Identification of biochemical adaptations in hyper- or hypocontractile hearts from phospholamban mutant mice by expression proteomics

Abstract

Phospholamban (PLN) is a critical regulator of cardiac contractility through its binding to and regulation of the activity of the sarco(endo)plasmic reticulum Ca²⁺ ATPase. To uncover biochemical adaptations associated with extremes of cardiac muscle contractility, we used high-throughput gel-free tandem MS to monitor differences in the relative abundance of membrane proteins in standard microsomal fractions isolated from the hearts of PLN-null mice (PLN-KO) with high contractility and from transgenic mice overexpressing a superinhibitory PLN mutant in a PLN-null background (I40A-KO) with diminished contractility. Significant differential expression was detected for a subset of the 782 proteins identified, including known membrane-associated biomarkers, components of signaling pathways, and previously uninvestigated proteins. Proteins involved in fat and carbohydrate metabolism and proteins linked to G protein-signaling pathways activating protein kinase C were enriched in I40A-KO cardiac muscle, whereas proteins linked to enhanced contractile function were enriched in PLN-KO mutant hearts. These data demonstrate that Ca²⁺ dysregulation, leading to elevated or depressed cardiac contractility, induces compensatory biochemical responses.

Cardiac muscle contraction is initiated by the entry of Ca²⁺ ions into the myoplasm from extracellular spaces and from stores in the sarcoplasmic reticulum (SR); relaxation is initiated by removal of myoplasmic Ca²⁺ by the combined activity of a sarco(endo)plasmic reticulum Ca²⁺ ATPase (SERCA2a), a plasma membrane Ca²⁺ ATPase, and a Na⁺/Ca²⁺exchanger (1). Of myocardial Ca²⁺, ≈70% fluxes through the SR in humans and ≈90% in mice. In both cases, a strong positive correlation exists between the size of the SR Ca²⁺ store and cardiac contractility (2). Thus the activity of SERCA2a is a crucial regulator of cardiac contractility and of the utilization of the cardiac reserve during periods of β-adrenergic stimulation of the heart (2).

Phospholamban (PLN), a 6-kDa transmembrane protein, regulates SR Ca²⁺ transport in cardiomyocytes by its ability to act as a reversible inhibitor of SERCA2a (3). The ability of PLN to determine cardiac contractility through its effects on the activity of SERCA2a has been demonstrated through the characterization of PLN-null mice, which are hypercontractile and healthy (4), and mice overexpressing either WT PLN (5) or superinhibitory mutant PLN molecules (6–8), which are hypocontractile and exhibit signs of cardiac hypertrophy and even heart failure. In contrast to mice, PLN-null humans manifest dilated cardiomyopathy (9), as do humans expressing chronically inhibitory forms of PLN (10). Substantial evidence also exists that the depressed contractility of failing human hearts results from deficiencies in basal Ca²⁺-signaling and in β-adrenergic stimulation (2, 11). In these cases, SERCA2a expression is diminished; the combination of lower SERCA2a levels and higher PLN-mediated inhibition of the remaining SERCA2a results in lower SR Ca²⁺ stores and diminished contractility.

It is likely that alterations in myocardial gene and protein expression underlie the pathophysiology of chronic cardiac dysfunction and determine the progression to an endpoint of heart failure. As one identified cause of progression to heart failure, alterations in PLN content or function are likely to induce compensatory changes in the composition of protein networks, especially those responsible for maintaining appropriate Ca²⁺-signaling transients within cardiomyocytes.

DNA microarrays have been developed for comprehensive screening of gene expression profiles, but the correlation between mRNA abundance and the quantity of functional protein is imperfect (12, 13). Hence, proteomic profiling procedures based on 2D gel electrophoresis of protein mixtures have been used toward the large-scale analysis of protein expression patterns associated with heart disease (14–16). A key limitation of this approach, however, is that membrane proteins do not resolve well in 2D gels and are poorly represented. Newer gel-free proteomic methods now offer the potential for characterization of the global protein composition of an entire tissue or organelle with significantly less detection bias (17). The use of gel-free HPLC, coupled online to highly sensitive electrospray tandem MS, makes it possible to measure alterations in the levels of even relatively low-abundance membrane proteins (18). All these approaches benefit directly from access to recently completed genomic sequences for humans (19) and mice (20).

