Paternal GNAS Mutations Lead to Severe Intrauterine Growth Retardation (IUGR) and Provide Evidence for a Role of XLαs in Fetal Development

Abstract

Context:

Heterozygous GNAS inactivating mutations cause pseudohypoparathyroidism type Ia (PHP-Ia) when maternally inherited and pseudopseudohypoparathyroidism (PPHP)/progressive osseous heteroplasia (POH) when paternally inherited. Recent studies have suggested that mutations on the paternal, but not the maternal, GNAS allele could be associated with intrauterine growth retardation (IUGR) and thus small size for gestational age.

Objectives:

The aim of the study was to confirm and expand these findings in a large number of patients presenting with either PHP-Ia or PPHP/POH.

Patients and Methods:

We collected birth parameters (ie, gestational age, weight, length, and head circumference) of patients with either PHP-Ia (n = 29) or PPHP/POH (n = 26) with verified GNAS mutations. The parental allele carrying the mutation was assessed by investigating the parents or, when a de novo mutation was identified, through informative intragenic polymorphisms.

Results:

Heterozygous GNAS mutations on either parental allele were associated with IUGR. However, when these mutations are located on the paternal GNAS allele, IUGR was considerably more pronounced than with mutations on the maternal allele. Moreover, birth weights were lower with paternal GNAS mutations affecting exons 2–13 than with exon 1/intron 1 mutations.

Conclusions:

These data indicate that a paternally derived GNAS transcript, possibly XLαs, is required for normal fetal growth and development and that this transcript affects placental functions. Thus, similar to other imprinted genes, GNAS controls growth and/or fetal development.

Mutations in GNAS, the imprinted gene complex encoding the α-subunit of the stimulatory G protein (Gαs) are responsible for several different, inherited disorders (1). Maternally inherited, heterozygous mutations affecting those GNAS exons that encode Gαs cause pseudohypoparathyroidism type Ia (PHP-Ia), a disorder characterized by developmental defects referred to as Albright's hereditary osteodystrophy (AHO), resistance toward PTH, and frequently resistance toward other hormones that mediate their actions through Gαs-coupled receptors. When inherited paternally, the same heterozygous GNAS mutations lead to pseudopseudohypoparathyroidism (PPHP), ie, AHO features in the absence of hormonal resistance (1).

GNAS is a complex imprinted gene locus that, through the use of alternative first exons, encodes the signaling proteins Gαs and XLαs and the neuroendocrine secretory protein 55 (NESP55) (2, 3). The GNAS locus furthermore gives rise to an antisense (AS) transcript as well as the A/B transcript (also referred to as 1A transcript) that appears to contribute to the regulation of Gαs expression. Gαs is derived in most tissues from both parental alleles; however, expression of this ubiquitously expressed signaling protein is derived in some tissues, such as the proximal renal tubules (and few other tissues), predominantly from the maternal allele (2, 4, 5). The mechanisms that silence paternal Gαs expression are largely unknown, but most likely involve active transcription from exon A/B. When mutations in GNAS exons 1–13 are inherited paternally, Gαs expression from the maternal allele is unaffected, thus leading to normal responsiveness to PTH and other hormones (1, 6).

Paternally inherited GNAS mutations that affect Gαs function lead to PPHP, namely lack of hormonal resistance, yet some AHO stigmata that are indistinguishable from those observed in patients affected by PHP-Ia. However, it became apparent that the incidence and/or severity of some of these AHO features differ considerably between PHP-Ia and PPHP patients. For example, only individuals with maternally inherited Gαs mutations, ie, PHP-Ia patients, are obese (7, 8), whereas the sc ossifications (referred to as progressive osseous heteroplasia [POH]) occur predominantly with paternally inherited mutations affecting Gαs function (9, 10). Furthermore, recent studies have suggested that paternal mutations affecting different GNAS exons, but not maternally inherited mutations, are associated with intrauterine growth retardation (IUGR) and thus small for gestational age (10, 11). To confirm and expand these findings, we studied the birth records of a large number of patients diagnosed with either PHP-Ia or PPHP/POH due to GNAS mutations with regard to weight and length at birth.

