Transcriptomics and sequencing analysis of gene expression profiling for major depressive disorder.

BACKGROUND: Major depressive disorder (MDD) is a common psychiatric disorder characterized by constant sadness and a lack of interest in work and social interactions. Maintaining the transcriptome levels via the controlled regulation of mRNA processing and transport is essential to alleviating MDD. Various molecular phenotypes such as aberrant RNA splicing and stability are identified as critical determinants of MDD. AIM: This study aims to compare the mRNA expression profiles between major depressive disorder non-suicide (MDD), major depressive disorder suicide (MDD-S), and control groups using RNA-Seq.
MATERIALS AND METHODS: A transcriptomics and sequencing analysis of gene expression profiling was conducted in 9 patients with MDD, 10 patients with MDD-S, and 10 control patients.
RESULTS: A comparison of the sample groups revealed that the PRKACB gene was upregulated in patients with MDD. At the same time, GRM3, DLGAP1, and GRIA2 were downregulated in these patients-these genes are majorly involved in the glutamatergic pathway. Five genes (GRIA1, CAMK2D, PPP3CA, MAPK10, and PPP2R2A) of the dopaminergic pathway were downregulated in patients with the MDD-S condition when compared with the MDD and control groups. Cholinergic synapses were altered in patients with MDD when compared to the control group due to the presence of dysregulated genes (KCNQ5, PLCB4, ADCY9, CAMK2D, PIK3CA, and GNG2).
CONCLUSION: The results provide a new understanding of the etiology of depression in humans and identify probable depression-associated biomarkers. Copyright:

INTRODUCTION

Neuropsychiatric diseases, which include major depressive disorder (MDD), are a prominent group of diseases.[1] Roughly 17% of the world's population suffers from MDD.[2] Furthermore, a survey showed that MDD was expected to be the second most common illness in the world by the end of 2020.[3] MDD often leads to mental disabilities among people of any age, thus adversely impacting their lives.[4] Both major depression and clinical depression bring about physical and emotional problems.[5] A depressed mood, general loss of interest, reduced appetite, and irregular sleep patterns are common symptoms experienced by patients with MDD; some patients also tend to have suicidal thoughts.[6] Cognitive impairment and memory loss are among other less prominent symptoms of MDD.[78] Initially, the guidelines provided by the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) were used to diagnose mental disorders.[69] Later advancements in biological and genetic studies improved the pharmacological aspects of detecting and treating MDD.

The World Health Organization (WHO) has significantly contributed to designing procedures for determining the prevalence of MDD worldwide. For instance, the organization performed a survey in the United States from 2001-2002 to learn about the prevalence of MDD.[10] The prevalence of MDD is also associated with age at onset, sociodemographic correlates, comorbidities, role impairments after 12 months, clinical severity after 12 months, and treatment after 12 months. The resultant progress report can be used to diagnose, treat, and verify the recovery status of patients with MDD after treatment.[11]

Despite the above-mentioned advancements, many details about MDD disease were still unknown, and many critical aspects of the disease remained unaddressed. Patients with MDD identification and many other physiological changes started being considered. Some physiological changes associated with MDD include sleep disorders, loss of appetite, constipation, sexual desire, a disinterest towards work, crying, suicidal thoughts, and less enthusiasm in the patient's speech and actions.[12] However, in contrast, some of the patients with MDD exhibited hyperactivity and an increased interest in daily activities, similar to manias experienced by patients with bipolar disorder.[13]

Next-generation sequence (NGS) technologies provide information about the whole transcriptome by sequencing all RNA transcripts produced by a given gene.[14] Advancements in personalized genomics and predictive medicine have lowered the cost of NGS while increasing its throughput.[151617] Today, national health agencies and institutions are carrying out sequencing projects to determine the genotypes of hundreds of thousands of individuals worldwide.[181920] Thus, such projects enhance diagnostic approaches and clinical screening applications. Furthermore, the widespread application of such technologies advances the broader field of medical science.

