Embryonic expression of AMPK γ subunits and the identification of a novel γ2 transcript variant in adult heart

Abstract

AMP-activated protein kinase (AMPK), the key sensor and regulator of cellular energy status, is a heterotrimeric enzyme with multiple isoforms for each subunit (α1/α 2; β1/β2; γ1/γ2/γ3). Mutations in PRKAG2, which encodes the γ2 regulatory subunit, cause a cardiomyopathy characterized by hypertrophy and conduction abnormalities. The two reported PRKAG2 transcript variants, γ2-short and γ2-long (encoding 328 and 569 amino acids respectively), are both widely expressed in adult tissues. We show that both γ2 variants are also expressed during cardiogenesis in mouse embryos; expression of the γ3 isoform was also detected unexpectedly at this stage. As neither γ2 transcript is cardiac specific nor differentially expressed during embryogenesis, it is paradoxical that the disease is largely restricted to the heart. However, a recently annotated γ2 transcript, termed γ2-3B as transcription starts at an alternative exon 3b, has been identified; it is spliced in-frame to exon 4 thus generating a protein of 443 residues in mouse with the first 32 residues being unique. It is increasingly expressed in the developing mouse heart and quantitative PCR analysis established that γ2-3B is the major PRKAG2 transcript (~ 60%) in human heart. Antibody against the novel N-terminal sequence showed that γ2-3B is predominantly expressed in the heart where it is the most abundant γ2 protein. The abundance of γ2-3B and its tissue specificity indicate that γ2-3B may have non-redundant role in the heart and hence mediate the predominantly cardiac phenotype caused by PRKAG2 mutations.

Highlights

► We have identified a novel PRKAG2 transcript of intermediate length (γ2-3B). ► γ2-3B is the most abundant cardiac AMPK γ2 at both mRNA and protein levels. ► Functional changes in AMPK containing γ2-3B may mediate PRKAG2 cardiomyopathy. ► γ2 and γ3 are the early embryonic AMPK γ subuits.

1. Introduction

AMP-activated protein kinase (AMPK) is an evolutionarily conserved sensor of nutritional and environmental stress. The heterotrimeric complex is composed of the catalytic α, the scaffolding β and the nucleotide binding γ subunits. There are two or three isoforms of each subunit in mammals (α1 and α2; β1 and β2; γ1, γ2 and γ3), each encoded by different genes [1,2]. The α1 and β1 subunits are ubiquitously expressed whereas α2 and β2 expression is relatively higher in cardiac and skeletal muscle than in other tissues [3]. Of the γ isoforms, γ1 and γ2 are expressed quite uniformly throughout different tissues whereas γ3 has only been detected in skeletal muscle [2]. The γ1 isoform is the major regulatory subunit, being present in complexes that account for 80–90% of total AMPK activity in all tissues [2]. In endothelial cells, AMPK containing the γ2 subunit has been localized to the cytokinetic apparatus where it may regulate mitotic processes [4].

Two major AMPK γ2 variants have been reported, produced by transcription from different promoters: γ2-short (also termed γ2b), a protein of 328 amino acids containing the four cystathione β-synthase (CBS) domains responsible for adenine nucleotide binding, and γ2-long (γ2a), which is composed of the γ2-short sequence plus a 241 residue N-terminal extension [2,5].

Mutations in the PRKAG2 gene have been shown to cause a cardiac specific phenotype of hypertrophy with associated glycogen deposition, Wolff-Parkinson-White syndrome (WPW) and conduction abnormalities [6–9]. All the reported mutations are located in the nucleotide-binding domains and our present understanding is that the AMP binding is lower or abolished in the mutant protein and in consequence, AMPK activation is impaired [10,11] but also that the basal activity is increased [12]. The largely cardiac-restricted nature of the disease suggests that AMPK γ2-containing complexes have a specific role, different subcellular localization and/or particular temporal expression in the heart. Certain PRKAG2 mutations cause death at the fetal or neonatal stage and therefore γ2-AMPK must be present in the developing heart where the relative expression of the γ isoforms has not been previously reported [13].

