Dystonia 16 (DYT16) mutations in PACT cause dysregulated PKR activation and eIF2α signaling leading to a compromised stress response

Abstract

Dystonia 16 (DYT16) is caused by mutations in PACT, the protein activator of interferon-induced double-stranded RNA-activated protein kinase (PKR). PKR regulates the integrated stress response (ISR) via phosphorylation of the translation initiation factor eIF2α. This post-translational modification attenuates general protein synthesis while concomitantly triggering enhanced translation of a few specific transcripts leading either to recovery and homeostasis or cellular apoptosis depending on the intensity and duration of stress signals. PKR plays a regulatory role in determining the cellular response to viral infections, oxidative stress, endoplasmic reticulum (ER) stress, and growth factor deprivation. In the absence of stress, both PACT and PKR are bound by their inhibitor transactivation RNA-binding protein (TRBP) thereby keeping PKR inactive. Under conditions of cellular stress these inhibitory interactions dissociate facilitating PACT-PACT interactions critical for PKR activation. While both PACT-TRBP and PKR-TRBP interactions are pro-survival, PACT-PACT and PACT-PKR interactions are pro-apoptotic. In this study we evaluate if five DYT16 substitution mutations alter PKR activation and ISR. Our results indicate that the mutant DYT16 proteins show stronger PACT-PACT interactions and enhanced PKR activation. In DYT16 patient derived lymphoblasts the enhanced PACT-PKR interactions and heightened PKR activation leads to a dysregulation of ISR and increased apoptosis. More importantly, this enhanced sensitivity to ER stress can be rescued by luteolin, which disrupts PACT-PKR interactions. Our results not only demonstrate the impact of DYT16 mutations on regulation of ISR and DYT16 etiology but indicate that therapeutic interventions could be possible after a further evaluation of such strategies.

Graphical Abstract

Introduction

Integrated stress response (ISR) is an evolutionarily conserved pathway activated in eukaryotic cells by many different types of stress stimuli in order to restore cellular homeostasis (Pakos-Zebrucka et al., 2016). The central event in this pathway is the phosphorylation of eukaryotic translation initiation factor 2 (eIF2α) on serine 51 by one of the four serine/threonine kinases (Donnelly et al., 2013, Taniuchi et al., 2016). This post-translational modification prevents the formation of the ternary complex during translation initiation, leading to a decrease in general protein synthesis while allowing induction of selected genes that promote cellular recovery (Wek, 2018). While transient eIF2α phosphorylation is favorable for cellular survival, prolonged eIF2α phosphorylation is pro-apoptotic due to the upregulation as well as preferential translation of pro-apoptotic transcripts (Donnelly et al., 2013). Thus, although ISR is primarily a pro-survival response to restore cellular homeostasis, exposure to severe stress drives signaling towards cellular death. Thus, ISR tailors the cellular stress response in a specific manner to the cellular context as well as the nature and severity of the stress signal.

The interferon (IFN) inducible double-stranded RNA (dsRNA)-activated protein kinase (PKR) is a ubiquitous eIF2α kinase (Garcia et al., 2007, Meurs et al., 1990) active under cellular stress conditions such as viral infection, oxidative and endoplasmic reticulum (ER) stress, and serum deprivation (Ito et al., 1999, Patel et al., 2000). In virally infected cells, PKR is activated by direct interactions with dsRNA, a viral replication intermediate for many viruses (Barber, 2001). However, in the absence of viral infections other stress signals activate PKR via its protein activator (PACT) (Patel and Sen, 1998) in a dsRNA-independent manner. Two evolutionarily conserved amino terminal dsRNA binding motifs (dsRBMs) of PKR mediate its interactions with dsRNA (Feng et al., 1992, Green and Mathews, 1992, Patel and Sen, 1992) as well as with PACT (Huang et al., 2002, Peters et al., 2001) and other regulatory proteins (Chang and Ramos, 2005). Upon binding dsRNA or PACT, PKR undergoes a conformational change which results in the autophosphorylation and activation of PKR (Cole, 2007, Nanduri et al., 1998). In the absence of stress, however, PKR is inhibited through direct interactions with the transactivation response element (TAR) RNA binding protein (TRBP) via the dsRBMs of each protein (Benkirane et al., 1997, Laraki et al., 2008). TRBP was initially discovered due to its strong affinity to the TAR RNA element of HIV (Benkirane et al., 1997) inhibits PKR both by sequestration of dsRNA and by direct interaction during viral infections (Daniels and Gatignol, 2012). In the absence of stress, TRBP inhibits PKR via the formation of both TRBP-PACT and TRBP-PKR heterodimers (Daher et al., 2009, Singh et al., 2011, Singh and Patel, 2012).

PACT is a stress-modulated activator of PKR that works via a direct, dsRNA-independent interaction in response to ER stress, oxidative stress, and serum deprivation (Bennett et al., 2004, Bennett et al., 2012, Ito et al., 1999, Patel et al., 2000). Similar to TRBP, PACT contains three copies of the conserved dsRBMs and the two amino terminal motifs, dsRBM1 and 2, are critical for dsRNA binding and protein-protein interactions and a carboxy terminal dsRBM3 motif that does not bind dsRNA being essential for PKR activation (Huang et al., 2002, Patel and Sen, 1998, Peters et al., 2001). Within dsRBM3, serines 246 and 287 serve as phosphorylation sites to promote PACT-PACT homomeric and PACT-PKR heteromeric interactions (Peters et al., 2006, Singh and Patel, 2012). In the absence of stress, PACT is constitutively phosphorylated on S246 (Peters et al., 2006), bound to TRBP (Daher et al., 2009) and is unable to activate PKR. In response to cellular stress, PACT is phosphorylated on S287 which promotes its dissociation from TRBP to trigger PACT-PACT homomeric interactions (Daher et al., 2009, Peters et al., 2006, Singh et al., 2011, Singh and Patel, 2012) which are required for PKR activation (Peters et al., 2006, Singh et al., 2011, Singh and Patel, 2012). Once activated, PKR phosphorylates eIF2α on serine 51 resulting in the attenuation of general protein synthesis (Garcia et al., 2006) and triggering downstream ISR events including ATF4 and CHOP induction that in turn regulate cellular fate either by restoring homeostasis or inducing apoptosis.

Recently, eight different mutations have been identified in Prkra gene (encoding PACT, OMIM: DYT16, 612067) in patients with a neuromuscular movement disorder dystonia 16 (DYT16) (Camargos et al., 2012, Camargos et al., 2008, de Carvalho Aguiar et al., 2015, Dos Santos et al., 2018, Lemmon et al., 2013, Quadri et al., 2016, Seibler et al., 2008, Zech et al., 2014). The dystonias are a heterogeneous group of movement disorders in which the affected individuals exhibit repetitive and painful movements of the affected limbs, as well as compromised posture and gait patterns (Bragg et al., 2011, Geyer and Bressman, 2006). DYT16 is a rare, early-onset dystonia parkinsonism disorder characterized by progressive limb dystonia, laryngeal and oromandibular dystonia and parkinsonism. Although DYT16 was originally described to have an autosomal recessive inheritance pattern (Camargos et al., 2008), four dominantly inherited variants of DYT16 have also been reported (Seibler et al., 2008, Zech et al., 2014). Previously, our lab reported that a recessively inherited P222L mutation increases cell susceptibility to ER stress through the dysregulation of eIF2α stress response signaling in DYT16 patient derived lymphoblasts (Vaughn et al., 2015). Furthermore, using an in-vitro approach we have demonstrated that a dominantly inherited frameshift mutation expresses a truncated PACT protein that disrupts PACT-TRBP heterodimers increasing PACT mediated PKR activation causing an enhanced sensitivity to ER stress via dysregulation of the eIF2α signaling pathway (Burnett et al., 2019). In accordance with our findings, subsequent reports identified dysregulation of eIF2α signaling in both DTY1 and DYT6 (Beauvais et al., 2016, Beauvais et al., 2018, Rittiner et al., 2016, Zakirova et al., 2018). Collectively, these findings indicate a potential common link among several forms of dystonia.

