Sarcomere Mutation-Specific Expression Patterns in Human Hypertrophic Cardiomyopathy

Human heart tissue procurement

Ventricular myocardial tissue was snap frozen in liquid N₂ or placed in formalin at the time of collection. This study had the approval of the University of Michigan Institutional Review Board (IRB) and subjects gave informed consent.

Transcript analysis

RT-PCR, cDNA sequencing, and RT-qPCR were performed by standard techniques (see on-line data supplement). An RT-qPCR assay was used to determine allelic specific expression in the samples containing splice-site mutations, while the single-base extension method was used in samples containing single nucleotide variants that resulted in premature stop codons (see on-line supplemental methods and Supplementary Table 1).

Protein preparation, quantification and immunolocalization

Immunoblotting, immunofluorescent imaging, myofilament fractionation, and extraction of insoluble proteins were performed by established methods (see on-line data supplement).

Absolute Quantification of Abundance (AQUA)

Myofilament proteins were separated using SDS-PAGE and Coomassie stained. The protein of interest was gel excised. After in-gel enzymatic digestion, samples were analyzed by nano LC/MS/MS performed in the Orbitrap at 70,000 FWHM and 17,500 FWHM resolution, respectively. In some cases, peptides were post-translationally modified (ex. Met oxidation) or contained missed cleavages. The abundance of mutant and wild type sarcomere proteins within each sample was then determined using isotopically-labeled synthetic AQUA peptides corresponding to each form of wild-type and mutant peptides for each individual sample. Samples were analyzed by LC-SRM/MS with a Waters NanoAcquity HPLC system interfaced to a ThermoFisher TSQ Quantum Ultra. Peak areas for the wild-type or mutant endogenous peptide were expressed as a ratio to their corresponding AQUA peptides allowing the mole ratio of wild type versus mutant peptide to be calculated. Molar amounts from all wild-type and mutant peptides were summed, and expressed as a percentage of total protein. More details can be found in the on-line data supplement.

cDNA sequencing of MYBPC3 splice site mutations confirm presence of PTCs

We sequenced cDNA from all myocardial samples containing known or putative splice site mutations in MYBPC3. All mutations located in a consensus splice donor or acceptor site created a PTC in the mutant transcript (Supplementary Table 3). All mutations in exon or intron splice donor sites (n=5) resulted in skipping of the exon containing, or preceding, the mutation (except c.2308 G>A for which a mutant transcript was not detectable), while mutations in intron splice acceptor sites (n=2) resulted in inclusion of the intron containing the mutation (Supplementary Figure 1).

Sarcomere gene mutations confer variable allele-specific expression

For truncating mutations, the mutant transcripts constituted a minority of total transcripts (range undetectable to 44%) for both the MYBPC3 splice site mutations as determined by RT-qPCR (Figure 1A, left panel), and for the non-splice site MYBPC3 truncating mutations (Figure 1A, right panel) as determined by single-base extension. These data are consistent with the hypothesis that truncating mutant transcripts with PTCs are susceptible to degradation by nonsense-mediated RNA decay (NMD).^{5, 6}

Open in a separate window

Figure 1

Sarcomere gene expression in human HCM samples. A. (Left) Allele-specific expression from wild-type and MYBPC3 splice-error PTC-containing transcripts quantified by RT-qPCR with custom designed primers (Supplementary Table 5). (Right) Allele-specific expression from wild-type and MYBPC3 PTC-containing transcripts (without splice errors) quantified by custom-designed single base extension reactions. These mutations are either single nucleotide substitutions that encode for a termination codon, or an insertion that causes a frameshift. B. Quantification of wild-type and single nucleotide substitution mutant sarcomere gene transcripts without PTCs by custom-designed single base extension reactions. For the MYBPC3 exon splice site mutations, the mutant fraction is the proportion of detectable missense transcript. The remainder of the samples contain true missense mutations. In instances where 2 samples carried the same mutation, these are designated as (a) and (b) respectively in this figure, and maintained throughout the manuscript. C. Total MYBPC3 transcript abundance (left panel) and total MYH7 transcript abundance (right panel) in carriers of MYBPC3 (n=18) or MYH7 (n=6) mutations, compared to HCM samples in which no sarcomere mutation (n=7) was identified and control (non-HCM donor, n=7) hearts. *P=0.02 vs. control hearts, †P=0.003 vs. non-sarcomere HCM. D. RT-PCR analysis showing uniformly increased intensity of amplified MYBPC3 sequence in samples containing MYBPC3 mutations compared to the other 3 groups.

