Severe biallelic loss-of-function mutations in nicotinamide mononucleotide adenylyltransferase 2 (NMNAT2) in two fetuses with fetal akinesia deformation sequence

Abstract

The three nicotinamide mononucleotide adenylyltransferase (NMNAT) family members synthesize the electron carrier nicotinamide adenine dinucleotide (NAD⁺) and are essential for cellular metabolism. In mammalian axons, NMNAT activity appears to be required for axon survival and is predominantly provided by NMNAT2. NMNAT2 has recently been shown to also function as a chaperone to aid in the refolding of misfolded proteins. Nmnat2 deficiency in mice, or in its ortholog dNmnat in Drosophila, results in axon outgrowth and survival defects. Peripheral nerve axons in NMNAT2-deficient mice fail to extend and innervate targets, and skeletal muscle is severely underdeveloped. In addition, removing NMNAT2 from established axons initiates axon death by Wallerian degeneration. We report here on two stillborn siblings with fetal akinesia deformation sequence (FADS), severely reduced skeletal muscle mass and hydrops fetalis. Clinical exome sequencing identified compound heterozygous NMNAT2 variant alleles in both cases. Both protein variants are incapable of supporting axon survival in mouse primary neuron cultures when overexpressed. In vitro assays demonstrate altered protein stability and/or defects in NAD⁺ synthesis and chaperone functions. Thus, both patient NMNAT2 alleles are null or severely hypo-morphic. These data indicate a previously unknown role for NMNAT2 in human neurological development and provide the first direct molecular evidence to support the involvement of Wallerian degeneration in a human axonal disorder. SIGNIFICANCE: Nicotinamide Mononucleotide Adenylyltransferase 2 (NMNAT2) both synthesizes the electron carrier Nicotinamide Adenine Dinucleotide (NAD⁺) and acts a protein chaperone. NMNAT2 has emerged as a major neuron survival factor. Overexpression of NMNAT2 protects neurons from Wallerian degeneration after injury and declining levels of NMNAT2 have been implicated in neurodegeneration. While the role of NMNAT2 in neurodegeneration has been extensively studied, the role of NMNAT2 in human development remains unclear. In this work, we present the first human variants in NMNAT2 identified in two fetuses with severe skeletal muscle hypoplasia and fetal akinesia. Functional studies in vitro showed that the mutations impair both NMNAT2 NAD⁺ synthase and chaperone functions. This work identifies the critical role of NMNAT2 in human development.

Figure 1.. Gross phenotype and histology of affected fetuses.

(A,B) Fetal MRI of Fetus II-1 in which hydrocephalus is noted by asterisk and cystic hygroma by arrowhead. (C) Dorsal view of fetus II-3 with notable edema and lack of skeletal muscle in the extremities. (D) Fetus II-3 displays malrotation of the gut with the appendix in the upper left quadrant. (E) Absence of the psoas muscles bilaterally is noted by the asterisks. (F,G) Fetus II-3 displays flattened hands with contractures of the elbow (F), and nearly complete absence of skeletal muscle of the leg (G). (H) Histology of the right radius and ulna shows reduced skeletal muscle fiber packing near the bone. (I) Histology of the interosseous muscles of the right hand show sparsely spaced muscle fibers with plump nuclei. (J) Histology of the hip joint shows fibrofatty tissue replacement of the musculature of the hip (K) Histology of the left ventricle shows normal architecture of the myocardium.

Figure 2.. Whole exome sequencing identifies compound heterozygous mutations in NMNAT2.

(A) Family pedigree. Stillborn infants are depicted as filled triangles with slashes (A). (B) Sanger sequencing of NMNAT2. (C) Conservation of arginine at aa232 in NMNAT2 homologues across distant phyla. (D) Diagram of functional domains of NMNAT2 with patient variant positions in red. (E) 3D structure model of NMNAT1. The conserved β strands and α helices important for enzymatic function are marked in yellow and cyan, respectively. The disordered region in NMNAT that contains the nuclear localization sequence is indicated by a dashed line. The R232 equivalent residue is labeled in red.

