Activation of Ca transport in cardiac microsomes enriches functional sets of ER and SR proteins

The importance of sarcoplasmic reticulum (SR) Ca-handling in heart has led to detailed understanding of Ca-release and re-uptake protein complexes, while less is known about other endoplasmic reticulum (ER) functions in the heart. To more fully understand cardiac SR and ER functions, we analyzed cardiac microsomes based on their increased density through the actions of the SR Ca-ATPase (SERCA) and the ryanodine receptor that are highly active in cardiomyocytes. Crude cardiac microsomal vesicles loaded with Ca oxalate produced two higher density subfractions, MedSR and HighSR. Analyses of protein enrichments from the 3 membrane preparations (crude microsomes, MedSR, and HighSR), showed that only a third of microsomal proteins in heart, or 354 proteins, were enriched ≥2.0-fold in SR. Previously studied SR proteins were all enriched, as were proteins associated with canonical ER functions. Contractile, mitochondrial, and sarcolemmal proteins were not enriched. Comparing the levels of SERCA-positive SR proteins in MedSR versus HighSR vesicles produced a range of SR subfraction enrichments signifying differing levels of Ca leak (ryanodine receptor) co-localized in the same membrane patch. All known junctional SR proteins were more enriched in MedSR, while canonical ER proteins were more enriched in HighSR membrane. Proteins from other putative ER/SR subdomains also showed characteristic distributions among SR subpopulations. We conclude that active Ca loading of cardiac microsomes, reflecting the combined activities of Ca uptake by SERCA, and Ca leak by RyR, permits evaluation of multiple functional ER/SR subdomains. Sets of proteins from these subdomains exhibited similar enrichment patterns across membrane subfractions, reflecting the relative levels of SERCA and RyR present within individual patches of cardiac ER and SR.


Introduction
The basic morphology of the endoplasmic reticulum (ER) in muscle has been known since the initial electron microscopic studies in the 1950's by Bennett and Porter [1] and Palade and Porter [2], when it was termed the sarcoplasmic reticulum (SR) 1 , given its unique structure that repeats across all sarcomeres in the myocyte. Studies over many decades have focused on its Ca handling properties, beginning with extensive characterization of the important Ca translocating ATPase activity, now termed SERCA, for SR, ER-Ca-ATPase [3,4]. Much later, a ryanodine-sensitive Ca-releasing activity was identi ed [5][6][7]. Extensive analyses of Ca transport in membrane preparations from skeletal muscle also led to established membrane subfractionation protocols, and isolation of membranes of distinct densities that contained different protein composition. Meissner [8] reported the rst fractionation of skeletal muscle membranes by isolating a fraction of heavy vesicles that contained the electron-opaque protein polymer calsequestrin [9], and later shown to contain foot processes consistent with morphological depictions of junctional SR [10]. The lighter membranes showed less of the dense luminal protein content, and was concluded to contain free (non-junctional) SR vesicles [8,9].
A technique for subfractionation of microsomes from heart tissue was later developed that took advantage of the ability of cardiac SR microsomes to concentrate Ca oxalate within their lumens [11][12][13]. While these studies were originally developed to biochemically separate sarcolemma and SR, subsequent work [5] showed that the Ca oxalate loading had actually produced two distinct types of SR membranes, with roughly half of membrane protein in the densest membrane fraction. And though it was known by then that the drug ryanodine could stimulate Ca accumulation into SR membrane vesicles at high concentrations, these dense membranes were ryanodine insensitive [5].
In contrast, a second membrane fraction of lower density (unable to traverse a 1.5 M sucrose cushion) was highly sensitive to ryanodine [5]. Indeed, by inhibiting Ca leak, 0.3 mM ryanodine added during membrane isolation converted SR membranes of lower density into high density (high Ca oxalate) vesicles [14]. The densities of SR membrane vesicles, as well as the distribution of SR proteins between the two subfractions, are regulated by the relative levels of SERCA and ryanodine receptor (RyR) contained in the fragmented SR patches.
Several studies have used SERCA-positive membrane subpopulations to either validate the identity of a putative junctional SR protein [5,[15][16][17], or non-junctional SR protein [14,[18][19][20][21]. In this study, we determined the complete protein compositions of membrane vesicles generated by SERCA-depending Ca oxalate loading. We found 354 proteins that were enriched as a result of SERCA activation. In addition, a co-enrichment of functionally related proteins resulted from their similar distribution among membrane patches, and re ect the enrichment of ER/SR subdomains in close proximity to SERCA2 and RyR2 activities in intact cardiomyocytes.

