Defective signaling through plexin-A1 compromises the development of the peripheral olfactory system and neuroendocrine reproductive axis in mice

Abstract

The olfacto-genital syndrome (Kallmann syndrome) associates congenital hypogonadism due to gonadotropin-releasing hormone (GnRH) deficiency and anosmia. This is a genetically heterogeneous developmental disease with various modes of transmission, including oligogenic inheritance. Previous reports have involved defective cell signaling by semaphorin-3A in the disease pathogenesis. Here, we report that the embryonic phenotype of Plxna1^-/- mutant mice lacking plexin-A1 (a major receptor of class 3 semaphorins), though not fully penetrant, resembles that of Kallmann syndrome fetuses. Pathohistological analysis indeed showed a strongly abnormal development of the peripheral olfactory system and defective embryonic migration of the neuroendocrine GnRH cells to the hypothalamic brain region in some of the mutant mice, which resulted in reduced fertility in adult males. We thus screened 250 patients for the presence of mutations in PLXNA1, and identified different nonsynonymous mutations (p.V349L, p.V437L, p.R528W, p.H684Y, p.G720E, p.R740H, p.R813H, p.R840Q, p.A854T, p.R897H, p.L1464V, p.K1618T, p.C1744F), all at heterozygous state, in 15 patients. Most of these mutations are predicted to affect plexin-A1 stability or signaling activity based on predictive algorithms and a structural model of the protein. Moreover, in vitro experiments allowed us to show the existence of deleterious effects of eight mutations (including a transcript splicing defect), none of which are expected to result in a complete loss of protein synthesis, targeting, or signaling activity, though. Our findings indicate that signaling insufficiency through plexin-A1 can contribute to the pathogenesis of Kallmann syndrome, and further substantiate the oligogenic pattern of inheritance in this developmental disorder.

Introduction

The olfacto-genital syndrome, also known as Kallmann syndrome (MIM 147950, 244200, 308700, 610628, 612370, 612702), is an inherited neurodevelopmental disorder defined as the association of congenital hypogonadotropic hypogonadism, due to gonadotropin-releasing hormone (GnRH) deficiency, with the inability to smell (anosmia or hyposmia), related to the abnormal development of the olfactory nerves and olfactory bulbs. Patients usually do not undergo spontaneous puberty, and later present with primary infertility. The genetics of Kallmann syndrome involves various modes of disease transmission, specifically, autosomal recessive, autosomal dominant with incomplete penetrance, X chromosome-linked, and also oligogenic inheritance (1,2). Studies of fetuses with olfactory bulb agenesis have shown that the reproductive phenotype of Kallmann syndrome results from a pathological sequence in embryonic life, whereby premature interruption of the olfactory, vomeronasal, and terminal nerve fibers disrupts the migration of neuroendocrine GnRH cells, which normally migrate from the nose to the brain along these nerve fibers (3–6). What causes the primary failure of these fibers to make proper contact with the forebrain is, however, still unknown. Since Kallmann syndrome is genetically heterogeneous, identification of the various genes involved and subsequent analysis of the appropriate animal models are expected to provide valuable clues at the molecular and cellular levels. Such genetic studies have indeed revealed the critical role of fibroblast growth factor (FGF)-signaling and prokineticin-signaling in the normal developmental process (7). However, fewer than half of the Kallmann syndrome patients have mutations in any of the causal genes identified so far, which include ANOS1 (ID: 3730) (8–10), FGFR1 (ID: 2260) and FGF8 (ID: 2253) (11,12), PROKR2 (ID: 128674) and PROK2 (ID: 60675) (13), HS6ST1 (ID: 9394) (14), FEZF1 (ID: 389549) (15), SOX10 (ID: 6663) (16), and CHD7 (ID: 55636) (17,18). Therefore, current efforts concentrate on the identification of other genes that contribute to this disorder.

One strategy is based on a close examination of targeted mutant mice that may reproduce the human abnormal phenotype. In this respect, various mouse mutants defective for proteins involved in semaphorin-signaling have been given careful consideration since semaphorin-3A (Sema3A) was shown to be involved in the guidance of olfactory and vomeronasal axons, more than 15 years ago (19,20). Class 3 semaphorins are secreted proteins that signal predominantly through transmembrane receptors called plexins (divided into four classes, A-D, on the basis of structural criteria), and with the contribution of neuropilin (Nrp1 or Nrp2) coreceptors (21–23). In mice GnRH cells begin to migrate from the epithelium of the vomeronasal organ through the frontonasal region around embryonic day 11.5 (E11.5), along the vomeronasal and terminal nerve fibers (24,25). From E13.5 on, GnRH cells penetrate the rostral forebrain with the central roots of these nerves and migrate caudally in the primitive cerebral hemispheres along a transient ventral branch of the vomeronasal nerve or along the fibers of the terminal nerve (26,27). They ultimately stop in the preoptic and hypothalamic regions, and project their axons to the median eminence where GnRH secretion will soon take place. The migration and axonal projection of neuroendocrine GnRH cells are completed before birth. Mutant mice lacking Sema3A (Sema3a^-/-) or Nrp2 (Nrp2^-/-), or harboring missense mutations affecting the semaphorin-binding domain of Nrp1 (Nrp1^sema/sema) displayed variable anomalies during the critical period of development corresponding to the outgrowth of the olfactory nerve fibers from the olfactory epithelium and the concomitant migration of neuroendocrine GnRH cells. In Sema3a^-/- mice, substantial errors in axon guidance of vomeronasal and terminal nerve fibers were associated with a strongly decreased number of GnRH cells in the hypothalamic region and the accumulation of these cells in the frontonasal mesenchyme (19,28). Likewise, Nrp1^sema/semaNrp2^-/- double mutants almost completely lacked GnRH cells in the basal forebrain (28). Nrp1^sema/sema and Nrp2^-/- single mutants showed milder defects in the patterning of vomeronasal/terminal nerve fibers and moderately decreased numbers of GnRH cells in the basal forebrain (28–30). The fertility of these mice was, however, reduced (20,30,31). Since it has been shown that Sema3A, Nrp1, and Nrp2 play important roles in vomeronasal/terminal axon guidance and GnRH cell migration, we investigated the role of plexin-A1 (PlxnA1), a major receptor of Sema3A, in this developmental process.

