| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Human Cancer Biology |
Authors' Affiliations: Departments of 1 Pathology and 2 Urology, McGill University Health Center, Departments of 3 Pathology and 4 Urology, Centre Hospitalier Universitaire de Montreal, and 5 Caprion Pharmaceuticals, Inc., Montreal, Quebec, Canada
Requests for reprints: Joel Lanoix, Caprion Pharmaceuticals, Inc., 7150 Alexander Fleming, Montreal, Quebec, Canada H4S 2C8. Phone: 514-228-3629; Fax: 514-940-3620; E-mail: jlanoix{at}caprion.com.
 |
Abstract |
|---|
 Top Abstract Materials and MethodsResultsDiscussionReferences |
|---|
Experimental Design: Radical prostatectomies from 84 patients (median, 61 years; prostate-specific antigen, 5.9; 66% nonpalpable) were processed with alternate, mirror image slices submitted for histology and tissue banking. Benign and malignant foci were macrodissected from the banked sections using the pathologically mapped, mirror image histology sections as a guide. Epithelial plasma membranes were isolated using novel immunomagnetic purification and their purity was assessed. Tissue homogenates were probed by Western blot for malignant (AMACR) and benign (p63) markers to test the accuracy of this protocol. Selected banked tissue slices were retrieved, thawed, and compared pathologically to their corresponding routinely processed alternate slices.
Results: Plasma membrane preparations showed the enrichment of epithelial plasma membrane markers (prostate-specific membrane antigen and epithelial-specific antigen) with minimal marker expression from nonepithelial cells or intracellular organelles. Cancer homogenates showed up-regulated AMACR and down-regulated p63, whereas benign homogenates showed up-regulated p63 and down-regulated AMACR. There was 30% benign (p63+) contamination in cancer slices and <6% cancer (AMACR+) contamination in benign slices. Retrieved tissues showed the retention of immunoreactivity while their histology was always adequate for diagnosis.
Conclusions: We have successfully isolated purified epithelial plasma membranes from benign and malignant human prostates and provided validation data for the accuracy of our protocol in a prostate-specific antigen–screened cohort. Our method also enabled the retrieval of banked tissues for clinical purposes with the retention of good histologic and immunohistochemical quality.
In the past, standard protocols to isolate plasma membranes have used the principle of differential centrifugation coupled to isopycnic centrifugation to generate a fraction enriched in plasma membrane (3–5). However, these protocols suffer from two drawbacks that limit their use for the proteomic identification of immunotherapeutic targets. The first is the overlapping density between different organelles that results in the presence of various contaminants in the plasma membrane fraction. The second is the inability to purify plasma membranes of specific cell types from primary tissue. To overcome these problems, the velocity/isopycnic centrifugation method needs to be coupled with another purification approach that employs a different principle. Immunoisolation is such an approach as it relies on the specific recognition by high-affinity antibodies of their cognate antigens localized at the surface of the cell types that one wishes to isolate. Hence, this approach should permit the isolation of a plasma membrane preparation that is markedly reduced in contaminant organelles as well as the purification of the surface of epithelial cells among the heterogeneous cell mixture present in primary prostatic tissues.
Obtaining prostate cancer tissue for proteomic research is especially challenging in today's prostate-specific antigen (PSA) screening era of low volume cancers where there is competition for tissue between pathologists who need it for clinical diagnosis, grading, and staging of tumors, and researchers who need fresh tissue. Often, the amount of "surplus" tissue, or tissue available after clinical needs are met, is scant, and pathologists are hesitant to allow this to be used for nonclinical purposes.
A large part of the pathologist's hesitation arises from the fact that prostate cancer does not generally form a discrete, grossly visible mass separate from the benign gland as do tumors of some other organs (6). The accuracy of gross identification of cancer from freshly cut prostate tissue has been correlated with increased tumor size, grade and stage, but overall, suffers poor sensitivity in cases where the tumor is <10 mm in diameter (7). In the present era of clinically nonpalpable, low volume disease, gross examination is an even less reliable means of procuring prostate cancer tissue. What is required, in practice, is direct microscopic correlation. Because prostate cancer is notorious for its capacity to infiltrate between benign glands, histologic mapping of benign and malignant areas of banked tissue is necessary for a high yield of pure malignant and benign cells.
