Organellar Proteomics

The zymogen granule (ZG) is the specialized organelle in pancreatic acinar cells for digestive enzyme storage and regulated secretion and has been a model for studying secretory granule functions. In an initial effort to comprehensively understand the functions of this organelle, we conducted a proteomic study to identify proteins from highly purified ZG membranes. By combining two-dimensional gel electrophoresis and two-dimensional LC with tandem mass spectrometry, 101 proteins were identified from purified ZG membranes including 28 known ZG proteins and 73 previously unknown proteins, including SNAP29, Rab27B, Rab11A, Rab6, Rap1, and myosin Vc. Moreover several hypothetical proteins were identified that represent potential novel proteins. The ZG localization of nine of these proteins was further confirmed by immunocytochemistry. To distinguish intrinsic membrane proteins from soluble and peripheral membrane proteins, a quantitative proteomic strategy was used to measure the enrichment of intrinsic membrane proteins through the purification process. The iTRAQ™ ratios correlated well with known or Transmembrane Hidden Markov Model-predicted soluble or membrane proteins. By combining subcellular fractionation with high resolution separation and comprehensive identification of proteins, we have begun to elucidate zymogen granule functions through proteomic and subsequent functional analysis of its membrane components.

The zymogen granule (ZG) is the specialized organelle in pancreatic acinar cells for digestive enzyme storage and regulated secretion and has been a model for studying secretory granule functions. In an initial effort to comprehensively understand the functions of this organelle, we conducted a proteomic study to identify proteins from highly purified ZG membranes. By combining two-dimensional gel electrophoresis and two-dimensional LC with tandem mass spectrometry, 101 proteins were identified from purified ZG membranes including 28 known ZG proteins and 73 previously unknown proteins, including SNAP29, Rab27B, Rab11A, Rab6, Rap1, and myosin Vc. Moreover several hypothetical proteins were identified that represent potential novel proteins. The ZG localization of nine of these proteins was further confirmed by immunocytochemistry. To distinguish intrinsic membrane proteins from soluble and peripheral membrane proteins, a quantitative proteomic strategy was used to measure the enrichment of intrinsic membrane proteins through the purification process. The iTRAQ TM ratios correlated well with known or Transmembrane Hidden Markov Model-predicted soluble or membrane proteins. By combining subcellular fractionation with high resolution separation and comprehensive identification of proteins, we have begun to elucidate zymogen granule functions through proteomic and subsequent functional analysis of its membrane components.

Molecular & Cellular Proteomics 5:306 -312, 2006.
Pancreatic acinar cells are the functional units of digestive enzyme synthesis, storage, and secretion and have been a classical model for studying regulated exocytosis (1,2). In acinar cells digestive enzymes are stored in a specialized organelle, the zymogen granule (ZG). 1 Stimulation of acinar cells by secretagogues triggers fusion of ZG membranes with the apical membrane and the subsequent release of digestive enzymes. Early studies using SDS-PAGE indicated a relatively simple protein structure for the ZG membrane. Several more recent studies using 2D GE led to the identification of two additional major ZG membrane components, GP3 (3) and membrane dipeptidase (4). In another study, 14 spots were identified as small GTPases by [ 35 S]GTP␣S overlay on a 2D gel of ZG membrane proteins, but the identities of the spots remained unknown (5). In addition, a number of low abundance proteins, including Rab3D and several SNARE proteins, have been identified on ZG membranes by immunoblotting and immunocytochemistry (2). Despite its importance, a comprehensive proteomic analysis of the membrane protein components of ZGs has not been achieved.
By combining 2D GE and 2D LC with tandem mass spectrometry, we report the identification of 101 proteins on ZG membranes, many of which had not been localized previously on ZGs including multiple small GTP-binding proteins, SNARE proteins, and molecular motor proteins. In addition, to distinguish intrinsic membrane proteins from peripheral and soluble proteins, a quantitative proteomic strategy was used to measure the relative abundances of ZG proteins during membrane purification. The intrinsic membrane proteins characterized based on iTRAQ TM (6) ratios correlated well with known or TMHMM-predicted membrane proteins.

