Analysis of the Plasma Peptidome from Pancreas Cancer Patients Connects a Peptide in Plasma to Overexpression of the Parent Protein in Tumors Kwasi Antwi,† Galen Hostetter,‡ Michael J. Demeure,‡ Benjamin A. Katchman,† G. Anton Decker,§ Yvette Ruiz,† Timothy D. Sielaff,| Lawrence J. Koep,⊥ and Douglas F. Lake*,† School of Life Sciences, Arizona State University, Tempe, Arizona 85287, Translational Genomics Research Institute, Phoenix, Arizona 85004, Mayo Clinic Scottsdale, Scottsdale, Arizona 85259, Virginia Piper Cancer Institute, Minneapolis, Minnesota 55407, and Banner Good Samaritan Medical Center, Phoenix, Arizona 85006 Received May 9, 2009
Blood circulates through nearly every organ including tumors. Therefore, plasma is a logical source to search for tumor-derived proteins and peptides. The challenge with plasma is that it is a complex bodily fluid composed of high concentrations of normal host proteins that obscure identification of tumorderived molecules. To simplify plasma, we examined a low molecular weight (LMW) fraction (plasma peptidome) using liquid chromatography-tandem mass spectrometry (LC-MS/MS) methods. In the plasma peptidome of patients with ductal adenocarcinoma of the pancreas (DAP), a prominent peptide was identified from the QSOX1 parent protein. This peptide is stable in whole blood over 24 h and was present in 16 of 23 DAP patients and 4 of 5 patients with intraductal papillary mucinous neoplasm (IPMN). QSOX1 peptides were never identified in the plasma peptidome from 42 normal healthy donors using the same methods. Immunohistochemical staining of DAP tissue sections with anti-QSOX1 antibody shows overexpression of QSOX1 in tumor but not in adjacent stroma or normal ducts. Three of four pancreas tumor cell lines also express QSOX1 protein by Western blot analysis. This is the first report of QSOX1 peptides in plasma from DAP patients and makes the rare connection between a peptide in plasma from cancer patients and overexpression of the parent protein in tumors. Keywords: Liquid chromatography • tandem mass spectrometry • collision-activated dissociation • plasma • peptidome • pancreatic cancer • immunohistochemistry • Western blot
Introduction Because blood circulates through almost every organ, it samples proteins from a multitude of tissues, including tumors. As a result, plasma is a complex collection of proteins, sugars, carbohydrates and lipids that span concentrations of almost 10 orders of magnitude.1 Even though diagnoses as well as response to therapeutic treatment are correlated with quantitative and physiological changes of plasma proteins,2,3 analysis of plasma usually requires overcoming highly abundant proteins (HAP). Plasma protein concentration is estimated to be 60-80 mg/mL, of which 22 “classical” HAP constitute approximately 99% of the total mass.1,4-6 The remaining 1% of the plasma proteome offers insights into normal versus pathological states. * To whom correspondence should be addressed. Douglas F. Lake, Arizona State University, School of Life Sciences, P.O. Box 4501, LSE S58, Tempe, AZ 85287-4501. E-mail:
[email protected]. Phone: 480-727-9579. Fax: 480-965-6899. † Arizona State University. ‡ Translational Genomics Research Institute. § Mayo Clinic Scottsdale. | Virginia Piper Cancer Institute. ⊥ Banner Good Samaritan Medical Center.
