Protein Kinase C ε Signaling Complexes Include Metabolism- and Transcription/Translation-related Proteins

The serine/threonine kinase protein kinase C ε (PKCε) has been shown to be a critical component in the heart’s resistance to cell death following ischemic insult. Recent studies have indicated that PKCε forms multi-protein signaling complexes to accomplish signal transduction in cardiac protection. Using two-dimensional electrophoresis (2DE), combined with matrix-assisted laser desorption ionization mass spectrometry (MS), the initial analysis of these complexes identified signaling molecules, structural proteins, and stress-activated proteins. The initial analysis, although fruitful, was limited by the number of proteins revealed on the 2D gels. It was also apparent that many known cardiac protective functions of PKCε could not be fully accounted for by the proteins identified in the initial analysis. Here we reported the identification of an additional 57 proteins in PKCε complexes using complimentary separation techniques, combined with high sensitivity MS. These techniques include 2DE or large format 1D SDS-PAGE followed by LC/MS/MS and solution trypsin digestion followed by LC/MS/MS, all of which yielded novel data regarding PKCε protein complexes. Nanoscale LC/MS/MS for the analysis of gel-isolated proteins was performed with sub-femtomole sensitivity. In contrast to 2DE analyses, the identification of proteins from 1D gels was independent of their visualization via staining and allowed for the identification of proteins with high isoelectric points. We found that PKCε complexes contain numerous structural and signaling molecules that had escaped detection by our previous analyses. Most importantly, we identified two new groups of proteins that were previously unrecognized as components of the PKCε complex: metabolism-related proteins and transcription/translation-related proteins.

Myocardial ischemia, or low blood flow to the heart, is the primary etiology of a heart attack and the accompanying cardiac cell death. Myocyte death consequently leads to impaired heart performance, and often mortality. As a result, mechanisms to reduce the deleterious effects of ischemia and to prevent myocardial cell death are of particular clinical importance (1)(2)(3). The serine/threonine kinase protein kinase C (PKC) 1 was first hypothesized to play a role in protection against ischemic injury in the heart by Downey and colleagues (4). In their 1994 report, inhibition of PKC was shown to block the salubrious effects of cardiac protection, whereas activation of PKC reduced myocardial infarction following ischemia (4). Subsequently, multiple investigations from a number of laboratories have demonstrated that the ⑀ isoform of PKC serves as a critical mediator of the resistance to ischemia in the heart (5)(6)(7)(8)(9)(10)(11)(12)(13) and other organs (14 -16). Importantly, PKC⑀ does not function in isolation in cardiac protection. Various ancillary studies have characterized a number of other proteins that participate in the resistance to ischemic cell death in a PKC⑀-dependent fashion (for review see Refs. 2 and 17). This line of evidence suggests the existence of a functional subproteome defined by the PKC⑀ cardiac protective signaling system.
Until recently, studies aimed at the identification of molecules participating in the PKC⑀ signaling system have taken a conventional approach, in which the role of an individual protein or kinase is examined (4 -7, 10 -12). As a result, despite the well established role of PKC⑀ in cardiac protection, the identity of the repertoire of proteins underlying the manifestation of a protective phenotype remains unknown. Unlike a classical approach, functional proteomic analyses enable an unbiased, yet focused, investigation of all participating molecules and thereby provide a detailed blueprint of the entire signaling network (8, 18 -22). This approach has been used to effectively map the N-methyl-D-aspartate receptor signaling complexes in the brain (19). Using a similar functional proteomic approach, our laboratory has demonstrated that PKC⑀ forms multiprotein signaling complexes in the heart (8,22,23).
Previous studies suggest that subcellular signaling often involves the engagement of multiple molecules in close vicinity (19, 24 -26). Accordingly, we hypothesized that PKC⑀ may form complexes with numerous types of proteins within the cell to accomplish task-specific signal transduction (27). To test this, our initial studies employed two-dimensional electrophoresis (2DE), coupled with peptide mass finger printing (8). These studies revealed that PKC⑀ associates with numerous structural proteins, signaling molecules, and stress-activated proteins (8). We found that these signaling complexes not only serve to bring the molecules into close vicinity but that they also facilitate signal transduction during the genesis of cardiac protection against ischemic injury (8,9,23,28).
