JOURNAJL OF BACTzRIOLOGY, Apr. 1977, p. 200-204 Copyright C) 1977 American Society for Microbiology
Vol. 130, No. 1
Printed in U.S.A.
Classification of Bacillus subtilis Flagellins MELVIN I. SIMON, SUZANNE U. EMERSON,1 JOEL H. SHAPER,2 PATRICIA D. BERNARD, AND ALEXANDER N. GLAZER3* Department of Biology, University of California, San Diego, La Jolla, California 92037, and Departments of Bacteriology and Biological Chemistry, University of California, Los Angeles, California 90024 Received for publication 7 January 1977
Purified flagellins derived from 16 strains of Bacillus subtilis were classified into at least five distinct groups on the basis of their reaction with antiflagellar filament antibody and antiflagellin antibody. This classification was in good accord with that derived independently on the basis of amino acid analyses of the flagellins. Flagellar antigenicity appears to provide a useful typological character in classifying B. subtilis strains. Flagellar serotype variation has been studied extensively in Salmonella and Escherichia coli. This variation is one of the basic criteria used in the classification of Enterobacteriaceae (7). A large degree of antigenic variation has also been observed in flagellar proteins derived from strains of Bacillus subtilis (5). The flagellin derived from B. subtilis strain 168 has been studied in great detail, and the entire amino acid sequence of this protein is known (1, 2, 9). It was therefore of interest to extend the studies on variation in flagellar serotype in B. subtilis both from the point of view of establishing methods to classify B. subtilis strains and to get a clearer picture of the biochemical basis of serotype variation. Two kinds of antisera can be prepared against flagellar protein. The first reacts primarily with determinants on the flagellar filament and is the most commonly used type of antiserum. Measurements of complement fixation and precipitation show that the antibody binds to the subunits in the filament. When the filament is disaggregated, the antibody crossreacts with the flagellin subunit, and the subunit can inhibit the reaction of the antibody with flagellar filaments (5). A second kind of antiserum can be prepared by injecting denatured flagellin. These sera react only with the denatured flagellin subunit and show no apparent reaction with the flagellar filament. In an initial survey, the two kinds of antisera were shown to respond differently to I Present address: Department of Microbiology, University of Virginia Medical School, Charlottesville, VA 22904. 2 Present address: Oncology Center and Department of Pharmacology and Experimental Therapeutics, The Johns Hopkins University School of Medicine, Baltimore, MD 21205. 3Present address: Department of Bacteriology and Immunology, University of California, Berkeley, CA 94720.
variation in flagellin structure. Antifilament antibody reacted with a restricted range of flagellar filaments, whereas antisubunit antibody showed extensive cross-reaction with many of the flagellins tested. These reactions were correlated with the variability observed in the peptide maps of flagellin. The peptide distribution suggested that some sequences were conserved and invariant, whereas others showed a very high degree of variability. Comparisons of partial amino acid sequences of B. subtilis strain 23 and 168 flagellin also suggest that there are "conserved" and "variable" regions in the molecule (3). The antibody response can be interpreted to mean that antifilament antibody reacts primarily with the more variable regions of flagellin, whereas antisubunit antibody is directed against the antigenic determinants of the primary sequence. Furthermore, to be available for reaction with the antifilament antibodies, the more variable regions would have to be on the outer surface of the filament. The determinants that react with the antisubunit antibody and which appear to be more conserved could be on the inner part of the filament, and some of these might be involved in subunit-subunit interactions. The present work was undertaken to extend the study of the variability of B. subtilis flagellins to a larger variety of strains and to try to correlate the serological measurements with the chemical properties of the various flagellin molecules. MATERIALS AND METHODS Bacterial strains, antigens, and antibodies. Most of the B. subtilis strains were from the American Type Culture Collection and are referred to by their American Type Culture Collection number. They were all grown from single colonies and harvested, and flagella were prepared by methods that were
200
VOL. 130, 1977
described previously (3, 5). The flagella filaments were purified to homogeneity as determined by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis. The filaments of all the strains, except 10774, were stable when stored in the cold. Flagella from strain 10774 dissociated in 0.01 M phosphate buffer (pH 7.9), at 40C. All of the flagella exhibited approximately the same wavelength and amplitude when examined by electron microscopy. The antisera were prepared as described previously (5). The antiflagella filament antibody was obtained from the same rabbit as had been immunized previously. The serum previously used was from early bleedings, whereas the serum used in the work described in this paper was from later bleedings. Antiflagellin antisera showed reactions only with flagellin subunits. The index of dissimilarity was determined by complement fixation as described previously (5). Amino acid analysis. Samples (approximately 1 mg of protein) were hydrolyzed in evacuated glass tubes at 110°C for 24 and 72 h with 1.5 ml of 6 N HCI containing 50 ,ul of 5% (wt/vol) phenol in water. Analyses were performed on the Spinco automatic amino acid analyzer model 120B (10). SDS-polyacrylamide gel electrophoresis. SDSpolyacrylamide gel electrophoresis was performed as described by Weber and Osborn (11).