In the present study we characterize the effects of three recessively inherited (C77S, C213F, C213R) and two dominantly inherited DYT16 point mutations (N102S and T34S) on their ability to regulate PKR activation and ISR. Our data demonstrates that although these mutations have no effect on PACT’s dsRNA binding ability and PACT-TRBP interactions, the dominant mutations show enhanced ability to interact with PKR. Most significantly, all the DYT16 mutations under study demonstrated a heightened capacity to form PACT-PACT homodimers in the absence of stress. Furthermore, using lymphoblasts derived from a compound heterozygous DYT16 patient containing both C213R and P222L mutations as independent alleles, we identified stronger binding affinity between PACT and PKR in the DYT16 patient cells and a dysregulation of the eIF2α stress response and ISR. The DYT16 patient lymphoblasts also demonstrated an increase in cell susceptibility to ER stress that could be rescued in the presence of luteolin, a potent inhibitor of PACT-PKR interactions. Our work further strengthens the case for involvement of dysregulated eIF2α signaling as a mechanism in the disease etiology and lays the groundwork for exploring possible therapeutic options for DYT16.

Results

DYT16 mutations do not affect PACT’s dsRNA-binding activity.

The majority of DYT16 mutations characterized in the present study occur outside of PACT’s highly conserved dsRBMs (Figure 1A). Four of the mutations associated with the recessively inherited DYT16 (C77S, C213F, C213R, and P222L) result in the loss of a cysteine or proline residues which could have dramatical consequences on the 3-dimensional conformation of the protein (Figure 1A). Furthermore, the two dominantly inherited mutations (N102S and T34S) occur on flanking ends of PACT’s first dsRBM that is most critical for dsRNA binding and protein-protein interactions (Figure 1A) (Chukwurah et al., 2018). As seen in Fig. 1 B and C, the DYT16 point mutants show no change in their dsRNA binding capabilities in comparison to the wt PACT (lanes 1–14). In order to ascertain the specificity of the dsRNA-binding assay, we used in vitro translated firefly luciferase, which has no dsRNA-binding activity as a negative control (lanes 19–20). Additionally, we demonstrate the specificity of the interaction for dsRNA by adding excess dsRNA or ssRNA as competitors. As seen in lanes 15–18, the binding to dsRNA immobilized on beads can be effectively competed by exogenously added dsRNA but not single-stranded (ss) RNA (lanes 15–18).

Figure 1: Effect of DYT16 mutations on dsRNA-binding.

(A) Schematic representation of DYT16 mutations: Of the three conserved dsRBMs, M1 and M2 are shown in grey and the third motif lacking dsRNA-binding (M3) is shaded blue with the two phosphorylation sites represented as dark blue lines. Dominant DYT16 mutations are indicated in red while recessive mutations are indicated in green. (B) dsRNA-binding assay: dsRNA binding activity of wt PACT and DYT16 point mutants was measured by a poly(I)·poly(C)-agarose binding assay with in vitro translated ³⁵S-labeled proteins. T, total input; B, proteins bound to poly(I)·poly(C)-agarose. Competition lanes (15–18): no competitor (−), competition with 100-fold molar excess of single-stranded RNA (ss) or dsRNA (ds). The minor bands below the full-length PACT bands represent products of in vitro translation from internal methionine codons and thus are not produced in similar quantities in all translation reactions and thus are of variable intensity. Lanes 19 and 20 represent binding of firefly luciferase protein to poly(I)·poly(C)-agarose, used as a negative control to demonstrate specificity. (C) Quantification of the dsRNA binding assay. Bands were quantified by phosphorimaging analyses, and % bound was calculated. Error bars: S.D. from three independent experiments. The p-values were calculated using statistical analyses indicated no significant difference between % dsRNA-binding of wt and point mutants.

DYT16 mutants activate PKR more efficiently.

PACT is best characterized for its ability to activate PKR under conditions of cellular stress (Patel et al., 2000, Singh et al., 2011, Singh et al., 2009, Singh and Patel, 2012). Therefore, we next evaluated the consequence of each of the DYT16 mutations for PACT’s ability to activate PKR using an in vitro PKR activity assay. Hexahistidine tagged wt PACT and DYT16 mutant proteins were expressed and purified from bacterial cells using nickel affinity chromatography. The purified recombinant proteins were used as activators in an in vitro PKR activity assay by adding in increasing amounts to PKR immunoprecipitated from HeLa cells. We are then able to determine efficiency of PKR activation by comparing PKR autophosphorylation in the presence of wt PACT and the various PACT mutants (Figure 2A). Some amount of basal levels of activated PKR are observed in lanes 1 and 10 (upper panel) and lanes 1 and 8 (lower panel) in the absence of any added activator. When the purified recombinant PACT proteins are added, a dose dependent increase (left: 400 pg, right: 4.0 ng) in activated autophosphorylated PKR is observed (lanes 2–9, 11–16 for upper panel and lanes 2–7, 9–12 for lower panel). The amount of radioactivity present in PKR bands was quantified using a phosphorimager analysis and is shown in Fig. 2 B. In all cases, recessive mutations demonstrated a slightly increased capacity to activate PKR (Fig. 2A, lanes 4–12 and Fig. 2 B) as compared to wt PACT (Fig. 2 A, lanes 2–3). Interestingly, when tested in combinations as reported in DYT16 patients, the recessive mutants showed significantly enhanced ability to activate PKR (Fig. 2 A upper panel lanes 13–16, and Fig. 2 B) The dominant mutants (lower panel) also showed enhanced ability to activate PKR at 400 pg (Fig. 2 A, lower panel lanes 4–7). Interestingly, when tested in combination with wt PACT, both the dominant mutants demonstrated significantly higher PKR activation (Fig. 2A lower panel: lanes 9–12, and Fig. 2 B). These results indicate that the DYT16 point mutants have enhanced ability to activate PKR as compared to wt PACT.

Figure 2: Effect of DYT16 mutations on PKR activation and cell fate:

(A) PKR kinase activity assay. Kinase activity assay was performed using PKR immunoprecipitated from HeLa cell extracts using a monoclonal PKR antibody (R&D Systems) and protein A-sepharose beads. Either 400 pg (lanes 2,4,6,8 top panel, and lanes 2,4, 6 bottom panel) or 4 ng (lanes 3,5,7,9 top panel, and lanes 3,5,7 bottom panel) of recombinant wt PACT or DYT16 mutant proteins were used as PKR activators. Lanes 13–16 (upper panel) and Lanes 9–12 (lower panel): PACT mutants in combinations reported in DYT16 patients were used as PKR activator with 200 pg (lanes 11,13,15 top panel and lanes 9,11 bottom panel) or 2 ng (lanes 12,14,16 top panel and lanes 10,12 bottom panel) of each mutant protein. (B) Quantification of kinase activity assay. Radioactivity in each band was quantified using phosphoimaging analysis and the relative signal intensities were plotted. Blue bars: PKR activity seen with 400 pg and orange bars: PKR activity seen with 4 ng of the corresponding pure recombinant PACT protein. The p values are as indicated. (C) Western blot analysis for cleaved PARP1. Whole cell extracts from normal (wt) and DYT16 patient derived lymphoblasts treated with 5 μg/ml of tunicamycin (TM) were analyzed at indicated time points using anti-cleaved PARP1 and anti-β-actin antibodies. (D) Caspase-Glo 3/7 activity. Lymphoblast lines established from wt and DYT16 patient were treated with 5 μg/ml tunicamycin and the caspase 3/7 activities were measured at indicated time points. Blue bars: wt cells, and red bars: DYT16 cells. The data is an average of three independent experiments and the p values are as indicated.

DYT16 patient derived lymphoblasts are more susceptible to ER stress.

As the DYT16 mutant proteins exhibited an increased ability to activate PKR, we next utilized the lymphoblast lines derived from a DYT16 patient and his normal, wt parent to determine the effect of one particular DYT16 mutation combination on cell viability in response to stress. It is important to note that DYT16 is a rare, early-onset movement disorder and patient cells are not available for most of the DYT16 patients. Here we characterize the effect of ER stress on DYT16 compound heterozygote patient derived lymphoblast cells expressing both P222L and C213R mutations as independent alleles. We compared these cells to wt lymphoblast cell lines derived from an unaffected family member. Consequently, we utilized the ER stress inducing agent, tunicamycin (TM), which results in the accumulation of misfolded proteins in the ER due to inhibition of protein glycosylation. In the case of wt lymphoblasts, over a 24-hour time course in response to TM treatment we observed a marginal increase in expression of cleaved PARP1, a marker of cellular apoptosis (Fig. 2 C lanes 2–7) (Oslowski and Urano, 2011). In contrast to this, in the DYT16 patient derived lymphoblasts, there was a dramatically significant increase in cleaved PARP1 in response to tunicamycin (Fig. 2 C, lanes 13–14). To further validate these results, we performed caspase 3/7 activity assays under the same conditions to measure apoptosis. In wt lymphoblasts we detect caspase activity at 24 h but not at 6 h post-treatment (Figure 2 D, blue bars). In contrast, the DYT16 patient lymphoblasts demonstrate significantly elevated caspase activity at 6 h which further increases at 24 h post-treatment (Figure 2 D, red bars). This further supports that the DYT16 patient lymphoblasts are significantly more susceptible to ER stress and exhibit increased apoptosis as compared to wt cells possibly due to a failure to restore homeostasis.

eIF2α phosphorylation and ISR is dysregulated in DYT16 patient lymphoblasts.

