Analysis of Multiple Families With Single Individuals Affected by Pseudohypoparathyroidism Type Ib (PHP1B) Reveals Only One Novel Maternally Inherited GNAS Deletion

Abstract

Proximal tubular resistance to parathyroid hormone (PTH) resulting in hypocalcemia and hyperphosphatemia are preeminent abnormalities in pseudohypoparathyroidism type Ib (PHP1B), but resistance toward other hormones as well as variable features of Albright’s Hereditary Osteodystrophy (AHO) can occur also. Genomic DNA from PHP1B patients shows epigenetic changes at one or multiple differentially methylated regions (DMRs) within GNAS, the gene encoding Gαs and splice variants thereof. In the autosomal dominant disease variant, these methylation abnormalities are caused by deletions in STX16 or GNAS on the maternal allele. The molecular defect(s) leading to sporadic PHP1B (sporPHP1B) remains in most cases unknown and we therefore analyzed 60 sporPHP1B patients and available family members by microsatellite markers, single nucleotide polymorphisms (SNPs), multiplex ligation-dependent probe amplification (MLPA), and methylation-specific MLPA (MS-MLPA). All investigated cases revealed broad GNAS methylation changes, but no evidence for inheritance of two paternal chromosome 20q alleles. Some patients with partial epigenetic modifications in DNA from peripheral blood cells showed more complete GNAS methylation changes when testing their immortalized lymphoblastoid cells. Analysis of siblings and children of sporPHP1B patients provided no evidence for an abnormal mineral ion regulation and no changes in GNAS methylation. Only one patient revealed, based on MLPA and microsatellite analyses, evidence for an allelic loss, which resulted in the discovery of two adjacent, maternally inherited deletions (37,597 and 1427 bp, respectively) that remove the area between GNAS antisense exons 3 and 5, including exon NESP. Our findings thus emphasize that the region comprising antisense exons 3 and 4 is required for establishing all maternal GNAS methylation imprints. The genetic defect(s) leading in sporPHP1B to epigenetic GNAS changes and thus PTH-resistance remains unknown, but it seems unlikely that this disease variant is caused by heterozygous inherited or de novo mutations involving GNAS.

Introduction

Resistance toward parathyroid hormone (PTH), the most prominent biochemical abnormality encountered in the different variants of pseudohypoparathyroidism (PHP), is caused by reduction or lack of Gαs expression in the proximal renal tubules leading to reduced or inappropriately normal synthesis of 1,25(OH)2 vitamin D (1,25(OH)2D) despite often severe hypocalcemia.^(1–4) In addition, impaired PTH-stimulated downregulation of the sodium-dependent phosphate cotransporters, NPT2a and NPT2c, reduces urinary phosphate excretion thus leading to hyperphosphatemia. A similar reduction in Gαs protein levels can occur in other tissues, such as thyroid and pituitary,^(5–8) where resistance occurs toward thyroid-stimulating hormone (TSH) and growth hormone-releasing hormone (GHRH), respectively, two hormones that also mediate their actions through Gαs-coupled receptors.

Different forms of PHP type Ib (PHP1B) are associated with characteristic epigenetic changes at the GNAS locus. Some autosomal dominant forms of PHP1B (AD-PHP1B) are caused by deletions affecting maternal GNAS exons NESP and/or AS3-4; ie, disease variants that are associated with a loss-of-methylation (LOM) at GNAS exons A/B, AS, and XL.^(4,9,10) Much more frequently, AD-PHP1B is attributable to maternal deletions within STX16, the gene encoding syntaxin 16, and in one family the disorder is caused by a maternal deletion of GNAS exon NESP and a large portion of the adjacent centromeric intron^(11–13); these AD-PHP1B variants are all associated with LOM at the GNAS exon A/B alone. However, most PHP1B patients are affected by a sporadic form of the disease. Few of these cases are caused by paternal uniparental isodisomy or heterodisomy involving chromosome 20q (patUPD20q and patUHD20q, respectively),^(14–16) but the majority of sporPHP1B patients remains undefined at the molecular level. These individuals typically display similarly broad GNAS methylation changes as patients with maternal NESP and/or AS3-4 deletions, namely a loss of all maternal methylation imprints, as well as biallelic methylation at exon NESP. However, the epigenetic GNAS changes, particularly LOM at exon XL, can be incomplete,^(17–21) thus raising the possibility that at least some of these sporPHP1B patients lack an as yet unknown trans-acting factor that is required for maintaining or establishing all of the maternal GNAS methylation imprints.

