New biological targets for treatment are urgently needed for patients with depressive disorders (both unipolar and bipolar), and the population where this need is most critical are those with medication-refractory depression. Identifying more of the diverse neuroimmune pathways that may be dysregulated in these disorders provides a first step in this process. As a screen for possible dysregulation, we looked for increased or decreased gene expression using qPCR from peripheral leukocytes in patients with medication-refractory depression vs. healthy controls with no history of depression. Specifically, we examined genes representing several biological pathways including glucocorticoid receptors, immune function, growth and transcription factors, and purinergic, transient vanilloid and acid sensing ion channel receptors. We restricted our sample to females to eliminate gender as a potential source of heterogeneity; however, we included depressed patients with both MDD and BPD diagnoses and secondarily examined differences in gene expression between these depression subgroups while controlling for medication use and depression severity. Finally, we examined relationships of increased or decreased expression to greater symptom severity, after controlling for several potential confounding factors, including medications.
Differentially expressed genes and their possible functional significance
Gene expression dysregulation may be indicative of imbalance or dysfunction in these biological pathways and therefore, implicate potential drug targets. Clinical studies previously have examined the effects of antidepressant and antipsychotic medications on biomarkers of inflammation [
36‐
38], glucocorticoid receptors [
39], glutamate [
40], and neuropeptides [
19], among others. Immune and HPA axis activity have thus been the most common targets of antidepressant intervention [
41‐
43] other than monoamine systems involving serotonin and norepinephrine which are primary targets of SSRI/SNRI/NRI and tricylic medications.
Consistent with previous studies on mRNA in blood of patients with depression, both medication responders and non-responders, we found upregulation in cytokines IL-6 and IL-10 in females with medication-refractory depression [
9,
12,
44]. Protein elevations of both cytokines in serum and CSF are a consistent finding in depression [
41,
45,
46], with some studies suggesting that IL-6 is specifically associated with medication-refractory depression, and that IL-10 levels decrease following successful treatments [
33‐
35]. Because we found that IL-10 levels were positively correlated with depression severity, it will be important for future studies to examine gene expression levels following symptom remission.
Our finding of increased peripheral OXTR mRNA in DD patients may reflect dysregulation in oxytocinergic signaling. Also, the trend shown in the present study for the association between the use of antidepressant medications and lower OXTR expression suggests that such decreases may contribute to beneficial effects of these medications. Oxytocin is a neuropeptide that is synthesized and released by the hypothalamus, and has been implicated in many biological functions, including social bonding, anxiety, pleasure-seeking, appetite, and stress response, all of which can be disrupted in depression [
47]. Most studies suggest that low plasma and CSF levels of oxytocin are associated with depressive behaviors and increased sensitivity to stressors, and that such differences may be specific to depressed females rather than to males [
18,
47,
48]. Furthermore, subjects with depression display greater fluctuations in oxytocin in response to stressful mental tasks, suggesting dysregulation [
16,
49]. In addition, polymorphisms in the OXTR can result in decreased receptor expression, symptoms of depression and anxiety, and sensitivity to stressors [
13‐
15]. Since intranasal oxytocin may have beneficial effects on depressive symptoms and anxiety, patients displaying oxytocin pathway mRNA dysregulation could benefit from such a targeted treatment [
17].
Of the growth factors that we examined, we observed that female patients with depression displayed elevated levels of amyloid precursor protein (APP) mRNA compared to healthy non-depressed controls. Subsequent analysis found that patients with BPD had higher levels of APP compared to MDD suggesting increases vs. CON are primarily due to the BPD subgroup. APP (most notable for its association with Alzheimer's Disease) plays multiple roles, for instance that of a neuronal growth factor and in transcriptional regulation, with complicated post-translational processing resulting in multiple cleaved peptides. Therefore, the increased APP gene expression that we observed may be due to compensation for low peripheral levels of peptides, or to non-functional peptides. For example, alterations in CSF APP peptides have been observed in patients with BPD, including decreased levels of sAPP-α and sAPP-β in Bipolar type I and changes in ratios for other Aβ peptide products in Bipolar type II [
50]. Decreased levels of Aβ40 and Aβ42 have also been observed in the CSF and serum of depressed patients with levels correlating to depression severity [
51‐
53], though both increased and decreased levels of Aβ42 have been found in human plasma [
54,
55]. Given the complex post-transcriptional processing of APP, future studies should combine mRNA and protein analysis to determine the relationship between APP transcript levels and peptide products, and if levels differ between medication-refractory and medication-responsive patients.
Our gene expression panel also consisted of several ligand-gated channel families important in detecting ATP and other metabolites that may be related to pain and fatigue, frequent symptoms in DD. Of the 9 ion channel receptors that we examined, P2RX7, TRPV1, and P2RY1 were upregulated in female patients during a depressive episode, and P2RY1 expression was positively related to depression severity along with 2 other ion channel receptors, TRPV4 and P2RX1. Our clinical findings in regard to P2RY1, P2RX1, TRPV1, and TRPV4 mRNAs are the first such observations in depressed humans; however they are consistent with a number of preclinical and genetic studies. Modulation of TRPV1 and P2RY1 receptors in animals has effects on depression and anxiety tasks [
23,
24,
56]. Furthermore, our research group previously found that following a moderate exercise task, when their fatigue and pain symptoms worsen, patients with CFS show increased expression of ion channels, with expression levels positively correlated with pain and fatigue symptoms [
21,
57]. While the role of peripheral P2RY1 receptor function in human depression is unknown, it is promising that we observed gene expression increases in DD that were also positively correlated to depression severity, suggesting either that increased expression may be involved in symptom presentation or that it is compensatory for receptor dysfunction.