In the present study, we have sought to isolate purified epithelial plasma membrane preparations from matched human benign and malignant prostatic tissues using novel immunoisolation techniques. In order to accomplish this, we applied a tissue banking protocol for handling radical prostatectomy specimens and tested it with respect to accuracy in its yield of benign and malignant cells, and whether it permitted the retrieval of banked tissues for clinical diagnostic purposes.
|
Materials and Methods |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
Tissue processing. Prostatectomy specimens were processed under the supervision of a dedicated pathologist as follows: fresh specimen was collected from the operating room as soon as feasible, usually within 5 to 10 minutes after removal of the gland. After weighing and measuring the gland, prosection was carried out in a cool environment (5°C). We inked the anterior prostatic moiety in red and the posterior half in black. A hair dryer exuding cool air was used to dry the ink rather than immersion in the standard Bouin's fixative or acetic acid. Shave margins of the bladder neck were taken and the seminal vesicles were amputated and submitted for histology.
The prostatic urethra was cannulated with a probe and the prostate was cut into right and left halves. Each half was then sectioned into anterior and posterior quadrants by cutting along the junction of the red (anterior) and black (posterior) ink. We were thus left with four quadrants: right anterior, right posterior, left anterior, and left posterior.
Each quadrant was then sectioned similarly from apex to base using a dissecting blade. The first 4 mm representing the apical portion of each quadrant was amputated, cut sagitally and entirely submitted for histology. The quadrants were then cut at 2 mm intervals with the odd-numbered slices submitted for histology and the even-numbered slices submitted for tissue banking (Fig. 1 ). Because the pathology sections were embedded and processed with the cut surface facedown, we dotted the "mirror image" face of the subsequent tissue-banked slice with green dye for correct orientation later. The final 4 mm of prostate representing the base of each quadrant was cut sagitally and entirely submitted for histology.
|
25 minutes. Those tissue slices sent for pathology that were mirror images of the tissue-banked slices were embedded flat and the first full face was sectioned with minimal trimming of the block. Areas of prostate cancer and benign tissue were mapped on H&E stained sections. Digital images of the mapped sections were taken using a scanner and the files transferred onto a compact disc. These digitized slices could then be correlated to their mirror image banked slices (Fig. 2 ). Tissues were only released for research once the case was signed out by the pathologist.
|
Tubes containing tumor sections were thawed in a 37°C water bath. Immediately after thawing, tubes were transferred on ice. Each tissue slice was washed by dipping it into a beaker containing DMEM. Tissue slices were then transferred onto a weighing dish and matched with their corresponding digital images. Posterior (black) and anterior (red) margins, and the upper side of the slice (green dot), were used to help with orientation. Tumor areas were then macrodissected and put into a beaker containing DMEM.
Tissue homogenization and crude plasma membrane generation. Benign and tumor tissues were cut into small pieces (cubes, 3 mm in size), and homogenization buffer [250 mmol/L sucrose, 10 mmol/L Tris-HCl (pH 7.4)], 100 units/mL of DNase 1 (Roche, Laval, Canada), 5 mmol/L MgCl2, Complete protease inhibitor EDTA-free cocktail (Roche) was added at a concentration of 10 mL/g of tissue. Tissues were homogenized twice for 20 seconds using a polytron (Kinematica, Newark, NJ) set at speed 8 (20,000 rpm).
