• E-mail
• Print
• Comment
• Font Size
• Digg
• del.icio.us
• Discuss article

Epithelial Cell Preparation for Proteomic and Transcriptomic Analysis in Human Pancreatic Tissue

Posted on: Tuesday, 14 September 2004, 06:00 CDT

Abstract

Standardized sample preparation procedures constitute a prerequisite for obtaining reliable and reproducible results in gene expression research in humans. In particular, in diseases such as pancreatic cancer and pancreatitis, isolating epithelial cells is an important step preceding such research. In pancreatic tissue, the high amount of RNAases is a further problem when it comes to obtaining high-quality RNA, and the presence of secreted proteases accelerates protein degradation. We developed a successful method that addresses these different problems. This method, which uses epithelial cell surface antibody Ber-Ep4, proteases, and RNAases inhibitors, leads to a significant enrichment (>95% purity) of epithelial cells from fresh human tissue samples and allows for both proteomics (Western Blot, 2D PAGE) and transcriptomics studies (rtPCR, cDNA microarray). Compared with other cell purification procedures, this method is characterized by several advantages: a large quantity of cells available for downstream analysis, combined transcriptomics and proteomics studies using the same samples, better reproducibility of proteomics studies, and an acceptable yield (63%) for gene expression arrays studies. Moreover, a quality control protocol addressing the needs of the industry and the requirements of regulatory agencies is proposed.

2004 Elsevier GmbH. All rights reserved.

Introduction

The identification of tumor-specific expression patterns at the protein and/or mRNA level may allow us to develop new therapeutic or diagnostic targets. For example, proteins that are overexpressed and essential for growth or cell survival are interesting targets for the development of specific inhibitors or antibodies.

However, there are several critical preconditions for the success of such a development. The outcome of tissue biopsy done in the operating room or in the pathology laboratory is usually influenced by the macroscopic aspect of tissue, but diagnosis must be confirmed by a surgical pathologist, using accepted techniques (histopathology, immunohistochemistry, etc.). Unfortunately, tissue located next to the biopsy is barely examined. Furthermore, diagnosis has to be classified and staged according to international standards (e.g. the International Classification of Diseases of the World Health Organization, or the Tumor-Node-Metastasis Classification of the International Union against Cancer) to permit later comparisons. Diagnosis of "cancer" vs. "normal" is clearly insufficient, in particular at a timepoint when the close molecular links between cancer and inflammation begin to be highlighted.

Purification procedures include the yield of homogeneous samples and isolation of individual cell types from clinical material. For gene expression studies in pancreatic cancer, samples of the tumor are usually compared with those obtained from corresponding non- neoplastic tissue of the same or of a different patient. The materials of choice have often been homogenates of whole tissue samples. However, these samples have some considerable disadvantages: they are highly complex, and are usually composed of several different cell types and extracellular matrices, for example non-neoplastic pancreatic tissue encloses ductal and acinar cells, various neuroendocrine cells, and mesenchymal cells. Pancreatic lesions, like ductal carcinomas, consist of carcinoma cells, a sometimes abundant desmoplastic stroma, and leukocytes. Thus, one has to be aware that using samples of tissue homogenates does not simply mean a comparison of neoplastic vs. non-neoplastic epithelial cells, but a complex mixture of proteins of diverse origin, some of them deriving from epithelial cells.

A further point of interest is the quality of isolated nucleic acid and protein samples. In all purification methods, the degradation of RNA is time-dependent. When rtPCR is possible, it is generally impossible to obtain good transcription data (e.g., cDNA or oligonucleotide microarray from clinical samples). Similar problems arise in studies of proteomics, but at a larger scale, because such studies usually require large amounts of material, with no protein amplification technology being available at the present time. Moreover, protein solutions are contaminated not only with foreign cells, but also with body fluids, such as serum, blood, secretions, and necrosis. In pancreatic tissue, the high amount of RNAases is a further problem for obtaining high-quality RNA, and the presence of secreted proteases accelerates protein degradation.

This article deals with the preparation of the appropriate biological material for expression studies at the protein and/or mRNA level in pancreatic diseases. For the following theoretical considerations and the description of technical approaches, we chose cancer as an example. However, the notes can be extrapolated to other diseases, in particular pancreatitis, as well as to other tissue types.

