- Email: [email protected]
Prostaglandins and the colonic epithelium
GASTROENTEROLOGY
Prostaglandins Epithelium
1991;101:1229-1234
and the Colonic
Effects of Misoprostol on Crypt Size, Cell Production, and Cell Migration in the Dog ROBERT A. GOODLAD, NIKKI MANDIR, STUART JAMES L. ALLEN, and NICHOLAS A. WRIGHT Histopathology
Unit, Imperial Cancer Research Fund. London. England
Colonic epithelial cell division, cell migration, and cell transit were investigated in dogs given 300 pg * kg-’ . day-’ of the prostaglandin E, analogue, misoprostol, for 11 weeks. The animals were then injected with [3H]thymidine and killed at timed intervals. The distribution of labeled and mitotic cells within the crypts was determined by scoring autoradiographs. There were no significant differences in mitotic index or labeling index between the two groups. The data were pooled and converted to give crypt cell production rates of 51.1 k 11.1 (control) and 58.24 + 8.6 cells per crypt per hour (test). However, the crypt length and cell population were slightly, but significantly, greater in the misoprostoltreated group (P < 0.01). The movement of the wave of labeled cells with time after injection was used to calculate the median cell migration rates, which were 23.50 ? 3.03 cell positions per day (control) and 18.30 ? 2.56 (test). Thus, misoprostol had no significant effect on either the cell migration rate or the transit rates.
rostaglandins of the E series can increase stomach mucosal mass by inducing hyperplasia. Proliferative responses in the colon have also been reported. Prostaglandins are short-lived derivatives of fatty acids with a wide range of biological actions. They may have an important role in the control of mucosal function and integrity in the gastric mucosa and can protect the stomach from the adverse effects of many noxious agents (l-4) in addition to being potent inhibitors of acid secretion (5). The study and use of prostaglandins has been limited by their short life span, but stable long-acting analogues are now available. One such analogue is misoprostol, a methyl ester of prostaglandin E,, which has a variety of prostaglandin-like effects on intestinal
P
LEVIN,
mucosa (6,7) and has been used as an antiulcer agent (8), especially when associated with nonsteroidal antiinflammatory agents (9). Misoprostol increases gastric mucosal mass (lo), which could either be the result of increased epithelial cell production (11,lZ) or decreased cell loss (13-15). We have recently shown that this gastric hyperplasia is the result of increased cell production (16) and that cell migration rates are in fact increased, but, because the cells have further to travel, transit times are not significantly altered (17). There are far less data available on the effects of prostaglandins on epithelial cell proliferation in the rest of the intestine, but, as in the stomach, conflicting data exist. Endogenous prostaglandin synthesis occurs in the colonic submucosa and can reach the epithelium (18). There are some reports of prostaglandin-associated increases in mucosal mass both in the small intestine (19,20) and the colon (21,22). In addition, prostaglandins may be able to protect the colonic epithelium from induced damage (23,24). The nature of any proliferative response of the distal intestine and colon to prostaglandins is confused by reports of inhibition of endogenous prostaglandin synthesis, by agents such as indomethacin and aspirin, increasing colonic proliferation (2 5,26), whereas exogenous prostaglandins may suppress proliferative activity (27); nevertheless, an antiproliferative role for indomethacin has also been reported in experimental colonic tumors (28). The present paper quantifies colonic epithelial cell proliferation and median migration and transit times
Abbreviation used in this paper: CCPR, crypt cell production rate. Q 1991 by the American Gastroenterological Association OOM-5085/91/$3.00
1230
GOODLAD ET AL
(29) as part of a major investigation of the effects of misoprostol on the gastrointestinal epithelium (16,17,30).
Methods Experimental
Plan
Thirty dogs were given a gelatin capsule containing 300 kg/kg misoprostol once a day for up to 98 days, whereas 30 (control] animals were given empty capsules. On day 77, all the dogs were injected IV with 0.25 mCi/kg of tritiated thymidine at timed intervals. Six test and 6 control animals were killed 1 hour after injection and used to generate the flash-labeling and mitotic index data. Three test and 3 control animals were killed on subsequent days for the migration study. On day 77, alternate control and test animals were killed between 8 AM and 2 PM. On subsequent days, they were killed between 8 AM and 11 AM. The last feed was given the afternoon before. The digestive tract of each dog was removed in toto, and samples of mucosa from various sites were fixed in Carnoy’s fluid and stored in 70% ethanol. The colonic tissue was taken from a position half the distance from the ileocecal valve to the rectum.
CASTROENTEROLOGY
it must have had four or more grains overlying its nucleus. Thirty crypts per animal were scored. The number of labeled and mitotic cells per crypt was recorded and used to calculate the labeling and mitotic indices (number of dividing cells per 100 crypt cells). All slides were coded before scoring. Examination of the histological sections and of microdissected crypts for another study showed that most crypts were relatively straight; thus, short or curved crypts would not contribute significantly to the crypt population. The crypt circumferential cell count was determined by counting the number of cells in 100 crypts cut in the transverse orientation from the day 77 animals. The crypt cell population for the day 77 dogs was calculated from the product of the crypt length and the crypt circumferential cell count (32). The position of a marker point on the distribution of labeled cells was calculated in two ways. In the first method, the labeling distribution in the crypt was taken as following a normal distribution (normal score tests showed that the distribution was not significantly nonnormal), and the point at which the labeled cells would reach a value of 50% of the maximum was calculated in terms of the normal distribution. Half-Maximum
Animals Male beagle dogs, 8-10 months old and weighing 11.2-14.9 kg, were used. They were housed individually and kept in environmentally controlled rooms with a 12hour light/la-hour dark cycle. Purina-certified canine diet no. 5007 (Ralston-Purina Co., St. Louis, MO) was provided 3 hours after the test agent for 2 hours. The animals were killed by IV injection of sodium pentobarbital and exsanguination. Tap water was available ad libitum. The antemortern part of this study was performed at Hazleton Laboratories America Inc., Madison, WI.
Histology Colonic mucosa was embedded in wax, and several 4-km sections, separated by at least 20 km, were cut and mounted. Autoradiographs were prepared by the dipping method (31) using Ilford K2 emulsion (Ilford Ltd., Knutsford, Cheshire, England). After 4-week exposure, the autoradiographs were developed in Kodak D19b (Hemel Hempstead, Hertfordshire, England) and fixed and stained with H&E.
Point = Mean + SD x &
In the second method, the marker position was the 60th percentile (the cell position that had 90% of the labeled cells below it]. This position was derived by determining the number of labeled cells per crypt column (n), and then finding the cell position from the crypt base (r), where r = (n + 1) x 0.9. It was easier to score the data starting from the uppermost labeled position using the rank r’, where r’ = (n + 1) - r. All values were rounded up.
Statistics All results are presented as the mean + SEM. Data were tested by a two-sided t test. Lines were fitted by least-squares linear regression. To give a measure of the statistical variation within crypts of the day 77 dogs, the standard deviations of each animal were squared, added, and divided by 6, and the square root was taken to give the pooled within animal standard deviation. The distribution of labeling within the crypt was tested for normality by testing the correlation of the labeling data and its calculated normal scores.
