A splice variant of cyclin D2 regulates cardiomyocyte cell cycle through a novel protein aggregation pathway

Cell division is an essential component of all stages of heart development. A high level of cell cycle activity is seen in the precardiac mesoderm as well as during early stages of heart formation (Erozhina, 1968; Rumyantsev, 1991). In later stages of heart development, the onset of cardiomyocyte differentiation is strictly accompanied by a rapid reduction in cardiomyocyte cell cycle activity (Li et al., 1996; Soonpaa and Field, 1998). As a result, the mammalian heart lacks intrinsic ability to replace ischemic myocardium with newly divided myocytes during myocardial infarction (MI) (Pasumarthi and Field, 2002). The D-type cyclins play an important role in the activation of cyclin-dependent kinases (CDKs) and thereby regulate the transition of cell cycle from G1 to S phase (Sherr, 1994). We have shown that targeted expression of cyclin D2 (cycD2; also known as CCND2) can induce cardiomyocyte cell cycle activity in the infarct border zone, partially reduce infarct size (Pasumarthi et al., 2005), and improve myocardial function (Hassink et al., 2008) post MI. A deeper understanding of mechanisms involved in the regulation of cardiomyocyte proliferation is required to further increase the magnitude of cardiac regeneration. However, progress in this area is further limited because of the scant information available on mechanisms regulating cardiomyocyte cell cycle.

A highly conserved splice variant of cyclin D2 (D2SV; originally designated `truncated cycD2') has been reported to exist in transformed cells as well as normal tissues such as brain and ovaries (Denicourt et al., 2003). Recent studies showed that D2SV can induce cellular transformation of mouse embryonic fibroblasts (MEF) when overexpressed in concert with activated H-Ras oncogene (Denicourt et al., 2008). However, D2SV alone was unable to induce cellular transformation (Denicourt et al., 2008). From these studies, the precise role played by this newly identified protein in cell cycle regulation is not clear. Furthermore, there is no information available on the expression pattern of D2SV during cardiac development. Given the role of cycD2 in cardiac regeneration (Pasumarthi et al., 2005), we sought to examine the role of D2SV in cardiac ontogeny and cardiomyocyte cell cycle regulation.

In the present study, we cloned the cDNA encoding D2SV from embryonic mouse heart and found that this protein is highly expressed in the primitive myocardium compared with the adult heart. We observed an inverse relationship between endogenous D2SV expression and cardiomyocyte cell cycle activity. The ability of D2SV to bind with other cell cycle proteins suggested a negative role for this protein in cell cycle regulation. Subcellular localization studies indicated that D2SV is subjected to protein aggregation and predominantly retained in the endoplasmic reticulum (ER), Golgi and lysosomal compartments. The process of D2SV aggregation appears to be sensitive to agents regulating cytoskeleton and ER stress. We found that enforced expression of D2SV but not cycD2 in embryonic cardiomyocytes leads to cell cycle exit through a unique protein aggregation pathway coupled with sequestration of several endogenous cell cycle proteins. Although coexpression of cycD2 and CDK4 significantly increased cardiomyocyte cell cycle activity, coexpression of D2SV and CDK4 did not have any significant effect. Furthermore, we showed that cell cycle exit induced by D2SV can be rescued by overexpression of D-type and B-type cyclins.

We amplified the coding sequence of D2SV from embryonic mouse heart RNA by RT-PCR and subcloned it into a pCR2.1 vector. The cDNA sequence of the cardiac D2SV was identical to that of the splice variant identified in Graffi murine leukemia virus (GMLV)-induced tumours (Denicourt et al., 2003). The cardiac D2SV cDNA contains the sequence of cyclin D2 exon 1 followed by a longer exon 2 resulting from the alternative splicing. The cDNA encodes a 156 amino acid protein, which includes the first 136 amino acids from the N-terminus of cyclin D2 and 20 novel residues at the C-terminus (Fig. 1A). It contains a stop codon in the 3′ end of the longer exon 2 region. Based on the known functional domains of cyclin D1 (Coqueret, 2002; Zwicker et al., 1999) (Fig. 1B), the primary structure of D2SV contains a binding site for members of the retinoblastoma family as well as partial binding sites for CDK4 and p21. Furthermore, D2SV shares a high degree of sequence homology (74-92%) with the N-terminus of members of the cyclin D family, but it lacks the majority of the C-terminus sequence found in cycD1, cycD2 or cycD3 (Fig. 1B).

