Methylation analysis of GNAS1. (a) The normal allele-specific methylation and expression patterns of the four known first exons of GNAS1, which splice onto exon 2 to produce transcripts encoding NESP55, XLαs, a transcript of unknown function (exon 1A), and G_sα (exon 1). Horizontal arrows indicate transcriptionally active promoters. The imprinting of NESP55 and XLαs have been defined previously (8, 9, 11). Exon 1 is probably paternally imprinted in some tissues, indicated by the dashed arrow. NESP55 protein is unrelated to G_sα, and its entire coding region is located within its first exon. In contrast, XLαs and G_sα proteins have identical COOH-terminal domains (encoded by exons 2–13), while their unique NH₂-terminal domains are encoded within their respective first exons. Exon 1A does not have a translational start site, and its transcripts are likely to be untranslated. The imprinting of the exon 1A region has been defined previously in mice (10). (b) Southern analyses of leukocyte genomic DNA from a normal subject (N), PHPIB patients 1, 3, and 7, the mother of patient 1 (1M), and the father of patient 1 (1F), using genomic DNA probes from the NESP55 (left panel; ref. 9), XLαs (middle panel; ref. 8), and exon 1A (right panel) regions. Above are the relevant restriction maps depicting each upstream exon as a black box and the position of the probes below. Patient 1 is abnormally methylated in all three regions, patient 3 is abnormally methylated only in the exon 1A region, and patient 7 is abnormally methylated in the NESP55 and exon 1A, but not the XLαs, regions. Bg, BglII; S, SacII; F, FspI, N, NgoMIV; A, AscI; Bs, BssHII; P, PstI; Mat, maternal allele; Pat, paternal allele.

The GNAS1 exon 1A region is methylated only on the maternal allele in normal subjects. (a) Genomic DNA from two normal subjects heterozygous for a 5-bp insertion polymorphism in GNAS1 exon 1A (N1 and N2) was digested with PstI and AscI, and fractions containing the 2.8-kb methylated (Meth) and 1.6-kb unmethylated (Unmeth) fragments were purified from agarose gels. The fractions were genotyped for the exon 1A polymorphism using PCR and direct sequencing and for parental assignment were run next to the same reaction performed on genomic DNA from a homozygous parent (mother for N1, father for N2). The presence of the polymorphic 5-bp insertion is indicated with brackets (although there is compression on these gels, bisulfite-modified genomic sequencing, shown in b, which removed these compressions, confirmed that the insertion is in fact 5-bp long). For both N1 and N2, the genotypes of methylated and unmethylated fragments corresponded to those of the mother and father, respectively. Identical results were also obtained for N3, the sister of N2 (data not shown). (b) Results of bisulfite-modified genomic sequencing of N3, showing the methylated maternal allele on the left and the unmethylated paternal allele (which includes the 5-bp insertion polymorphism) on the right. In the paternal allele, all Cs are unmethylated and are therefore converted to Ts. In the maternal allele all Cs within CpG dinucleotides (eight shown in figure; 25 total in sequencing reaction) are methylated and therefore remain as Cs (Cs that are not within CpG dinucleotides are unmethylated and are therefore converted to Ts).

RT-PCR analysis of blood RNA. (a) RT-PCR was performed on total RNA isolated from blood of two normal subjects (N1 and N2) and two PHPIB patients (patients 1 and 3) using exon 1A-specific upstream and exon 2-specific downstream primers. Direct sequencing of RT-PCR products is shown above with the position of the polymorphic 5-bp insertion indicated by brackets. For N1 and N2, only RT-PCR products with the 5-bp insertion are identified (which for both N1 and N2 is the paternal allele; see Figure 2a). In contrast, RT-PCR products both with and without the polymorphic 5-bp insertion are amplified from patients 1 and 3, indicating biallelic expression of exon 1A mRNAs in these patients. Below each sequence are results of genotyping of subcloned products. For patient 1, clones with the 5-bp insertion are derived from the paternal allele, whereas for patient 3 these clones are derived from the maternal allele. (b) RT-PCR was performed on total RNA isolated from blood of normal subjects (N) and four PHPIB patients (patients 1, 5, 8, and 12) using NESP55-specific primers (above) or β-actin–specific primers (below). In patients 1, 5, and 8 both GNAS1 alleles are methylated in the NESP55 upstream region, whereas in patient 12 only one allele is methylated in this region (see Table 1). The presence or absence of enzyme in the RT reaction is shown above each lane.

Genotyping of 20q polymorphic markers in PHPIB patient 1 and her parents. The relative position of each marker in megabases is indicated on the left (http://cedar.genetics.soton.ac.uk/public_html/) from centromere (Cen) to telomere (Tel). Results and interpretation for each marker are shown on the right. The GNAS1 polymorphism is a 5-bp insertion within exon 1A (+ and – indicates presence or absence of the insertion, respectively). NI, not informative; R/O UPD, rule out uniparental disomy; R/O UPID, rule out uniparental isodisomy; R/O UPHD, rule out uniparental heterodisomy. (UPID is inheritance of two copies of the same chromosome from a single parent; UPHD is inheritance of one copy each of both chromosomes from a single parent).

Potential mechanisms for tissue-specific imprinting of G_sα. (a) The exon 1A DMR may contain a binding site for a tissue-specific repressor (Rep) that binds to the paternal allele (Pat) in renal proximal tubules, leading to silencing of the G_sα (exon 1) promoter. Binding to the maternal allele (Mat) is prevented by methylation of the binding site. The model predicts that the repressor is expressed in renal proximal tubules, but not in other tissues where G_sα is biallelically expressed. (b) The exon 1A DMR may contain one or more boundary elements (B) that block activation of the G_sα promoter by an upstream enhancer (E) in the paternal allele, but do not block enhancer-promoter interactions in the maternal allele due to methylation. This is the mechanism by which Igf2 is imprinted (36, 37). This model predicts that expression of G_sα in other tissues is not dependent on the upstream enhancer.