Pathogenic Mechanism of the FIG4 Mutation Responsible for Charcot-Marie-Tooth Disease CMT4J

Impaired interaction of FIG4-I41T with the scaffold protein VAC14

The corresponding yeast mutation, Fig4-I59T, results in impaired vacuole morphology and defective regulation of PI(3,5)P₂ [1], [13]. We tested the interaction of Fig4p-I59T with the Fig4p binding partners Vac14p and Fab1p using a directed yeast two hybrid assay. Fig4p was fused to the DNA binding domain of GAL4 and Vac14p was fused to the transcription activation domain. The I59T mutant did not support growth under stringent selection in the presence of 3AT (Figure 1A). This result demonstrates reduced binding of the Fig4p-I59T mutant to Vac14p. This result was confirmed in a co-immunoprecipitaton assay using myc-tagged Fig4p and GFP-tagged Vac14p [15]. Wildtype and mutant yeast proteins were expressed at similar levels, but co-precipitation of Vac14p was reduced by approximately 75% for Fig4p-I59T (Figure 1B). Co-immunoprecipitation of GFP-labeled Fab1p was reduced to a similar extent by the Fig4p-I59T mutation (Figure 1C). The latter could be an indirect effect of impaired interaction with Vac14p.

To confirm the effect of the I41T mutation in the context of the mammalian proteins, we tested the interactions of human FIG4 and human VAC14 in the yeast two hybrid system. The I41T mutation significantly impaired the interaction between the two human proteins, preventing growth on both of the selective media (Figure 1D). These experiments demonstrate that the isoleucine-to-threonine substitution reduces the direct interaction of FIG4 with VAC14.

Interaction with VAC14 is required for stability of wildtype FIG4 protein in vivo

To investigate the in vivo dependence of FIG4 on interaction with the VAC14 scaffold protein, we examined FIG4 levels in tissues from a Vac14 null mouse [7]. The absence of VAC14 in the null mouse was confirmed by Western blot (Figure 2A). To detect FIG4 protein we generated a monoclonal antibody to a bacterially-expressed 220 amino acid fragment from the C-terminus of FIG4 (Materials and Methods). The monoclonal antibody recognizes a single protein of ∼100 kDa in homogenates of mouse tissues, consistent with the calculated molecular weight of 103 kDa for the 907 amino acid FIG4 protein (Figure S1). Remarkably, the abundance of FIG4 protein was greatly reduced in the Vac14 null mouse (Figure 2B), although the level of Fig4 mRNA was normal (Figure 2C, 2D). Although FIG4 protein was undetectable in the tissue extract, a very low level of protein could be detected in cultured fibroblasts by Western blot (Figure S2). The data demonstrate that wildtype mammalian FIG4 protein is dependent on interaction with VAC14 for stability.

Generation of Fig4-I41T transgenic mice

To investigate the in vivo function of the mutant protein, we generated a transgene construct containing the mouse Fig4-I41T cDNA under the direction of the ubiquitously expressed chicken β-actin promoter (Figure 3A). Expression of Fig4-I41T on a wildtype background did not cause any visible abnormality in two independent transgenic lines, which exhibited normal fertility and life span. The transgene copy-number was measured by analysis of genomic DNA by qPCR and demonstrated a transgene copy number of 4 copies in line Tg705 and 2 copies in line Tg721 (Figure S3).

Quantitation of transgene transcript

Brain RNA was prepared from mice carrying the transgene on a wildtype genetic background. qRT-PCR reactions were carried out with Taqman primers in exon 3 and 4 and the product was detected with a probe spanning the junction between exon 3 and exon 4. Both endogenous transcripts and transgene-derived transcripts are detected by this assay. The abundance of Fig4 transcripts was compared with Tata binding protein (Tbp) transcripts as an internal control (see Materials and Methods). The ratio of Fig4 transcripts to Tbp transcripts in Fig4 ^+/+,Tg705 brain was 3× higher than in nontransgenic Fig4 ^+/+ mice (Figure 3B). Subtracting the 1× contribution from the endogenous Fig4 alleles, the contribution of the 705 transgene is 2× the level of endogenous expression. In the brain of the higher expressing line Tg721, Fig4 transcript was 6× higher than in Fig4^+/+ mice, indicating that the transgene transcript is expressed at 5× endogenous expression (Figure 3B).