Figure 3.. NMNAT2^R232Q and NMNAT2^Q135Pfs*44 both have reduced capacity to maintain neurite survival.

(A) Representative images (of n = 4 independent experiments) of cut neurites of SCG neurons co-injected with expression vectors for Flag-NMNAT2^WT, or Flag-NMNAT2^R232Q or NMNAT2^Q135Pfs*44 (10 ng/μl) and DsRed (pDsRed2, 40 ng/μl). Neurites were cut 48 hours after injection when DsRed expression allows clear visualization of the distal neurites of the injected neurons. Images show transected neurites, just distal to the lesion, immediately after (0h) and 24 hours after cut. The lesion site is located the bottom edge of each field. Brightness and contrast have been adjusted for optimal visualization of neurites. (B) Quantification of neurite survival at 24 hours after cut for experiments described in panel A. The number of intact neurites with continuous DsRed fluorescence at 24 hours is shown as a percentage of intact neurites at 0h. Individual values and means ± SEM are plotted (individual values represent the average of two fields per separate culture). n.s. = not significant (p > 0.05), *** p < 0.001, one-way ANOVA with Tukey–s multiple comparisons test. (C) Relative expression level of Flag-NMNAT2 variants in injected SCG neuron cell bodies. Representative fluorescent images of SCG neurons 24 hours after co-injection with expression vectors for Flag-NMNAT2^WT, Flag-NMNAT2^R232Q or Flag- NMNAT2^Q135Pfs*44 and DsRed (each at 25 ng/μl). DsRed identifies injected neurons, Flag immunostaining shows expression of the Flag-NMNAT2 proteins, and DAPI labels nuclei. Relative intensities (± SEM) of Flag immunostaining and DsRed signal are shown after transformation to the mean of levels in neurons injected with the Flag-NMNAT2^WT construct. The data for WT, R232Q and Q135Pfs*44 were calculated from 47, 62 and 40 injected neurons (DsRed positive) of which 87.2%, 81,3% and 22.5% were Flag-positive respectively.

Figure 4.. Relative stabilities and activities of NMNAT2^R232Q and NMNAT2^Q135Pfs*44 in HEK 293T cells.

(A) Representative immunoblots (of n = 3) of extracts of HEK 293T cells cotransfected with expression vectors for Flag-NMNAT2^WT, Flag-NMNAT2^R232Q or Flag-NMNAT2^Q135Pfs*44 and eGFP at the indicated times after addition of 10 μM emetine. Emetine was added 24 h after transfection. Extract from non-transfected cells is also shown (NT). Blots were probed with Flag, eGFP and α-Tubulin antibodies. To avoid saturation of the protein degradation machinery that might artificially slow rates of turnover, expression of the Flag-NMNAT2 proteins was kept relatively low by including empty vector as part of the transfection mix (see Materials and Methods). Co-transfected eGFP or endogenous α-Tubulin (present in transfected and non-transfected cells) are both relatively stable proteins and were respectively used as a reference for Flag-NMNAT2 protein turnover (to control for transfection efficiency) and for loading. Arrows indicate the positions of bands corresponding to Flag-NMNAT2^WT (black, ~34 kDa), Flag-NMNAT2^R232Q (red, ~37 kDa), and Flag-NMNAT2^Q135Pfs*44 (green, ~22 kDa). The position of a faint non-specific band is also marked (*). (B) Relative steady-state Flag-NMNAT2 protein band intensities (0h, just before emetine addition) after normalization to co-transfected eGFP for blots described in panel A. Individual values (n = 4-5) and means ±SEM are plotted. n.s. = not significant (p > 0.05), ** p < 0.01, one-way ANOVA with Tukey’s multiple comparisons test (only comparisons to Flag-NMNAT2^WT. (C) Representative immunoblots (of n = 4) of extracts of HEK 293T cells, as in panel A, but transfected with a higher concentration of Flag-NMNAT2^Q135Pfs*44 expression vector and with increased loading per lane to maximize Flag band intensity at the 0h time point so that its level at 0h is similar to that of Flag-NMNAT2^WT in panel A. This allows for a more accurate comparison of turnover rates. Only a ~22 kDa band corresponding to the predicted size of Flag-NMNAT2^Q135Pfs*44 is seen. (D) Relative turnover rates of Flag-NMNAT2 proteins after emetine addition. Flag-NMNAT2 band intensities on blots described in panel A (Flag-NMNAT2^WT and Flag-NMNAT2^R232Q) and panel C (Flag-NMNAT2^Q135Pfs*44) were normalized to co-transfected eGFP and intensities at each time point after emetine addition were calculated as a proportion of the intensity of the 0h, untreated band. Means ± SEM (n = 4) are plotted. n.s. = not significant (p > 0.05), **p < 0.01 and *** p < 0.001, two-way AN OVA with Sidak’s multiple comparisons test for effects between variants. One-phase decay curves were fitted to the data sets for Flag-NMNAT2^wr and Flag-NMNAT2^R232Q using non-linear regression. The R² value and half-life (t_½) are reported. No intensity values could be obtained for Flag-NMNAT2^Q135Pfs*44 at any timepoint assessed after emetine addition precluding curve fitting and statistical analysis.