Patients and Methods

Patients

For this retrospective study, we enrolled 55 patients who had been referred to our Reference Center for Rare Disorders of Calcium and Phosphorus Metabolism. For each family, written informed consent was obtained from the patients and/or their parents for collection of clinical and laboratory data and for DNA collection to conduct molecular studies. Weight (W) and term data at birth were available for all patients, but data for length (L) and head circumference (HC) was lacking for 6 and 13 patients, respectively. Pregnancies complicated by pre-eclampsia were excluded from the study.

The clinical criteria used for diagnosis of PHP-Ia (OMIM 103580) were previously described and include AHO, ie, short stature, rounded face, brachydactyly, and mental retardation associated with multihormonal resistance toward multiple hormones, particularly PTH and TSH (1). The clinical criteria used for diagnosis of POH (OMIM 166350) are those described by Kaplan et al (12, 13), ie, the presence of heterotopic ossifications progressing from cutaneous and sc tissue into deep connective and muscular tissue; 11 of these patients had been described previously (10). The criteria used for PPHP (OMIM 612463) diagnosis included clinical features indicative of AHO that were associated with decreased erythrocyte Gαs activity without laboratory evidence for PTH or TSH resistance.

Outcome of pregnancy and biometric data at birth were collected retrospectively using birth certificates or hospital records. Gestational age, HC, L, and W were compared to the French reference chart AUDIPOG specific for girls and boys (14). All data are expressed as Z-scores. Preterm birth was defined as delivery before 37 weeks gestation, and IUGR was defined as a Z-score of less than −1.28 (<10th percentile) as generally accepted (15).

Histopathological examination of placenta was obtained in 5 cases (4 in the PPHP/POH cohort and 1 in the PHP-Ia cohort). The weight was expressed as a percentile compared to placental weight at the corresponding term (16).

Laboratory investigations

Biochemical and endocrine analyses were performed using standard methods. PTH resistance was defined as abnormally low total calcium concentration (<2.2 mmol/L) in the presence of elevated, age-corrected serum phosphorus levels and/or increased serum PTH levels (normal range, 10–65 pg/mL). Gαs bioactivity was evaluated in erythrocyte membranes, as described by Marguet et al (17); results were expressed as percentage of activity in erythrocyte membranes from healthy adult controls. Gαs bioactivity values less than 80% were considered to be reduced.

Molecular analysis

Genomic DNA was isolated from peripheral blood leukocytes using standard methods. We designed several sets of primers for amplification of GNAS exons 1–13 and their intron-exon junctions as previously described (6) (GenBank ID: NM_001077488.2). PCRs were performed according to standard procedures except for exon 1, which required different conditions due to highly GC nucleotide-rich sequences (10). PCR products were purified and sequenced using the CEQ DTCS Quick Start Kit (Beckman Coulter) on a Beckman Coulter DNA Sequencer. When available, DNA samples from parents were analyzed to determine the parental origin of the mutations. If a de novo mutation was identified, the parental origin of the mutated allele was defined through parent-specific intragenic polymorphisms: rs2295583 (GenBank accession number) in intron 3 (T/A), rs234629 in intron 6 (G/A), rs7121 in exon 5 (T/C), and rs919196 (T/C) in intron 6. DNA was PCR-amplified across one of these polymorphisms, and the PCR product was cloned using the pGEM-T Easy Vector kit (Promega), as previously described (6, 10). Ten to 16 independent clones were chosen at random for nucleotide sequence analysis or restriction enzymatic analysis to search for the parent-specific polymorphism and thereby to determine whether the mutation was on the maternal or the paternal allele.

Statistical analysis

We used the Mann-Whitney nonparametric test.

Results

We investigated 55 patients (28 males, 27 females) from unrelated families who are carriers of GNAS mutations on either the maternal or the paternal allele. Clinical and laboratory phenotypes consistent with either PPHP/POH (numbered P-01 to P-26) or PHP-Ia (numbered M-01 to M-29) were identified in 26 (12 males, 14 females) and 29 subjects (16 males, 13 females), respectively (Tables 1 and 2). Three patients with PHP-Ia (M-08, M-27, and M-28) and associated hypothyroidism had been identified through the newborn TSH screening.

Table 1.

