IKKα and IKKβ Regulation of DNA Damage-Induced Cleavage of Huntingtin

Background Proteolysis of huntingtin (Htt) plays a key role in the pathogenesis of Huntington's disease (HD). However, the environmental cues and signaling pathways that regulate Htt proteolysis are poorly understood. One stimulus may be the DNA damage that accumulates in neurons over time, and the subsequent activation of signaling pathways such as those regulated by IκB kinase (IKK), which can influence neurodegeneration in HD. Methodology/Principal Findings We asked whether DNA damage induces the proteolysis of Htt and if activation of IKK plays a role. We report that treatment of neurons with the DNA damaging agent etoposide or γ-irradiation promotes cleavage of wild type (WT) and mutant Htt, generating N-terminal fragments of 80–90 kDa. This event requires IKKβ and is suppressed by IKKα. Elevated levels of IKKα, or inhibition of IKKβ expression by a specific small hairpin RNA (shRNA) or its activity by sodium salicylate, prevents Htt proteolysis and increases neuronal resistance to DNA damage. Moreover, IKKβ phosphorylates the anti-apoptotic protein Bcl-xL, a modification known to reduce Bcl-xL levels, and activates caspases that can cleave Htt. When IKKβ expression is blocked, etoposide treatment does not decrease Bcl-xL and activation of caspases is diminished. Similar to silencing of IKKβ, increasing the level of Bcl-xL in neurons prevents etoposide-induced caspase activation and Htt proteolysis. Conclusions/Significance These results indicate that DNA damage triggers cleavage of Htt and identify IKKβ as a prominent regulator. Moreover, IKKβ-dependent reduction of Bcl-xL is important in this process. Thus, inhibition of IKKβ may promote neuronal survival in HD as well as other DNA damage-induced neurodegenerative disorders.


Introduction
Huntington's disease is a neurodegenerative disorder caused by expansion of a CAG repeat, which is translated into a polyglutamine (polyQ) stretch in the N-terminus of Htt protein [1]. Neurotoxicity in HD is attributed to the cleaved N-terminal fragments of mutant Htt [2][3][4]. Wild type Htt is also cleaved and inactivated by proteases, and its deletion in the central nervous system (CNS) promotes neurodegeneration and is deleterious for development [5][6]. The neuroprotective functions of WT Htt include inhibition of caspase-3, induction and transport of brainderived neurotrophic factor (BDNF), and guarding against DNA damage and excitotoxicity [7][8][9][10][11].
DNA double-stranded breaks progressively accumulate in the aging brain, and elevated DNA damage is found in HD patients and HD animal models [11][12][13][14]. The vulnerability of neurons to DNA damage in HD is further exemplified by reduced expression of nuclear proteins such as high mobility group B 1(HMGB1), which protects against genotoxic stress [12,15]. Elevated expression of HMGBs protects against polyQ-induced neurotoxicity in primary neurons and in a Drosophila polyQ model [15]. DNA damage induced by oxidative stress is also an important factor in the development of neurotoxicity and phenotypic changes in a chemical model of HD [16]. Thus, accumulation of DNA damage is a potential regulator of HD pathology. However, the mechanism for how DNA damage influences neurotoxicity in HD is not well understood. One contributing factor may be the DNA damage-induced activation of p53 and IKK signaling pathways, which have been implicated in HD neurotoxicity [14,[17][18].
DNA damage is a potent inducer of IKK [17]. The core complex has two kinases, IKKa and IKKb, and a regulatory subunit, IKKc [19]. As an activator of NF-kB, the IKK complex regulates inflammation, cytokine production and cell survival. The IKKb subunit is the predominant kinase responsible for inflammatory responses [19]. Excessive IKKb activity is, however, implicated in several neurodegenerative disorders, including HD, Alzheimer's disease (AD), and Parkinson's disease (PD) [18,[20][21]. It is relevant that HD patients have chronically elevated levels of inflammatory cytokines in the serum and CNS long before the onset of symptoms [22], implying that persistent dysregulation of IKK may occur early in the disease. In contrast, IKKa can repress IKKb activity and reduces the production of inflammatory cytokines [23]. IKKa also has neuroprotective properties and promotes memory reconsolidation in the hippocampus [24]. Moreover, nuclear IKKa inhibits the activity of tumor suppressor p53 that is induced by DNA damage, and increases cellular resistance to genotoxic stress [25]. P53 is elevated in HD brains and reducing its activity ameliorates HD symptoms in animal models [14].
We previously showed that HD mouse models have elevated IKKb/NF-kB in the CNS, and blocking IKKb activity prevents degeneration of medium-sized spiny neurons caused by a toxic fragment of mutant Htt [18]. Here we report that DNA damageinduced IKKb is an important activator of Htt proteolysis, while IKKa inhibits this event. We present evidence for a signaling network that involves phosphorylation and reduction of Bcl-xL by IKKb, and subsequent activation of caspases, which can cleave Htt.

