Dysfunction of X-linked inhibitor of apoptosis protein (XIAP) triggers neuropathological processes via altered p53 activity in Huntington’s disease

Abstract

Mitochondrial dysfunction is associated with neuronal damage in Huntington’s disease (HD), but the precise mechanism of mitochondria-dependent pathogenesis is not understood yet. Herein, we found that colocalization of XIAP and p53 was prominent in the cytosolic compartments of normal subjects but reduced in HD patients and HD transgenic animal models. Overexpression of mutant Huntingtin (mHTT) reduced XIAP levels and elevated mitochondrial localization of p53 in striatal cells in vitro and in vivo. Interestingly, XIAP interacted directly with the C-terminal domain of p53 and decreased its stability via autophagy. Overexpression of XIAP prevented mitochondrially targeted-p53 (Mito-p53)-induced mitochondrial oxidative stress and striatal cell death, whereas, knockdown of XIAP exacerbated Mito-p53-induced neuronal damage in vitro. In vivo transduction of AAV-shRNA XIAP in the dorsal striatum induced rapid onset of disease and reduced the lifespan of HD transgenic (N171-82Q) mice compared to WT littermate mice. XIAP dysfunction led to ultrastructural changes of the mitochondrial cristae and nucleus morphology in striatal cells. Knockdown of XIAP exacerbated neuropathology and motor dysfunctions in N171-82Q mice. In contrast, XIAP overexpression improved neuropathology and motor behaviors in both AAV-mHTT-transduced mice and N171-82Q mice. Our data provides a molecular and pathological mechanism that deregulation of XIAP triggers mitochondria dysfunction and other neuropathological processes via the neurotoxic effect of p53 in HD. Together, the XIAP-p53 pathway is a novel pathological marker and can be a therapeutic target for improving the symptoms in HD.

Introduction

Huntington’s disease (HD) is a progressive autosomal-dominant neurological disorder caused by expanded CAG repeats that code for glutamine in the Huntingtin (HTT) gene of chromosome 4. The HTT gene is ubiquitously expressed in several types of neurons (Huntington’s Disease Collaborative Research Group, 1993; Mangiarini et al., 1996). Previous reports have shown that both mutant HTT (mHTT) protein aggregates and its proteolytic fragments lead to neuronal damage by altering genetic and epigenetic components and molecular signal transduction pathways (Ferrante et al., 2003; Gardian et al., 2005; Lee et al., 2013; Nucifora et al., 2001; Ryu et al., 2006; Sadri-Vakili et al., 2006; Stack et al., 2007). Mitochondrial dysfunction is associated with neuropathological changes in HD patients, HD transgenic animal models, and HD cell lines (Lee et al., 2011). However, the mitochondria-mediated neuropathological mechanism in HD is not fully understood yet.

It is known that p53, a tumor suppressor protein, plays a specific yet an indirect role in the mitochondria-associated cellular dysfunction and behavioral abnormalities of HD mice (Bae et al., 2005). Mutant huntingtin (mHTT) binds to p53 and up-regulates p53 transcriptional activity (Bae et al., 2005). This study suggests that p53, in part, links nuclear and mitochondrial pathological features of HD. Interestingly, it has been well-established that p53 localizes to the mitochondria under conditions that provoke apoptosis, and that mitochondria-localized p53 is sufficient to launch cell death via the release of cytochrome c (Mahyar-Roemer et al., 2004; Marchenko et al., 2000; Sansome et al., 2001). Furthermore, it is found that p53 is continually localized in the mitochondria with or without stress stimuli. Mitochondrially localized p53 directly interacts molecularly via its DNA-binding domain with the anti-apoptotic Bcl-2 and Bcl-xL proteins (Mihara and Moll, 2003). In this context, we may propose that mitochondrial trans-localization of p53, due to mHTT and other cellular stresses, may trigger or accelerate cell death cascades in neurons. In spite of the abundance of p53 in the mitochondria of neurons and its importance in neurodegenerative disorders, the exact mechanism of mitochondrial p53-dependent neuronal activity is elusive.

Inhibitor of apoptosis (IAP) family proteins were first identified as baculovirus proteins that inhibit apoptosis in host cells (Birnbaum et al., 1994; Crook et al., 1993), and their homologs are found in many animal species including mammals (Deveraux and Reed, 1999; Miller, 1999). IAPs block cell death by inhibiting caspase activity (Deveraux and Reed, 1999; Miller, 1999). As one of human IAP homologs, XIAP (X-chromosome linked IAP) can bind to active caspase-3, caspase-7, and caspase-9 and subsequently inhibit their activities (Deveraux et al., 1998; Deveraux et al., 1997; Roy et al., 1997). SMAC (second mitochondria-derived activator of caspase) is released from the mitochondria into the cytosol with cytochrome c during apoptosis, then binds directly to IAPs, and eliminates the interaction of IAPs with caspases (Du et al., 2000; Verhagen et al., 2000). On the other hand, IAPs are structurally characterized as ubiquitin ligases because they contain a C-terminal RING finger domain in addition to having one to three copies of the baculoviral IAP repeat (BIR) domain as their N-termini (Deveraux and Reed, 1999; Miller, 1999). The BIR domains are important regions that are necessary for binding to and inhibiting caspases. Not only is the BIR2 domain of XIAP, together with its N-terminal linker region, sufficient to bind to caspase-3, but also the BIR3 domain selectively targets caspase-9 (Srinivasula et al., 2001; Sun et al., 1999; Sun et al., 2000; Takahashi et al., 1998). RING finger domains in IAPs have shown to promote auto-ubiquitination and self-degradation in response to apoptotic stimuli (Yang et al., 2000). Moreover, previous studies suggest that these IAPs also function as E3s with respect to other intracellular protein substrates. For instance, caspase-3 has been shown to be poly-ubiquitinated by XIAP in intact cells (Huang et al., 2000; Suzuki et al., 2001). However, the pro-survival role of XIAP against neuronal damage and the molecular mechanisms of XIAP-dependent neuronal survival in HD have not been investigated. Otherwise, the pro-survival role of XIAP against mitochondrial p53-depependent neuronal damage and the molecular interaction of XIAP with p53 are not fully understood yet.

In the current study, we aimed to determine whether XIAP levels are affected by mHTT and whether its levels are altered in HD patients and HD transgenic mouse models. Furthermore, we examined whether XIAP interacts with p53, and if so, whether it modulates p53-mediated neuronal death via mitochondria-dependent pathway. In addition, we verified how a loss of XIAP function contributes to neuropathological sequelae and behavioral phenotypes as well as disease onset and survival rate in the HD animal model. Collectively, our study identified a novel molecular mechanism of neuropathogenesis via XIAP and p53-dependent pathways in HD.

MATERIALS AND METHODS

Human brain samples.

Neuropathological processing of normal and HD postmortem brain samples was performed using procedures previously established by the Boston University Alzheimer’s Disease Center (BUADC). The striatal histopathology was verified and graded according to the criteria as described previously (Myers et al., 1988; Vonsattel et al., 1985). This study was approved for exemption by the Institutional Review Board of the BU School of Medicine (Protocol H-28974) as it included post-mortem tissues that were not classified as human subjects. The study was undergone in accordance with institutional regulatory guidelines and principles of human subject protection in the Declaration of Helsinki. The information of brain tissues is described in Supplementary Table 1.

Animals and virus injection

Male transgenic HD mice (N171-82Q strain) were purchased from The Jackson Laboratory (Bar Harbor, ME, USA) and maintained as a colony at the KIST SPF Animal Facility. Brain specimens of R6/2 mice and YAC128 mice were prepared as previously described (Hwang et al., 2014; Kim et al., 2016; Lee et al., 2012; Stack et al., 2005). AAV-shRNA-XIAP virus was injected by stereotaxic micro-injector (Stoelting Co.) at 14 weeks of age. AAVs were delivered into striatum (AP: 0.86 mm, ML: +/− 2 mm, DV: 2.9 mm) of wild-type mice and HD transgenic mice as previously described. Control groups were injected with AAV-scrambled-shRNA injection. Behavioral, neuropathological, and biochemical analysis were performed at 2 weeks after injection. The sequence information of shRNA-XIAP and miRNAs is described in Supplementary Table 3 and 4, respectively. A limited number of deaths occurred overnight and were recorded the following morning. For the calculation of Kaplan-Meier survival plots, time point of death was based on either last survival date or date of euthanasia administered when mice were unable to initiate movement and right themselves after being gently prodded for 30 s. All animal experiments were performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by KIST Animal Care Committees.

