REGULATION OF INTERSTITIAL CELL DIFFERENTIATION IN HYDRA ATTENUATA

Transcription

1 J. Cell Sri. 5a, (98) 85 Printed in Great Britain Company of Biologists Limited 98 REGULATION OF INTERSTITIAL CELL DIFFERENTIATION IN HYDRA ATTENUATA VL POSITIONAL PATTERN OF NERVE CELL COMMITMENT IS INDEPENDENT OF LOCAL NERVE CELL DENSITY SHELLY HEIMFELD* AND HANS R. BODE Department of Developmental and Cell Biology, and the Developmental Biology Centre, University of California at Irvine, Irvine, Calif. 9277, U.S.A. SUMMARY The interstitial cell of hydra is a multipotent stem cell, which produces nerve cells as one of its differentiated cell types. The amount of interstitial cell commitment to nerve differentiation varies in an axially dependent pattern along the body column. The distribution of nerve cell density has the same equivalent axial pattern. These facts have led to speculation that the regulation of nerve cell commitment is dictated by the nerve cell density. We examined this question by assaying interstitial cell commitment behaviour in 2 cases where the normal nerve cell density of the tissue had been perturbed: () in epithelial hydra in which no nerve cells were present; and (2) in hydra derived from regenerating-tip isolates in which the nerve density was increased nearly 4-fold. We found no evidence of regulation of nerve cell commitment in response to the abnormal nerve cell densities. However, the typical axial pattern of nerve commitment was still obtained in both sets of experiments, which suggests that interstitial cell commitment to nerve differentiation is dependent on some parameter of axial location that is not associated directly with the local nerve cell density. INTRODUCTION The interstitial cell of hydra is a multipotent stem cell, producing 2 classes of differentiated cell types during asexual growth: nerve cells (e.g., see Davis, 974) and nematocytes (Lehn, 95; Slautterback & Fawcett, 959). David & Gierer (974) have calculated that under steady-state growth conditions the average behaviour of the interstitial cell population is such that, daily, 0% of the cells are committed to the nerve pathway, 30 % to nematocyte differentiation, and the remaining 60 % divide to maintain the stem cell population. This raises the question as to what is responsible for the regulation of interstitial cell commitment. Several authors have proposed that interstitial cell differentiation and proliferation behaviour is regulated by interactions with other cell types (see Bode & David, 978). Yaross & Bode (978 a) measured the distribution of cell types and interstitial cell commitment to nerves and nematocytes in various axial regions of the hydra. They found a strong correlation between the axial position and both the interstitial cell commitment and the density of certain cell types. In the specific case of nerve differen- Author for correspondence.