Here, we describe the development and application of a high-performance gel-free expression-profiling methodology optimized for assessing the differential accumulation of membrane-associated proteins in cardiac tissue samples derived from mouse model systems. The approach is based on high-resolution multidimensional capillary-scale liquid chromatography-MS of purified cardiac membrane protein fractions and the use of peptide sequence coverage as a quantitative assay (21). Using this approach, we have carried out a comprehensive proteomic comparison of the protein composition of microsomal membrane fractions of cardiac tissue from mutant mice with a high level of contractility achieved by ablation of PLN (PLN-KO) (4) and a reduced level of contractility achieved by the transgenic overexpression of a superinhibitory, monomeric form of PLN in the PLN-KO background (I40A-KO) (4, 22). The study revealed differential accumulation of select functional categories of proteins in the cardiac muscle of the PLN-KO mutant as compared with the I40A-KO mutant, providing additional insight into pathophysiological changes associated with cardiac hypertrophy.

Experimental Procedures

Materials. Poroszyme bulk immobilized trypsin was purchased from Applied Biosystems. Endoproteinase Lys-C was obtained from Roche Diagnostics. SPEC Plus PT C18 solid-phase extraction pipette tips were purchased from Ansys Diagnostics (Lake Forest, CA). All other reagents were obtained from Sigma.

Mouse Breeding and Genotyping. Transgenic mice overexpressing the I40A PLN mutant (4) and phospholamban-null (PLN-KO) mice (6) were mated at least five generations to obtain I40A transgenic mice on a PLN-null background (I40A-KO). The strain of both sets of mice was 129SVJ:C57BL6. Age-matched littermate offspring were used in all experiments. FVB/N and SVJ129:C57BL6 mice were used as WT controls in cardiac functional analysis. In vivo left ventricular (LV) pressure hemodynamics were recorded from anesthetized animals by direct LV catheterization through the carotid artery, and M mode and Doppler echocardiography were performed noninvasively for assessment of LV function and dimensions, as described (6).

Microsomal Membrane Protein Fractions. Ventricular cardiac muscles from three individual isogenic mice were excised, combined, and homogenized in ice-cold buffer consisting of 10 mM Tris·HCl (pH 7.4), 150 mM KCl, 20 μM CaCl₂, 0.25 M sucrose, 2 mM DTT, and protease inhibitor (Complete; Roche Diagnostics). The extracts were centrifuged at 7,700 × g to remove insoluble debris, nuclei, and mitochondria, followed by ultracentrifugation at 100,000 × g for 60 min to isolate the membrane fraction. Ca²⁺ uptake assays were performed as described (6).

Western Blot Analysis. Protein fractions were separated by SDS/PAGE using standard procedures. Antibodies to SERCA2, RyR2 (Affinity BioReagents, Golden, CO), Na⁺-Ca²⁺ exchanger, plasma membrane Ca²⁺ ATPases (Research Diagnostic, Flanders, NJ), Integrin-β1, calreticulin, PKC-β, RACK-1, DGK-θ (BD Transduction Laboratories, Lexington, KY), and PKC-α, PKC-θ, and PKC-ε (Santa Cruz Biotechnology), were used, as recommended by the supplier. Signals were visualized by using horseradish peroxidase-conjugated goat anti-mouse secondary antibody and enhanced chemiluminescence (Super Signal, Pierce) and quantified by using a Fluor-S System and quantity one software (Bio-Rad).

Quantitative RNA Analyses. Total RNA was isolated from ventricles by using TRIZOL (Invitrogen), and 2 μg were blotted onto nitrocellulose filters by using a slot-blot filtration manifold. Filter preparation and hybridization to specific oligonucleotide probes, end-labeled with [γ-³²P]ATP. Hybridization signals were quantified by a Personal Molecular Imager FX System and quantity one software. The oligonucleotides used were: atrial natriuretic factor (ANF), 5′-AATGTGACCAAGCTGCGTGACACACCACAAGGGCTTAGGATCTTTTGCG, and β-myosin heavy chain (β-MHC), 5′-GGCTCCAGGTCTGAGGGCTTCACGGGCACCCTTAGAGCTGGGTAGCAC.