In NGS, transcriptome (RNA-Seq) detects previously identified transcripts and produces novel transcripts, including several tissue-specific nonprotein-coding RNAs. Microarrays have some limitations, including high background noise and signal intensity saturation. RNA-sequencing is beneficial in that it overcomes all the limitations associated with microarrays. For example, RNA-Seq can map sequenced paired-end reads to the reference genome, thus recognizing the expressed region and ultimately generating a transcriptome map.[21] Moreover, cDNA libraries do not contain an intronic region—however, the aligner employed by RNA-sequencing performs all mapping tasks via the splice alignment feature. Another advantage of RNA-sequencing is that the obtained data provide information about the gene expression count while the quantitative polymerase chain reaction (qPCR) technique relatively quantifies the control sample.[21] qPCR can also be used to validate vital data related to gene expressions obtained through NGS.[22] Yet another noteworthy advantage of RNA-sequencing is that it can detect splice junctions, making it extremely helpful when measuring differences in splicing events (i.e., differential exon usage).[23]

Thus, the present study analyzed RNA-Seq data collected from patients with MDD and a control group. All data were processed to find any differentially expressed genes. A comparison between the two groups identified some key genes involved in certain neurological pathways responsible for maintaining brain health conditions.[24] According to previous research, alterations to these pathways and associated functions can change a person's behavior. The resultant neural imbalances cause the person to become unsocial and isolated, which can lead to suicide.[25] Specific genes and functional alterations are found in patients with MDD and MDD-S.[26] Fortunately, the appropriate and early diagnosis and treatment of MDD based on this knowledge can reduce the suicidal tendencies of patients.[27]

People of all ages, countries, and professions can experience depression. The WHO estimated that around 4.3% of the world's population (roughly 300 million people) suffered from depression in 2015. Moreover, 800,000 people commit suicide due to depression worldwide each year. In India alone, 258,000 suicides were registered in 2012 among people aged 15-29 years old. Moreover, according to national mental health survey reports, one in every 20 Indians suffers from depression.[28]

Thus, the present study considers the prevalence of depression and focuses on identifying the most common ways pathways by which patients are affected by differentially expressed genes. For this purpose, variations in the transcripts of patients with MDD, MDD-S patients, and control participants were investigated. The results of the comparison revealed probable biomarkers that can be considered when diagnosing MDD.

MATERIALS AND METHODS

Study sample

Twenty-nine samples out of which 9 MDD (patients with major depressive disorder non-suicide), 10 MDD-S (patients with major depressive disorder suicide), and 10 control patients were considered in this study. Raw data in FASTQ format from the Sequence Retrieve Archive (SRA) were retrieved from NCBI (National Center for Biotechnology Information), which is a public data source (Submission-ID: SRA587411). In the current study, the mRNA profiles of the three groups were developed and a differential gene expression study was performed using RNA-Seq data.

Data processing

The quality of all raw data was checked using FastQC software (Babraham Bioinformatics, Cambridge, United Kingdom). Cutadapt (version 1.18) was employed for data processing. Raw data (FASTQ files) were processed by removing sequencing adapters and trimming low-quality bases with a specific Phred quality score cutoff. All high-quality processed RNA-Seq data were then aligned to the reference genome (Human hg19), and all GFF files used in the study were downloaded from Ensembl. Alignment for the 29 samples against the reference genome was carried out using HISAT2 (version 2.1.0). We included the parameters of 'spliced alignment’ and ‘reported alignments’ tailored specifically for Cufflinks during alignment in HISAT2. We also specified the library as ‘paired-end’ and ‘forward, reverse’ as additional parameters while running HISAT2. The alignment resulted in averages of 92.79% in the MDD group, 92.46% in the MDD-S group, and 93.97% in the control group. Reference-assisted assembly was carried out using Cufflinks (version 2.2.1). Through this process, the number of transcripts present in each sample was determined. We also carried out a whole transcriptome gene-level analysis to obtain expression counts from the data using Cuffdiff v2.2.1. This analysis also allowed us to examine and compare the differential expression genes between the three groups. The main results reported throughout the text were derived from the full set of 29 samples.

Differential gene expression analysis

Cuffdiff was used to assess the differential expression between groups. This tool normalizes the read mapped count by the method ‘total hits normalization.’ It also calculates fragments per kilobase of transcript per million mapped reads (FPKM) to find the log2Fold change ratio, which expresses information about the over-expressed and under-expressed transcripts. Six comparisons were carried out in total as follows: Control versus MDD-S, Control versus MDD, MDD-S versus MDD, Control + MDD versus MDD-S, Control + MDD-S versus MDD, and MDD + MDD-S versus Control. All conditions were considered to obtain all possible interpretations concerning the biological phenomenon.

Statistical analysis:

FPKM was calculated by mapping the reads to the reference human genome to obtain the number of reads in the exonic region. Cuffdiff was utilized for differential expression and to determine the P value. The formula E[ log[Y] ]/Var[ log[Y] ] is the test statistic. This is calculated by taking a log (a fold change distribution by dividing the FPKM distributions of diseased by that of control condition and taking the log of the distribution). According to the formula, the mean divided by the calculated distribution variance gives the test statistic.

