In this study, we identified BDNF as a target gene of HD and the RNA data revealed that BDNF expression was down-regulated. The GSEA analyses engaging 19,414 background genes showed that DEGs in HD / control and BDNF-low / high groups were significantly enriched in regulation of angiogenesis, protein maturation and protein processing. Pathologically, the accumulation of N-terminal mHtt was observed to be responsible for protein processing impairment, protein misfolding and neuronal degeneration [13, 14]. It was worth noting that these biological processes were potentially related to HD as well as the low BDNF expression. Thenceforth, the global regulatory network and co-expression modules of DEGs interacting with BDNF were constructed to investigate the genome-level pathogenesis of BDNF in HD.

The results emerging from co-expression network analysis demonstrated that the blue and turquoise modules had the greatest correlation with HD and BDNF expression, in which the DEGs were involved in GABAergic synapse, phagosome, Rap1, cAMP, MAPK, Ras and retrograde endocannabinoid signaling pathways. Additional experiments from animal HD models linked the MAPK signaling pathway to the neurotoxicity of mHtt [15–17]. More specifically, the role for MAPK signaling in HD was supported by mHtt-induced inhibition of fast axonal transport through the activation of MAPKs, which provided a molecular basis for HD neuropathology [18]. An increase in BDNF was sufficient to preserve synaptic vesicle proteins and facilitate behavioral recovery in post-stroke mice, partially via MAPK signaling [19]. Several indirect evidences confirmed the linkage between BDNF and MAPK signaling, namely, the targeted deletion of MAPK kinases inhibited the neuroprotective action of BDNF, leading to neuronal apoptosis and brain developmental defects [20–22]. On the other hand, pretreatment with BDNF prevented MAPK phosphorylation activated by amyloid-beta peptide in the entorhinal cortex of Alzheimer’s disease, suggesting a negative correlation of BDNF with MAPK signaling [23]. Similarly, our results supported the likelihood that low expression of BDNF was involved in the hyperactivation of MAPK signaling and that enhancing BDNF expression could be neuroprotective in HD.

With except of the MAPK signaling, enrichment analysis of intersection pathways revealed that BDNF jointly participated in Ras and cAMP signaling pathways. The involvement of Ras has been increasingly reported in the pathophysiology of neurodegenerative diseases, such as Alzheimer’s disease [24] and HD [25]. Angiotensin converting enzyme (ACE) is a dipeptidase that belongs to the Ras and cleaves angiotensin I to generate angiotensin II (Ang II) [26]. In mouse models of Alzheimer’s disease [27] and HD [25], ACE inhibitors and Ang II antagonists were administrated to improve cognitive impairment by reducing mitochondrial oxidants.

Recently, the protective effects of candesartan (an inhibitor of Ang II) on BDNF loss and neuronal apoptosis has also been demonstrated in cognitively impaired rats, pointing to the reduction in BDNF on Ras signaling as a mechanism of neurodegeneration [28]. Moreover, BDNF participation in Ras / MAPK signaling pathways was extended by binding to and activating the tyrosine receptor kinase B, giving rise to the differentiation and survival of knock-in striatal cells in HD [29]. For cAMP signaling, it is essential for mHtt-induced energy metabolism deficits, especially in early stage of HD [30]. There was convincing evidence that inhibition of cAMP expression and mitochondrial respiratory chain dysfunction emerged in HD brain as early as 12 hours following mHtt transgene induction [31]; intriguingly, the administration of forskolin to raise cAMP levels attenuated the neurotoxicity of mHtt [32]. In addition, experiment from primary cultures of hippocampal neurons showed the deteriorative damage of cAMP signaling with the down-regulation of BDNF [33], which was in line with our findings on the involvement of cAMP signaling in low BDNF-mediated HD pathogenesis.

Further scatterplot of the relationship between MM and GS confirmed that DEGs in the blue and turquoise modules were strongly interacting with the BDNF expression. Based on these DEGs, the global regulatory network was constructed to predict the intersection pathways of BDNF, which supported the potential roles of BDNF reduction in HD pathophysiology via MAPK, Ras and cAMP signaling pathways. Owing to the low expression of BDNF, the susceptibility of these presented pathways to defects might be obvious, resulting in the development of HD under a variety of pathogenic mechanisms [29, 34]. The AUC analysis exhibited a good diagnostic performance of low BDNF in differentiation of HD cases from non-dementia controls, implying BDNF to be a potential biomarker of HD. It was consistent with previous animal experiment that low expression of BDNF was found in pre-symptomatic HD, and this pathological decline could be up-regulated through beneficial interventions, such as wheel operation and environmental enrichment [35]. Moreover, the analyses of Pearson correlation showed significant correlation of BDNF with signature genes, which suggested that alterations in BDNF expression led to changes in signature genes of each intersection pathway, thus providing computational statistical evidence that low BDNF expression related HD pathogenesis was mediated via cAMP, MAPK and Ras signaling pathways. Further investigation in vivo or in vitro is expected to verify the relevant pathways proposed in this study underlying pathological process of HD.