APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Sept. 1986, p. 425-429 Vol. 52, No. 3 0099-2240/86/090425-05$02.00/0 Copyright C) 1986, American Society for Microbiology Identification of Eubacteria Isolated from Leaf Cavities of Four Species of the N-Fixing Azolla Fern as Arthrobacter Conn and Dimmick W. H. WALLACE' AND J. E. GATESL12* Department ofBiology' and Department of Microbiology and Immunology,2 Virginia Commonwealth University, Richmond, Virginia 23284 Received 12 March 1986/Accepted 3 June 1986

The nutritional and physiological characteristics of 15 isolates from four species of the Azolla fern were determined. Although some minor variation existed in levels of activity, ability to utilize xylose, and formation of N2 gas from NO3, all 15 isolates were rather similar and believed to represent a single species. These eubacteria exhibited aminopeptidase activity and became viscous when treated with KOH, similar to gram-negative organisms; however, the absence of lipopolysaccharide and 2-keto-3-deoxyoctonate in cell walls indicated that they are truly gram-positive organisms. They are unusual because peptidoglycan could not be detected during most of their growth cycle. The presence of lysine as the major diamino acid in cell wall hydrolysates, the inability to hydrolyze cellulose, and the distinctive developmental pattern with rods and "V" forms present during log phase, becoming progressively shorter until cocci dominated during stationary and death phases, indicated that these organisms belong to the genus Arthrobacter Conn and Dimmick. With the exception of the inability to hydrolyze gelatin, their characteristics are consistent with those of the type species, Arthrobacter globiformis Conn and Dimmick.

Previously, we reported that coryneform bacteria were and maintained as stock cultures. The ferns were cultured present in the leaf cavities of the nitrogen-fixing fern Azolla under continuous illumination by white fluorescent lights at caroliniana Willd. and that the numbers of these eubacteria an intensity of 10 klx at a temperature of 26°C in modified exceeded those of the cyanobacteria responsible for the IRRI medium (16) containing K2SO4 (87 ppm), CaCl2 - 2H20 nitrogen fixation process (8). These aerobic eubacteria were (147 ppm), MgSO4- 7H20 (394 ppm), NaH2PO4 (83 ppm), gram variable, gram positive for a brief duration and FeC6H507 -5H20 (12 ppm) supplemented with during log phase but gram negative throughout the rest of micronutrients recommended by Allen (1). The final pH of their growth cycle. Their cell shape and developmental the medium was adjusted to 5.5 with 0.1 N NaOH. Cultures pattern with "V" forms produced during log phase and cocci were transferred to fresh medium every 12 to 15 days. produced during stationary and death phases were typical of of eubacteria. Eubacteria were isolated either by the genus Arthrobacter Conn and Dimmick. homogenizing surface-sterilized dorsal lobes of fern leaves Although we described these bacteria as being gram as previously described (8) or by aseptically removing cavity positive, transmission electron micrographs have indicated contents from surface-sterilized leaves, using a very fine an atypical cell wall structure with little to no peptidoglycan capillary tube. Cavity homogenates or contents were plated evident. Moreover, others have reported isolation of the out on plate count agar (Difco Laboratories, Detroit, Mich.) gram-negative bacteria Pseudomonas (3), Caulobacter, and plates, using sterile buffered water (2) as the diluent. Single- Alcaligenes (11) spp. from Azolla sp. The eubacteria isolated colony isolations were made after 5 days at 30°C. Fifteen in our lab (8) key to either Pseudomonas or Alcaligenes sp. isolates from the four fern species were compared by using if they are considered to be gram negative and are identified the following procedures. via the Roche Oxi-Ferm multitest system. Because other Gram classification. Methods used to clarify the Gram bacteria initially described as coryneforms have been reclas- classification of these bacteria included Gregersen's KOH sified as gram-negative organisms (18, 19), it was imperative test (9), Wiegel and Quandt's polymyxin B procedure for to determine the Gram classification of these eubacterial detecting lipopolysaccharide (LPS) in the cell wall (19), isolates from Azolla sp. by means other than conventional analysis of cell wall for 2-keto-3-deoxyoctonate (KDO) (10, staining. 12, 17), and testing intact cells for aminopeptidase activity This report also provides additional information on the (5). physiological and nutritional characteristics of these bacteria (i) KOH test. Five drops of 3% KOH were placed on a and compares isolates from four fern species: A. caroliniana clean glass slide and a small amount of a 24-h-old plate count Willd., A. mexicana Presl., A. filiculoides Lam., and A. agar culture was added with a toothpick. After gentle mixing pinnata R.Br. for about 10 s the toothpick was raised slowly from the slide to determine if the KOH-cell suspension became viscous and MATERIALS AND METHODS a thread of slime followed the toothpick as it was raised, Fern culture methods. Azolla isolates were obtained from indicating a gram-negative organism. Gram-positive orga- D. W. Rains, University of California, Davis. Samples were nisms form only a watery suspension and no thread of slime either screened immediately for the eubacteria or cultured (9). All 15 isolates were tested along with the gram-positive controls Arthrobacter globiformis, Corynebacterium * Corresponding author. xerosis, Bacillus subtilis, Staphylococcus aureus, and 425 426 WALLACE AND GATES APPL. ENVIRON. MICROBIOL.