4722 Journal of Proteome Research 2009, 8, 4722–4731 Published on Web 08/22/2009
For screening purposes, interest has focused on proteins that are secreted or shed from tissues.1,7 For example, increased plasma levels of CA19-98 prostate-specific antigen (PSA)9 CA125,10 and cardiac myoglobin11 have been used as markers for the detection of pancreatic cancer, prostatic cancer, ovarian cancer, and myocardial infarction, respectively. The effort of searching for disease markers within the past few decades has achieved limited success partly due to HAP and the large dynamic range of plasma protein concentration. Many investigators have tried to separate HAP away from less abundant, but more biologically informative molecules by using twodimensional electrophoresis (2-DE), liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS), and immunoaffinity approaches.1,7,12-21 However, instrument sensitivity and capacity still fall short of reliably and reproducibly detecting the lower range of plasma proteins. Because a diagnosis of cancer late in the disease process may lead to poor prognoses for patients and significant healthcare expense, it is important to identify plasma-based biomarkers to detect cancer early or to monitor recurrence after surgery. As normal cells transition to tumor cells, they go through many genetic transformations that result in production of protein variants or overproduction of proteins that can be distinguished 10.1021/pr900414f CCC: $40.75
2009 American Chemical Society
From the Plasma Peptidome to Tumor from wild-type proteins produced by normal cells due to their increased abundance in tissues and plasma.22,23 If a protein is overexpressed by a tumor, it may be shed into circulation and subject to proteolytic digestion, resulting in LMW peptides from the parent protein. Several plasma-based proteolytically derived peptides have been reported;6,19,24-28 however, their use for disease diagnosis has been underexplored. For example, fibrinogen A-derived peptides have been reported to correlate with hepatocellular carcinoma,29 ovarian carcinoma,30 colorectal carcinoma,31,32 and severe acute respiratory syndrome (SARS).33 Using mass spectrometry alone to identify potential biomarkers in plasma has been criticized because of lack of confirmatory nonmass spectrometry based evidence to support plasma-based results.26,34-38 Here we report the use of mass spectrometry methods followed by confirmatory immunohistochemistry (IHC) and Western blotting. Mass spectrometry analyses led us to identify peptide fragments from QSOX1 in pancreatic cancer patient plasma, and IHC with anti-QSOX1 antibodies indicated that the QSOX1 peptides identified in plasma are likely derived from pancreas tumors.
Experimental Section Materials. LC-MS grade water, acetonitrile (ACN), and formic acid were obtained from Fisher Scientific (Fair Lawn, NJ). Blood Collection and Processing for Mass Spectrometric Analysis. Blood samples from 42 normal subjects were collected under an IRB-approved protocol at Arizona State University. Participants included 22 male and 20 female normal subjects whose ages were between 18 and 63 years old. The age and sex of normal donors partially overlaps with the age bracket of pancreas cancer patients. Blood samples from normal subjects were drawn directly into sterile blood collection tubes containing EDTA (Fisher, Fair Lawn, NJ), as an anticoagulant. Blood samples were separated into plasma, lymphocytes and red cells by performing standard Ficoll blood separation. Approximately 30% of blood samples collected after 4 p.m. were processed the next morning after sitting at room temperature overnight (∼16-18 h later). Approximately 3-5 mL of plasma was collected, placed into cryovials (Wheaton, Millville, NJ), and stored at -80 °C for analysis. Prior to LC-MS/MS analysis, frozen plasma was thawed on ice and filtered through 0.45 µm polyvinylidene fluoride (PVDF) membrane ultrafilter (Millipore, Bedford, MA), and the filtrate was further passed through a 3 kDa filter (Millipore, Bedford, MA). With the use of NanoDrop spectrophotometer (Thermo Scientific), the concentrations of the 0.45 µm and 3 kDa filtrates were measured to be 64.5 ( 0.8 and 6.4 ( 0.6 mg/mL, respectively. Blood samples from patients undergoing surgery for pancreas neoplasms were obtained from the Tissue Acquisition core of Dr. Daniel Von Hoff’s pancreas cancer PO1 under the supervision of Dr. Demeure. Whole blood was collected from patients undergoing surgical resection for DAP at the Virginia Piper Cancer Institute in Minneapolis, MN, and Banner Good Samaritan Medical Center in Phoenix, AZ, under approved IRB protocols. Patients included 12 males and 16 females whose ages were between 34 and 84 years old. Blood samples were collected into tubes containing EDTA anticoagulant and shipped to us by overnight express at ambient temperature (except in summer when cold packs were used). Patient’s blood samples shipped to us were processed between 18 and 24 h post phlebotomy. Blood samples collected locally from pancreas cancer patients were processed from 1 to 4 h post phlebotomy.