Results and Discussion

Plexin-A1 is expressed in the frontonasal region of mouse embryos, along the migratory path of neuroendocrine GnRH cells

Differential expression of plexin A subfamily members has previously been reported in the mouse nervous system (32). However, the study was not focused on the developing olfactory system, and was carried out from E16.5, at a time when GnRH cells have almost completed their migration to the preoptic area of the brain. We showed that Plxna1 is expressed in the frontonasal region during the period of GnRH cell migration, by in situ hybridization using an antisense RNA probe. On E11.5, Plxna1 transcripts were detected in the vomeronasal organ, along the vomeronasal nerves, in the migratory mass (a transient embryonic structure made of various types of neuronal and non-neuronal cells that accumulate in the frontonasal mesenchyme (33)), and in the neuroepithelium of the rostral forebrain (Fig. 1A). On E13.5, Plxna1 transcripts were also present in the olfactory epithelium (Fig. 1B). Immunohistofluorescence experiments with antibodies specific to PlxnA1 showed a strong labeling along the vomeronasal and terminal nerve fibers and in the migratory mass. Coimmunolabeling with an anti-βIII tubulin antibody (a neuronal marker) and an anti-GnRH1 antibody showed that Plxna1 expressing cells were neuronal cells (Fig. 1C), and that some of them were GnRH-synthesizing neurons (Fig. 1D). We also confirmed the previously reported presence of Nrp1 and Nrp2 along the vomeronasal/terminal nerve pathway (28–30). However, while the Nrp2 immunostaining was clearly located along the nerve fibers, Nrp1 was also detected in neighboring cells (Fig. 1E and F). Some of these cells could be unambiguously identified as olfactory ensheathing cells, a subpopulation of glial cells involved in the growth and guidance of the olfactory axons, by their coimmunoreactivity for the brain lipid-binding protein (BLBP) (34,35) (Fig. 1G). These observations indicated that Sema3A holoreceptors composed of Nrp1 or Nrp2 and the signal-transducing protein PlxnA1 are present along the vomeronasal/terminal nerve pathway, therefore suggesting that Sema3A signaling through PlxnA1 could be involved in the guidance of these nerve fibers and consequently (perhaps even directly) of migrating GnRH cells.

Figure 1

Plxna1, Nrp1, and Nrp2 expression profiles along the GnRH neuron migratory pathway in the mouse embryonic frontonasal region. (A,B) In situ hybridization with a Plxna1 antisense probe on coronal and sagittal sections of the frontonasal region in E11.5 (A) and E13.5 (B) embryos, respectively. Plxna1 is expressed in the vomeronasal organ (vno), along the vomeronasal nerves (vnn), in the migratory mass (mm), and in the neuroepithelium of the rostral forebrain (fb) on E11.5 (arrows), and also in the olfactory epithelium (oe) on E13.5. (C) Double immunolabeling with an anti-PlxnA1 antibody (green) and a pan-neuronal marker (Tuj1, red) on E11.5. The arrow points to the colocalization of the immunostainings along the vomeronasal nerve fibers. (D) Double immunolabeling with an anti-PlxnA1 antibody (green) and an anti-GnRH1 antibody (red) on E12.5 shows that some of the PlxnA1-immunoreactive cells are GnRH-synthesizing neurons (arrowheads). (E) Double immunolabeling with anti-Nrp1 (red) and anti-Nrp2 (green) antibodies confirms the presence of both coreceptors in the frontonasal region on E12.5, and shows their partial colocalization along the vomeronasal nerve fibers. (F) Double immunolabeling with anti-Nrp1 (red) and anti-PlxnA1 (green) antibodies shows the colocalization of these receptors along the vomeronasal nerve fibers (arrow). (G) Nrp1 (red) is also present in BLBP-immunoreactive (green) olfactory ensheathing cells (arrows). Other abbreviation: bv, blood vessel. Scale bar = 200 µm in (A) and (C), 400 µm in (B), 30 µm in (D), 100 µm in (E) and (F), and 50 µm in (G).

Olfactory sensory axons and neuroendocrine GnRH cells fail to reach the forebrain in some but not all Plxna1^-/- mouse embryos