In order to understand the development of the cardiac disease with γ2 mutations we studied the embryonic expression of the γ regulatory subunits, γ1, γ3 and the two transcript variants of γ2 in mouse embryos. In the developing heart we detected a third, largely cardiac specific γ2 transcript variant that becomes, with γ2-short, the major γ2 protein in adult heart.

2. Materials and methods

2.1. Animals, tissue collection

All experiments were conducted in accordance with the UK Home Office Animals (Scientific Procedures) Act of 1986 and the Guide for the Care and Use of Laboratory animals published by the US National Institutes of Health (NIH Publication No. 85-23, revised 1996). CD1 mice were sacrificed, and embryos were collected at different embryonic time points for immunohistochemistry studies and for RNA extraction from isolated hearts. Hearts were also obtained from new-born and adult CD1 mice. Organ samples for protein extraction were taken from a 10 week-old male C57BL/6 mouse.

2.2. In situ hybridization

Riboprobes were amplified by PCR from mouse heart cDNA, using transcript specific primers. The PCR fragments were cloned into pGEM-T Easy. Sense and anti-sense riboprobes were labelled with DIG-11-dUTP using the DIG RNA labeling kit (Roche). The primers for the γ2-long probes were: GGACGAGGCACCGGAATTAAC (forward)–GTATGTCGGGGATGCCAGCGGAGGTC (reverse); the probe is 698 bp. The primers for the γ2-short probe were: CGCCCCCCGCCACTG (forward)–GCCGCCAGCACCGCCTGAG; the probe is 107 bp and anneals to a segment in the 5′ UTR.

In situ hybridization was performed in whole mount embryos, using a published procedure [14].

2.3. Immuno-histochemistry

Mouse embryos were fixed in 4% paraformaldehyde-phosphate buffered saline (PFA-PBS), then dehydrated in increasing ethanol concentration, cleared in Histoclear, embedded in paraffin and 5–10 μm sections were cut. Sections were rehydrated, and microwaved in a citrate solution (Vector Laboratories) for antigen retrieval. We used Antibody Diluent (DAKO) for blocking before incubating the sections with rabbit-anti-PRKAG2 (1:25); rabbit anti-PRKAG1 (1:50); rabbit anti-PRKAG3 (1:50) for 1 h at room temperature. (All antibodies were purchased from Zymed Laboratories.) Alkaline phosphatase conjugated secondary antibodies (Vector laboratories) were applied for 1 h, and staining was visualized using the Vector® Blue Alkaline Phosphatase Substrate Kit III (Vector laboratories). Section was dehydrated and permanently mounted in Vectamount (Vector laboratories).

2.4. Real-time PCR

Semi-quantitative real-time PCR with QuantiTect SYBR Green RT-PCR reagents (Qiagen) were carried out in triplicates using 20 ng RNA isolated from developing and adult mouse hearts. Data were normalized against the housekeeping gene, metastatic lymph node 51 (MLN51), that is expressed more consistently during development than other, traditionally used housekeeping genes [15,16]. Primers for MLN51: CGCCGAGGAGTCTGAGTGTG (forward) and TCGTTAGCTTCTGATTTCAGC (reverse). Primers for γ2-short: CCCGATGCGAGAAGCCCG (forward) and CTCTGCTTTTTACTTTCCCACAGCGGC (reverse); for γ2-long: ATCTATGCTTCCTCGTCCCCTCCAG (forward) and AAAGCCACTTTCTGAGTCTTCTT (reverse).

The primers to detect γ1: GCTGAGGAACTGGCGGGCG (forward) and GGGAATTAGGTCATAGCAGC (reverse); to detect γ3: CTGTTCCCTTGGCTGAAGCGGAGACC (forward) and TGGGGCTGGGAACTCTATGGTCA (reverse).

Relative expression of the γ2 transcript variants was quantified by TaqMan® real-time PCR method using a human cDNA panel (Clontech). Oligonucleotides, designed with the Primer Express 3.0 software are listed in Supplement Table S1. Taqman MGB probes span exon-exon or 5′ UTR-exon junctions. Beta-2 microglobulin was the endogenous control. All primers and probes were custom made by Applied Biosystems unless it is stated otherwise. Reactions were carried out in StepOnePlus Real-Time PCR system using MicroAmp Fast Optical 96-well plates with Fast Master Mix, both purchased from Applied Biosystems. Reactions (20 μl each) were set up in duplicates with 1 ng cDNA, 200 nM probe and 900 nM primers. Expression was quantified using the relative standard curve method (Applied Biosystems, user bulletin #2).

