Associations of Escherichia coli K-12 OmpF trimers with rough and smooth lipopolysaccharides.

The associations of both rough and smooth lipopolysaccharides (LPS) with the OmpF porin of Escherichia coli K-12 were examined in galE strains deleted for ompC. Transformation with pSS37 and growth with galactose conferred the ability to assemble a Shigella dysenteriae O antigen onto the core oligosaccharide of E. coli K-12 LPS. The association of LPS with OmpF trimers was assessed by staining, autoradiography of LPS specifically labeled with [1-14C]galactose, and Western immunoblotting with a monoclonal antibody specific for OmpF trimers. These techniques revealed that the migration distances and multiple banding patterns of OmpF porin trimers in sodium dodecyl sulfate-polyacrylamide gels were dictated by the chemotype of associated LPS. Expression of smooth LPS caused almost all of the trimeric OmpF to run in gels with a slower mobility than trimers from rough strains. The LPS associated with trimers from a smooth strain differed from the bulk-phase LPS by consisting almost exclusively of molecules with O antigen.     

Analysis of lipopolysaccharides of Bacteroides fragilis by sodium dodecyl sulphate-polyacrylamide gel electrophoresis and electroblot transfer

Abstract Lipopolysaccharides (LPS) from three strains of Bacteroides fragilis were run on SDS-polyacrylamide gels and stained with silver. Each LPS produced a similar pattern, consisting of a series of regularly spaced discrete bands which decreased in intensity as they increased in M _r value. Electroblot transfer from duplicate SDS gels onto nitrocellulose membrane were reacted with antisera raised to whole cells of two of the strains and antigens were visualised with horse-radish peroxidase-antirabbit-IgG conjugate and colour reagent. Results revealed that the two lowest M _r bands of the LPS preparation (rough LPS) represented common antigens.     

Antigenic analysis of three strains of Mycoplasma arginini by two-dimensional immunoelectrophoresis.

The occurrence of species-specific and strain-specific antigens in three strains of Mycoplasma arginini (G-230, leonis and 23243) was studied by two-dimensional immunoelectrophoresis. Approximately 20 antigenic components could be detected in each strain. It was possible to analyze 6 to 7 major and distinct components from each strain by two techniques: "enhancement" where antigen to an additional strain is added to the first phase of the electrophoresis which increases the size of common peaks and "suppression" where antiserum to an additional strain is incorporated in the second phase whereby peak size of components to which both sera have antibody are decreased. A total of 10 distinct antigens were recognized. Electrophoretic mobilities relative to bovine albumin ranged from 0.2 to 1.08. Three components were common to all strains; two of these represented major amounts of material. Four components represented strain-specific components. Unique fast components were found both in strains 23243 and G-230. Three antigens were distributed into only two of the three strains. The electrophoretic mobilities of some common antigens were quite different between strains.     

Differences Between Brucella Antigens Involved in Indirect Hemagglutination Tests with Normal and Tanned Red Blood Cells

Brucella antigens capable of sensitizing normal and tanned sheep red blood cells for indirect hemagglutination were compared with antigens involved in agglutination, gel diffusion, and immunoelectrophoresis. Hyperimmune rabbit sera, before and after absorption with various antigenic preparations from smooth and rough B. abortus, were used in the tests. Normal erythrocytes could be sensitized with an NaOH-treated ether-water extract (EW-T) of smooth Brucella. Tanned erythrocytes could be sensitized with a water-soluble extract from ultrasonically disrupted smooth or rough Brucella. The EW-T produced a single precipitation band and the water-soluble antigens produce 6 to 23 bands in immunoelectrophoresis with unabsorbed sera. After absorption of antisera with water-soluble extracts from smooth or rough Brucella cells or from smooth or rough cell walls, the hemagglutinins for sensitized tanned erythrocytes and the precipitins for water-soluble antigens were removed. Absorption with living smooth or rough Brucella cells or with EW-T did not remove these antibodies. The precipitins and hemagglutinins for the antigen EW-T, and agglutinins for smooth cells, were absorbed by smooth antigens but not by rough antigens. It appears that the antigens which sensitize tanned erythrocytes and diffuse through agar gels are present in both smooth and rough forms and may be situated in the cytoplasm or in the internal part of the cell wall, whereas the agglutinogen and the antigen which attaches to normal erythrocytes are surface antigens found only on the smooth Brucella cell.     