Previous analyses of vomeronasal axons and GnRH cell migration in Sema3a^-/-, Nrp1^sema/sema, or Nrp2^-/- simple or compound mutant mice have revealed Kallmann syndrome-like phenotypes, suggesting that a defective cell signaling activity of Sema3A can be involved in the disease (28–30). To investigate the role of PlxnA1 in Sema3A signaling during early development of the peripheral olfactory system, we undertook a detailed analysis of Plxna1 knockout mice (Plxna1^-/-) (36), which are viable. We first looked at the formation of olfactory nerves and the location of GnRH cells during the period of their migration. In E14.5 wild-type embryos, olfactory nerve fibers had already contacted the rostral forebrain, olfactory bulbs had formed, and most GnRH cells had entered or were entering the forebrain, caudally to the olfactory bulbs, subsequently turning posteriorly toward the ventral forebrain (Fig. 2A–C). The absence of PlxnA1 resulted in a strongly abnormal phenotype in 5 of the 18 Plxna1^-/- embryos analyzed, while the 13 remaining embryos had a phenotype similar to the wild-type embryos. In the most severely affected embryos, olfactory nerve fibers and associated nerve fibers did not contact the forebrain, and instead terminated in the upper frontonasal region. As a consequence, the olfactory bulbs had not formed and GnRH cells had remained outside of the brain, in the frontonasal region, where they had accumulated in the migratory mass (Fig. 2A–C). No GnRH cells were observed in the brain of these embryos. This drastic embryonic phenotype, although not fully penetrant, was remarkable since most of the Kallmann syndrome mouse models defective for semaphorin signaling so far only exhibit a partial misguidance of GnRH cells in the brain (28–30). Notably, one of the Plxna1^-/-embryos showed a mixed phenotype, having all the aforementioned defects of the olfactory structures and GnRH cell migration on the right side, but normally developed olfactory nerve and olfactory bulb on the left side. However, misguided vomeronasal/terminal axons were detected in the left cerebral hemisphere, growing dorsally toward the cortex instead of ventrally as in wild-type embryos, and GnRH cells were spotted along these misoriented vomeronasal/terminal axons (Fig. 2B, right panel, and data not shown). We did not find the drastic phenotype described above in any of the 8 Plxna1^-/- mice analyzed on P0, which could result from an early death of the most severely affected mice soon after birth owing to starvation. Suckling is indeed highly dependent on olfaction in rodents (37), and we noticed that most of the Plxna1^-/- newborn mice had little or no milk in their stomach. Nevertheless, the number of GnRH cells present at their normal location in the ventral forebrain on P0 was on average 27% lower in Plxna1^-/- mice (n = 8) than in wild-type mice (n = 6) (Student’s t-test, P < 0.01). Accordingly, GnRH immunolabeling in the median eminence of the hypothalamus, the axonal projection field of neuroendocrine GnRH cells, was also decreased on P0 (Fig. 2D). In addition, a higher number of these cells were ectopically located in the dorsal or lateral parts of the forebrain in the mutant mice than in wild-type mice (Student’s t-test, P < 0.01) (Fig. 2E). We conclude that cell signaling through PlxnA1 is involved in the formation of the neuroendocrine reproductive axis by organizing the ‘olfactory-associated’ axonal tracts that guide the migrating GnRH-synthesizing neurons from the olfactory epithelium to the rostral telencephalon, and from there, to their final location in the diencephalon.

Figure 2

Abnormal targeting of olfactory, vomeronasal, and terminal nerve fibers and neuroendocrine GnRH cells in Plxna1^-/- embryos and newborn mice. (A–C) Coronal (A) and sagittal (B,C) sections of the frontonasal region in wild-type (left panels) and Plxna1^-/- (middle and right panels) embryos on E14.5. (A) Immunostaining of neuronal cells (anti-βIII-tubulin, red) and GnRH-synthesizing cells (green). Note the abnormal location of the GnRH-immunoreactive cells in the Plxna1^-/- embryo: these cells are not entering the rostral forebrain as in the wild-type embryo (arrowheads), and instead have accumulated in the migratory mass. (B) Olfactory, vomeronasal, and terminal nerve fibers are labeled with an anti-peripherin antibody. After entering the rostral forebrain, the fibers of the terminal nerve and/or caudal branch of the vomeronasal nerve (filled arrowhead) travel posteriorly toward the diencephalon in the wild-type embryo (left panel). In the Plxna1^-/- embryos harboring the most severe phenotype, the olfactory nerves have not contacted the brain and consequently, the olfactory bulbs have not formed (middle panel). Notably, one of these embryos (right panel) has an almost normal left cerebral hemisphere with a well-developed olfactory bulb and caudal branch of the vomeronasal nerve on this side (filled arrowhead); however, a small proportion of these nerve fibers make their way ectopically toward the dorsal forebrain (empty arrowhead). The inset is a higher magnification of the boxed area, showing the normal and ectopic nerve fibers in the left forebrain. (C) GnRH immunostaining on alternate sections of the same embryos as in (B). In the Plxna1^-/- embryo, GnRH cells accumulate in the migratory mass, beneath the forebrain, instead of entering the forebrain and migrating posteriorly. (D) GnRH immunostaining on coronal sections of the median eminence (arrows), showing the axonal projections of neuroendocrine GnRH cells in wild-type and Plxna1^-/- newborn (P0) mice. The GnRH-immunoreactive area is reduced in the Plxna1^-/- mouse. (E) Quantitative analysis of GnRH cell distribution in wild-type and Plxna1^-/- mice on P0. Abbreviations: fb, forebrain; mm, migratory mass; ob, olfactory bulb; oe, olfactory epithelium; V3, third ventricle. Scale bar = 200 µm in (A) and (C), 400 µm in (B), and 100 µm in (D).

