Biomedical Research

Journal Banner

De Novo transcriptome assembly for analysis and functional annotation of genes expressed in Alport syndrome iPSCs

Wenbiao Chen1#, Jianrong Huang2#, Yong Dai3*

1Shenzhen Guangming New District People’s Hospital, Shenzhen, Guangdong, PR China

2The Third People’s Hospital of Shenzhen, Shenzhen, Guangdong, PR China

3Second Clinical Medical College of Jinan University, Shenzhen People’s Hospital, Shenzhen, Guangdong, PR China

#These authors contributed equally to this work

*Corresponding Author:
Yong Dai
The Third People’s Hospital of Shenzhen
Shenzhen, Guangdong, PR China

Accepted on June 04, 2016

Visit for more related articles at Biomedical Research

Abstract

Alport syndrome (AS) is an inherited disorder of collagen that affects the kidney, eye and cochlea. About 85% of AS cases are caused by a mutation in X-linked COL4A5, which encodes the alpha 5 chain of type IV collagen. AS patients inevitably develop end-stage renal disease and need replacement therapy. The mechanism by which the gene mutation results in AS is not completely known, in part because of a lack of genomic and transcriptome information about AS. In this study, an AS family contained three generations was subjected to comprehensively analyse. We performed high-throughput transcriptome sequencing on induced pluripotent stem cells (iPSCs) from AS renal tubular cells. Transcript sequences were used for gene analysis and functional characterization. Using an Illumina sequencing platform, 26,886,745 raw reads were acquired from AS cells and 29,252,903 from normal control (NC) cells. After quality control and filtering of raw reads, we obtained 26,021,874 clean reads from AS cells and 27,551,343 from NCs. Clean reads were analyzed for differences in gene expression, gene ontology (GO) analysis, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment, alternative splicing, and novel transcript prediction. Analyses showed 1168 differentially expressed genes between AS and NC samples, with 786 upregulated and 382 down regulated. GO analysis showed that the largest proportions of differentially expressed genes were in membranes and membrane components. The mitogen-activated protein kinase (MAPK) signalling pathway had the most differentially expressed genes by KEGG analysis. We predicted 881 novel transcripts in AS cells and 963 in NCs. Novel transcripts were assessed for protein-coding potential using a coding potential calculator. We used SOAP splice to detect alternative splicing of mRNA. This study lays a foundation for further research on population genetics and gene function analysis in AS.

Keywords

Alport syndrome, Transcriptome, iPSCs, Gene ontology, KEGG pathways, Alternative splicing.

Introduction

Alport syndrome (AS) is a hereditary disease that leads to kidney failure is caused by mutations to the COL4A3, COL4A4, COL4A5 genes, and absence of collagen α3α4α5 () networks found in mature kidney glomerular membrane (GBM). About 80% of AS is X-linked, due to mutations in COL4A5, the genetic encoding the alpha 5 chain of type collagen. . Induced Pluripotent Stem Cells (iPSCs) are selfrenewable and can differentiate to different types of adult cells, which has shown great promises in the field of regenerative medicine. AS is often accompanied by progressive, high-tone sensorineural hearing loss and ocular changes in form of macular flecks and lenticonus [1,2]. AS is a heterogeneous genetic disease caused by mutations in collagen type IV. Changes in podocytes and the GBM lead to early kidney fibrosis [3,4]. AS has a prevalence of 1 in 5000, and 85% of patients have the X-linked form [5]. Patients with AS commonly require renal replacement therapy by age 20 or 30 years [6]. AS diagnosis is mainly made by history and physical examination, detailed family history, urinalysis, immunohistochemical analysis of basement membranes, and examination of renal biopsy specimens by electron microscopy [7,8]. Since early stage AS leads to end-stage renal disease, early diagnosis is important [9]. Because AS is genetically heterogeneous, it can be caused by mutations in one of several genes [10]. Molecular genetics could be a powerful tool for definitive AS diagnosis [11]. We have successfully generated iPSCs from renal tubular cells previously [12]. In this study, we performed De Novo transcriptome assembly to analyze the AS family transcriptome base on iPSCs. Our goal was to understand differences in gene expression and perform bioinformatics analysis including gene ontology (GO) enrichment and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis. Our data could be an important step in establishing genetic research on AS, as well as providing insights into AS pathogenesis. Our results could contribute to potentially using genes as diagnostic or prognostic tools, or therapeutic targets for AS.

Materials and Methods

Clinical sample collection

We have clinically identified AS family contained three generations (Figure 1). The propositus (III 3) who is female and 26 years old was clinically observed gross hematuria and albuminuria. She was diagnosed AS in Second Clinical Medical College of Jinan University in 2013. The propositus was subjected to kidney pathological examination and the biopsy specimens were examined under light microscope and electron microscope. The propositus grandmother (I2) was also diagnosed AS and passed away of kidney failure. The propositus mother (III 3) behaving AS symptom, included kidney failure, gross hematuria, albuminuria, sensorineural hearing loss and pathognomonic ocular lesions. Propositus sister (III 3) was also clinically observed mild gross hematuria and albuminuria. Six members from the AS family were participated in our molecular research. This study protocol and consent forms were approved by Jinan University and adhere to Helsinki Declaration guidelines on ethical principle for medical research involving human subjects. Both participants provided written informed consent.

biomedres-Alport-pedigree

Figure 1: X-linked Alport system pedigree chart and the result of restriction fragment length palymorphism. normal male; normal female; male patient; female pateient; male′ death; female′ death; proposita.

