Methylation levels upstream of the XL exon are differentially affected according to underlying defects in patients with PHP1B. To analyze the GNAS methylation status of patients with PHP1B (n = 31 in total) with a variety of defined genetic alterations (n = 20 in total) (Figure 2A), genetically undefined sporadic cases (n = 11), and unaffected controls (n = 21), we first utilized a commercially available MS-MLPA kit, which has 3 probes in the region between AS exon 1 and the XL exon, hereafter referred to as AS 256, 166, and 320 probes based on the amplicon length (Figure 2B and Supplemental Table 1; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.177190DS1). Consistent with previous findings (4, 5), methylation levels at the A/B DMR were significantly lower in all PHP1B cases than in unaffected controls (Supplemental Figure 1). Among patients with PHP1B, sporadic cases with undefined defects, UPDpat, and patients with a deletion involving NESP55-AS exons 3/4 region showed broad methylation defects, including hypomethylation at AS and XL (Supplemental Table 1), consistent with previous findings (4, 5, 10, 16). The remaining patients with PHP1B showed normal methylation levels at all MS-MLPA probes, except for 1 probe, the “320 probe,” which showed varying degrees of methylation (Figure 2, B–E, and Supplemental Table 1). The 320 probe is located in the region between AS and XL DMRs, slightly centromeric (upstream) of the XL exon (Figure 2B). Among patients with PHP1B who did not show GNAS broad methylation defects, the 320 probe methylation levels were significantly lower than normal in patients with maternal STX16 deletion and a chromosomal duplication comprising the maternal NESP55-AS exon 1, which does not involve the region between the AS exon 1 and the XL exon (Figure 2E). On the other hand, the methylation levels did not differ significantly from normal in patients with PHP1B with 2 other defined genetic causes, a retrotransposon insertion telomeric (downstream) of the XL exon and a chromosomal inversion involving A/B and all Gsα exons with a centromeric breakpoint close to the retrotransposon insertion (but not involving the XL exon) (Figure 2E). These results suggest that, on top of conventionally analyzed GNAS methylation patterns, methylation levels at the 320 probe could have diagnostic potential for narrowing the location of the PHP1B-causing genetic mutation.

Differential effects of PHP1B genetic alterations on MS-MLPA probes. (A) Schematic representations of underlying genetic alterations in PHP1B patient samples studied. Red and white circles depict methylated and unmethylated DMRs, respectively. (B) A UCSC genome browser track showing the chromosomal locations of MS-MLPA probes in the AS-XL region. The 256, the 166, and the 320 probes are the 3 probes designed in the AS DMR region. (C–E) MS-MLPA results of patients with PHP1B (n = 31) and unaffected controls (n = 21). Methylation levels at the 256 probe (C), the 166 probe (D), and the 320 probe (E) are shown. In E, values in the STX16 deletion group were significantly higher than in the sporadic-c group (P = 0.0003). Intergroup comparisons were performed by 1-way ANOVA with post hoc Dunnett multiple comparison test. ****P < 0.0001. Sporadic-c, sporadic cases with complete methylation defects; Sporadic-i, sporadic cases with incomplete methylation defects; UPDpat, paternal uniparental disomy of chromosomal 20; NESP55-AS3/4 del, maternal deletion of NESP55-AS exons 3/4 region; STX16 del, maternal STX16 deletion; NESP55-AS1 duplication, duplication comprising the maternal NESP55-AS exon 1 (excluding the region between the AS exon 1 and the XL exon). Retrotransposon insertion idicates retrotransposon insertion telomeric (downstream) of the maternal XL exon. Inversion indicates maternal inversion involving A/B and all Gsα exons with a centromeric (upstream) breakpoint between XL and A/B.

The MS-MLPA probe of diagnostic potential reflects AS2 methylation levels. Since the 320 probe is located ~200 bp centromeric (upstream) of the AS2 DMR (Figure 3A), we measured AS2 methylation levels using methylation-sensitive restriction enzyme quantitative PCR (MSRE-qPCR) in a subset of the cohort analyzed by MS-MLPA (19 unaffected controls and 12 patients with PHP1B with a variety of underlying genetic alterations or undetermined causes; Supplemental Table 1). The methylation level at the AS2 DMR in unaffected controls was highly variable (mean ± SEM, 27.5% ± 2.1%) but was, on average, lower than the expected 50% (Figure 3B). AS2 methylation was almost completely lost in patients with PHP1B who showed broad GNAS methylation defects (Supplemental Table 1) and in those with a maternal STX16 deletion or a chromosomal duplication comprising NESP55-AS exon 1 (Figure 3B and Supplemental Table 1). On the other hand, AS2 methylation levels were close to normal levels in patients with PHP1B with a retrotransposon insertion telomeric (downstream) of the XL exon or a chromosomal inversion involving A/B and all Gsα exons (Figure 3B and Supplemental Table 1). Based on the similarity of methylation patterns at the 320 probe and the AS2 DMR, we compared methylation levels at the AS2 DMR and the 320 probe. AS2 methylation levels were significantly correlated with methylation levels at the 320 probe but not with other nearby probes in unaffected control (Figure 3, C–E). A similar correlation between AS2 methylation and the 320 probe was also observed in all (unaffected + PHP1B) samples (Figure 3F). These results indicate that methylation levels at the 320 probe and the AS2 DMR are equivalently regulated and are affected only by specific genetic causes underlying PHP1B.

