Whole population cell kinetics of mouse duodenal jejunal iieal and colonic epithelia as determined by radioautography and flow cytometry.код для вставкиСкачать
THE ANATOMICAL RECORD 203:251-264 (1982) Whole Population Cel1 Kinetics of Mouse Duodenal, Jejunal, Ileal, and Colonic Epithelia as Determined by Radioautography and Flow Cytometry HAZEL CHENG A N U MATTHEW BJERKNES Department of Anatomy, Medical Sciences Building, University of Toronto, Toronto, Ontario, Canada M5S IA8 ABSTRACT We report methods for the determination of whole population cell kinetics in the mouse intestinal epithelium by means of radioautography and flow cytometry. Epithelium was isolated from the four regions of the mouse intestine by the perfusion method described in Bjerknes and Cheng (1981a).Two experimental series were performed. In the first series, the tissue from one set of animals was fixed in 3.5% paraformaldehyde, dissociated by serial filtration, and then processed for flow cytometry. In the second series, fixed epithelium from a second set of animals was dissociated by gentle pipetting and then used to prepare dried cell suspensions on slides which were radioautographed. The whole population kinetic parameters determined by the two techniques, flow cytometry and radioautography, were not significantly different, indicating the reliability of both techniques for whole population kinetics determinations. In al1 regions of the intestinal tract, 12-14% of epithelial cells were in the S-phase. From this value, the whole epithelial turnover time was calculated to be about 60 hours in al1 regions of the intestine. The whole epithelial growth fraction was calculated to be 0.23 for the small intestinal epithelium and 0.31 for the colonic epithelium. Detailed analysis of the flow cytometric data showed that there were significantly more cells in early S-phase than in mid and late S-phase. From the isolated epithelial sheets a mean number of crypts per villus was determined for the duodenum, jejunum, and ileum. Single villi and crypts were microdissected for the preparation of squashes. From the squashes a mean number of cells per villus as wel1 as per crypt was determined. From the results, the ratio of the number of epithelial cells in the villus population to the number of cells in the crypt population was determined to be 1.41, 1.31, 1.36 in duodenum, jejunum, and ileum, respectively. The proportion of the three main epithelial cell types, columnar, mucous, and Paneth, was determined with the dried cell suspension preparations. There was a decreasing gradient in the proportion of columnar cells from the proximal to the distal intestine while an increasing gradient was observed in the proportion of mucous cells. We found that mucous cell divisions account for only one half of mucous cell production in the small intestine. In the colonic epithelium, mucous cell divisions account for twothirds of mucous cell production. This was in agreement with previous findings. The intestinal epithelium has a complex three-dimensional arrangement. This complexity has made the study of whole epithelial kinetics difficult. This is reflected in the range of values reported in the few attempts at such measurements in the rat (Leblondand Stevens, 1948; Bertalanffy,1960; Altmann and Enesco. 1967). The difficulty in making accurate whole pop- 0003-276X18212032-0251$04.00 1982 ALAN R. LISS. INC ulation measurements is primarily technological. Conventionalmethods do not allow accurate measurement of a tissue with the topological complexity of the intestinal epithelium. For ex~ - ~ ~ ~ Keceived Fehrudry 20. 1981 accepted December 14. 1981 Matthew Hjerknes's present address is Department of Anatomy. University of Minnesota. 4~135Jackson Hall. 321 Church Street S.E.. Minneapolis. Minnesota 55455. 252 H. CHENG AND M. BJERKNES ample, if whole population kinetic measurements are made on tissue sections, many complex errors wil1 be introduced. What is needed is a method that allows the study of the epithelium as a whole, and that is free of selection artifacts. In this paper we introduce one such approach. Recently, we reported methods for the isolation of structurally intact epithelium, free of underlying nonepithelial elements, from any segment of mouse intestine (Bjerknes and Cheng, 1981a).The integrity of the preparation insures the true representation of al1 epithelial elements. This makes it suitable for whole population kinetic measurements. In this paper we report the application of this preparation to the study of whole population intestinal epithelial cell kinetics using radioautography and flow cytometry. These methods allow US to make accurate whole population measurements of epithelial labeling and mitotic indices. In addition, methods are als0 presented for differential counting of the main epithelial cell types with resulting whole population measurements of their behavior. MATERIALS AND METHODS Animals Al1 animals were adult male Swiss albino mice (25-35 gm). They were housed in animal facilities, maintained on a light-dark cycle of 7:OO to 19:OO hours light, and were given free access to Purina laboratory chow and water. Isolation procedure for intestinal epithelium The isolation procedure has been reported previously in detail (Bjerknes and Cheng, 1981a). Briefly, animals were given an intraperitoneal injection of 1 &i of 3H-thymidine per gm body weight 1 hour before they were killed (al1 animals were killed between 11:OO A.M. and noon). They were anesthetized with an intraperitoneal injection of 60 mglkg sodium pentobarbital and were killed by intraventricular perfusion with a 37" C solution of 30 mM ethylenediamine tetraacetic acid (EDTA) in calcium-magnesium-free Hank's balanced salt solution (CMF).The duration of perfusion was 3 minutes. At the end of the perfusion, a 5-cm segment of duodenum (immediately distal to the pylorus), jejunum (16 cm distal to the pylorus), ileum (immediately proximal to the ileocecal valve), and colon (proximal to the rectum) were collected and placed in cold CMF. The segments were everted and placed on glass rods. The rods were mounted in a stirrer-vibrator (Buchler Scientific) and the gut was vibrated in tubes of cold CMF. The vibration released large sheets (up to 2 cm2)of pure epithelium. Al1 of the epithelium from each segment was isolated. Fixation The epithelium was allowed to settle on ice. The CMF was aspirated and replaced with 30 cc of a cold 3.5% solution of paraformaldehyde in CMF (pH 7.4). The samples were spun for 4 minutes at 500 g in a refrigerated centrifuge. Then the fixative was aspirated and the samples were washed four times with cold CMF. The samples were then stored in about 5 cc of CMF at 4" C at least overnight before use. Preparation of slides for radioautography An aliquot of the epithelium, about 0.5 ml (0.5 X 106 cells), was removed, pelleted, the CMF aspirated, and the epithelium resuspended in about 0.25 ml of distilled water. Then the epithelium was gently pipetted with a pasteur pipette to disperse the cells. After about five strokes, a smal1 drop of the sample was examined with phase-contrast microscopy. Usually, at this point most of the sample was dissociated. If the cells were not mostly dispersed, the pipetting could be prolonged until dispersion was complete. (The sample was checked routinely to ensure that no broken cells were foiind. If broken cells were observed the sample was discarded and the dissociation process repeated with a new aliquot.) A drop or two of the sample was placed on a 37" C slide warming tray to dry. Therefore the preparation used was a dried cell suspension. The average time for the procedure, from dissociation to preparation of slides, was about 3 minutes. The slides were then dipped in methanol. After they were dried they were prestained with alcian blue for 30 minutes then processed for radioautography (Kodak NTB-2 emulsion; 2 weeks exposure; Kodak D-19 development; Kopriwa and Leblond, 1962). The slides were poststained through the emulsion with Harris's hematoxylin for 30 minutes, xanthene for 1 minute, and then mounted for study. Labeling and mitotic index counts The radioautographs were examined at a magnification of X 1,000. The fields scored were 1 mm apart from each other as measured by the stage micrometer. Cells with ten or more silver grains over their nuclei were con- INTESTINAL EPITHELIAL CELL KINETICS sidered labeled. Every cell in the field was scored as being either labeled or not. Mitotic figures were als0 recorded. Three slides per animal were used and about 1,000 cells were scored from each slide. Therefore a total of about 3,000 cells were scored per animal. From the results, the labeling and mitotic indices were determined for the whole epithelium. A total of 13 animals was studied for duodenum, 19 for jejunum, 16 for ileum, and 14 for colon. Differential counts of the epithelial cells Another series of animals were used for differential cell type count determination. The radioautographs were analyzed in the manner described above. Every cell in the field was scored; its cell type and whether or not it was labeled was recorded. The percentage of each cell type and its labeling index (when applicable) was calculated for each animal and the mean for al1 the animals determined. A total of seven animals were studied for duodenum, six for jejunum, nine for ileum, and nine for colon. Preparation of cells for flow cytometry Another series of animals were used for flow cytometry. A total of five animals were studied for duodenum, eight for jejunum, eight for ileum, and seven for colon. The cells were dispersed by passage through a series of nylon meshes (20 pm Nitex). Each sample was first passed five times through one layer of Nitex in a 13-mmsweeney filter holder (Millipore Inc.), then three times through two layers of Nitex, three times through three layers, and finally five times through four layers. The filters were routinely checked after each step for residual clumps of cells, and if clumps were found, the sample was passed through that filter three additional times. This was usually sufficient to disperse the clumps. If not, the sample was discarded and another aliquot was used. After dispersion, a drop of each sample was examined for doublets with phase-contrast microscopy. Samples with more than 1% doublets were discarded and another aliquot was used. For this method, the sample volume should be kept to about 1 cc (1-2 x i 0 6 cells). If a Iarger sample volume is necessary, the sample should first be dispersed by gentle pipetting and then passed through the series of filters. The filtration is necessary after the pipetting to break up smal1 clumps and doublets; without it. doublets are as high as 3%. After dispersion, the cells were pelleted and 253 resuspended in a solution of 0.1Y0 RNAase (Sigma type IA) in Dulbecco’s phosphate-buffered saline (PBS).They were incubated at 37” C for 2 hours. Then the cells were pelleted and resuspended in a solution of 0.025% propidium iodide (Sigma)in 0.037 M sodium citrate. After 5 minutes staining in the dark, the cells were pelleted and resuspended in PBS. The cells were routinely examined with ultraviolet microscopy and no background cytoplasmic staining was observed. The samples were then kept in the dark until analysis with the flow cytometer. Flow cytometry The samples were first filtered five times through four layers of Nitex and then run in our TPS-1 cell sorter (Coulter Electronics Inc.). Our sorter has a 30 mW argon ion laser at 488 nm. The sheath fluid was PBS. The sample flow rate was kept under 300 cellslsecond. For each animal, about 20 runs (a total of 100.000 to 200,000 cells) were analyzed. The data were transferred int0 an interfaced Z-80-based microcomputer (Arisia Microsystems Inc., Mississauga, Ontario, Canada) running under