However, the initial functional proteomic analysis was limited by the number of proteins revealed on the 2D gels. It was also apparent that many known cardiac protective functions of PKC⑀ could not be accounted for fully by the proteins identified in this previous study, such as the role of PKC⑀ in the regulation of transcription (10,29) and in the modulation of cellular metabolism (30 -32). To test the hypothesis that PKC⑀ forms complexes with a distinct subset of proteins to regulate these aforementioned processes, we tailored the present investigation with the goal of a more thorough description of the PKC⑀ complex.
Several approaches have been used previously by other investigators for the analysis of protein complexes (21,(33)(34)(35). These include separation of the isolated complex by 2DE or 1D SDS-PAGE (33,34) followed by excision and analysis of visible spots or bands using matrix-assisted laser desorption ionization (MALDI)/MS (35), low-flow rate electrospray (nanospray) tandem MS (34,35), or reverse-phase LC/MS/MS (35). More recent experiments have eliminated gel electrophoresis and utilized solution digestion of the entire protein mixture followed by direct long gradient reverse-phase LC/MS/MS (36) or two-dimensional LC with cation exchange chromatography preceding the reverse-phase separation (37,38). In the present study, we used 1D SDS-PAGE and 2DE with in-gel digestion and LC/MS/MS and solution digestion with long gradient reverse-phase LC/MS/MS because of the complimentary information each has the potential to provide. MALDI/MS and low flow rate electrospray tandem MS are common protein identification techniques following in-gel digestion of 2D gel plugs or 1D SDS-PAGE gel bands. However, previous studies by our laboratory and our preliminary data in the present study suggest that neither approach alone is optimal because of the sample complexity and dynamic range of protein quantity. Nanoscale LC/MS/MS with fast gradients, as demonstrated herein, serves as a high sensitivity method for protein identification, allowing sub-femtomole detection limits, and was therefore the technique of choice to charac-terize proteins that may have escaped detection by our previous analyses (8).
Here we report another 57 proteins in the PKC⑀ signaling system, some of which are involved in metabolism, others in the control of cellular energy production, as well as a group of proteins known to be important in transcription and translation processes. Additional signaling molecules and structural proteins have been identified to supplement our initial list (8). Taken together with our initial analysis reporting the identification of 36 molecules (8), this provides a total of 93 proteins characterized in the functional subproteome of the PKC⑀ cardiac protective signaling system.

EXPERIMENTAL PROCEDURES
Isolation and Purification of the Functional Subproteome Defined by PKC⑀ Signaling Complexes-To characterize the sensitivity of our analysis, we used tissue samples extracted from hearts of transgenic mice expressing known amounts of cardiac PKC⑀ protein. PKC⑀ mice were generated as described previously (39 -41), and hearts were excised and frozen in liquid nitrogen until protein analyses were conducted. For large format 1D SDS-PAGE and 2DE analyses, hearts were homogenized in 1-2 ml (per heart) of immunoprecipitation buffer (20 mM Tris-HCl, 150 mM NaCl, 10 mM EDTA, 1 mM sodium orthovanadate, 1% Nonidet P-40, and protease inhibitor; Roche Molecular Biochemicals) using a glass-glass technique (see Fig. 1, C and D regarding immunoprecipitation protocol). Tissues were preincubated (1 h at 4°C) solely with protein G-Sepharose beads, and the mixture was centrifuged at 5,000 rpm for 10 min to pellet these beads and any nonspecific interacting proteins. 500-g proteins of the supernatant were incubated (12 h at 4°C) with 0.4 g of PKC⑀ antibody and 50 l of protein G-Sepharose beads, whereas 500-g proteins from the same source were incubated with IgG and protein G-Sepharose beads (as negative controls). After incubation, immunocomplexes were pelleted by centrifugation at 5,000 rpm for 5 min at 4°C. The pellets were then re-suspended and washed three additional times to remove nonspecific interactions. This protocol was arrived at through extensive pilot studies, which indicated that this amount of antibody and degree of washing was optimal for complex isolation (see Fig. 1, C and D). The protein complexes were then eluted from the beads either via low pH (50 mM glycine, 1% Nonidet P-40, pH 2.5) or thiourea buffer (Genomic Solutions, Inc., Ann Arbor, MI). Pilot experiments indicated that protein visualization was optimal for 1D SDS-PAGE following low pH elution and for 2DE following thiourea elution. In separate studies, it has been shown previously that PKC⑀ signaling complexes can alternatively be isolated via GST-based affinity pull-down method using recombinant GST-PKC⑀ proteins (8,28).