Mutations in exon splice sites could yield both missense and nonsense transcripts, depending on the extent to which the mutation alters splicing.⁷ Therefore, missense transcripts were analyzed using the single base extension reaction in samples with exon splice site mutations. A missense transcript for c.1624G>C (Glu542Gln) was present, in addition to the mutant splice variant transcript, and constituted 13 and 16% of total full-length MYBPC3 transcripts for each sample respectively (Figure 1B). However, no missense transcript was detected for either c.772G>A (Glu258Lys) sample, indicating that 100% of the transcripts from the mutant allele exist as truncating mutant transcripts. The Sequenom software was not able to design primers for the MYBPC3 mutation c.2308G>A (Asp770Asn). However, amplification of cDNA using primers in exons 22 and 24 yielded a single band containing only wild-type sequence (Supplemental Figure 1C). No band corresponding to the mutant transcript was detectable with use of a forward primer that spanned the junction of exon 22–24, suggesting that this particular mutation efficiently alters splicing (likely resulting in skipping of exon 23), but that the mutant PTC-containing transcript is undetectable due to instability. Less likely is the possibility that splicing is altered at a distant splice site not detected using these primers.

The ratio of mutant to wild-type transcript for MYBPC3 c.3742_3759dup (Gly1248_Cys1253dup) could not be precisely quantified because the wild-type transcript could not be uniquely amplified (addition of “A” after extension primer could be either wild-type or second repeat of mutant transcript). However, the mutant transcript was identified in a standard sequencing reaction (Supplementary Figure 2A) and constituted at least 30% of the total by single base extension (i.e. 30% of the extension product contained the first nucleotide of the duplicated sequence).

Using single base extension, ratios of all true missense mutations in MYBPC3, MYH7, MYL2 and TNNT2 were present at the expected ~1:1 ratio with wild-type transcripts (Figure 1B). The Sequenom software could not design primers for TPM1 c.850A>G (Ile284Val), perhaps because it is in the terminal codon. However, cDNA sequencing showed polymorphism at that site (Supplementary Figure 2B).

Overall, PTC-containing transcripts in MYBPC3 are significantly less abundant than transcripts from non-PTC-containing mutant alleles from either MYBPC3 or other sarcomere genes (mean 16 ± 3% vs. 47 ± 1% of the total respectively, P<0.00001), consistent with their recognized targeting for NMD. However, the degree of susceptibility to NMD appears to be highly variable given the wide range observed for the relative proportion of MYBPC3 truncating mutant transcripts.

Increased total MYBPC3 mRNA abundance in samples containing MYBPC3 mutations

To determine effects of mutant transcript instability on total MYBPC3 transcript abundance we performed RT-qPCR in samples from 4 groups: control donor hearts (n=7), HCM hearts from patients without sarcomere gene mutations (n=7), HCM hearts with MYBPC3 mutations (n=18), and HCM hearts with MYH7 mutations (n=6). Unexpectedly, mean total MYBPC3 transcript levels were increased in the MYBPC3 mutation group compared to non-failing donor hearts (p=0.02) and HCM without sarcomere mutations (p=0.003, Figure 1C, left panel). Increased MYBPC3 expression was consistent across all samples, with at least a 2-fold increase in total transcript level in each individual sample, including the 2 samples with missense mutations (Arg495Gln). This contrasts with MYH7 abundance for which no differences were observed among the groups (Figure 1C, right panel). The magnitude of increased MYBPC3 expression did not appear to correlate with the domain location of the mutation. MYBPC3 expression was highly variable in the samples with MYH7 mutations, but with no significant difference in the mean transcript level compared to controls or sarcomere negative HCM. These results were also supported by simple RT-PCR analysis of representative samples with a separate pair of PCR primers (Figure 1D). These results suggested that transcriptional upregulation of MYBPC3 compensates for degradation of the mutant transcript.

