DEVELOPMENTAL
BIOLOGY
141,412-420 (1990)
Patterning Processes in Aggregates of Hydra Cells Visualized with the Monoclonal Antibody, TS19 MIKA SATO,’ HANS R. BODE, AND YASUJI SAWADA* Developmental Biology Center, Department of Developmental and Cell Biology, University of Calzfbrnia at Irvine, Irvine, Ca&fornia 92717, USA; and *Research Institute of Electrical Communication, Tohoku University, 2-1-l Katahira, Send& 980, Japan Accepted June 12, 1990 The monoclonal antibody, TS19, (Heimfeld et aL, 1985), labels the apical surface of ectodermal epithelial cells of tentacles and lower peduncles in Hydra To investigate the patterning process in a tissue whose original pattern was completely destroyed, the TS19 staining pattern was examined in developing aggregates of Hydra cells. Two types of aggregates were prepared. G-aggregates were made from tissue of the gastric portion of animals and RG-aggregates from gastric tissue allowed to regenerate for 24 hr before making aggregates. G-aggregates were initially TSlS-negative, and later dim and uniformly TSlS-positive. Thereafter, TS19 staining broke up into brightly stained and unstained regions. The brightly staining regions developed into head or foot structures. The TS19 pattern in RG-aggregates developed differently. Since the initial aggregates contained cells of regenerating tips, they started with TSlS-positive cells as well as TSlS-negative cells. The numbers of brightly staining TSlS-positive cells increased with time. Some patches of these cells developed into head or foot structures, while others did not. These results and a simulation using a reaction-diffusion model suggest that the changes in activation levels affected the temporal changes in the pattern of TS19 staining, and that the de nouo pattern formation in hydra can be explained in terms of a process involving activation and inhibition properties. o lgw Academic press, Inc. INTRODUCTION
A great deal is known about the processes underlying pattern formation in Hydra (see Bode and Bode, 1984 for review). As the animal has a simple body plan (a head and a foot on opposite ends of the body column, which is a tube), the central question is how the head and foot are patterned. The question is even more interesting because of the phenomenon of regeneration polarity. An excised piece of the body column will always regenerate a head at the original apical end and a foot at the original basal end. An extensive body of grafting and regeneration experiments have demonstrated that the primary processes involved in locating as well as forming the head and foot are two pairs of developmental gradients. One pair, of head activation and head inhibition, controls head formation (MacWilliams, 1983a,b) and the other pair with similar properties controls foot formation (MacWilliams et aZ., 1970; Hicklin and Wolpert, 1973). Further, the dynamics of the gradients explain the polarity of regeneration (Wolpert, 19’74; Bode and Bode, 1984). A large fraction of these data has been explained quite well in terms of a reaction-diffusion model (Gierer and Meinhardt, 1972; MacWilliams, 1982; MacWilliams, 198313). 1 Present address: Research Institute of Electrical Communication, Tohoku University, 2-l-l Katahira, Sendai 980, Japan. 0012-1606/90 $3.00 Copyright All rights
0 1990 by Academic Press, Inc. of reproduction in any form reserved.
More recently the use of region-specific monoclonal antibodies has refined the understanding of the processes. All of the previous work relied on the final structures formed to deduce the nature of the patterning processes. Two of the monoclonal antibodies stain regenerating heads long before any morphological structures have formed (Javois et aL, 1986; Heimfeld et al, 1985; Bode et aZ., 1986). By following the temporal changes in the staining patterns during regeneration, one can get an idea of the dynamics of the patterning processes involved, and thereby increase the precision of the deductions made about the processes. For example, an unanswered question concerned the order of patterning of the two parts of the head. The head consists of a hypostome at the apex above a ring of tentacles. The monoclonal antibody, TS19, recognizes an antigen which is restricted to the apical surface of the ectodermal epithelial cells of the tentacles and the lower part of the body column (Heimfeld et al, 1985; Bode et al, 1986). The temporal pattern of changes of TS19 staining during head regeneration indicated that the tissue was first committed to form tentacles, and later the hypostome (Bode et ak, 1988). All of this information has been obtained by perturbing, but not destroying, an established pattern. Hence, it is not known whether the discerned patterning processes are sufficient for setting up the pattern of head and foot at opposite ends of the animal, or if additional processes are necessary. One approach to this problem 412
SATO, BODE, AND SAWADA
Patterning
is to dissociate Hydra tissue into a suspension of cells, form aggregates of cells by centrifugation, and allow the aggregates to regenerate complete animals (Noda, 1971; Gierer et al, 1972). In this approach the original pattern is completely destroyed, and a new pattern must arise de nova. Not only does this provide a means for determining if other patterning processes are involved, but it also provides a different context for studying the known processes required for head and foot formation. The aggregation technique has been used to show that (1) in the absence of organized gradients new heads or single tentacles form in a spacing pattern which can be explained in terms of lateral inhibition, a prominent feature of existing models for Hydra patterning (Sato and Sawada, 1989), (2) tentacle formation is unstable unless associated with a developing head (Sato and Sawada, 1989), and (3) head activation levels can be maintained at the level of individual cells for a significant period of time (Gierer et al, 1972). To gain more information on the patterning processes operating in aggregates we have examined the temporal changes of the TS19 staining pattern, which will appear long before any morphological structures, during aggregates development. The results suggest that in the absence of an organized pattern the entire tissue of the aggregate initiates head and/or foot resulting in the observed spacing pattern of heads and feet. MATERIALS
Animals
AND
G-AGGREGATE ,
Dis.vr$tion Aggregation
RG-AGGREGATE Dissz;dion Aggregation
FIG. 1. Two types of aggregates used in the present work. G-aggregates were made from gastric tissue. RG-aggregates were made from gastric tissue allowed to regenerate for 24 hr before dissociation.
regions were pooled, dissociated, and aggregated to form aggregates in the size expected to produce one or two hypostomes after regeneration (Sato and Sawada, 1989). (2) RG-aggregates: aggregates made from gastric tissue allowed to regenerate for 24 hr before dissociation. The beginning of the regeneration period was defined here as the time when an aggregate was transferred from the dissociation medium at 4°C into DM at 18°C soon after centrifugation. The “steady state” of regeneration was defined as the condition in which the numbers of structures remained unchanged for more than 3 days, which was the same definition used previously (Sato and Sawada, 1989).