After the AS family analysis, we selected 6 members (3 AS patient and 3 healthy people) from AS family to further research. Propositus (III 3), his mother (II 3) and his sister (III 3) acted as experimental group (AS). His sister (III 4), his brother (III 1) and his father 250 ml (II 4) acted as normal control (NC). Aseptic midstream urine was collected in the morning from each participant and was bottled into glass vials, which had been freeze-dried, c-irradiated, and filled with 5 ml penicillin-streptomycin antibiotics in temperature. Then, we separated out the renal tubular cells from urine. The renal tubular cells were reprogrammed to generate human iPSCs [12]. The iPSCs were our ultimate specimen that been used to further research in our study.

Total RNA isolation, cDNA library preparation

Total RNA was extracted from iPSCs using TRIzol reagent (Invitrogen, USA) according to the manufacturer’s protocol. We pool equivalent amount of total RNA from each sample into a single large combination to maximize the diversity of transcriptional unit’s according to same groups, respectively. DNase I (Ambion, USA) was used to remove genomic DNA from RNA samples. mRNA was purified from total RNA using oligo(dT) magnetic beads and fragmented in fragmentation buffer at 70ºC for 5 min. Cleared RNA fragments were copied into first-strand cDNA using reverse transcriptase and random primers. Second-strand cDNA synthesis used DNA polymerase I and RNase H. Synthesized cDNA was subjected to end-repair and phosphorylation, and 3’-adenylated with Klenow Exo-(3’- to-5’ exo minus, Illumina). Illumina paired-end adapters were ligated to the ends of the 3’-adenylated cDNA fragments. After agarose gel electrophoresis, cDNA libraries were constructed with 200 bp insection fragment. After validating on an Agilent 2100 bioanalyzer (Agilent Technologies, Santa Clara, CA, USA), mRNA-sequence libraries were sequencing on an Illumina HiSeq 2000 sequencing platform.

Sequence data analysis and alignment

Primary sequencing data as raw reads were subjected to quality control (QC) to determine if the raw reads were suitable for mapping. QC examined base composition and base quality. Base composition is in Figure 2. The quality distribution of bases among reads is in Figure 3.

biomedres-composition-reads

Figure 2: Base composition of reads.

biomedres-Quality-reads

Figure 3: Quality distribution of bases along reads.

After QC procedures, raw reads were filtered by removing adapter sequences, reads in which unknown bases were greater than 10%, and reads in which more than 10% of bases had a quality score <20. The resulting clean reads were aligned to the references using SOAPaligner (http://soap.genomics.org.cn) as described by Li et al. [13]. Alignment data were used to calculate the distribution of reads on reference and for coverage analysis. Alignments that passed alignment quality control were used in downstream analyses including differential expression analysis, GO enrichment, and KEGG pathway assignment and prediction of alternative splicing and novel transcripts.

Differential gene expression

Relative transcript abundance was determined as read per kilobase of an exon model per million mapped reads (RPKM) [14], calculated as RPKM=109C/NL, where C is the number of reads that are uniquely aligned to a gene, N is the total number of reads uniquely aligned to all genes, and L is the number of bases in the gene. Differentially expressed genes between AS and NC cells were detected by IDEG6 software (http:// telethon.bio.unipd.it/bioinfo/IDEG6/) [15] using a general chisquare test based on RPKM values. Test results were corrected for false discovery rate (FDR) using FDR ≤ 0.001 and an absolute value of (log2Ratio) ≥ 1 as the threshold for significance for gene expression differences. log2Ratio was analysed as the RPKM values of the gene in one sample was at least 2 times that of the gene in another sample.

Bioinformatics analysis

The GO international gene function classification system was used to map all genes with significantly different expression in AS and NC cells to GO terms (http://www.geneontology.org/), calculating a gene number for every term. A hypergeometric test found significantly enriched GO terms. Bonferroni correction [16] was performed on calculated p-values with corrected p-value ≤ 0.05 as a threshold. GO terms with p ≤ 0.05 were defined as significantly enriched. KEGG was used to analyze for pathway enrichment of genes with significantly different expression in AS and NC cells to understand the biological functions of genes and identify significantly enriched metabolic pathways or signal transduction pathways compared with the whole genome background. Calculations were as for GO analysis with p ≤ 0.05 as the standard for significance.

Alternative splicing

Alternative splicing generates different mRNA transcripts from a single gene that can be translated into different proteins [17]. We used SOAP splice (http://soap.genomics.org.cn/ soapsplice.html) software to identify splice junctions, predicting seven alternative splicing types: exon-skipping (ES), intron-retention, alternative 5’splice site, alternative 3’splice site, alternative first exon, alternative last exon, and mutually exclusive exon (MXE) (Figure 4) as described in Zhang et al. [18].

biomedres-Seven-splicing

Figure 4: Seven types of alternative splicing.