Decreased fertility in Plxna1^-/- males

Fertility analysis was carried out in male and female mutant mice by breeding Plxna1^-/- young adults (P40) with wild-type littermates during 7 months. Both the number of litters and their size were monitored. No difference in litter size was noted between the breeding pairs whatever the genotypes. In addition, the fertility index (number of litters per month) values in Plxna1^-/- females were similar to those in wild-type females (1.08 ± 0.06 vs. 1.06 ± 0.06, mean ± s.e.m., respectively, n = 6 in both groups), and so were the GnRH immunostainings in the median eminence of these mice (Fig. 3A and B and data not shown). By contrast, the mean fertility index value in Plxna1^-/- males was significantly reduced (0.57 ± 0.12 vs. 1.06 ± 0.05 in wild-type males; Student’s t-test, P < 0.001) (Fig. 3A). Among the 15 Plxna1^-/- males monitored, 5 had normal fertility index values (1.13 ± 0.04), whereas 5 had reduced fertility index values between 0.28 and 0.8 (0.58 ± 0.09), and 5 were unable to reproduce. Plxna1^-/- males with an impaired fertility (fertility index value < 0.8) showed moderately decreased GnRH immunostainings in the median eminence of the hypothalamus compared to wild-type males or Plxna1^-/- males with fertility index values ≥ 1 (Fig. 3B and C; Student’s t-test, P < 0.05 for both comparisons). In addition, the seminal vesicles, which produce a substantial proportion of the seminal fluid and are highly dependent on androgens to reach their adult size, were hypoplastic in 8 out of 9 Plxna1^-/- males analyzed (Fig. 3D). The testes, however, had fairly normal sizes in all Plxna1^-/- males (13 900 ± 469 pixels, mean ± s.e.m., vs. 14 700 ± 795 in wild-type mice, and Fig. 3D). Such a different impact of the Plxna1 knockout on fertility in females (fertile) and males (often hypofertile or infertile) is reminiscent of previous reports on Nrp1 or Nrp2 deficient mice (28,30,31), and has so far remained unexplained, but might be somehow related to the long reported three to five times higher proportion of males than females among Kallmann syndrome patients (1,2). Moreover, the moderate decrease in the number of neuroendocrine GnRH cells and of their axonal projections shown here may not be sufficient to account for the complete infertility of some Plxna1^-/- males: an additional defect in GnRH neurosecretion and/or testosterone secretion in the testes might also be involved (38). Indeed, the contrast between the normal testis volume, which largely depends on FSH levels, and the markedly reduced size of the androgen-dependent seminal vesicles in these mice argues in favor of an additional defect of testosterone secretion or action.

Figure 3

Impaired reproductive capacity of Plxna1^-/- males. (A) Comparison of the fertility index values in wild-type and Plxna1^-/- mice. Plxna1^-/- males, but not females have decreased fertility index values. (B) Quantitative analysis of the GnRH-immunoreactive area in the median eminence of wild-type and Plxna1^-/- mice. Plxna1^-/- males, but not females show a 20% decrease of the area stained with the anti-GnRH antibody. (C) Coronal sections of the median eminence region stained with an anti-GnRH antibody in wild-type and Plxna1^-/- adult males. There are fewer axonal projections of GnRH neuroendocrine cells in Plxna1^-/- than in wild-type males (arrows). (D) Genitalia of wild-type and Plxna1^-/- adult males. In Plxna1^-/- males the testes (t) have normal sizes, but the seminal vesicles (sv) are hypoplastic. Other abbreviation: V3, third ventricle. Scale bar = 200 µm in (C), and 10 mm in (D).

PLXNA1 mutations in Kallmann syndrome patients

Plxna1, Nrp1, and Nrp2 expression profiles in the embryonic nasal region of the mouse, the Kallmann syndrome-like phenotypes of the corresponding mutant mice, and the previously reported involvement of SEMA3A mutations in Kallmann syndrome (28–30,39), prompted us to search for mutations of PLXNA1 (ID 5361), NRP1 (ID 8829), NRP2 (ID 8828), and SEMA3A (ID 10371) in Kallmann syndrome patients. We thus sequenced all the coding exons and abutting splice sites of these genes in 237 unrelated patients who did not have identified mutations in nine causal genes routinely tested: ANOS1, FGFR1, FGF8, PROKR2, PROK2, HS6ST1, FEZF1, SOX10, and CHD7 (2). Different missense mutations of PLXNA1 were found in 15 patients, all at heterozygous state. Eight mutations (p.R528W, p.H684Y, p.G720E, p.R740H, p.R840Q, p.L1464V, p.K1618T, p.C1744F) have not been reported in the Exome Variant Server database, while the other five mutations (p.V349L, p.V437L, p.R813H, p.A854T, p.R897H) have allele frequencies below 0.2% in the European American population. All the mutations except two (p.V349L and p.R897H) are predicted to be deleterious/damaging or possibly damaging by SIFT and/or PolyPhen-2 algorithms (Table 1 and Fig. 4). Deleterious effects of the mutations on the protein stability and dynamics were also suspected from the analysis of models of the human PlxnA1 ecto- and cyto-domains based on previously reported crystal structures ((23,40,41) and Supplementary Material, Fig. S1). In addition, the c.5231G > T (p.C1744F) mutation, which affects the last nucleotide of exon 28, is predicted to affect splicing of the primary transcript by three different algorithms (MaxEntScan::scoresplice, NNSplice, and Human Splicing Finder). Three missense mutations of NRP1 (p.R207H, p.G734S, p.I857M) and three missense mutations of NRP2 (p.T242M, p.R334C, p.G484S) were found at heterozygous state in 5 patients and in 4 patients, respectively. These mutations either are absent from the Exome Variant Server database or have allele frequencies below 0.2% in the European American population. All six mutations are predicted to have a deleterious/damaging effect on the protein according to SIFT and/or PolyPhen-2 (Table 1 and Fig. 4). Notably, one of these patients carried monoallelic mutations of PLXNA1 (p.R740H) and NRP1 (p.I857M), both of which, however, were inherited from his clinically unaffected mother, therefore indicating that the two mutations are not sufficient to produce the disease phenotype. Finally, 12 patients carried missense mutations of SEMA3A (p.N153S, p.M342I, p.V435I, p.T506A), all at heterozygous state. The p.N153S, p.M342I, and p.V435I mutations are predicted to be deleterious according to SIFT but not PolyPhen-2, while p.T506A is predicted to be tolerated/benign by both algorithms (Table 1 and Fig. 4). In addition, p.V435I and p.N153S have previously been shown to affect protein secretion and signaling activity, respectively (30). We also searched for mutations of PLXNA1, NRP1, or NRP2 in 13 patients who had previously been diagnosed with monoallelic mutations of SEMA3A (30), and did not find additional mutations in any of them.