Comparable MYBPC3 protein levels between HCM samples with MYBPC3 mutations and control hearts

To determine if transcriptional upregulation of MYBPC3 in samples containing MYBPC3 mutations prevented haploinsufficiency at the protein level, full length MYBPC3 protein was detected by immunoblotting using an antibody targeting the N-terminus. None of the HCM groups (MYBPC3 mutations, non-MYBPC3 mutations, no sarcomere mutation) demonstrated a significant difference in MYBPC3 protein abundance compared to control hearts (Figure 2A and B). However, among the HCM groups, samples with mutations in genes other than MYBPC3, or without sarcomere gene mutations, had a significantly greater abundance of MYBPC3 protein in comparison to samples with MYBPC3 mutations (P<0.001). Other sarcomere proteins were similar in abundance among different HCM groups and controls (Supplementary Figure 3A). MYBPC3 protein levels were not significantly different between control heart samples and samples obtained from failing hearts due to non-ischemic or ischemic cardiomyopathies (n=14, Supplementary Figure 3B), suggesting that advanced heart failure itself does not alter MYBPC3 abundance.

Open in a separate window

Figure 2

MYBPC3 protein level. A. Representative western blot and B. Graph summary of densitometric analysis for MYBPC3 protein normalized to GAPDH and expressed as a % of the average of controls run on each gel. Results are shown for control, non-failing hearts (N=10), and HCM hearts without sarcomere mutations (N=14), with MYBPC3 mutations (N=22), and with sarcomere mutations other than MYBPC3 (N=10). *P<0.001 for HCM with MYBPC3 mutations vs HCM with or without other sarcomere mutations, †P<0.05 for HCM without sarcomere mutations vs controls and for HCM with non-MYBPC3 sarcomere mutations vs controls.

The presence of a MYBPC3 mutation did not appear to cause MYBPC3 mislocalization since cell fractionation experiments showed MYBPC3 to be present only in the myofilament fraction, and immunofluorescence imaging showed only normal localization of the protein in the sarcomere A-band (Figure 3A and B). Consistent with previous reports,^{8, 9} no truncated MYBPC3 proteins were detected by immunoblot. We also analyzed several samples by LC/MS/MS of a trypsin digest of a myofilament fraction in solution or gel-extracted fragments at the expected truncated molecular mass with and without extraction of insoluble proteins, but did not detect peptides containing the corresponding mutation (data not shown).

Open in a separate window

Figure 3

MYBPC3 localization. A. Representative western blot for MYBPC3 of samples containing various MYBPC3 mutations shows that all MYBPC3 protein is detected in the myofilament fraction, and none in the supernatant. B. Representative immunofluorescent images of formalin-fixed heart tissue from human hearts, reacted with antibodies to MYBPC3 (green) and α-actinin (red) showing correct localization of MYBPC3 to the A-b and of the sarcomere in a doublet pattern. This localization pattern was observed in all truncating and missense mutant samples examined.

Given the variability in mutant transcript stability, we considered the possibility of translation re-initiation that could stabilize truncating mutant transcripts. We probed samples containing truncating MYBPC3 mutations with an antibody raised against the C-terminal domain of MYBPC3. However, there were no unique bands containing MYBPC3 peptides detectable by LC/MS/MS (data not shown).

