Laforin is Required for the Functional Activation of Malin in Endoplasmic Reticulum Stress Resistance in Neuronal Cells

Abstract

Mutations in either EPM2A, the gene encoding a dual-specificity phosphatase named laforin, or NHLRC1, the gene encoding an E3 ubiquitin ligase named malin, cause Lafora disease (LD) in humans. LD is a fatal neurological disorder characterized by progressive myoclonus epilepsy, severe neurological deterioration, and accumulation of poorly branched glycogen inclusions, called Lafora bodies (LBs) or polyglucosan bodies (PGBs), within the cell cytoplasm. The molecular mechanism underlying the neuropathogenesis of LD remains unknown. Here we present data demonstrating that in the cells expressing low levels of laforin protein, overexpressed malin and its LD-causing missense mutants are stably polyubiquitinated. Malin and malin mutants form ubiquitin-positive aggregates in or around the nuclei of the cells in which they are expressed. Neither wild type (WT) malin nor its mutants elicit endoplasmic reticulum (ER) stress, although the mutants exaggerate the response to ER stress. Overexpressed laforin impairs the polyubiquitination of malin and recruits malin to PGBs. The recruitment and activities of laforin and malin are both required for the PGB disruption. Consistently, targeted deletion of laforin in brain cells from Epm2a knockout (KO) mice increases polyubiquitinated proteins. Knockdown of Epm2a or Nhlrc1 in neuronal Neuro2a cells shows that they cooperate to allow cells to resist ER stress and apoptosis. These results reveal that a functional laforin-malin complex plays a critical role in destroying LB and relieving ER stress, implying that a causative pathogenic mechanism underlies their deficiency in LD.

Introduction

Laforin, encoded by the epilepsy of progressive myoclonus type 2A gene (EPM2A), is highly expressed in adult brain [1, 2]. Loss-of-function mutations of EPM2A in humans cause Lafora disease (LD), an early-onset fatal epileptic neurodegenerative disorder marked by the accumulation of poorly branched glycogen inclusions called Lafora bodies (LBs) or polyglucosan bodies (PGBs), in brain neurons [3-6]. Although LBs are also distributed in glycogen metabolism-active tissues, such as liver and muscle, LD is not classified as a glycogen storage disease. Loss-of-function mutations of NHL repeat-containing 1 (NHLRC1), which encodes an E3 ubiquitin ligase named malin, can also cause LD [7]. Interestingly, although mutation of either gene can cause the development of LD, some LD patients do not harbor either mutation [8]. Mice with targeted deletion of either Epm2a or Nhlrc1 do not recapitulate the early-onset lethal neurological features of LD [9-11], which indicates that the mutations themselves should be considered in the quest to elucidate the molecular mechanism(s) of LD development.

Laforin has two functional domains: a carbohydrate-binding domain (CBD) and a dual-specificity phosphatase domain (DSPD) [3, 12-14]. The CBD is critical for the in vitro binding of laforin to glycogen [13], the in vivo binding of laforin to polyglucosan [12, 15], and the in vivo binding of laforin to itself [16]. Malin also has two functional domains: a RING finger E3 ubiquitin ligase domain and six repeats of NHL that are defined by (and named after) amino acid sequence homologies with NCL-1, HT2A and LIN41 proteins [17]. Three missense mutations of NHLRC1 in the RING domain have been reported to impair malin’s E3 ligase activity, while two missense mutations in the NHL repeats have been reported to impair the association of malin with laforin [18, 19].

