Fig 1.
Functional analysis of differentially regulated genes.
(A) For differential gene expression between RR, BR and AT whole blood samples we calculated fold change analysis and performed a paired statistical test using DEseq2. A cut of fold change ≥ 1.2 and p ≤ 0.05 was applied for RR vs. BR and fold change ≥ 1.2 was applied for RR vs. AT. In examining the RR temporal stages: BR (before reaction); RR (reversal reaction), AT (after treatment), we identified 1017 genes upregulated in RR vs BR and AT and 193 genes downregulated in RR vs. BR and AT. Top 5 GO terms and canonical pathways identified by ClueGO and IPA respectively of significantly upregulated genes (B) and downregulated genes (C) for RR whole blood samples. Graphs show the number of associated genes and -log p-value for each GO terms. Padj was calculated with B-H multiple testing for the association of the functional term with the gene-expression data. (D) Upregulated RR genes in the IPA interferon signaling canonical pathway are presented in red.
Fig 2.
IFN-β and IFN-γ signature on RR samples.
(A) IFN-β and IFN-γ gene expression signature by SaVant. For each time point of the IFN-β and IFN-γ MDM gene expression data, signature enrichment scores were calculated using average gene expression for each RR temporal stages and normalized to Z scores. Columns correspond to RR temporal stages and rows correspond to IFN signature; each individual square corresponds to the enrichment for one IFN signature in a specific RR sample for each subtype. P-value was calculated by one-tailed ANOVA, followed by Tukey's multiple comparisons test. (B) Heatmap showing the normalized counts for the 38 and 13 overlapped genes between RR upregulated genes and IFN-β and IFN-γ MDM specific genes, respectively. (C) Enrichment analysis of overlap between IFN-β and IFN-γ- specific upregulated genes identified in human MDMs and RR and BR/AT specific leprosy whole blood transcripts (fold change ≥ 1.2 and P ≤ 0.05). Dotted lines indicate either the expected fold enrichment of one (left) or the hypergeometric enrichment p-value of 0.05 (log P = 1.3, right). Hypergeometric analyses were performed to determine fold enrichment (observed/expected) and signed log enrichment p- value (negative for de-enriched). The Bonferroni multiple hypothesis test correction was applied for each group. RR temporal stages: BR = before reaction, RR = during reversal reaction and AT = after treatment. N = 10 for each RR subtype.
Fig 3.
IFN-β and IFN-γ specific downstream gene signature for all leprosy subtypes.
(A-C) Enrichment analysis of overlap between IFN-β and IFNγ–specific upregulated genes identified in human MDMs and MB, BR, RR and PB leprosy whole blood transcripts (PM = top 500 genes). The top graph shows the number of overlapped genes between each leprosy clinical form and IFN-β and IFN-γ specific genes (A). Dotted lines indicate either the expected fold enrichment of one (middle—B) or the hypergeometric enrichment p-value of 0.05 (log P = 1.3, bottom—C). Hypergeometric analyses were performed to determine fold enrichment (observed/expected) and signed log enrichment p-value (negative for de-enriched). The Bonferroni multiple hypothesis test correction was applied for each group. (D) IFN-β and IFN-γ specific gene voting summation scores were calculated for an individual patient blood sample in leprosy states MB (n = 8), BR (n = 10), RR (n = 10) and PB (n = 8). MB = multibacillary, BR = before reaction, RR = reversal reaction and PB = paucibacillary.
Fig 4.
Gene expression in whole blood from RR temporal stages.
(A) Identification of RR whole blood subtype gene modules. WGCNA eigengene modules correlated to at least one RR temporal subtype (p ≤ 0.05). Red indicates a positive correlation, and green indicates an inverse correlation. Module eigengenes, as well as the corresponding number of genes in each module, are labeled on the y axis, and RR temporal stages are labeled on the x axis. BR (before reaction), RR (reaction) and AT (after reaction treatment). (B) Integration of WGCNA gene modules with cell-type-specific gene signatures. For the two significant modules derived from WGCNA, enrichment for MDM IFN-β and IFN-γ specific downstream genes (2h, 6h and 24h) were calculated and displayed in a heatmap of Z scores. Hypergeometric analyses were performed to determine enrichment p-value. * p<0.001.
Fig 5.
Functional analysis (434 common genes between bisque4 and upregulated genes).
(A) Top 5 functional GO terms for the 434 common genes between the RR positively correlated module and RR upregulated genes. Graphs show the number of associated genes, -log p-value and 5 hits for each GO term. Padj was calculated with B-H multiple testing for the association of the functional term with the gene-expression data. (B) Top 5 upstream transcriptional factors for the 434 common genes. The upstream analysis was performed by IPA upstream regulator analysis. Graphs show the number of target genes, -log p-value and z-score for each transcriptional factor. Orange and positive z-score for activation and blue and negative z-score for inhibition. (C) Integrated network of gene expression, upstream regulators and functional analysis terms. Gephi was used to create a functional annotation network, showing connections among significant GO terms, IPA upstream regulator analysis and 434 common genes between bisque4 module and upregulated significantly expressed genes. Genes are colored by term a, transcription factors in orange and connections are represented by dotted lines.
Fig 6.
GBPs are upregulated in the peripheral blood of leprosy patients during RR.
(A) Heatmap of GBPs 1 to 6 normalized counts from all 10 RR temporal stages (BR; RR and AT). Significance was determined by paired one-tailed ANOVA and post-hoc (Tukey multiple comparison test). (B) GBPs 1 to 6 correlation of fold change (FC) and p-value (-log p-value) between RNAseq and qPCR data. Squares represent the mean FC (RNAseq and qPCR) of the 10 RR pairs (RR vs. BR). Pearson correlation was used to calculate r values and the two-tailed p-value was calculated for correlation significance. (C) Distribution of GBPs 1 to 6 normalized counts in the whole blood for all groups of leprosy patients. The graph shows the mean per clinical type. MB = 8, BR = 10, RR = 10, AT = 10 and PB = 8 (D) Distribution of GBP 1 to 6 mRNA by microarray analysis of skin lesions shown in arbitrary units (AU). The graph shows the mean per clinical type. L-lep/MB = 6; RR = 7 and T-lep/MB = 10. Significance was determined by one-tailed ANOVA using GraphPad Prism software and post-hoc (Tukey multiple comparison test). (E) Distribution of GBPs 1 to 6 mRNA in the RNAseq analysis of human MDMs stimulated for 24h, media, yellow; IFN-β, red; IFN-γ, blue and TLR2/1 ligand, green. The graph normalized counts for all 5 individuals per stimuli for GBPs 1 to 6. Significance was determined by paired one-tailed ANOVA using GraphPad Prism software and post-hoc (Tukey multiple comparison test). * P <0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. MB = multibacillary, BR = before reaction, RR = reversal reaction, AT = after treatment, PB = paucibacillary, L-lep = lepromatous leprosy and T-lep = tuberculoid leprosy.
Fig 7.
IFN-γ is the major upstream regulator for GBP expression in the RR peripheral blood.
IPA pathway and path designer were used to create a functional IFNG downstream causal network. Upstream regulators with predicted activation are in orange and RR downstream upregulated GBPs are in red. Bold lines mean direct activation and dotted lines mean indirect activation.