EXPERIMENTAL PROCEDURES
Isolation of ZGs and Purification of ZG Membranes-A previously validated Percoll gradient procedure was utilized (7). Briefly rat pancreata were homogenized in ice-cold buffer containing 0.25 M sucrose, 25 mM MES, 2 mM EGTA (pH 6.0). Homogenates were centrifuged at low speed to generate a crude particulate enriched in ZGs. The particulate was resuspended in homogenization buffer and mixed with an equal volume of Percoll. The mixture was centrifuged at 60,000 ϫ g for 20 min. The dense white ZG band was collected and washed with homogenization buffer. To purify ZG membrane, the isolated ZGs were lysed with nigericin and centrifuged at 100,000 ϫ g for 1 h. The membrane pellet was washed with 250 mM KBr and then 0.1 M Na 2 CO 3 (pH 11.0) each for 1 h.
2D Gel Electrophoresis, In-gel Protein Digestion, iTRAQ Labeling, 2D LC-MALDI, Tandem Mass Spectrometry, and Data Analysis-These are described in detail in the Supplemental Experimental Procedures.

RESULTS
The ZG membrane purification and iTRAQ tagging are outlined in Fig. 1, left. The high purity of isolated ZGs using Percoll gradient was confirmed in a previous study by electron microscopy and enrichment of the ZG enzyme marker amylase (7). In the current study, a similar degree of amylase enrichment was achieved, and the ZGs were highly homogenous as indicated by Nomarski images and positive staining for the known ZG membrane marker Rab3D (Fig. 1, right). To characterize the ZG membrane proteome, 2D gel maps were generated for Na 2 CO 3 -washed ZG membrane. In a representative gel, about 130 spots were visualized among which 61 were identified by tandem mass spectrometry (Supplemental Fig. 1). These 61 spots represented 29 unique proteins (Table  I). To uncover additional ZG membrane proteins, the same samples were also run on one-dimensional gels as well as broad range and basic 2D gels (data not shown). Sixteen additional proteins were identified from these gels, the majority of which were basic proteins. Altogether a total of 45 unique proteins were identified by the gel-based proteomic analysis (supplemental table).
To distinguish intrinsic ZG membrane proteins from other identified proteins, a quantitative proteomic strategy was implemented using iTRAQ reagents and 2D LC separation (Supplemental Fig. 2). The peak areas of reporter ions, 114, 116, and 117, in the MS/MS spectra correspond to the relative abundances of ZG proteins in the crude, KBr-, and Na 2 CO 3washed membranes, respectively. Because of their enrichment through wash steps, it is expected that intrinsic ZG membrane proteins will have 117:114 ratios greater than one, whereas soluble or peripheral membrane proteins will have ratios less than one. A representative MS/MS spectrum of an iTRAQ-labeled peptide from the intrinsic membrane glycoprotein GP2 is shown in Supplemental Fig. 3. The iTRAQ ratios of all proteins in Table I with at least two independent measurements are summarized in Table II. These proteins are classified in two groups with 117:114 ratios in group 1 less than one (not membrane proteins) and in group 2 greater than one (membrane proteins). In another column, TM domains, the presence or absence of transmembrane structures in these proteins is summarized based on literature or predictions using TMHMM (10). It was found that the two columns, 117: 114 ratios and TM domains, correlated very well for the characterization of intrinsic membrane proteins. The only exception in group 1, syncollin, has been reported to be washed off ZG membranes when incubated with 0.1 M Na 2 CO 3 (11). In group 2, 18 proteins had either transmembrane domains or FIG. 1. Outline of ZG membrane purification. Left, rat pancreata were homogenized and then centrifuged in two low speed steps to generate a crude particulate fraction (P2). The ZGs were purified by Percoll gradient. To purify ZG membrane, the isolated ZGs were lysed and centrifuged. The membrane pellet was then washed sequentially with 250 mM KBr and 0.1 M Na 2 CO 3 (pH 11.0). The proteins were extracted from crude, KBr-, and Na 2 CO 3 -washed ZG membrane, and the digested peptides were labeled with 114, 116, and 117 iTRAQ TM reagents, respectively. Right top shows a schematic to illustrate the Percoll gradient ultracentrifugation; at the bottom are Nomarski and fluorescent images of purified ZGs to demonstrate the purity of ZGs and the positive