Identification of PSME3 as a Novel Serum Tumor Marker for Colorectal Cancer by Combining Two-dimensional Polyacrylamide Gel Electrophoresis with a Strictly Mass Spectrometry-based Approach for Data Analysis*

The purpose of this study was to identify and validate novel serological protein biomarkers of human colorectal cancer (CRC). Proteins from matched CRC and adjacent normal tissue samples were resolved by two-dimensional gel electrophoresis. From each gel all spots were excised, and enveloped proteins were identified by MS. By comparison of the resulting protein profiles, dysregulated proteins can be identified. A list of all identified proteins and validation of five exemplarily selected proteins, elevated in CRC was reported previously (Roessler, M., Rollinger, W., Palme, S., Hagmann, M. L., Berndt, P., Engel, A. M., Schneidinger, B., Pfeffer, M., Andres, H., Karl, J., Bodenmuller, H., Ruschoff, J., Henkel, T., Rohr, G., Rossol, S., Rosch, W., Langen, H., Zolg, W., and Tacke, M. (2005) Identification of nicotinamide N-methyltransferase as a novel serum tumor marker for colorectal cancer. Clin. Cancer Res. 11, 6550–6557). Here we describe identification and initial validation of another potential marker protein for CRC. Comparison of tissue protein profiles revealed strong elevation of proteasome activator complex subunit 3 (PSME3) expression in CRC tissue. This dysregulation was not detectable based on the spot pattern. The PSME3-containing spot on tumor gels showed no visible difference to the corresponding spot on matched control gels. MS analysis revealed the presence of two proteins, PSME3 and annexin 4 (ANXA4) in one and the same spot on tumor gels, whereas the matched spot contained only one protein, ANXA4, on control gels. Therefore, dysregulation of PSME3 was masked by ANXA4 and could only be recognized by MS-based analysis but not by image analysis. To validate this finding, antibody to PSME3 was developed, and up-regulation in CRC was confirmed by Western blot analysis and immunohistochemistry. Finally by developing a highly sensitive immunoassay, PSME3 could be detected in human sera and was significantly elevated in CRC patients compared with healthy donors and patients with benign bowel disease. We propose that PSME3 be considered a novel serum tumor marker for CRC that may have significance in the detection and in the management of patients with this disease. Further studies are needed to fully assess the potential clinical value of this marker candidate.

diagnosed only in an advanced stage, and prognosis is accordingly poor (1). Various screening methods for early detection of colorectal tumors are available (for a review, see Ref. 2). However, their diagnostic value is limited regarding sensitivity (fecal occult blood test) (3,4) or regarding costs, risks, and inconvenience (colonoscopy) (5)(6)(7)(8).
Serological biomarkers can be analyzed relatively easily and economically and therefore have the potential to greatly enhance screening acceptance. Various serum markers for CRC are available among which carcinoembryonic antigen (CEA) is the most commonly used. However, this marker lacks sensitivity as well as specificity for screening an average risk population (9,10). Therefore, genomics-and proteomicsbased approaches have been used to identify new biomarkers for CRC. Particularly 2-DE has been used in many of the proteomics studies (11). Despite a number of limitations, the resolution of 2-DE gels is impressive, rendering this technology still a preferred tool in many proteomics studies (12). To distinguish differentially expressed proteins, spot patterns of 2-DE gels from colon cancer samples are matched and compared with those from control samples. However, this comparative image analysis may be hampered by several factors, such as moderate reproducibility or partly suboptimal resolution of 2-DE gels and the resulting difficulties of spot matching of different gels. In addition, spot overlapping may also represent a problem. Back in 1984, Young (13) switched from running conventional size 2-DE gels to giant gels to resolve a higher number of proteins. The author observed that "many of the major spots visible on the smaller gels, and especially streaks, were resolved into multiple spots on the larger gels. Thus, spots presumed to represent single proteins on smaller gels often in fact represent several." Later Gygi et al. (14) proved by mass spectrometric analysis that spots may envelope more than one protein. They even found one spot to be composed of a sextuplet. Recently Campostrini et al. (15) systematically analyzed the extent of spot overlapping and concluded from their work that for a typical tissue homogenate under normal loading conditions (1 mg of protein) and standard gel sizes (18 ϫ 20 cm), "the singlets would be by far the least abundant species." Because the problems listed mainly affect comparative studies of 2-DE gels by image analysis, we chose an alternative strategy to identify novel biomarkers for CRC. For this purpose, proteins from tumor tissue samples and healthy controls were resolved on 2-DE gels, and all spots were further analyzed applying a strictly MS-based approach. By comparing the protein profiles of the analyzed samples, elevated proteins in neoplastic tissue can be identified. In a previous publication (16) we reported the complete proteome of colon normal and neoplastic tissues as identified by this methodology. In addition, identification and preliminary validation of five exemplarily selected proteins, which were elevated in CRC tissue, was described. However, significantly elevated serum levels in CRC patients could only be demon-strated for one of the five proteins. In the present report, we describe for the first time identification and validation of another novel cancer-associated protein from that study, PSME3, which was significantly up-regulated in CRC tissue. Importantly due to overlapping of the PSME3 spot with an ANXA4-containing spot, dysregulation of PSME3 was undetectable using comparative image analysis only.