RESULTS Polyacrylamide gel electrophoresis. The results of SDS-polyacrylamide gel electrophoresis of a number of B. subtilis flagellins are illustrated in Fig. 1. In general, only a single component was observed. The apparent molecular weights of the various flagellins determined in this manner, after treatment with 1% SDS at 1000C for 1 min, ranged from 28,000 to 40,000 (Table 1). It was noted, however, that the mobility of certain of these proteins on SDS gels depends on the details of sample pretreatment. For example, after the usual dissociation in SDS, 23 flagellin exhibits an apparent molecular weight of 28,000. If the treatment with SDS is extended to 10 min at 1000C prior to application to the gel, 23 flagellin migrates as a single component of apparent molecular weight of 33,000. Similarly, 9799 flagellin migrates with an apparent molecular weight of 34,000 if it is heated to 90°C in SDS before it is applied to the gel. After boiling for 5 to 10 min, the mobility of 9799 flagellin corresponds to a molecular weight of 40,000. Such observations do not hold true for all of the flagellins. For example, extended boiling in SDS does not influence the migration of 168 flagellin in gels. It should be noted that the apparent molecular weight of 168 flagellin calculated from mobility in SDS gels is 40,000, whereas the actual molecular weight of this macromolecule is 33,000 as calculated from the complete amino acid sequence.
BACILLUS SUBTILIS FLAGELLINS
A
B
201
C D E F G H I J K
FIG. 1. SDS-polyacrylamide gel electrophoresis of flagellins derived from various strains of B. subtilis (A) 7059; (B) 11774; (C) 12100; (D) 12432; (E) 10783; (F) 12695; (G) 4944; (H) 10774; (I) 9466; (J) W23; (K) 168.
TABLE 1. Apparent molecular weights of B. subtilis f1agellins as determined by SDS-polyacrylamide gel electrophoresis Strain no. Apparent mol wt 168 40,000 4944 40,000 9466 39,000 23 10783 4529 12695
33,000 33,000 34,000 33,000
9799
34,000
13542 11838
39,000 40,000
The anomalous mobilities observed for the flagellins suggest that there are residual interactions within flagellin molecules in the presence of SDS and, consequently, these proteins do not behave as ideal random coils. Effects similar to those described above have been observed in comparisons of E. coli and Salmonella flagellins (8). Amino acid analysis. The amino acid compositions are presented in Table 2. The flagellins can be divided into four groups on the basis of their compositions. Group 1 includes strains 168, 4944, 9466, and 7059. This group is charac-
202
J. BACTERIOL.
SIMON ET AL.
