Proc. Nati. Acad. Sci. USA Vol. 76, No. 11, pp. 5952-5956, November 1979 Microbiology

Social gliding is correlated with the presence of pili in Myxococcus xanthus (motility/swarming/cell interactions/fimbriae/Myxobacterales) DALE KAISER Department of Biochemistry, Stanford University, Stanford, California 94305 Contributed by A. Dale Kaiser, August 20, 1979

ABSTRACT Myxococcus xanthus, an organism whose mutation in any one of the loci of system S, exhibit (pure) A- motility involves cell interactions, normally bears pili. Myxo- motility. A-S+ mutants, which have a mutation in any one of coccal pili are found only at cell poles, are less than 10 nm in the loci of system A, exhibit (pure) S-motility. A-S- diameter, and may be longer than a cell. Myxococcus has two mutants, basic patterns of cell movement, adventurous (A-motility) and which have a mutation in any locus of system A and a mutation social (S-motility). Pili are found to be completely correlated in any locus of system S, are nonmotile. with the presence of S-motility. (The S-motility pattern has many There is evidence that pili (fimbriae), first reported on groups of cells, almost no single cells, and is governed by a set myxobacteria by MacRae and McCurdy (8), are associated with of genes called system S.) On the other hand, A-motility is in- motility. Many, but significantly not all, motile myxobacteria dependent of piliation. (The A-motility pattern has many single, and other gliding bacteria that have been examined have polar isolated cells and it is governed by a second set of genes called system A.) Electron microscopic examination of more than 40 pili (8-11) and several nonmotile mutants of M. xanthus were genetically different strains shows that all A+S+ (wild-type) and shown to lack them. But if pili are associated with motility, how A-S+ strains have pili, but A+S- and A-S- strains lack them. can there be nonpiliated motile strains? The existence of two Mutations in four different loci belonging to system S were patterns of movement in M. xanthus, controlled by different tested and were found to stop production of pili: the loci sglA, sets of genes, suggested the possibility that pili might be asso- sglB, sglG, and tgl. When brought into contact with tgl+ cells, ciated with only one of the two patterns. The experiments re- cells of a tgl- strain, which lack pili, become phenotypically ported here were designed to test that possibility. S+ produce pili, and become S-motile. Both motility and the prouction of pili are transient when initiated in this way. Thus it appears that pili permit cells that are close to one another to MATERIALS AND METHODS move. Bacteria. The origin and description of strains DK1200- DK1292 may be found in refs. 6 and 7 and of strains Myxobacteria are rod-shaped cells that move by gliding on DK801-DK898 in ref. 12. DZ2 was described by Campos and surfaces, an activity that helps them carry out their primitive Zusman (13). DK1050 is a single-colony isolate of the motile kind of multicellular development (1, 2). When starved, prototype M. xanthus strain FB (14). DK1622 and DK1253SR thousands of cells assemble to form a fruiting body whose shape are A+S+ transductants of DK1217 and DK1253, respectively, is genetically determined (3). Not only when fruiting but also isolated by David Morandi. DK1805, DK1808, DK1811, when growing in the presence of ample nutrient, myxobacteria DK1813, DK1818, and DK1819 are A+S- transductants of the move in multicellular units (4, 5). A thin halo of cells forms at A-S- strains DK1246, DK1251, DK1252, and DK1292, pro- the edge of a colony on agar and this halo continually expands duced by using phage Mx8 grown on the A+S- strain DK1253. outward. When viewed at higher magnification the edge of the A+ transductants were selected by their capacity to glide halo is seen to consist of a single-layer filigree of cells in which through a Millipore membrane filter 150 tm thick with single cells and groups of cells move and reassort (5). The work 0.45-,gm pores. Transductants were picked and purified and to be reported here is aimed at understanding how myxobac- their A-motility was confirmed by microscopic examinations terial cells interact to regulate and coordinate their move- of the halo of cells around single colonies. ment. The general conditions for growth of bacterial cultures, An investigation of more than 100 mutants of Myxococcus transduction, and stimulation have been described (15). The xanthus has revealed that its movement on agar is composed