Adhesion of Dictyostelium discoideum cells to carbohydrates immobilized in polyacrylamide gels. I. Evidence for three sugar-specific cell surface receptors.

Dictyostelium discoideum cells appear to be able to recognize particular carbohydrate prosthetic groups at different stages in their life cycle. We therefore used our previously developed model system (consisting of polyacrylamide gels containing putative ligands covalently linked to the polymer) to determine the receptors on these cells capable of recognizing carbohydrates. D. discoideum cells, at different developmental stages from growth phase to late aggregation, were incubated with the derivatized gels, and the number of adherent cells was determined by measuring alanine transaminase after cell lysis. From 70 to 100% of the cells firmly adhered to gels derivatized with glucose, maltose, or cellobiose. The cells were also capable of binding to N-acetylglucosamine and mannose, but both the rate and the extent of binding to these sugars were less than those observed with the glucose derivatives. Furthermore, binding to N-acetylglucosamine decreased to negligible levels during the aggregation stage of development. The cells did not bind to the glucose-derivatized gels in the presence of glucose and a variety of carbohydrates containing glucose at the nonreducing termini, whereas binding was not inhibited by N-acetylglucosamine, mannose, and derivatives of these sugars. Adhesion to all sugars was blocked by 2,4-dinitrophenol. This inhibitor also reversed the binding to gels containing N-acetylglucosamine and mannose, but not to glucose. Differential binding to the three monosaccharides was also observed under conditions affecting the normal amoeboid shape of the cells. In addition, adhesion to N-acetylglucosamine and mannose was trypsin-sensitive, whereas adhesion to glucose was only slightly affected by treating the cells with trypsin (and cycloheximide). These and other results suggest that D. discoideum cell adhesion to derivatized gels is mediated by three different receptors, one highly specific for glucose and two (probably less specific) for N-acetylglucosamine and mannose.

In the present experiments, starved cells were placed on these and other immobilized sugars to determine whether the sugar derivatives influenced normal development in this organism. When D. discoideum cells are on a solid surface under water, they form aggregation centers and strands of cells (which radiate from the center), send "signals" i.e. pulses of cyclic AMP from the center down the strands, and finally, after cells in the strands migrate to the center, form tight aggregates. These results were obtained on all polyacrylamide gel derivatives tested except one class, derivatives of D-glucose (0and S-glucosides, cellobiosides, and maltosides). On these gels, aggregation centers and strands formed normally, but at a certain point stopped "signaling" and suddenly dissociated, with the cells rapidly migrating away from one another by negative chemotaxis (see Appendix to this report). Furthermore, a simultaneous dissociation of several centers was often observed.
Following a brief period of random movement after dissociation, aggregation centers once again formed and the cycle was repeated. This cycle was repeated as often as 30 times or more over a 24-h period. The cells on the glucoside gels became aggregationcompetent at the same time as the control cells, and the adhesion-dissociation cycle appeared to have no effect on the synthesis of some developmentally regulated proteins, such as UDP-glucose pyrophosphorylase.
Interpretations of the phenomenon and its potential for studying gene regulation in this organism are discussed.
In a study of the interactions between cells of the slime mold Dictyostelium dkcoideum and analogues of cell surface * This work was supported by Grant CA 21901 from the National Institutes of Health and by the Deutsche Forschungsgemeinschaft. This is Paper 1221 from the McCollum-Pratt Institute. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. complex carbohydrates, we showed that binding to D-glUCOSe and glucosides, N-acetyl-D-glucosamine, and D-mannose derivatives was apparently mediated by three separate receptors (1). We also showed that the receptor responsible for the binding to N-acetylglucosamine disappeared as the cells developed from vegetative to "aggregation-competent.'' This and other observations (see accompanying paper (1)) suggested to us that carbohydrates might play a regulatory role during development of the organism. The present paper describes studies designed to determine whether cell contact with carbohydrate-derivatized gels affects differentiation of D. discoideum.
