Cause or compensation?—Altered neuronal Ca²⁺ handling in Huntington's disease

Summary

Huntington's disease (HD) is a hereditary neurodegenerative disorder of typically middle‐aged onset for which there is no disease‐modifying treatment. Caudate and putamen medium‐sized spiny projection neurons (SPNs) most severely degenerate in HD. However, it is unclear why mutant huntingtin protein (mHTT) is preferentially toxic to these neurons or why symptoms manifest only relatively late in life. mHTT interacts with numerous neuronal proteins. Likewise, multiple SPN cellular processes have been described as altered in various HD models. Among these, altered neuronal Ca²⁺ influx and intracellular Ca²⁺ handling feature prominently and are addressed here. Specifically, we focus on extrasynaptic NMDA‐type glutamate receptors, endoplasmic reticulum IP3 receptors, and mitochondria. As mHTT is expressed throughout development, compensatory processes will likely be mounted to mitigate any deleterious effects. Although some compensations can lessen mHTT's disruptive effects, others—such as upregulation of the ER‐refilling store‐operated Ca²⁺ channel response—contribute to pathogenesis. A causation‐based approach is therefore necessary to decipher the complex sequence of events linking mHTT to neurodegeneration, and to design rational therapeutic interventions. With this in mind, we highlight evidence, or lack thereof, that the above alterations in Ca²⁺ handling occur early in the disease process, clearly interact with mHTT, and show disease‐modifying potential when reversed in animals.

1. INTRODUCTION

Huntington's disease (HD) is a monogenic, autosomal dominantly inherited neurodegenerative disease caused by a CAG repeat expansion in exon 1 of the Huntingtin gene. The wild‐type (WT) huntingtin protein (wHTT) contains a 6‐35 amino acid N‐terminal polyglutamine repeat (CAG‐coded), which is expanded to 36 or more in mutant huntingtin (mHTT).1 Humans expressing a mutant allele with at least 40 CAG repeats invariably manifest disease within a normal life span, typically with a middle‐aged onset (mean 30‐50 years).2 However, age of onset is inversely correlated with the CAG repeat length, and childhood cases are seen with more extensive expansions.3 HD is progressive, with patients showing cognitive decline and troubling motor symptoms—these involve chorea (involuntary movements), as well as motor incoordination, bradykinesia, dystonia, rigidity, dysphagia, dysarthria, and postural instability with progression. Psychiatric symptoms, including depression and psychosis, also are present commonly. Following diagnosis, patients progress to death in approximately 15‐30 years.

The HTT protein is ubiquitously expressed throughout the body and appears to serve an essential developmental function, as evidenced by the embryonic lethality of germline knockouts in mice.4, 5 Based on this, and HD's dominant pattern of inheritance, mHTT appears to cause disease primarily via a toxic gain of function. However, wHTT exerts neuroprotective actions,6 the partial loss of which likely also contributes to disease. Striatal medium‐sized spiny projection neurons (SPNs) of the caudate and putamen are first affected and most extensively degenerate in HD.7, 8 Other brain areas, including the neocortex and hippocampus, are impacted also, particularly in later stages. Interestingly, although HTT is widely expressed, many brain areas such as the cerebellum are largely spared.8 Even within the striatum, cholinergic and aspiny interneurons are relatively unaffected.9, 10 The reasons underlying this relatively selective striatal SPN vulnerability are incompletely understood.

Although HD's genetic underpinnings are well defined, the precise mechanisms linking protein alteration to symptomatology and underlying neurodegeneration remain unclear. Complicating matters, wHTT's normal function is still incompletely understood. However, its large size (347 kD) and multiple protein‐protein interaction sites suggest a scaffolding role,11 which is supported by its large protein interactome.12

Mutant huntingtin's polyglutamine expansion causes it to form amyloid‐like aggregates, seen as cytoplasmic and intranuclear inclusions, but also facilitates interactions with proteins possessing similar domains.13, 14 A quantitative proteomic analysis suggests that 200 proteins preferentially associate with mutant vs WT HTT. Conversely, mHTT's altered conformation impedes 149 normal WT protein interactions.14 Therefore, despite HD's monogenic origins, mechanisms underlying neurodegeneration are likely multifactorial.