Homogenates were filtered through a 180 µm nylon mesh and centrifuged at 900 x g (2,000 rpm) for 10 minutes at 4°C. Supernatants were collected, brought to 12.5 mL with the homogenization buffer and transferred into 12.5 mL/Ultraclear centrifuge tubes (Beckman Coulter, Mississauga, Ontario, Canada). For each tube, a cushion made of 100 µL of 50% w/v sucrose was placed at the bottom. Samples were then centrifuged at 35,000 rpm (100,000 x g) for 60 minutes at 4°C to pellet the membranes. Membrane pellets were resuspended at 1 mL of homogenization buffer per gram of tissue and incubated with 500 units/mL of micrococcal nuclease (US Biologicals, Swampscott, MA) and 1 mmol/L of CaCl2 for 15 minutes at 4°C. To the resuspended membranes, 2.55 mol/L of sucrose solution was added to obtain a final sucrose concentration of 1.7 mol/L. To isolate crude plasma membranes, isopycnic centrifugation using discontinuous sucrose gradients was done as follows: on top of the 1.7 mol/L sucrose fraction containing membranes, the 1.5, 1.3, and 0.5 mol/L sucrose layers were overlaid and samples were then centrifuged at 35,000 rpm (100,000 x g) for 18 hours at 4°C. After centrifugation, the crude plasma membrane fraction located at the 0.5 to 1.3 mol/L sucrose interface was collected. The amounts of protein were determined using the bicinchoninic acid assay according to the manufacturer's instructions (Pierce, Rockford, IL). Following bicinchoninic acid assay, the crude plasma membrane fractions were snap-frozen in liquid nitrogen and stored at –80°C.
Immunomagnetic purification of epithelial plasma membranes from benign and malignant prostate tissues. Crude plasma membranes were thawed and incubated with mouse antiepithelial plasma membrane antibody cocktail for 60 minutes at 4°C. For 1 mg of crude plasma membranes, 2 µg of CD164 (BD PharMingen, San Jose, CA), 100 µg of epithelial-specific antigen (Neomarkers, Fremont, CA), and 100 µg of epidermal growth factor receptor antibodies (Neomarkers) were added and the incubation was done in 10 mL of isolation buffer [PBS, 0.5 mg/mL PVP-40T (Sigma, St. Louis, MO), 0.5 mg/mL skimmed milk, and Complete protease inhibitor EDTA-free cocktail]. Samples were then transferred into 12.5 mL ultracentrifuge tubes. A cushion of 100 µL of 50% sucrose was placed at the bottom of the tubes and samples were centrifuged at 40,000 rpm for 60 minutes at 4°C to pellet the membranes. Membranes were resuspended in 2 mL of isolation buffer per milligram of crude plasma membrane and incubated for 30 minutes at 4°C with goat anti-mouse MACS immunomagnetic beads (Miltenyi Biotech, Auburn, CA) at a ratio of 1 µL of beads/µg of crude plasma membrane in a total volume of 10 mL isolation buffer. To reduce cytoskeletal protein content associated with plasma membrane, potassium iodine was added to the samples to obtain a final concentration of 600 mmol/L, and then incubated for 30 minutes at 4°C. In a cold room, samples were applied on magnetic LS columns according to the manufacturer's instructions (Miltenyi Biotech). Columns were washed twice with 8 mL of isolation buffer containing 600 mmol/L of potassium iodine and once with 8 mL of 250 mmol/L sucrose, 10 mmol/L Tris-HCl (pH 7.4) buffer. Columns were then removed from the magnet and purified epithelial plasma membranes were eluted with 3.5 mL of 250 mmol/L sucrose, 10 mmol/L Tris-HCl (pH 7.4) buffer into 15 mL tubes. To determine the amount of protein using the micro-bicinchoninic acid assay, 350 µL of the eluted plasma membrane fraction was reserved. The remaining eluate was transferred into a 4 mL Ultraclear centrifuge tube. A cushion of 50 µL of 33% sucrose was placed at the bottom of the tube and samples were spun at 50,000 rpm (337,000 x g) for 30 minutes at 4°C to pellet the plasma membranes. Plasma membrane pellets were snap-frozen in liquid nitrogen and stored at –80°C.