Some prior studies have suggested that certain polymorphisms in P2RX7 may predispose individuals to MDD, although these findings have been disputed by others [
58‐
60]. Our results of increased P2RX7 mRNA expression in patients with medication refractory DD are contrary to those reported by Zhang et al., who found decreased peripheral expression of P2RX7 in psychiatric patients (primarily female) with post-traumatic stress disorder or MDD, and similar mRNA decreases in postmortem tissue from suicide victims in a meta analysis [
28]. Our samples are clearly different in several characteristics, but it is also possible that our increased peripheral receptor mRNA reflects secondary compensatory increases to diminished central activity in this pathway. Still, our observations as those of Zhang et al. are consistent in indicating dysregulation in this purinergic pathway in DD.
Although purinergic P2X and P2Y, acid-sensing ASIC, and transient vanilloid TRPV ion channels have been the target for the treatment of multiple other disorders, including thrombosis, cardiac arrhythmias, and neuropathic pain [
61,
62], these pathways have not yet been examined for antidepressant effects in patients with MDD or BPD. Given the numerous animal studies implicating ion channels including P2X, P2Y, and TRPV receptor function in anxiety- and depression-like states, and our studies showing dysregulation in these ion channel receptors in refractory depression, further investigation of drugs or other interventions targeting ion channels for depressive symptoms may be warranted.
Though still in their infancy, gene expression studies may one day help to identify individuals that would most benefit from drugs targeting non-monoamine pathways.
Effects of medications, clinical diagnosis, and depression severity on gene expression
In order to explore possible confounders in our DD sample, we included medication use, depression severity, and DD diagnosis in a five-factor regression model. Medications have been shown to affect immune function and alter gene expression [
10,
37,
63‐
66]. We found significant effects of medication for NR3C2, ASIC1, and APP with ASIC3, OXTR, and SIRT1 showing marginal effects. Importantly, with the exception of APP and OXTR, none of these genes were identified in either our DD vs. CON or BPD vs. MDD analyses. Decreased expression of OXTR with medication use suggests that medications may have minimized rather than contributed to the higher expression in DD vs. CON. Conversely, anticonvulsant use was associated with higher expression of APP and therefore medication use could have contributed to elevated levels in DD vs. CON. Future work should continue to consider the effects of medication use on gene expression and how it may accentuate or mask group differences.
In our sample we focused on patients that were currently in a depressive state. This included patients with BPD and MDD diagnosis. We identified several genes that showed group differences after controlling for concurrent psychotropic medication use, age, and depression severity. These included BPD elevations in transcription factors CREB1, NR3C1, and NFKB1, matrix-associated genes APP and SPARC, and downregulation of proinflammatory cytokine TNF. Of these genes, only APP was also found to be confounded by anticonvulsant use. However, robust differences persisted even after controlling for medication use. The same genes displayed differences between BPD and MDD when using HRSD. These are promising results that suggest possible biomarker differences between these diagnostic groups, but future replication studies are necessary.
Finally, we examined associations between depression severity and gene expression using both the clinical interview HRSD and QIDS self report scores. Both the QIDS and HRSD displayed positive associations for depression and gene expression for ion channels P2RX1 and P2RY1, further strengthening the possible relationship of ion channels and depression. In addition, QIDS had a positive association for another ion channel, TRPV4, and for the cytokine IL10, which was also identified as up-regulated in DD vs. CON in the primary analysis. HRSD score was not associated with these latter two genes (Table
5). Although QIDS and HRSD scores were strongly related to each other (R
2 = 0.586,
p = 0.003), as is always true, they were not identical. Future mRNA studies may benefit by including both assessment methods.
Limitations
There are several noteworthy limitations to this study. Because we examined female subjects only, it is not possible to conclude whether the observed gene expression differences would extend to depressed males or represent true gender-specific differences. Secondly, in this study we did not include control subjects taking antidepressants or antipsychotics. For example, since there were only 4 DD patients that were not taking antidepressants, it is possible that gene expression differences were influenced by antidepressant use. Medication-responsive control subjects likely represent a distinct group from healthy non-depressed controls and must be treated differently. In our regression analysis of depressed patients only, we did find that medication use led to differences in expression when controlling for age, depression severity, and diagnosis, including increased expression of ASIC receptors, transcription factor NR3C2, and growth factor APP. It is critical for future studies to examine gene expression in light of medication use. Thirdly, our primary analysis combined patients with BPD and MDD into a single group when comparing to non-depressed controls. This was done partially because all of our patients were currently in a depressed state. Further regression analysis controlling for depression severity, age, and medication, revealed several potential differences based on diagnosis. These results are promising in the search for potential diagnostic biomarkers, but our small sample size warrant future studies where BPD and MDD groups with adequate power are also examined separately compared to controls. Finally, it is important to note that depression remains enigmatic and difficult to treat, partially because of the diversity of disease onset, symptom presentation, comorbidities, and disease evolution. Current studies are under way in patients with depression to examine gene expression changes following moderate exercise tasks, found to differentiate patients with CFS and FM from controls. Similar to the study by Cattaneo et al. [
12], future research should compare gene expression over time for treatment-resistant and treatment-responsive patients in prospective studies before and after symptom remission. By amassing a panel of differentially expressed genes, researchers could continue to identify novel biological mechanisms contributing to depression and potentially pave the way to new drug targets and personalized medicine.