Deletion hotspots in AMACR promoter CpG island are cis-regulatory elements controlling the gene expression in the colon

Abstract

Alpha-methylacyl-coenzyme A racemase (AMACR) regulates peroxisomal beta-oxidation of phytol-derived, branched-chain fatty acids from red meat and dairy products -- suspected risk factors for colon carcinoma (CCa). AMACR was first found overexpressed in prostate cancer but not in benign glands and is now an established diagnostic marker for prostate cancer. Aberrant expression of AMACR was recently reported in Cca; however, little is known about how this gene is abnormally activated in cancer. By using a panel of immunostained-laser-capture-microdissected clinical samples comprising the entire colon adenoma-carcinoma sequence, we show that deregulation of AMACR during colon carcinogenesis involves two nonrandom events, resulting in the mutually exclusive existence of double-deletion at CG3 and CG10 and deletion of CG12-16 in a newly identified CpG island within the core promoter of AMACR. The double-deletion at CG3 and CG10 was found to be a somatic lesion. It existed in histologically normal colonic glands and tubular adenomas with low AMACR expression and was absent in villous adenomas and all CCas expressing variable levels of AMACR. In contrast, deletion of CG12-16 was shown to be a constitutional allele with a frequency of 43% in a general population. Its prevalence reached 89% in moderately differentiated CCas strongly expressing AMACR but only existed at 14% in poorly differentiated CCas expressing little or no AMACR. The DNA sequences housing these deletions were found to be putative cis-regulatory elements for Sp1 at CG3 and CG10, and ZNF202 at CG12-16. Chromatin immunoprecipitation, siRNA knockdown, gel shift assay, ectopic expression, and promoter analyses supported the regulation by Sp1 and ZNF202 of AMACR gene expression in an opposite manner. Our findings identified key in vivo events and novel transcription factors responsible for AMACR regulation in CCas and suggested these AMACR deletions may have diagnostic/prognostic value for colon carcinogenesis.

Figure 1. Detection of AMACR expression level by immunostaining.

A–H: the typical AMACR immunostaining found in normal and neoplastic colon sections from our case materials. I: AMACR expression scores of these foci, representing the above groups, depicted in a scatter plot. One-way analysis of variance, followed by Tukey's HSD post hoc test, indicated the significant difference among different groups (p<0.0001). The normal crypt group served as a reference. Foci with normal cryptal glands had very low AMACR expression (score: 0.22±0.22), with 8 of 9 foci scored negative. Foci with normal apical surface epithelium had mildly elevated expression (1.1±0.55) that was not statistically different from that of the former group (p = 0.67). Expression at foci harboring TA glands with a mild degree of dysplasia (0.22±0.15, p = 1.00) was not statistically different from that in normal cryptal glands. However, VAs had elevated expression (2.3±0.62, p = 0.007) comparable to that of well- (2.8±0.60, p = 0.001) and moderately differentiated carcinomas (2.7±0.49, p = 0.002); the three groups (open ellipses) have higher AMACR expression scores than normal and TAs. In marked contrast, AMACR expression scores in poorly differentiated cancers were low (0.14±0.14, p = 1.00), with 6 of 7 foci devoid of AMACR immunostaining.

Figure 2. The organization of AMACR gene 5′-flanking region.

A: The location of the CpG island upstream translation start site (designated as +1). Individual CG sites are indicated as red vertical lines and numbered from 1 to 18. The 222-bp nested PCR-amplified CGI is illustrated. B: Partial exon 1 and the promoter sequence encompassing the CGI. The first exon is indicated by a bent arrow (TSS). Predicated transcription factor binding sites of Sp1 and ZNF202, together with the CCAAT enhancer binding site, are underlined. The locations of two pairs of primers for bisulfite sequencing PCR are blocked with different colors. The ChIP assay-amplified region is marked with brackets. The DNA upstream -4 (bent arrow, promoter clone) was cloned for promoter analysis. Two direct repeats of up to 7 nt (5′-GGCGCCG-3′) that may related to the deletion caused by slipped-strand mispairing are marked by dotted lines. Primers PolyF/PolyR for polymorphism study are marked with the arrows. The wild type (WT) probe for gel shift assay targeting on putative ZNF202 binding site is boxed. Two short putative ZNF202 core sequences identified by MatInspector were highlighted in red.

Figure 3. Bisulfite sequencing analysis of iLCM.

