BD-II is a common variant of bipolar disorder (BD) characterized by major depressive and hypomanic episodes (Vieta, 2019). Although BD-II received increasing attention, the diagnosis of BD-II is clinically challenging. It is reported that BD-II has an initial misdiagnosis rate of approximately 40%, therefore, contributes to a 10-year lag of treatment (Manning et al., 1997). Such high misdiagnosis/delayed diagnosis rate may be due to 1) because 1) initial episodes and long-term clinical course are usually dominated by depressive episodes; 2) patients usually request medical attention only during depressive episodes whereas hypomanic episodes are considered as normal positive mood state (Angst, 2007; Ghaemi et al., 2000, 2003). Evidence indicated that Misdiagnosis of BD-II may result in more fluctuated course and mood episodes, less asymptomatic duration, and an increased suicide rate (Ghaemi et al., 2000; MacQueen and Young, 2001). Therefore, a prompt and accurate diagnosis of BD-II is essential to enable a further treatment plan and benefit prognosis (Suppes et al., 2008). Nowadays, the diagnosis of BD-II rests on the clinical evaluation. The discovery of reliable and clinically applicable biomarkers for BD-II would provide better diagnostic accuracy and lead to timely treatment (Fernandes et al., 2017).

In our previous study, we explored peripheral diagnostic biomarkers including small noncoding RNA (miRNAs) and novel protein biomarkers. We have reported significantly elevated serum levels of miR-7-5p, miR-142–3p, miR-221–5p, and miR-370–3p in BD-II patients than controls (Lee et al., 2020). In addition, we constructed a diagnostic panel implementing the support vector machine (SVM) by combining the above significant miRNAs. The model showed a good diagnostic accuracy (Lee et al., 2020). We have also reported several candidate protein biomarkers for BD-II, including PRDX2, CA-1, FARSB, MMP9, and PCSK9 (Lee et al., 2021c). The combination of these five candidate proteins could differentiate BD-II from controls with good validity, even in independent validation groups. However, the mechanisms underlying these candidate miRNAs and proteins with the pathogenesis of BD-II remain unclear. Because neurodegeneration may be one of the pathogeneses for BD (Lim et al., 2013; Soeiro-de-Souza et al., 2012); peripheral BDNF has been regarded as both trait and state biomarker for BD (Fernandes et al., 2011). We have previously reported positive correlations between plasma BDNF levels with the levels of miR-7-5p, miR-221–5p, and miR-370–3p (Lee et al., 2021b). We have also found the level of one of the proteins, FARSB, positively correlated with the BDNF level (Lee et al., 2021a). Inferring from these two correlation studies, we suspect that the candidate miRNAs and proteins share the same pathomechanism for BD-II.

MiRNAs may repress proteins synthesis of their targeting mRNAs. Bioinformatics such as TargetScan has been employed to predict the targets of miRNA, based on the confirmed rules of interaction between miRNAs and their targets (Witkos et al., 2011). Although we have identified both candidate miRNAs and proteins with no prior predilection, how their interaction remains unknown. We plan to identify the protein targets of the candidate miRNAs using TargetScan. We will compare our candidate proteins with the identified target proteins. We will further explore whether the target proteins identified from TargetScan interact with our candidate proteins, using BioGrid. Finally, it will be of interest to analyze the interaction of peripheral levels of candidate miRNAs and proteins in our previous samples, as a validation of above prediction.

In the current study, we plan to explore the correlations between our candidate miRNAs and these candidate proteins. We will further validate these associations by using TargetScan and BioGrid to identify the target proteins of our candidate miRNAs. We proposed that the miRNA and proteins may interact and contribute to their association with BD-II.