The complexity of the promoters for SMAD4
has been under-appreciated until now. Minami et al
) evaluated 1285 bases immediately upstream of the 5′-UTR of SMAD4
plus 45 in the 5′-UTR (comparable to Promoter D in this study) and described a peak luciferase activity of 38 light units compared to the control of 2.2 light units. However, this region lacked typical promoter region characteristics, such as high CG content (CpG islands) or a TATA-box, but did have some TATA-like structures (TAAAAT) and other potential transcriptional binding sites. Current evidence suggests that low CG content promoters are more likely to be cell specific, while CG rich are more likely to be ubiquitously expressed (25
Zhou et al
) found that six endometrial carcinoma specimens had LOH of chromosome 18q21 markers and altered transcription of SMAD4
and sequencing for mutations of a part of promoter D revealed substitutions in two patients. Functional assays (chloramphenicol acetyltransferase) of promoter activity revealed these substitutions led to significantly reduced activity relative to the wild-type sequence. We found that promoter D had minimal activity in the four different cell lines we tested, despite the inclusion of the 5′-UTR and part of coding exon 1 where the majority of predicted TFBS were concentrated and similar results were obtained when the 5′-UTR and this portion of exon 1 were left out (data not shown). The lack of RNA polymerase II binding sites and DNaseI sites around promoter D support the notion that this is not a very active site of transcription, at least in the cell lines examined.
Promoter C was found to have significant luciferase activity in all four cell lines evaluated and this region has sparked interest from investigators in the past. This began with Roth et al
) who screened an overlapping region of 700
bp from position 46
540 to 46
239 for methylation in colorectal cancer specimens. They selected this region because it was CG-rich and an unpublished manuscript by Hagiwara et al
. apparently examined the sequence immediately upstream from a newly discovered NC exon. The region they looked at is primarily within and downstream of NC exon 4 (which starts at position 46
582 and ends at 46
991). However, it did contain 42 bases of promoter C upstream from the TSS and while these bases were found to have negligible luciferase activity in our deletion constructs, the sequence examined would have contained four of six phylogenetically conserved sites. Roth et al
. did not find methylation in 42 colorectal cancer samples, but these studies may have been inconclusive since this promoter appears to include a larger sequence upstream, as evident by our luciferase models.
Onwuegbusi et al
) also screened the segment studied by Roth et al
. and found that 70% of esophageal adenocarcinomas had methylation. The same region was evaluated in prostate cancer specimens and although no evidence of methylation was found in benign prostatic hypertrophy samples, 45% of prostate cancers had methylation. Furthermore, they found that patients with lymph node metastasis had a higher incidence of methylation (63%). They also looked for mutations, but none were identified within the 40 bases of promoter C that were screened, or in the sequences of NC exon 4 (20
). Again, these studies examined primarily NC exon 4 and the intron downstream and therefore, the consequence of finding methylated CpGs here upon tumor formation is unclear.
Ando et al
. screened sequences between 46
524 and 46
769 for methylation in CRC specimens and found none. The area they studied is mostly within NC exon 4, with only 58 bases upstream of the TSS included (26
). Wang et al
) examined a larger area of promoter C by looking at a region of 200 bases upstream from position 46
611, which was believed to be the TSS of NC exon 4 (although current evidence suggests the TSS is now at position 46
581). They screened gastric carcinoma specimens and found that 4 specimens out of 75 (5%) had methylation that was associated with decreased expression of SMAD4. Kloth et al
) looked at an even larger area of promoter C in cervical cancer specimens, by screening for methylation including up to 270 bases upstream from the TSS, to 155 bases downstream and they found no evidence of methylation. This study did not include the other 255 bases upstream to position 46
882. It remains to be determined whether the incidence of methylation in gastric and cervical cancer patients would have been higher if these additional bases had been screened. It should also be noted that prior to the current study, there have been no luciferase studies published that confirmed this region had promoter activity.
In the cell lines tested, Promoters A and C were the most functionally active. Promoter C has a greater number of mRNA isoforms that might potentially be regulated by it, a higher CpG content relative to promoter A and showed luciferase activity in all four cell lines, while promoter A had no promoter activity in a normal colon fibroblast cell line (CRL-1459). ChIP-Seq and DNaseI data revealed a greater number of sites at promoter C, but the most abundant RNA isoform found by hybridization was one predicted to be under the control of promoter A. In contrast, none of these studies suggested an important role for either promoter B or D in the cell lines we tested. However, different promoters may play distinct roles in various tissues, during stages of development, or physiologic conditions. Presumably this activity is influenced by the differing context and abundance of transcription factors which are present in each circumstance. Analyzing promoter C with MatInspector (www.genomatix.de/products/MatInspector
) revealed several potentially important TFBS that could be involved in regulation of SMAD4
. From the −500
bp to the −414
bp construct, there was a >45–50% drop in luciferase activity and this region has one zinc finger homeodomain transcription factor binding site (ZFHX), an AP-2 site, an SP-1 site, several C-abl DNA binding sites (CABL), ZBPF sites and a TATA box. This area was not examined in all of the previous studies attempting to screen the SMAD4
promoter for methylation. Between the −500
bp and −34
bp constructs, there was a loss of >85–95% of the luciferase activity in both CRL-1459 and HEK-293 cell lines. Between the −234
bp and −34
bp deletion constructs, there is an additional SP-1 site, six ZBPF, two AP-2 sites, a CABL site and a possible core promoter element for RNA pol II transcription binding site for TATA-less promoters (). Furthermore, phylogenetic data show that there are multiple TFBS that are conserved between species, which is not only a testament to the importance of these regions, but also provides insight into the transcriptional regulatory elements that might play a role in the expression of this important tumor suppressor gene. We focused further attention on the sequence of Promoter C because more of the mRNA isoforms are likely regulated by this region and the deletion seen in one JP patient affected this region. Further analysis by SDM showed how important the TATA box, ZBPF and the SP-1 sites could be in influencing the transcription of SMAD4
, for when these sites were mutated the promoter activity was significantly diminished ().
Although screening of our 65 JP probands revealed no germline mutations within Promoter C, we did find two JP patients with germline deletions affecting this region, one of which involved only promoter C and NC exon 4 (and none of the coding exons). Aretz et al
) also found four JP patients with deletion of these four MLPA probes, but they also had deletions involving all the coding exons as well. These data suggest that promoter alterations play a role in the genesis of JP, as recently reported for BMPR1A
), and therefore, further evaluation of promoter A in JP patients will be of interest. Whether epigenetic inactivation of the normal copy of SMAD4
leads to polyp formation is another important question to be examined in JP patients. Now that these promoter regions have been characterized, follow-up studies in colon, gastric, cervical, pancreatic and other sporadic cancers will be imperative to define their importance in tumorigenesis.