Inter-relationship studies of laforin and malin have demonstrated that the combination of these proteins is responsible for reducing glycogen content in neuronal cells that ectopically express protein targeting to glycogen (PTG), a glycogenesis activator that induces protein phosphatase 1 to dephosphorylate glycogen synthase [20, 21]. The combination of laforin and malin has also been shown to degrade PTG [22, 23], misfolded proteins [24], and even laforin itself [18]; it can be stabilized and activated by AMP-activated protein kinase in hepatoma cells [25, 26]. Laforin has been shown to be an in vitro phosphatase of glycogen; it removes phosphates from phosphate-labeled amylopectin, isolated muscle glycogen and muscle glycogen synthesized by muscle glycogen synthase (GS1) [15, 27-29]. We and the Minassian laboratory showed that laforin dephosphorylates and inactivates GSK3β at serine 9 in serum-starved, growth factor-stimulated cells [19, 30]. However, under physiological conditions in Epm2a KO mice, increased serine 9 phosphorylation in GSK3β was not observed in the soluble portion of tissue lysate [28, 31]. This suggests that laforin dephosphorylates GSK3β at serine 9 in a context-dependent manner.

We have also shown that laforin reduces and its mutants exaggerate neuronal cell response to ER stress stimulation, and that laforin protects cells from apoptosis induced by energy deprivation stress [32, 33]. Consistently, increased ER stress in both Epm2a KO liver cells and an autopsy sample from LD patient has been revealed [34]. Combination of laforin with malin protects cells from thermal stress via activation of heat-shock factor 1, a transcriptional factor that activates heat shock genes [35]. These findings show that one function of laforin and malin is associated with resistance to stress.

Here, we present data demonstrating that laforin recruits malin to PGBs, where it activates functional malin for PGB disruption. Both the PGB-binding ability and the phosphatase activity of laforin are required for the recruitment and activation of malin to destroy PGBs. Likewise, both the laforin-binding ability and the E3 ligase activity of malin are required for the disruption of PGBs by the laforin-malin complex. The functional assembly of the laforin-malin complex alleviates ER stress and prevents the apoptosis of neuronal cells exposed to stress stimuli.

Results

Malin and its mutants aggregate and are polyubiquitinated but do not contribute to ER stress

To determine the relationship between malin and laforin, we generated malin mutations at the same sites found in LD patients. These included three sites near the RING finger E3 ubiquitin ligase domain and nine sites in the first five of the six NHL repeats (Fig. 1A). Similar to some mutants previously reported in transformed monkey kidney fibroblast COS7 cells [36, 37], we found that overexpressed WT malin and its mutants in mouse neuroblastoma Neuro2a (N2A) cells and human embryonic kidney HEK293 cells were expressed as monomers as well as polymerized aggregates (Fig. 1B). No statistically significant differences in protein stability were seen between WT malin and its mutants in these two types of cells, although transfection efficiency varied in separate experiments. Immunocytochemistry revealed that in HEK293 cells, Flag-tagged malin’s or its mutants’ aggregates mostly localized in the cytoplasm (Fig. 1C). Immunoprecipitation of the Flag-tagged malin or its mutants demonstrated that the aggregates were polyubiquitinated (Fig. 1D). RING domain mutants (C26S, C68Y, and L87P) and NHL mutants (D146N, D245N, and Q308A) were polyubiquitinated to a lesser extent than WT protein, as quantified by densitometry (Fig. 1D). Some aggregates observed in the ligase-inactive RING mutant cells were polyubiquitin-negative (Fig. 1E), indicating that the polyubiquitination of aggregates is likely attributable to malin autoubiquitination. This is consistent with the characteristics of the RING finger E3 ubiquitin ligase family [38] and with the result showing malin protein is autoubiquitinated in vitro[18]. In support of this, our results showed that the three RING finger mutants had much lower levels of ubiquitination than did WT or other malin mutants (Fig. 1D).

Fig. 1. Malin and its mutants aggregate and are polyubiquitinated.