2.5. Protein extraction and Western blot analysis

Organ samples were homogenized in 50 mM Tris–HCl, pH 7.5, 150 mM NaCl, 2% SDS, 1 mM PMSF and 1 mM dithiothreitol containing complete protease inhibitor mix (Roche). Polyclonal antibody, G2-3B was raised in rabbit against the KHL-conjugated peptide of mouse γ2-3B (MKRFGSLRGTKKPKDQNRSTQRR-C) by Harlan UK (Hillcrest). Rabbit polyclonal GAPDH antibody was obtained from Abcam and rabbit polyclonal anti-PRKAG2 antibody was purchased from Atlas-Antibodies.

2.6. Cell fractionation

The FractionPREP™ cell fractionation kit was purchased from BioVision and used according to manufacturer's instructions. Hearts were obtained from C57BL/6 mice and processed immediately.

3. Results

3.1. Expression of PRKAG2 transcripts in adult mouse tissues

PRKAG2 is widely expressed throughout different tissues in human and therefore the reason for the cardiac-restricted nature of the disease caused by PRKAG2 mutations is unclear [5]. In order to examine whether either of the two major γ2 variants, γ2-short and γ2-long, showed cardiac specific expression, we examined the expression of PRKAG2 transcripts by real-time PCR in mouse heart, skeletal muscle, liver, brain and kidney (Fig. 1). The primers were designed to distinguish between the transcripts: there is a unique 5′ untranslated sequence (UTR) in the γ2-short transcript in front of exon 5 where the forward primer anneals. While γ2-short expression is absent from liver, it is present in the other tissues examined as is the transcript of γ2-long, and therefore neither γ2 variant is cardiac specific in mouse.

Fig. 1.

Transcription of γ2-short and γ2-long in mouse — real-time PCR.

Relative expression of γ2-short and γ2-long in adult tissue samples; SYBR Green detection, n = 3. Cartoons of the DNA sequences show the unique 5′ UTR (black) of γ2-short, the extended N-terminal region (white) of γ2-long and the nucleotide-binding region (gray). The locations of the PCR products are indicated by solid lines.

3.2. Embryonic expression of the AMPK-γ subunits — in situ hybridization

Differential expression of the PRKAG2 transcript variants may occur during embryogenesis. Our first choice to investigate this was in situ hybridization since a riboprobe that hybridises to the 5′ UTR segment of the γ2-short transcript can detect the γ2-short transcript separately from γ2-long. However, early mouse embryos (8.5 dpc — day post coitum) were similarly stained around the fusion of the neural fold, folding ectoderm, somites, developing forebrain with both the γ2-long and γ2-short riboprobes (Fig. 2) and at least at this stage, the expression pattern of the two transcript variants appears to be very similar.

Fig. 2.

Whole mound in situ hybridization of mouse embryos — 8–8.5 dpc.

A: γ2-long, B: γ2-short, dorsal view; sm — somite; ne — neural ectoderm; ↙ — sense probe as control.

The γ1 transcript, which is the major AMPK regulatory subunit in adult, could not be detected in these embryos; even prolonged incubation with the γ1-specific probe produced no colouration (not shown). In contrast, hybridization with the γ3 riboprobe resulted in strong, overall staining (not shown) that indicates an unexpected, ubiquitous expression of γ3 at 8.5 dpc; γ3 expression has only been shown in skeletal muscle in adults [2,17].

3.3. Mouse embryo sections — immunohistochemistry

Since both γ2 transcripts and γ3 but not γ1 were detected in early embryos, we switched to immunohistochemistry to investigate the detailed expression pattern of all γ subunits in the developing embryonic tissues. Antibodies specific to each AMPK γ subunits are available but only γ2-long can be detected separately with antibodies that recognize unique epitopes in the N-terminal extension of the γ2-long protein. We sectioned and stained 10.5–12.5 dpc embryos with an N-terminal γ2 antibody and carried out experiments in parallel with γ1 and γ3 antibodies (Fig. 3).