Micrococcus luteus and the gram-negative controls Pseu- Pa.) for 4 min at maximum setting. The disrupted cells were domonas aeruginosa, Alcaligenes fecalis, Escherichia coli, immediately heated to 60°C for 10 min to inactivate autolytic Enterobacter aerogenes, Salmonella typhimurium, and enzymes. Whole cells were removed by centrifuging at 1,000 Proteus mirabilis. x g for 8.5 min. The suspension containing cell walls and (ii) Aminopeptidase test. A small amount of bacteria from a cytoplasmic components was centrifuged at 25,000 x g for 24-h-old plate count agar culture was placed in the well of a 20 min to pellet the crude cell wall fraction, which was spot plate containing 0.1 ml of 4% L-alanine-4-nitroanilide. suspended in 100 ml of chloroform-saturated 0.5 M phos- The formation of a yellow color after 5 min indicated the phate buffer (pH 7.6). DNase (0.5 ml of a 1-mg/ml solution) presence of aminopeptidase activity (5). The above-listed and RNase (1 ml of a 5-mg/ml solution) were added to the gram-negative and gram-positive bacteria served as controls. crude wall suspension and incubated at 37°C for 30 min. (iii) Detection of LPS with polymyxin B. Eubacteria were Crystalline trypsin (0.1%, wt/vol) was added, and the sus- cultured in plate count broth (PCB) (1 g of dextrose [Difco], pension was shaken at 37°C until the A280 remained constant, 5 g of tryptone [Difco], and 2.5 g of yeast extract per liter of indicating complete hydrolysis of contaminating cytoplasmic distilled H20). Cells were harvested by centrifugation proteins (ca. 18 h). The mixture was centrifuged at 1,000 x (10,000 x g) at late log phase (determined turbidimetrically), g for 10 min and the cell wall material was recovered from washed twice in Tris hydrochloride buffer (10 mM, pH 7.2), the supernatant by centrifugation at 18,000 x g for 16 min. suspended in cold (0°C) Tris hydrochloride buffer containing The pelleted wall material was washed twice in 0.9% saline 0, 100, or 300 ppm of polymyxin B, and incubated at 35°C for in 0.1 M phosphate buffer (pH 7) and pelleted by centrifuging 45 min. The cells were centrifuged at 3,000 x g and washed for 16 min at 18,000 x g. Cell wall hydrolysis was accom- three times with cold (0°C) Tris hydrochloride buffer to plished by boiling in 6 N HCl for 18 to 20 h in sealed tubes. remove the antibiotic. The final pellet was fixed in glutaral- The hydrolysate was filtered through a 0.45-,um membrane dehyde for 3.5 h at 4°C and prepared for transmission filter (Millipore Corp., Bedford, Mass.) and evaporated to electron microscopy (15). Electron micrographs were exam- dryness in a rotary evaporator at 45°C. The residue was ined for bleb formation on the bacterial cell surface, indicat- taken up in 10 ml of 10% isopropanol, and the amino acid ing the presence of LPS and therefore a gram-negative composition was determined with a model 116 amino acid classification (19). The Azolla eubacteria were compared to a analyzer (Beckman Instrument, Inc., Fullerton, Calif.) with gram-positive organism, B. subtilis, and a gram-negative ninhydrin detection. Resulting absorbance peak height times organism, P. aeruginosa. peak width at half the peak height divided by a predeter- (iv) KDO analysis. Eubacterial cells were cultured and mined constant for each amino acid determined the concen- harvested as described above, Wvashed twice in 0.9% saline tration of each amino acid in the sample added to the at 0°C, and frozen. About 20 g of frozen cells was suspended column. The equipment had been previously standardized in 350 ml of 65 to 68°C H20. The LPS was extracted with by using a synthetic mixture of amino acids. Each compo- 90% phenol