Pathobiology of Christianson Syndrome: Linking Disrupted Endosomal-Lysosomal Function with Intellectual Disability and Sensory Impairments

Abstract

Christianson syndrome (CS) is a recently described rare neurogenetic disorder presenting early in life with a broad range of neurological symptoms, including severe intellectual disability with nonverbal status, hyperactivity, epilepsy, and progressive ataxia due to cerebellar atrophy. CS is due to loss-of-function mutations in SLC9A6, encoding NHE6, a sodium-hydrogen exchanger involved in the regulation of early endosomal pH. He19re we review what is currently known about the neuropathogenesis of CS, based on insights from experimental models, which to date have focused on mechanisms that affect the CNS, specifically the brain. In addition, parental reports of sensory disturbances in their children with CS, including an apparent insensitivity to pain, led us to explore sensory function and related neuropathology in Slc9a6 KO mice. We present new data showing sensory deficits in Slc9a6 KO mice, which had reduced behavioral responses to noxious thermal and mechanical stimuli (Hargreaves and Von Frey assays, respectively) compared to wild type (WT) littermates. Immunohistochemical and ultrastructural analysis of the spinal cord and peripheral nervous system revealed intracellular accumulation of the glycosphingolipid GM2 ganglioside in KO but not WT mice. This cellular storage phenotype was most abundant in neurons of lamina I-II of the dorsal horn, a major relay site in the processing of painful stimuli. Spinal cords of KO mice also exhibited changes in astroglial and microglial populations throughout the gray matter suggestive of a neuroinflammatory process. Our findings establish the Slc9a6 KO mouse as a relevant tool for studying the sensory deficits in CS, and highlight selective vulnerabilities in relevant cell populations that may contribute to this phenotype. How NHE6 loss of function leads to such a multifaceted neurological syndrome is still undefined, and it is likely that NHE6 is involved with many cellular processes critical to normal nervous system development and function. In addition, the sensory issues exhibited by Slc9a6 KO mice, in combination with our neuropathological findings, are consistent with NHE6 loss of function impacting the entire nervous system. Sensory dysfunction in intellectually disabled individuals is challenging to assess and may impair patient safety and quality of life. Further mechanistic studies of the neurological impairments underlying CS and other genetic intellectual disability disorders must also take into account mechanisms affecting broader nervous system function in order to understand the full range of associated disabilities.

1. Introduction:

1.1. Christianson Syndrome: clinical features and genetic basis

Christianson syndrome (CS) is a rare X-linked monogenic disorder with variable expressivity and a broad range of neurodevelopmental and neurologic features. Core clinical features (occurring in >85% of affected males) present in early childhood and include moderate to severe intellectual disability (ID), nonverbal status, epilepsy, truncal ataxia, postnatal microcephaly, and hyperactivity [1]. Secondary symptoms frequently experienced by CS patients include cognitive and motor regressions including loss of walking ability [1–6], cerebellar atrophy as detected by MRI [1–5, 7–9], low weight for age [1–3, 7, 10–11], eye movement problems [1–4], and features of Angelman syndrome such as a happy demeanor [1, 3, 5–8].

CS was first clinically described in 1999 in a 5-generation South African family with an X-linked intellectual disability (XLID) syndrome comprising 16 affected males and 10 carrier females [2]. The causative gene was identified in 2008, when DNA sequencing in a separate family with an XLID syndrome resembling Angelman Syndrome (but without UBE3A mutations) identified a 6 base pair deletion in the SLC9A6 gene, residing on Xq26.3 [7]. DNA was then examined from affected males within the original family described as having CS, due to the shared clinical features and overlapping X chromosome interval identified in an earlier linkage analysis. This uncovered a separate deletion mutation in SLC9A6 [7]. To our knowledge, over 60 CS patients have now been reported in the literature, with 31 distinct mutations in SLC9A6 [1–17]. The majority of mutations in SLC9A6 that cause CS are protein truncating mutations thought to lead to a loss of protein function [1, 18]; for a detailed review of SLC9A6 mutations and their relationship to CS and other neurological disorders, we direct the reader to Kondapalli et al 2014 [18] and Pescosolido et al 2014 [1]. Data from recent genomic studies suggest that mutations in SLC9A6 may be among the most common causes of XLID, potentially affecting thousands of patients worldwide [14, 19, reviewed by Pescosolido et al [1]].

SLC9A6 encodes NHE6, an endosomal sodium-hydrogen exchanger thought to be involved in the regulation of endosomal pH (Figure 1), as well as the processing and trafficking of intracellular cargo [18]. NHE6 is one of nine members of the SLC9A subgroup of the SLC9 gene family [20–21]. SLC9A comprises genes encoding the sodium hydrogen exchangers (NHEs), of which there are nine members. NHE1-NHE5 mostly localize to the plasma membrane, while NHE6–9 localize to membranes of intracellular organelles, although the relative distribution among organelles for each NHE may depend on cell type [21–22]. Plasma membrane NHEs are involved with regulation of cytoplasmic pH, while organellar NHEs participate in the regulation of organelle-specific pH, helping to establish the pH gradients along the secretory and endocytic pathways that are essential for intracellular trafficking and sorting of cargo along these pathways [18, 21–22]. A notable difference between plasma membrane and organellar NHEs is their ion selectivity; while plasma membrane NHEs are selective for Na⁺, organellar NHEs may transport either Na⁺ or K⁺ [22–23]. This enables their ability to regulate the pH within organellar compartments, through exchange of cytosolic K⁺ for luminal H^+, thus promoting luminal alkalization (Figure 1). In addition to NHE6, NHE9 is associated with endosomal populations, and mutations in the cognate gene SLC9A9 have also been linked to neurological disease (autism, epilepsy, and attention deficit hyperactivity disorder) [24–31]. Interestingly, a study analyzing microarray datasets from cerebral cortices of patients with idiopathic autism found that expression levels of NHE6 and NHE9 trended away from average, with NHE6 expression downregulated and NHE9 upregulated compared to healthy controls [32]. These clinical and genetic studies clearly demonstrate that endosomal sodium-hydrogen exchangers have important roles in the development and function of the nervous system. To date, no other NHE proteins have been linked to neurological disease in humans.

Figure 1:

Endosomal NHEs (NHE6 and NHE9) are thought to participate in regulation of luminal pH within the early and recycling endosomes on which they reside, through electroneutral exchange of luminal H⁺ with cytoplasmic Na⁺ or K ⁺. This exchange has an overall alkalizing effect, to counterbalance the acidification provided by the H⁺-pumping V-ATPase.

1.2. Neuropathogenesis of Christianson Syndrome: A Review

The molecular and cellular roles of NHE6, as well as the consequences of its loss of function, have been studied both in vitro and in vivo. Here we will briefly review the progress that has been made in elucidating the roles of NHE6 in the nervous system, as well as understanding the pathogenic cascades contributing to CS and related phenotypes.