Figure 4

Diagrams of PlxnA1, Nrp1, Nrp2, and Sema3A, showing the positions of the missense mutations found in Kallmann syndrome patients. Abbreviations: sema, semaphorin; PSI, plexin/semaphorin/integrin; IPT, immunoglobulin-like fold in plexins/transcription factors; TM, transmembrane domain; GAP₁ & GAP₂, Ras GTPase-activating protein domain (formed by these two GAP homology regions folded together); RBD, Rho family GTPase-binding domain; CUB, complement components C1r and C1s/uEGF/BMP-1; CF5 and CF8, coagulation factors V and VIII; MAM, meprin/A5 antigen/receptor-like protein tyrosine phosphatase μ; Ig, immunoglobulin-like.

Table 1

PLXNA1, NRP1, NRP2, and SEMA3A mutations identified in Kallmann syndrome patients.

Number (gender) of patients	Nucleotide change (cDNA)	Predicted amino acid change	Protein domain	SIFT score	PolyPhen-2 score	dbSNP	Allele frequency in control subjects
							EVS	ExAC
2 (M)	PLXNA1:c.1045G>C	V349L	sema	0.26	0.01	rs41266465	0.07%	0.13%
1 (M)*	PLXNA1:c.1309G>C	V437L	sema	0.04	0.16	rs150194563	0.01%	0.008%
1 (M)	PLXNA1:c.1582C>T	R528W	PSI1	0	0.93			0.004%
1 (M)	PLXNA1:c.2050C>T	H684Y	PSI2	0.01	0.93
1 (M)	PLXNA1:c.2159G>A	G720E	IPT2	0	0.45			0.004%
2 (M + F)	PLXNA1:c.2219G>A	R740H	IPT2	0.05	0.70
1 (F)	PLXNA1:c.2438G>A	R813H	PSI3	0.01	0.32	rs368277644		0.003%
1 (M)	PLXNA1:c.2519G>A	R840Q	PSI3	0.36	0.88			0.003%
1 (M)	PLXNA1:c.2560G>A	A854T	PSI3	0.16	0.48	rs371145234	0.04%	0.004%
2 (M)	PLXNA1:c.2690G>A	R897H	IPT3	0.13	0.01	rs138276286	0.14%	0.10%
1 (M)	PLXNA1:c.4390C>G	L1464V	RasGAP	0	0.99
1 (F)	PLXNA1:c.4853A>C	K1618T	RBD	0.04	0.85
1 (M)*	PLXNA1:c.5231G>T	C1744F	RasGAP	0.01	0.92
1 (M)	NRP1:c.620G>A	R207H	CUB	0	0.99	rs148308681	0.01%	0.007%
1 (M)	NRP1:c.2200G>A	G734S	MAM	0	1			0.0008%
3 (M + 2 F)	NRP1:c.2571C>G	I857M	transmembrane	0	1	rs144845322	0.05%	0.02%
1 (M)	NRP2:c.725C>T	T242M	CUB	0	1	rs138033888	0.10%	0.05%
2 (M)	NRP2:c.1000C>T	R334C	CF1	0	0.82	rs114144673	0.23%	0.16%
1 (F)	NRP2:c.1450G>A	G484S	CF2	0	0.99			0.03%
1 (M)	SEMA3A:c.458A>C	N153S	sema	0	0.06	rs139295139	0.04%	0.23%
1 (M)	SEMA3A:c.1026G>A	M342I	sema	0	0.01	rs142186928	0.01%	0.04%
9 (8 M + F)	SEMA3A:c.1303G>A	V435I	sema	0	0.01	rs147436181	1.4%	0.01%
1 (F)	SEMA3A:c.1516A>G	T506A	sema	0.49	0	rs369182739	0.01%	0.002%

*

 These two PLXNA1 mutations were found in the same patient.

Cell signaling through receptors of the plexin family involves multiple pathways including GTPases, kinases, and oxidoreductases (42,43). To assess the effect of the PLXNA1 mutations on the signaling activity of the receptor, we performed a cell collapse assay on cotransfected COS7 cells producing Nrp1 and either wild-type PlxnA1 or PlxnA1 harboring each of the 13 different mutations found in the patients, in the presence of the ligand Sema3A. The COS7 cell collapse assay has previously been used as a heterologous experimental system for axonal growth cone collapse (23,44). We did not observe a reduced cell collapse response with any of the mutations, which might argue against a major deleterious effect of these mutations on cell signaling (data not shown). However, the mutations may affect signaling pathways that depend on semaphorins other than Sema3A, or on coreceptors other than Nrp1, both of which have been shown to be involved in GnRH cell migration (29,45) but were not explored by the cell collapse assay. In addition, PlxnA1 has been shown to trigger various signaling pathways that presumably make uneven contributions to the response of axonal growth cones (including pathways through Rho-family GTPases, Ras-family GTPases, mitogen-activated protein kinase, phosphatidylinositol 3-kinase, Src-family kinases, or collapse response mediator proteins (42)), and these signaling pathways might be differentially affected by the different PLXNA1 mutations, an issue that could be addressed by additional experiments specifically assessing each pathway.