SDS-PAGE and Western blotting. Tissue homogenates from benign tissues and from macrodissected tumor areas were probed by Western blotting for the expression of p63 (1:100 dilution; US Biologicals) and AMACR (1:100 dilution; Chemicon, Temecula, CA), antigens known to be expressed by benign and malignant prostate tissues, respectively (8, 10). The purity of epithelial plasma membrane preparations was assessed by probing for epithelia-specific plasma membrane markers, prostate-specific membrane antigen (1:150 dilution; Northwest Biotherapeutics, Bothell, WA) and epithelial-specific antigen (1:2,000 dilution; Neomarkers), plasma membrane markers for stromal cells [leukocyte marker CD45, 1:50 dilution (Neomarkers); fibroblast marker CD90, 1:100 dilution (Calbiochem, San Diego, CA)], as well as for major intracellular organelles [all antibodies were from BD PharMingen; nuclear marker P62 (1:3,000 dilution); endoplasmic reticulum marker, calnexin (1:2,000 dilution); and mitochondrial marker, Tom20 (1:1,000 dilution)].
H&E and immunohistochemical staining of retrieved tissue banked sections. Selected malignant tissues initially banked for research that were too small for macrodissection were retrieved, placed in a plastic bag, thawed by immersion in warm water for 2 minutes, and subsequently formalin fixed and paraffin embedded before staining with H&E for pathologic evaluation. Benign slices from these patients were also retrieved.
Three pathologists (K. Sircar, L. Gaboury, and C. Lussier) independently and blindly evaluated the histology of thawed tissue and its routinely processed, mirror image on 41 pairs of tissue slices from 17 patients after a period of freezing that varied between 1 and 20 weeks. Tissue staining quality was scored based on architectural and cytologic features using the following system: 0 (poor, diagnosis/Gleason grading severely limited by histology), 1 (fair, diagnosis/Gleason grading somewhat limited by histology), 2 (good, adequate for diagnosis/Gleason grading), and 3 (excellent, optimal tissue staining). The pathologic features of the paired thawed and routinely processed slices were also recorded.
Immunohistochemistry was done on thawed tissue banked sections using a standard avidin-biotin technique with diaminobenzine as the chromogen. An automated immunostainer (Benchmark, Ventana Medical Systems, Tucson, AZ) was used, as were the following antibodies: PSA (polyclonal, 1:1,500; Dako, Carpinteria, CA), prostatic acid phosphatase (polyclonal, 1:6,000; Dako), 34ßE12 (monoclonal, 1:200; Dako), and PIN-4 Cocktail or P504S (AMACR) + p63 (mouse and rabbit, prediluted; Biocare, Walnut Creek, CA).
|
Results |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
Homogenates from benign and macrodissected malignant tissues were then probed by Western blotting for the expression of AMACR, a marker that is significantly up-regulated in cancer (8, 9), and p63, a prostatic basal cell marker present in benign glands but absent in cancer (10), to assess the levels of tissue cross-contamination. In 62% of the macrodissected cases (32 of 52), a low level of tissue cross-contamination was observed as benign tissues contained on average <6% AMACR+ (tumor) cells, whereas malignant tissues were contaminated, on average, by 30% p63+ (benign) cells (Fig. 3
). Therefore, our data indicates a higher cross-contamination rate for benign prostatic cells infiltrating into putatively malignant-labeled banked tissue. However, in 38% of the macrodissected cases (20 of 52), a high degree (>50%) of tissue cross-contamination was assessed. These were not processed further as the ability to identify differentially expressed proteins is significantly reduced when the levels of tissue cross-contamination reached >50%. Hence, a final cohort of 32 cases showing minimal tissue cross-contamination and sufficient cancer volume was selected to undergo a plasma membrane isolation procedure.
|
|
|
|
Pathologic comparison of routinely processed and retrieved tissue-banked sections. Of 41 slices that were retrieved and thawed, their routinely processed mirror image showed the following histology: 29 slices were entirely benign with the dotted focus showing a mean diameter of 14.8 mm, 12 slices had cancer with a mean cancer diameter of 3.5 mm.