Results
Data Gathering From Autoradiographs Slides were examined systematically until wellorientated crypts (sectioned along the axis of the lumen) were found. The low probability of finding such axially sectioned crypts necessitated the sampling of many fields from fields in several slides. The crypt was then scored, working from the base to the top, and the presence and location of labeled and mitotic cells were recorded. The crypt length (the cell count from the base to the top of the crypt) was also noted. For a nucleus to be scored as labeled,
Vol. 101. No. 5
There was no significant difference in body weight or food intake between the control and the test groups. Misoprostol treatment was associated with a small, but highly significant, increase in crypt length of 8.1% (P < 0.01; Figure 1). The crypt diameter (circumferential cell count) was also, but very marginally, increased in the test group (from 39.64 _’ 0.73 to 40.47
* 1.40).
The
crypt
cell population
(crypt
col-
November
.,,,:, I MISOPROSTOL
1991
::
\‘\‘.’ \‘<‘\’ \‘%‘\’ \‘\‘\’ , ,~/ :,>\, \‘\‘\’ \‘\‘\’ ,I,
I
I
CCPR =
umn length times crypt diameter] was consequentially also significantly increased (by 10.5%) with misoprostol treatment (P < 0.01). There were no significant changes in either the number of mitotic cells or labeled cells per crypt or in the the mitotic index or the labeling index of the two groups (Figure 2). The pooled within-animal standard deviation of the day 77 animal control and test groups was 5.73 and 6.53 for the labeled cell per crypt, 0.99 and 0.93 for the mitotic data, and 20.15 and 18.12 for the crypt length. Although all the data were scored by the same observer, the slides [but not identical crypts) from 6 animals were also scored by a second observer. The correlation coefficient for the crypt length data between the observers was 0.988; for the labeling data, 0.803; and for the cell position of half-maximum labeling, 0.92 1. aooT5
6 k a
600'
,rg o.a-
2
0.6'
E .g 0 c I
0.4-
1231
Proliferative Index X Cell Population 100
X-
Phase Duration 24
I
Figure 1. The effects of V-day treatment with misoprostol on crypt length (measured in cell positions from base to colonic lumen), crypt diameter, and crypt cell population (length X diameter). There were six control and six test dogs. Shaded bum, Control: cross-hatched bars, test. **Significantly greater than control (P < 0.01).
501T
CELL PROLIFERATION
The labeling indices and mitotic indices were converted to crypt cell production rates (CCPR) as follows:
>>:,
\‘\‘.’ \‘\‘\’ \‘\‘\’ \‘<‘\’ \‘\‘~’ \‘\‘\’ \‘\‘\’
AND COLONIC
The duration of the S phase was taken as 8 hours, and the mitotic duration was taken as 1 hour. The CCPR of the control group was calculated as 64.4 ? 10.5 and 37.8 1 11.6 cells per crypt per hour for the labeling- and mitoses-based data and 79.76 2 8.68 and 30.30 + 8.46 for the test group, giving a mean control CCPR of 51.1 and test CCPR as 55.03 cells per crypt per hour. Labeled cells were seen at the top of the crypt on day 80; thus, only crypts for days 77, 78, and 79 were scored. The labeling index per position was plotted, and normal scores tests confirmed that this approximated to a normal distribution (P < 0.05). The mean and standard error of the distribution were then calculated and used to define the half-maximum labeling position, which is shown in Figure 3A. The median migration rates (the slopes of the curves) were 22.403 IfI 1.212 and 18.303 +- 1.751 cell positions per day for the control and the test groups, respectively (P = 0.057). The location of the halfmaximum labeling point in the two groups at day 77 was at position 43.08 5 1.21 and 48.40 & 1.05 for the control and test groups, respectively; consequently, the time taken for this point in the distribution to reach the luminal surface would be 5.4 and 7.1 days for the control and the test groups, respectively. The data were also analyzed without making assumptions concerning the normality of the data. The changes in the position of the 90th percentile are shown in Figure 3B. The migration rate derived from these slopes was 23.50 t 3.03 for the control and 19.35 ? 2.55 for the test group (P = 0.10). The data from the above are summarized in Table 1. Thus, migration appeared to be decreased by an average of 22% in the misoprostol-treated group. The mean crypt lengths of the control and test animals were 164.18 ? 2.6 and 177.55 + 3.1 cells; consequently, cells could transit the entire crypt in 7.3 days and 9.7 days, respectively.
0.2'
Discussion Figure 2. The effects of 77-day treatment with number of mitoses per crypt, mitotic index, cells per crypt, and labeling index. There were test dogs. Shaded bars, Control; cross-hatched
misoprostol on the number of labeled six control and six bars, test.
The modest increase in colonic crypt cell population size of 8% with misoprostol treatment was statistically significant but far less than the increase of 42% observed in the stomach of the same dogs (16).
1232
GOODLAD ET AL.
GASTROENTEROLOGY
100,
20_ 77
A
78 Day
79
76
79
loo-
40.
20_
B
, 77
W Figure 3. The migration of the leading edge of labeled cells based on the half-maximum labeling cell position (A) and on the 90th percentile (B). Dogs were dosed with misoprostol for 77 days and then injected with 0.25 mCi/hg [3H]thymidine. There were six control and six test dogs on day 77 and three control and three test dogs on the subsequent days. -0--, Control; --•--, test.
The question of whether this was the consequence of increased proliferation or of reduced migration could not be resolved, but the calculated crypt cell production rates were slightly increased in the test group, whereas the migration rate was slightly decreased.
Table 1. Effects of Misoprostol on the Crypt Length, Cell Migration, and Transit Rates Crypt length
(cells] Using the 50% data Control Test Using the 90th percentile data Control Test
Migration
(cell positions/day)
Transit time
(days)
164.20 ? 2.55 177.55 ” 3.07
22.40 -c 1.21 18.30 2 1.75
7.33 T 0.48 9.70 2 1.01
164.20 k 2.55 177.55 2 3.07
23.50 f 3.03 19.35 2 2.56
6.99 2 0.94 9.18 2 1.27
NOTE. The dosage of misoprostol was 300 kg kg-’ day-’ for 11 weeks; crypt length was measured in cells from the base to the luminal surface. The last two measures were derived either assuming that the positions’ labeled data followed a normal distribution or not making such an assumption and using the 90th percentile.