at 68°C for 15 min with vigorous stirring. The nent of this mixture had been found to be quantitatively suspension was cooled to 10°C and centrifuged at 10,000 x g recoverable in a single peak (14). for 30 min. The upper (aqueous) layer was removed and Antibiotic susceptibility. Fifteen isolates from four fern dialyzed for 3 days against distilled H20 to remove the species were cultured in PCB for 16 h. Plate count agar pour phenol and low-molecular-weight contaminants. The dialy- plates were prepared by using a 5% (vol/vol) inoculum from sate was concentrated to 10 ml, using a rotary evaporator at these cultures. Disks of ampicillin (10 jig), penicillin G (10 37°C, and the concentrate was centrifuged at 3,000 x g for 10 pLg), carbenicillin (100 [ig), bacitracin (10 U), aureomycin (10 min. The supernatant was then centrifuged at 100,000 x g for ,ug), erythromycin (15 pRg), kanamycin (30 pLg), neomycin (10 1 h, and the LPS pellet was washed five times in 0.9% saline. Rg), streptomycin (10 ,ug), polymyxin B (300 U), and tetra- The final precipitate was suspended in a small volume of cycline (10 jig) were aseptically applied with a Difco auto- distilled H20 and then lyophilized to dryness (12). KDO was matic dispensor. The plates were incubated at 30°C for 48 h, assayed by treating 2 nig of the LPS with 1 ml of 0.2 N and the diameters of the zones of inhibition were measured. H2SO4 at 100°C for 36 min. A 0.5-ml sample of this clear Standard determinative procedures. The 15 isolates were solution was mixed with 0.25 ml of 0.04 M H104 in 0.125 N subjected to a range of standard determinative bacteriologi- H2SO4. After 20 min at room temperature, 0.25 ml of 2.6% cal procedures (13). The Oxi-Ferm multitest system (Roche NaAsO2 in 0.5 N HCl was added. The mixture was allowed Diagnostic Systems, Nutley, N.J.) was also used to charac- to stand at room temperature until the brown color disap- terize these bacteria. peared. A 0.5-ml volume of 0.6% aqueous thiobarbituric acid Nutritional analyses. Growth studies were conducted to was added, and the mixture was heated for 15 min at 100°C. determine the minimal nutritional requirements of these A red product was formed which was kept in solution at bacteria. Rates of growth and cell yield in a variety of room temperature by adding 1 ml of dimethyl sulfoxide, chemically defined culture media were determined measured spectrophotometrically at 548 nm, and compared turbidimetrically, using a Klett-Summerson colorimeter with with a KDO standard curve prepared in the same manner a no. 66 (red) filter, and compared with those in PCB. Linear (10). An LPS extract from an isolate of the eubacteria from regressions determined the best fit of log-phase points to the A. filiculoides and P. aeruginosa, a posititve control, were line used in calculating generation times. assayed for KDO by this method. Peptidoglycan extraction and amino acid analysis. The method of Cummins and Harris (7) as modified by Work (20) RESULTS was used to extract cell wall peptidoglycan from an isolate of Gram classification. The Azolla eubacteria stained gram the eubacteria from A. filiculoides. Approximately 15 g negative during most of the growth cycle, resisting destain- (fresh weight) of log-phase cells was harvested from PCB by ing for only a brief period during early log phase. These centrifuging at 5,000 x g, suspended in cold 0.9% saline organisms formed a slimy, viscous suspension when treated (4°C), placed in an ice bath, and disrupted with a Sonic with KOH, similar to the gram-negative control organisms Dismembrator model 300 (Fisher Scientific Co., Pittsburgh, tested. The gram-positive control organisms consistently