Digestion of Cardiac Microsomal Fractions. One hundred micrograms of ventricular microsomal proteins, as determined by Bio-Rad assay, were dissolved in 1% SDS, precipitated with 5 volumes of acetone at –20°C overnight, and centrifuged at 21,000 × g for 20 min to remove lipids. The pellets were solubilized in 0.5% SDS at 37°C for 2 h before dilution with an equal volume of 200 mM NH₄HCO₃. Endoprotease Lys-C was added to a final substrate-to-enzyme ratio of 50:1, and the mixture was incubated at 37°C for 15 h. An equal volume of 1 M guanidine-HCl was then added to the digests, and the samples were centrifuged at 21,000 × g for 10 min to remove SDS precipitates. The supernatant was diluted with an equal volume of 100 mM NH₄HCO₃ (pH 8.5), and CaCl₂ was added to a final concentration of 1 mM. The sample was further digested with Poroszyme-immobilized trypsin overnight at 30°C with continuous agitation. The resulting peptide mixtures were absorbed on SPEC-Plus PT C18 cartridges, washed, eluted, and stored at –80°C.

Liquid Chromatography Tandem MS. As reported by Florens et al. (21), the total cumulative amino acid coverage detected by MS provided an approximate quantitative measure of relative abundance between I40A-KO- and PLN-KO-derived tissue. To ensure the reliability of the measured sequence coverage, an equal amount of peptide (≈50 μg, based on total protein precipitated and digested) generated for each sample was analyzed by 2D capillary scale liquid chromatography coupled online to automated data-dependent electrospray ion-trap tandem MS, as reported by Yates and colleagues (23), with the adaptations described in Kislinger et al. (24).

Protein Identification and Informatics. Peptide fragmentation product ion mass spectra were sequence-mapped against a nonredundant set of human and mouse protein sequences obtained from the SWISS-PROT and TrEMBL (25) databases by using the sequest software algorithm (26) running on a multiprocessor computer cluster. A probability-based evaluation algorithm, statquest (24), was used for filtering of all putative matches based on a maximum P value threshold corresponding to a ≥90% likelihood of correct identification. The perl program goclust was used for automatic annotation and sorting of the positively identified proteins into specific functional categories by using the Gene Ontology (GO) annotation schema obtained from the European Bioinformatics Institute (25, 27).

Statistical Analysis. Data are presented as mean ± standard error. Statistical analysis was performed by Student's t test between WT and transgenic mice, with a P value of <0.05 considered to be significant.

Results

Functional Characteristics of I40A-KO and PLN-KO Mice. I40A-KO and PLN-KO littermates were compared directly, and FVB/N age-matched mice were used as WT controls in functional analyses. I40A-KO mice showed 1.8-fold overexpression of only monomeric PLN in the heart, whereas PLN-KO mice showed no PLN expression, based on Western blots using the anti-PLN monoclonal antibody, 1D11 (Fig. 1A). The apparent Ca²⁺ affinity (K_Ca) of SERCA2a in mouse ventricles, determined by measurement of the Ca²⁺ dependence of Ca²⁺ uptake, was 0.62 ± 0.03 μM for WT, 1.18 ± 0.05 μM for I40A-KO, and 0.18 ± 0.02 μM for PLN-KO mice (Fig. 1B); results were consistent with previous reports (4, 6).

Fig. 1.

Functional characteristics of I40A-KO and PLN-KO mice. (A) Western blot analysis of a microsomal fraction from ventricles of I40A-KO, PLN-KO, and WT mice demonstrating 1.8-fold overexpression of I40A PLN protein in I40A-KO transgenic mice and no expression of PLN in PLN-KO transgenic mice, with the anti-PLN monoclonal antibody ID11. p, PLN pentamer; m, monomer. (B) The apparent Ca²⁺ affinity of SERCA2a in microsomal fractions of ventricular tissue was determined by measurement of the Ca²⁺ dependence of Ca²⁺ transport. (C) LV +dP/dt and –dP/dt were measured by direct LV catheterization in vivo in I40A-KO, PLN-KO, and WT mice. *, P < 0.05 (a significant difference from WT).

To quantify cardiac contractility in I40A-KO, PLN-KO, and WT hearts, hemodynamic measurements and echocardiographic assessment in live animals were used, as described (6). Hemodynamic measurements of cardiac contractility and relaxation demonstrated that both +dP/dt and –dP/dt were significantly (P < 0.05) depressed in the hearts of I40A-KO mice compared with hearts of WT or PLN-KO mice (in I40A-KO, +dP/dt and –dP/dt were 3,256 ± 290 mmHg/s and 2,516 ± 422 mmHg/s; in WT, +dP/dt and –dP/dt were 4,242 ± 540 mmHg/s and 4,004 ± 368 mmHg/s; and in PLN-KO, +dP/dt and –dP/dt were 7,809 ± 967 mmHg/s and 4,753 ± 694 mmHg/s); results were consistent with previous reports (4, 6) (Fig. 1C).

