Introduction
Huntington’s disease (HD) is an autosomal dominant neurodegenerative disorder, which is typically characterized by psychiatric disturbances, motor dysfunction and cognitive decline (Wilson et al., 1987). In HD, a gene located on the short arm of chromosome 4 named IT15 is mutated. IT15 encodes for the protein huntingtin (The Huntington’s Disease Research Collaborative Group, 1993), which mutation leads to a CAG expansion beyond the normal 10–35 triplet repeat range (Albin and Tagle, 1995). From a neuropathological point of view, a dramatic degeneration of neurons in the striatal part of the basal ganglia occurs in HD, accounting for the progressively severe motor dysfunction (Vonsattel et al., 1985; Hedreen et al., 1991; Storey et al., 1992).
Aside from the projection neurons, HD also affects interneurons such as parvalbumin and calretinin-containing neurons. These interneurons appear to degenerate at almost the same rate as the projection neurons (Ferrer et al., 1994). Conversely, the degeneration process relatively spares other subclasses of striatal interneurons, such as somatostatin-NPY and cholinergic interneurons.
Mutated huntingtin impairs the function of cAMP response element-binding protein (CREB; Steffan et al., 2000; Sugars and Rubinsztein, 2003; Sugars et al., 2004), suggesting that inhibition of CREB-mediated gene transcription could play an important role in HD (Kazantsev et al., 1999; Steffan et al., 2000; Nucifora et al., 2001; Mantamadiotis et al., 2002; Jiang et al., 2003). Accordingly, it was observed that cAMP cerebrospinal fluid levels are reduced in HD patients (Cramer et al., 1984) and that CREB-regulated gene transcription is down regulated in the R6/2 HD transgenic mouse (Luthi-Carter et al., 2000; Wyttenbach et al., 2001).
Of note, we previously observed that in parvalbumin-containing neurons, levels of activated CREB are markedly reduced after excitotoxic lesions, which could, at least in part, account for their selective vulnerability (Giampà et al., 2006).
In HD, the N-terminal fragment of mutated huntingtin forms ubiquitinated aggregates named neuronal intranuclear inclusions (NIIs; DiFiglia et al., 1997). These aggregates were shown to be able to interact with several transcription factors, thereby impairing their functions (Martindale et al., 1998; Gutekunst et al., 1999).
CREB binding protein (CBP) is a transcriptional co-activator that was shown to mediate neuronal survival signals (Bonni et al., 1999; Walton and Dragunow, 2000). Moreover, it was observed that NIIs sequestrate CBP in the R6/2 HD mice. Therefore, striatal function is impaired by the interaction between mutated huntingtin and CBP, causing a disruption of transcription, and leading to toxicity for the neuron.
Recently, we showed that a poly (ADP-ribose) polymerase 1 (PARP-1) inhibitor was neuroprotective in the R6/2 mouse model of HD, where sparing of striatal neurons was associated with an increased level of pCREB (Cardinale et al., 2015).
PARP-1, a 116-kD protein, member of PARP family, is a nuclear enzyme, consists of three main domains: the N-terminal DNA-binding domain (DBD), the automodification domain (AMD) and the C-terminal catalytic domain (Kameshita et al., 1984; Kurosaki et al., 1987) involved in the poly(ADP-ribosylation) reaction. The relationship between the inappropriate activation of PARP-1 and neurodegeneration has been demonstrated (Strosznajder et al., 2005; Kauppinen et al., 2011; Martire et al., 2013). In fact, the abnormal activation of PARP induces the release and the translocation of apoptosis-inducing factor (AIF) from the mitochondria to the nucleus causing a programmed cell death, caspase-independent, named parthanatos (Wang et al., 2009).
Many authors have focused the attention on PARP-1 inhibition by common inhibitors used in clinical for cancer treatment, able to restore the physiological cell functions such as mitochondrial activity, or the regulation of transcription factors including p53 required for neuronal survival (Martire et al., 2015).
In an earlier study, our group had shown that, in the R6/2 mouse of HD, the beneficial effects exerted by phosphodiesterase inhibitors on phenotype and on projection neurons sparing was associated to a rescue of parvalbumin positive interneurons and to an inhibition of CBP sequestration into NIIs (Giampà et al., 2009).