(A) Diagram depicts the distribution of missense mutations identical to the sites in LD patients. (B) Aggregation of malin and its mutants. Malin and its mutants were transiently transfected into N2A cells for 24 hrs. The cell lysate was resolved by reducing heat-denaturing SDS-PAGE. (C) Aggregates of malin and its mutants are formed in or around the nuclei of transfected HEK293 cells. 24 hrs after transfection with malin or its mutants, HEK293 cells were fixed and stained with primary antibody to V5 followed by a secondary Texas Red-conjugated IgG. (D) Aggregates of malin and its mutants are polyubiquitinated. Immunoprecipitation (IP)-Western blot shows the polyubiquitinated aggregates of malin and its mutants in their Flag-tagged plasmid-transfected HEK293 cells 24 hrs before lysis for IP. The bottom digits of the gel show the densitometric quantitation of polyubiquitinated malin divided by the total malin as a relative ratio to WT malin that was arbitrarily set to 1. (E) Aggregates of malin and its representative mutants are polyubiquitin-positive. HEK293 cells transfected as in D were fixed and double-stained with antibodies to Flag and polyubiquitin. In all photos nuclei were stained with DAPI and merged photos are shown. Scale bars, 10μM.

To test the effect of malin and its mutants on ER stress, we transfected them into N2A cells and treated the transfected cells with either thapsigargin or tunicamycin, two molecules that induce ER stress by depleting ER calcium pump activity and dysglycosylating ER proteins, respectively [39, 40]. We then measured two common ER stress markers, ER chaperone 78-kDa glucose-regulated protein (GRP78) and the transcription factor proapoptotic C/EBP homologous protein (CHOP). Our results show that in the absence of thapsigargin or tunicamycin, N2A cells expressing WT malin or its mutants expressed little or no CHOP protein. However, in the presence of either stressor, significant levels of CHOP protein were induced (Fig. 2A, B). Induced CHOP levels were higher in most of the cells transfected with malin mutants than in cells transfected by WT malin. The inconsistent induction of GRP78 and CHOP in some mutants’ transfected cells after exposure to stressors indicate that the mutants elicited signs of both early and late ER stress at different stages. Usually, GRP78 is an early stage marker of ER stress, while CHOP is a late stage marker. N2A cells transfected with malin mutants showed a marked increase in cleaved activating transcriptional factor 6 (ATF6), an active form of ATF6 and another ER stress marker, regardless of whether the cells were stressed. Neither WT nor mutants of malin affected the levels of phosphor-eIF2α, an early-stage marker of ER stress, in transfected N2A cells exposed to stress stimulation (Fig. 2C). It is interesting to note that although overexpressed WT malin formed polyubiquitinated aggregates in the transfected N2A cells (Fig. 1C), cells transfected with malin showed a decreased ER stress response to stimuli in comparison to empty vector transfected cells (Fig. 2B), suggesting that WT malin plays a role in the prevention of ER stress. Taken together, these results showed that aggregated, polyubiquitinated WT malin alone does not induce ER stress. In contrast, the malin mutants themselves induce slight ER stress but exacerbate this response upon the addition of exogenous ER stress inducers.

Fig. 2. WT malin prevents and malin mutants exacerbate ER stress in N2A cells.

(A-C) Protein expression of ER stress markers. N2A cells were transiently transfected with V5-tagged malin or its mutants for 24 hrs and then treated with 2μM thapsigargin (A, C) or 2 μg/ml tunicamycin (B) or DMSO vehicle control. 8 hrs after treatment, the cells were directly lysed with 1% Triton X-100 lysis buffer. The ER stress marker proteins GRP78, CHOP, cleaved ATF6 (cATF6) and phosphor-eIF2α, were detected by Western blot. All blots were performed on the same membrane after stripping. Representatives of two separate experiments are shown.

Malin mutants have impaired binding to laforin

We next characterized the malin mutants’ ability to bind to laforin, and, although WT malin is known as a binding partner of laforin, the precise domains required for binding of malin to laforin remain undefined [18, 19]. We constructed plasmids that expressed different truncated versions of laforin and malin (Fig. 3A, B). Via analysis of immunoprecipitation data and cotransfection of these plasmids into HEK293 cells, we were able to define the reciprocal binding regions of laforin and malin. The region of laforin that binds malin was near the CBD end and included the entirety of exon 2; the region of malin that binds laforin began near the RING domain end and spanned the first five NHL repeats (Fig. 3A, B). The critical region of malin that binds laforin encompasses the end of RING domain and extends to the first NHL repeat. Due to the maintenance of glycogen-binding ability of the exon deleted mutants of laforin, cellular glycogen could cause a nonspecific binding between the deleted forms of laforin and malin (Fig. 3A). Consistent with data known about the region of malin that binds laforin, the RING domain mutants C26S, C68Y and L87P did not differ in their binding to laforin. In contrast, other NHL repeat mutants demonstrated impaired ability to bind to laforin, with the exception of the two mutants D245N and R253K (Fig. 3C). These results demonstrated that most NHL repeat mutants of malin were defective in their ability to bind laforin.