Materials and methods

Preparation procedure

A previously described method for isolating cells [1] has now been developed further, and is evaluated for proteomics (Western Blots, 2D-electrophoresis) and transcriptomics (cDNA arrays) studies. The main quality criteria applied to the samples are number and purity of cells, quality of immunological detection of proteins in Western Blots, quality of electrophoresis, and quantity and quality of extracted RNA.

Patients and samples

From May 2002 to February 2004, 60 samples were obtained during surgery for pancreatic diseases at the Universities of Magdeburg, of Berlin (Charit), of Erlangen, and at the Academic Hospital Cottbus (Germany). Research protocols were approved by the Institutional Review Boards of the respective institutions. Informed consent was obtained from each patient.

Mechanical isolation procedure

Immediately after surgery, all tissue samples were placed on ice. Samples obtained from the tumor and matching non-tumorous mucosa were placed in wash buffer no later than 20 min after resection. Selection was done by an experienced consultant histopathologist. After sufficient samples were obtained for diagnostic purposes, the pathologist decided which portion of the neoplastic and non- neoplastic tissue can be used for cell preparation (Fig. 1A). All samples were washed in ice cold preparation buffer. Tumor and corresponding non-tumorous tissue were cut into slices of about 1 mm (Fig. 1C), pressed through a steel mesh (Fig. 1D), and collected in ice cold preparation buffer (Fig. 1E). The suspensions were centrifuged at 300g for 10min at 4C, followed by two washings of the pellets in 15ml preparation buffer. Two hundred microliters of the cell suspension was removed from the last washing step for cytological control (Fig. 1B). The pellets were collected in a 15ml conical tube and used for immunological separation.

Immunological separation

Prior to use, the Ber-Ep4 beads (Dynabeads Epithelial Enrich, Dynal, Norway) were washed twice with water in 1.6ml micro- centrifuge tubes. This procedure was carried out according to the manufacturer's recommendations using a magnetic particle concentrator (MPC-1, Dynal). After the last washing step, the beads were resuspended in separation buffer to give the original concentration. Eighty microliters of BerEp4 beads (~3.2 10^sup 7^ beads) was used for the immunological separation of 15ml cell suspension. The suspension was moderately shaken for 30 min at 4C. The dynabeads were then separated from suspension by using a magnetic particle concentrator (MPC-E-1, Dynal) (Fig. 1G). The beads were washed twice with separation buffer, resuspended in 8 ml separation buffer, and divided into 1 ml aliquots. After centrifugation, the supernatant was removed, and the samples were stored at -80C for further analysis (Fig. 1H).

Fig. 1. Mechanical preparation steps. Surgical specimen after duodenopancreatectomy (so-called Whipple procedure) is shown in (A). The areas where the tissue was removed for preparation is indicated, including margins for histological controls. To explain how many different cells are in tissue, and to show Ber-Ep4 antigen expression, Ber-Ep4 stain is shown in (B). These different cell types represent normal duct cells (nd), lymphocytes (ly), connective tissue (ct), and tumor cells (tu). Steps of the preparation method are indicated in C-H. Tissue slices (C) are passed through a sieve (D). Collected cells (E) result in differently sized pellets (F). After magnetic separation of Ber-Ep4-bound cells (G), normal tissue sample (NT) and tumor sample (Tu) are aliquoted (H).

Quality control of cells with proteomics and transcriptomics procedures

Protein preparation for Western blotting and 2D electrophoresis

The purified epithelial cells were denatured with four pellet volumes of denaturation buffer (urea 420.42g, thiourea 152.24g, Amberlite IRN-150L (Pharmacia) 100 g, CHAPS 40g, dithiothreitol (DTT) 15.42g, Servalyt 4-9 T 50 ml, Bromophenol blue trace, HPLC- grade water up to 11). After addition of the denaturation buffer, the samples were vortexed, sonicated three times for 10s on ice, and centrifuged for 10 min at 4C with maximum speed. The protein concentration was determined using the Bradford assay.

2D electrophoresis

For 2D electrophoresis, 3 mg of protein sample was solubilized in 300 l rehydration buffer a\nd transferred onto the focusing tray. The IPG strips were positioned upside down onto the tray. The gel and the sample were covered with about 2ml low viscosity oil (Mineral oil from Bio-Rad) to avoid evaporation. The strips were rehydrated for 8 h at low voltage (50 V).

Focusing

The strips were focused at 20C under increasing voltage from 300 to 3500 V for 3h, followed by an additional 3 h at 3500 V (intensity of 50 A per strip (2 mA max. in total)). Voltage was increased to 10,000V until a volt hour product of 80-100kV h was achieved.

