Myxococcus xanthus (chemotaxis/pattern formation/morphogenesis/multicellular development/cell movement)

Transcription

1 Proc. Natl. Acad. Sci. USA Vol. 93, pp , April 1996 Developmental Biology C factor, a cell-surface-associated intercellular signaling protein, stimulates the cytoplasmic Frz signal transduction system in Myxococcus xanthus (chemotaxis/pattern formation/morphogenesis/multicellular development/cell movement) LOTrE S0GAARD-ANDERSEN* AND DALE KAISERt Department of Biochemistry, Stanford University School of Medicine, Stanford, CA Contributed by Dale Kaiser, December 5, 1995 ABSTRACT C factor, an intercellular signaling protein, is required for aggregation and sporulation of the social bacterium Myxococcus xanthus. We report that C factor, which normally is associated with the cell surface, provides input to the Frz signal transduction cascade. Elements of this cascade have sequence homology to bacterial chemotaxis systems and are known to control the frequency of gliding reversal. Exposure of developing cells of a C-factor-less mutant (csga) to purified C factor increases the ratio of methylated to nonmethylated FrzCD protein, the Frz homolog of the methylaccepting chemotaxis proteins. Methylation depends on the cognate methyltransferase FrzF, and its extent increases with the concentration of C factor. C-factor-induced methylation also depends on the product of a gene, called class II, which is necessary in vivo for all known responses to C factor. A model for aggregation is proposed in which C factor stimulates the Frz cascade and thereby decreases cell reversals in a way that preferentially leads cells into an aggregate. In morphogenesis, organized cell movements often precede cellular differentiation. The extensive cell movements of gastrulation are followed by differentiation of the three germ layers (e.g., ref. 1). Neural crest cells migrate before differentiation (2), and Dictyostelium discoideum amoebae aggregate prior to their differentiation (3). In the Gram-negative bacterium Myxococcus xanthus, C factor, the cell-surface-associated (4, 5) intercellular signaling protein, induces aggregation and then sporulation (5-7). By using C factor, we address the question of how a cell-bound signaling protein can direct organized cell movements. M. xanthus cells build fruiting bodies by using their gliding motility, which permits them to move over one another. Starting from a lawn of disordered cells, within 4-8 hr after initiation of development, small asymmetric aggregates are formed (8). These foci become symmetrical mounds as they accumulate up to 105 cells. By 24 hr, spores have begun to differentiate within these mounds. Cells tend to move as sheets evident later in and in spiral or circular streams that are still the cell arrangements seen in mounds and fruiting bodies (8-12). C factor is a cell-surface-associated protein encoded by the csga gene (4, 5, 13). csga mutants begin to aggregate but fail to form large symmetrical mounds, and they fail to sporulate (14, 15). During the early stages of aggregation, cells outside the nascent aggregates often move in series of equally spaced ridges of cells coordinated as traveling waves, called ripples (14). csga mutants also fail to ripple. Moreover, as measured by the expression of transcriptional lac fusions to developmentally regulated genes, gene expression induced at 6 hr or later is reduced or abolished in csga mutants, including a staged The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C solely to indicate this fact increase in expression of the csga gene itself (7, 16). However, the aggregation, sporulation, and developmental gene expression of a csga mutant is rescued by codevelopment with wild-type (wt) cells or by addition of purified C factor at 1 nm (5, 6). Rescue by wt cells requires the motility of both cell types (17, 18). It has been suggested that movement establishes specific contacts between cells (thought to be end-to-end contacts) that are required for intercellular C-factor signal transmission (19, 20). Isolation of mutants that, like csga, start but fail to complete aggregation brought forward two classes of functions in the C-factor signal transduction pathway: Frz mutants and class II mutants (L.S.