To determine the molecular mass of the D2SV protein, we transfected a pcDNA3.1-D2SV construct in HEK293 cells and analyzed the protein lysate using polyclonal antibodies specific for the C-terminus (CT; Fig. 1C). In agreement with the predicted size of D2SV protein, we found a major 20 kDa band under denaturing conditions. In addition, we also detected several prominent immunoreactive bands with apparent molecular masses ranging from 32-45 kDa, using CT-specific antibodies. The specificity of D2SV antibodies was confirmed by preincubation of antibodies with CT-region peptide and also by the absence of immunostaining in cells transfected with the cycD2Myc construct (supplementary material Fig. S1). Presence of these higher molecular mass species could be the result of dimerization, stable intra- or intermolecular disulfide bond formation or also because this protein has been subjected to ubiquitylation (K.W., Q.S. and K.B.S.P., unpublished data).

We assessed expression profiles of D2SV in histological sections using immunofluorescence analysis. In E11.5 and E14.5 mouse embryos, D2SV protein was predominantly localized in the ventricular myocardium compared with other tissues (Fig. 2A-D). We found that immunoreactive D2SV protein was much higher and also more evenly distributed between compact and trabecular zones in cardiac sections of E11.5 than E14.5 embryos. Interestingly, D2SV staining was predominantly in the trabecular zone and less in the compact zone in some sections of E14.5 stage (Fig. 2D). Further analyses of these sections using confocal microscopy indicated that the D2SV protein localizes as micro-aggregates in the cytoplasmic compartment of the majority of myocardial cells (Fig. 2E). In the adult mouse heart, we observed a similar D2SV micro-aggregate staining pattern in less than 1% of ventricular myocardial cells (Fig. 2F-H).

Mitotic activity in D2SV-expressing and nonexpressing cells was assessed in E14.5 cardiomyocyte cultures by immunostaining with phosphohistone H3 antibodies. Using this assay, none of the cardiomyocytes expressing D2SV showed any mitotic activity compared with nonexpressing cardiomyocytes which showed positive staining in ∼2% of the cells counted (Fig. 3A). Subsequently, the ability of D2SV-expressing cells to transit through G1-S phase and synthesize DNA was monitored in vitro. Dispersed cell preparations from E14.5 hearts were incubated with [³H]thymidine for 6 hours and processed for anti-D2SV immunostaining and in situ thymidine autoradiography (Fig. 3B-D). The [³H]thymidine labeling index (LI) is the proportion of the total number of D2SV-positive or -negative myocardial cells counted that have nuclear silver grains (Fig. 3E). D2SV-expressing cardiomyocytes had a 3% LI, and nonexpressing cells a 12% LI (Fig. 3E). This result indicates a considerable reduction (75%) in the ability of D2SV-expressing cells to enter the cell cycle. The inverse relationship between D2SV expression and cardiomyocyte cell cycle activity combined with the ability of this protein to form micro-aggregated masses in vivo suggests that D2SV is capable of abrogating the activities of positive cell cycle regulators.

The appearance of D2SV micro-aggregates in tissue sections may be explained by the intrinsic ability of this protein to form aggregated masses. To verify the intrinsic ability of this protein to aggregate, lysates obtained from HEK293 cells transfected with D2SV or a control vector were analyzed in basic native gels. Samples were run on basic gels because of the overall acidic nature of D2SV protein (pI=5.3). Our results revealed that non-denatured D2SV protein had a significantly reduced mobility, producing an immunoreactive smear with apparent molecular mass ranging from 60 to 150 kDa (Fig. 3F). Furthermore, the majority of high molecular D2SV protein aggregates present in non-denatured samples could be readily separated into a monomeric form with the exception of few stable higher mobility complexes when subjected to denaturing gel electrophoresis (Fig. 3G).