Low level of FIG4-I41T protein in transgenic lines on the Fig4 null background

In order to generate a model of CMT4J expressing I41T in the absence of wildtype Fig4 protein, the FIG4-I41T transgenic mice were crossed with heterozygous mice carrying the Fig4 null (plt) allele. Fig4-null mice carrying the I41T transgene were generated in the expected Mendelian proportions from standard two-generation crosses with both transgenic lines. When extracts from brain and kidney of Fig4^−/−,Tg705 and Fig4^−/−,Tg721 mice were examined by Western blotting, the abundance of FIG4-I41T protein was substantially lower than in wildtype tissues (Figure 3C, 3D). With longer exposure, a low level of FIG4-I41T protein could be detected in the Tg705 line, and a higher level in the Tg721 line, consistent with their relative transcript levels (Figure 3E). Comparison with a 1∶10 dilution of wildtype brain indicated that the level of FIG4-I41T protein in line Tg721 is approximately 10% of wildtype (Figure 3F). The low abundance of I41T protein was confirmed with a second antibody, a rabbit polyclonal antibody generated to the same C-terminal antigen (Figure S4).

The abundance of VAC14 protein is normal in transgenic and Fig4 null mice (Figure 3G, 3H). Thus low VAC14 protein is not responsible for the low level of FIG4-I41T protein in the transgenic lines. The data are consistent with the evidence above from VAC14 null mice, and indicate that the FIG4-I41T protein is destabilized in vivo by its reduced affinity for VAC14.

Dose-dependent survival in the I41T transgenic lines

Inheritance of the low expressing transgene Tg705 increased survival of Fig4 null mice from 1–2 months to 3–6 months (Figure 4). These mice provide a model of human CMT4J, as described below. In the high-expressing Tg721 line, lethality was completely corrected (Figure 4). The oldest cohort of Fig4^−/−,Tg721 mice have survived more than 28 months with no visible abnormalities. The data demonstrate that a relatively low level of Fig4-I41T protein can rescue the lethality of Fig4 null mice.

Rescue of neurodegeneration by the Fig4-I41T transgene

Fig4 null mice exhibit a reproducible pattern of spongiform degeneration in the brain and extensive loss of neurons from peripheral ganglia (Figure 5, top panel). Neurons in layers 4 and 5 of the cortex, the deep cerebellar nuclei, and the dorsal root ganglia (DRG), are severely affected, with accumulation of vacuoles that fill the cytoplasm [1]. In null mice carrying the Tg721 transgene, these abnormalities are almost completely eliminated. Spongiform degeneration of the brain is minimal and DRG neurons are intact at P90 (Figure 5). In line Tg705, an intermediate level of degeneration is visible at P90. In both transgenic lines degeneration of the cerebellar nuclei is visible, indicating that this region is extremely sensitive to PI(3,5)P₂ levels (Figure S5). Overall, neurodegeneration is rescued in a dose-dependent manner.

Enlarged lateral ventricles and hydrocephalus are seen in both transgenic lines (Figure 5, left panels). High pressure hydrocephalus is indicated by the compression of the cerebellum and hippocampus observed in all of the Fig4^−/−,Tg705 mice near the end of their lifespan (Figure 5, bottom panel). This is also indicated by the domed appearance of the head and the expulsion of cerebrospinal fluid during dissection. The hydrocephalus in these mice is very similar to that in the L156R (ingls) mutant of Vac14 [3].