DNA damage has opposite effects on IKKa and IKKb
In this study we used a human embryonic neuronal stem cell line (MESC2.10) isolated from the midbrain of an eight-week-old embryo [26], to characterize the signaling between DNA damage, IKKb activation and Htt turnover. To establish the model, MESC2.10 neuroblasts were differentiated and examined for expression of neuronal markers. Upon differentiation, MESC2.10 cells acquire neuronal morphology (Fig. 1A) and express neuron-specific proteins such as PSD-95, b-catenin and the neurofilament Tuj-1 (Fig.1B, top two panels and 1C). These neurons can be maintained for more than two weeks without significant apoptosis (Fig. 1B, third panel). To induce DNA damage, we first used the topoisomerase inhibitor etoposide, which produces DNA doublestranded breaks in post-mitotic neurons [27]. The induction of DNA damage in MESC2.10 neurons was confirmed by nuclear staining of phosphorylated histone H2aX (c-H2aX), a surrogate marker of DNA damage (Figs. 1C and 1D) [28]. We used an acute etoposide treatment of 6 hr to avoid neuronal death, which occurs after prolonged incubation [27].
To determine if DNA damage activates IKK in neurons we first performed in vitro kinase assays using recombinant IkBa as the substrate for IKKb complexes immunoprecipitated from the extracts of etoposide-treated neurons. We find that IKKb is activated 2 hrs after etoposide treatment and remains active for up to 4 hrs ( Fig. 2A top panel). These results are consistent with the effects of etoposide on IKKb in non-neuronal cells [17]. We were also interested in whether DNA damage influences IKKa activity, since this subunit appears to play a prominent role in survival and neuronal plasticity [24][25]. To obtain IKKa complexes devoid of IKKb, neuronal extracts were first depleted by prior incubation with anti-IKKc and -IKKb antibodies, and IKKa complexes were subsequently immunoprecipitated. IKKa activity is routinely tested in vitro by phosphorylation of IkBa, although in vivo this phosphorylation is predominantly performed by IKKb [19]. Interestingly, MESC2.10 neurons display constitutive IKKa activity (Fig. 2B, top lane 1). MESC2.10 neurons are grown in complex medium with several growth factors, some of which might be responsible for activation of IKKa. Longer treatment with etoposide, however, reduces IKKa activity (Fig. 2B, lanes 4). We also observe a decrease in the level of IKKa protein, which may contribute to the low IKKa activity (Fig. 2B, bottom panel). Thus, induction of DNA damage has opposite effects on the activity of IKKa and IKKb in neurons.