METHODS
and Culture
Hydra vulgar&x (formerly H. attenuata) were used for all experiments. Stock cultures of animals were maintained as described by Shimizu and Sawada (1987) and by Dunne et al. (1985). Animals used for experiments were starved for 1 day before use. Preparation
413
in Aggregates of Hydra Cells
of Aggregates
The procedure described previously (Sato and Sawada, 1989), based on the method of Flick and Bode (1983), was used for making aggregates. Briefly, tissue pieces were dissociated into a cell suspension in hyperosmotic medium (dissociation medium, DM; Flick and Bode, 1983), filtered through a nylon cloth with a mesh of 53 pm and aggregated by low-speed centrifugation (200g). The medium was gradually diluted according to the degree of regeneration, In the present work, two types of aggregates were prepared (Fig. 1). (1) G-aggregates: aggregates made from only the tissue of gastric region. The gastric region, the cross-hatched area of the animals in Fig. 1, of more than 50 animals were excised by removing the apical eighth and basal half of each animal. The gastric
Immunocytochemistry Live aggregates or animals were stained with the monoclonal antibody TS19 (Heimfeld et aL, 1985; Bode et al., 1986) using an avidin-biotin amplification procedure. Aggregates or animals were incubated in a 1:lOO dilution of a TS19 ascites fluid in Hydra culture medium (HM) (Dunne et al, 1985) for 5 min. After being washed three times in HM, they were incubated with 1:50 dilution of a biotinylated goat anti-mouse IgG (TAGO) in HM for 5 min, washed three times again, incubated with a 1:50 dilution of a FIT&labeled avidin (Vector Lab. Inc.) for 5 min, and, finally, rinsed three times in HM. To stain aggregates at an early stage before a cavity developed, dissociation medium (Flick and Bode, 1983) diluted 1:l with HM was used instead of HM, because aggregates at this stage tended to disintegrate in HM. The diluted dissociation medium did not affect TS19 staining as no difference was observed between samples stained in HM and diluted dissociation medium. Measurement
of Relative
The relative intensity TSlS-stained aggregates
Intensity
of Fluorescence
of fluorescence was quantitated
of areas of with a pho-
414
DEVELOPMENTALBIOLOGY VOLUME141,199O
tometer attached to a Zeiss photomicroscope equipped with epifluorescence. An excitation frequency of 365 nm and an emission frequency of 450 nm were used. Stained aggregates were relaxed in 2% urethan for less than 1 min, fixed in 70% ethanol, and flattened between two coverslips (Bode et aZ., 1988). Since the entire surface of the ovoid aggregates was stained, and the intensity of stain was maximal at the edges for geometric and optical reasons, measurements were made at the edges. The diameter (16 pm) of the measured area was defined by a shutter on the photometer. During early stages of regeneration when the stain was uniform across the aggregate, the fluorescence intensity was measured at four to eight arbitrarily chosen locations around the periphery of each of five aggregates. At later stages when the staining pattern was nonuniform, the intensities of all maxima and minima around the periphery were measured. Numerical
Simulations
The particular version of the reaction-diffusion model (MacWilliams, 1982) was used. In this version, the body of intact Hydra is divided into 21 parts (#l-#21), where #l-#5 are head parts and #6-#21 are body parts. In the simulation for G-aggregates, activator values of #lo-#21 were used to decide the initial values in aggregates. As for RG-aggregates, activator values of #lO#21 after 1440 iterations of calculation were used assuming the regeneration before dissociation, where #lO exhibited the regeneration tip. Aggregates were assumed as 42 columns. Here, column #1 was next to both columns #2 and #42. For each column, activator values of intact Hydra (#lo-#21) were randomly chosen 30 times and the averaged value of them was used as the initial activator value of the column. The calculated values were plotted per 355 iterations. The same value throughout the columns was used as the initial inhibitor value of the aggregate because of its big diffusion constant. RESULTS
Development
of Animals
from Two Types of Aggregates
Two types of aggregates (Fig. 1) were examined in the present work. One type, G-aggregates, was made from tissue of the gastric region, and therefore, contained no TSlS-positive cells when they were made. TSlS-positive cells are found only in the tentacles and lower peduncle of adult Hydra (Heimfeld et aL, 1985). The other type, RG-aggregates, was made from isolated gastric regions, that were allowed to regenerate for 24 hr before dissociation, Thus, RG-aggregates contained cells from regenerating tips which are TSlS-positive by 24 hr.
The overall regeneration of animals from aggregates was as previously described (Gierer et al., 1972; Graf and Gierer, 1980). Within 6-8 hr of pelleting the cells, ectodermal epithelial cells sorted out to the surface to form a continuous epithelium covering the entire surface. Within 24 hr, a two-layered hollow sphere had formed consisting of ectoderm and endoderm surrounding a fluid-filled cavity. By 2-4 days tentacles and hypostomes begin to form, and by 10 days complete animals had formed. There were two primary differences between the development of the two types of aggregates. First, tentacles appeared earlier in RG-aggregates than in G-aggregates. Second, in G-aggregates, most tentacles formed near a hypostome as part of a developing head. Few body tentacles, which are not associated with a head, formed. In contrast, many body tentacles were found in RG-aggregates during the early stages of regeneration. Later, the isolated body tentacles were absorbed, and most of the remaining tentacles were associated with developing heads. These differences between the two types of aggregates were observed previously (Sato and Sawada, 1989), although in a different species. Temporal Changes in the Pattern in G-Aggregate
of TS19 Binding
To examine the staining pattern of TS19 at various times during the development of G-aggregates, samples of aggregates at different stages of regeneration were stained with TS19 and visualized with indirect immunofluorescence. The temporal changes are shown in Fig. 2. By 6 hr, when the outer layer, the ectodermal epithelium, had formed, no TS19 staining was visible (Fig. 2a). By 30 hr when the aggregate had developed into a hollow sphere, the entire surface exhibited a dim uniform stain (Fig. 2b). A day later, the stain was still uniform but the intensity had increased (Fig. 2~). At this point the intensity was much less than that found in the tentacles and lower peduncle of intact animals. By the third day, the pattern began to change. The aggregates developed protrusions, some of which were clearly identified as tentacle bumps, which are evaginations that subsequently developed into tentacles. Further, the TS19 staining pattern was no longer uniform. Protrusions or tentacle bumps were brightly stained with TS19 whereas other regions were only dimly stained (Fig. 2d). Compared to the previous day, the intensity of the TS19 stain had increased on protrusions and had decreased in other areas. Over the next several days this trend continued. As the tentacles developed, the intensity of TS19 staining increased reaching that of complete animals (Fig. 2e). In regions that developed into the body column, the stain continued to decrease
Patterning
SATO,BODE,ANDSAWADA
and disappear to the background adult body column. Quantitative Temporal on G-Aggregates
level typical
of the
Changes in TS19 Binding
To obtain a quantitative and more detailed description of the changes in TSlS-binding pattern, the fluorescence intensity around the periphery of the aggregates was measured at different times. Figure 3 shows results of the quantitative measurements on G-aggregates stained with indirect immunofluorescence (see Materials and Methods). By 6 hr after aggregation when an ectodermal epithelium had formed, the level of fluorescence intensity was the background level observed in the body columns of intact unstained animals. Over the next 48 hr, it steadily increased loo-fold. By 52 hr, the aggregates changed from a uniform sphere, or ovoid, shape to an ovoid with protrusions (Fig. 2~). Within variation, all cells were uniformly stained at this time. By ‘78 hr, significant changes in staining intensities occurred. The developing tentacles were more intensely stained (crosses at 78 hr; Fig. 3) than the overall epithelium at 52 hr. Further, regions of the ectoderm that were not part of the protrusions or developing head structures decreased in intensity. Both of these trends continued over the next several days so that by the time the aggregates had reached a steady state in terms of the number of head structures to be formed (here, 10 days), the tentacles had acquired a staining intensity typical of tentacles in intact animals. The lower parts of the body column, the peduncles and feet had also formed on the aggregates. By the steady state, the peduncles had also reached the staining level typical of intact animals. Conversely, the staining of the regions of the epithelium destined to form the body column had dropped to the background levels typical of the body column of control. Temporal
Changes in TS19 Binding
on RG-Aggregates
Tentacles formed earlier and in larger number in RGaggregates compared to G-aggregates. These differences are easily attributed to the head regeneration processes that occur in the excised piece during the 24 hr before dissociation and aggregation (e.g., Bode and Bode, 1984). To determine if this 24-hr period of regeneration affected the general pattern of the TS19 staining, RG-aggregates were stained and observed at 6,30, 54, and 78 hr. Twenty four hours after excision isolated gastric regions exhibit TSlS-positive cells at the original apical and basal ends. Thus, it was not surprising that in 6-hr RG-aggregates there were isolated TSlS-positive cells
in Aggregates of Hydra Cells
415
in a background of unstained cells (Figs. 4a and 4b). By 30 hr the number of brightly staining TSlS-positive cells had increased, and were occasionally found in small patches (Fig. 4~). In addition there were dimly staining cells as well as many TSlS-negative cells. This is in strong contrast to G-aggregates at this time in which all cells were uniformly dimly stained (Fig. 2b). By 54 hr the size and number of patches had increased (Figs. 4d and 4e). Patches of labeled cells were next to patches of unlabeled cells (Fig. 4e), indicating that the labeling could be discontinuous instead of necessarily graded over an area. All tentacles and tentacle bumps that appeared were brightly stained (Fig. 4d). However, not all patches of TSlS-positive cells were associated with developing head or foot structures. Nor were all protrusions uniformly stained with TS19. The protrusion in the foreground of the aggregate in Fig. 4e has a patch of dimly stained, or unstained, cells at the tip. Further, the patch of stained cells below the tip extends away from the protrusion into a region of the aggregate where no head or foot structures seem to be forming. These observations will be dealt with in the Discussion. By 78 hr numerous intensely stained tentacles had formed (Fig. 4f). The patches of positive cells not associated with head or foot structures tended to disappear by this time. Numerical
Simulation
with a Reaction-R$sion
Model
Much of the regeneration and grafting data has been explained in terms of a reaction-diffusion model that postulates a stable activator molecule with a small diffusion constant, and an unstable inhibitory molecule with a large diffusion constant (Gierer and Meinhardt, 1972; MacWilliams, 1982). The parameters of the model were chosen to explain the dynamics of the gradients as exhibited in those experiments. It is of interest to know whether the particular version of the reaction-diffusion model (MacWilliams, 1982) can explain the observed TS19 results, since the temporal changes in the TS19 staining during regeneration are correlated with the changes in head activation (Bode et al, 1988). In the following only head formation will be simulated. In an intact body column, the head activation gradient is very shallow, and the level of head activation is very low compared to the head (MacWilliams, 1982). Thus, for G-aggregates it was assumed that the overall head activation level was initially low and uniform throughout the aggregate. This assumption is reasonable given the complete disruption of the tissue into cells, and thus, the disruption of the gradient in the process. In addition, it is known that the cells could retain their original positional memory such as head activation level (Gierer et ah, 1972). The simulation (Fig. 5a)
416
DEVELOPMENTALBIOLOGY
vOLUME141.1%0
FIG. 2. Temporal changes in the TSlS-binding pattern on G-aggregates visualized by indirect immunofluorescence on a whole mount preparation. To summarize the binding pattern, the periphery of samples were focused. (a) At 6 hr from the beginning of the regeneration period. (a, right) A bright field image of the same aggregate as the fluorescent image, (a, left). (b) At 30 hr. (c) At 52 hr. (d) At 78 hr. (e) At the steady state (lo-day), where arrows show the head (left) and the foot (right) of the animal. Bars show 0.1 mm.