1.2.1. Insights from in vitro studies of NHE6 expression, subcellular localization and molecular function:

The spatiotemporal expression of NHE6 within the nervous system, as well as its subcellular distribution, may provide clues as to its functions as well as the regions that may be compromised by NHE6 deficiency and thus implicated in CS pathogenesis. In situ hybridization and RNA-seq studies [Allen Brain Institute] have revealed that NHE6 is highly expressed in the human (and mouse) brain, is enriched in regions critical for cognitive and intellectual function (hippocampus, cortex) and that its expression levels vary throughout prenatal and postnatal life (for a detailed review of this data, see Kondapalli et al 2014, [18]). Overall, NHE6 expression is highest during the prenatal period, decreases postnatally and peaks again during adulthood (after age 21).

Studies in several cell types have established that NHE6 localizes mainly to early and recycling endosomal populations [23, 33–37] and also appears transiently on plasma membranes [34, 38]. In developing neurons, NHE6 also localizes to growing axon tracts [36]. In hippocampal slice cultures as well as primary dissociated neurons, NHE6 expression in pyramidal neurons has been found to increase during synaptogenesis [35]. In mature neurons, NHE6 has been found in dendritic spines and co-localizes with pre- and post-synaptic markers [35–36]. During a chemical induction of NMDA receptor-mediated long-term potentiation (LTP), NHE6-containing vesicles have been shown to move into dendritic spine heads [35]. These findings are consistent with NHE6 being involved in cellular processes underlying neuronal development, synaptogenesis, and synaptic function.

Loss of NHE6 has been demonstrated to impact endosomal pH [22, 36, 39], and to have consequences for the processing and turnover of intracellular cargo. In HeLa cells, NHE6 was found to colocalize with clathrin and transferrin, and knockdown of NHE6 led to endosomal overacidification and reduced uptake of transferrin, with no effect on the endocytosis of epidermal growth factor or the cholera toxin B subunit. Conversely, over-expression of WT NHE6 (but not a mutant form defective in ion transport) led to an increase in endosomal pH as well as increased transferrin uptake [39]. In addition to affecting endocytosis, trafficking of internalized cargo is likely to be affected by a reduced endosomal pH. While the mechanisms regulating the trafficking of cargo throughout the endocytic pathway are not completely understood, it is well established that endosomal compartments maintain tightly regulated pH ranges [40–41]. In cells treated with Bafilomycin A1, which inhibits the vacuolar H⁺-ATPase and has an alkalizing effect on endosomal compartments, the movement of cargo from early endosomes to late endosomes and lysosomes is slowed [42–43], which suggests that pH-dependent mechanisms help to regulate cargo trafficking through these pathways. Studies of the function of Nhx1, the yeast orthologue to the endosomal NHE proteins, have further helped to shed light on the importance of pH regulation by Na⁺/H⁺ exchangers for vesicular trafficking. Loss of function mutations in Nhx1 led to abnormal intracellular trafficking of surface receptor Ste3, which accumulated in aberrant endosomal compartments rather than being degraded in the vacuole [44].

Identification of specific NHE6-dependent endosomal cargo has helped to shed light on potential mechanisms underlying neuronal dysfunction in CS. One such cargo is TrkB, and reduced intra-endosomal TrkB signaling is a proposed mechanism impacting neuronal circuit development in CS. Neurons cultured from Slc9a6 KO mice were shown to have reduced dendritic and axonal arbors, increased endosomal acidification, and attenuated TrkB signaling in response to BDNF application compared to WT neurons. Transfection with WT NHE6 (but not a mutant form lacking transporter function) rescued these phenotypes. Interestingly, application of leupeptin, a protease inhibitor, also rescued the attenuated TrkB signaling, suggesting that premature degradation of TrkB within acidified endosomes, or enhanced degradation within lysosomes, is responsible for the reduction in signaling [36]. Intra-endosomal BDNF signaling through TrkB helps to drive dendritic and axonal growth [45], and an attenuation in this process may thus explain some of the developmental phenotypes of CS, including postnatal microcephaly and intellectual disability. NHE6 has also been shown to colocalize with AMPARs in a chemical model of LTP [35], raising the possibility that alterations in neurotransmitter receptor trafficking may also occur in CS. Indeed, recycling endosomes supply AMPARs for LTP, and endosomal trafficking is thought to be a mechanism involved in regulating this process [46–47]. Disruption in mechanisms of synaptic plasticity could thus provide a partial explanation for intellectual disability in CS. A study in HEK293 cells expressing human amyloid precursor protein (APP) found that knockdown of NHE6 caused an elevation in Aβ production. Conversely, NHE6 over-expression led to endosomal alkalization and decreased APP processing/ Aβ secretion by shifting APP away from the trans-Golgi network into early and recycling endosomes. Interestingly, this study also found that NHE6 mRNA and protein levels are reduced in AD [48]. Thus, aberrant processing of APP due to endosomal pH dysregulation could be a mechanism contributing to neurodegeneration in CS as well as AD and related amyloid disorders.

Taken together, these studies highlight specific endosomal mechanisms that are disrupted by NHE6 loss of function, and implicate processes related to neurodevelopment, synaptic function, and neurodegeneration in the pathobiology of CS.

1.2.2. Insights from in vitro studies of specific SLC9A6 mutations:

A few studies have begun to define the molecular consequences of specific SLC9A6 mutations. The majority of CS-causing mutations lead to premature translation termination, which is presumed to result in a complete loss of function due to clearance of the truncated protein or nonsense-mediated decay. Indeed, nonsense-mediated decay was confirmed in a study of one such protein-truncating mutation (c.441delG, p.S147fs), resulting in the absence of detectable NHE6 protein in RT-PCR and western blot experiments performed on lymphoblastoid cells cultured from the patient [6]. SLC9A6 is alternatively spliced, with two resulting isoforms, isoform a (701 amino acids), and isoform b (669 amino acids). The S147 frame shift mutation affected only isoform a, yet the patient phenotype was described to be highly similar to the original CS cases reported by Gilfillan et al [7], suggesting that this isoform is important for brain function.

In addition to protein truncating mutations, two small in-frame deletion mutations in SLC9A6 have been identified, which have been extensively studied in vitro. A 6-bp mutation that results in deletion of amino acids E255-S256 of the NHE6 protein was the first SLC9A6 mutation to be identified by Gilfillan et al [7]. These residues are situated in a region within the predicted 7^th transmembrane domain that is highly conserved among all human NHEs, and mutation of the analogous glutamate residue in the plasma membrane transporter NHE1 impacts both the subcellular distribution of this transporter as well as its intrinsic catalytic activity [49]. In HeLa cells, exogenous expression of NHE6v2Δ²⁵⁵ES²⁵⁶ (v2ΔES), the product of this deletion within the shorter NHE6 isoform b, led to rapid degradation through both lysosomal and proteasomal pathways [50]. Subsequent studies of NHE6v1Δ²⁸⁷ES²⁸⁸ (v1ΔES), the analogous mutation in the longer splice variant (NHE6v1) giving rise to isoform a, corroborated these earlier findings and in addition found that this mutation impacts oligosaccharide maturation, reducing sorting to the plasma membrane and recycling endosomes [51]. In addition to reduced stability, ΔES apparently impacts the catalytic activity of NHE6, as v1ΔES-associated vesicles were found to be more acidic than WT-vesicles in a pH assay performed in transfected AP-1 cells [51]. Another in-frame deletion mutation, which results in the loss of three residues (³⁷⁰Trp-Ser-Thr³⁷², v1ΔWST) bordering the predicted 9^th transmembrane domain, was found in a family with a clinical presentation comprising a subset of CS features as well as late-onset neurodegenerative features, including widespread neuronal loss and tau inclusions [15]. This mutation was similarly characterized and found to be retained in the ER, with reduced oligosaccharide maturation, and diminished sorting to the plasma membrane and recycling endosomes [52]. Both v1ΔES and v1ΔWST mutations were associated with impaired clathrin-mediated uptake of transferrin, and decreased dendritic arborization in neuronal cultures when exogenously expressed [51–52].