In order to elucidate the underlying mechanism driving heightened sensitivity to ER stress in DYT16 lymphoblasts, we performed western blot analysis on cells treated with TM under the same conditions probing for markers of cellular stress response (Fig. 3). We compared the kinetics of both eIF2α phosphorylation and PKR activation in the DYT16 lymphoblasts to the wt lymphoblasts from the unaffected family member. In wt lymphoblasts (left) we observe a low basal level of eIF2α phosphorylation in the untreated cells (Fig. 3 A, lane 1) followed by increased eIF2α phosphorylation at 1–4 hours post treatment (lanes 2–4) and then restoration to basal levels by 8 hours (lane 5). In contrast to this, in the DYT16 lymphoblasts (right), we observe a similar increase in eIF2α phosphorylation 1 hour after treatment (lane 7), however, the eIF2α phosphorylation is sustained even at 8-hours post treatment (lanes 8–10). We also studied the time course of PKR activation in DYT16 patient lymphoblasts under the same conditions. In wt lymphoblasts (left) we observe PKR activation at 1 hour after TM treatment that is sustained until 4 hours (lanes 1–4) and shows a slight decrease by 8 hours (Figure 3A). In contrast to this, the DYT16 lymphoblasts (right) exhibit a dramatically elevated level of activated PKR even in untreated cells (lane 6) that does not show any stress-dependent increase after treatment with TM (lanes 7– 10). As we noted the differences in eIF2α and PKR phosphorylation responses between wt and DYT16 lymphoblasts, we examined if the downstream effects of eIF2α phosphorylation also show similar differences. In wt lymphoblasts (left), ATF4 is undetectable in untreated cells (Fig. 3 B, lane 1) and its expression increases in a time dependent manner from 1–8 hours post treatment (lanes 2–5) and declines at 12 and 24 hours after treatment (lanes 6–7). In contrast, in the DYT16 patient lymphoblasts (right) although we observe increased expression of ATF4 from 1–8 hours post treatment (lanes 9–11), it persists at high levels even at 12 hours post treatment and shows only a small decline at 24 hours after treatment. Finally, we compared levels of CHOP, an ATF4-induced pro-apoptotic protein, in response to TM treatment in wt and DYT16 lymphoblasts. CHOP is undetectable in untreated cells (Fig. 3 B, lane 1) and its expression increases in a time dependent manner from 2–8 hours post treatment (lanes 3–5) and declines at 12 and 24 hours after treatment (lanes 6–7). In contrast, in the DYT16 patient lymphoblasts (right) we observe a delay in expression of CHOP and it is not detected until 4 hours post treatment (lane), and it persists at high levels at 8–24 hours post treatment (lanes 12–14). Collectively these results demonstrate a dysregulation of ISR pathway, prolonged phosphorylation of eIF2α, elevated levels of activated PKR, prolonged elevated levels of ATF4 translation, and delayed but sustained induction of CHOP.

Figure 3: PKR activation and ISR in response to tunicamycin in normal and DYT16 patient lymphoblasts.

(A) Western blot analysis for p-PKR and p-eIF2α. Whole cell extracts from normal (wt) and DYT16 patient derived lymphoblasts treated with 5 μg/ml of tunicamycin (TM) were analyzed at indicated time points. Blots were probed for p-eIF2α, total eIF2α, p-PKR, and total PKR. Best of four representative blots are shown. (B) Western blot analysis for ATF4 and CHOP. Whole cell extracts from normal (wt) and DYT16 patient derived lymphoblasts treated same as in 3A were analyzed at indicated time points. Blots were probed for ATF4, and CHOP. Best of four representative blots are shown. β-actin was used as a loading control to ensure equal amounts of protein was loaded in each lane.

Effect of DYT16 mutations on PACT-PKR interactions.

In light of the heightened basal levels of PKR activation observed in the DYT16 patient cells (Fig. 3A), we next wanted to investigate the effect of these DYT16 mutations on PACT-PKR interactions. To address this, we performed co-immunoprecipitation (co-IP) assays using cells expressing a combination of myc-epitope tagged wt or DYT16 mutant PACT and flag-epitope tagged PKR. PKR is expressed at low basal levels in cells and both increased PKR activation and increased PKR expression levels are toxic to cells as it induces apoptosis. Thus, in order to evaluate PACT-PKR heterodimer formation we utilized an expression vector encoding Flag-tagged K296R, a catalytically inactive PKR mutant, which inactivates PKR’s catalytic activity without affecting PACT-PKR or any other interactions (Cosentino et al., 1995). Previously our lab has reported that the recessively inherited DYT16 mutation, P222L, shows an increased ability to form PACT-PKR heterodimers relative to wt PACT (Vaughn et al., 2015). Here our results show that the other recessively inherited mutations (C77S, C213F, and C213R) show no difference in their ability to interact with PKR relative to wt PACT (Figure 4 A, lanes 2–5). In the absence of myc-PACT, no flag-PKR is immunoprecipitated confirming that there is no non-specific binding of flag-PKR to the beads in the absence of myc-PACT (co-IP panel, lane 1). Lanes 7–10 demonstrate equal amounts of myc-PACT proteins were immunoprecipitated in each lane (top panel) while input gels (lower panel) demonstrate equal expression of each myc-PACT expression construct (lanes 7–10) and flag-PKR (lanes 1–5). In contrast, we do observe an increase in the PACT-PKR heterodimer formation in case of dominantly inherited mutations (N102S and T34S) under the same conditions (Fig. 4 B). As compared to wt PACT (lane 2), co-IP of the dominant mutants N102S and T34S (lanes 3–4) is significantly increased. No co-IP of myc-PACT is seen in the absence of flag-PKR (lane 1), thus demonstrating that there is no non-specific interaction of PACT proteins with the beads in the absence of flag-PKR. Lanes 6–8 (upper IP panel) demonstrate equal amounts of flag-PKR was immunoprecipitated in each lane, while input panels demonstrate equal expression of all constructs (lower panel, lanes 1–4, and 6–8).

Figure 4: Effect of DYT16 mutations on PACT-PKR interaction.

(A & B) Co-IP assays: HeLa cells were co-transfected with flag-PKR and myc-PACT expression plasmids in pCDNA3.1-. 24 hours post-transfection, cells were harvested and myc-PACT (A) or flag-PKR (B) was immunoprecipitated using myc-agarose or flag agarose beads. The immunoprecipitates were analyzed by western blot analysis with anti-flag (A) or anti-myc (B) antibodies (co-IP panel) and with anti-myc (A) or anti-flag (B) for IP panels. Input gels show expression levels of proteins without immunoprecipitation. (A) Recessive DYT16 mutants and (B) Dominant DYT16 mutants. (C and D) Mammalian two-hybrid assays. HeLa cells were transfected with 250 ng of each of the two test plasmids encoding proteins to be tested for interaction, 50 ng of the reporter plasmid pG5Luc, and 1 ng of plasmid pRL-Null to normalize transfection efficiency. Cells were harvested 24 h after transfection, and cell extracts were assayed for luciferase activity. The plasmid combinations are as indicated, PKR was expressed as a GAL4 DNA-binding domain fusion protein (bait) and all PACT proteins were expressed as VP16-activation domain fusion proteins (preys). The experiment was repeated twice with each sample in triplicate, and the averages with standard error bars are presented. The p values are as indicated for samples with significant differences in interaction. RLU, relative luciferase units.

In order to validate the co-IP results, we tested the PACT-PKR interactions using the mammalian two-hybrid (M2H) assay. In agreement with co-IP data, our results demonstrate that the recessively inherited mutations C77S, C213F and C213R have no difference in their ability to interact with PKR (Fig. 4 C). Consistent with our previously reported data, the P222L mutant demonstrates a stronger binding to PKR as indicated by greater induction of the luciferase reporter gene compared to wt PACT (Fig. 4 C). In the case of the P222L mutation, we observed about 2.5-fold increase in the PKR interaction as compared to wt PACT, whereas, the other recessive mutants showed similar PKR interaction as the wt PACT. Similarly, our results from the co-IP data were confirmed in case of the dominant mutations (Fig. 4 D). The T34S mutant showed about 2.25-fold increase and the N102S mutant showed about 4.25-fold increase in PKR interaction relative to wt PACT (Fig. 4 D).