Methods
Preparation of Crude Cardiac Sarcoplasmic Reticulum Vesicles -Cardiac microsomes were isolated from female mongrel dog left ventricular tissue, and loaded with Ca oxalate as previously described, with minor modi cations [5]. Brie y, heart tissue was homogenized in 10 mM NaCO 3 at 1:20 dilution (buffer volume/tissue wet weight). Cardiac microsomes were isolated by differential centrifugation, isolating microsomes between 10,000-75,000 x g max . In contrast to the previously published method, a nal wash of the pellet in 0.6 M KC1, 30 mM histidine, pH 7.0, was not carried out. The investigation conforms to 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). Animal research was approved by the Wayne State University Animal Investigation Committee (protocol #A 04-02-13).
Ca oxalate loading of SR -Loading of crude cardiac microsomes, 75 mg of protein were resuspended in 40 ml of an ice-cold medium containing 50 mM histidine, 100 mM KCI, 65 mM MgCl 2 , 60 mM Na 2 ATP, 25 mM Tris/EGTA, 20 mM CaCl 2 , and 5 mM Tris oxalate (pH 7.1). The suspension was rapidly warmed to 37 o C to initiate active and rapid Ca uptake, and the incubation was conducted for 10 min with 5 mM additional Tris/oxalate added after the rst 5 min of incubation. The suspension was then immediately centrifuged at 4 ºC for 30 min at 100,000 x g max . The resulting brownish membrane pellet included a central white region, indicative of the Ca oxalate precipitate inside the highest density membranes. The whitish center was not present if ATP was not included in the Ca oxalate loading step (Fig. 1A).
Isolation of medium and high density membrane vesicles after Ca oxalate loading -Subfractionation of Ca loaded membranes was carried out essentially as previously described [5]. Brie y, membrane pellets were resuspended in 0.25 M sucrose containing 300 mM KC1, 50 mM sodium pyrophosphate, and 100 mM Tris (pH 7.2). This material was layered over a discontinuous sucrose gradient of 0.6 M, 0.8 M, 1.0 M, and 1.5 M sucrose dissolved in the same buffer. After centrifugation at 125,000 x g max for 2 h, membranes were collected from both the 1.5 M sucrose cushion (MedSR) or as the white pellet at the bottom of the gradient (HighSR). These two subfractions were not present in membranes not Ca loaded (Fig. 1B). MedSR was diluted with 4 volumes of ice-cold H 2 O, and then sedimented at 105,000 x g max for 60 min. MedSR and HighSR were resuspended in 0.25 M sucrose, 10 mM histidine, and stored frozen at -20 o C. Protein was determined by the method of Lowry et al. (17).
Preparation of protein samples for mass spectrometric analysis -To evaluate the protein composition of the two SERCA-positive membrane fractions, relative to the crude cardiac membranes, we analyzed exactly 20 µg each of MVs, MedSR, and HighSR membranes by SDS-PAGE. The wet gel was stained with SPYRO Ruby red to visualize the protein separations qualitatively (Fig. 1C).
LC-MS/MS analysis and database search -Tryptic peptides were separated by reverse phase chromatography (Magic C18 column, Michrom), followed by ionization with the ADVANCE ion source (Michrom), and then analyzed in an LTQ-XL mass spectrometer (Thermo Fisher Scienti c). For each MS scan, up to seven MS/MS scans were obtained using collision-induced dissociation. Data analysis was performed using Proteome Discoverer 1.1 (Thermo) which incorporated the Mascot algorithm (Matrix Science). The NCBI database was used against mammalian protein sequences and a reverse decoy protein database was run simultaneously for false discovery rate (FDR) determination. Secondary analysis was performed using Scaffold (Proteome Software). A xed modi cation of + 57 on cysteine (carbamidomethylation) and variable modi cations of + 16 on methionine (oxidation) and + 42 on protein N-terminus (acetylation) were included in the database search. Minimum protein identi cation probability was set at ≥ 95%, with 1 unique peptide at ≥ 95% minimum peptide identi cation probability. From 62,000 individual peptide spectra obtained, we identi ed 1105 different proteins, along with their relative levels within each of the three different membrane fractions, when levels were su cient. Raw data and search results have been submitted to the ProteomeXchange via PRIDE with accession PXD022455 (Username: reviewer_pxd022455@ebi.ac.uk, Password: G1wlk66N).
Results And Discussion 1. Protein composition of crude cardiac microsomes An abundance index (A SR ) was calculated for each protein, calculated as the number of its peptide spectra, divided by protein molecular weight as a rough correction for available target peptides, then normalized to the spectral level for SERCA2, which we de ned as A SR =100.0. A SR values are a crude surrogate for protein abundance, since factors unrelated to protein abundance will affect the quantifcation of tryptic peptides. Nonetheless, A SR values across sets of functionally related proteins can provide useful comparisons.
In a preparation of cardiac microsomes (MVs), we found 1102 proteins (see Online Resource 1). The most abundant peptides from crude cardiac MV membranes were from known mitochondrial proteins. The three proteins with the highest A SR values in cardiac microsomes were ATP/ADP translocase and ATP synthase, α and β subunits, with known mitochondrial proteins accounted for 7 of the top 10 proteins. The other three major proteins among the top ten highest A SR values were SERCA2, a keratin 1 isoform, and cardiac α1 actin. The focus of our experiments, however, was not directed at crude cardiac microsomal proteins, but instead, at proteins that are enriched in cardiac SR. Of the top 10 microsomal proteins detected, only SERCA2 was enriched in the denser membrane vesicles that result when Ca transport is activated by addition of ATP.