Rescue of astrocytosis and autophagy in transgenic mice

Spongiform neurodegeneration in Fig4 null brain is accompanied by accumulation of p62 and other autophagy intermediates, predominantly in activated astrocytes [11]. In brain from transgenic mice, there is an intermediate level of accumulation of the autophagy markers p62 and LAMP-2 (Figure 6A). Accumulation of the astrocyte protein GFAP is also corrected in a dose-dependent manner (Figure 6A). Reduction of astrocytosis in Tg721 mice is indicated by the decreased number of GFAP positive cells (Figure 6B). Accumulated p62 and lysosomal membrane protein LAMP-2 are localized in astrocytes of the trangenic mice (Figure 6B), as previously shown in null mice [1]. In the high expressing Tg721 line, astrocytosis and GFAP accumulation are almost completely corrected (Figure 6B).

Sciatic nerve myelination is rescued in transgenic mice

The defective myelination of peripheral nerves characteristic of Charcot-Marie-Tooth disease is also seen in Fig4 null mice ([1] and Figure 7A, top panel, arrows). In the two transgenic lines, sciatic nerve myelination was comparable to wildtype (Figure 7A). To quantitate axonal myelination we calculated the g-ratio, the inner axon diameter divided by the diameter of the nerve fiber (Figure 7B). At postnatal day 21, the g-ratio of 0.5 for wildtype sciatic nerve is increased to 0.7 for Fig4 null sciatic nerve by the thinning of the myelin sheath. In the transgenic mice the g-values were restored to wildtype (Figure 7B).

Slowed nerve conduction secondary to defective myelination is another characteristic of Charcot-Marie-Tooth disease that is reproduced in Fig4 null mice [1]. In sciatic nerve of null mice at 1 month of age, the conduction velocity was 50% of the wildtype value [1]. In contrast, conduction velocities in sciatic and sural nerves from the two transgenic lines, measured at 4 months and 14 months of age, varied between 65% and 80% of wildtype (Figure 7C, 7D), demonstrating partial rescue of the defect. The amplitude of the compound action potential in sciatic nerve was was also restored in the transgenic mice (data not shown).

Rescue of coat color in transgenic mice

Fig4-null mice have diluted pigmentation due to reduced numbers of melanosomes in the mature hair follicle and clumping of melanosomes within the hair shaft [1]. Pigmentation is partially rescued in the higher expressing line Tg721, but not in line Tg705 (Figure S6). In view of the role of Fig4 in autophagy, it is interesting that autophagy components may play a role in melanosome biogenesis [16].

Expression of FIG4-I41T in patient fibroblasts

In order to determine whether the low abundance of the FIG4-I41T protein in transgenic mice was representative of patient tissues, we examined fibroblasts from a CMT4J patient who is a compound heterozygote for I41T in exon 2 and the null allele R183X in exon 6 [2]. Exon 2 was amplified from patient and control fibroblast RNA (Figure 8A). The sequence of the control RNA contains a T nucleotide at position 122, encoding the wildtype isoleucine allele. The patient RNA contains a C nucleotide encoding threonine (Figure 8B). The absence of the wildtype nucleotide in patient RNA indicates that the R183X transcript is not stably expressed, probably due to degradation by nonsense-mediated decay [17], consistent with its location at a distance from the C-terminal exon 23.

Western blot analysis of fibroblast extracts demonstrated that FIG4-I41T protein in patient fibroblasts is dramatically reduced in comparison with wildtype (Figure 8C). The very low abundance of FIG4-I41T protein, well below the expected 50% of normal, indicates that the mutant protein is unstable in patient tissues, as it is in the transgenic mice.