indicates that there was an initial slow rise followed by a more rapid rise in activation that was uniformly distributed over the entire aggregate. This would correspond to the observed uniform changes in TS19 staining. Later, several peaks of activation emerged and grew to a maximum value, while other areas declined in activation. RG-aggregates were also simulated. Because RG-aggregates at 6 hr had a range of TS19 staining intensities, a larger range of head activation levels was chosen and these were distributed randomly along the lines of cells. The simulation (Fig. 5b) indicates that the rise to high values of head activation occurred earlier than in G-aggregates, and that the location of activation peaks was to some extent correlated with initially high values of
activation. Thus, assuming the temporal changes in TS19 reflect changes in head activation in developing G-aggregates and RG-aggregates, the changes are simulated by the model quite well. DISCUSSION
TSl9 ReJEects the Activation
Patterning
Process
Two pairs of developmental gradients, activation and inhibition gradients controlling head formation and foot formation, have been known in Hydra (see Bode and Bode, 1984 for review). Activation is a stable tissue property that is graded along the axis. The head activation gradient has a maximum in the head while the foot activation is maximal in the foot. In contrast, inhibition is
SATO, BODE, AND SAWADA
Patter&Q
in
AQQWQateS
of
417
Hydra cek!s
head activation measured in the regenerating tip (Bode et aL, 1988), and are not correlated with the dynamics of the head inhibition gradient. Behavior of Activatim of G-Aggregates
I j
1830 52 78 I+-% 240 REGENERATION TIME (hr)
FIG. 3. Quantitative measure of the temporal changes in TSlS fluorescence intensity in G-aggregates. Regeneration times on the abscissa show the time when the aggregates were treated with TSlS. The ordinate shows the amount of fluorescence (in arbitrary intensity units) measured with a photometer. Note that they are on the logscale. Values equal to or less than 0.2 are exhibited as “0.2.” Five aggregates were stained and measured for each time point. Each row of points at a time point represents a single aggregate. Closed circle, nonprotruding portions of aggregates. Closed triangles, protrusion portions. Crosses, tentacles or tentacle bumps. Open triangles, lower peduncles. Open circles, body column. The mean value of intensity measured in tentacles, lower peduncles, and body columns of stained intact animals are shown in A, and those of body columns of unstained animals are shown in B (bars indicate the standard deviations).
an unstable property with a graded distribution. Head inhibition is produced in the head, is graded with maximum in the head, and is transmitted by diffusion down the body column, possibly through gap junctions (Fraser et aL, 1987). Foot inhibition, which has similar properties, is produced in the foot. Following decapitation, head inhibition levels fall rapidly throughout body column. Head activation at the apical tip rises to a head level within 16 hr (see Bode and Bode, 1984 for review). The tissue at the tip is then determined to make a head, produces the head inhibition substance to prevent other regions from forming head, and undergoes head morphogenesis and differentiation (e.g., Webster and Wolpert, 1966; Hicklin et a& 1975). A similar process occurs upon removal of the foot. Temporal changes in the TS19 staining pattern following decapitation are closely correlated with the rise in
Processes during
Regeneration
During head regeneration following decapitation (Bode et CCL,1988) and the regeneration of head and foot in an excised piece of tissue from gastric region (O’Neill, Awad, and Bode, unpublished experiments), TS19 staining observed is confined to the regenerating end(s). It does not appear in other regions at any time during regeneration. The staining, which initially appears as a dim stain and increases in intensity with time, correlates with the local rises in activation (Bode et al, 1988). In contrast, the TS19 staining in the G-aggregates is not confined to specific areas when it becomes clearly discernible. Instead, it is uniformly distributed over the entire surface of the aggregate (Fig. 2b). The succeeding temporal pattern is similar to that seen during head regeneration following decapitation (Bode et al, 1988). Assuming that the TS19 antigen is related to the activation process, the temporally changing TS19 patterns in G-aggregates (Figs. 2 and 3) could be interpreted as follows. (a) Head (and probably foot) activation levels are rising all over the aggregate. (b) The increasing intensity of staining reflects an increasing rate of change of the activation level. (c) When protrusions and tentacle bumps appear activation levels continue to rise there, but in areas where the stain is lessening or has disappeared, activation no longer rises. These qualitative results were simulated using MacWilliams’ version of a reaction-diffusion model (MacWilliams, 1982) that has been applied to the regeneration in hydra. The observed changes were consistent with rise and fall in TS19 staining intensity in different areas starting around 3 days. Thus, assuming that the temporal changes in TS19 reflect changes in head activation in the developing aggregate, these changes fit the model quite well. This interpretation has limits. It is unlikely that the very high level of the TS19 antigen in the adult tentacles has anything to do with activation properties. Adult tentacles are not capable of inducing secondary axes or regenerating other parts of the animal. Instead, it is more likely that the antigen also plays a role in the physiology of the adult animal. Behavior of Activation of RGAggregates
Processes during Regeneration
The differences in the regeneration of the RG-aggregates and of the G-aggregates can be explained in terms
418
DEVELOPMENTAL BIOLOGY
VOLUME 141.1990
FIG. 4. Temporal changes in the TSlS-binding pattern in RG-aggregates. (a and b) At 6 hr. The periphery of the sample was focused in a, while b is higher magnification of another sample. (c) At 30 hr. (d and e) At 54 hr. (f) At 78 hr. Arrow shows a tentacle expected to be a “body tentacle.” Bars show 0.05 mm.