research articles All blood samples were processed using the same methods as healthy donor plasma. Immunohistochemistry (IHC). Formalin-fixed, paraffinembedded tissue blocks from patients who underwent surgical resection for DAP were sectioned at 5 µm thickness using water flotation for tissue section transfer and dried overnight at room temperature. The slides were dewaxed, rehydrated and antigen retrieved on-line on the Bond autostainer (Leica Microsystems, Inc., Bannockburn, IL). All slides were cut at 5 µm and baked at 60 °C for 60 min. Slides were subjected to heat induced epitope retrieval using a proprietary citrate based retrieval solution for 20 min. Endogenous peroxidase was blocked. For QSOX1, the tissue sections were incubated for 30 min with antiQSOX1 polyclonal antibody at 1:75 from Proteintech Group, Inc. (Chicago, IL). The sections were visualized using the Bond Polymer Refine Detection kit (Leica) using diaminobenzidine chromogen as substrate. IHC optimization and staining parameters were evaluated by Dr. Hostetter, a board-certified pathologist, with standard scoring based on stain intensity (0-3) with score of 0 representing no staining and score 3 intense staining. Criteria for IHC stain localization in tumor cells were nuclear, cytoplasmic or membranous. SDS-PAGE-Western Blotting. Cell lysates from Panc-1, CFPac-1, Bx-PC3, and healthy donor PBMC were generated using an established protocol.39 Twenty micrograms of cell lysate protein was loaded into each lane of a 4-12% gradient SDSPAGE gel and electrophoresed under reducing conditions. The gel was transferred to PVDF membrane, blocked with 5% nonfat powdered milk and probed with anti-QSOX1 rabbit polyclonal antibody (same as was used in IHC) at a concentration of 0.34 µg/mL (1:1000 dilution) for 3 h. The blot was washed free of anti-QSOX1 primary antibody followed by addition of goat antirabbit IgG coupled with horseradish peroxidase (HRP) (Jackson Immunolabs, West Grove, PA) at a 1:5000 dilution and incubated with the blot for 1 h followed by washing. BCIP/NBT substrate (Pierce Chemical, Rockford, IL) was added and the blot was developed for 1 h at room temperature. Separation and Analysis of Peptides. Plasma samples from pancreatic cancer patients and normal subjects were filtered through centricon centrifugal filter devices with MWCO of 3 kDa at 10 000g at 4 °C. The filtrates were either analyzed immediately or transferred into siliconized tubes (VWR Scientific) and stored at -20 °C until analysis. The filtrates were resolved on an Agilent 1100 HPLC-Chip Cube Interface (Agilent Technologies). The Agilent HPLC-Chip integrates enrichment column (ZORBAX 300SB-C18, 40 nL, 5 µm), analytical column (ZORBAX 300SB-C18, 75 µm × 43 mm, 5 µm), and a nanospray emitter on a single and reusable microfluidic chip. Four microliters of each plasma sample was injected on the enrichment column using mobile phase containing 97% water, 3% ACN, and 0.1% formic acid at a flow rate of 4 µL/min. The enrichment column was washed with 12 µL of mobile phase to remove salts in the plasma sample that got trapped alongside peptides. Peptides that were trapped on the enrichment column were then eluted onto the analytical column for separation with a gradient from the nano-HPLC pump system. The gradient ran from 5 to 50% in 14 min and 50-100% in 4 min with mobile phase B (90% ACN, 0.1% formic acid in LCMS grade water) versus mobile phase A (3% ACN, 0.1% formic acid in LC-MS water) at a flow rate of 0.3 µL/min. The eluted gradient was analyzed on-line with nanoelectrospray ionization (nano-ESI) ion trap mass spectrometer (LC/ MSD Trap XCT Ultra, Agilent Technologies) in the positive ion Journal of Proteome Research • Vol. 8, No. 10, 2009 4723
