Haploinsufficiency of Dmxl2, encoding a synaptic protein, causes infertility associated with a loss of GnRH neurons in mouse

Abstract

Characterization of the genetic defects causing gonadotropic deficiency has made a major contribution to elucidation of the fundamental role of Kisspeptins and Neurokinin B in puberty onset and reproduction. The absence of puberty may also reveal neurodevelopmental disorders caused by molecular defects in various cellular pathways. Investigations of these neurodevelopmental disorders may provide information about the neuronal processes controlling puberty onset and reproductive capacity. We describe here a new syndrome observed in three brothers, which involves gonadotropic axis deficiency, central hypothyroidism, peripheral demyelinating sensorimotor polyneuropathy, mental retardation, and profound hypoglycemia, progressing to nonautoimmune insulin-dependent diabetes mellitus. High-throughput sequencing revealed a homozygous in-frame deletion of 15 nucleotides in DMXL2 in all three affected patients. This homozygous deletion was associated with lower DMXL2 mRNA levels in the blood lymphocytes of the patients. DMXL2 encodes the synaptic protein rabconnectin-3α, which has been identified as a putative scaffold protein for Rab3-GAP and Rab3-GEP, two regulators of the GTPase Rab3a. We found that rabconnectin-3α was expressed in exocytosis vesicles in gonadotropin-releasing hormone (GnRH) axonal extremities in the median eminence of the hypothalamus. It was also specifically expressed in cells expressing luteinizing hormone (LH) and follicle-stimulating hormone (FSH) within the pituitary. The conditional heterozygous deletion of Dmxl2 from mouse neurons delayed puberty and resulted in very low fertility. This reproductive phenotype was associated with a lower number of GnRH neurons in the hypothalamus of adult mice. Finally, Dmxl2 knockdown in an insulin-secreting cell line showed that rabconnectin-3α controlled the constitutive and glucose-induced secretion of insulin. In conclusion, this study shows that low levels of DMXL2 expression cause a complex neurological phenotype, with abnormal glucose metabolism and gonadotropic axis deficiency due to a loss of GnRH neurons. Our findings identify rabconectin-3α as a key controller of neuronal and endocrine homeostatic processes.

Figure 1. DMXL2 is mutated in affected patients.

(A) Pedigree of the affected family. Closed symbols indicate affected individuals. (B) Linkage analysis delineated two candidate regions on chromosomes 13 and 15 with a LOD score of 2.5. (C) Next-generation sequencing characterized a deletion of 15 nucleotides (c.5824_5838del) in exon 24 of DMXL2. This deletion removes five amino acids (p.1942_1946del). Rbcn-3α is a protein with 17 WD domains (green box) and one Rav1p_C domain, which is involved in regulating the glucose-dependent assembly and disassembly of the V1 and V0 subunits of the vacuolar ATPase (purple box) . (D) Quantification, by RT-qPCR, of DMXL2 mRNA levels relative to RNA 18S levels in blood lymphocytes. Error bars, SEM. * p<0.05. Numerical data used to generate graph 1D may be found in Table S5.

Figure 2. Rbcn-3α is expressed in exocytosis vesicles in the external layer of the median eminence.

(A) ISH with a mouse Dmxl2 antisense probe (AS) and a sense probe (S). (B) Immunolabeling with an antibody against Rbcn-3α revealed high levels of Dmxl2/Rbcn-3α expression in the dentate gyrus, the CA1 and CA3 regions of the hippocampus, and the cortex (black arrowheads). Scale bars, 200 µm. (C and D) Rbcn-3α was found to be strongly expressed in the external layer of the ME (C) and the OVLT (D). Scale bars, 100 µm. (E) Confocal analysis with an antibody against Rbcn-3α showed punctate staining in the median eminence and staining of the long processes extending from the cell bodies lining the third ventricle. (F and G) Rbcn-3α immunoreactivity was observed in small clear vesicles and LDCVs at the extremities of the axons in the ME (white arrow). Scale bar, 0.2 µm.

Figure 3. Rbcn-3α is specifically expressed in gonadotropes in the pituitary.

(A) Double-immunostaining for Rbcn-3α and LH or FSH in the rat pituitary gland showed that Rbcn-3α was expressed in gonadotropes. Scale bars, 10 µm. (B) Double-immunostaining for Rab3GAP or Rab3GEP revealed that these two proteins were present in both LH- and FSH-expressing cells. (C) Immunostaining with antibodies against Rbcn-3α, Rab3GAP, or Rab3GEP and ACTH, TSH, and GH showed that Rbcn-3α, Rab3GAP, and Rab3GEP were not expressed in corticotropes, thyreotropes, and somatotropes. Scale bars, 40 µm. Red staining, antibodies against Rbcn-3α, Rab3GAP, and Rab3GEP. Green staining, antibodies against pituitary hormones.

Figure 4. Female Nes-cre;Dmxl2 ^–/wt mice displayed delayed puberty and were infertile.