SATO, BODE, AND SAWADA
Patterning
a
No.
posit
posit
ion
ion
FIG. 5. Evolution of head activator for (a) G-aggregates and (b) RG-aggregates; results of the simple numerical simulation using a reaction-diffusion model (MacWilliams, 1982). Column No. 1 was next to both columns No. 2 and No. 42. See text.
of differences in head activation levels and their distributions. (Differences in foot activation levels and distribution also occurred, but to keep it simple only the head will be considered.) Since the gastric regions used for RG-aggregates regenerated for 24 hr before making aggregates, the activation level at the apical end rose to the level of the head (e.g., MacWilliams, 1983b). Hence, instead of starting with a population of cells that has levels of head activation very similar to those in G-aggregates, the head activation levels differed greatly among the cells from the beginning. Because the cells would retain their activation level for a while (Gierer et al., 1972), the distribution of head activation over the surface of the aggregate would be heterogeneous. The changes in the TS19 staining pattern in RG-aggregates shown in Fig. 4 can be explained as follows. The initial aggregates contained bright TSlS-positive cells isolated in a background of unstained cells (Figs. 4a and 4b). Since 24 hr regenerating tips have bright TSlS-positive cells; this is not surprising. More telling is the pattern observed at 30 hr (Fig. 4~). The number of bright TSlS-positive cells has increased and these cells are clustered. The early appearance of head structures in RG-aggregates could be due to the sorting out of cells with high-level head activation into areas where they rapidly formed a tentacle or, if enough, a complete head. However if the loose clustering were simply due to sorting out of cells with a high head activation level, the number would not be expected to increase. Therefore, it is deduced that small numbers of high activation level cells that by chance were in one area could, through a metabolic process such as a reaction-diffusion mechanism, raise the head activation level of sufficient surrounding cells to form a head structure. The patches of very bright cells found in the tentacle protrusions at
in Aggregates of Hydra Cells
419
54 hr (Figs. 4d and 4e) provide a vivid correlation between high activation levels and the appearance of structure. Simulation of these results with MacWilliams’s model (1982) showed that the model correlates well with the data. The protrusion in Fig. 4e has a patch of TSlS-positive cells as well as a patch of TSlS-negative cells near the apex. If the protrusion were to form a tentacle, one would expect the entire protrusion to be TSlS-positive. The simulation in Fig. 5b indicates that some activation peaks that appear early will later vanish. Such peaks could correspond to patches of TSlS-positive cells that appear on the surface of the aggregate not associated with head or foot structures (Figs. 4e and 4f), or they could be correlated with the protrusion as a whole. CONCLUSION
The following conclusions may be obtained the results of the present work.
based on
(1) The temporal changes in the patterns of TS19 staining in G-aggregates and RG-aggregates can be satisfactorily explained in terms of changes in activation levels. (2) The same general patterning processes involving activation and inhibition properties which can explain head and foot regeneration appear, to a first approximation, to be sufficient to explain the de nova formation of an animal from tissue in which the established pattern has been destroyed. We are grateful to P. Bode, 0. Koizumi, T. Sugiyama, H. Ide, H. Shimizu, T. Itayama, and T. Awad for helpful suggestions about this work. One of the authors (M.S.) expresses her hearty gratitude to the hospitality extended to her during her stay in Developmental Biology Center, U. C. Irvine. This work was partially supported by the JapanUS Cooperation Program of Japan Society for Promotion of Science and by a grant from the National Institute of Health (USA; GM29130). REFERENCES BODE, H. R., DUNNE, J., HEIMFELD, S., HUANG, L., JAVOIS, L., KOIZUMI, O., WESTERFIELD, J., and YAROSS, M. (1986). Transdifferentiation occurs continuously in adult hydra. Current Topics Dev. Bill. 20, 257-280. BODE, P. M., and BODE, H. R. (1984). Patterning in hydra. In “Pattern Formation, a Primer in Developmental Biology” (G. M. Malacinski and S. V. Bryant, Eds.), pp. 213-241. MacMillan, New York. BODE, P. M., AWAD, T. A., KOIZUMI, O., NAKASHIMA, Y., GRIMMELIKHUIJZEN, C. J. P., and BODE, H. R. (1988). Development of the twopart pattern during regeneration of the head in hydra. Development 102,223-235. DUNNE, J. F., JAVOIS, L. C., HUANG, L. W., and BODE, H. R. (1985). A subset of cells in the nerve net of Hydra oligactis defined by a monoclonal antibody: Its arrangement and development. Dew. Biol. 109, 41-53. FLICK, K. M., and BODE, H. R. (1983). Dissociated tissues into cells and
420
DEVELOPMENTAL BIOLOGY
development of hydra from aggregated cells. In “Hydra: Research Methods.” (H. M. Lenhoff, Ed.), pp. 251-260. Plenum, New York. FRASER, S. E., GREEN, C. R., BODE, H. R., and GILULA, N. B. (1987). Selective disruption of gap junctional communication interferes with a patterning process in Hydra. Science 237,49-55. GRAF, L., and GIERER, A. (1980). Size, shape and orientation of cells in budding hydra and regulation of regeneration in cell aggregates. Wilhelm Rowx’s Arch, Dev. Biol l&X3,141-151. GIERER, A., BERKING, S., BODE, H., DAVID, C. N., FUNK, K. M., HANSMANN, G., SCULLER, H., and TRENKNER, E. (1972). Regeneration of Hydra from reaggregated cells. Nature New Biol 239,98-101. GIERER, A., and MEINHARDT, H. (1972). A theory of biological pattern formation. Kybernetik. 12,30-39. HEIMFELD, S., JAVOIS, L. C., DIJNNJZ,J. L., LITIZEFIEXD, C. L., HUANG, L., and BODE, H. R. (1985). Monoclonal antibodies: A new approach to the study of hydra development. Arch. Sci. Phys. Nat. 38,429-438. HICKLIN, J., and WOLPERT, L. (1973). Positional information and pattern regulation in hydra: Formation of the foot end. J. Embryo1 Exp. Mwphol. 30,727-740. HICKLIN, J., HORNBRUCH,A., and WOLPERT, L. (1975). Positional information and pattern formation in Hydra Dynamics of regions away from the boundary. J. E-01. Exp. Morph01 33,511-521. JAVOIS, L., WOOD, R. D., and BODE, H. R. (1986). Patterning of the head in hydra as visualized by a monoclonal antibody. Dew. Bid 117, 607-618.
VOLUME 141.1990
MACWILLIAMS, H. K. (1982). Numerical simulations of hydra head regeneration using a proportion-regulating version of the GiererMeinhardt model. J. Theor. Biol 99,681-703. MACWILUAMS, H. K. (1988a). Hydra transplantation phenomena and the mechanism of Hydra head regeneration. I. Properities of the head inhibition. De-u. Biol 96,217-238. MACWJLLIAMS, H. K. (198.3b). Hydra transplantation phenomena and the mechanism of Hydra head regeneration. II. Properities of the head activation. Dev. Biol 96,239-257. MACWIIUAMS, H. K., KAFATOS, F. C., and BOSSERT, W. H. (1970). The feedback inhibition of basal disk regeneration in Hydra has a continuously variable intensity. De-v. Biol 23, 380-398. NODA, K. (1971). Reconstitution of dissociated cells of hydra. (abstract in English). Zoo1 Magazine 80,99-101. SATO, M., and SAWADA, Y. (1989). Regulation in the numbers of tentacles of aggregated hydra cells. Deu. BioL 133,119-127. SIMIZU, H., and SAWADA, Y. (1987). Transplantation phenomena in hydra: Cooperation of positio-dependent and structure-dependent factors determines the transplantation result. Dev. BtiL 122,113119. WEBSTER,
C., and WOLPERT, L. (1966). Studies on pattern regulation in hydra. I. Regional differences in time required for hypostome determination. J. Embryo1 Exp. Morph01 16,91-104. WOLPERT, L., HORNBRUCH, A., and CLARKE, M. R. B. (1974). Positional information and positional signaling in hydra. Amer. ZooL 14,647663.