DNA damage-induced proteolysis of Htt is regulated by IKKs
Induction of p53 by DNA damaging agents promotes the expression of Htt [29]. Consistent with these observations, we find that treatment of MESC2.10 neurons with etoposide increases the level of endogenous, full-length Htt ,4-fold (Fig. 3A, top panel, asterisk, lanes 1-3). The elevation of Htt overlaps in time with accumulation of nuclear p53 (Fig. 3B, top panel, lanes 1-4). Longer exposure of neurons to etoposide induces proteolysis of endogenous Htt, however, generating N-terminal fragments of ,80 kDa (Fig. 3A, top panel, lane 4, arrow). These data suggest that accumulation of DNA damage in neurons activates proteolytic enzymes that can cleave Htt.
To confirm that Htt cleavage is induced by DNA damage and not other, secondary effects of etopoisde, we examined whether DNA damage induced by c-irradiation of neurons could have a similar effect on Htt cleavage. As expected, c-irradiation also generates an Htt fragment similar in size to that produced by etoposide treatment (Fig. S1). Interestingly, induction of Htt proteolysis by c-irradiation occurs faster than etoposide treatment and is prominent by 4 hr post-irradiation, whereas maximal Htt proteolysis induced by etoposide requires ,6 hr. The difference maybe due to rapid induction of double stranded DNA breaks by irradiation.
Since etoposide treatment reduces the level and activity of IKKa (Fig. 2B), we examined whether increasing its level could rescue the effects of DNA damage on Htt protein. MESC2.10 neurons were transduced with an IKKa lentivirus, which increases the level of IKKa by ,3-fold (Fig. 3C). Indeed, this enables the IKKa + neurons to resist etoposide-induced Htt proteolysis (Fig. 3A, top panel, lanes 5-8). IKKa + neurons display a higher basal level of full-length WT Htt than controls (Fig. 3A, compare lanes 1 and 5), and accumulation of p53 in IKKa+ neurons (Fig. 3B, top panel, lanes [5][6][7][8] has no additional effect on Htt levels. The lack of further Htt elevation in IKKa + neurons may be due to inhibition of p53 transcriptional activity by elevated IKKa [25]. While the interaction between p53, IKKa and Htt expression remains to be explored in detail, these findings support a protective role for IKKa in reducing proteolysis of endogenous, WT Htt in neurons with DNA damage. In contrast, IKKb activity is induced by DNA damage (Fig. 2A). Thus, we hypothesized that reducing its level may be protective. Towards this end, we silenced IKKb expression with a specific, anti-IKKb small hairpin RNA (shRNA) expressed from a lentivirus (Fig. 4A), and treated the neurons with etoposide. We find that, similar to the effect of elevating IKKa, silencing IKKb expression reduces the proteolysis of Htt (Fig. 4B). Silencing of IKKb expression or inhibition of its kinase activity by sodium salicylate [30] also blocks Htt cleavage induced by c-irradiation (Fig. S1, lanes 8 and 5, respectively). Thus, IKKb activation by DNA damage promotes Htt cleavage, and increasing IKKa or reducing IKKb blocks this event.
It is relevant that etoposide-induced IKKb activates p65 NF-kB DNA binding in MESC2.10 neurons (Figs. 4C and D). Interestingly, the DNA binding activity of p65 is significantly reduced in neurons with elevated IKKa (Fig. 4C). While IKKa is known to inhibit IKKb/NF-kB in immune cells [23], this has not been reported for post-mitotic neurons. Therefore, the protective effects of IKKa in response to DNA damage may include inhibition of IKKb activity. However, inhibitors of NF-kB do not influence Htt proteolysis, suggesting that IKKb regulation of Htt proteolysis is NF-kB independent (see below). Etoposide has no effect on the activation of p52 NF-kB in MESC2.10 neurons (Fig. 4D).

IKKs influence the DNA damage-induced activation of pro-apoptotic caspases
Several caspases, including caspase-3 and -6, are known to cleave Htt between amino acids 500-600 generating fragments similar in size to those observed in our DNA damage paradigm ( Fig. 3A) [2][3][4][5]. To better understand the role of the IKKs in . Moreover, extracts of etoposide-treated IKKa + neurons contain reduced caspase-3 and caspase-6 activity (Figs. 5B and 5C, column 6, respectively). Blocking IKKb activity with sodium salicylate, a potent inhibitor of IKKb [30], or silencing its expression by shRNA also lowers the etoposideinduced activation of caspases (Figs. 5B and 5C, columns 4 and 8, respectively). Thus, the opposite effects of IKKa and IKKb on caspase activation in the context of DNA damage may underlie their differential effects on Htt cleavage.