Stabilization of FIG4 by co-transfection of VAC14 in HEK293 cells

The level of endogenous FIG4 in HEK293 cells was increased by transfection of VAC14 (Figure 8D, lane 1 and 2). The level of tranfected wildtype mouse FIG4 in HEK cells was also increased by co-transfection of VAC14 (Figure 8D, lane 3 and 4). The abundance of FIG4-I41T after transfection was much lower than for the wildtype protein, and the effect of co-transfected VAC14 was much smaller (Figure 8D, lane 5 and 6). These experiments demonstrate that wildtype FIG4 can be stabilized by coexpression of VAC14 in cultured cells, and that mutant FIG4-I41T is less effectively stabilized.

Effect of proteasome inhibition on FIG4-I41T

The level of FIG4 protein in primary fibroblasts from Tg721 transgenic mice is lower than in fibroblasts from wildtype mice (Figure 8D), but is sufficient to prevent vacuolization (data not shown). It was recently reported that the proteasome inhibitor MG-132 increases the level of endogenous FIG4 protein in COS7 cells, indicating that there is turnover of FIG4 in the proteasome [14]. To determine whether the mutant FIG4-I41T protein could be stabilized, we cultured primary fibroblasts for 8 hrs with 10 uM MG-132. The level of FIG4-I41T protein was significantly increased by this treatment in fibroblasts from Tg721 transgenic mice (Figure 8E). In control fibroblasts the level of wildtype FIG4 was also increased by culture with MG-132 (Figure 8F). The MW of protein produced in the presence of MG-132 was higher than in untreated cells (Figure 8E, 8F). Since MG-132 does not prevent ubiquitination of substrates, the higher MW may be a consequence of ubiquitination of the FIG4 protein.

Ethics statement

This study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The protocol was approved by the University Committee on the Use and Care of Animals (UCUCA) of the University of Michigan (Protocol No. 08629).

Fig4 null mice

The Fig4 null mutant “pale tremor” (plt) arose spontaneously in a mixed strain background that included C57BL/6J, C3H, 129 and SJL. This stock was crossed for one generation to strain C57BL/6J. Heterozygous plt/+ offspring were crossed to strain CAST/Ei to produce an F2 generation for genetic mapping [1]. plt/+ F2 mice were intercrossed to initiate a recombinant inbred line that is maintained by brother×sister breeding and is now at generation F12. The genetic background of this line, designated CB.plt, is approximately 50% CAST/Ei and 25% C57BL/6J, with smaller contributions from strains C3H, 129 and SJL.

Transgenic mice

The mouse Fig4 transcript was amplified from brain RNA isolated from strain C57BL/6J using two primers containing SphI sites: F, ACG CAT GCT ATG CTA TGT GTC TGG TGT GCT GGA GGT CTG and R, TCG CAT GCA GTC CTT TAC CCA TGA GCT GCA TC. The product was digested with SphI and subcloned into the corresponding site of the vector pCAG3z [23], [24]. The I41T mutation was incorporated into the Fig4 cDNA clone by site-directed mutagenesis and the construct was completely sequenced. Plasmid DNA was isolated with the Qiagen MaxiPrep Kit and digested to generate a linear fragment containing the cDNA sequence with the promoter and polyadenylation site. Transgenic mice carrying the Fig4-I41T cDNA construct were generated by microinjection of (C57BL/6J X SJL)F2 mouse oocytes at the Transgenic Animal Model Core at the University of Michigan (www.med.umich.edu/tamc). Transgenes were maintained in heterozygous state during breeding to prevent unequal recombination between multi-copy inserts.

Yeast functional tests

The yeast two-hybrid test and immunoprecipitation analysis were carried out as previously described [3]. Human and yeast VAC14 were subcloned into the ClaI–BglII and XmaI–SalI sites of pGAD, respectively. Human and yeast wild type and mutant FIG4 were subcloned into the XmaI–SalI and BamHI–PstI sites of pGBD, respectively. pGAD and pGBD plasmids were cotransformed into the yeast strain PJ69-4A. Transformants were initially plated onto SC-LEU-TRP media for plasmid selection and replica-plated onto selective plates with SC-LEU-TRP-ADE-HIS+3AT media or SC-LEU-TRP-ADE-HIS media and grown at 24°C for 4 to 14 days for colony formation. The yeast strain PJ69-4A and the pGAD and pGBD vectors have been described [25]. The Venus-derivative of GFP was previously described [15]. Selective media contained 3-amino-1,2,4-triazole (3AT), an inhibitor of His3p.