Fig. 3. Most NHL repeat mutants of malin impair their binding to laforin.

(A-B) The reciprocal binding regions of malin and laforin are defined. HEK293 cells were transiently cotransfected with protein-expressing plasmids as indicated. 24 hrs after transfection the cells were directly lysed by 1% Triton X-100 plus 0.02% SDS lysis buffer and supernatants were immunoprecipitated and then blotted with the indicated antibodies. The areas that are critical for binding between malin and laforin are shown with bold lines. (C) Impaired binding ability of malin mutants to laforin. HEK293 cells were transiently co-transfected with V5-tagged malin or malin mutants together with Flag-tagged laforin, and then lysed directly with Triton X-100 plus 0.02% SDS buffer. Resultant supernatants were immunoprecipitated with Flag antibody and the immunoprecipitates were subjected to Western blot. Densitometric quantitation of total malin divided by total malin input of the lysate is represented as a relative ratio to WT malin that was arbitrarily set to 1.

Laforin is required for malin recruitment to polyglucosan

Subsequent experiments were undertaken to reveal the target that laforin-malin complex might work on. Based on the finding demonstrating that transgenic mice expressing GS1 under the control of skeletal muscle-specific promoter display elevated polyglucosan levels in the tissue [41], we constructed and transfected GS1 into HEK293 cells and found that it synthesized polyglucosan that was glycogen positive, Periodic Acid Schiff stain (PAS) positive, and resistant to α-amylase hydrolysis, as previously reported [41]. After cotransfecting or singly transfecting GS1 with laforin, laforin mutant C265S, malin, or malin mutants L87P or D233A into HEK293 cells, we found that laforin and C265S, but none of the WT malin and its mutants, bound GS1-synthesized polyglucosan (Fig. 4A). In the triple combination indicated in Fig. 4A, only WT laforin and WT malin bound to and destroyed PGBs into relatively small granules. In the cells containing these small granules, GS1 was still distributed within the cytoplasm and nucleus. This indicates that if the laforin-malin complex degrades GS1 to limit polyglucosan formation, this process takes place in polyglucosan bodies only (Fig. 4A). Neither phosphatase-dead mutant C265S, E3 ligase-inactive mutant L87P, nor laforin binding-deficient mutant D233A in combination with WT malin or WT laforin was able to destroy the PGBs, demonstrating that the disruption of PGBs requires the activities of both laforin phosphatase and malin E3 ligase, as well as the appropriate recruitment of the laforin-malin complex. To prove the requirement of laforin for malin in binding to PGBs, we combined purified laforin protein and malin immunoprecipitate with the isolated GS1 polyglucosan in vitro and found that the malin beads pulled down the GS1 polyglucosan only in the presence of laforin (Fig. 4B, lane 2 compared to lane 4). Also, by determining glycogen content in HEK293 cells that possessed endogenous GS1 glycogen, we found that only cells transfected with both malin and laforin had reduced glycogen content (Fig. 4C), a finding consistent with results observed in neuronal cells that ectopically express PTG [20]. These results demonstrate that GS1 polyglucosan is a target of the laforin-malin complex, and that polyglucosan disruption requires the bindings and activities of laforin and malin. This conclusion is supported by previous data showing that it is GS1, not PTG, that accumulates in LBs [42].

Fig. 4. Laforin recruits malin and both destroy polyglucosan.