Effect of DYT16 mutations on PACT-PACT interactions.

As PACT-PACT interactions are critical for activation of PKR, it is most relevant to assess if the DYT16 mutations affect PACTs ability to form homomeric interactions. Consequently, using the same protein-protein interaction studies outlined in Fig. 4 we addressed whether PACT-PACT interactions were affected by the DYT16 mutations (Fig. 5). We co-expressed myc- or flag-epitope tagged PACT proteins transiently by co-transfection of the respective expression constructs in combinations seen in patients. Our data shown in Fig. 5 A–C demonstrates that all DYT16 mutants show a dramatic increase in their ability to form PACT-PACT homodimers in the absence of stress as compared to wt PACT. We observe minimal wt PACT homodimerization (Fig. 5 A, lane 2) with this being variable and no interaction being detected in few experimental repeats as it is established that in the absence of stress, PACT-PACT dimerization is usually absent. The recessively inherited DYT16 mutations show enhanced C77S-C213F and P222L-C213R interactions as compared to wt PACT-wt PACT interactions (compare lanes 3–4 to lane 2). In case of the dominantly inherited mutations we tested their ability to form wt PACT-mutant dimers (Fig. 5 B), as well as mutant-mutant dimers (Fig. 5 C). We did not observe any wt PACT homodimerization in the absence of stress (Fig. 5 B and C, lane 2), however, both the dominant DYT16 mutants N102S and T34S showed enhanced interaction with wt PACT (Fig. 5 B, lanes 3–4) with N102S showing the strongest interaction with wt PACT. When evaluating these dominant mutations for their ability to interact with themselves, we observe very strong interaction between N102S-N102S and T34S-T34S (Fig. 5 C, lanes 3–4) as compared to wt PACT-wt PACT with the strongest interaction being T34S-T34S. We do not observe any co-IP of myc-tagged wt PACT in the absence of flag-tagged wt PACT (lane 1) demonstrating the absence of any non-specific binding to the beads (Fig. 5 A–C). The IP panels show that equal amounts of flag-tagged PACT protein was immunoprecipitated in each lane (Fig. 5 A–C, upper panel, lanes 5–8), and input blots indicate equal expression of each construct (Fig. 5 A–C, lower panel, lanes 1–8).

Figure 5: Effect of DYT16 mutations on PACT-PACT interactions.

(A-C) Co-IP assays to measure PACT-PACT interaction with mutant protein combinations as present in DYT16 patients. HeLa cells were co-transfected with flag-PACT and myc-PACT expression plasmids in pCDNA3.1-. 24 hours post-transfection, cells were harvested and flag-PACT was immunoprecipitated with flag-agarose beads. The immunoprecipitates were analyzed by western blot analysis with anti-myc antibodies (co-IP panel) or anti-flag antibodies (IP panel). Input gels show expression levels of proteins without immunoprecipitation. (A) Recessive DYT16 mutants, (B) Dominant DYT16 mutant interactions with wt PACT and (C) Dominant DYT16 mutant interactions with dominant mutants (homomeric interactions). Input gels show expression levels of proteins without immunoprecipitation. (D and E) Mammalian two-hybrid assays. HeLa cells were transfected with 250 ng of each of the two test plasmids encoding proteins to be tested for interaction, 50 ng of the reporter plasmid pG5Luc, and 1 ng of plasmid pRL-Null to normalize transfection efficiency. Cells were harvested 24 h after transfection, and cell extracts were assayed for luciferase activity. The plasmid combinations are as indicated, various PACT proteins were expressed as a GAL4 DNA-binding domain fusion proteins (bait) and also as VP16-activation domain fusion proteins (preys). The experiment was repeated twice with each sample in triplicate, and the averages with standard error bars are presented. The p values are as indicated. RLU, relative luciferase units.

To further confirm our co-IP data, we tested the interaction between DYT16 PACT mutants utilizing the M2H (Fig. 5 D–E). As seen in Fig. 5 D and E, in the patient specific combinations all the recessive mutants show enhanced interactions relative to wt PACT-wt PACT interaction (Fig. 5 D). The P222L-C213R and C213F-C77S interactions are ~5-fold and ~9-fold higher than wt PACT-wt PACT interaction respectively (Fig. 5 D). Furthermore, the dominant mutants T34S and N102S also show enhanced interactions (Fig. 5 E). The wt PACT-T34S and N102S-wt PACT interactions are ~10-fold and ~30-fold higher than wt PACT-wt PACT interactions. Finally, the T34S-T34S and N102-N102S interactions are enhanced ~10-fold and ~20-fold respectively compared to wt PACT-wt PACT interactions. These results further strengthen our co-IP data that the DYT16 mutations enhance PACT’s ability for forming PKR activating homomeric interactions.

PACT’s ability to interact with TRBP is not affected by the DYT16 mutations.

In the absence of stress, TRBP binds PACT and prevents the formation of PACT-PACT homodimers that could result in PKR activation. Thus, changes in PACT’s interaction with TRBP can consequentially affect PKR activation and previously, our lab has reported that the recessively inherited DYT16 P222L mutation increases PACT’s binding affinity to TRBP ultimately resulting in delayed PKR activation (Vaughn et al., 2015). We thus determined the consequence of the DYT16 mutations under study on PACT-TRBP heterodimer formation and our results indicate that the recessively inherited mutations, C77S, C213F, and C213R (Fig. 6 A, lanes 2–5) as well as the dominantly inherited N102S and T34S mutations (Fig. 6 B, lanes 2–4) have similar binding affinity to TRBP relative to wt PACT. As we do not detect the presence of myc-wt PACT in the absence of flag-TRBP expression (lane 1, Fig. 5 A and B) we can rule out any nonspecific binding of myc-PACT to the beads. Finally, IP blots indicating that equal amount of myc-TRBP protein was immunoprecipitated (Fig. 5 A and B, IP panels, lanes 7–10 and lanes 6–8) and input blots demonstrating equal protein expression are shown (Fig. 5 A and B, input panels, lanes 1–10 and lanes 1–8).

Figure 6: Effect of DYT16 mutations on PACT-TRBP interactions.

(A, B) Co-IP assays. HeLa cells were co-transfected with flag-TRBP and myc-PACT expression plasmids in pCDNA3.1-. 24 hours post-transfection, cells were harvested and flag-TRBP was immunoprecipitated with flag-agarose beads. The immunoprecipitates were analyzed by western blot analysis with anti-myc antibodies (co-IP panel) and anti-flag antibody (IP panel). Input gels show expression levels of proteins without immunoprecipitation. (A) Recessive DYT16 mutants, (B) Dominant DYT16 mutants. (C and D) Mammalian two-hybrid assays. HeLa cells were transfected with 250 ng of each of the two test plasmids encoding proteins to be tested for interaction, 50 ng of the reporter plasmid pG5Luc, and 1 ng of plasmid pRL-Null to normalize transfection efficiency. Cells were harvested 24 h after transfection, and cell extracts were assayed for luciferase activity. The plasmid combinations are as indicated, TRBP protein was expressed as a GAL4 DNA-binding domain fusion protein (bait) and various PACT proteins as VP16-activation domain fusion proteins (preys). The experiment was repeated twice with each sample in triplicate, and the averages with standard error bars are presented. RLU, relative light units. The p values are as indicated, n.s. indicates not significant.

We also validated these results using the M2H to determine the relative strengths of PACT-TRBP interactions. Consistent with our previously reported data, the P222L mutation shows ~2-fold increase in interaction with TRBP relative to wt PACT (Fig. 6 C). The other recessively inherited mutations, however, show no difference in their ability to interact with TRBP relative to wt PACT (Fig. 6 C). Consistent with the co-IP data, we also do not observe any difference in the PACT-TRBP interaction for the dominantly inherited mutations relative to wt PACT (Fig. 6 D). These results confirm that the new DYT16 mutations examined here do not change PACT’s interactions with TRBP and validate our earlier report that the P222L DYT16 mutation enhances PACT’s interaction with TRBP (Vaughn et al., 2015).

DYT16 Patient lymphoblasts show stronger PACT-PKR interactions and its disruption rescues their higher sensitivity to ER stress.