We transfected E14.5 cardiomyocyte cultures with D2SVMyc or cycD2Myc plasmids and processed them for protein expression using Myc immunostaining (Fig. 4). Approximately 90% of myocytes transfected with the D2SVMyc plasmid revealed a distinct micro-aggregated staining pattern predominantly in the cytoplasmic compartment (Fig. 4A-F,J,K). In addition, confocal imaging of transfected cells co-stained for D2SV and ConA (a membrane-specific probe) revealed a close association of D2SV protein with the cell membrane (Fig. 4J). However, we did not find any detectable levels of D2SV protein in the conditioned medium of either myocardial cells or transfected HEK 293 cells (data not shown). In contrast to D2SV-transfected cells, myocytes expressing cycD2Myc revealed strong and diffuse Myc staining in both nuclear and cytoplasmic compartments (Fig. 4G-I). Although ∼15% of cells transfected with cycD2Myc also displayed a punctate staining pattern resembling that of the protein aggregates, it was not comparable to the magnitude of protein aggregation seen consistently with D2SV expression (compare Fig. 4G and 4A). We confirmed that cardiomyocytes overexpressing D2SV do not undergo apoptosis as determined by the absence of TUNEL staining (data not shown).

To directly examine the role of D2SV in cell cycle control, myocyte cultures were transfected with D2SVMyc, cycD2Myc or control vectors, maintained for 24 hours and pulsed with [³H]thymidine for an additional 6 hours. Subsequently cells were processed for anti-Myc immunostaining and in situ thymidine autoradiography. The labeling index (LI) is the proportion of the total number of Myc-positive myocardial cells counted that have nuclear silver grains (Fig. 5A-C). We also monitored the LI of non-transfected cells using Hoechst 33342 nuclear staining and [³H]thymidine in each transfected cultures. Compared with the LI of cultures transfected with the vector alone or that of non-transfected cells, overexpression of cycD2Myc did not have any significant effect on embryonic cardiomyocyte cell cycle activity. By contrast, overexpression of D2SVMyc led to a reduction (90%) in cardiomyocyte cell cycle activity compared with the vector control or cycD2Myc groups (Fig. 5D). Addition of a SV40-derived nuclear localization signal (NLS) to the D2SVMyc sequence did not have any impact on the cell cycle exit mediated by this protein in cardiomyocytes (Fig. 5D).

We next co-transfected E14.5 mouse ventricular myocyte cultures with expression vectors harboring CDK4 and D2SVMyc or cycD2Myc cDNA and assessed [³H]thymidine uptake. Compared with the LI of myocytes transfected with the D2SVMyc alone, co-expression of D2SVMyc and CDK4 did not have any significant effect on cardiomyocyte cell cycle activity (Fig. 6G). Furthermore, both CDK4 and D2SVMyc proteins colocalized in the cytoplasmic compartment and formed protein aggregates in the majority of co-transfected cells (Fig. 6A-C). By contrast, coexpression of cycD2Myc and CDK4 led to the colocalization of both proteins in the nucleus (Fig. 6D-F) and significantly increased cardiomyocyte cell cycle activity compared with the cycD2Myc- or vector-transfected groups (∼2.6-fold; Fig. 6G).

Intracellular protein aggregation is considered as a typical hallmark of problems associated with protein folding in the ER or cytoplasm (Bernales et al., 2006; Garcia-Mata et al., 2002; Kopito, 2000). To more precisely define the distribution of D2SV aggregates, we performed double-labeling experiments for D2SV and markers specific for an ER chaperone (protein disulfide isomerase), Golgi (anti-giantin) and lysosomes (Lysotracker red) on cultures transfected with epitope-tagged D2SV constructs. In these experiments, we observed colocalization of D2SV protein predominantly with the ER and lysosomal compartments (∼30% of the cells) and to a lesser extent with the Golgi (∼5% of the cells) (Fig. 7A-I). To further verify whether accumulation of D2SV aggregates is due to ER stress and impairment of ER-associated degradation (ERAD), we employed a well characterized ERAD reporter gene, the TCRα-GFP fusion gene (DeLaBarre et al., 2006). Using this system, impairment of ERAD can be readily assessed by the accumulation of GFP fluorescence. Cardiomyocytes transfected with TCRα-GFP construct alone exhibited a baseline GFP fluorescence 30 hours after transfection (Fig. 7J-L), but cardiomyocytes co-transfected with D2SV and TCRα-GFP consistently displayed a much higher level of GFP fluorescence (Fig. 7J-L). These results indicate that D2SV-expressing cells are subjected to ER stress and impaired ERAD.