TABLE 2. Amino acid composition of flagellins from various strains of B. subtilisa No. of amino acid residues/moleculeb
168c 4944
9466
7059
W23
12695 10783
Group
Group 3
Group 2
Group 1
Amino acid
4529 11774 12432 13542d 11838
9799
12100
6984
10774
20 16 3 3 3 3 3 13 14 16 16 16 15 46 46 52 50 48 50 48 53 32 32 24 24 25 26 26 26 34 33 25 24 21 21 22 24 27 28 Serine 36 46 46 44 45 46 46 45 46 46 Glutamic acid 1 2 1 1 1 1 2 1 1 2 Proline 19 18 20 24 27 27 26 24 25 24 Glycine 37 41 43 37 42 41 41 39 38 38 Alanine 7 10 11 15 11 10 10 10 11 11 Valine 9 13 8 10 11 12 12 11 12 11 9 8 Methionine 16 21 22 23 23 21 22 22 22 20 25 26 Isoleucine 30 32 29 32 32 36 33 34 34 33 31 30 Leucine 1 0 5 0 0 0 0 0 1 1 1 1 Tyrosine 2 2 7 7 7 7 7 7 5 5 5 Phenylalanine 5 a The number of amino acid residues for molecule of B. subtilis 168 flagellin was calculated from the amino acid sequence (2). The composition of the other flagellins is based on an assumed monomer molecular weight of 35,000. b The numbers represent the average of determinations performed after 24 and 72 h of hydrolysis and are given to the nearest whole number except for the serine and threonine values, which were obtained by extrapolation to zero time. c Strain number. d The flagellin was oxidized with performic acid prior to analysis. e Determined as methionine sulfone.
Lysine Histidine Arginine Aspartic acid Threonine
15 4 14 49 18 24 41 2 19 39 14
15 4 14 54 19 24 45 2 21 43 15
15 4 14 54 19
14
15 4 17 51 24
14
13 3 15
15 4
14 3 16
14
18 3 12 47 33 38 39 2 20 42 0 12e 19 29 4 2
22 3 13 44 31 37 45 1 23 36 9 12 17 31 4 2
20 3 13 49 34 37 38 2 20 38 10 11 17 31 5 2
21 3 12 49 33 34 40 2 20 39 10 12 18 32 5 2
terized by two proline, one tyrosine, and five study, either double diffusion in agar or comphenylalanine residues. A second group in- plement fixation was used to test for flagellar cludes strains W23, 4529, 10783, 11774, 12432, antigenicity. A number of clear-cut classes and 12695. The members of this group are char- were established. Anti-168 reacted only with acterized by the presence of a proline residue, flagella from 168, 9466, 4944, and 7059. Antiseven phenylalanines, and the absence of tyro- W23 reacted with W23, 4529, 10783, and 12695. sine. The third group, higher in tyrosine than Anti-13542 reacted with 13542, 12432, and phenylalanine, includes strains 13542, 11838, 11774. Anti-9799 reacted with 9799, 6984, and 9799, 12100, and 6984. Finally, a fourth group 12100. These groupings also corresponded well with a very low tyrosine plus phenylalanine with the groupings established by comparing content is represented by strain 10774. Whereas similarities in amino acid composition. Reaction with antiflagellin antibody. Spethe differences in aromatic amino acid content are most pronounced, the distribution of the cific anti-168 flagellin antibody was prepared. other amino acids is also consistent with this The serum reacted with the flagellin subunit rough classification. For example, there is vari- but not with the filament. In initial experiation in the arginine and lysine content be- ments it was tested by double diffusion in agar tween flagellins of different groups but very against all of the flagellin antigens. It showed a positive reaction against each of these. The anlittle variation within a group. Reaction with antiflagellar filament anti- tigens were then arranged in adjacent wells in body. We showed in previous work that tests a variety of combinations. Antigen 4944 and with antisera prepared against different flagel- 9466 showed lines of ideptity with 168, whereas lar filament antigens can be used to define at all of the other antigens gave a spur with 168. least five distinct serological groups. The anti- These results suggested that the antibody was sera that were used were made against 168, recognizing the antigenic determinants in the W23, 9799, and 13542. It is difficult to do simple 168 group as identical to 168 and was showing agglutination tests since many B. subtilis cross-reaction with all of the other antigens. In strains have a tendency to clump even in the an attempt to quantitate these relationships, absence of specific antibodies. Therefore, in this the antiserum was used to measure the index of
BACILLUS SUBTILIS FLAGELLINS
VOL. 130, 1977
