Protein Secretions in Hamsters with Pancreatic Carcinoma BARBARA J. ALLAN, MS, THOMAS T. WHITE, MD, HOWARD A. REBER, MD, CARL ROBERTS, MB, FRCS, and JOHN J. SCHILLING

Pancreatic secretions from the hamster model for pancreatic ductal adenocarcinoma were analyzed to determine whether alterations had occurred in the protein composition. CCK- and secretin-stimulated secretions were collected from 13 animals with cancer and 16 normal controls. Subsequent separation of the proteins by isoelectric focusing showed the following: (1) Significant changes occurred in the protein composition from animals with carcinoma. An unusually dark band was present at pH 7.5 just below amylase (pH 7.65); two unidentified bands, present in the normals at pH 6.9 and 7.0, were missing; and marked decreases occurred in the cathodic proteins with isoelectric points above pH 9. (2) CCK provided the optimal stimulus for differentiating specimens from animals with carcinoma from the normal controls. (3) The protein concentrations of CCK-stimulated secretions of animals with carcinoma were significantly lower than the controls. We have concluded that the protein alterations which have occurred in the hamster model for pancreatic ductal adenocarcinoma warrant further investigation.

Carcinoma of the pancreas has been aptly described as the "dismal disease" (1). It is relatively asymptomatic until its later stages and often difficult to diagnose. Early diagnosis could greatly improve the survival rate. For this reason considerable effort has been directed towards the identification of tumor-markers which could serve as the basis for a

Manuscript received May 27, 1981; revised manuscript received October 12, 1981; accepted October 26, 1981. From the Department of Surgery, RF 25, University of Washington, Seattle, Washington 98195; and Department of Surgery, University of Missouri School of Medicine, Columbia, Missouri 65102. A preliminary report of portions of this study has been given. Fed Proc 40:1868, 1981. The authors gratefully acknowledge support for this study by PHS grant CA AM 14380 awarded by the National Cancer Institute, DHHS, and from the National Pancreatic Cancer Project, CA 25058, and from the Medical Research Service of the Harry S. Truman Memorial Veterans Hospital. Address for reprint requests: B.J. Allan, Department of Surgery RF 25, University of Washington, Seattle, Washington 98195.

diagnostic test and ultimately, perhaps, as a screening test for this disease. Secretory abnormalities have been known to occur in carcinoma of the pancreas in man for some time and have been thought to be due to obstruction caused by the malignancy (2). More recently in the hamster model for pancreatic ductal adenocarcinoma (3), secretory abnormalities have been found to antedate the appearance of neoplasms where no obstruction was present (4). These abnormalities included impairment in the secretion of fluid, electrolytes, and proteins in the pancreatic juice. On the basis of these results, we wanted to know whether the protein composition of the secretions was altered in the animals with carcinoma. In this study we have investigated this question using the same hamster model. Pancreatic secretions from animals with carcinoma and from normal controls were collected following secretin and CCK stimulation and were analyzed by isoelectric focusing (IEF). The resultant protein patterns were compared with computer assistance.

Digestive Diseases and Sciences, Vol. 27, No. 5 (May 1982)

0163-2116/82/0500-0437503.00/1 e 1982DigestiveDiseaseSystems,Inc.

437

A L L A N ET A L

MATERIALS AND M E T H O D S Thirteen maie Golden Syrian hamsters (weight 40-100 g) were given weekly subcutaneous injections of diisopropyl nitrosamine (DIPN) (250 mg/kg). Sixteen control animals were administered weekly injections of 0.15 M NaC1. Secretory Studies. Before a study the animals were fasted for 24 hr and anesthetized with ethylcarbamate (100 mg/100 g body weight). They were immobilized supine on a heated platform, and Ringer's lactate solution (10 ml/kg/hr) was infused through a femoral vein for the duration of the experiment. The abdomen was opened, and the common bile-pancreatic duct was cannulated extraduodenally with a fine polyethylene tube (PE-10, Clay-Adams, Parsippany, New Jersey). To exclude bile from the collections, the duct was ligated near the liver hilum proximal to the pancreas. The prepared animal was allowed to rest for 1 hr, and then the experiment was begun. Pancreatic juice was collected on ice under mineral oil in tared cups. After a 1-hr collection of basal secretion, 2 units/kg of secretin or cholecystokinin (CCK) were injected intravenously as a bolus. The hormones were purchased from the Gastro-intestinal Hormone Research Laboratory, Karolinska Institutet, Stockholm, Sweden. After the hormone stimulus, four 15-rain collections were made. No juice was collected for the next 30 min; then the second of the hormones was given and collections were resumed for another hour. Previously we had determined that these doses of secretin and CCK produced maximal secretory responses in hamsters. The order in which the hormones were given was alternated in different animals. After the secretory collections were completed, the animal was killed, the organs were removed, and the pancreas, liver, and lungs were prepared for histologic examination by light microscopy (hematoxylin and eosin stain). Pancreatic juice was analyzed for concentration of HCO3 and C1 (secretin-stimulated juice), or total protein. The results of those studies formed the basis of a previous publication (4). F o r the present experiments, the 1-hr collections were pooled, aliquots were removed for protein determination (5), and the specimens were stored frozen at - 3 5 ~ C.