As previously described by Gerisch (2) and shown in Fig. 1, A and B, when maintained at an air-water interface, the cells normally undergo a specific cycle of development. About 5 h after the removal of nutrients, they become aggregationcompetent, show chemotaxis toward cyclic AMP, and form aggregation centers followed by tight aggregates ( A ) , proceeding to slugs and fruiting bodies (B). If the cells are maintained under buffer, however, development is normally arrested at the tight aggregate stage (A, panel 9). In the present studies, we found that normal development was observed a t both airwater interfaces and under buffer on all gels tested, except those derivatized with glucose or glucosides. Under buffer on glucoside gels, the cells formed aggregation centers, but were unable to proceed to the formation of tight aggregates. Instead, both centers and streams abruptly disintegrated to single cells, which underwent the same process of aggregation and dissociation many times. Speculations on the mechanism of this unexpected behavior are presented under "Discussion." EXPERIMENTAL PROCEDURES'

RESULTS
Development of D. discoideum on Deriuatized Polyacrylamide Gels-We have reported (1, 10) that D. dkcoideum attaches nonspecifically to polyacrylamide and to the derivatized polyacrylamide gels at low ionic strength (phosphate buffer, 0.017 M, pH 6.0). Cells were therefore permitted to attach to the different gels, and the developmental process was followed by time-lapse cinematography for periods of 24 h or longer. rpm on a gyratory shaker. After 1 h, a sample was taken, washed once, and resuspended in phosphate buffer a t a density of 5 X 106/ml, and 50-pl aliquots were transferred to derivatized gels. After the cells had attached to the gels, 6 ml of phosphate buffer were pipetted into the dish to cover the gels with a layer of buffer. T o allow complete differentiation in the control, after aggregation (between 8 and 9 h), the buffer was carefully aspirated to the level of the gels, giving the desired air-water interface. A and B, Control gels, copolymer of 6-acrylamidohexanol and acrylamide. The number in each panel is the time in hours after removal of the bacteria. Normal development of the slime mold D. discoideum on a solid surface proceeds through the following cycle. Vegetative cells become aggregation-competent about 5 h after removal of nutrients. Certain cells in the population then become foci of "aggregation centers" by emitting pulses of cyclic AMP. The remaining cells respond to this chemotactic signal by streaming toward the centers, forming strands of cells. At this stage, waves of movement are evident in the strands in response to repetitive cyclic AMP signals emanating from the aggregation centers ("signaling"). When essentially all of the cells have migrated into the center, a "tight" aggregate is formed. It is important to emphasize that the term "tight aggregate" used here and throughout this paper is meant to designate clusters of cells of the type shown in panel 9. Tight aggregates formed both under water and at an air-water interface are distinguishable from aggregation centers shown in panel 6, because the latter contain strands of migrating cells. When observed under identical conditions, the two types of aggregates show only minor morphological differences. The difference in the two panels is an optical effect resulting from the fact that panel 6 is a micrograph of an aggregate under water, while 9 is a micrograph at an air interface. The tight aggregate develops a tip and may form a slug or may directly form a stalk and fruiting (spore) body. The slug ultimately attaches to the surface and forms a stalk and fruiting body containing spores. The cycle is complete when the spores germinate to form vegetative cells. As noted above, the complete sequence from vegetative cells to mature fruiting bodies ( A and E) requires that the cells he exposed to an air-water interface, whereas when under water, they develop only through the formation of tight aggregates ( A ) . C, Cellobiose-derivatized gels. These photographs of cells under buffer were taken a t 24 h. The bur in each photograph = 100 pm. No significant differences were noted on any of the gels when the cells were exposed to an air-water interface. Fig. 1, A and B presents typical results, showing development of an aggregation center (6 h), tight aggregates (9 h), a slug and a tight aggregate with a tip (13 h), and mature stalks and fruiting bodies (20 h).
In the second set of experiments, the cells were placed on various polyacrylamide gels, and permitted to develop under buffer. Under these conditions, normal development proceeds for about 5 h at which time the cells become aggregationcompetent. At this point, the cells form aggregation centers and cell strands (Fig. lA, panel 6 ) , and then develop further to the tight aggregate stage (Fig. lA, panel 9). These results were obtained with all of the gels tested with the exception of glucoside gels, as described below. The only other difference was that on mannose-derivatized gels the cells seemed to be delayed for about 2 h in forming tight aggregates.