Fig. 3.

Immunohistochemistry — mouse embryos.

Transverse (A, C, E) and sagittal (B, D, F) serial sections were stained with γ1 — A–B; γ2 N-terminal — C–D; and γ3 — E–F antibodies. Labeling: fba — first branchial arch; int — intestine; lv — liver; mrl — marginal layer neural tube; mtl — mantle layer neural tube; se — surface ectoderm; sg — sympathetic ganglion; urg — urogenital ridge; v — ventricle.

In mouse embryos, γ1 expression can be detected around 10.5 dpc in distinct neural structures at first instance, while γ2-long and γ3 expression continues. Strong γ2-long expression was detected in the surface ectoderm and in the neurotube of 10.5 dpc embryos (Fig. 3C). While γ1 becomes detectable at this stage it is not as widely expressed as γ2 and γ3; the expression is mainly restricted to a few structures showing overlapping staining with the γ3 antibody (10.5 dpc — e.g. sympathetic ganglion, Figs. 3A, E). There is some γ1 staining in the developing heart but it is still weak (Figs. 3A, B). Staining of the sympathetic ganglion was confirmed with the neural specific NeuN antibody (not shown).

The staining pattern with the γ3 and the γ2-long antibodies somewhat overlaps (ventricle — Figs. 3C, D, E, F) but there are differences: γ2-long is detected in the developing liver at 11.5 dpc while γ3 staining is not present there (Figs. 3D, F). The γ2 antibody also depicts structures that are stained with the γ3 antibody as well (heart, tongue, facial structures and kidney — not all are demonstrated in Fig. 3), structures where abnormalities have been reported accompanying the cardiac phenotype with certain γ2 mutations [13,18].

3.4. Cardiac development in mouse embryos — immunohistochemistry

Focusing on the developing heart, here we show which specific cardiac structures are stained with the different γ antibodies (Fig. 4). The γ2-long antibody staining depicts the atria, common ventricular chamber, epicardium and pericardium (Figs. 4D, C) and γ2-long is also present in the atrio-ventricular junction that separates the atria from the ventricle (Fig. 4C). The latter could be particularly relevant as defects in this region are implicated in the development of accessory pathways in patients with WPW. Strong γ3 staining was detected in the developing heart: structures stained with the γ2-long antibody are also stained with the γ3 antibody (Figs. 4E, F). In addition, γ3 is also present in the outflow tract (Fig. 4E). In contrast, the γ1 staining, both in the atria and ventricles, was still weak at 12.5 pc (Fig.4B); the strong staining of the vagus nerve overlapped with γ3 staining and there was no γ2-long detected here (Figs. 4B, F).

Fig. 4.

AMPK γ proteins in developing mouse heart — immunohistochemistry.

Sagittal (A, D, E) and transverse (B, C, F) embryo sections stained with γ1, γ2 N-terminal (γ2) and γ3 antibodies. Labelling: a — atria; avj — atrio-ventricular junction; cv — common ventricular chamber; ep — epicardium; lv — left ventricle; oft — outflow tract; pc — pericardium; rv — right ventricle; vg — vagus nerve.

3.5. Expression of AMPK-γ subunits during cardiogenesis by real-time PCR

In order to compare the expression of the γ subunits, real-time PCR was performed on cDNA prepared from developing hearts from 13.5 dpc, this being the earliest time point when dissection of the heart and extraction of sufficient RNA was possible in our hands. Each γ transcript, including γ1 was already clearly detectable at 13.5 dpc (Fig. 5). The expression of both γ2 transcripts, γ2-long transcription especially, increased until birth after which the expression levels of both, but in particular the short form, declined. Embryonic expression of γ3 was unexpected as it had previously been reported to be only present in adult skeletal muscle [2,17]. The γ3 expression was unchanged from 13.5 dpc to birth when it sharply declined, thus γ3 appears to function as an “embryonic” isoform in the heart. The transcription of γ1 increased in the developing heart and remained high, accounting for the reported predominance of γ1 in the adult heart [2].