Furthermore we found that addition of a SV40 NLS to the D2SV sequence did not enhance nuclear localization of this protein (data not shown). Analysis of the D2SV primary sequence using a Faucher and Pliska hydrophobicity plot indicated that the novel 20-amino acid C-terminus harbors the highest degree of hydrophobicity of the whole sequence (Fig. 8A). To verify the role of this motif in intracellular targeting and protein aggregation, we generated a mutant construct (D2SVΔCT) encoding the first 136 amino acids of D2SV but lacking the novel 20 amino acids, and transfected this construct into embryonic cardiomyocytes (Fig. 8D,E). Similar to D2SV, D2SVΔCT also formed protein aggregates, but these were found predominantly in the nucleus (Fig. 8B-E). Nuclear aggregation was found in approximately 75% of cells transfected with the D2SVΔCT construct (Fig. 8F,G), in contrast to only 15% of the cells transfected with D2SV. D2SVΔCT overexpression also caused cell cycle arrest in cardiomyocytes, similar to D2SV (data not shown).

Enforced expression of dynamitin, one of the dynactin subunits, was shown to prevent dynein-mediated mobilization of aggregation-prone proteins into structures called aggresomes (Garcia-Mata et al., 1999; Ramanathan et al., 2007). To test whether D2SV aggregation can be blocked, we co-transfected cardiomyocytes with D2SV and dynamitin. In cells expressing low levels of dynamitin (Fig. 9A), D2SV continued to form aggregates (Fig. 9B). However, in cells displaying a high level of dynamitin expression (Fig. 9D), D2SV aggregation was not observed; instead, diffuse staining of D2SV was observed throughout the cytoplasm (Fig. 9E). The percentage of cells showing this diffuse D2SV staining phenotype was significantly greater in cells co-expressing D2SV and dynamitin, compared with cells expressing only D2SV (Fig. 9G). Our results suggest that as the expression level of dynamitin increases, D2SV fails to form visible aggregates.

Although mammalian embryos are known to express a number of ER chaperones (Ni and Lee, 2007), it is not known whether embryonic cardiomyocytes possess all three branches of ER stress pathway (ATF6, Ire1 and PERK) that are known to regulate intracellular protein folding. All three pathways are known to activate the C/EBP homologous protein, CHOP/GADD153/DDIT3 and induce apoptosis in an attempt to cope with ER stress (Oyadomari and Mori, 2004). Given the absence of apoptosis and prevalence of cytoplasmic protein aggregation in cardiomyocytes expressing D2SV, we examined the effects of CHOP coexpression in cells transfected with D2SV or cycD2. In embryonic cardiomyocytes transfected with CHOP alone, the CHOP protein localized to the nucleus but we did not find any abnormal signs of cell death. Furthermore, co-transfection of CHOP promoted D2SV nuclear localization (Fig. 9H-J) and decreased protein aggregation (95%) in a dose-dependent manner (Fig. 9N). Similar to this result, expression of CHOP also promoted nuclear localization of cycD2 and decreased (50%) the incidence of protein aggregates in cardiomyocytes co-transfected with CHOP and cycD2 (not shown).

Given the negative effect of D2SV expression on cardiomyocyte cell cycle activity, we examined whether D2SV protein aggregates can sequester other cell cycle proteins in the cytoplasmic compartment, reduce their nuclear import and thereby regulate the progression of cell cycle activity. In the first series of experiments, we transfected D2SV into E14.5 cardiomyocytes and examined whether D2SV aggregates colocalize with endogenous cell cycle proteins. Double immunostaining experiments revealed that D2SV protein aggregates also contained endogenous proteins such as cyclin D2, cyclin B1 (Fig. 10A-F) and CDK4 (not shown) in 5-10% of transfected cells. Since the endogenous cell cycle proteins are rapidly degraded, cardiomyocytes were also co-transfected with D2SV and expression constructs coding for various cell cycle proteins and examined to see whether D2SV protein aggregates also contained the transfected cell cycle proteins. In each co-transfection experiment, approximately 50-70% of the co-transfected cells had a striking co-localization of D2SV protein aggregates with the transfected cycD1, cycB1 (Fig. 10G-L), CDK4 (Fig. 6A-C) or cycD2 (not shown).