Diagnostic relevance of AS2 methylation levels and their correlation with the methylation levels at MS-MLPA probes in the AS region (A) A UCSC genome browser track showing the chromosomal locations of AS2 MSRE-qPCR amplicons, MS-MLPA probes, XL exon, and AS exon 1. (B) AS2 methylation levels measured by MSRE-qPCR in unaffected controls (n = 19) and various patients with PHP1B (n = 12). Intergroup comparisons were performed by 1-way ANOVA with post hoc Dunnett multiple comparison test. *P < 0.05, **P < 0.01. (C–F) Correlation between AS2 methylation levels measured by MSRE-qPCR and those at 3 MS-MLPA probes. Correlation of methylation levels between AS2 and the 256 (C), the 166 (D), and the 320 (E) probes in unaffected controls (n = 17). Correlation of methylation levels between AS2 and the 320 probes in all available samples (n = 29), including patients with PHP1B and unaffected controls (F). The Pearson correlation coefficients and the p values are shown.

AS2 methylation depends on both STX16- and NESP-ICRs. Based on the result that AS2 methylation levels are almost entirely lost in patients with PHP1B with deletions in either the STX16 or NESP55-AS exons 3/4 regions (Figure 3B), we hypothesized that these regions are involved in a previously uncharacterized mechanism regulating AS2 methylation. Two ICRs, NESP-ICR and STX16-ICR, are required for A/B methylation on the maternal allele in an early embryonic period and possibly during oogenesis (18, 21). Accordingly, we used hESCs with either STX16-ICR or NESP-ICR deletion (18) to test which ICR affects AS2 methylation levels. WT hESCs showed lower AS2 methylation levels (~3.6%) by MSRE-qPCR compared with those observed in the leukocyte DNA from unaffected controls (3.73%–46.7%) (Figure 3B and Figure 4, A–C). Remarkably, AS2 methylation levels were significantly lower in the absence of either the STX16- or the NESP-ICR, specifically on the maternal allele but not on the paternal allele (Figure 4, B and C). The methylation levels at the flanking AS1 and XL DMRs were not reduced, as we have shown previously (18). These findings demonstrate that STX16- and NESP-ICRs are indispensable for AS2 methylation in an early embryonic period and, furthermore, suggest that STX16 enhancer–driven NESP55 transcription specifically affects the AS2 DMR within the AS-XL region on the maternal allele.

AS2 methylation levels in hESCs with GNAS ICR deletions. (A) A UCSC genome browser track showing the chromosomal locations of MSRE-qPCR amplicons (AS1, AS2, and XL) used for methylation analysis of hESCs. The AS2 DMR was analyzed in this study, and surrounding DMRs, AS1 and XL, were analyzed in our previous study (18). Locations of MS-MLPA probes and XL exon and AS exon 1 are also shown. (B and C) Methylation levels at the AS2 DMR in hESCs analyzed by MSRE-qPCR. Each dot represents an independent hESC clone. Results in wild-type (WT) hESCs and STX16-ICR maternally (ΔMat, n = 4) or paternally (ΔPat, n = 3) deleted hESC clones are shown (B). Results in WT hESCs and NESP-ICR maternally (ΔMat, n = 4) or paternally (ΔPat, n = 3) deleted hESC clones are shown (C). WT vs ΔPat or ΔMat values were compared using the one-sample t-tests. ***P < 0.001, ****P < 0.0001.

Retrotransposon insertion attenuates transcription. AS2 methylation was not reduced in patients with PHP1B with a retrotransposon insertion or a chromosomal inversion, as opposed to those with STX16 deletions (Figure 2E and Figure 3B). Given that both the retrotransposon and the chromosomal inversion involve the maternal GNAS region downstream (telomeric) of the AS2 DMR, we hypothesized that AS2 methylation, unlike A/B methylation, was preserved in these cases because the NESP55 transcript was truncated between AS2 and A/B. To test this hypothesis, we focused on 1 of our PHP1B kindreds (family number 208, including 6 patients in Supplemental Table 1) with a retrotransposon insertion between AS2 and A/B, which is 1 of 2 such familial cases described to date (12, 14) (Figure 5A). Affected patients in both kindreds showed equivalent methylation patterns — i.e., A/B DMR was variably hypomethylated, but AS2 DMR was not hypomethylated (Figure 5A) (12, 14). The inserted sequences of both kindreds share substantial homology (~93% identity) over a ~600 bp region, followed by several tandem consensus polyadenylation signals at the telomeric end (Figure 5B and Supplemental Figure 2). To examine if these sequences might prematurely truncate the NESP55 transcript, we cloned the polyadenylation signals and the flanking sequences (from kindred #1 in Figure 5A) into luciferase reporter plasmids driven by the NESP55 promoter with STX16 enhancer (18) and tested its effect in hESCs. Remarkably, including these patient-derived sequences suppressed the luciferase activity driven by the NESP55 promoter. This effect was significantly more profound when inserted in the same orientation as present in the patients’ genomes compared with the inverted orientation (Figure 5C). Furthermore, we generated various reporter constructs with truncated insertions to identify the critical portion that blunts transcription (Figure 5D). The middle segment alone, including the tandemly repeated polyadenylation signal, nearly abrogated the STX16 enhancer/NESP55 promoter reporter activity. These results collectively support our hypothesis that the presence of the homologous portion in the inserted retrotransposon efficiently suppresses read-through transcription. Although the tandem polyadenylation signals in the retrotransposon likely play a critical role in impeding NESP55 transcription, additional surrounding sequences, especially the centromeric (upstream) portion, may also contribute to this effect.

