Identification of the peanut-agglutinin binding pancreatic cancer serum marker in pancreatic tissue extracts.

ImagesFigure 1

Pancreatic tissue glycoprotein extraction was performed according to Rao and Shinozuka (1984) with minor modification. Approximately I g wet weight pancreatic tissue was used per specimen. Samples were cut into small pieces and then ultrasonicated (Ultrasonicator KS 100, Kerry Ultrasonics, UK) for 1 min in 10 ml Tris HCI (20 mM, pH 7.4)/ EDTA (1 mM) buffer containing 200Lg ml-' soybean trypsin inhibitor. This was followed by homogenisation in a Polytron homogeniser (PCU, Kriens-Luzern, Switzerland) and then centrifugation at 13,000g (Sorvall RC5, DuPont instruments, USA) for 20 min. Supernatants were discarded because preliminary analysis of concentrated supernatants from two normal pancreatic tissues did not reveal any PNA binding glycoproteins identifiable on lectin blotting from the gel. The pellets were washed x 5 and then each sample rehomogenised separately in I ml of the same buffer for 30 s. One ml of the homogenised tissue was mixed with 9 ml of chloroform/methanol (2: 1) mixture and stirred vigorously for 30 min. The aqueous phase was separated from the lipid phase and the solid residue by centrifugation at 300 g (Centaur 2, Fison instrumentation services, UK) for 20 min. The aqueous phase was then concentrated by gentle evaporation under nitrogen to approximately 1/4 of its original volume and glycoprotein precipitation was carried out using nine equivalent volumes of the aqueous phase of absolute ethanol. The precipitate was obtained by centrifugation at 200 g for 10min.
Glycoprotein precipitates were reconstructed in 400 l of de-ionised, distilled water. An aliquot was used for Lowry protein estimation (Lowry et al., 1951). SDS-PAGE (using approximately 100 iLg protein per sample), and then high intensity transfer of proteins and glycoproteins on to nitrocellulose papers and finally identification of PNA binding glycoproteins on the blots were performed as described before . The high molecular weight PNA binding glycoprotein identified in the tissue extracts was then further characterised using other lectins, UEA I (25 jgml-'), LFA (12.5 ig ml-') and GS 2 (25 JLgml-').
The mean yield of water soluble protein obtained from one extraction step ranged between 2-4.5 mg g-' pancreatic tissue. A high molecular weight (approximately 3.5 million Da) PNA binding glycoprotein (lane 3, Figure 1) having identical electrophoretic mobility to the serum marker, both before and after purification, was identified in tissue extracts from 3/3 pancreatic cancers, 1/4 chronic pancreatitis and 2/5 normals. The sole ampullary carcinoma extract studied did not contain the high molecular weight glycoprotein at 3.5 x 106 Da but it showed a strong PNA binding region around 1 x 106 Da, indicating the possibility of another tumour related PNA + glycoprotein in this epithelial carcinoma. A lower molecular weight (50,000 Da approximately) peanut agglutinin binding glycoprotein was also present. This was co-purified with the 3.5 million Da glycoprotein and was still present after the 3.5 million Da glycoprotein had been cut out of the gel, eluted and rerun. Characterisation of the water soluble pancreatic tissue PNA binding 3.5 million Da  glycoprotein by the use of other lectins showed that it bound LFA but not UEA I and GS 2 lectins (Table I) indicating the expression of the epitopes gal 1-3 gal NAc (blood group T antigen), and sialic acid but not L-fucose (blood group H antigen) or GIc NAc (blood group Tk antigen). Other PNA binding glycoproteins present on the blots are probably normal pancreatic epithelial structure components as we have previously shown in a lectin histochemical study that PNA binding glycoproteins other than the secreted mucus can be identified . This study has demonstrated that a peanut lectin (PNA) binding glycoprotein previously found in pancreatic cancer serum is also present in the pancreatic tissue itself not only in pancreatic cancer but also in benign pancreatic disease and in the normal pancreas. Both serum and tissue PNA binding glycoproteins had identical electrophoretic mobility. The lectin binding characteristics of the tissue glycoprotein extracted from benign and malignant pancreas indicate that it possesses gal I1-3 gal NAc (PNA binding) and sialic acid (LFA binding) side chains. These epitopes have been demonstrated on the serum glycoprotein (Table I) which variably bears additional epitopes namely L-fuCose (H antigen, UEA I binding) and glc NAc (Tk antigen, GS2 binding) (Ching & Rhodes, 1987b).