Besides the mRNA encoding Gαs, the GNAS locus gives rise to additional sense and antisense transcripts that utilize alternative first exons and promoters. These include the A/B transcript, which may encode an amino-terminally truncated form of Gαs⁽²²⁾ and noncoding antisense transcripts (AS), as well as transcripts encoding the extra-large Gαs variant (XLαs) and a 55-kDa neuroendocrine secretory protein (NESP55).^(1–4) Studies in genetically manipulated mice have suggested that some of these transcripts contribute to the regulation of Gαs expression. For example, ablation of exon 1A (murine equivalent of exon A/B) on the paternal allele was shown to enhance Gαs expression sufficiently to improve or reverse some of the laboratory abnormalities observed in mice with a maternal mutation in Gnas exon 6 (Oed-Sml mouse) or with ablation of Gnas exon 1, respectively.^(23,24) Likewise, ablation of exon Xl on the paternal allele corrected early lethality, but not the laboratory abnormalities, in mice with a maternal deletion comprising Gnas exons Nesp and AS2-4.^(25,26) These findings could imply that the promoters giving rise to the mRNA encoding Gαs and other GNAS-derived messages, such as 1A and Xl, are directly or indirectly competing for a tissue-specific transcription factor involved in regulating Gαs expression. Alternatively, ablation of the paternal exons 1A and Xl as well as adjacent intronic sequences may disrupt interaction with a protein that binds preferentially to this nonmethylated GNAS region. Such a protein has been postulated to be a tissue-specific “silencer” that prevents a transcription factor from binding to the nonmethylated paternal allele thereby reducing Gαs synthesis.^{(2,4,23,24,27,28)} Alternatively, the putative protein may be an “enhancer” that is required for efficient Gαs transcription from the nonmethylated region of the paternal allele. This “enhancer” would have to be widely expressed, except for tissues where this signaling protein is derived predominantly from the maternal allele. Thus, expression of a “silencer” would most likely be restricted to selected tissues, such as proximal renal tubules, thyroid, and pituitary, whereas an “enhancer” would be lacking in tissues with predominantly maternal Gαs expression. Based on observations in different groups of patients, it appears plausible that the cellular levels of a transcriptional modifier of Gαs expression change over time as PTH-resistance in PHP1B and PHP1A patients usually does not become apparent until early childhood^(15,28,29) and some individuals with documented STX16 deletions appear to develop only mild or no hormonal resistance.^(29,30)

In order to help determine what kind of genetic defect might cause sporadic PHP1B, we have now studied 60 patients affected by this disorder, as well as available parents and siblings, by MLPA, MS-MLPA, and microsatellite marker analyses. Differences in GNAS methylation changes were observed when using genomic DNA from peripheral blood cells and immortalized lymphocytes (LCLs) that were available for some patients.

Patients and Methods

Patients and healthy family members

We investigated 60 sporPHP1B cases (37 females and 23 males), who had presented with PTH-resistant hypocalcemia and hyperphosphatemia, and with mild AHO features in some; 23 of these patients had been previously reported.^(29,31,32) Gender, age at diagnosis, and biochemical results for each patient are presented in Supplemental Table 1; all biochemical measurements had been performed in the hospital laboratories of the referring physicians. None of the available parents, siblings, and children showed abnormalities in calcium and phosphate homeostasis. Lymphoblastoid cells (LCLs) were generated from 6 PHP1B patients and 4 healthy controls, which were maintained as described.^(9,10) Genomic DNA was extracted from peripheral blood leukocytes or LCLs using standard protocols.

The study was approved by the Institutional Review Board of the Massachusetts General Hospital. Analyses were performed after obtaining informed consent from the patients or their parents.

MLPA of STX16 and GNAS

MLPA was performed using the SALSA MLPA kit ME031 GNAS (MRC-Holland, Amsterdam, The Netherlands) following the manufacturer’s instructions. PCR products were electrophoretically separated using the ABI3730xl Genetic Analyzer at the DNA Core Facility of the Massachusetts General Hospital.

GNAS methylation analysis

Epigenetic GNAS changes were assessed by Southern blot analysis, methylation-specific polymerase chain reaction (MS-PCR),^(11,33) or MS-MLPA, as described.⁽¹⁵⁾ The methylation status for each patient and most of their mothers is presented in Supplemental Table 2.

Analysis of microsatellite markers

To exclude a large deletion comprising GNAS, patUPD20q, or patUHD20q, respectively, several previously described microsatellite markers (D20S86, 907-rep2, 261P9-CA, 806M20-CA, 543J19-TTA, D20S171) were analyzed for the patients by the Center for Human Genetic Research of the Massachusetts General Hospital. When markers showed homozygosity for a PHP1B patient, maternal DNA was also analyzed; paternal DNA was investigated to exclude patUHD20q (Supplemental Table 3).