2 86 S. Heimfeld and H. R. Bode tiation, they found that the interstitial cell commitment pattern closely paralleled the distribution of nerve cell density. Where the local nerve cell density was high, such as in the head and foot, nerve cell commitment was also high. In the gastric region where the local nerve cell density was much lower, interstitial cell commitment to nerve differentiation was correspondingly reduced. Based on this kind of data, Yaross & Bode (978 a) proposed that the positionrelated differences in nerve cell commitment were dependent on the differences in local nerve cell density at these axial locations. We tested this idea by assaying the nerve cell commitment behaviour of the interstitial cell population in milieux in which the normal nerve cell densities had been modified experimentally. We found interstitial cell commitment to nerve differentiation to be uninfluenced by the local nerve cell density, but commitment still showed an axial dependence. MATERIALS AND METHODS Culture of animals Hydra attenuata were cultured at 8 (± ) C in media consisting of x io~ 3 M-CaCl t, x io~ 3 M-NaHCO 3) x io~ s M-Na,EDTA in distilled H,O. Stock cultures were fed 6 days per week with Artemia nauplii and washed approximately 8 h after feeding. Hydra with two or more buds were selected for experimentation 24 h after the last feeding. They were not fed during the course of any experiment. Epithelial hydra (Campbell, 976) were generously provided by Dr D. Rubin. They were force-fed shrimp 5 days per week and flushed 8 h later as described by Campbell (976). Cultures were maintained in the media described above with the addition of 50/ig/ml rifampicin. Nomenclature and quantitation of cell composition The nomenclature used for interstitial cells is that described by Bode & David (978). Interstitial cells are those formerly designated as big interstitial cells (David, 973) that occur singly or in pairs. This class includes the multipotent stem cells as well as cells committed to nerve or nematocyte differentiation (David & Gierer, 974). Specific regions or tissue fragments were designated using the nomenclature of Wolpert (969): the head region (H), 4 equal fragments of the gastric region (, 2, 3, 4), the budding zone (B), the upper and lower peduncle (5, 6), and the basal disk or foot region (F). The cell composition of whole hydra or of specific regions was analysed using the maceration technique of David (973). For each sample, 0-30 hydra or tissue fragments were macerated together. To determine the density of any particular cell type (i.e. the ratio to epithelial cells) a differential cell count was made until a total of 000 epithelial cells was reached per determination. Epithelial cell numbers per hydra or tissue fragment were determined with a Neubauer Cell Counting Chamber using phase microscopy. [ 3 H]thymidine administration and autoradiography Whole hydra were radioactively pulse-labelled by injecting [ 3 H]thymidine (so/tci/ml; 6 Ci/mmol; Schwarz-Mann) into the gastric cavity using a polyethylene needle (David & Campbell, 972). Hydra regenerated from isolated pieces were labelled by soaking the fragments in a solution of [ s H]thymidine for -2 h. Slides of macerated, labelled cells were processed for autoradiography as follows: slides were washed in distilled H,0 for 30 min, dipped in :, distilled H,O/Kodak NTB, Nuclear Track Emulsion, and then dried. After 7 days exposure in the dark at 4 C, autoradiographs were developed for 7 min in Kodak D-9, rinsed in Stop-bath, fixed for 5 min in Kodak Rapid-fix, and then washed thoroughly with distilled H a O for 30 min.

3 I-cell commitment uninfluenced by nerves 87 Transplantation techniques Hydra pieces were grafted together by threading appropriate fragments, as described in Results, on to nylon fishline (2 pound test). Pieces of polyethylene tubing (3-5 mm); (PE 50, Clay-Adams), just large enough tofitsnugly on the fishline, were used as sleeves to keep the cut edges in contact for -2 h to permit healing prior to removal from the fishline (Herlands & Bode, 974). Formation of regenerating-tip hydra Head regeneration was initiated in all cases by removal of the hypostomal region immediately below the tentacle whorl. The regenerating tip (defined as the apical /8 to /20 of the remaining gastric region) was excised 24 h following decapitation. After removal of the regeneratingtip tissue, a control region (consisting of the subjacent /3 to /4 of the gastric region) was also excised. The regenerating-tip and control isolates were then placed in small Petri dishes containing hydra medium and allowed to undergo regeneration and proportion-regulation for 4 days, to form miniature hydra. Only those isolates that had regenerated into hydra containing a single head, body column and foot region were used for further experimentation. Assay for nerve cell commitment Nerve cell commitment was assayed as described previously (Yaross & Bode, 978a). Hydra or tissue isolates were pulse-labelled with [ s H]thymidine to label a cohort of the interstitial cell population. After sufficient time to allow differentiation of newly committed nerve cells (26 28 h), 0-30 hydra or tissue fragments were macerated. The resulting cell suspensions were spread on slides and processed for autoradiography. The fraction of the pulse-labelled stem cells committed to nerve cell differentiation was determined from the distribution of the radioactively labelled cells. After h, labelled interstitial cells in nests of and 2 cells (i* and 2 # ), dividing nematoblasts in nests of 4 cells (4*), and labelled nerve cells (Nv # ) would all be derived from the initially labelled cohort (see Yaross & Bode, 978a). The nerve cell commitment fraction was therefore defined as follows: Nerve cell commitment fraction (NvC) = Nv # From 500 to 000 total labelled cells were counted per determination. RESULTS Regional pattern of interstitial cell commitment to nerve differentiation When hydra are continuously exposed to [ 3 H]thymidine the percentage of labelled nerve cells rises equally in the hypostome, the gastric region and the peduncle (David & Gierer, 974). Since the density of nerve cells varies considerably in these regions (Bode et al. 973), this indicates that the regional differentiation of interstitial cells into nerves must occur at a rate proportional to the existing nerve cell density. By measuring regional interstitial cell commitment to nerve differentiation, Yaross & Bode (978 a) directly demonstrated this point. This measurement was repeated to obtain a more detailed analysis of the axial pattern of interstitial cell commitment and its possible relationship to the local nerve cell density. To measure the interstitial cell pattern of differentiation to nerves along the axis of the hydra body column, whole hydra were pulse-labelled with [ 3 H]thymidine, incubated for 28 h, and then cut into the 9 tissue fragments shown in the abscissa of Fig.. These 9 regions were macerated separately and processed for autoradiography.