the CPlM operating system (Digital Research Inc., California). Each run from each animal was normalized to a population of 100,000 cells. A mean DNA histogram for each animal was then calculated from the normalized runs. Analysis of the D N A histograms For the duodenum and colon the mean DNA histogram from each animal was analyzed independently by the method of Fried and Mandel (1979)using a copy of their multigaussian curve-fitting routine. The program was run on the DEC-10 computer at the University of Toronto Computing Center. The results were als0 analyzed on our microcomputer system using the multigaussian curve-fitting routine described in the Appendix. In addition to the kinetic parameters obtained with the Fried program, the program described in the Appendix provides conclusions about the distribution of cells in four different subcompartments within the S-phase (see Appendix). Since we found no significant differences in kinetic parameters as determined by the Fried program and those determined on our microcomputer (see Results), data for the jejunum and ileum were analyzed by the microcomputer-based program alone. We als0 derived an “average”epithelial histogram for each region of the intestine by taking 254 H. CHENG AND M. BJEHKNES the average of the mem DNA histograms from each of the animals used. Direct determination of the crypt to uillus ratio Five animais were used in this study. The epithelium from a segment of duodenum, jejunum, and ileum was isolated directly int0 2.5% glutaraldehyde for maximum preservation of the structural integrity. The isolated epithelium was fixed for 30 minutes in the Same fixative. An epithelial sheet was chosen from each animal and the total number of villi and crypts in the sheet was counted under a stereoscope. From these results the number of crypts per villus was determined for each region of the small intestine. Another epithelial sheet from each animal was stained for 7 minutes with Harris hematoxylin. Villi and crypts were microdissected (great care was taken to leave the surface epithelium with the villi). Four villi and four crypts were randomly selected from each animal, dehydrated with graded ethanol, and mounted individually with permount on microscope slides. The villi or crypts were then slowly squashed to disperse the epithelial cells. The number of cells per villus or crypt was determined from these slides. The ratio of villus cells to crypt cells was calculated for each region of the small intestine by multiplying the number of crypts per villus by the number of cells per crypt and then dividing the result int0 the number of cells per villus. Statis tical methods Results from different animals were kept distinct. Means and associated error values were determined for any given parameter by considering the mean of the results from each animal as independent samples, determining sums and sums of squares, and then using as N the number of animals studied. Means were compared with “Student’s’’t-test. The probability level used for a decision of a significant difference was P s 0.05. RESULTS A typical example of a radioautograph of a dried cell suspension is shown in Figure 1. Three of the four main epithelial cell types (columnar, mucous, and Paneth) were easily identified (Fig. 2). Enteroendocrine cells were not clearly defined in these preparations. We have not developed methods to clearly demonstrate them. However, in terms of the whole population kinetic parameters measured here, the inability to distinguish this population from the columnar cells was not significant (enteroendocrine cells represent less than 1% of the epithelial population and have essentially the Same turnover characteristics as do the columnar and mucous cells; Cheng and Leblond, 1974a,b). Whole population kinetic parameters of mouse intestinal epithelium as derived from the radioautographs are shown in Table 1. The Fig. 1. Photomicrograph of an average radioautograph of a dried cell suspension. X 550. 255 INTESTINAL EPITHELIAL CELL KINETICS Fig. 2. Photomicrographs of b i e d cell suspension preparation showing in a) a mucous cell, b) a Paneth cell (arrow- TABLE I . Whole population kinetic parameters head), and c) a Paneth cell (arrowhead)and columnar cells. X 900. of mouse intestinal epithelium determined by radioautography S-phase ‘70 (X i S.E.) G2-phase% (X i S.E.) M-phase % (X I S E . ) G ,-phase ‘70 (X I S.E.) Duodenum 13.0 f 0.52 1.0 f 0 . 2 4 83.2 I 1.11 Jejunum 12.3 f 0.72 0.8 I 0.05 84.1 f 0.73 Ileum 13.3 f 0.43 0.8 f 0 . 0 5 83.0 f 0.61 Colon 12.2 i 0.52 3.0 2.6 2.5 2.6 2.5 2.7 2.5 2.5 0.8 i 0.03 84.3 f 0.67 f 0.72’ f 0.10’ f 0.16’ I 0.12’ f0.15‘ IO.10’ f 0.10’ f0.11’ ‘Value for G2-phase calculated from the M-phase a s follows: M.I. X Gi-phase duration M-phase duration = E 0.5 2Value for Gi-phase calculated frorn the S-phase a s follows: % S-phase Gi-phase duration = 9% S-phase X E . 7.4 S-phase duration proportion of cells in G2-phasewas calculated from both the S-phase and M-phase as shown in the footnotes to Table 1. The proportion of cells in the G1-phase was then determined. There were no significant differences in the kinetic parameters of al1 the regions of the intestinal tract studied. With our procedure for flow cytometry, the G peak of the DNA histograms routinely had a coefficient of variance (CV) in the 4-570 range. The kinetic parameters derived from analysis of the duodenal and colonic data with Frieds program were not significantly different (Table 2) from those derived with the micro- computer-based program (see Appendix). Therefore the jejunal and ileal data were only analyzed with the microcomputer-based program. As in the case of the radioautographic study there were no significant differences in the kinetic parameters of the four regions of the intestine. There were no significant differences between the results derived with flow cytometry and those derived with radioautography