staining of a previously established ZG marker, Rab3D. Note essentially all granules stain for Rab3D. ER, endoplasmic reticulum. posttranslational modifications such as prenylations and GPI linkages. The high 117:114 ratio of ubiquitin may suggest that some ZG membrane proteins were preferentially ubiquitinated. However, the specific proteins that are ubiquitinated have not been identified. The very large 117:114 ratio observed for myosin Vc indicates that it is very strongly bound to the ZG. The mechanism of this strong attachment is unknown and requires further investigation. In contrast to 117:114 ratios, the 116:114 ratios were less informative in distinguishing membrane from soluble proteins, reflecting the fact that the KBr wash did not remove proteins from crude ZG membranes very efficiently. In addition to the quantitative results, the 2D LC experiments also led to the identification of more proteins than from 2D gels. In these experiments, a total of 2,498 peptides were fragmented for MS/MS analysis, and 630 peptides were assigned to 84 non-redundant proteins. By combining the LCand gel-based approaches, a total of 101 proteins was identified from purified ZG membranes among which 17 proteins were uniquely identified by the gel-based approach, 56 were uniquely identified by the LC-based approach, and 28 were identified by both approaches. The complete list of identified proteins is summarized in the supplemental table. After excluding likely contaminants from other organelles including known mitochondrial enzymes and ribosomal proteins, 60 proteins were categorized based on their generally accepted biological functions and are listed in Table I. These categories included Proton pumps and ion channels, Enzymes, Small GTP-binding proteins, Vesicular trafficking proteins, Matrix and glycoproteins, and a group of hypothetical proteins categorized as Unknown. In addition, 11 digestive enzymes from the ZG content were also present as contaminants in the ZG membrane fraction. Nine proteins that do not have an obvious relation to other categories were categorized as Others. In these 60 proteins, 28 of the known ZG membrane proteins were included that verified our approach. In addition, a large number of proteins previously unknown on ZG membranes were also identified including several hypothetical proteins such as Similar to c20orf178.
The new discoveries come primarily from the small GTPbinding proteins and vesicular trafficking categories. These include eight small G proteins, Rab1, Rab6, Rab8A, Rab14, Rab11A, Rab27B, Rac1, and Rap1; one SNARE protein, SNAP29; and the molecular motor protein myosin Vc. Six of these proteins were further confirmed by immunocytochemistry. In isolated acini, Rab27B was localized throughout the granule region as described previously (9). SNAP29 and myosin Vc shared this distribution (Fig. 2, left). Although Rap1 was present in most of the ZG area, staining appeared more intense in the deeper ZG regions, likely including Golgi elements. In contrast, Rab6 was sharply confined to the deepest region of the apical ZGs, and the reticular nature of the staining suggested localization primarily to Golgi. Although Rab11A was present to varying degrees throughout the ZG region, intense staining was generally seen immediately proximal to the luminal area. The presence of Rap1, Rab11A, Rab6, VAMP 2, and Rab27B on ZGs was confirmed by immunolocalization at the purified ZG level. As shown in Fig. 2, right, essentially all ZGs were stained with Rab27B or Rap1 antibodies, whereas only a moderate number showed strong staining for Rab11A, and just a few granules were clearly stained for Rab6. DISCUSSION Here we report the first comprehensive proteomic study of ZG membrane proteins with the identification of 101 proteins. As a verification of our approach, 28 previously reported ZG membrane proteins were identified in this study including the low abundance SNARE proteins VAMP 2 and VAMP 8 (12) but not Syntaxin 3, another SNARE protein localized on ZG membrane previously by immunocytochemistry (13). Seventythree new proteins were identified on ZG membrane for the first time, nine of which were confirmed by immunocytochemistry at both the isolated acini and ZG levels. Despite the high  Fig. 1 are listed. b Proteins identified by 2D GE using IPG strips pH 3-10 or 7-11 nonlinear are labeled with "x" (gel images are not shown).