The aim of the present study was to deepen our knowledge of the effects of PARP-1-inhibition. In particular, we aimed at investigating the effects of PARP-1 inhibition on CBP localization, by reducing CBP sequestration into the NIIs, and thus decreasing cellular toxicity resulting from it. Moreover, we studied the effects of the compound on calretinin and parvalbumin containing striatal cells, two subsets of interneurons that are particularly vulnerable to HD. In summary, our aim was to verify if the modulation of CBP activity (Oliveira et al., 2006), on one hand, and the sparing of selected striatal interneurons (Torres et al., 1994), on the other, were associated with the previously described neuroprotective effects and the rescue motor deficits in HD mice.
Discussion
Our findings demonstrate that chronic treatment with the PARP inhibitor INO-1001 has a protective effect for striatal parvalbumin and calretinin-containing neurons in terms of neurodegeneration, and prevents CBP sequestration into NIIs.
Neurodegeneration affects neuronal subpopulations of the striatum in different ways. In fact, PARV-positive GABAergic interneurons display a cell vulnerability that is comparable to that of projection neurons (Ferrer et al., 1994; Liang et al., 2005). Notably, projection neurons and interneurons are different in their sensitivity to both ischemic and excitotoxic insults (Roberts and DiFiglia, 1989; Chesselet et al., 1990). However, the molecular basis of such different behavior remains to be fully elucidated. (Beal et al., 1989; Figueredo-Cardenas et al., 1994). One possibility that our group had investigated was that some interneurons tendency to survive longer (such as cholinergic interneurons) was related to their ability to maintain sufficient levels of BDNF (Fusco et al., 2003).
Striatal parvalbumin interneurons are GABAergic neurons that have a spiny varicose dendrites and indented nuclei (Celio, 1990; Bennett and Bolam, 1994) and axons with typically dense collateral arborizations (Cicchetti and Parent, 1996) and because they fire phasically at high frequency in response to cortical stimulation, they are considered fast-firing neurons (Kita, 1993; Kawaguchi et al., 1995). Striatal PV neurons are considered interneurons because of their morphological characteristics, which include a medium-sized cell body, aspiny varicose dendrites and a strongly indented nucleus (Celio, 1990; Bennett and Bolam, 1994).
GABAergic PV-positive interneurons only constitute 3%–5% of the cells in the rodent neostriatum (Kawaguchi et al., 1995), and yet they are able to exert a very powerful inhibitory control on the activity of cortical projecting neurons (Koos and Tepper, 1999). Cepeda et al. (2013) described how such a small population of neurons plays a pivotal role in HD. Indeed, by means of optogenetics and electrophysiology, they described that parvalbumin neurons are an important source for increased frequency of spontaneous GABA synaptic activity in the striatum.
Interestingly, these neurons are not only as vulnerable as projection neurons to HD but they also tend to form NIIs to the same extent as projection neurons, whereas other types of more resistant interneurons, such as cholinergic and somatostatin interneurons, form NIIs less frequently (Kosinski et al., 1999). Thus, the observation that PARP inhibition significantly spared the striatal PARV interneurons confirms the beneficial effect of the treatment on HD pathology.
However, a sparing of parvalbumin neurons could be desirable not only for HD, but also in other hyperkinetic neurological disorders. Indeed, striatal parvalbumin expression is decreased in a dystonic hamster model (Bode et al., 2017). Thus, disorders where parvalbumin interneurons fast spiking activity is involved such as dystonia could benefit from such neuroprotective effect (Reiner et al., 2003).
Calretinin is a calcium-binding protein containing a structural domain that binds calcium (Henzi et al., 2009). A protective role for calretinin has been postulated, as it was observed that CALR neurons are spared in a 6-OHDA-lesion model (Tsuboi et al., 2000) and also in HD patients (Massouh et al., 2008). Moreover, CALR was shown to be correlated to an impaired neurogenesis in adult HD (Fedele et al., 2011). In a recent report, Dong et al. (2012) demonstrated that CALR can interact with mutant huntingtin and that it can decrease cytotoxicity caused by mutant huntingtin in cellular models of HD.
We observed that CALR neurons were rescued by INO-1001 treatment, supporting the idea that functionally active CALR interneurons may participate in the survival of striatum in HD.