We also examined the proteins that associate with IgG by mass spectrometry following 1D SDS-PAGE in the present investigation (see Supplemental Table VII). It was interesting to note that this approach identified only three proteins that were found in PKC⑀ complexes (vimentin, vinculin, and tropomyosin (identified in Ref. 8)). Thus, the vast majority of the proteins were not shared between the IgG and PKC⑀ complexes. Furthermore, subsequent studies using immunoprecipitation followed by immunoblotting verified vimentin, vinculin, and tropomyosin as members of the PKC⑀ complex.
Complementary Protein Separation-Following elution from beads, the protein mixture was split into two aliquots. The first was concentrated using an Amicon YM3 microconcentrator into SDS loading buffer and run on a 5% stacking/10% resolving large format 1D SDS-PAGE gel (Duracryl; Genomic Solutions, Inc., Ann Arbor, MI) with standard Laemmli buffers; three parallel lanes were run (Fig. 2A), each corresponding to immunoprecipitant from 1.8 mg of tissue protein. The second aliquot was buffer-exchanged with 25 mM ammonium bicarbonate for solution digestion. Following thiourea buffer elution from beads, one-third of the sample was used to rehydrate a pH 3-10 immobilized pH gradient strip overnight, followed by isoelectric focusing and 2DE on a 10% Duracryl gel with Tricine chemistry. The 2DE analysis corresponded to immunoprecipitant from 6.4 mg of tissue protein. Protein features (for both 2DE and 1D SDS-PAGE gels) were visualized with Sypro Ruby stain (Molecular Probes, Inc., Eugene, OR) and imaged using a fluorescent imager (ProExpress; Genomic Solutions, Inc., Ann Arbor, MI). Observed features from the 2D gel were excised robotically (ProPic; Genomic Solutions, Inc., Ann Arbor, MI) following triangulation mapping (HT Analyzer Software; Genomic Solutions, Inc., Ann Arbor, MI) of gel spots to the picking robot (Fig. 3A). Note: Because of the fact that not all proteins were identified from 2D gels and therefore do not have observed pIs, the calculated pIs for all proteins are reported (see Tables I-IV).
Because we have observed previously that protein identification is often limited by the sensitivity of the stain, gel plugs were excised robotically from the entire continuum of the SDS gel as opposed to excising only the visible bands ( Fig. 2A). Gel plugs were digested robotically (ProGest; Genomic Solutions, Inc., Ann Arbor, MI) using a modified version of the method originally described by Rosenfeld et al. (42). Briefly, a reduction/alkylation step was utilized only for 1D SDS-PAGE plugs; 100 ng (10 l at 10 ng/l) of trypsin (Promega Corporation, Madison, WI) was allowed to rehydrate the gel pieces, 15 l of 25 mM ammonium bicarbonate was subsequently added, and the mixture was incubated at 37°C for 4 h. Digested supernatant was acidified with 1% formic acid and analyzed directly, without organic extraction of the gel plugs. For solution digestion, proteins were reduced with dithiothreitol and alkylated with iodoacetamide followed by digestion using 1 g of trypsin at 37°C for 18 h.
Mass Spectrometry Analysis-Fast gradient nanoscale LC/MS/MS was utilized for the analysis of in-gel digests (both 1D SDS-PAGE ( . C and D, pilot study for immunoprecipitation protocol. PKC⑀ complexes were isolated from mouse hearts by immunoprecipitation (IP), eluted following wash with different increasing amounts of the detergent Nonidet P-40, and immunoblotted for PKC⑀. As seen in the blot (C) and summarized in the graph (D), increased detergent in the wash buffer results in decreased PKC⑀ protein following elution (i.e. more is lost in the wash phase).
