The Nuclear Factor k B–Activator Gene PLEKHG5 Is Mutated in a Form of Autosomal Recessive Lower Motor Neuron Disease with Childhood Onset

Lower motor neuron diseases (LMNDs) include a large spectrum of clinically and genetically heterogeneous disorders. Studying a large inbred African family, we recently described a novel autosomal recessive LMND variant characterized by childhood onset, generalized muscle involvement, and severe outcome, and we mapped the disease gene to a 3.9-cM interval on chromosome 1p36. We identiﬁed a homozygous missense mutation (c.1940 T r C [p.647 Phe r Ser]) of the Pleckstrin homology domain–containing, family G member 5 gene, PLEKHG5. In transiently transfected HEK293 and MCF10A cell lines, we found that wild-type PLEKHG5 activated the nuclear factor k B (NF k B) signaling pathway and that both the stability and the intracellular location of mutant PLEKHG5 protein were altered, severely impairing the NF k B transduction pathway. Moreover, aggregates were observed in transiently transfected NSC34 murine motor neurons overexpressing the mutant PLEKHG5 protein. Both loss of PLEKHG5 function and aggregate formation may contribute to neurotoxicity in this novel form of LMND.

Lower motor neuron diseases (LMNDs) are clinically characterized by progressive paralysis with amyotrophy and loss of deep tendon reflexes and fasciculations, because of motor neuron degeneration in the anterior horn of the spinal cord and the brainstem. Diagnosis is confirmed by electrophysiological or histological evidence of muscle denervation, with normal or subnormal motor nerve conduction velocities and normal sensory potentials. The classic form of autosomal recessive proximal spinal muscular atrophy (MIM 253300) is linked to the SMN1 gene, 1 but numerous LMND variants have been described, differing by localization of motor weakness, mode of inheritance, and age at onset. 2,3 To date, six other causative genes were reported in pure LMNDs: five with autosomal dominant inheritance (HSPB8/HSP22 and HSPB1/HSP27 in distal hereditary motor neuronopathy type II [d-HMNII {MIM 158590 and 608634}], GARS and BSCL2 in d-HMNV [MIM 600794], and DCTN1 in d-HMNVII [MIM 607641]) and one with autosomal recessive inheritance (IGHMBP2 in d-HMNVI, also known as "spinal muscular atrophy with respiratory distress" [SMARD {MIM 604320}]). [4][5][6][7][8][9] All encode proteins that are directly or indirectly involved in two important intracellular pathways: axonal transport and RNA processing. 10,11 Elsewhere, we described a novel clinical variant of au-tosomal recessive LMND with childhood onset in a large inbred family originating from Mali. 12 Affected individuals presented with generalized muscle weakness and atrophy with denervation and normal sensation. Bulbar symptoms and pyramidal signs were absent. In four of the five affected children, the outcome was severe, with loss of walking and the need for permanent respiratory assistance before adulthood. Genetic analyses with the use of a homozygosity mapping strategy assigned this LMND locus to a 3.9-cM (or 1.5-Mb) interval on chromosome 1p36, between loci D1S508 and D1S2633 ( a t Z p 3.79 v p max at locus D1S253). Here, we report the identification 0.00 of the Pleckstrin homology domain-containing, family G member 5 gene, PLEKHG5 (GenBank accession number NM_020631), located within the candidate region, which is mutated in this pedigree.

Clinical Study
Elsewhere, we reported the linkage data on the African family. 12 Informed consent was obtained from all family members, and the study was approved by the ethics committee of the Catholic University of Louvain.