for any of the four regions. Table 3 shows the distribution of cells in four subcompartments within the S-phase as determined by the microcomputer-based program. In al1 four regions of the intestine there was a signifi- 256 H. CHENG AND M. BJERKNES TABLE 2. Whole population kinetic parameters of mouse intestinal epithelium determined by f i w cytometry G2 + M-phase 90 (X f SE.) - _ ~ _ _ S-phase % (X f S.E.) Duodenum 13.1 12.4 12.7 13.6 13.2 12.6 Jejunum Ileum Colon 4.1 3.9 3.6 3.6 4.5 4.2 f 0.94’ f 1.43’ I 1.02’ f 1.01’ f 0.78’ f 0.85’ G ‘-phase % (X f .S.E.) - ~- f0.23’ f 0.34’ f 0.38’ i 0.24’ f 0.50’ f 0.51’ 82.9 83.8 83.8 82.8 82.4 83.2 f 0.95’ f 1.20’ f 1.35’ f 0.91’ i 1.19’ f 1.32’ ’ Analyzed hy the Fried and Mande1 program. ’Andyzed by the microcomputer-haced program. TABLE 3. Uistribution of epithelial cells in the four subcompartrnents within the S-phase Duodenum Jejunum Ileum Colon Subcompartment 1 Subcompartment 2 Subcompartment 3 70 90 % (X I S.E.) (X f S.E.) (X f S.E.) 20.5 1 0 . 9 2 19.1 I 2.33 13.4 I 1.12 17.1 I 1.50 23.3 f 2.16 18.0 f 1.67 15.1 f 1.84 23.9 k 1.18 37.6 40.6 54.6 40.3 f 5.14 f 5.05 f 5.10 I 2.94 cantly larger proportion of cells in the first subcompartment than in the remaining three subcompartments of S-phase. Figures 3-10 show plots of the “average”DNA histograms determined from the mem DNA histograms studied for duodenum, jejunum, ileum, and colon. The proportion of the three main cell types, columnar, mucous, and Paneth, in the whole epithelial population is shown in Table 4. The proportion of labeled cells within each cell types is ais0 included in the table. The proportion of columnar cells was the Same in duodenum and jejunum but was significantly lower in ileum and colon. There was, however, no difference in the proportion of labeled cells in the columnar cell populations throughout the intestinai tract. The proportion of mucous cells was the same in duodenum and jejunum but was significantly higher in the ileum and even higher in the colon. However, no significant differences were found between regions in the proportion of labeled cells in the mucous cell population. The percentage of Paneth cells was significantly higher in ileum than in the duodenum and jejunum. In the duodenal epithelium there were large variations in the size of both villi and crypts. Some villi were about two to three times as wide as the average duodenal villus while some crypts were iiiuch longer than the average duo- 4 20 40 Subcompartment 4 90 (if S.E.) ~ 18.7 22.3 16.9 18.8 GO 8[! f 2.24 i 2.07 f 2.31 f 2.01 i I!? i .2 ’ CHRhlUFl NclVRFFi Fig. 3. DNA histogram of mouse duodenai epithelium. The data presented are the average derived from the means of five animals and are fitted by the microcomputer-based prugraixi. ~ 257 INTESTINAL EPITHELIAL CELL KINETICS 2500 u7 J J 2000 LL o U D cr 1 SU0 W u7 z 3 z i000 500 __L.li 20 GO 80 i00 i20 CHFlhNEL hdMBER Fig. 4. The data presented in Figure 3 were plotted on a different scale to show details of the S-phase and G * + M phase curves. Fig. 5. DNA histogram of mouse jejunal epithelium. The data presented are the average derived from the means of eight animals and are fitted by the microcomputer-based program. i50~ci 2500 m i 2 Cu i) 2003 LL O I ! 1 , i 508 i 00ii 100 CYcihhFI [U,JPQrCi Fig. 6 The data presented in Figure 5 were plotted on a different scale to show details of the S-phase and Gz+M phase curves. Fig. 7. DNA histogram. of mouse ileal epithelium. The data presented are the average derived from the means of eight animals and are fitted by the microcomputer-based program. denal crypts. This did not appear to be the case in jejunum and ileum. Both the viUi and crypts in those locations appeared to be of relatively constant size. These observations are reflected in the variation of the duodenal measurements shown in Table 5. The cel1 count studies for crypts and villi confirmed previous reports that the size of H. CHENG AND M. BJERKNES 2500 2000 1500 1000 500 20 40 GO 80 i00 1 2 ~ CHQNNCL VdWSF? Fig. 8. The data presented in Figure 7 were plotted on a different scale to show details of the S-phase and Gz+M phase curves. 200UU s, i 15OCO Fig. 10. The data presented in Figure 9 were plotted on a different scale to show details of the S-phase and G,+M phase curves. both villi and crypts decreases from the proximal to the distal region of the small intestine. Such a decreasing gradient was als0 observed in the number of crypts associated with a villus. The ratio of villus cells to crypt ceìls, however, was similar throughout the length of the small intestine. (Table 5 ) . Lu o DISCUSSION LL D 5000 Fig. 9. DNA histogram of mouse colonic epithelium. The data presented are the average derived kom the means of seven animals and are fitted by the microcomputer-based program. Accurate whole population epitheiial kinetic parameters are significant. Following perturbation (physiologicalor otherwise),a response observed in the crypt population may be reflected very differently at the level of the whole population. For example, if there was an increase in the size of the nonproliferative compartment (largely made up of the villus compartment) while the proliferative compartment remained constant, whole population measurements would detect the change while crypt measurements probably would not. Conversely, if there was a decrease in the proliferative population and a proportional decrease in the size of the nonproliferating population, no detectable change would be seen at the whole population level. However, at the crypt level, significant changes would be observed. So, while crypt measurements are important, whole population measurements are als0 necessary for a 259 INTESTINAL EPITHELIAL CELL KINETICS T A B L E 4. Proportion of the three main cell tyDes in mouse intestinal epithelium Columnar Labeling index (Ef S.E.) (X f S.E.) Whole population labeling index (i f S.E.) Duodenum Jejunum Ileum Colon 12.6 11.9 12.4 12.2 f 0.64 i 0.96 f 0.62 f 0.52 % 97.0 96.6 93.4 84.2 I 0.22 f 0.48 I 0.66* I 0.713 12.9 I 0.68 12.2 f 1.03 12.9 f 0.66 13.1 I 0.56 90 (Xf S.E.) Mucous Labeling index (X I S.E.) 2.5 i 0.13 2.9 f 0.49 5.5 I 0.53* 15.8 f 0.713 5.1 5.1 6.1 6.9 Paneth 70 (Xf S.E.) f 0.99 f 0.85 f0.89 10.52 0.5 f 0.12 0.6 10.18 1.1 i 0.20' O 'Significantly different from the duodenal population iP < 0.05). 