Variability in altered splicing for different MYBPC3 splice site mutations

A substantial proportion of mutations in MYBPC3 are located in splice site junctions. Correct predictions of whether and to what degree these mutations alter splicing has critical implications for disease mechanisms elicited by these mutations. Splice predictions by in silico models, in vitro splice reactions,¹⁰ or sequencing of peripheral blood leukocyte mRNA^{4, 11} may not accurately predict the consequences of a splice site mutation in the relevant tissue.¹² In this study, we confirmed altered splicing and generation of a PTC from myocardium of all HCM patients harboring MYBPC3 mutations in consensus splice sites. The same mutation present in different samples consistently produced the same splice defect, implying that the manner in which splicing is altered is inherent to a particular mutation. For all splice donor mutations, we confirmed previous predictions of exon skipping.⁴ However, both splice acceptor mutations demonstrated intron inclusion as opposed to exon skipping.^{10, 11} For the exon splice site mutation Glu542Gln, we demonstrated not only a splice-defective transcript, but also a full-length missense transcript that was not previously appreciated.¹³ Confirmation of the splice products for each mutation also prevented misclassification of mutations in exon splice sites as missense,^{8, 14, 15} with important implications for disease pathogenesis.

MYBPC3 transcript analysis supports variable efficiency of nonsense-mediated RNA decay for truncating mutant transcripts

Disease mechanisms may also be influenced in part by the relative instability of PTC-containing MYBPC3 mutant transcripts. In relation to wild-type transcripts, PTC-containing mutant MYBPC3 transcripts were markedly lower in abundance compared to non-PTC transcripts, consistent with previous reports^{8, 9} but with notable differences. First, the mean fraction of mutant transcripts was lower than in previous studies (16% vs 25–33%), and with considerably greater patient to patient variability (0–44% of the total), implying differential mutation-specific susceptibility to NMD. For most mutations, similar levels of truncating transcripts were observed for the same mutation studied in 2 unrelated HCM patient samples, suggesting that NMD efficiency is primarily intrinsic to a particular mutation. All PTCs were located more than 55 nucleotides upstream of the last intron and therefore all mutant transcripts would be expected to be targets for NMD.^{16, 17} Despite the tendency for PTCs situated more 5′ to elicit NMD more efficiently than those downstream,¹⁷ there was no obvious association between PTC location and abundance of the mutant transcript. Translation re-initiation can rescue a mutant transcript from NMD as observed in β-thalassemia¹⁸ and long QT syndrome.¹⁹ However, we could not identify any stable MYBPC C-terminal peptides that would implicate this mechanism as an explanation for the variability in mutant transcript stability.

Regardless of the mechanism, variable NMD efficiency has important consequences for disease pathogenesis and potential therapies. NMD can be beneficial in conditions where the mutant protein is toxic²⁰ or has a dominant-negative effect,²¹ but detrimental when the mutant protein retains partial functionality.^22–24 The role of NMD and mechanisms underlying variable susceptibility of individual MYBPC3 mutant transcripts will also be key determinants of the efficacy of future targeted therapeutics for HCM, such as PTC read-through strategies.⁶

Evidence for compensatory upregulation of MYBPC3 challenges haploinsufficiency disease model

Complete absence of the full-length MYBPC3 protein is clearly sufficient to cause severe dilated cardiomyopathy both in mice and in humans.^25–27 However, these observations do not necessarily implicate loss of functional protein as the primary mechanism underlying the heterozygous condition in HCM. While total MYBPC3 transcript abundance has not been previously assessed in human HCM samples, only a modest reduction in total MYBPC3 protein (~25–30%)^{8, 9, 28} or even an increase¹⁵, has been observed in MYBPC3 mutant samples compared to control hearts, reflecting at least a partial compensatory increase in wild-type protein expression. A similar modest reduction in total protein is observed in heterozygous MYBPC3 knock-out or knock-in mouse models, ~10–25%.^{7, 25, 29} In the present study comprising a larger number of HCM samples with unique mutations and control samples than in previous reports, MYBPC3 protein levels were no different in human HCM samples with MYBPC3 truncating mutations compared to control hearts. These data strongly challenge the hypothesis that haploinsufficiency of MYBPC3 is the primary mechanism of disease.