The Antigens Associated with the Cell Walls of Members of the Genus Pseudomonas

Three antigens were associated with the cell walls of pseudomonads. A highly antigenic, strain-specific antigen of high molecular weight and protein or lipoprotein in nature, occurred as an envelope around the cells. It could be washed off the cells closely associated with carbohydrate material but its antigenicity was not dependent on the carbohydrate present. Another antigen, common to all strains tested, was situated below the first antigen. This was less antigenic than the strain-specific antigen and was polysaccharide or lipopolysaccharide in nature. A second common antigen was the mucopeptide of the cell walls. This had an antigenicity similar to that of the second antigen and was dependent on both the carbohydrate and polypeptide components of the macromolecule. There appears to be some correlation between these findings and the structure of cell walls of pseudomonads are shown by electron microscopy.     

Pseudomonas aeruginosa lipopolysaccharide: evidence that the O side chains and common antigens are on the same molecule.

We investigated whether Pseudomonas aeruginosa produces two distinct lipopolysaccharides (LPS) containing either serologically variable O side chains or a neutral polysaccharide common antigen, designated A bands, that reacts with monoclonal antibody (MAb) E87. Immunoprecipitation of LPS and free O side chains with O-side-chain-specific antibodies or MAb E87 resulted in coprecipitation of both polysaccharides when antibody of either specificity was employed. Chromatography of LPS and free O side chains in a disaggregating deoxycholate buffer indicated the two polysaccharide antigens cochromatograph when eluates were analyzed by sensitive and specific enzyme-linked immunosorbent assay inhibitions. The LPS from a mutant of strain PAO1 that lacks polymerized O side chains but retains the common antigen eluted in fractions containing smaller LPS molecules, indicating the necessity of polymerized O side chains for elution in early fractions containing large LPS monomers. A phosphomannomutase mutant of P. aeruginosa PAO1 makes a rough LPS lacking both O side chains and common antigen but still produces a small (< 6-kDa) common antigen component detectable in cell lysates. Introduction of the cloned pmm gene into this strain restored production of a smooth LPS expressing large MAb E87-reactive common antigen. Destruction with NaOH of O side chains on recombinant LPS molecules eluting early from the molecular sieve column resulted in a shift of the MAb E87-reactive antigen to the late-eluting fractions. These results indicate that on most P. aeruginosa LPS molecules, O side chains and neutral polysaccharide common antigens are both present.     

Epitope identification for a panel of anti-Sinorhizobium meliloti monoclonal antibodies and application to the analysis of K antigens and lipopolysaccharides from bacteroids.

In two published reports using monoclonal antibodies (MAbs) generated against whole cells, Olsen et al. showed that strain-specific antigens on the surface of cultured cells of Sinorhizobium meliloti were diminished or absent in the endophytic cells (bacteroids) recovered from alfalfa nodules, whereas two common antigens were not affected by bacterial differentiation (P. Olsen, M. Collins, and W. Rice, Can. J. Microbiol. 38:506-509, 1992; P. Olsen, S. Wright, M. Collins, and W. Rice, Appl. Environ. Microbiol. 60:654-661, 1994). The nature of the antigens (i.e., the MAb epitopes), however, were not determined in those studies. For this report, the epitopes for five of the anti-S. meliloti MAbs were identified by polyacrylamide gel electrophoresis-immunoblot analyses of the polysaccharides extracted from S. meliloti and Sinorhizobium fredii. This showed that the strain-specific MAbs recognized K antigens, whereas the strain-cross-reactive MAbs recognized the lipopolysaccharide (LPS) core. The MAbs were then used in the analysis of the LPS and K antigens extracted from S. meliloti bacteroids, which had been recovered from the root nodules of alfalfa, and the results supported the findings of Olsen et al. The size range of the K antigens from bacteroids of S. meliloti NRG247 on polyacrylamide gels was altered, and the epitope was greatly diminished in abundance compared to those from the cultured cells, and no K antigens were detected in the S. meliloti NRG185 bacteroid extract. In contrast to the K antigens, the LPS core appeared to be similar in both cultured cells and bacteroids, although a higher proportion of the LPS fractionated into the organic phase during the phenol-water extraction of the bacteroid polysaccharides. Importantly, immunoblot analysis with an anti-LPS MAb showed that smooth LPS production was modified in the bacteroids.     