Analysis of SNPs for family 82

To determine the boundaries of the suspected deletion in patient 82/II-1, seven SNPs were analyzed to search for loss-of-heterozygosity (LOH). The PCRs to analyze these SNPs and to confirm the identified deletions were performed using QIAGEN Inc. (Valencia, CA, USA) Taq DNA polymerase following the manufacturer’s protocols; cycler program: denaturation at 94°C for 5 min followed by 35 cycles at 94°C for 1 min, 59°C for 1 min, and 72°C for 1 min, followed by an additional elongation step at 72°C for 10 min. PCR primers are listed in Supplemental Tables 4 and 5. The PCR products were purified using ExoSap-IT (Affymetrix) and sequenced at the DNA Core Facility of the Massachusetts General Hospital.

Statistical and data analyses

All data are presented as mean ± SE. Age-related reference values (RV, 95% confidence intervals) were obtained from http://cclnprod.cc.nih.gov/dlm/testguide.nsf/HRRAll?OpenForm&Count=1187#C. Serum phosphate levels, as well as the levels for calcium and magnesium were transformed into standard deviations scores (SDSs) by using the age-appropriate mean reference value (RVmean) and the standard deviation (SD) (where SD is RV/4) according to the formula: SDS = (level − RVmean)/SD. Differences between groups were calculated using unpaired t test. A two-tailed p value of <0.05 was considered statically significant. Pearson’s chi square test was used to assess the likelihood of autosomal dominant or recessive traits as causes of sporPHP1B.

Results

Laboratory findings in our cohort of sporPHP1B patients

Thirty-seven female and 23 male sporPHP1B patients were investigated, 23 of whom had been reported previously (see Supplemental Table 1). The age when PTH-resistance was first diagnosed varied considerably (2.9 to 28 years), but there was no statistically significant difference for the mean age at diagnosis for male and female patients (10.3 ± 1.2 versus 13.1 ± 1.4 years; p = 0.209). Average blood calcium levels tended to be lower in male than in female patients (6.33 ± 0.32 versus 6.66 ± 0.19 mg/dL; p = 0.342), but the SDSs, ie, calcium levels corrected for the age-appropriate mean, did not show any significant difference between both genders (−6.72 ± 0.69 for males and −5.92 ± 0.44 for females; p = 0.438). Serum phosphate levels were higher in male than in female patients (7.81 ± 0.40 versus 6.75 ± 0.28 mg/dL; p = 0.0315) and even the SDSs were significantly different, albeit marginally (+5.63 ± 0.66 for males and +4.22 ± 0.66 for females; p = 0.047). PTH and 1,25(OH)2D levels were indistinguishable in males and females (PTH: 458 ± 135 versus 574 ± 59 pg/mL, p = 0.370; 1,25(OH)2D: 50.9 ± 8.7 versus 44.9 ± 5.7 pg/mL; p = 0.5519). The serum magnesium levels were on average at the lower end of the normal range, but no differences were observed for both genders, even when corrected for SDS (data not shown). Likewise, there were no differences for 25OHD, TSH, and fT4 levels for male and female PHP1B patients, which were on average all within the respective normal ranges, except for a slightly elevated TSH. Note that patient 82/II-1 was excluded from calculating the means ± SE because a GNAS microdeletion had been identified in this individual during the course of this study.

GNAS methylation analysis and search for genetic abnormalities involving the STX16/GNAS region in our cohort of sporPHP1B patients

For some of the previously studied PHP1B patients, genomic DNA had been studied by Southern blot analysis after incubation with different methylation-sensitive endonucleases.^(11,33) These studies had revealed GNAS methylation changes, but had required large amounts of DNA and no or only very small amounts had been preserved for some patients. For most individuals, however, sufficient amounts of DNA were available to repeat GNAS methylation analyses by MS-MLPA, thus allowing quantitative assessments at different sites. These studies confirmed the broad GNAS methylation changes for the previously studied PHP1B patients and established epigenetic changes for all more recently investigated individuals (see Supplemental Table 2). Similar to previously reported findings by others,^(17–21) several patients showed incomplete GNAS methylation changes. Only 1 of the 60 patients, namely individual 82/II-1, revealed evidence for an allelic loss by MLPA (Fig. 1A). Analysis with five adjacent probes suggested a deletion extending from exon NESP to antisense exon 4. This conclusion was substantiated through the analysis of several microsatellite markers, which revealed loss of heterozygosity at marker 806M20-CA and suggested an even larger deletion. Some of the other sporPHP1B patients showed homozygosity for one or several microsatellite markers, but analysis of the maternal DNA provided no evidence for discordance (see Supplemental Table 3). These studies thus excluded for all sporPHP1B patients paternal uniparental isodisomy involving large portions of the STX16/GNAS region. For 24 patients, we furthermore analyzed microsatellites using paternal DNA, which excluded paternal uniparental heterodisomy of the chromosome 20q region.