Novel transcript prediction and assessment of proteincoding potential

To discover novel transcribed regions, we compared assembled transcripts and annotated genomic transcripts to reference sequences. Predicted novel transcripts were required to meet three requirements: at least 200 bp from an annotated gene, more than 180 bp length, and sequencing depth no less than 2. After predicting novel transcripts, we investigated their functions. We distinguished protein-coding RNAs from noncoding RNA. We used the support-vector machine-based classifier Coding Potential (CPC) (http://cpc.cbi.pku.edu.cn/) to assess protein-coding potential. Negative scores indicated noncoding transcripts. Scores between 0 and 1 indicated weak potential for coding. Scores ≥ 1 indicated strong potential for coding.

Expression profiling by qRT-PCR

The differential expression of selection of 6 genes identified as being differentially expressed was validated by applying qRTPCR. GAPDH was selected as the internal control. In brief, 2 μg of total RNA from each sample was reverse transcribed for CDNA synthesis using a reverse transcription kit according to the manufacturer's protocol (Promega, Madison, WI). Amplification of CDNA was performed in the presence of genes specific primers and the SYBR Green PCR master mix (Applied Biosystems, Foster city, CA, USA) in MicroAmp Optical 96-well reaction plates with optical cover using an ABI prism 7500 Sequence Detector (Applied Biosystems). The sequences of the primer pairs designed using Primer Express Software V2.0 were listed in Table 1. The PCR amplification was carried out as follows: 95 for 2 min, followed by 40 cycles of amplification (94 for 10 seconds, 59 for 10 seconds, 72 for 45 seconds). The expression of each gene was confirmed in three rounds of independent qRT-PCR reaction. The relative expression level of each gene was normalized by against GAPDH. Fold change was calculated according to the 2-Ct method.

Primer name sequences 5' to 3' TM
XIST-F AAAGTGGCCGCCATTTTAGA 57
XIST-R CAACAATCACGCAAAGCTCC
CX3CL1F CGTGCAGCAAGATGACATCA 59
CX3CL1R TCCTTGACCCATTGCTCCTT
LRRC55F AATGGACACCCGAAACCTCA 57
LRRC55R TGGCACATGGCTGAAATTGT
FAM18B1 F AATGGTTGGCCTACGTTGGT 59
FAM18B1 R TGGACAGGCAATAAGTCCCA
AURKC F AGCGCACAGCCACGATAATA 59
AURKC R TGCACAGACCAGCCAAAATC
RPS4Y1 F ATGGCAAGGTTCGAGTGGAT 59
RPS4Y1 R GATGCGGTGAACAGCAAAAC
GAPDH F ACCACAGTCCATGCCATCAC 59
GAPDH R TCCACCACCCTGTTGCTGTA  

Table 1. qRT-PCR primer used in the validation assays.

Results

Sequencing data

After QC analysis and filtering of raw reads, we obtained 26,021,874 clean reads from AS and 27,551,343 from NCs. We aligned clean reads onto the reference gene and reference genome. As shown in Table 2, 61.16% of AS reads mapped to the gene and 88.54% mapped to the genome, while 63.90% of NC reads mapped to the gene and 88.24% to the genome. For AS samples, 47.97% perfectly matched the reference gene and 61.12% perfectly matched the genome; For NC reads, 52.07% perfectly matched the gene and 62.84% the genome. RNA sequencing methods chemically fragmented mRNAs into short segments. If fragmenting was not random, read preference for specific gene regions might affect subsequent bioinformatics analysis. Therefore, we used the read distribution of genes to evaluate randomness and found that the reads were evenly distributed over genes from, 5’ to 3’ (Figure 5). We also determined the distribution of gene coverage in the AC and NC transcriptome. Gene coverage was defined as the percent of genes covered by the reads and the value was determined as the ratio of total bases of a gene covered by uniquely mapped reads to total bases of the gene. A high percentage of genes (46% in AS, 51% in NC) showed 90–100% coverage and most gene coverage was higher than 50% (70% in AS, 77% in NC) (Figure 6). The raw sequencing data were submitted to NCBI Sequence Read Archive (SRA, http://www.ncbi.nlm.nih.gov/ Traces/sra_sub/sub.cgi) under the accession number of SRP041474.