Disordered neuronal Ca²⁺ signaling is a recurrent finding in the HD literature, observed across multiple model systems.15 This takes many forms and, in in vitro models, sensitizes striatal SPNs to Ca²⁺‐mediated apoptosis. Synaptic plasticity is also intimately linked to Ca²⁺‐dependent processes16 and is likewise aberrant in HD models,17 preceding frank neuronal loss. Importantly, animal models support the disease‐modifying potential of treatments targeting some aspects of deranged Ca²⁺ handling. Aside from directly targeting mHTT, we argue Ca²⁺‐dependent processes present some of the most promising therapeutic targets and might additionally prove useful for treating more common neurodegenerative disorders such as Alzheimer's and Parkinson's disease.15

Mutant huntingtin is expressed throughout development, while symptomatic onset is relatively late in life; this provides ample opportunity for compensatory processes to occur. A major challenge thus becomes discerning root pathology from compensation, particularly when considering the myriad of documented mHTT‐mediated effects on neuronal function. The principle focus of this review will be highlighting mHTT‐mediated alterations in neuronal Ca²⁺ handling with an emphasis on discerning causation from compensatory processes. Rodent HD models have been instrumental in addressing these questions, including transgenic mice expressing fragments of mHTT (eg, R6/2, R6/1) or full‐length mHTT (using yeast or bacterial artificial chromosomes—YAC and BAC models, respectively) and mice in which the CAG expansion is knocked in to the murine HTT gene; we direct the reader to the following publications for comprehensive reviews of prominent HD animal models.18, 19 As attention will be given to those alterations seen early in the disease process, much of the evidence reviewed is from the YAC transgenic mouse model, which shows slow progression allowing for investigation of disease mechanisms involved in synaptic and neuronal dysfunction preceding neurodegeneration. Additionally, emphasis will be placed on mechanisms whose inhibition reverses or prevents disease progression in animals and on highlighting molecular interactions between mHTT and affected cellular processes. Finally, we strive to identify current knowledge gaps meriting further investigation.

2. ER Ca²⁺ STORES

Inositol 1,4,5‐trisphosphate receptors (IP3R) are widely expressed endoplasmic reticulum (ER) Ca²⁺ channels gated by IP3 and cytosolic Ca²⁺. Functional receptors are hetero‐ or homotetramers of three known IP3R subtypes (1‐3).20 IP3Rs mediate regenerative Ca²⁺‐induced Ca²⁺ release; however, an IP3 binding requirement and channel inhibition at high cytosolic Ca²⁺ concentrations normally constrain this positive feedback.20 Multiple cell surface receptors, including the Gq‐coupled type‐1/5 metabotropic glutamate receptors, regulate IP3R activity via phospholipase C (PLC), which hydrolyzes membrane bound phosphatidylinositol 4,5‐bisphosphate (PIP₂), releasing cytosolic IP3. mHTT directly binds a cytosolic portion of the type‐1 IP3R's carboxy terminus, increasing its IP3 responsiveness.21 Large signaling complexes position IP3Rs close to relevant cell surface receptors; this likely delivers high local IP3 concentrations for fast, spatially discrete signaling.22 By allowing Ca²⁺ release at lower IP3 concentrations, mHTT might elicit more widespread, sustained IP3R Ca²⁺ release in response to PLC‐coupled receptor activation. Consistent with this, primary rat striatal SPN cultures transfected with mHTT show greatly enhanced Ca²⁺ responses to low doses of the type‐1/5 metabotropic glutamate receptor agonist 3,5‐dihydroxyphenylglycine (DHPG).21 As neurons ubiquitously express both HTT protein and IP3Rs, this interaction by itself cannot explain the selective pattern of neurodegenerative observed in HD. However, evidence suggests that in the striatum SPNs selectively express type‐5 metabotropic glutamate receptors, whereas its expression is notably absent in subsets of relatively HD‐resistant interneurons.23 Glutamate might therefore preferentially engage aberrant IP3R signaling in SPNs via their metabotropic receptor expression. Likewise, DHPG substantially increased NMDA‐mediated intracellular Ca²⁺ release in SPNs, but not large cholinergic interneurons in rat brain slice experiments.24

The direct interaction of mHTT with type‐1 IP3Rs suggests causation. If true, IP3R responses should be altered early in the disease time course, prior to neurodegeneration. It remains unclear when IP3R responses first deviate in HD. However, immunoprecipitation experiments performed on striatal lysates from YAC128 mice suggest an enhanced type‐1 IP3R‐mHTT association in 4‐month‐old mice,25 a largely asymptomatic time point.