BIOLOGY
141,412-420 (1990)
Patterning Processes in Aggregates of Hydra Cells Visualized with the Monoclonal Antibody, TS19 MIKA SATO,’ HANS R. BODE, AND YASUJI SAWADA* Developmental Biology Center, Department of Developmental and Cell Biology, University of Calzfbrnia at Irvine, Irvine, Ca&fornia 92717, USA; and *Research Institute of Electrical Communication, Tohoku University, 2-1-l Katahira, Send& 980, Japan Accepted June 12, 1990 The monoclonal antibody, TS19, (Heimfeld et aL, 1985), labels the apical surface of ectodermal epithelial cells of tentacles and lower peduncles in Hydra To investigate the patterning process in a tissue whose original pattern was completely destroyed, the TS19 staining pattern was examined in developing aggregates of Hydra cells. Two types of aggregates were prepared. G-aggregates were made from tissue of the gastric portion of animals and RG-aggregates from gastric tissue allowed to regenerate for 24 hr before making aggregates. G-aggregates were initially TSlS-negative, and later dim and uniformly TSlS-positive. Thereafter, TS19 staining broke up into brightly stained and unstained regions. The brightly staining regions developed into head or foot structures. The TS19 pattern in RG-aggregates developed differently. Since the initial aggregates contained cells of regenerating tips, they started with TSlS-positive cells as well as TSlS-negative cells. The numbers of brightly staining TSlS-positive cells increased with time. Some patches of these cells developed into head or foot structures, while others did not. These results and a simulation using a reaction-diffusion model suggest that the changes in activation levels affected the temporal changes in the pattern of TS19 staining, and that the de nouo pattern formation in hydra can be explained in terms of a process involving activation and inhibition properties. o lgw Academic press, Inc. INTRODUCTION
A great deal is known about the processes underlying pattern formation in Hydra (see Bode and Bode, 1984 for review). As the animal has a simple body plan (a head and a foot on opposite ends of the body column, which is a tube), the central question is how the head and foot are patterned. The question is even more interesting because of the phenomenon of regeneration polarity. An excised piece of the body column will always regenerate a head at the original apical end and a foot at the original basal end. An extensive body of grafting and regeneration experiments have demonstrated that the primary processes involved in locating as well as forming the head and foot are two pairs of developmental gradients. One pair, of head activation and head inhibition, controls head formation (MacWilliams, 1983a,b) and the other pair with similar properties controls foot formation (MacWilliams et aZ., 1970; Hicklin and Wolpert, 1973). Further, the dynamics of the gradients explain the polarity of regeneration (Wolpert, 19’74; Bode and Bode, 1984). A large fraction of these data has been explained quite well in terms of a reaction-diffusion model (Gierer and Meinhardt, 1972; MacWilliams, 1982; MacWilliams, 198313). 1 Present address: Research Institute of Electrical Communication, Tohoku University, 2-l-l Katahira, Sendai 980, Japan. 0012-1606/90 $3.00 Copyright All rights
0 1990 by Academic Press, Inc. of reproduction in any form reserved.
More recently the use of region-specific monoclonal antibodies has refined the understanding of the processes. All of the previous work relied on the final structures formed to deduce the nature of the patterning processes. Two of the monoclonal antibodies stain regenerating heads long before any morphological structures have formed (Javois et aL, 1986; Heimfeld et al, 1985; Bode et aZ., 1986). By following the temporal changes in the staining patterns during regeneration, one can get an idea of the dynamics of the patterning processes involved, and thereby increase the precision of the deductions made about the processes. For example, an unanswered question concerned the order of patterning of the two parts of the head. The head consists of a hypostome at the apex above a ring of tentacles. The monoclonal antibody, TS19, recognizes an antigen which is restricted to the apical surface of the ectodermal epithelial cells of the tentacles and the lower part of the body column (Heimfeld et al, 1985; Bode et al, 1986). The temporal pattern of changes of TS19 staining during head regeneration indicated that the tissue was first committed to form tentacles, and later the hypostome (Bode et ak, 1988). All of this information has been obtained by perturbing, but not destroying, an established pattern. Hence, it is not known whether the discerned patterning processes are sufficient for setting up the pattern of head and foot at opposite ends of the animal, or if additional processes are necessary. One approach to this problem 412
SATO, BODE, AND SAWADA
Patterning
is to dissociate Hydra tissue into a suspension of cells, form aggregates of cells by centrifugation, and allow the aggregates to regenerate complete animals (Noda, 1971; Gierer et al, 1972). In this approach the original pattern is completely destroyed, and a new pattern must arise de nova. Not only does this provide a means for determining if other patterning processes are involved, but it also provides a different context for studying the known processes required for head and foot formation. The aggregation technique has been used to show that (1) in the absence of organized gradients new heads or single tentacles form in a spacing pattern which can be explained in terms of lateral inhibition, a prominent feature of existing models for Hydra patterning (Sato and Sawada, 1989), (2) tentacle formation is unstable unless associated with a developing head (Sato and Sawada, 1989), and (3) head activation levels can be maintained at the level of individual cells for a significant period of time (Gierer et al, 1972). To gain more information on the patterning processes operating in aggregates we have examined the temporal changes of the TS19 staining pattern, which will appear long before any morphological structures, during aggregates development. The results suggest that in the absence of an organized pattern the entire tissue of the aggregate initiates head and/or foot resulting in the observed spacing pattern of heads and feet. MATERIALS
Animals
AND
G-AGGREGATE ,
Dis.vr$tion Aggregation
RG-AGGREGATE Dissz;dion Aggregation
FIG. 1. Two types of aggregates used in the present work. G-aggregates were made from gastric tissue. RG-aggregates were made from gastric tissue allowed to regenerate for 24 hr before dissociation.
regions were pooled, dissociated, and aggregated to form aggregates in the size expected to produce one or two hypostomes after regeneration (Sato and Sawada, 1989). (2) RG-aggregates: aggregates made from gastric tissue allowed to regenerate for 24 hr before dissociation. The beginning of the regeneration period was defined here as the time when an aggregate was transferred from the dissociation medium at 4°C into DM at 18°C soon after centrifugation. The “steady state” of regeneration was defined as the condition in which the numbers of structures remained unchanged for more than 3 days, which was the same definition used previously (Sato and Sawada, 1989).