dissimilarity with all of the antigens. The results are shown in Table 3. Whereas the precise numbers are different than those from previous results (5), presumably because a different antiserum was used, the order and the grouping that emerge are consistent with earlier results. The order of relatedness is consistent both with the grouping found using antifilament antibody and with the grouping that emerged from the amino acid analysis. The correlation is not absolutely perfect. Thus, while 6984 shows considerable difference from 9799 in its reaction with antiflagellin antibody, they are apparently the same when examined by amino acid analysis or with antifilament antibody. This test suggests that strain 11838 is part of the 9799 group. However, by using the antifilament antibody it could not be assigned to any of the groups. The antisubunit antibody permits establishment of relationships between flagellins that are classified into different groups by the other criteria. DISCUSSION The data presented here extend previous results that demonstrated extensive variation in the serological properties and amino acid composition ofB. subtilis flagellins. The serological properties can be monitored in two ways: either by using a variety of group-specific antisera, which are made against intact flagellar filaments, or by using antibody against the denatured subunit. Since the latter antibody population detects the extent of homology between the sequences of various flagellins, the wide spectrum of cross-reaction allows for a graded estimate of the relationships among the flagellar antigens. It is interesting that the groupings established on the basis of amino acid compositions correlate well with the serological classifiTABLE 3. Antigenic cross-reactivity of flagellins of B. subtilis Dilution for 50% re- Index of dissimiStrain
action
168 9466 4944 7059 12695 10783 23 4529 11774 12432 11838 12100 9799 6984 10774
1:2,100 1:2,100 1:2,100 1:1,600 1:1,300 1:1,300 1:1,300 1:1,300 1:900 1:900 1:500 1:500 1:500 1:300
Vol. 130, No. 1
Printed in U.S.A.
Classification of Bacillus subtilis Flagellins MELVIN I. SIMON, SUZANNE U. EMERSON,1 JOEL H. SHAPER,2 PATRICIA D. BERNARD, AND ALEXANDER N. GLAZER3* Department of Biology, University of California, San Diego, La Jolla, California 92037, and Departments of Bacteriology and Biological Chemistry, University of California, Los Angeles, California 90024 Received for publication 7 January 1977
Purified flagellins derived from 16 strains of Bacillus subtilis were classified into at least five distinct groups on the basis of their reaction with antiflagellar filament antibody and antiflagellin antibody. This classification was in good accord with that derived independently on the basis of amino acid analyses of the flagellins. Flagellar antigenicity appears to provide a useful typological character in classifying B. subtilis strains. Flagellar serotype variation has been studied extensively in Salmonella and Escherichia coli. This variation is one of the basic criteria used in the classification of Enterobacteriaceae (7). A large degree of antigenic variation has also been observed in flagellar proteins derived from strains of Bacillus subtilis (5). The flagellin derived from B. subtilis strain 168 has been studied in great detail, and the entire amino acid sequence of this protein is known (1, 2, 9). It was therefore of interest to extend the studies on variation in flagellar serotype in B. subtilis both from the point of view of establishing methods to classify B. subtilis strains and to get a clearer picture of the biochemical basis of serotype variation. Two kinds of antisera can be prepared against flagellar protein. The first reacts primarily with determinants on the flagellar filament and is the most commonly used type of antiserum. Measurements of complement fixation and precipitation show that the antibody binds to the subunits in the filament. When the filament is disaggregated, the antibody crossreacts with the flagellin subunit, and the subunit can inhibit the reaction of the antibody with flagellar filaments (5). A second kind of antiserum can be prepared by injecting denatured flagellin. These sera react only with the denatured flagellin subunit and show no apparent reaction with the flagellar filament. In an initial survey, the two kinds of antisera were shown to respond differently to I Present address: Department of Microbiology, University of Virginia Medical School, Charlottesville, VA 22904. 2 Present address: Oncology Center and Department of Pharmacology and Experimental Therapeutics, The Johns Hopkins University School of Medicine, Baltimore, MD 21205. 3Present address: Department of Bacteriology and Immunology, University of California, Berkeley, CA 94720.