These studies have helped to establish a clear link between the loss of function of NHE6 and the CS clinical phenotype. In addition to a loss of function, the molecular characterization of the ΔES and ΔWST mutants suggest that issues with post-translational processing within the ER may be occurring in some cases. ER stress is thought to contribute to the pathophysiology of a number of neurodegenerative disease in which misfolded and aggregated proteins accumulate [53–54], and the resulting unfolded protein response (UPR) is a newly recognized target for therapeutic intervention [55–56]. Further studies aimed at elucidating the specific cellular consequences of individual SLC9A6 mutations are warranted to gain insight into the pathogenesis of CS and to identify therapeutic targets.

1.2.3. Insights from the Slc9a6 KO mouse

Extensive investigation of the neuropathology and behavioral phenotype of Slc9a6 knockout (KO) mice have led to the identification of selective cellular vulnerabilities within brain regions that may be key drivers of the clinical phenotype of CS. Intriguingly, Slc9a6 KO (male hemizygous Slc9a6 ^-/Y and female homozygous Slc9a6 ^−/−) mice have neuropathology similar to that occurring in numerous lysosomal diseases of the brain. This includes the late endosomal/lysosomal accumulation of GM2 ganglioside and unesterified cholesterol. Importantly, this occurs in neurons only within select brain regions, notably the basolateral amygdala, dentate gyrus, CA3/CA4 pyramidal neurons of the hippocampus, and parts of the cerebral cortex [57–58]. Within the hippocampus, a reduced activity of β-hexosaminidase, the lysosomal enzyme that normally degrades GM2, was detected by histochemical assay, suggesting that NHE6 deficiency could lead to alterations in the synthesis, catalytic activity, or trafficking of this and possibly other lysosomal enzymes. Additionally, Purkinje neurons within the cerebellum of Slc9a6 KO mice were shown to degenerate in a progressive and patterned manner associated with the formation of axonal spheroids [57–58]. This cerebellar pathology has been described in mouse models of numerous lysosomal storage diseases such as Niemann-Pick type C [59–61]. Progressive changes have also been identified in additional areas of the brain in Slc9a6 KO mice, including the hippocampus, striatum, and cortex, which occur in a manner consistent with both developmental (undergrowth) and degenerative causes [62].

At the behavioral level, male KO mice exhibit several of the relevant phenotypic features of CS, such as hyperactivity [57–58], visuospatial memory deficits [58], and progressive motor dysfunction as manifested by increased slips while crossing a balance beam compared to wild type littermates [58]. Heterozygous female Slc9a6 KO (Slc9a6 ^X/-) mice also suffer from visuospatial memory deficits as well as motor incoordination, with a less severe phenotype than hemizygous males [58]. This correlated with mosaic neuropathology and Slc9a6 expression due to X chromosome inactivation. These findings raise the possibility that female SLC9A6 carriers may also exhibit clinical features resulting from their mutations. Indeed, there is clinical evidence of neuropsychiatric features in certain SLC9A6 carrier females, including intellectual disability and behavioral disturbances [1–3, 13]. Further clinical studies are needed to appreciate the full phenotypic range of this SLC9A6 carrier population.

1.3. Christianson Syndrome and Aberrant Pain Perception

Disrupted sensory function, including an apparent insensitivity to pain, is a phenotypic feature of CS that has recently come to our attention. While only a single study mentions this issue [1], unanimous concerns were brought up by parents of sons with CS who were in attendance at the 2015 International Clinical and Basic Science Conference on Christianson Syndrome. These included anecdotes of serious injuries such as broken bones, which in some cases would go unidentified by parents and caretakers. Pain is particularly challenging to assess in nonverbal and/or intellectually disabled patients, and CS patients frequently exhibit a characteristic happy demeanor, further complicating its identification. Our goal was to explore sensory function at the behavioral level in Slc9a6 KO mice, as well as to identify any region-specific pathologies that may help to explain the cellular basis of noted phenotypes. Here we demonstrate deficits in behavioral responses to noxious thermal and mechanical stimuli in Slc9a6 KO mice. Our neuropathological examination revealed evidence of lysosomal dysfunction in a select population of neurons in the superficial dorsal horn of the spinal cord, a major relay site in the processing of painful stimuli, as well as changes in astroglial and microglial populations throughout the gray matter of the spinal cord. These findings establish the Slc9a6 KO mouse as a relevant tool for studying the basis of altered sensory processing in CS, and highlight regions of selective vulnerability that may be contributing to this phenotype.

2. Materials and Methods

2.1. Mouse generation:

The Slc9a6 KO mice used in this study were from a colony maintained at the Albert Einstein College of Medicine, originally obtained from Jackson Laboratories (Stock# 005843, strain name B6.129P2-Slc9a6<tm1Dgen). This mutant was originally created by Deltagen through insertion of the LacZ reporter gene, encoding β-galactosidase, into the Slc9a6 genomic locus. See our previous work for validation of this knockout model [57]. Heterozygous female (Slc9a6^+/X) mice were crossed to WT male (Slc9a6^+/Y) mice to obtain littermate WT and hemizogous (Slc9a6^-/Y) mice (referred to here as Slc9a6 KO). Mice were genotyped upon weaning at postnatal day 21 (P21), prior to conducting behavioral assessments. All procedures used in experiments involving mice were approved by the Institutional Animal Care and Use Committee of the Albert Einstein College of Medicine.

2.2. Behavioral studies:

Only male KO (N=8) and WT (N=10) mice were used for behavioral assays. Behavioral assays were performed by an experimenter blind to genotypes.

2.2.1. Hargreaves thermal assay:

The Hargreaves assay was used to measure sensitivity to thermal stimulation, using the Plantar Analgesia Meter (IITC, CA, USA). Awake, unrestrained mice were placed on the glass base heated to 35°C for one hour prior to behavioral assessment. The radiant heat light source was set at fixed intensities (30%, 40%, and 50% of maximum) and focused on the plantar surface of each mouse’s hind paw in order to cause warming and the paw withdrawal latency was recorded. A total of four trials at each intensity were included for each mouse, with at least five minutes between trials, and a cutoff time of 20 seconds was imposed to avoid tissue damage.