Echocardiographic studies revealed that I40A-KO mice displayed mild cardiac hypertrophy, including significant increases in LV end-diastolic and systolic dimensions, significant reductions in LV fractional shortening and velocities of circumferential shortening, and a significant increase in LV weight and the LV weight-to-body weight ratio, when compared with WT (Table 1). By contrast, PLN-KO littermates exhibited a significant decrease in ejection time and an increase in velocity of circumferential shortening, but LV weight and LV weight-to-body weight ratios were similar between PLN-KO and WT mice (Table 1). These differences are consistent with impaired contractile function and cardiac hypertrophy in the I40A-KO hearts and support previous findings in mice with 2-fold overexpression of PLN I40A on a WT PLN background (6).

Table 1. Base-line echocardiographic measurements of 140A-KO, PLN-KO, and WT control mice.

Parameter	WT, n = 6	140A-KO, n = 8	PLN-KO, n = 6
HR, beats/min	397 ± 11.5	408 ± 16.3	385 ± 27.1
LV EDD, mm	3.58 ± 0.05	4.32 ± 0.46*	3.82 ± 0.38
LV ESD, mm	2.05 ± 0.05	3.35 ± 0.51*	2.29 ± 0.31
LV AW	0.65 ± 0.01	0.67 ± 0.03	0.62 ± 0.03*
LV PW, mm	0.64 ± 0.03	0.67 ± 0.03*	0.62 ± 0.03
LV mass, mg	77.16 ± 1.65	111 ± 19.97*	82.46 ± 16.30
LV/body mass, mg/g	2.67 ± 0.15	3.37 ± 0.53*	2.44 ± 0.53
FS, %	42.98 ± 1.83	22.97 ± 5.23*	40.50 ± 3.08
Ejection time, ms	78.35 ± 7.21	95.25 ± 11.10*	51.00 ± 8.69*
Vcfc, circ/s	4.71 ± 0.55	2.46 ± 0.63*	8.23 ± 2.02*

M mode and Doppler echocardiography were performed as described (6) and demonstrate LV hypertrophy in the 140A mouse. Values represent mean ± standard error. HR, heart rate; Vcf_c, velocity of circumferential shortening; EDD, end-diastolic dimension; ESD, end-systolic dimension; AW, anterior wall; PW, posterior wall; FS, shortening fraction. *, P < 0.05, significant difference from WT.

Gel-Free Protein Expression Profiling. To investigate global differences in the accumulation of proteins in I40A-KO and PLN-KO cardiac muscle, a high-throughput gel-free tandem MS-based methodology was used for systematic monitoring and identification of the numerous proteins present in heart microsomal protein fractions isolated from the mutant strains.

More than 200,000 uninterpreted peptide fragmentation product ion mass spectra were sequence-mapped against a nonredundant set of human and mouse protein sequences obtained from the SWISS-PROT and TrEMBL (25) databases by using the sequest software algorithm (26). A total of 782 proteins, including 170 known membrane-associated proteins, were identified collectively in the microsomal fractions from the two mouse lines. A detailed summary of the proteomics data is provided in Table 3, which is published as supporting information on the PNAS web site.

Each of these proteins was then mapped against the GO annotation database (24). Of the 782 proteins, 572 (73.1%) could be matched to one or more GO terms within the GO database, whereas 210 proteins could not be assigned to any defined GO term (26.9%), suggesting that they are functionally uncharacterized (Fig. 2A). The number of proteins assigned to a specific molecular function, biological process, or cellular component category within the GO database is shown in Fig. 2B. The subcellular localization of 432 proteins (55.2%) assigned to “cellular component” is shown in Fig. 2C. The assignment of many proteins to the endoplasmic reticulum (ER)/SR (29), Golgi (8), plasma membrane (30), or mitochondria (116), indicated that at least a portion of each of these membrane systems was present in our microsomal preparation. The number of membrane proteins identified by our gel-free proteomic method is more than an order of magnitude greater than that seen in typical experimental systems based on 2D gel separation of proteins (28). Indeed, nearly 130 of these proteins were predicted to contain one or more transmembrane domains by TMpred (www.ch.embnet.org/software/TMPRED_form.html; Fig. 2D).