The demonstration of the high molecular weight PNA binding glycoprotein in pancreatic tissue makes it very likely (Ching~~"5 &'Rhodes', 1988 in paceai cace sera 9is indeed:: coming~~~....... fromC the pacra. Shddn of muci ino seu appears to besamoereuarfatr of parxds-N etnbotncreaftic canero (+) (+), variable binding of the high mol. wt serum glycoprotein to the lectins (7/12 to UEA I and 4/12 to both LFA and GS 2) (Ching & Rhodes, 1987b). compared with gastrointestinal tumours such as colonic and gastric tumours as shown by the higher rate of positive serum tests in pancreatic cancer using both enzyme-linked PNA assay and CA19-9 radioimmunoassay, even though the CA19-9 antibody was raised against a colorectal cancer cell line. In a previous study, we have shown that the PNA binding pancreatic cancer-related serum mucus glycoprotein sometimes but not always expresses the CA19-9 epitope . The binding sites for PNA (gal 1-3 gal NAc) and CA19-9 (sialylated N-fucopentaose II oligosaccharide) cannot occur on the same oligosaccharide side chains so we envisage the tumour-related mucin as a complex glycoprotein that may variably express the sialylated Lewis antigen (CA19-9 epitope) on some side chains, the PNA epitope (T antigen) on others and UEA I (H antigen) and GS2 (Tk antigen) binding sites on yet other side chains. Simultaneous demonstration of an additional carbohydrate epitope (CA-242) on the tumour marker glycoprotein CA5O has also been reported recently (Nilsson et al., 1988).
The sialylated Lewisa antigen has been found in normal colon (Afdhal et al., 1987), bile (Albert et al., 1987) and pancreatic juice (Kalthoff et al., 1986) so seems to be a normal tissue and mucin glycoprotein that is abnormally expressed in the serum in cancer rather than an oncofetal antigen. The status of the PNA binding site (T antigen) is more uncertain. It behaves more as an oncofetal antigen in colon (Boland et al., 1982;Cooper, 1984;Rhodes et al., 1986), breast (Howard et al., 1981), stomach (Kuhlmann et al., 1983), ovary (Soderstrom, 1988) and lymphoid (Ree & Hsu, 1983) tissue. It can be predicted from the known structure of mucin that the T antigen can only be present as the base pair of the oligosaccharide side chain (Hounsell & Feizi, 1982) which is usually concealed by further glycosylation or sialylation. It seems likely that its expression at least reflects a relatively immature mucin side chain. In a previous study using lectin histochemistry, PNA binding has however been found variably in normal pancreatic cytoplasm  and in normal large bile ducts  and the study presented here confirms that it can be variably expressed in normal pancreas.
The presence of mucin in serum is perhaps surprising, but the CA19-9 epitope bearing mucin has also been found in pancreatic cancer (Haglund et al., 1986) and in patients with cystic fibrosis . In pancreatic cancer, this might reflect either early invasion of this tumour into blood vessels or early ductal obstruction with reflux. It is clear from our study that this mucin contains at least four different oligosaccharide side chain structures and probably many more so development of a panel of monoclonal antibodies against different epitopes on this mucin may lead to the development of a more sensitive and specific test for pancreatic cancer.
C.K.C. was an Amelie Waring research fellow of the British Digestive Foundation.