Genetic, Clinical, Biochemical Findings, and Birth Parameters for 26 Patients With PPHP/POH (Paternal)

ID	Age at Molecular Diagnostic	Exon	Mutation	Sex	Disease Transmission	Phenotype	Biochemistries				Birth Parameters
							Ca, mmol/L (normal range, 2.2–2.6)	P, mmol/L (normal range, 1.1–2.0)	PTH, pg/mL (normal range, 10–65)	Gαs Activity (normal range, 85–115%)	Term, wk	Weight, g [Z-scores]	Length, cm [Z-scores]	Head Circumference, cm [Z-scores]	IUGR
P-01	22 y	1	c.-4_2del p. (0?)	F	—	PPHP	2.63	—	19	53	>37	2250 [<−1.49]	45 [<−1.22]	33 [<−0.17]	Yes
P-02b	5 y	1	c.3G>A p. (Met1Iso)	F	De novo	PPHP	2.37	1.71	42	—	40 (full term)	2500 [−2.40]	47 [−1.55]	—	Yes
P-03	1 y 1 mo	1	c.34C>T p. (Gln12*)	M	De novo	PPHP/POH	2.6	1.76	67	—	38	2210 [−2.66]	44 [−2.67]	33 [−0.93]	Yes
P-04	11 y 2 mo	1	c.85C>T p. (Gln29*)	M	De novo	POH	2.31	—	55	73	40	2670 [−2.24]	46 [−2.53]	33 [−1.52]	Yes
P-05	52 y	1	c.85C>T p. (Gln29*)	F	—	POH	2.19	0.6	31.7	74	>37	2100 [<−1.96]	—	—	Yes
P-06	5 mo	1	c.103C>T p. (Gln35*)	M	Father PHP-Ia	POH	2.61	1.86	—	—	39	3345 [−0.01]	47 [−1.31]	—	No
P-07	11 y 11 mo	int 1	c.139 + 1G>C p. (?)	F	De novo	POH	N	1.41	N	60	41	2780 [−1.81]	46 [−2.43]	34 [−0.88]	Yes
P-08	12 y 9 mo	5	c.320T>C p. (Ile107Thr)	F	De novo	PPHP	2.32	N	38	65	37	1720 [−3.32]	41 [−3.19]	30 [−2.31]	Yes
P-09	9 mo	5	c.348_349insT p. (Pro117Serfs*24)	F	De novo	POH	—	—	—	82	38	1880 [−3.45]	43 [−2.79]	33 [−0.54]	Yes
P-10	22 y	int 5	c.435 + 1G>A p. (?)	M	De novo	PPHP/POH	2.35	1.03	—	79	36	1180 [−5.36]	38 [−4.33]	29 [−2.84]	Yes
P-11	14 y 1 mo	6	c.436del p. (Glu146Asnfs*27)	M	De novo	PPHP	2.61	1.05	13	63	37	1740 [−3.64]	45 [−1.60]	30 [−2.63]	Yes
P-12a	13 y 4 mo	6	c.472_474del p. (Glu158del)	F	De novo	POH	2.52	N	28	40	33	1150 [−2.56]	37 [−2.44]	26 [−2.83]	Yes
P-13	11 y 9 mo	int 6	c.533 + 5_533 + 8del p. (?)	F	—	PPHP	N	N	35	—	39	1990 [−3.72]	—	—	Yes
P-14	7 mo	7	c.568_571del p. (Asp190Metfs*14)	M	De novo	POH	2.68	2.07	26	84	38	2200 [−2.69]	46 [−1.89]	32 [−1.99]	Yes
P-15a	7 y 2 mo	7	c.568_571del p. (Asp190Metfs*14)	F	De novo	POH	2.51	1.63	50.1	93	39	2310 [−2.55]	46 [−1.70]	33 [−0.85]	Yes
P-16	2 y 5 mo	7	c.568_571del p. (Asp190Metfs*14)	M	De novo	PPHP/POH	2.36	1.89	37	75	37	1930 [−2.92]	43 [−2.59]	31 [−2.53]	Yes
P-17	33 y	7	c.568_571del p. (Asp190Metfs*14)	F	De novo	PPHP/POH	2.35	1.13	22.4	—	>37	2480 [<−0.83]	47 [<−0.24]	34 [<0.82]	No
P-18	9 y	7	c.574_575del p. (Val192Alafs*18)	M	De novo	POH	2.6	1.76	21.8	71	40	2200 [−3.85]	45 [−3.08]	32 [−1.94]	Yes
P-19	18 y	int 7	c.588 + 1G>C p. (?)	F	Father	PPHP/POH	N	N	N	84	37 + 2	2020 [−2.22]	42 [−2.45]	32 [−0.89]	Yes
P-20	44 y	int 7	c.588 + 2T>G p. (?)	F	De novo	PPHP	2.42	1.02	39.4	—	36	1420 [−3.75]	40 [−3.02]	—	Yes
P-21a	7 y 10 mo	int 7	c.588 + 2T>G p. (?)	M	De novo	POH	2.46	1.5	54.1	—	36 + 4	1230 [−5.09]	41 [−3.16]	31 [−1.49]	Yes
P-22a	1 y 6 mo	8	c.627dup p. (Glu210*)	M	De novo	POH	—	—	50	87	38	2030 [−3.29]	45 [−2.41]	32 [−1.64]	Yes
P-23	19 y	9	c.694C>T p. (Arg232Cys)	M	De novo	PPHP	2.47	0.96	29	64	38	2265 [−2.47]	45 [−2.15]	34 [−0.57]	Yes
P-24	35 y	9	c.704G>A p. (Trp235*)	F	Father	PPHP	2.42	0.71	29	—	>37	2300 [<−1.34]	—	—	Yes
P-25	25 y	10	c.817C>T p. (Leu273Phe)	F	—	PPHP	2.14	N	64	—	>37	1820 [<−2.93]	43 [<−2.21]	34 [<−0.18]	Yes
P-26	1 mo	12	c.1027C>T p. (Arg343*)	M	Father PHP-Ia	PPHP	2.51	1.99	62	—	38	2660 [−1.27]	47 [−1.11]	32 [−1.64]	No