-A., F. Slack, H. Kimsey, and D.K., unpublished results). Phenotypic analysis of these mutants suggested a branched signal transduction pathway for C factor. In this pathway (Fig. 1), the activating signaling event is the interaction between C factor on one cell with a C-factor sensor on a second cell. The C-factor signal is transduced through the product of the class II gene(s), the earliest known component in the pathway, to the Frz proteins in a branch that leads to the motility responses to C factor; a second branch from class II leads to the sporulation response and csga gene expression. Class II is defined by three closely linked Tn5 lac insertions, but it is not known whether these are insertions in a single gene or adjacent genes with related functions. The Frz genes regulate the reversal frequency of gliding cells, and it has been suggested that cells orient the direction of their movement in response to chemical stimuli thereby (21-23). In the present report, we provide biochemical evidence that C factor regulates FrzCD activity, by showing that C factor increases the level of FrzCD methylation. We propose that such excitation of the Frz system by C factor decreases the reversal frequency of appropriately starved cells and that this change helps to bring about cell aggregation for development. MATERIALS AND METHODS M. xanthus Strains and Growth. Strains DK5204 (Tn5 lac f4435) (24), DK5253 (csga::tn5-132 fls205, Tn5 lac f4435) (16), DK11050 (csga::tn5-132 fls205,frzf::tn5 lac f7522), and DK11074 (csga::tn5-132 fls205, Tn5 lac f7540) (L.S.- A., F. Slack, H. Kimsey, and D.K., unpublished results) were grown in CTT medium (25). Purification of Proteins. C factor was purified as described (5, 6). Briefly, two fractions were obtained: FII (104 C-factor units/mg of protein) after solubilization in 1.2% 3-[(3-cholamidopropyl)dimethylammonio]-l-propanesulfate (CHAPS) and ultrafiltration through a 10-kDa molecular mass cut-off filter. FIV (1389 C-factor units/mg of protein) was obtained from FII by Abbreviation: wt, wild type. *Present address: Department of Molecular Biology, University of Odense, Campusvej 55, 5230 Odense M, Denmark. tto whom reprint requests should be addressed.

2 2676 Developmental Biology: S0gaard-Andersen and Kaiser FIG. 1. Frz Rippling v' Aggregation / l II / Frz Class "11-, - Sporulation csga expression Model of the C-factor signal transduction pathway (see text for additional details). Cell-surface-bound C factor (-f>) and C-factor sensor (>) are as indicated. For simplicity, the pathway is shown in only one of the cells, but it should be imagined in both. ion-exchange chromatography, ultrafiltration as above, and gel filtration chromatography. MalE-C-factor fusion protein (16 C-factor units/mg of protein) was purified from Escherichia coli by affinity chromatography as described by Lee et al. (26). MalE protein was purchased from New England Biolabs. Protein concentrations were determined by the Bio-Rad protein assay by the manufacturer's suggestions using bovine IgG as standard. C-Factor and FrzCD Methylation Assays. Each sample to be tested for C-factor activity was dialyzed against 4 liters of A50 buffer (10 mm Mops/1 mm CaCl2/4 mm MgCl2/50 mm NaCI, ph 7.2) for 18 hr at 4 C. Dialyzed samples were serially diluted in A50, each dilution (400,l) was warmed to 32 C and added to 2.5 x 108 csga responder cells that had been starved in submerged culture for 6 hr at 32 C in A50. Prior to addition of the samples, the A50 overlying the responder cells was removed. Fruiting bodies and spores were detected as described (5). One C-factor unit of activity is the amount of C factor that restores wt fruiting body formation and sporulation to 2.5 x 108 csga cells (5). To test a sample for an effect on FrzCD methylation, responder cells were allowed to develop for 6 hr as described above. Buffer overlying the cells was removed and replaced with prewarmed A50. Cells were harvested by gently scraping them off the bottom of the well and then transferred to a 12-ml polypropylene tube. Cells were shaken vigorously to break clumps and the sample to be tested for an effect on FrzCD methylation was added to give a final volume of 400 tul per 2.5 x 108 cells. Cells were incubated for 60 min with shaking at 32 C, harvested, and resuspended in 50,ul of SDS loading buffer. Proteins released from the lysed cells were separated in SDS/PAGE as described (22, 27). Protein from 5 x 107 cells was loaded per lane. To determine FrzCD methylation status, immunoblot analysis with polyclonal rabbit anti-frzcd serum was performed (27, 28): Proteins were transferred from SDS/ polyacrylamide gels to Immobilon-P (Millipore) membranes. To detect bound primary antibody, horseradish peroxidaseconjugated secondary antibody was used. Bound secondary antibody was detected by using a chemiluminescense kit from Dupont/NEN. RESULTS In enteric bacteria a change in the methylation state of the MCP proteins reflects a flux of phosphorylation through the two-component cascade of the Che proteins (29). By analogy changes in the methylation status of FrzCD, the MCP analog in the Frz system, are indicative of signaling through the