Vol. 101. No. 5
The proliferative results have been expressed as both proliferative indices and as number of dividing cells per crypt. Proliferative indices can be affected by changes in the denominator; thus, more meaningful information is obtained when the data are expressed per crypt, as has been shown in our study of the gastric fundus (16).There was no significant change in labeling index of the gastric mucosa, but because the gland cell population (denominator) had also increased, there was in fact a significant effect on the number of labeling cells per gland and, hence, on the gland cell production rate. The discrepancy between the mitotic and the labeling data-based CCPRs could be explained if there was a circadian rhythm in cell division, which has been well documented in several other species (29); also, such rhythms can be particularly pronounced in the colon (33). The labeling data should be the more robust measure, because the number of labeled and mitotic cells seen is inversely proportional to the duration of the respective phases; thus, about 8-10 times more labeled than mitotic cells should be seen. The recognition of labeled cells in section is also more positive, because some mitotic figures, especially prophase, are difficult to distinguish. The conversion of the data per crypt into crypt cell production rates assumes that the duration of the mitotic and DNA synthesis phases does not differ between groups; this calculation assumed that S-phase duration was 8 hours and and the mitotic duration 1 hour (29) and took the mean of the mitotic and labeling data to give the CCPR of the control group as 51.1 cells per crypt per hour and the CCPR of the test group as 58.3 cells per crypt per hour. The values per 1000 cells are 7.9 and 8.07 cells produced per 1000 cells per hour, which is in good accord with the values of 6-12 cell per 1000 quoted for the rat colon (34) and not greatly dissimilar to the 13-17 cell per 1000 values reported in the mouse (35). The selection and study of well-orientated crypts introduces a degree of bias into all cytokinetic studies, because one of the fundamental principles of morphometric analysis, viz., the random choice of test area, is not fulfilled. Some short crypts may thus be excluded from the analysis, because it would be assumed that they are long crypts sectioned at an angle. Tortuous crypts would also be excluded, but they are rare outside of disease. Nonetheless, the preparation of several histological sections, cut at different levels of the block, introduces a large random element into the selection of the crypts. A large variation in crypt length may necessitate the use of a crypt normalization program (36) in which the effect of length on distribution is corrected for by projecting the data onto a standardized crypt, the length of which is that of the mean value for the
November
1991
group. Examination of the crypt lengths in this study and in microdissected canine colonic crypts investigated in another study showed most of the crypts to be relatively straight and uniform; thus, problems related to variation in crypt size were minimized in the present study. However, the distribution of labeled cells in the crypts is not random but is in fact highly ordered, with moderate labeling indices seen at the base of the crypt, which then increase in the basal two thirds of the crypt where most cells are in the cell division cycle; labeling then falls off as cells leave the cell cycle and start to differentiate (37). Consequently, to obtain the maximum amount of useful kinetic data, proliferative index distribution data must be gathered from well-orientated, axially sectioned crypts (29). The importance of this was shown in our study of the gastric mucosa (16), where concomitant changes in the denominator of the labeling index can, and have, distorted data based on a simple proliferative indices. Nevertheless, it should be borne in mind that the use of well-orientated crypts in kinetic studies could introduce the risk that the results are not always totally representative of the entire cell population. Whereas no significant effect of misoprostol on cell migration could be shown, the effect approached statistical significance (P = 0.~157) when the halfmaximum position was used, which would tend to support the hypothesis that decreased migration is the cause of colonic hyperplasia. This directly contrasts our findings in the stomach (17). The assumption that the distribution of the labeled cells in the crypt approximated a normal distribution may not be valid, and when the data were analyzed using the 90th percentile, no significant difference was noted. The two sets of curves were, nevertheless, very similar, and the lack of significance in the percentile data could be attributed to the greater error seen in the calculated marker positions. The migration (23.5 +- 3.03 cell positions per day for the controls and 19.4 ? 2.56 for the test animals) and transit times (7.33 of: 1.5 days for the controls and 9.2 * 2.5 days for the test animals) reported here are, to the best of our knowledge, the first median transit and migration times reported for the dog colon. The presence of labeled cells at the top of the crypt was noted at day 80; thus, the minimum transit time was about 3 days. Minimum time can only show how long it takes for the vanguard of the labeled cells to travel from their starting positions (almost half way up the crypt) to the luminal surface. Therefore, median transit times are the far more meaningful measure. Values for transit times in other species are also scarce but most estimates quote a value of 2-8 days (29) in the upper part of the crypt. Transit time in the rodent stomach has been given as l-3 days (34,35),
MISOPROSTOL
AND COLONIC
CELL PROLIFERATION
1233
which is considerably quicker than reported in the dog or in humans in whom values of 4-6 days have been derived (29); nonetheless, these values are usually minimum transit times or are based on the presence of label in 50% of the cell positions immediately below the surface and, thus, would underestimate the true, median transit time. Whether the proliferative effects of misoprostol and other E-series prostaglandins in the stomach, and other sites, are direct or indirect remains to be seen. Cell culture has also provided conflicting data; whereas some workers have found no effect of prostaglandins on intestinal epithelial cultures (38), others report the inhibition of proliferation in gastric-derived cell cultures (39). The results of our previous study indicate that most of the proliferative response to misoprostol is seen in the stomach, which could either reflect the sensitivity of the stomach to this agent or that the effects in the stomach are of a local nature. Although much information has been gathered on the influence of systemic agents on cell division rates in the small intestine (40), there is far less information on the effects of such agents on the colon. There is some evidence that the colon and small intestine are subject to different control mechanisms (29); the different results obtained in the colon and in the stomach of the present series of studies would tend to support such a hypothesis. To conclude, misoprostol has a small, but significant, effect on the cell population of colonic crypts. In the current study, the effect could have been either the result of an increase in cell production or a decrease in cell migration. The magnitude of the response was such that no significant change in proliferative indices or migration rates could be observed; however, because of the precision of the techniques used, it is unlikely that changes will be observed unless considerably more animals are studied.
References 1. Jacobson
2.
3.
4.
5.
6.
ED. Direct and adaptive cytoprotection. Dig Dis Sci 1986;31(Suppl):28S-31s. Lacy ER. Effects of absolute ethanol, misoprostol. cimetidine, and phosphate buffer on the morphology of rat gastric mucosa. Dig Dis Sci 1986;31(Suppl):lOlS-107s. Liss RH, Letourneau RJ, Schepis JP. Evaluation of cytoprotection against ethanol induced injury in gastric mucosa pretreated with misoprostol. cimetidine. or placebo. Dig Dis Sci 1986;31(Suppl):108S-114s. Yeomans ND. The protective effects of prostaglandins in the human gastrointestinal tract. Ther Today 1986;5(Suppl):1925. Tsai BS, Kessler LK, Butchko GM. Bauer RF. Effects of misoprosto1 on histamine stimulated acid secretion and cyclic AMP formation in isolated canine parietal cells. Dig Dis Sci 1987;32: 1010-1016. Smedfors B, Johansson C. Stimulation of duodenal bicarbonate
1234
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
GOODLAD
ET AL.
