Gaucher disease: transcriptome analyses using microarray or mRNA sequencing in a Gba1 mutant mouse model treated with velaglucerase alfa or imiglucerase

Abstract

Gaucher disease type 1, an inherited lysosomal storage disorder, is caused by mutations in GBA1 leading to defective glucocerebrosidase (GCase) function and consequent excess accumulation of glucosylceramide/glucosylsphingosine in visceral organs. Enzyme replacement therapy (ERT) with the biosimilars, imiglucerase (imig) or velaglucerase alfa (vela) improves/reverses the visceral disease. Comparative transcriptomic effects (microarray and mRNA-Seq) of no ERT and ERT (imig or vela) were done with liver, lung, and spleen from mice having Gba1 mutant alleles, termed D409V/null. Disease-related molecular effects, dynamic ranges, and sensitivities were compared between mRNA-Seq and microarrays and their respective analytic tools, i.e. Mixed Model ANOVA (microarray), and DESeq and edgeR (mRNA-Seq). While similar gene expression patterns were observed with both platforms, mRNA-Seq identified more differentially expressed genes (DEGs) (∼3-fold) than the microarrays. Among the three analytic tools, DESeq identified the maximum number of DEGs for all tissues and treatments. DESeq and edgeR comparisons revealed differences in DEGs identified. In 9V/null liver, spleen and lung, post-therapy transcriptomes approximated WT, were partially reverted, and had little change, respectively, and were concordant with the corresponding histological and biochemical findings. DEG overlaps were only 8-20% between mRNA-Seq and microarray, but the biological pathways were similar. Cell growth and proliferation, cell cycle, heme metabolism, and mitochondrial dysfunction were most altered with the Gaucher disease process. Imig and vela differentially affected specific disease pathways. Differential molecular responses were observed in direct transcriptome comparisons from imig- and vela-treated tissues. These results provide cross-validation for the mRNA-Seq and microarray platforms, and show differences between the molecular effects of two highly structurally similar ERT biopharmaceuticals.

Figure 1. Correlations of microarray and mRNA-Seq and their DE analytic methods.

(A) Correlation of signal intensity of saline treated 9V/null tissues in microarray platform with mRNA-Seq platforms. The panels show the (Log₂) mRNA-Seq read counts for each gene plotted on the X-axis compared with the (Log₂) intensities from the microarray data on the Y-axis. To avoid log of 0, 1 was added to each of the average counts prior to taking logs. The Pearson's coefficients (at the top of each panel) for each tissue show high correlation between the microarray and mRNA-Seq data. (B) Correlations of three DE analytic methods. edgeR and DESeq for mRNA-Seq and Mixed Model ANOVA for microarray were employed to pick a common subsets of genes from mRNA-Seq and microarray platforms. The genes that met the cut-off criteria (FDR  = 0.05, and a FC ≥ ±1.5) by all three DE methods were interrogated.

Figure 2. Comparisons of the DEGs between microarray and the mRNA-Seq.

DEGs were identified in 9V/null vs. WT spleen by Mixed Model ANOVA (microarray) and DESeq and edgeR (mRNA-Seq). The colors indicate the analytic methods. (A) saline-treated, (B) imig-treated, and (C) vela-treated 9V/null spleen. (D) The number of DEGs in 9V/null spleen identified by the different analytic methods in the saline-, imig-, and vela- treated groups. The genes with increased expression levels are shown in dark grey and the genes with decreased expression levels are in light grey with the corresponding number of genes indicated below.

Figure 3. Functional classifications of the DEGs in spleen.

(A) Functional relationship of spleen core DEGs associated with each treatment. An abstracted view shows the interaction of the biological functions by the core DEGs in 9V/null spleen compared with WT under different treatment conditions. The biological functions associated with the core DEGs from saline (pink node), vela (blue node) and imig (green node) treated 9V/null mouse spleens. Merged nodes indicate the shared functions between treatments. (B) 3-way Venn diagram presents the distribution of the biological functions by the core DEGs in spleen with different treatments. Each color represents a treatment as labeled. The GO were identified with DAVID. There were 16 functions common for 3 treatments. The unique functions for saline were 10, imig were 4, and vela were 56. The top biological functions are listed against each treatment.

Figure 4. Spleen core DEGs forming network of mitochondrial dysfunction, oxidative phosphorylation, and ubiquinone biosynthesis.

The network consists of 50 mitochondrial genes related to dysfunction, oxidative phosphorylation and ubiquinone biosynthesis. Genes colored with green or red indicate altered expression in saline-treated 9V/null spleen. Genes circled in black indicate the expression at WT level. (A) In the saline-treated 9V/null spleen, all genes in the network were abnormally expressed, shown in green or red. (B) The expression level of genes in imig-treated spleen. The ATPase and heme oxygenase circled in black were at WT levels. (C) In vela-treated spleen the expression of ATPase (circled in black) was at WT levels. Red indicates expression above normal and green indicates expression below normal levels.

Figure 5. Comparative analyses of the DEGs identified by direct comparison of imig- vs. vela-treatment.

For the analyses in this figure imig- and vela- treated samples were directly compared without normalization to WT gene expression in the corresponding tissue. (A) The Y-axis represents the number of DEGs in imig-treatment compared to the number with vela-treatment. The X-axis represents the different tissues. Three times more genes were detected by mRNA-Seq than that by microarray analysis. The number of genes are color coded for increased expression (dark grey) and decreased expression (light grey). Liver showed smaller DEG differences. In lung, The number of DEGs in imig- and vela-treated samples were not different. (B) Common and unique DEGs identified by microarray and mRNA-Seq in spleens. The Venn diagram of DEGs in the spleen compares the number of identified DEGs from microarray and mRNA-Seq that were different in imig- vs. vela-treatment. Compared to vela-treated spleen, 50 and 243 unique genes were identified in microarray (left) or mRNA-Seq (right) data sets in the imig-treated spleens. Forty seven genes (intersection) were common to both platforms. The GO annotation was performed using DAVID and the number of increased and decreased DEGs in the top functions identified by IPA are listed in Table S13.

Figure 6. DEGs in splenic networks.

The networks were generated using IPA software and were from the direct comparisons of imig- vs. vela-treated data sets without normalization to WT. The pathway included DEGs with decreased expression (imig/vela, green symbols) and the DEGs with expression level-increased (imig/vela, red symbols). The gene symbols and their interactions are as indicated. (A) The cell division/proliferation network is composed from total 42 DEGs determined by microarray (12 genes, red star) and mRNA-Seq (37 genes, blue star) which includs 7 common genes (blue and red stars) (see gene list in Table S14a). A general decrease in DEG expression levels was found in cell division/proliferation network from imig-treated vs. vela-treated spleen. (B) Hematopoietic system network was composed of total 54 DEGs determined by microarray (16 genes, red star) and mRNA-Seq (49 genes, blue star). Among them, 11 were common genes (red and blue stars) (Table S14b). (C) Inflammatory response/macrophage network was composed of total 41 DEGs determined by microarray (5 genes, red star) and mRNA-Seq (40 genes blue star), of those 4 were common genes (red and blue stars) (Table S14c).