DNA damage-induced IKKb regulates Bcl-xL
Activation of caspases can be triggered by several mechanisms, including reduction in Bcl-xL [31]. The prosurvival protein Bcl-xL, which is abundant in the CNS, has recently been implicated in Htt proteolysis in a chemical model of HD [32][33]. We find that the level of intact Bcl-xL is reduced in extracts of neurons treated with etoposide (Fig. 6A, lane 6). In contrast, etoposide treatment does not affect Bcl-xL in neurons with elevated IKKa or reduced IKKb expression (Fig. 6A, lanes 2 and 4, respectively). To confirm that Bcl-xL is important for blocking of DNA damage-induced Htt proteolysis, we increased its expression in MESC2.10 neurons using a recombinant lentivirus (Fig. 6B, second panel, lanes 3 and  4). As predicted, elevated Bcl-xL prevents Htt proteolysis by DNA damage and this overlaps in time with prevention of caspase-3 activation (Fig. 6B first and third panels, respectively). Although Bcl-xL expression prevents DNA damage-induced Htt cleavage, some reduction in the level of full-length Htt is also observed. Whether etoposide treatment induces production of other Htt fragments that are not detected by this anti-Htt antibody and not protected by Bcl-xL, remains to be investigated. Overall, these studies indicate that Bcl-xL level is a critical component of caspasemediated Htt proteolysis induced by DNA damage, and is likely influenced by IKKb.
The level of Bcl-xL mRNA in MESC2.10 neurons is not affected by etoposide (data not shown), indicating that Bcl-xL reduction is likely due to enhanced protein turnover. Although phosphorylation-induced degradation of Bcl-xL in the presence of genotoxic agents is a necessary step for induction of apoptosis, it is unclear which kinase(s) mediates this event [31]. We tested the possibility that activated IKKb phosphorylates Bcl-xL. To avoid contamination with other neuronal kinases that may co-immunoprecipitate with IKKb, we used recombinant IKKs for in vitro kinase assays with Bcl-xL as the substrate. Indeed, IKKb phosphorylates Bcl-xL (Fig. 6C, top panel), and this is specific since inhibition of IKKb by sodium salicylate prevents this reaction, and IKKa does not phosphorylate Bcl-xL. Thus, IKKb is a novel kinase that modifies Bcl-xL and reduces its level in stressed, post-mitotic neurons. This is consistent with the unchanged levels of Bcl-xL in etoposide-treated neurons in which IKKb expression has been silenced (Fig. 6A).

DNA damage-induced Htt proteolysis is polyQindependent
Generation of N-terminal fragments of mutant Htt is thought to initiate neurotoxicity, culminating in HD [2][3][4]. We tested whether DNA damage also promotes the proteolysis of full-length mutant Htt. In a striatal neuronal line obtained from HD knock-in mice (Hdh Q111/Q111 ) [34], we find that etoposide treatment also promotes cleavage of full-length, mutant Htt protein (Fig. 7, top  panel, lane 2). This proteolysis is blocked by inhibition of caspase-3 or IKKb activity (Fig. 7 top panel, lanes 3 and 4). Moreover, cleavage of Htt overlaps in time with reduction of Bcl-xL, which is prevented by inhibition of IKKb with sodium salicylate (Fig. 7,  second panel). Taken together, these results indicate that, in postmitotic neurons, DNA damage-activated IKKb facilitates Htt proteolysis indirectly by promoting Bcl-xL turnover and activating a caspase pathway. Thus, in the context of neuronal DNA  (Fig. 3C). Bars indicate S.E.M. and asterisk shows significant difference between control and IKKa+ neurons treated with etoposide for 4 hr, P ,0.01 using a student's t test. (D) Competition of etoposide-induced p65 NF-kB binding by consensus oligonucleotides. Nuclear extracts of etoposide treated neurons were pre-incubated with 100 ng of competitor NF-kB oligonucleotides (Clontech) on ice for 1 hr (Column 3) before treatment with etoposide for 6 hrs. Wells with mutated NF-kB DNA oligonucleotides were used to ensure specificity of the binding (column 4). (E) Binding of p52 NF-kB to consensus oligonucleotides is not changed by DNA damage. Experiments were similar to in part C, except binding was examined for p52. Bars indicate S.E.M. and asterisks show significant difference in binding between samples without or with the competitor oligonucleotides or with p65 binding to mutated NF-kB oligonucleotides, P value ,0.01, using a student's t test. doi:10.1371/journal.pone.0005768.g004 damage, IKKb activation is detrimental and its inhibition may be protective in HD and potentially in other neurodegenerative disorders where DNA damage plays a role.