Sample AS NC
Map to Gene    
Total Reads 52043748 (100.00%) 55102686(100.00%)
Total BasePairs 4683937320 (100.00%) 4959241740(100.00%)
Total Mapped Reads 31828347 (61.16%) 35208207(63.90%)
Perfect match 24963326 (47.97%) 28689725(52.07%)
≤5bp mismatch 6865021 (13.19%) 6518482(11.83%)
Unique match 30472073 (58.55%) 33651851(61.07%)
Multi-position match 1356274 (2.61%) 1556356(2.82%)
Total Unmapped Reads 20215401 (38.84%) 19894479(36.10%)
Map to Genome
Total Reads 52043748 (100.00%) 55102686(100.00%)
Total BasePairs 4683937320 (100.00%) 4959241740(100.00%)
Total Mapped Reads 46077502 (88.54%) 48621918(88.24%)
Perfect match 31810738 (61.12%) 34625794(62.84%)
≤5bp mismatch 14266764 (27.41%) 13996124(25.40%)
Unique match 41632944 (80.00%) 43818058(79.52%)
Multi-position match 4444558 (8.54%) 4803860(8.72%)
Total Unmapped Reads 5966246 (11.46%) 6480768(11.76%)

Table 2. Alignment for AS and NC.

biomedres-reads-mapped

Figure 5: Distribution of reads mapped to reference genes.

biomedres-gene-coverage

Figure 6: Distribution statistics for gene coverage.

Analysis of differentially expressed genes

We used deep RNA sequencing to determine genes differentially expressed between AS and NC. Using FDR ≤ 0.001 and the absolute value of log2ratio ≥ 1 as threshold values, 1168 genes were found to be differentially expressed, with 786 upregulated and 382 downregulated. In addition, 33 genes were expressed only in AS cells and 26 were expressed only in NC cells. Among upregulated genes, XIST was the most changed, with an expression level that increased about 11-fold (log2ratio) in AS cells compared with NCs. The most changed downregulated gene was RPS4YI, with an expression level that decreased about 17-fold (log2ratio). The top 20 upregulated and downregulated genes are in Table 3. To validate gene expression profiles, we conducted qRT-PCR to confirm the expression level of 6 selected gene (Table 4). We can see from Table 4, the genes exhibited high abundance and were differentially expression between AS and NC. The expression pattern of 6 genes was consistent with the reads abundance of deep sequencing, suggesting that the robustness of deep sequencing based expression analysis. For example, gene AURKC, RYS4Y1 and FAM18B1 were downregualted, gene XLST, LRRC55 and CX3CL1 were upregualted in microarray analysis. The qRT-PCR verification had the same expression and proved the genes were reasonable and probable.

GeneID Name Gene length N-RPKM AS-RPKM log2 Ratio(AS/N) Mode P-value
up-regulated
7503 XIST 19271 0.003 7.22 11.19 Up 0.000110
9506 PAGE4 493 0.001 1.60 10.64 Up 0.000000
100132288 TEKT4P2 1640 0.001 1.40 10.45 Up 0.000031
3127 HLA-DRB5 1171 0.001 1.12 10.13 Up 0.000000
727764 MAFIP 2293 0.001 1.09 10.09 Up 0.001174
440224 CXADRP3 1613 0.001 1.06 10.05 Up 0.000234
343172 OR2T8 939 0.001 1.05 10.03 Up 0.000500
164668 APOBEC3H 1164 0.001 0.65 9.34 Up 0.000044
154790 CLEC2L 1288 0.001 0.61 9.26 Up 0.000355
127064 OR2T12 963 0.001 0.55 9.09 Up 0.000990
1116 CHI3L1 1867 0.001 0.46 8.84 Up 0.001990
100130827 SGK110 1270 0.001 0.39 8.60 Up 0.000320
79190 IRX6 2337 0.001 0.35 8.46 Up 0.002000
5178 PEG3 8765 0.001 0.34 8.43 Up 0.002622
440695 ETV3L 1977 0.001 0.33 8.37 Up 0.000000
78989 COLEC11 1399 0.001 0.33 8.36 Up 0.001174
129804 FBLN7 2329 0.001 0.28 8.14 Up 0.000006
54346 UNC93A 2132 0.001 0.26 8.03 Up 0.000622
146212 KCTD19 2911 0.001 0.26 8.02 Up 0.000340
50964 SOST 2322 0.001 0.23 7.82 Up 0.001430
down-regulated            
90665 TBL1Y 2407 0.605 0.00 -9.24 Down 0.000010
64641 EBF2 2297 0.673 0.00 -9.39 Down 0.000001
400655 LOC400655 2674 0.811 0.00 -9.66 Down 0.000100
3211 HOXB1 1014 0.821 0.00 -9.68 Down 0.160143
7652 ZNF99 3111 1.614 0.00 -10.66 Down 0.000002
360205 HOXB13-AS1 564 2.476 0.00 -11.27 Down 0.000107
83869 TTTY14 515 2.712 0.00 -11.41 Down 0.000003
64595 TTTY15 5262 2.999 0.00 -11.55 Down 0.000000
3212 HOXB2 1614 3.277 0.00 -11.68 Down 0.015301
55410 NCRNA00185 1508 3.311 0.00 -11.69 Down 0.000000
246126 CYorf15A 872 4.123 0.00 -12.01 Down 0.001004
9087 TMSB4Y 1702 4.295 0.00 -12.07 Down 0.004300
8287 USP9Y 10048 5.223 0.00 -12.35 Down 0.003184
8284 KDM5D 5595 6.814 0.00 -12.73 Down 0.001083
6736 SRY 897 8.779 0.00 -13.10 Down 0.003529
84663 CYorf15B 3315 10.192 0.00 -13.32 Down 0.000067
84366 PRAC 403 11.134 0.00 -13.44 Down 0.002000
8653 DDX3Y 4648 23.956 0.00 -14.55 Down 0.000140
9086 EIF1AY 1399 36.364 0.00 -15.15 Down 0.000003
6192 RPS4Y1 910 222.870 0.00 -17.77 Down 0.000000

Table 3. Top 20 differentially expressed genes between AS and NC.