To determine potential interactions of D2SV with other proteins in vivo, lysates of whole E14.5 embryos were immunoprecipitated (IP) with D2SV antibodies and immune complexes were collected using protein A-Sepharose beads. IP and supernatant fractions were subjected to western blot analysis using D2SV, CDK4, cycD2 and cycB1 antibodies. These IP and western experiments indicated that endogenous D2SV forms complexes mainly with endogenous CDK4 and cycD2 and also to a lesser extent with cycB1 (Fig. 10M). Although D2SV protein was able to sequester both endogenous and transfected cyclins, we reasoned that increased levels of intracellular cyclins could rescue D2SV-induced cell cycle arrest. To this end, we assessed [³H]thymidine LI in cultures co-transfected with D2SV and cycD2, cycD1 or cycB1 expression vectors. Compared with the LI of myocytes transfected with the D2SV or pcDNA vector alone, there was a significant increase in the LI of myocytes co-transfected with D2SV and cycD2 (11-fold), cycD1 (eightfold) or cycB1 (eightfold; Fig. 10N).

In this study, we showed an inverse relationship between D2SV expression and embryonic cardiomyocyte cell cycle activity in vivo. In addition, enforced expression of D2SV led to cell cycle exit in embryonic cardiomyocytes through a novel aggregation pathway coupled to ER stress. Collectively our findings indicate that D2SV is a negative cell cycle regulator involved in the control of early myocardial cell proliferation.

Similar to other D-type cyclins, D2SV lacks a consensus NLS. Phosphorylation at Thr286 of cyclin D1 by GSK-3β increases the nuclear export of this protein (Diehl et al., 1998). In contrast to the nuclear and cytosolic localization of D-type cyclins, D2SV is predominantly retained in subcellular compartments such as ER, Golgi and lysosomes, presumably through a cytoplasmic retention signal. Although addition of a SV40 NLS to D2SV did not alter its predominantly cytoplasmic localization, deletion of the last 20 amino acids significantly increased its nuclear localization. These results indicate that the C-terminus of D2SV harbors a cytoplasmic or organellar retention signal. Similar cytoplasmic or ER retention signals were also identified in other cell cycle proteins such as cyclin B1 (Pines and Hunter, 1994) and myt1 (Liu et al., 1997).

At present, it is also not clear how D2SV enters and exits the ER in the absence of a signal sequence. Studies from both yeast and mammalian systems identified signal-independent posttranslational translocation mechanisms for the import of cytoplasmic proteins into ER using components such as Sec61p, Sec62/63 and BiP (Eser and Ehrmann, 2003; Ng et al., 1996). It is possible that the hydrophobic 20-amino-acid D2SV C-terminus may be involved in such a process. Signal or ER retention sequences have been mapped to the N-terminus in the majority of cellular proteins (Martoglio and Dobberstein, 1998). Although, the C-terminal tails (∼20 residues) of secretory IgM heavy chain (Shapira et al., 2007) and COX2 (Mbonye et al., 2006) proteins have been shown to harbor ER retention and ERAD signals. Although D2SV C-terminus does not contain any known ER retention sequences, it does contain several amino acids with hydrophobic aliphatic side chains (e.g. leucine, isoleucine and valine) that are commonly found in several SRP-dependent signal sequences. Consistent with this notion, deletion of the C-terminus resulted in nuclear localization of D2SV protein.

Protein aggregation generally occurs as a result of misfolding (Garcia-Mata et al., 2002) or exposed hydrophobic regions (Wedegaertner, 2002). Misfolding of D2SV and D2SVΔCT could expose an aggregation prone hydrophobic stretch of amino acids that is not normally exposed in cycD2. Native gel electrophoresis combined with immunoblotting provides evidence that D2SV protein has intrinsic capacity to form high molecular mass complexes. Formation of D2SV aggregates in transfected cells or embryonic cardiomyocytes in vivo could also be explained by its interactions with other cell cycle proteins in a CDK4-dependent or -independent manner. We found that deletion of the last 20 C-terminal amino acids of D2SV is not sufficient for the abrogation of protein aggregation. Further mutational analysis of the N-terminus is required to map the region of D2SV responsible for protein aggregation. Abrogation of D2SV aggregation in cells over producing dynamitin clearly supports a role for dynein and dynactin in the retrograde transport of D2SV protein aggregates. Dynamitin, a subunit of the dynactin complex, has been shown to play a role in linking cargo to the dynein motor protein (Johnston et al., 2002). In the presence of excessive dynamitin, the dynactin complex undergoes a structural modification, which in turn triggers dissociation of dynein-dynactin interaction (Schroer, 2004). Similar to our study, enforced expression of dynamitin was also used to prevent aggresome formation of other aggregation prone proteins (Garcia-Mata et al., 1999; Ramanathan et al., 2007).