Abbreviations: F, female; M, male; —, unknown; N, normal values. For patient P-13, the variation c.533 + 5_533 + 8del is close to the donor splice-site and yields an aberrant transcript.

^aSubjects for whom the placentae were analyzed.

^bReference 18.

Table 2.

Genetic, Clinical, and Biochemical Findings and Birth Parameters for 29 Patients With PHP-Ia (Maternal)

ID	Age at Molecular Diagnostic	Exon	Mutation	Sex	Disease Transmission	Phenotype	Biochemistries				Birth Parameters
							Ca, mmol/L (normal range, 2.2–2.6)	P, mmol/L (normal range, 1.1–2.0)	PTH, pg/mL (normal range, 10–65)	Gαs Activity (normal range, 85–115%)	Term, wk	Weight, g [Z-scores]	Length, cm [Z-scores]	Head Circumference, cm [Z-scores]	IUGR
M-01	1 y 2 mo	1	c.85C>T p. (Gln29*)	M	De novo	PHP-Ia POH-like	2.24	2.09	178	52	39	2835 [−1.30]	—	—	Yes
M-02	6 y 11 mo	1	c.100A>T p. (Leu34*)	M	—	PHP-Ia	1.93	2.82	640	52	40	3280 [−0.53]	50 [−0.32]	38 [2.16]	No
M-03	40 y	1	c.103C>T p. (Gln35*)	M	—	PHP-Ia	2.27b	1.36	—	—	40	4400 [1.90]	51 [0.51]	38 [1.79]	No
M-04	2 y 9 mo	1	c.125G>A p. (Arg42His)	M	Mother	PHP-Ia	—	—	—	66	38	2870 [−0.70]	47 [−1.11]	37 [1.91]	No
M-05	10 y 2 mo	1	c.125G>A p. (Arg42His)	F	Mother	PHP-Ia	N	—	59	74	>37	2890 [<0.22]	47 [<−0.24]	35 [<1.60]	No
M-06	1 y 3 mo	int 1	c.139 + 1G>C p. (?)	M	Mother PPHP	PHP-Ia	2.49	2.08	25b	60	38 + 5	2515 [−1.69]	45 [−2.41]	34 [−0.22]	Yes
M-07	4 y 11 mo	int 1	c.139 + 1G>C p. (?)	M	Mother PPHP	PHP-Ia	1.67	3.03	398	67	>37	2250 [<-1.84]	—	—	Yes
M-08	2 mo	int 1	c.139 + 2T>G p. (?)	M	Mother	PHP-Ia	2.49	1.77	30	—	36	2670 [−0.05]	47 [−0.11]	33 [−0.15]	No
M-09	7 y 6 mo	2	c.155G>A p. (Gly52Asp)	F	De novo	PHP-Ia	2.46	1.30	115	53	35	1420 [−2.93]	38 [−3.27]	29 [−2.01]	Yes
M-10	2 y 1 mo	4	c.287T>A p. (Ile96Asn)	F	Mother PHP-Ia	PHP-Ia	2.50	2.33	738	—	38	2470 [−1.46]	43 [−2.79]	32 [−1.27]	Yes
M-11	2 mo	5	c.347C>T p. (Pro116Leu)	M	Mother PHP-Ia	PHP-Ia	2.46	2.70	120	—	38 + 5	2890 [−0.65]	48 [−0.59]	35 [0.49]	No
M-12	29 y	5	c.347C>T p. (Pro116Leu)	F	Mother PPHP	PHP-Ia	2.26	1.45	230	—	39	2600 [−1.63]	47 [−1.16]	36 [1.04]	Yes
