Transcriptomic profiles in major depressive disorder: the role of immunometabolic and cell-cycle-related pathways in depression with different levels of inflammation

Abstract

Transcriptomic profiles are important indicators for molecular mechanisms and pathways involved in major depressive disorder (MDD) and its different phenotypes, such as immunometabolic depression. We performed whole-transcriptome and pathway analyses on 139 individuals from the observational, case-control, BIOmarkers in DEPression (BIODEP) study, 105 with MDD and 34 controls. We divided MDD participants based on levels of inflammation, as measured by serum high-sensitivity C-reactive protein (CRP), in n = 39 'not inflamed' (CRP < 1 mg/L), n = 31 with 'elevated CRP' (1-3 mg/L), and n = 35 with 'low-grade inflammation' (>3 mg/L). We performed whole-blood RNA sequencing using Illumina NextSeq 550 and statistical analyses with the Deseq2 package for R statistics (RUV-corrected) and subsequent pathway analyses with Ingenuity Pathway Analysis. Immunometabolic pathways were activated in individuals with CRP > 1 mg/L, although surprisingly the CRP 1-3 group showed stronger immune activation than the CRP > 3 group. The main pathways identified in the comparison between CRP < 1 group and controls were cell-cycle-related, which may be protective against immunometabolic abnormalities in this 'non-inflamed' depressed group. We further divided MDD participants based on exposure and response to antidepressants (n = 47 non-responders, n = 37 responders, and n = 22 unmedicated), and identified specific immunomodulatory and neuroprotective pathways in responders (especially vs. non-responders), which could be relevant to treatment response. In further subgroup analyses, we found that the specific transcriptional profile of responders is independent of CRP levels, and that the inhibition of cell-cycle-related pathways in MDD with CRP < 1 mg/L is present only in those who are currently depressed, and not in the responders. The present study demonstrates immunometabolic and cell-cycle-related transcriptomic pathways associated with MDD and different (CRP-based and treatment-based) MDD phenotypes, while shedding light on potential molecular mechanisms that could prevent or facilitate an individual's trajectory toward immunometabolic depression and/or treatment-non-responsive depression. The recognition and integration of these mechanisms will facilitate a precision-medicine approach in MDD.

Fig. 1. Differentially expressed transcripts (FDR p-adjusted <0.1) in CRP-based group comparisons.

A UpSet plots to summarise key differentially expressed (DE) transcripts. These panels summarise the DE transcript overlap between comparisons for up-or down-regulated DE transcripts (in black), for down-regulated DE transcripts (in blue) and, upregulated DE transcripts (in red). In each panel, the bottom left horizontal bar graph labelled “Set size” shows the total number of DE transcripts per comparison. The circles in each panel’s matrix represent what would be the different Venn diagram sections (unique and overlapping DE transcripts). Connected circles indicate a certain intersection of DE transcripts between comparisons. The top bar graph in each panel summarises the number of DE genes for each unique or overlapping combination. In the top left panel, for example, the first vertical bar shows those DE transcripts that are unique to MDD with CRP > 3 mg/L vs. CRP < 1 mg/L (1508 DE transcripts). The second shows those DE transcripts that are shared only between MDD with CRP > 3 mg/L vs. CRP < 1 mg/L and MDD with CRP > 3 mg/L vs. controls (457 DE transcripts). B, C, D Volcano plots of RNA-seq expression analysis (C, D MDD with CRP <1 mg/L vs. controls with CRP <1 mg/L). Each dot represents a transcript comparing the conditions stated in the heading. The horizontal line corresponds to a Benjamini-Hochberg FDR-adjusted significance value of <0.1. The vertical lines in D correspond to a log2 FC value of |0.26| corresponding to an FC value of |1.2|.

Fig. 2. Canonical pathways differentially activated in CRP-based group comparisons.

Heatmap of the activation z-score (IPA) of statistically significantly enriched pathways for at least one of the comparisons. Hierarchical clustering was used to group pathways and comparisons. A MDD CRP 1–3 and >3 vs. MDD CRP < 1 and controls. B MDD CRP < 1 vs. all controls and controls with CRP < 1.

Fig. 3. Canonical pathways differentially activated in treatment-based group comparisons.

Heatmap of the activation z-score (IPA) of statistically significantly enriched pathways for at least one of the comparisons. Hierarchical clustering was used to group pathways and comparisons.

Fig. 4. Schematic representation of canonical pathways in group comparisons.

Schematic representation of the different pathways involved in group differences A in the CRP-based and B in the treatment-based MDD groups and all MDD cases compared with controls and one another. Gene transcripts have been selected based on p <0.05 and FC > | 1.2 | . The groups analysed in different comparisons are written above the respective box, activation or inhibition refers to the first group. Each box contains pathways for a single comparison, ordered based on p-values (lower on the top and higher on the bottom). Orange indicates a predicted activation of the pathway (positive z-scores), blue a predicted inhibition (negative z-scores); the intensity of colours reflects the z-score values. Canonical pathways have been selected based on p < 0.05 and z-scores ≥ |2|.