Fig. 1.

(A) MLPA for patient 82/II-1 (upper panel) and his healthy mother 82/I-2 (lower panel); the MLPA probes cover several exons of STX16, the alternative first GNAS exons NESP, XL, and A/B, and different Gsα exons (see also Patients and Methods); note that the antisense (AS) exon 1 is covered by three distinct probes, whereas AS exons 3 and 4 are each covered by one probe, as indicated. (B) Laboratory findings, analysis of different microsatellite markers, and GNAS methylation status for patient 82/II-1 (black square) and his unaffected parents, 82/I-1 (white square) and 82/I-2 (gray circle); the patient’s paternal allele is indicated by italic numbers, whereas the maternal allele is shown with gray background. Marker 806M20-CA shows discordance (bold numbers) between 82/II-1 and his mother 82/I-2; percent methylation at GNAS exons NESP, AS, XL, and A/B is provided for both individuals. (C) Schematic representation of the region extending from the syntaxin 16 (STX16) to the GNAS locus. Boxes represent different STX16 or GNAS exons; horizontal arrows, direction of transcription; **Locations of differentially methylated regions. Vertical arrows indicate locations of different microsatellite markers. P = paternal; M = maternal.

Discovery of a novel inherited GNAS microdeletion

The MLPA findings for patient 82/II-1 and the discordance between him and his mother at marker 806M20-CA (Fig. 1B) suggested a large deletion that could potentially extend from exon AS1 to STX16 exon 8 (Fig. 1C). The healthy mother, 82/I-2, showed a similar reduction in copy number for the same region as the patient (see Fig. 1A). However, analysis of her DNA revealed no LOM at GNAS exons AS, XL, and A/B, yet an apparently complete gain-of-methylation (GOM) at GNAS exon NESP (data not shown), indicating that her deletion resides on the paternal allele. These findings suggested that patient 82/II-1 had inherited the mutant allele from his mother, who in turn had presumably inherited it from her father, thus explaining that she is an unaffected carrier.

Subsequent analysis of several SNPs between GNAS exons AS2 and AS3, and between exon NESP and marker 806M20-CA showed discordance between the patient and his mother for rs11481507, rs6123832, and rs1800905, whereas four other SNPs, including rs6064709 and rs6026557, revealed heterozygosity for either the patient or his mother (Fig. 2A, Table 1). PCR using the forward primer for rs6064709 (primer a) and the reverse primer for rs6026557 (primer b) allowed amplification of genomic fragments from DNA of patient 82/II-1 and his mother, but not for DNA from his father or from a healthy control. Nucleotide sequence analysis of the amplified genomic DNA indicated that the patient and his mother carry two deletions; one large deletion of 37,597 bp (assembly hg19, chr. 20 (GRCh37); g.57,380,466-57,418,062,del) and a small deletion of 1427 bp (assembly hg19, chr. 20 (GRCh37); g.57,418,522-57,419,948) (Fig. 2B, C; panel a). To confirm these findings, duplex PCR was performed using primers located at both deletion breakpoints, which allowed amplification of the wild-type 240-bp fragment with primers d and e for DNA from the patient and his mother, as well as for DNA from the healthy father and from a control sample (Fig. 2C; panels b and c). Using primers c and e, only DNA from the patient and his mother allowed amplification of a 189-bp fragment covering the telomeric breakpoint of the large deletion. Likewise, when using primers f and g, or primers h and i, the wild-type allele could be amplified from DNA of the patient, both parents, and the control sample. In contrast, when using primers f and i, a PCR product was amplified only for 82/II-1 and his mother.

Fig. 2.

(A) Schematic representation of GNAS exons NESP and AS1-5. Horizontal black lines indicate the identified deletions. (B) Nucleotide sequences across the two deletions are shown; numbering according to assembly hg19 (GRCh37). (C) (a), PCR amplification across the deletion of the mutant allele using primers located centromeric and telomeric, respectively, of the breakpoints; (b) and (c), Duplex PCR by using primers at each of the deletion breakpoints (horizontal arrows in A).

Table 1.

Single Nucleotide Polymorphism for Family 82

SNP	Position	Father	Patienta	Mother
rs412181	57298627	A/G	A/G	A/G
rs6100228	57374642	A/T	T	A/T
rs6064709	57379850	G	G	A/G
rs1800905	57415745	A/G	G	A
rs6123832	57418071	C/T	T	C
rs11481507	57419741	–/T	T	–
rs6026557	57420401	A/C	A/C	C

Positions are based on hg19 (GRCh37) assembly.