Gene Cт Cт Mean Control Cт Mean ⊿Ct Normal control⊿Ct ⊿⊿Ct 2-⊿⊿Ct
AURKC AS 24.982 24.851 17.813 7.038 7.038 0 1
AS 24.896 24.851 17.813 7.038 7.038 0 1
AS 24.675 24.851 17.813 7.038 7.038 0 1
NC 22.562 22.700 18.127 4.573 7.038 -2.465 5.521
NC 22.545 22.700 18.127 4.573 7.038 -2.465 5.521
NC 22.994 22.700 18.127 4.573 7.038 -2.465 5.521
RPS4Y1
AS 24.539 24.368 17.813 6.555 6.555 0 1
AS 24.438 24.368 17.813 6.555 6.555 0 1
AS 24.127 24.368 17.813 6.555 6.555 0 1
NC 21.994 22.093 18.127 3.966 6.555 -2.588 6.015
NC 22.115 22.093 18.127 3.966 6.555 -2.588 6.015
NC 22.171 22.093 18.127 3.966 6.555 -2.588 6.015
FAM18B1
AS 24.231 24.334 17.813 6.521 6.521 0 1
AS 24.307 24.334 17.813 6.521 6.521 0 1
AS 24.464 24.334 17.813 6.521 6.521 0 1
NC 23.665 23.761 18.127 5.634 6.521 -0.887 1.849
NC 23.968 23.761 18.127 5.634 6.521 -0.887 1.849
NC 23.651 23.761 18.127 5.634 6.521 -0.887 1.849
XLST
AS 18.073 21.075 17.813 3.262 6.023 -2.762 6.782
AS 18.158 21.075 17.813 3.262 6.023 -2.762 6.782
AS 18.025 21.075 17.813 3.262 6.023 -2.762 6.782
NC 24.030 24.150 18.127 6.023 6.023 0 1
NC 24.148 24.150 18.127 6.023 6.023 0 1
NC 24.273 24.150 18.127 6.023 6.023 0 1
LRRC55
AS 16.770 16.583 17.813 -1.230 4.065 -5.295 39.260
AS 16.611 16.583 17.813 -1.230 4.065 -5.295 39.260
AS 16.368 16.583 17.813 -1.230 4.065 -5.295 39.260
NC 22.365 22.192 18.127 4.065 4.065 0 1
NC 22.006 22.192 18.127 4.065 4.065 0 1
NC 22.206 22.192 18.127 4.065 4.065 0 1
CX3CL1
AS 17.421 17.525 17.813 -0.288 5.266 -5.554 46.978
AS 17.740 17.525 17.813 -0.288 5.266 -5.554 46.978
AS 17.414 17.525 17.813 -0.288 5.266 -5.554 46.978
NC 23.523 23.393 18.127 5.266 5.266 0 1
NC 23.473 23.393 18.127 5.266 5.266 0 1
NC 23.184 23.393 18.127 5.266 5.266 0 1

Table 4. qRT-PCR confirmation data.

GO and KEGG classification

GO analysis for cellular components, molecular function and biological processes was applied to determine significantly enriched functions for differentially expressed genes. In total, 41 GO terms were significantly enriched: 9 in cellular components, 2 in molecular function, and 30 in biological processes, using corrected p-value ≤ 0.05 as a threshold (Table 5). Among the enriched GO terms, terms intrinsic to membranes and membrane components were the most abundant in the cellular component category. The terms ion binding and cation binding were the most abundant in the molecular function category. The terms signalling, development process, multicellular organismal process, and anatomical structure development were the most abundant term in the biological process category. GO analysis networks are in Figure S1, Figure S2, Figure S3.

