Mutant Muscle LIM Protein C58G causes cardiomyopathy through protein depletion

Abstract

Cysteine and glycine rich protein 3 (CSRP3) encodes Muscle LIM Protein (MLP), a well-established disease gene for Hypertrophic Cardiomyopathy (HCM). MLP, in contrast to the proteins encoded by the other recognised HCM disease genes, is non-sarcomeric, and has important signalling functions in cardiomyocytes. To gain insight into the disease mechanisms involved, we generated a knock-in mouse (KI) model, carrying the well documented HCM-causing CSRP3 mutation C58G. In vivo phenotyping of homozygous KI/KI mice revealed a robust cardiomyopathy phenotype with diastolic and systolic left ventricular dysfunction, which was supported by increased heart weight measurements. Transcriptome analysis by RNA-seq identified activation of pro-fibrotic signalling, induction of the fetal gene programme and activation of markers of hypertrophic signalling in these hearts. Further ex vivo analyses validated the activation of these pathways at transcript and protein level. Intriguingly, the abundance of MLP decreased in KI/KI mice by 80% and in KI/+ mice by 50%. Protein depletion was also observed in cellular studies for two further HCM-causing CSRP3 mutations (L44P and S54R/E55G). We show that MLP depletion is caused by proteasome action. Moreover, MLP C58G interacts with Bag3 and results in a proteotoxic response in the homozygous knock-in mice, as shown by induction of Bag3 and associated heat shock proteins. In conclusion, the newly generated mouse model provides insights into the underlying disease mechanisms of cardiomyopathy caused by mutations in the non-sarcomeric protein MLP. Furthermore, our cellular experiments suggest that protein depletion and proteasomal overload also play a role in other HCM-causing CSPR3 mutations that we investigated, indicating that reduced levels of functional MLP may be a common mechanism for HCM-causing CSPR3 mutations.

Keywords: Hypertrophic cardiomyopathy; In vivo phenotyping; MLP C58G mutation; Mouse knock-in model; Proteasome; Protein depletion; RNAseq transcriptome analysis; Sarcomere.

Fig. 1

Cardiac phenotype of KI/+ and KI/KI mice in comparison to WT mice. A – Characterisation of cardiac dimensions and function by echocardiography: Fractional shortening and end-diastolic dimensions are shown. For cohort characteristics and a wider set of echocardiographic parameters please refer to Table S1. B – Heart weight is increased in KI/KI mice, values are normalised to tibial length. ****p < 0.0001 versus WT. C – Invasive haemodynamic assessment of LV performance of the three genotypes: dP/dt_max, dP/dt_min and relaxation constant Tau are shown at baseline conditions, and under adrenergic stress (dobutamine infusion at 4 ng g⁻¹ BW min⁻¹ and 16 ng g⁻¹ BW min⁻¹). Where error bars are missing, they are smaller than the symbols. For cohort characteristics and a wider set of parameters refer to Table S2.

Fig. 2

Molecular phenotype of KI/+ and KI/KI hearts. A, B – Transcriptome profiling by RNAseq: A – KEGG Pathway analysis: The top five statistically (p < 0.001) enriched GO terms for KI/KI hearts when compared to WT hearts are shown. B – Heat map shows genes in the KEGG HCM gene set with the strongest up-regulation in the KI/KI hearts at the top and the genes showing the strongest down-regulation at the bottom. Colours range from dark red to dark blue representing respectively the highest and lowest expression of a gene. C – Assessment of transcriptional changes by qPCR for genes related to the fetal gene programme (top), hypertrophic signalling (middle) and fibrosis (bottom). All measurements are normalised to Gapdh; significant changes are observed in the hearts of KI/KI mice; *p < 0.05, ***p < 0.001, ****p < 0.0001 versus WT, n = 6 per group. D – Changes in hypertrophic signalling proteins assessed by Western blotting: Fhl1, Carp (Ankrd1) and Dscr1 (Rcan1) are induced in KI/KI hearts, in line with the activation of hypertrophic signalling pathways. Anti-hypertrophic Fhl2 is down-regulated in these hearts (for quantification see Fig. S5B). Gapdh serves as loading control. The position of marker proteins is indicated (molecular weight in kDa). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 3

MLP depletion in KI/KI hearts. A – Measurement of Csrp3 transcript by qPCR (left) and MLP protein (right) in the three groups, normalised to Gapdh transcript and protein, respectively. In the KI/KI mice, Csrp3 transcript is 2.5-fold up-regulated. The abundance of MLP decreased in KI/KI mice by 80% and in KI/+ mice by 50%. MLP level is down-regulated to 50% in KI/+ mice and to 20% in KI/KI mice. ***p < 0.001, ****p < 0.0001 versus WT, n = 6 per group. B – Representative Western blot demonstrating reduced MLP levels in KI/+ and KI/KI mice. Gapdh serves as loading control. The position of marker proteins is indicated (molecular weight in kDa). N.B., both TaqMan probe target site and epitope of the anti-MLP antibody 79D2 are not overlapping with the site of the C58G mutation. C – Immuno-fluorescence on isolated adult cardiomyocytes from mice of the three genotype groups. Cells are stained for MLP (left), titin M-band epitope m8 (middle) and merged images are shown on the right (MLP red, titin m8 green). Scale bar represents 20 μm. Insert: Magnified area, scale bar represents 5 μm.