Initial validation studies confirmed the relevance of PSME3 as a tumor-associated protein. Polyclonal antibody to recombinantly expressed PSME3 was generated, and up-regulation of the protein in CRC tissue was confirmed by Western blot analysis and immunohistochemistry. Importantly the marker could also be measured in serum using a highly sensitive immunoassay and was significantly elevated in serum of CRC patients compared with healthy individuals and patients with benign bowel disease.

EXPERIMENTAL PROCEDURES
Chemicals and Reagents-All chemicals used were analytical grade from Merck and Fluka (Buchs, Switzerland) if not indicated differently. Deionized water produced by a Milli-Q system (Millipore, Billerica, MA) was used for all buffers.
Patients and Blood Donors-Clinical tissue samples were obtained from the Institute for Pathology, Klinikum Kassel, Kassel, Germany, in accordance with an ethic vote. All patients were diagnosed by histopathology. Fresh tumor tissues and paired tumor-adjacent normal colon tissues were obtained and frozen in liquid N 2 immediately after surgery. Stripped mucosa was rapidly prepared from normal colon tissue immediately before freezing. Starting time of surgical resection and time of freezing were recorded for each sample. Long term storage of tissue samples was at Ϫ80°C. In total, tissue specimens of 16 patients with different stages of CRC were analyzed. 40 serum samples of patients with CRC were purchased from Impath (Franklin, MA), and 69 further serum samples of CRC patients were obtained from three clinical centers in Germany following the ethical rules of the respective institution: Stadtkrankenhaus Rü sselsheim, Rü sselsheim; Medizinische Klinik II, Hochtaunuskliniken, Bad Homburg; and Krankenhaus Nordwest, Frankfurt. 317 specimens of healthy individuals were donated on a voluntary basis by Roche Diagnostics GmbH employees and collected at the three clinical sites mentioned above, respectively. In addition, 87 benign bowel disease control samples of patients with one or more of the following diagnosis were collected (frequency of diagnosis is given in parentheses): diverticulosis (53), colitis (18), diverticulitis (9), morbus Crohn (3), and others (9). Written informed consent of all patients and blood donors was documented. Mean age with S.D. and gender distribution of the patients are listed in Table I. Further characterization of the clinical tissue samples is documented elsewhere (16).
Sample Preparation and Protein Extraction-Venous blood sampling and protein extraction from tissue samples were performed as described before (16).
2-DE-IEF and SDS-PAGE for 2-DE were carried out as described in detail before (16).
Peptide Mass Fingerprinting and Identification of Proteins-Peptide mass fingerprinting analysis was essentially performed as described previously (17).
Briefly all spots on the gels were excised and placed into 96-well microtiter plates. The excised spots were destained using 180 l of 100 mM NH 4 HCO 3 in 30% acetonitrile, and the gel piece was dried in a SpeedVac evaporator. The dried gel piece was rehydrated with 5 l of 20 g/ml recombinant trypsin (proteomics grade, Roche Diagnos-tics GmbH, Mannheim, Germany) solution. After 5 h at room temperature, 20 l of 50% acetonitrile containing 0.3% trifluoroacetic acid were added, and the gel pieces were incubated for 15 min with gentle shaking. Sample application to a target plate and analysis as well as peptide matching and protein searching were carried out as described previously (17,18).