research articles mode. The high voltage capillary was set at 1880 V, while the dry gas flow and dry temperature were set at 5 L/min and 320 °C, respectively. Mass spectrometer (MS) full scans were acquired from 350 to 2000 m/z in data-dependent mode. Six most abundant ion peaks above the background in each mass spectrum were selected as precursor ions for tandem mass spectrometry (MS/MS) using collision-activated dissociation (CAD). MS/MS scan of the same ion was not allowed in more than two MS spectra that were obtained within a period of 1 min. MS/MS Data Interpretation. MS/MS data were searched against NCBI refseq protein database (May 2008) which contained approximately 30 000 protein entries using Spectrum Mill software Rev A.03.03.078 (Agilent Technologies). Software parameters were adopted per manufacturer’s recommendations except changing the manufacturer’s recommended precursor mass tolerance of (2.5 to (1.5 Da to minimize false positive rate. Fragment ion mass tolerance of (0.7 Da was used. MS/MS spectra with at least four detected peaks and sequence tag length greater than 3 were considered for sequence interpretation. A peptide which had a charge of +2 with a score threshold of at least 10, or a charge of +1 or +3 with a score threshold of at least 13 was considered a good hit if it also has a scored peak intensity (SPI) of at least 70, and both forwardreverse score threshold and rank 1-2 score threshold of at least 2. Because we analyzed a low molecular weight fraction of the plasma proteome (peptidome), trypsin digestion was not employed. However, all proteins that were identified by only one peptide as well as all peptides that were ambiguously assigned to more than one protein were rejected from our list. A false positive rate of less than 7% was obtained by running our MS/MS spectra against a decoy database that was generated by shuffling target sequences in the NCBI refseq protein database.40 All peptide sequences reported in this study were again searched using blastp (http://www.ncbi.nlm.nih.gov/ blast/) to make sure each sequence was not found in more than one protein. Synthetic Peptides. Because of the relatively large precursor mass tolerance ((1.5 Da) used in the database search, 30 peptides reported in this study were chemically synthesized and run on the LC-MS/MS under the same conditions as described above to make sure their retention times and spectra matched those of the natural peptides. Peptides were synthesized at the proteomics core facility at Arizona State University on a Milligen 9050 peptide synthesizer (Millipore, Bedford, MA). Stable heavy isotope labeled peptide NEQEQPLGQWHL*S, which had all six carbon-12 and nitrogen-14 in Leu12 (C6H13NO2) substituted with carbon-13 and nitrogen-15, respectively, was synthesized by Anaspec, Inc. (San Jose, CA). Mass of the peptide was determined by amino acid analysis. After HPLC purification, purity of all synthetic peptides was estimated to be greater than 95%. Mass spectrometric analysis was used to confirm amino acid composition of peptides. Sequence Confirmation by MALDI-TOF/TOF. Ion trap mass spectrometer was initially used to determine peptide sequences by searching NCBI refseq protein database. Because ion trap mass spectrometry has low mass resolution, the peptide sequence NEQEQPLGQWHLS (from QSOX1), identified in plasma from pancreas cancer patients, was further confirmed by MALDI-TOF/TOF. MALDI-TOF/TOF analysis was performed as follows. Twenty picomoles of heavy isotope labeled version of NEQEQPLGQWHLS, which had all six carbon-12 and the nitrogen-14 4724
Journal of Proteome Research • Vol. 8, No. 10, 2009
Antwi et al. in Leu12 substituted with carbon-13 and nitrogen-15, respectively (NEQEQPLGQWHL*S), was spiked into 1 mL of plasma from P1032 (patient 1032) and filtered through a 3 kDa ultrafiltration device before analyzing the sample on Agilent 1100 Series HPLC system. The HPLC system was equipped with a quaternary pump, Vydac C18 reversed-phase column (10 × 250 mm, 10 µm, 90 Å pore), a manual injector with a 1 mL sample loop, multiple wavelength detector and a fraction collector. The plasma was analyzed in four replicates by analyzing 0.25 mL at a time at a flow rate of 4 mL/min with the following gradient: 10-28% B (0.1% TFA in ACN) versus A (0.1% TFA in LCMS grade water) for 16 min, 28-90% B for next 8 min, and held the gradient at 90% B for another 6 min to remove hydrophobic compounds from the column before reequilibrating the system with 10% B for next injection. Eluting peaks from the column were detected at a wavelength of 280 nm. A peak that eluted between 13.5 and 14.3 min which contained both light and heavy forms of NEQEQPLGQWHLS was collected from all four HPLC analyses, combined into a 15 mL polypropylene centrifuge tube and lyophilized for further analysis by MALDI-TOF/TOF. The retention time of the plasmabased peptide was predetermined by running sample of the heavy isotope