Immunofluorescence and Western blots

Immunocytochemistry of p62, LAMP2 and GFAP on fresh frozen cryosections of mouse brain and Western blotting of brain extracts were carried out as previously described [11]. For immunoblotting of FIG4, tissues were homogenized in 0.25 M sucrose, 0.05 M Tris, pH 7.5 and the 21,000× g soluble fraction was analyzed. Immunofluorescence images were captured on an Olympus BX51 microscope equipped with epifluorescence and processed and merged with Adobe Photoshop software.

FIG4 antibodies

A rabbit polyclonal antiserum and a mouse monoclonal antibody were generated to a C-terminal fragment of FIG4 containing the final 220 amino acids, residues 688 to 907, which is encoded by exons 18 to 23. A 660 bp cDNA fragment was amplified from a full length Fig4 cDNA derived from strain C57BL/6J and cloned into the XhoI and EcoRI sites of the expression vector pRSETA (Invitrogen Corporation, Calrsbad, CA, USA), adding a polyhistidine tag (6× His) to the N-terminus. Recombinant protein was expressed in E. coli and purified from inclusion bodies using Novagen BugBuster Protein Extraction Reagent (EMD Chemicals, Gibbstown, NJ, USA). Rabbit polyclonal antiserum to the purified protein was generated and affinity-purified by Pocono Rabbit Farm and Laboratory, Inc. (Canadensis, PA). The purified polyclonal antiserum was used at 1∶50 dilution for western blots. The monoclonal antibody was generated using the same protein fragment at the UC Davis/NIH NeuroMab Facility (Clone N202/7). The monoclonal was purified from cell culture and used at 1∶200 for western blots. A broad crossreacting band of 50 to 55 kDa is ocassionally observed on Western blots of control and mutant tissues but there is no consistent association with genotype.

Histology

Brain and spinal cord were fixed for 24 hours at 4°C in phosphate buffered 10% formalin and then in 70% ethanol for an additional 24 hours at 4°C. Paraffin embedding, decalcification of spine, and H&E staining were carried out at Histoserv Inc (Maryland). Images were obtained with an Olympus BX51 microscope and DP50 camera. Image capture settings were identical for sections from −/− and −/−,Tg705 mice that exhibited immunostaining for autophagy markers. To detect the outlines of brain sections from −/−,Tg721 and +/+ mice that lacked immunostaining, a longer image capture time was used.

qRT-PCR of Fig4 transcripts

Five ug aliquots of total RNA from whole brain was treated with DNAse I (Invtrogen) and cDNA was prepared using the SuperScript First Strand Synthesis System for RT-PCR (Invitrogen). The Fig4 transcript was quantitated using TaqMan gene expression probe Mm00506074_m1 (ABI) which spans the junction between exon 3 and exon 4. As an internal control, the TATA binding protein (Tbp) transcript was quantitated using TaqMan probe Mm00446971_m1. Fluorescence was measured on a Step One Real Time PCR System (ABI) at the Microarray Core at the University of Michigan. A linear relationship between copy number and C_T, the threshold cycle for detection of fluorescence, was observed (Figure S3). The mean C_T value was determined from quadruplicate assays of each sample. The value of ΔC_T was calculated by subtracting the C_T for Tbp from the C_T for Fig4. The ratio of Fig4 transcripts to Tbp transcripts was calculated as 2^−ΔCT.