Running the second dimension using the Ettan Dalt II (Amersham)

The Buffer Kit from Amersham (Cat. No. 17-6002-36) was used according to the instructions. The strip was placed on top of a 12.5% gel. Protein molecular weight standard was loaded onto a wick at the right end of the strip. The sealing solution is pipetted on the top of the gel. Gel was run at 2.5W/gel for 30min and 19W/gel for 5h.

Western blotting

For Western blotting, cell pellets were homogenized in a buffer at pH 6.0 containing 50mM sodium phosphate, 0.2 M NaCl, SmM EDTA, 100 M E-64, and 1 mM PMSF using first a homogenizer and then a sonicator. Cell lysates were centrifuged for 10 min at 4C at 10,000g. The supernatant was mixed with 5 Laemmli-buffer containing 20% DTT and boiled for 5 min. Proteins were quantified in all samples according to the Bio-Rad DC Protein Assay (30 g protein was used for SDS-PAGE).

Proteins were transferred onto nitrocellulose membranes (Schleicher & Schuell, Dassel, Germany) by electroblotting for 60 min at 100mA. Membranes were incubated with the mouse anti- cytokeratin 7 antibody (1:500), followed by incubation with a secondary, peroxidase conjugated goat anti-mouse antibody. For development, the Super Signal West Pico Chemiluminescent Substrate Kit was used.

RNA quality control and gene expression analysis on cDNA microarrays

Total RNA was extracted from the purified epithelial cells using the RNAeasy kit, Qiagen (Hilden, Germany), according to the manufacturer's recommendations. RNA quality was checked with the RNA 6000 Nano Lab Chips (Agilent, Palo Alto, USA) using Agilents 2100 Bioanalyzer system. Cyanine-3 and cyanine-5 labeled cDNA targets were generated from 10 g total RNA using the LabelStar kit (Qiagen). In each experiment, Cy-3-labeled cDNA from non-tumor epithelial cells was mixed with Cy-5-labeled cDNA from tumor epithelial cells and hybridized to the human-1 cDNA microarray from Agilent. The hybridization conditions were chosen according to the manufacturer's recommendations.

Fig. 2. Immunological purification of epithelial cells and comparison of expression patterns from epithelial cells w/o RNALater. Comparison of purified epithelial cells before (A) and following (B) epithelial cell isolation using BerEp-4 epitope bound to magnetic beads. Cellular stains were performed using Vimentin and pan-Cytokeratin antibodies. Magnification 20 . Gel like view (picture from Bioanalyzer) of RNA quality visualized by 18S and 28S RNA quantity and purity (C). Lane 1: Standard Size Marker. Lane 2: RNA quality without RNALater incubation. Lane 3: RNA quality using 50% RNALater in preparation buffer. Lane 4: RNA quality using 40% RNALater in preparation buffer. Lane 5: RNA quality using 30% RNALater in preparation buffer. Lane 6: RNA quality using 20% RNALater in preparation buffer. Lane 7: RNA quality using 10% RNALater in preparation buffer. Western Blot analysis of cytokeratin 7 expression level (D). Lane 1: Cytokeratin 7 staining from epithelial cells prepared without RNALater. Lane 2: Cytokeratin 7 staining from epithelial cells prepared in the presence of 50% RNALater. Lane 3: Cytokeratin 7 staining from whole tumor tissue extracts prepared without RNALater. 2 D PAGE analysis of epithelial cell extracts from pancreas tissue (E). Left 2D gel shows protein expression pattern from pancreatic epithelial cells prepared without RNALater, right 2D gel shows protein expression pattern from pancreatic epithelial cells prepared in the presence of 50% RNALater.

Results

In 2001, Ott et al. described a method that led to a significant enrichment (>95% purity) of epithelial cells obtained from fresh human tissues [1]. Briefly, immediately after surgery and histological classification, tumor and corresponding non-neoplastic tissue is mechanically dissociated and incubated with magnetic beads coupled to epithelial cell surface antigen Ber-Ep4. Subsequent purification of the bead-cell-complexes results in a specific epithelial subpopulation (Fig. 2A and B) and in a left over mess- culture of the remaining non-BerEP4-immunoreactive cell types, most of which are probably non-epithelial in origin.

