Introduction

In mammals, male sex determination starts by activation of the testis-determining factor gene, SRY (sex-determining region on Y chromosome) 1. This gene encodes an HMG domain transcription factor that triggers the differentiation of Sertoli cells, and therefore testis formation. Although several genes involved in gonadal differentiation have been identified including SOX9 (an SRY-related gene that contains a similar HMG box domain), DAX1 (an X-chromosome-located orphan nuclear receptor), SF1 (steroidogenic factor, another orphan nuclear receptor) and WT1 (Wilms tumor suppressor), many genes involved in sex determination and differentiation genes are unidentified. A wild-type SRY gene has been observed in over 80% of patients with 46,XY pure gonadal dysgenesis, suggesting that other genes are involved 2. A gene termed DMRT1 (d sx- and m ab3-related transcription factor 1) is located at the distal portion of chromosome 9p and this gene is deleted in the monosomy 9p syndrome that includes defective testis development and occasional male-to-female sex reversal 3, 4, 5. The human DMRT1 gene shows significant molecular similarity to doublesex (dsx) (Drosophila) and mab-3 (Caenorhabditis elegans) 6, 7, 8. These genes encode putative transcription factors that have a common DNA-binding domain, termed the DM domain. Both dsx and mab-3 control aspects of sex-specific differentiation and are functionally related 7. Male-specific expression for the DMRT1 gene in early gonadogenesis is consistent with its role in human testis development 9. A Dmrt1 knockout mouse also demonstrated that the gene is required for testis differentiation after determination, but dispensable for ovary development 10. Furthermore, the Dmrt1 gene transposed to the Y chromosome in some fish (medaka) has become a master regulator gene (Dmrt1y/DMY) in male determination 11, 12. In birds, the Dmrt1 is sex-linked 13 on the Z chromosome, and there is a higher dosage of the gene in the male (ZZ) as compared with the female (ZW) 8, 14, 15, 16. These data show that the expression of the DM genes from invertebrates to vertebrates is sex-specific (or sex differential) and (with the exception of Drosophila) associated with male-specific development. However, we know remarkably little about the evolution and functions of the DM genes.

Here we identify and characterize multiple isoforms of human DMRT1 that were spliced alternatively in the testis. These transcripts contain coding regions by exonization of intronic sequences, especially by the exonization of Alu elements with the DMRT1 gene. The alternatively spliced Alu exons (AExs) enrich the gonad transcriptome and enhance the coding capacity, suggesting that Alu exonization is an evolutionary pathway that creates human-specific transcriptomic diversity.

Materials and methods

RACE analysis and cloning of alternatively spliced cDNAs of DMRT1

On the basis of the sequence information of the DM domain of human DMRT1, RACE primers were designed. 5' RACE was performed using human testis cDNAs (Marathon™ cDNA, CLONTECH Laboratories) as template. The primers were AP2, 5′ ACT CAC TAT AGG GCT CGA GCG GC 3′ and DM primer 3-1, 5′ACA AGC AGC TCG GCC GCA CTG 3′. We performed 3′ RACE using primers AP1 and DM primer 5-1, 5′ TGG GTG CCG GGA GCA AGA AGT3′. After the PCR, nest PCR was carried out using primers AP2, and DM primer 5-N, 5′ TTT GCG GCC GCA AGT GCG CAC GCT GCA GGA AC3′. RACE cycling conditions were as follows: 35 cycles, each with 30 s, 94 °C; 2 min, 70 °C in a 20 ml Advantage 2 Polymerase Mix (CLONTECH Laboratories). All bands amplified were gel-purified, cloned into the pCRII (Invitrogen, USA) and sequenced using an ABI 377 autosequencer.