Transgene copy number determination by qPCR

Transgene copy number was determined using the TaqMan probe described above. A standard curve was generated using pCAG3Z-Fig4 diluted into wildtype mouse DNA for copy-number standards of 0, 0.5, 1, 2, 4, 8, and 16 copies per haploid genome using 2.83 pg/ug for one haploid copy equivalent for this 3 kb transgene. Genomic DNA (50 ng) from Tg705/+ mice, Tg721/+ mice, and standards were assayed in quadruplicate. The Taqman probe includes the junction between exon 3 and exon 4, which are separated by 4 kb in genomic DNA. No fluorescent signal was obtained from 8 replicates of wildtype genomic DNA. R² for the standard curve was 0.999. Observed copy numbers for both transgenic lines are within the linear range of the assay.

Primary fibroblasts and HEK293 cells

HEK293 cells (passage 15) were maintained in DMEM/F-12 supplemented with 10% fetal bovine serum, penicillin, streptomycin and amphotericin. Subconfluent cells in 10 cm dishes were transfected with 6 ug plasmid DNA and 18 uL FuGene (Roche) for 24 hours in complete culture media. Cells were lysed in RIPA buffer and lysates were spun at 15,000× g for 10 minutes. Supernatant protein was quantitated using the BCA kit (Pierce). Mouse fibroblasts were isolated from P0 mouse tail biopsy by digestion with collagenase type 2 (Worthington labs) and cultured in RPMI 1640 supplemented with 15% fetal bovine serum and containing penicillin, streptomycin and amphotericin. Experiments were carried out at passage 3 to 5. Treatment with 10 uM MG-132 (Sigma) was carried out in complete media for 8 hours. Primary cultured fibroblasts from CMT4J patient 2 were previously described [8]; passage 4 cells were kindly provided by Dr. Jun Li, Department of Neurology, Vanderbilt University.

Nerve conduction velocity

Nerve conduction velocity (NCV) was measured as previously described [26]. Mice were anesthetized and maintained at a 34°C core temperature with a heating lamp. Sural sensory NCV was determined by recording at the dorsum of the foot and antidromically stimulating with supramaximal stimulation at the ankle. NCV was calculated by dividing the distance by the take-off latency of the sensory nerve action potential. Sciatic-tibial motor NCV was determined by recording at the dorsum of the foot and orthodromically stimulating with supramaximal stimulation first at the ankle, then at the sciatic notch. Latencies were measured in each case from the initial onset of the compound muscle action potential. The sciatic-tibial motor NCV was calculated by subtracting the measured ankle distance from the measured notch distance. The resultant distance was then divided by the difference in the ankle and notch latencies for a final nerve conduction velocity.

Electron microscopy of sciatic nerve axons

Mice were anesthetized with ketamine and xylazine and perfused transcardially with 3% paraformaldehyde (Electron Microscopy Sciences) and 2.5% glutaraldehyde (Ted Pella, Inc.). Sciatic nerves were dissected and post-fixed for several hours in perfusion solution at 4°C. Tissues were incubated in a 1% solution of OsO₄ and embedded in epoxy resin. Ultrathin (75 nm) sections were cut and visualized with a Philips CM-100 TEM. Images were analyzed at 2600-fold magnification using Image-J software for quantification of g-ratio (defined as the diameter of an axon within its myelin sheath divided by the diameter of the axon outside of its myelin sheath). Two measurements were made per axon in order to account for elongated or irregular shape, and the mean was used. Only myelinated axons were used in for the calculation of g-ratio. Data were compared using Microsoft Excel.

Accession numbers

Accession numbers for the genes referred to in this paper are: human FIG4 (Entrez Gene ID 9896), mouse Fig4 (Entrez Gene ID 103199), yeast Fig4p (Entrez Gene ID 855392), mouse Vac14 (Entrez Gene ID 234729), yeast Vac14p (Entrez Gene ID 851102), mouse Fab1/PIKfyve (Entrez Gene ID 18711), yeast Fab1p (Entrez Gene ID 850574).