Genotyping and Mutation Analysis
We isolated genomic DNA (gDNA) from blood samples, using standard protocols. We performed genotyping for microsatellites as described elsewhere. 12 We used the Primer3 program to design PCR primers that flanked all exons of candidate genes, on the basis of the chromosome 1 draft human sequence. Details on PLEKHG5 primers are given in table 1. We performed direct sequencing, using the dideoxy chain-termination method (ABI Big Dye 3.1) on a 3100 automated sequencer (ABI Prism [Applied Biosystems]) in accordance with standard procedures and the manufacturer's recommendations. The mutation was verified bidirectionally on gDNA and cDNA. We extracted whole mRNA from fibroblast cell lines of patients (RNeasy [Qiagen]), and we obtained cDNA products by RT-PCR, using oligo(dT) and random hexamers (Transcriptor First Strand cDNA Synthesis Kit [Roche]).

Plasmid Constructions
PLEKHG5 full-ORF expression clones, containing the complete coding cDNA for isoforms BC042606 or BC015231, were sequence verified (Deutsches Ressourcenzentrum fü r Genomforschung [RZPD]). To introduce the c.1940 TrC amino acid substitution in the PLEKHG5 plasmids, we amplified cDNA of the patients with primers framing the mutation (table 1). We then restricted the mutated cDNA amplification product and the full-ORF expression clones with two single-cut endonucleases, BstEII and PflMI (New England Biolabs). After ligation, we screened for the presence of the mutated insert, and we verified the absence of additional mutation by sequencing analysis. pCMVlacZ plasmid was constructed by insertion of the Escherichia coli lacZ coding region into the multiple cloning site of pCMX-PL1. 13 Cell Culture and Transfection Conditions HEK293 cells (human embryonic kidney 293 cells) and MCF10A cells (human mammary epithelial cells) were maintained at 37ЊC in a humidified 5% CO 2 atmosphere, in Dulbecco's modified Eagle medium (DMEM) F12 (Invitrogen) supplemented with 5% (for MCF10A) or 10% (for HEK293) horse serum (Cambrex), 100 IU/ ml penicillin, and 100 mg/ml streptomycin (Sigma). Also added to the medium for MCF10A cells was 0.1 mg/ml cholera toxin (Calbiochem), 10 mg/ml insulin (Invitrogen), 0.5 mg/ml hydrocortisone (Sigma), and 20 ng/ml human epidermal growth factor (Invitrogen). Before transfection, exponentially proliferating cells were trypsinized, and MCF10A cells or 5 5 1.6 # 10 2.6 # 10 HEK293 cells were plated in each well of a 6-well plate. Twentyfour hours after plating, cells were transfected using 3 ml of Fugene 6 (Roche) in accordance with the manufacturer's instructions. For luciferase-reporter gene assays and real-time RT-PCR analyses, 0.5 mg of the expression vectors was transfected together with 1 mg of the reporter plasmid (nuclear factor kB [NFkB] cis-reporting system [Stratagene]) and with 20 ng of constitutive reporter plasmid (pCMVlacZ) for luciferase activity normalization. Expression vectors were replaced by 50 ng of pFC-MEKK (Stratagene) and 0.5 mg of carrier DNA (pCat), to provide a positive control for NFkB activation. For western-blot analyses and immunofluorescence studies, cells were transfected with 1.5 mg of expression vector DNA.
NSC34 cells (a mouse embryonic spinal cord-neuroblastoma cell line with a motor neuron phenotype, kindly provided by Dr. Neil Cashman, University of Toronto 14 ) were cultured at 37ЊC under 5% CO 2 and 95% air in DMEM supplemented with 10% fetal calf serum. NSC34 cells in a 24-well plate were transfected with either the wild-type or the mutant PLEKHG5 plasmids with use of Lipofectamine (Invitrogen). The transfection efficacy was controlled using cotransfection with a green fluorescent protein (GFP)-expressing plasmid.

Enzymatic Assays
Cells were harvested 48 h after transfection. Lysis and enzymatic activity dosages were performed with the b-Gal Reporter Gene Assay (chemiluminescent) kit (Roche) and the Luciferase Reporter Gene assay (high-sensitivity) kit (Roche).