To prove causation, mHTT‐mediated neurodegeneration must be prevented, or at least ameliorated, by restoring normal IP3R function. In vitro models provide some evidence to this effect. YAC128 mouse‐derived striatal neuronal cultures show greatly enhanced Ca²⁺ responses to repetitive glutamate application, associated with enhanced apoptotic cell death.26 Direct IP3R or metabotropic glutamate receptor blockade prevents enhanced glutamate‐mediated apoptosis,26 while IP3R knockdown with antisense RNA prevents synapse loss in a YAC128‐derived cortical‐striatal coculture model.27 In vivo, the interaction between IP3Rs and mHTT has been blocked, via transgenic expression of a soluble 122‐peptide portion of the type‐1 IP3R, which binds mHTT (termed IC10).25 IC10 normalized elevated Ca²⁺ responses to glutamate in YAC128 striatal neuronal cultures and, when expressed in the striatum of YAC128 mice, reduced disease‐mediated behavioral declines and striatal SPN loss. IC10 also reduced nuclear mHTT aggregates.25 However, this study must be interpreted with some caution. By virtue of its mHTT‐binding domain, the IC10 peptide might directly reduce mHTT aggregation and/or other protein interactions and exert neuroprotective effects independent of actions on IP3Rs. Future in vivo experiments in which IP3R‐mediated Ca²⁺ release is normalized with pharmacological agents devoid of direct mHTT interactions will be necessary to clarify the IP3R role.

Experiments in YAC128‐derived cortical‐striatal cocultures suggest an enhanced IP3R leak that chronically depletes ER Ca²⁺ stores. This elevates the store‐operated Ca²⁺ channel (SOC) response, a mechanism by which ER Ca²⁺ depletion is counteracted by specific, closely ER‐apposed, plasma membrane Ca²⁺ channels.27, 28 In aged YAC128 SPNs, SOC enhancement is mediated by upregulation of stromal interacting protein type‐2 (STIM2)—a Ca²⁺‐sensitive, ER‐resident protein involved in SOC recruitment. Interestingly, pharmacological SOC inhibition normalizes SPN dendritic spine loss in vitro and in vivo,27 but does not normalize enhanced glutamate‐mediated apoptosis.28 Nonetheless, this suggests that neurodegeneration in HD is partly mediated by SOC upregulation, a compensatory response to ER Ca²⁺ depletion (Figure 1). In Parkinson's disease models, Ca²⁺ influx through voltage‐gated Ca²⁺ channels causes loss of dendritic spines in D₂ receptor‐expressing SPNs.29 Presumably, SOC‐mediated Ca²⁺ influx similarly elicits spine loss in HD, although precise mechanistic details are unclear. However, the Ca²⁺‐dependent phosphatase calcineurin is a potential candidate, as it is expressed in dendritic spines throughout the brain and mediates spine removal.30 Furthermore, calcineurin inhibition is neuroprotective in some HD models.31, 32

Figure 1.

Summary of Ca²⁺ handling abnormalities in Huntington's disease (Left) Representation of a healthy wild‐type (WT) striatal medium‐sized spiny projection neuron (SPN) [not expressing mutant huntingtin protein (mHTT)]. Endoplasmic reticulum (ER) inositol 1,4,5‐trisphosphate receptor (IP3R) activity is low, resulting in minimal ER Ca²⁺ leak; consequently, stromal interacting protein type‐2 (STIM2) activity is suppressed. Low STIM2 activity reduces activation of the ER‐refilling store‐operated Ca²⁺ channels (SOCs). Dendritic spines, contacted by glutamatergic cortical afferents, are present at high density in healthy SPNs. Glutamate release robustly activates synaptic N‐methyl‐D‐aspartate receptors (NMDARs), which conduct Ca²⁺ when the postsynaptic membrane is coincidently depolarized. By contrast, synaptic glutamate release only modestly activates extrasynaptic NMDARs which are expressed at appropriate levels. Ca²⁺ influx through activated synaptic NMDARs is transduced to the nucleus where, by causing cyclic AMP response element‐binding protein (CREB) phosphorylation, it induces procell‐survival gene transcription. (Right) Representation of a mHTT‐expressing, degenerating SPN. By directly binding to type‐1 IP3Rs, mHTT increases their IP3 responsiveness, causing a persistent ER Ca²⁺ leak. ER Ca²⁺ depletion activates STIM2, which subsequently activates SOCs. Activated SOCs increase cytosolic Ca²⁺, which facilitates ER refilling, but also causes dendritic spine loss. Reduced dendritic spine expression likely diminishes prosurvival signaling. In addition, extrasynaptic NMDAR expression is enhanced in HD. Increased Ca²⁺ influx via extrasynaptic NMDARs is taken up by mitochondria, causing membrane depolarization and release of proapoptotic mediators such a cytochrome C (Cyt C). By directly binding to mitochondria, mHTT might also increase their susceptibility to Ca²⁺ overload