M-13	3 mo	5	c.347C>T p. (Pro116Leu)	F	Mother PHP-Ia	PHP-Ia	2.38	1.79	37b	—	39	2450 [−2.09]	44 [−2.79]	33 [−1.23]	Yes
M-14	28 y	6	c.481C>T p. (Arg161Cys)	F	Mother PHP-Ia	PHP-Ia	2.00	0.86	86	—	41	3200 [−0.61]	50 [−0.14]	—	No
M-15a	3 y 2 mo	6	c.481C>T p. (Arg161Cys)	M	Mother PHP-Ia	PHP-Ia	2.43	1.75	130	55	38	2130 [−2.93]	44 [−2.67]	—	Yes
M-16	9 y 9 mo	6	c.497G>A p. (Arg166His)	M	De novo	PHP-Ia	2.49	1.67	75.6	—	38	2560 [−1.56]	47 [−1.11]	34 [−0.57]	Yes
M-17	1 y 1 mo	6	c.510C>G p. (Tyr170*)	M	Mother	PHP-Ia	2.42	1.76	26b	—	35	2850 [0.93]	—	—	No
M-18	4 mo	6	c.529C>T p. (Gln177*)	M	De novo. somatic mosaicism	PHP-Ia, POH-like	2.61	2.02	39b	—	41	3530 [−0.15]	51 [−0.04]	36 [0.51]	No
M-19	8 mo	int 6	c.534–1G>C p. (?)	M	Mother PHP-Ia	PHP-Ia	2.47	2.18	115	79	36	3000 [0.71]	47 [−0.11]	35 [1.20]	No
M-20	2 y 5 mo	07	c.568_571del p. (Asp190Metfs*14)	M	Mother PHP-Ia	PHP-Ia	1.40	2.68	600	67	39	4390 [2.15]	49 [−0.23]	—	No
M-21	1 y 2 mo	07	c.568_571del p. (Asp190Metfs*14)	F	Mother PHP-Ia	PHP-Ia	2.43	—	51.6b	—	38	3050 [0.08]	45 [−1.75]	36 [1.67]	No
M-22	12 y 9 mo	09	c.694C>T p. (Arg232Cys)	F	—	PHP-Ia	2.46	1.10	82.7	71	35	1700 [−1.86]	42 [−1.51]	29.5 [−1.68]	Yes
M-23	9 y 10 mo	09	c.682C>T p. (Gln228*)	M	De novo	PHP-Ia	1.82	2.52	956	—	38	2570 [−1.53]	45 [−2.15]	32 [−1.99]	Yes
M-24	2 y 9 mo	10	c.724G>A p. (Val242Met)	F	Mother	PHP-Ia	2.50	1.70	80.1	61	35 1/2	2760 [1.01]	46 [0.25]	—	No
M-25	5 mo	10	c.838A>G p. (Asn280Asp)	F	Mother PPHP	PHP-Ia	2.39	2.09	226	74	37	3020 [0.52]	47 [−0.24]	36 [1.96]	No
M-26	2 y 11 mo	int 10	c.843–2del p. (?)	F	Mother	PHP-Ia	2.35	2.08	108	49	38	3320 [0.70]	46 [−1.23]	34 [0.20]	No
M-27	8 mo	int 10	c.843–2del p. (?)	F	Mother	PHP-Ia	2.40	2.02	60	73	36	2830 [0.62]	45 [−0.70]	32 [−0.47]	No
M-28	8 mo	int 10	c.843–2del p. (?)	M	Mother	PHP-Ia	2.50	2.28	58	—	37	3100 [0.39]	48 [−0.12]	33 [−0.35]	No
M-29	24 y	13	c.1099G>A p. (Ala367Thr)	F	Maternal germinal mosaicism	PHP-Ia	2.29	1.85	416	—	39	2720 [−1.27]	47 [−1.16]	34 [−0.10]	Yes