Discussion
The novel aspects of present work are the triggering of Htt proteolysis by DNA damage and the identification of IKKb as a prominent regulator of enzymes known to cleave Htt. We propose that in post-mitotic neurons, IKKb activation by DNA damage promotes Htt cleavage by influencing Bcl-xL and a caspase pathway. Conversely, our data support a protective role for IKKa, whose expression in neurons prevents DNA damage-induced Htt proteolysis (Fig. 8).
Our data indicate that activation of IKKb by DNA damage may injure neurons by increasing the turnover of WT Htt. It is notable that both WT and mutant Htt are cleaved in our DNA damage paradigm. Neurons in HD brains, which have WT and mutant copies of Htt, may therefore be susceptible to DNA-induced damage on both counts. In addition to loss of WT Htt function, cleavage of the mutant Htt could also generate toxic N-terminal fragments that are amyloidogenic [4]. Thus, DNA damageinduced IKKb in the HD brain may set in motion one of the earliest events in HD pathology. Other insults such as chronic exposure to pro-inflammatory cytokines, which are elevated in the plasma and brain tissue of pre-symptomatic HD patients, may also cause persistent activation of IKKb in neurons and further exacerbate the cleavage of Htt [19,22]. A potential source for these cytokines is the activated microglia that are detected in various parts of HD brains, including striatum and hypothalamus [39,40]. It is interesting that deletion of IKKb in microglia reduces neurodegeneration in a chemical model of HD by suppressing the expression of pro-inflammatory cytokines [41]. While the role of cytokines in IKKb-dependent Htt proteolysis remains to be characterized, we can speculate that activation of IKKb in neurons may trigger Htt proteolysis cell autonomously, as in the induction of DNA damage, or non-autonomously by chronic exposure to pro-inflammatory cytokines produced by microglia (41). It is also of interest that inhibitors of IKKb reduce proteolysis of amyloid precursor protein (APP) and subsequent Ab42 production [38]. The regulation of amyloidogenic proteins such as Ab and N-terminal Htt fragments by activated IKKb is an intriguing area for future work.
In contrast to IKKb, IKKa promotes neuronal survival and prevents Htt proteolysis. This could simply result from suppression of IKKb activity by IKKa (Fig. 4C) [23]. However, IKKa has several other neuroprotective properties that could be important. For example, IKKa activates CREB binding protein (CBP), a transcriptional co-activator of many neuronal signaling pathways including BDNF expression [25,42]. Enhancement of the histone acetyltransferase activity of CBP is known to reduce the toxicity of mutant Htt [43]. IKKa also inhibits the transcriptional activity of p53, which has been implicated in HD pathogenesis, and promotes neuronal plasticity and memory formation in the hippocampus [14,[24][25]. Thus, IKKa may prevent Htt proteolysis independent of IKKb by modulating expression and/or activation of gene products that antagonize the toxic effects of  figure 3A and examined for the levels of procaspase-3 (top panel) or procaspses-6 (middle panel) by Western blotting. Arrow shows the cleaved products of procaspase-3. (B and C) Caspase-3 (B) and caspase-6 (C) activities are shown in MESC2.10 neuronal lysates. For the specific inhibitors neurons were first pretreated with 20 mM of Ac-DEVD-CHO, caspase-3 inhibitor (C3I) or 20 mM of Ac-VEID-CHO, caspase-6 inhibitor (C6I), or 5 mg/ml of sodium salicylate (NaSal) one hr prior to etoposide treatment for 6 hrs. Extracts were incubated with either caspase-3 substrate (DEVD conjugated to p-nitroanaline) or caspase-6 substrate (VEID conjugated to p-nitroanline) in a 96 well plate at 37uC for 1 hr. Enzyme activities for caspase-3 (B) or caspase-6 (C) were measured in a microplate reader. Results are shown as relative enzyme activity and represent averages of three experiments. Bars indicate S.E.M. and asterisk shows significant difference from etoposide treated control neurons (column 2), p,0.01, using a student's t test. doi:10.1371/journal.pone.0005768.g005 DNA damage. Further examination of IKKa using gene therapy or compounds that selectively enhance its functions in vivo is worthy of investigation in HD animal models.
The mechanism of Htt proteolysis is complex and probably involves integration of multiple signaling pathways. Our results identify Bcl-xL as a potential mediator of DNA damage-induced Htt proteolysis (Fig. 6). Bcl-xL participates in neuronal survival,  and its degradation results in synaptic degeneration [32]. Our findings indicate that DNA damage reduces Bcl-xL levels in an IKKb-dependent manner. A role for IKKb is supported by the inability of etoposide to reduce Bcl-xL in neurons expressing an shRNA targeting IKKb, or following pretreatment with sodium salicylate, a potent inhibitor of IKKb (Figs. 6A and 7). We propose that IKKb reduces Bcl-xL levels by phosphorylation (Fig. 6C), a modification known to promote Bcl-xL degradation [31]. Phosphorylation-dependent reduction of Bcl-xL is suggested to play a role in spinal cord neuronal injury and is implicated in Htt proteolysis in the striatum of 3-nitropropionic acid injected mice [33,44]. Thus, reduction of Bcl-xL may be the signal that leads to activation of caspases that target Htt (Fig. 8). Inhibition of caspase-3 activation and Htt proteolysis in neurons with elevated Bcl-xL further supports this notion (Fig. 6B).
Environmental factors appear to play a role in the development of HD, since the age of onset is variable among patients with similar polyQ expansions [45]. Accumulation of DNA damage as well as activation of modifiers that influence Htt proteolysis could impact disease progression. Our data indicate that IKKb induced by DNA damage in neurons can be very significant, and inhibition of IKKb can prevent depletion of WT Htt as well as the production of toxic N-terminal fragments of mutant Htt. It is noteworthy that IKKb is itself activated by N-terminal fragments of mutant Htt [18]. Thus, Htt proteolysis and IKKb activation can form a toxic feedback loop that could promote neurodegeneration. This cycle can be influenced by cytokines that are induced by IKKb activation and which are elevated in HD patients [22]. IKKa, on the other hand, may act as a brake to suppress excessive IKKb activation and reduce neurotoxicity (Fig. 8). Thus, the ratio of IKKa to IKKb activity may be a determinant of Htt proteolysis in stressed neurons. Identification of the IKKs as regulators of Htt proteolysis offers novel targets to search for molecules that prevent one of the earliest events in HD pathogenesis.