Gene-Expression Analysis by Quantitative RT-PCR
Total RNA was isolated 48 h after transfection and was purified using the Trizol procedure (Invitrogen) in accordance with the manufacturer's instructions. cDNA was transcribed from 3 mg of total RNA by use of random hexamers and M-MLV Reverse Transcriptase (Invitrogen), in the presence of RNase inhibitor (Promega). Quantitative real-time RT-PCR amplification was performed on cDNA by use of the qPCR Master Mix Plus for SYBR Green I (Eurogentec) in the presence of PLEKHG5-specific primers (Eurogentec) (table 1). The measurement of the b2 microglobulin gene provided an amplification control that allowed PLEKHG5 expression to be normalized. Each reaction was performed in triplicate, by use of an MX3000P Real-Time PCR System (Stratagene).

Synthesis of Polyclonal Antibody to PLEKHG5
Polyclonal antibodies to PLEKHG5 were obtained by immunization of two rabbits with two synthesized specific peptides (NH2-CYLRVKAPAKPGDEG-CONH2 and NH2-CKVDIYLDQSNTPLSL-CONH2) and were purified on a sepharose column (Covalab). High reactivity of immunopurified antibodies was confirmed by ELISA.

Western-Blot Analysis
Proteins were harvested 48 h after transfection. Cells were washed three times with ice-cold PBS and then were lysed in 0.32 M sucrose or in lysis buffer (10 mM Hepes [pH 7.8], 10 mM KCl, 2 mM MgCl 2 , 0.1 mM EDTA, and 1 mM dithiothreitol) and 10% Nonidet P40 (Sigma), with the addition of the Protease Inhibitors cocktail (Roche). Cell extracts were boiled for 5 min in Laemmli buffer and were submitted to a 7.5% SDS-PAGE. Proteins were electroblotted onto nitrocellulose membranes. After electroblotting, membranes were treated with 5% dry milk in TBS buffer (50 mM Tris [pH 8.1] and 150 mM NaCl) containing 0.05% Tween-20, for 1 h at room temperature, and then were hybridized overnight at 4ЊC with anti-PLEKHG5 antibodies (1:1,000) diluted in blocking solution. The antigen-antibody complexes were revealed by secondary incubation with horseradish peroxidase-coupled goat anti-rabbit immunoglobulin G antibodies (1:10,000 [Sigma]). Immunoreactive proteins were visualized using enhanced chemiluminescence reagents (Perkin Elmer). Hybridiza-

Immunocytochemistry and Microscopy
Transfected HEK293 and MCF10A cells were fixed in 4% formaldehyde 48 h after transfection. We permeabilized cells with 1% Triton X-100 in PBS for 15 min and blocked aspecific fixation sites in 5% nonfat milk for 2 h at room temperature. Cells were incubated with anti-PLEKHG5 polyclonal antibodies (1:100) for 1 h and then with fluorescein isothiocyanate-conjugated goat anti-rabbit antibody (1:1,000 [Sigma]) for an additional 1 h. Cells were examined using a Zeiss Axioplan 2 imaging microscope equipped with ISIS 3 software (MetaSystem).
Transfected NSC34 murine cells were treated for PLEKGH5 immunofluorescence with the polyclonal anti-PLEKGH5 antibody (1:100) coupled with diamidino-4 ,6-phenylindole-2 dichlorhydrate staining of the nuclei. Cells were incubated with the pri- Figure 2. Expression of PLEKHG5 in the human nervous system. PLEKHG5 expression was evaluated in several human nervous tissues by real-time quantitative RT-PCR. All reactions were performed in triplicate, and PLEKHG5 expression was normalized according to b2 microglobulin expression. Results were expressed as amplification rates in comparison with PLEKHG5 expression in untransfected HEK293 cell line. PLEKHG5 transcripts were detected in all studied tissues, supporting its ubiquitous expression pattern in the human nervous system, with a predominance in peripheral nerve and spinal cord. Sm p skeleton muscle; Br p brain; Sc p spinal cord; Pn p peripheral nerve.