The observation that PARP-inhibition was effective in decreasing the size of NIIs was in agreement with our previous report about the reduced frequency of NIIs in the INO-1001 treated mice. The role of NIIs in HD has been discussed for many years now, and the question whether aggregates could be the basis of clinical manifestations and neurodegeneration in HD still remains unsolved (Davies and Scherzinger, 1997; Kim and Tanzi, 1998; Saudou et al., 1998; Sisodia, 1998). It is conceivable that HD aggregates could trigger cellular dysfunction leading to the death of those neurons in which they are contained. Indeed, NIIs formation can impair cellular functions by their ability to sequester proteins and transcription factors, thereby causing a dysfunction of a number of cellular transcriptional mechanisms (Maat-Schieman et al., 1999; Perutz, 1999; Preisinger et al., 1999; Meade et al., 2000). However, there is no clear demonstration that aggregates are the direct cause of HD neuronal death. Moreover, Saudou et al. demonstrated that neuronal death caused by the mutation is not directly related to NIIs formation in cultured cells transfected with mutant huntingtin. Additionally, Reiner and co-workers (Reiner et al., 1988) observed that the presence of NIIs was related to a prolonged survival in chimeric R6/2 mice.
However, our studies have shown that every treatment that proved beneficial for HD phenotype and neurodegeneration (DeMarch et al., 2008) was also effective in decreasing the frequency and the size of NIIs. Moreover, a toxic effect for aggregates was demonstrated. Therefore, polyQ aggregation could be considered as a therapeutic target for HD (Bates, 2003).
Transcriptional responses to cAMP are mediated by the interaction of CREB with CBP, and cAMP activated transcription is regulated by CREB phosphorylation (Azuma et al., 1996).
However, CBP is not only a transcriptional coactivator for CREB (Arany et al., 1994). Indeed, CBP is able to acetylate promoter proximal nucleosome histones, which results in an augmented availability of DNA for other important factors (Ogryzko et al., 1996). Of note, several studies clearly demonstrated that histone acetyltransferase activity could be altered by mutated huntingtin. Moreover, the enzyme dysfunction could be at the basis of the transcription deregulation occurring in HD. Interestingly, histone deacetylase inhibitors have proven to display a beneficial effect in several HD models (Sadri-Vakili and Cha, 2006).
CBP is abundant in the nuclei of 100% of wild type cells, but only in 18% of HD mutant cells (Gines et al., 2003). The nuclear localization of CBP is therefore a feature of normal cells, and its displacement away from the nucleus and sequestration into the aggregates accounts for the cellular dysfunction in HD. The phenomenon described is indeed typical of HD pathology, as it has been observed in several HD models, from cell cultures to human patients (Nucifora et al., 2001). An association between the depletion of CBP and mutant huntingtin-mediated cellular toxicity was previously demonstrated (Jiang et al., 2006).
Therefore, the physiological role of CBP is markedly impaired in HD and it participates in the neuronal degeneration not only via the altered CREB function but also through the detrimental effect on the mechanism of histone acetylation. Correspondingly, the improved functioning of CBP might be related not only to the restored CREB functions, but also to rescued physiological functioning of CBP per se.
Accordingly, we are reporting that INO-1001 is able to prevent CBP segregation into the aggregates. This confirms our recent report on neuroprotection exerted by INO-1001 in the R6/2 mice, and supporting the idea that reinstating the nuclear localization of CBP is able to participate to the correct cellular functioning. In line with our earlier report, we can establish that there is a strong correlation between CBP recruitment in NIIs and neuronal death.
Thus, we can speculate that the reinstated CREB function, along with the physiological localization of CBP in the nucleus, can concur in keeping adequate BNDF expression levels that are necessary to protect striatal neurons from HD neurodegeneration (Bemelmans et al., 1999; Pérez-Navarro et al., 2000).
We also provide the first evidence that parvalbumin and calretinin-containing interneuron are rescued by PARP inhibition in the R6/2 mouse model of HD. Therefore, these results demonstrate the possibility of ameliorating HD neuropathology by chronic peripheral administration of INO-1001 and, more generally, underline the potential therapeutic value of PARP inhibitors in HD.