METHODS
and Culture
Hydra vulgar&x (formerly H. attenuata) were used for all experiments. Stock cultures of animals were maintained as described by Shimizu and Sawada (1987) and by Dunne et al. (1985). Animals used for experiments were starved for 1 day before use. Preparation
413
in Aggregates of Hydra Cells
of Aggregates
The procedure described previously (Sato and Sawada, 1989), based on the method of Flick and Bode (1983), was used for making aggregates. Briefly, tissue pieces were dissociated into a cell suspension in hyperosmotic medium (dissociation medium, DM; Flick and Bode, 1983), filtered through a nylon cloth with a mesh of 53 pm and aggregated by low-speed centrifugation (200g). The medium was gradually diluted according to the degree of regeneration, In the present work, two types of aggregates were prepared (Fig. 1). (1) G-aggregates: aggregates made from only the tissue of gastric region. The gastric region, the cross-hatched area of the animals in Fig. 1, of more than 50 animals were excised by removing the apical eighth and basal half of each animal. The gastric
Immunocytochemistry Live aggregates or animals were stained with the monoclonal antibody TS19 (Heimfeld et aL, 1985; Bode et al., 1986) using an avidin-biotin amplification procedure. Aggregates or animals were incubated in a 1:lOO dilution of a TS19 ascites fluid in Hydra culture medium (HM) (Dunne et al, 1985) for 5 min. After being washed three times in HM, they were incubated with 1:50 dilution of a biotinylated goat anti-mouse IgG (TAGO) in HM for 5 min, washed three times again, incubated with a 1:50 dilution of a FIT&labeled avidin (Vector Lab. Inc.) for 5 min, and, finally, rinsed three times in HM. To stain aggregates at an early stage before a cavity developed, dissociation medium (Flick and Bode, 1983) diluted 1:l with HM was used instead of HM, because aggregates at this stage tended to disintegrate in HM. The diluted dissociation medium did not affect TS19 staining as no difference was observed between samples stained in HM and diluted dissociation medium. Measurement
of Relative
The relative intensity TSlS-stained aggregates
Intensity
of Fluorescence
of fluorescence was quantitated
of areas of with a pho-
414
DEVELOPMENTALBIOLOGY VOLUME141,199O
tometer attached to a Zeiss photomicroscope equipped with epifluorescence. An excitation frequency of 365 nm and an emission frequency of 450 nm were used. Stained aggregates were relaxed in 2% urethan for less than 1 min, fixed in 70% ethanol, and flattened between two coverslips (Bode et aZ., 1988). Since the entire surface of the ovoid aggregates was stained, and the intensity of stain was maximal at the edges for geometric and optical reasons, measurements were made at the edges. The diameter (16 pm) of the measured area was defined by a shutter on the photometer. During early stages of regeneration when the stain was uniform across the aggregate, the fluorescence intensity was measured at four to eight arbitrarily chosen locations around the periphery of each of five aggregates. At later stages when the staining pattern was nonuniform, the intensities of all maxima and minima around the periphery were measured. Numerical
Simulations
The particular version of the reaction-diffusion model (MacWilliams, 1982) was used. In this version, the body of intact Hydra is divided into 21 parts (#l-#21), where #l-#5 are head parts and #6-#21 are body parts. In the simulation for G-aggregates, activator values of #lo-#21 were used to decide the initial values in aggregates. As for RG-aggregates, activator values of #lO#21 after 1440 iterations of calculation were used assuming the regeneration before dissociation, where #lO exhibited the regeneration tip. Aggregates were assumed as 42 columns. Here, column #1 was next to both columns #2 and #42. For each column, activator values of intact Hydra (#lo-#21) were randomly chosen 30 times and the averaged value of them was used as the initial activator value of the column. The calculated values were plotted per 355 iterations. The same value throughout the columns was used as the initial inhibitor value of the aggregate because of its big diffusion constant. RESULTS
Development
of Animals
from Two Types of Aggregates
Two types of aggregates (Fig. 1) were examined in the present work. One type, G-aggregates, was made from tissue of the gastric region, and therefore, contained no TSlS-positive cells when they were made. TSlS-positive cells are found only in the tentacles and lower peduncle of adult Hydra (Heimfeld et aL, 1985). The other type, RG-aggregates, was made from isolated gastric regions, that were allowed to regenerate for 24 hr before dissociation, Thus, RG-aggregates contained cells from regenerating tips which are TSlS-positive by 24 hr.
The overall regeneration of animals from aggregates was as previously described (Gierer et al., 1972; Graf and Gierer, 1980). Within 6-8 hr of pelleting the cells, ectodermal epithelial cells sorted out to the surface to form a continuous epithelium covering the entire surface. Within 24 hr, a two-layered hollow sphere had formed consisting of ectoderm and endoderm surrounding a fluid-filled cavity. By 2-4 days tentacles and hypostomes begin to form, and by 10 days complete animals had formed. There were two primary differences between the development of the two types of aggregates. First, tentacles appeared earlier in RG-aggregates than in G-aggregates. Second, in G-aggregates, most tentacles formed near a hypostome as part of a developing head. Few body tentacles, which are not associated with a head, formed. In contrast, many body tentacles were found in RG-aggregates during the early stages of regeneration. Later, the isolated body tentacles were absorbed, and most of the remaining tentacles were associated with developing heads. These differences between the two types of aggregates were observed previously (Sato and Sawada, 1989), although in a different species. Temporal Changes in the Pattern in G-Aggregate
of TS19 Binding
To examine the staining pattern of TS19 at various times during the development of G-aggregates, samples of aggregates at different stages of regeneration were stained with TS19 and visualized with indirect immunofluorescence. The temporal changes are shown in Fig. 2. By 6 hr, when the outer layer, the ectodermal epithelium, had formed, no TS19 staining was visible (Fig. 2a). By 30 hr when the aggregate had developed into a hollow sphere, the entire surface exhibited a dim uniform stain (Fig. 2b). A day later, the stain was still uniform but the intensity had increased (Fig. 2~). At this point the intensity was much less than that found in the tentacles and lower peduncle of intact animals. By the third day, the pattern began to change. The aggregates developed protrusions, some of which were clearly identified as tentacle bumps, which are evaginations that subsequently developed into tentacles. Further, the TS19 staining pattern was no longer uniform. Protrusions or tentacle bumps were brightly stained with TS19 whereas other regions were only dimly stained (Fig. 2d). Compared to the previous day, the intensity of the TS19 stain had increased on protrusions and had decreased in other areas. Over the next several days this trend continued. As the tentacles developed, the intensity of TS19 staining increased reaching that of complete animals (Fig. 2e). In regions that developed into the body column, the stain continued to decrease
Patterning
SATO,BODE,ANDSAWADA
and disappear to the background adult body column. Quantitative Temporal on G-Aggregates
level typical
of the
Changes in TS19 Binding
To obtain a quantitative and more detailed description of the changes in TSlS-binding pattern, the fluorescence intensity around the periphery of the aggregates was measured at different times. Figure 3 shows results of the quantitative measurements on G-aggregates stained with indirect immunofluorescence (see Materials and Methods). By 6 hr after aggregation when an ectodermal epithelium had formed, the level of fluorescence intensity was the background level observed in the body columns of intact unstained animals. Over the next 48 hr, it steadily increased loo-fold. By 52 hr, the aggregates changed from a uniform sphere, or ovoid, shape to an ovoid with protrusions (Fig. 2~). Within variation, all cells were uniformly stained at this time. By ‘78 hr, significant changes in staining intensities occurred. The developing tentacles were more intensely stained (crosses at 78 hr; Fig. 3) than the overall epithelium at 52 hr. Further, regions of the ectoderm that were not part of the protrusions or developing head structures decreased in intensity. Both of these trends continued over the next several days so that by the time the aggregates had reached a steady state in terms of the number of head structures to be formed (here, 10 days), the tentacles had acquired a staining intensity typical of tentacles in intact animals. The lower parts of the body column, the peduncles and feet had also formed on the aggregates. By the steady state, the peduncles had also reached the staining level typical of intact animals. Conversely, the staining of the regions of the epithelium destined to form the body column had dropped to the background levels typical of the body column of control. Temporal
Changes in TS19 Binding
on RG-Aggregates
Tentacles formed earlier and in larger number in RGaggregates compared to G-aggregates. These differences are easily attributed to the head regeneration processes that occur in the excised piece during the 24 hr before dissociation and aggregation (e.g., Bode and Bode, 1984). To determine if this 24-hr period of regeneration affected the general pattern of the TS19 staining, RG-aggregates were stained and observed at 6,30, 54, and 78 hr. Twenty four hours after excision isolated gastric regions exhibit TSlS-positive cells at the original apical and basal ends. Thus, it was not surprising that in 6-hr RG-aggregates there were isolated TSlS-positive cells
in Aggregates of Hydra Cells
415
in a background of unstained cells (Figs. 4a and 4b). By 30 hr the number of brightly staining TSlS-positive cells had increased, and were occasionally found in small patches (Fig. 4~). In addition there were dimly staining cells as well as many TSlS-negative cells. This is in strong contrast to G-aggregates at this time in which all cells were uniformly dimly stained (Fig. 2b). By 54 hr the size and number of patches had increased (Figs. 4d and 4e). Patches of labeled cells were next to patches of unlabeled cells (Fig. 4e), indicating that the labeling could be discontinuous instead of necessarily graded over an area. All tentacles and tentacle bumps that appeared were brightly stained (Fig. 4d). However, not all patches of TSlS-positive cells were associated with developing head or foot structures. Nor were all protrusions uniformly stained with TS19. The protrusion in the foreground of the aggregate in Fig. 4e has a patch of dimly stained, or unstained, cells at the tip. Further, the patch of stained cells below the tip extends away from the protrusion into a region of the aggregate where no head or foot structures seem to be forming. These observations will be dealt with in the Discussion. By 78 hr numerous intensely stained tentacles had formed (Fig. 4f). The patches of positive cells not associated with head or foot structures tended to disappear by this time. Numerical
Simulation
with a Reaction-R$sion
Model
Much of the regeneration and grafting data has been explained in terms of a reaction-diffusion model that postulates a stable activator molecule with a small diffusion constant, and an unstable inhibitory molecule with a large diffusion constant (Gierer and Meinhardt, 1972; MacWilliams, 1982). The parameters of the model were chosen to explain the dynamics of the gradients as exhibited in those experiments. It is of interest to know whether the particular version of the reaction-diffusion model (MacWilliams, 1982) can explain the observed TS19 results, since the temporal changes in the TS19 staining during regeneration are correlated with the changes in head activation (Bode et al, 1988). In the following only head formation will be simulated. In an intact body column, the head activation gradient is very shallow, and the level of head activation is very low compared to the head (MacWilliams, 1982). Thus, for G-aggregates it was assumed that the overall head activation level was initially low and uniform throughout the aggregate. This assumption is reasonable given the complete disruption of the tissue into cells, and thus, the disruption of the gradient in the process. In addition, it is known that the cells could retain their original positional memory such as head activation level (Gierer et ah, 1972). The simulation (Fig. 5a)
416
DEVELOPMENTALBIOLOGY
vOLUME141.1%0
FIG. 2. Temporal changes in the TSlS-binding pattern on G-aggregates visualized by indirect immunofluorescence on a whole mount preparation. To summarize the binding pattern, the periphery of samples were focused. (a) At 6 hr from the beginning of the regeneration period. (a, right) A bright field image of the same aggregate as the fluorescent image, (a, left). (b) At 30 hr. (c) At 52 hr. (d) At 78 hr. (e) At the steady state (lo-day), where arrows show the head (left) and the foot (right) of the animal. Bars show 0.1 mm.