4 88 S. Heimfeld and H. R. Bode A i i l I I I i I B I < I I i i i i \ \ \ / F Fig. i. Axial pattern of nerve cell commitment and nerve cell distribution in hydra. Nine regions (as indicated on the abscissa) were excised from 5-20 hydra per determination. Each value ( ) represents the mean of 2 independent experiments. A. Regional variation in nerve cell commitment fraction, B, Regional variation in nerve cell density H, head; F, foot. The nerve cell commitment fraction (NvC) was determined for each region and is plotted in Fig. A. A high degree of regional variance in NvC was obtained as had been found before (Yaross & Bode, 978a). The head and foot extremities have values of NvC between O-4O-O-46 while the mid-gastric region was 0 to 20-fold lower ( ). Nerve cell density (ratio of nerve cells to epithelial cells) was measured concurrently in these same 9 regional fragments and the values obtained are shown in Fig. B. The axial distribution of nerve cell density is strikingly similar to that found for nerve cell commitment. This similarity becomes particularly evident when the 2 parameters, NvC and nerve density, are plotted against one another as in Fig. 2. Linear regression analysis of these data gives a correlation coefficient r = 0-92, P < o-ooi. Based on results similar to Fig. 2 and other data, Yaross & Bode (978 a) formed a hypothesis that this relationship between NvC and nerve cell density was not coincidental, but causal in nature. They proposed that the regulation of the amount of interstitial cell commitment to nerve differentiation was a function of the local nerve cell density. High local levels of nerve density, such as found in the head and foot regions, were influencing the interstitial cell population in these regions to commit a

5 I-cell commitment uninfluenced by nerves Nerve cell density Fig. 2. Correlation of nerve cell commitment with nerve cell density. Each point represents the mean value of one axial region using the data from Fig. i. The solid line represents the linear regression curve of these points. The correlation coefficient r = 0-92, P < o-ooi. large fraction of their cells to nerve differentiation. Lower levels of nerve density exerted a much smaller influence. Thus, the positional pattern of NvC was a direct reflexion of the existing axial distribution of nerve cell density in the hydra. However, the hypothesis is based on a correlation. To test the idea directly we modified the normal axial distribution of nerve cell densities and then assayed the commitment behaviour of the interstitial cells to see whether they would respond according to the new nerve cell densities or would continue to yield the normal values of NvC characteristic of their axial location. Interstitial cell commitment to nerve differentiation in epithelial hydra The data shown in Fig. A, B indicate that the head region of a normal hydra has the highest levels of NvC (0-45) and nerve cell density (0-5) found anywhere in the animal. If the level of interstitial cell commitment to nerve differentiation is dictated by the local concentration of nerves then one would predict that removal of all the nerves in the head region should lead to a significant reduction in NvC. We tested this prediction using epithelial hydra. Hydra treated with colchicine will rapidly lose their interstitial cells (Campbell, 976). These hydra will then undergo a gradual loss of their nerve cells by dilution and tissue turnover due to epithelial cell growth with no concommitant nerve cell replacement, since the precursor interstitial cell population has been extinguished. They finally reach a state where they consist only of ectodermal and endodermal cells. These are then termed epithelial hydra (Campbell, 976). CEL 52