Fig. 2.

Distribution of the identified proteins. (A) Fraction of the identified proteins that could be linked to an annotation term within the GO schema. (B) Number of proteins linked to the three main GO subcategories. (C) Subcellular localization of 432 proteins that mapped to the GO category “cellular compartment.” (D) Number of proteins predicted to have one or more transmembrane domains.

Differential Expression of Membrane and Membrane-Associated Proteins in I40A-KO Cardiac Tissue Relative to PLN-KO Mutant Hearts. Select subsets of the proteins detected were expressed differentially in the hypertrophic I40A-KO animals relative to the PLN-KO mutant strain. Our proteomic analysis detected increased levels of ANF and β-MHC, both well established markers of hypertrophy (29) in cardiac tissue from the I40A-KO line, as compared with the PLN-KO strain. We verified these observations by RNA slot-blot analysis, which showed that significantly higher levels of ANF and β-MHC transcripts were expressed in the I40A-KO line (Fig. 3A). The levels of other known markers of cardiomyopathy, including plasma membrane Ca²⁺ ATPase, ryanodine receptor (RyR), and integrin-β1, were also increased in I40A animals, with no significant changes in SERCA2, Na⁺/Ca²⁺exchanger 1, or calreticulin levels. These results were confirmed by Western blot analysis (Fig. 3B). They are consistent with the results of previous studies of PLN transgenic mice, and other models of cardiac hypertrophy, which showed that SERCA2 protein levels do not change (8) and that RyR protein levels are perturbed by changes in PLN expression (4). These results validate our proteomic methods and confirm the hypertrophic state of these animals (Fig. 1).

Fig. 3.

Adaptive levels of transmembrane and membrane-associated proteins in I40A-KO ventricular muscle. (A Left) Data obtained from the proteomic analyses indicating undetectable (n.d.) levels of ANF and β-MHC in PLN-KO hearts, with significant protein levels in I40A-KO cardiac muscle. (A Right) RNA slot-blot analysis demonstrating elevated transcript levels of ANF and β-MHC in the I40A-KO mice. (B Left) Relative protein levels (as reflected by total sequence coverage) detected by MS. (B Right) Consistent with the proteomic data, Western blot analysis of ventricular membrane protein fractions from I40A-KO and PLN-KO mice demonstrating increased expression of plasma membrane Ca²⁺ ATPase, RyR2 (RYR), and integrin-β1, but no change in the levels of SERCA2, Na⁺/Ca²⁺exchanger (NCX1), or calreticulin (CRT).

We grouped the remaining differentially expressed proteins into annotation categories. The data for a select subset is shown in Table 2. Several of the results are intriguing. First, the I40A transgenic tissue showed increased levels of ER protein-sorting molecules, such as endoplasmin, SEC22, and COP-coated vesicle protein, whereas the levels of other stably resident ER/SR proteins, such as SERCA2, calsequestrin, calumenin, and calreticulin, were not perturbed.

Table 2. Functional grouping of differentially expressed proteins.