Gene ontology

Gene ontology (GO) was carried out using the Database for Annotation, Visualization and Integrated Discovery (DAVID); version 6.8 web portal. DAVID provides comprehensive functional annotation tools that researchers can use to gain various types of information about a gene or list of genes with their biological annotations. The dysregulated genes for different comparisons (Control versus MDD, Control versus MDD-S, MDD-S versus MDD, Control + MDD-S versus MDD, Control + MDD versus MDD-S, and MDD + MDD-S versus Control) were individually annotated using DAVID. GO includes biological process (BP), molecular function (MF), and cellular component (CC). Pathway analysis was carried using the Kyoto Encyclopedia of Genes and Genomes (KEGG), considering the dysregulated genes for different comparisons.

RESULTS

Demographics

A total of 9 MDD, 10 MDD-S, and 10 control samples were included in the analysis. Reference-based alignment resulted in transcript assembly from RNA-Seq data. The average numbers of transcripts observed were 15,527 (Control), 16,711 (MDD), and 14,532 (MDD-S).

Differential gene expression

The identification of different transcripts for each sample using Cuffdiff resulted in an abundance of transcripts in the individual sample. Because of this, the ‘total hits normalization’ method was used to identify significant transcripts from an individual sample. The log ratios of “Control versus MDD-S, Control versus MDD, MDD-S versus MDD, Control + MDD versus MDD-S, Control + MDD-S versus MDD, and MDD + MDD-S versus Control” indicate differential gene expression between the sample group. The log2 (fold change) values were positive for overexpressed genes and negative for under-expressed genes. Meanwhile, for neutrally regulated genes, both positive and negative values of log2 (fold change) were found.

Gene ontology and pathway analysis

DAVID provides functional enrichment results related to the differentially expressed genes for different comparison groups. The GO analysis and pathway enrichment of results were obtained based on the input gene list. The GO analysis of dysregulated genes in the Control versus MDD comparison showed that 49 BPs, 26 CCs, and 24 MFs were involved. The top 10 GO analysis indicates that the BP ‘chemical synaptic transmission’ is essential to neurons and might affect nine genes. Specifically, SLC12A6: solute carrier family 12 members 6; KCNQ5: potassium voltage-gated channel subfamily Q member 5; SYT1: synaptotagmin1; DLGAP1: disk large-associated protein 1; SLC1A3: excitatory amino acid transporter 1; GRIA2: glutamate receptor 2; FGF12: fibroblast growth factor 12; NRXN1: neurexin-1; LRFN5: leucine-rich repeat, and fibronectin type-III domain-containing protein 5 were dysregulated. Of these nine genes, the SYT1 gene was upregulated, while the other eight genes were downregulated.

Among dysregulated BPs, chemical synaptic transmission stands out as being highly significant (P < 0.001). Twelve genes with the MF ‘calcium ion binding’ were dysregulated. Genes involved in neurons and neurological function-related CCs—such as ‘neuron projection’ (eight genes) and ‘neuronal cell bodies’ (seven genes) were dysregulated in Control versus MDD. Fifteen related neuronal genes were dysregulated in total. The pathway analysis conducted using the DAVID bioinformatics tool (V6.8) revealed gene involvement in 25 different dysregulated pathways. Among the dysregulated pathways, the dopaminergic synapse pathway (hsa04728), cholinergic synapse pathway (hsa04725), and glutamatergic synapse pathway (hsa04724) are related to neurons and neurological functions.

DISCUSSION

Mental disorders are pervasive among humans, and much research has explored contributing factors to enhance our understanding of these disorders. This study investigated a previously underexplored biological mechanism using RNA-Seq transcriptomics analysis in light of previous findings. Previous reports have expressed that proline-rich protein 5-like (PRR5 L), neurocalcin delta (NCALD), and other genes are differentially expressed genes in patients with MDD when compared with control samples.[29] Also, PRR5 L is more downregulated in patients with MDD-S than patients with MDD. This gene is also associated with the PI3K/Akt and mTOR signaling pathways. Meanwhile, the NCALD gene, which was downregulated in both MDD and MDD-S, is present during transmissions across chemical synapses and is involved in presynaptic function.