3. NMDA RECEPTORS

Aberrant N‐methyl‐D‐aspartate receptor (NMDAR) signaling has long been causally implicated in HD. Indeed, intrastriatal injections of the NMDAR agonist quinolinic acid were used as a chemical model of HD and largely recapitulate the motor deficits and histological hallmarks of HD in rodents and nonhuman primates.33, 34

Functional N‐methyl D‐aspartate receptors (NMDARs) are heterotetramers typically containing two GluN1 and two GluN2 subunits.35 Eight distinct GluN1 isoforms and four GluN2 types (A‐D) exist, which variably combine, creating numerous functionally distinct NMDAR‐types. Additionally, a third GluN3 subunit class incorporates into some NMDARs, modulating Ca²⁺ permeability and Mg²⁺ block.36 Glutamate binds the GluN2 subunit, while the GluN1 and GluN3 subunits bind the obligatory coagonists D‐serine or glycine.

Although activated NMDARs are highly Ca²⁺ permeable, a voltage‐dependent extracellular Mg²⁺ block largely prevents Ca²⁺ conductance at resting membrane potentials. Postsynaptic depolarization relieves this Mg²⁺ block and, when paired with presynaptic glutamate release, causes NMDAR‐mediated Ca²⁺ influx. In this way, NMDARs detect coincident neuronal activity. Synaptic NMDAR Ca²⁺ currents, in turn, form critical neuronal signals evoking long‐term changes in synaptic strength [see16 for a recent excellent review]. Beyond affecting individual synapses, synaptic NMDAR Ca²⁺ signals propagate to the nucleus where they modulate gene expression via transcriptional regulators such as CREB.37 This is necessary for late‐stage synaptic long‐term potentiation (LTP), but also elicits transcription of key prosurvival genes that confer neuronal resistance in the face of various stressors. Indeed, blocking NMDARs is detrimental to the health of numerous neuron types.38, 39

As alluded to earlier, NMDAR activation also promotes neuronal death under certain contexts.40 This process, termed excitotoxicity, can manifest as apoptosis due to Ca²⁺‐dependent caspase and calpain recruitment and can be modeled in vitro by applying glutamate to neuron preparations. A sizable body of evidence suggests that excitotoxicity is mediated by extrasynaptic NMDARs—those receptors located outside the postsynaptic density and not contacted by synaptic glutamate release except under conditions of high‐frequency presynaptic activity.41 By contrast, synaptic NMDARs appear to mediate a trophic, neuroprotective action of glutamate41, 42, 43 This dichotomy is reflected at the molecular level in opposing effects on nuclear CREB, which is phosphorylated and activated following synaptic NMDAR signaling but dephosphorylated by extrasynaptic NMDAR activation.44

Importantly, mHTT‐expressing striatal SPNs are more vulnerable to glutamate‐mediated toxicity.45 This appears to be due to enhanced NMDAR currents, seen in SPNs in a variety of HD models.46 Furthermore, NMDAR current density correlates with the mHTT polyglutamine repeat length in the transgenic YAC mouse model.47 Enhanced NMDAR currents appear to be largely mediated by GluN2B subunit‐containing NMDARs, which are preferentially extrasynaptic in adult neurons. Indeed, brain slice electrophysiology experiments strongly suggest the selective enhancement of extrasynaptic GluN2B‐containing NMDAR currents in striatal SPNs of YAC128 mice; here, evoked NMDAR currents were only greater than those in WT or YAC18 control at high stimulation intensities, conditions expected to cause extrasynaptic glutamate spillover.48 Enhanced extrasynaptic NMDARs were also reported in striatal SPNs from other HD mouse models,49 but whether this change is present in striatal interneurons is unknown. Furthermore, striatal extracts from YAC128 mice show decreased nuclear CREB activity, consistent with enhanced extrasynaptic NMDAR signaling in vivo.48 YAC128 and YAC72 SPNs show enhanced surface expression of GluN2B‐containing NMDARs, which localize to extrasynaptic membranes.47, 48, 50 However, total striatum expression of these proteins is similar in YAC46, YAC72, YAC128, and WT littermates.51, 52 Thus, increased surface GluN2B expression is not driven by enhanced protein synthesis, but rather appears to be due to higher rates of ER to cell surface export in YAC72 mice (where this question was investigated).47 Interestingly, YAC72 SPNs also show enhanced endocytic removal of cell surface GluN1‐GluN2B‐containing NMDARs,47 in the context of this review, perhaps as a compensatory response. Beyond increased surface receptor expression, the molecular mechanism(s) underlying facilitated NMDAR signaling in HD are unresolved. In contrast to IP3Rs, no direct interaction between mHTT protein and any NMDAR subunits has been reported, although proteins associated in a complex with GluN2B‐containing NMDARs have been implicated.12, 53, 54

