Comp. Biochem. Physiol. Vol. 86A.No.
3, PP. 423428, 1987
0300-9629/X7 $3.00+ 0.00 0 1987PergamonJournals Ltd
Printed in Great Britain
EPITHELIAL AND PILLAR CELL REPLACEMENT IN GILLS OF JUVENILE TROUT, SALMO GAIRDNERI RICHARDSON W. G. E. ZENKER,* H. W. FERGU~ON,*$I. K. BARKER*and B. WCI~DWARD~ *Department of Pathology and TDepartment of Nutrition, University of Guelph, Guelph, Ontario, Canada NIG 2W1 (Received 6 May 1986)
Young rainbow trout, Salmo gairdneri (Richardson) were injected intraperitoneally with tritiated thymidine, and killed at intervals between 2 hr and 16 days after inoculation. 2. Labelled epithelial cells were first detected autoradiographically along the base of gill lamellae. Epithehal cells proliferated here and then migrated toward the tips of the lamellae. Uniform labelhng along the length of the filaments at the base of lamellae indicated that cells were dividing at a constant rate. 3. Transverse sections of filaments showed that epithelial proliferation was also uniform across the base of the lamellae. 4. The interior of the lamellae often had labelled pillar cells, indicating that these cells also divide. The high intensity of the label in animals killed 16 days after inoculation with tritiated thymidine suggests that division probably occurs slowly, less than once every 16 days,
Abstract-l.
INTRODUCrION The general structure of teleostean gills is documented at both the light microscope and ultrastructural levels (Hughes and Grimstone, 1965; Morgan and Tovell, 1973; Hughes and Mittal, 1980; Laurent and Dunel, 1980; Karlsson, 1983; Laurent, 1984), but there have been only a few attempts to determine the origin and rate of turnover of gill epithelium. Only by understanding normal branchial epithelial kinetics will it be possible to investigate the pathogenesis of gill disease, the most serious production problem of farmed fish in southern Ontario (Daoust and Ferguson , 1983), and a serious disease problem in fish worldwide (Snieszko, 1981). Conte and Lin (1967) undertook some preliminary kinetic studies on gill epithelium in conjuction with their work on osmoregulation in coho salmon, Onto rhynchus kisutch (Walbaum) and chinook salmon, 0. tshawytscha (Walbaum). These authors identified the major site of cellular proliferation to be at the base of the lamellae, along the filaments. By contrast, Mackinnon and Enesco (1980) reported the site of proliferation of the epithelial cells in the gills of rosy barbs, But&us conchonius (Hamilton-Buchanan) to be along the base of the filaments, close to the gill arch, with subsequent migration of cells out along the filaments and over the lamellae. As part of an ultrastructural study of gill development in embryonic rainbow trout, Salmo gairdneri (Richardson), Morgan, (1974) also made observations on the proliferation of pillar cells. Based on the presence of mitotic figures, she suggested that new pillar cells were formed by division of a proximal row of mesen-
chymal cells, located close to the main body of the filament. The purpose of the present study was to determine, using tritiated thymidine, the location of epithelial proliferation and the transit time of epithelial cells on the gills of young rainbow trout. Observations were also made on the proliferation of pillar cells in lamellae.
MATERIALSAND
METHODS
Experimental animals
Rainbow trout, 2.4 (SD = l.O)g in weight and 5.8 (SD = 0.5) cm long, were obtained from a commercial trout farm free of the specific pathogens notifiable in Canada. The fish were maintained in aerated 40 1 tubs with a continuous through-flow system using dechlorinated water at a flow rate of 400 ml/min at 9.8 +_O.S”C. The photoperiod was 12 hr light and 12 hr dark, supplied by fluorescent lights in a windowless laboratory. The fish were fed (Salmonid Grower Ration, Martin Feed Mills. Elmira. Ontario) at a rate of 2% of body wt once daily. Radioisotopic labelling
Fish were netted, wrapped in the net, and without use of anaesthetic, given an intra-peritoneal injection of 0.1 ml of a sterile aqueous solution of tritiated thymidine [3H-TdR] (0.1 mCi/ml, sp. act. 20ci/mmol, New England Nuclear Products, North Billerica, MA, USA.) using a 27 gauge S/8 inch needle. To minimize leakage, injections were made through the dorsal axial musculature, just lateral to the dorsal fin, rather than through the thinner abdominal wall. Digital pressure was applied briefly at the injection site to occlude it. Injected fish were then placed in a separate tank. Experimental design
Fish were arbitrarily assigned to a group of 20 fish to be sampled at 2 hr or to six groups of ten fish to be killed I, 2, 4, 8, 12 and 16 days after injection. The fish were killed by severing the spine with a scalpel behind the head.