Quite different results were observed with gels derivatized with glucose, maltose, or cellobiose. In these cases, aggregation centers and strands were formed in the usual manner, but the cells never formed tight aggregates. Instead, the cells were arrested at the aggregation center stage. Typical results on cellobiose-derivatized gels are shown in Fig. 1B adhesive bonds. This latter property is due to the formation of specific cell surface components, thought to be glycoproteins (11,12), which are designated as contact sites A (51.2 Thus, it is possible that cells in contact with glucoside gels were unable to develop such sites. As can be seen in Fig. 2, however, cells on cellobiose gels developed contact sites A in the normal manner, and did not lose these sites over 24 h, the duration of the experiment. In addition, contact with the cellobiose gels did not affect normal chemotaxis toward cyclic AMP. Cells incubated on cellobiose gels for 5,10, and 24 h, and challenged by treatment with capillaries filled with 0.1 mM cyclic AMP, reacted immediately by moving toward the capillaries (data not shown).
Time-Lapse Studies: Dissociation of Aggregation Centers on Glucoside Gels-A more detailed analysis of the behavior of the cells on the glucoside and control gels was undertaken with time-lapse cinematography.
Cells washed free of bacteria were transferred to polyacrylamide gels derivatized with different ligands, and followed for periods generally up to 24 h (sometimes as long as 50 h). I n these experiments, the cells were maintained under buffer. During the first 4 h of starvation, there were no apparent differences in the behavior of the cells on any gel. The cells moved randomly, and when they touched each other they did not form stable adhesive bonds; this behavior was identical * Vegetative cells are also capable of adhering to one another, but these bonds are sensitive to EDTA and the cell surface components responsible for this process are designated contact sites B (5). Contact sites B persist throughout development.
to that shown by cells on glass coverslips.
In the period 4-6 h of starvation, the typical pattern of early aggregation was established. Cells elongated and collected into aggregation streams and centers, and end-to-end contacts were formed; the latter are considered to be typical of aggregation-competent cells (Fig. 3, panel 0). The strands sometimes appeared thicker on the glucose compared to the control gels. Also, the aggregation centers adhered more tightly to these gels in that the aggregation centers on the glucose gels could not be shaken off, whereas they were easily released by similar shaking of the control gels. Aggregation was virtually complete on the control gels at about ts: at which time, most of the strands and cells had collected into aggregation centers.
It was at this stage that a surprising effect was found with the cells on glucoside gels. Typical results are shown in Fig.  3. The aggregation center continued to increase in size, and the strands became thicker for about 18 min. At about 21 min, there was an abrupt change, and the center began to dissociate, and by 33 min the center had totally dissociated.
The dissociation of aggregation centers was studied in detail4 and the observations are summarized as follows. (i) The abbreviation t is used to designate the time after removal of the cells from the bacteria; the subscript numbers refer to time in nours after removal of the bacteria.
The process shown in Fig. 3 and descrmea above should not be confused with the dissociation of aggregation centers which is sometimes observed when two centers me close to one another. In the latter, the centers compete for cells and disintegration of one center Dissociation generally started at aggregation centers and propagated from the centers down the strands (Fig. 3). In many instances, dissociation was observed a t centers while strands were still growing distally. (ii) Dissociation, which was a rapid, active process ( Fig. 3), was observed throughout the center, often preceded by intense cell movement and pseudopodial activity, after which the cells migrated away from the centers and one another. Dissociation of an entire center usually took 5-15 min (Fig. 3). (iii) The response of the cells resembled negative chemotaxis; cell migration in the center was radial and away from the center, while it was lateral in the strands. Most of the cells became single cells, but some small clusters persisted for a longer period of time. Once separated, the single cells migrated more or less randomly on the gels, so that eventually they were approximately evenly distributed over the surface. Evidence that the initial rapid dissociation resulted from negative chemotaxis is given in the Appendix to this paper. (iv) The process was frequently found to be synchronous for many centers over a wide area, and sometimes over the entire gel. (v) Once the centers had dissociated, the cells migrated randomly for varying periods of time, and then began to form new aggregation centers and strands. As before, the newly formed centers were incapable of developing further to tight aggregates, and abruptly dissociated. This process of forming and reforming aggregation centers was repeated many times by the cells on a single gel. A more detailed description of this behavior is given in the following section. (vi) Dissociation was not due to loss of viability of the cells; even after 50 h on the gels, we were unable to detect round, nonmotile cells in any of the films examined.