Recombinant Antigen Production and Generation of Antibodies-The expression and purification of full-length PSME3 was performed as described in detail previously (16). For generation of polyclonal antibody, recombinant antigens were used to immunize rabbits, and antisera were collected after 3 months.
SDS-PAGE and Western Blot-SDS-PAGE and Western blot analysis were carried out as described elsewhere (16).
Immunohistochemistry-Paraffin-embedded tissue sections were purchased from BioCat GmbH (Heidelberg, Germany; catalog numbers T8235090 and COCA2406-2-OL). Tissues were deparaffinized in xylene (3 ϫ 5 min) and rehydrated through a series of graded EtOH followed by two washing steps with deionized H 2 O. Antigen retrieval was performed with 10 mM Na 3 -citrate buffer, pH 6.0 (30 min at 97°C). Endogenous peroxidase activity was blocked by incubation in 0.3% H 2 O 2 in methanol for 20 min, and the slides were washed twice with H 2 O and once with PBS ϩ 0.05% Tween 20. For PSME3-specific staining the slides were incubated with the polyclonal antibody (5 g/ml in antibody diluent (Dako, Hamburg, Germany)) for 1 h and washed three times with PBS ϩ 0.05% Tween 20. After incubation with anti-rabbit Ig-horseradish peroxidase (Dako) for 30 min, the slides were washed again three times in PBS ϩ 0.05% Tween 20. Afterward the slides were incubated for 10 min in diaminobenzidine chromogen solution (Dako) and rinsed twice with H 2 O. After counterstaining with hematoxylin tissues were microscopically analyzed using the magnification indicated.
ELISA for PSME3-For detection of PSME3 in human serum, a sandwich ELISA was developed using streptavidin-coated 96-well microtiter plates. 24 l of human serum sample or diluted HT29 cell lysate as calibrator antigen were incubated with 216 l of antibody reagent containing biotinylated and digoxigenylated affinity-purified polyclonal anti-PSME3 antibody (0.83 g/ml each) from different animals in 40 mM phosphate buffer, pH 7.4, 0.9% NaCl, 0.1% bovine IgG, 0.022% polymerized rabbit IgG, 1.025% polyethylene glycol 40,000, 1.1% normal rabbit serum, 0.6% Synperonic F68, 0.01% N-methylisothiazolone, and 0.1% chloroacetamide. After incubation overnight at room temperature, 100 l were transferred to a streptavidin-coated microwell plate and incubated for 1 h. Subsequently the plates were washed three times with 0.9% NaCl, 0.1% Tween 20. For the detection of bound antigen-antibody complexes, 100 l of a monoclonal anti-digoxigenin horseradish peroxidase conjugate (30 milliunits/ml in Universal Conjugate Buffer, Roche Diagnostics GmbH, Mannheim, Germany) was added and incubated for 1 h. The excess of conjugate was removed by washing the plates three times with 0.9% NaCl, 0.1% Tween 20. The amount of bound conjugate was determined by adding tetramethylbenzidine substrate solution (Roche Diagnostics, Mannheim, Germany) and incubating for 1 h. The reaction was stopped by adding sulfuric acid, and absorbance was measured at 450 nm with a correction wavelength of 620 nm using an ELISA reader. A lysate of HT29 tumor cells was used for calibration. The PSME3 content of this material had already been estimated by Western blot by comparing the intensity of the PSME3 band with known amounts of recombinant full-length PSME3.
CEA Assay-CEA was measured by a commercially available assay (Roche Diagnostics GmbH, Mannheim, Germany).

Identification of PSME3 in Colon Cancer Tissue-Colon
tumor tissue, adjacent normal tissue, and adjacent normal stripped mucosa from 16 patients were analyzed in triplicate by 2-DE. The 2-DE gels were stained with colloidal Coomassie, all visible spots were excised, and the corresponding proteins were analyzed by peptide mass fingerprinting to identify the protein. This generated a list of 735 distinct proteins identified in either colon tumor, colon normal tissue, or both. In a pilot study five proteins up-regulated in colon cancer tissue were exemplarily selected and further validated as reported previously (16). In addition, all identified proteins were reported, and whether they were identified in tumor tissue, normal tissue, or both tissue types was indicated.