$To whom correspondence
should be addressed: Department of Pathology, Ontario Veterinary College, University of Guelph, Guelph, Ontario, Canada NlG 2Wl. 423
424
W. G. E. ZENKERet al.
Histological methods Immediately after the fish were killed their opercula were removed and the animals were immersed in Bouin’s solution for 24 hr at room temperature. They were then placed in 70% ethanol to await further processing. The second and third gill arches from both sides of each fish were removed. Each gill arch was placed in a separate identified tissue processing cassette. Gill arches from the right side were laid flat so as to give longitudinal sections of the arch and filaments, and transverse sections of the lamellae. Those from the left side were placed vertical to the plane of section, thus providing a transverse view of the filaments. Specimens were routinely processed for paraffin embedding and 5 pm sections were cut and stained with haematoxylin and eosin. At least three slides, with three or four sections of tissue per slide, were taken from each of the four blocks per fish. Autoradiograph) All slides were coated to Kodak NTB2 liquid nuclear track emulsion (Eastman Kodak Company, Rochester, NY, USA) using a modification of the method of Kopriwa and Leblond (1962). Modifications involved the use of complete darkness instead of the Wratten No. 2 filter; the dilution of the emulsion I : 1 with distilled water; and the use of clean slides without re-cleaning and pre-coating with celloidin. To monitor background exposure, one control slide without tissue was processed for every ten tissue slides. Slides were allowed to air dry and placed together with small sachets of silica gel crystals, into light-proof plastic slide boxes which were wrapped with a double layer of aluminium foil. The slides were exposed at 4’C for 3G-96 days. Exposed slides were developed with Dectol (Eastman Kodak Company, Rochester, NY, USA) at a 1: 1 dilution for 2 min at 17~18°C. They were then rinsed in distilled water, fixed for 5 min with Kodak general purpose hardening fixer (Eastman Kodak Company, Rochester, NY, USA) and washed in running tap water for about 15 min. The slides were allowed to air dry overnight and then cover slips were mounted with Diatex (Scientific Products, McGaw Park, IL, USA). The dense clusters of silver grains over the nuclei of gill epithelium which had taken up the [lH]TdR were readily distinguishable under the microscope. Only nuclei marked with at least six silver grains were judged to be labelled. Measurements All slides were scanned at 400 x magnification seeking labelled nuclei. Slides were deemed to be positive if a cell in any of the sections on the slide was labelled. While at least ten fish were sampled at each time, there were not ten fish with positive slides in each group, presumably due to the small numbers of mitotic cells and variation in uptake of label. The distance from the base of a lamella to each labelled cell was measured in micrometers km] using an eye piece graticule calibrated with an objective micrometer. Since the nuclei of epithelial cells tend to orient themselves over pillar cells (Morgan and Tovell, 1973) these measurements were also expressed in terms of a “pillar cell unit”; one mean pillar cell unit equals the mean length of the 2347 lamellae seen to have labelled cells, divided by the mean number of pillar cells in those lame!lae and thus reflects both the width of the pillar cell and the associated blood sinus. The mean lamellar length was 79.0 pm (SD = 22.1) with 6.3(SD = 1.9) pillar cell units per lamella; thus the mean pillar cell unit was 12.8 pm (SD = 2.1). The speed of migration of the epithelial cells along the lamellae was estimated using a ranking method to determine the position with the greatest number of labelled cells, and then estimating the time required for cells to arrive at these positions. The 2-hr sample was not utilized to estimate rate because less than half of the positive
slides had labelled cells on lamellae. If two or more loci were not markedly different then the mean value was used (Table I. 2-day sample). Slalislical analysis The results were analyzed using a Chi-square distribution, at a level of significance of P < 0.05, to determine whether the numbers of marked cells at the various positions along the lamella varied significantly with time.