labeled peptide on the HPLC under gradient conditions described above. Prior to MALDI-TOF/TOF analysis, the lyophilized sample was reconstituted in 50 µL of deionized water of which 3 µL was mixed with 3 µL of CHCA matrix solution (180 µL ACN, 120 µL deionized water, 1 mg CHCA and 0.2% TFA) and 1.5 µL of the mixture was spotted onto microScout MALDI target for analysis. An Ultraflex III MALDITOF/TOF instrument (Bruker Daltonics, Bremen, Germany) was used to analyze the sample in reflectron mode in the mass range of 800-3000 m/z. MS/MS analysis was performed on peaks that corresponded to masses of plasma-derived and heavy isotope labeled forms of NEQEQPLGQWHLS using the LIFT function of MALDI-TOF/TOF Ultraflex III instrument. FlexAnalysis 3.0 and Biotools 3.0 software were used for data analysis. The MALDI-TOF/TOF analysis was repeated with plasma from three other patient samples, P1027, P1001 and P1015 with the same result. Stability of QSOX1 Peptide. To determine the stability of the QSOX1 peptide, blood was drawn from a pancreas cancer patient just prior to surgical resection of tumor into a 7 mL EDTA tube and provided to us within 1 h of phlebotomy. It was split into three aliquots and processed into plasma as previously described after 0, 6, and 24 h incubation at room temperature. Plasma from each time point was transferred into siliconized tubes in 100 µL aliquots and stored at -80 °C for future use. Once ready for analysis, a tube from each time point was removed from freezer, thawed on ice and filtered through 3 kDa ultrafilters. Six microliters of plasma sample that was filtered through 3 kDa ultrafilter and spiked with 46 fmol of stable heavy isotope labeled peptide, NEQEQPLGQWHL*S, as an internal standard was analyzed to determine the levels of QSOX1 peptide (NEQEQPLGQWHLS) using LC-MRM (multiple reaction monitoring) technique. Plasma samples were injected on a Finnigan Surveyor LC system (Thermo Electron Corp., San Jose, CA) online with a linear ion trap (LTQ, Thermo Electron Corp., San Jose, CA). The Surveyor LC system was equipped with a standard autosampler. The peptides were desalted and concentrated on a Paradigm Platinum Peptide Nanotrap (Michrom Bioresources, Inc.) precolumn (0.15 × 50 mm) and subsequently on fused silica microcapillary C18 column (75 mm internal diameter, 10
From the Plasma Peptidome to Tumor cm in length, 5 µm particle size) (PFC7515-PP2-10, Michrom Bioresources, Inc.), at a flow rate of 160 µL/min. The samples were subjected to a 35 min (3-90% ACN) gradient and directly eluted into the microcapillary column set to 2.32 kV. The LTQ was operated in the positive-ion mode with collision energy of 35%. Each plasma sample was analyzed in three technical replicates. The QSOX1 peptide precursor-to-product represents transition for the doubly charged parent ion (NEQEQPLGQWHLS, 2H+) with m/z 783.51 to the single-charge product y ions with m/z 937.37 and 1194.42, whereas the precursor-to-product of the internal standard represents transition for the doubly charged parent ion (NEQEQPLGQWHL*S, 2H+) with m/z 787.31 to single-charge product y ions with m/z 944.52 and 1201.47. Data analysis and integration of peak areas were accomplished using Xcalibur 2.0. Peak area of the QSOX1 peptide was normalized to the signal of the internal standard. Quantitatitve ELISA for QSOX1 Peptide. Rabbit antisera were raised against NEQEQPLGQWHLS peptide by Epitomics, Inc. (Burlingame, CA) and affinity-purified on an NEQEQPLGQWHLS peptide column. A 50 ng/mL affinity purified antiNEQEQPLGQWHLS antibody was incubated for 1 h with plasma such that the plasma was diluted 1:1 in the antibody preparation. This mixture was then added to ELISA plates precoated with streptavidin-biotin-NEQEQPLGQWHLS peptide and further incubated at room temperature. After 1 h, the plate was washed free of unbound antibody-peptide followed by addition of a predetermined dilution of goat anti-rabbit IgGHRP (Pierce, Rockland, IL). After another hour of incubation, the ELISA plate was washed as before, and tetramethyl benzidine substrate was added followed by termination of the HRP enzyme activity with 1N H2SO4. Plates were read at OD 450 nm and the concentration of NEQEQPLGQWHLS peptide in plasma was calculated based on the linear portion of a standard curve of NEQEQPLGQWHLS peptide from 500 to 1.2 ng/mL.