Peptides were analyzed using Q-Tof2 (Micromass UK Ltd, Simonsway, Manchester, United Kingdom) and LCQ-Deca (ThermoFinnigan, San Jose, CA) electrospray tandem mass spectrometers. Electrospray was facilitated in both cases by the application of a pre-column voltage (typically 1700 V); a PicoView source (New Objectives, Woburn, MA) was used on the LCQ-Deca. Data-dependent acquisitions were performed with both instruments, and fast LC employed MS survey and product ion spectra integration times of 0.3 and 1.5 s, respectively, for Q-Tof; collision offset was automatically FIG. 2. A, example of sequential robotic gel spot selection from a 1D gel, followed by LC/MS/MS analysis (B). Here, spot 85 was identified as 2,4-dienoyl CoA reductase, a molecule with a high pI value (pI ϳ9.10). Notice that there was no distinct protein staining in the gel at spot 85, indicating that mass spectrometric identification was independent of protein visualization. Data were acquired on the LCQ-Deca, and product ion spectra were matched to three peptides. Products of m/z 563.82 are shown, and observed product ions are indicated. determined based upon precursor mass and ion charge state. Acquisitions of 2 and 3 microscans (with maximum ion fill times limited to 50 and 100 ms using automatic gain control) were used for MS and tandem MS, respectively, with LCQ-Deca. Longer gradients (Q-Tof2 only) utilized integration times of 1 and 2 s for MS and MS/MS, respectively.
One factor often overlooked is the importance of high resolution and mass measurement accuracy of fragment ion data. Currently the only commercially available instrument that allows for this resolution and mass accuracy and that can be operated in a data-dependent LC/MS/MS mode is a quadrupole time-of-flight hybrid instrument such as the Q-Tof2. These advantages (high resolution and mass accuracy) increase the confidence of search results. Using a locally stored copy of the Mascot search engine (www.matrixscience.com), product ion data were searched against the publicly available NCBI nonredundant protein database (www.ncbi.nlm.nih.gov), allowing for cross-species protein identification. The protein identification criteria that we use are 1) at least two peptides must match to each protein, FIG. 3. A, gel spot selection from a 2D gel, followed by LC/MS/MS analysis. Example shows the identification of a protein with low abundance on the 2D gel, isocitrate dehydrogenase (DH) (B). The gel image (A) has been contract-stretched to allow visualization of these low intensity spots, and product ion data were from Q-Tof2 matched to five peptides (B). Products of m/z 584.3 (upper spectrum) and 752.4 (lower spectrum) for isocitrate dehydrogenase and observed product ions are indicated. and 2) in each of those spectra, the abundant fragment ions must be assigned.

Optimization of Mass Spectrometry for the Analysis of PKC⑀
Complexes-The sensitivity of LC/MS/MS can be correlated directly to flow rate and chromatographic peak width. Typical conditions described for proteomic applications involve 20 -60-min gradients that yield peak widths of ϳ20 -30 s (fullwidth half-maximum) (43) when analyzing simple protein mixtures, such as those derived from 1D SDS-PAGE or 2DE experiments. This imposes limitations on the sensitivity attainable when only a few fmol of a digest is loaded on-column. Here, we utilize 200 nl/min flow rates with 6-min gradients; this fast gradient affords peak widths of 2.5 s (full-width half-maximum) and resultant peak volumes of less than 20 nl (Fig. 4A). Such sample concentration allows subsequent routine data-dependent sensitivity of ϳ500 attamol of a digest loaded on-column, dependent on peptide response factor by electrospray (Fig. 4B). With optimized transfer lines and connections, cycle times (injection to injection) from 12 to 20 min (depending on the volume of sample injected) are achieved (Fig. 4A).