In 27 of 29 benign slices, the histology of the thawed tissues matched their corresponding routinely processed mirror image sections, i.e., areas labeled benign on the formalin-fixed tissues were benign in the thawed tissues without evidence of high-grade prostatic intraepithelial neoplasia or cancer. In two benign routinely processed slices, the thawed mirror image showed a small focus of cancer (Gleason score 6). Thus, the pathologic cross-contamination rate by cancer into putatively "benign" banked slices was 6.9%.
In 8 of 12 routinely processed malignant slices, the histology of the thawed tissue matched in that it also showed cancer. In four malignant routinely processed slices, the thawed mirror image did not show cancer, i.e., the small cancer focus was lost in the mirror image slice. Hence, the cross-contamination rate by benign tissue into putatively "malignant" banked slices was 33% in a sample with small diameter cancer (median, 3.25 mm).
|
Discussion |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
To our knowledge, the present series is the first to evaluate epithelial plasma membrane preparations from matched human benign and malignant prostatic tissue. A key variable for the generation of such plasma membrane preparations was the use of highly specific antibodies that bound their cognate antigens at the cell surface with sufficient specificity to purify epithelial plasma membranes of benign and tumor cells away from plasma membranes of stromal cells and other contaminating organelles. This was clearly achieved as the epithelial plasma membrane preparations displayed little contamination from stromal plasma membranes or from other intracellular organelles (Fig. 4).
Moreover, our method enabled purification of the entire plasma membrane and not merely a portion of it. This is particularly relevant to epithelial cell–derived malignancies or carcinomas because epithelial cells are polarized, implying that their plasma membranes are not homogenous but rather separated into two distinct domains: the apical and basolateral domains. Figure 4 shows the presence of both the apical marker, prostate-specific membrane antigen (14, 15) and the basolateral marker, epithelial-specific antigen (16, 17). Hence, the purified membranes should allow for the identification by mass spectrometry of the entire complement of the plasma membrane. Such information will be critical for the optimal identification of novel immunotherapeutic targets.
The mirror image banking protocol was tested with respect to pathologic diagnosis and found to be highly satisfactory. Specifically, tissue-banked sections that were retrieved, thawed, and processed after various periods of time showed histologic features that were always diagnostically adequate, albeit qualitatively slightly inferior to routinely processed tissues (Table 1). This is important, as pathologists may need to recover selected tissue-banked slices to fully assess equivocal areas of extracapsular invasion or surgical margin positivity seen on the routinely processed mirror image sections. Moreover, cancer may only be present in the slices sent for tissue banking and thus may be required for cancer diagnosis and Gleason scoring. We only encountered one case in which cancer was not seen on routinely processed sections and all the banked slices were returned with minimal cancer identified on one such retrieved, banked section. However, our study sample was relatively small and this scenario of extremely low volume prostate cancer will inevitably, if rarely, occur. Thawed tissues also showed the retention of immunoreactivity with antibodies routinely used in diagnosing prostate cancer.
With respect to prostate cancer, certain limitations inherent to the disease and the procedure need to be addressed. First, even the minimal amount of tumor (0.3 g) required for plasma membrane isolation and downstream proteomic applications are often unavailable due to low volume cancer. Thirty-eight percent of our patients were excluded on this basis. This is not surprising considering our PSA-screened patient cohort in which only 34% had clinically palpable disease. The second issue in a multifocal, highly infiltrative neoplasm such as prostate cancer is cross-contamination. Even after careful screening, our data suggests rates of cross-contamination of 30% for benign into malignant-labeled banked tissue and <6% for tumor into benign-labeled banked tissue. When evaluating a 2-mm-thick tissue slice, mirror image histology can be used as a guide in selecting malignant areas, but infiltration between benign glands is intrinsic to prostate cancer and some degree of contamination cannot be entirely eliminated.