2.2.2. Von Frey:

Mechanical sensitivity was assessed using Von Frey filaments over a range of diameters (Stoelting, Wood Dale, IL, USA). Awake, unrestrained mice were placed over a wire grid that enabled access to their hindpaws. Each filament was applied to the plantar surface of the hindpaw with just enough pressure to induce buckling of the filament, which occurs at a known force at each diameter. For each mouse, ten trials of each filament diameter were applied, in order of ascending diameter, with at least 30 seconds between trials. Stimuli were applied only when mice were stationary, with all four paws flat (i.e., not rearing). Behaviors that were scored as responses to the applied stimuli included paw withdrawals, toe curls, and lifting or moving the foot. Once an animal had produced responses to at least 8 out of 10 trials for two consecutive filament diameters, no further stimuli were applied and the animal was scored as 8/10 for subsequent filaments.

2.3. Tissue collection:

Mice used for tissue collection were deeply anesthetized with ketamine/xylazine. Once insensate, animals were perfused transcardially with 0.9% sodium chloride, followed by 4% paraformaldehyde (PFA) in 0.1 M phosphate buffer. Neural tissues were collected, including brain, spinal cord, dorsal root ganglia, and sciatic nerves. These tissues were immersed for overnight fixation in the 4% PFA solution at 4°C, then stored in 0.1 M phosphate buffer at 4°C until further processing.

2.4. Tissue preparation and histochemistry:

Tissues used for immunolabeling studies were cryoprotected by placing in sucrose solutions (15% then 30% sucrose in PBS) until sinking. Tissues were then embedded in OCT by freezing in ethanol over dry ice. Tissues were then sectioned (35 μm for spinal cord, 20 μm for dorsal root ganglia) using a cryostat (Leica CM3050 S). For immunofluorescent staining, tissues were first blocked for 60 minutes at room temperature with shaking (10% goat serum, 0.02% saponin [anti-β-galactosidase, -GM2, -GFAP, -CD68, -NeuN] or 0.4% Triton [anti-NHE6], 1x PBS). Sections were then incubated with primary antibody in the same blocking solution overnight at 4°C with shaking. Primary antibodies used were rabbit antiNHE6 (generously provided by Dr. John Orlowski, Mcgill University, Montreal, Canada [35]); mouse anti-GM2 (1:50, mouse IgM, cell culture supernatant produced in-house by Dr. K. Dobrenis [63]), chicken anti-E.coli β-galactosidase (ab9361, 1:1000, Abcam, Cambridge, MA); mouse antiglial fibrillary acidic protein (GFAP) mAb G-A-5 (G3893, 1:1000, Sigma-Aldrich); rat anti-CD68 mAb (MCA1957, 1:1000; AbD Serotec, Kidlington, UK); mouse anti-NeuN (neuronal nuclei) IgG mAb (MAB377, 1:1000, Chemicon). Species-specific secondary antibodies conjugated to Alexa Fluor 488 and 546 dyes (Invitrogen, Carlsbad, CA, USA) were used for detection of primary antibodies. For IB4 staining, some spinal cord sections were subsequently incubated with IB4 lectin (FITC-conjugated, 25 μg/mL in PBS, L2895, Sigma) for 2 hours. For Luxol fast blue staining of spinal cord and sciatic nerve, tissues were processed using an ASP300 automated processor, hand-embedded in paraffin blocks, and sectioned on a microtome at 5 μm. Sections were stained for myelin by Luxol fast blue and counter-stained with cresyl violet.

2.5. Confocal microscopy:

Confocal imaging was performed with a Zeiss Meta 510 Duo V2 laser scanning confocal microscope using Plan-Apochromat 20X/0.8, C-Apochromat and 40 X/1.2 water UV-VIS-NIR objectives and 351 nm, 488 nm, 543 nm laser lines. Acquisition channels were combined to produce merged images, and analyzed using ImageJ software and MetaMorph software (Molecular Devices).

2.6. Light microscopy:

Myelin-stained sections of spinal cord and sciatic nerve were examined and photographed using an Olympus AX70 upright epifluorescence microscope equipped with MagnaFire CCD camera from Optronics (Goleta, CA).

2.7. Electron microscopy:

After perfusion and fixation described above, spinal cord, dorsal root ganglia, and sciatic nerve tissue used for electron microscopy were post-fixed in 2.5% glutaraldehyde in 0.1 M cacodylate buffer. Tissue samples were then transferred to 2% osmium tetroxide, followed by alcohol dehydration and plastic embedding in Epon. Sections were cut on an ultramicrotome and stained with uranyl acetate and lead citrate. A Philips CM10 electron microscope equipped with an 11 Megapixel Morada G2 TEM camera (Olympus) were used for tissue analysis.

2.8. Statistical analysis and graphical representation:

All data were compiled and statistically analyzed using GraphPad Prism Software Version 7.0a and JMP software from SAS Version 13.0.0 (Figure 2A–2C, 6B). To determine the statistical significance of the sensory behavioral data, repeated measures ANOVA (genotype x stimulus intensity) was performed and a difference was deemed significant if p<0.05. When the repeated measures ANOVA detected significant differences, a post hoc t-test analysis with Bonferroni correction was performed at each stimulus intensity. For the quantitative analysis of myelin in sciatic nerves, G-ratios were calculated from TEM micrographs as the diameter of the axon divided by the diameter of the axon and myelin. A minimum of 80 axons per mouse, and 3 mice per genotype, were analyzed.

Figure 2: Sensory behavioral phenotype.

A-i: Results from the Hargreaves thermal assay performed in mice (N=10 WT, 8KO) at 2 months of age and 8 months of age show that KO mice have significantly increased latencies to hindpaw withdrawal from noxious thermal stimuli at 2 months of age, compared with their WT littermates (repeated measures ANOVA, p = 0.0026). Despite the apparent trend toward higher response latencies at 8 months, this was not significant (repeated measures ANOVA, p = 0.18). Post hoc t-test comparisons are shown at individual stimulus intensities, adjusted (black asterisk and font) and non-adjusted (blue asterisk and font) via Bonferroni method for multiple comparisons. A-ii: Bar graphs with individual data points overlaid for the 50% stimulus intensity at 2 months and 8 months. Post hoc t-tests with Bonferroni correction. B-i: Tactile sensitivity at 8 months of age (N=10 WT, 8 KO) was assessed with Von Frey filaments, revealing a statistically significant reduction in the responsiveness of KO mice compared with WT over a range of fiber diameters (repeated measures ANOVA, p = 0.023). Post hoc t-test comparisons are shown at individual stimulus intensities, adjusted (black asterisk and font) and non-adjusted (blue asterisk and font) via Bonferroni method for multiple comparisons. B-ii: Bar graphs with individual data points overlaid at representative filament diameters (2g, 4g and 6g). Post hoc t-tests with Bonferroni correction. In A and B, gray circles represent values for KO, white triangles represent values for WT, and scale bars indicate standard error. s = seconds, ns = nonsignificant

Figure 6:

Analysis of myelination in spinal cord (A) and sciatic nerve (B) in 8 month old WT and Slc9a6 KO mice. A. Luxol fast blue staining of paraffin embedded spinal cord sections revealed no apparent differences in myelination of WT (i) and Slc9a6 KO mutant (ii) spinal cord tracts. The region shown here contains the spinothalamic tract and adjacent ventral horn. B. Myelination of sciatic nerve neurons was assessed by analyzing TEM micrographs and calculating the g ratio, defined as the diameter of the axon divided by the diameter of the axon and myelin. i. Example

3. Results

3.1. Slc9a6 KO mice have reduced behavioral responses to noxious thermal and mechanical stimuli:

We conducted behavioral assays to assess sensory functioning in Slc9a6 KO mice. Mice (N=10 WT, 8 KO) were evaluated for responsiveness to thermal stimuli at 2 months and 8 months of age using the Hargreaves assay. Overall, KO mice showed greater latencies to paw withdrawal from a heat stimulus compared with WT over a range of stimulus intensities at 2 months of age (Figure 2A; repeated measures ANOVA, F_(1,16) = 12.7455, p = 0.0026). While there was a trend toward higher response latency in the KO animals compared to WT through all stimulus intensities at the 8-month time point, this was not significant (Figure 2B; repeated measures ANOVA, p = 0.18). In this assay, we found that the age at testing had a significant effect on withdrawal latency in our cohort, independent of genotype (within-subjects analysis, F_(2,32) = 7.0036, p =0.003).

To further examine potential sensory processing abnormalities at 8 months of age, we used Von Frey filaments to evaluate responses to mechanical stimuli. We recorded the number of paw withdrawals elicited during ten applications of each filament to the plantar surface of the hindpaw (the force required to cause the filament to buckle is designated in units of gravity (g)) (Figure 2B). Our results indicate that, compared to WT, KO mice have significantly reduced behavioral responses to filaments over a range of diameters (repeated measures ANOVA F_(1,16) = 5.893, p = 0.023).

Taken together, our findings suggest that Slc9a6 KO mice are less sensitive to noxious thermal and tactile stimuli than their WT littermates.

3.2. Spinal cords of Slc9a6 KO mice have lysosomal disease-like pathology:

Next, we conducted immunohistochemical (IHC) studies to evaluate pathology in the dorsal root ganglia (DRG) and spinal cord, areas that have not been previously investigated in Slc9a6 KO mice.

Given our previous findings of GM2 ganglioside storage occurring in neurons within the brains of Slc9a6 KO mice [57], we looked for its accumulation in DRG and spinal cord, which might indicate regions of selective vulnerability to NHE6 deficiency. Interestingly, GM2 was detected within select neurons of the superficial dorsal horn (Figure 3A-C). GM2-containing neurons were found throughout lamina I-III of the dorsal horn, with the majority concentrated in laminae I-II, just dorsal and ventral to the IB4-labeled terminals of a subset of nonpeptidergic primary nociceptive afferents that arborize in lamina II (Figure 3C).

Figure 3:

A-B: Lumbar sections of spinal cords from 2-month old Slc9a6-KO and WT male mice were labeled with primary antibodies against GM2 (red) and NeuN (green). A. Maximum intensity projections from 21 confocal planes are shown. A select population of cells within the dorsal horn accumulates GM2 (see arrowheads). B. Higher magnification confocal planes through three neurons, one of which exhibits notable GM2 accumulation. C. GM2 pathology (punctate labeling, white arrowheads) occurred most prominently in superficial laminae of the dorsal horn, both dorsal (D) and ventral (V) to IB4 (green), which labels the terminals of a subset of primary nociceptive afferents in laminae I and II. Scale bars: A = 25 μm, B = 25 μm, C ii. 25 μm, 15 μm (inset)

Ultrastructural analysis of lamina I-II revealed evidence of a lysosomal storage process, with loose-lamellar inclusions characteristic of ganglioside storage bodies occurring within the soma (Figure 4A) and dendritic processes (Figure 4B). In addition, some neurons exhibited an abnormal distribution of organelles similar to that previously reported in the brains of Slc9a6 KO mice [55], with corralling of vesicular inclusions near the Golgi and trans-Golgi network (Figure 4A). The same IHC and ultrastructural analyses performed in DRG tissue did not reveal notable differences between WT and KO mice (data not shown).

Figure 4:

Electron microscopy studies of spinal cord neurons within superficial dorsal horn laminae of Slc9a6-KO mice reveal notable cellular pathologies, including aggregated vesicular storage bodies corralled above the nucleus in one of two neurons shown (boxed region, Ai). Aii and Aiii depict higher magnifications of the region boxed within Ai, highlighting the ultrastructure of these vesicular inclusions, which exhibit both membranous (pink arrowhead) and granular (blue arrowhead) morphology. B depicts a lamina I neuron with notable electron dense storage bodies occurring within a primary dendritic process (higher magnification shown in Bii, circled in orange). C. A representative dorsal horn neuron of similar size from a WT animal is shown for comparison. Scale bars: Ai = 10 μm, Aii = 2.5 μm, Aiii = 1 μm, Bi = 5μm, Bii = 2.5 μm, C = 5 μm. Nuc = nucleus, TGN = trans-Golgi network, rER = rough endoplasmic reticulum. Green arrowheads show post-synaptic densities.

Given the observed lysosomal storage pathology, we looked for evidence of astroglial and microglial involvement, which are common features of lysosomal diseases indicative of neuroinflammatory and neurodegenerative changes [64–66]. Spinal cords 2-month old KO mice had a significantly elevated presence of GFAP⁺ astrocytes throughout the gray matter of the spinal cord (Figure 5), as well as CD68⁺ microglia/brain macrophages, compatible with an inflammatory response. Unlike the GM2 accumulation, which was concentrated within the superficial dorsal horn, these altered astroglial and microglial populations were present in the gray matter of dorsal and ventral horns, as well as near the pericentral canal. These differences were also present at 8 months (data not shown).

Figure 5:

Elevated presence of GFAP⁺ astrocytes and CD68⁺ microglia detected throughout the gray matter of 2-month old Slc9a6-KO mice. Sample images of GFAP (red) and CD68 (green) immunolabeled sections of spinal cord from 2-month old WT (top row) and Slc9a6-KO (bottom row) mice. Images depict maximum intensity projections through an equal number of planes. Scale bar = 100 μm.

3.3. Myelination of peripheral nerve and spinal cord is normal in Slc9a6 KO mice:

One plausible explanation for the longer response times to noxious stimuli by Slc9a6 KO mice would be a myelin deficit that could impact nerve conduction velocity. We thus examined the myelination status within the spinal cord and peripheral nerves of aged (8 month) Slc9a6 KO mice through a gross histologic analysis using Luxol fast blue staining and quantitative comparison of g ratios (ratio of the inner to outer myelin diameter) from ultrastructural data. No differences between WT and KO mice were detected in these assays (Figure 6).