symbol A+ indicates that all loci of system A are wild-type and of two distinct patterns of movement, named A (for adventu- A- indicates that at least one locus in system A is mutant. S+ and rous) and S (for social). In A-motility, single cells move (6), re- S- have analogous meanings for the loci of system S. sulting in a spatial distribution with many single cells. In S- Assay of Pili. Bacterial cultures were grown in 1% casitone/8 motility, isolated cells do not move, but cells that are close to mM MgSO4/10 mM Tris.HCl, pH 7.6/1 mM potassium phos- one another do move (7). The net result is a spatial distribution phate, pH 7.6 (CTT broth) (15) at 250C with rotary shaking to of cells with many clusters and very few isolated single cells (4, cell densities of 1-4 X 108 cells per ml. (Several cultures were 7). The two patterns are governed by different sets of genes: also grown at 330C, but no differences in piliation from 25'C A-motility by one set of 21 loci, called gene system A and S- cultures were noted, except for temperature-sensitive mutants.) motility by another set of 9 loci, called gene system S (6, 7). One The cells were sedimented, the supernatant fluid was poured motility locus, mgl, is necessary for both kinds of movement. off, and the cell pellet was resuspended in 10 mM Tris-HCl, pH Wild-type M. xanthus has both gene systems and shows a 7.6, at a density of 2 X 108 cells per ml. One drop of cell sus- mixture of the two patterns. However, because all but one of pension was placed on a carbon-coated electron microscope grid the motility loci affect either one or the other of the two patterns but not both, by using mutations it is possible to separate the two Abbreviations: A-motility, adventurous movement; S-motility, social patterns from each other. Thus A+S- mutants, which have a movement. 5952 Microbiology: Kaiser Proc. Natl. Acad. Sci. USA 76 (1979) 5953

and cells were allowed to settle and adhere to the carbon for 1-2 Table 1. Piliation of wild-type strains min. The liquid droplet on the grid was washed by addition of Cell ends with several successive drops of the negative stain- 1% uranyl ace- pili* tate mixed with 30 Mg of bacitracin per ml, serving as a Strain Origin Number % spreading agent (16). The stain solution was immediately re- DK801 Soil (Tracy, CA) 28/79 35 moved with blotting paper and the grid was dried in air. Grids DK804 Soil (Yosemite, CA) 44/53 83 were imaged at an electron optical magnification of X12,600 DK805 Soil (Palo Alto, CA) 49/76 65 and the image was viewed through a X10 telescope. Only those DK806 Soil (Merced, CA) 12/33 36 cell ends were scored that were free of adhering extracellular DK813 Soil (South Dakota) 45/74 61 material and for which the quality of staining was such that in DK816 Soil (Ontario, Canada) 20/30 67 the judgment of the observer it could have revealed pili. Each DK854 Soil (St. Louis, MO) 40/69 58 cell end was scored individually because very often cells were DK879 Soil (King City, CA) 16/34 47 joined by one end to a chain or clump of cells or partly over- DK891 Soil (Ames, IA) 79/110 72 lapped the metallic portion of the grid. To the extent that cells DK897 Soil (Maryhill, WA) 17/41 42 adhere to other cells by their piliated ends, the number of pi- DK898 Soil (Fiji) 35/57 61 hated ends will be underestimated. However, nonpiliated cells DZ2 FB (13) 53/98 54 disperse well and offer no problems for counting. Generally, DK1050 FB (From M. Dworkin in 1973) 68/153 44 only piliated strains form clumps but cultures of these strains DK1622 Transduction of DK1217 41/70 58 also contain many free piliated cells. Therefore, enumeration DK1253SR Transduction of DK1253 157/373 42 of cell ends with pili is most accurate at low degrees of piliation * Under "Number" is given a fraction whose numerator is the number and may underestimate the high degrees of piliation. of cell ends with pili and whose denominator is the total number of ends scored. RESULTS fewer than three at the other. The two poles of a cell thus differ Pili of M. xanthus from each other in a way that might be explained by the dif- Fig. 1 shows a piliated cell of M. xanthus, illustrating several ference in their ages. characteristic features. Pili arise from cell ends. Whenever the Pili observed on negatively stained cells from liquid culture anchor point of a pilus in the cell surface was clearly visible in vary in length. Most are around one cell length long,-3-5 the negatively stained image, it was found in the curved region Am-but some pili are at least 