Aggregation-Dissociation Cycle on Gels Derivatized with Glucosides-Since the process of aggregation-dissociation-reassociation continued for a t least 50 h on the glucoside gels, it was of interest to determine whether the time course of these processes changed with increasing time of incubation on these gels. The results obtained with five different cell preparations are given in Table I. We measured the length of time that the aggregation centers remained, the signal period (the time between which two successive pulses migrated down the strand), the time for disintegration of the centers, and the lag period before initiation of new centers.
The results were highly variable for each of these parameters, but nevertheless, certain trends are apparent (Table I).
With increasing time of incubation of the cells on the gels, each phase in the cycle appeared to become shorter. It was possible that the centers (or cells acting as aggregation centers) become progressively less capable of spontaneously secreting cyclic AMP as they repeatedly proceeded through the cycle, which would account for the shorter lifetimes of the centers and the dissociation times.
Effect of Exogenous Cyclic AMP-The data presented above show that regardless of the length of time that the cells were in contact with glucoside-derivatized gels, they remained capable of secreting pulses of cyclic AMP. Close study of the time-lapse films provided one clue to what is undoubtedly a complex process. Just before onset of dissociation, the optically refractile changes which traveled in waves from the center through the strands suddenly stopped. Based on earlier studies of others (13)(14)(15)(16)(17), this is probably because the centers stopped emitting pulses of cyclic AMP. If this were the case, exogenously supplied cyclic AMP would be expected to prevent dissociation. Therefore, in three separate experiments, a may occur as cells migrate from it (and its associated strands) to the strands of a second center. This is quite different from the process which takes place on the glucoside gels.

TABLE I
Aggregation-dissociation cycles of D. discoideum on cellobiosederivatized gels Cells were placed on cellobiose-derivatized gels and observed by time-lapse cinematography as described under "Experimental Procedures." The films were analyzed to obtain the values reported below. The data are presented as a function of time of incubation of the cells on the gels a t room temperature (first column). The following designations are used "lifetime of aggregation centers," time fromqnitial spontaneous pulsing of a cluster of cells until beginning of dissociation; "signal period," time interval between two waves moving down the strands from the center; "dissociation time," the time for complete dissociation of a center (not necessarily including its strands) from the point when the center stops pulsing until it is completely dissociated; and "interphase lag," the time from dissociation of a center to the formation of a new center from the same cell population. The data presented give the ranges of the measurements using five different cell preparations. capillary filled with 1 mM cyclic AMP solution was positioned near t8 cells, which, although they were not actively engaged in aggregation, had already been through one or two cycles of aggregation-dissociation. The cyclic AMP solution diffusing from the capillary caused the cells to migrate to the capillary tip, and within 45 min a dense cluster of cells was gathered around the tip, with strands of cells evident at the periphery of the cluster (Fig. 4). If the capillary was removed a t this point, the cells continued to aggregate for about 5 min and then dissociated from one another. If, on the other hand, the capillary containing the cyclic AMP was left in continuous contact with the cell mass for many hours (generally overnight), a large aggregate was formed which resembled a tight aggregate, and which did not immediately dissociate upon removal of the capillary. Eventually, however, cells at the periphery of these aggregates migrated to neighboring aggregation centers.
One additional observation was in accord with the hypothesis that dissociation of centers on glucoside gels under buffer takes place because the cells stop signaling. As noted above, microscopic observation suggested that glucoside gels did not influence the developmental process of cells at an air-water interface. Time-lapse cinematography was therefore used to confirm this conclusion. Cells on the glucoside gels were placed in plastic multiwells as described under "Experimental Procedures." Under these conditions, cells at the center of the gel were at the air-water interface while those located peripherally were under a layer of buffer. As expected, the cells at the air-water interface completed the process of differentiation, while those under buffer did not. More important, the time-lapse films showed that the aggregation centers at the meniscus continued to send the pulsating signals without interruption, while the centers under buffer stopped signaling just before dissociation of the centers.