3.4. Expression of NHE6 in dorsal root ganglia and spinal cord:

In order to gain further insight into the cellular basis of aberrant sensory processing in CS patients, we examined the expression of NHE6 within the DRG and the spinal cord of adult (2-month old) mice. Antibody labeling of bacterial nuclear-targeted β-galactosidase (which serves as a transcriptional reporter for expression of the mutant Slc9a6 allele in our KO model, highlighting nuclei of cellular populations that express Slc9a6 [58]), suggested that NHE6 is typically expressed in DRG neurons (Figure 7A). In addition, direct immunohistochemical detection of NHE6 protein in DRG from WT mice showed a similar pattern of expression among neurons, in this case with robust punctate staining of cell bodies (Figure 7A). In the spinal cord, co-labeling studies with primary antibodies to β-galactosidase and GFAP (to label astrocytes) or NeuN (to label neuronal nuclei) in Slc9a6 KO mice suggested that Slc9a6 may be more highly expressed in astrocytes, as the majority of β-galactosidase⁺ nuclei were found to be within GFAP⁺ astrocytes, while NeuN⁺ nuclei were infrequently co-labeled by β-galactosidase (Figure 7B).

Figure 7.

DRG (Ai) and spinal cord (Bi-Bii) sections of 2-month old Slc9a6-KO mice were immunolabeled with a primary antibody to β-galactosidase (Bgal) to determine the cell populations that express NHE6. DRG from WT mice were also immunolabeled with an antibody to NHE6 (Aii), which gave an expression pattern among neurons similar to the β-galactosidase marker. Spinal cord tissue was co-labeled with primary antibodies to NeuN (to label neuronal nuclei, Bi) and GFAP (to label astrocytes, Bii), which revealed infrequent association of Bgal⁺ nuclei with NeuN, and frequent association of Bgal⁺ nuclei with GFAP⁺ astrocytes, suggesting a primarily glial expression of NHE6 within the spinal cord. Scale bars: Ai = 20 μm, Aii = 10 μm, Bi = 40 μm, Bii = 100 μm

4. Discussion

Our behavioral data indicate that Slc9a6 KO mice have baseline deficits in the processing of noxious stimuli, which parallel the apparently high pain thresholds observed in patients with Christianson syndrome. These deficits are apparent in two separate modalities (thermal and mechanical). In a repeat test conducted to assess any potential progression to the detected thermal sensory deficits present at 2 months, we found no significant differences between WT and KO mice at 8 months, although KO mice trended toward higher latencies. Age had a significant effect on the responses of both WT and KO mice, which exhibited significantly increased baseline response latencies at all stimulus intensities. This is consistent with reports in the literature of decreases in thermal sensitivity with age [67–68]. In addition, there was greater variability in the response latencies for all animals at 8 months. Thus, while the number of animals used in our study was comparable to that used in similar studies [66–69], it may not be sufficient to detect significant differences reliably at aged time points in this assay. Importantly, the Von Frey assay administered at 8 months revealed deficient behavioral responses to noxious mechanical stimuli, corroborating a measurable sensory phenotype at this age. Taken together, these findings provide important preliminary insights into an under-investigated comorbidity afflicting children with CS and potentially other genetic intellectual disability and autism-related disorders, and highlight the utility of the Slc9a6 KO mouse as an experimental tool to explore the basis of these sensory issues.

In addition to our behavioral findings, we have identified lysosomal storage pathology selectively occurring within superficial laminae of the dorsal horn, an early site within the pain processing pathway, which we hypothesize may be related to the observed phenotypes. We have also identified evidence of neuroinflammatory changes occurring throughout the gray matter of the spinal cords of KO mice. In addition to brain, we report that NHE6 appears to be expressed in the dorsal root ganglia and the spinal cord. Thus, a lack of functional NHE6 could conceivably have consequences not only for brain development and function, but may also affect the spinal cord and peripheral nervous system to produce the complex neurological manifestations of CS.

4.1. Endosomal-lysosomal dysfunction:

Our findings presented here as well as our previous studies [57–58] demonstrate that the neuropathology of Slc9a6 KO mice is highly similar that occurring in primary lysosomal diseases, suggesting a potential overlap in the pathogenic mechanisms underlying nervous system dysfunction in CS and in this family of disorders.

Lysosomal diseases are a group of approximately 50 individually rare hereditary metabolic disorders that result from defects in lysosomal function, and are most often characterized by the storage of undegraded compounds within lysosomes [70–71]. These diseases have been broadly classified by the primary storage material accumulating (sphingolipidoses, gangliosidoses, mucopolysaccharidoses, glycoproteinoses, etc). Importantly, a host of secondary storage products has been identified in a majority of lysosomal diseases, most prominently glycosphingolipids (e.g. GM2 ganglioside) and cholesterol [72–73]. In the normal, mature brain, GM2 typically constitutes less than 2% of total gangliosides, and is usually undetectable via immunohistochemistry [73]. In disease states, it may accumulate as either a primary storage product (e.g. GM2 gangliosidosis, where the degradative enzyme β-hexosaminidase is directly impacted) or secondarily to a different lysosomal defect (e.g. Niemann-Pick type A or C, mucolipidosis type IV, and many others) and can be detected using a variety of techniques including immunolabeling and chromatography. Secondary substrate accumulation may denote an overlap in the pathogenic cascade of several individually rare lysosomal diseases of differing etiologies. Understanding its causes and consequences may thus further our understanding of lysosomal diseases as a whole and inform common therapeutic strategies.

The select neuronal populations prone to GM2 accumulation in Slc9a6 KO mice are highly relevant for the neurological manifestations of CS, and we hypothesize that GM2 is an indicator of some of the neuronal populations most compromised by NHE6 deficiency. Affected areas include regions of the hippocampus (CA3, CA4, dentate gyrus), basolateral amygdala, cerebral cortex, and the superficial dorsal horn. Hippocampal dysfunction is implicated in numerous forms of intellectual disability [74–75], and severe intellectual disability is one of the core clinical features of CS [1] as well as numerous lysosomal diseases [76]. The basolateral amygdala is involved with fear learning and also with generating appropriate behavioral expression in response to fear-inducing conditions [77–78]. Brain pathology in Urbach-Wiethe disease, an ultra-rare autosomal recessive disorder (lipoid proteinosis, caused by loss of function of ECM1) and a putative lysosomal storage disease [79], also has a proclivity for the amygdala [80–82]. In addition to neuropsychiatric symptoms, an indifference to pain has been reported in some patients [83], which is potentially of relevance to the reduced reactions to pain exhibited by CS patients as well as our behavioral findings in Slc9a6 KO mice. Our findings in the superficial dorsal horn also suggest that the lower level circuitry involved in the transmission and processing of nociceptive stimuli may be compromised.