'Significantly different from the duodenal and jejunal populations (P < 0.011. 'Significantly different from the duodenal. jejunal. and ileal populations (P i0.01) T A B L E 5. EDithelial cell counts of intestinal uilli and crvuts # cells per villus # cells per crypt # crypts per villus a (i I S.E.) b (X f S.E.) C (X f S.E.) 881 f 93.7 620 i 47.7' P < 0.05 403 f 47.g2 P < 0.02 10.2 i 0.43 8.5 i 0.34' P < 0.02 6.9 f 0.312 P < 0.01 ~~~~~~~~ Duodenum Jejunum Ileum ~~~ ~~~~~~~~ 12.638 f 2.369 6:918 I '8073 P < 0.02 3.775 f 3162 P < 0.02 ~ ~~ ratio of villus cells to crypt cells aibc ~~~ ~ 1.41 1.31 1.36 'Significantly different from duodenum and ileum. 2Significantly different from duodenum and jejunum 'Significantly different from ileum. true picture of the overall response of the intestinal epithelium. This paper is an attempt to derive accurate whole population epithelial kinetic parameters using isolated epithelium from various regions of the intestinal tract of normal adult mice. The parameters reported were derived with two different techniques, radioautography and flow cytometry. There were no significant differences between the two sets of data (Tables 1, 2). The correspondence of the two sets of data demonstrates that both techniques, radioautography and flow cytometry, can be used to derive reliable whole population kinetic parameters of intestinal epithelium. While flow cytometry is a more rapid procedure, radioautography does not require specialized equipment, which may not be available to some laboratories. Pipetting was used for the dissociation of cells in preparation for radioautography while the epithelium was dissociated by filtration for flow cytometry. The similarity of the vahes derived from radioautography and flow cytometry indicates that we were not selectively losing cells during the filtration-dissociation step in flow cytometry. Whole population kinetic parameters Kinetic parameters for the crypt population determined by radioautographic studies on tissue sections of mouse intestine give a labeling index of about 33yo for the crypt population in duodenum, 30% in jejunum, and 29% in ileum (derived from the data reported in Cheng and Leblond, 1974a).The ratio of villus cell number to crypt cell number was determined to be about 1.41 for duodenum, 1.31 for jejunum, and 1.36 for ileum (Table 5). Therefore the labeiing index for the whole epithelium can be estimated to be 13.7% (3311.41 1)for duodenum, 13% (3012.31) for jejunum, and 12.3% (29/2.36)for ileum. These numbers correspond wel1 with the direct measurements of labeiing index made in the present study (Tables 1, 2) and further indicate the reliability of the whole population kinetic parameters reported in this paper. As shown in Tables 1and 2 there were no significant differences in the kinetic parameters + 260 H. CHENG AND M. BJERKNES of the four regions studied. At first glance, this homogeneity in kinetic parameters might be surprising. I t is known that there is a clear reduction in villus size from the proximal to the distal portion of the small intestine to the extreme of no villus compartment in the colon (Altmann and Leblond, 1970). Intuitively, such an obvious decrement in the overall size of this substantial nonproliferating compartment should be reflected in regional variation in whole epithelial kinetic parameters. However, the reader must realize that there is a corresponding reduction ;n the size of the crypts and in the number of crypts per villus (Table 5; Fujii, 1972; Wright, 1980). Thus there is a simultaneous reduction in the size of both the proliferating and the nonproliferating compartments. As a result the ratios of villus cells to crypt cells were similar throughout the small intestine (Table 5; Wright, 1980).The kinetic parameters at the whole epithelial level were therefore not biased by variation in the size of the nonproliferating villus compartment. In the case of colon, one might expect a higher labeling index at the whole epithelial level than that in the small intestine because of the absence of the nonproliferating villus compartment. However, in the colon there is the surface epithelium, a nonproliferating compartment of significant size. Furthermore, the cell cycle time for the proliferating cells in colon is longer than that in the small intestine (Appleton et al., 1980). These factors were probably responsible for the value of the whole population labeling index observed in the colonic epithelium. When al1 of these factors are taken int0 account, our finding of a homogeneity in whole population kinetic parameters throughout the intestinal tract appears reasonable. Relatiue sizes of the crypt and uilluspopulations The ratio of the number of epithelial cells in the villus population to the number of cells in the crypt population was found to be similar along the length of the small intestinal epithelium (Table 5 ) . This finding is in agreement with the report of Wright (1980). The villusi crypt cell ratio can als0 be estimated algebraically from the labeling indices of the crypt population and of the whole epithelium (WLI, whole population labeling index; CLI, crypt labeling index; NLC, number of labeled cells; NVP, number of cells in villus population; NCP, number of cells in crypt population; using the equations WLI = NLCI(NCP NVP), and CLI = NLCINCP and then solving + for NVPINCP we derive the equation NVPI NCP = (CLI - WL1)IWLI).Using a crypt labeling index of about 3370, 30%. and 29% for duodenum, jejunum, and