10 ttm long. The elementary at the end of a cell. Sometimes pili appeared to emerge from pilus filament appears to be less than 10 nm in diameter. On the side of a cell, but closer examination showed that these stained carbon films the pili are cohesive and are often seen filaments lie beneath cells. No unambiguous lateral insertions adhering to each other. The cohesiveness of pili for each other have been observed. Typically, 4-10 pili emerge from the same or of pili for cells may account for the tendency of piliated end. The majority of cells have pili at one end, but in heavily strains to grow nondispersed in liquid culture. Cultures of pi- piliated strains many cells have pili at both ends. For example, liated strains growing in CTT broth with rotary agitation at in one sample of strain DK1217, 73 cells bore pili at one end, 25°C form a macroscopic ring of cells adhering to the culture 59 cells had pili at both ends, and 2 cells had no pili. Bipolarly flask at the uppermost intersection of the meniscus of culture piliated cells usually had more than five pili at one end and liquid with the surface of the vessel. Also, in the bulk phase of the liquid, spherical clumps of hundreds of cells form. The piliation of wild-type strains of Myxococcus are reported in Table 1. These strains are highly motile and fruit well. The list includes 11 Myxococcus species recently isolated from soil by the method of Singh (17), a procedure that selects for strains that fruit. Strain FB of M. xanthus, which has been propagated in the laboratory since 1963 (14), retains the capacity to fruit. The distribution of cells at the edge of a colony of FB on agar includes many single cells and many groups of cells, which is diagnostic of the A+S+ state (7). Two cultures of FB, DK1050 and DZ2, that had been propagated separately for more than 10 years in the laboratories of M. Dworkin and of D. Zusman show the A+S+ phenotype, and both strains have pili (Table 1). Finally, two A+S+ strains, DK1622 and DK1253SR, generated

Table 2. Piliation of A-motile (A+S-) strains Cell endswith pili Genotype Strain Number % sglAl DK101 23/97 23 sglB3 DK1818 1/112 1 sglB46 DK1805 6/112 5 sglB46 DK1808 0/116 <1 sglGl DK1300 0/77 <2 tgl-1 DK1253 2/97 2 FIG. 1. Electron micrograph of a piliated M. xanthus cell nega- tgl-3 DK1817 1/129 1 tively stained with uranyl acetate. Bar, 0.5,m. Data were tabulated as in Table 1. 5954 Microbiology: Kaiser Proc. Natl. Acad. Sci. USA 76 (1979) Table 3. Piliation of S-motile (A-S+) strains Table 4. Piliation of nonmotile (A-S-) strains Cell ends with pili Cell ends with pili Genotype Strain Number % Genotype Strain Number % ag1Ji DK1212 16/20 80 ag1Bi sg1G1 DK1259 0/57 <2 ag1Ni DK1213 12/20 60 aglE2 sglB46 DK1246 2/1046 <1 aglBi DK1217 145/242 60 aglR2 sglB3 DK1292 0/66 <2 aglE2 DK1222 27/41 66 agiBi tgl-1 DK1250 2/1110 <1 aglR2 DK1236 59/77 77 cglFI tgl-3 DK1252 0/91 <1 cglB2 DK1218 17/27 63 agiJi tgl-2 DK1251 4/58 7 cglEl DK1230 11/23 48 as in Table 1. cg1Fi DK1234 42/59 71 Data were tabulated cglDi cglB DK1237 38/64 59 cglDi cglC DK1238 14/20 70 that system A nonstimulatable mutants lack pili,* which was cglDl cglB DK1239 11/21 52 based on strains that have since been shown to be double mu- Data were tabulated as in Table 1. tants (6, 7). None of the loci that belong to system A appear to control piliation, in sharp contrast to the effects of mutation in the loci of system S. from nonmotile A-S- strains by transduction are also found to If mutations in the loci of system S control piliation, then have pili. Thus all A+S+ strains examined are found to be pi- A-S- double mutants should lack pili. Accordingly, mutants hated. having different combinations of A and S system mutations were examined (Table 4). All were found to have no or very low degrees of piliation. As was the case with A+S-, those A-S- cells Piliation of motility mutants that had any pili had very few, usually only one. From a piliated As described in the Introduction, motility in M. xanthus com- wild-type strain, MacRae and McCurdy selected nonmotile bines two patterns: single-cell or A-motility and group or S- mutants (10). Genetic analysis of four of the mutants showed motility which are controlled by two sets of multiple genes, them to be A-S- (7) and all four lack pili (8). system A and system S (6, 7). To examine the connection be- Thus, in all the groups of strains examined, piliation is cor- tween piliation and the genes of system A and of system S, related with a complete set of system S genes and appears to