Separation and characterization of two chemically distinct lipopolysaccharides in two Pectinatus species.

Lipopolysaccharides (LPS) from the type strains of the anaerobic beer spoilage bacteria Pectinatus cerevisiiphilus and P. frisingensis were extracted with the 5:5:8 volume ratio modification of the phenolchloroform-petroleum ether method (H. Brade and C. Galanos, Eur. J. Biochem. 122:233-237, 1982). Sequential precipitations of LPS with water and acetone from the phenol phase yielded LPS which differed in that water-precipitable material (LPS-H2O; 0.1 to 0.4% of the dry weight of the cells) was rough-type LPS, whereas acetone-precipitable material (LPS-Ac; 4.6 to 5.8% of the dry weight) contained both rough-type LPS and high-molecular-weight material resembling smooth LPS. The LPS were chemically characterized, and they contained D-glucosamine, 4-amino-4-deoxy-L-arabinose, 3-deoxy-D-manno-2-octulosonic acid, D-fucose, D-galactose, D-glucose, D-mannose, and phosphate. D-Fucose was present mostly in LPS-Ac, suggesting that it is a constituent of the O antigen. The major fatty acids were ester- and amide-linked (R)-3-hydroxytridecanoic and ester-linked undecanoic acids, with minor amounts of ester-linked tridecanoic and (R)-3-hydroxyundecanoic acids. The chemical compositions of LPS-H2O and LPS-Ac suggested that they differ not only in their smooth or rough nature but also in the structure of their core regions. This may explain their different precipitabilities from the extraction mixture. The extraction method was also shown to be applicable to the isolation of smooth-type LPS from Salmonella enterica serovar Typhimurium. Extraction of two Typhimurium strains carrying chemically different O antigens resulted in high yields (8% of the dry weight) of LPS. Strain SH2183, which contains the relatively hydrophobic O-4,5,12 antigen yielded almost exclusively LPS-Ac, whereas the LPS of strain SH5770, which has a hydrophilic O-6,7 antigen, was exclusively LPS-H2O. No fractionation to smooth and rough LPS occurred with the Typhimurium strains.     

Distribution of the rol gene encoding the regulator of lipopolysaccharide O-chain length in Escherichia coli and its influence on the expression of group I capsular K antigens.

The rol (cld) gene encodes a protein involved in the expression of lipopolysaccharides in some members of the family Enterobacteriaceae. Rol interacts with one or more components of Rfc-dependent O-antigen biosynthetic complexes to regulate the chain length of lipopolysaccharide O antigens. The Rfc-Rol-dependent pathway for O-antigen synthesis is found in strains with heteropolysaccharide O antigens, and, consistent with this association, rol-homologous sequences were detected in chromosomal DNAs from 17 different serotypes with heteropolysaccharide O antigens. Homopolymer O antigens are synthesized by a pathway that does not involve either Rfc or Rol. It was therefore unexpected when a survey of Escherichia coli strains possessing mannose homopolymer O8 and O9 antigens showed that some strains contained rol. All 11 rol-positive strains coexpressed a group IB capsular K antigen with the O8 or O9 antigen. In contrast, 12 rol-negative strains all produced group IA K antigens in addition to the homopolymer O antigen. Previous research from this and other laboratories has shown that portions of the group I K antigens are attached to lipopolysaccharide lipid A-core, in a form that we have designated K(LPS). By constructing a hybrid strain with a deep rough rfa defect, it was shown that the K40 (group IB) K(LPS) antigen exists primarily as long chains. However, a significant amount of K40 antigen was surface expressed in a lipid A-core-independent pathway. The typical chain length distribution of the K40 antigen was altered by introduction of multicopy rol, suggesting that the K40 group IB K antigen is equivalent to a Rol-dependent O antigen. The prototype K30 (group IA) K antigen is expressed as short oligosaccharides (primarily single repeat units) in K(LPS), as well as a high-molecular-weight lipid A-core-independent form. Introduction of multicopy rol into the K30 strain generated a novel modal pattern of K(LPS) with longer polysaccharide chains. Collectively, these results suggested that group IA K(LPS) is also synthesized by a Rol-dependent pathway and that the typically short oligosaccharide K(LPS) results from the absence of Rol activity in these strains.     