ileum, respectively, and the whole population labeling indices determined in this study (Table l),the ratio of villus cells to crypt cells was calculated to be 1.5, 1.4, and 1.2 for duodenum, jejunum, and ileum, respectively. These estimates correspond wel1 with the direct measurements made in this study (Table 5). Since the two sets of numbers were determined by two independent methods (one from the labeling indices, the other from actual cell counts) their similarity again indicates the reliability of the whole population kinetic parameters reported in this paper. Whole epithelial turnover time With conventional methods it is difficult to accurately determine the turnover time for the whole epithelium. Using the labeling indices determined in this paper (Tables 1, 2 ) , and a mean S-phase duration of 7.4 hours (ts)for intestinal epithelial cells (Schultze et al., 1972; Kovacs and Potten, 1973; Appleton et al., 1980), the turnover time (100 X t,ILI) for the whole epithelial cell population can be calculated. The results are displayed in Table 6. Al1 four regions had a whole epithelial turnover time of about 60 hours. By definition, turnover time is the time it takes to produce a number of cells equal to the total number of cells present in the population (Cheng and Leblond, 1974a).Therefore, whole epithelial turnover time determinations should include al1 cell types. Al1 cells were included in the determination of the 60-hour turnover time discussed above. However, the specific turnover characteristics of each of the cell types may vary. For example, the Paneth cellpopulation has a turnover time of about 3 weeks (Bjerknes and Cheng, 1981b) while the columnar cells have a turnover time of about 58 hours (Table 6). Whole epithelial growth fraction The whole epithelial growth fraction may als0 be approximated from our results by the equation GF = LII(100 X t4tc).With a small intestinal cell cycle time (tc)of about 13 hours (Al-Dewachiet al., 1975; Schultze et al., 1979), and a whole population labeling index (LI) of about 13%(Tables 1 , 2 ) ,thegrowth fraction for the small intestinal epithelium can be estimated to be about 0.23. In other words, in the small intestinal epithelium, about 23% of the 26 1 INTESTINAL EPITHELIAL CELL KINETICS TABLE 6. Whole population, columnar cell. and mucous cell turnouer characteristics Duodenum (k t S.E.) Whole population turnover time' Columnar cell turnover time' Mucous cell turnover time'.2 Jejunum _ _ (X f S.E.) Ileum - (i f S.E.) -. 59.7 k 3.01 63.6 f 4.30 60.9 I 2.94 62.0 ? 2.73 58.2 f 3.02 62.1 f 4.34 58.4 I 2.85 58.0 t 2.55 181.3 -+ 26.3 191.5 f 45.1 143.5 I 21.5 124.0 I 16.4 'Turnover time for each population is calculated as follows: x 100 where tS is the duration of S-phase 17.4 hoursi and L.I. is the labeling index of the cel1 population 'This is not the true mucous cell turnover time because columnar cells contribute heavily to the mucus cell h e . cells are proliferating. The growth fraction in the colon is higher. With a mean tc of about 18.5 hours (Appletonet al., 1980)and a labeling index of about 12.5% (Tables 1, 2), the colonic growth fraction is 0.31. Thus in the colon, about 31% of al1 epithelial cells are proliferating. Mucous cell renewal I t has been demonstrated that in the crypt, both the columnar and the mucous cell populations turn over at about the same rate (Merzel and Leblond, 1969; Cheng, 1974; Cheng and Leblond, 1974b; Chang and Nadler, 1975). Cheng (1974)as well as Chang and Nadler (1975) further demonstrated that in the crypt, the rate of proliferation of mucous cells is too low to account totally for mucous cell renewal. I t was concluded that the stem cell and thus the columnar cell population contribute to the renewal of the mucous cell population (Cheng, 1974; Cheng and Leblond, 1974b).Our present data confirm these previous findings. If we assume that the mucous cell population receives no external contribution to its numbers, then using the columnar cell S-phase duration (according to Thrasher and Greulich, 1966, mucous and columnar cells have the same S-phase duration), and the labeling index of mucous cells (Table 4), the turnover time of the mucous cell population could be calculated (this value is not the true mucous cell population turnover time because, as is discussed below, columnar cells contribute heavily to the mucous cell line).These values were included in Table 6. This calculated turnover time was much longer than the turnover time of the columnar cell population. Since the mucous cell population is actually renewing at about the same rate as the columnar cell population we confirmed our previous finding (Cheng, 1974; Cheng and Leblond, 1974b) that the mucous cell population did not produce cells at a rate that will explain their rate of renewal. In order for the columnar and mucous cell populations to be renewed at about the same rate, cells other than mucous cells must contribute to the renewal of the mucous cell population. If the turnover time of the columnar cell line is calculated (Table 6), it may be seen that they are renewing at a slightly higher rate than the whole population (the difference is not significant; however, since the mucous cell population is much smaller than the columnar cell population it will only need a very smal1portion of the columnar cells to meet the need of the mucous cell renewal rate). Thus we confirm that the columnar cell population is participating in mucous cell renewal. I t is possible to calculate the proportion of the mucous cell population derived from mucous cell divisions (as opposed to those derived from columnar cells). Following the rationale outlined in Cheng (1974), given that the S-phase duration and rate of renewal are similar for both columnar and mucous cells, if the mucous cell population alone was responsible for its renewal then the frequency of cells in either line (columnar or mucous) should