In light of the increased PACT-PACT interaction independent of cellular stress observed in fig. 5 and the elevated basal levels of activated PKR observed in fig. 3, we next wanted to investigate if the PACT-PKR interaction in patient derived lymphoblasts is also stronger as compared to wt lymphoblasts derived from the unaffected family member. In order to address this question, we treated these cells with luteolin, a flavonoid that has been previously been established to efficiently inhibit or disrupt PACT-PKR interactions (Dabo et al., 2017). As seen in Fig. 7 A, in the wt lymphoblasts we can detect some PACT-PKR interaction (upper panel) prior to luteolin treatment (lane 2), and at 1 h after luteolin treatment PACT-PKR interactions are barely detectable (lane 3) and a further time dependent decrease in the PACT-PKR interaction is seen from 1–8 h (lanes 3–5), with the interaction no longer be detected at 24 h post-treatment (lane 6). In the DYT16 patient lymphoblasts, we observe much higher PACT-PKR interaction prior to luteolin treatment (lane 8) and the interaction persists until 2 h (lanes 9–10) then decreasing slowly at 4 h and 8 h after luteolin treatment (lanes 11–12). We do see a complete loss of PACT-PKR interactions at 24 h after treatment in the DYT16 patient lymphoblasts (lane 12). IP blots (lower panel) demonstrate that equal amounts of PKR were immunoprecipitated in all lanes (lanes 1–12), and the input blots demonstrate that equal amount of protein was present in all IP samples. We do not detect the presence of PACT or PKR in samples incubated overnight in the absence of PKR antibody thus demonstrating that there is no nonspecific binding of PKR or PACT to the beads in the absence of PKR antibody (lanes 1 and 7). These results confirm that PACT-PKR interaction is stronger in DYT16 patient cells as compared to the wt cells and that a 24 h treatment with luteolin disrupts the interaction in wt as well as DYT16 cells.

Figure 7: Effect of luteolin on PACT-PKR interaction and caspase activation in response to tunicamycin in DYT16 patient lymphoblasts.

(A) Co-IP of endogenous PKR and PACT proteins. Lymphoblasts from unaffected family member (wt) or DYT16 patient (patient) were treated with 50μM luteolin. The cell extracts were prepared at the indicated times, and endogenous PKR protein was immunoprecipitated using anti-PKR mAb and protein A-sepharose, which immunoprecipitates total PKR. The immunoprecipitates were analyzed by western blot analysis with anti-PACT monoclonal antibody (Co-IP panel). The blot was stripped and re-probed with anti-PKR mAb to ascertain an equal amount of PKR was immunoprecipitated in each lane (IP panel). Input blot: Western blot analysis of total proteins in the extract with anti-PACT and anti-PKR mAbs showing equal amount of PACT and PKR in all samples. (B) Effect of luteolin on Caspase 3/7 activity in lymphoblasts. Lymphoblasts from unaffected family member (wt) and DYT16 patient (patient) were treated for 24 hours with 50 μM Luteolin (red) or left untreated (blue) followed by treatment with 5 μg/ml tunicamycin (TM). Caspase 3/7 activity was measured at indicated time points after tunicamycin treatment. The p values are as indicated.

Prevalence of PACT-TRBP heteromeric interactions promote cell survival while both PACT-PACT homomeric and PACT-PKR heteromeric interactions promote apoptosis. Our data indicates that there is an increase in both PACT-PACT (Fig. 5) and PACT-PKR interactions (Fig. 4 and Fig. 7A), and the DYT16 patient lymphoblasts are more susceptible to ER stress induced apoptosis (Fig. 2 D). Therefore, we next wanted to determine if disrupting the PACT-PKR interaction in the DYT16 patient lymphoblasts would lead to an increase in cell viability in response to ER stress. As a 24 h treatment with luteolin of both wt control and DYT16 patient lymphoblasts could significantly disrupt PACT-PKR interactions (Fig. 7 A, lanes 6 and 12), we tested if prior luteolin treatment would be protective for DYT16 patient lymphoblasts after ER stress. As seen in Fig 7 B, the wt control lymphoblasts we do not detect any caspase 3/7 activity in our untreated samples or at 6 h after TM treatment but there is a significant increase in caspase activity at 24 h post treatment (Fig. 7 B, blue bars). The cells treated for 24 h with luteolin prior to TM treatment show a marked reduction in caspase 3/7 activity (Fig. 7 B, red bars). In contrast, the DYT16 patient cells show higher basal levels of caspase 3/7 activity prior to TM treatment that is enhanced at 6 h post-treatment and a further increase is seen at 24 h (Fig. 7 B, blue bars). This increase is dramatically reduced, especially at 24 h post treatment, when cells are treated with luteolin 24 h prior to TM treatment (Fig. 7 B, red bars). These results demonstrate that disrupting PACT-PKR interactions with luteolin in DYT16 cells can protect the cells from ER stress-induced apoptosis.

Discussion:

DYT16 is an early-onset, generalized dystonia caused by mutations in the Prkra gene, which encodes for PACT (Patel and Sen, 1998), a stress-modulated activator of PKR (Patel et al., 2000). In response to cellular stress, PACT activates PKR leading to eIF2α phosphorylation and inhibition of general protein synthesis (Patel et al., 2000). Although primarily a protective response to restore homeostasis, if PKR remains active for prolonged periods it triggers cell death via apoptosis (Gil and Esteban, 2000, Singh et al., 2009). Our previous work has established that the regulation of PKR activation in response to stress depends on shifting the PKR inhibitory (PACT-TRBP and TRBP-PKR) interactions to PKR-activating (PACT-PKR and PACT-PACT) interactions soon after the cell encounters the initial stress signal (Daher et al., 2009). This is regulated by stress-induced PACT phosphorylation at serine 287, which dissociates PACT from TRBP and allows for its interaction with PKR (Singh et al., 2011, Singh and Patel, 2012). Previously we investigated the effects of a recessively inherited DYT16 missense mutation P222L on PACT-induced PKR activation in response to ER stress (Vaughn et al., 2015). Our results indicated that P222L activates PKR more robustly and for longer duration but with initial lag and slower kinetics as compared to wt PACT. In addition, the affinity of PACT-TRBP, PACT-PACT as well as PACT-PKR interactions was also enhanced in DYT16 patient lymphoblasts homozygous for P222L mutation. The initial lag in PKR activation and eIF2α phosphorylation was due to stronger TRBP-PACT interaction ultimately leading to a delayed but prolonged and intense PKR activation due to stronger PACT-PACT and PACT-PKR interactions causing enhanced cellular death. In addition, our previous work on a dominant frameshift DYT16 mutation that results in truncation of the PACT protein after 88 amino acids (Seibler et al., 2008) also demonstrated a dysregulation of PACT-PKR-eIF2α pathway (Burnett et al., 2019). The truncated mutant PACT protein formed aggregates in cells and caused PKR activation by displacing TRBP from PACT-TRBP complexes to promote PACT-PKR interaction, eIF2α phosphorylation, caspase activation and apoptosis.

In the present study we evaluated the effects three recessive (C77S, C213R, and C213F) (de Carvalho Aguiar et al., 2015) and two dominant DYT16 mutations (T34S and N102S) (Zech et al., 2014) on PKR activation. Our results establish that similar to two previously examined DYT16 mutants (Burnett et al., 2019, Vaughn et al., 2015), dysregulation of ISR is a common feature of all five DYT16 mutations. However, there are some important differences in the mechanism by which the dysregulation of ISR is brought about by these five mutants. Similar to the P222L mutant, all five mutants show stronger PACT-PACT interactions as well as enhanced PKR activation but unlike the P222L and the frameshift DYT16 mutants, none of these five mutants exhibited changes in PACT-TRBP interactions. The recessive mutants tested in combinations as found in DYT16 patients, as well as the two dominant mutants exhibited marked enhancement of PACT-PACT interactions in both co-IP and mammalian two-hybrid assays (Fig. 5). Using DYT16 lymphoblasts from a compound heterozygote patient, we observe that the enhanced PACT-PKR interactions (Fig. 7) and elevated PKR kinase activity (Fig. 2 A and B) leads to a dysregulation of ISR and increased apoptosis in response to ER stress (Fig. 2, C and D).

Based on our results, we predict that the neuronal cells carrying mutant PACT proteins will be unable to cope well with cellular stress and restore homeostasis. Although the observed enhanced apoptosis in lymphoblast cells is significant and indicative of possible increased neuronal apoptosis, it is of limited scope and must be taken with caution and more meaningful molecular studies on neurons derived from DYT16 patient cells should be undertaken in the future. There is significant expression of PACT as well as PKR in neurons based on our studies on murine brain as well as human cultured neurons (data not shown). The PACT-PKR stress response pathway functions ubiquitously in all cell types including neurons (Chen et al., 2006, Paquet et al., 2012, Vaughn et al., 2014) and PACT mediated PKR activation and its involvement in neurodegeneration has been noted in Alzheimers patients and mouse models (Paquet et al., 2012). As currently there are no derived DYT16 neurons available, our studies on patient lymphoblasts indicate that considerable efforts involved in undertaking in-depth studies using DYT16 derived neurons would be worthwhile in future.