Analytical Methods. Protein concentrations were determined by the Lowry protein method (5). Proteolytic activity was monitored with chymotrypsin as previously described (6). Pancreatic ductal secretions were then analyzed by isoelectric focusing. The isoelectric focusing techniques used minimize proteolysis of the pancreatic enzymes and are the subject of a separate publication (10). Samples were used immediately after thawing without concentration or prior processing. This preserves the native pancreatic secretory trypsin inhibitor present in the secretions. Amylase, RNase, and procarboxypeptidase A were identified by histochemical methods in freshly prepared polyacrylamide gels immediately following the separation of the hamster pancreatic secretions by isoelectric focusing. One half of a given protein pattern was used for enzyme detection and the other half was fixed and stained for identification purposes. To detect the amylase bands, a filter paper strip containing the dyed substrate (7) was placed over the unstained gel and incubated in a moist chamber at room temperature. Bands of amylase activity appeared as white areas on the pink filter paper, Bands were marked by snipping the paper and by punching a corresponding hole in the gel. When the test was completed the gel was fixed, stained, and compared with the freshly stained gel. RNase bands were detected similarly in unstained gels using microscope slides coated with an RNA film (8). Procarboxypeptidase A and carboxypeptidase A were located by methods used for the detection of these enzymes on cellulose acetate strips (9). The isoelectric focusing techniques have been described in detail (10). A brief description of these techniques is given here. Analytical isoelectric focusing on polyacrylamide gels was performed at 4 ~ in an L K B Multiphor 2117 using pH 3.5-9.5 gel slabs provided by LKB (LKB 1804-101). Samples which had been stored at - 3 5 ~ C were kept frozen while the gel was prefocused and then thawed immediately before application to the gel surface in Lucite frames (inner well dimensions of 0.5 x 1.0 cm, or 0.5 x 0.5 cm for smaller samples). Protein loads between 80 and 220 ~g/cm were found to give reproducible patterns using this method, and linearity between the load applied and the absorbancy was found between 80 and 180 ~g/cm (10). Patterns obtained in this

TABLE I. COMPARISON OF PROTEIN CONCENTRATIONS OF CCK- AND SECRET1N-STIMULATED PANCREATIC SECRETIONS FROM HAMSTERS WITH AND WITHOUT CARCINOMA OF THE PANCREAS

CCK-stimulated Protein concentration (mg/ml)

I II III IV V

Above 20.0 7.0-20.0 3.6-6.9 0.1-3.5 None obtained Total number of animals Average protein concentration (mg/ml)

Secretin-stimulated

Normals

Cancer

Normals

Cancel"

14 2 0 0 0 16 39.7 -+ 16.5

1" 3* 1" 6 2 13 81. -+ 9.8t

2 4 6 4 0 16 10.2 -+ 12.9

0 5 7 1 0 13 6.4 -+ 3.5

*Specimens used for Figure lB, tFor 11 animals.