secretion by misoprostol. Dig Dis Sci 1986:31(Suppl):96S100s. Wilson DE, Quadros E, Rajapaksa T, Adams A, Noar M. Effects of misoprostol on gastric acid and mucus secretion in man. Dig Dis Sci 1986;31(Suppl):126S-129s. Bright Asare P, Sontag SJ, Gould RJ, Brand DL, Roufail WM. Efficacy of misoprostol [twice daily dosage) in acute healing of duodenal ulcer. A multicenter double blind controlled trial. Dig Dis Sci 1986;31(Suppl):63S-67s. Levi S, Goodlad RA, Lee CY, Stamp G, Walport MJ, Wright NA, Hodgson H. Inhibitory effect of non-steroidal anti-inflammatory drugs on mucosal cell proliferation associated with gastric ulcer healing. Lancet 1990;336:840-843. Dembinski A, Konturek SJ. Effects of E, F, and I series prostaglandins and analogues on growth of gastrointestinal mucosa and pancreas. Am J Physiol1985;248:G170-G175. Assad RT, Dodd DC, Dajani EZ. Effects of misoprostol (prostaglandin E, analogue) on DNA synthesis of canine fundic mucosa. Gastroenterology 1985;88:1310. Assad RT. Dodd DC, Dajani EZ. Effects of misoprostol (prostaglandin E, analogue) epithelial proliferation of canine fundic mucosa. Gastroenterology 1985;88:1310. Uribe A, Johansson C, Rubio C. Effects of prostaglandin E, on cell kinetics of the rat gastrointestinal mucosa. Gastroenterology 1985;88:1621. Fich A, Arber N, Sestieri M, Zajicek G, Rachmilewitz D. Effects of misoprostol and cimetidine on gastric cell labelling index. Gastroenterology 1985;85:57-61. Arber N, Fich A, Zajicek G, Rachmilewitz D. Effects of misoprosto1 on human gastric cell turnover. Gastroenterology 1985;88: 1710. Goodlad RA, Madgwick AJA, Moffatt MR, Levin S, Allen JL, Wright NA. Prostaglandins and the gastric epithelium: effects of misoprostol on gastric epithelial cell proliferation in the dog. Gut 1989;30:316-321. Goodlad RA, Madgwick AJA, Moffatt MR, Levin S, Allen JL, Wright NA. Prostaglandins and the gastric epithelium: effects of misoprostol on cell migration and transit in the dog stomach. Gastroenterology 1990;96:1-6. Craven PA, deRubertis FR. Patterns of prostaglandin synthesis and degradation in isolated superficial and proliferative coionic epithelial cells compared to residual colon. Prostaglandins 1983;26:583-586. Helander HF, Johansson C, Blom H, Uribe A. Trophic effects of E, prostaglandins in the rat gastrointestinal mucosa. A quantitative morphological study. Gastroenterology 1985;89:13931399. Uribe A, Tribukait B, Johansson C. Cell cycle distribution of proliferative and functional cells of the rat jejunum after treatment with oral E, prostaglandins. Stand J Gastroenterol 1987;22:177-184. Gilbertson TJ, Ruwart MJ, Stryd RP, Brunden MN, Friedle NM, Rush BD, Christianson CA. Partial characterization of the gastrointestinal weight changes produced in the female rat by 16.16 dimethyl prostaglandin E,. Prostaglandins 1983;26:745759. Gilbertson TJ, Stryd RP, Brunden MN, Christianson CA, Rush BD. Changes in thymidine uptake in the gastrointestinal tract of the rat following treatment with l6,16 dimethyl prostaglandin E,. Prostaglandins 1984;27:887-898. Wallace JL, Whittle BJR, Boughton-Smith NK. Prostaglandin protection of rat colonic mucosa from damage induced by ethanol. Dig Dis Sci 1985:30:866-876.
GASTROENTEROLOGY
Vol. 101. No. 5
24. Psaila JV, Myers B, Jones IR, Rhodes J. Effects of prostaglandin PGE, on alcohol induced ulceration in the rat colon. Digestion 35:224-228. 25. Craven PA, Thornburg K, DeRubertis FR. Sustained increase in the proliferation of rat colonic mucosa during chronic treatment with aspirin. Gastroenterology. 1988;94:567-575. 26. Craven PA, Saito R, DeRubertis FR. Role of local prostaglandin synthesis in the modulation of proliferative activity of rat colonic epithelium. J Clin Invest 1983;72:1365-1375. 27. DeRubertis FR, Craven PA, Saito R. 16,16 dimethylprostaglandin E,, suppresses the increases in the proliferative activity of rat colonic epithelium induced by indomethacin and aspirin. Gastroenterology 1985;89:1054-1063. 28. Narisawa T, Takahashi M, Masuda T, Nagasawa 0, Ogata N. Niwa M. Anti cancer treatment with prostaglandin synthetase inhibitor in animal models. Gan To Kagaku Ryoho 1986;13: 1329-1335. 29. Wright NA. Alison MR. The biology of epithelial cell populations. Vol 2. Oxford, England: Clarendon, 1984. 30. Goodlad RA, Madgwick AJA, Moffatt MR, Levin S, Allen JL, Wright NA. Prostaglandins and the dog stomach: effects of misoprostol on the proportions of mucosa to muscle and on the proportions of different epithelial cell types. Digestion 1990;45: 212-217. 31. Rogers AW. Techniques of autoradiography. New York: Elsevier, 1979. 32. Wright NA, Carter J, Irwin M. The measurement of villus cell population size in the mouse small intestine in normal and abnormal states: a comparison of absolute measurements with morphometric estimators in sectioned immersion fixed material. Cell Tissue Kinet 1989;22:425-450. 33. Goodlad RA, Wright NA. The effects of starvation and refeeding on intestinal cell proliferation in the mouse. Virchows Arch [B] 1984;45:63-73. 34. Sunter JP, Appleton DR, Rodriguez MSB. Wright NA, Watson AJ. A comparison of cell proliferation at different sites in the large bowel of the mouse. J Anat 1979:129:833-842. 35. Sunter JP, Watson AJ, Wright NA, Appleton DR. Cell proliferation at different sites in the colon of the male rat. Virchow Arch [B] 1979;26:275-287. 36. Cairnie AB, Bently J. Cell proliferation studies in the intestinal epithelium of the rat. Hyperplasia during lactation. Exp Cell Res 1967;46:428-440. 37. Goodlad RA. Gastrointestinal
epithelial
cell proliferation.
Dig
Dis 1989;7:169-177. 38. Stange EF, Schneider A, Preclik G, Ditschuneit H. Prostaglandins are not involved in the differentiation or growth of cultured small intestinal cells. Digestion 1986;35:217-223. 39. Nakamura A, Chiba T, Yamatani T, Yamaguchi A, Inui T, Morishita T, Kadowaki S. Fujita T. Prostaglandin E, and F, alpha inhibit growth of human gastric carcinoma cell line KATO III with simultaneous stimulation of cyclic AMP production. Life Sci 1989;44:75-80. 40. Goodlad RA, Wright NA. Peptides and epithelial growth regulation. In: Polak JM, ed. Regulatory peptides. 1989:180-192.
Received March 7,199O. Accepted April 24,1991. Address requests for reprints to: Robert A. Goodlad, Ph.D., Histopathology Unit, Imperial Cancer Research Fund, 35-43 Lincoln’s Inn Fields, London WCZA 3PN, England.