Abbreviations: F, female; M, male; —, unknown; N, normal values.

^aSubjects for whom the placentae were analyzed.

^bTreatment with calcium carbonate and vitamin D.

PPHP/POH patients

GNAS mutations were identified in the 26 PPHP/POH patients in most of the 13 exons encoding Gαs, including exon 1 and the splice donor site in intron 1 (n = 7) (Table 1). Only 4 patients (P-06, P-19, P-24, and P-26) had inherited the mutation from their father, whereas 18 patients carried a de novo defect on the paternal allele. For 4 patients without evidence for hormonal resistance (P-01, P-05, P-13, and P-25), parental DNA was not available. Four of the 26 patients (P-10, P-12, P-20, and P-21) were born prematurely (gestational age, <37 wk).

For each patient the Z-scores for W, L, and HC at birth are shown according to the sex and gestational age (Figure 1, A, B, and C, respectively). With the exception of 2 patients (P-06 and P-17), GNAS mutations on the paternal allele were associated with severe retardation of intrauterine development. The average Z-scores (expressed as mean ± SEM, [range], number of individuals) were: −2.69 ± 0.24, [−5.36/−0.01], n = 26 for W (Figure 1A); −2.26 ± 0.18, [−4.33/−0.24], n = 23 for L (Figure 1B); and −1.36 ± 0.22, [−2.84/+0.82], n = 20 for HC (Figure 1C), respectively.

Figure 1.

A–C, Comparison between patients affected by PHP-Ia (circles) and PPHP/POH (squares) for W (A), L (B), and HC (C) at birth. Data for each subject are expressed as Z-score. Open circles and open squares referred to exon 1/intron 1 GNAS mutation in PHP-Ia and PPHP/POH patients, respectively. The horizontal and vertical bars for each group represent the mean ± SEM; statistically significant differences between PPHP/POH (paternal) and PHP-Ia (maternal) patients are indicated for each parameter. D, Birth W of PHP-Ia (circles) and PPHP/POH (squares) patients according to the GNAS mutation in exon 1 or exons 2–13. The birth W was significantly higher for PPHP/POH patients with a GNAS mutation in exon 1/intron 1 than in patients with mutations in the other GNAS exons. However, there was no obvious difference in birth W of PHP-Ia patients with mutations in any of the 13 GNAS exons. ***, Very high significance level; *, high significance level.

For 4 patients with PPHP/POH, pathological examination of the placenta was performed, and all were hypotrophic (Table 3). No significant placental pathology was noted; in particular, there was no macroscopic evidence for inflammation or fibroid necrosis (ie, findings that are usually associated with vascular disease) or placental dysmaturity.

Table 3.

Placental Weights and Gestational Age for 5 Newborns Presenting With Either PPHP/POH or PHP-Ia

Patients	P-12	P-15	P-21	P-22	M-15
Term, wk	33	39	36 + 4	38	38
Weight placenta, g	220	300	226	343	260
Centile	5	<3rd	<3rd	5 to 10	<3rd

PHP-Ia patients

Similarly to the patients with PPHP/POH, we identified GNAS mutations in the 29 patients with PHP-Ia in most of the 13 Gαs-encoding exons, including exon 1 and intron 1 (n = 8) (Table 2). In 21 patients, the mutation was inherited from a mother who had been diagnosed with either PPHP (n = 4) or PHP-Ia (n = 8), or had an unknown phenotype (n = 8), or presented with germinal mosaicism (n = 1). Mutations had occurred de novo in 5 patients, and for 3 patients, parental DNA was not available. Interestingly, we noticed that 7 patients (24%) were born before 37 weeks gestation, and that 12 patients (41%) were small for gestational age (Table 2). The Z-scores for W, L, and HC were: −0.54 ± 0.25, [−2.93/+2.15], n = 29; −1.05 ± 0.2, [−3.27/+0.51], n = 26; and 0.20 ± 0.28, [−2.01/+2.16], n = 22, respectively (Figure 1, A–C). There was no statistical difference for W or L depending on the maternal phenotype, namely PHP-Ia or PPHP/de novo mutations (expressed as mean ± SEM): −0.60 ± 0.57 (n = 8) vs −1.35 ± 0.33 (n = 9) (P = .29), and −1.38 ± 0.44 (n = 8) vs −1.44 ± 0.39 (n = 8) (P = .96), respectively.