438

Digestive Diseases and Sciences, Vol. 27, No. 5 (May 1982)

PANCREATIC SECRETIONS IN CARCINOMA

NORM-CC~ [15]

4

9

IE)

c~

(12

~8. I]

2.1~

4.8 G.O 8,111 CENTIMETERS FROM THE ANOOE

II~.I~

Fig 1. IEF patterns of CCK-stimulated hamster pancreatic juice. A densitometric plot of the absorbance at 570 nm vs the protein band position (cm from the anode). The shaded area represents the 95% confidence interval. (A) Patterns from 15 normal animals. (B) Patterns from 13 specimens from animals with carcinoma. The averaging, statistics, and plots were prepared with a Computer following isoelectric focusing (see Materials and Methods).

protein load range were adjusted to the standard load (180 i~g/cm) by the computer (see below). The standard protein load was used unless otherwise noted. Protein markers were included in tWO tracks in each run. Immediately following the focusing run. the pH was measured at 4~ C with a surface electrode, The gel was then fixed in a hot sulfosalicylic acid-TCA mixture at 65 ~ C and stained with Coomassie brilliant blue R-250. After destaining, colorpositive transparencies of the gels were prepared at 7/10 scale for scanning in the densitometer. Computer Averaging of Data. The color transparencies were scanned in a Beckman CDS-100 computing densitometer interfaced with a North Star Horizon II-64K DQ microcomputer. Each of the pattern and absorbance standards were scanned, and the data were recorded on 51h-in. Verbatim floppy disks. The data were entered and processed interactively on the computer from a LearSiegler ADM 3A terminal, using the TIBIAS program package, a group of programs in the BASIC language which will be the subject of a publication elsewhere. The tasks accomplished by these programs are: entry and processing of the scan data, removal of the blank gel backgrounds to obtain more accurate areas under the peaks, matching Of the patterns, adjustment of the height, and averaging of the patterns to give the average pattern. The following statistical calculations are also performed: + / - the standard error, +/ the standard deviation, and + / - the 95% confidence limits. Plots were prepared with a Houston Instruments DMP-4 plotter. All these proceDigestive Diseases and Sciences, Vol. 27, No. 5 (May 1982)

dures were carried out with a microcomputer located in the laboratory, unlike our earlier methods (10, 11). RESULTS Significant decreases occurred in both the protein outputs and concentrations in animals with pancreatic c a r c i n o m a when C C K was the s e c r e t o r y stimulus. Seven of 13 specimens had protein c o n c e n t r a tions less than 8 mg/ml, well below the normal range, and two were not obtainable as a result of the l o w outputs. The remaining four specimens (38%) had concentrations overlapping the lower normal range (Table 1). W h e n secretin was used to stimulate secretion, no significant differences in protein concentration were found (Table 1). Specimens from the control animals given secretin showed a wider range of protein concentrations than those from the animals with cancer, p r e s u m a b l y because of the greater v o l u m e r e s p o n s e to this h o r m o n e in normal animals (4). Analysis of Protein Patterns in Controls. The protein composition of h a m s t e r pancreatic secretions was analyzed by isoelectric focusing (IEF) as described in Materials a n d Methods. C C K - and

439

ALLAN ET AL B o

CANC-ECK [~3]

9

LJ Z O2

o0

4.0 6.0 8.0 CENTIMETERS FROM THE ANODE Fig l . Continued,

secretin-stimulated secretions from normal animals were compared first. Computer-aided matching and averaging of 15 CCK patterns resulted in plots of + and - the 95% confidence limits of the average pattern versus the distance from the acidic wick (Figure 1A). The same comparison was used for the analysis of 14 normal specimens obtained with secretin (Figure 2A). The IEF protein patterns found with CCK were much more uniform than those obtained with secretin, and they gave much more consistent densitometric patterns. This is evident from the narrower width between confidence limits seen with the CCK samples as opposed to the secretin specimens (cf Figures 1A and 2A). Analysis of Protein Patterns in Cancer. Significant differences in the IEF patterns were seen when CCK-stimulated specimens from animals with carcinoma were analyzed: Five carcinoma specimens which had the highest protein concentrations of this group (see the values with an asterisk in Table 1) were compared with the normal controls. Four of these specimens overlapped the normal protein concentration range. Patterns from these specimens were distinctly different from any of the normal CCK-stimulated specimens (Figure 1B). These patterns were characterized by an unusually dark band located at pH 7.5 just below the major amylase band, A2, at pH 7.65 (Figure 3). Other changes