Prostaglandins Epithelium
1991;101:1229-1234
and the Colonic
Effects of Misoprostol on Crypt Size, Cell Production, and Cell Migration in the Dog ROBERT A. GOODLAD, NIKKI MANDIR, STUART JAMES L. ALLEN, and NICHOLAS A. WRIGHT Histopathology
Unit, Imperial Cancer Research Fund. London. England
Colonic epithelial cell division, cell migration, and cell transit were investigated in dogs given 300 pg * kg-’ . day-’ of the prostaglandin E, analogue, misoprostol, for 11 weeks. The animals were then injected with [3H]thymidine and killed at timed intervals. The distribution of labeled and mitotic cells within the crypts was determined by scoring autoradiographs. There were no significant differences in mitotic index or labeling index between the two groups. The data were pooled and converted to give crypt cell production rates of 51.1 k 11.1 (control) and 58.24 + 8.6 cells per crypt per hour (test). However, the crypt length and cell population were slightly, but significantly, greater in the misoprostoltreated group (P < 0.01). The movement of the wave of labeled cells with time after injection was used to calculate the median cell migration rates, which were 23.50 ? 3.03 cell positions per day (control) and 18.30 ? 2.56 (test). Thus, misoprostol had no significant effect on either the cell migration rate or the transit rates.
rostaglandins of the E series can increase stomach mucosal mass by inducing hyperplasia. Proliferative responses in the colon have also been reported. Prostaglandins are short-lived derivatives of fatty acids with a wide range of biological actions. They may have an important role in the control of mucosal function and integrity in the gastric mucosa and can protect the stomach from the adverse effects of many noxious agents (l-4) in addition to being potent inhibitors of acid secretion (5). The study and use of prostaglandins has been limited by their short life span, but stable long-acting analogues are now available. One such analogue is misoprostol, a methyl ester of prostaglandin E,, which has a variety of prostaglandin-like effects on intestinal
P
LEVIN,
mucosa (6,7) and has been used as an antiulcer agent (8), especially when associated with nonsteroidal antiinflammatory agents (9). Misoprostol increases gastric mucosal mass (lo), which could either be the result of increased epithelial cell production (11,lZ) or decreased cell loss (13-15). We have recently shown that this gastric hyperplasia is the result of increased cell production (16) and that cell migration rates are in fact increased, but, because the cells have further to travel, transit times are not significantly altered (17). There are far less data available on the effects of prostaglandins on epithelial cell proliferation in the rest of the intestine, but, as in the stomach, conflicting data exist. Endogenous prostaglandin synthesis occurs in the colonic submucosa and can reach the epithelium (18). There are some reports of prostaglandin-associated increases in mucosal mass both in the small intestine (19,20) and the colon (21,22). In addition, prostaglandins may be able to protect the colonic epithelium from induced damage (23,24). The nature of any proliferative response of the distal intestine and colon to prostaglandins is confused by reports of inhibition of endogenous prostaglandin synthesis, by agents such as indomethacin and aspirin, increasing colonic proliferation (2 5,26), whereas exogenous prostaglandins may suppress proliferative activity (27); nevertheless, an antiproliferative role for indomethacin has also been reported in experimental colonic tumors (28). The present paper quantifies colonic epithelial cell proliferation and median migration and transit times
Abbreviation used in this paper: CCPR, crypt cell production rate. Q 1991 by the American Gastroenterological Association OOM-5085/91/$3.00
1230
GOODLAD ET AL
(29) as part of a major investigation of the effects of misoprostol on the gastrointestinal epithelium (16,17,30).
Methods Experimental
Plan
Thirty dogs were given a gelatin capsule containing 300 kg/kg misoprostol once a day for up to 98 days, whereas 30 (control] animals were given empty capsules. On day 77, all the dogs were injected IV with 0.25 mCi/kg of tritiated thymidine at timed intervals. Six test and 6 control animals were killed 1 hour after injection and used to generate the flash-labeling and mitotic index data. Three test and 3 control animals were killed on subsequent days for the migration study. On day 77, alternate control and test animals were killed between 8 AM and 2 PM. On subsequent days, they were killed between 8 AM and 11 AM. The last feed was given the afternoon before. The digestive tract of each dog was removed in toto, and samples of mucosa from various sites were fixed in Carnoy’s fluid and stored in 70% ethanol. The colonic tissue was taken from a position half the distance from the ileocecal valve to the rectum.
CASTROENTEROLOGY
it must have had four or more grains overlying its nucleus. Thirty crypts per animal were scored. The number of labeled and mitotic cells per crypt was recorded and used to calculate the labeling and mitotic indices (number of dividing cells per 100 crypt cells). All slides were coded before scoring. Examination of the histological sections and of microdissected crypts for another study showed that most crypts were relatively straight; thus, short or curved crypts would not contribute significantly to the crypt population. The crypt circumferential cell count was determined by counting the number of cells in 100 crypts cut in the transverse orientation from the day 77 animals. The crypt cell population for the day 77 dogs was calculated from the product of the crypt length and the crypt circumferential cell count (32). The position of a marker point on the distribution of labeled cells was calculated in two ways. In the first method, the labeling distribution in the crypt was taken as following a normal distribution (normal score tests showed that the distribution was not significantly nonnormal), and the point at which the labeled cells would reach a value of 50% of the maximum was calculated in terms of the normal distribution. Half-Maximum
Animals Male beagle dogs, 8-10 months old and weighing 11.2-14.9 kg, were used. They were housed individually and kept in environmentally controlled rooms with a 12hour light/la-hour dark cycle. Purina-certified canine diet no. 5007 (Ralston-Purina Co., St. Louis, MO) was provided 3 hours after the test agent for 2 hours. The animals were killed by IV injection of sodium pentobarbital and exsanguination. Tap water was available ad libitum. The antemortern part of this study was performed at Hazleton Laboratories America Inc., Madison, WI.
Histology Colonic mucosa was embedded in wax, and several 4-km sections, separated by at least 20 km, were cut and mounted. Autoradiographs were prepared by the dipping method (31) using Ilford K2 emulsion (Ilford Ltd., Knutsford, Cheshire, England). After 4-week exposure, the autoradiographs were developed in Kodak D19b (Hemel Hempstead, Hertfordshire, England) and fixed and stained with H&E.
Point = Mean + SD x &
In the second method, the marker position was the 60th percentile (the cell position that had 90% of the labeled cells below it]. This position was derived by determining the number of labeled cells per crypt column (n), and then finding the cell position from the crypt base (r), where r = (n + 1) x 0.9. It was easier to score the data starting from the uppermost labeled position using the rank r’, where r’ = (n + 1) - r. All values were rounded up.
Statistics All results are presented as the mean + SEM. Data were tested by a two-sided t test. Lines were fitted by least-squares linear regression. To give a measure of the statistical variation within crypts of the day 77 dogs, the standard deviations of each animal were squared, added, and divided by 6, and the square root was taken to give the pooled within animal standard deviation. The distribution of labeling within the crypt was tested for normality by testing the correlation of the labeling data and its calculated normal scores.