Lastly, all 3 parameters (weight, length, and head circumference) were lower in PPHP/POH patients than in PHP-Ia patients (P < .0001; P = .0003; and P = .0004, respectively).

Mutations in GNAS exon 1/intron 1

GNAS exon 1 is unique to Gαs because it does not contribute to the maternally derived NESP55 protein or the paternally derived XLαs; maternal and paternal mutations in this exon could therefore provide important information regarding the contribution of XLαs to intrauterine development. The mean birth weight for PHP-Ia patients carrying a GNAS mutation in exon 1/intron 1 was indistinguishable from that of PHP-Ia patients with mutations in the other exons: −0.50 ± 0.43, [−1.84/+1.90], n = 8 vs −0.55 ± 0.30, [−2.93/+2.15], n = 21; P = .90 (Figure 1D). In contrast, PPHP/POH patients with paternal GNAS mutations in exon 1/intron 1 had birth weights that were significantly higher than those of patients with mutations in the other GNAS exons: −1.80 ± 0.33, [−2.66/−0.01], n = 7, vs −3.01 ± 0.27, [−5.36/−0.83], n = 19; P = .0152 (Figure 1D).

Furthermore, the birth weight was significantly lower in patients with a paternal exon 1/intron 1 mutation compared to patients with mutations anywhere on the maternal allele: −0.54 ± 0.25, [−2.93/+2.15], n = 29, vs −1.80 ± 0.33, [−2.66/−0.01], n = 7; P = .0165 (Figure 1D).

Discussion

For the present study, we collected birth parameters (ie, gestational age, W, L, and HC) of patients with either PHP-Ia or PPHP/POH due to GNAS mutations on the maternal or the paternal allele, respectively. We obtained compelling evidence for the conclusion that heterozygous GNAS mutations contribute significantly to IUGR. When these mutations were located on the paternal allele, IUGR was more pronounced than with mutations on the maternal allele. Importantly, paternal GNAS exon 2–13 mutations were associated with birth W that were significantly lower than those recorded for newborns with paternal GNAS exon 1 mutations. In particular, these latter findings suggest that Gαs haploinsufficiency may not be the only reason for IUGR and that impaired activity of the paternally derived XLαs could also contribute to this phenotype.

GNAS encodes, through the use of alternative first exons, the signaling proteins Gαs and XLαs and the NESP55. The GNAS locus furthermore gives rise to an AS transcript as well as the A/B transcript that contributes to the regulation of Gαs expression and may be translated into an amino-terminally truncated Gαs variant that appears to antagonize the actions of the full-length signaling protein (18). The mRNA encoding Gαs is expressed in most tissues from both parental alleles (2), including human embryonic stem cells and mouse placenta (19, 20). However, in some tissues, like the proximal renal tubules, paternal Gαs expression is silenced (4); consequently, maternal GNAS mutations that affect exons 1–13 lead to little or no Gαs expression. In contrast to the unknown mechanisms restricting Gαs expression to the maternal allele in some tissues, parent-specific methylation at the 4 differentially methylated regions clearly limits transcription of the other GNAS-derived mRNAs to the nonmethylated parental allele (3). Consequently, NESP55 is derived only from the maternal allele, whereas XLαs, AS, and A/B are expressed only from the paternal allele. Methylation of the XLαs differentially methylated region is established very early after fertilization (19).