Enrichment of membrane vesicles by SERCA activation
The two enriched SR subfractions analyzed in this study have previously been described as junctional SR and free SR vesicles; however, with the much greater number of proteins identi ed through the use of our GeLC-MS/MS analysis, we have substituted the terms "medium-density SR membranes" (MedSR) and "high-density SR membranes" (HighSR), for these two membrane subpopulations (Table 1). This terminology allows us to discuss the larger collection of proteins, whether or not they participate directly in Ca handling functions.
Comparing the abundance of each protein (A SR ) in MedSR and HighSR membranes with its abundance in MVs, we determined protein enrichments in SR (E SR ) ( Table 1). Calculated from ratios of A SR values, the enrichment values were relatively unaffected by the sampling issues for any given protein. We used the value of E SR to de ne SERCA-positive SR proteins as those proteins that were enriched by SERCA activation at least 2.0-fold (E SR ≥ 2.0). For 354 SERCA-positive SR proteins, the enrichments varied, but the average enrichment was 8.0-fold (avg E SR = 8.0).
In contrast to enrichment of SR proteins in total SR (HighSR plus MedSR), enrichments in HighSR versus MedSR membranes (E sub ), re ect differences in the way small membrane patches are pulled apart, yielding small vesicles with varying levels of Ca release activity (RyR) on top of their very active Ca accumulation. Thus, individual proteins that segregate with MedSR were more likely to contain membrane patches that contain RyR; that is, closer to junctional SR sites in vivo. To evaluate this distribution as a single number, we de ned E sub as the difference between levels in the two SR fractions divided by the total (Table 1) For 14 of the best characterized SR proteins (Fig. 2B, green bars), the average fold enrichment in SR over crude cardiac microsomes (E SR ) was 5.0 ±1.6 (mean ±S.D.). When sorted based on their relative distribution between SR subcompartments (E sub ), all of the known junctional SR proteins were more enriched in MedSR (Fig. 3A), consistent with the idea that MedSR membrane patches probably contain more RyR molecules, possibly combined with less SERCA. Luminal ER/SR proteins, on the other hand, were highly enriched in HighSR (except for calsequestrin-2). SERCA2, SERCA1, and phospholamban were more equally distributed among the membrane vesicles, consistent with a more equal activation of Ca loading in both SR populations. Despite the similar enrichments of known SR proteins, their relative abundances (A SR , Fig. 3B) varied greatly, consistent with data widely known from protein staining of SDSgels [15,18].
For the entire set of 354 proteins in SERCA-positive membranes, E sub spanned a continuous range of values from + 0.90 to -0.90, corresponding to proteins at a 90% greater level in HighSR, or in MedSR, respectively (Fig. 4). Interestingly, for sets of proteins with related function, E sub values were clustered together. 2, in support of the idea that membrane fragments are generated from ER/SR membrane patches that contain characteristic ratios of SERCA and RyR levels.