indicates that there was an initial slow rise followed by a more rapid rise in activation that was uniformly distributed over the entire aggregate. This would correspond to the observed uniform changes in TS19 staining. Later, several peaks of activation emerged and grew to a maximum value, while other areas declined in activation. RG-aggregates were also simulated. Because RG-aggregates at 6 hr had a range of TS19 staining intensities, a larger range of head activation levels was chosen and these were distributed randomly along the lines of cells. The simulation (Fig. 5b) indicates that the rise to high values of head activation occurred earlier than in G-aggregates, and that the location of activation peaks was to some extent correlated with initially high values of
activation. Thus, assuming the temporal changes in TS19 reflect changes in head activation in developing G-aggregates and RG-aggregates, the changes are simulated by the model quite well. DISCUSSION
TSl9 ReJEects the Activation
Patterning
Process
Two pairs of developmental gradients, activation and inhibition gradients controlling head formation and foot formation, have been known in Hydra (see Bode and Bode, 1984 for review). Activation is a stable tissue property that is graded along the axis. The head activation gradient has a maximum in the head while the foot activation is maximal in the foot. In contrast, inhibition is
SATO, BODE, AND SAWADA
Patter&Q
in
AQQWQateS
of
417
Hydra cek!s
head activation measured in the regenerating tip (Bode et aL, 1988), and are not correlated with the dynamics of the head inhibition gradient. Behavior of Activatim of G-Aggregates
I j
1830 52 78 I+-% 240 REGENERATION TIME (hr)
FIG. 3. Quantitative measure of the temporal changes in TSlS fluorescence intensity in G-aggregates. Regeneration times on the abscissa show the time when the aggregates were treated with TSlS. The ordinate shows the amount of fluorescence (in arbitrary intensity units) measured with a photometer. Note that they are on the logscale. Values equal to or less than 0.2 are exhibited as “0.2.” Five aggregates were stained and measured for each time point. Each row of points at a time point represents a single aggregate. Closed circle, nonprotruding portions of aggregates. Closed triangles, protrusion portions. Crosses, tentacles or tentacle bumps. Open triangles, lower peduncles. Open circles, body column. The mean value of intensity measured in tentacles, lower peduncles, and body columns of stained intact animals are shown in A, and those of body columns of unstained animals are shown in B (bars indicate the standard deviations).
an unstable property with a graded distribution. Head inhibition is produced in the head, is graded with maximum in the head, and is transmitted by diffusion down the body column, possibly through gap junctions (Fraser et aL, 1987). Foot inhibition, which has similar properties, is produced in the foot. Following decapitation, head inhibition levels fall rapidly throughout body column. Head activation at the apical tip rises to a head level within 16 hr (see Bode and Bode, 1984 for review). The tissue at the tip is then determined to make a head, produces the head inhibition substance to prevent other regions from forming head, and undergoes head morphogenesis and differentiation (e.g., Webster and Wolpert, 1966; Hicklin et a& 1975). A similar process occurs upon removal of the foot. Temporal changes in the TS19 staining pattern following decapitation are closely correlated with the rise in
Processes during
Regeneration
During head regeneration following decapitation (Bode et CCL,1988) and the regeneration of head and foot in an excised piece of tissue from gastric region (O’Neill, Awad, and Bode, unpublished experiments), TS19 staining observed is confined to the regenerating end(s). It does not appear in other regions at any time during regeneration. The staining, which initially appears as a dim stain and increases in intensity with time, correlates with the local rises in activation (Bode et al, 1988). In contrast, the TS19 staining in the G-aggregates is not confined to specific areas when it becomes clearly discernible. Instead, it is uniformly distributed over the entire surface of the aggregate (Fig. 2b). The succeeding temporal pattern is similar to that seen during head regeneration following decapitation (Bode et al, 1988). Assuming that the TS19 antigen is related to the activation process, the temporally changing TS19 patterns in G-aggregates (Figs. 2 and 3) could be interpreted as follows. (a) Head (and probably foot) activation levels are rising all over the aggregate. (b) The increasing intensity of staining reflects an increasing rate of change of the activation level. (c) When protrusions and tentacle bumps appear activation levels continue to rise there, but in areas where the stain is lessening or has disappeared, activation no longer rises. These qualitative results were simulated using MacWilliams’ version of a reaction-diffusion model (MacWilliams, 1982) that has been applied to the regeneration in hydra. The observed changes were consistent with rise and fall in TS19 staining intensity in different areas starting around 3 days. Thus, assuming that the temporal changes in TS19 reflect changes in head activation in the developing aggregate, these changes fit the model quite well. This interpretation has limits. It is unlikely that the very high level of the TS19 antigen in the adult tentacles has anything to do with activation properties. Adult tentacles are not capable of inducing secondary axes or regenerating other parts of the animal. Instead, it is more likely that the antigen also plays a role in the physiology of the adult animal. Behavior of Activation of RGAggregates
Processes during Regeneration
The differences in the regeneration of the RG-aggregates and of the G-aggregates can be explained in terms
418
DEVELOPMENTAL BIOLOGY
VOLUME 141.1990
FIG. 4. Temporal changes in the TSlS-binding pattern in RG-aggregates. (a and b) At 6 hr. The periphery of the sample was focused in a, while b is higher magnification of another sample. (c) At 30 hr. (d and e) At 54 hr. (f) At 78 hr. Arrow shows a tentacle expected to be a “body tentacle.” Bars show 0.05 mm.