^aThe homozygous nucleotides shown in bold letters indicate discordance between the patient and his mother.

Evaluation of siblings and children of sporadic PHP1B patients

Information regarding siblings was available for 14 male and 27 female patients affected by sporPHP1B (Table 2). DNA could be investigated for 5 siblings of male and 23 siblings of female PHP1B patients. Forty-seven siblings were reportedly healthy, but no DNA could be obtained for analysis. Thus, although only about one-third of the siblings could be tested by MS-MLPA, our findings indicated that none of the 75 siblings of the 41 sporadic patients is affected by PHP1B.

Table 2.

Siblings and Offspring of Investigated PHP1B Cases

Sporadic PHP1B: healthy siblings (n = 75) (by MS-MLPA/by report)
Male	14	Male	17 (4/13)
		Female	9 (1/8)
Female	27	Male	29 (10/19)
		Female	20 (13/7)
Sporadic PHP1B: healthy offspring (n = 21) (by MS-MLPA/by report)
Male	1	Male	1 (0/1)
		Female	0
Female	10	Male	9 (7/2)
		Female	11 (7/4)

Patients and offspring include cases in Refs. ¹⁵ and ³⁴.

Only 8 PHP1B patients (1 male, 7 females) in our series were old enough to have children. None of their 15 children are affected by PHP1B, as determined by MS-MLPA (n = 9) or by report (n = 6). This is consistent with the findings in two previous reports,^(15,34) in which all 6 children of 3 female PHP1B patients, all 3 with broad GNAS methylation changes, are healthy. Thus, information on 21 children of 11 patients affected by PHP1B suggests that this disease variant is not readily transmitted to the next generation (see Table 2).

DMR-specific quantification of GNAS methylation changes in sporPHP1B

Quantification of methylation at the four DMRs of the GNAS locus was available for 47 out of 60 sporPHP1B patients; little or no DNA was available for all other patients, for whom GNAS methylation had been assessed previously through Southern blot analysis of genomic DNA that had been digested with different methylation-sensitive endonucleases.^(11,33) As previously shown by other investigators,^(17–21) we identified different patterns of GNAS methylation in 47 cases (Fig. 3). Group 1 is characterized by complete methylation changes at GNAS exons NESP, AS, XL, and AB (n = 34). Group 2 is characterized by partial LOM at exon XL (>17%), and complete methylation changes at exons NESP, AS, and A/B (n = 6). Group 3 is characterized by incomplete LOM at exon AB (>7%) and incomplete changes at the other three DMRs (n = 7).

Fig. 3.

Percent methylation for each DMR of the GNAS locus in different subtypes of sporPHP1B cases. Horizontal lines and error bars in the graphs are the mean ± SEM. Data for the four DMRs as well as the corresponding p values are presented for each group in the table below the figure.

DNA from peripheral blood cells or LCLs show differences in GNAS methylation

The incomplete epigenetic changes at several GNAS DMRs in some sporPHP1B patients could indicate that different mechanisms are involved in regulating GNAS methylation. We therefore investigated the GNAS methylation pattern in lymphoblastoid cells derived from 3 patients with epigenetic changes characteristic of groups 1 to 3; LCLs from 4 healthy controls and from 3 patients with PHP1B due to deletions within STX16 or GNAS served as controls.

Genomic DNA from peripheral blood of patient 83/II-1 had shown partial epigenetic changes at all four DMRs of the GNAS locus (see Supplemental Table 2) and thus did not fit entirely into any of the three groups. However, her LCLs showed complete LOM at GNAS exons A/B and AS, incomplete LOM at exon XL, and complete GOM at exon NESP (Fig. 4); ie, findings that are similar to the epigenetic changes observed in the patients of group 3. In contrast, DNA derived from LCLs of patient 140/III-1 showed in comparison to DNA derived from his peripheral blood cells only a small decrease in methylation at exon XL, but epigenetic changes at A/B, AS, and NESP that were indistinguishable from those observed for LCL-derived DNA of patient 83/II-1; these findings are similar to the epigenetic changes observed in the patients of group 2. Patient S2/II-3⁽³²⁾ revealed similarly complete GNAS methylation changes in DNA from blood and LCLs; ie, changes similar to those in group 1 that encompasses most sporPHP1B patients. Likewise DNA derived from LCLs of 4 healthy controls showed methylation patterns at all four DMRs that were indistinguishable for DNA from peripheral blood of numerous healthy controls (n = 22). No change in GNAS methylation was observed when using DNA derived from either LCLs or from blood cells from 3 patients with autosomal dominant PHP1B due to STX16 or GNAS deletions (see legend to Fig. 4).