Generation of MESC2.10 human neurons
The methods for generation and differentiation of MESC2.10 human neuroblasts have been reported previously [26]. Briefly, neuroblasts obtained from an 8 week-old human embryo were transduced with a retrovirus encoding a tetracycline-regulated vmyc to promote proliferation. These neuroblasts were grown on poly-lysine and laminin coated plates in DMEM/F12 in the presence N2 and B-27 neuronal supplements (Invitrogen) and 20 ng/ml FGF-2 (Promega). To differentiate MESC2.10 cells, proliferation medium was replaced DMEM/F12 medium containing 2 mg/ml of doxycycline and 5 mM cAMP.
Etoposide treatment, cell fractionation and Western blotting MESC2.10 cells, differentiated for 9 days, were treated with 10 mM etoposide for the periods indicated in each figure. Neurons were harvested and separated into cytoplasmic and nuclear fractions using the NE-PER kit (PIERCE) according to instructions. For most experiments we used ,120 mg of lysate for SDS-PAGE and Western blotting with the indicated antibodies. Reactive bands in Western blots were detected by enhanced chemiluminescence (ECL) using a gel documentation system. Sodium salicylate (5 mg/ml), caspase-3 inhibitor (Ac-DEVD-CHO) and casapse-6 inhibitor (Ac-VEID-CHO) were added at Bcl-xL and enhance its degradation (arrow 1). Reduction of Bcl-xL levels triggers the activation of caspases, which cleaves Htt. IKKb inhibition block degradation of Bcl-xL, caspases activation, and proteolysis of Htt. Similar to the inhibition of IKKb, elevation of Bcl-xL also prevents caspase activation and Htt proteolysis. On the other hand, etoposide treatment reduces the activity of IKKa (arrow 2). This may enhance IKKb activation and/or block expression of neuroprotective proteins that are essential for interfering with caspase activation and maintaining Htt levels. Elevated IKKa expressed from a lentivirus overcomes these deficiencies and prevents Htt proteolysis. doi:10.1371/journal.pone.0005768.g008 20 mM 1 hr prior to etoposide treatment. To examine the effects of etoposide on mutant Htt differentiated mouse striatal cells (Hdh Q111/Q111 ) were treated with etoposide and analyzed as described for MESC2.10 cells.

Immunochemistry
Differentiated neurons on coverslips were treated with 10 mM etoposide for 4 hr. Cells were fixed and stained with a rabbit antibody that specifically recognizes H2aX phosphorylated at Ser 139 (c-H2aX) (1:500) (R&D systems). Anti-Tuj-1 was used to label the cytoplasm (1:1000). Goat anti-rabbit conjugated to FITC (green) and goat anti-mouse conjugated to rhodamine (red) was used as secondary antibodies. Pictures were captured with a confocal microscope.