Gene Ontology term Genome frequency of use Corrected P-value
cellular component
intrinsic to membrane  5163 out of 16150 genes, 32.0% 0.036250
membrane  7064 out of 16150 genes, 43.7% 0.001843
extracellular region part  1092 out of 16150 genes, 6.8% 0.005600
extracellular region  1113 out of 16150 genes, 6.9% 0.002655
extracellular matrix  327 out of 16150 genes, 2.0% 0.000584
membrane part  5889 out of 16150 genes, 36.5% 0.000003
integral to membrane  1448 out of 16150 genes, 9.0% 0.000003
plasma membrane  1313 out of 16150 genes, 8.1% 0.011000
cell periphery  1328 out of 16150 genes, 8.2% 0.011000
molecular function
ion binding  3916 out of 15658 genes, 25.0% 0.020000
cation binding  3866 out of 15658 genes, 24.7% 0.027000
biological process  
system development  2497 out of 14935 genes, 16.7% 0.014789
cell communication  778 out of 14935 genes, 5.2% 0.032728
anatomical structure development  3031 out of 14935 genes, 20.3% 0.000837
multicellular organismal development  2812 out of 14935 genes, 18.8% 0.000002
cell-cell signaling  454 out of 14935 genes, 3.0% 0.001069
multicellular organismal process  4818 out of 14935 genes, 32.3% 0.003210
nervous system development 1029 out of 14935 genes, 6.9% 0.000163
central nervous system development  487 out of 14935 genes, 3.3% 0.000822
developmental process  3662 out of 14935 genes, 24.5% 0.000047
cell-cell adhesion  267 out of 14935 genes, 1.8% 0.010889
regulation of system process  364 out of 14935 genes, 2.4% 0.010555
organ development  1615 out of 14935 genes, 10.8% 0.000299
signaling  4026 out of 14935 genes, 27.0% 0.001280
behavior  381 out of 14935 genes, 2.6% 0.000000
cell differentiation  1403 out of 14935 genes, 9.4% 0.043200
tissue development  825 out of 14935 genes, 5.5% 0.000804
brain development 293 out of 14935 genes, 2.0% 0.001000
regulation of transmission of nerve impuls 196 out of 14935 genes, 1.3% 0.001000
regulation of synaptic transmission  176 out of 14935 genes, 1.2% 0.001000
regulation of neurological system process  212 out of 14935 genes, 1.4% 0.002000
regulation of multicellular organismal process 1000 out of 14935 genes, 6.7% 0.003000
cell adhesion 467 out of 14935 genes, 3.1% 0.010000
biological adhesion 467 out of 14935 genes, 3.1% 0.010000
cellular developmental process  1841 out of 14935 genes, 12.3% 0.012000
anatomical structure morphogenesis 1355 out of 14935 genes, 9.1% 0.014000
pattern specification process  305 out of 14935 genes, 2.0% 0.019000
cell surface receptor linked signaling pathway  1703 out of 14935 genes, 11.4% 0.028000
regionalization  265 out of 14935 genes, 1.8% 0.032000
regulation of cell communication 606 out of 14935 genes, 4.1% 0.034000
tissue morphogenesis  304 out of 14935 genes, 2.0% 0.041000

Table 5. GO classifications of genes significantly differentially expressed between AS and NC cells.

The KEGG database was used to categorize gene functions into biochemical pathways. The 1168 differentially expressed genes were assigned to 208 KEGG pathways. However, only 15 pathways contained significantly enriched terms by corrected p ≤ 0.05 (Table 6). The pathways with the highest representation of genes were MAPK signalling pathways (46 genes, 4.79%), neuroactive ligand-receptor interactions (39 genes, 4.06%), and cell adhesion molecules (38 genes, 3.95%). The MAPK signalling pathway is in Figure 7.

biomedres-signalling-pathway

Figure 7: MAPK signalling pathway. Upregulated genes in red, downregulated genes in green and genes that did not change in black.

Pathway Gene with pathway annotation (961) Corrected P-value
Cell adhesion molecules 38 (3.95%) 0.001100
Neuroactive ligand-receptor interaction 39 (4.06%) 0.002500
MAPK signaling pathway 46 (4.79%) 0.002600
Allograft rejection 12 (1.25%) 0.002600
Type I diabetes mellitus 12 (1.25%) 0.008000
VEGF signaling pathway 21 (2.19%) 0.008000
Autoimmune thyroid disease 13 (1.35%) 0.008000
B cell receptor signaling pathway 22 (2.29%) 0.015200
Graft-versus-host disease 12 (1.25%) 0.022400
Axon guidance 32 (3.33%) 0.022700
Gap junction 19 (1.98%) 0.023500
Calcium signaling pathway 29 (3.02%) 0.028100
Rheumatoid arthritis 15 (1.56%) 0.030900
Wntsignaling pathway 28 (2.91%) 0.031300
Glioma 17 (1.77%) 0.038200

Table 6. KEGG pathways with genes significantly differentially expressed between AS and NC cells.

Identification of alternative splicing

Alternative splicing generates different mRNA transcripts that are translated into distinct proteins. We determined the number of alternative splicing possibilities and the number of genes involved in alternative splicing (Figure 8). As many as 10,567 genes could be alternatively spliced in 19,530 ways in the AS libraries and 11,616 genes could be alternatively spliced in 22,319 ways in the NC libraries. ES was the major class of alternative splicing events, accounting for 35.8% (6987 in AS, 7981 in NC) of all alternative splicing in the AS and NC libraries. Only 5 examples of MXE were found in the AS and NC libraries.

biomedres-splicing-genes

Figure 8: Alternative splicing and genes involved.