Results
We sequenced the 25 candidate and predicted genes located in the 3.9-cM mapped interval on chromosome 1p36 (NPHP4, KCNAB2, CHD5, RPL22, RNF207, C1orf188,  ICMT, C1orf211, HES3, GPR153, ACOT7, HES2, ESPN,  TNFRSF25, PLEKHG5, NOL9, TAS1R1, HKR3, KLHL21, PHF13, THAP3, DNAJC11, CAMTA1, AK098599, and AK128074), as well as two additional candidate genes known to cause Charcot-Marie-Tooth disease (CMT) and located very close to this chromosomal region (MFN2 and KIF1Bb) (fig. 1a). In the five affected members of the Malian family, we found a homozygous mutation (c.1940 TrC) resulting in an amino acid substitution (p.647 PherSer) in the pleckstrin homology (PH) domain of the PLEKHG5 protein ( fig. 1c). The mutation was not detected in 300 healthy controls (600 chromosomes), of whom 250 originated from Mali (500 chromosomes). The mutant phenylalanine is highly conserved across species and is a canonical amino acid residue of the PH consensus sequence ( fig. 1d and 1e). We then screened PLEKHG5 in a sample of four unlinked families and 16 isolated patients with a close phenotype, but we identified no additional mutations.
PLEKHG5 spans 53,917 bp on human chromosome 1 and codes for a member of the Dbl protein family that shares a PH domain and a guanine nucleotide exchange factor for Rho protein (RhoGEF) domain. 16 Interestingly, it has been suggested elsewhere that the PLEKHG5 protein has an NFkB-activating function. 17 Five mRNA isoforms annotated by National Center for Biotechnology Information (NCBI) genome browser and two Mammalian Gene Collection Full ORF mRNAs have been reported to code for proteins that mainly differ by their N-terminal end (GenBank accession numbers NM_020631, NM_198681, NM_001042663, NM_001042664, NM_001042665, BC042606, and BC015231) ( fig. 1b). According to microarray databases (UniGene, Expression Profile Viewer), PLEKHG5 is ubiquitously expressed, but predominantly in the peripheral nervous system and brain. Using western blotting and immunofluorescence, we failed to detect endogenous PLEKHG5 protein in various human cells (fibroblastic cell lines, lymphoblastoid cell lines, cultured amniocytes, and cultured primary hepatocytes) (data not shown). However, using real-time quantitative RT-PCR, we confirmed the ubiquitous expression of PLEKHG5 in the human nervous system, with a predominance in peripheral nerve and spinal cord ( fig. 2).
To further investigate the impact of the c.1940 TrC mutation, two isoforms of PLEKHG5 (BC042606 and BC015231) ( fig. 1b) were transfected in HEK293 and MCF10A cell lines. The wild-type PLEKHG5 protein was not visualized by western blotting or immunofluorescence before transfection. By contrast, after transfection of expression vectors encoding wild-type BC015231 and BC042606 isoforms, we detected protein fragments of 130 and 150 kDa in cell lysates, using polyclonal rabbit anti-PLEKHG5 antibodies ( fig. 3a). In addition, immunofluorescence analysis revealed that wild-type PLEKHG5 proteins were diffusely localized in cytoplasm ( fig. 3b). On the contrary, in cells transfected with the corresponding c.1940 TrC mutant constructs, mutant PLEKHG5 proteins were consistently undetectable by western blotting, suggesting an instability of the mutant variants ( fig. 3a). By immunofluorescence, mutant PLEKHG5 proteins were not detected with a classic exposure time (0.04 s). However, a light, diffuse cytoplasmic signal could be detected after a longer exposure time (0.34 s) (fig. 3b).