Given the impairment of ERAD in myocytes expressing D2SV, absence of key ER stress handling proteins in the embryonic heart could also explain persistent levels of D2SV protein aggregation. The expression of CHOP, a key regulator of ER stress (Oyadomari and Mori, 2004), was detected only during later stages of embryonic heart development in the rat fetus (Rees et al., 1999). In this study, we reasoned that enforced expression of CHOP in E14.5 myocytes expressing D2SV would sensitize D2SV-expressing cells and subsequently induce cell death. However, we observed only abrogation of D2SV protein aggregation but no cell death in co-transfected myocytes. This data further suggests that downstream factors linking CHOP to apoptosis may also be absent in embryonic mouse heart (Oyadomari and Mori, 2004). A striking nuclear colocalization of CHOP and D2SV or cycD2 also suggests that CHOP may be involved in proper folding and nuclear import of D-type cyclins either by direct or indirect interactions. This notion is consistent with the previous observation that the CHOP binding protein C/EBPα can interact with CDK2 and CDK4 and directly regulate their stabilities (Wang et al., 2002; Wang et al., 2001).

We showed that D2SV protein aggregation leads to cell cycle arrest without causing any overt cytotoxic effects in embryonic cardiomyocytes. This effect is distinct from other classes of aggregation prone proteins which have been shown to cause cytotoxic effects (de Cristofaro et al., 2000; Garcia-Mata et al., 2002). Several groups reported that aggregation prone proteins can sequester other non-aggregating proteins, provided that they are structurally similar to each other (Garcia-Mata et al., 2002; Suhr et al., 2001). Since D2SV harbors the majority of well-conserved cyclin box sequence (Abrahams et al., 2001), we reasoned that the intracellular aggregates formed by this protein might sequester other cell cycle proteins containing a cyclin box or cyclin box interacting domains by a process previously described as analogue co-aggregation (Garcia-Mata et al., 2002). In cells overexpressing D2SV, we found that this protein co-aggregates with endogenous or transfected cell cycle proteins such as cycD1, cycD2, cycB1 and CDK4. Taken together, these results provide a mechanism by which D2SV causes cell cycle arrest. If endogenous cyclins and CDKs are somehow trapped within D2SV aggregates, they will be unable to fulfill their role as positive regulators of the cell cycle and the outcome would be cell cycle inhibition. Consistent with this notion, we showed that overexpression of D-type or B-type cyclins can abrogate D2SV induced cell cycle arrest.

A previous study from Rassart's group found that D2SV is capable of binding to CDK4 but the D2SV and CDK4 complex is unable to phosphorylate pRb in the presence of CDK activating kinase (Denicourt et al., 2008). Rb phosphorylation is a prerequisite for the release of E2F transcription factors and progression of the cell cycle (Sherr, 1994). Inability of D2SV to phosphorylate Rb suggests that D2SV protein bound to CDK4 can act as a dominant negative for the normal D-type cyclins. Although Rassart's group did not directly examine the effects of D2SV in G1-S transit control, certainly their results on the lack of pRb phosphorylation are in agreement with the negative cell cycle effects of D2SV observed in the present study. Furthermore, the fact that coexpression of CDK4 with D2SV did not increase cell cycle activity similar to the CDK4 and cycD2 experiment also suggests that at least a part of the C-terminal region located downstream of cyclin box in D-type cyclins is also essential for the G1-S transit control. Moreover, absence of cellular transformation in MEFs transfected with D2SV alone (Denicourt et al., 2008) can be explained by the growth inhibiting role played by this protein. However, it is not clear how this protein is involved in the facilitation of H-Ras-mediated transformation of MEFs (Denicourt et al., 2008). It is generally accepted that oncoproteins such as T-antigen and E1A can induce cellular transformation by directly binding and abrogating the activities of growth suppressor proteins such as Rb and p53 (Dyson et al., 1989). It is probable that a similar mechanism may be responsible for the D2SV and H-Ras mediated cell transformation.