Frz system (22, 27). Because the electrophoretic mobility of FrzCD increases with methylation, the fraction of methylated FrzCD protein can be determined on immunoblots with polyclonal FrzCD antibodies after separation of total cellular protein in SDS/PAGE (22). Six forms of FrzCD protein have been identified: the two with the lowest mobility correspond to nonmethylated FrzCD, and the remaining four correspond to FrzCD proteins with one or more methyl groups added (22). Level of FrzCD Methylation Is Decreased in a csga Mutant. Densitometric scans of the immunoblots from a wt strain and a csga mutant revealed that the total amount of FrzCD protein per cell was the same in both strains until 9 hr of development (Fig. 2, compare lanes 1-4 and 9-12). From 9 hr on, the csga Proc. Natl. Acad. Sci. USA 93 (1996) wildtype Hr of development csga Hr of development ~ r , FIG. 2. Developmental expression and methylation of FrzCD. Approximately 2.5 x 108 DK5204 (wt) or DK5253 (csga) cells resuspended in 400 Mi of A50 were induced to develop in submerged culture. At the times indicated, cells from a well were harvested and frozen. Protein from 5 x 107 cells was added to each lane. Open and solid arrows indicate nonmethylated and methylated FrzCD, respectively. cells contained more FrzCD protein than the wt strain (Fig. 2, compare lanes 4-8 and 12-16). FrzCD methylation in the two strains was similar from 1 to 6 hr: FrzCD is demethylated within the first hour of starvation followed by slightly increased methylation at 6 hr. From then on, FrzCD became increasingly methylated in wt cells, at 9 hr a large fraction was methylated, and at 12, 18, and 24 hr all detectable FrzCD protein was present in three methylated forms (Fig. 2, lanes 4-7). At 36 hr, FrzCD protein was no longer detectable in wt cells (lane 8). In csga cells, however, nonmethylated FrzCD could be detected until 24 hr of development (Fig. 2, lane 15). In csga cells, FrzCD never reached the level of methylation found in the wt strain, as indicated by the presence of three bands of methylated FrzCD from wt cells late in development compared to only one band from csga cells (Fig. 2, compare lanes 5-6 and 15-16). These results agree with previous analyses of FrzCD methylation during development (30), except that differences in levels of FrzCD methylation observed between a wt and a csga strain were more significant here. The previous report used a different csga allele and different conditions to induce development, which might contribute to the differences observed. Two conclusions can be drawn from Fig. 2. (i) csga is not required for accumulation of FrzCD protein during development. This observation and the finding that Tn5 lac fusions to frzcd, frze, and frzf are expressed at similar levels in wt and csga cells (L.S.-A. and D.K., unpublished results) suggest that C-factor control is not exerted at the level of frz gene expression. (ii) wt and csga cells differ in their patterns of FrzCD methylation. The major shift in FrzCD from nonmethylated to methylated forms in wt cells between 6 and 12 hr of development is almost completely absent in csga cells. Within this time interval, the divergence of gene expression and of aggregation behavior of csga mutants from wt is first detected, suggesting that they are related and that C factor could be an input signal to the Frz system during aggregation. Induction of FrzCD Methylation with Purified C Factor. If C factor is an input to the Frz system, then addition of C factor to csga cells that have been starved for 6 hr should increase FrzCD methylation. In testing this prediction, the experimental protocol was set up to distinguish direct effects of C factor on the Frz system from indirect effects that depend on cell aggregation. For this purpose, csga mutant cells that had been starved for 6 hr in submerged culture were harvested, brought into suspension, and exposed to C factor for 60 min with shaking. Microscopic examination of the cultures revealed that C factor did not induce aggregate formation under these conditions (data not shown). Therefore, any effect of C factor