A view of HD has thus emerged wherein excessive extrasynaptic NMDAR signaling promotes neurodegeneration (Figure 1). Mature SPNs express GluN2B‐containing NMDARs at higher levels compared to other neuronal populations, possibly contributing to their vulnerability in HD.51 Moreover, GluN2B‐subunit overexpression exacerbates striatal SPN loss in Hdh^(CAG)150 knock‐in‐model mice.55 Furthermore, several HD‐resistant striatal interneuron types predominantly express GluN2D‐containing NMDARs with little GluN2B subunit expression,56 perhaps conferring resistance to disease‐related excitotoxicity.57 As well, similar amplitudes of NMDAR‐mediated current elicit smaller mitochondrial ROS (reactive oxygen species) responses in striatal nNOS‐expressing interneurons compared with SPNs.58 Consistent with a proximal role in pathogenesis, deranged NMDAR signaling appears to precede disease onset. For instance, YAC72 striatal SPNs, isolated from early postnatal mice, show increased NMDA‐mediated cell death, although here prolonged culture time (9‐12 days) might have caused receptor expression to diverge from in vivo.45 A crucial study, using more physiologically representative acute brain slices, provided clarity revealing enhanced extrasynaptic NMDAR currents (mediated largely by GluN2B‐containing receptors) in YAC128 SPNs at 1 month's age, long preceding symptomatic stages.48

The above speaks compellingly to causation, however definitive proof requires mitigating HD pathology by blocking excessive extrasynaptic NMDAR activation. The current lack of perfectly selective pharmacological tools, however, makes this practically challenging. The best evidence thus far is provided by studies using memantine, a preferential extrasynaptic NMDAR blocker at low doses.59 Early low‐dose memantine treatment improved behavioral and histological pathology in YAC128 mice48, 60 and normalized nuclear CREB function.48, 61 Conversely, high‐dose memantine, which also blocks synaptic NMDARs, was detrimental.60 Antagonists selective for GluN2B subunit‐containing NMDARs are available and therefore might achieve a relatively selective extrasynaptic receptor blockade. Indeed, the GluN1‐GluN2B‐specific antagonist ifenprodil largely blocked NMDA‐mediated apoptosis in YAC72, YAC128, and WT SPN cultures.26, 45 However, some GluN1‐GluN2B receptors are synaptic, the blockade of which might negatively impact neuronal health. Importantly, the above striatal SPN cultures lacked cortical neurons and therefore had no trophic glutamatergic input to be blocked. When administered In vivo, ifenprodil failed to improve survival or motor outcomes in the R6/2 mouse HD model, as did two other GluN2B subunit‐specific NMDAR antagonists62; this is perhaps owing to adverse effects on synaptic GluN2B‐containing receptors. Consistent with this, remacemide, an NMDAR antagonist whose low affinity and use‐dependence preferentially targets extrasynaptic receptors, improved survival and behavioral measures in R6/2 mice.63 However, these data do not entirely preclude the utility of GluN2B antagonists. Unlike YAC models, R6/2 mice are insensitive to NMDAR‐mediated excitotoxicity64 and are arguably of lower construct validity (discussed below). Future in vivo GluN2B‐selective antagonist studies in YAC model mice would help clarify the utility of these drugs.

