCL13, a chemokine and B-lymphocyte chemoattractant associated to calcium influx was downregulated in both sarcoidosis (lymph node) and TB granulomas but upregulated CM granulomas.
Functional biologic pathway enrichment analysis. Pathway analysis in sarcoidosis identified enriched transcripts associated with protein binding, regulation of biological processes, and cellular processes. Comparison of the pathway enrichment showed a marked difference between lung and lymph nodes in sarcoidosis, delineated by a clear immunological response, involving leucocyte migration and neutrophil chemotaxis in the lymph nodes while a structural regenerative response, characterized by cell migration and angiogenesis, was observed at the lung level (Fig. 4A). At the disease level, we identified pathways that were disease specific for each gene set as seen in Fig. 4B. We also identified common biological pathways that were present in both tissues in sarcoidosis. Furthermore, our analysis identified common dysregulated pathways in the three disease categories (p < 0.006). Among these pathways, cytokine-cytokine receptor interaction, chemokine signaling, and chemokine receptors bind chemokines were the top common dysregulated pathways present in the three diseases.
Independent validation of disease-specific granuloma transcriptome results. We further validated our results using independent cohorts of microarray expression in lymph node and lung tissues from sarcoidosis and tuberculosis (Supplementary Fig. 1). Validation analysis in these microarrays data sets confirmed the dysregulation of 90 genes (FC > 2 FDR < 0.012) of the DEG identified in our panel. Among those dysregulated transcripts, we validated 46 in TB and 44 in sarcoidosis, 14 of them differentially expressed in lung, 30 in lymph nodes and 4 genes dysregulated in both tissues ADAMST1, CXCL2, CXCL9, FABP4. However, CXCL9 was also upregulated in the TB granulomas. The direction of the expression defined by fold change was confirmed in all the dysregulated genes in the microarray data.