purity of the ZG membrane preparations, a number of potential contaminating proteins including mitochondrial enzymes and ribosomal proteins were observed but only in the LCbased identification. This is most likely due to the mixture of crude and purified ZG membranes as well as the higher sensitivity of the LC-based approach. To further validate our results, the iTRAQ reagents were used to distinguish intrinsic ZG membrane proteins from other identified proteins. Traditionally this would require quantitative Western blots for each individual protein. Our result indicates that isotope tagging represents an effective mass spectrometry-based method to achieve the same goal systematically. The newly identified ZG membrane proteins provide new directions for studying the molecular mechanisms of ZG trafficking and regulated exocytosis. According to the current model for regulated exocytosis, secretory vesicles are regulated at multiple steps during vesicular transport from the Golgi network to the plasma membrane. It is generally believed that two families of proteins govern this process with the SNARE proteins mediating plasma membrane docking and probably also fusion of secretory vesicles with plasma membrane and at least one specific Rab protein regulating each vesicular transport step (14). Although it is believed that ZGs share the above common mechanisms with other secretory vesicles, most of the players in this model have been missing for ZGs including most of the Rab proteins and the complete set of SNARE complexes.
Rab3D has been the only candidate for regulating ZG exocytosis (15), although recently stimulated amylase release was reported to be normal in Rab3D-deficient mice (16), implying the presence of additional Rab(s) to regulate ZG exocytosis. Here Rab27B, the closest homologue of Rab3, was identified on ZG membranes for the first time and was further demonstrated to play a positive role in regulating acinar exocytosis in a separate study (9). It is of interest that Rab27A was found in a complex with myosin Va in melanosomes and thus tethered melanosomes to the apical actin cytoskeleton (17). The fact that myosin Vc was also identified  on ZG membranes in this study leads us to hypothesize that myosin Vc and Rab27B may form a complex and thus tether ZGs at the apical actin web. While Rab27B is very likely responsible for ZG tethering, Rab6 may play a role in budding from Golgi or in regulating ZG transport along microtubules; Rab11A, based on work in other cell types, is possibly in-FIG. 2. Immunolocalization of novel proteins identified from ZG membranes. ZG localization of a subset of the novel proteins was demonstrated by immunocytochemistry and confocal microscopy using corresponding antibodies (red). The subluminal actin is stained with Oregon Green-conjugated phalloidin, and nuclei are stained with 4Ј,6-diamidino-2-phenylindole (DAPI) (blue). Left, immunostaining of isolated acini for Rab27B, SNAP29, myosin Vc, Rap1, Rab6, and Rab11A. Right, immunostaining of purified isolated ZGs for Rab27B, VAMP 2, Rap1, Rab6, and Rab11A. ZG fluorescence images are paired with the corresponding Nomarski images. volved in ZG membrane recycling after exocytosis. The fact that Rab6 and Rab11A localized to only a fraction of ZGs may indicate that different Rabs target to ZGs at different stages in the secretory pathway. The roles of Rab8 and Rab14 are currently unknown, although it was recently reported that Rab8 was partially associated with mature melanosomes and regulated actin-dependent movement of melanosomes (18). In addition to the small G proteins, a novel SNARE protein, SNAP29, was identified on ZGs. SNAP29 has been considered to be a ubiquitous cytoplasmic SNARE protein involved in general membrane trafficking steps (19), and its role in ZG exocytosis needs to be further investigated.
In summary, the catalog of protein components comprising the ZG membrane described in this study evokes a range of new hypotheses regarding mechanisms of ZG function and will aid future efforts to identify higher order architectural components of the ZG, including protein complexes, leading the field to a better understanding of ZG architecture and functions.