Proteome analysis of human gastric cardia adenocarcinoma by laser capture microdissection

IPG strips (pH 3–10 nonlinear) and IPG buffer solutions (pH 3–10 nonlinear) were purchased from Amersham Pharmacia Biotech (APB, Sweden). DTT, urea, thiourea, Tris base, TX-100, CHAPS, glycine, acrylamide, methylenebisacrylamide, SDS, TEMED, ammonium persulfate, silver nitrate, Trypsin (sequencing grade), ACN, and TFA were purchased from Sigma (St. Louis, MO, USA). Finally, the complete protease inhibitor cocktail was from Roche (Lewes, UK). Milli-Q grade water was used for all the solutions.

Nine pairs of human gastric cardia adenocarcinoma and their adjacent nontumourous cardia tissues were obtained within 30 minutes after a surgical resection at the second affiliated hospital of Xi'an Jiaotong University in 2005 (Table 1). In order to avoid obvious areas of ulcer or necrosis, the samples were procured from the edges of the tumours. They were immediately frozen in liquid nitrogen and stored at -80°C until use. Informed consents were obtained from all patients. Meanwhile, two experienced pathologists evaluated the tumour grading by a microscopic examination of the samples.

Sections (8 um thick) were cut onto slides (precleaned using a detergent, washed with deionised water, and dipped in ethanol) using a Leica CM 1900 cryostat (chamber temperature -28°C). The sections were either stored at -80°C or kept in the cryostat chamber prior to LCM. Haematoxylin staining was carried out only for monitoring of tissue sections and was not used in conjunction with LCM. Sections were fixed (70% ethanol for 1 min), haematoxylin stained and dehydrated (70% ethanol for 45 s, 95% ethanol for 45s, 2 × 100% ethanol for 30s, and followed by xylene 2 × for 5 min). Meanwhile, unstained sections were used for LCM. They were fixed (70% ethanol for 30s), washed in deionised water for 10s, and dehydrated (70% ethanol for 15s, 95% ethanol for 15s, 2 × 100% ethanol for 15s, and followed by xylene 2 × for 1 min). Complete protease inhibitor cocktail tablets were

After fixing and dehydration, unstained sections were air dried and microdissected using the Arcturus PixCellα system (Arcturus, Mountain View, CA, USA). LCM microscope imaged a haematoxylin stained section adjacent to the one of interest. This image was displayed on the LCM computer screen using LCM image analysis software (Arcturus, USA) and was used as a map to delimit the region for dissection on an adjacent unstained section. The sections were then captured using a 15 um diameter laser beam typically at 50–80 mV power with pulse duration of 3–10 ms in machine gun mode and with a laser firing frequency of 1 shot per 500 ms. On the average, between 10,000–12,000 shots were taken per cap, and approximately 25,000–30,000 cells were obtained per cap. Each cap was captured within one hour. Based on a careful review of the histologic sections, each dissection was estimated to contain ≥ 95% of the desired cells.

Microdissection caps were inserted into 0.5 mL microcentrifuge tubes containing 50 uL of a lysis buffer (7 M urea, 2 M thiourea, 4% w/v CHAPS, 65 mM DTT, 0.5% v/vTX-100, 40 mMTris-Base, and complete protease inhibitor cocktail). Then the cells were solubilised by inversion of tubes followed by vortex-mixing for 1 min and brief pulse-centrifugation at 12, 000 × g. Afterwards, tissues from multiple caps (about 800,000~1,000,000 cells) were extracted into the same lysis buffer until sufficient material had been collected. The samples were centrifuged at 40,000 × g for 1 hour at 4°C. Where necessary, supernatants were stored at -80°C until further use. Protein concentration was determined by the Bradford method.