RESULTS In the samples taken at 2 hr and 1 day after injecting the [3H]TdR, if labelled cells were present on sections of gill, they were always at the base of the lamellae (Fig. 1; Table 1). These cells were usually one cell layer below the surface and sometimes adjacent to the cartilagenous ray of the filament, at the interlamellar space. This arrangement occurred along the full length of the filaments. The amount of label appeared to be uniform, whether the lamellae were located close to the base of the filament, in the mid-section of the filament or toward its distal end. Usually only a single cell was labelled; rarely was label present over two adjacent cells. Labelled cells were large and round (2&30 pm), with a prominent, round nucleus. On transverse sections of filaments there were usually four or five labelled cells uniformly distributed across the base of lamellae (Fig. 2). The interfilamental space adjacent to the gill arch had few, if any, labelled cells. As time progressed, while all positive gills still had labelled cells at the base of some of the lamellae, labelled cells were also present further up the lamellae and fewer grains were seen per labelled nucleus. Even as early as 2 hr post-injection a few labelled cells were present on the proximal portion of the lamellae (pillar cell units 1-3, Table I), and not just in the interlamellar space. With time, however, more cells were found further up the lamellae, in a higher proportion of slides, while cells at the base still retained a faint label. There was a significant tendency for labelled cells to be located more distally on the lamellae with time (Table 1); however, the prevalence of labelled cells dropped with increasing distance from the lamellar base. Table 1 depicts this phenomenon. In addition the number of slides with label on lamellae (at pillar cell unit 1 and beyond) increased with time. The results of the ranking method to identify with the highest concentration of labelled cells are sumarized in Table 2. The mean migratory rate of epithelial cells using the I-. 2-, 4- and 8-day samples was 12.8 (SD = 0.3) pm/day. Pillar cells were also labelled (Fig. 3). This was more obvious at the later time periods because the less heavily labelled epithelial cells did not obscure the labelled pillar cells. No measurements were taken, but the pattern appeared similar at all sampling times, with the pillar cells remaining heavily labelled and distal pillar cells often being labelled at early sample times. Only rarely were two adjacent pillar cells labelled, or was the intensity of label reduced, indicating that the majority of labelled pillar cells did not divide after the initial uptake of [3H]TdR. Goblet cells were not labelled, and because the developer tended to make the tissues eosinophilic, it
Epithelial and pillar cell replacement
425
Fig. I. Gill from trout killed (2 hr) after injection of [3H]TdR. Labelled cells are present in the progenitor compartment of the inter-lamellar spaces.
was also difficult to recognize chloride cells or the other cellular constituents of gill filaments. An occasional cartilage cell in the filament was labelled, and these, like the pillar cells, tended to retain a heavy label long after most label in epithelial cells had become diluted. The tip of the filaments appeared to have increased numbers of labelled cells which resembled those found in the interlamellar spaces but there were not enough samples with the filament tip visible to assess this quantitatively.