What could be the relationship between GM2 ganglioside accumulation and neuronal compromise in CS? Insights from the study of lysosomal disease may provide clues. Storage of gangliosides, and in particular of GM2, has been correlated with the phenomenon of ectopic dendritogenesis [84], a pathological change unique to lysosomal disease whereby mature neurons (especially cortical pyramidal neurons) initiate the formation of new neurites, a process which normally occurs only within a narrow developmental window. The mechanisms underlying ectopic dendritogenesis in lysosomal storage diseases remain elusive. As a component of plasma membrane lipid rafts [85–86], it is hypothesized that GM2 may interact with growth-factor related signaling complexes, such as neurotrophin receptors, which are involved in the normal developmental processes underlying dendritogenesis [87]. Indeed, alterations to the lipid milieu of plasma membrane lipid rafts may be a general mechanism underlying neuronal dysfunction across many lysosomal diseases. In addition to these changes in neuronal structure and function, progressive lysosomal storage may lead to neuronal cell death, and lysosomal diseases collectively comprise one of the most common causes of childhood neurodegenerative disease [88–89]. The mechanisms linking lysosomal storage and neurodegeneration are complex, and may involve disruptions to various pathways including autophagy, mTORC signaling, mitochondrial function, and calcium homeostasis (for a review, see Onyenwoke et al 2015 [64]). In a mouse model of GM2 gangliosidosis (Hexb^−/−), GM2 accumulation leads to decreased calcium uptake via the sarco/endoplasmic reticulum Ca²⁺-ATPase (SERCA) [90], a mechanism which could ultimately trigger ER stress and apoptosis [91]. The reduced rate of Ca²⁺ uptake was abolished in mice treated with a partial inhibitor of ganglioside synthesis. In our results presented here, we did not find evidence of a progression to the sensory phenotype using the Hargreaves thermal sensitivity assay, which would have been suggestive of a neurodegenerative process occurring. It is possible that we were unable to detect progressive changes with the settings we used in this assay, since both WT and KO mice were seemingly less responsive to thermal stimuli at the 8-month time point, and exhibited greater variability in their response latencies. Additionally, we observed GM2 accumulation as early as 3 weeks of age, prior to our first behavioral assay, thus it is also possible that any cellular changes related to this regional pathology occur earlier in life, i.e. as a developmental phenotype. We anticipate that further studies assessing the impact of GM2 storage over time on neuronal viability, function, and resulting behavioral phenotypes will help further our understanding of CS pathogenesis.

Another important question which arises is how a lack of NHE6, a protein localized mostly to early and recycling endosomes, leads to altered lysosomal function. The endosomal-lysosomal system is a continuum of organelles, and it is conceivable that a disruption in the compartments feeding into lysosomes (e.g. endosomes, autophagosomes) can lead to lysosomal dysfunction. A major function of NHE6 is to regulate early endosomal pH. One possibility is that a pH disturbance is transmitted throughout the endosomal maturation process to affect lysosomal pH. If this happens, lysosomal enzymes, which function within a certain pH range, may be compromised, resulting in the buildup and storage of undegraded metabolites. Another way a pH disruption within early endosomes could affect lysosomal function pertains to intracellular trafficking of proteins that are important for lysosomal function. The major pathway for lysosomal enzymes to reach lysosomes is directly from the trans Golgi network through mannose-6-phospate receptor dependent trafficking [92–93]. However, a fraction of lysosomal enzymes may be secreted from the cell and subsequently re-endocytosed, trafficking through endosomes to reach the lysosome [94]. If there is a deficit in this pathway (e.g., disrupted clathrin-mediated endocytosis and/or altered trafficking throughout endosomal populations, as discussed in 1.2.1), enzymes may not make it to their destination. We have previously reported evidence suggesting reduced β– hexosaminidase activity in hippocampal neurons of Slc9a6 KO mice [57]. In the histochemical assay employed on hippocampal sections, pH was fixed to that optimal for β-hexosaminidase A catalytic activity, yet results still showed reduced reactivity for the KO mice, arguing a lysosomal pH disturbance was not sufficient explanation. Instead these results would be more compatible with the trafficking hypothesis, though they would not strictly exclude possible other factors inhibiting catalytic activity or potential alterations in enzyme production or stability. Evidence from studies of Nhx1, the yeast ortholog to endosomal NHEs, points to potential roles in trafficking of lysosomal proteins. Nhx1 has a high degree of colocalization with Vps10, the yeast ortholog to sortilin, which has a similar function to the mannose-6-phosphate receptor in lysosomal protein targeting [95]. Further studies are warranted to determine if and how NHE6 function is involved with the trafficking of lysosomal proteins, potentially through the trafficking of receptors such as mannose-6-phosphate receptor or sortilin.

4.2. Spinal cord glial pathology:

The increased GFAP immunoreactivity and numbers of CD68⁺ macrophages/microglia within spinal cords of Slc9a6 KO mice is consistent with a neuroinflammatory process occurring, and was already present at 2 months of age, corresponding with our earliest identification of significant behavioral deficits in sensory responses (via thermal assay, figure 2A) in KO mice. Gliosis is a common response to a variety of CNS insults, such as trauma, ischemia, and neurodegenerative disease, and may have both protective and harmful effects [96–97]. Unlike GM2 accumulation, which was localized mostly to superficial dorsal horn laminae, the observed changes in astroglial and microglial populations occurred throughout the entire gray matter. This distribution is indicative of a disease process affecting both sensory and motor regions of the spinal cord, which may contribute to the complex and progressive sensorimotor phenotype in CS.

What could be the reason for the observed changes in astroglial and microglial populations? A possible explanation for these findings could be a typical secondary reaction to neuronal loss [96–97]. Indeed, in the cerebellum of Slc9a6 KO mice, gliosis directly correlates spatially with the anterior-posterior patterned Purkinje cell death observed [57–58]. In several lysosomal diseases, gliosis is a commonly observed secondary change occurring in response to neurodegeneration [98–99]. An exception to this sequence occurs in Sandhoff disease, in which microglial activation occurs prior to neuronal loss [100], and is thought to perpetuate the neurodegenerative disease process [101–102]. Future studies are needed in order to determine if the observed changes in astroglial and microglial populations in spinal cords of Slc9a6 KO mice are associated with neuronal death. An alternative reason for these observations could be due to cell-intrinsic changes within affected astrocytes. In support of this possibility, our histologic analyses of the spinal cord using the β-galactosidase reporter suggest that Slc9a6 may be more highly expressed in astrocytes than in neurons, though it is important to note that these findings were made in the KO condition/disease state which may itself have an influence on expression. Given the many ways in which astrocytes participate in regulation of neuronal function and metabolism (e.g. buffering of extracellular K+, neurotrophins, and glutamate, neurotransmitter synthesis) [18], it is possible that sensory and motor phenotypes in Slc9a6 KO mice may be due in part to neuronal dysfunction or death that is driven or exacerbated by glial dysfunction. If and how changes within glia alter the function of sensory and motor circuits within the spinal cord are open questions. Appropriate conditional knockout models may be useful in answering these. Such models in lysosomal diseases with similar glial pathology have helped to elucidate the contributions of different cell types to the disease processes and to distinguish between autonomous and non-autonomous neurodegeneration [103–104]. Such an approach, in combination with analysis of relevant neuropathological changes and behavioral studies that include cognitive, motor, and sensory assays, would be instrumental in helping to uncover the predominant cell types involved in various aspects of nervous system dysfunction in CS.