Since cells at the interface are exposed to higher concentrations of O2 than cells under buffer, attempts were made to increase the oxygen tension of cells under buffer. Cells on glucoside gels, but under phosphate buffer, were subjected to pure oxygen atmospheres for as long as 15 h, but they displayed repetititve cycles of aggregation-dissociation, similar to controls.
We also tested the effects of inhibitors of oxidative phosphorylation, such as 2,4-dinitrophenol, on the aggregationdissociation process. In these experiments, 5 x lo6 aggregation-competent cells/ml of phosphate buffer were incubated on glass coverslips; the experimental solutions contained from 0.1 to 50 ~L M 2,4-dinitrophenol. We could not find a 2,4dinitrophenol concentration which induced the aggregationdisintegration process described above. At concentrations below 10 ~L M inhibitor, cells aggregated normally, whereas at concentrations of 12 ~L M or above they became round and did not aggregate at all. These results agree with an earlier report of Gerisch (18) who found that any concentration of dinitrophenol which inhibited cell aggregation also caused the cells to become round.
Effect of Glucoside Gels on the Synthesis of UDP-glucose Pyrophosphorylase-A commonly used marker for monitoring the development of slime mold cells is the enzyme UDPglucose pyrophosphorylase. Increased activity of this enzyme accompanies differentiation of stalk and spore cells, and it is one of the key enzyme markers of the postaggregation period of differentiation (19-25). UDP-glucose pyrophosphorylase activity was therefore measured in cells kept under water and in contact with cellobiose and control gels. The results, presented in Fig. 5, show no significant differences between the two sets of cells, indicating that contact with the cellobiose gels did not interfere with the synthesis of UDP-glucose pyrophosphorylase.

Is Dissociation of Aggregation Centers Related to the
Strength of Cell-Gel Adhesion?-If the cells adhered avidly to the glucoside gels, it seemed possible that cell-gel binding might compete with cell-cell binding and prevent the normal formation of tight aggregates.
We tested this possibility by examining aggregate formation on solid surfaces treated with poly-L-lysine and bovine serum albumin to which the cells are known to bind very t i g h t l~.~ The cells behaved normally on both the poly-L-lysine and albumin substrata. They completed the developmental process beyond the formation of tight aggregates, and the time course was about the same as observed with control cells. Although we have not yet measured the relative strengths of adhesion of the cells to the glucoside derivatives compared to poly-L-lysine and albumin, it appeared that binding to each of these substrata was comparable, and if so, then tight binding to the substratum cannot explain the effect seen with the glucoside gels. If quantitative binding affinity measurements support this conclusion, it appears that the aggrega-S. Bozzaro, personal observations.

Effects of D-Glucoside Derivatives on D.
discoideum Development tion-disintegration cycle is somehow triggered by specific sugars in the substratum.

Effects of Free Sugars and Thioglucoside-derivatized
Gek-In all of the experiments described above, the gels were derivatized with either glucose, maltose, or cellobiose linked to the gels via 0-glycosidic bonds. Since the cells produce both a-and p-glucosidases (26, 27), it was possible that the sugars were hydrolyzed from the gel sites where they were in contact with the cells. The liberated sugar could then be taken up and metabolized, thus reversing the differentiation process. High concentrations of metabolizable sugars do, in fact, induce dedifferentiation (28, 29) and inhibit aggregation (30-33). (It should, however, be stressed that there is no clear-cut sugar specificity in these phenomena, and the high concentrations of sugars needed for inhibition have toxic effects on the cells (32)). 5 To determine whether hydrolysis of the 0-derivatives with production of free sugars could account for the observed effects on aggregation, we utilized gels derivatized with Dthioglucopyranoside (34), which is similar to the corresponding 0-glucoside except that thioglycosides are not hydrolyzed by glycosidases. The results obtained with the thioglucoside were identical with those obtained with the 0-glucosides. In addition, we tested the effect of a number of free sugars in solution on cells on glass coverslips or on control gels. Glucose, maltose, cellobiose, mannose, galactose, and lactose were tested at concentrations ranging from 5 to 100 mM with the cells under buffer. No inhibition of tight aggregate formation was observed a t concentrations below 50 mM. Above 50 mM, the sugars affected the shape of the cells and clear-cut results were not obtained.