Immunohistochemistry and neuropathology

Immunohistochemistry (IHC) was performed to determine the distribution of anti-XIAP (1:200, sc-55550, Santa Cruz) and anti-p53 (1:200, sc-6243, Santa Cruz) in the striatal sections as previously described (Ferrante et al., 2003; Lee et al., 2013; Ryu et al., 2005). Changes of immunoreactivity in HD group was normalized to the background signal (pixel) in the Control group. The neuropathological assessment was performed as previously described (Lee et al., 2013; Ryu et al., 2005). The intensity of immunoreactivity and soma size of striatal neurons were analyzed using Multi-Gauge Software (Fuji photo film Co, Ltd. Japan) as described previously (Ferrante et al., 2003).

In situ proximity ligation assay (PLA)

The tissue sections were blocked with blocking solution (1% H₂O₂), and incubated with anti-XIAP (1:200, sc-55550, Santa Cruz) and anti-p53 (1:200, sc-6243, Santa Cruz) overnight at 4°C. After washing three times, the sections were incubated with the secondary oligonucleotide-linked antibodies (PLA probes) provided in the kit (Duolink, Sigma-Aldrich Korea, Seoul, South Korea). The oligonucleotides bound to the antibodies were hybridized, ligated, amplified, and detected using a brightfield probe. For counter stain, the sections were incubated Nuclear stain buffer for 3 minutes at room temperature. The results were viewed using a brightfield microscopy.

Statistics

Data is represented as the mean ± standard error of the mean (SEM). For single comparisons, the significance of differences between means was assessed by unpaired Student’s t-test; for multiple comparisons, data were analyzed by ANOVA followed by Fisher’s protected least significant difference (LSD) test using Prism 8 (GraphPad Software, San Diego, CA, USA). Data was considered significant with a value of p < 0.05.

Methods for primary cell cultures and cell lines, Tet-Inducible Mito-p53 cell lines, Induced neuron by direct conversion and lentivirus generation, isolation of mitochondria, construction of XIAP (full length and truncated) expression vectors and GST fusion protein purification, Mito-p53 promoter assay, confocal microscopy and image analyses, transmission electron microscopy (TEM) and quantitative real–time PCR (qPCR), cell apoptosis and mitochondria function assay, and behavioral tests are provided in the online Supplementary Materials.

RESULTS

Aberrant levels of XIAP and p53 are found in the striatum of HD postmortem brains and the directly converted neurons from HD patient fibroblast

In order to investigate the association of XIAP and p53 in the pathogenesis of HD, at first, we determined how XIAP and p53 protein levels were altered in the postmortem brains of HD patients and normal subjects. We performed immunohistochemistry for XIAP and p53 in the striatum (caudate and putamen) of HD patients and normal subjects. As expected, XIAP signals were found as punctate perinuclear structures within the cytosolic compartment of medium spiny neurons (MSN) in the caudate and putamen (Fig. 1A). In addition, XIAP colocalized with mitochondrial marker and cytochrome c in the caudate and cortex (Supplementary Fig. 1A and B). Interestingly, densitometry analysis showed that XIAP immunoreactivity was significantly decreased in HD patients compared to normal subjects (Fig. 1B and C). XIAP mRNA levels were also lowered in HD striatal tissues compared to age-matched normal striatal tissues (Supplementary Fig. 1C). Consistent with the decreased XIAP immunoreactivity in HD, Western blot analysis of striatal extracts from HD brains revealed the marked decrease in total XIAP levels compared to normal brains. Whereas, p53 protein levels were significantly increased in the HD brain (Fig. 1D–F; Supplementary Fig. 1D and E). Both p53 and phosphorylated-p53 (p-p53) immunoreactivity were increased in the caudate and putamen of HD brain (Supplementary Fig. 1E–H). Intriguingly, a linear regression analysis showed an inverse correlation between XIAP and p53 protein levels in the striatum of HD and normal subjects. (Fig. 1G). To verify the association between XIAP and p53 molecules, we performed in situ a proximity ligation assay (PLA) assay (Duolink method) that can detect colocalization of two proteins (Fig. 1H and I). Colocalization foci of XIAP and p53 were found as punctate perinuclear structures in the MSN of normal subjects, but the colocalization signals were markedly decreased in both the caudate and putamen of HD due to the unbalanced levels of two molecules (Fig. 1J and K).

Figure 1. Inverse correlation of XIAP and p53 protein levels are found in the striatum of HD patients.

(A) XIAP Immunoreactivity was decreased in the striatum of HD postmortem brain compared to normal subject (Nor). The nuclei were counterstained with hematoxylin. Scale bars: black, 20 μm; white, 5 μm. (B and C) The intensity of XIAP was significantly decreased in caudate and putamen of HD, respectively [each case, N=3; cell counting, n=60 (20 cells/case)]. (D) Western blot analysis confirmed the reduction of XIAP protein and the increase of p53 protein in the putamen of HD. Arrowhead (red) (E) The XIAP protein level (normalized to GAPDH) was significantly decreased in HD. (F) Densitometry analysis showed that p53 levels (normalized to GAPDH) are significantly increased in HD (each Normal and HD case, N=5). (G) The XIAP and p53 protein level was inversely correlated in the striatum of HD patients (N=4) and normal subjects (N=4). (H) A scheme for the determination of XIAP and p53 colocalization by proximity ligation assay (PLA). (I) Colocalization foci of XIAP and p53 were decreased both in the caudate and putamen of HD. (J and K) The intensity of colocalization signal between XIAP and p53 was significantly decreased in caudate and putamen of HD, respectively [each case, N=3; cell counting, n=39 (13cells/case)]. (L) A scheme of experimental procedures for a direct conversion of neurons from HD patient fibroblasts. (M) XIAP immunoreactivity (blue) was decreased in the MAP2-positive induced neurons (red) of HD patient compared to normal. Scale bars (white): 20μm. Arrowheads (white) indicate the induced neurons with XIAP expression. (N) The intensity of XIAP immunoreactivity was significantly decreased in induced neurons from HD patient fibroblasts (each N=7, data collected from 7 slides). (O) Western blot analysis confirmed the reduction of XIAP protein in the direct conversion of neurons from HD patient fibroblasts. (P) The XIAP protein level (normalized to ACTIN) was significantly decreased in induced neurons of HD.

To further examine whether the levels of XIAP and p53 were altered in human HD cell model by direct lineage reprogramming, we prepared Fuw-ASCL1-BRN2-MYT1L (ABM) lentiviral-transduced human HD patient-derived fibroblasts (Fig. 1L and Supplementary Fig. 2A–J). From 15 days after lentiviral infection, we observed that cell morphologies were shown as normal fibroblasts (Supplementary Fig. 2D). On the other hand, neuronal morphological transition from HD patient-derived fibroblasts was observed from 25 days after lentiviral transduction (Supplementary Fig. 2A–E). The number of induced neurons was reduced in HD cases (Supplementary Fig. 2B). Moreover, compared to normal fibroblasts, smaller soma size and shorter dendrites were observed in HD patient fibroblast-derived induced neurons at 15 and 25 days after neuronal induction (Supplementary Fig. 2C–E). Further examination confirmed that not only the fibroblasts transduced with ABM factors acquired the neuronal morphology, but also they expressed neuronal markers. qRT-PCR analysis revealed that both normal and HD patient fibroblast-derived induced neurons expressed an increased level of neuronal genes compared to non-induced fibroblast controls (Supplementary Fig. 2F). Additionally, XIAP immunoreactivity was significantly decreased in the HD patient fibroblast-derived induced neurons compared to the normal fibroblast-derived induced neurons (Fig. 1M and N). The mRNA level of XIAP was significantly low (by 3- to 4-fold) in the HD patient fibroblast-derived induced neurons compared to normal fibroblast-derived induced neurons (Supplementary Fig. 2G). The level of TP53/p53 mRNA, in contrast, increased in HD patient fibroblast-derived induced neurons (Supplementary Fig. 2H). The phenotype of induced neurons by ABM was verified as glutamatergic neurons that mainly express VGLUT1 gene (Supplementary Fig. 2I and J).