We also tested the effect of the PLXNA1 mutations on receptor synthesis and targeting to the plasma membrane, by ELISA on cotransfected HEK293T cells producing both PlxnA1 and Nrp1. This assay provided us with semi-quantitative estimates of the PlxnA1 whole cell content (ELISA after permeabilization of the plasma membrane) and of the amount of receptor present at the cell surface (ELISA in the absence of membrane permeabilization). Seven mutations (R528W, H684Y, G720E, R740H, R840Q, A854T, and K1618T) resulted in reduced total amounts of the protein (one-way ANOVA test, P < 0.01 for all comparisons). In addition, the A854T mutation also affected the cell surface targeting of the receptor (Kruskal-Wallis test, P < 0.05) (Fig. 5A and B). Finally, we tried to confirm experimentally the predicted effect of the PLXNA1:c.5231G > T (p.C1744F) mutation on the splicing of the primary transcript, using a minigene comprised of exons 28, 29 and introns 27-29 in transfected COS7 cells. We found that the presence of the mutation indeed increased the ratio of the abnormally spliced transcript retaining intron 28 to the normal transcript by a factor of 3 (Fig. 5C). Retention of intron 28 in the mature transcript is predicted to result in a premature stop codon at codon position +1 from the mutated codon.

Figure 5

Functional assays on the PLXNA1 mutations. (A) Bar chart illustrating the total amount of the Flag-tagged PlxnA1 mutant proteins (normalized to wild-type protein) in cotransfected HEK293T cells producing both Flag-tagged PlxnA1 and Nrp1, quantified by ELISA after cell permeabilization. (B) Bar chart illustrating the amount of the Flag-tagged PlxnA1 wild-type or mutant protein present at the cell surface (quantified by ELISA on non-permeabilized cells), expressed as a percentage of the total amount of the protein. Data are mean ± s.e.m. values from three independent experiments, each performed in triplicate. *P < 0.05; **P < 0.01; ***P < 0.001 (one way ANOVA or Kruskal-Wallis’s multiple comparison tests). (C) Diagram of the PLXNA1 ‘minigene’ construct, and of the major two ‘minitranscripts’ produced in transfected NIH3T3 cells. The black boxes correspond to exon sequences. The positions of the forward and reverse primers used for RT-PCR analysis are indicated by horizontal arrows. The arrowhead indicates the position of the PLXNA1:c.5231G>T mutation. The ratio of the abnormally spliced transcript containing intron 28 (2578 bp amplicon) to the correctly spliced transcript (604 bp amplicon) was quantified after agarose gel electrophoresis of the RT-PCR products derived from the ‘minitranscripts’. This ratio is only 1.5 for the wild-type minigene (lane 1), but is 4.5 for the minigene containing the mutation (lane 2). The first two lanes on the left side of the gel contain molecular size markers (100 bp DNA ladder and EcoRI/HindIII digest of phage λ DNA, respectively).

From the developmental anomalies found in Plxna1 knockout mice and the functional analyses of the PLXNA1 mutations identified in some Kallmann syndrome patients, we therefore suggest that insufficient cell signaling through PlxnA1 is involved in the abnormal olfactory and reproductive phenotypes of these patients. Together, the in vitro experiments allowed us to show the existence of deleterious effects of eight out of thirteen PLXNA1 mutations even though none are expected to result in a complete loss of protein synthesis, targeting, or signaling activity, again indicating a non-monogenic mode of transmission of the disease phenotype in the patients (30) and calling for additional studies to clarify the complex genetics of this developmental disorder.

Materials and Methods

Animals

All experiments on mice were performed in accordance with the French and European Community guidelines, and reviewed by the local ethics committee of Université Paris-Descartes (86/809/EEC and 01319.03). The Plxna1^-/- mouse line (36) was established from breeding pairs of animals obtained from V. Castellani on a mixed genetic background (CD-1 and C57BL/6). Mouse embryos were obtained from timed mating of Plxna1^+/- heterozygous mice or outbred Swiss mice, with the date of vaginal plug counted as E0.5. In the fertility assays, at least 6 couples of mice for each combination of genotypes (Plxna1^+/+ ♂ x Plxna1^+/+ ♀; Plxna1^+/+ ♂ x Plxna1^-/- ♀; Plxna1^-/-♂ x Plxna1^+/+ ♀; Plxna1^-/-♂ x Plxna1^-/-♀) were crossed during a minimum period of 7 months, and the number and size of the litters were monitored.

In situ hybridization

E11.5 and E13.5 mouse embryos were fixed by immersion in 4% paraformaldehyde/phosphate buffered saline (PBS), pH 7.4, for 3 h. Embryos were then washed in PBS, incubated in 30% sucrose (PBS), and embedded in 7.5% gelatin, 10% sucrose (PBS). They were frozen and serially sectioned in the frontal or horizontal plane, using a cryostat. In situ hybridization was carried out on cryosections with a digoxigenin-labeled Plxna1 antisense riboprobe (46) or a sense riboprobe as negative control, according to the protocol published by Strähle et al. (47), with the following modifications: hybridization was done at 72°C; sections were blocked in a blocking solution (Tris 0.1 M, NaCl 0.15 M, 0.1% tween 20, pH 7.5) containing 10% heat-inactivated normal goat serum; they were then incubated overnight at room temperature with an alkaline phosphatase-conjugated anti-digoxigenin antibody (1:5000, Roche) in the blocking solution containing 1% normal goat serum, and stained several hours with 5-bromo-4-chloro-3-indolyl phosphate (BCIP)/nitroblue tetrazolium (NBT) (Roche) used as a chromogenic substrate to detect the alkaline phosphatase activity, according to the published protocol (47).