As an alternate hypothesis to haploinsufficiency, we suggest that the primary mechanism driving disease progression in MYBPC3-mutation HCM may be related to the burden of truncated mutant proteins. Even though the relative abundance of mutant transcripts is limited by NMD, the marked increase in total MYBPC mRNA predicts excess production of protein from both alleles. Two specific mechanisms are supported by previous investigations. First, truncated proteins are inherently misfolded and may induce cytotoxicity by overwhelming protein quality control systems. Truncated proteins derived from MYBPC3 truncating mutations are substrates for the proteasome, cause proteasome dysfunction and form cytosolic aggregates.^{7, 30, 31} Proteasome impairment is also evident in human HCM hearts, and is more pronounced in patients with sarcomere gene mutations.³² A second possible mechanism supported by experimental models is that truncated MYBPC3 may incorporate into the sarcomere inefficiently, causing disruption of myofibrillar architecture, functional cardiac impairment, heart failure and increased mortality.^33–35 The inability to identify truncated proteins in hearts with MYBPC3 mutations does not exclude these possibilities as low level or transient expression of truncated peptides could escape detection by available methods. Marked alterations in protein structure and folding could reduce binding affinity of mutant proteins for available MYBPC3 antibodies, even those recognizing N-terminal epitopes.

It is certainly possible that haploinsufficiency of MYBPC3 drives early cellular remodeling with other mechanisms manifesting once full-length protein levels are restored. This mechanistic transition has been previously reported in mice with a frameshift lamin A (LMNA) mutation, in which early remodeling was attributed to haploinsufficiency, while later disease progression occurred despite normalization of LMNA levels and was associated with accumulation of toxic aggregates and proteasome impairment.³⁶ Since our analyses utilize human samples with a mature form of the disease, we cannot rule out haploinsufficiency early in the disease course. It is also important to recognize that although MYBPC3 protein levels were the same in MYBPC3 mutant HCM samples as in control hearts, they were lower compared to HCM samples harboring other mutations or without sarcomere mutations. The significance of this finding is unclear at this time but merits further consideration. We would contend that haploinsufficiency and dominant negative effects of mutant proteins are not mutually exclusive and it is unlikely that a single mechanism can explain how different truncating mutations in MYBPC3 trigger disease onset and drive progressive remodeling over a human lifespan.

The marked increase in MYBPC3 mRNA abundance in MYBPC3 mutant samples compared to control hearts and HCM samples without sarcomere mutations in the present study may be driving the compensatory response at the protein level to maintain stoichiometry between full-length MYBPC3 and other myofilament proteins. Since the mechanisms that regulate MYBPC3 transcription, translation and protein turnover are largely unknown, it is difficult to speculate on the precise signals through which such a response may be occurring. However, it is possible that instability of mutant transcripts and/or proteins may be somehow involved. The discrepancy in the magnitude of increase in transcript to protein abundance can probably be explained by the tight regulation of sarcomere stoichiometry which precludes a significant over-abundance of any one component.^{37, 38} For example, in a transgenic mouse model expressing a mutant MYBPC3 truncated peptide, the mRNA level is 5-fold increased without any difference in MYBPC3 protein stoichiometry.³⁴

An increase in MYBPC3 transcript abundance was not limited to samples harboring MYBPC3 truncating mutations, but was also evident in 2 samples with MYBPC3 missense mutations and in myectomy samples with MYH7 mutations (2 transplanted hearts had decreased expression). This upregulation, unique to sarcomere HCM, reveals the complexity of MYBPC3 gene regulatory pathways in human HCM and warrants further exploration in model systems.

Funding Sources: NIH R01HL093338 (SMD), R01HL105826 and K02HL114749 (SS), private donation (Help4hcm.com)