(A) Laforin is required for malin recruitment to GS1-synthesized polyglucosan. HEK293 cells were transfected with a plasmid of GS1-Flag alone or in combination with indicated plasmids of laforin-myc and malin-V5 or their mutants for 24 hrs. The transfected cells were fixed and double-stained with antibodies to glycogen or Flag, V5, or Myc tags. (B) Laforin is essential for malin binding to polyglucosan in vitro. Isolated GS1 polyglucosan was added to binding buffer containing malin-protein G beads or empty beads in the presence or absence of purified laforin protein. After 2 hrs of binding at 4°C, the washed beads were lysed with 1X SDS loading buffer for Western blot to detect GS1 polyglucosan. (C) Laforin and malin together decrease glycogen content in vivo. HEK293 cells transfected with the indicated plasmids were directly lysed in NaAc buffer for glycogen determination. Results are presented as mean ± SEM of three separate experiments. Scale bars, 10μM.

Laforin and malin functionally depend on each other to prevent ER stress

Laforin recruits malin to PGBs, which both proteins destroy together. Disruption of PGBs either provides glucose (energy) to cells to counteract ER stress stimulation (which results in cellular energy decline), or releases laforin and malin for recycling. Based on this hypothesis, knockdown of either laforin or malin in N2A cells could enhance cell sensitivity to ER stress stimulation. As expected, knockdown of either in N2A cells increased thapsigargin-induced CHOP protein levels (Fig. 5A, B). Laforin restoration did not diminish the increased sensitivity of malin-silenced N2A cells to CHOP induction by thapsigargin; malin restoration did not diminish the increased sensitivity of laforin-silenced N2A cells to CHOP induction, even though they increased the resistance of cells transfected with scrambled shRNA to ER stress (Fig. 5B). To determine whether both phosphatase and E3 ligase activities were required for the prevention of ER stress-induced apoptosis, we transfected malin or its mutants into N2A cells expressing either laforin or C265S, and observed that only a combination of WT laforin and WT malin significantly prevented an ER stress-induced increase of Annexin V-positive (apoptotic) cells (12.68% in WT laforin-WT malin versus 23.23% in WT laforin-mutant L87P). Other combinations could not prevent apoptosis (Fig. 5C). These results show that a functional laforin-malin complex is necessary for cellular resistance to ER stress.

Fig. 5. Laforin and malin cooperate in ER stress relief and apoptosis prevention.

(A) Knockdown of laforin or malin in N2A cells. Silencer of small hairpin RNA of laforin (sh-L) or malin (sh-M) was cotransfected with Flag-tagged laforin or malin into N2A cells for 24 hrs. Knockdown efficiency was determined by Western blot using anti-Flag antibody. (B) Codependence of laforin and malin in the prevention of ER stress. Scrambled sh (Sr), sh-L or sh-M cells of N2A were transiently transfected with vector, malin or laforin for 24 hrs and then treated with 2 μM thapsigargin or vehicle for an additional 8 hrs. After treatment the cells were lysed for Western blot and probed for ER stress proteins. (C) Activities of both laforin and malin are required for the inhibition of apoptosis induced by ER stressors. N2A cells expressing WT laforin or its mutant C265S were transiently transfected with malin-EGFP or malin mutant L87P-EGFP. 24 hrs after transfection, the cells were divided into two test groups and treated with 1 μM thapsigargin or vehicle for 24 hrs in 2.5% FBS DMEM medium. Treated cells were stained with Annexin V (for apoptotic cells) and DAPI (for dead cells). The % Annexin V of EGFP-positive gate was analyzed by flow cytometry.