XIAP and p53 levels are also inversely modulated in animal and cell line models of HD

Next, to verify whether XIAP and p53 levels were altered in mouse models and cell line of HD, we performed immunostaining of XIAP and p53 in the striatum of three HD transgenic mouse models (N171-82Q, R6/2, and YAC128) and in mouse striatal cell lines, respectively (Fig. 2A–K and Supplementary Fig. 3A–N). Similar to the finding in HD postmortem brains, XIAP immunoreactivity was significantly decreased in all three HD transgenic mice models (Fig. 2A, B, and Supplementary Fig. 3A–N). In contrast, p53 immunoreactivity was significantly increased in MSN in N171-82Q mice (Fig. 2A and C). The altered levels of XIAP and p53 were inversely correlated in N171-82Q mice (Fig. 2D). The Western blot analysis showed a decrease of XIAP and an increase of p53 levels in the striatal lysates of R6/2 mice (Fig. 2E–H). Additionally, we performed immunostaining of XIAP and p53 in HD (STHdh^Q111/Hdh^Q111) and control (STHdh^Q7/Hdh^Q7) mouse striatal cell lines (Fig. 2I). Intensity of XIAP was significantly decreased in HD (STHdh^Q111/Hdh^Q111) compared to the control (STHdh^Q7/Hdh^Q7) mouse striatal cell lines. Western blot analysis verified the reduction of XIAP and the increase of p53 protein level in STHdh^Q111/Hdh^Q111 cells (Fig. 2J and S3K and L). The regression analysis of XIAP and p53 levels from STHdh^Q7/Hdh^Q7 and STHdh^Q111/Hdh^Q111 cell lysates showed an inverse correlation between endogenous XIAP and p53 protein levels (Fig. 2K). p-p53 protein level were also significantly increased in N171-82Q mice (Supplementary Fig. 3I and J). p53 signal was colocalized with cytochrome c, a mitochondrial marker, and increased in N171-82Q mice compared to WT mice (Supplementary Fig. 3M and N). Interestingly, in vitro overexpression of XIAP significantly decreased p53 level in striatal cells (Supplementary Fig. 3O and P). Next, we underwent PLA assay to examine the colocalization of the XIAP and p53 proteins in WT and N171-82Q mice (Fig. 2L). Notably, colocalization signals (brown-colored cytosolic puncta structures) of XIAP and p53 were lowered in the striatum of N171-82Q mice and it was consistent with the result from HD patients (Fig. 2L and M). The colocalization signals of XIAP and p53 were high in MSNs of WT mice but significantly lower in MSNs of N171-82Q mice (Fig. 2M).

Figure 2. XIAP immunoreactivity is decreased in the striatum of HD transgenic animal models and moue HD striatal cell line.

(A) XIAP immunoreactivity (red) was decreased in the striatum of N171-82Q HD transgenic mice compared to wild-type (WT). The nuclei were counterstained with DAPI (blue). White line indicates the colocalization analysis of XIAP and p53 (green). Scale bars (white): 5μm. (B and C) Colocalization analysis showed that XIAP intensity was decreased while p53 intensity was increased in the MSNs of N171-82Q HD transgenic mice. (D) The XIAP and p53 protein level was inversely correlated in N171-82Q and WT mice. (E) Western blot analysis confirmed the reduction of XIAP protein and the increase of p53 protein in the striatum of R6/2 (Q153) HD transgenic mice. (F) The intensity of XIAP was significantly decreased in striatum of R6/2 mice (WT, N=3; R6/2, N=3). (G) The intensity of p53 was significantly increased in striatum of R6/2 mice (WT, N=3; R6/2, N=30). (H) The XIAP and p53 protein level was inversely correlated in the striatum of WT (N=3) and R6/2 mice (N=3). (I) XIAP immunoreactivity was reduced while p53 immunoreactivity was elevated in HD (STHdh^Q111/Hdh^Q111) striatal cell line compared WT (STHdh^Q7/Q7) cell line. Scale bars (white): 5μm. (J) Western blot verified that XIAP protein levels were decreased but p53 protein levels were increased in mouse HD (STHdh^Q111/Q111) striatal cells compared to controls (STHdh^Q7/Q7) striatal cells. (K) The level of XIAP and p53 was inversely correlated in HD (STHdh^Q111/Q111) and WT (STHdh^Q7/Q7) cells (STHdh^Q7/Q7, N=4; STHdh^Q111/Q111, N=4). (L) XIAP and p53 colocalization was decreased in the striatal neurons of N171-82Q HD transgenic mice. Scale bars (black): 3 μm. (M) The intensity of colocalization signal between XIAP and p53 was significantly decreased in the striatal neurons of N171-82Q HD transgenic mice [WT, N=3; N171-82Q, N=3; cell counting, n=90 (30 cells/case)]. Significantly different from WT mice at *p < 0.05, **p < 0.005.

Overexpression of mutant Huntingtin (mHTT) reduces XIAP level but increases p53 level

A transgene expressing 103 consecutive glutamines (Q103) is well known to mimic mutant Huntingtin (mHTT)-induced neurodegeneration. In order to define whether mHTT affects XIAP and p53 levels, we transfected wild type HTT (wtHTT) (Q25) and mHTT (Q103) in striatal cells (STHdh^Q7/Hdh^Q7), and measured the change in target protein levels and mitochondrial oxidative stress (Fig. 3A). Importantly, overexpression of mHTT significantly decreased XIAP immunoreactivity while increasing p53 immunoreactivity and mitochondrial oxidative stress in mHTT positive cells compared to control cells (Fig. 3B–E). Interestingly, XIAP and mitochondrial oxidative stress levels were inversely correlated in mHTT transfected cells, while p53 levels were positively correlated with oxidative stress (Fig. 3D and E). Western blot analysis confirmed that mHTT (Q103) reduced XIAP but increased of p53 protein level in striatal cells (Fig. 3F–H). Next, to examine whether in vivo transduction of mHTT also affects XIAP and p53 levels, we delivered AAV-Q25-EGFP and AAV-Q103-EGFP viruses into the dorsal striatum of C57BL/6J mouse (Fig. 3I–K). As we expected, overexpression of mHTT significantly reduced XIAP immunoreactivity in MSNs (Fig. 3L and M). On the other hand, p53 immunoreactivity was significantly increased by mHTT expression (Fig. 3N and O). Moreover, Western blot analysis confirmed that XIAP protein level was significantly decreased while p53 was significantly increased in the striatum of AAV-Q103 expressing mice (Fig. 3P–R). XIAP and p53 levels were inversely correlated in the striatum of mHTT expressing mice (Fig. 3S).

Figure 3. Mutant Huntingtin (mHTT) reduces XIAP levels and increases p53 levels both in the striatal cell lines and the mouse striatum.

(A) A scheme illustrating transient expression of mHTT in striatal cells. (B) Overexpression of mHTT (Q103) decreased XIAP immunoreactivity but increased MitoSox signal (red) in striatal cells. The nuclei were counterstained with DAPI (blue). Scale bars (white): 5 μm. (C) Overexpression of mHTT (Q103) elevated p53 protein levels and mitochondrial oxidative stress (MitoSOX) in striatal cells. Scale bars (white): 5 μm. (D) The XIAP level and mitochondrial oxidative stress level was inversely correlated. (E) The p53 level and mitochondrial oxidative stress level was positively correlated. (F) Western blot showed that mHTT (Q103) reduced XIAP protein level but elevated p53 protein level in striatal cells. (G) The level of XIAP was significantly decreased by mHTT (Q103) (Q25, N=3; Q103, N=3). (H) The level of p53 was significantly increased by mHTT (Q103). (I and J) A workflow scheme illustrating in vivo delivery of AAV-Cont (Q25) and AAV-mHTT (Q103) viruses into the striatum of WT mice. (K) Detection of GFP signal verified that AAV is delivered to the dorsal striatum of WT mice. Scale bars (white): 20 μm. (L) Transduction of AAV-mHTT (Q103) decreased XIAP levels in the dorsal striatum of mice compared to AAV-HTT (Q25). Scale bars (white): 5 μm. (M) The intensity of XIAP in GFP-positive cell was significantly decreased by AAV-mHTT (Q103) [Q25, N=5; Q103, N=5; cell counting, n=40 (8 cells/case)]. (N) Transduction of AAV-mHTT (Q103) increased p53 levels in the dorsal striatum of mice compared to AAV-HTT (Q25). Scale bars (white): 5 μm. (O) The intensity of p53 in GFP-positive cell was significantly increased by AAV-mHTT (Q103) [Q25, N=5; Q103, N=5; cell counting, n=40 (8 cells/case)]. Significantly different from AAV-HTT (Q25)-transduced mice at *p < 0.05, **p < 0.005. (P) Western blot showed that mHTT (Q103) reduced XIAP protein level but elevated p53 protein level in striatal cells in vivo. (Q) The intensity of XIAP was significantly decreased by mHTT (Q103) (Q25, N=3; Q103, N=3). (R) The intensity of p53 was significantly increased by mHTT (Q103). (S) The XIAP and p53 protein levels were inversely correlated in AAV-HTT (Q25) (N=3) and AAV-mHTT (Q103)-transduced striatal cells (N=3).