440

10.0

included the loss of a large number of bands, notably two major bands at pH 6.9 and 7.0, not yet identified in the normal secretions. A considerable decrease in the cathodic proteins, those with pH values above 9 also occurred: The remaining seven CCK specimens were analyzed by IEF, but were not included in the averaged data. The protein concentrations of these specimens were well below (less than 7 mg/ml) any of the normal controls and, of necessity, were applied to the gels below 80 txg/cm. All of these specimens showed extremely faint patterns which were similar to the more concentrated specimens. The characteristic IEF pattern seen with CCKstimulated secretions from animals with carcinoma was also clearly present in one of the secretin specimens from the same animals (Figure 3, column 7, and Figure 2B). All of the remaining secretin specimens from the treated animals appeared to have the same characteristic pattern. These patterns appeared somewhat faint, even with protein loads above 80 ~g/cm. However, computer-aided comparison of secretin-stimulated specimens from the two groups of animals showed no significant differences in protein patterns (cf Figs 2A and 2B). Some preliminary experiments were performed with secretions from animals with ductal hyperplasia [as defined earlier (4)]. Patterns found were very Digestive Diseases and Sciences, Vol. 27, No. 5 (May 1982)

PANCREATIC SECRETIONS IN CARCINOMA A

NORM-SEE 1141 9

.4

U.I k.J Z

rn c12

,,--t

~D

~0.0

2.0

4.0 6.0 8.0 CENTIMETERS FROM THE ANODE

I0,0

Fig 2. IEF patterns of secretin-stimulated hamster pancreatic juice. The shaded area represents the 95% confidence interval for (A) 14 specimens from the normal control group, and (B) 13 specimens from animals with carcinoma. The conditions and experimental details are given in Figure 1 and in Materials and Methods.

similar to the corresponding normal controls. The secretions did not exhibit the abnormal patterns seen for any of the animals with carcinoma. DISCUSSION Recent studies with the hamster model for ductal adenocarcin0ma have shown that the enzyme and protein outputs of the pancreatic ductal secretions have been impaired (4). In the present investigation we have used the same hamster model to determine whether the protein composition of the secretions differs in the animals with cancer, or whether decreased outputs of the "normal" protein secretions are the only change seen. There were three striking findings in this study. First, the protein composition was significantly abnormal when CCK was the secretory stimulus. Abnormal protein secretions were found also after secretin stimulation (Figure 3, column 7), but the patterns were not significantly different from the controls. We have attributed the greater significance of the CCK data to the "reproducibility" of the protein patterns when this hormone was used, as opposed to the much greater variability found with secretin-stimulated specimens. Thus in those Digestive Diseases and Sciences, Vot. 27, No. 5 (May 1982)

experiments where secretin was given, no significant differences were apparent between normal animals and those with cancer (Table 1 and Figures 2A and 2B). This suggests that the source of the abnormal proteins is the acinar cells themselves, since CCK is known to stimulate them primarily. This, however, does not exclude the possibility that abriormal secretions have also originated from other cells such as the mucin-secreting cells, tumor cells, and goblet cells. Second; there were differences in the protein patterns and the levels of several protein bands in secretions from animals with carcinoma. An unusually dark band was present at pH 7.5 just below the major amylase band (A2), found in normal secretions at pH 7.65. It is very possible that this band contains several proteins. Separation of this band by two-dimensional electrophoresis will be necessary to identify and characterize the parent band and any substituent proteins by size, charge, and enzymatic activity. Two bands, present in normal secretions at pH 6.9 and 7.0, were missing in secretions from animals with carcinoma. Marked decreases also occurred in proteins having isoelectric points greater than pH 9. Positive identification

441

ALLAN ET AL

CANC-SEC [13] 9

E~

~JJ U Z

Pr~ C~

~ZI. CI

4.0 B.O 8.0 CENTIMETER5 FROMTHE ANODE Fig 2. Continued.

2.0

10,0

2

3

4

5

6

CA

N

CA

N

CA

7

RNa,,

A1 A2 A3

OPC -1

+

N

CA

Fig 3. IEF protein patterns of hamster pancreatic juice. CCK-stimulated secretions (columns 1-6) from normal controls (columns 1, 3, and 5), and from animals with carcinoma (columns 2, 4, and 6). Column 7 is a secretin-stimulated specimen from a hamster with carcinoma. The protein loads applied at the origin (0-) were: 183 Ixg, 139 Ixg, 178 ~,g, 176 Ixg, 180 ~g, 179 t,g, and 180 p,g, for columns 17, respectively. Twenty-four protein bands were present. Amylase (A), ribonuclease (RNase), and procarboxypeptidase A (PC-A) were identified as described in Materials and Methods. One major (A2) and two minor amylase bands (A3 and AI) were identified by enzymatic activity. Traces of additional amylase bands were seen only after a prolonged (18-hr) incubation.