RNA isolation, cDNA synthesis and RT-PCR

All the RNAs were prepared by the RNeasy Mini kit (Qiagen, USA) according to the manufacturer's instructions,digested by Rnase-free DNase I and purified. About 3 μg RNAs were used as template for reverse transcription using 0.5 μg poly(T)20 primer and 200 U MMLV reverse transcriptase (Promega, USA). RT-PCR (non quantification) was used to amplify individual isoforms of DMRT1 from the adult testis cDNA. PCR cycling condition were as follows: 35 cycles, each with 30 s, 94 °C; 30 s, 68 °C; 30 s, 72 °C in a 20 μl reaction mix containing 10 mM Tris-HCl pH 8.3, 1.5 mM MgCl2, 50 mM KCl, 200 μM dNTP, 0.2 μM each primer, 1 U Taq DNA polymerase. The primers were as follows: sense primer for DMRT1a, b and c, 5′ CGG CTG CCC AAG TGC GCA 3′, and antisense primers, 5′ CTC GTC CTC CTC GAT GAC GG 3′ for DMRT1a; 5′ CAA GTG CGT GTG TGC CTC TG 3′ for Dmrt1b, and 5′ CTA CTT GGG AGG CTG AGG CAG 3′ for Dmrt1c, and antisense nest primer 5′ GCG CTT GTG GCC CTT GAG CG 3′ for Dmrt1c, and GAPD primers: 5′ TCC AAA ATC AAG TGG GGC GA 3′ and 5′ AGT AGA GGC AGG GAT GAT GT 3′.

Real time fluorescent quantitative RT-PCR

Real time RT-PCR was used for the quantification of the expression of both DMRT1a and b using the multi-channel RotorGene 3000 (Corbett Research, Australia), according to the supplied protocol. PCR cycling condition were as follows: 5 min at 95 °C; 40 cycles of 30 s at 95 °C, 30 s at 68 °C, 30 s at 72 °C in a 25 μl reaction mix containing 0.5×Sybr Green I. The primers used were the same as those in RT-PCR. A serial dilution of cDNA samples was simultaneously used to amplify GAPD to determine the quantity of cDNAs as standards. Based on the standards, the DMRT1 cDNAs were then amplified using a serial dilution of the cDNAs as templates, and the relative expression levels of the DMRT1 to GAPD were determined. For robustness issues, each sample was performed in triplicate or more. Data were analyzed by the software Rotor-gene version 4.6.

DMRT1 gene structure analysis

All sequences were analyzed with the Vector NTI software. Sequence alignments were performed using the ClustalW program. Genomic structure of the DMRT1 was analyzed by searching GenBank, NCBI online, and by comparison of cDNAs with genomic contig, which encodes the DMRT1 cDNA.

Alu element analysis

The Alu element similarity searches were carried out by BLAST online. All Alu elements within the genomic sequence of DMRT1 were counted and their directions on both strands were determined.

Results

Alternative splicing of the DMRT1

In an attempt to isolate the DMRT1 transcripts from the human testis to determine transcriptional diversity, the Marathon cDNAs of the human testis synthesized from the AP primer were used as template for both 5' RACE and 3' RACE by combined use of the primer pairs of DM domain and AP primer and nested PCR. RACE PCR results indicated several bands in 3′ RACE, while a specific band was observed for the 5' RACE. Sequencing analysis demonstrated that there are three transcripts (Figure 1). Interestingly, sequencing comparisons indicated two kinds of alternative splicing events of the DMRT1 transcript in the human testis, all of which occurred in the region 3′ to the DM domain. Three spliced transcripts, DMRT1a, DMRT1b and DMRT1c, were identified in the testis and they encoded three predicted proteins of 373, 275 and 175 aa, respectively (Figure 2). The expression patterns of these spliced isoforms were analyzed by RT-PCR and quantitated using real-time fluorescent quantitative RT-PCR from adult testis. Expression level of DMRT1a is the highest among these transcripts and 4.4 times higher than DMRT1b expression (Figure 3A and B). DMRT1c expressed much lower and can only be detected by nest PCR (Figure 3A).

Figure 1
figure 1

Alignment of predicted amino-acid sequences of three isoforms of human DMRT1; GenBank accession numbers are AY442914 for DMRT1b and AY442915 for DMRT1c. Triangles indicate the splicing sites. Arrows show the primer positions for RACE analysis.

Figure 2
figure 2

Diagram of alternatively spliced transcripts of human DMRT1 gene. DMRT1 is spliced to form different mRNAs transcripts: DMRT1a, DMRT1b and DMRT1c, which encode DM proteins with different amino acids (numbers above each line). DM domains are indicated by shaded boxes. Alternatively spliced regions in 3' region are showed by different color. In Dmrt1c, the splicing event occurred within the DM domain. Therefore, the new transcript missed the distal part of the DM domain. The numbers in the end under the lines indicate nucleotide numbers of these cDNAs.