High abundance proteins illustrate wider functions in cardiac SR than only Ca handling
A SR values are based upon the number of peptide spectra attributed to a particular protein, but are only semi-quantitative, as they assume similar coverage of every protein sequence by LC-MS/MS, which does not occur [22]. Yet, it was very interesting to look at the proteins with the highest A SR values, as it illustrates the scope of protein functions in cardiac SR. The 5 most abundant proteins in our SERCApositive SR sample were SERCA2, desmin, sarcalumenin, phospholamban, and calsequestrin-2 (CSQ2), accounting for about a quarter of SR protein mass (percent of total spectra) (Fig. 5). SERCA2a and phospholamban are well known major constituents of cardiac SR [5,13], and, as expected, their very high A SR values were a predictable consequence of their activity in Ca oxalate loading of SR vesicles.
Sarcalumenin is a known luminal constituent of cardiac SR membranes [18, 23], but of uncertain function. It is comprised of two well known splice variants: one generally de ned by its roughly 150-kDa apparent molecular weight on standard SDS-gels, and one de ned by a roughly 53-kDa [24]. Desmin, a muscle-speci c intermediate lament protein of striated muscle [25], was found at A SR levels comparable to SERCA2 (Fig. 5).
Surprisingly high A SR values were also found for many proteins barely discussed in the cardiac research  (Table 2, Appendix) may suggest a role as a secreted antimicrobial protein [27,28] or copper reductase enzyme [29].

Functional sets of proteins in cardiac ER/SR
The identi cation of sets of proteins related to known SR and ER functions, added consistency and context to the 354 proteins found in our analysis. Proteins involved in related ER/SR functions were present as complete sets in SERCA-positive membranes. And, while these ER/SR subdomains were present in widely varying abundances (A SR values), their enrichment factors (E SR and E sub ) were remarkably similar, consistent with their distribution together in membrane patches (vesicles after homogenization). The vast majority of these 354 proteins have not been previously identi ed in cardiac SR. We brie y summarize the major functional sets of proteins found, and compare their constituent protein enrichments and relative abundances.

SR Ca pumping
The most abundant peptide spectra, not surprisingly, were from SERCA2. Peptides from phospholamban, a subunit of SERCA2a that dissociates with PKA-dependent phosphorylation were detected at roughly 50% of SERCA2. The need for inclusion of the MW factor to the calculation of A SR values can be easily appreciated in this case, with the 20-fold difference in mass between SERCA and phospholamban molecules. While the relative levels of SERCA2 and phospholamban from this single mongrel canine heart sample will be subject to error and biological variability, the set of three proteins that function in Ca pumping (SERCA2, phospholamban, and SERCA1) were the 1st, 4th, and 15th highest ASR values, accounting for roughly 11% of spectra, supporting the view that Ca accumulation is a primary function of the cardiac ER/SR. Their enrichments in SR (E SR ), and distribution between MedSR and HighSR membrane subcompartments (E sub ) were also highly similar ( Fig. 5 and Online Resource S1).