The results were summarized from the status of 4302 CG sites from a total of 239 alleles in the entire set of 55 microdissected samples. A: Cluster analyses of the deletion and methylation to establish clusters of CG sites based on the entire sequencing data set. The average linkage was employed for hierarchical clustering (sites 1–18). The absolute number of co-occurrences of different CG deletions was used as the similarity measure. The higher the number of clones with two specific CG deletions, the closer they are in the dendrogram. The hotspots were restricted to CG3, 10, and 12-16. B: Typical bisulfite sequencing results of AMACR CGI with deletions highlighted in grey. The sequences of AMACR promoter variants have been deposited in Genbank and are described in Materials and Methods.

Figure 4. Deletion hotspots in AMACR promoter CGI are the cis-acting elements.

A: Promoter assays showed that AMACR599 with the CGI in it had promoter activity comparable to that of AMACR1818, suggesting that the 599 bp region is critical for the gene regulation. Thus, we selected AMACR599 for further investigation. The promoter activity was normalized as relative light units. B: The location of the deletion hotspots in AMACR promoter. Various sequence variants were compared with the wild-type promoter. A previously identified CCAAT box is illustrated. C: Compared with the wild-type AMACR599, deletion of CCAAT enhancer element at CG5, or in combination with other deletion hotspots at CG3, 10 and 12-16, significantly reduced the promoter activity (58∼67%, p<0.0001, one-way analysis of variance, followed by Tukey's HSD post hoc test). No significant differences among the CCAAT deletion groups (p = 0.014 to 1) were observed. When the CCAAT enhancer was maintained intact, the single deletion at CG12-16 or in combination with deletion hotspots at CG3 and 10 resulted in an increase in the promoter activity by 83−105% (p<0.0001) but no significant difference among the deletion groups (p = 0.060 to 1). Further, when the CCAAT enhancer was maintained and CG12-16 was intact, deletion of either CG3 or CG10 did not change the promoter activity significantly (p = 0.26 and 0.69, respectively). In contrast, double-deletion at CG3 and 10 decreased the promoter activity by 69% (p<0.001).

Figure 5. Transcription factors Sp1 is involved in AMACR gene regulation in HCT 116 cells.

Putative Sp1 binding site at CG3 and 10 were identified. A: ChIP assay with Sp1 antibody targeting AMACR CGI (Figure 2B). A PCR signal was detected in the Sp1 antibody ChIP with genomic DNA and normal IgG-immunoprecipitated DNA as the PCR input and negative control, respectively (Figure 5A, top panel). As a ChIP negative control, amplification of a region in the last exon of AMACR gene distant to the putative Sp1 sites was included in the experiment. Only the DNA input showed the amplification (Figure 5A, lower panel). B: siRNA-mediated Sp1 knockdown decreased the AMACR transcript level. Real-time RT-PCR demonstrated that the first-round siSp1 decreased the Sp1 transcript level 48% (p<0.001). With the second-round siSp1, the Sp1 transcript level further decreased 64% (p<0.001). In parallel, the first-round siSp1 resulted in little change in AMACR mRNA level (p = 0.66). Notably, the second-round siSp1 decreased the AMACR transcript level 53% (p = 0.002). In the negative control experiments, the same set of cDNA was used and siRNA knockdown of Sp1 did not affect GUSB and PP1A gene expression (data not shown).

Figure 6. ZNF202 is involved in AMACR gene regulation in HCT 116 cells.

A: The ³²P labeled wild type (WT) probe corresponds to a sequence containing CG12-16 and its flanking regions (Table 5). Gel shift assays showed a single, specific shifted band (arrow) whose signal intensity could be impeded by co-incubation with 100× excess cold WT probe or a ZNF202 consensus sequence (GnT) but not by 100× excess mutated (Mut) or CG12-16 deleted (Del) ODNs. The other shifted bands represent unknown protein-DNA complexes formation. B, Left: Using labeled WT as a probe three major shifted bands were identified. Signal intensities of these bands could not be reduced by co-incubation with excess cold Del ODN that has deletion of CG12-16. Right: Using labeled Del as probe, one major band that differs from those observed with the labeled WT was identified. Its signal intensity was not diminished by co-incubation with excess cold WT. C: Ectopic expression of ZNF202 decreased AMACR promoter activity and mRNA level in a dose-dependent manner. Co-transfection of the ZNF202 expression plasmid with AMACR599 (10 ng) decreased the promoter activity (p = 0.007); in parallel, ectopic expression of ZNF202 (2 µg) decreased the level of AMACR mRNA (p = 0.009). The asterisks indicate a significant difference in the group compared with the control.