Figure 3
figure 3

Non-quantitative RT-PCR (A) and Real time fluorescent quantitative RT-PCR (B) analysis of DMDRT1. (A) Non-quantitative RT-PCR of DMRT1 transcripts shows their expression in adult testis. Both DMRT1 a and b can be amplified by first PCR; however, DMRT1c showed much lesser expression and can only be detected by nest PCR. PCR amplification of GAPD was used as a control. 100 bp DNA ladder was loaded on the right lane as a molecular weight marker. Real-time fluorescent quantitative RT-PCR analysis (B) shows the relative expression of both DMRT1a and b to GAPD in the testis. Data were analyzed by the Rotor-gene software, version 4.6.

Gene structure and exonization of intronic sequences of human DMRT1

After comparing the genomic sequence with mRNA sequences, we found that there were seven exons, which were alternatively spliced to generate multiple DMRT1 isoforms (Figure 4). In isoform DMRT1b, exon 3 continues to read through intron 3 to form a new exon 6, while in DMRT1c, exon 1 continues to read through intron 1 to form another new exon 7, suggesting that alternative splicing enriches the gonad transcriptome and enhances the coding capacity.

Figure 4
figure 4

Schematic representation of the genomic organization and mRNA sequences of human DMRT1. Lines at the top show the genomic contig from gene database. Solid bars represent exons (E1-E7), whereas introns are indicated by the straight lines. Non random distribution of Alu elements on both directions of both DNA strands of human DMRT1 gene. Arrows and numbers indicate numbers and locations of Alu elements on sense and antisense strand, respectively. Arrows indicate the directions of Alu elements.

Exonization of Alu elements of the human DMRT1

A BLAST search against these isoforms indicated that half of the DMRT1c exon 7 was generated by the exonization of an Alu sequence in intron 1. A further screen for Alu elements within the whole gene revealed that 99 Alu elements in the DMRT1 gene are distributed among the non-coding regions in both directions, especially within introns, except for the exon 7 of the DMRT1c (Figure 4), which suggested that the locations of the Alu elements within the DMRT1 are not random.

Discussion

The surprisingly low number of genes in the human genome suggested that alternative splicing could account for biological complexity. Indeed, bioinformatic analyses indicate that at least 50% of all human genes participate in alternative splicing, which contributes significantly to human proteome complexity 17, 18. Alternative splicing is often regulated according to tissue type, developmental stage, sex or physiological conditions. Aberrant regulation of alternative splicing has been implicated in an increasing number of human diseases, including cancer 19. Human DMRT1 in testis tends to be spliced alternatively at 3′ region. Interestingly, alternative splicing events at 3′ termini of several DM-domain genes were also observed: for example, dsx of Drosophila 20 and zebrafish Dmrt1 21. The dsx is alternatively spliced at 3′ region, by default, into the male form dsxm, and the female dsxf is generated by a female-specific TRA. Diverse isoforms of the DM genes generated by alternative splicing at 3′ termini may provide potentially diverse targets for different upstream and downstream interacting factors in sexual regulation. Few targets of regulation by DM factors have yet been clearly identified, and most of the data are obtained from dsx of Drosophila and mab-3 of C. elegans 22, 23. Further identification of interactive factors with DMRT1 in vertebrates will determine how these spliced DMRT1s operate.

Exonization of intronic sequences is a mechanism that not only enriches the transcriptome and enhances the coding capacity of human genome for a diversity of biological activities 24, 25, but also provides genetic sources for evolution. Exonization of intronic sequences of human DMRT1 was generated by alternative splicing. It seems that alternative splicing might be a force for both enhancement of the coding capacity of human genome and molecular evolution, and DMRT1 as a target of these processes would have an important impact on the evolution of sexual development mechanisms. Alu sequences are short interspersed elements (SINEs), typically 300 nucleotides in length. Alu elements multiply within the genome through RNA polymerase III-derived transcripts in a process termed retroposition. Despite their lack of known function, recent findings suggests that Alu elements have a broad evolutionary impact 26. In human DMRT1, non random distribution of Alu elements on both directions of both DNA strands suggest a biological role. Alu-mediated homologous unequal recombination may result in genetic defects. Exonization of Alu elements, that is, insertion into mature mRNAs by alternative splicing, generated isoforms of DMRT1, which might provide a redundant/alternative way for functions of the gene in spermatogenesis, as sexual development is an essential process of reproduction.