XIAP directly interacts with and modulates p53 stability

Since we observed an inverse correlation of XIAP and p53 levels and their colocalization in the striatum, we further characterized whether XIAP physically interacts with p53. We subcloned and expressed GST fusion constructs of the full length XIAP and its fragments (BIR-1+2+3, BIR-2+3, BIR-2, BIR-3 and RING domain) using pGEX-5X1 vector (Supplementary Fig. 4A). Equal volume of human brain lysates was incubated with Glutathione Sepharose bound GST-XIAP proteins including full-length XIAP and its fragments, respectively. As a result, we confirmed that the full length XIAP and majority of the XIAP fragments (except BIR-2) retained binding ability to p53 molecule (Supplementary Fig. 4B and C). In order to verify the interaction of XIAP with p53 and p53 fragments, we generated Flag-p53 full length and several deletion constructs including Flag-p53-N-terminal, Flag-p53-Mid-terminal, Flag-p53-DNA-binding domain (DBD), and Flag-C-terminal construct (Fig. 4A). We found that p53 C-terminal has strong interactions with XIAP while p53 N-terminal has weak interactions with XIAP (Fig. 4B). We also transfected pcDNA-XIAP-Myc, performed Myc-antibody immunoprecipitation, and determined XIAP interaction with p53 in neuronal cells (Fig. 4C). Additionally, a reverse immunoprecipitation experiment using anti-p53 antibody showed that p53 interacts with XIAP in neuronal cells (Fig. 4D). Together, it is obvious that XIAP interacts with the p53 Mid- and C-terminal while p53 interacts with XIAP BIR-1 and -3 domains, and vice versa.

Figure 4. XIAP interacts directly with and modulates p53 stability in striatal neurons.

(A) A scheme of full length p53 and its deletion mutant constructs. (B) XIAP interacted with the C-terminal domain of p53. Flag-p53 full length, Flag-p53 N-terminal, Flag-p53 Mid-terminal, Flag-p53 C-terminal, and Flag-p53 DBD constructs were transfected in 293T cells. Equal protein amounts of cell lysates were immunoprecipitated with anti-FLAG M2 antibody followed by immunoblotting for anti-FLAG M2 antibody, in order to detect p53 and anti-XIAP antibody for XIAP detection. (C) Immunoprecipitation of XIAP showed an interaction with p53. SH-SY5Y cells were transfected with Myc-tagged full length XIAP and mock vector for 48hrs. Equal protein amounts of cell lysates were immunoprecipitated with anti-Myc antibody followed by immunoblotting for anti-p53 antibody to detect p53 and lgG. (D) Reverse immunoprecipitation of p53 showed an interaction with XIAP. Mito-p53 Tet-on-inducible cells were treated by doxycycline (Doxy) to induce Mito-p53 expression. (E) Knockdown of XIAP by shRNA elevated TP53/p53 mRNA (each case, N=4). (F) Knockdown of XIAP increased mitochondrial p53 binding element (BE) reporter activity (each case, N=6). Significantly different from pcDNA control at *p < 0.05, **p < 0.01. (G) Cycloheximide (CHX) (50 μg/ml) pulse chase study and Western blot showed that overexpression of XIAP destabilized and reduced p53 protein levels. (H and I) XIAP destabilize and reduced p53 protein levels in SH-SY5Y (each case, N=3) and STHdh^Q7/Q7 cells (each case, N=3), respectively. Significantly different from pcDNA control at *p < 0.05, **p < 0.005. (J) Overexpression of XIAP reduced p53 protein levels in SH-SY5Y cells. Proteasome and autophagy functions were inhibited by MG132 (10 μM) and E64D/Pepstatin (10 μg/ml) for 12 hr, respectively. An arrow (red) indicates the bands of exogenous XIAP expression. (K) Reduction of p53 protein levels by XIAP was significantly inhibited by E64D/Pepstatin and MG132 (each case, N=3).

After verifying the interaction between XIAP and p53, we further questioned whether XIAP is likely to affect mitochondrial activity of p53. In order to address this hypothesis, we applied shRNA-XIAP constructs to knockdown XIAP protein level and found that the p53 level was increased (Supplementary Fig. 4D–H). Accordingly, the levels of XIAP protein and p53 protein were inversely correlated due to the knockdown of XIAP (Supplementary Fig. 4I). In addition, the knockdown of XIAP elevated p53 mRNA level (5 fold) in comparison to the controls (Fig. 4E). To examine whether XIAP can modulate p53-dependent mitochondrial transcriptional activity, we also constructed a reporter vector with p53 consensus DNA-binding element in the mitochondria genome and determined whether XIAP modulates mitochondrial transcriptional activity of p53 (Fig. 4F) (Lee et al., 2018). Knockdown of XIAP increased the transcriptional activity of Mito-p52 BE1 (Fig. 4F). Otherwise, Mito-p53 increased the transcriptional activity of Mito-p52 BE1 by 2-fold, but overexpression of XIAP decreased the reporter activity to the basal level (Supplementary Fig. 4J). In contrast, the knockdown of XIAP increased the transcriptional activity of the Mito-p52 BE1 (Supplementary Fig. 4K). To address whether XIAP regulates p53 stability in a time-dependent manner or not, we performed cycloheximide (CHX)-pulse chase experiment in vitro. Our results showed that XIAP overexpression markedly decreased the half-life of endogenous p53, indicating that XIAP destabilizes p53 protein (Fig. 4G). Three separate experiments confirmed that XIAP significantly decreased p53 protein levels in both SH-SY5Y cells and striatal cells (STHdh^Q7/Hdh^Q7) in a time-dependent manner (Fig. 4H and I). It seems likely that p53 turnover is a slow process in neuronal cells such as SH-SY5Y and striatal cells compared to other cell types HEK (~60 min) and U2OS cells (~40 min) (Dat et al., 2013; Sun et al., 2009). In this context, the p53 turnover may be regulated in a cell-type specific manner and a stress/stimulus dependent manner. Next, to determine how XIAP facilitates the turnover of p53 molecule, we examined both proteasome function-mediated and autophagy pathway-mediated turnover of p53 (Fig. 4J and K). Interestingly, our results showed that inhibition of the autophagy pathway by E64D plus pepstatin (PEP) significantly prevented the p53 turnover by XIAP, whereas inhibition of the proteasome function by MG132 did not prevent p53 turnover by XIAP in SH-SY5Y cells (Fig. 4J). Previous studies show that inhibition of proteasome function by MG132 slightly increases LC3II level but reduces p62 protein level (Choudhury et al., 2013; Wang et al., 2019). Considering that two molecules are involved in different steps of the autophagy pathway, their turnover may be modulated through a different rate of degradation. Further study is necessary for verifying the different modulatory processes of LC3 versus p62. To further verify XIAP-dependent p53 turnover through the autophagy pathway, we treated chloroquine, an autophagy and lysosome inhibitor, to XIAP and p53-overexpressed SH-SY5Y cells, and detected the level change of p53 protein. As a result, we found that chloroquine blocks p53 turnover by XIAP (Supplementary Fig. 5A–C). The modulation of p53 turnover in the mitochondria by XIAP was determined (Supplementary Results and Supplementary Fig. 6A–H).

XIAP modulates p53 activity through the C-terminal lysine residues of p53

It is well known that lysine (K) residues in the C-terminal domain of p53 are potential sites for post translational modification. Those C-terminal lysine amino acids are implicated in MDM2-mediated ubiquitination of p53. In this molecular perspective, we tested whether XIAP regulates p53 activity through the lysine residues of C-terminal domain of p53. We utilized XIAP, WT-p53, and lysine residue site-directed p53 mutant constructs (3NKR, 3CKR, 6KR, K370R, K372/373R, K381/382R, and K386R) (Fig. S6A). We co-transfected XIAP and various p53 mutant constructs and verified the localization of p53 in the mitochondrial fraction by sucrose density centrifugation (Fig. S6B). Western blot analysis showed that overexpression of XIAP reduced WT-p53 level in the mitochondria but made no difference with p53-6KR mutant constructs. On the other hand, knockdown of XIAP increased both WT-p53 level and p53-6KR mutant level (Fig. S6C and D). In addition, an overexpression of XIAP rescued WT-p53-induced mitochondrial dysfunction and cell death but it did not affect p53-6KR mutant-induced mitochondrial dysfunction and cell death (Fig. S6E and G). Moreover, knockdown of XIAP accelerated WT-p53-induced mitochondrial dysfunction and cell death but it did not affect mutant p53-6KR mutant-induced mitochondrial dysfunction and cell death, which were analyzed by Annexin V/PI and JC-1 staining followed by flow cytometry (Fig. S6F and H).