referred to this mechanism of epithelial cell replacement in gills, but gave no reference to support the concept. In contrast, MacKinnon and Enesco (1980) using rosy barbs, obtained significantly different results in which epithelial cells migrated over the gill lamellae from a proliferative compartment at the base of the filaments close to the gill arches. In our study, few labelled cells were seen at the base of the filaments. It remains to be determined if these differences are the result of small numbers of fish examined by Mackinnon and Enesco or if there is a real variation as a result of differences in species, age, size or water temperature. In the present study the uniform intensity of cellular labelling at the base of lamellae along the entire length of the filament indicates that these cells were dividing at a similar rate and is not consistent with the concept that cells originate near the gill arch and migrate toward the tip of the filaments. The transverse sections also showed that there was uniformity
DISCUSSION Lamellar epithelial cells proliferated at the base of the lamellae and with time labelled cells were present on the lamellae in increased numbers and at progressively greater distances from the apparent progenitor compartment. These findings resemble those of Conte and Lin (1967) who found, in juvenile coho and chinook salmon, that epithelial cells proliferated at the base of the lamellae. Tovell et al. (1970)
Table I. Prevalence of epithelial cells with labelled nuclei at various pillar cell loci along gill lam&e at different times after inoculation of tritiated thymidine. Expressed as the percentage of labelled slides with labelled nuclei at the locus in question Pillar cell unit Time _~ 2 hr I day 2 days 4 days 8 days 12 days I6 days
n
0
32 35 37 37 I6 II 25
100 100 100 100 100 100 100
I
2
21.9 9.4 48.6 22.9 37.8 40.5 35.1 24.3 68.8 50.0 18.2 36.4 42.9 48.6
3
4
5
6
7
8
9
9.4 11.4 43.2 45.9 50.0 54.5 51.4
0.0 14.3 29.7 29.7 31.3 36.4 48.6
0.0 0.0 27.0 29.7 56.3 45.5 60.0
0.0 0.0 21.6 27.0 37.5 27.3 34.3
3.1 0.0 10.8 24.3 37.5 45.5 31.4
0.0 0.0 5.4 13.5 37.5 36.4 25.7
3.1 0.0 5.4 13.5 12.5 27.3 14.3
lo+ 0.0 0.0 0.0 0.0 6.3 0.0 11.4
*n refers to the number of slides with label at the respective loci. tLocus 0 refers to the base of the lamella. All positive lamellae were labelled at the base.
W. G. E. ZENKER et al.
5g. 2. Cross
section
from trout
killed 2 hr after injection of [2H]TdR across width of lamella.
showing
labelled
epithelial
Table 2. Pillar cell unit with maximum label at various times after injection of tritiated tbymidine. Distance from the base is used to calculate migratory rate of labelled epitbelial cells Time
Locus with maximum labelling’
I day 2 days 4 days 8 days
I 2 3 5
Distance from base (pm)
Migration rate (um/day) 12.9 12.9 26.4 13.2 37.8 12.6 62.0 12.4 Mean Migratory Rate 12.8 (SD = 0.3)
*Expressed as pillar cells units.
Fig. 3. Gill from trout
killed after injection
of [jH]TdR
showing
levelled pillar cell (arrow)
cells
Epithelial and pillar cell replacement
of labelling across the base of the lamellae. From this we conclude that the epithelial progenitor compartment is uniformly distributed at the base of lamellae along the full length of the filaments. In the sense that the proliferative compartment is at the base of lamellae and epithelial migration from base to tip of lamellae occurs, the phenomenon is analogous to epithelial cell replacement occurring in mammalian small intestine (LeBlond and Stevens, 1948; Williamson, 1978). The finding of labelled epithelial cells in the lamellae shortly (2 hr) after the injection of [3HJTdR indicates that the cells can leave the progenitor compartment during or shortly after the S phase (Quastler and Sherman, 1959) or else that cells can divide at the base of the lamellae. In the intestine the “decision” to differentiate and migrate is usually made shortly after mitosis, and is dependent on the presence or absence of an inhibitory substance called a “chalone” (Bullough, 1975) the presence of the chalone being inhibitory for proliferation (Williamson, 1978; Moon, 1983). There is a short time lapse between mitosis and movement into the functional compartment in the gut (Quastler and Sherman, 1959). In contrast to the findings of Conte and Lin (1967) we did not find a time lapse between labelling and movement of cells into lamellae. It is not known if the epithelial cells migrating onto the surface of lamellae retain the capacity for mitosis, but contrary to the suggestions of Laurent, (1984), in our experiments this did not appear to be the case since labelled cells were largely limited to the base of lamellae up to pillar cell unit 3 (Table 1, 2 hr). The short ( < 2 hr) time lag between labelling and movement onto the lamellae (Pillar Cell Unit I) in the present study may be att~buted to the species studied and the age and size differences in our work in contrast to that of Conte and Lin. it is also unclear whether Conte and Lin used an anaesthetic when they gave an intracardiac injection of thymidine; they used MS-222 when they gave an intraperitoneal injection to their fish. Preliminary work at this institution seemed to indicate that MS-222 may inhibit mitosis on gills (Daoust, personal communication). Lamellae of teleost fish are roughly triangular in shape (Laurent, 1984). The villi of the mammalian small intestine also taper at their distal end and as a consequence the migratory rate increases as the epithelial cell approaches the tip of the villus (Clark, 1970). If we compare the rate of migration in these two morphologically similar systems it follows that we are less likely to see labelled cells at the tip than at the base because the cells would be moving faster as they approach the tip. In addition, there is the possible complicating factor of exfoliation. Moreover. while some lamellae were as long as 10 pillar cell units, the prevalence of labelled cells on distal parts of lamellae is artificially reduced because values were expressed as per cent prevalence of all labelled slides when in fact many lamellae did not extend beyond 6 or 7 pillar cell units in length (mean 6.3, SD = 1.9). These three factors probably combined to make it difficult to visualize labelled cells toward the tip of the lamellae and may explain the small number of slides with labeiled cells in this location.