variation in flagellin structure. Antifilament antibody reacted with a restricted range of flagellar filaments, whereas antisubunit antibody showed extensive cross-reaction with many of the flagellins tested. These reactions were correlated with the variability observed in the peptide maps of flagellin. The peptide distribution suggested that some sequences were conserved and invariant, whereas others showed a very high degree of variability. Comparisons of partial amino acid sequences of B. subtilis strain 23 and 168 flagellin also suggest that there are "conserved" and "variable" regions in the molecule (3). The antibody response can be interpreted to mean that antifilament antibody reacts primarily with the more variable regions of flagellin, whereas antisubunit antibody is directed against the antigenic determinants of the primary sequence. Furthermore, to be available for reaction with the antifilament antibodies, the more variable regions would have to be on the outer surface of the filament. The determinants that react with the antisubunit antibody and which appear to be more conserved could be on the inner part of the filament, and some of these might be involved in subunit-subunit interactions. The present work was undertaken to extend the study of the variability of B. subtilis flagellins to a larger variety of strains and to try to correlate the serological measurements with the chemical properties of the various flagellin molecules. MATERIALS AND METHODS Bacterial strains, antigens, and antibodies. Most of the B. subtilis strains were from the American Type Culture Collection and are referred to by their American Type Culture Collection number. They were all grown from single colonies and harvested, and flagella were prepared by methods that were
200
VOL. 130, 1977
described previously (3, 5). The flagella filaments were purified to homogeneity as determined by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis. The filaments of all the strains, except 10774, were stable when stored in the cold. Flagella from strain 10774 dissociated in 0.01 M phosphate buffer (pH 7.9), at 40C. All of the flagella exhibited approximately the same wavelength and amplitude when examined by electron microscopy. The antisera were prepared as described previously (5). The antiflagella filament antibody was obtained from the same rabbit as had been immunized previously. The serum previously used was from early bleedings, whereas the serum used in the work described in this paper was from later bleedings. Antiflagellin antisera showed reactions only with flagellin subunits. The index of dissimilarity was determined by complement fixation as described previously (5). Amino acid analysis. Samples (approximately 1 mg of protein) were hydrolyzed in evacuated glass tubes at 110°C for 24 and 72 h with 1.5 ml of 6 N HCI containing 50 ,ul of 5% (wt/vol) phenol in water. Analyses were performed on the Spinco automatic amino acid analyzer model 120B (10). SDS-polyacrylamide gel electrophoresis. SDSpolyacrylamide gel electrophoresis was performed as described by Weber and Osborn (11).
RESULTS Polyacrylamide gel electrophoresis. The results of SDS-polyacrylamide gel electrophoresis of a number of B. subtilis flagellins are illustrated in Fig. 1. In general, only a single component was observed. The apparent molecular weights of the various flagellins determined in this manner, after treatment with 1% SDS at 1000C for 1 min, ranged from 28,000 to 40,000 (Table 1). It was noted, however, that the mobility of certain of these proteins on SDS gels depends on the details of sample pretreatment. For example, after the usual dissociation in SDS, 23 flagellin exhibits an apparent molecular weight of 28,000. If the treatment with SDS is extended to 10 min at 1000C prior to application to the gel, 23 flagellin migrates as a single component of apparent molecular weight of 33,000. Similarly, 9799 flagellin migrates with an apparent molecular weight of 34,000 if it is heated to 90°C in SDS before it is applied to the gel. After boiling for 5 to 10 min, the mobility of 9799 flagellin corresponds to a molecular weight of 40,000. Such observations do not hold true for all of the flagellins. For example, extended boiling in SDS does not influence the migration of 168 flagellin in gels. It should be noted that the apparent molecular weight of 168 flagellin calculated from mobility in SDS gels is 40,000, whereas the actual molecular weight of this macromolecule is 33,000 as calculated from the complete amino acid sequence.