Although neurodegeneration is the expected long-term outcome of enhanced neuronal apoptosis, neither apoptosis not neurodegeneration has so far been systematically investigated in blood or brain of dystonia patients, and this lack of information is usually interpreted as neurodegeneration generally being absent in dystonia patients. On the other hand, there is some evidence of increased neuronal apoptosis in the context of DYT16. Most DYT16 patients develop symptoms in early childhood and in the case of the compound heterozygous patient carrying P222L and C213R mutant alleles included in this study, imaging studies had revealed progressive MRI abnormalities with significant bilateral volume loss in the basal ganglia (Brashear, 2013, Lemmon et al., 2013), which agrees with the enhanced apoptosis observed in our experiments. This patient developed dystonia after a febrile illness, which could be a possible cellular stress event that may have triggered progressive cellular dysfunction or loss. In accordance with our earlier in vitro studies on three Brazilian P222L homozygous patients that showed enhanced apoptosis (Vaughn et al., 2015), the imaging studies performed on one Portuguese P222L homozygous patient showed marked bilateral loss of striatal presynaptic dopamine transporters, suggesting nigrostriatal neurodegeneration as a possible feature of the disease (Pinto et al., 2020). In addition to these imaging studies on DYT16 patients, there is evidence of neuronal apoptosis in dystonic mice with Prkra mutations. A spontaneously arisen, recessive insertion mutation in Prkra was identified at the Jackson Laboratory (JAX) that results in a progressive dystonia, kinked tails, and mortality (Palmer et al., 2015). Some neurons in the dorsal root ganglia and the trigeminal ganglion were noted to be apoptotic in the homozygous mutant mice, consistent with the observed neurodegenerative phenotype. In agreement with this, our in vitro studies on a similar frameshift mutation reported in a German patient indicated increased apoptosis as an outcome, although patient cells were not included in our study (Burnett et al., 2019). Thus, it would be very informative if neurodegeneration is investigated in DY16 patients in the future in order to shed light on DYT16 pathogenesis.

This study further strengthens the case for a maladaptive ISR as possible disease etiology for DYT16 as our previous report with homozygous P222L patients was the first on dysregulated eIF2α signaling in any type of dystonia (Vaughn et al., 2015). Subsequently DYT1 (Beauvais et al., 2018, Rittiner et al., 2016), DYT6 (Zakirova et al., 2018) as well as DYT11 (Xiao et al., 2017) studies also suggested the maladaptive ISR pathway as a point of convergence for neuronal dysfunctions observed in dystonia. Two independent studies support the involvement of aberrant eIF2α signaling in brain to DYT1 synaptic defects. Using an unbiased proteomics approach abnormal eIF2α pathway activation in DYT1 mouse and rat brain was identified, which also correlated with human brain samples (Beauvais et al., 2018). Rittiner et al. used an RNAi-based functional genomic screening in HEK293T cells that also indicated dysregulated eIF2α pathway in DYT1. Moreover, in this study, pharmacological restoration of eIF2α signaling was reported to restore the cortico-striatal LTD in DYT1 knock-in mice (Rittiner et al., 2016). In addition, this report also examined patients with focal cervical dystonia and reported sequence variants in ATF4, which is a direct target of eIF2α signaling (Rittiner et al., 2016). RNA-Seq analysis to identify the effect of heterozygous DYT6 Thap1 mutations on the gene transcription signatures in neonatal mouse striatum and cerebellum identified eIF2α signaling as one of the top dysregulated pathways. The neuronal plasticity defects in DYT6 could also partially be corrected by salubrinal, a selective inhibitor of the eIF2α phosphatase which downregulates the ISR in a timely manner (Zakirova et al., 2018). A gene-expression analysis in adult cerebellar tissue from a mouse model of DYT11 also have identified genes associated with protein translation among the top down-regulated mRNAs (Xiao et al., 2017).

Stress-induced eIF2α phosphorylation by any of the four ISR kinases results in a suppression of general translation, but at the same time selectively stimulates the translation of some specific mRNAs (Pakos-Zebrucka et al., 2016). Typically, these mRNAs have long 5′-UTR with complicated secondary structure and one or more short upstream open reading frames (uORFs). Such mRNAs are preferentially translated when eIF2α is phosphorylated and initiation from other mRNAs is suppressed. Thus, eIF2α phosphorylation during cell stress not only achieves conservation of energy by a reduction of total translation but also allows new synthesis of a few proteins such as transcription factors ATF4 and CHOP whose translation is upregulated by eIF2α phosphorylation (Sano and Reed, 2013, Wek, 1994). These in turn induce the transcription of several genes either coding for ER enzymes and chaperones to cope with the accumulated unfolded proteins in the ER, or trigger apoptosis when homeostasis cannot be achieved due to intense or prolonged stress (Tabas and Ron, 2011). The dysregulation of ISR observed in DYT16 patient lymphoblasts although present in all cell types of the patients, it is likely to be especially detrimental to neuronal function. There is large amount of evidence indicating that in neurons, eIF2α phosphorylation driven translational changes are an essential feature of normal neuronal functions in the absence of stress and all four eIF2α kinases participate either individually, synergistically or even interchangeably in regulating neuronal activity (Chesnokova et al., 2017). The eIF2α phosphorylation dependent translation regulation allows the neurons to quickly change protein compositions at the synapse in a stimulus-dependent manner, and such regulation is known to be important for maintaining healthy neuronal functions. For example, ATF4, which presumably is the most important protein known to be regulated at translational level by eIF2α phosphorylation, is known to be associated with regulation of neuronal activity in the absence of stress (Chen et al., 2003). When PKR-mediated eIF2α phosphorylation was specifically increased in hippocampal CA1 pyramidal cells by a chemical inducer, ATF4 expression increased significantly (Jiang et al., 2010). Increased levels of ATF4 led to impairment of hippocampal long-term potentiation (LTP) and memory consolidation. Despite the well-established role of ATF4 as a suppressor of synaptic plasticity, it has to be understood that the changes in ATF4 concentrations are complicated and sometimes can be bidirectional. For example, the GCN2−/− mice have decreased eIF2α phosphorylation, and thus have decreased ATF4 in hippocampal neurons (Costa-Mattioli et al., 2005). These mice showed strong and sustained L-LTP and their spatial memory was improved compared to control wt mice. Thus, it may seem that low levels of ATF4 make neurons more sensitive to stimulation and their potentiation occurs too easily. It is certainly possible that neuronal activity-dependent shifts in ATF4 levels are important for LTP to take place normally. Any perturbation in such shifts, in either direction due to lower or higher ATF4 may be detrimental for normal neuronal functions. This becomes relevant to dystonia as both higher and lower ATF4 levels seem to be detrimental in different forms of dystonia. Rittiner et al observed reduced ATF4 induction in DYT1 cells and also identified presence of inactivating mutations in ATF4 in sporadic cervical dystonia patients (Rittiner et al., 2016). In our DYT16 lymphoblasts, we observe a sustained and higher level of ATF4 expression in response to ER stress (Fig 3). This was true both in P222L homozygous (data not shown) as well as P222L and C213R compound heterozygous DYT16 patients. Our studies are thus strongly indicative that a dysregulation of ATF4 expression occurs in DYT16 and this could derail normal healthy neuronal function.