In the present work, we report identification and validation of an additional tumor-associated protein, PSME3. It was detected in seven of 10 tumor tissues analyzed on pH 4 -7 2-DE gels. PSME3 was not detectable on gels displaying adjacent normal colon tissue and adjacent normal stripped mucosa. This finding clearly indicated up-regulation of PSME3 in colorectal tumors compared with adjacent normal tumor tissue. Fig. 1 shows a representative 2-DE gel from a colon tumor and adjacent normal stripped mucosa. The spot indicated enveloped two proteins in the tumor sample, PSME3 and ANXA4, whereas the same spot enveloped one single protein, ANXA4, in the adjacent normal control tissue. By visual inspection of the 2-DE gels, this spot showed no difference with regard to intensity and position in the tumor sample when compared with matched normal colon tissue. Image analysis would have therefore failed to recognize dysregulation of PSME3. All mass peaks derived from the MALDI mass spectrum of a representative PSME3/ANXA4-containing spot excised from a 2-DE gel loaded with a CRC tissue lysate are listed in Table II. The predicted molecular mass/pI for PSME3 was 29.5 kDa/ 5.69 and fits well to the position of the corresponding spot on the 2-DE gel. ANXA4 was identified in two separate spots on the 2-DE gels (Fig. 1). Besides its identification in the spot overlapping with PSME3, it was also identified in a second spot with a slightly higher apparent molecular mass and a more basic pI. The predicted molecular mass/pI for full-length ANXA4, 35.8 kDa/5.85, are well in accordance with the position of this spot on the gel. Therefore, ANXA4 identified in the spot overlapping with PSME3 may represent a fragment of the full-length protein. For example, deletion of the C-terminal 44 amino acids of ANXA4 would shift the predicted molecular mass/pI to 30.8 kDa/5.67, which would be very similar to the molecular mass/pI of PSME3. In line with this argumentation are the findings of the peptide mass fingerprint analysis. In the less mobile spot featuring the presumable full-length ANXA4 the most C-terminal tryptic fragment found was peptide 293-299. In contrast, in the PSME3/ANXA4 spot the most Cterminal ANXA4 fragment found was peptide 259 -269, suggesting the presence of a C-terminally deleted ANXA4 fragment in this spot.
Differential PSME3 Expression Detected by Western Blot-To confirm the differential expression of PSME3 in normal and colon tumor tissue, antibody against PSME3 was developed for the use in immunoblot analysis and immunohistochemistry. The full-length protein was expressed in Escherichia coli as recombinant protein, and polyclonal antiserum was obtained from immunized rabbits. Fig. 2 shows a representative immunoblot analysis for six exemplary patients. The blot shows a strong signal at an apparent molecular mass of 31 kDa from all six tumor tissue lysates, whereas only weak signals were obtained from four of the adjacent normal colon tissues. In the lysates of the other two normal tissues, difference of expression compared with the matched tumor tissue was somewhat less pronounced. A signal of the same size was obtained with the recombinant antigen. Therefore, Western blot analysis confirmed elevated levels of expression in neoplastic colon compared with normal colon tissue.
Analysis of PSME3 Expression by Immunohistochemistry-Paraffin-embedded tissue was stained using the polyclonal rabbit antiserum raised against recombinant PSME3. Normal colon displayed only weak staining for PSME3 (Fig. 3A). Obviously only single epithelial cells were positive for the protein (Fig. 3B). In contrast, epithelial cells of colonic adenoma were uniformly stained by anti-PSME3 antibody (Fig. 3, C and D), and only very few epithelial cells within the adenoma were negative for PSME3. In the case of invasive adenocarcinoma, almost all cancer cells showed a strong staining for PSME3 (Fig. 3, E and F). Both in adenocarcinoma and adenoma as FIG. 1. Representative two-dimensional gel of human primary colorectal cancer. Tissue lysates of malignant (tumor) and healthy (control) stripped mucosa were prepared as described under "Experimental Procedures." 1.5 mg of protein were subjected to 2-DE with a first dimension pH gradient of 4 -7. Gels were stained with colloidal Coomassie Blue. Spots were excised, processed as described under "Experimental Procedures," and analyzed by peptide mass fingerprinting. Protein identification of selected spots is indicated. ANXA3, annexin A3; CTSB, cathepsin B; CTSD, cathepsin D.
well as in normal colon, staining for PSME3 was confined to the cell nucleus. The histologic data thus corroborate the observation of strong elevation of PSME3 expression in neoplastic colon tissue compared with normal colon.