BACILLUS SUBTILIS FLAGELLINS
A
B
201
C D E F G H I J K
FIG. 1. SDS-polyacrylamide gel electrophoresis of flagellins derived from various strains of B. subtilis (A) 7059; (B) 11774; (C) 12100; (D) 12432; (E) 10783; (F) 12695; (G) 4944; (H) 10774; (I) 9466; (J) W23; (K) 168.
TABLE 1. Apparent molecular weights of B. subtilis f1agellins as determined by SDS-polyacrylamide gel electrophoresis Strain no. Apparent mol wt 168 40,000 4944 40,000 9466 39,000 23 10783 4529 12695
33,000 33,000 34,000 33,000
9799
34,000
13542 11838
39,000 40,000
The anomalous mobilities observed for the flagellins suggest that there are residual interactions within flagellin molecules in the presence of SDS and, consequently, these proteins do not behave as ideal random coils. Effects similar to those described above have been observed in comparisons of E. coli and Salmonella flagellins (8). Amino acid analysis. The amino acid compositions are presented in Table 2. The flagellins can be divided into four groups on the basis of their compositions. Group 1 includes strains 168, 4944, 9466, and 7059. This group is charac-
202
J. BACTERIOL.
SIMON ET AL.
TABLE 2. Amino acid composition of flagellins from various strains of B. subtilisa No. of amino acid residues/moleculeb
168c 4944
9466
7059
W23
12695 10783
Group
Group 3
Group 2
Group 1
Amino acid
4529 11774 12432 13542d 11838
9799
12100
6984
10774
20 16 3 3 3 3 3 13 14 16 16 16 15 46 46 52 50 48 50 48 53 32 32 24 24 25 26 26 26 34 33 25 24 21 21 22 24 27 28 Serine 36 46 46 44 45 46 46 45 46 46 Glutamic acid 1 2 1 1 1 1 2 1 1 2 Proline 19 18 20 24 27 27 26 24 25 24 Glycine 37 41 43 37 42 41 41 39 38 38 Alanine 7 10 11 15 11 10 10 10 11 11 Valine 9 13 8 10 11 12 12 11 12 11 9 8 Methionine 16 21 22 23 23 21 22 22 22 20 25 26 Isoleucine 30 32 29 32 32 36 33 34 34 33 31 30 Leucine 1 0 5 0 0 0 0 0 1 1 1 1 Tyrosine 2 2 7 7 7 7 7 7 5 5 5 Phenylalanine 5 a The number of amino acid residues for molecule of B. subtilis 168 flagellin was calculated from the amino acid sequence (2). The composition of the other flagellins is based on an assumed monomer molecular weight of 35,000. b The numbers represent the average of determinations performed after 24 and 72 h of hydrolysis and are given to the nearest whole number except for the serine and threonine values, which were obtained by extrapolation to zero time. c Strain number. d The flagellin was oxidized with performic acid prior to analysis. e Determined as methionine sulfone.