The first-dimensional isoelectric focusing (IEF) was carried out on Pharmacia Immobiline IPG DryStrip system (Uppsala, Sweden). For the first dimension of electrophoresis, the samples containing 60 ug protein for analysis gels were diluted to 250 uL with a rehydration solution (7 M urea, 2% w/v CHAPS, 50 mM DTT, 0.5% v/v IPG buffer (pH 3–10 nonlinear), and trace bromophenol blue) before loading onto 13 cm IPG strips (pH3–10 nonlinear). IEF was then performed using IPGphor electrophoresis unit according to the manufacturer's instructions. Thereafter, the strips were equilibrated with a solution (6 M urea, 30% v/v glycerol, 2% w/v SDS, and 50 mM Tris-HCl, pH 8.8), reduced with 1% w/v DTT for 15 min, and alkylated with 2.5% w/v iodoacetamide for 15 min. Strips were then rinsed in electrophoresis buffer (25 mM Tris base, 192 mM glycine, and 0.1% w/v SDS), applied to 11% acrylamide gels, and sealed with melted agarose (0.5% w/v agarose in electrophoresis buffer containing a trace of bromophenol blue). SDS-PAGE was carried out using Hoefer SE 600 vertical chambers and a Tris-glycine buffer (25 mM Tris and 192 mM glycine) containing 0.1% w/v SDS, with initial separation at a constant 10 mA/gel for 30 min followed by 20 mA/gel. The second-dimensional SDS-PAGE was developed until the bromophenol blue dye marker had reached the bottom of the gel. The total run time was typically 4 to 4.5 hours. Gels were fixed in 10% v/v acetic acid, 40% v/v ethanol before sensitisation for 30 min in a buffer containing 30% v/v ethanol, 0.2% w/v sodium thiosulphate, and 0.83 M sodium acetate. This was followed by three 15 min washes in deionised water. The proteins were then stained with 0.1% w/v silver nitrate for 20 min, washed twice in deionised water for 1 min, and developed in 2.5% w/v sodium carbonate containing 0.04% v/v formaldehyde (37% solution). The development was stopped with 1% v/v acetic acid, and the gels were washed three times in water.

Consistently and significantly different spots selected for analysis by MALDI-TOF MS. Protein spots of interest were excised and minced into small pieces followed by destaining and washing with deionised water for several times until the yellow color disappeared. Gel pieces were then rinsed with 25 mM ammonium bicarbonate, dehydrated with ACN, and dried in a vacuum centrifuge. Afterwards, proteins in-gel were digested with 0.02 ug/uL trypsin overnight at 37°C. Tryptic peptides were then re-dissolved in a solution containing 50% v/v ACN and 0.2%v/v TFA. A 2 uL aliquot was spotted onto a sample plate with 4 uL of matrix solution CHCA (a-cyano-4-hydroxcinnamic acid, 10 mg/mL in 50% v/v ACN, 0.2%v/v TFA) and allowed to air dry. The dried spots were then analysed in a Voyager DE MALDI-TOF MS (Framingham, MA, USA). This was followed by running the spectrometer in a positive ion mode and in reflector mode with the following settings: accelerating voltage, 20 KV; gride voltage, 94%; guide wire voltage, 0.01%; and a delay of 200 ns. Spectra were acquired manually with the laser intensity set at 3200 with 80 shots per spectrum. Then the mass range was set between m/z 500 and m/z 5000. Internal calibration was applied using angiotensin II and insulin B chain peaks at 912.08 Da and 3495.95 Da, respectively.

Proteins were identified by peptide mass fingerprinting using the search program Aldente [14] with the following search parameters applied: SWISS-PROT and TrEMBL were used as the protein sequence databases; a mass tolerance of 100 ppm and one incomplete cleavage were allowed; carboxyamidomethylation, oxidised methionine, and phosphorylation were considered as possible modifications; the minimum number of matched-peptides was 4; and the peptide ion was [M + H]⁺.

To validate the expression patterns of three proteins in GCA tissues, immunohistochemistry was performed using formalin-fixed and paraffin-embedded tissue specimens that were matched with 2-DE samples. Dewaxed 5 um thick sections were treated with a 0.3% hydrogen peroxidase for 3 min and with a blocking antibody for 30 min. After heat-mediated antigen retrieval, sections were incubated with primary antibody at 4°C overnight as follows: HSP27 (Sigma, St. Louis, MO, USA), 1:500; HSP60 (Santa Cruz, America), 1:300; and Prx-2 (Santa Cruz, America), 1:150. Sections were then incubated with a peroxidase-labeled antibody (1:500), developed with diaminobenzine, and counterstained with hematoxylin.