Cardiac Tissue Samples with Known Amount of PKC⑀ Protein-Previous studies from our laboratory have indicated that cardiac-specific transgenic activation of PKC⑀ is sufficient to produce an innate protection against myocyte death due to ischemia (8,39). The hearts of these mice are inherently protected, in the absence of exogenous stimuli, against cardiac cell death because of a prolonged ischemic episode (8,39). Seen in Fig. 1A is a Western blot for PKC⑀ in hearts of nontransgenic mice compared with hearts of PKC⑀ cardiacprotected mice. The increase in myocardial PKC⑀ expression is ϳ6.2-fold (Fig. 1B).
Complementary Separation Techniques Enhance Sensitivity of Protein Complex Characterization-In the present investigation, a combined approach involving 1D SDS-PAGE and 2DE with LC/MS/MS has enabled us to identify numerous novel proteins that reside in PKC⑀ complexes. The target of immunoprecipitation, PKC⑀, was quantitated (via Western immunoblotting) after solubilization. This experiment revealed that the gels contained a maximum of 8 ng of PKC⑀ protein, of which our LC/MS/MS method allowed detection even in the presence of excess myosin heavy chain. A total of 57 proteins not identified by our previous analyses were found in the present study, only three of which were common between 1D SDS-PAGE and 2DE approaches. Those exclusive to 1D SDS-PAGE could be accounted for on the basis of pI, as most were considerably basic (pI Ͼ 8.5). Those exclusive to 2DE were likely not observed by 1D SDS-PAGE because of comigration of multiple isoforms of abundant structural proteins

Mass Spectrometric Analysis of PKC Signaling Subproteome
(such as actin and myosin) or IgG.

Identification of Metabolism-, Transcription-, and Translation-related Proteins in Myocardial PKC⑀ Complexes-Protein
immunocomplexes were isolated with either PKC⑀ monoclonal antibody or IgG. IgG pulled down a significant number of nonspecific proteins that were not detected in PKC⑀ complexes and also a few proteins that are present in PKC⑀ complexes (see Supplemental Table VII for IgG-associated proteins identified by mass spectrometry). Specific members of PKC⑀ complexes are discussed in detail below. The identification of two proteins, 2,4 dienoyl CoA reductase and hnRNP A/B, are shown in Fig. 2 and Fig. 3. These examples are representative of the mass spectrometric data obtained from low intensity gel spots in this study.
PKC⑀-mediated cardiac protection has been shown to involve enhanced energy production and glucose uptake by the heart (30 -32), but the mechanism by which this is accomplished is unknown. The present analysis identified 15 proteins whose functions influence directly ATP production on multiple levels (Table II). These proteins include numerous glycolytic enzymes such as glyceraldehyde-3-phosphate dehydrogenase and enolase, specific subunits of various citric acid cycle enzymes such as isocitrate dehydrogenase and succinate dehydrogenase, and proteins related to mitochondrial metabolism, including the adenine nucleotide transporters and the voltage-dependent anion channel.
In addition, the resistance to ischemia engendered by activation of PKC⑀ is known to involve increased synthesis of protective proteins (2). However, the means by which PKC⑀ accomplishes this up-regulation are unknown. In the current investigation, we found that histones H1.3 and H4 (proteins that regulate chromatin structure) and numerous ribosomal proteins (S3, S18, and S19) reside in PKC⑀ complexes (Table  III). Notably, we also identified 10 hnRNPs (Table III) that have been shown to be involved in transcription, mRNA transport and regulation, and in the control of translation (44). The identification of these molecules highlights a previously uncharacterized function for PKC⑀. One study described the ability of PKC␦ to interact with hnRNP K (45); however virtually nothing is known regarding the functional consequences of an interaction between any of the hnRNPs and PKC⑀. The findings of the present study show that PKC⑀ complexes host numerous mRNA processing enzymes and suggest that these complexes may control protein expression and possibly splic- triggered the instrument to switch to MS/MS mode, and the subsequent product ion spectrum was integrated over 1.5 s (lower panel); observed product ions are indicated. The protein was identified using Mascot from a total of three peptides.