PKR has emerged as a major player in several neurodegenerative diseases in recent years as aberrant elevated PKR activation has been observed in human patients in post-mortem studies as well as in several mouse models (Gal-Ben-Ari et al., 2018, Hugon et al., 2017, Marchal et al., 2014). Increased levels of PKR phosphorylation have been detected in the brains of patients with neurodegenerative diseases such as Alzheimer’s disease (AD) (Chang et al., 2002), Parkinson’s disease, Huntington’s disease (Peel and Bredesen, 2003, Peel et al., 2001), dementia (Taga et al., 2017), prion disease (Paquet et al., 2009). It is important to note that DYT16 patient lymphoblasts show higher levels of apoptosis in the absence of a stress signal in our analyses and it is unclear at present if this results from chronic low levels of PKR activation we noted in our studies. Activated PKR was recently shown to be responsible for the behavioral and neurophysiological abnormalities in a mouse model of Down syndrome and PKR inhibitory drugs partially rescued the synaptic plasticity and long-term memory deficits in mice (Zhu et al., 2019). Drugs that target the eIF2α signaling pathway have shown benefits in many mouse models for neurodegenerative diseases and in particular, inhibiting PKR has proven to be effective, showing rescue of synaptic and learning deficits in two different AD mouse models (Hwang et al., 2017). In case of DYT16 we wanted to take a more specific approach as C16, a widely used chemical inhibitor of PKR has been documented to have off target effects (Chen et al., 2008) thereby questioning its suitability in treating DYT16. We have previously reported that luteolin disrupts the PACT-PKR interaction efficiently and can inhibit stress-induced ISR and inflammation (Dabo et al., 2017). We tested if luteolin is able to rescue the DYT16 cells from stress induced apoptosis. Luteolin was able to dissociate PACT-PKR interactions efficiently in both normal as well as DYT16 lymphoblasts (Fig. 7). The observation that it takes significantly longer to disrupt PACT-PKR interactions in DYT16 patient cells as compared to normal wt cells, further supports that PKR interacts much stronger with P222L and C213R mutant PACT molecules. Luteolin treatment rescues the higher apoptosis phenotype in DYT16 cells and offers a promising lead into possible future therapies aimed at disrupting PACT-PKR interactions. In addition to DYT16, such therapies may also show promise in AD as PACT mediated PKR activation has been implicated in AD (Paquet et al., 2012).

The results presented here not only strengthen our previous research on DYT16, they also demonstrate the merit in developing drugs to disrupt PACT-PKR interactions for possible clinical application in future. Further research is essential on exploring the efficacy of luteolin on dystonic symptoms and the recently described Prkra mutant mouse model with dystonia symptoms would prove valuable and needs to be evaluated carefully (Palmer et al., 2015). Further efforts to discover compounds similar to luteolin that disrupt PACT-PKR interactions at lower concentrations or to develop specific peptides that disrupt PACT-PKR interaction may also be fruitful. In this regard, it is worth a mention that an interaction between dsRBM3 of PACT with PKR’s catalytic domain is essential for PKR activation (Li et al., 2006) and a disruption of such an interaction can be beneficial as it would block PKR’s catalytic activity even after PACT-PKR interactions have taken place. In combination with luteolin such peptides could offer valuable therapeutic options by lowering the effective luteolin dose significantly.

Methods and Materials:

Cell lines and antibodies:

Both HeLaM and COS-1 cells were cultured Dulbecco’s Modified Eagle’s Medium (DMEM) containing 10% Fetal Bovine Serum and penicillin/streptomycin. wt and DYT16 Patient B-Lymphoblasts were cultured in RPMI 1640 medium containing 10% FBS and penicillin/streptomycin. Both wt and DYT16 patient lymphoblast cell lines were Epstein-Barr Virus-transformed to create stable cell lines as previously described^{(Anderson and Gusella, 1984, Vaughn et al., 2015)}. All transfections were carried using Effectene transfection reagent (Qiagen) per manufacturer protocol. The antibodies used were as follows: PKR: anti-PKR (human) monoclonal (71/10, R&D Systems), P-PKR: anti-phospho-PKR (Thr-446) monoclonal (Abcam, [E120]), eIF2α: anti-eIF2α polyclonal (Invitrogen, AHO1182), p-eIF2α: anti-phospho-eIF2α (Ser-51) polyclonal (CST, #9721), PACT: Anti-PACT monoclonal (Abcam, ab75749), ATF4: Anti-ATF4 monoclonal (CST, #11815), CHOP: anti-CHOP monoclonal (CST, #2895), Cleaved PARP: anti-Cleaved-PARP monoclonal (CST, #32563), FLAG-HRP: anti-FLAG monoclonal M2-HRP (Sigma A8592), MYC-HRP: anti-MYC monoclonal (Santa Cruz, 9E10), β-Actin: Anti-β-Actin-Peroxidase monoclonal (Sigma-Aldrich, A3854).

Generation of DYT16 point mutations:

We generated each DYT16 mutant construct using site specific mutagenesis through PCR amplification changing the codon within the PRKRA gene to be consistent with DYT16 patients and code for the appropriate amino acid substitution. The following site-specific mutagenic primer pairs were used:

C77S Sense:

5’-GCT CTA GAC ATA TGG AAA TGT CCC AGA GCA GGC AC-3’

C77S Antisense:

5’-GCC TCT GCA GCT CTA TGT TTC GCC AGC TTC TTA CTT GTA CCT TCA CCT GTG GAG GTT ATG TCA CCA ACG G-3’

C213F Sense:

5’-GCT CTA GAC ATA TGG AAA TGT CCC AGA GCA GGC AC-3’

C213F Antisense:

5’-GGA GAA TTC CTC AAG GAA TGC CAA GTA AAT CCT AAA GAA TGT CC-3’

C213R Sense:

5’-GCT CTA GAC ATA TGG AAA TGT CCC AGA GCA GGC AC-3’

C213R Antisense:

5’-GGA GAA TTC CTC AAG GAA TGC CAA GTA CGT CCT AAA GAA TGT CC-3’

N102S Sense:

5’-GCT GCA GAG GCT GCC ATA AAC ATT TTG AAA GCC AGT GCA AGT ATT TGC TTT GC-3’

N102S Antisense:

5’-GGG GAT CCT TAC TTT CTT TCT GCT ATT ATC-3’

T34S Sense:

5’-GCT CTA GAC ATA TGG AAA TGT CCC AGA GCA GGC AC-3’

T34S Antisense:

5’-CGT GTA ATA CCT GAA TCG GTG ATT TCC CTG GCT TAG C-3’

To generate each construct, we performed PCR amplification in order to mutate the corresponding wild type sequence to code for the amino acid residue consistent with the DYT16 patients. Each PCR product was then sub-cloned into pGEMT-easy vector (Promega) and sequences were validated through DNA sequencing. After sequence validation, we generated full length DYT16 ORFs through cutting: (i) partial DYT16 ORF in pGEMT-easy with construct specific restriction enzymes, (ii) Amino terminal FLAG or Myc-tagged wt PACT in BSIIKS+ with compatible restriction sites. Cloning scheme was as follows: C77S in pGEMT-easy cut with NdeI-PstI ligated into FLAG/Myc-PACT-BSIIKS+ cut with PstI-BamHI. C213F and C213R in pGEMT-easy cut with NdeI-EcoRI ligated into FLAG/Myc-PACT-BSIIKS+ cut with EcoRI-BamHI. N102S in pGEMT-easy cut with NdeI-PstI and ligated into FLAG/Myc-PACT-BSIIKS+ cut with PstI-BamH1. T34S in pGEMT-easy cut with NdeI-TfiI ligated into FLAG/Myc-PACT-BSIIKS+ cut with TfiI-BamHI. Once full length DYT16 ORFs were generated with amino terminal FLAG or myc tags we then sub-cloned each ORF into pCDNA3.1-using XbaI-BamHI restriction sites. All DYT16 constructs were also cloned into Mammalian two-hybrid system vectors and pET15b (Novagen) using NdeI-BamHI restriction sites. TRBP and Flag-PKR constructs were as previously described (Singh et al., 2011).

Expression and purification of PACT from E. coli:

The ORFs of both wt PACT and all DYT16 point mutations were sub-cloned into pET15b (Novagen) to generate an in-frame fusion protein with a histidine tag. Recombinant proteins were then expressed and purified as previously described (Patel and Sen, 1998).

dsRNA binding assays:

Both wt PACT and DYT16 PACT constructs in pCDNA3.1-were in vitro translated using the TNT-T7-coupled rabbit reticulocyte system from Promega while incorporating an ³⁵S-Methionine radiolabel and the dsRNA binding ability was measured using poly(I:C) conjugated agarose beads. We diluted 4 μl of in vitro translation in 25 μl of binding buffer (20 mM Tris–HCl, pH 7.5, 0.3 M NaCl, 5 mM MgCl₂, 1 mM DTT, 0.1 mM PMSF, 0.5% NP-40, 10% glycerol) and incubated in 25 μl of poly(I:C)-agarose beads and incubated at 30°C for 30-minutes. We then washed the beads 4 times with 500 μl of binding buffer and bound proteins were analyzed via SDS-PAGE gel electrophoresis and autoradiography. The competition assay was performed incubating either soluble single-stranded RNA, poly(C), or dsRNA, poly(I:C), with the poly(I:C)-agarose beads before the adding the in vitro translated proteins. To ensure the presence of PACT was due to the dsRNA binding capacity we assayed in vitro translated ³⁵S-Methionine labeled firefly luciferase which has no dsRNA binding ability. Bands in bound and total lanes were quantified using Typhoon FLA7000 by analyzing relative band intensities of both T and B lanes. Percentage of PACT bound to beads was calculated and plotted as bar graphs.