We observed that the birth W for PPHP/POH patients were significantly higher when the GNAS mutations were located in exon 1/intron 1 than for PPHP/POH patients with mutations in any of the other 12 paternal GNAS exons. Exon 1 is used only for Gαs transcription. The significantly lower birth W of patients with paternal mutations in GNAS exons 2–13 therefore suggests that XLαs could have a growth-promoting role during fetal development. Consistent with the “conflict hypothesis” (21), a lack of the paternally derived XLαs transcript and protein could contribute to the severe placental hypotrophy observed in our PPHP/POH patients. Consistent with this hypothesis, patients presenting with paternal uniparental isodisomy for chromosome 20 (patUPD20) and thus with double dosage of XLαs transcripts have birth weights that are well above the 50th percentile (22). XLαs mimics, when tested in vitro, the role of Gαs with regard to cAMP production. In fact, because XLαs fails to internalize and thus remains in the cell membrane, this G protein variant may have in some tissues a higher efficacy than Gαs (23).

Besides potentially beneficial effects on placental functions, it is conceivable that XLαs mediates specific fetal actions. Analysis of XLαs-null mice revealed that XLαs and Gαs have opposite effects with regard to central nervous system activity (24), as well as energy and glucose metabolism, which is consistent with the conflict hypothesis (25).

Mice lacking Gnas exon 1 on the paternal allele, thus expressing Gαs only from the maternal allele, are lean and recapitulate the leanness observed in mice (E2^m+/p−) with mutations of Gnas exon 2 common to all protein-coding Gnas transcripts (11). Similar to these findings in rodents, patients with paternal exon 1/intron 1 GNAS mutations have lower birth weights than patients with mutations on the maternal allele of this gene. Although our results are based on only a small number of patients, they implicate GNAS imprinting as a requirement for fetal growth.

An intact placenta is critically important for normal fetal growth because it brings into close contact maternal and fetal blood circulation, thereby facilitating nutrient exchange and determining resource allocation. For some of our patients, it was possible to study the placenta, which revealed severe hypotrophy; however, microscopic examination revealed no obvious abnormalities and showed normal vascular development, thus providing no clues regarding the underlying mechanisms.

We can also eliminate a maternal effect associated with hypocalcemia or short stature because the birth weight and length are not different between babies born to PHP-Ia or PPHP mothers or to healthy mothers, (ie, carriers of de novo mutations).

Gαs and XLαs couple different G protein-coupled receptors to the cAMP-generating enzyme adenylyl cyclase, and increased cAMP levels are observed in trophoblast cells, which turn on expression and secretion of numerous mediators (26). In humans, it is well established that increased intracellular cAMP levels contribute to the fusion process associated with syncytium and microvilli formation, 2 processes that are essential for facilitating the secretory and transfer functions of the syncytiotrophoblast (26–29). We speculate that reduced cAMP signaling due to Gαs/XLαs haploinsufficiency alters the normal fusion process leading to metabolic consequences that impair intrauterine development. Consistent with these data, GNAS expression is reduced in the placentae from IUGR newborns, but not in placentae from preeclamptic or term controls (30, 31).

Patients with IUGR due to vascular abnormalities typically show normal cranial development resulting in a normal HC, which is similar to what was observed in our PHP-Ia patients. However, length at birth was more reduced in these patients (Z-score <−1.05) than weight (Z-score <−0.54), suggesting that Gαs haploinsufficiency could have important biological consequences for the developing fetus. Alternatively, these findings could indicate that this signaling protein is expressed predominantly from the maternal allele in some fetal tissues, including the pituitary where reduction in Gαs expression contributes in older PHP-Ia patients to GH deficiency (7).

In conclusion, our current findings in a relatively large cohort of PPHP/POH patients extend previous observations that had suggested more severe IUGR when a GNAS mutation is located on the paternal allele (10, 11). These data indicate also that a paternally derived GNAS transcript, most likely XLαs, could be essential for normal fetal growth and development and that this transcript affects essential placental functions. Consistent with this conclusion, paternal GNAS exon 1/intron 1 mutations, which do not affect XLαs expression or function, are associated with milder growth retardation. Because IUGR was also observed in PPHP/POH patients with exon/intron 1 GNAS mutation, it is likely that paternally derived Gαs contributes also to nutrient exchange between mother and fetus by affecting placental functions and growth, albeit to a lesser extent than XLαs.

It is therefore conceivable that paternally expressed gene products favor the development of the fetus at the expense of the mother, whereas other maternally expressed placental genes favor the well-being of the mother over that of the fetus, as previously proposed (21).