TABLE II Novel metabolism-related proteins identified in PKC⑀ complexes
Gel refers to the type (1D SDS-PAGE or 2DE) of gel from which the protein was identified. All molecular masses (Mm) and isoelectric points (pI) were calculated using the ExPASy Compute MW/pI tool online. DH denotes dehydrogenase. ing and thereby regulate the synthesis of protective proteins. Additional Structural and Signaling Proteins Identified in the PKC⑀ Complexes-A primary goal of the present experimental design was to determine which, if any, proteins in PKC⑀ complexes had escaped detection in our previous proteomic studies (8). Using MALDI/MS, these previous investigations identified multiple signaling kinases, structural proteins, and stress-activated molecules in PKC⑀ complexes (8). We found signaling molecules such as mitogen-activated protein kinases and tyrosine kinases (8) that had been shown previously to be involved in PKC⑀-mediated signaling in the heart. In addition, stress-activated proteins, such as the inducible isoform of nitric-oxide synthase and cyclooxygenase-2, and structural proteins, such as desmin and myosin light chain, were identified in the preceding studies (8). Hence, we sought in the present study to employ more high sensitivity methods to detect proteins that may have evaded identification previously. The 2DE and 1D SDS-PAGE approach, coupled with LC/MS/MS, allowed for identification of six additional signaling proteins and 13 additional structural proteins within PKC⑀ complexes (solution digestion allowed for the identification of vinculin; see Table I). For example, the present study identified a novel PKC⑀-associated Hsp, Hsp 60 (Table I). Furthermore, the LIM and PDZ domain-containing protein, Oracle, was found to reside in PKC⑀ complexes (Table I), a finding that suggests involvement of this relatively uncharacterized protein in PKC⑀-dependent signaling. In addition, five unknown proteins were also identified in the present study (Table IV). Furthermore, the present investigation uncovered the presence of various disease biomarkers in PKC⑀ complexes. The ␤ myosin heavy chain (in addition to the ␣ isoform of this protein), a marker of cardiac hypertrophy (46), was found in PKC⑀ complexes, as was the ischemic heart disease marker troponin I (47).

Salient Findings of the Present Study-
The necessary involvement of the ⑀ isoform of PKC in multiple forms of protection against ischemia has been well established (4,5,11,12). Despite this, the molecular nature of the PKC⑀ signaling system and the cellular mechanisms underlying its ability to engender a phenotype resistant to ischemic injury are poorly understood. The present investigation provides a framework within which PKC⑀ may regulate biological processes relating to cellular energy production and storage, which are well characterized as critical parameters regulating cell fate during ischemia and other stresses. Furthermore, we have identified additional signaling molecules and structural proteins, which strengthens our ongoing hypothesis regarding the role of subcellular task-specific signaling modules in the development of a physiological phenotype (8,22,23,27,48). Last of all, we found that PKC⑀ complexes contain various molecules involved in the handling of mRNA and in the regulation of protein synthesis. These findings, taken with what has been shown previously regarding the up-regulation of multiple proteins following PKC⑀ transgenesis, suggest that the role of

TABLE III Novel transcription-and translation-related proteins identified in PKC⑀ complexes
Gel refers to the type (1D SDS-PAGE or 2DE) of gel from which the protein was identified. Histone H1.3 was identified solely by the solution digest method. All molecular masses (Mm) and isoelectric points (pI) were calculated using the ExPASy Compute MW/pI tool online. hnRNP denotes heterogeneous nuclear ribonucleoprotein. PKC⑀ as a mediator of protection against ischemia involves specific regulation of protein synthesis. A comprehensive list of proteins identified in PKC⑀ complexes now contains a total of 93 proteins, which includes proteins reported in our previous investigation (8) and those reported in the current work (Supplemental Material lists all PKC⑀-associated proteins identified in these two studies). These PKC⑀-associated proteins can be classified into five functional groups, including 24 structural proteins (see Table  I and Ref. 8), 21 signaling molecules (see Table I and Ref. 8), 11 stress-activated molecules (see Table I (Table II), 17 transcription-and translation-related proteins (Table III), and five proteins of unknown function (Table IV).