Mitochondrially targeted-p53 (Mito-p53) induces mitochondrial dysfunction and cell death

To prove whether HD-related up-regulation of p53 can be mimicked through the generation of an in vitro cell line that selectively over-expresses mitochondrially targeted (Mito-p53), and to gain an insight into the role of mitochondrial p53, we developed a Tet-inducible Mito-p53 cell line (Supplementary Fig. 7A) (Lee et al., 2018). Western blot analysis showed that induction of Mito-p53 is well driven by doxycycline (Doxy) (Supplementary Fig. 7B). Our results showed that an induction of Mito-p53 induced cell death of around 70% compared to control cells, 12 h after the treatment of Doxy (Supplementary Fig. 7C). To determine whether overexpression of Mito-p53 can affect mitochondrial functions and morphology, we checked cytochrome c protein levels using specific antibodies and measured the mitochondrial length change. Interestingly, mitochondria were aggregated and mitochondrial length was significantly reduced by Mito-p53 (Supplementary Fig. 7D and E). BAX protein level was also significantly increased 24 h after the induction of Mito-p53 (Supplementary Fig. 7F and G). MitoTracker (red) staining intensity representing the mitochondrial membrane potential (MMP) was significantly decreased 24 h after the induction of Mito-p53 (Supplementary Fig. 7H and I). Similarly, JC-1 monomer (green) was significantly increased but JC-1 aggregate (red) was decreased 24 h after the induction of Mito-p53 (Supplementary Fig. 7J–L). Quantitative analysis showed that the relative ratio of JC-1 aggregate/ monomer is 2-fold lower after the induction of Mito-p53 (Supplementary Fig. 7M). Together, mitochondria-specific overexpression of p53 led to mitochondrial dysfunctions (MMP reduction, cytochrome c release, and mitochondrial fission).

XIAP modulates Mito-p53-induced mitochondrial dysfunctions and cell death

Since we identified a molecular interaction of XIAP with p53, we proposed that XIAP may prevent Mito-p53-induced mitochondrial dysfunctions and cell death. In this context, we transfected XIAP in Tet-inducible Mito-p53 cells and measured cell death by Annexin V/PI staining and followed by flowcytometry. Overexpression of XIAP reduced the ratio of early apoptosis and late apoptosis to 29.0% and 25.8% in Mito-p53 cells, respectively (Fig. 5A). Three independent experiments confirmed that XIAP significantly decreased Mito-p53-induced cell death (Fig. 5B). XIAP also rescued Mito-p53-induced reduction of MMP (Fig. 5C and D). Mito-p53 increased JC-1 monomer (R1 area) to more than 30%, and overexpression of XIAP reduced its level to 20% (Fig. 5D). Induction of Mito-p53 elevated both Bax monomer and oligomer levels, but overexpression of XIAP reduced Bax oligomerization (Fig. 5E and F). Because active caspase-3 is well known to play an executioner role for apoptosome formation by triggering down stream of cell death pathways, we further detected active caspase-3 levels. Overexpression of XIAP significantly reduced active caspase-3 levels. (Fig. 5G and H).

Figure 5. XIAP modulates mitochondrially targeted p-53 (Mito-p53)-induced mitochondrial dysfunctions, Bax oligomerization, caspase-3 activation, and cell death.

(A) Annexin V/PI staining and flow cytometry analysis showed that XIAP rescued mito-p53-induced cell death. Mito-p53 Tet-oninducible cells were transfected with XIAP and control vector. After 24 h, cells were treated with 1 μg/ml doxycycline (Doxy) for 24 h. (B) XIAP reduced the proportion of neuronal cell death by Mito-p53 (N=3). (C) JC-1 staining and flow cytometry showed that XIAP rescued mitochondrial damage by mito-p53. Note that the R2 area shows the accumulation of JC-1 monomer. (D) XIAP reduced the proportion of JC-1 monomer (N=3). (E, F) XIAP significantly reduced Bax oligomer formation. β-actin was used as a loading control. The blot is a representation of three repeated experiments. (G, H) XIAP significantly reduced cleaved caspase-3 (C Cas-3) levels. β-actin was used as a loading control. The blot is a representation of three repeated experiments. (I) Annexin V/PI staining and flow cytometry analysis showed that shXIAP exacerbated mito-p53-induced cell death. (J) shXIAP increased the proportion of neuronal cell death by mito-p53 (N=3). (K) JC-1 staining and flow cytometry showed that shXIAP exacerbated mitochondrial damage by mito-p53. (L) shXIAP elevated the proportion of JC-1 monomer and multimers (N=3). (M, N) shXIAP significantly increased Bax monomer and oligomer formation. β-actin was used as a loading control. (O, P) shXIAP significantly increased cleaved caspase-3 (C Cas-3) levels. β-actin was used as a loading control.

Based on the protective role of XIAP against p53-induced mitochondrial dysfunctions and cell death, we hypothesized that a loss of XIAP function may promote and exacerbate p53-induced mitochondrial dysfunction and neuronal damage. In this context, we applied Annexin V/PI staining and flow cytometry and verified that knockdown of XIAP significantly elevated Mito-p53-induced cell death to as close as 80% (Fig. 5I and J). In addition, knockdown of XIAP increased the percentage of JC-1 monomer formation in the mitochondria to more than 50% compared to the control (Fig. 5K and L). The knockdown of XIAP also increased Mito-p53-induced Bax oligomerization and caspase-3 activation (Fig. 5M–P). Collectively, loss of XIAP function enhanced Mito-p53-induced cell death and mitochondria dysfunction.

We further evaluated the neuroprotective role of XIAP in primary cultured neurons (Fig. 6A and Supplementary Fig. 8A). As we proposed, overexpression of Mito-p53 induced nuclear DNA fragmentation, but co-transfection of XIAP prevented Mito-p53-induced nuclear DNA fragmentation (Fig. 6B). Quantified analysis showed that more than 90% of Mito-p53-transfected cells exhibited DNA fragmentation, but co-transfection of XIAP decreased Mito-p53-inuced DNA damage to as close as 40% in primary cultured neurons (Fig. 6C). The cytochrome c immunofluorescence staining showed that Mito-p53 elevated cytochrome c release to the cytosol, but co-transfection with XIAP significantly prevented cytochrome c release and restored cytochrome c levels in the mitochondria of primary cultured neurons (Fig. 6D and E).

Figure 6. XIAP prevents Mito-p53-induced neuronal damage and mitochondrial dysfunction in primary striatal neuron cultures and striatal cell lines.

(A) A scheme illustrating transient transfections in primary neuron cultures. (B) XIAP reduced the proportion neuronal DNA damage by Mito-p53 (N=3). Rat primary medium spiny neurons (MSNs) at 7 days in vitro (DIV) were transfected with pDsRed vector, pDsRed-mito-p53 and XIAP for 36 h. Scale bars (white): 10μm. Arrowheads (white) indicate the nuclear morphology of transfected neurons. (C) Overexpression of XIAP significantly prevented Mito-p53-induced neuronal damage in primary MSNs. (D) XIAP prevented Mito-p53-induced neuronal DNA damage. Scale bars (white): 5 μm. Arrowheads (white) indicate the cytosolic compartment. (E) Western blot analysis showed that XIAP prevented mitochondrial cytochrome c release by Mito-p53. (F) Knockdown of XIAP by shRNA-XIAP elevated p53 levels and mitochondrial oxidative stress in striatal cells (STHdh^Q7/Q7). Scale bars (white): 5 μm. (G and H) Colocalization analysis showed that knockdown of XIAP increased mitochondrial p53 levels and oxidative stress marker levels.

To verify the loss of XIAP function in primary cultured neurons, we transfected mitochondrially targeted DsRed -Mito-p53 construct with or without shRNA-XIAP. Knockdown of XIAP led to more than a 1.5-fold increase of p53 protein level and significant DNA fragmentation in primary neurons (Supplementary Fig. 8B–D). In addition, knockdown of XIAP elevated p53 level and mitochondrial oxidative stress in mouse striatal cells (STHdh^Q7/Hdh^Q7) (Fig. 6F). Interestingly, line measurement analysis exhibited increased p53 immunoreactivity co-localized with MitoSOX, a mitochondrial oxidative stress marker, in XIAP knockdown cells compared to control cells (Fig. 6G and H).