(A) Postnatal growth curve of female nes-Cre;Dmxl2 ^–/wt mice (black line, Dmxl2^lox/wt; hatched gray line, nes-Cre;Dmxl2 ^–/wt). (B and C) VO and first estrus occurred significantly later in female nes-Cre;Dmxl2 ^–/wt mice than in their WT littermates. (D) The interval from VO to first estrus was significantly longer in female nes-Cre;Dmxl2 ^–/wt mice than in WT mice, suggesting a defect in maturation of the HPG axis. (E) Very few complete estrous cycles were observed in female nes-Cre;Dmxl2 ^–/wt mice. (F) Female nes-Cre;Dmxl2 ^–/wt spent less time in high estradiol concentration (M, miestrus; E, estrus; D, diestrus; P, proestrus). (G) A significant difference in AGD was observed between male nes-Cre;Dmxl2 ^–/wt mice and their WT littermates. Black line, Dmxl2^lox/wt; hatched gray line, nes-Cre;Dmxl2 ^-/wt. White bars, Dmxl2^lox/wt. Black bars, nes-Cre;Dmxl2 ^–/wt. Numerical data used to generate graphs 4B, 4C, 4D, 4E, 4F, or graphs 4A, 4G may be found in Table S5 and Table S6, respectively. Error bars are SEM. *p<0.05, **p<0.01, ***p<0.001.

Figure 5. nes-Cre;Dmxl2 ^–/wt mice displayed a partial gonadotropin deficiency.

(A and B) Weights of testes and ovaries were low in nes-Cre;Dmxl2 ^–/wt mice. (C) Histological analysis of ovaries showed a normal number of antral follicles but very few corpora lutea in female nes-Cre;Dmxl2 ^–/wt mice. (D) Estradiol concentrations were normal in female nes-Cre;Dmxl2 ^–/wt mice. (E) Plasma testosterone concentration was low in male nes-Cre;Dmxl2 ^–/wt mice. (F) Plasma LH concentrations were moderately high in female nes-Cre;Dmxl2 ^–/wt mice. (G) Despite their lower testosterone concentrations, male nes-Cre;Dmxl2 ^–/wt mice had normal plasma LH concentrations. (H) The GnRH-induced increase in LH concentration was normal in nes-Cre;Dmxl2 ^–/wt mice. (I) The administration of PMSG to young mice induced a normal increase in estradiol concentration in nes-Cre;Dmxl2 ^–/wt mice, similar to that observed in their WT littermates (asterisks indicate significant differences: * p<0.05, ** p<0.001; ***p<0.0001). Error bars: SEM. P, postnatal day. White bars, Dmxl2^lox/wt; black bars, nes-Cre;Dmxl2 ^–/wt. Numerical data used to generate these graphs may be found in Table S5.

Figure 6. Hypothalamic GnRH mRNA and GnRH-IR neuron levels are lower in the hypothalamus of nes-Cre;Dmxl2 ^–/wt mice.

(A) Rbcn-3α is expressed in GnRH neurons in the median eminence. (B) Rbcn-3α is located in small clear vesicles and in LDCVs in GnRH neurons. White arrows indicate Rbcn-3α DAB staining; white arrow heads indicate GnRH nanogold staining. (C) GnRH1 mRNA levels relative to RNA18S were lower in the hypothalamus of nes-Cre;Dmxl2 ^–/wt male mice than in WT mice. (D) The total number of GnRH-ir neurons in the brain was lower in nes-Cre;Dmxl2 ^–/wt male mice than in WT mice. (E) GnRH immunostaining in the OVLTs in Dmxl2 ^lox/wt and nes-Cre;Dmxl2 ^–/wt male mice. (F) An analysis of the rostral–caudal distribution of GnRH-ir neurons in the hypothalamus revealed that nes-Cre;Dmxl2 ^–/wt male mice had fewer GnRH-ir cell bodies in the OVLT (see inset) than their WT littermates. * p<0.05, *** p<0.0001. White bars, Dmxl2^lox/wt; black bars, nes-Cre;Dmxl2 ^–/wt. Numerical data used to generate graphs 6C, 6D, 6F, and 6F inset may be found in Table S5.

Figure 7. Dmxl2 knockdown in INS-1E cells decreases glucose-induced insulin release.

(A) Rbcn-3α is expressed in insulin-secreting cells in the islets of Langerhans in the mouse pancreas. (B) Seventy-two hours after the transfection of INS-1E cells with Dmxl2-siRNA, Dmxl2 mRNA levels had decreased by 75% (error bars: SEM; from three independent experiments). (C) Seventy-two hours after siRNA transfection, the release of insulin was evaluated by quantifying insulin concentrations in cell supernatants. In the absence of glucose, insulin concentrations in Dmxl2-siRNA–transfected cells were double those in cells transfected with NT siRNA (NT-siRNA). In the presence of various concentrations of glucose (5 and 20 mM), Dmxl2-siRNA–transfected cells displayed only a small increase in insulin release, at a glucose concentration of 20 mM only, whereas a 2- to 3-fold increase was observed with NT siRNA-transfected cells. Error bars: SEM from one representative experiment performed twice, in hextuplicate. * p<0.05, ** p<0.001, *** p<0.0001. Numerical data used to generate graphs 7B and 7C may be found in Table S5.