421
The migration of the epithelial cells from the progenitor compartment at a rate of about 10.9 pm/day, resulted in a calculated average transit time from lamellar base to tip of 7.9 days. This contrasts with a transit time varying between 1.0 to 5.0 days from crypt to tip of villus in mammalian small intestine (Clark, 1973; Moon, 1983; Eastwood, 1977). The present observations suggest a second wave of labelled nuclei appearing about day 8 (Table 1, pillar cell unit 1). On days 12 and 16 there were no discernable patterns to the labelling. This is consistent with results on the proliferative cycle in gut where, as a result of desynchronization of the cells and a decline in the size of successive cycles, the biological system did not always follow the theoretical curve (Eastwood, 1977; Lipkin, 198 1). Ultrastructural studies often describe a double epithelial layer on the lamellae of teleostean gills (Hughes and Grimstone, 1965; Morgan and Tovell, 1973; Karlsson, 1983), but it was not possible in our light microscopic study to recognize two layers. Therefore it is not clear whether only one, or both, of these epithelial layers is migrating, or if the rate of migration of the two layers is different. The paucity of labelled cells in lamellae 2 hr after inoculation suggests that virtually no epithelial celi division takes place there. Hence if the inner layer replaces the outer as suggested by Laurent (1984), it probably does so without division. Labelled pillar cells were present throughout lamellae, from base to tip, indicating that these cells too were dividing. The fact that few adjacent pillar cells were labelled, and that they retained a heavy label even after 16 days, indicates that the rate of mitosis was low. A very small increase in numbers of pillar cells could easily “support” a significantly larger number of epithelial cells. One pillar cell in section is associated with x epithelial cells, therefore in two dimensions it would be associated with an area counted by x2 epithelial cells. Since there are two epithelial surfaces associated with one pillar cell, pillar cell numbers would be roughly proportional to twice the square of the number of epithelial cells (i.e. 2 x2). The rate of division of pillar cells therefore might be relatively slow, proportional to 1/2x2 of the rate of division of the epithelial cells. This may explain why other investigators have not found many mitotic figures in pillar cells. The fact that Tovell et al. (1970) found dividing pillar cells only in the mesenchymal layer, at the base of the lamellae, close to the basement membrane, may be explained by the fact that they were studying early post-embryonic growth and development of the lamellae. a period of rapid proliferation and differentiation of tissues. Kendall and Dale (1979) found unusually closely adjacent pillar cells which they classified as Type I and Type II pillar cells. They had no explanation for these two types of pillar cells, but our findings suggest they could have been the result of recent cell division. Erratic labelling of the fish by 13H]TdR was seen. In each group of fish an average of only 60% were identifiable as labelled. This can be ascribed to leakage of the label injected, despite the application of digital pressure, and injection through the thicker dorsal musculature. Leakage through the abdominal