Protein	Accession no.	140A-KO	PLN-KO
SR/ER associated proteins
Sarcoplasmic/endoplasmic reticulum Ca2+ ATPase	ATA2 MOUSE	23	23.7	—
Cardiac Ca2+ release channel	Q9ERN6	5.2	2.4	▴
Calsequestrin, cardiac muscle isoform	CAQC MOUSE	21.9	14.2	—
Calreticulin	CRTC MOUSE	11.5	14.7	—
FKBP-rapamycin-associated protein	FRAP MOUSE	1	1	—
Cop-coated vesicle membrane protein p24	P24 MOUSE	6	3	▴
Calcium/calmodulin-dependent protein kinase II	Q92991	10.8	10.8	—
Transitional endoplasmic reticulum ATPase	TERA MOUSE	8.3	2.9	▴
SEC22, vesicle trafficking protein-like 1	Q91VU3	3.3	0	▴
Endoplasmin	ENPL MOUSE	9.2	0	▴
Calumenin	CALU MOUSE	8.3	3.2	▴
Calnexin	CALX MOUSE	11.7	11.3	—
Intracellular signaling cascade
Tyrosine-protein kinase BLK	BLK HUMAN	2.4	0	▴
Guanine nucleotide-binding protein, α-11	GB11 MOUSE	3.3	0	▴
Guanine nucleotide-binding protein G, α-2	GBI2 MOUSE	6.5	0	▴
Guanine nucleotide-binding protein β subunit	GBLP HUMAN	7.6	0	▴
Guanine nucleotide-binding protein G, α	GBQ MOUSE	3.4	0	▴
Myosin-IXa	Q9UNJ2	0.2	0	—
Multiple PDZ domain protein	Q9Z1K3	2	2	—
Ras-related protein RAL-A	RALA MOUSE	9.7	0	▴
Ras-related protein RAP-1A	RAPA HUMAN	12.5	6.5	▴
Transforming protein P21/H-RAS-1	RASH MOUSE	5.8	0	▴
Ras-related protein Rab-10	RB10 HUMAN	6	0	▴
Ras-related protein Rab-1A	RB1A HUMAN	8.8	0	▴
Ras-related protein Rab-21	RB21 MOUSE	24.5	14.3	▴
Ras-related protein Rab-5B	RB5B HUMAN	5.6	0	▴
Ras-related protein R-Ras	RRAS MOUSE	5	0	▴
Superoxide dismutase [Cu-Zn]	SODC MOUSE	30.1	17	▴
Cell surface receptor-linked signal transduction
Adenylate kinase iscenzyme 1	KAD1 MOUSE	26.3	4.1	▴
Glucagon receptor precursor	GLR HUMAN	1.7	0	▴
Integrin β-1 precursor	ITB1 MOUSE	4.1	0	▴
Carbohydrate metabolism
Fructose-bisphosphate aldolase A	ALFA MOUSE	7.7	3.3	▴
α enolase	ENOA MOUSE	2.5	0	▴
Phosphoglycerate kinase 1	PGK1 MOUSE	8.4	0	▴
Phosphoglycerate mutase 2	PMG2 MOUSE	4.4	0	▴
Triosephosphate isomerase	TPIS MOUSE	13.7	0	▴
Fatty acid regulation
Fatty aldehyde dehydrogenase	DHA4 MOUSE	2.9	0	▴
Fatty acid-binding protein, heart	FABH MOUSE	49.2	0	▴
Long-chain fatty acid transport protein precursor	FATP MOUSE	2.2	2.2	—
Apolipoprotein A-1 precursor	APA1 MOUSE	26.1	12.5	▴
Very-low-density lipoprotein receptor precursor	LDVR MOUSE	2.3	0	▴
Ion transport
Sarcoplasmic/endoplasmic reticulum Ca2+ ATPase	ATA2 MOUSE	23	23.7	—
Sodium/calcium exchanger 1 precursor	NAC1 MOUSE	5.4	5.7	—
ATPase, Na+K+ transporting, α1 polypeptide	Q8VDN2	12	9.3	▴
Cardiac Ca2+ release channel	Q9ERN6	5.2	2.4	▴
Plasma membrane calcium-transporting ATPase 1	ATB1 HUMAN	1.1	0	▴
Inward rectifier potassium channel 2	IRK2 MOUSE	5.6	2.1	▴
Voltage-dependent anion-selective channel	POR1 MOUSE	50.3	66.6	—
Voltage-dependent anion-selective channel	POR2 MOUSE	34.9	25.8	—
Voltage-dependent anion-selective channel	POR3 MOUSE	33.6	30	—
Cytoskeleton
Actin, cytoplasmic 1	ACTB HUMAN	7.7	4.8	▴
Desmin	DESM MOUSE	22.4	12.4	▴
Ezrin	EZRI MOUSE	3.1	0	▴
Similar to ARP1 actin-related protein 1 homolog B	Q8R5C5	0	6.1	▾
Cytovillin 2	Q9UJU1	0	10.6	▾
β-sarcoglycan	SGCB MOUSE	13.8	0	▴
Spectrin α chain, erythrocyte	SPCA MOUSE	7.1	0	▴
Spectrin β chain, erythrocyte	SPCB MOUSE	3.5	0	▴
Spectrin α chain, brain	SPCN MOUSE	6.5	5.5	—
Spectrin β chain, brain 1	SPCO MOUSE	4.1	0.9	▴
Vesicle-associated membrane protein 2	VAM2 MOUSE	14.8	0	▴
Vesicle-associated membrane protein 3	VAM3 MOUSE	16.5	0	▴
Vimentin	VIME MOUSE	15.3	2.8	▴
Class 1 β-tubulin. Tubulin β-5 chain	TBBX HUMAN	16	12.8	—
Developmental processes
α-2-HS-glycoprotein precursor	A2HS MOUSE	0	6.1	▾
Annexin II	ANX2 HUMAN	0	5.3	▾
Myosin light-chain alkali, nonmuscle isoform	MLEN MOUSE	17.7	6.4	▴
Myosin light-chain 1, slow-twitch/ventricular	MLEV MOUSE	0	45.1	▾
Myosin regulatory light chain 2, ventricular isoform	MLRV MOUSE	15.2	7.3	▴
Myosin heavy chain, last skeletal muscle,	MYH3 MOUSE	14.5	0	▴