D2SV is expressed at higher levels at developmental stages when primitive myocardial cell proliferation rates are higher (e.g. E11.5) than at later stages of heart development (e.g. E14.5) when cell cycle activity gradually declines and ceases after birth. It was shown that trabecular cardiomyocytes have lower proliferative activity (∼twofold) compared with the compact zone myocytes during heart development (Erozhina, 1968; Rumyantsev, 1991). The presence of a higher level of D2SV protein in the trabecular zone compared with the compact zone in E14.5 myocardium is also consistent with a critical role for this protein in the negative regulation of myocardial cell cycle activity. The unique ability of D2SV protein to aggregate as well as sequester several endogenous cell cycle proteins underscores the physiological significance of a novel protein aggregation pathway involved in the control of myocardial cell proliferation. Although aggregation of misfolded proteins is implicated as a pathogenic mechanism in several neurodegenerative diseases, a similar role for protein aggregates has been speculated in cardiovascular diseases (Patterson and Cyr, 2002). We have recently shown that D2SV protein is upregulated in the adult heart in response to isoproterenol induced cardiac hypertrophy (not shown). Indeed, the presence of D2SV protein in normal or hypertrophic adult hearts suggests that mechanisms similar to D2SV protein aggregation may serve as a major block for therapeutic cardiac regeneration.

The breeding colony of CD1 mice (Charles River Laboratories, Montreal, Canada) was maintained in-house and female mice were mated with males under a 12 hour light:dark cycle. Noon on the day when the copulation plug was found was designated as embryonic 0.5 day (E0.5). All animal procedures were performed according to the guidelines set by the Canadian Council on Animal Care. Whole embryos and cardiac tissue were cryoprotected in PBS containing 30% sucrose and frozen in Tissue-tek OCT medium (Sakura Finetek, Japan). Tissue sections (10 μm) were obtained using a cryostat (Leica Microsystems, Canada).

Total RNA from embryonic mouse hearts was isolated using TRIzol reagent (Invitrogen, Carlsbad, CA). RNA was reverse transcribed using a Superscript II reverse transcription (RT) kit (Invitrogen). RT reaction was used to amplify the coding sequence of D2SV with sense (D2ex1S) and antisense (D2altAS) primers (see supplementary material Table S1 for sequences). The amplified sequence was cloned in pCR2.1 and pcDNA3.1A vectors (Invitrogen) and the sequence was confirmed. Primer sequences were selected based on the cycD2 splice variant cloned from the GMLV-transformed cell lines (Denicourt et al., 2003). D2SV or the sequence coding for the first 136 amino acids (D2SVΔCT) were fused in-frame with either a Myc epitope or an EGFP reporter to generate pcDNA-D2SVMyc, pcDNA-D2SVΔCTMyc or pEGFP-D2SV constructs. A nuclear localization signal (NLS) from SV40 was added to D2SV to generate pcDNA-NLS-D2SVMyc. The cDNA sequences coding for the mouse cyclin D1 (Soonpaa et al., 1997) or CDK4 (Matsushime et al., 1992) were also subcloned in pcDNA3.1A vector. All recombinant DNA manipulations were performed according to standard methodologies (Sambrook et al., 1989). The plasmids expressing a Myc-tagged CHOP (Kubota et al., 2005), a Myc-tagged dynamitin (Echeverri et al., 1996), cyclin B1 (Pines and Hunter, 1989) and TCRα-GFP (DeLaBarre et al., 2006) were described previously.

A 15 amino acid peptide corresponding to the C-terminus of D2SV (LLTLPFPITLRPPH) was synthesized and conjugated to keyhole limpet hemocyanin using a commercial system (Sigma-Genosys, Texas). The synthetic peptide was reconstituted in Freund's complete adjuvant and injected subcutaneously in New Zealand white rabbits. After two booster doses, serum was collected and affinity purified using a protein G-Sepharose column (GE Healthcare, NJ). Sections (10 μm) derived from embryos and adult hearts were fixed in methanol and blocked with 10% goat serum and 1% BSA in PBS for 1 hour. The specimens were incubated with polyclonal antibodies for D2SV and phosphohistone H3 Ser10 antibodies (63G; Cell signaling Technology, MA), followed by anti-rabbit and anti-mouse antibodies conjugated to Alexa Fluor 555 and Alexa Fluor 488 (Invitrogen). Nuclei were stained with 10 μg/ml of Hoechst 33342 stain for 5 minutes. Samples were examined using a Leica DM2500 fluorescence microscope or Zeiss LSM 510 Meta confocal microscope and images were digitized.