3 Developmental Biology: S0gaard-Andersen and Kaiser on FrzCD methylation cannot be secondary to a primary effect on aggregation. In response to the addition of 1 unit of crude C factor (fraction FII, 104 C-factor units/mg of protein), the level of FrzCD methylation in csga cells strikingly increased relative to the control with no added C factor (Fig. 3A, lane 1 and 3). Addition of 0.5 unit of C factor stimulated FrzCD methylation to a lesser extent (Fig. 3A, lane 2). More highly purified C factor (fraction FIV, 1389 C-factor units/mg of protein) had similar effects per unit of C factor added (Fig. 3A, lanes 4-6). To test whether the effect of partially purified C factor on FrzCD methylation should be attributed to C factor rather than some other substance, csga cells were exposed to a MalE-C-factor fusion protein purified to homogeneity from E. coli. Fusion protein has C-factor activity measured by the standard C-factor bioassay (26). Addition of the fusion protein (specific activity, 16 C-factor units/mg of protein) also increased FrzCD methylation (Fig. 3A, lanes 7-10). While 0.5 or 1 unit of fusion C factor had detectable effects on FrzCD methylation, 4 units converted most of the FrzCD to methylated forms. To be sure that the effect of the fusion protein was due to C factor, MalE protein was added equal in weight to the largest amount of fusion protein tested. Pure MalE protein had no more activity than the buffer control (Fig. 3A, lane 11 vs. lane 1). V8 protease destroys the ability of native C factor to rescue aggregation and sporulation (6). To test whether proteolysis also destroys the ability of MalE-C-factor fusion protein to induce FrzCD methylation, the fusion protein was incubated with V8 protease. Exposure of fusion protein to V8 protease rendered it inactive in the C-factor bioassay whereas a fusion protein incubated under similar conditions in the absence of protease retained C-factor activity (data not shown). As shown in Fig. 3B, treatment of MalE-C-factor protein with protease greatly decreased its capacity to induce FrzCD methylation, whereas untreated protein retained this capacity. The results of Fig. 3 rule out active nonproteinaceous contaminants, strengthening the argument that C factor itself increases FrzCD methylation in csga cells. A FII FIV MalE-C-factor C-factor units MalE B MalE-C-factor (V8) MalE-C-factor (no V8) * FIG. 3. Effect of exogenous C factor on FrzCD methylation in developing csga cells. (A) The indicated amounts of three C-factor preparations (FII, FIV, and MalE-C factor) were added to cells. For lane 11, MalE was added to a final concentration of 625,ig/ml, the same as the final concentration of MalE-C-factor protein in lane 10. (B) One C-factor unit of MalE-C-factor protein treated with V8 protease and 1 unit of untreated fusion protein were added as indicated to csga cells that had been induced to develop. Fusion protein and V8 protease (Boehringer Mannheim) were mixed in a 20:1 weight ratio in A50 at 32 C for 18 hr. In parallel, fusion protein was incubated in the absence of V8 protease, and V8 protease was incubated in the absence of fusion protein. The fusion protein/v8 protease mixture was added directly to cells. An identical amount of V8 protease incubated in the absence of fusion protein was added to cells along with the untreated fusion protein. Open and solid arrows indicate nonmethylated and methylated FrzCD, respectively. Relevance of FrzCD Methylation Induced by Addition of C Factor. To what extent are the changes in methylation of FrzCD in response to addition of purified C factor applicable to the cell-to-cell C-factor signaling in vivo? Fig. 1 summarizes knowledge derived from biochemical, cell biological, and genetic experiments on C-factor signaling. If the FrzCD methylation induced by addition of purified C factor conforms to this circuit, then definite predictions can be made as to the way frzf csga and class II csga double mutants should respond to the addition of C factor. The FrzCD methyltransferase is encoded by frzf (27), and the yeast extract-induced FrzCD methylation is frzf-dependent (22). In a frzf csga double mutant, FrzCD was present in the two nonmethylated forms and FrzCD methylation did not change in response to addition of yeast extract (Fig. 4, lanes 5 and 6), whereas yeast extract had its reported effect in a frzf strain (Fig. 4, lanes 1 and 2). To determine whether the effect of C factor on the level of FrzCD methylation was dependent on FrzF, 1 unit of C factor (fraction FII) was added tofrzf csga cells that had developed for 6 hr in submerged culture. Addition of C factor had no effect on FrzCD methylation in this strain (Fig. 4, lanes 7 and 8). Thus, the increase in FrzCD methylation induced by addition of C factor also depends on the cognate FrzCD methyltransferase. The class II component, which is required for both motility and sporulation responses to C factor, is located downstream of the C-factor signaling event but upstream of the Frz proteins (Fig. 1). It follows that addition of C factor should not increase FrzCD methylation in a strain with both a class II mutation and a csga mutation. As predicted, 1 unit of fraction II C factor did not change the level of FrzCD methylation in Tn5 lac 17540, csga cells (Fig. 4, lanes 11 and 12) (Tn5 lac Q7540 