Results
Data Gathering From Autoradiographs Slides were examined systematically until wellorientated crypts (sectioned along the axis of the lumen) were found. The low probability of finding such axially sectioned crypts necessitated the sampling of many fields from fields in several slides. The crypt was then scored, working from the base to the top, and the presence and location of labeled and mitotic cells were recorded. The crypt length (the cell count from the base to the top of the crypt) was also noted. For a nucleus to be scored as labeled,
Vol. 101. No. 5
There was no significant difference in body weight or food intake between the control and the test groups. Misoprostol treatment was associated with a small, but highly significant, increase in crypt length of 8.1% (P < 0.01; Figure 1). The crypt diameter (circumferential cell count) was also, but very marginally, increased in the test group (from 39.64 _’ 0.73 to 40.47
* 1.40).
The
crypt
cell population
(crypt
col-
November
.,,,:, I MISOPROSTOL
1991
::
\‘\‘.’ \‘<‘\’ \‘%‘\’ \‘\‘\’ , ,~/ :,>\, \‘\‘\’ \‘\‘\’ ,I,
I
I
CCPR =
umn length times crypt diameter] was consequentially also significantly increased (by 10.5%) with misoprostol treatment (P < 0.01). There were no significant changes in either the number of mitotic cells or labeled cells per crypt or in the the mitotic index or the labeling index of the two groups (Figure 2). The pooled within-animal standard deviation of the day 77 animal control and test groups was 5.73 and 6.53 for the labeled cell per crypt, 0.99 and 0.93 for the mitotic data, and 20.15 and 18.12 for the crypt length. Although all the data were scored by the same observer, the slides [but not identical crypts) from 6 animals were also scored by a second observer. The correlation coefficient for the crypt length data between the observers was 0.988; for the labeling data, 0.803; and for the cell position of half-maximum labeling, 0.92 1. aooT5
6 k a
600'
,rg o.a-
2
0.6'
E .g 0 c I
0.4-
1231
Proliferative Index X Cell Population 100
X-
Phase Duration 24
I
Figure 1. The effects of V-day treatment with misoprostol on crypt length (measured in cell positions from base to colonic lumen), crypt diameter, and crypt cell population (length X diameter). There were six control and six test dogs. Shaded bum, Control: cross-hatched bars, test. **Significantly greater than control (P < 0.01).
501T
CELL PROLIFERATION
The labeling indices and mitotic indices were converted to crypt cell production rates (CCPR) as follows:
>>:,
\‘\‘.’ \‘\‘\’ \‘\‘\’ \‘<‘\’ \‘\‘~’ \‘\‘\’ \‘\‘\’
AND COLONIC
The duration of the S phase was taken as 8 hours, and the mitotic duration was taken as 1 hour. The CCPR of the control group was calculated as 64.4 ? 10.5 and 37.8 1 11.6 cells per crypt per hour for the labeling- and mitoses-based data and 79.76 2 8.68 and 30.30 + 8.46 for the test group, giving a mean control CCPR of 51.1 and test CCPR as 55.03 cells per crypt per hour. Labeled cells were seen at the top of the crypt on day 80; thus, only crypts for days 77, 78, and 79 were scored. The labeling index per position was plotted, and normal scores tests confirmed that this approximated to a normal distribution (P < 0.05). The mean and standard error of the distribution were then calculated and used to define the half-maximum labeling position, which is shown in Figure 3A. The median migration rates (the slopes of the curves) were 22.403 IfI 1.212 and 18.303 +- 1.751 cell positions per day for the control and the test groups, respectively (P = 0.057). The location of the halfmaximum labeling point in the two groups at day 77 was at position 43.08 5 1.21 and 48.40 & 1.05 for the control and test groups, respectively; consequently, the time taken for this point in the distribution to reach the luminal surface would be 5.4 and 7.1 days for the control and the test groups, respectively. The data were also analyzed without making assumptions concerning the normality of the data. The changes in the position of the 90th percentile are shown in Figure 3B. The migration rate derived from these slopes was 23.50 t 3.03 for the control and 19.35 ? 2.55 for the test group (P = 0.10). The data from the above are summarized in Table 1. Thus, migration appeared to be decreased by an average of 22% in the misoprostol-treated group. The mean crypt lengths of the control and test animals were 164.18 ? 2.6 and 177.55 + 3.1 cells; consequently, cells could transit the entire crypt in 7.3 days and 9.7 days, respectively.
0.2'
Discussion Figure 2. The effects of 77-day treatment with number of mitoses per crypt, mitotic index, cells per crypt, and labeling index. There were test dogs. Shaded bars, Control; cross-hatched
misoprostol on the number of labeled six control and six bars, test.
The modest increase in colonic crypt cell population size of 8% with misoprostol treatment was statistically significant but far less than the increase of 42% observed in the stomach of the same dogs (16).
1232
GOODLAD ET AL.
GASTROENTEROLOGY
100,
20_ 77
A
78 Day
79
76
79
loo-
40.
20_
B
, 77
W Figure 3. The migration of the leading edge of labeled cells based on the half-maximum labeling cell position (A) and on the 90th percentile (B). Dogs were dosed with misoprostol for 77 days and then injected with 0.25 mCi/hg [3H]thymidine. There were six control and six test dogs on day 77 and three control and three test dogs on the subsequent days. -0--, Control; --•--, test.
The question of whether this was the consequence of increased proliferation or of reduced migration could not be resolved, but the calculated crypt cell production rates were slightly increased in the test group, whereas the migration rate was slightly decreased.
Table 1. Effects of Misoprostol on the Crypt Length, Cell Migration, and Transit Rates Crypt length
(cells] Using the 50% data Control Test Using the 90th percentile data Control Test
Migration
(cell positions/day)
Transit time
(days)
164.20 ? 2.55 177.55 ” 3.07
22.40 -c 1.21 18.30 2 1.75
7.33 T 0.48 9.70 2 1.01
164.20 k 2.55 177.55 2 3.07
23.50 f 3.03 19.35 2 2.56
6.99 2 0.94 9.18 2 1.27
NOTE. The dosage of misoprostol was 300 kg kg-’ day-’ for 11 weeks; crypt length was measured in cells from the base to the luminal surface. The last two measures were derived either assuming that the positions’ labeled data followed a normal distribution or not making such an assumption and using the 90th percentile.