Detection of PSME3 in Human Serum-After having established up-regulation of PSME3 in neoplastic colon tissue we assessed the potential release of this protein into the periphery and hence its potential value as serologic biomarker for the disease. For this purpose, a highly sensitive immunoassay for PSME3 was established, and the serum levels of the protein in healthy controls (n ϭ 317), in patients with benign bowel disease (n ϭ 87), and in patients with CRC (n ϭ 109) CEA serum levels were stage-dependent, being dramatically higher in UICC stage IV disease (median, 41.5 ng/ml) as compared with stage I disease (median, 2.7 ng/ml; Wilcoxon two-sample test, p Ͻ 0.0001). In contrast, PSME3 abundance was less stage-dependent with serum levels of 119 and 207 ng/ml in UICC stages I and IV, respectively, and the difference did not reach the same level of statistical significance (Wilcoxon two-sample test, p ϭ 0.0245).

FIG. 2. Up-regulation of PSME3 in primary colorectal cancer.
Tissue lysates of tumor (T) and matching adjacent healthy tissue (N) of six patients were prepared as described under "Experimental Procedures," and 10 g of protein were resolved on 4 -12% NuPAGE gels. The proteins were blotted onto a nitrocellulose membrane and detected with polyclonal antisera raised against PSME3. Ag, 1 ng of recombinant PSME3 was loaded. Lane M, molecular mass markers.  The relationship between the specificity and sensitivity of PSME3 measurements for the detection of CRC is represented by a receiver-operating characteristic curve (19) (Fig.  4). The area under the curve was 0.77 for PSME3 and 0.76 for CEA. Therefore, diagnostic accuracy of both markers was in a comparable range. DISCUSSION In the present report we show elevated expression of PSME3 in CRC when compared with normal colon tissue. Differential expression of the protein was identified using a proteomics approach based on protein separation by 2-DE. Comparative analysis of the resulting 2-DE gels was not done by visual comparison of spot pattern and spot intensities ("image analysis"). Instead all spots from every gel were excised followed by protein identification using peptide mass fingerprinting by MALDI-TOF. Using this strategy, PSME3 was identified in CRC tissue from seven of 10 patients and in none of the adjacent normal tissue samples, indicating elevated expression of the protein in CRC compared with normal colon tissue.
Intriguingly visual appearance and intensity of the PSME3containing spot on the tumor gels and the matching spot on the control gels were essentially the same (Fig. 1). However, mass spectrometric analysis revealed that the PSME3-containing spot on the tumor gels enveloped a second protein, ANXA4. On the other hand, in the matching spot on the control gels, only ANXA4 was detectable. Presumably ANXA4 contributed to most of the spot intensity of the PSME3/ ANXA4 spot on the tumor gel as well, impeding identification of dysregulation of PSME3 by image analysis strategies. In contrast, differential expression of PSME3 was easily detectable by the mass spectrometry-based approach. This example corroborates the recent work by Campostrini et al. (15).