We tested and established a method that allows us to conduct parallel proteomics and transcriptomics studies of samples obtained from the same surgical specimen. In doing so, it is possible for the first time to compare RNA and protein data from a single patient and tumor patient cohorts, and to make a comparison between tumor and non-tumorous tissues.

RNALater is routinely used during RNA purifications to stabilize RNA and to avoid its degradation. Its effect on cellular proteins and antigen-antibody binding (fundamental to the above described purification method) still needs to be clarified.

In a first set of experiments, fresh pancreatic tumor and corresponding non-neoplastic tissues were incubated (immediately after histological classification using frozen sections; ~15min after resection) in preparation buffer containing different amounts of RNALater (between 0% and 50%). Consequently, the epithelial cells were prepared in the presence of respective percentages of RNALater.

Comparing the amount and the purity of the isolated epithelial cells, pellet size and specific epithelial control staining did not show any effect, as RNALater (data not shown) was used. In a next step, the quality of isolated RNA was controlled. As illustrated by the 'gel-like view' image obtained from the Bioanalyzer (Agilent Inc.), the quality of isolated RNA decreased with lower amounts of RNALater (Fig. 2C). Best results could be obtained using a 1:2 mixture of the preparation buffer and RNALater (Fig. 2C, lane 3), while concentrations of 20% and 10% RNALater in the preparation buffer hardly protected the RNA (Fig. 2C, lanes 6 + 7, respectively) from degradation. Intermediate levels of RNALater showed partial protection of the isolated RNA (Fig. 2C, lanes 4 + 5).

To test the quality of isolated RNA for experimental purposes, RNA from pancreatic epithelial cells purified in the presence of 50% RNALater and from cells without RNALater was labeled with Cy3 and Cy5 fluorescence dye and hybridized to a 12 K cDNA chip from Agilent Inc. A very high background was found when RNA preparation was not protected by RNALater. In particular, no different RNA expression patterns were found. By contrast, protection of RNA by RNALater resulted in reproducible and differential expression patterns and in a complete removal of the experimental background (Fig. 3B).

These experiments show that the implemented modifications and improvements of our cell purification method allow us to screen freshly purified human tissues for differential gene expression. Out of 60 samples prepared, 38 (63.3%) fulfilled the industrial quality control standards requested for downstream analysis, in particular for gene expression arrays. Finally, we addressed the question of how RNALater influences parallel proteomics studies. The quality and success of protein preparation was assessed by Western blotting using an antibody directed against cytokeratin 7, which is commonly expressed in non-neoplastic and neoplastic pancreatic epithelia. As shown in Fig. 2D (lane 1 + 2), a set of similar bands of 35-50 kDa were detected by Western blotting using either our improved cell preparation technique or conventional homogenates of whole tissue samples. However, by adding RNALater to the preparation procedure (Fig. 2D, lanes 1+), additional lower molecular weight bands were identified, which might be due to either protein stabilization or degradation products (Fig. 2D, lane 2).

We then applied our samples to 2D PAGE analysis, which allows for the resolution of far more proteins. The use of RNALater significantly influenced neither the overall number of proteins detected nor that of the expression levels from representative proteins (Fig. 2E). The results of both Western blotting and 2D PAGE analysis show that the addition of RNALater does not impair proteomics studies. Thus, the same starting material can be applied to proteomics and transcriptomics studies.

Discussion

Proteomics and transcriptomics technologies are now widely used to study the pathology and pathogenesis of malignant pancreatic tumors. The quality of the data obtained from these studies largely depends on the starting material. An accurate sampling and characterization of diseased and non-diseased pancreatic tissue samples is mandatory. In practice, a number of obstacles influence the quality of the sample, some of which are addressed by our cell preparation technique.