Fig. 5.

Real-time PCR — γ transcripts in developing and adult heart.

A: γ2-short and -long; B: γ1 and γ3 transcription. SYBR Green detection; transcription level is expressed as fractions of the values at birth in each case; the positions of the PCR products from the γ2 transcripts as in Fig. 1; n = 3.

3.6. A third, cardiac specific γ2 transcript variant

In an earlier report by Lang et al. [5], detection of PRKAG2 transcripts in a human tissue panel by northern blot revealed the presence of an abundant band in the region of the short γ2 message (2.4–3 kb), with the long transcript (3.8 kb) accounting for only a small fraction of the total signal. Interestingly, the authors noted that the band of the short transcript was actually a doublet, and the slightly larger component (3 kb) was suggested to have a longer 3′ UTR. However, since that report two further PRKAG2 variants have been identified and it is likely that the intermediate 3 kb transcript represents one of these. One form, in which transcription starts in exon 2 and encoding a protein of 525 amino acids, has been annotated in human Ensembl; this has been referred to as γ2c in a recent report [19]. However, we found very low levels of γ2c in human tissues and in mouse its expression appeared to be restricted to the brain (Supplement Fig. S2). Therefore we propose that the 3 kb γ2 transcript is most likely to be the intermediate-sized γ2 mRNA encoding a protein of 443 (mouse) or 445 (human) residues recently annotated in both mouse and human Ensembl databases (http://www.ensembl.org). This mRNA is transcribed from another alternative promoter in the PRKAG2 locus within intron 3 (Fig. 6). The transcript has a unique starting exon (“3b”), which is spliced in-frame with exon 4, and the rest of the sequence is therefore identical with that of γ2-long (see Supplement Fig. S1). We refer to this new intermediate form as γ2-3B.

Fig. 6.

Schematic representations of the PRKAG2 locus — transcription of the γ2 variants.

We amplified the entire coding sequence of γ2-3B by PCR using cDNA from mouse and human heart (Supplement Fig. S3) and verified the sequence in both cases. We applied a TaqMan-based real-time PCR approach to quantify the contribution of γ2-3B transcript to total γ2 mRNA in cDNA from a selection of human tissues (Fig. 7). Both fetal and adult hearts contain substantially more γ2 than the other human tissues and approximately 50% of it is γ2-3B (Fig. 7A). The level of γ2-3B in the heart is at least five times greater than that found in skeletal muscle and brain, and over ten times higher than the levels in other tissues. A similar pattern of expression was found using semi-quantitative PCR analysis of mouse tissues (Supplement Fig. S3).

Fig. 7.

The major γ2 transcript is γ2-3B in human heart — quantitative PCR.

A: Human cDNA panel — expression of γ2 or γ2-3B is shown as fraction of the γ2 transcription in adult heart. B: Contribution of each γ2 transcript variant to the overall γ2 level in adult human heart. C: The annealing sites of primers and probes.

Transcription of the total γ2, along with the γ2a, γ2b, γ2c and γ2-3B variants was also quantified by TaqMan® real-time PCR method using primers that anneal to the 5′ UTR of γ2-long and γ2-short with probes annealing to the overlapping sequence of 5′ UTR and exon 1 in each cDNA (Fig. 7C; sequences listed in Supplement Table S1). Analysis of adult human heart cDNA demonstrated that γ2-3B and γ2-short transcripts are the major cardiac γ2 mRNAs (Fig. 7B, Supplement Table S2). The γ2-long form appears to be a minor component (less than 10% of total) and expression of γ2c was found to be negligible (Fig. 7B, Supplement Table S2). The γ3 mRNA level was also negligible in heart but, as expected, high in skeletal muscle (Supplement Table S2).