For western blot analysis, the protein samples (30 ug) used in 2-DE were run on 12% SDS-PAGE, transferred onto PVDF membranes, and then blocked with PBS/5% skim milk/0.01% Tween for 2–4 hours at room temperature. Primary antibody was diluted in blocking buffer and was added as follows: HSP27, 1:200; HSP60, 1:300; and Prx-2, 1:200. Afterwards, it was incubated with a horseradish peroxidase (HRP)-conjugated secondary antibody and a HRP-conjugated anti-GAPDH/β-actin antibody to confirm equal protein loading in each lane for 1–2 hours at 37°C or room temperature. The samples were washed and detected with enhanced chemiluminescence for 30–60 s (Minipore).

To evaluate the effects of navigated LCM on the profiles of tissue protein, we performed 2-DE protein profiles of whole undissected cryostat tissues and LCM-samples of GCA. Figure 1 shows an example of the navigated LCM process. Majority of the spots detected in the dissected non-tumour and tumour tissues could be observed in the whole undissected cryostat tissues, except for several spots. However, there are still some spots found in the whole undissected cryostat tissue which could not be observed in both LCM samples. Thus, this technology could not only enrich non-tumour and tumour epithelial cells but could also diminish the contamination in tissues such as hemoglobin [10, 11]. Moreover, the navigated LCM process could eliminate the effects of staining on protein separation by 2-DE [15]. Thus, our data support the need to perform navigated LCM in the proteomic studies of GCA tissues.

We performed a navigated LCM and 2-DE for matched pairs of tumour tissues and the surrounding non-tumour tissues from GCA patients. Approximately 800–1,000 protein spots were detected by silver staining. To measure reproducibility, each sample underwent the experiment three times. There were 905 ± 74 and 867 ± 51 protein spots in the map of GCA and non-tumour gastric cardiac tissue, respectively. The matched spots were 799 ± 29 and 727 ± 34 separately, and the respective average matching rate was 88.2% and 83.8%. In addition, the average position deviation of matched spots was 1.031 ± 0.205 mm and 1.44 ± 0.11 mm in the IEF and SDS-PAGE direction, respectively. Although the image analysis showed that these 2-DE maps were reproducible, there were slight differences in the intensity and number of spots. Nonetheless, an analysis of the gels revealed 27 protein spots whose intensities varied substantially and consistently between non-tumour and tumour tissues (Fig. 2). The ratios of normalised spot intensities of cancer to paraneoplasis tissues for each of the proteins of interest were calculated. Spots showing a two-fold difference and with statistical significance (p < 0.05) were selected. Three of the 2-DE results were confirmed by immunohistochemical and western blot analyses.

Both whole undissected cryostat specimens and LCM-specimens were used for protein identification. Twenty-seven different protein spots between GCA tissues and surrounding non-tumour tissues were excised. However, only 23 proteins were identified by MALDI-TOF-MS (Table 2). This phenomenon is likely due to post-translational modifications like peroxiredoxin 1 (Prx-1), which was present in two adjacent spots (Fig. 2, spot 11). These proteins were classified into any of the following: cell proliferation and differentiation (ANXA2, ANXA4, hnRNPA2/B1), apoptosis (Prx-1, Prx-2, GSTP, VDAC, ETFB), metabolism (ADH1C, AKR1C3, CA2, GATM), protease related (CTSD, PACSIN1), cystoskeleton (Actin 1, Keratin 8), chaperones (HSP27, 60, 70, PDIA3), and RNA binding and transcription (hnRNPH3, PCBP1, ENO1). Figure 3 shows the PMF maps of HSP60.

To further investigate whether HSP27, 60, and Prx-2 are expressed in tissues and to determine which cells express these proteins, immunohistochemical analysis was performed using formalin-fixed and paraffin-embedded tissue specimens that were matched with 2-DE samples. The expression of HSP27 and 60 could be seen in all the cytoplasms of carcinoma cells, while it could be seen in some of the cytoplasms of columnar epithelial (Fig. 4A–D) in non-tumour tissues. In contrast, Prx-2 was almost not found to be expressed in the carcinoma cells but rather in some of the cytoplasms of columnar epithelial cells (Fig. 4E, F). To examine these protein expression levels, western blot analysis was also performed using the same tissues (Fig. 5). As expected, HSP27 and 60 were found to be consistently highly expressed in tumour tissue, whereas Prx-2 was suppressed.