Results A Peptide from QSOX1 Protein Is Prominent in Plasma from Patients with DAP. In our LC-MS/MS analysis, two peptides from the C-terminus of the long isoform of QSOX1 (QSOX1-L) were detected. The first peptide, NEQEQPLGQWHLS (amino acid 631-643), is from the C-terminus of QSOX1-L and occurred in 16/23 patients with DAP and 4 of 5 patients with IPMN (Supplemental Tables 1A and 1B). The other peptide, AAPGQEPPEHMAELQR (residues 615-630), is also derived from the C-terminus of the protein, but just upstream of NEQEQPLGQWHLS peptide. It occurred much less frequently in DAP and IPMN plasmas and will not be further discussed in this report. No peptide from QSOX1 was found in plasma from 42 normal donors using the same processing and detection methods. With the use of Fisher’s exact t test, this is highly significant with a p-value of less than 0.001. Of particular interest is detection of QSOX1-L peptide corresponding to QSOX1 in patients with IPMN. There is controversy about whether IPMN is a premalignant condition,41,42 but if IPMN predisposes patients to DAP, QSOX1 could be a useful predictor for which patients with IPMN might progress to DAP. We are planning a retrospective study to address this hypothesis. Many other peptides such as complement C4A, apolipoprotein A2 and fibrinogen were also identified in plasma from cancer patients as well as normal donors, demonstrating that the plasma
research articles peptidome is a viable source of tissue-derived peptides (see Supplemental Table 2). The QSOX1 gene locus is on human chromosome 1q24 in proximity to the prostate cancer locus (HPC1).43 It has short and long isoforms that are produced as splice variants. The long isoform (QSOX1-L) is 747 amino acids and the short isoform (QSOX1-S) is 604 amino acids (Supplemental Figure 1). QSOX1 is a sulfhydryl oxidase thought to participate in redox reactions during intracellular protein folding43 and has FADbinding domains homologous to ERV1p from yeast.44,45 Although there are relatively few publications related to QSOX1, one cancer-related study reported that overexpression of QSOX1 might protect cells against oxidative stress-induced apoptosis.46 Using LC/MS-MS methods, we analyzed the plasma peptidome and did not perform trypsin digestion on plasma proteins. Therefore, identification of multiple peptides from the same protein may not occur. To confirm identification of NEQEQPLGQWHLS, stable heavy isotope labeled version of the peptide (NEQEQPLGQWHL*S) was chemically resynthesized and run on LC-MS/MS, and LC retention time and MS/MS spectrum were compared to those of the natural peptide detected in plasma. The retention times of the labeled peptide and natural NEQEQPLGQWHLS overlap perfectly. The mass spectra from the synthetic and natural sequences also show the same ion peaks with an increase in m/z values of y ions by +7 Da in the labeled peptide spectrum as expected (Figure 1), confirming that the peptide is not a false positive. To further confirm the sequence in plasma-based NEQEQPLGQWHLS, stable heavy isotope labeled version of the peptide, NEQEQPLGQWHL*S, was spiked into aliquots of whole plasma from four pancreas cancer patients, then filtered through 3 kDa ultrafilters before fractionating the low molecular weight fraction using HPLC. The HPLC fraction that contained the sequence NEQEQPLGQWHLS and its heavy isotope labeled version was collected and analyzed by MALDI-TOF/TOF as described in Experimental Section above. Analysis of MALDITOF/TOF spectra obtained from the plasma-derived peptide and its stable heavy isotope labeled version again confirms the sequence as NEQEQPLGQWHLS (Figure 2). Because there is a single dalton difference between N and D and E and Q, it is possible that the ion trap used in our initial analysis provided an incorrect mass that could have resulted in an incorrect sequence. To address this concern, we bioinformatically substituted D for N and E for Q and vice versa at each position where N, E and Q occur in the sequence. Each substituted sequence was then searched using blastp for matches. No matches better than 6 continuous amino acids were identified, strongly suggesting that NEQEQPLGQWHLS is the correct sequence that is derived from QSOX1. Immunohistochemical Detection of QSOX1. Because peptides corresponding to QSOX1 were identified frequently in plasma from patients with DAP, but not in plasma from healthy donors, we hypothesized that QSOX1 might be overexpressed in DAP. To address our hypothesis, immunohistochemistry was performed with polyclonal anti-QSOX1 antibody to stain tissue sections from patients with DAP. As shown in Figure 3A, antiQSOX1 antibodies stain DAP tumor cells, but not normal adjacent ducts or surrounding stroma. In Figure 3B, antiQSOX1 antibodies stained a micrometastasis from a peripancreatic lymph node. Upon closer examination of the type of staining (Figure 3C), it appears that QSOX1 expression in DAP is cytoplasmic. Figure 3D demonstrates lack of specific staining Journal of Proteome Research • Vol. 8, No. 10, 2009 4725
research articles
Antwi et al.
Figure 1. Tandem mass spectra of (A) NEQEQPLGQWHLS (precursor m/z 783.36; +2 charge state) QSOX1-L peptide identified from