Differential regulation of select subsets of functionally related proteins was found in the ventricular membrane fractions of 140A-KO transgenic mice as compared with PLN-KO mutant mice. Protein-abundance ratios are based on the percent peptide coverage obtained in each protein expression and were obtained by averaging three experiments in pooled ventricular tissues. Each protein was classified to a specific annotation category according to the GO schema. For a more complete listing of the proteomic data, see Table 3. ▴, up-regulated, ratio of 140A-KO/PLN-KO ≥ 2; ▾ down-regulated, ratio of 140A-KO/PLN-KO ≤ 2; —, no change.

Strikingly, the levels of proteins linked to intracellular signaling were up-regulated in the I40A mutant animals. These included numerous guanine nucleotide-binding proteins (e.g., G protein α subunit Gq/G11) and Ras-related proteins (e.g., R-Ras, Rab-10, and RAP-1A) (Table 2). These data are consistent with compensatory activation of G protein-dependent and Ras-related signaling pathway(s) in the hypertrophic I40A mutant hearts (11, 30). These pathways may be linked to cytoskeletal reorganization (11, 30). Consistent with this, we also observed marked changes in the levels of cytoskeletal proteins, such as actin, desmin, vimentin, spectrin, and tubulin in the I40A heart, changes consistent with pernicious growth (Table 2).

The abundance of certain structural/contractile proteins were also altered in the I40A-KO heart tissue relative to the PLN-KO, including elevated levels of embryonic myosin heavy chains and reduced levels of adult ventricular myosins (Table 2), consistent with the reversion of adult cardiac tissue to the fetal patterns of expression observed during cardiac hypertrophy (29). The I40A-KO animals showed increased levels of proteins linked to carbohydrate metabolism, such as the glycolytic enzymes α-enolase, triosephosphate isomerase, fructose-biphosphate aldolase A, and proteins linked to fat metabolism, including fatty acid-binding protein, apolipoprotein A, very-low-density lipoprotein receptor, and fatty aldehyde dehydrogenase, consistent with substantive physiological remodeling of cell metabolism (Table 2).

Activation Profile of PKC in I40A-KO. Activation of the G proteins Gq/G11 is normally associated with a characteristic pattern of PKC isoform activity (11, 30). Because the proteomic data predicted activation of G protein-signaling cascades in the I40A mutant heart, we examined the levels of PKC in cardiac muscle tissue by Western blot analysis. We probed both cytosolic and microsomal fractions because PKC translocation into membranes is a key feature of PKC activation (11). The levels of both membrane-associated and soluble cytosolic fractions of PKCα, PKCβ, PKCθ, and RACK1 were all increased significantly in the I40A-KO mice relative to PLN-KO (Fig. 4). By contrast, the cytosolic fraction of PKCε was increased significantly in I40A-KO hearts. These observations suggest that the activation of PKC is closely linked to the development of cardiomyopathy and the emergence of cardiac hypertrophy.

Fig. 4.

Activation profile of PKC in I40A-KO hearts. Western blot analysis of PKC-α, PKC-β, PKC-ε, PKC-θ, and RACK-1 in both cytosolic and membrane protein fractions. Each band reflects fractions isolated from three pooled mutant hearts. The experiment was performed in duplicate.