W. G. E. ZENKER et al.
428
pores found in some salmonids (George et al. 1982) may also have occurred. These pores are believed to be involved in elimination of foreign materials from the abdominal cavity. Future experiments must allow for these factors. The present study indicates that epithelial replacement on the gills of trout is somewhat analogous to that found in small intestine, epithelial cell dividing at the base of lamellae, and migrating toward the top, with little or no evidence of division. More work must be done to ascertain how the epithelial transit time varies with fish species, age, environmental temperature or with disease conditions, and to determine epithelial generation time and the time cells spend in each of the compartments. Acknowledgements-This work was supported by the Ontario Ministry of Agriculture and Food. Dr. Hicks is thanked for critically reading the manuscript and helpful discussion. Technical assistance by Mr. Michael BakerPearce is also gratefully acknowledged. REFERENCES
Bullough W. S. (1975) Mitotic control in adult mammalian tissues. Biol Rec. 50, 99-127. Clark R. M. (1970) Cell migration in the epithelium of the small intestine. J. Anal. Land. 106, 175.. Clark R. M. (1973) Progress in measuring epithelial turnover in the villus of the small intestine. Digestion 8, 161-175. Conte F. P. and Lin D. H. Y. (1967) Kinetics of cellular morphogenesis in gill epithelium during sea water adaptation of Oncorhynchus (Walbaum). Comp. Biochem. Ph_vsiol. 23, 945-957. Daoust P.-Y. and Fereuson H. W. (19831 Gill diseases of cultured salmonids m Ontario. C&t. J.’ camp. Med. 47, 358-362. Eastwood G. L. (1977). Gastrointestinal epitheliai renewal Gastroenterology 72, 962-975. George C. J.. Ellis A. E. and Bruno D. W. (1982) On rememberance of the abdominal pores in rainbow trout, Salmo Gairdneri Richardson, and some other salmonid spp. J. Fish Biol. 21, 643-647. Hughes G. M. and Grimstone A. V. (1965) The fine structure of the secondary lamellae of the gills of Gadus pollachius. Q. Jl microx. Sci. 106, 343-353.
Hughes G. M. and Mittal A. K. l(l980) Structure of the gills of Barbus sophor (Ham), a cyprinid with tertiary lamellae. J. Fish Biol. 16, 461467. Karlsson L. (1983) Gill morphology in the zebrafish, Branchydanio (Hamilton-Buchanan). J. Fish Biol. 23, 51 lL514. Kopriwa B. M. and Leblond C. P. (1962) Improvements in the coating technique of radioautography. J. Histochem. Cvtochem. 10. 269-284.
Kendall M. W. and Dale J. E. (1979) Scanning and transmission electron microscopic observations of rainbow trout (Salmo Gairdneri gill. J. fish. Res. Ed Can. 36, 1072-1079. Laurent P. (1984) Gill internal morphology. In Fish Physiology (Edited by Hoar and Randall), p. 167. Academic Press, Orlando. Laurent P. and Dune1 S. (1980) Morphology of ail1 epithelia in fish. Am. J. Physiol: 238; R14?-R157. * LeBlond C. P. and Stevens C. E. (1948) The contrast renewal of the intestinal epithelium in the albino rat. Anat. Rec. 100, 357-371. Lipkin M. (1981) Proliferation and differentiation of gastrointestinal cells in normal and disease states. In Phvsioloev of the Gastrointestinal Tract (Edited by Johnson'L. RT), pp. 145-168. Raven Press, New York. Mackinnon M. and Enesco H. E. (1980) Cell renewal in the gills of the fish Barbus conchonius. Can. J. 2001. 58, 65&653. Moon H. W. (1983) Intestine. In Cell Pathology, 2nd Edition (Edited bv Cheville N.). DD. 503-529.Iowa State University Press, Ames. ” ’. Morgan M. (1974) Development of secondary lamellae of the gills of the trout Salmo gairdneri (Richardson). Cell Tiss. Res. 151, 509-523. Morgan M. and Tovell P. W. A. (1973) Structure of the gill of the trout Salmo gairdneri (Richardson). Z. Zellforsch. 142, 147-162. Quastler, H. and Sherman F. G. (1959) Cell population kinetics in the intestinal epithelium of the mouse. Exp. Cell Res. 17, 420-438. Snieszko S. F. (1981) Bacterial gill diseases of freshwater fishes. U.S. Fish and Wildlife Service, Fish Disease Leaflet 62. 1lpp. Washington DC. Tovell P. W., Morgan M. and Hughes G. M. (1970) Ultrastructure of trout gills. Sot. Franc. Microsc. Electron., Proc. Seventh Int. Cong. III, 601-602. Williamson, R. C. N. (1978) Intestinal adaptation. New Engl. J. Med. 298, 1393-1402; 1444-1450.