SATO, BODE, AND SAWADA
Patterning
a
No.
posit
posit
ion
ion
FIG. 5. Evolution of head activator for (a) G-aggregates and (b) RG-aggregates; results of the simple numerical simulation using a reaction-diffusion model (MacWilliams, 1982). Column No. 1 was next to both columns No. 2 and No. 42. See text.
of differences in head activation levels and their distributions. (Differences in foot activation levels and distribution also occurred, but to keep it simple only the head will be considered.) Since the gastric regions used for RG-aggregates regenerated for 24 hr before making aggregates, the activation level at the apical end rose to the level of the head (e.g., MacWilliams, 1983b). Hence, instead of starting with a population of cells that has levels of head activation very similar to those in G-aggregates, the head activation levels differed greatly among the cells from the beginning. Because the cells would retain their activation level for a while (Gierer et al., 1972), the distribution of head activation over the surface of the aggregate would be heterogeneous. The changes in the TS19 staining pattern in RG-aggregates shown in Fig. 4 can be explained as follows. The initial aggregates contained bright TSlS-positive cells isolated in a background of unstained cells (Figs. 4a and 4b). Since 24 hr regenerating tips have bright TSlS-positive cells; this is not surprising. More telling is the pattern observed at 30 hr (Fig. 4~). The number of bright TSlS-positive cells has increased and these cells are clustered. The early appearance of head structures in RG-aggregates could be due to the sorting out of cells with high-level head activation into areas where they rapidly formed a tentacle or, if enough, a complete head. However if the loose clustering were simply due to sorting out of cells with a high head activation level, the number would not be expected to increase. Therefore, it is deduced that small numbers of high activation level cells that by chance were in one area could, through a metabolic process such as a reaction-diffusion mechanism, raise the head activation level of sufficient surrounding cells to form a head structure. The patches of very bright cells found in the tentacle protrusions at
in Aggregates of Hydra Cells
419
54 hr (Figs. 4d and 4e) provide a vivid correlation between high activation levels and the appearance of structure. Simulation of these results with MacWilliams’s model (1982) showed that the model correlates well with the data. The protrusion in Fig. 4e has a patch of TSlS-positive cells as well as a patch of TSlS-negative cells near the apex. If the protrusion were to form a tentacle, one would expect the entire protrusion to be TSlS-positive. The simulation in Fig. 5b indicates that some activation peaks that appear early will later vanish. Such peaks could correspond to patches of TSlS-positive cells that appear on the surface of the aggregate not associated with head or foot structures (Figs. 4e and 4f), or they could be correlated with the protrusion as a whole. CONCLUSION
The following conclusions may be obtained the results of the present work.
based on
(1) The temporal changes in the patterns of TS19 staining in G-aggregates and RG-aggregates can be satisfactorily explained in terms of changes in activation levels. (2) The same general patterning processes involving activation and inhibition properties which can explain head and foot regeneration appear, to a first approximation, to be sufficient to explain the de nova formation of an animal from tissue in which the established pattern has been destroyed. We are grateful to P. Bode, 0. Koizumi, T. Sugiyama, H. Ide, H. Shimizu, T. Itayama, and T. Awad for helpful suggestions about this work. One of the authors (M.S.) expresses her hearty gratitude to the hospitality extended to her during her stay in Developmental Biology Center, U. C. Irvine. This work was partially supported by the JapanUS Cooperation Program of Japan Society for Promotion of Science and by a grant from the National Institute of Health (USA; GM29130). REFERENCES BODE, H. R., DUNNE, J., HEIMFELD, S., HUANG, L., JAVOIS, L., KOIZUMI, O., WESTERFIELD, J., and YAROSS, M. (1986). Transdifferentiation occurs continuously in adult hydra. Current Topics Dev. Bill. 20, 257-280. BODE, P. M., and BODE, H. R. (1984). Patterning in hydra. In “Pattern Formation, a Primer in Developmental Biology” (G. M. Malacinski and S. V. Bryant, Eds.), pp. 213-241. MacMillan, New York. BODE, P. M., AWAD, T. A., KOIZUMI, O., NAKASHIMA, Y., GRIMMELIKHUIJZEN, C. J. P., and BODE, H. R. (1988). Development of the twopart pattern during regeneration of the head in hydra. Development 102,223-235. DUNNE, J. F., JAVOIS, L. C., HUANG, L. W., and BODE, H. R. (1985). A subset of cells in the nerve net of Hydra oligactis defined by a monoclonal antibody: Its arrangement and development. Dew. Biol. 109, 41-53. FLICK, K. M., and BODE, H. R. (1983). Dissociated tissues into cells and
420
DEVELOPMENTAL BIOLOGY
development of hydra from aggregated cells. In “Hydra: Research Methods.” (H. M. Lenhoff, Ed.), pp. 251-260. Plenum, New York. FRASER, S. E., GREEN, C. R., BODE, H. R., and GILULA, N. B. (1987). Selective disruption of gap junctional communication interferes with a patterning process in Hydra. Science 237,49-55. GRAF, L., and GIERER, A. (1980). Size, shape and orientation of cells in budding hydra and regulation of regeneration in cell aggregates. Wilhelm Rowx’s Arch, Dev. Biol l&X3,141-151. GIERER, A., BERKING, S., BODE, H., DAVID, C. N., FUNK, K. M., HANSMANN, G., SCULLER, H., and TRENKNER, E. (1972). Regeneration of Hydra from reaggregated cells. Nature New Biol 239,98-101. GIERER, A., and MEINHARDT, H. (1972). A theory of biological pattern formation. Kybernetik. 12,30-39. HEIMFELD, S., JAVOIS, L. C., DIJNNJZ,J. L., LITIZEFIEXD, C. L., HUANG, L., and BODE, H. R. (1985). Monoclonal antibodies: A new approach to the study of hydra development. Arch. Sci. Phys. Nat. 38,429-438. HICKLIN, J., and WOLPERT, L. (1973). Positional information and pattern regulation in hydra: Formation of the foot end. J. Embryo1 Exp. Mwphol. 30,727-740. HICKLIN, J., HORNBRUCH,A., and WOLPERT, L. (1975). Positional information and pattern formation in Hydra Dynamics of regions away from the boundary. J. E-01. Exp. Morph01 33,511-521. JAVOIS, L., WOOD, R. D., and BODE, H. R. (1986). Patterning of the head in hydra as visualized by a monoclonal antibody. Dew. Bid 117, 607-618.
VOLUME 141.1990
MACWILLIAMS, H. K. (1982). Numerical simulations of hydra head regeneration using a proportion-regulating version of the GiererMeinhardt model. J. Theor. Biol 99,681-703. MACWILUAMS, H. K. (1988a). Hydra transplantation phenomena and the mechanism of Hydra head regeneration. I. Properities of the head inhibition. De-u. Biol 96,217-238. MACWJLLIAMS, H. K. (198.3b). Hydra transplantation phenomena and the mechanism of Hydra head regeneration. II. Properities of the head activation. Dev. Biol 96,239-257. MACWIIUAMS, H. K., KAFATOS, F. C., and BOSSERT, W. H. (1970). The feedback inhibition of basal disk regeneration in Hydra has a continuously variable intensity. De-v. Biol 23, 380-398. NODA, K. (1971). Reconstitution of dissociated cells of hydra. (abstract in English). Zoo1 Magazine 80,99-101. SATO, M., and SAWADA, Y. (1989). Regulation in the numbers of tentacles of aggregated hydra cells. Deu. BioL 133,119-127. SIMIZU, H., and SAWADA, Y. (1987). Transplantation phenomena in hydra: Cooperation of positio-dependent and structure-dependent factors determines the transplantation result. Dev. BtiL 122,113119. WEBSTER,
C., and WOLPERT, L. (1966). Studies on pattern regulation in hydra. I. Regional differences in time required for hypostome determination. J. Embryo1 Exp. Morph01 16,91-104. WOLPERT, L., HORNBRUCH, A., and CLARKE, M. R. B. (1974). Positional information and positional signaling in hydra. Amer. ZooL 14,647663.