Intraluminal ER/SR proteins Intraluminal proteins represented an abundant set of proteins based on their number and high A SR values,
constituting about 12% of total SR protein, with about half of the mass due to sarcalumenin and calsequestrin (Table 5). Many luminal SR proteins are those carrying a C-terminal sequence Lys-Asp-Glu-Leu (KDEL), which interacts with KDEL receptors to retrieve these proteins from the Golgi back to ER compartments [49][50][51][52]. Protein disul de isomerase (PDI) isoforms, although not previously identi ed in heart tissue, were among the proteins most enriched in SR (Table 5); also, more highly enriched in HighSR than MedSR (Fig. 3A) consistent with the common notion that free SR is essentially cardiac smooth ER, and that it contains less RyR [53].

Junctional SR proteins
The Ca oxalate loading method was previously used to identify several major junctional SR proteins, including calsequestrin-2, cardiac triadin, junctin, and RyR2 [15,16,54,55]. These 4 major junctional SR proteins, along with junctophilin-2, all presented here with relatively high levels of A SR , accounting for roughly 5% of total SR peptide spectra. Levels of couplon proteins reported in rabbit fast-twitch skeletal muscle SR [56] show interesting parallels (Fig. 6) previously reported that no gross differences occurred immunologically between junctin and junctate levels in heart homogenates, suggesting that junctin may be roughly half of the A SR level reported here.
Co-enrichment of L-type Ca channel with junctional SR markers was further evidence of a stable protein complex of couplon proteins with the sarcolemmal T-tubule membrane [21].

Peroxisomal proteins and rab proteins in MedSR
In addition to the enrichment of junctional SR proteins, MedSR membranes were also enriched in peroxisomal proteins, rab proteins, and caveolar proteins (Fig. 7), but the three types of protein-containing vesicles varied in their membrane enrichments. Peroxisomal proteins were very uniquely enriched in SERCA-positive membranes, and their enrichment was highly variable (E SR =26.7 ±22.3) (Fig. 8A). In When E sub values were averaged for several sets of functionally related proteins, average E sub values covered a range of distribution between the two SR subpopulations (Fig. 8A), suggesting that individual membrane patches are enriched in separate functional subdomains, with each subdomain exhibiting particular Ca transport properties that re ect its inclusion of SERCA and RyR protein.
Rabs and other small GTPases were also highly enriched in MedSR, exhibiting E sub values similar to those of known junctional SR proteins (Fig. 8A ]. Yet, SERCA-positive SR membrane vesicles also contained a complete collection of known rough ER proteins involved in translation, translocation, and N-linked glycosylation (Table 6, Appendix). Rough ER proteins were of relatively low abundance (A SR = 4.0 ± 1.7), but very highly-enriched over crude MVs (E SR = 6.0 ± 2.5) (Fig. 8A).
5.6 Lipid metabolism and lipid modi cations of proteins -Numerous enzymes involved in lipid metabolism were enriched in SERCA-positive membranes. The highest A SR value (= 20.7) resulted from CDITP, which appends inositol-3-phosphate to diacylglycerol, with numerous lipid metabolizing proteins present at lower A SR levels ( Fig. 9, Appendix).

5.7
Proteins involved in ER membrane structure and dynamics -Many cardiac ER/SR proteins are those thought to play roles in distributing and tra cking proteins across the biosynthetic pathway. Several act by guiding membrane patches along transport laments; these include Ca-binding protein p22, vesicletra cking protein Sec22b, vesicle-associated membrane protein 2 (VAMP-2), and vesicle transport protein Sec20 [67 -70]. The mammalian proteins of the p24 family (TMED 10, 9, 2, 1) are involved in selective loading of cargo in transport vesicle between membrane compartments, and the co-enrichment and relative abundances of its known subunits support a role in cardiac ER/SR protein distribution [71,72]. Finally, other ER proteins may function in the maintaining the structure of ER subcompartments, such as reticulons-2 and − 4 [73], lunapark-3 [74], and climp-63 [75]. Proteins involved in ER/SR dynamics are tabulated, along with enrichment values, in Table 7, Appendix.