6 9o S. Heimfeld and H. R. Bode Inject [ 3 H]thymidine 24 h Excise 24 n > Measure NvC Fig. 3. Grafting procedure used to analyse nerve cell commitment of interstitial cells that migrated into normal or epithelial hydra tissue (see text for explanation). Crosshatching represents radioactively labelled tissue. The epithelial hydra used in these experiments were derived from animals originally produced by Dr D. Rubin. They have been completely devoid of nerve cells for over 3 years (Rubin & Bode, 98). However, in order to attempt to answer the above experimental prediction we reintroduced interstitial cells into these nerve-free hydra. Interstitial cells are known to migrate from region to region along the hydra body column (Tardent & Morgenthaler, 966; Campbell, 974). Furthermore, Herlands & Bode (974) demonstrated, by grafting labelled and unlabelled halves of hydra together, that some of the labelled migratory interstitial cells differentiate into nerves and nematocytes in the unlabelled host tissue. We used this ability of interstitial cells to migrate and subsequently differentiate to test the commitment behaviour of the interstitial cell population in tissue with a nerve cell density equal to zero. The procedure employed for analysing interstitial cell commitment after migration is shown in Fig. 3. Hydra were pulse-labelled with [ 3 H]thymidine, the Hi region (head plus apical /4 of the gastric region; see Materials and methods for nomenclature) removed and an equivalent unlabelled Hi region from either a normal or epithelial hydra was grafted in place. The grafts were left together for 24 h to allow sufficient numbers of labelled interstitial cells to migrate into the unlabelled host tissue. The hypostomal regions were then excised and incubated for an additional 24 h to permit time for the migratory interstitial cells to differentiate into nerves. Thereafter, the hypostomes for a sample were macerated, processed for autoradiography, and the nerve cell commitment fraction determined. In a second parallel set of experiments, larger tissue fragments consisting of the Hi234 regions (head plus entire gastric region) were grafted to the same pulse-labelled donor tissue (region 2-F). These grafts were treated as above except that after 24 h the grafts were broken at the region 3/4 boundary of the unlabelled host tissue. In this second procedure 80-90% of the migrating interstitial cells end up in host body region. For each of the 4 host tissues (H or H23 of normal or epithelial hydra) the number

7 I-cell commitment uninfluenced by nerves 9 Table. Nerve cell commitment of migratory interstitial cells in normal and epithelial host tissue Nerve cell Nerve cell Host type Region density commitment fraction Normal H Normal H ± Epithelial H 00 o-o 0-45 i 0-04 Epithelial Hi23 o-o ±o-o Each value represents the mean standard deviation of 2-3 independent experiments. Ten graft regions were pooled together per determination. of input cells was constant. After 24 h of grafting, approximately 200 labelled interstitial cells per graft migrated into the unlabelled hosts (unpublished data). Those cells that entered a normal head region showed a high amount of nerve commitment (Table ). Migrating cells that moved into the normal H23 host, and thus were localized mainly in body-region tissue, showed far lower levels of NvC. Thus, the nerve commitment behaviour of these migrating interstitial cells is very similar to that found for interstitial cells in the intact animal. The behaviour again reflects the axial distribution of nerve cell density (Table ). Examination of interstitial cell commitment behaviour in epithelial hydra showed that those cells that moved into the head region yielded a high NvC, while Hi23 interstitial cells had a very low nerve cell commitment (Table ). Thus, as in the normal grafts, the migrating interstitial cell population again showed positional differences in nerve commitment, but in this case the differences occurred in the absence of nerve cells. Removal of all the nerves from the head region had no effect on the amount of nerve cell commitment that subsequently took place, since the normal and nerve-free heads had the same level of NvC (Table ). These results suggest that: () the high density of nerve cells is not the factor responsible for high nerve cell commitment in the head; and (2) the positionally related differences in NvC values are not dependent on the distribution of nerve cell density in the tissue. Interstitial cell commitment to nerve differentiation in hydra derived front regenerating tips The above results using the epithelial hydra demonstrated that lowering the nerve cell density to zero did not lead to a reduction in the levels of interstitial cell commitment to nerve differentiation. To gain more information the converse experiment, of raising the nerve cell density above normal values to see whether modifications in the upward direction could influence nerve cell commitment, was carried out. To accomplish this we took advantage of 2 aspects of hydra regeneration. Removal of the apical portion of the hydra body column results in a regeneration of the missing hypostomal area at the apical end of the remaining tissue. This morphogenetic process is confined to the tip of the column and in its early stages is accompanied by major changes in the cellular composition of the tissue that will undergo this head regeneration (Bode et al. 973; Berking, 974; Schaller, 976). Specifically, 4-2