Fig. 4.

Comparison of GNAS methylation using genomic DNA extracted from peripheral blood samples and from LCLs of three sporPHP1B patients. Using peripheral blood DNA of healthy controls (n = 22), a reference range for normal methylation at all four GNAS DMRs was generated for exons A/B (52.14% ± 1.11%), XL (52.55% ± 0.77%), AS (54.45% ± 1.16%), and NESP (52.68% ± 0.88%); percent methylation of healthy controls was divided by 50 thus setting normal methylation at 1; the gray horizontal bar represents the mean ± SE of healthy controls. Percent methylation for the different GNAS DMRs of each sporPHP1B patient was divided by the average methylation of healthy controls; colored open circles, blood DNA; colored closed circles, LCL-derived DNA (140/III-1: blue symbols; 83/II-1: red symbols; S2-II/3: green symbols). Results for genomic DNA extracted from LCLs of four healthy controls are shown by black symbols. As expected, LCL-derived DNA from two patients with autosomal dominant PHP1B due to a 3-kb or a 4.4-kb STX16 deletion showed normal ratios of methylation at GNAS exons XL (0.9), AS (1.0), and NESP (1.1), but it was reduced at GNAS exon A/B (0). In contrast, LCL-derived DNA from a patient with autosomal dominant PHP1B due to a NESP/AS deletion (C-II/1 (9)) showed methylation ratios at A/B, XL, AS, and NESP of 0, 0.1, 0, and 1.9, respectively.

Correlation between GNAS methylation status and clinical parameters

Serum calcium, PTH, and TSH concentrations were similar in all three groups and no significant difference was observed for these biochemical parameters and the GNAS methylation status (Fig. 5). The SDS of serum phosphate levels were significantly higher for group 1 (+5.18 ± 0.42; n = 31) than for the other two groups combined (+2.97 ± 0.64; n = 10) (group 1 versus groups 2 and 3; p = 0.0109). The patients in group 1 presented on an average at a younger age than the patients in the other two groups combined (10.7 ± 1.2 versus 17.3 ± 2.7 years; p = 0.0131); however, there were only few patients in groups 2 and 3. In fact, only comparisons between patients in groups 1 and 2 revealed a significant difference in age at diagnosis (10.7 ± 1.2 versus 19.6 ± 4.1; p = 0.0095). For patients with complete LOM at exon A/B, age at diagnosis correlated significantly with percent methylation at the exon XL DMR; no such correlation was apparent for the percent methylation at the DMRs for GNAS exons NESP and AS (Fig. 6).

Fig. 5.

Age at diagnosis and different biochemical parameters for the investigated sporPHP1B patients with different GNAS methylation patterns. *p < 0.05 between the groups. PTH = parathyroid hormone; Ca = calcium; P = phosphate; SDS = standard deviation score.

Fig. 6.

Correlation between the age at diagnosis and % methylation at GNAS exons NESP, AS, and XL; r² and p values are presented in the table below the graphs.

Discussion

We investigated a large number of sporPHP1B patients, who all showed broad epigenetic changes at the GNAS locus; ie, loss of the maternal and gain of the paternal methylation imprints. Although the number of females affected by this rare disorder was higher in the current study, others had reported that more males than females are affected,^(15,18–21) making it likely that both sexes are affected with equal frequency. There was a wide range for the ages when patients were first diagnosed with sporPHP1B. However, the mean age at diagnosis was indistinguishable from that of AD-PHP1B patients, who are carriers of the previously described 3-kb STX16 deletion on the maternal allele.⁽²⁹⁾ This enhances further the likelihood that LOM at exon A/B, which is shared among all PHP1B variants with abnormal GNAS methylation, is particularly important for reducing Gαs transcription and thus the development of hormonal resistance.

At diagnosis, calcium, PTH, and 1,25(OH)2D levels were not significantly different in males and females. Even when corrected for age, phosphate levels were marginally higher in male than in female PHP1B patients. However, such a difference between both genders had not been previously observed.⁽²⁹⁾ It thus appears more likely that the higher phosphate levels of male patients in the current study are related to their younger age at diagnosis, rather than gender-specific differences in phosphate levels. The average TSH level was slightly elevated with a mean fT4 concentration at the lower end of the reference range, but there were no gender-specific differences; these findings are consistent with previous data indicating that Gαs is derived in the human thyroid preferentially from the maternal allele.^(5,6,35) In our sporPHP1B cohort magnesium levels were on average at the lower end of normal and significantly reduced in some patients, making it plausible that the hypocalcemia-induced elevation of PTH levels enhances the excretion of magnesium in the distal renal tubules; ie, a portion of the kidney where Gαs is expressed from both parental alleles.

