A novel exon was identified between canine BRCA2 exons 1 and 2
We identified a novel exon between canine BRCA2 exons 1 and 2 via RT-PCR using samples from two ovaries and one testis (Fig. 1 A). These PCR products were sequenced to determine their splicing patterns (GenBank accession no. LC547963, LC547964, LC547965, LC547966, LC547967 and LC547968). We identified six splicing variants in addition to the annotated transcript (Fig. 1 B). These variants were identified between originally reported exon 1 and exon 2 in 5′ UTR of canine BRCA2 gene and complied with the GU-AG rule (Supplemental Fig. 1). These splice variants resulted in mRNA transcripts that varied in size from 248 bp to 988 bp. We referred to the splicing variants as I-VI, I being the variant that encodes the longest transcript and VI being the shortest. Novel splicing variant I contained 8 possible start codons, which potentially translate short and truncated proteins, and splicing variant VI, the shortest variant, contained 5 possible start codons, whereas the registered sequence does not have a start codon in the 5′ UTR (Supplemental Fig. 1). It is not easy to determine the precise expression level by RT-PCR, but splicing variants I, II, V, VI and registered sequence were the main transcripts (Fig. 1 A). We also tested the expression of these splicing variants in cultured cells derived from dogs. All four cell lines showed the presence of these spliced variants, similar to testis and ovary tissues (Fig. 1 C) and the splicing variants I, II, V, VI and the registered variant showed higher mRNA levels.
Expression pattern and levels of splicing variants in cultured cells were affected by serum starvation
The expression level of BRCA2 is regulated by the cell cycle and the condition of the cells. Thus, we tested the effect of serum starvation, which introduced G1 phase arrest, on cultured cell lines. The expression pattern of splicing variants in CHM-p cells changed due to serum starvation. However, MDCK and CIP-p cells maintained the same expression pattern even after serum starvation (Fig. 2 A). In CNM-p cells, expression levels of splicing variants III and IV and the registered sequence were reduced after serum starvation. In CHM-p cells, levels of splicing variants VI and the registered sequence were reduced after serum starvation (Fig. 1 and 2 A). CNM-p and CHM-p cells also showed relatively increased expression levels of splicing variant II compared to splicing variant I upon serum starvation (Fig. 2 A). In all the four cell lines mentioned above, the expression level of BRCA2 was reduced by serum starvation (Fig. 2 B). The relative ratio of expression levels of the BRCA2 variants with the shortest registered sequence to total BRCA2 was also changed (Fig. 2 C). MDCK and CIP-p cells showed an increased relative expression ratio of registered sequence, but it was decreased in the cases of CNM-p and CHM-p (Fig. 2 C).
Human cultured cell lines also expressed similar splicing variants
Canine tissues and cultured cells expressed splicing variants in canine BRCA2 5′ UTR, and the expression ratio of these variants was regulated by the condition of the cells. We hypothesized that regulation of BRCA2 gene expression via splicing variants could also be present in humans. We tested this using two human derived cell lines, HeLa and 293T. Splicing variants including the entire intron 1 sequence were expressed in these cell lines (Fig. 3 A and B, GenBank accession no. LC547969). The novel splicing variant contained 9 possible start codons, which potentially translate short and truncated proteins, whereas the registered sequence did not have a start codon in the 5′ UTR (Supplemental Fig. 1). Although the expression pattern of the splicing variants was not changed by serum starvation, the relative ratio of registered sequence and shorter splicing variant was reduced or increased with a slight increase in BRCA2 expression level in HeLa or 293T cells (Fig. 3 C). We also tested the effect of X-ray irradiation, as some genes have been reported to change the ratio of splicing variants to regulate expression level in response to DNA damage [23,24]. However, the relative ratio of the registered sequence did not show any drastic change until 60 min after 10 Gy X-ray treatment (Fig. 3 E).