They analyzed the extent of spot overlapping in complex 2-DE gels in detail and conclude that on typical 2-DE gels, loaded with 1 mg of total protein, "the singlets will be the minority, rarely exceeding 30% of all spots analyzed." The relevance of their findings has been challenged by Hunsucker and Duncan (20). They argued that "the enormous dynamic range covered by proteins in biological samples works to our advantage" and pointed out that for those proteins that are visible on a 2-DE gel "in all but a few instances the measured (total) intensity will be derived from essentially one principal component." In principle, we are in agreement with this evaluation. However, the dynamic range covered by proteins in biological samples may not always work to our advantage but also to our disadvantage. We show in the present work that spot overlapping can indeed lead to masking of relevant dysregulated proteins and that a strictly MS-based approach for the analysis of proteins displayed on 2-DE gels is superior to image analysis-based strategies in such cases. However, one can only speculate how frequently spot overlapping actually may lead to masking of dysregulated proteins on 2-DE gels, impeding their recognition by image analysis, and whether our results on PSME3 represent rather a rare exception. Importantly the reported finding of up-regulation of PSME3 in tumor tissue based on mass spectrometric analysis of 2-DE gels was unambiguously confirmed by independent immunological methods, namely Western blotting and immunohistochemistry (Figs. 2 and 3). Applying these technologies, we found a generally weak expression of PSME3 in normal colon tissue and strongly elevated expression of the protein in neoplastic tissue from the colon. The fact that we did not detect PSME3 expression in normal colonic tissue on 2-DE gels and MALDI mass spectrometry at all simply reflects a lower limit of detection of the latter technology compared with antibody-based approaches.
PSME3 is a member of the PA28 family of proteins (for a review, see Ref. 21), which have been shown to bind specifically to 20 S proteasomes and stimulate the hydrolysis of peptides. Generally proteasomes are responsible for the degradation of cellular proteins in the cytosol and nucleus of eukaryotic cells and thereby play an important role in many cellular processes, including cell cycle progression (22). The 20 S proteasome consists of 28 protein subunits and possesses three different proteolytic active sites with different specificities. Proteasomes are activated by protein complexes that bind to the outer rings of the complex. The best known activator is PA700, also known as 19 S, which binds to the 20 S proteasome to form the 26 S proteasome. The 26 S proteasome recognizes ubiquitin-conjugated proteins, which are subsequently degraded in an ATP-dependent manner (for a review, see Ref. 23). In contrast, the PA28 protein complexes, also known as 11 S, bind to the 20 S core, yielding an active, ATP-independent peptide degrading complex that does not require ubiquitinated proteins (24).
There are three PA28 homologs, called PA28␣, -␤, and -␥. PA28␥, also known as PSME3, is expressed at high levels in the brain compared with moderate levels in other organs. In contrast, PA28␣ and -␤ are virtually absent from the brain and are particularly abundant in immune tissues. PA28␣ and -␤ are mainly expressed in the cytoplasm and seem to play a role in the immune system, enhancing the production of peptides for loading of major histocompatibility complex class I molecules. Expression of PSME3 is confined to the nucleus (25). Several studies indicate a potential role of PSME3 in cell cycle traverse, apoptosis, or both, and it was suggested that PSME3 is an antiapoptotic factor (21, 26 -30) The proposed antiapoptotic activity could explain its high abundance in the adult brain because neurons should be well protected against self-destruction. In addition, an antiapoptotic function could also explain the findings by Okamura et al. (31), who reported abnormally high expression of PSME3 in thyroid cancer as estimated by immunohistochemistry and Western blotting.
In this study, we describe elevated expression of PSME3 protein in neoplastic tissue of the colorectum, whereas normal colonic epithelium expressed PSME3 only at weak levels. It is conceivable therefore that PSME3 might have an antiapoptotic function in this tumor entity as well. Immunohistochemical analysis demonstrated PSME3 expression both in colonic adenoma and invasive cancer (Fig. 3) suggesting that PSME3 may play an important role during all phases of tumorigenesis. However, more detailed immunohistochemical analysis including tissue from all tumor stages and pathological classifications will be necessary to further elucidate the correlation between PSME3 expression and neoplastic transformation of epithelial cells in the colorectum. In accordance with previous studies and a potential role in cell proliferation, expression of PSME3 was confined to the nucleus. It is tempting to speculate that PSME3 may have a general role in tumorigenesis and is not confined to thyroid and colorectal cancers. Future studies have to show whether the findings for colorectal and thyroid cancer can be generalized to other neoplasms.
Interestingly a second member of the PA28 family of proteins, PSME1, also known as PA28␣, was also detected in our analysis. However, this protein, which binds to the 20 S proteasome in the cytosol to form the so called "immunoproteasome," was not elevated in colon cancer (Fig. 1). This finding underlines the specificity of the reported elevated expression of PSME3 in colon cancer.