The majority of our investigated sporPHP1B patients showed a complete loss of all maternal methylation imprints, namely at GNAS exons A/B, XL, and AS, and a complete GOM at exon NESP. With the exception of 82/II-1, who was subsequently found to carry a novel maternally inherited GNAS deletion (see below; second to last paragraph), none of the investigated patients provided evidence for an allelic loss in the STX16-GNAS region as determined by MLPA. Furthermore, most affected individuals were heterozygous for several of the six investigated microsatellite markers, yet 22 sporPHP1B cases showed homozygosity for at least one microsatellite marker comprising the STX16-GNAS region, namely 261P9-CA, 806M20-CA, and/or 543J19-TTA. However, the latter patients revealed for these markers no evidence for discordance with their mother, thus making it unlikely that they are affected by a dominantly inherited disorder outside the GNAS locus or by patUPD20q involving a large portion of the paternal chromosome 20q. Inheritance of both paternal alleles, ie, patUHD20q, which is generally a very rare event, was experimentally excluded in 24 patients; it thus appears unlikely that sporPHP1B is caused by the presence of two paternal GNAS alleles.

None of the parents and none of the 75 siblings in our collection of sporPHP1B patients showed clinical or laboratory evidence for an abnormal regulation of calcium homeostasis, and none of those siblings for whom genomic DNA was available revealed methylation changes (n = 27; data not shown). These findings make it unlikely that the sporadic disease variant represents yet another autosomal dominant form of PHP1B, even when considering the possibility of incomplete penetrance. Likewise, inherited heterozygous mutations involving maternal portions of the GNAS locus^(9,10) or mutations in another imprinted gene elsewhere in the genome are equally unlikely. Although it is conceivable that de novo point mutations or deletions had been acquired on the maternal GNAS allele by some of the sporPHP1B patients, the different MLPA probes failed to detect such deletions, which would be predicted to involve GNAS exons AS3 and AS4, ie, those exons that are shared among the 3 previously reported familial AD-PHP1B cases with broad GNAS methylation changes^(9,10) and for patient 82/II-1 described herein.

If de novo heterozygous mutations elsewhere in the genome were responsible for PHP1B in our patients, those mutations could have been transmitted as a dominant trait; consequently one-half of the children of affected individuals would also be affected. However, this was not the case for 11 adult sporPHP1B patients with a total of 21 children (8 sporPHP1B patients in this report and 3 previously reported cases^(15,34)); it therefore appears unlikely that a de novo heterozygous mutation in GNAS or elsewhere occurred in the investigated sporPHP1B patients, which would then be transmitted as a dominant trait (Pearson’s chi squared test p value: 0.00032). Instead, it seems more plausible that a recessive mutation somewhere in the genome is responsible for our and most other sporPHP1B cases. This hypothesis is consistent with previously reported findings in family S2, 2 affected and 3 unaffected sisters, for whom we had excluded a mutation within the GNAS region as the cause of their PHP1B.⁽³²⁾ GNAS mutations were furthermore unlikely in several other previously reported families^(32,34) because several siblings and/or children of affected patients were healthy and analysis of their DNA revealed no epigenetic changes, yet the same maternal haplotype.

However, 4 of the sporPHP1B patients investigated herein, 7/III-1, 67/III-1, 72/II-1, and 78/III.1, have 4 to 7 siblings each, yet none of the siblings for whom genomic DNA was available (n = 11; data not shown) revealed GNAS methylation changes; likewise, none of the siblings of other reported sporPHP1B cases had clinical or laboratory evidence for the disorder.^(32,34) Assuming a recessive mode of inheritance, sporPHP1B patients would likely be carriers of homozygous or compound heterozygous mutations in a transacting gene that establishes or maintains GNAS methylation. The parents of sporPHP1B cases would thus have to be healthy heterozygous carriers and 25% of their children should be affected; this is not statistically different from our findings, namely 41 PHP1B patients among a total of 116 siblings (Pearson’s chi squared test to assess whether a recessive disease can be excluded; p value: 0.204). However, it is also conceivable that a mutation in one parental allele had been inherited, whereas a second hit had occurred de novo on the other allele. In either case, the term sporPHP1B would no longer be appropriate.