Laforin and ER stressor impair malin autoubiquitination

The question of how autoubiquitinated malin becomes functionally activated once it has bound laforin remained. To determine the mechanism whereby this occurs, we co-transfected laforin along with malin or one of its mutants into HEK293 cells, and found that in malin immunoprecipitate where laforin existed, malin became less ubiquitinated, while other mutants of malin did not (despite the presence or absence of laforin in their immunoprecipitates) (Fig. 6A). To ascertain if an ER stressor activates malin by impairing its autoubiquitination, we treated the malin-transfected N2A cells with tunicamycin or thapsigargin. Autoubiquitination of malin was significantly induced by both stressors (Fig. 6B). Detection of endogenous malin activation by laforin or ER stress stimulation cannot be performed, because at this time, no convincing, commercially-available malin antibodies exist. However, via immunoprecipitation using an anti-polyubiquitin antibody [43], we were able to detect significantly more polyubiquitinated proteins in the brain cells of Epm2a KO mice than in age-matched WT mice (Fig. 6C, D). This result indicates that laforin plays a role in the prevention of the accumulation of polyubiquitinated proteins, which is consistent with the result showing that increased polyubiquitinated proteins were detected in the lysate of laforin-deficient human fibroblasts by Western-blot [44]. Taken together, these data demonstrated that laforin is required not only for the binding of malin to PGBs, but also for the activation of malin by prevention of its autoubiquitination.

Fig. 6. Laforin and ER stressors promote malin deubiquitination.

(A) Laforin favors the deubiquitination of malin. HEK293 cells cotransfected with malin or its mutants together with vector or laforin were lysed 24 hrs after transfection to detect polyubiquitination of malin by IP-Western blot using tag antibodies. (B) ER stress stimulation activates malin deubiquitination. N2A cells expressing malin-Flag were treated with 2 μg/ml tunicamycin (Tu) or 1 μg/ml thapsigargin (Th). 8 hrs after treatment, the cells were lysed for anti-Flag IP and anti-polyubiquitin blotting to detect malin ubiquitination. (C) Polyubiquitinated proteins accumulated in brain cells from Epm2a KO mice. Equal amounts of Triton X-100-soluble proteins in the brain lysate of 6-month-old Epm2a KO and WT mice were immunoprecipitated with polyubiquitin Lys48-linkage antibody. Ubiquitinated proteins were detected by mouse anti-ubiquitin and rabbit anti-Lys48-linkage polyubiquitin antibodies. (D) Graphical representation of polyubiquitinated protein levels in the two strains.

Discussion

Through a systematic analysis of malin, laforin, and their missense mutants, and by using knockdown approaches, we demonstrate here, for the first time, the co-operation of laforin and malin that confers neuronal protection against ER stress. Although the exact mechanism underlying the prevention of ER stress by laforin-malin complexes remains to be defined, we hypothesize that polyglucosan disruption governed by laforin-malin complexes provides not only endogenous energy in the form of glucose, but also lowers the level of stress within the cytoplasm of stressed cells, easing the ER burden. We are the first to demonstrate that the process of polyglucosan disruption absolutely requires both laforin and malin. However, alone these proteins are not sufficient. Thus, for destroying PGBs, a consecutive process may be required. First, laforin recruits malin and other proteins or enzymes to PGBs, where the functional assembly of a laforin-malin complex destroys PGBs into relatively small, “normal” glycogen granules that can be degraded by conventional glycogen metabolic enzymes. GS1 is the key enzyme involved in PGB formation, and may be the first enzyme targeted by the functional complex of laforin and malin. The binding and recruitment of laforin and malin to GS1 polyglucosan suggests that the laforin-malin complex degrades GS1 and thus inhibits GS1 to synthesize polyglucosan, which process takes place in an insoluble glycogen pool. Degradation of GS1 by the complex has been hinted at by a decrease in GS1 protein levels in the lysate of N2A cells expressing laforin, malin and PTG [20], and in the accumulation of GS1 protein in LBs from Epm2a or Nhlrc1 KO mice [10, 42, 45, 46]. Disruption of PGBs by the laforin-malin complex may subsequently prevent laforin and malin from becoming trapped in PGBs. The requirement of both laforin phosphatase activity and malin E3 ligase activity for PGB disruption suggests that dephosphorylation and ubiquitination of key components in PGBs may take place simultaneously. These processes occurring in the insoluble glycogen pool may create a situation in which detection of target protein alteration becomes difficult [45].