Novel splicing variants of canine and human BRCA2 suppressed translational efficiency
The novel splicing variants of BRCA2 were expressed in canine tissues and both canine and human derived cell lines and these ratios were changed by the conditions in which the cells were maintained. We speculated that these splicing variants regulate the translational efficiency of BRCA2. To evaluate the translational efficiency of BRCA2, luciferase assay with quantitative PCR were employed (Fig. 4 A). To evaluate the translational efficiency, we designed plasmids containing CMV promoter followed by each 5′ UTR variant sequence and luciferase. These constructs might also contribute to the suppression of the transcriptional efficiency. Thus, to avoid this effect, we quantified the expression level of firefly luciferase and the transfection efficiency of the plasmid DNA construct containing firefly luciferase gene and genomic DNA of human BRCA2 exon 27 was determined via quantitative PCR. In canine BRCA2, the splicing variants, I, II, V, VI, and the registered sequence were the major variants in tissues and cell lines. Thus, we evaluated these five variants. Figure 4 B and D shows the relative expression level of firefly luciferase. Unexpectedly, splicing variants I and II of canine BRCA2 and splicing variant I of human BRCA2 suppressed the CMV promoter activity (Fig. 4 B and D). In parallel, luciferase assay was performed to evaluate the translational activity. The values of the transcriptional activity were used to normalize the relative translational activity. The splicing variants of canine BRCA2 excluding splicing variant I and splicing variant I of human BRCA2 were found to be associated with decreased translational activity (Fig. 4 C and E).
Intron 1 of canine and human BRCA2 also suppressed BRCA2 promoter activity
Novel splicing variants of canine and human BRCA2 suppressed CMV promoter activity in addition to the translational suppression. Thus, we next tested the effect of the DNA sequence of various splicing variants on BRCA2 promoter activity. Prior to this, canine and human BRCA2 intron 1 regions were compared to find conserved DNA sequences, because the novel exon region identified in intron 1 was different from the registered sequence and conserved sequence had the greatest potential to be a silencer element (Fig. 5). Three conserved regions were identified in canine and human BRCA2. A conserved 15 bp 5′-terminal region was located in the BRCA2 promoter. The other two regions, human BRCA2 +460-+617 bp and +732-+942 bp, were also found to be conserved within canine BRCA2 intron 1. Thus, we expected these two conserved DNA sequences to have a silencer element. Because canine BRCA2 splicing variant II, which did not have 3′-terminal 95 bp of intron 1, still suppressed the effect of CMV promoter activity, thus the sequence around +460-+841 bp in human BRCA2 (corresponding +413-+787 bp in canine BRCA2) should also be associated with silencer elements. To avoid translational effects, the human BRCA2 intron 1 region without promoter sequence (-187-+310 bp) was inserted upstream of human BRCA2 promoter (Fig 6 A). As expected, +443-843 bp region in human BRCA2 decreased the promoter activity similar to human BRCA2 intron 1 region (+311-942 bp). Next, we attempted to identify where the silencer element was located within the 500 bp sequence. Sequential deletion mutants in 5′ and 3′ ends were compared with the promoter activity (Fig 6 B and C). The +743-+842 bp region contained a cis-regulatory silencer element (Fig 6 C). The region of +443-+542 bp also contained a silencer, as the presence of this region showed significantly reduced promoter activity than that of the +543-+842 bp region (Fig 6 C) and the +443-+542 bp region tended to reduce the promoter activity but it was not significant (Fig 6 B; p=0.0295, by Holm’s method; p value < 0.0125 was considered significant.). Unexpectedly, the region around +643-+742 bp harbored an enhancer element, and the +543-+642 bp region contained a silencer element (compare Figure 6 C to Figure 6 B). Thus, we speculated that cis-regulatory elements could be separated in some constructs. To study this region in detail, three distinct constructs were tested (Fig 6 D). As we expected, the +593-+692 bp region contained a silencer element. However, surprisingly, the +543-+642 bp region acted as an enhancer element. To identify the enhancer elements around +742 bp in intron 1, we also analyzed the +693-+792 bp region (Fig 6 E). This region showed slightly decreased BRCA2 promoter activity. Thus, we concluded that the silencer and enhancer elements were too close to be separated, and the effect of the silencer element in this region was stronger than the enhancer element.