is a class II insertion). However, a strain with an intact class II component responded to C factor with increased methylation (Fig. 4, lanes 3 and 4). Yeast extract resulted in the same quantitative increase in FrzCD methylation in the Tn5 lac csga strain as observed in the csga mutant (Fig. 4, compare lanes 9 and 10 with 1 and 2). Thus, a strain with a class II mutation remains fully competent to methylate FrzCD in response to yeast extract. Proc. Natl. Acad. Sci. USA 93 (1996) 2677 DISCUSSION Unlike its E. coli and Salmonella che gene homologs, FrzCD lacks transmembrane and extracellular domains and it is a soluble protein found in the cytoplasm of M. xanthus cells (21, 22). Several related experiments combine to show that extracsga frzf, csga.27540, csga <---..*4<-- >- > C-factor (F II) - Yeast extract Buffer Prestarv. (hr) FIG. 4. Effect offrzf and class II mutations on C-factor-induced FrzCD methylation. As indicated, 1 unit of C factor from fraction FII, yeast extract, or starvation buffer A50 was added to the three strains. Cells were exposed to 0.5% yeast extract for 1 hr (23) after being prestarved for 1 hr with shaking in A50 at 32 C. Cells in the experiments shown in lanes 1, 5, and 9 were incubated for a total of 2 hr in A50 with shaking. In the experiments in lane 3, 4, 7, 8, 11, and 12, cells were prestarved for 6 hr in submerged culture before exposure to C factor or buffer in suspension. Tn5 lac Q7540 is abbreviated Q7540; csga strain is DK5253;frzFcsgA strain is DK11050; csga strain is DK Open and solid arrows indicate nonmethylated and methylated FrzCD, respectively.

4 2678 Developmental Biology: S0gaard-Andersen and Kaiser cellular C factor provides an input signal to the Frz signal transduction system, as indicated by increases in FrzCD methylation after 6 hr of starvation. (i) Fraction II and fraction IV preparations of partially purified native C factor induce an increase in FrzCD methylation. (ii) Pure MalE-C-factor fusion protein from E. coli added to M. xanthus cells that have been starved on a surface for 6 hr increases FrzCD methylation. The fusion protein has C-factor activity as measured in the C-factor bioassay (26). Fusion protein, fraction II, and fraction IV induced similar quantitative increases in FrzCD methylation (Fig. 3) by the bioassay unit of C factor, within a factor of 4, which compares to the 2- to 4-fold precision range of the bioassay (5, 6). (iii) Addition of MalE protein did not have any effect on FrzCD methylation, arguing that the C-factor moiety of the fusion protein is crucial for the FrzCD response. V8 protease inactivates C-factor fusion protein as measured by bioassay as well as by FrzCD methylation (ref. 6 and Fig. 3). Therefore, increases in FrzCD methylation are due to a protein constituent of the fusion protein preparation. (iv) The time at which FrzCD becomes heavily methylated during development corresponds to the time at which C factor becomes essential for aggregation and for developmentally regulated gene expression in vivo. Two observations indicate that extracellular C-factor signals FrzCD by the route shown in Fig. 1. (i) The effect of extracellular C factor on FrzCD methylation depends on the FrzCD methyltransferase FrzF (27). (ii) C factor did not induce an increase in FrzCD methylation in a class II mutant. Genetic evidence has shown that class II is located downstream of the cell-to-cell signaling event and upstream of the Frz proteins in the C-factor signaling pathway. These experiments establish a firm logical connection between extracellular C factor and the cytoplasmic FrzCD. The class II component could be a physical link between the cell's exterior and its cytoplasm. To our knowledge, C factor is the first described compound from M. xanthus that has been shown to stimulate the Frz system. Based on a study of rippling, it has been suggested that C-factor signaling modulates the frequency of gliding reversals (20). Methylation of FrzCD correlates with a low reversal frequency and demethylation with a high one (22, 23). This correlation predicts that C factor, which increases FrzCD methylation in cells starved for 6 hr, should decrease reversal frequency, and we assume that to be the case. However, Sager and Kaiser (20) observed that the addition of low levels of C factor ( unit) to wt cells starved for 48 hr increased the reversal frequency of peripheral rod cells 3-fold, suggesting that the nature of the reversal response to C factor may depend on its concentration and the regulatory state of the responsive cells (6-hr vs. 48-hr starved cells, c.f. Fig. 2). A Model for Aggregation. Proper regulation of the cell reversal frequency is an important element in the formation