In summary, the results with the free sugars were entirely different from those obtained with the glucoside-derivatized gels. At concentrations in the range that did not affect osmolarity, the free sugars (including glucose and cellobiose) showed no effect at all. At extremely high concentrations (relative to the osmolarity of the phosphate buffer) aggregation was inhibited for the most part, but the cycle of events was very different from that seen on glucoside gels or during the normal cycle. Moreover, no clear-cut sugar specificity was found.
All of these results taken together indicate that the association-dissociation phenomenon described above cannot be ascribed to release and utilization of the active sugars from the gels, and it was the sugars bound to the gels which caused disintegration of the aggregation centers.

DISCUSSION
These studies were designed to investigate the effect(s) of cell contact with sugar-derivatized gels on the development of D. discoideum. We found that cells proceeded through a normal cycle of development at an air-water interface on all gels tested. When the gels were maintained under buffer, normal development to tight aggregates was observed on all gels tested except those derivatized with glucose and glucosides. On these gels, the cells formed aggregation centers and strands of cells but were unable to proceed to form tight aggregates. Rather the centers and strands abruptly dissociated into single cells; the latter reassociated into aggregation centers and again dissociated. This cycle was repeated many times over a 24-h period, and the only change observed was a slight tendency for all phases of the cycles to become shorter. The phenomenon was dependent upon cyclic AMP; not only did the cells stop sending pulses of this nucleotide to the attached strands before they dissociated, but also dissociation was not observed if the cells were continuously exposed to an exogenous source of cyclic AMP. Dissociation appears to involve negative chemotaxis (see Appendix) which begins when the cells stop sending pulses of cyclic AMP and possibly start producing a negative chemotactic agent.
The observations reported here are not a general phenomenon exhibited by all slime molds since it was not detected with Polysphondylium pallidurn, which formed tight aggregates on all gels tested (under buffer). However, we do not know whether P. pallidum contains the glucose-specific receptor detected in D. discoideum.
There is evidence that dissociation of centers is a normal part of the process of development of D. discoideum. That is, behavior similar to that described here, a spontaneous dissociation of aggregation centers, has been reported by Bonner (35) and particularly by Shaffer (36). The only difference is that in the normal sequence, the dissociation process is a rare event, while it is highly exaggerated on the glucose gels since all of the centers undergo dissociation.
All of the results reported here are consistent with the interpretation that the cycle of association-dissociation is initiated only when D. discoideum cells are in contact with immobilized glucosides. If this interpretation is correct, then it supports the idea of membrane messengers (37).
What is the underlying teleological reason for the association-dissociation phenomenon? We tend to favor the speculation that the process resembles one which may occur naturally. When nutrients are added to starving D. discoideum cells, the cells reverse development (28, 29). The glucoside gels used in the present studies may mimic naturally occurring nutrients. We would speculate that the starved cells begin the process of development as usual, proceed to the aggregation center stage, and then, because they are in contact with a putative nutrient (the glucoside gels), they reverse the developmental process, and dissociate to single cells. Since they are then unable to utilize the immobilized glucosides, they remain starved, and once again form aggregation centers, thereby completing the cycle.
Despite the fact that the aggregation centers dissociate, two of the usual markers of cell development (contact sites A and UDP-glucose pyrophosphorylase) increase in the cells on glucoside gels, just as though the cells were proceeding through the normal developmental process. Thus it appears that there are two sets of events under genetic and metabolic control which can be separated with the glucoside gels. The reversal of synthesis of the two markers, contact sites A and UDPglucose pyrophosphorylase, may require ingestion of nutrients (catabolite repression?), whereas the dissociation phenomenon, the cessation of signaling, and the possible secretion of negative chemotactic agents may well be under the control of different regulatory genes which are triggered by the binding of the immobilized glucosides to their membrane receptors.
The synthetic gels should prove to be useful in studying these important events.