Deregulation of XIAP exacerbates neuropathogenesis in a HD animal model

In order to examine whether in vivo deregulation of XIAP affects the neuropathology and behaviors in mice, we generated AAV-shRNA-control and AAV-shRNA-XIAP constructs. We delivered AAV-shRNA-Control and AAV-shRNA-XIAP viruses into the mouse striatum by bilateral stereotaxic injection method (Fig. 7A and B). It has previously been observed that the DARPP32-positive MSNs exhibit “neuronal atrophy” accompanying the reduction of cell size and neurites in the putamen of HD (Guo et al., 2013). In this perspective of neuronal atrophy, we analyzed the reduction of size and the alteration of number of MSNs in the striatum. As we expected, knockdown of XIAP induced neuronal atrophy in the striatum of WT mice and further reduced neuronal size in N171-82Q mice (Fig. 7C). We performed transmission electron microscopy (TEM) and confirmed the nuclear size change in MSNs (Fig. 7C). Semiquantitative analysis showed that the knockdown of XIAP significantly reduced the nucleus size of striatal neurons (Fig.7D). To determine whether in vivo knockdown of XIAP induces abnormal elevation of p53 levels in MSNs, we evaluated p53 levels in the striatum of mice (Fig. 7E). Western blot analysis verified that knockdown of XIAP elevated p53 and cleaved caspase-3 protein levels under mHTT (Q103) expression in primary striatal neurons (Fig. 7F). Densitometry analysis showed that knockdown of XIAP significantly increased p53 immunoreactivity in WT and N171-82Q mice (Fig. 7G). We verified that knockdown of XIAP increased localization of p53 in TOM20-positive mitochondrial foci in the MSNs of WT mice (Supplementary Fig. 8E–G). In contrast, knockdown of XIAP reduced DARPP32 immunoreactivity in the MSNs of WT mice (Supplementary Fig. 8H and I). Western blot analysis showed that knockdown of XIAP elevated both p53 protein level in WT and N171-82Q mice (Supplementary Fig. 8J and K). We also observed that knockdown of XIAP induced ultrastructural changes of mitochondrial cristae in the MSNs of WT and N171-82Q mice (Fig. 7H and I). Knockdown of XIAP significantly increased active caspase-3 immunoreactivity in the MSNs of WT mice and further enhanced its level in N171-82Q mice (Fig. 7J and K). Cresyl violet (Nissl) staining showed the reduction of neuronal size consistent with EM image data (Supplementary Fig. 9A and B). BAX staining and densitometry analysis showed that knockdown of XIAP significantly increased BAX immunoreactivity in the MSNs of WT and N171-82Q mice (Supplementary Fig. 9C and D). Moreover, TEM image analysis exhibited that XIAP dysfunction reduced synaptic density and morphology (Supplementary Fig. 9E and F). Previous studies have shown that decreased autophagy activity is associated with increased mHTT immunoreactivity in HD animal models and, otherwise, XIAP regulates autophagy activity (Fu et al., 2018; Lin et al., 2015; Poetsch et al., 2018). Since we found that XIAP downregulation contributes to a reducing autophagy activity, we proposed that knock-down of XIAP results in an increase of mHTT immunoreactivity in the striatum and cortex of N171-82Q mice. As expected, knockdown of XIAP significantly elevated mHTT immunoreactivity in the striatum and cortex of N171-82Q mice compared to the level in shRNA-Control-injected mice (Supplementary Fig. 10A–D). knockdown of XIAP also increased mHTT immunoreactivity in HD (STHdh^Q111/Hdh^Q111) cell line (Supplementary Fig. 10E–G).

Figure 7. In vivo knock down of XIAP induces neuronal damage in the striatum of WT and HD mice.

(A) A scheme illustrating delivery of AAV-shRNA-Control)-mCherry (shCont) or AAV-shRNAXIAP-mCherry (shXIAP) viruses into the striatum of WT and HD (N171-82Q) mice. (B) Detection of mCherry and DAPI signal verified that AAV was delivered to the dorsal striatum of WT and HD mice. (C) Knock down of XIAP altered the ultrastructure of neuron in the striatum. Scale bars (white): 1000nm. (D) Not only did knock down of XIAP significantly reduce the size of striatal neurons in WT mice, but also exacerbated the reduction of neuronal size in HD (N171-82Q) mice [WT+shCont, N=3; WT+shXIAP, N=3; N171-Q82+shCont, N=3; N171-Q82+shXIAP N=3; cell counting, n=18 (6 cells/case)]. *, Significantly different from control at p <0.01. (E) Knockdown of XIAP increased p53 immunoreactivity in the striatum of WT and HD mice. Tissues were counterstained with DAPI. Scale bars (white): 10μm. (F) Western blot verified that knockdown of XIAP elevates p53 level in primary MSNs under mHTT (Q103) overexpression in vitro. (G) Densitometry analysis showed that shXIAP significantly increased p53 level in primary striatal neurons under mHTT (Q103) overexpression (N=3). *, Significantly different from control at p <0.05. (H) Mitochondria ultrastructure in the striatal neurons of WT and N171-82Q mice. The image represents one of four brain specimens. Scale bars (white): 200nm. Arrowheads (yellow) indicate the cristae morphology. (I) Knockdown of XIAP significantly increased mitochondrial damage in the striatal neurons of WT and HD mice [WT+shCont, N=3; WT+shXIAP, N=3; N171-Q82+shCont, N=3; N171-Q82+shXIAP N=3; cell counting, n=48 (16 cells/case)]. (J) Knockdown of XIAP increased cleaved caspase-3 (Cas-3) immunoreactivity in the striatum compared to control. Tissues were counterstained with hematoxylin. Scale bars (black): 20μm. Arrowheads (black) indicate the cleaved Cas-3 immunoreactivity in MSNs. (K) Densitometry analysis showed that shXIAP significantly increased the active caspase-3 immunoreactivity in the striatal neurons [WT+shCont, N=3; WT+shXIAP, N=3; N171-Q82+shCont, N=3; N171-Q82+shXIAP N=3; cell counting, n=60 (20 cells/case)]. **, Significantly different from control at p <0.005.

In order to investigate the loss of XIAP function, we further transduced AAV-shRNA-XIAP and AAV-shRNA-Control viruses to directly converted neurons from the fibroblasts of normal and HD patients (Supplementary Fig. 11A). Notably, knockdown of XIAP decreased the number of HD patient fibroblast-derived induced neurons (Supplementary Fig. 11B). Additionally, knockdown of XIAP significantly reduced the soma size of induced neurons, indicating that XIAP played a role in maintaining neuronal morphology maturation process (Supplementary Fig. 11C and D). On the other hand, knockdown of XIAP increased p53 and BAX levels in fibroblast-derived induced neurons from normal and HD patients (Supplementary Fig. 11E–G). Together, the data strongly supports that the deregulation of XIAP affects cellular processes involved in the pathogenesis in HD.

Deregulation of XIAP exacerbates motor dysfunction and reduces the life span of HD mice

To verify the loss of XIAP function on the motor coordination and locomotor activity of WT and N171-82Q mice, we performed rotarod, tail suspension and cylinder tests, respectively (Fig. 8A). As we supposed, not only did the knockdown of XIAP impair rotarod performance, but it also significantly affected motor coordination in both WT mice and N171-82Q mice (Fig. 8B). Tail suspension test showed that knockdown of XIAP significantly increased forelimb, hind limb clasping, and torso movement in WT and N171-82Q mice, indicating that XIAP dysfunction induced an imbalanced state of motor coordination (Fig. 8C–F). Furthermore, cylinder test showed that knockdown of XIAP significantly affected spontaneous movements (first wall touch, the number of wall touch (rears), and rearing) both in WT and N171-82Q mice (Fig. 8G–J). Lastly, we determined the lifespan of N171-82Q mice through Kaplan-Myer survival rate analysis in response to XIAP deregulation. The knockdown of XIAP significantly reduced the survival rate of N171-82Q mice (average age: 128 ± 3 day) compared to control N171-82Q mice injected with the control virus (average age: 142 ± 4 day) (Fig. 8k).

Figure 8. In vivo knock down of XIAP impairs motor behavior and reduces survival rate in HD transgenic mice.

(A) A scheme illustrating experimental designs for behavioral tests, neuropathology, and survival rate analysis. (B) shXIAP significantly impaired rotarod performance in WT (N=6) and HD transgenic (N171-82Q) mice (N=6). Significantly different from WT mice at *, p<0.05; **, p<0.005. Significantly different from N171-82Q mice at ^#, p<0.05; ^##, p<0.005. (C, D, E, and F) Tail suspension test showed that shXIAP significantly increased forelimb, hind limb clasping, and torso movement, in WT (N=5) and HD transgenic (N171-82Q) mice (N=5). (G, H, I, and J) Cylinder test showed that shXIAP significantly affected spontaneous movements (first wall touch, the number of wall touch (rears), and total rearing) in WT (N=6) and was exacerbated in HD transgenic (N171-82Q) mice (N=6). Significantly different at *, p<0.05; **, p<0.005 (K) Kaplan-Meier survival analysis showed that shXIAP significantly reduced the life span of N171-82Q mice (WT+shCont, N=10; WT+shXIAP, N=10; N171-Q82+shCont, N=10; N171-Q82+shXIAP, N=10). Significantly different from WT mice at *, p<0.05. Significantly different from N171-82Q mice at ^##, p<0.005. (L) A scheme proposing that reduction of XIAP levels increased mitochondria p53 activity, mitochondria dysfunction, and ultimately leads to neuronal damage in the medium spiny neuron of HD.