The two isoforms of PLEKHG5 were then tested for their ability to activate the NFkB pathway. HEK293 and MCF10A cell lines were transfected with a luciferasereporter gene responsive to NFkB, together with expression vectors encoding either wild-type or c.1940 TrC mutated PLEKHG5 cDNAs. Although the PLEKHG5 transcript was barely detected in untransfected control cells by use of real-time quantitative RT-PCR, its abundance increased up to 1,000-fold after transfection in cells transfected with either wild-type or mutant PLEKHG5 variants ( fig. 4). Consistently, the luciferase activity was more than sixfold higher in the wild-type PLEKHG5-transfected cells than in the control cells. By contrast, induction of the luciferase activity was markedly reduced in cells transfected with the mutant isoforms ( fig. 5a), therefore demonstrating that the amino acid substitution severely impaired PLEKHG5 ability to activate the NFkB pathway. This loss of activity clearly reflects, at least in part, the instability of the mutant protein.
Finally, we transiently transfected murine motor neuronal NSC34 cells with the two wild-type BC015231 and BC042606 isoforms and the corresponding mutant constructs. Western-blotting experiments confirmed the in- Figure 4. Quantification, by real-time RT-PCR, of PLEKHG5 expression levels in untransfected and transfected HEK293 and MCF10A cells. At 48 h after transfection, the cells were harvested and were analyzed by real-time RT-PCR for PLEKHG5 mRNA. The fluorescence was measured in every PCR cycle and was proportional to the accumulation of PCR product. Endogenous expression of PLEKHG5 was detected in untransfected control cells but was markedly increased in transfected cells (up to 1,000-fold, estimated by DDCt method). Overexpression of wild-type (WT) and mutated PLEKHG5 in transfected cells was similar. Triplicate analyses were performed for each sample. stability of the mutant PLEKHG5 proteins, which accumulated to much lower level than that of the wild-type proteins ( fig. 3c). Detection of the mutant PLEKHG5 proteins by immunofluorescence revealed formation of aggregates in the motor neuron somas close to the nucleus in the majority (60%-70%) of the transfected cells ( fig.  3d). This was observed neither in NSC34 cells transfected with the wild-type counterparts nor in nontransfected cells.
In conclusion, transfection experiments using PLEKHG5 variants in distinct cellular models supported instability and loss of NFkB-activating function of the mutant PLEKHG5 proteins and revealed their involvement in aggregate formation in transfected murine motor neuron cells.