In order to confirm the abundance of the γ2-3B variant at the protein level we raised a rabbit polyclonal antibody against the unique N-terminal peptide of γ2-3B; this antibody (G2-3B) only recognizes γ2-3B and its specificity was verified using recombinant γ2-3B in western blot (Supplement Fig. S4). Mouse heart contains 5–10 times more of the γ2-3B protein than any other tissue we have sampled; no γ2-3B was detected in skeletal muscle or in kidney, and it appears that the protein is largely cardiac specific (Fig. 8A). In order to estimate the relative abundance of the various γ2 variants in the heart and their subcellular localization, western blots were performed on mouse heart fractionated using a differential detergent extraction method to generate nuclear, membrane, cytoskeletal and soluble fractions. A pan-γ2 antibody, raised against a peptide that overlaps the junction between the N-terminal extension of γ2-long/γ2-3B and the N-terminus of γ2-short, was used. The γ2 proteins were almost exclusively located in the cytoskeletal fraction (Supplement Fig. S5A) and within this fraction bands corresponding to γ2-3B and γ2-short were the major γ2 proteins (Fig. 8B). Three γ2 proteins were also detected with a γ2 C-terminal antibody, kindly provided by D. Carling (Supplement Fig. S5B).

Fig. 8.

Expression of the γ2-3B protein in mouse tissues — Western blots.

A: G2-3B antibody, specific to γ2-3B; the intensities of the protein bands were quantified and plotted as fractions of the level in heart. Equal amount of protein (45 μg) was loaded in each lane. B: Proteins in the cytoskeletal fraction of mouse heart were probed with anti-PRKAG2 antibody (Atlas Antibodies) that recognizes all three γ2 proteins. Bands were quantified and expressed as % of total γ2 (n = 3).

4. Discussion

4.1. Cardiac-specific γ2 transcript variant

This study set out to gain insight into the expression of PRKAG2 in the heart and how mutations in this apparently ubiquitously expressed gene cause a cardiac-restricted disease. Neither of the two well characterized transcripts, γ2-long and γ2-short, are cardiac-specific in human [5] or in mouse (Fig. 1) and there is no embryonic preference for either γ2 transcript as both are expressed in early embryos (Fig. 2). However, there is an intermediate γ2 transcript variant that is also transcribed from the PRKAG2 locus; it appears to be increasingly expressed in the developing mouse heart and is largely cardiac-specific in adult. Since transcription of this γ2 variant starts in exon 3b, an alternative exon within intron 3, we have named this transcript γ2-3B.

Quantitative PCR confirmed that γ2-3B is a major transcript in both fetal and adult human heart. It accounts for 50–60% of total γ2 mRNA, with γ2-short being the other significant contributor. It is likely that this newly identified γ2-3B transcript corresponds to the 3 kb γ2 message earlier detected in a northern blot of human cardiac mRNA [5].

4.2. The major AMPK γ2 protein is γ2-3B and γ2-short in heart

Consistent with the transcript data (Fig. 7B), the γ2-3B and γ2-short variants comprise the large majority of γ2 protein in mouse heart (Fig. 8B). This finding contradicts earlier western blot data which indicated that γ2-long is almost the only form of γ2 protein in the heart [2,17]. However, our protein data are not only consistent with our transcript results but are also supported by the northern blot data of Lang et al. [5] which clearly shows the γ2-long to be a minority transcript in the heart.

While γ2-short is more uniformly expressed in mouse tissues, γ2-3B is largely cardiac-specific (Fig. 8A), suggesting that alterations to the function of γ2-3B may be responsible for the predominant cardiac phenotype produced by mutations in the PRKAG2 gene. The PRKAG3 mutation which causes glycogen accumulation in skeletal muscle in both pig and human [20,21], does so by affecting the γ subunit specific to skeletal muscle; we suggest that, in an analogous way, PRKAG2 mutations may act via γ2-3B, a cardiac specific γ subunit.

4.3. Embryonic expression pattern of γ1 and γ3

The investigation of the expression pattern of the other γ regulatory subunits yielded unexpected observations. Surprisingly, γ3 was detected in early embryos by in situ hybridization and by immunohistochemistry. In early mouse embryos (8–8.5 dpc) there was no expression of γ1, the major AMPK regulatory subunit in adults [2]. Both short and long γ2 transcriptions were detected in distinct structures of the developing neuromuscular system (e.g. somites, neural ectoderm). The γ2 and γ3 subunits appear to act as the early “embryonic” AMPK regulatory subunits, rather than γ1. We first detected the γ1 protein in 10.5 dpc embryos in which the staining was localized only to certain neural structures. However, increased γ1 expression was observed at later stages of embryogenesis in the developing mouse heart and γ1 becomes the major AMPK regulatory subunit in adult [2]. The γ1 knockout mouse does not suffer embryonic lethality nor has any marked cardiac phenotype in adult [22], and this may reflect the limited embryonic role of γ1.