XIAP overexpression improves mHTT-induced neuropathology and motor dysfunction in mice

To verify a gain of XIAP function on the motor coordination and locomotor activity of AAV-Control and AAV-mHTT-transduced mice, we performed rotarod, tail suspension, and cylinder tests, respectively (Fig. 9A, Supplementary Fig. 13A and B). As we supposed, XIAP overexpression improved rotarod performance in mHTT-transduced mice (Fig. 9B). Tail suspension test showed that XIAP overexpression significantly decreased forelimb, hind limb clasping, and torso movement in mHTT-transduced mice, indicating that XIAP overexpression rescued an balanced the state of motor coordination (Supplementary Fig. 13C–F). Furthermore, cylinder test showed that XIAP overexpression significantly affected spontaneous movements [the number of wall touch (rears) and rearing] in mHTT-transduced mice (Supplementary Fig. 13G–J). Next, we examined alterations of neuropathology in mHTT-transduced mice by XIAP. XIAP overexpression improved neuronal atrophy in the striatum of mHTT-transduced mice (Fig. 9C). We performed Nissl-staining with cresyl violet (CV) and confirmed the nuclear size change in MSNs (Fig. 9C). Semiquantitative analysis showed that XIAP overexpression significantly rescued the nucleus size of MSNs (Fig.9D). To determine whether XIAP overexpression in vivo reduces abnormal elevation of p53 levels in MSNs, we examined p53 levels in the striatum of mHTT-transduced mice (Fig. 9E). The intensity of XIAP and p53 levels was inversely correlated in three groups of mice (AAV-Cont, AAV-mHTT, and AAV-mHTT+AAV-XIAP) (Fig. 9F). XIAP overexpression rescued cytochrome c levels in the MSNs of AAV-mHTT-transduced mice (Supplementary Fig. 13K and L). Furthermore, XIAP overexpression improved neuronal atrophy in the striatum of N171-82Q mice (Supplementary Fig. 14A and B). Lastly, XIAP overexpression significantly decreased active caspase-3 immunoreactivity in the MSNs of N171-82Q mice (Supplementary Fig. 14C and D).

Figure 9. XIAP overexpression improves neuropathology and motor behavior in mHTT-transduced mice.

(A) A scheme illustrating experimental designs for behavioral tests and neuropathology. (B) XIAP overexpression significantly rescued rotarod performance in mHTT-transduced mice (N=5). Significantly different from Control mice at *, p<0.05; **, p<0.005. Significantly different from mHTT-transduced mice at ^#, p<0.05; ^##, p<0.005. (C) XIAP overexpression ameliorates neuronal atrophy. Nissl-staining with cresyl violet (CV) was performed in the striatum of AAV-Cont (Q25), AAV-mHTT (Q103) and AAV-XIAP viruses injected mice. Scale bars (black): 2.5μm. (D) XIAP overexpression prevented mHTT-induced neuronal atrophy in the striatum of AAV-XIAP co-injected mice compared to mice injected with only AAV-mHTT (Q103). (E) Transduction of AAV-XIAP decreased p53 levels in the dorsal striatum of mice compared to AAV-mHTT (Q103). Scale bars (white): 10μm. (F) The intensity of XIAP and p53 in GFP-positive cell was inversely correlated in AAV-HTT (Q25), AAV-mHTT (Q103) and AAV-mHTT+XIAP-transduced mice. [AAV-Q25, N=5; AAV-Q103, N=5; AAV-Q103+XIAP, N=5; cell counting, n=30 (6 cells/case)].

DISCUSSION

Abnormally expanded polyglutamine repeats in HD lead directly or indirectly to mitochondrial dysfunction and oxidative damage in medium spiny neurons. However, the exact mechanisms of the mitochondria-mediated HD pathology are not fully understood. In the current study, we made a series of novel findings in which alterations of XIAP and mitochondrial p53 activity account for the pathogenesis of HD in cellular and animal models.

XIAP and p53 levels are inversely altered in HD patients, mouse models, and mHTT striatal cells

First, we discovered that XIAP and p53 protein levels are inversely correlated in human, animal, and cell line models of HD. XIAP levels were significantly decreased in HD patients, three types of HD transgenic mouse models (N171-82Q, R6/2, and YAC128), and HD (STHdh^Q111/Hdh^Q111) cells compared to controls, whereas, both non-phosphorylated p53 and phosphorylated p53 (p-p53) protein levels were significantly increased in the mitochondria of all HD models (Bae et al., 2005; Goffredo et al., 2005). It has been reported that mHTT affects p53 activity in HD (Bae et al., 2005; Steffan et al., 2000). In our study, we verified that mHTT overexpression significantly reduced XIAP levels but increased p53 levels in both in vivo animal and in vitro cell systems. Based on the previous reports, we proposed that the reduction of XIAP level may be due to impaired transcriptional regulation via a negative feedback of abnormally increased p53 by mHTT (Bae et al., 2005; Steffan et al., 2000; Ye et al., 2015). But a precise mechanism of XIAP down-regulation by mHTT needs to be verified in the future study. p53 contributes to mitochondria-associated neuronal dysfunction and behavioral abnormalities of HD (Bae et al., 2005; Steffan et al., 2000). Herein, we found that p53 is highly increased in the mitochondria of striatal neurons of the HD brain, suggesting a direct and potential role of p53 in the mitochondria, which is associated with the HD pathogenesis. The precise mechanism on how p53 localizes to the mitochondria is not known yet (Mahyar-Roemer et al., 2004; Marchenko et al., 2000). It is reasonable to propose that certain posttranslational modifications of the p53 protein may facilitate its localization to the mitochondria (Bergeaud et al., 2013; Erster and Moll, 2004; Lee et al., 2018; Vaughn et al., 2007; Zhao et al., 2005). Although the functioning of ubiquitin-proteasomal system (UPS) in mHTT protein aggregate and inclusion formations has been regarded as an important and essential part of the cellular pathogenic mechanism (Li and Li, 2011), the presence and function of UPS in the mitochondria has not been demonstrated in neurons of HD. Since there is substantial evidence suggesting an interplay between the aberrant mitochondrial dysfunction and the pathogenesis of HD (Kim et al., 2010), we proposed that the impairment of mitochondrial survival components may contribute to the pathogenesis of striatal cell death in HD. In this regard, our data shows that the change of XIAP and p53 levels in human HD patients can be a neuropathological marker, indicating the mitochondrial and neuronal damage in HD (Goffredo et al., 2005).