Recent evidence suggests that a population of largely extrasynaptic GluN3A subunit‐containing NMDARs mediate excitotoxicity in HD. Striatal SPNs strongly express these GluN3A‐containing NMDARs during development, but increased endocytic removal substantially decreases adulthood expression. This is mediated by PACSIN1 an adapter protein to which huntingtin binds with a polyglutamine repeat length‐dependent affinity.65 In multiple HD mouse models, this interaction is sufficiently strong to sequester PACSIN1, thereby persistently elevating surface GluN3A‐receptor expression.66 GluN3A expression is likewise enhanced in human HD patient striatal tissue. Moreover, enhanced striatal SPN GluN3A expression is causally implicated in HD. Similar to HD mouse models, dendritic spine loss is seen in mouse SPNs overexpressing GluN3A in vivo. Conversely, the behavioral and histological YAC128 phenotype was substantially mitigated in GluN3A subunit lacking mice. The apparent contradiction between these findings and studies implicating extrasynaptic GluN2B‐containing NMDARs in HD could be reconciled if heteromultimeric NMDARs expressing both GluN2B and GluN3A subunits underlie enhanced extrasynaptic currents. Although this has not been directly confirmed, GluN2B and GluN3A surface expression increases in parallel in YAC128 SPNs. Furthermore, enhanced extrasynaptic NMDAR currents were not seen in SPNs from combined YAC128, GluN3A subunit knocked‐out mice.66 These findings further suggest the PACSIN1‐mHTT interaction, by reducing endocytic removal of putative GluN2B‐GluN3A NMDARs, also drives the enhanced GluN2B extrasynaptic surface expression (discussed above). Although appealing, this model is inconsistent with findings of Fan et al 2007, suggesting both expedited ER to cell surface transport and cell surface GluN2B‐receptor removal in YAC72 SPNs.47

Surprisingly, R6/1 and R6/2 HD model mice are resistant to NMDAR‐mediated excitotoxicity.64 Despite this, these mice show rapid progression of behavioral features and widespread cytoplasmic inclusions. Interestingly, resistance to excitotoxicity appears to parallel inclusion appearance,67 which is generally viewed as a compensatory response. HD patients show lower levels of cytoplasmic mHTT inclusions, more akin to that seen in YAC model mice. However, some human postmortem studies suggest relative sparing of those striatal SPNs with greater visible mHTT aggregates.68 Although controversial, this could be explained if larger aggregates sequester more toxic soluble mHTT oligomers. Furthermore, the R6 models suggest that aggregates might somehow mitigate excitotoxicity. In this regard, a recent seminal study suggests that mHTT aggregation does indeed reduce apoptotic SPN death and is enhanced by synaptic NMDAR activation.60 Increased mHTT aggregation formation improved cell survival despite the persistence of enhanced extrasynaptic NMDAR currents; however, extrasynaptic NMDAR activation caused mHTT disaggregation.60 These findings suggest that enhanced synaptic NMDAR activation, by sequestering toxic mHTT oligomers into aggregates, facilitates a beneficial compensation against enhanced extrasynaptic NMDAR activity.

4. MITOCHONDRIA

Mitochondria play a fundamental role in translating increased cytosolic Ca²⁺, mHTT‐induced or otherwise, into cell death signals.69 Normally, mitochondria take up excessive cytosolic Ca²⁺, facilitating its tight intracellular regulation.70 However, when the mitochondria's finite Ca²⁺ handling capacities are exceeded, Ca²⁺ influx collapses the mitochondria's membrane potential, causing the inner membrane permeability transition pore to open, releasing apoptotic mediators including cytochrome C into the cytosol.

Cytosolic cytochrome C release is greatly enhanced in cultured YAC128 SPNs exposed to a 250 μmol/L glutamate challenge and parallels increased apoptotic cell death in these neurons.26 Apoptosis rates in this model were restored to WT levels when mitochondrial Ca²⁺ uptake was prevented with the mitochondria uniporter blocker ruthenium 360. Blocking the permeability transition pore and mitochondrial‐dependent caspase activation similarly normalized apoptosis.26, 45 R6/2 mice show increased cytosolic cytochrome C levels in vivo.71 In addition, minocycline, an antibiotic that prevents mitochondrial release of prodeath molecules including cytochrome C,72 ^, 71 increased R6/2 mouse survival time.73