The effect of retrotransposon sequences on the passing-through transcription (A) A UCSC genome browser track showing the locations of retrotransposon insertions identified in 2 kindreds (#1 and #2; refs. 12, 14). Locations of the AS2 MSRE-qPCR amplicon and the 320 probe are also shown. (B) A schematic representation of the location of a highly homologous sequence in the retrotransposon identified in kindreds #1 and #2. The red arrow indicates the location of the tandemly repeated polyadenylation signal. Blue and yellow arrows indicate surrounding cloned regions for the reporter assay. (C and D) Luciferase assays in hESCs. Forty-eight hours following the transfection of each reporter plasmid in WT hESCs, firefly counts were measured and normalized using Renilla counts. The polyadenylation signal portion with surrounding sequences derived from kindred #1 was cloned into the STX16-ICR/NESP55 promoter-driven firefly luciferase plasmid (n = 4). Rightward and leftward arrows indicate sense and antisense orientation, respectively. STX16-ICR, NESP55 promoter and STX16-ICR; STX16-ICR+SVA, NESP55 promoter and STX16-ICR with sense-oriented insertion of transposon sequence; STX16-ICR+SVAinv, NESP55 promoter and STX16-ICR with antisense-oriented insertion of transposon sequence (C). Inserted kindred #1–derived sequence used in C was truncated as indicated (TR1-TR3) (n = 3) (D). Intergroup comparisons were performed by 1-way ANOVA with post hoc Dunnett multiple comparison test. *P < 0.05, ****P < 0.0001.

Mechanistic categorization of patients with PHP1B based on GNAS methylation defect patterns. Based on the current findings, methylation levels at the AS2 DMR appear to reflect the locations of GNAS cis-regulatory mutations in PHP1B. Normal AS2 methylation is likely to be preserved only when read-through transcription from the NESP55 promoter is intact. Therefore, by combining AS2 methylation levels, which can be inferred from those measured by the MS-MLPA 320 probe, with those at the conventionally analyzed DMRs, patients with PHP1B can be categorized to reflect the location of cis-regulatory defects affecting the GNAS locus (Figure 6). Category 1 represents patients with broad methylation defects — i.e., a loss of methylation at all maternally methylated DMRs, including AS2, and a gain of methylation at the NESP55 DMR. Sporadic patients with PHP1B and those with maternal AS exon 3/4 deletion belong to this category. Category 2 comprises patients with a loss of methylation restricted to the A/B and the AS2 DMRs that is caused by impaired NESP55 transcription centromeric (upstream) of the AS2 DMR. This group includes patients with maternal STX16 deletions and those with a maternal duplication of a region that extends from upstream of the NESP55 exon to downstream of AS exon 1 (excluding the AS2 DMR). Although we had no DNA samples from patients with deletions restricted to maternal NESP55 exon, their GNAS methylation pattern was reported to be similar to that of patients with STX16 deletions (15, 22), suggesting that these patients should fall into category 2. Category 3 patients have bona fide isolated A/B hypomethylation with normal methylation status at the remaining DMRs, including the AS2 DMR. NESP55 transcription through the AS2 DMR should be intact in those cases. Therefore, this category includes maternal retrotransposon insertions telomeric (downstream) of the AS2 DMR and maternal chromosomal inversions with the centromeric breakpoint between XL and A/B.

PHP1B categories based on pathogenic mechanisms and corresponding GNAS methylation patterns. Patients with PHP1B are classified into 3 categories based on GNAS methylation patterns. Category 1 is characterized by broad methylation defects caused by undetermined underlying causes (sporadic) with the exception of paternal uniparental disomy of chromosome 20 (UPDpat) and deletions comprising the NESP55-AS exons 3/4 region. Category 2 cases show a loss of methylation at AS2 and A/B while AS and XL methylation levels are preserved. Transcriptional attenuation of NESP55 centromeric (upstream) of the AS2 DMR causes this pattern, in which maternal STX16 deletions are the most frequent cause. Category 3 is characterized by isolated A/B loss of methylation with preserved AS2 methylation levels, suggesting that NESP55 transcription is blunted telomeric (downstream) of the AS2 DMR. Asterisk indicates apparent hypomethylation due to copy number gain.