be proportional to their corresponding fraction within the proliferating compartment of the epithelium. For example, the percentage of mucous cells in the epithelium should be proportional to the percentage of the labeled epithelial cells that contain mucus (i.e.,if there are 100 labeled epithelial cells and ten of them contain mucus, then the percentage of the labeled epithelial cells that contain mucus will be 10% and this should be equal to the percentage of mucous cells in the epithelium). Similarly, the labeling 262 H. CHENG AND M. BJERKNES indices of the two cell lines (columnar and mucous) should be the same. On the other hand, if the columnar cell line were contributing to the mucous cell population, then the percentage of mucous cells would be higher than the percentage of labeled epithelial cells with mucus, and the labeling index of the mucous cell population would be lower than that of the columnar cell line. The latter case was observed (Tables 4,7).The ratios (labelingindex of mucous cellsi labeling index of columnar cells, percentage labeled cells with mucuslpercentage epithelial cells with mucus) should indicate the relative contribution of mucous cell divisions to the mucous cell population. The remainder would come from the columnar cell line. These number are shown in Table 7. Thus, only about half of the small intestinal mucous cell line is derived from mucous cell divisions. This is in agreement with our previous determination for the crypt mucous cell population using conventional radioautographic studies on tissue sections (Cheng, 1974). In the colonic epithelium approximately two-thirds of the mucous cells were derived from mucous cell divisions (Table 7). Thus, the mucous cell line in the colon is more actively proliferating than that in the small intestine. This is in agreement with the observation of Chang and Nadler (1975)for the colonic crypt mucous cell population using conventional radioautographic studies on tissue sections. These results again demonstrate the agreement of the findings derived from kinetic parameters of the crypt population (determined by conventional radioautographic studies on tissue sections) with those derived from kinetic parameters of the whole epithelium (determined by flow cytometry and radioautography on dried cell suspension). Therefore we are confident that the whole population kinetic parameters, and thus, the techniques reported in this paper are reliable. In conclusion, we would like to stress the potential utility of the techniques presented in this paper for the determination of whole epithelial kinetic parameters. They should be useful for rapid assessment of the effects of an experimental or physiological manipulation on cell proliferation in the intestinal epithelium. APPENDIX Multigaussian curve fitting (such as the algorithms developed by Fried and his collegues; Fried, 1976;Fried and Mandel, 1979)has proved to be a sensitive method for the analysis of DNA histograms from flow cytometric studies. However, the usual methods for deriving these fits require the solution of large sets of simultaneous linear equations. Because of this, these programs require the use of large computer systems. This has meant that most users could not have analyses computed “on the spot” but rather had to transmit the data to a large mainframe computer and then get the results printed at some remote location. To get around this difficulty some effort has been expended in the development of algorithms that would allow the analysis to be performed locally on a microcomputer. The most successful of these has been the polynomial fitting routines of Dean and collegues (Dean and Jett, 1974; Dean, 1980). However, we have found that the polynomial fitting algorithms do not provide satisfactory results with our tissue. We report here a microcomputer-compatible algorithm we have developed which allows the fitting of multigaussian curves to flow cytometric data. The program fits as many curves as there are channels between the G, and the G2 peaks. The program runs on our system in TABLE 7. PercentaEe of mucous cell population deriued from mucous cell diuisions % labeled cells with mucus 1 hour after 3H-thymidine % epithelial cells with mucus % of mucous cell population contributed by proliferative mucous cells Duodenum (X f S.E.) (X I S E . ) 1.1 i 0.16 1.2 f 0.27 2.9 i 0.55 10.4 f 1.12’ 2.5 f 0.13 2.9 f 0.49 5.5 f 0.53 15.8 f 0.71 45.6 I 6.523 44.7 i 6.68‘ 42.3 f 6.453 41.4 I 6 . 4 4 ‘ 53.0 f 7.923 52.1 f 7.454 ‘Significantly different P < 0.01) from duodenum. jejunum. and ileum ’Significantly different ( P < 0.05) from duodenum and jejunum. 2ca,culated as Yo labeled cells with mucus x 100. YOepithelial ceik with mucous 4C.culated as labeling index of mucous cells x 100. labeling inaex of columnar cells Jejunum Ileum Colon (X I_ S E ._ ) - ~ _ _ (X _i S.E.) ’ 63.6 i 4.46’.3 60.2 I 4.554 INTESTINAL EPITHELIAL CELL KINETICS about 90 seconds on the average (our system is composed of the 8-bit microprocessor made by Zilog Corp., the Z-80A CPU, 64K bytes of RAM and an Intel C8231A arithmetic processor chip, which doubles the speed of most arithmetic computations). The program is written in PLlI (Digital Research Corp.) and requires about 50K bytes to run. The algorithm is a twopart fitting procedure. First, an initial estimate of the population is made using the quadratic polynomial algorithm of Dean (1980).The coefficient of variance is estimated by using the method outlined by Fried and Mandel (1979). This initial estimate is then transformed into a multigaussian approximation by using the result obtained from the polynomial fit for each channel as the area that must be found under the gaussian curve that will represent the data of that channel (each of the curves is assumed to have the same coefficient of variance). After this approximation is obtained, the program proceeds to the second step. In the