PKC⑀ Complexes Contain Metabolism-, Transcription-and Translation-related Proteins-Although previous work by our laboratory and others (10, 29 -32, 50, 51) had indicated that PKC⑀ is involved in cellular metabolism and the regulation of protein expression and synthesis, biochemical evidence supporting these roles was lacking. Specifically, it was unclear whether PKC⑀ complexes, which have been shown previously to regulate distinct tasks within the cell (9,23,28), were also a mechanism by which PKC⑀ influences energy production and transcription/translation processes. It has been shown by numerous other studies that alterations in transcription factor activation (10,19,52) and protein expression (51) are necessary to protect the heart from ischemia. Furthermore, it has also been well established that ischemic injury to the heart and other organs involves altered metabolic processes (30,50). In the present study, we found that PKC⑀ complexes indeed contain molecules that are known to be essential for metabolism and the regulation of protein synthesis.
For instance, the finding that PKC⑀ complexes host glycolytic enzymes, such as enolase and glyceraldehyde-3-phosphate dehydrogenase, supports a role for PKC⑀ in the regulation of glucose metabolism. The presence of the citric acid cycle enzymes succinate dehydrogenase and isocitrate dehydrogenase further implicates PKC⑀ in processes relating to the metabolism of fats, carbohydrates, and proteins. Finally, the multiple mitochondrial enzymes involved in ATP production and transport (such as the ␣ subunit of the F1 ATPase) that were found to reside in PKC⑀ complexes suggest that this isozyme performs specific functions in the mitochondria relating to cellular energy balance.
The abundance of hnRNPs identified in PKC⑀ complexes suggests that PKC⑀ plays an extensive role in mRNA processing and perhaps in control of transcription (44), a function of PKC⑀ that has been indicated previously (10,29). The fact that we identified 10 of these proteins makes tenable the notion that the protective PKC⑀ signaling system in the heart involves extensive control over protein expression in part by regulation of mRNA. Our finding is consistent with a previous report that one of these hnRNPs, hnRNP K, co-resides with PKC (45). The multiple ribosomal proteins that appeared in PKC⑀ complexes support the concept that PKC⑀ may also regulate protein expression through specific tasks at the ribosome. Further studies will be required to characterize effectively the specific manner in which association of these factors with the PKC⑀ complex leads to altered regulation of protein synthesis and ultimately resistance to ischemic injury.
Identification of Additional Structural, Signaling, and Stressactivated Proteins in PKC⑀ Complexes-In addition to the identification of the aforementioned novel subgroups of proteins within PKC⑀ complexes, the present investigation also expanded our understanding of the structural, signaling, and stress-activated proteins that are involved in the PKC⑀ signaling system (8). The finding that PKC⑀ interacts with structural proteins such as Lamin, a nuclear structural protein, and vimentin, a protein associated with the contractile apparatus, supports the concept that there is subcellular compartmentdependent organization of these complexes (22). In addition, the identification of the signaling protein Oracle and stressactivated protein glucose-regulated protein, suggests a role of PKC⑀ in cardiac development (49) and molecular chaperoning processes (53), respectively.
Methodological Considerations-Characterization of protein mixtures has been recognized as technically challenging by many investigators. Our previous studies and preliminary data here indicated three issues would influence the success of the analytical approaches adopted: sample complexity (the sample was considered to consist of a discreet number of proteins (Ͻ100)), low abundance of constituent proteins, and high dynamic range of protein amounts (abundant co-isolated proteins impeding the identification of low abundance pro- teins). Gel electrophoresis-based approaches and in-gel digestion, along with experiments utilizing solution digestion of the entire protein mixture followed by direct long gradient reverse-phase LC/MS, were compared in the present study. Regarding the sensitivity for the analysis of gel digests, it is clear that MALDI and nanospray incur a limitation with respect to concentration; both techniques require a final sample volume of ϳ1 l. LC/MS/MS approaches with gel-isolated proteins, even with simple (n Ͻ 5) mixtures, have commonly exploited 30 -60-min gradients at 200 nl/min and subsequent peak widths of 30 s (full-width half-maximum) and peak volumes of ϳ100 nl. At the lowest levels of protein visualized on a gel, and considering the efficiency of an in-gel digestion procedure, all three mass spectrometric approaches are limited. LC/MS/MS with fast gradients gives peak volumes of Ͻ20 nl and sufficient chromatographic resolution to deal with simple protein mixtures (considering two or three peptides can yield protein identification) and was therefore utilized here. Using this method, many proteins were identified from regions of the 1D SDS-PAGE gel that had no visible protein and from 2DE spots that were at the detection limit of the fluorescent stain (54,55).