Isolation of epithelial cells

Pancreatic tissue consists of various epithelial cell types: duct cells and acinar cells; a subtype of them has neuroendocrine functions and epithelial stem cells. The latter can be differentiated not only into epithelial cells, but also into islet cells [2]. Techniques that allow us to obtain only epithelial cells are, e.g., laser capture microdissection (LCM) and immunological techniques, such as fluorescent assisted cell sorting (FACS) or procedures using immobilized antibody approaches. LCM has been widely used for studying proteins and RNA in selected cell populations. The main limitation for its use in proteomics and transcriptomics studies is the length of dissection, which is required to produce sufficient cells for analysis. Another problem lies in the stainingof cells to distinguish between different cell types. Staining with e.g., Hematoxylin and Eosin interferes with proteins and is not appropriate for proteomics analyses [3]. In general, staining is time-consuming and exposes cells to oxygen, which might lead to post-translational modifications of cellular proteins. For mRNA assays, this drawback might be negligible in certain tissues. However, in the case of pancreatic tissue, time is an important factor owing to the high levels of degrading enzymes, such as RNases. Nevertheless LCM is able to produce suitable cells for subsequent analysis of proteomics and transcriptomics. Another technical approach to obtaining distinct cell populations is FACS. FACS-based immunological approaches are simple and fast, and sufficient amounts of cells can be collected [4]. An advantage of FACS is that not only surface molecules, but also intracytoplasmatic antigens can be used for separation. This makes the method more specific for a distinction between different epithelial cells, such as acinar or ductal cells from pancreatic tissue. The FACS approach is similar to the method (immobilized antibody approach) described herein. Using our cell preparation method, it is possible to isolate pure epithelial cells from pancreatic tissue omitting the limitations mentioned above. The method allows us to purify 5 10^sup 6^-10^sup 8^ epithelial cells (>95% purity) of most epithelial tumors and corresponding healthy tissues within 30-60 min. In addition, an expensive equipment is not necessary. Ber-Ep4 surface protein is expressed in nearly all epithelial cells, including those of the pancreas, making it a suitable choice for isolating epithelial cells from pancreatic and other tissues. As yet, pancreatic cancer-surface proteins, which are specific for ductal cells, have not been identified.

Fig. 3. Microarray analysis of RNA purified from epithelial cells w/o RNALater. RNA from epithelial cells purified in the presence of 50% RNALater and from epithelial cells prepared without RNALater was labeled with Cy3 and Cy5 fluorescence dye and hybridized to a 12K cDNA chip. While 'unprotected' RNA does not show differential gene expression between tumor and healthy tissue (A), as well as high unspecific background (A), 'protected' RNA shows reproducibly differential gene expression patterns (B) and no unspecific background (B). To visualize differences better, each experiment is shown in two different magnifications.

Analysis of proteomics and transcriptomics

The complexity of neoplastic and non-neoplastic tissue will undoubtedly influence any comparative study. When tissue homogenates are used, artifacts are possibly introduced. This study bias can be overcome by immunoisolation and LCM of specific cell populations. Our strategy, i.e., the immunoisolation described here, allowed us to extract proteins and mRNA from the same tissue in a simple, fast, and high quality manner. The procedure of isolating proteins from such samples was already described by Ott et al. [1]. Pancreatic tissue is rich in RNase that will impair RNA extraction. By modifying this standard protocol, we were able to obtain good yields of high-quality RNA suitable for analysis of transcriptomics (Fig. 2D and E). Therefore, our method greatly facilitates comparisons of specific cell types at the protein and DNA levels, and reduces the risk of tissue and cell contamination. In addition, simultaneous analysis of the transcriptome and proteome of the same sample largely improves studies of gene translation and also confirms the origin of the proteins found (Figs. 2 and 3).

Global approaches

Analysis of a whole system of molecular components, so-called system biology, is useful only when the analyses are based on well- defined perturbations of well-standardized systems. The pathologist is now confronted with problems that were not encountered in in vivo studies of animal models or in in vitro studies using cell lines. As mentioned above, successful isolation of individual cell populations from a complex tissue sample bears great advantages, because analysis of whole tissue lysates may be more difficult to interpret, as the relative amount of the individual components, as well as their contribution to the study results, is often unknown. Conversely, in theory, any sample preparation procedure can endanger sample integrity and stability. Thus, whole tissue samples may bear the advantage of only minimal manipulation representing more closely the in vivo situation. In addition, tumor-host interactions are undoubtedly important regarding tumor biology. Analysis of whole tissue lysates may provide information about these interactions, which will probably be missed when only single cell populations are studied. In the present study, we analyzed different steps of sample preparation and can now make precise statements regarding method- induced modifications. Of course, we cannot exclude any further undetected modification. However, by applying the same preparation procedure to paired samples, we expect that such modifications will have no effect on differential display results.