Previous studies found that MDD and MDD-S patients have dysregulated glutamatergic-associated genes.[30] Glutamatergic pathway-related observations were found in the current study by comparing Control + MDD-S versus MDD. Furthermore, the PRKACB gene was upregulated in patients with MDD, whereas GRM3, DLGAP1, and GRIA2 were downregulated. In the glutamatergic synapse pathway and it shows how the presynaptic and postsynaptic junction signaling systems can be affected. Another comparison (MDD-S versus MDD) revealed that glutamatergic synapse pathway genes—such as DLGAP1, ADCY9, PRKACB, and GRIA4—were upregulated, while two genes—SLC1A3, GRM7—were downregulated. Glutamate is stored in the presynaptic vesicles and released upon neuronal depolarization and the opening of calcium channels. Dysregulation was seen in the synaptic cleft[31] due to an imbalance in the synaptic signaling system. This dysregulation is one of the major differences between patients with MDD and MDD-S.

The sphingolipid signaling pathway is also affected. This pathway is associated with phosphorylation, which is essential to cell cycles, growth, apoptosis, and signal transduction.[32] When MDD-S was compared to Control + MDD, the genes that we observed as being downregulated in the sphingolipid signaling (hsa04071) pathway were MAPK1, PRKCZ, PIK3CA, MAPK10, PRKCE, PTEN, and PPP2R2A. When we compared MDD-S with MDD, we found that the PRKCZ, PIK3CA, MAPK10, and SGMS1 genes were upregulated. Hence, PRKCZ, PIK3CA, and MAPK10 are potentially important genes among people with MDD, as they are commonly dysregulated and affect the sphingolipid signaling pathway.

Improving dopamine production and reuptake at synapses is critical in treating neurological brain disorders.[33] When we compared the dopaminergic synapse (hsa04728) of Control + MDD with that of MDD-S, were observed the downregulation of GRIA1, CAMK2D, PPP3CA, MAPK10, and PPP2R2A genes. Dopamine hormone levels are likely restricted by this downregulation. Furthermore, a break-in signal between the nervous communications might happen. Whereas high levels of dopamine can cause anxiety, low levels are associated with a depressive state.

The Control versus MDD comparison revealed that the cholinergic synapse (hsa04725) pathway is affected in patients with MDD. Specifically, the KCNQ5, PLCB4, ADCY9, CAMK2D, PIK3CA, and GNG2 genes were affected by cholinergic synapse dysregulation. Out of the six examined genes, KCNQ5 and GNG2 were downregulated in the cholinergic synapse according to the Control versus MDD comparison Cholinergic synapses are vital to maintaining the body's regular activity, as they control the body's cells and are present at all neurotransmitter junctions. Furthermore, acetylcholine, which is associated with learning, serves as a neurotransmitter in cholinergic synapses.[34]

The retrograde endocannabinoid signaling pathway and hippo signaling pathway are dysregulated according to the Control versus MDD-S comparison. We found that the CCND3, SMAD3, FBXW11, CTNNA3, CTNNA2, and DLG1 genes were dysregulated in the hippo signaling pathway. Meanwhile, in the retrograde endocannabinoid signaling pathway, the GABRA2, ADCY2, GNG2, GRIA4, and MAPK10 genes were dysregulated These findings are worth considering because cannabinoid receptors, which are essential to neural functioning and behavior, are active in retrograde endocannabinoid signaling.[35] Also, the hippo signaling pathway seems to be involved in neurodegeneration, regulating organ size, and suppressing tumors.[36] The current findings provide several avenues for future comparative analyses of brain transcripts among MDD and MDD-S patients. Such research could lead to the identification of depression-associated biomarkers.

Limitations

This study examined a small sample—the robustness of similar studies can be improved by increasing the sample size. A larger number of samples can be attained in future studies via sample size estimation. Furthermore, within-group gene expression studies would improve our knowledge of the essential pathways and functions that are affected when gene expressions are altered. Moreover, for the altered genes, expression can be validated in the laboratory using larger samples than that used in the current study.

CONCLUSION

The current study revealed that many genes and pathways are dysregulated and altered in the patients with MDD. Genes whose expressions are altered could serve as biomarkers to aid the discovery of drugs and the diagnosis of MDD and MDD-S. Our results show that the hippo signaling pathway is altered by the dysregulation of CCND3, SMAD3, FBXW11, CTNNA3, CTNNA2, and DLG1 genes. Similarly, retrograde endocannabinoid signaling is altered due to the dysregulation of GABRA2, ADCY2, GNG2, GRIA4, and MAPK10 genes. This study provides new ways to identify biomarkers associated with depression.