Identification of novel transcript and annotation

Alignment of the sequencing data to the reference indicated 881 novel transcripts in the AS libraries and 963 in the NC libraries. About 80% of the novel transcripts (710 in AS, 794 in NC) were longer than 500 bp. Among the novel transcripts, 218 in AS and 243 in NC had the potential to encode proteins. We used CPC scores to assess protein-coding potential. The strongest potential for coding among the novel transcripts was seen for novel transcript 819 (chromosome 9, 226 bp) in the AS libraries with a score of 14, and novel transcript 586 (chromosome 22, 2771 bp) in the NC libraries with a score of 13. In the AS libraries, 115 novel transcripts had strong potential as coding transcripts (score ≥ 1), and 103 had weak potential as coding transcripts (score value 0-1). In the NC libraries, 126 transcripts had strong coding potential and 117 had weak coding potential. The top 20 novel transcripts with coding potential are in Table 7.

Novel transcriptome ID Chromosome Length Cpc score
AS
NovelTr_819 chr9 2226 14.8984
NovelTr_83 chr10 2504 14.2614
NovelTr_667 chr6 1474 12.7469
NovelTr_407 chr19 3249 12.6963
NovelTr_283 chr16 1857 11.8018
NovelTr_692 chr6 2965 11.586
NovelTr_47 chr1 1259 11.3246
NovelTr_221 chr14 1009 10.9114
NovelTr_410 chr19 1424 10.5572
NovelTr_409 chr19 1057 10.5202
NovelTr_517 chr3 752 9.49248
NovelTr_693 chr6 1614 9.3311
NovelTr_37 chr1 1441 9.07706
NovelTr_380 chr19 2272 8.55492
NovelTr_815 chr9 731 8.0968
NovelTr_783 chr8 728 7.71177
NovelTr_710 chr7 4357 7.63221
NovelTr_845 chrX 990 7.59434
NovelTr_795 chr9 1967 6.88168
NovelTr_576 chr4 970 6.79338
NC      
NovelTr_586 chr22 2771 13.6726
NovelTr_585 chr22 2773 13.3342
NovelTr_464 chr19 3154 12.7105
NovelTr_601 chr3 840 10.4092
NovelTr_873 chr8 789 9.32909
NovelTr_465 chr19 661 9.1925
NovelTr_34 chr1 1002 8.95671
NovelTr_620 chr3 2624 8.83077
NovelTr_872 chr8 639 8.6662
NovelTr_33 chr1 1005 7.98502
NovelTr_336 chr16 695 7.46555
NovelTr_582 chr22 687 6.82787
NovelTr_632 chr3 2311 6.63984
NovelTr_441 chr19 4786 6.26961
NovelTr_550 chr20 1143 6.18412
NovelTr_427 chr19 1560 6.05214
NovelTr_47 chr1 1578 6.01541
NovelTr_337 chr16 597 5.85511
NovelTr_269 chr14 670 5.55965
NovelTr_691 chr4 2443 5.34056

Table 7. Predicted novel transcripts and their protein-coding potential.

Discussion

In previous study, we generated iPSCs from renal tubular cells in urine samples from AS and NC on AS family [12]. Unlike traditional experimental samples, iPSCs have potential for investigating human illnesses through disease modeling, tissue engineering, drug discovery, and cell therapy [19]. IPSCs can be used to analyze gene networks, microRNA, signalling pathways and transcription factors using high-throughput sequencing platforms [20]. Our study compared AS and NC iPSCs to find differentially expressed genes and their GO enrichments and KEGG pathway assignments. The results yielded an integrated and accurate database for investigation of AS pathogenesis and potential genetic therapy.

In our preliminary analysis of genes differentially expressed between AS and NC, 1168 genes showed differential expression, with 786 genes upregulated and 382 genes downregulated. We also found that 33 genes were unique to AS and 26 genes were unique to NC. Of interest, all unique genes were among those that showed the greatest upregulation or downregulation by log2ratio. AS is a genetically heterogeneous disease associated with mutations in the COL4A5, COL4A3, COL4A4, and COL4A6 genes [21], which are important in the pathogenesis of AS. The 33 genes we found that were unique to AS showed strong upregulation relative to NCs and might be related to AS pathogenesis. This hypothesis requires further study and evaluation. In our study, the COL4A5, COL4A3, COL4A4, and COL4A6 genes were fully covered in the AS and NC libraries but were not significantly differentially expressed. We hypothesize that the analysis methods and threshold values we used to define significantly different expression excluded the COL4A5, COL4A3, COL4A4, and COL4A6 genes. Our data suggest new research directions for understanding AS pathogenesis. The gene with the greatest upregulation was XIST, a model for understanding the formation of facultative heterochromatin in mammalian development and a paradigm for RNA-mediated regulation of gene expression [22]. XIST is important to human disease; XIST deregulation is associated with human tumors and has potential for development in to diagnostic markers [23]. Our results with the XIST gene as the most differentially expressed gene between AS and NC samples suggests new potential research directions. Several researchers have spearheaded the research of XIST's interactome and the factors involved in silencing. Several novel proteins have now been shown to be required for the transcriptional silencing of the X chromosome and/or XIST spreading and localization to the inactive X chromosome. AS is a hereditary disease. About 80% of AS is X-linked. Just as the previous research that gene XIST played the important role in transcriptional silencing. However, the gene XIST was the greatest upregulation. We suspected weather the gene XIST in X chromosome was mutation and lost the inactivation and the outcome was upregulation without limitation. The gene XIST of AS was more energetic than NC. Of course, this suspect need to further research.