XIAP interacts with and destabilizes p53 protein

Secondly, we identified that XIAP interacts directly with p53 in the brain. Full length XIAP and its fragments including BIR3 bound with WT-p53, p53 Mid-, and C-terminals in vitro. Importantly, we further found that XIAP was involved in the regulation of p53 half-life by reducing the stability of p53. It has been shown that XIAP is structurally characterized as E3 ubiquitin ligases since it contains a C-terminal RING finger domain, and one to three copies of the baculoviral IAP repeat (BIR) domain as their N-termini (Deveraux and Reed, 1999; Miller, 1999). Suzuki et al., reported that XIAP ubiquitinates caspases-3 and induces its degradation (Suzuki et al., 2001). The ubiquitination and degradation of p53 is a well-known process for the suppression of p53-induced apoptosis (Marchenko and Moll, 2007; Marchenko et al., 2007; Vaseva and Moll, 2009; Yang et al., 2000). Based on previous reports and our findings, we proposed that the loss of XIAP function may relieve p53 molecules to be stabilized and diverted into the mitochondria by shunting the ubiquitin proteasome system (UPS) (Gradzka et al., 2018). Indeed, interestingly, our study indicates that XIAP reduces the stability of p53 by facilitating autophagy activity and, in part, modulates UPS activity in the neuronal system (Poetsch et al., 2018). It seems likely that the autophagy function is a major turnover machinery for p53 in neuronal cells while the UPS is partially working for p53 destabilization. It is known that XIAP is involved in the activation of autophagy by elevating Beclin 1 expression through the NFKB signaling pathway in lymphoma-derived cell lines (Lin et al., 2015). Importantly, Poetsch et al., found that a loss of XIAP function leads to the accumulation of mature autophagosomes in mice and human brain cells (Poetsch et al., 2018). They suggest that XIAP regulates autophagic flux by promoting the fusion of autophagosome-lysosome. In addition, the XIAP pathway modulates antibacterial autophagy and its deficiency is associated with intestinal inflammation (Schwerd et al., 2017). As such, XIAP dysfunction by HD stress impaired the autophagic destabilization of p53 molecules. Consequently, increased p53 activity led to mitochondrial and neuronal damage in both in vitro cell systems and in vivo animal models (Carter et al., 2010). Our supplementary data supports that inhibition of autophagy function hinders XIAP-dependent p53 turnover in neurons (Supplementary Fig. 5). It has previously been shown that autophagic activity is altered in cortical and striatal neurons of HdhQ200 mutant mice (Heng et al., 2010). Not only do mHTT aggregates colocalize with LC3, but they also colocalize with ubiquitin conjugates in the striatal neurons of HdhQ200 mutant mice (Heng et al., 2010). While XIAP exhibits a negative regulatory effect on the turnover of p53 via autophagy in in vitro and in vivo models of HD, other previous studies have shown that XIAP plays a positive role in the turnover of p53 by reducing MDM2 level in different cell types such as HCT116 (a human colon cancer cell line) and PC12 (a rat adrenal pheochromocytoma) cells (Huang et al., 2013; Zhao et al., 2020). It seems likely that the role of XIAP on the turnover of p53 (or vice versa) is differentially regulated in a cell-type specific manner and a stress/stimulus-dependent manner (Huang et al., 2013; Zhao et al., 2020). Considering how both autophagic and UPS functions are abnormally changed in HD, it will be important to identify an exact molecular network and cross-talk pathway on which molecules other than XIAP are further involved in p53 turn-over and translocation to the mitochondria during the pathogenesis of HD in future studies.

XIAP prevents p53-induced mitochondrial dysfunction and neuronal damage in vitro and in vivo

Lastly, we confirmed a bona fide protective role of XIAP against p53-induced mitochondria and neuronal damage in HD cell lines and animal models. After the initial finding in the year 2000 that p53 is localized to mitochondria and modulates pro-apoptotic functions, others and we have shown that mitochondrially targeted-p53 (Mito-p53) triggers mitochondrial permeability transition pore opening and leads to cell death in solid tumor and ischemic brain injury model (Erster and Moll, 2004; Lee et al., 2018; Marchenko et al., 2000; Marchenko and Moll, 2007; Marchenko et al., 2007; Mihara and Moll, 2003; Vaseva and Moll, 2009). In the previous study we reported that SIRT3 prevents Mito-p53-triggered cytochrome c release and caspase-3 activation in primary neuron cultures (Lee et al., 2018). It seems likely that therapeutic regulation of Mito-p53-associated mitochondrial dysfunction could be a reasonable strategy to rescue neuronal damage, but a further study is necessary. The significant finding of current study is that the gain of XIAP function is to decrease the stability and half-life of endogenous p53, while the loss of XIAP function is to increase p53 level and its mitochondrial localization. In order to address the specific role of p53 in mitochondria, we constructed Mito-p53 and verified how Mito-p53 affects mitochondrial activities that involve the mitochondrial membrane potential, mitochondrial morphology, and cytochrome c release (Lee et al., 2018). Several studies have shown that the reactive oxygen species level in mitochondria was one of the damage markers in neurodegenerative diseases (Lee et al., 2011; Marchenko et al., 2000; Marchenko and Moll, 2007; Marchenko et al., 2007; Mihara and Moll, 2003).

Deregulation of mitochondria-dependent energy metabolism is one of pathological features in HD (Orr et al., 2008; Panov et al., 2002; Sawa et al., 1999). Mutant HTT protein directly interacts with the mitochondria membrane, alters mitochondrial calcium (Ca²⁺), and reduces mitochondrial membrane potential (Orr et al., 2008; Panov et al., 2002). In the perspective of direct interaction between mHTT and mitochondrial membrane, mHTT-induced mitochondrial dysfunction leads to the oxidative damage of MSN (Lee et al., 2011). Considering multiple effects of mutant HTT on cellular function, not only does the direct interaction of mHTT with mitochondria membrane affects mitochondrial toxicity in MSN, but also, modulation of other molecular pathways such as XIAP and p53 by mHTT can affect mitochondria function (Beal et al., 2004). In this context, our current study suggests that impaired XIAP and p53 activity-induced mitochondria dysfunction may be a thread of downstream pathways that are negatively and indirectly affected by mHTT. Interestingly, XIAP levels and mitochondrial oxidative stress levels were inversely correlated in mHTT transfected cells, whereas, p53 and mitochondrial oxidative stress levels were positively correlated in mHTT transfected cells. XIAP exhibited a neuroprotective function against Mito-p53-induced mitochondrial dysfunctions such as disruption of mitochondria membrane potential, Bax oligomerization, and caspases-3 activation, which result in cell death. In contrast, knockdown of XIAP exacerbated mito-p53-induced mitochondrial dysfunctions and cell death. Knockdown of XIAP increased both the protein and mRNA levels of p53, indicating that an impaired XIAP function disinhibited the turnover of p53 molecules. We further verified the protective role of XIAP against p53-induced neuronal death and survival in a primary neuronal culture system. Not only did XIAP prevent mito-p53-induced mitochondrial dysfunction by reducing cytochrome c release, but it also rescued the mitochondrial membrane potential and neuronal damage. Together, these findings suggest that XIAP plays an antagonistic role against p53-induced mitochondrial dysfunction and consequently provides a salubrious role for neuronal survival. In this paradigm, deregulation of XIAP increases p53 activity and mitochondrial dysfunction, and consequently induces cellular damage in MSNs during the pathogenesis of HD (Fig. 8l).

Loss of XIAP function exacerbates neuronal damage and motor symptom, and reduces the life span of HD mice

In order to examine the loss of XIAP function in vivo, we delivered an AAV-shControl or AAV-shXIAP virus bilaterally to the striatum of WT and HD transgenic (N171-82Q) mice, respectively. The N171-82Q mice (containing an N-terminally truncated human HTT cDNA) expressed the first 171 amino acids including 82 glutamine repeats (82Q) (Schilling et al., 1999). The truncated HTT cDNA expression is driven by a mouse prion protein promoter. N171-82Q mice were normal at birth through 1–2 months, but after that period, they lost body weight and showed HD-like phenotypes, which are motor dysfunctions such as an abnormal gait, tremors, hypokinesis, and increased hind limb clasping (Luthi-Carter et al., 2000; Schilling et al., 1999;, Schilling et al., 2001). The N171-82Q mice were dying prematurely with a lifespan of around 5 and 6 months. When we knocked down XIAP in the dorsal striatum, we found an increase of atrophic morphological change in medium spiny neurons in both WT and N171-82Q mice. As expected, the loss of XIAP function markedly exacerbated neuronal damage in N171-82Q mice compared to WT littermate mice. Knock-down of XIAP also significantly increased mHTT aggregation and active aspase-3 levels, while reducing DARPP32 levels in the striatum, indicating XIAP dysfunction as a trigger to striatal neuron damage. Importantly, the loss of XIAP function elevated both non-phosphorylated p53 and p-p53 levels in striatal neurons. Concurrent with in vitro data, impaired XIAP function increased p53 levels and resulted in mitochondrial dysfunction. Thus, the abnormal molecular balance between XIAP and p53 levels found in the striatal neurons of HD patients were feasibly mimicked in shXIAP-delivered HD (N171-82Q) mouse models. As a result, due to the knock-down of XIAP, motor abnormalities were significantly increased even in WT mice and further exacerbated in N171-82Q mice. XIAP dysfunction reduced the lifespan of N171-82Q mice. It seems likely that the neuropathological sequelae are associated with XIAP dysfunction and correlated with the shortening of the lifespan. Furthermore, our study verified that gain of XIAP function improves neuronal damage and motor symptom in both AAV-mHTT-transduced mice and HD (N171-82Q) mice.

In conclusion, our findings provide a new molecular mechanism that XIAP directly interacts with and modulates p53 stability. XIAP ameliorated p53-induced mitochondrial and neuronal damage. Notably, impaired XIAP function led to an abnormal increase in p53 activity and mitochondrial dysfunction, and consequentially damaged striatal neurons associated with motor behavior and neuropathology of HD. Together, modulation of the XIAP-p53 pathway may be a challengeable therapeutic target in HD (Goffredo et al., 2005).