Immunohistofluorescence experiments and counting of neuroendocrine GnRH cells

Cryosections (20 µm thickness) were permeabilized with 0.1% Triton X-100 in PBS, blocked with 1% normal goat serum in PBS, and incubated with the antibodies in PBS containing 0.1% Triton X-100 and 1% normal goat serum, either overnight (primary antibodies) or for 2 to 3 h (secondary antibodies), at room temperature. The following primary antibodies were used: rabbit anti-GnRH1 (1:500 dilution, Abcam ab5617), rabbit anti-peripherin (1:1000 dilution, Millipore AB1530), rabbit anti-plexinA1 (1:250 dilution, Chemicon International AB9602), goat anti-neuropilin1 (1:400 dilution, R&D systems AF566), rabbit anti-neuropilin2 (1:400 dilution, Santa Cruz Biotechnology sc-5542), mouse anti-βIII tubulin (TUJ1) (1:1000 dilution, Promega), rabbit anti-BLBP (1/500 dilution, Millipore ABN14). Immunoreactive structures were detected with the appropriate secondary antibodies conjugated to Alexa 488 or Alexa 594 fluorophore (1:1000 dilution, Molecular Probes). DAPI (1 μg/ml, Vector) was used to visualize the cell nuclei.

GnRH1-immunoreactive cell bodies were counted in newborn (P0) mouse embryos, on every second section (frontal plane) of the forebrain. Likewise, the area of axonal projection of the neuroendocrine GnRH cells was quantified (in pixels) in adult mice on every second section (frontal plane) of the hypothalamus median eminence. Statistical analyses were performed on six P0 and three adult mice of each genotype, using Student’s t test.

Patients and DNA sequencing

A total of 250 Kallmann syndrome patients (179 males and 71 females), who displayed both anosmia or severe hyposmia (based on self-report, questioning, or formal olfactory testing) and congenital hypogonadotropic hypogonadism, underwent genetic analysis. The patients had a reported Caucasian origin, with few exceptions. Clinical data were obtained from the referring endocrinologists. Informed consent was obtained from all individuals included in the study.

Genomic DNA was extracted either from the patient’s blood cells or from a derived lymphoblastoid cell line. Mutations were sought in the coding exons and abutting splice sites of PLXNA1 (31 exons), NRP1 (17 exons), NRP2 (16 exons), and SEMA3A (17 exons), using the Ion Torrent semiconductor sequencing technique. Primers were designed with the Ampliseq designer software. Library preparation was carried out with the Ion Plus fragment library kit, with 50 ng of genomic DNA. Adapter ligation, nick repair, and amplification were performed according to the Ion Torrent protocol (Life Technologies). Emulsion PCR and enrichment steps were carried out with the Ion One Touch template kit. Sequencing of the amplicon libraries was carried out on the Ion Torrent PGM system with 316 chips, and bar coding with the Ion Xpress bar code adapters kit. The Ion sequencing kit version 2 was used for all sequencing reactions, according to the recommended protocol. After sequencing, reads were mapped to the human genome 19 assembly with the Torrent mapping alignment program. Single-nucleotide variants and small insertions/deletions (indels) were identified with the Torrent Variant Caller (Life Technologies) and Nextgene software. We used Sanger sequencing on new PCR products to confirm the mutations.

Nomenclature and in silico analyses of the mutations

The GenBank accession numbers of the relevant cDNA sequences used for mutation nomenclature are: NM_032242 (PLXNA1), NM_045259 (NRP1), NM_201266 (NRP2), and NM_006080 (SEMA3A).

Two different algorithms were used to predict the effects of the missense variants: Sorting Intolerant From Tolerant (SIFT) (sift.jcvi.org/www/SIFT_enst_submit.html) and PolyPhen-2 (genetics.bwh.harvard.edu/pph2/). Both algorithms result in scores ranging from 0 to 1. A mutation is predicted as “deleterious” by SIFT if its score is not greater than 0.05; otherwise it is predicted as “tolerated”. A mutation is predicted as “possibly damaging” by PolyPhen-2 if its score is greater than 0.15, and as “probably damaging” if it is greater than 0.85; otherwise it is predicted as “benign”. The values of SIFT and PolyPhen-2 scores predicting possible or likely deleterious effects of the mutations are highlighted in bold in Table 1. The dbSNP reference numbers of the missense variants (when available), their presence and allelic frequencies in the approximately 4000 European American (EA) control individuals of the Exome Variant Server (EVS) database and 60000 individuals of the Exome Aggregation Consortium (ExAC) database are also indicated in Table 1. The effect of the PLXNA1:c.5231G > T mutation presumably affecting splicing of the primary transcript was predicted by the MaxEntScan::scoresplice (genes.mit.edu/burgelab/maxent/Xmaxentscan_scoreseq.html), NNSplice (omictools.com/nnsplice-tool), and Human Splicing Finder (rd-connect.eu/tools-resources/human-splicing-finder) algorithms. To study the consequences of the missense mutations on the protein stability and dynamics, we made homology models of human PlxnA1 with Modeller 9.17 (48) and the loop optimization method, based on crystal structures of the ectodomain and cytodomain of murine PlxnA1 and PlxnA3, respectively (Protein Data Bank IDs: 5L56 (23) and 3IG3 (41)). In addition, we used a crystal structure of the murine PlxnA1 cytodomain in complex with Rac1 (Protein Data Bank ID: 3RYT (40)) as a template to model this protein complex. The model was constructed by superimposition of the cytodomain model and the cytodomain present in the crystal structure, first by sequence alignment and later by matching close Cα atoms of the model with Cα atoms of the reference protein within a cut-off distance of 2.5 Å. Then, the Rac1 atomic coordinates were transferred to the model to build the complex. The protein sequences were obtained from the UniProt database (48), and aligned with the template using ClustalW2 (49,50). The top ten of a hundred models classified by DOPE score (51) were visually inspected, and the best-scored structure with suitable loops was selected. Discovery Studio Visualizer 2016 (BIOVIA, Dassault Systèmes) was used for protein structure visualization and Protein Data Bank (PDB) file editing purposes. The images were produced with UCSF Chimera software (52). The multiple sequence alignment visualization and analysis were performed with Jalview software (53).