be several series of mutants defective in one, the other, or both of independent of the genes of system A. This correlation is also these systems were examined. evident in specific lines of strain descent. Starting with wild- The first series tested were A+S- mutants, which have a type M. xanthus, which has pili, many different A-S+ mutants complete system A but are defective in different genes of system were isolated. As shown in Table 3 all of them have pili. Con- S. The data of Table 2 summarize measurements on eight A+S- sider, for example, the piliated A-S+ mutant ag1B1 (strain strains and-show that all eight have fewer cell ends with pili than DK1217). After mutagenesis of DK1217, a nonmotile A-S- A+S+ strains. Moreover, those A+S- strains that had some pi- mutant, DK1250, was isolated. DK1250 is ag1B1 tgl-1 and it hated cells have fewer pili per cell, generally 1-3 pili compared lacks pili. DK1250 was transduced to agiB + tgl-1 and the with 4-10 pili in corresponding S+ strains. Seven of the A+S- resulting A+S- strain DK1253 lacks pili. Finally, DK1253 was strains had 5% or fewer ends with pili; A+S+ strains had about transduced to A+S+ and those transductants have pili. Nine 5t% (Table 1). The eighth strain, which carries the sglAl similar though less complete lines of descent can be traced mutation, had 23% cell ends with pili, but it had some residual among the existing strains, and in every case the correspondence S-motility (15) and it is likely that its higher level of piliation between system S and piliation is observed. than other S strains parallels its retention of some S-motility. Wild-type strains of Myxococcus grow nondispersed in liquid Thus, at least four loci of system S control production of pili: the culture, forming a ring at the top of a shake flask culture and loci sglA, sglB, sglG, and tgl. Both stimulatable (tgl) and nonstimulatable (sgl) mutants are represented here (7). In the * Dobson, W. J. & McCurdy, H. D. (1978) Abstracts of the Annual course of these experiments it was also observed that the A+S- Meeting of the American Society on Microbiology, abstr. J3, p. strains grew dispersed in liquid culture in contrast to A+S+ 77. strains which showed nondispersed growth. Thus, the S+ phe- notype is correlated with nondispersed growth as well as the presence of pili. Table 5. Piliation induced by stimulation of system S To test whether any of the genes of system A control piliation, Cell ends with a series of A-S+ strains was examined. Eleven strains having pili a complete set of genes of system S but defective in different Cells Motility Number % genes of system A were tested (Table 3). All eleven were found agi - S+ S-motile 172/283 61 to be piliated at levels comparable to A+S+ strains. The first agiBi tgl-1 Nonmotile 1/945 0.1 group of mutants in Table 3 carry agl mutations that cannot aglE2 sglB46 Nonmotile 2/968 0.2 be corrected by stimulation (6). Other mutants reported in agiBi tgl -1 plus aglE2 sglB46 S-motile Table 3 are cgl mutants, which are stimulatable system A Experiment 1 mutants that can move if mixed with a wild-type strain or an- Picked from S-motile fringe 80/228 35 other mutant (7, 15). It should be noted that stimulation is locus, Experiment 2 not allele, specific. So, for example, all mutants isolated so far Picked from S-motile fringe 72/367 20 at the cglB locus are stimulatable (6). Mutants representing all Cells from fringe grown: of the five known stimulatable types of system A mutants are 3 gen. in liquid culture 12/726 1.7 included in Table 3 (15). In every case, whether stimulatable 8 gen. in liquid culture 4/1342 0.3 or not, A-S+ strains had pili. This finding refutes the proposal gen., Generations. Microbiology: Kaiser Proc. Natl. Acad. Sci. USA 76 (1979) 5955 a time when there are no further contacts with sgl - cells to renew stimulation. The transient appearance of motility is paralleled by a transient presence of pili. Bacteria from the swarming fringe were inoculated into liquid shaker culture at 330C. When the density of this culture had increased from 1 X 108 cells per ml to 8 X 108 cells per ml, samples were examined by electron microscopy. It was observed that the number of piliated cell ends had fallen to 1.7% (Table 5). After five more generations of growth in liquid culture the number of piliated ends had