8 92 S. Heimfeld and H. R. Bode 24 h / I Excise ^ 4 days Measure [ 3 H] thymidine Fig. 4. Procedure for the formation of hydra with elevated levels of nerve cell density and subsequent regional nerve-cell commitment analysis (see text for explanation). Cross-hatching represents radioactively labelled tissue. interstitial cell commitment to nerve differentiation rises 0 to 20-fold from control body levels by 3 h after decapitation (Yaross & Bode, 9786), leading to a large increase in nerve cell density in the region that will eventually form the new head. A second property of hydra regeneration concerns the behaviour of an isolated fragment of body column. When a piece of hydra is excised it undergoes a process by which it rearranges and reshapes its tissue to form a complete hydra. The size of this regenerated hydra is dependent on the size of the initial isolate but the proportion of tissue that forms head or body tissue is always a constant (Bode & Bode, 980). When isolated, the regenerating tip also shows this property of proportion-regulation. This implies that the tissue that was fated to become totally head region has now been partially respecified to form some body region. Early in tip formation, however, nerve cell commitment rises dramatically. Therefore, some of these committed interstitial cells differentiate into nerves in what will become body region tissues. This leads to an elevated level of nerve cell density for this region. This was exploited to determine interstitial cell commitment behaviour in regions of raised nerve cell density. The experimental protocol used is outlined in Fig. 4. Hydra were decapitated directly below the tentacle whorl. Twenty four hours following head removal the regenerating tip and the control region below the tip (see Materials and methods for details) were excised. These isolates were incubated for 4 days to allow regeneration and formation of complete hydra. The resulting animals, termed regenerating tip and control hydra, respectively, were then cut into separate head and body regions, pulselabelled by soaking the fragments in a solution of [ 3 H]thymidine for -2 h, incubated for an additional 26 h, and then macerated and counted. To determine whether this decapitation/regeneration procedure had resulted in a

9 I-cell commitment uninfluenced by nerves 93 Type Regenerated tip Regenerated control Normal Regenerated tip Regenerated control Normal Table 2. Nerve cell commitment in regenerated hydra Region Head Head Head Body Body Body Epithelial cell no. 590 ± ± ± ± ± Interstitial cell density 0-09 ±o-oi O'I4±O-O2 008 ± ± ±O'O2 Interstitial cell L.I. (%) 42 ±3 45 ±! ± Nerve cell density Io-O Io-o O-II 0-04 Nerve cell commitment fraction lo-oi io-02 Each value represents the mean standard deviation of 2-4 independent experiments. A total of 0-30 tissue fragments were pooled per determination. Labelling index (L.I.) was determined 26 h after pulse administration. disturbance of the normal characteristics of the interstitial cell population, the interstitial cell density and labelling index were measured. As shown in Table 2, the levels of these 2 parameters in the head and body regions of the 2 types of regenerates were very similar to values found for the normal animal. This is a good indication that in these experimental hydra the interstitial cell population is normal. This was the case despite the fact that epithelial cell numbers were reduced 0- to 5-fold. For elevating the levels of nerve cell density this procedure was extremely successful. The fourth column in Table 2 shows that in the head regions of the tip and control regenerates the nerve cell density was only slightly increased as compared to the normal head (0-58, 0-62 versus 0-5). The body regions, on the other hand, showed much larger increases. In the control there was approximately 50% elevation in nerve cell density (0-7 versus o-i ), which may reflect an increased nerve differentiation occurring during regeneration. In the tip body region, however, there was an almost 400% increase (0-4 versus o-i ). In fact, the nerve cell density of the regenerating-tip body region was much closer to normal head values than to the value for body. Interstitial cell commitment to nerve differentiation in the regenerated hydra (last column of Table 2), once again, exhibited a positional dichotomy. Those interstitial cells that are in head-region tissue show a high level of nerve commitment while those in the body column have very low amounts. This occurred despite the nearly 4-fold differences in nerve cell densities between the various body regions and the nearly headlike level of nerve cell density in the tip body region. These results therefore yield similar conclusions to the previous data from epithelial hydra: () a high density of nerves is not a sufficient condition for elevated levels of nerve cell commitment; and (2) the positionally related differences in nerve cell commitment are not a function of local differences in nerve cell density.