Lentivirus production
IKKa was cloned into the lentiviral FUGW vector under the control of a ubiquitin promoter [46]. An EGFP-lentivirus was used as a control. The Bcl-xL cDNA was cloned from MESC2.10 neurons by RT-PCR using standard procedures and its identity confirmed by sequencing. The cDNA was subsequently inserted into FUGW lentiviral vector. The shRNAs for IKKb cloned in a lentiviral backbone was purchased from Open Biosystems (Huntsville, AL). Lentiviruses were produced by transfection of 293 cells using calcium phosphate precipitation [46]. Supernatants of virus-producing cells were harvested 48 hr post-transfection and concentrated on Amicon Ultra columns (Millipore, MA). An EGFP virus was used as a control to monitor viral titer. A multiplicity of infection of 4:1 was used to infect MESC2.10 neuroblasts. Expression of IKKa, Bcl-xL and reduction of IKKb were determined by Western blotting.

Kinase assay
IKKb activity was assayed as described previously [18]. To determine IKKa activity, lyastes were first depleted of IKKb complexes using 4 mg of anti-IKKc and anti-IKKb antibodies coupled to protein G. The depleted lysates were used to isolate IKKa complexes with 2 mg of anti-IKKa antibody. GST-IkBa was used as a substrate to measure kinase activity. To examine whether IKKs phosphorylate Bcl-xL, 0.5 mg of either IKKa or IKKb were incubated with 1 mg of full-length Bcl-xL in the presence of 32 P-c-ATP for 30 min at 30uC. GST protein was used as a negative control. All kinase products were examined by SDS-PAGE and autoradiography.
Assay of NF-kB binding to consensus DNA oligonucleotides Nuclear extracts for control or etoposide-treated MESC2.10 neurons were obtained using the cell fractionation kit from pierce according to the instructions provided. Fifty mg of each nuclear extract was incubated for an hour at RT on 96 well plates (Clonetech, Mountain View, CA) coated with consensus NF-kB oligonucleotides (p65 or p52). Mutated NF-kB oligonucleotides were used to confirm binding specificity. For competition assays, nuclear extracts were preincubated with NF-kB oligonucleotides for 1 hr on ice and then added to the coated wells. After washing, each well was incubated with anti-p65 or anti-p52 antibodies for 30 min at 37uC. Wells were washed as recommended by the manufacturer (Clontech), followed by incubation with a secondary antibody conjugated to HRP. TMB (3, 39, 5, 59-tetramethylbenzidine) was added for 10 min. Binding was measured in a microplate reader at 655 nm.

Caspase assay
To measure the activity of caspase-3 or -6, the colorimetric assay kits from R&D Systems were used according to manufacturer's instructions. Briefly, MESC2.10 neurons were pre-incubated with caspase-3 inhibitor (Ac-DEVD-CHO), casapse-6 inhibitor (Ac-VEID-CHO) or sodium salicylate (5 mg/ml) 1 hr prior to etoposide treatment, which was for an additional 6 hrs. Cells were lysed as instructed. Fifty mg of each lysate was incubated with either caspase-3 substrate (DEVD conjugated to p-nitroanaline) or caspase-6 substrate (VEID conjugated to p-nitronalaine) in a 96 well plate at 37uC for 1 hr. Enzyme activities for caspase-3 or -6 were measured with a microplate reader at 405 nm. Results are shown as relative enzyme activity and represent average of three experiments. Figure S1 c-irradiation induced DNA damage promotes Htt cleavage in MESC2.10 neurons. c-irradiation was carried out using a MARK-I c-irradiator with a 137Cs source at a specific dose rate of 1.22 Gy/min. Cells were irradiated with 5 Gy and further incubated for the indicated time. Sodium salicylate (NaSal, lane 5) was added at a concentration of 5 mg/ml 1 hr prior to irradiation and incubated for 6 hr. Etoposide treatment (10 mM) was used as a positive control and was carried out for 6 hr (lane 6). Lanes 7 and 8 represent MESC2.10 neurons transduced with an IKKb shRNA lentivirus (Fig. 4A) and irradiated with 5 Gy for 6 hr. The asterisk shows full-length Htt and the arrow indicates the cleaved products. Tubulin was used as a loading control. Found at: doi:10.1371/journal.pone.0005768.s001 (4.20 MB TIF)