'i fallen to 0.3%. Stimulation does not occur in liquid culture to any substantial extent (15) and, as shown above, few sgl - cells s% i f would have been present in the liquid culture to stimulate tgl - Y.- cells in any case. If the number of piliated ends had simply been i. diluted by growth of new nonpiliated cell ends, one would have * 4%l expected 2.5% (1/8 of 20%) after three generations and 0.1% ('/A56 of 20%) after eight generations, in good agreement with the FIG. 2. Stimulation of nonpiliated nonmotile mutants. Droplets values observed. Thus cells that have been stimulated to pro- (1 pil) of cell suspensions containing 4 X 105 cells per Al were placed duce pili do not reproduce piliated cells. on CTT agar and incubated for 15 hr at 330C. The spot at lower left is DK1246 (agJE2 sglB46) alone; the spot at lower right is DK1250 DISCUSSION (agiBi tgl-J) alone; the spot at the top is a 1:1 mixture of DK1246 Myxococcus combines two patterns of cell movement on agar, plus DK1250. Bar, 1 mm. each controlled by its own set of genes: the A-motile pattern with many isolated single cells controlled by the genes of system spherical clumps of cells in the liquid phase. A-S+ strains also A and the S-motile pattern with few single cells but many clump, but A+S- and A-S- strains grow dispersed in liquid groups of cells controlled by the genes of system S. The data culture. Thus, cell-cell adhesion is also correlated with pili and presented in this paper reveal a 3-fold correlation between S- a complete set of S genes. motility and the presence of pili. First, all strains examined that exhibit S-motility are S+ and have pili. Twenty-six different S+ Pili can be produced by stimulation strains have been tested and all 26 have pili. If the strain is S+, Temporary S-motility can arise in genetically S- cells when pili are present whether the strain has a complete set of system there is stimulation (7). For example, when tgl - mutant cells A genes (15 strains tested) or has a mutation in one (8 strains are mixed with sgl - mutant cells, the tgl - cells are stimulated tested) or two (3 strains tested) A genes. Second, all 12 S- strains to move transiently. Such stimulated motility is evident in Fig. examined lack S-motility and lack pili. Again, the absence of 2. Suspensions of 4 X 105 cells of an sgl - mutant, of a tgl - pili is independent of whether the S- strain is A+ (six strains mutant, and of a mixture of the two were placed on agar and tested) or A- (six strains tested). Third, genotypically S- cells incubated. The edges of the sgl - mutant and of the tgl - which, as a result of stimulation, transiently show S-motility also mutant were sharp, showing no cell movement beyond the edge become transiently piliated. It is as if stimulation produces the of the original drop. Although most cells were concentrated in S+ phenotype in genotypically S- cells and pili are a component an annulus around the edge of the original drop, this was a of the S+ phenotype. consequence of drying, not motility, because carbon grains also Four loci of system S were examined and all four were found form an annulus. The mixture of tgl - and sgl - mutants de- to control the production of pili. At the tgl locus, three mutant veloped a fringe of bacteria which moved beyond the edge of alleles were examined and all three reduced or eliminated the the original drop. Viewed at higher magnification, the fringe number of pili. Strains that had two different alleles of sglB, was observed to have a typical S-motility arrangement of cells. one allele at the sglA locus, and one at sglG were examined; A-motility was not seen because both strains carry nonstimu- all these mutant alleles reduced or eliminated pili. Because the latable mutations in system A loci. first four loci of system S that were examined proved to control To look for pili, bacteria were picked up from the fringe at the production of pili, it is likely that a large fraction of all S loci the edge of the mixed drop and from the control drops and were will prove to do so. examined by electron microscopy. The data presented in Table What role do pili play in the motility of Myxococcus? On the 5 show that, whereas cells from the pure (control) drops had no one hand, pili do not appear to be required for gliding motility pili, one-fifth to one-third of the ends of cells from the edge of or for the stimulation of motility. A+S- cells glide, yet they have the mixed drop had pili. These pili had the same morphology no pili. Unless A-motile gliding is fundamentally different from as those found on genotypically S+ cells. S-motile gliding, pili are not necessary for gliding per se. In an sgl/tgl mixture, only the tgl - cells are expected to However, because