10 94 S. Heimfeld and H. R. Bode DISCUSSION One characteristic of stem cell systems is their ability to read the levels of various external signals. They then respond by regulating the fraction of stem cells that will be committed to a particular differentiation pathway. The amount of response is determined by the strength of the signal (Lajtha, 979). The interstitial cells of hydra clearly demonstrate this property with regard to nerve differentiation: the level of nerve commitment displays a 20-fold range in values in response to differences in axial position (see Fig. A). The question of interest is how does an interstitial cell determine its axial location? What external signals are the stem cells reading in order to determine the amount of nerve differentiation they will undergo? A logical property of such a signal would be that it varies in intensity in a pattern equivalent to that of nerve commitment. The local nerve cell density fits this criterion very well (see Fig. A, B). This led Yaross & Bode (978a) to propose that the regulatory signal emanated from nerve cells, and that signal intensity was a direct function of the local nerve cell density. Thus, they suggested that the positional dependence of interstitial cell commitment to nerve differentiation had its basis in the axial distribution of nerve cell density in the hydra. This is in effect a positive feedback loop, which would appear to be unsuitable for maintaining a steady-state density of nerve cells and interstitial cells. However, the dynamics of these 2 populations are wholly consistent with such a mechanism (Yaross & Bode, 978 a). One consequence of this idea is that perturbations in signal intensity (i.e. changes in nerve cell density) should lead to alterations in nerve cell commitment. Assuming that the linear regression line of Fig. 2 is a good approximation of the proposed functional relationship between nerve cell density and signal intensity, we arrive at the following experimental predictions. () If the nerve cell densities in similar positions (e.g. head regions) of 2 animals are unequal, then interstitial cell commitment to nerves should be different. (2) If the nerve cell densities in different positions (e.g. head versus body regions) of the same animal are similar, then the NvC values should also be equivalent. (3) If the nerve cell density is raised or lowered, the direction of perturbation should also be the direction of alteration of nerve commitment. This is a consequence of the linear relationship in Fig. 2. These predictions were tested in 2 different kinds of experiments. In one case epithelial hydra totally devoid of nerve cells were used. These animals have normal regeneration, budding and patterning properties (Marcum, Campbell & Romero, 977). By grafting normal and epithelial hydra to one another, interstitial cells were transplanted via migration into tissue where they were confronted with nerve cell densities drastically different from normal values. Their NvC behaviour was compared with that of [ 3 H]thymidine-labelled interstitial cells transplanted into normal tissue. This is a more direct, and possibly more accurate, comparison than using the values obtained for the intact animal, as both interstitial cell populations examined arrived in the head and body tissue by migration. If nerve cell density dictates the intensity of the commitment signal then interstitial

11 I-cell commitment uninfluenced by nerves 95 / / t 0 / / A - o! 0-2 e / / / o Nerve cell density Fig. 5. Comparison between nerve cell commitment of regions of normal and experimental hydra. The solid line is the linear regression relationship between nerve-cell commitment fraction and nerve cell density found in normal regions of hydra (Fig. 2). The symbols represent experimental data (Tables,2): epithelial head (O), body ( ); regenerating-tip head (A), body (A). cells in the head regions of normal and epithelial hydra should show unequal amounts of nerve commitment (prediction (i)), with the epithelial head having the lower NvC value (prediction (3)). Since the nerve cell density is zero throughout the epithelial tissue, nerve commitment should be equivalent in the head and body regions (prediction (2)). In the second type of experiment hydra were generated in which the nerve cell density of the body column was elevated to a level nearly 4-fold higher than normal. In this case one would expect interstitial cell commitment to show unequal values of NvC when comparing the control to the experimental hydra body columns (prediction ()), and the regenerating-tip body should have the higher value (prediction (3)). In neither of these experimental modifications of nerve cell density did any of the above predictions hold (Tables and 2). However, an experimental consistency was observed in all cases. Head regions, regardless of their local nerve cell density, showed high levels of nerve commitment. Body-region tissue yielded low amounts of nerve commitment, again independent of local nerve cell density. This point is emphasized graphically in Fig. 5. The experimental data from the epithelial and regenerating-tip hydra are presented with the NvC versus nerve density curve found for the normal animal (F.ig. 2). These points clearly do not correlate with the linear relationship found previously but they do emphasize the positional dichotomy in NvC values between head and body regions. This suggests that signal intensity is still functionally