The differentially expressed genes in our study were assigned to a range of GO categories, suggesting a diversity of transcripts from the AS cell genome. Membranes and membrane components were the most abundant GO categories. The kidney GBM is a specialized extracellular matrix that supports and informs adherent cells of the glomerular endothelium and podocytes [24]. AS is a genetic disease of the GBM involving the COL4A5, COL4A3, COL4A4 andCOL4A6 network of type IV collagen genes [7,25]. Cosgrove et al. used an AS animal model to determine how the molecular makeup of the GBM affects glomerular function [26]. The finding that membranes and membrane components were the most abundant classes in our GO enrichment analysis indicated that AS pathogenesis included the absence of the subepithelial network of three chains in GBM. However, we could not determine whether the membranes and membrane components in the GO analysis indicated adherent cells, glomerular endothelium or podocytes. More research is needed on this topic. Differentially expressed genes were subjected to KEGG pathway analysis. The MAPK signaling pathway had the most representatives with 46 genes (4.79%). MAPKs are serine and threonine protein kinases that are activated by phosphorylation in response to extracellular stimuli such as mitogens, growth factors, cytokines and osmotic stress [27]. The activation of MAPK pathways is a potential mechanism in kidneys and kidney disease can be ameliorated by inhibiting MAPK signalling pathways [28,29]. Our results suggested that the MAPK signalling pathway is important in AS pathogenesis. This hypothesis is a basis for further research. The expression of kidney injury molecule-1 (KIM-1), a very sensitive and specific urinary biomarker for acute renal injury, was markedly upregulated in injured and regenerating renal proximal tubular epithelial cells following ischemic or toxic insults. However, the function of KIM-1 expression was regulated likely mediated via ERK MAPK signaling pathway [35-37]. Renal fibrosis results from an excessive accumulation of extracellular matrix that occurs in most types of chronic kidney disease. Transforming growth factor-β1 (TGF-β1) and inflammation after injury played critical roles in renal fibrotic processes. Inhibitory effects of TGF-β1-mediated myofibroblast activation were associated with down-regulation of MAPK [38]. Most of the research on the relationship between MAPK pathway and kidney disease proved that inhibition effects of MAPK was beneficial to the recovered of injured kidney and protected kidney disease from degenerating. In this research, MAPK signal pathway was active in AS, we certainly believed that MAPK signal pathway contributed to the development of AS. The method of inhibition MAPK signal pathway may be the effective therapy to treat AS. However, this was the imagination that also need to further research.

Alternative splicing is essential for protein diversity and function [30]. Alternative splicing is widespread in eukaryotes, but its biological function is incompletely understood. Palusa et al. found that serine/arginine-rich protein genes generate a large transcriptome that is altered by stresses and hormones [31]. Kalsotra et al. reported that alternative splicing drives physiological changes and can provide mRNA variability for other regulatory mechanisms [32]. We hypothesized that alternative splicing regulated or was associated with responses to a different environment. Therefore, we analyzed and found 19,530 alternative splicing possibilities corresponding to 10,567 genes in AS cells; and 22,319 alternative splicing possibilities corresponding to 11,616 genes in NC cells. AS had fewer alternative splicing possibilities than NC. Pritsker et al. highlighted alternative splicing regulation as important in signalling pathways for stem cell function [33]. The frequency of alternative splicing is high in tissue-specific genes compared to genes ubiquitous in stem cells. The negative regulation of constitutively active splicing sites could be a model for generation of splice variants and alternative splicing is generally not conserved between orthologous genes in humans and mice [33,34]. Since our samples were iPSCs, comprehensive identification of all biological molecules produced in iPSCs will be an important step to understanding AS pathogenesis and possible genetic therapies.

We characterized the transcriptome of AS iPSCs, comparing expression of AS and NC on an AS family. Genes were functionally annotated by comparison with databases such as GO and KEGG. We predicted novel transcripts and determine alternative splicing possibilities. To our knowledge, we attempt to assemble and characterize the transcriptome of AS iPSCs using an Illumina sequencing method. Our study on the AS transcriptome is a valuable resource for understanding of AS pathogenesis and for research on potential genetic therapies. The next research phase will focus on microRNAs; transcriptome proteomics; and long, non-coding RNAs in AS iPSCs. We will combine microRNA, transcriptome, transcriptome proteomics, and long non-coding RNA databases for network correlation analysis. However, our research as the basic research and there were a lot of question was unknown. Make advantage of iPSCs to research disease is rare and we have no reference materials to made comparison. The research on AS on genetic level was just the beginning and there was further research need to prove the previous outcome. Not to said the use of genetic material to diagnosis and treatment. Anyway, the research laid a foundation for us to try to study AS in genetic level and these databases will serve as references for understanding AS pathogenesis and potential genetic therapy.

Acknowledgement

We are grateful for the AS family for their willingness to participate in to the research. The research was supported by Guangdong Shenzhen Knowledge Innovation Program basic Research items (JXYJ20140416122812045).

References