Stresses that result in intracellular energy decline, such as energy deprivation and ER stress, may induce PGB formation; subsequently, the activation of laforin-malin complexes may destroy PGBs, thereby supplying energy for cell recovery from the stresses. Further, under homeostatic conditions, the laforin-malin complex plays a critical role in the surveillance and prevention of PGB formation; thus, deletion of either gene causes PGB accumulation [9, 45-47]. The laforin-malin complex also prevents the formation of and destroys PTG-activated GS1-synthesized abnormal, but not normal, glycogen [20, 22, 23], because laforin preferentially binds to polyglucosan over normal glycogen [15]. We also predict that polyubiquitination of malin under physiological conditions may be a form of self-inactivation to control its activity when it is not needed. Because laforin is not a deubiquitinating enzyme, an unknown deubiquitinating enzyme may be recruited to PGBs or perhaps even earlier in the malin-laforin pathway to counteract the cellular effects of ER stress. In addition, increased protein levels of ER stress markers were not observed in brain extracts from 9-month-old Epm2a KO mice, probably because it is difficult to extract brain proteins from tissues with massive PGB accumulation [34].

Since PGBs are found in other neuronal disorders, including Alzheimer’s disease [48] and temporal lobe epilepsy [49], decreased functional laforin and/or malin could also contribute to the progression of these diseases. Therefore, our study has clinical implications across a broad range of neurological disorders.

Experimental procedures

Mice and Cells

The Epm2a KO mice (from a 129Sv strain [9]) used in this study have been backcrossed onto a C57BL/6 background for more than 10 generations. Experiments were performed using WT and Epm2a KO mice that were littermates born from homozygous breeding pairs. All the mice were kept and used according to the procedures approved by the Unit for Laboratory Animal Medicine (ULAM) at the University of Michigan.

Human embryonic kidney 293 FT (HEK293) and mouse neuroblastoma (Neuro2a, N2A) cell lines were from Invitrogen and ATCC, respectively. The HEK293 cells were cultured in DMEM medium supplemented with 4.5g glucose, 2 mM glutamine, 2% penicillin, and 10% fetal bovine serum (FBS). The N2A cells were cultured in MEM medium supplemented with 2 mM glutamine, 2% penicillin, and 10% FBS.

Reagents and Antibodies

Sources of rabbit polyclonal antibodies to specific proteins are as follows: Flag (Sigma), phosphor-Ser52-eIF2α (Cell Signaling), and polyubiquitin Lys48 linkage for immunoprecipitation (clone Apu2, Millipore). The sources of mouse monoclonal antibodies to specific proteins are as follows: ubiquitin for Western blot (P4D1, Santa Cruz), polyubiquitin for immunocytochemistry (Ubi-1, Thermo Scientific), ATF6 (IMGENEX), V5 and Myc tags (Invitrogen), Flag tag (M2, Sigma), Flag-Cy3 (Sigma), GRP78/Bip (BD Transduction Lab), S6K (H-9, Santa Cruz), β-actin (Sigma), and CHOP/GADD153 (MA1-250, Thermo Scientific). The flow cytometry antibody to APC-Annexin V used in the present study was from BD Pharmingen. The monoclonal anti-glycogen IgM antibody was from Dr. Otto Baba. Thapsigargin and tunicamycin, enzymes and standard for amyloglucosidase (A7420), amylase (A6814) and glycogen (G0885), and glucose (GO) assay kit were all purchased from Sigma.