442

Digestive Diseases and Sciences, Vol. 27, No. 5 (May 1982)

P A N C R E A T I C S E C R E T I O N S IN C A R C I N O M A

of these substances may lead us to a tumor marker that can be used in humans. Third, the protein concentrations and outputs of CCK-stimulated secretions of animals with carcinoma were significantly lower than the controls (Table 1). Specimens obtained with secretin stimulation showed no differences in the protein concentration. This supports earlier investigations with the same hamster model (4). Our results show that specimens collected with CCK are much more useful for this type of comparison than those obtained with secretin. CCK-stimulated secretions are rich in protein and may be used directly without prior processing to concentrate the protein and to remove the salt. Preprocessing is time-consuming, and is very likely to introduce artifacts which were not present in the original collections. This directly relates to the analysis of human pure pancreatic juice by isoelectric focusing. It is pertinent to the collection of human ductal pancreatic juice by endoscopy where secretin is the only stimulation used. Finally, if our findings with CCK in the hamster model are relevant to the clinical situation, it may be that a CCK stimulus would indeed demonstrate a secretory abnormality. This possibility should be investigated. Experiments are in progress to separate hamster secretions by two-dimensional electrophoresis so that we can identify the abnormal protein bands seen in the carcinoma specimens and characterize the changes in the protein composition which have taken place. In addition to these studies, we are investigating abnormalities in protein secretion which may occur over the period of time during treatment with the carcinogen, but before the tumors actually develop. Since it has been shown that duct obstruction also caused abnormalities of protein secretion independent of the development of tumors (13), this abnormality, too, should be characterized further.

Digestive Diseases and Sciences, Vol. 27, No. 5 (May 1982)

ACKNOWLEDGMENTS The authors would like to acknowledge Drs. J.A. Beeley and P.D. Eckersall (University of Glasgow, Scotland) for kindly supplying the amylase detection strips used for this study.

REFERENCES 1. Fitzgerald PJ: Pancreatic cancer: The dismal disease. Arch Pathol 100:513-515, 1976 2. Dreiling DA, Greenstein A: Diagnosis of pancreatic disease. In The Pancreas. LC Carey, (ed). St Louis, CV Mosby, 1973, p 61 3. Pour P, Kruger FW, Cardesa A, Althoff J, Mohr U: A new approach for induction of pancreatic neoplasms. Cancer Res 35:2259-2268, 1975 4. Reber HA, Johnson FE, Montgomery C, Carl WR: Pancreatic secretion in hamsters with pancreatic cancer. Surgery 82:34-41, 1977 5. Bailey JL: Techniques in Protein Chemistry. New York, Elsevier, 1962, pp 234-294 6. Allan BJ, Tournut R, White TT: Intraductal activation of pancreatic zymogens behind a carcinoma of the pancreas. Gastroenterology 65:412-418, 1973 7. Burdett PE, Kipps AE, Whitehead PH: A rapid technique for the detection of amylase isoenzymes using an enzyme sensitive test-paper. Anal Biochem 72:315-319, 1976 8. Tournut R, Allan BJ, White TT: Cancer, pancreatitis, and the detection of the isoenzymes of DNAase, RNAase and amylase. Clin Chim Acta 88:345-353, 1978 9. Allan BJ, White TT: A refined cellulose acetate electrophoretic technique for analysis of human pancreatic juice. Biochem Med 12:166-182, 1975 10. Allan BJ, White TT, Kirk J, Schilling JJ: Techniques for reproducible transient-state isoelectric focusing of human pancreatic secretory proteins with computer-assisted pattern matching, averaging and analysis. Anal Biochem 113:1-12, 1981 11. Allan BJ, White TT, Schilling JJ: A simplified method for computer-aided analysis of matched IEF patterns of pancreatic secretory proteins. Fed Proc 40:1868, 1981 12. Doria JC, Mosley JG, Reber HA: Effect of a single exposure of carcinogen on pancreatic function in hamsters. J Surg Res 30:21-26, 1981 13. Austin JL, Roberts C, Rosenholz MJ, Reber HA: Effects of partial duct obstruction and drainage on pancreatic function. J Surg Res 28:426-433, 1980

443