Lysine Histidine Arginine Aspartic acid Threonine
15 4 14 49 18 24 41 2 19 39 14
15 4 14 54 19 24 45 2 21 43 15
15 4 14 54 19
14
15 4 17 51 24
14
13 3 15
15 4
14 3 16
14
18 3 12 47 33 38 39 2 20 42 0 12e 19 29 4 2
22 3 13 44 31 37 45 1 23 36 9 12 17 31 4 2
20 3 13 49 34 37 38 2 20 38 10 11 17 31 5 2
21 3 12 49 33 34 40 2 20 39 10 12 18 32 5 2
terized by two proline, one tyrosine, and five study, either double diffusion in agar or comphenylalanine residues. A second group in- plement fixation was used to test for flagellar cludes strains W23, 4529, 10783, 11774, 12432, antigenicity. A number of clear-cut classes and 12695. The members of this group are char- were established. Anti-168 reacted only with acterized by the presence of a proline residue, flagella from 168, 9466, 4944, and 7059. Antiseven phenylalanines, and the absence of tyro- W23 reacted with W23, 4529, 10783, and 12695. sine. The third group, higher in tyrosine than Anti-13542 reacted with 13542, 12432, and phenylalanine, includes strains 13542, 11838, 11774. Anti-9799 reacted with 9799, 6984, and 9799, 12100, and 6984. Finally, a fourth group 12100. These groupings also corresponded well with a very low tyrosine plus phenylalanine with the groupings established by comparing content is represented by strain 10774. Whereas similarities in amino acid composition. Reaction with antiflagellin antibody. Spethe differences in aromatic amino acid content are most pronounced, the distribution of the cific anti-168 flagellin antibody was prepared. other amino acids is also consistent with this The serum reacted with the flagellin subunit rough classification. For example, there is vari- but not with the filament. In initial experiation in the arginine and lysine content be- ments it was tested by double diffusion in agar tween flagellins of different groups but very against all of the flagellin antigens. It showed a positive reaction against each of these. The anlittle variation within a group. Reaction with antiflagellar filament anti- tigens were then arranged in adjacent wells in body. We showed in previous work that tests a variety of combinations. Antigen 4944 and with antisera prepared against different flagel- 9466 showed lines of ideptity with 168, whereas lar filament antigens can be used to define at all of the other antigens gave a spur with 168. least five distinct serological groups. The anti- These results suggested that the antibody was sera that were used were made against 168, recognizing the antigenic determinants in the W23, 9799, and 13542. It is difficult to do simple 168 group as identical to 168 and was showing agglutination tests since many B. subtilis cross-reaction with all of the other antigens. In strains have a tendency to clump even in the an attempt to quantitate these relationships, absence of specific antibodies. Therefore, in this the antiserum was used to measure the index of
BACILLUS SUBTILIS FLAGELLINS
VOL. 130, 1977
dissimilarity with all of the antigens. The results are shown in Table 3. Whereas the precise numbers are different than those from previous results (5), presumably because a different antiserum was used, the order and the grouping that emerge are consistent with earlier results. The order of relatedness is consistent both with the grouping found using antifilament antibody and with the grouping that emerged from the amino acid analysis. The correlation is not absolutely perfect. Thus, while 6984 shows considerable difference from 9799 in its reaction with antiflagellin antibody, they are apparently the same when examined by amino acid analysis or with antifilament antibody. This test suggests that strain 11838 is part of the 9799 group. However, by using the antifilament antibody it could not be assigned to any of the groups. The antisubunit antibody permits establishment of relationships between flagellins that are classified into different groups by the other criteria. DISCUSSION The data presented here extend previous results that demonstrated extensive variation in the serological properties and amino acid composition ofB. subtilis flagellins. The serological properties can be monitored in two ways: either by using a variety of group-specific antisera, which are made against intact flagellar filaments, or by using antibody against the denatured subunit. Since the latter antibody population detects the extent of homology between the sequences of various flagellins, the wide spectrum of cross-reaction allows for a graded estimate of the relationships among the flagellar antigens. It is interesting that the groupings established on the basis of amino acid compositions correlate well with the serological classifiTABLE 3. Antigenic cross-reactivity of flagellins of B. subtilis Dilution for 50% re- Index of dissimiStrain
action
168 9466 4944 7059 12695 10783 23 4529 11774 12432 11838 12100 9799 6984 10774
1:2,100 1:2,100 1:2,100 1:1,600 1:1,300 1:1,300 1:1,300 1:1,300 1:900 1:900 1:500 1:500 1:500 1:300