The pathologic discrepancies that we encountered between routinely processed and banked/thawed mirror image slices could be explained by the trimming effect: prostatic quadrant slices are relatively large—approximately the size of a standard cassette—and because three-dimensional tissues are never perfectly flat, 100 to 150 µm needs to be trimmed off the pathology block before getting a full face section that the pathologist can read. Thus, when evaluating the pathology slide, we do not see the exact mirror image of the tissue banked slice, but rather, a section that is 100 to 150 µm deeper. This creates space for cross-contamination, especially in a highly infiltrative, multifocal neoplasm such as prostate cancer. It was magnified by the trimming of both blocks when histologically comparing thawed and routinely processed mirror image pairs.
There is currently no standard, universally accepted method for processing radical prostatectomy specimens even for clinical diagnosis. Both complete and partial submission of the prostate are widely practiced and both are recognized as valid by the College of American Pathologists (18). Methods of procuring prostate tissue for research purposes are similarly variable. Common techniques include punch biopsies of fresh prostatic tissue with pathologic examination of the tissue surrounding this hole used to infer the histology of the procured area (11, 19). Although histologic correlation of the procured area with its surrounding tissue has been validated in a recent study (20), the basic drawback of this technique is the inaccuracy of knowing which area to punch: cancer was missed in roughly 50% of prostatectomy specimens. Where data is available correlating the size of prostate cancer with accuracy in identifying it grossly, the results are worse. The large series by Renshaw showed that 61% of cancers were missed when they were <10 mm in diameter (7). More exotic methods of identifying cancer include the use of a dissecting microscope (21), which is more accurate than gross examination but which requires the presence of a trained pathologist at the time of prosection. Another technique involves freezing and banking alternate slices of prostate. Frozen sections can be cut later and the procured tissue can be histologically mapped.
Our mirror image banking protocol is more accurate than punch biopsy or any other method that relies on gross examination to guess where the tumor might be. This is particularly relevant in today's era of nonpalpable, low volume prostate cancer. The utility and practicality of a dissecting microscope is limited by the expert pathologic knowledge required for its use. Essentially, a pathologist trained to use the dissecting microscope would have to prosect and evaluate each prostate shortly after its removal. If the whole prostate is evaluated rather than just one representative slice, this technique also becomes time-consuming. Mirror image banking, by contrast, can easily be done by pathology assistants or junior residents. It is comparable to frozen section banking and, in some respects, advantageous. Placing tissue in DMEM/fetal bovine serum/DMSO medium improves cellular viability and better simulates a physiologic milieu. The histologic and immunohistochemical quality of the thawed tissue is very good and always adequate for clinical diagnosis (Table 1). From a technical standpoint, it is also cheaper and faster because a cryostat and technician time in cutting frozen sections is not required. No special expertise associated with cutting frozen sections is needed.
In summary, we have used novel immunomagnetic purification techniques to successfully isolate the entire epithelial plasma membrane from matched benign and malignant human prostatic tissues. To accomplish organelle isolation from human prostates, we have described and provided validation data for the technique of mirror image tissue banking of radical prostatectomy specimens in a PSA-screened patient population. In the future, our techniques will prove useful in identifying proteins that are overexpressed on plasma membranes of malignant prostatic cells that can putatively serve as therapeutic targets for prostate cancer treatment.
|
Footnotes |
|---|
Received 2/10/06; revised 5/ 1/06; accepted 5/10/06.
|
References |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
-Methylacyl-CoA racemase: a novel tumor marker over-expressed in several human cancers and their precursor lesions. Am J Surg Pathol 2002;26:926–31.[CrossRef][Medline]-Methylacyl coenzyme A racemase as a tissue biomarker for prostate cancer. JAMA 2002;287:1662–70.
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Cancer Research | Clinical Cancer Research |
| Cancer Epidemiology Biomarkers & Prevention | Molecular Cancer Therapeutics |
| Molecular Cancer Research | Cancer Prevention Research |
| Cancer Prevention Journals Portal | Cancer Reviews Online |
| Annual Meeting Education Book | Meeting Abstracts Online |