Expression vectors, cell transfection, and ELISA

cDNAs containing the entire coding region of the mouse Plxna1 or Nrp1 were inserted into the pRK5 plasmid expression vector. Recombinant plasmids containing the Plxna1 cDNAs harboring each of the 13 mutations identified in the Kallmann syndrome patients were then engineered using the QuickChange mutagenesis protocol (Stratagene).

HEK293T cells were cultivated at 37 °C (5% CO₂) in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal calf serum, 100 µg/ml streptomycin and 100 U/ml penicillin, and transfected using Lipofectamine^TM2000 (Invitrogen). 10⁵ cells (for the ELISA experiments) were cotransfected with equal amounts of recombinant plasmids encoding wild-type or mutant plexinA1 and neuropilin1, in polyornithine-coated 96‐well plates. The cells were then cultivated for 20 h after transfection. For ELISA experiments, they were fixed with 4% paraformaldehyde in PBS for 5 min. They were then either permeabilized using 0.05% Triton X-100 for 5 min or not permeabilized, and incubated in PBS containing 1% fetal calf serum (blocking solution) for 30 min. The Flag-tagged PlxnA1 proteins were detected with the anti-Flag M2 monoclonal antibody (Sigma-Aldrich), conjugated with horseradish peroxidase. When indicated, statistical analyses of the data were performed on results from at least three independent experiments, each performed in triplicate. Mean ± s.e.m. values were plotted, and a one-way ANOVA test was performed, followed by a Kruskal-Wallis’s multiple comparison test, using the analysis software GraphPad Prism. A P value < 0.05 was required to decide on a statistically significant difference.

Cell collapse assay

COS7 cells were cultivated at 37 °C in DMEM supplemented with 10% fetal calf serum, penicillin-streptomycin, and L-glutamine. Cells were plated on glass coverslips at a density of 10⁴ cells/cm² in a 12-well plate. After 24 h, the cells were either cotransfected with 0.5 μg of the recombinant pRK5 expression vectors containing the Plxna1 cDNAs (wild-type or harboring the mutations identified in the patients) and 0.25 μg of the NP1-pcDNA1 expression vector containing a murine Nrp1 cDNA (54) or transfected with the NP1-pcDNA1 plasmid alone (used as a negative control), according to the Lipofectamine^TM2000 protocol. One day after transfection, cells were incubated with or without 3 nM of Sema3A at 37 °C for 1 h as previously described (44). Cells were then washed and fixed in 4% paraformaldehyde (in PBS), and immunostained for Nrp1 and PlxnA1. Images were acquired on a wide-field fluorescence microscope (40× objective) with the Apotome structured illumination optical sectioning system (Zeiss). Cell areas were measured using ImageJ, as previously described (44). No significant cell collapse was detected in the absence of Sema3A or in the presence of Nrp1 alone.

Construction of a PLXNA1 ‘minigene’ and RT-PCR analysis of the ‘minitranscripts’

A 2.6 Kb genomic fragment containing the PLXNA1 sequence from position c.5038 of exon 27 to position c.5465 of exon 30 (GenBank accession number NM_032242) was amplified by PCR on human genomic DNA using the XL-PCR kit (Roche diagnostics) and a specific primer set (PLXNA1-forward: 5’ TTGAC AC GGCTACTGGCCA 3’ and PLXNA1-reverse: 5’ TGGCGA TGT CT GCA T AGTAC 3’), and cloned into the pcDNA3.1 TOPO TA expression vector (Invitrogen). The c.5231G > T mutation was then introduced using the QuickChange mutagenesis protocol (Stratagene). NIH3T3 cells were transfected with either the wild-type construct or the mutated construct according to the Lipofectamine^TM2000 protocol (Invitrogen). Total RNA was extracted 24 h after transfection, and reverse-transcription was performed using the Super Script II reverse transcriptase kit (Invitrogen), followed by PCR-amplification and sequencing of the PLXNA1 ‘minitranscript’ cDNAs with the same above-mentioned primers. PCR products were separated by agarose gel electrophoresis, stained with ethidium bromide, and the ratio between the major two amplicons (i.e. with and without intron 28) was quantified using the Image J software threshold tool.

Supplementary Material

Supplementary Material is available at HMG online.

Acknowledgements

We thank the patients for their contribution to the study. We also thank the animal platform (CRP2, CNRS UMS3612, Inserm US25, Paris-Descartes University) for housing of the mice.

Conflict of Interest statement. None declared.

Funding

Agence Nationale pour la Recherche (grant number ANR-14-CE12-0015 to S.M., C.M., F.A., C.D., and J.-P.H.).