12 96 S. Heimjeld and H. R. Bode associated with axial location, but that nerve cell density is not responsible for this regulation. Certain aspects of the experimental data are subject to interpretational difficulties that could weaken the above conclusion. In epithelial hydra only the migrating subpopulation of the interstitial cells was assayed. If these cells constitute a special class that was precommitted to differentiation prior to migration, then the elevated level of NvC detected in the head may be a reflexion of this precommitment rather than a response to a signalling mechanism. This explanation is unlikely for the following reasons. Migrating cells transplanted from the same donor tissue showed large differences in nerve cell commitment in different axial locations. We should have found similar values in the different locations in order to argue for precommitment. To continue the idea further, one would have to assume different classes of migrating cells that only move in response to the correct host tissue: head or gastric region. While this point has not been rigorously tested the fact that equal numbers of cells migrated into each region puts another constraint on this idea. Another argument against precommitted migration is the fact that it is possible to repopulate a nerve-free animal to recover a normal hydra, indicating that some migrating cells are multipotent (Campbell, 976). The experiment with epithelial hydra may have another weakness in that functions normally performed by nerve cells may be taken over by the epithelial cells in nerve-free hydra. This type of behaviour has recently been suggested for the head activator that affects head regeneration in hydra (Schaller, Grimmelikhuizzen, Schmidt & Bode, 979). One could thus explain the interstitial-cell commitment behaviour by assuming that in these nerve-free hydra the signalling mechanism has been taken over by the epithelial cells. All of these objections, however, are not applicable to the regenerating-tip data. Here we used tissue that is normal, in that it contains nerve cells. Also we assayed the commitment behaviour of the total interstitial-cell population, not a migrating subclass. These results clearly rule out a role for nerve cell density in regulating signal intensity. This extends the observation that simple feedback loops from the differentiation products or intermediates to the interstitial cells do not play a role in controlling interstitial-cell commitment behaviour. It is shown here for nerves and has been demonstrated previously for nematocytes (Bode& David, 978; Yaross & Bode, 978 a). If signal intensity is not governed by feedback from the product cells, how then does the position-dependent pattern of nerve differentiation arise? A clue may be found in the observed relationship between nerve commitment and head regeneration. Following decapitation, the apical tip forms a new head in 2-3 days. However, within hours after cutting, the tip tissue acquires the capacity to induce a secondary axis when transplanted into the body region of an intact host. This increase in capacity for head formation, termed activation, is restricted to the tip. It does not occur in the regions below (e.g., see MacWilliams, 98). As mentioned previously, the rise in nerve cell commitment during head regeneration shows a similar positional restriction (Bode et al. 973; Yaross & Bode, 9786). Even stronger support for this relationship is provided by the results of Yaross, Baca, Bode & Chow (unpublished). They found that the kinetics of the rise in interstitial cell commitment to nerve differentiation