Epitope Identification for a Panel of Anti-Sinorhizobium meliloti Monoclonal Antibodies and Application to the Analysis of K Antigens and Lipopolysaccharides from Bacteroids

In two published reports using monoclonal antibodies (MAbs) generated against whole cells, Olsen et al. showed that strain-specific antigens on the surface of cultured cells of Sinorhizobium meliloti were diminished or absent in the endophytic cells (bacteroids) recovered from alfalfa nodules, whereas two common antigens were not affected by bacterial differentiation (P. Olsen, M. Collins, and W. Rice, Can. J. Microbiol. 38:506–509, 1992; P. Olsen, S. Wright, M. Collins, and W. Rice, Appl. Environ. Microbiol. 60:654–661, 1994). The nature of the antigens (i.e., the MAb epitopes), however, were not determined in those studies. For this report, the epitopes for five of the anti-S. meliloti MAbs were identified by polyacrylamide gel electrophoresis-immunoblot analyses of the polysaccharides extracted from S. meliloti and Sinorhizobium fredii. This showed that the strain-specific MAbs recognized K antigens, whereas the strain-cross-reactive MAbs recognized the lipopolysaccharide (LPS) core. The MAbs were then used in the analysis of the LPS and K antigens extracted from S. meliloti bacteroids, which had been recovered from the root nodules of alfalfa, and the results supported the findings of Olsen et al. The size range of the K antigens from bacteroids of S. meliloti NRG247 on polyacrylamide gels was altered, and the epitope was greatly diminished in abundance compared to those from the cultured cells, and no K antigens were detected in the S. meliloti NRG185 bacteroid extract. In contrast to the K antigens, the LPS core appeared to be similar in both cultured cells and bacteroids, although a higher proportion of the LPS fractionated into the organic phase during the phenol-water extraction of the bacteroid polysaccharides. Importantly, immunoblot analysis with an anti-LPS MAb showed that smooth LPS production was modified in the bacteroids.     

Heterogeneity of lipopolysaccharides from Pseudomonas aeruginosa: analysis of lipopolysaccharide chain length.

Lipopolysaccharide (LPS) from smooth strains of Pseudomonas aeruginosa 503, PAZ1, PAO1715, PAO1716, and Z61 was fractionated by gel filtration chromatography. LPS samples from the first four strains, all PAO1 derivatives, separated into three major size populations, whereas LPS from strain Z61, a Pac K799/WT mutant strain, separated into two size populations. When column fractions were applied to sodium dodecyl sulfate-polyacrylamide gels in their order of elution, molecules of decreasing size were resolved, and the ladder of molecules with different-length O antigens formed a diagonal across the gel. The LPS from the PAO1 derivatives contained two distinct sets of bands, distinguished on the gels as two sets of diagonals. The set of bands with the faster mobility, the B bands, was found in column fractions comprising the three major amino sugar-containing peaks. In the sample from strain 503, a fourth minor peak which contained B bands was resolved. The slower-moving set of bands, the A bands, were recovered in a minor peak. LPS from strain Z61 contained only one set of bands, with the higher-molecular-weight molecules eluting from the column in a volume similar to that of the B bands of the PAO1 strains. Analysis of the fractions of LPS from all strains indicated that less than 8% of the LPS molecules had a long, attached O antigen. Analysis of the peak that contained mainly A bands indicated a lack of reactive amino sugar and phosphate, although heptose and 2-keto-3-deoxyoctulosonic acid were detected. Reaction of isolated fractions with monoclonal antibody specific for the PAO1 O-antigen side chain indicated that only the B bands from the PAO1 strains were antigenically reactive. The bands from strain Z61 showed no reactivity. The data suggest that the A and B bands from the PAO1 strains are antigenically distinct. We propose that PAO1 strains synthesize two types of molecules that are antigenically different.     

Association of a 38 kDa bovine serum protein with the outer membrane of Bordetella pertussis

A series of isogenic mutants lacking either the O1 (O-:K66) or K66 (O1:K-) antigens or both (O-:K-), some of which had additional defects in their LPS core polysaccharide was used to examine the interaction between polymorphonuclear leucocytes (PMNLs) and K. pneumoniae serotype O1:K66. In the absence of serum complement, only a O-:K- strain with a deep rough LPS chemotype elicited a PMNL-dependent chemiluminescent (CL) response. However, following opsonization of the non-capsulated strains by complement, the largest CL response was to the O1:K- mutant. This mutant also activated and bound more complement C3 than any of the other encapsulated or non-capsulated strains examined. Despite the surface exposure of smooth and rough LPS in the encapsulated parent and mutant strains, the K66 antigen reduced the binding of C3 and prevented PMNL activation. Both anti-LPS and anti-K66 antibodies, however, stimulated a PMNL-dependent CL response to the K66 bearing strains.     