Vol. 101. No. 5
The proliferative results have been expressed as both proliferative indices and as number of dividing cells per crypt. Proliferative indices can be affected by changes in the denominator; thus, more meaningful information is obtained when the data are expressed per crypt, as has been shown in our study of the gastric fundus (16).There was no significant change in labeling index of the gastric mucosa, but because the gland cell population (denominator) had also increased, there was in fact a significant effect on the number of labeling cells per gland and, hence, on the gland cell production rate. The discrepancy between the mitotic and the labeling data-based CCPRs could be explained if there was a circadian rhythm in cell division, which has been well documented in several other species (29); also, such rhythms can be particularly pronounced in the colon (33). The labeling data should be the more robust measure, because the number of labeled and mitotic cells seen is inversely proportional to the duration of the respective phases; thus, about 8-10 times more labeled than mitotic cells should be seen. The recognition of labeled cells in section is also more positive, because some mitotic figures, especially prophase, are difficult to distinguish. The conversion of the data per crypt into crypt cell production rates assumes that the duration of the mitotic and DNA synthesis phases does not differ between groups; this calculation assumed that S-phase duration was 8 hours and and the mitotic duration 1 hour (29) and took the mean of the mitotic and labeling data to give the CCPR of the control group as 51.1 cells per crypt per hour and the CCPR of the test group as 58.3 cells per crypt per hour. The values per 1000 cells are 7.9 and 8.07 cells produced per 1000 cells per hour, which is in good accord with the values of 6-12 cell per 1000 quoted for the rat colon (34) and not greatly dissimilar to the 13-17 cell per 1000 values reported in the mouse (35). The selection and study of well-orientated crypts introduces a degree of bias into all cytokinetic studies, because one of the fundamental principles of morphometric analysis, viz., the random choice of test area, is not fulfilled. Some short crypts may thus be excluded from the analysis, because it would be assumed that they are long crypts sectioned at an angle. Tortuous crypts would also be excluded, but they are rare outside of disease. Nonetheless, the preparation of several histological sections, cut at different levels of the block, introduces a large random element into the selection of the crypts. A large variation in crypt length may necessitate the use of a crypt normalization program (36) in which the effect of length on distribution is corrected for by projecting the data onto a standardized crypt, the length of which is that of the mean value for the
November
1991
group. Examination of the crypt lengths in this study and in microdissected canine colonic crypts investigated in another study showed most of the crypts to be relatively straight and uniform; thus, problems related to variation in crypt size were minimized in the present study. However, the distribution of labeled cells in the crypts is not random but is in fact highly ordered, with moderate labeling indices seen at the base of the crypt, which then increase in the basal two thirds of the crypt where most cells are in the cell division cycle; labeling then falls off as cells leave the cell cycle and start to differentiate (37). Consequently, to obtain the maximum amount of useful kinetic data, proliferative index distribution data must be gathered from well-orientated, axially sectioned crypts (29). The importance of this was shown in our study of the gastric mucosa (16), where concomitant changes in the denominator of the labeling index can, and have, distorted data based on a simple proliferative indices. Nevertheless, it should be borne in mind that the use of well-orientated crypts in kinetic studies could introduce the risk that the results are not always totally representative of the entire cell population. Whereas no significant effect of misoprostol on cell migration could be shown, the effect approached statistical significance (P = 0.~157) when the halfmaximum position was used, which would tend to support the hypothesis that decreased migration is the cause of colonic hyperplasia. This directly contrasts our findings in the stomach (17). The assumption that the distribution of the labeled cells in the crypt approximated a normal distribution may not be valid, and when the data were analyzed using the 90th percentile, no significant difference was noted. The two sets of curves were, nevertheless, very similar, and the lack of significance in the percentile data could be attributed to the greater error seen in the calculated marker positions. The migration (23.5 +- 3.03 cell positions per day for the controls and 19.4 ? 2.56 for the test animals) and transit times (7.33 of: 1.5 days for the controls and 9.2 * 2.5 days for the test animals) reported here are, to the best of our knowledge, the first median transit and migration times reported for the dog colon. The presence of labeled cells at the top of the crypt was noted at day 80; thus, the minimum transit time was about 3 days. Minimum time can only show how long it takes for the vanguard of the labeled cells to travel from their starting positions (almost half way up the crypt) to the luminal surface. Therefore, median transit times are the far more meaningful measure. Values for transit times in other species are also scarce but most estimates quote a value of 2-8 days (29) in the upper part of the crypt. Transit time in the rodent stomach has been given as l-3 days (34,35),
MISOPROSTOL
AND COLONIC
CELL PROLIFERATION
1233
which is considerably quicker than reported in the dog or in humans in whom values of 4-6 days have been derived (29); nonetheless, these values are usually minimum transit times or are based on the presence of label in 50% of the cell positions immediately below the surface and, thus, would underestimate the true, median transit time. Whether the proliferative effects of misoprostol and other E-series prostaglandins in the stomach, and other sites, are direct or indirect remains to be seen. Cell culture has also provided conflicting data; whereas some workers have found no effect of prostaglandins on intestinal epithelial cultures (38), others report the inhibition of proliferation in gastric-derived cell cultures (39). The results of our previous study indicate that most of the proliferative response to misoprostol is seen in the stomach, which could either reflect the sensitivity of the stomach to this agent or that the effects in the stomach are of a local nature. Although much information has been gathered on the influence of systemic agents on cell division rates in the small intestine (40), there is far less information on the effects of such agents on the colon. There is some evidence that the colon and small intestine are subject to different control mechanisms (29); the different results obtained in the colon and in the stomach of the present series of studies would tend to support such a hypothesis. To conclude, misoprostol has a small, but significant, effect on the cell population of colonic crypts. In the current study, the effect could have been either the result of an increase in cell production or a decrease in cell migration. The magnitude of the response was such that no significant change in proliferative indices or migration rates could be observed; however, because of the precision of the techniques used, it is unlikely that changes will be observed unless considerably more animals are studied.
References 1. Jacobson
2.
3.
4.
5.
6.
ED. Direct and adaptive cytoprotection. Dig Dis Sci 1986;31(Suppl):28S-31s. Lacy ER. Effects of absolute ethanol, misoprostol. cimetidine, and phosphate buffer on the morphology of rat gastric mucosa. Dig Dis Sci 1986;31(Suppl):lOlS-107s. Liss RH, Letourneau RJ, Schepis JP. Evaluation of cytoprotection against ethanol induced injury in gastric mucosa pretreated with misoprostol. cimetidine. or placebo. Dig Dis Sci 1986;31(Suppl):108S-114s. Yeomans ND. The protective effects of prostaglandins in the human gastrointestinal tract. Ther Today 1986;5(Suppl):1925. Tsai BS, Kessler LK, Butchko GM. Bauer RF. Effects of misoprosto1 on histamine stimulated acid secretion and cyclic AMP formation in isolated canine parietal cells. Dig Dis Sci 1987;32: 1010-1016. Smedfors B, Johansson C. Stimulation of duodenal bicarbonate
1234
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
GOODLAD
ET AL.