PatUPD20q can be responsible for some cases of sporPHP1B,^(14,15) which would explain why the disorder in these patients is not transmitted to the next generation. Likewise, it is conceivable that a gene conversion event leads to duplication of a small portion of the paternal GNAS allele and thus complete epigenetic changes.^(36,37) Such a limited duplication, which would be predicted to comprise the imprinting control region, could provide an explanation for the sporadic PHP1B variant, just like patUPD20q. Unfortunately, such gene conversations would be very difficult to identify, because these presumably small regions may not contain an informative SNP that could provide evidence for discordance between a sporPHP1B patient and his/her mother. Because paternal duplications through gene conversion can occur postzygotically, the resulting epigenetic changes may lead to incomplete GNAS methylation changes. However, most sporPHP1B patients showed a complete loss of all maternal GNAS methylation imprints and a complete GOM at GNAS exon NESP. It is, therefore, conceivable that sporPHP1B is caused only occasionally by gene conversion, possibly in those patients with incomplete LOM at GNAS exon XL alone (group 2) or with incomplete methylation changes at all four differentially methylated GNAS regions (group 3); ie, findings that are similar to those described by others.^(17–21)

Incomplete epigenetic changes at GNAS exons A/B, AS, and/or XL in DNA from peripheral blood cells could also be consistent with mosaic defects in an as-of-yet undefined gene. In fact, it was surprising that DNA obtained from LCLs of patients 140/III-1 and 83/II-1 revealed, in contrast to the partial epigenetic changes in DNA from peripheral blood cells, a complete LOM at GNAS exons A/B and AS, as well as a complete GOM at exon NESP. However, in contrast to patient 83/II-1, who showed at exon XL an almost complete LOM when testing LCL-derived DNA, methylation at this DMR was largely preserved in LCL-derived DNA of patient 140/III-1. Both individuals are thus different from the previously reported patient S2/II-3,⁽³²⁾ who showed indistinguishable epigenetic changes at all four DMRs of the GNAS locus, irrespective of the source of genomic DNA. Likewise, genomic DNA from blood cells or LCLs from healthy controls, and from patients with PHP1B due to STX16 or GNAS deletions revealed no obvious differences in the pattern of GNAS methylation. These data, which are consistent with our previous findings,⁽⁹⁾ indicate that immortalization of control lymphocytes does not change the normal parent-specific GNAS methylation pattern and that epigenetic GNAS changes caused by germline STX16/GNAS mutations are faithfully maintained in cell culture.

Patients 83/II-1 and 140/III-1 had shown incomplete GNAS methylation changes at all four DMRs in DNA samples extracted from two independent blood samples that had been collected at least 1 year apart. In contrast, immortalized cells from 83/II-1 and 140/III-1 revealed a complete LOM at exons AS and A/B, and a complete GOM at exon NESP. Surprisingly, only LCLs from patient 83/II-1 showed a close to complete LOM at GNAS exon XL, whereas patient 140/III-1 maintained a methylation pattern at this DMR that was similar to that of healthy controls. It is conceivable that these two sporPHP1B patients are mosaic for different as-of-yet unknown genetic mutations, which are responsible for their epigenetic changes at the GNAS locus. Furthermore, the mutant LCLs could have a significant growth advantage thus rapidly outcompeting the normal cell population and explaining why methylation abnormalities become complete after prolonged time in culture. Alternatively, the epigenetic abnormalities in sporPHP1B patients may be caused by the partial or complete lack of bioactivity of a protein that is normally required for maintaining the normal GNAS methylation pattern. Hence, DNA from peripheral blood cells showed only partial GNAS methylation abnormalities whereas reduced expression of this protein in LCLs resulted in complete epigenetic changes.

The novel, large GNAS deletion identified in 82/II-1 overlaps with the previously observed deletions only for the region comprising exons AS3 and AS4, which had been shown to be associated with complete loss of all maternal GNAS methylation imprints.^(9,10) The AS3-4 portion of GNAS thus appears to be of particular importance for establishing or maintaining the epigenetic changes on the maternal allele. The mutation was inherited from the patient’s unaffected mother suggesting that she carries the genetic defect on her paternal allele. Consistent with this conclusion, her DNA showed no LOM at GNAS exons A/B, XL, and AS, but an apparent GOM at exon NESP; apparent because the portion of her paternal GNAS region that remains nonmethylated has been deleted, thus giving the false impression of GOM (data not shown).

In summary, detailed genetic and epigenetic investigations in a large number of families with only a single patient affected by PHP1B has lead to the discovery of one novel, maternally inherited GNAS deletion. For all other investigated PHP1B patients, the genetic cause(s) underlying the broad GNAS methylation changes and associated hormonal resistance remains to be determined. However, unless a stochastic developmental process targets exclusively the GNAS locus, but not other differentially methylated regions throughout the genome, our findings predict that the sporadic PHP1B variant is caused either by recessive mutations or by duplications involving very small portions of the paternal GNAS locus that are undetectable with the technology employed in this study.