Beyond simply responding to increased cytosolic Ca²⁺, mitochondria themselves might be dysfunctional in HD. There is a general consensus that mHTT strongly associates with mitochondria and mitochondria‐related proteins.14, 74, 75, 76, 77, 78 In humans, functional PET studies indicate metabolic changes in striatum,79 while HD patients show persistent weight loss, despite appropriate caloric intake, suggesting global metabolic dysfunction.80, 81 Isolated HD patient‐derived lymphoblasts show lower ATP/ADP ratios proportionate to the mHTT polyglutamine repeat length82; likewise, cyanide (an electron transport chain inhibitor) caused greater mitochondrial membrane depolarizations in juvenile HD patient lymphoblasts, again mHTT repeat length proportionate.74 SPNs from symptomatic R6/2 mice show depolarized resting membrane potentials, likely partly mediated by mitochondrial dysfunction.83 Moreover, rodent models suggest mHTT directly modulates permeability transition pore gating, causing opening and cytochrome C release at lower Ca²⁺ concentrations77 ^. 84 Reduced Ca²⁺ uptake capacity has also been reported in mitochondria from YAC72 mice.78

Despite the above evidence, an exhaustive set of studies by Nickolay Brustovetsky's group, using multiple mouse HD models, failed to observe any metabolic or Ca²⁺ handling abnormalities in SPN‐derived mitochondria.75, 76, 85 These studies mainly relied on isolated mitochondria, suggesting other cellular constituents might be required to reveal mHTT‐mediated mitochondrial dysfunction. Consistent with this, several mitochondrial‐related abnormalities, including reduced ATP/ADP ratios, mitochondrial depolarization, and increased Ca²⁺ content, were exacerbated by exogenous NMDA applications in HD knock‐in line‐derived cultured SPNs. Blocking NMDARs or chelating extracellular Ca²⁺ normalized the above parameters nearly to WT levels.82

Mitochondrial Ca²⁺ uptake is increased via large IP3R‐mediated Ca²⁺ release events occurring at domains of ER‐mitochondria membrane contact.86 The mHTT‐IP3R interaction (outlined above) could therefore increase mitochondrial Ca²⁺ uptake, a hypothesis supported by some experimental evidence87; this potentially contributes to mitochondria depolarization and limits further Ca²⁺ uptake.

Although inhibiting mitochondrial Ca²⁺ uptake normalizes glutamate‐mediated apoptosis in YAC128 SPNs, this does not prove causation. Enhanced apoptosis, although dependent on mitochondrial processes, might be entirely caused by increased cytosolic Ca²⁺ entry (via mechanisms outlined above) (Figure 1). Given conflicting results from reduced preparations, in vivo studies may be needed to form a consensus. Ideally, confining the effects of mHTT to mitochondria in an intact biological system could clarify the organelle's causative role; however, this is not readily achieved with genetic models. Alternatively, as mHTT‐mitochondrial protein interactions become better understood, these could perhaps be selectively disrupted. Nonetheless, classic toxin studies provide some insight into the mitochondria's potential role in HD pathology. The mitochondrial toxin 3‐nitropropionic acid (3NP) elicits relatively selective striatal SPN toxicity in humans, rats, and nonhuman primates,88 producing a neurological syndrome strikingly reminiscent of HD. Likewise, multiple mitochondrial mutations produce relatively selective striatal pathology.89 3NP studies further suggest a potentially destructive interaction between dysfunctional mitochondrial and pathological SPN NMDAR activation. Application of 3NP (300 μmol/L) strongly potentiated evoked NMDA currents in rat‐ or mouse‐derived brain slice striatal SPNs, but not large cholinergic interneurons; this was dependent on cytosolic Ca²⁺ elevation.90 Higher 3NP concentrations deplete cytosolic ATP in both striatal SPNs and large cholinergic interneurons. However, this effect differentially couples to membrane currents, depolarizing SPNs (to a greater extent in slices derived from the R6/2 mice than in WT) while hyperpolarizing large cholinergic interneurons.91 These data imply a damaging SPN‐specific positive feedback. Mitochondrial dysfunction, by depolarizing SPNs and directly enhancing their NMDAR currents, potentially enhances NMDAR‐mediated Ca²⁺ influx further compromising mitochondrial function. Thus, although its causative role is not definitive, mitochondrial dysfunction can preferentially harm striatal SPNs, consistent with a pathogenic role in HD.