secretion by misoprostol. Dig Dis Sci 1986:31(Suppl):96S100s. Wilson DE, Quadros E, Rajapaksa T, Adams A, Noar M. Effects of misoprostol on gastric acid and mucus secretion in man. Dig Dis Sci 1986;31(Suppl):126S-129s. Bright Asare P, Sontag SJ, Gould RJ, Brand DL, Roufail WM. Efficacy of misoprostol [twice daily dosage) in acute healing of duodenal ulcer. A multicenter double blind controlled trial. Dig Dis Sci 1986;31(Suppl):63S-67s. Levi S, Goodlad RA, Lee CY, Stamp G, Walport MJ, Wright NA, Hodgson H. Inhibitory effect of non-steroidal anti-inflammatory drugs on mucosal cell proliferation associated with gastric ulcer healing. Lancet 1990;336:840-843. Dembinski A, Konturek SJ. Effects of E, F, and I series prostaglandins and analogues on growth of gastrointestinal mucosa and pancreas. Am J Physiol1985;248:G170-G175. Assad RT, Dodd DC, Dajani EZ. Effects of misoprostol (prostaglandin E, analogue) on DNA synthesis of canine fundic mucosa. Gastroenterology 1985;88:1310. Assad RT. Dodd DC, Dajani EZ. Effects of misoprostol (prostaglandin E, analogue) epithelial proliferation of canine fundic mucosa. Gastroenterology 1985;88:1310. Uribe A, Johansson C, Rubio C. Effects of prostaglandin E, on cell kinetics of the rat gastrointestinal mucosa. Gastroenterology 1985;88:1621. Fich A, Arber N, Sestieri M, Zajicek G, Rachmilewitz D. Effects of misoprostol and cimetidine on gastric cell labelling index. Gastroenterology 1985;85:57-61. Arber N, Fich A, Zajicek G, Rachmilewitz D. Effects of misoprosto1 on human gastric cell turnover. Gastroenterology 1985;88: 1710. Goodlad RA, Madgwick AJA, Moffatt MR, Levin S, Allen JL, Wright NA. Prostaglandins and the gastric epithelium: effects of misoprostol on gastric epithelial cell proliferation in the dog. Gut 1989;30:316-321. Goodlad RA, Madgwick AJA, Moffatt MR, Levin S, Allen JL, Wright NA. Prostaglandins and the gastric epithelium: effects of misoprostol on cell migration and transit in the dog stomach. Gastroenterology 1990;96:1-6. Craven PA, deRubertis FR. Patterns of prostaglandin synthesis and degradation in isolated superficial and proliferative coionic epithelial cells compared to residual colon. Prostaglandins 1983;26:583-586. Helander HF, Johansson C, Blom H, Uribe A. Trophic effects of E, prostaglandins in the rat gastrointestinal mucosa. A quantitative morphological study. Gastroenterology 1985;89:13931399. Uribe A, Tribukait B, Johansson C. Cell cycle distribution of proliferative and functional cells of the rat jejunum after treatment with oral E, prostaglandins. Stand J Gastroenterol 1987;22:177-184. Gilbertson TJ, Ruwart MJ, Stryd RP, Brunden MN, Friedle NM, Rush BD, Christianson CA. Partial characterization of the gastrointestinal weight changes produced in the female rat by 16.16 dimethyl prostaglandin E,. Prostaglandins 1983;26:745759. Gilbertson TJ, Stryd RP, Brunden MN, Christianson CA, Rush BD. Changes in thymidine uptake in the gastrointestinal tract of the rat following treatment with l6,16 dimethyl prostaglandin E,. Prostaglandins 1984;27:887-898. Wallace JL, Whittle BJR, Boughton-Smith NK. Prostaglandin protection of rat colonic mucosa from damage induced by ethanol. Dig Dis Sci 1985:30:866-876.
GASTROENTEROLOGY
Vol. 101. No. 5
24. Psaila JV, Myers B, Jones IR, Rhodes J. Effects of prostaglandin PGE, on alcohol induced ulceration in the rat colon. Digestion 35:224-228. 25. Craven PA, Thornburg K, DeRubertis FR. Sustained increase in the proliferation of rat colonic mucosa during chronic treatment with aspirin. Gastroenterology. 1988;94:567-575. 26. Craven PA, Saito R, DeRubertis FR. Role of local prostaglandin synthesis in the modulation of proliferative activity of rat colonic epithelium. J Clin Invest 1983;72:1365-1375. 27. DeRubertis FR, Craven PA, Saito R. 16,16 dimethylprostaglandin E,, suppresses the increases in the proliferative activity of rat colonic epithelium induced by indomethacin and aspirin. Gastroenterology 1985;89:1054-1063. 28. Narisawa T, Takahashi M, Masuda T, Nagasawa 0, Ogata N. Niwa M. Anti cancer treatment with prostaglandin synthetase inhibitor in animal models. Gan To Kagaku Ryoho 1986;13: 1329-1335. 29. Wright NA. Alison MR. The biology of epithelial cell populations. Vol 2. Oxford, England: Clarendon, 1984. 30. Goodlad RA, Madgwick AJA, Moffatt MR, Levin S, Allen JL, Wright NA. Prostaglandins and the dog stomach: effects of misoprostol on the proportions of mucosa to muscle and on the proportions of different epithelial cell types. Digestion 1990;45: 212-217. 31. Rogers AW. Techniques of autoradiography. New York: Elsevier, 1979. 32. Wright NA, Carter J, Irwin M. The measurement of villus cell population size in the mouse small intestine in normal and abnormal states: a comparison of absolute measurements with morphometric estimators in sectioned immersion fixed material. Cell Tissue Kinet 1989;22:425-450. 33. Goodlad RA, Wright NA. The effects of starvation and refeeding on intestinal cell proliferation in the mouse. Virchows Arch [B] 1984;45:63-73. 34. Sunter JP, Appleton DR, Rodriguez MSB. Wright NA, Watson AJ. A comparison of cell proliferation at different sites in the large bowel of the mouse. J Anat 1979:129:833-842. 35. Sunter JP, Watson AJ, Wright NA, Appleton DR. Cell proliferation at different sites in the colon of the male rat. Virchow Arch [B] 1979;26:275-287. 36. Cairnie AB, Bently J. Cell proliferation studies in the intestinal epithelium of the rat. Hyperplasia during lactation. Exp Cell Res 1967;46:428-440. 37. Goodlad RA. Gastrointestinal
epithelial
cell proliferation.
Dig
Dis 1989;7:169-177. 38. Stange EF, Schneider A, Preclik G, Ditschuneit H. Prostaglandins are not involved in the differentiation or growth of cultured small intestinal cells. Digestion 1986;35:217-223. 39. Nakamura A, Chiba T, Yamatani T, Yamaguchi A, Inui T, Morishita T, Kadowaki S. Fujita T. Prostaglandin E, and F, alpha inhibit growth of human gastric carcinoma cell line KATO III with simultaneous stimulation of cyclic AMP production. Life Sci 1989;44:75-80. 40. Goodlad RA, Wright NA. Peptides and epithelial growth regulation. In: Polak JM, ed. Regulatory peptides. 1989:180-192.
Received March 7,199O. Accepted April 24,1991. Address requests for reprints to: Robert A. Goodlad, Ph.D., Histopathology Unit, Imperial Cancer Research Fund, 35-43 Lincoln’s Inn Fields, London WCZA 3PN, England.