Discussion

Physiological features of end-stage heart failure include decreases in the activity of SERCA2a, in the size of SR Ca²⁺ stores, in the amplitude of depolarization-induced Ca²⁺ transients, and in cardiac contractility (2). Thus, the potential role of Ca²⁺ regulatory proteins in human congestive heart failure is of particular mechanistic interest. However, changes in the function of a single, critical regulatory protein are likely to result in adaptive changes in the expression and activity of an entire network of functionally related proteins, ultimately leading to systemic biochemical remodeling of the heart and progression to heart failure. In this respect, knockout and transgenic animals that develop heart failure provide the opportunity for investigation of a series of alterations in signaling pathways that may lead to a single endpoint, such as dilated cardiomyopathy. With this in mind, we have embarked on a proteomic screening approach to identify global perturbations in cardiac protein expression patterns that occur in mice exhibiting different degrees of contractility, some of which progress to cardiac hypertrophy.

To date, studies describing protein expression profiles in the pathogenesis of heart disease have focused mainly on soluble cytosolic proteins and have used 2D gel electrophoresis (14, 15). Characteristically, these studies lack comprehensive data on membrane-bound proteins. Our study describes the development and successful application of a robust and sensitive gel-free method for the direct identification and assessment of changes in the levels of membrane-associated proteins in heart tissue from animals exhibiting differential contractility. This form of proteome mapping circumvents many of the technical limitations associated with gel-based proteomics technology (31) and enables comprehensive characterization of the global protein components of isolated membranous fractions (23, 24).

We have started with two well characterized models of dysregulation of Ca²⁺ signaling: the hypercontractile PLN-null mouse and the PLN superinhibitory transgenic model that develops mild cardiac hypertrophy. These animals were chosen because they represent polar extremes of cardiac contractility in mice, neither of which progress to heart failure. Mice lacking PLN display an augmentation of both cardiac contractility and rate of relaxation, without any long-term pathological effects (5, 32). In contrast, the I40A-KO null mutant mice exhibit ventricular dilation and decreased contractile function, as observed by echocardiography (Table 1), and impairment of force development and relaxation, as assessed by hemodynamic measurements. They also develop mild cardiac hypertrophy, manifested by a statistically significant increase in LV mass (Table 1), and up-regulation of embryonic and contractile protein genes such as ANF and β-MHC, as confirmed by MS (Fig. 3).

Within this set of differentially expressed proteins, we found altered abundance of a number of components of specific G protein-signaling pathways in I40A-KO cardiac muscle, as compared with PLN-KO mice. The accumulation of Gq/11 and RACK1 detected in I40A-KO (Table 2 and Fig. 4) suggests that activation of Gq/G11/PKC-dependent signaling pathways occurs during impaired cardiac function in the I40A-KO heart. The increased levels of PKCα, PKCβ, PKCε, PKCθ, and RACK1 in both cytosolic and membrane fractions in the I40A-KO heart (Fig. 4) are consistent with this model. This finding is similar to previous reports on the response of cultured cardiomyocytes to agonist binding to Gq/G11 protein-coupled receptors (33, 34) and in vivo transgenic overexpression of Gq (35, 36) that also implicate PKC in the development of cardiac hypertrophy. However, because we examined only one subcellular fraction and did not carry out a global proteomic assessment of all other cellular fractions, interpretation of these data must be considered tentative.

The significant changes observed in cytoskeletal, structural, and contractile proteins in the I40A heart are consistent with a tissue that is undergoing substantial structural remodeling, as would be expected in a hypertrophic heart. Indeed, increases in protein levels of collagens, integrins, myosin light chains, and tubulins have all been documented in mouse or human models of cardiac hypertrophy (37, 38). Our ability to detect changes in the levels of these proteins (Table 2) is entirely consistent with previous reports and further validates our proteomics analyses.

The apparent changes in mitochondrial components, including proteins involved in carbohydrate and fat metabolism, indicate substantial changes in energy metabolism in the I40A heart. Again, such changes would be anticipated given extensive documentation in other models of cardiac hypertrophy (39). However, because these proteins were detected in a preparation that is enriched for microsomes, and not mitochondria, it will be particularly appropriate to confirm these changes in isolated mitochondrial fractions.

In summary, we have successfully adapted a gel-free MS procedure for the investigation of cardiomyopathy. The most significant advantage of such an approach is the ability to investigate membrane proteins, a limitation of gel-based systems. This investigation is an example of applying emerging mass spectrometric technology to the proteomic analysis of heart disease. Continuation of these studies through the careful investigation of additional models of heart disease and other subcellular compartments, in particular, mitochondrial, cytosolic, and nuclear fractions, and further refinements to the quantitative nature of the procedure should provide additional insight into the pathophysiological mechanisms of cardiac hypertrophy leading to heart failure.