Plasmids and RT-PCR

The coding regions of cDNAs of human malin and mouse GS1 (Gys1) were amplified from reverse transcription mRNA of human cord blood cells and C57BL/6J bone marrow cells, respectively, and cloned into a pcDNA vector with Myc, V5, or Flag tag at the N-terminus (malin) or C-terminus (GS1). After sequencing confirmation of the WT malin, all point mutants used here were made by site-directed mutagenesis using WT malin as a template. Truncated forms of malin were generated by PCR using full-length malin as template. Plasmids of laforin and its mutants have been described previously [16]. Gene-silenced sequences were: Epm2a, 5-ttccagactgaatgggata-3, and Nhlrc1, 5-attctcttcttgtgctgga-3. These were cloned into small hairpin (sh) RNA lentiviral vectors with EGFP as a reporter. The lenti-sh vector was made by substituting the CMV prompter of plenti6-TOPO/V5 (Invitrogen) with the U6-siRNA-PGK-EGFP cassette. Malin and its mutants were cloned into a lenti-EGFP vector that was made by substituting the T7-blasticidin cassette of plenti6-TOPO/V5 with the PGK-EGFP cassette.

Transfection, Western blot and Immunoprecipitation

Generally, HEK293 and N2A cells were transiently transfected with a total of 0.25 μg plasmid and 0.75 μl of Lipofectamine 2000 (Invitrogen) per well on a 24-well plate for 24 hrs in 0.5 ml Opti-MEM medium containing 10% FBS. To prevent cells from detaching when the culture medium was changed to 0.5 ml Opti-MEM medium, one third of the original culture medium was not removed. Overnight-passaged cells grown to ~75% confluence were used for transfection. In double or triple transfections, the total DNA plasmids did not exceed 0.375 μg (double) or 0.5 μg (triple) per well in 24-well plates. Transfected cells were lysed with 1% Triton X-100 lysis buffer containing 20mM Tris-HCl, pH7.4, 150mM NaCl, 40mM NaF, 1mM DTT, and a protease and phosphatase inhibitor cocktail (Sigma). Supernatants of the lysate were used for Western blots and resolved on a reducing and heat-denaturing 10% SDS-PAGE gel. Immunoprecipitation was carried out with protein G beads at 4°C, overnight, with rotation. The supernatant was pre-incubated with protein G beads for 2 hrs, and cleared supernatant was used for immunoprecipitation. After washing 3 times with 1% Triton X-100 lysis buffer, the immunoprecipitates were dissolved in SDS loading buffer and resolved on a 10% SDS-PAGE gel.

Polyglucosan body (PGB) isolation and malin ubiquitination

To isolate PGBs from transfected cells, 0.55% NP-40 in HEPES buffer containing 10 mM HEPES, 1mM EDTA and 1mM EGTA was used for lysing cells. Nuclei were removed by centrifugation at 1000g for 5 min and washed twice with HEPES buffer. The combined supernatants were centrifuged at 8000g for 15 min to remove debris and were then centrifuged again at 18,000g for 45-60 min to get the pellets containing insoluble PGBs. For detecting malin ubiquitination and the binding of malin to laforin, the pellets were digested with 1U/ml amyloglucosidase and 5U/ml amylase in PBS, pH7.4, at 37°C for 2 hrs and then dissolved by 5X 1% Triton X-100 plus 0.02% SDS buffer. The pellet-digesting solution in combination with cytosolic supernatant was centrifuged and resultant supernatant was used for immunoprecipitation in a final solution containing 1% Triton X-100, 0.5% NP-40 and 0.02% SDS.

Immunofluorescence staining

Cells were fixed by cold methanol for 10 min and then permeabilized with 0.3% Triton X-100 in 10mM Tris-HCl buffer for 30 min. Immunofluorescence staining with the primary antibody was performed overnight in 10 mM Tris-HCl buffer containing 2% bovine serum albumin (BSA) at 4°C. Secondary antibody staining was carried out in 2% BSA Tris-HCl buffer at room temperature for 2 hrs.

Abbreviations

CBD

carbohydrate-binding domain

DSPD

dual specificity phosphatase domain

EPM2A

epilepsy of progressive myoclonus type 2A

ER

endoplasmic reticulum

GS1

glycogen synthase 1

HEK293

human embryonic kidney 293

LB

Lafora body

LD

Lafora disease

PGB

polyglucosan body

PME

progressive myoclonic epilepsy

PTG

protein targeting to glycogen