Discussion
The mechanism by which the mutation in the PH domain of PLEKHG5 leads to an autosomal recessive, generalized LMND is unclear. PH domains are protein modules of ∼100 aa, found in a wide range of eukaryotic proteins, many of which are involved in cell signaling and cytoskeletal regulation. They play a membrane-anchoring function because of their ability to bind to phosphoinositides. 18 In Dbl-family proteins, including PLEKHG5, PH domains also independently contribute to the allosteric regulation of the RhoGEF domain. This latter domain activates GTPases by stimulating the exchange of GDP to GTP, thereby initiating various signaling mechanisms that regulate neuronal shape and plasticity, dendrite growth, synapse formation, and neuronal survival. [19][20][21][22] Recent experiments show that mutations in the PH domain, impairing phosphoinositide binding, do not systematically affect protein subcellular localization. However, in all cases, these mutations significantly reduce the guanine nucleotide exchange activity of the Dbl proteins. 23 Two proteins sharing a PH domain or a PH/RhoGEF domain have already been shown to account for human neurodegenerative diseases: Dynamin 2 (encoded by DNM2) and alsin (encoded by ALS2  [26][27][28][29][30][31][32] The alsin protein is composed of three guanine nucleotide exchange factor domains, which result in Ran, Rho/Rac1, and Rab5 guanine nucleotide exchange activities, involved in motor neuron maintenance, axonal transport, and neurite outgrowth. 10,11 The crucial role of the PH/ RhoGEF domain for the alsin-mediated neuroprotection against the toxic effect of the mutant form of SOD1 has recently been demonstrated. 33,34 Similar to PLEKHG5, endogenous alsin is a low-abundance protein enriched in neural tissues. 35 The reported mutations of the ALS2 gene are rarely located in the PH domain. However, all but one of the causative mutations generate alsin protein truncation, often leading to the loss of the PH domain. Except for the recently reported missense G540E mutation, a lossof-function mechanism was attributed to the causative homozygous mutations in the ALS2 gene. Indeed, when expressed in cultured HEK293 cells, most of the diseaseassociated mutant forms are unstable and are rapidly degraded by the proteasome, as observed with mutant PLEKHG5. 35 The Rho family of GTPases are known to activate the NFkB signaling pathway. 36,37 Here, we confirm the NFkBactivating function of PLEKHG5. The NFkB signaling pathway has not been found to be directly involved in human motor neuronopathies. However, its neuronal antiapoptotic role has been documented in various cellular models, and several studies showed that inhibition of the NFkB pathway promoted apoptosis in neurons. [38][39][40][41][42][43] Moreover, down-regulation of NFkB activity has recently been observed in the expanded polyglutamine protein-expressing neuro-2a cells, suggesting that this pathway could be involved in the pathogenesis of polyglutamine-degenerative disorders, including Kennedy disease (MIM 313200), a spinobulbar muscular atrophy. 44 Therefore, considering its wide CNS expression pattern and its NFkB-stimulating activity, PLEKHG5 might play a role in neuronal maintenance. Indeed, a mutation in the PH domain leading to the loss of NFkB activation could fail to protect neurons against apoptosis and could compromise the neuronal survival.
Finally, we showed that mutant PLEKHG5 variants formed aggregates in transfected NSC34 murine motor neurons. Interestingly, aggregates were never observed in wild-type-transfected NSC34 cells, and formation of mutant PLEKHG5 aggregates appeared to be specific to the neuronal cell line, because they were not observed in transfected HEK293 and MCF10A cells. This last finding argues against the hypothesis of an artifact of overexpression. 45 Animal models or human postmortem histological studies would be useful for evaluating whether this observation in vitro is correlated with intracellular inclusions in vivo. Although the neurotoxic or neuroprotective effect of aggregates is still debated, abnormal aggregation is a common feature in several neurodegenerative diseases, including motor neuron diseases. [45][46][47][48][49] In particular, protein aggregation has been recognized as a characteristic change in degenerating motor neurons of amyotrophic lateral sclerosis (MIM 105400) with mutant SOD1. 50 Moreover, transfection experiments revealed abnormal aggregation of various mutated proteins involved in autosomal dominant distal hereditary motor neuropathies, such as the small heat-shock proteins 1 and 8 (HSPB1 and HSPB8), the seipin protein (Bscl2), and the dynactin protein (DCTN1). 8,9,[51][52][53] In these examples, there is increasing evidence that the misfolded protein and aggregate structures lead to a dominant pathological phenotype by a toxic gain-of-function mechanism or by the combination of both loss of function and toxic gain of function. 9,51,[53][54][55][56] Conversely, specific aggregation of mutant proteins involved in autosomal recessive LMNDs has not yet been reported.
Whether PLEKHG5 is directly or indirectly involved in RNA processing or axonal transport in motor neuron, as postulated for the other LMND-causative genes, remains to be further explored. As already described in neurodegenerative diseases with aggregate accumulation, mutant PLEKHG5 could generate toxic interaction with various intracellular partners, including specific components of the axonal cytoskeleton, leading to the disruption of the anterograde transport pathway in motor neurons. 7,8,11,33,51,[56][57][58][59][60] In conclusion, we have demonstrated that a form of LMND with childhood onset is caused by a homozygous mutation (c.1940 TrC [p.647 PherSer]) in the PLEKHG5 gene. This mutation caused protein instability, impaired the ability of PLEKHG5 to activate the NFkB pathway in transfected HEK293 and MCF10A cell lines, and eventually led to aggregate formation in a transfected murine NSC34 motor neuron cell line. The observation of a mutation of the PLEKHG5 gene in a motor neuron degenerative disorder suggests that the RhoGEF-mediated NFkB signaling pathway plays an important function in motor neuron maintenance and could be involved in other human neuronopathies as well.