Fig. 4

The Csrp3 C58G allele cannot rescue the phenotype of Csrp3-/- knock out mice. A – Representative examples of M-mode echocardiography of Csrp3+/- mice with normal systolic function (top), Csrp3 KI/- mice (middle) and Csrp3-/- mice (bottom), the latter both with impaired systolic function. The calculated value of LV fractional shortening (FS) is given for each mouse. B – Cardiac morphology of the genotypes. Haematoxylin-eosin stained paraffin heart sections are shown.

Fig. 5

Protein depletion is a hallmark of HCM-associated MLP mutations. A – Recombinant HA-tagged MLP WT, HCM-associated MLP mutants (L44P, S44R/E54G, C58G), a DCM-associated MLP mutant (K69R) and empty parental vector were expressed in HEK cells, by transfection with a bicistronic vector system (see Section 1.6 of Material and Methods in Appendix A Supplementary data). Cell lysates were blotted for MLP (using either anti-HA antibody against the N-terminal tag, light grey, or anti-MLP antibody 79D2 against the C-terminus of the protein, dark grey) and hrGFP. In addition, CSRP3 and hrGFP transcripts were measured by qPCR in parallel transfected cells (black bars). The ratios of CSRP3 transcript or MLP protein to hrGFP were set to 100% for WT transfections and all other samples expressed relative to WT, n = 3 per group. Note that all HCM-associated mutants are affected by protein depletion, while the DCM-associated mutation is not affected. B – NRC were transduced with adenoviral particles coding for MLP WT, L44P, C58G and empty vector (see A) and cell lysates were blotted for MLP (using antibody 79D2) and hrGFP. The HA-tag on the recombinant MLP allows separation of this protein from endogenous rat MLP by size. Pan-actin serves as a loading control. The position of marker proteins is indicated (molecular weight in kDa). C – Quantification of Western blots from panel B (grey bars) and measurement of CSRP3 and hrGFP transcripts by qPCR to confirm equal transduction efficiency (back bars). Data are normalised as in panel A. The TaqMan probe is specific for the recombinant (human) MLP constructs and does not recognise the endogenous rat Csrp3 transcript.

Fig. 6

HCM-associated MLP mutant proteins are subjected to depletion by proteasomal pathways: A – NRC were transduced with adenoviral particles as in Fig. 5B and treated with proteasomal inhibitors MG-115 and MG-132. Cell lysates were blotted for HA, detecting the recombinant MLP WT, L44P and C58G. Pan-actin served as loading control and blotting for ubiquitin confirmed efficient inhibition of the proteasome by accumulation of ubiquitinated proteins. In control (ctr, DMSO-treated) cells, MLP L44P and C58G are destabilised, however they are restored to WT protein levels upon proteasomal inhibition. B – TUBE assay: Ubiquitinated proteins were pulled down from Csrp3 KI/KI heart lysate (treated in vivo with proteasomal inhibitor MG-262 1 μmol/kg BW for 20 h) using immobilised tandem ubiquitin-binding entities (TUBE). Agarose matrix without TUBE served as control (ctr). Lysate controls are shown (LC, 1% of input). Blotting for ubiquitin (ubi) indicates the enrichment of ubiquitinated proteins in the TUBE pulldown. Blotting for MLP shows specific pulldown of MLP with TUBE, indicating ubiquitination of the protein. Please note the higher molecular weight (>20 kDa) than normal endogenous MLP (< 20 kDa, see panel C). C – Proteasomal overload in the KI/KI mice: The ubiquitin G76 V GFP reporter (GFP-ub) was crossed onto mice of the three genotypes. Samples of hearts of 6 month old mice were blotted for GFP and increased accumulation of GFP-ub was observed in the KI/KI mice, indicating proteasomal overload. Blotting for MLP illustrates the genotype of the mice and Gapdh served as loading control. The position of marker proteins is indicated (molecular weight in kDa). D – Quantification of GFP-ub induction (blots relating to panel C), expressed relative to Gapdh (left) or MLP (right) protein levels, n = 6 per group. A striking induction of the GFP-ub reporter is observed in KI/KI hearts; **p < 0.01, ***p < 0.001.

Fig. 7

Proteotoxic stress response in the KI/KI mice: A – Co-immunoprecipitation of MLP and Bag3 from KI/KI heart lysate (see Fig. 6B). MLP C58G was precipitated using anti-MLP 79D2 antibody (IP) or isotype antibody control (ctr). Lysate controls are shown as detection controls (LC, 1% of input); * indicates signal from antibody chains. Bag3 co-precipitates with MLP C58G, evidencing a complex formation between the endogenous proteins. The position of marker proteins is indicated (molecular weight in kDa). B – In the KI/KI mice with proteasomal overload, the protein quality control complex Bag3 - Hsp70 – Hsc70 is induced and small heat shock proteins Hsp27, HspB7 and αβ-crystallin (ABcryst) are upregulated. Please note, the blots for MLP and loading control Gapdh on the same membrane are already shown in Fig. 3B. The position of marker proteins is indicated (molecular weight in kDa). C – Left: Assessment of transcriptional changes by qPCR for genes related to protein quality control. All measurements are normalised to Gapdh (n = 6 per group); significant changes are observed in the hearts of KI/KI mice for all transcripts apart from Hsc70 and Hsp70. Right: Quantification of protein levels of proteins involved in protein quality control (blots of panel 7B). All measurements are normalised to Gapdh (n = 4 per group); significant changes are observed in the hearts of KI/KI mice for all proteins apart from αβ-crystallin. **p < 0.01, ***p < 0.001, ****p < 0.0001 versus WT.