Conclusions
In this study, we used the technique of Ca oxalate loading of cardiac SR membrane vesicles to produce two SR membrane subpopulations (MedSR and HighSR), thereby enriching patches of cardiac ER/SR membrane that contain combinations of SERCA and RyR levels. Roughly one third of microsomal proteins were enriched in SERCA-positive membranes, and about a third of those were more enriched in the MedSR membranes, indicating the presence of enriched RyR in the same SR patch, and suggesting a relative proximity to junctional SR, or at least a biochemically distinct membrane subdomain. The activation of SERCA activity in our current study, an historical measure of SR function, led to the enrichment of proteins from every ER subcompartment, supporting a view that cardiac ER and SR should cannot be demarcated based only upon their ability to function in Ca handling.
Proteins enriched in SERCA-positive SR vesicles were de ned as any protein that was enriched ≥2.0-fold over its level in crude heart microsomes. This single enrichment criterion selected 354 proteins of 1102 total proteins in crude microsomes, while excluding all mitochondrial, contractile protein, and other known organellar contaminants. Even major mitochondrial and contractile protein contaminants were eliminated by this simple measure of enrichment. SERCA-positive SR proteins encompassed proteins from all known functional ER and SR subdomains, leading us to conclude that SERCA-positive membranes represent cardiac ER and SR. Plotting A SR values for the top 2-4 proteins of different groups of proteins (Fig. 10), provided some semi-quantitative insight into how different ER/SR subdomains may contribute to overall SR function.
In spite of these substantial variations in A SR among functional sets of proteins, the enrichment properties E sub and E SR among the same sets were remarkably consistent; both because of their physical segregation in membrane vesicles, but also because enrichment values are derived from ratios of A SR values in two preparations. For example, a large number of known KDEL proteins were found in SR, exhibiting a wide range of A SR values (6.3 ±9.0, N = 7 proteins). For the same 7 KDEL proteins, however, E sub values were 0.47 ±0.08, and E SR values were 7.1 ±3.7.
The enrichment of junctional SR proteins in MedSR membranes was previously demonstrated by immunoblot analyses to be a feature of this SR preparation [5,15,18,21,76]. In the present study, we found that the values of E sub for 354 proteins formed a continuous range from − 0.9 (largely detected only in MedSR) to + 0.9 (largely detected only in HighSR membranes) (Fig. 4). Other functional protein groups exhibited similarly segregated values (Figs. 3,4,8), suggesting that they too were physically distributed into at least two divergent subcellular sites: those closer to junctional SR and those further removed from such sites. The segregation of protein functional groups in terms of E SR and E sub shows that vesicles are derived from small enough membrane surfaces to fractionate with different enrichment patterns, and do not simply represent huge sections of membrane surface.
In summary, enrichment of cardiac membranes by Ca oxalate loading leads to the enrichment of ER/SR proteins, distributed between membrane fractions that differ in Ca leak through RyR. The distribution of individual proteins between the two fractions (Esub), and the consistent enrichments found among different functional sets of proteins, re ects the connections between ER/SR subdomains and the wellstudied spatial relationships between junctional and free SR. Our data present for the rst time a reliable estimation of cardiac ER/SR protein content, along with a semi-quantitative assessment of prominent sets of functional ER/SR subdomains present in a microsomal preparation of canine ventricular tissue.  Puri cation of SR membranes from heart tissue using Ca oxalate loading.

Declarations
A,Crude microsomal vesicles (MV) were isolated by differential sedimentation, then incubated under Ca loading conditions without ATP (-ATP) or +ATP. A, Ca oxalate precipitate inside SR vesicles appears as a whitish pellet, following activation of SERCA2a (arrow). B,Separation of MVs on a discontinuous sucrose gradient showed banding of MedSR membrane vesicles a oat on a sucrose concentration (suc conc) of    Distributions between HighSR and MedSR are similar for sets of proteins with similar functions.
E sub values for all 354 SERCA-positive SR proteins (shaded gray area) range from +0.90 to -0.90 (90% more enriched in HighSR or in MedSR, respectively). Distribution of any one protein between the two density layers re ects in part, its physical proximity to junctional SR sites. Co-enrichment of proteins with similar function within common membrane fragments therefore re ect similar relative levels of (nearby) SERCA2 and RyR.