After having shown strong elevation of PSME3 in CRC tissue compared with adjacent normal colon we speculated that PSME3 might be released from tumor cells, giving rise to elevated PSME3 levels also in serum of CRC patients. Therefore, we developed a highly sensitive immunoassay and tested the level of PSME3 in a large number of sera from CRC patients, healthy blood donors, and patients with benign bowel disease. As a benchmark, we also assessed serum levels of CEA, the established tumor marker for CRC. Median serum levels of CEA were elevated from 1.3 ng/ml in healthy donors and 1.8 ng/ml in benign bowel disease to 3.6 ng/ml in CRC patients (Table I). With regard to PSME3, we found a strong elevation of PSME3 in the blood of cancer patients. The median serum level for PSME3 was 4 times higher in the cohort of CRC patients (114.0 pg/ml) compared with healthy blood donors (28.1 pg/ml) and about twice as high as in the patients with benign bowel diseases. Therefore, PSME3 could prove to be a new, sensitive, and specific marker that assists in the detection of CRC. The present study represents the initial validation of PSME3 as a potential serological surrogate marker for CRC, and it is self-evident that more patient sera need to be tested, including pre-and postoperative samples, other cancers, etc., to understand the potential value of this new biomarker.
In a previous report of this study we reported identification and validation of five exemplarily selected proteins that were elevated in CRC tissue samples (16). Up-regulation was confirmed for all five proteins by immunoblot analysis of tissue lysates. However, association of elevated serum levels with the presence of CRC could only be shown for one protein, nicotinamide N-methyltransferase (NNMT), by developing a sensitive immunoassay and testing serum samples. Two other proteins tested of the five selected proteins showed no association with the disease when tested in serum samples. Elevation of NNMT in sera was highly significant, and discrimination between CRC patients and healthy donors was comparable to that of CEA (area under the curve was 0.84 and 0.78, respectively, using a receiver-operating characteristic curve).
The present report is the first to demonstrate elevation of PSME3 in colorectal tumors and to show a correlation of PSME3 serum levels with the presence of CRC. Due to spot overlapping with ANXA4, dysregulation of PSME3 in tumor tissue (by means of comparative 2-DE) could only be detected by applying MS-based analysis instead of the widely used image analysis. Together with the previously described cancer marker NNMT, this is the second example of a novel validated serologic cancer biomarker that could be identified by means of combining 2-DE with a strictly MS-based data analysis of tissue samples from CRC patients. These two examples indicate that this approach may have high potential in biomarker discovery.
The potential clinical value of PSME3 might be best discussed in the context of CEA, which is probably the best current single tumor marker for CRC. Although CEA is not recommended for early detection of CRC due to a lack of specificity and sensitivity (32), it is still one of the most valuable tumor markers we currently have. Measurement of serum CEA levels is recommended preoperatively as an independent prognostic factor, for surveillance of recurrent disease after tumor resection, and for monitoring treatment of advanced disease (33). Since the discovery of CEA as a cancer marker in 1965 (34,35) and the development of a first immunoassay to measure circulating CEA in serum (36), no biomarker has been established that fulfills the requirements for early detection of CRC, namely sufficiently high sensitivity and specificity for the disease, regardless of considerable research efforts during the past 4 decades. Not even could any novel marker so far displace CEA in the management of the disease. Nevertheless "omics" technologies represent a new and promising avenue to identifying additional novel biomarkers for CRC and other diseases. Two such markers, PSME3 and NMMT, are being described in this report and by Roessler et al. (16), respectively. Both display a comparable diagnostic performance to CEA based on a preliminary analysis in a relatively small serum panel. Although from the experience of the past 37 years of research the expectations to find the "golden bullet" (i.e. one single biomarker suitable for early detection of CRC) are low, novel markers may well enhance our capability of detecting early CRC by serum assays. A number of reports have clearly shown that the combination of several markers for a disease by applying multivariate analysis can significantly improve the diagnostic performance (37)(38)(39)(40). It yet has to be investigated and shown by using appropriate algorithms in multivariate analysis whether the combination of known markers and of novel omics markers can improve the early detection of CRC. However, for statistically significant results still much larger sample numbers are required than for univariate analysis. Respective studies have been initiated in our laboratory.
* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.