5. FROM THE BENCH TO THE BEDSIDE—PROMISING Ca²⁺‐CENTRIC THERAPEUTIC TARGETS

5.1. Current treatment of Huntington's disease

Current HD treatments are entirely symptomatic and do not halt disease progress or increase longevity. Tetrabenazine (TBZ) is used to manage chorea and functions by reducing synaptic dopamine release, but also depletes serotonin and norepinephrine.92 Unfortunately, monoamine depletion often worsens anxiety and depression; TBZ is therefore contraindicated in patients with previous suicide attempts or uncontrolled depression. Other motor symptoms, which typically emerge with disease progression, are difficult to manage, with available treatments based on low‐quality evidence. HD patients show high rates of depression and suicidality,93, 94, 95 and while some evidence supports SSRI or venlafaxine treatment, methodological problems limit the conclusions of these studies.96, 97, 98 Overall, evidence for the pharmacological treatment of depression in HD is lacking.99 Treatments for HD cognitive symptoms are similarly lacking, and little evidence supports the use of the cholinesterase inhibitors used to treat cognitive symptoms in Alzheimer disease.100, 101

5.2. Agents Normalizing ER Ca²⁺ and the SOC Response

Pridopidine has shown promise as a neuroprotective agent in models of Parkinson's and Alzheimer's disease.102, 103 Originally thought to stabilize dopaminergic transmission, pridopidine rather acts as an agonist at the sigma‐1 receptor (S1R),104 a predominantly ER‐expressed transmembrane protein involved in Ca²⁺ store regulation. S1R protein expression is enhanced in the striatum of aged YAC128 mice and in HD patients, consistent with its compensatory upregulation.106 Treatment of YAC128 cortical‐striatal cocultures with pridopidine (or another S1R agonist) normalized striatal ER Ca²⁺ levels, prevented SOC response upregulation, and mitigated SPN spine loss.106 Additionally, high‐dose pridopidine treatment substantially improved the motor phenotype in R6/2 model mice.105 Clinical HD pridopidine trials suggest its use in the management of some motor symptoms, but failed to show cognitive improvements.107, 108, 109 Modest improvements in human trials contrast with animal model findings, but might be accounted for by the time point of the intervention. Clinical trials administered pridopidine to symptomatic, diagnosed HD patients, at which point neurodegeneration has typically progressed for more than 10 years before a diagnosis (110see Fig 3). We suggest that trials directing pridopidine treatment toward premanifest HD gene mutation carriers (based on predictive genetic testing) are more likely to address disease‐modifying potential.

Alternatively, agents blocking downstream effects of enhanced IP3R activity such as the enhanced SOC response (see above) might prove to be useful HD treatments.27 Although pharmacological blockers of the SOC response in SPNs are available,27, 28 greater preclinical work is likely needed to validate this treatment approach.

5.3. Memantine and Other Modulators of NMDAR Signaling

HD animal models suggest blocking enhanced extrasynaptic NMDAR signaling is neuroprotective (see above). However, as yet there are no reports of randomized, placebo‐controlled low‐dose memantine studies in HD patients. Some pilot clinical studies suggest benefit, but these are not definitive.111, 112 As with pridopidine, those studies were restricted to symptomatic patients. By contrast, animal studies initiated treatment at early, largely asymptomatic stages. Indeed, enhanced NMDAR signaling may be a first step in a larger pathological cascade of events. If so, blocking extrasynaptic NMDARs might be futile at late disease stages. Supporting this idea, GluN2B‐containing NMDARs may be downregulated in late‐stage HD, as compensation for persistently elevated extrasynaptic receptor signaling. In YAC128 mice calpain‐mediated GluN2B‐containing receptor cleavage significantly reduces total striatum GluN2B protein expression,52 an effect that is progressive with age and disease stage.

6. CONCLUSIONS

In pursuing a mechanistic, cause‐based understanding of mHTT‐induced pathology, our ultimate goal is to translate these insights into novel disease‐modifying therapies, a currently unmet need. Agents that normalize the disrupted Ca²⁺ handling (outlined above) show particular promise. It will be critical to target physiological processes causally related to disease progress, otherwise disrupting beneficial compensations might hasten disease progression. Greater benefit will likely also be derived from early interventions, prior to large‐scale neuronal loss. A lack of early diagnostic markers makes this impractical in most neurodegenerative disorders; however, identifying HD patients before symptom onset is feasible given the relative ease of genetic testing.

Mackay JP, Nassrallah WB, Raymond LA. Cause or compensation?—Altered neuronal Ca²⁺ handling in Huntington's disease. CNS Neurosci Ther. 2018;24:301–310. 10.1111/cns.12817