The recent publication of Kim and co-workers analyzed developmental and disease-related changes in AMPK subunit expression [19]. However, the analysis of γ2 expression focused on γ2-short, γ2-long and γ2c and did not include γ2-3B. We have not detected substantial γ2c expression in human or mouse heart, and in mouse γ2c was mostly detected in brain (Supplement Fig. S2).

The “γ2-long” primer pair used in the early stages of our work was capable of amplifying a fragment from both γ2-3B and γ2-long transcripts; as γ2-long is a small fraction of total γ2 in the heart (Fig. 8B), the increase of “γ2-long” expression in the embryonic mouse heart was likely due principally to the increasing expression of the γ2-3B form. We tested the presence of γ2-3B transcript in the embryonic developing mouse heart (Supplement Fig. S6) and found that transcription of γ2-3B occurred from 13.5 dpc and increased during development; moreover the pattern resembled the increase of “γ2-long" observed earlier (Fig. 5).

The fetal-adult transition from γ3 (which is present in glycolytic, white skeletal muscles in the adult [17]) and γ2-short to γ1 and γ2-3B might be attributed to a metabolic switch at birth when oxidative metabolism replaces the mainly glucose and lactate utilization via glycolysis [23,24].

4.4. Implications of the specific staining patterns in the heart

The similar staining of cardiac structures during development with γ3 and γ2-long antibodies (e.g. the atrio-ventricular junction) suggests that AMPK trimers containing these γ subunits might have similar functions. Lesions in the atrio-ventricular junction have been previously shown in patients with γ2 mutations and it was also found in the γ2 Asn488Ile transgenic mouse [25,26]. Therefore these lesions might allow for accessory pathways and later conduction block. However, it has to be noted that no conduction defects have been observed in either pigs with γ3 mutations [20,27] or in the transgenic mice that mimic the porcine phenotype [28,29] suggesting that γ2 may either have a different or more significant role during development in this region.

4.5. Implications of the cardiac specificity of AMPK γ2-3B

Since the heart has relatively more AMPK γ2 than other tissues with the majority being γ2-3B, it is possible that the mainly cardiac phenotype caused by γ2 mutations is mediated via the γ2-3B variant. In addition, AMPK complexes containing the different γ2 variants may have different functions and localizations, interacting with different proteins and linking the specific AMPK trimers to different signalling pathways. It has been reported that γ1 directs AMPK to the Z-line in differentiated myofibres [30] and α2/β2/γ3 complexes are preferentially activated upon certain exercise regime [31]. By extracting proteins from different cellular fractions we found all γ2 proteins almost exclusively in the cytoskeletal fraction of mouse cardiomyocytes (Supplement Fig. S6) suggesting that γ2-AMPK is mainly associated with sarcomeric structures. Using a yeast two-hybrid screen of a human heart cDNA library, we have identified cardiac troponin I as an interactor with amino acids 1–273 of γ2-long, a region that includes overlap with γ2-3B [32]. However, further investigation is required to detect the possible differential compartmentalization of the γ2-AMPK complexes and to establish whether AMPK complexes with different γ2 variants are linked to different cellular processes.

5. Conclusions

We have demonstrated that a third AMPK γ2 variant, γ2-3B, is increasingly expressed in the developing heart. Since the expression is largely cardiac-specific, PRKAG2 mutations most likely cause the cardiac-restricted phenotype via changes in the function and/or localization of γ2-3B-AMPK. The embryonic expression profile of the other AMPK regulatory subunits (γ1, γ3) has also been investigated and established that γ2 and γ3 are the early embryonic AMPK regulatory subunits.

Disclosures

None declared.

Appendix A. Supplementary data

Supplementary material