A Yersinia enterocolitica serotype 0:3 lipopolysaccharide-specific monoclonal antibody reacts more strongly with bacteria cultured at room temperature than those cultured at 37 degrees C

It has been suggested that the O-side chains of the lipopolysaccharide (LPS) of serotype 0:3 strains of Yersinia enterocolitica vary quantitatively and qualitatively depending on whether they are cultured at 37 degrees C or 25 degrees C. It is uncertain whether this affects the expression of the serotype-specific antigens that are probably carried on the LPS. We studied this question with a serotype 0:3-specific monoclonal antibody, 2C1. This monoclonal antibody immunoprecipitated a 39K major protein from detergent-solubilized 125I-labeled Yersinia preparation. However, results of Western blot experiments demonstrated that the antigens reactive with 2C1 were not actually the 39K protein but the O-side chains of the LPS. Significantly, reactivity could be detected whether the bacteria were cultured at room temperature or at 37 degrees C. However, absorption experiments confirmed that there were less accessible antigenic determinants on the 37 degrees C-LPS. The LPS preparations were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and silver staining. These SDS-PAGE profiles showed that less O-side chains were present in the 37 degrees C-LPS. Hence, the most likely explanation for our observation is that the 37 degrees C incubation condition induced a partial smooth to rough transition of the Yersinia LPS with a decrease in the amount of 2C1-reactive antigen.     

Characterization of polyagglutinating and surface antigens in Pseudomonas aeruginosa

Mutants of Pseudomonas aeruginosa PAC1R (serotype O:3) which were resistant to bacteriophage D were isolated and shown to react with O:5d, O:9 and O:13 antisera as well as O:3. Antisera to the parent strain and to the three polyagglutinating (PA) mutants also showed cross-reactions. The mutants differed from the parent strain in their lipopolysaccharide (LPS) composition. The LPS from two of the three mutants yielded high molecular weight polysaccharide fractions. Although the high molecular weight fraction from one of the mutants contained the amino sugars and other components characteristic of the O:3 serotype strains, its mobility on Sephadex G75 was different from that of the parent strain. The high molecular weight material from the second mutant lacked the O-antigenic determinants but these were present in a semi-rough LPS fraction. The third mutant appeared rough and completely lacked the O-antigenic components. These three mutants were compared with the parent strain and with a non-agglutinating LPS-defective mutant which lacked both O-antigenic side chains and all neutral sugars in the outer core. Agglutination with absorbed sera and haemagglutination using purified LPS and ELISA detection suggested that wall components other than LPS were responsible for some of the cross-reactions observed. The components responsible were detected after SDS-PAGE of crude outer membrane fractions by a combination of Coomassie blue and silver-staining and antigenic components were detected by immunoelectrophoresis and ELISA-linked immunoblotting of the gels. The main antigenic determinants detected by antiserum to the parent strain were in the high molecular weight O-polysaccharide fractions and in the semirough fractions of the LPS, with some activity due to the H protein of the outer membrane. O:5d antisera reacted with unidentified high molecular weight polysaccharide fractions. Cross-reactions with the O:9 antiserum appeared to be due mainly to the F porin and, to a lesser extent, to the G and E proteins of the outer membrane. O:13 antiserum reacted with high molecular weight polysaccharide fractions but also with the rough core and F and H protein. Cross-reactivity of the other three mutant antisera could largely be interpreted in terms of the outer membrane components exposed in each strain. One reacted strongly with the F porin and the rough core, while the others reacted with a number of protein and LPS-derived fractions.(ABSTRACT TRUNCATED AT 400 WORDS)     

Surface antigens of Proteus mirabilis revealed by electroblotting from sodium dodecyl sulphate-polyacrylamide gels

Western Blotting of whole cell preparations of three strains of Proteus mirabilis after separation by electrophoresis on SDS-polyacrylamide gels revealed a complex pattern of antigens. Similar antigen profiles were obtained with isolated outer membranes indicating that the majority of cell surface antigens are located in the outer membrane. Major outer membrane proteins were strongly antigenic and cross-reactive. The highly immunogenic flagella were detected in whole cell preparations and visible in isolated outer membranes. Whereas the protein and flagellar antigens were cross-reactive, lipopolysaccharide (LPS) could only be detected as immunoreactive material using homologous antisera for each strain. The LPS appeared as two broad bands (high and low Mr, respectively) in immunoblots of whole cells, isolated outer membranes and purified LPS. However, isolated LPS could be resolved into multiple sharp bands when 4 M urea was included in the gel system. These discrete bands are assumed to represent differing O antigen chain lengths of the LPS as reported for other Gram-negative organisms.     