Ventricles were dissected from E14.5 embryonic hearts as reported previously (McMullen et al., 2008; Zhang and Pasumarthi, 2007). Ventricular myocardial cells were isolated by digestion with 0.2% type I collagenase (Worthington, NJ) and counted using a hemocytometer. Approximately 1-3×10⁶ cells were plated on fibronectin-coated chamber slides (Nalge Nunc, NY) in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal bovine serum (FBS). Human embryonic kidney cell line (HEK293) was obtained from the American Type Culture Collection and cultured in 10 cm dishes (Corning) with DMEM plus 10% FBS. Cells were transfected with 2-8 μg of plasmid DNA using a Lipofectamine reagent (Invitrogen) for 6 hours as recommended by the manufacturer.

Transfected HEK293 cells or whole embryos were lysed in tumor lysis buffer [1% NP40, 5 mM EDTA, 50 mM Tris-HCl pH 8.0, phenylmethylsulphonyl fluoride (PMSF; 10 μg/ml) and aprotinin (10 μg/ml)]. Lysate was sonicated and centrifuged at 13,500 × g for 15 minutes at 4°C. Supernatants were collected and protein concentrations were estimated by the Bradford assay (Pierce). 40 μg of protein was denatured in Laemmli buffer and separated in denaturing conditions on a 12.5% SDS-PAGE gel. Proteins (50 μg) were also separated on a 5% basic-native PAGE gel as described by the Wolfson Centre for Applied Structural Biology (http://wolfson.huji.ac.il). For immunoprecipitation (IP) analysis, 1 mg of protein lysate was incubated with D2SV antibodies overnight at 4°C, and immune complexes were collected using protein A-Sepharose beads (GE Healthcare, Canada) as described previously (Tsai et al., 2000). Supernatant and IP fractions were resolved on 10% SDS-PAGE gels. The separated proteins were transferred from the gel to a nitrocellulose membrane, blots were blocked (5% milk, 3% BSA, 0.1% Tween in PBS), and incubated with antibodies specific for D2SV (in-house), cycD2 (Santa Cruz Biotechnology, sc-593), CDK4 (sc-260) and cycB1 (sc-752) followed by incubation with anti-rabbit or anti-mouse antibodies conjugated to peroxidase. Signals were visualized using the ECL method (GE Healthcare).

Cells were fixed in ice cold methanol or acetone for 15 minutes, blocked with 10% goat serum and 1% BSA in PBS for 1 hour. The specimens were incubated with antibodies (4 μg/ml) for D2SV (in-house), c-Myc (sc-40), sarcomeric myosin (MF20, DSHB, University of Iowa), cyclin D2 (sc-593), CDK4 (sc-260), PDI (sc-20132), giantin (a generous gift from Neale Ridgeway, Dalhousie University), cyclin B1 (sc-752) or cyclin D1 (sc-717), followed by anti-rabbit and or anti-mouse secondary antibodies conjugated to Alexa Fluor 555 and/or 488 (Invitrogen). Cell membranes were stained using a glycoprotein-specific probe, Alexa Fluor 647-labeled concanavalin-A stain (Invitrogen; 50 μg/ml). Nuclei were stained with 10 μg/ml of Hoechst 33342 stain for 5 minutes. For visualization of lysosomes, live cells were stained with Lysotracker Red DND-99 (Invitrogen) at a concentration of 75 nM in the growth medium for 30 minutes at 37°C.

E14.5 myocardial cells were pulsed with [³H]thymidine (GE Healthcare, 1.0 μCi per 1 ml of medium) and incubated at 37°C for 6 hours. Cells were washed with PBS, fixed and processed for immunodetection and Hoechst staining procedures. Chamber slides with the cells attached were then air dried, coated with Ilford photographic emulsion (L4, Polysciences, PA) and placed in a light-tight box at 4°C for 2 days. Subsequently, slides were developed in Kodak-D19 developer (Sigma) for 4 minutes, fixed with Ilford rapid fixer (Polysciences) and mounted using a 1% propyl gallate solution as reported earlier (Pasumarthi et al., 1996). Cells undergoing DNA synthesis were readily identified as those that showed colocalization of the Hoechst and more than 15 silver granules.

Data are presented as means ± s.e.m. Multiple group comparisons were performed using ANOVA with Tukey's multiple comparisons tests. Between groups comparisons were performed using Student's t-test. All statistical analyses were performed using GraphPad InStat version 3.06 for Windows (Graphpad, San Diego, CA). Significance was assumed at P<0.05.