the mechanism of gliding is unknown, some be stimulated to move (7). To verify this point 100 clones were caution should be exercised on this point. Nor are pili needed isolated from the fringe of stimulated cells and streaked twice for stimulation. In the stimulation of tgl - cells, pili are a con- for purification. To test their genotypes, each unknown isolate sequence of rather than a condition for stimulation. On the was mixed on the one hand with standard sgl - cells (DK1246) other hand, pili do appear to be necessary for an interaction and on the other hand with standard tgl - cells (DK1250). The between cells that leads to S-motility. There is the strong cor- result: 96 of 100 clones gave a stimulation fringe in the mixture relation between pili and S-motility documented above and with sgl - and no fringe in the mixture with tgl -; they were S-motility is only expressed among cells that are in close prox- therefore tgl -. Only 4 of 100 clones gave the opposite response imity. Single A-S+ cells do not move on agar, but the same cells and thus were sgl -. Neither sgl + tgl + nor sgl - tgl - clones in dense culture on agar do move (7). Time-lapse photographic were found. Because the cells present in the swarming fringe studies of these cultures show the movement to be intense are almost exclusively tgl -, the fringe stops expanding after (unpublished results). 5956 Microbiology: Kaiser Proc. Natl. Acad. Sci. USA 76 (1979) Pili might bring about cell association and S-motility in 3. McCurdy, H. D. (1974) in Bergey's Manual of Determinative several different ways: (a) Henrichsen (18) has suggested that Bacteriology, eds. Buchanan, R. E. & Gibbons, N. E. (Williams pili are drawn by a hydrophobic quality to the surface of the and Wilkins, Baltimore), 8th Ed., pp. 76-98. 4. Burchard, R. P. (1970) J. Bacteriol. 104, 940-947. thin water film that overlies a group of cells on agar. As a con- 5. Kuhlwein, H. & Reichenbach, H. (1968) Encyclo. Cinemato- sequence of the presence of pili at the ends of cells only, cells graphica Film C893/1965 (Inst. Wiss. Film, Gottingen, W. would become oriented and concentrated by the surface and Germany). these factors may facilitate S-motility. (b) If there are binding 6. Hodgkin, J. & Kaiser, D. (1979) Mol. Gen. Genet. 171, 167- sites for pili on the surface of cells, or on the pili themselves, as 176. is the case in many kinds of bacteria (19), then pili would be able 7. Hodgkin, J. & Kaiser, D. (1979) Mol. Gen. Genet. 171, 177- to bind cells together. In support of this possibility is the ob- 191. servation that piliated strains of Myxococcus grow in liquid 8. MacRae, T. H. & McCurdy, H. D. (1976) Can. J. Microbiol. 22, shake cultures as roughly spherical clumps of hundreds of cells. 1589-1593. (c) Pili might be sensory hairs that initiate the movement of a 9. MacRae, T. H., Dobson, W. J. & McCurdy, H. D. (1977) Can. J. touch another cell. In the stimulation of S- Microbiol. 23, 1096-1108. cell when they 10. MacRae, T. H. & McCurdy, H. D. (1976) Can. J. Microbiol. 22, motility described here, it is the piliated tgl - cells that become 1282-1292. motile. Moreover, time-lapse cinematographic studies show that 11. Henrichsen, J. & Blom, J. (1975) Acta Pathol. Microbiol. Scand. groups of S-motile cells are not permanent, but that cells fre- Sect. B 83, 161-170. quently move from one to a neighboring group. (d) Pili might 12. Martin, S., Sodergren, E., Masuda, T. & Kaiser, D. (1978) Virology be correlated with S-motility because both depend on the same 88, 44-53. underlying cell structure. For example, the genes of system S 13. Campos, J. M. & Zusman, D. R. (1975) Proc. Natl. Acad. Sci. USA might control the structure of cell surface components that are 72, 518-522. needed for S-motility and for the assembly of pili. Further ex- 14. Dworkin, M. (1963) J. Bacteriol. 86, 67-72. 15. Hodgkin, J. & Kaiser, D. (1977) Proc. Natl. Acad. Sci. USA 74, periments are required to distinguish among these possibili- 2938-2942. ties. 16. Gregory, D. W. & Pirie, B. J. S. (1973) J. Microsc. (Oxford) 99, Cathy Crosby capably assisted in these experiments. 'rhe work was 251-265. supported by National Science Foundation Grant PCM 77-19382. 17. Singh, B. N. (1947) J. Gen. Microbiol. 1, 1-10. 1. Dworkin, M. (1966) Annu. Rev. Microtnol. 20, 75-106. 18. Henrichsen, J. (1975) Acta Pathol. Microbiol. Scand. Sect. B 83, 2. Kaiser, D., Manoil, C. & Dworkin, M. (1979) Annu. Rev. Mi- 187-190. crobiol. 33, 595-639. 19. Ottow, J. C. G. (1975) Annu. Rev. Microbiol. 29, 79-108.