Intensive bioinformatics represents another method for overcoming limitations of an unknown relative and absolute composition of a whole tissue sample. However, experimental designs must then analyze molecular components with the use of different technologies across a set of different tissues. In this approach, a normalization step is implemented. Prerequisite for this is the knowledge about expression levels (proteins or mRNA) in the different cell types accompanying a tumor. These cellular protein maps might be obtained by studying individual cell populations. Thus, our technique may prove to be a valuable adjunct to proteomics studies of cancer biology. It will probably not answer all questions related to tumor biology, as tumor- host interactions might be missing. However, when a more global examination is wanted, most of the cell types not enriched in a first step can be isolated by using our method in a following step.

In summary, the above described procedure represents an advanced and highly efficient method for profiling human epithelial tissues, performing parallel transcriptomics and proteomics analysis on the same tumor specimen.

This method, using the epithelial cell surface antigen Ber-Ep4, as well as proteases and RNAases inhibitors, leads to a significant enrichment (>95% purity) of epithelial cells obtained from fresh human tissue samples, allowing for both proteomics (Western Blot, 2D PAGE) and transcriptomics studies (rtPCR, cDNA microarray). Compared with other cell purification procedures, this method is characterized by several advantages: a large quantity of cells available for downstream analysis, a combination of transcriptomics and proteomics studies using the same samples, a better reproducibility of proteomics studies, and an acceptable yield (63%) for gene expression arrays studies. Moreover, a quality control protocol addressing the needs of the industry and the requirements of regulatory agencies is now available.

In our experience, this method has expedited the process of identifying new, significant, and relevant therapeutic targets in cancer research, in particular by enabling sample standardization between different clinical institutions, and it has facilitated collaboration between academia and industry.

Acknowledgements

We thank Chr. Rcken for his critical and productive discussion of this manuscript; also M. Stoklasek for her excellent technical assistance, and L. Fels, T. Buschmann, and S. Krger for their significant contributions to this work.

Duality of interest

This study was supported by Europroteome AG, Hennigsdorf, Germany. The procedures described in this communication are protected by the following patents and patent applications: WO9843091, EP 0970377, AU 726738, CA 2,284,272, US 09/381,266, Japan 10-545-232, DE 102 22 494.3, WO03097874. UK, RS, HUS and MAR are consultants to Europroteome AG. The terms of these arrangements are transparent to the employers and managed by the academic institutions in accordance with their conflict of interest policies.

References

[1] V. Ott, K. Guenther, R. Steinert, S. Tortola, B. Borisch, W. Schlegel, M.A. Reymond, Accuracy of two-dimensional electrophoresis for target discovery in human colorectal cancer, Pharmacogenomics J. 1 (2) (2001) 142-151.

[2] J. Peters, A. Jurgensen, G. Kloppel, Ontogeny, differentiation and growth of the endocrine pancreas, Virchows Arch. 436 (6) (2000) 527-538.

[3] J. Mojsilovic-Petrovic, M. Nesic, A. Pen, W. Zhang, D. Stanimirovic, Development of rapid staining protocols for laser- capture microdissection of brain vessels from human and rat coupled to gene expression analyses, J. Neurosci. Methods 133 (1-2) (2004) 39-48.

[4] T. Ordog, D. Redelman, L.J. Miller, V.J. Horvath, Q. Zhong, G. Almeida-Porada, E.D. Zanjani, B. Horowitz, K.M. Sanders, Purification of interstitial cells of Cajal by fluorescence- activated cell sorting, Am. J. Physiol. Cell. Physiol. 286 (2) (2004) C448-456.

Udo Kellnera(a),*, Ralph Steinert(b), Volker Seibert(c), Steffen Heim(c), Angela Kellner(a), Hans- Ulrich Schulz(d), Albert Roessner(e), Sabine Krger(e), Marc Reymond(d)

a Department of Pathology & Cytology, Klinikum Minden, Friedrichstrasse 17, 32427 Minden, Germany

b Department of Surgery, Karl-Thiem-Klinikum, Thiemstrasse 111, 03048 Cottbus, Germany

c Europroteome AG, Neuendorfstrasse 24b,16761 Hennigsdorf, Germany

d Department of Surgery, Otto-von-Guericke University Magdeburg, Leipziger Strasse 44, 39120 Magdeburg, Germany

e Department Pathology, Otto-von-Guericke University Magdeburg, Leipziger Strasse 44, 39120 Magdeburg, Germany

Received 4 February 2004; accepted 1 March 2004

* Corresponding author. Tel.: +49-571-801-2600; fax: + 49-571- 801-2611.

E-mail address: udo.kellner@klinikum-minden.de (U. Kellner).

Copyright Urban & Fischer Verlag 2004

More News in this Category