Sample complexity and dynamic range proved pivotal in the success of 1D SDS-PAGE versus 2DE; the techniques were complimentary, as 29 proteins were identified only from 1D SDS-PAGE and 18 only from 2DE, with merely three common proteins between the two methods. The third method, direct solution digestion, identified only two proteins (Vinculin and Histone H1.3) that were not found by either 1D SDS-PAGE or 2DE. Proteins that were unique to the 2DE approach could presumably not be identified from 1D gels because of the co-migration of abundant proteins (e.g. actin, myosin, and IgG) at the same molecular mass on the gel. Proteins exclusive to 1D SDS-PAGE could not be identified using 2DE because of their high pI. The abundant proteins dominated LC/MS/MS analysis from solution digest; accordingly, a two-dimensional LC approach is now ongoing to improve this problem.
Target Verification-The determination of the biochemical, and moreover, the physiological importance of these individual proteins as they relate to PKC⑀ signaling is a labor-intensive, yet extremely important, pursuit that is the focus of ongoing studies in our laboratory. We have designed a specific platform that deciphers the functional roles of protein complex members identified by proteomic analyses (such as the one conducted in this study).
First, the physical association (and quantitative amount) of these proteins with PKC⑀ is verified by co-immunoprecipitation experiments (immunoblotting of PKC⑀ immunoprecipitates) and in vitro GST-PKC⑀-based affinity pull-down assays. The subcellular distribution of these proteins is determined via confocal microscopic analyses (28,48). As reported by others (56), mass spectrometric analysis of protein-protein interactions may lead to false positive results. In our studies, using either MALDI/MS (8) or LC/MS/MS, a total number of 93 proteins were identified as members of PKC⑀ signaling complexes. To verify whether these proteins are members of PKC⑀ signaling complexes or artifacts due to the mass spectrometric analysis, we immunoprecipitated PKC⑀ signaling complexes and verified the presence of these members via immunoblotting. Using this co-immunoprecipitation approach, we have verified a total number of 60 proteins that co-reside in the PKC⑀ signaling complexes (representing two-thirds of the total 93 proteins identified). These 60 proteins include 36 proteins reported in our previous studies (8) and 24 proteins reported in the current studies (see Tables I, II, and III). Among the 33 proteins that remain to be confirmed, five are of unknown function, and antibodies to the other 28 proteins are currently unavailable.
Second, the biochemical features of these proteins are characterized by kinase activity assays, phosphorylation assays and/or phospho-antibodies, and other protein activity assays (such as DNA binding assays for transcription factors (10)). Third, the contribution of these proteins to the genesis of a physiological phenotype is determined using cell culture and transgenic mouse models. We have aggressively developed and implemented this protocol to functionally characterize members of the PKC⑀ complex (9,23,28,48) and will continue to do so for the members of the PKC⑀ complex identified in the present investigation.

CONCLUSIONS
The finding that a combined approach of 1D SDS-PAGE, 2DE, and direct solution digest yielded complementary data regarding the members of PKC⑀ complexes in the present study underscores the importance of using multiple separation techniques to study protein complexes. To fully characterize a dynamic complex of proteins, an approach that combines complimentary 1D SDS-PAGE and 2DE allows more complete mapping of the complex. Utilizing high sensitivity, fast gradient LC/MS/MS, combined with 1D SDS-PAGE or 2DE, as described herein, is a powerful alternative to MALDI, nanospray, or long gradient LC/MS/MS in isolation. In the present study, this approach enabled the identification and characterization of two novel groups of proteins (metabolism and transcription/translation) that possess biological functions essential to the protective signaling role of PKC⑀ in the heart.