4.3. NHE6 and sensory function: insights from studies of autism and hereditary pain processing disorders:

There is an increasing recognition of sensory abnormalities, including aberrant pain processing, in patients with autism spectrum disorder (ASD), yet the nature of these abnormalities and their etiology are not well understood. There is a substantial degree of overlap between CS and ASD. Clinical descriptions of CS patients have often included features of autism, [1], and SLC9A6 expression in idiopathic autism is down-regulated [32]. Elucidating how deficiency of NHE6 affects sensory function may thus enhance our understanding of sensory disturbances not only in CS but in other forms of ASD as well.

In the Diagnostic and Statistical Manual V (DSM-V), “apparent indifference to pain/temperature,” is included in a list of diagnostic criteria for ASD, yet systematic studies of thresholds and responses to specific sensory modalities in patients with autism is lacking. The majority of reports within the literature linking pain insensitivity with autism are based on self/parental accounts, which are prone to memory and interpretation biases, and a systematic review of these accounts concluded that there is insufficient evidence for difference in pain thresholds or experiences in patients with autism [105]. To date, the most comprehensive direct examination of sensory thresholds in autistic patients detected increased pain sensitivity in a thermal pain assay [106]. Another study measured pain perception in response to mild electric shocks as well as pain anticipation using MRI techniques, and found that autistic subjects, compared with healthy controls, designated lower levels of electric stimulation as being painful and also showed greater activation in the anterior cingulate cortex during anticipation of the painful stimulation, findings which are consistent with increased pain sensitivity and anticipation [107].

There are several potential reasons for the apparent disagreement between patient/caretaker accounts and these clinical research findings. The causes of autism are complex, with no single gene responsible for greater than 3% of cases, and most cases are currently idiopathic [108–109] thus, the limited number of patients (8 in the cohort studied by Cascio et al [106], 17 in the cohort studied by Gu et al [107]) included in these studies are not necessarily representative of the phenotypic range of sensory disturbances in ASD. Along those lines, both studies included only high functioning individuals. Lower IQ individuals are more prone to self-injurious behaviors, possibly due in some cases to increased pain tolerance [110–112]. Pain processing abnormalities in autism may thus encompass a spectrum of sensory findings from low sensitivity to high, depending on the individual and the etiology. It has been suggested that pain phenotypes can be used as a means of subtyping individuals with ASD [105].

Basic research into monogenic causes of autism has begun to shed light on potential mechanisms of altered pain perception, and support a decreased sensitivity to pain in at least a subset of ASD. SHANK genes encode scaffolding proteins located within the post-synaptic densities of glutamatergic synapses, and mutations in these genes have been associated with ASD [113]. There are established KO mouse models for all three Shank proteins, which have helped to elucidate the alterations in synaptic function and plasticity at glutamatergic synapses within the brain that may contribute to ASD and related ID phenotypes [114–115]. Importantly, roles for Shank proteins in synaptic function of nociceptive circuits are starting to emerge as well. Similar to our findings in Slc9a6 KO mice, Shank2 KO mice have reduced responses to noxious thermal and mechanical stimuli, and in addition they have reduced sensitivity to chronic neuropathic and inflammatory pain [116]. Another study found that Shank2 KO mice have deficits in NMDA-induced pain hypersensitivity, and that Shank2 is involved with the transmission of pain within glutamatergic synapses in the spinal cord [117]. In contrast to Shank2 KO mice, Shank3 KO mice do not have deficits in baseline pain, but do show reductions in heat hyperalgesia [118]. In this study, loss of SHANK3 reduced the synaptic responsiveness of DRG neurons to capsaicin, and KO of Shank3 specifically in Nav1.8-expressing primary sensory neurons was sufficient to impair heat hyperalgesia. Thus, Shank proteins appear to be involved in the lower level (i.e., spinal cord and DRG) synaptic transmission of nociceptive information.

We have shown that NHE6 is expressed in DRGs, thus it is possible that, like the Shank proteins, NHE6 may have roles outside of the CNS to contribute to aberrant sensory phenotypes. A recently described rare genetic disorder featuring insensitivity to pain, inability to feel touch, and severe intellectual disability, is due to homozygous mutation in clathrin heavy chain-22 (CHC22) [119]. Knockdown studies in human IPSC-derived nociceptors suggest a role for this protein in the negative regulation of neurite outgrowth [119], thus directly linking endosomal processes to nociceptor development and pain function. Congenital insensitivity to pain with anhidrosis (CIPA) is another rare genetic disorder and is due to mutations in NTRK1, which encodes tyrosine kinase receptor A (TrkA). CIPA patients have severe insensitivities to pain due to impaired development of primary nociceptive neurons, which rely on TrkA signaling [120–121]. Given the roles of NHE6 with TrkB signaling and neurite outgrowth of central neurons [36], it is conceivable that NHE6 deficiency in developing primary nociceptive neurons may affect intra-endosomal TrkA signaling, which is required for the survival, outgrowth, development, and postnatal function of nociceptive neurons [122–126]. Interestingly, knockdown of NHE5 has been shown to reduce the cell surface abundance of TrkA, leading to reduced downstream signaling and impaired neurite outgrowth in PC-12 cells [127]. Further studies are clearly warranted to establish the roles of NHE6 and endosomal pH regulation in the development and functioning of nociceptive circuits.

Conclusion:

Christianson syndrome (CS) is a complex neurogenetic intellectual disability disorder, with neurodevelopmental as well as neurodegenerative features, and clinical/neuropathological overlap with autism as well as lysosomal diseases of the nervous system. Here we have demonstrated that Slc9a6 KO mice, a model of CS, exhibit reduced responsiveness to noxious thermal and mechanical stimuli, consistent with clinical reports of apparently high tolerance to pain. Notable neuropathology included the accumulation of GM2 ganglioside within select neurons of the superficial dorsal horn, an area involved with the transmission and processing of nociceptive information, as well as abnormal vesicular organelle distribution within perikarya in this region. The spinal cords of KO mice also exhibited changes in astroglia and microglia, consistent with a neuroinflammatory response, throughout the gray matter. While we do not know whether the observed sensory phenotypes are due to dysfunction in the cerebrum, spinal cord, or peripheral nervous system, our findings indicate a disease process occurring within the spinal cord. Lastly, we have shown that, in addition to the cerebrum, NHE6 appears to be expressed in cell populations of dorsal root ganglia and the spinal cord, and thus may be involved with the normal functioning and/or development of these structures. In addition to presenting these findings, we have reviewed the state of knowledge of the pathobiology of CS, and discussed potential areas of overlap with neurodevelopmental disorders that similarly affect intellectual and sensory functions.

Highlights.

• Sensory impairments found in mouse model of rare intellectual disability disorder.

• Lysosomal disease phenotypes in dorsal horn neurons relevant to pain transmission.

• Neuroinflammatory changes found uniformly throughout spinal cord gray matter.

• Reviews cellular mechanisms linking intellectual disability and pain insensitivity.