of a fruiting body, as shown by the inability of frz mutants to construct them (21). A model for aggregation that emphasizes results described here and changes in reversal frequency dependent on C-factor signaling is diagrammed in Fig. 5. The elementary event in this model is an aligned (19) contact between cells (here diagrammed as end to end; ref. 20) followed by C-factor signaling. C-factor signaling increases the level of FrzCD methylation (results above); the observation that increased FrzCD methylation correlates with a decrease in reversal frequency (22, 23) is assumed to apply. C-factor signaling would be expected (7) to increase production of C factor in both contacting cells, thus enhancing and prolonging the initial response of a decrease in reversal frequency. Applied to a field of starved developing cells that have just begun to aggregate, this model may explain how initially small asymmetric (8) aggregates accrete cells, grow larger, and become symmetric. When cells have developed to a stage of competence for C-factor signaling, it would be expected to initiate within some kind of aggregate because the higher C, CZ._o -'E CD (._ E =a 00 CD cn Proc. Natl. Acad. Sci. USA 93 (1996) FIG. 5. Model for C-factor-induced aggregation. C-factor signaling by end-to-end contacts results in decreased reversal frequencies and in sequential recruitment of cells to chains moving toward an aggregation center. Cells are indicated by the labeled rounded boxes. Arrows indicate direction of movement of chains of cells. Cell 1 has end-to-end contact with a cell in an aggregation center. This interaction and the center is indicated by the left-hand jagged edge of cell 1. Hatching indicates cells that have been recruited to a chain and are engaged in C-factor signaling. Open cells are competent for C-factor signaling but are not yet engaged in it (see text for additional explanation). density of cells there would engender more aligned contacts. Consider cell 1 at the edge of an initial aggregate that is C-factor signaling with a cell in the aggregate and is moving into the aggregate (Fig. 5). If cell 2 happens to move into end-to-end contact with cell 1, then the two cells will transmit the C-factor signal to each other, decreasing the reversal frequency of both. Consequently, both cells 1 and 2 will continue to move in the same direction, which is into the aggregate. This sequence of events could be repeated sequentially adding cell 3, etc., thus creating a chain whose cells are moving in line toward the same aggregation center. By contrast, if cell 1 had been moving away from the aggregate, it would have lost its contact with an intensely C-factor signaling cell in the aggregate and, thus, have returned to the behavioral mode of frequent reversals, decreasing its chance of initiating or continuing a chain. This is a key point: cells 1-4 maintain C-factor signaling with each other only if they continue to move as a chain into the aggregate. Although reversal of gliding direction of any cell in a chain would break that chain, reversals are unlikely for several reasons. (i) Such reversals should have been suppressed by the C-factor signaling that created the chain. (ii) The C-factorinduced increase in csga expression (7) would increase the total level of C factor on the cells of the chain, further increasing the intensity of C-factor signaling and further decreasing their probability of reversal. Once a cell had achieved this exhalted level of C-factor signaling, it might continue to move into the aggregate even after its connection through a chain of cells to the aggregate had broken. (iii) Other known cell-cell interactions could help to keep chains from breaking up, including slime trails and side-by-side contacts directed by the products of the A and S systems of genes that govern the density dependence of cell movement (21). One potential gap in the scheme of Fig. 5 turns on whether there is adaptation and desensitization of cells to C-factor signaling. In enteric bacteria, methylation of the MCP proteins is part of an adaptation to the higher levels of chemoattractant that are experienced as a cell moves up a spatial gradient of attractant (31). If FrzCD methylation in response to C-factor signaling is associated with desensitization, then signaling would be expected to cause only a transient decrease in reversal frequency. There are as yet no experimental data on adaptation to C factor in M. xanthus. However, the fact that M. xanthus responds distinctly and progressively to low, medium, and high levels of C factor (7, 28) would seem to argue against adaptation. Clearly, experimental tests of this point as well as other aspects of the model are required.