Figure 5.. Bacterial expression and in vitro characterization of the activity of recombinant NMNAT2^R232Q and NMNAT2^Q135Pfs*44.

(A) Coomassie blue stained 12 % SDS polyacrylamide gel loaded with similar amounts (~3 μg) of NMNAT2^WT and each indicated recombinant NMNAT2 variant arising from His-tag affinity chromatography. (B) Immunoblots of ~0.3 μg of the same protein samples as in A probed with anti-His and anti-NMNAT2 antibodies as indicated. As in FIEK cells, bacterially-expressed NMNAT2^R232Q migrates slower than NMNAT2^WT and NMNAT2^Q135Pfs*44, which lacks the epitope recognized by the NMNAT2 antibody (raised against the C-terminus of the full-length protein), is expressed at a low level. The NMNAT2^Q135Pfs*44 preparation contains a ~34 kDa protein recognized by both anti-His and anti-NMNAT2 antibodies that is likely to be His-tagged NMNAT2^WT (red boxes). (C) NMNAT specific activity of His-tag purified preparations measured at 37 °C with saturating concentrations of substrates. The less pure NMNAT2^Q135Pfs*44 preparation is omitted despite some activity found since it was not associated with the His-tagged 22 kDa truncated protein arising from the frame shift mutation (see text). (D) Magnesium-dependent rates of NMNAT activity referred to 1 mM MgCl₂ (arbitrary 100 % value). (E) Enzyme stability after 1 hour treatment at different temperatures. Treated enzyme solutions were then assayed at 37 °C. Relative rates are expressed as percentages of the untreated enzyme kept at 4 °C (100 % not shown). (f) Enzyme stability at 37 °C as function of time. Rates are relative to time zero. (g) Optimum temperature after heating of whole assay mixtures at the indicated temperatures. Relative rates are expressed as percentages of the maximum observed (42 °C for both enzymes). All data presented are the mean ± SEM from n = 3 independent measures. T test p values vs corresponding WT are marked by (*) p < 0.015 or by (**) p < 0.005 (Two Sample t Test, unequal variances).

Figure 6.. NMNAT2^R232Q and NMNAT2^Q135Pfs* have reduced chaperone activities.

HEK 293T cells were co-transfected with luciferase and one of the following plasmids: DsRed2 vector (control), Hsp70, NMNAT3, NMNAT2^WT, NMNAT2^R232Q, and Nmnat2^Q135Pfs*44. At 48 hrs after transfection, protein synthesis was inhibited, and cells were subjected to heat shock at 42 °C for 45 mins, and then recovered at 37 °C for 3 hours. Quantification of luciferase activity measured without heat shock (blue bars), after heat shock (red bards), and after recovery (green bars). Luciferase activity in each group was normalized to no heat shock (set to 1). All data were presented as mean ± SD, n=4. Statistical significance was established by two-way ANOVA post hoc Tukey’s multiple comparison test. ***P<0.001, ****P<0.0001, NS: not significant.