13 I-cell commitment uninfluenced by nerves 97 closely parallel the rise in activation of the tip tissue. These results suggest the intriguing idea that perhaps these 2 phenomena, the rise both in activation and nerve cell commitment during head regeneration, are responses to the same positiondependent signal. This could explain how a change in signal intensity would govern pattern regulation and at the same time affect the spatial distribution of a differentiated cell type. This research was supported by a research grant from the National Institute of Child Health and Human Development (HD 08086) and the N.I.H. Training Grant for Mechanisms underlying Development (HD 07024). REFERENCES BERKING, S. (974). Nachweis eines morphogenetisch aktiven Hemstoffs in Hydra attenuata und Untersuchung seinen Eigenschaften und Wirkungen. Ph.D. thesis, University of Tubingen. BODE, H. R., BERKING, S., DAVID, C. N., GIERER, A., SCHALLER, H. & TRENKNER, E. (973). Quantitative analysis of cell types during growth and morphogenesis in Hydra. Wilhehn Itoux Arch. EntwMech. Org. 7, BODE, P. M. & BODE, H. R. (980). Formation of pattern in regenerating tissue pieces of Hydra attenuata. I. Head-body proportion regulation. Devi Biol. 78, BODE, H. R. & DAVID, C. N. (978). Regulation of a multipotent stem cell, the interstitial cell of Hydra. Prog. Biophys. molec. Biol. 33, CAMPBELL, R. D. (974). Cell movements in Hydra. Am.Zool. 4, CAMPBELL, R. D. (976). Elimination of Hydra interstitial and nerve cells by means of colchicine.,7. Cell Set. 3, -3. DAVID, C. N. (973). A quantitative method for maceration of hydra tissue. Wilhelvi Roux Arch. EntwMech. Org. 7, DAVID, C. N. & CAMPBELL, R. D. (972). Cell cycle kinetics and development of Hydra attenuata. I. Epithelial cells. J. Cell Set., DAVID, C. N. & GIERER, A. (974). Cell cycle kinetics and development of Hydra attenuata. III. Nerve and nematocyte differentiation.^. Cell Set. 6, DAVIS, L. E. (974). Ultrastructural studies of the development of nerves in Hydra. Am. Zool. I4> HERLANDS, R. L. & BODE, H. R. (974). Oriented migration of interstitial cells and nematocytes in Hydra attenuata. Wilhehn Roux Arch. EntwMech. Org. 76, LAJTHA, L. G. (979). Stem cell concepts. Differentiation 4, LEHN, H. (95 I). Teilungsfolgen und Determination von I-Zellen fur die Cnidenbildung bei Hydra. Z. Naturf. 6b, MACWILLIAMS, H. K. (98). Hydra transplantation phenomena and the mechanism of Hydra regeneration Devi Biol. (In Press.) MARCUM, B. A., CAMPBELL, R. D. & ROMERO, J. (977). Polarity reversal in nerve-free hydra. Science, N. Y. 97, RUBIN, D. I. & BODE, H. R. (98). Nerve cells and epithelial cells are both involved in the inhibition gradient of Hydra. Devi Biol. (In Press). SCHALLER, H. C. (976). Head regeneration in Hydra is initiated by release of head activator and inhibitor. Wilhelm Roux Arch. EntwMech. Org. 80, SCHALLER, H. C, GRIMMELIKHUIZZEN, C, SCHMIDT, R., & BODE, H. (979). Morphogenetic substances in nerve-depleted hydra. Wilhelm Roux Arch. EnttvMech. Org. 87, SLAUTTERBACK, D. B. & FAWCETT, D. W. (959). The development of cnidoblasts of Hydra. An electron microscope study of cell differentiation. J. biophys. biochein. Cytol. 5, TARDENT, P. & MORGENTHALER, U. (966). Autoradiographische Untersuchungen zum Problem der Zellwanderungen bei Hydra attenuata (Pall.) Revue suisse Zool. 73,

14 98 S. Heimfeld and H. R. Bode WOLPERT, L. (969). Positional information and the spatial pattern of cell differentiation. J. theor. Biol. 25,-47. YAROSS, M. S. & BODE, H. R. (978a). Regulation of interstitial cell differentiation in Hydra attenuate. III. Effects of i-cell and nerve cell densities. J. CM Sci. 34, -26. YAROSS, M. S. & BODE, H. R. (9786). Regulation of interstitial cell differentiation in Hydra attenuata. IV. Nerve cell commitment in head regeneration is position-dependent. J. Cell Sci. 34, {Received 7 April 98 - Revised 25 June 98)