5 Developmental Biology: S0gaard-Andersen and Kaiser We gratefully acknowledge Mark McBride for providing us with the FrzCD antibodies and many helpful suggestions about the methylation assay. We are indebted to Lawrence Shimkets for providing the plasmid that encodes the MalE-C-factor fusion protein and for advice on its purification. This investigation was supported by Public Health Service Grant GM to D.K. from the National Institute of General Medical Sciences and a stipend to L.S.-A. from the Danish Natural Science Research Council. 1. Stern, C. D. & Ingham, P. W. (1992) Gastrulation (Company of Biologists, Cambridge, U.K.). 2. Le Douarin, N. M. (1984) Cell 38, Gross, J. D. (1994) Microbiol. Rev. 58, Shimkets, L. J. & Rafiee, H. (1990) J. Bacteriol. 172, Kim, S. K. & Kaiser, D. (1990) Cell 61, Kim, S. K. & Kaiser, D. (1990) Proc. Natl. Acad. Sci. USA 87, Kim, S. K. & Kaiser, D. (1991) J. Bacteriol. 173, Kuner, J. M. & Kaiser, D. (1982) J. Bacteriol. 151, O'Connor, K.A. & Zusman, D. R. (1989) J. Bacteriol. 171, Sager, B. & Kaiser, D. (1993) Proc. Natl. Acad. Sci. USA 90, Reichenbach, H. (1965) Ber. Deutsch. Bot. Ges. 78, Reichenbach, H. (1966) in Myxococcus spp. (Myxobacteriales). Schwarmentwicklung und Bildung von Protocysten, ed. Wolf, G. (Inst. Wissenschaftlichen Film, Gottingen, Germany), Vol. 1A, pp Hagen, T. J. & Shimkets, L. J. (1990) J. Bacteriol. 172, Proc. Natl. Acad. Sci. USA 93 (1996) Shimkets, L. J. & Kaiser, D. (1982) J. Bacteriol. 152, Shimkets, L. J., Gill, R. E. & Kaiser, D. (1983) Proc. Natl. Acad. Sci. USA 80, Kroos, L. & Kaiser, D. (1987) Genes Dev. 1, Kim, S. K. & Kaiser, D. (1990) Genes Dev. 4, Kroos, L., Hartzell, P., Stephens, K. & Kaiser, D. (1988) Genes Dev. 2, Kim, S. K. & Kaiser, D. (1990) Science 249, Sager, B. & Kaiser, D. (1994) Genes Dev. 8, McBride, M. J., Hartzell, P. & Zusman, D. R. (1993) in Motility and Tactic Behaviour ofmyxococcusxanthus., eds. Dworkin, M. & Kaiser, D. (Am. Soc. Microbiol., Washington, DC), pp McBride, M. J., Kohler, T. & Zusman, D. R. (1992) J. Bacteriol. 174, Shi, W., Kohler, T. & Zusman, D. R. (1993) Mol. Microbiol. 9, Kroos, L., Kuspa, A. & Kaiser, D. (1986) Dev. Biol. 117, Hodgkin, J. & Kaiser, D. (1977) Proc. Natl. Acad. Sci. USA 74, Lee, B.-U., Lee, K., Robles, J. & Shimkets, L. J. (1995) Genes Dev. 9, McCleary, W. R., McBride, M. J. & Zusman, D. R. (1990) J. Bacteriol. 172, Li, S., Lee, B.-U. & Shimkets, L. J. (1992) Genes Dev. 6, Parkinson, J. S. (1993) Cell 73, McBride, M. J. & Zusman, D. R. (1993) J. Bacteriol. 175, Schnitzer, M. J., Block, S. M., Berg, H. C. & Purcell, E. M. (1990) Symp. Soc. Gen. Microbiol. 46,