PKR activity assays:

HeLa M cells treated with IFN-β for 24-hours and harvested at 70% confluency, washed using ice-cold PBS and centrifuged at 600 g for 5-minutes. Cell were resuspended in lysis buffer (20 mM Tris–HCl pH 7.5, 5 mM MgCl2, 50 mM KCl, 400 mM NaCl, 2 mM DTT, 1% Triton X-100, 100 U/ml aprotinin, 0.2 mM PMSF, 20% glycerol) and incubated on ice for 5 minutes. Lysates were centrifuged at 10,000 g for an additional 5-minutes. PKR was immunoprecipitated from 100 μg of this protein extract using anti-PKR monoclonal antibody (R&D Systems: MAB1980) in a high salt buffer (20 mM Tris–HCl pH 7.5, 50 mM KCl, 400 mM NaCl, 1 mM EDTA, 1 mM DTT, 100 U/ml aprotinin, 0.2 mM PMSF, 20% glycerol, 1% Triton X-100) at 4°C on a rotating wheel for 30-minutes. We then added 10 μL of protein A-Sepharose beads to each immunoprecipitate followed by an additional 1 hour incubation under the same conditions. Protein A-Sepharose beads were washed 4 times in high salt buffer followed by an additional two washes in activity buffer (20 mM Tris–HCl pH 7.5, 50 mM KCl, 2 mM MgCl2, 2mM MnCl2, 100 U/ml aprotinin, 0.1 mM PMSF, 5%, glycerol). PKR activity assay using PKR bound to protein A-Sepharose beads was conducted by incorporating: 0.1 mM ATP, 10 μCi of [γ-³²P] ATP, and increasing amounts of either pure recombinant wt PACT or DYT16 PACT (400 pG – 4 ng) as the PKR activator. Reaction was incubated at 30°C for 10 min and resolved on a 12% SDS-PAGE gel followed by phosphorimager analysis on Typhoon FLA7000.

Western blot analysis:

Lymphoblasts derived from a compound heterozygous DYT16 patient containing both P222L and C213R mutations as independent alleles were cultured alongside lymphoblasts derived from a family member containing no mutations in PACT as our control cells. Cells were plated at a concentration of 300,000 cells/ml of RPMI media containing 10% fetal bovine serum and penicillin/streptomycin. To analyze cellular response to ER stress, we treated cells with 5 μg/ml of tunicamycin (Santa Cruz) over a 24-hour time course and harvested cells in RIPA (150 mM NaCl, 1.0% IGEPAL^® CA-630, 0.5% sodium deoxycholate, 0.1% SDS, 50 mM Tris, pH 8.0) buffer containing a 1:100 dilution of protease inhibitor cocktail (Sigma) and phosphatase inhibitor (Sigma). Concentration of total protein extract was then determined using BCA assay and appropriate amounts of extracts were analyzed by western blot analyses using appropriate antibodies as indicated.

Co-Immunoprecipitation assays with endogenous proteins:

For Co-Immunoprecipitation (co-IP) of endogenous proteins DYT16 and wt lymphoblasts were seeded at a concentration of 300,000 cells/ml of RPMI complete media and treated with 50 μM of luteolin (Santa Cruz) over a 24 hour time course. Cells were harvested at indicated time points and whole cell extract was immunoprecipitated overnight at 4°C on a rotating wheel in IP buffer (20 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 20% Glycerol) using anti-PKR antibody (71/10, R&D Systems) and protein A sepharose beads (GE Healthcare). Immunoprecipitation was carried out using 100 ng of anti-PKR antibody and 10 μl of protein A sepharose beads slurry per immunoprecipitation. Immunoprecipitates were washed 3 times in 500 μl of IP buffer followed by resuspension and boiling for 5 minutes in 1X Laemmli buffer (150 mM Tris–HCl pH 6.8, 5% SDS, 5% β-mercaptoethanol, 20% glycerol). Samples were resolved on 10% SDS-PAGE denaturing gel and probed with anti-PACT antibody to determine co-IP efficiency and anti-PKR antibody to determine equal amounts of PKR were immunoprecipitated in each sample. Input blots of whole cell extract without immunoprecipitation are shown to indicate equal amounts of protein in each sample.

Co-Immunoprecipitation Assays in HeLa Cells:

In all cases HeLa M cells were seeded at 20% confluency in 6-well dishes 24-hours prior to co-transfecting 250 ng of each flag- and/or myc-tagged constructs using Effectene reagent (Qiagen). Cells were harvested 24-hours post transfection and harvested in IP buffer. Whole cell extract was then immunoprecipitated overnight at 4°C on a rotating wheel with either flag-agarose (Sigma) or myc-agarose beads (Thermo Scientific). Immunoprecipitates were then washed 3–5 times in IP buffer followed by resuspension and boiling for 5 minutes in 1X Laemmli buffer. Samples were then resolved on 10% SDS-PAGE denaturing gels and transferred to PVDF membranes. To evaluate PACT-PACT homodimerization and PACT-TRBP heterodimerization, flag-tagged constructs were immunoprecipitated using 15 μl of flag-agarose beads and blots were initially probed with anti-myc antibody to detect co-IP (PACT), followed by re-probing with anti-flag antibody to detect efficiency of IP (PACT or TRBP). PACT-PACT homodimerization co-IP blots were incubated at 50°C for 30 minutes in stripping buffer (62.5 mM Tris-HCl pH 6.8, 10% SDS, 0.75% β-mercaptoethanol) prior to re-probing with anti-flag antibody. To evaluate PACT-PKR interactions, we co-transfected myc-tagged PACT constructs in pCDNA3.1-with a flag-tagged dominant negative PKR mutant, K296R, also in the pCDNA3.1-. We immunoprecipitated the cell lysates in 15 μl of myc-agarose beads and resolved on 10% SDS-PAGE denaturing gels and transferred to PVDF membranes. Blots were initially probed anti-flag antibody to detect co-IP (PKR) followed by re-probing with anti-myc antibody to determine equal amount of IP (PACT) per sample. Input blots of whole cell lysate exempt from immunoprecipitation are shown to demonstrate equal expression of each construct prior to immunoprecipitation.

Mammalian 2-hybrid interaction assays:

In all cases, wt PACT, DYT16, TRBP, or PKR ORFs were sub-cloned into both pSG424 expression vector such that it created an in-frame fusion to a GAL4 DNA binding domain (GAL4-DBD), and pVP16AASV19N expression vector such that it maintains an in-frame fusion to the activation domain of the herpes simplex virus protein VP16 (VP16-AD). COS-1 cells were then transfected with: (i) 250 ng each of the GAL4-DBD and the VP16-AD constructs, (ii) 50 ng of pG5Luc a firefly luciferase reporter construct, and (iii) 1 ng of pRLNull plasmid (Promega), to normalize for transfection efficiencies. Cells were then harvested 24-hours post transfection and assayed for both firefly and renilla luciferase activities using Dual Luciferase^® Reporter Assay System (Promega). Fusion proteins were assayed for interaction in all combinations.

Caspase 3/7 activity assays:

Both wt and patient derived lymphoblasts were seeded at a concentration of 300,000 cells/ml of RPMI complete medium and treated with a concentration of 5 μg/mL of tunicamycin over a 24-hour time course. Samples were collected at indicated time points and mixed with equal parts Promega Caspase-Glo 3/7 reagent (Promega G8090) and incubated for 45 minutes. Luciferase activity was measured and compared to cell culture medium alone and untreated cells as the negative controls. To address the effect of inhibiting PACT-PKR interaction on cell viability, we cultured wt and patient lymphoblasts as described above in 50 μM of luteolin for 24 hours followed by treatment with 5 μg/ml of tunicamycin in luteolin free media over the same 24-hour time course.

Abbreviations:

PKR

Protein kinase RNA-activated

IFN

interferon

ds

double-stranded

eIF2

eukaryotic initiation factor

ER

endoplasmic reticulum

dsRBM

dsRNA-binding motif

TRBP

TAR RNA-binding protein

PACT

PKR activator protein, Prkra, Protein Activator of interferon-induced protein kinase

DYT

dystonia

ISR

integrated stress response

HIV

human immunodeficiency virus