The chemical composition of the lipopolysaccharide from Pseudomonas aeruginosa strain PAO and a spontaneously derived rough mutant.

The chemical composition of the lipopolysaccharide (LPS) of the smooth strain Pseudomonas aeruginosa PAO 307 and a spontaneously derived rough mutant, obtained by selection for resistance to the LPS-specific phage E79, are compared. The rough LPS was shown to contain lipid A, heptose, 2-keto 3-deoxyoctonic acid, galactosamine, alanine and phosphate but lacked glucose, rhamnose and fucosamine which were important constituents, on a weight basis, of the smooth LPS. These results, and chromatographic analysis of the polysaccharide fraction indicate that the rough strain lacked side chain material and was defective in its inner core region. The chemical date obtained were consistent with a core in the PAO strain similar to that of strain NCTC 1999, enhancing the evidence for a common core polysaccharide in the LPS of P. aeruginosa strains.     

Mutants of Escherichia coli O9:K30 with altered synthesis and expression of the capsular K30 antigen

Escherichia coli K30 produces a thermostable group I capsular polysaccharide. Two classes of mutants were isolated with defects in the synthesis or expression of capsule. The most common mutant phenotype was acapsular (K-), with no K-antigen synthesized. A second class of mutants, termed Ki or intermediate forms, produced colonies which were indistinguishable from those of acapsular forms yet K-antigenicity was expressed. Previous studies had demonstrated that E. coli strains that produce K30 antigen synthesize a lipopolysaccharide (LPS) fraction that is recognised by monoclonal antibodies against the K30 antigen. Synthesis of this LPS fraction was not affected in Ki forms. The results of morphological examination, LPS analysis and phage sensitivity studies are consistent with the interpretation that the defect in Ki strains results from an inability to polymerize the K30 antigen. Using plasmid pULB113 (RP4::mini-Mu), mutations resulting in both K- and Ki phenotypes were localized near the his region of the chromosome.     

Antigenic determinants and specificity of antisera against acidophilic bacteria

Polyclonal antisera, produced against whole cells of Thiobacillus thiooxidans, T. ferrooxidans and Leptospirillum ferrooxidans, gave highly specific reactions when cross-reacted with 23 strains of acidophilic bacteria using an immunofluorescence (IF) staining technique. Strains of identical serotype exhibited maximum cross-reaction whereas strains of different serotype reacted only weakly. Lipopolysaccharides (LPS) examined by SDS-PAGE showed different, serotype-specific migration patterns indicating their rough or smooth character. LPS patterns may therefore be used for serological classification of acidophilic bacteria. Surface antigens of four strains were identified by immunoblot staining; LPS and some proteins were antigenic determinants with LPS the most specific.The authors are with the Universität Hamburg, Institut für Allgemeine Botanik, Abteilung für Mikrobiologie, Ohnhorststrasse 18, D-22609 Hamburg, Germany     

Chemical basis of rough and smooth variation in mycobacteria.

Rough and smooth colony variants of Mycobacterium kansasii were compared with respect to surface glycolipid composition. Thin-layer chromatography of the native glycolipid antigens, gas chromatography of the constituent sugars, and in situ probing with an appropriate monoclonal antibody by colony dot blot enzyme-linked immunosorbent assay and immunogold labeling demonstrated that all M. kansasii strains of smooth colony morphology contain on their surfaces the recently described trehalose-containing lipooligosaccharides, whereas all rough variants were devoid of such surface antigens. Yet all strains, rough and smooth, contained another glycolipid, the M. kansasii-specific phenolic glycolipid. Previous studies by others had shown that the rough forms of M. kansasii persist longer than smooth variants in experimentally infected mice. Therefore, this study may provide some insight into the question of the chemical basis of pathogenesis in certain mycobacteria.