second step, the program searches for the point of maximum deviation of the data from the fit. The area under the gaussian curve whose mean is at the channel of maximum deviation is then adjusted in a direction that reduces the chi square of the fit. After that channel has been adjusted to the optimum point, the program proceeds until each channel has been adjusted. Then the program starts over again, looping back through the second stage of the program until adjustments no longer make significant changes in the fit. In most cases it requires four loops to fit the data. At this stage the fitting is completed and all that remains is the calculation of the S-phase subcompartments. With a coefficient of variance of 4-5%, we are not justified in dividing the S-phase into more than about four subcompartments. For this reason, the Gaussian curves that represent the S-phase component of the cell cycle are placed into four groups. For example, if the G , peak was centered on channel 41, and the GZ+ M peak was centered on channel 82, then the 40 curves that represented S-phase cells would be subdivided into four groups. Thus, the proportion of cells in the first subcompartment of S-phase would be calculated from the area under the Gaussian curves centered on channels 42-51, the next compartment would correspond to the curves of channels 52-61, and so forth. This would result in S-phase being broken down into four steps, corresponding to early, early-mid, latemid, and late S-phase. These subcompartments are comparable and may be used to 263 make comparisons of the relative proportion of cells in early, mid, or late S-phase. ACKNOWLEDGMENTS We thank R. Kuk for technical assistance. Supported by the Canadian Foundation for Ileitis and Colitis and the Medical Research Council of Canada. LITERATURE CITED Al-Dewachi, H.S., N.A. Wright, D.R. Appleton, and A.J. Watson (1975) Cell population kinetics in the mouse jejunal crypt. Virchows Arch. [Cell Pathol.], 18:225-242. Altmann, G.G., and M. Enesco (1967)Cell number as a measure of distribution and renewal of epithelial cells in the small intestine of growing and adult rats. Am. J. Anat., 121:319-336. Altmann, G.G., and C.P. Leblond (1970)Factors influencing villus size in the small intestine of adult rats a s revealed by transposition of intestinal segments. Am. J. Anat., 127;15-36. Appleton, D.R., J.P. Sunter, M.S.B. de Rodriguez, and A.J. Watson (1980)Cell proliferation in the mouse large bowel, with details of the analysis of the experimental data. In: Cell Proliferation in the Gastrointestinal Tract. D.R. Appleton, J.P. Sunter, and A.J. Watson, eds. Pitman Medical, Tunhridge Wells, Kent, England, pp. 40-53. Bertalanffy, F.D. (1960) Mitotic rates and renewal times of the digestive tract epithelia in the rat. Acta Anat., 40:130-148. Bjerknes, M., and H. Cheng (1981a) Methods for the isolation of intact epithelium from the mouse intestine. Anat. Rw., 1!?9:565-574. Bjerknes, M., and H. Cheng(1981b)Thestem-cellzoneof the small intestinal epithelium. I. Evidence from Paneth cells in the adult mouse. Am. J. Anat., 160:51-63. Chang, W.W.L., and N.J. Nadler (1975) Renewal of the epithelium in the descending colon of the mouse. IV. Cell p o p ulation kinetics of vacuolated-columnar and mucous cells. Am. J. Anat., 144:39-56. Cheng, H. (1974) Origin, differentiation and renewal of the four main epithelial cell types in the mouse small intestine. 11. Mucous cells. Am. J. Anat., 141:481-502. Cheng, H., and C.P. Leblond (1974a)Origin, differentiation and renewal of the four main epithelial cell types in the mouse small intestine. I. Columnar cell. Am. J. Anat., 141:461-480. Cheng, H., and C.P. Lehlond (1974b)Origin, differentiation and renewal of the four main epithelial cell types in the mouse small intestine. V. Unitarian theory of the origin of the four epithelial cell types. Am. J. Anat., 141:537-562. Dean, P.N. (1980)A simplified method of DNA distribution analysis. Cell Tissue Kinet., 13299-308. Dean, P.N., and J.H. J e t t (1974) Mathematical analysis of DNA distributions derived from flow microfluorometry. J. Cell Biol., 60;523-527. Fried, J. (1976) Method for the quantitative evaluation of data from flow microfluorometry. Comp. Biomed. Res., 8263-276. Fried, J., and M. Mandel(1979)Multi-user system for analysis of data from flow cytometry. Computer Programs in Biomedicine, 10218-230. Fujii, R. (1972)Quantitationof thenumber ofvilli andcrypts in the intestine of rodent animals. Experientia, 28;1209-1210. Kopriwa, B.M., and C.P. Leblond (1962)Improvement in the coating technique of radioautography. J. Histochem. Cytochem., 10:269-284. 264 H. CHENG AND M. BJEHKNES Kovacs, L., and C.S. Potten (1973)An estimation of proliferative population size in stomach, jejunum, and colon of DBA-2 mice. Cel1 Tissue Kinet., 6:125-134. Leblond, C.P., and C.E. Stevens (1948)The constant renewal of the intestinal epithelium in the albino rat. Anat. Rec., 100..357-371. Merzel, J., and C.P. Leblond (1969) Origin and renewal of goblet cells in the epithelium of the rnouse smal1 intestine. Am. J. Anat., 124:281-306. Schultze, B., V. Haack, A.C. Schmeer, and W. Maurer (1972) Autoradiographic investigation on the cel1 kinetics of crypt epithelia in the jejunum of the mouse. Cell Tissue Kinet., 5:131-145. Schultze, B.. A.M. Kellerer, and W. Maurer (1979) Transit times through the cycle phases of jejunal crypt cells of the mouse. Analysis in terms of the mean values and the variances. Cel1 Tissue Kinet.. 12:347-359. Thrasher, J.D., and R.C. Greulich (1966)The duodenal progenitor population. 11. The progenitor cel1 cycle of principal. goblet and Paneth cells. J. Exp. Zool., 161.9-20. Wright, N.A. (1980) Cel1 proliferation in the normal gastrointestinal tract. Implications for proliferative responses. In: Cell proliferation in the gastrointestinal tract, D.R. Appleton, J.P. Sunter, and A.J. Watson, eds. Pitman Medical, England, pp. 3-25.