VOL. 52, 1986 AZOLLA EUBACTERIA 427 yielded only watery suspensions which did not adhere to the TABLE 1. Results of tests used to determine the Gram toothpick. Aminopeptidase was detected in the Azolla classification of the Azolla eubacteria eubacteria and the gram-negative control organisms but was Test Result Interpretation universally absent in the gram-positive controls. Electron micrographs of the polymyxin B (100 and 300 Gram staining Gram variable Gram positive control cells of P. aeruginosa KOH test KOH penetrates cell Gram negativea ppm)-treated gram-negative Aminopeptidase test Activity in intact cells Gram negativeb produced typical blebs, indicating the presence of an LPS Polymyxin B test LPS absent Gram positive outer envelope (Fig. la). Such blebs were absent on un- KDO assay KDO absent Gram positive treated control cells. The Azolla eubacteria (Fig. lb) and the gram-positive control, B. subtilis (Fig. lc), did not form such a Absence of a thick peptidoglycan layer would allow KOH to penetrate a that LPS was in outer gram-positive cell as well. blebs, suggesting lacking the envelope bArthrobacter species have been demonstrated to have activity in intact of these microorganisms. Untreated controls (no polymyxin cells. B) of Azolla eubacteria were indistinguishable from treated (100 or 300 ppm of polymyxin B) cells. A thick peptidoglycan layer was present on the Bacillus cells but was not evident normally found in peptidoglycan, both lysine and ornithine on the cells of the Azolla eubacteria. were present; however, the concentration of lysine was The cell walls of the gram-negative control, P. aeruginosa, about three times that of ornithine. The four amino acids contained 13.5 ,ug of KDO per mg of LPS. KDO was not expected to be present in peptidoglycan, glycine, alanine, detected in the cell walls of the Azolla eubacterium, although glutamic acid, and presumably lysine, made up about 44% of the assay was sensitive to concentrations as low as 1 ,ug/mg the total numbers of amino acid residues detected. These of LPS. Table 1 summarizes the results of the various tests amino acids were present in a molar ratio of approximately used to determine the Gram classification of these bacteria. 5:3:2:1 (glycine-alanine-glutamic acid-lysine). Typical pepti- Relatively small amounts (nanograms per gram [fresh doglycan subunits exhibit a ratio of 5:2:1:1. The source of weight] of cells) of 16 different amino acids were found in the the other 12 amino acids was not determined. cell wall hydrolysates. The relative frequencies of these Antibiotic susceptibility. All 15 isolates exhibited identical amino acids are given in Table 2. Of the diamino acids susceptibilities to the various antibiotics. They were unsusceptible (0-mm-diameter zones of inhibition) to aureomycin, bacitracin, carbenicillin, and penicillin, moder- ately susceptible to polymyxin ( 8 + 1 mm) and neomycin (10 + 1 mm), and relatively susceptible to streptomycin (15 + 2 mm), ampicillin (17 ± 3 mm), kanamycin (19 ± 4 mm), tetracycline (20 ± 5 mm), and erythromycin (22 ± 3 mm). Determinative tests. All 15 isolates were non-fermentative, oxidase positive, and weakly positive and utilized citrate, glucose, and fructose. They were unable to use arabinose, inositol, rhamnose, melibiose, or inulin. They did not hydrolyze starch, cellulose, or gelatin, nor did they form H2S or indole from protein. They lacked phenylalanine deaminase, arginine decarboxylase, and lysine decar- boxylase activities. Nine of the isolates were unable to utilize xylose and did not produce N2 gas from NO3. The remaining six isolates produced N2 gas; of these, five were able to utilize xylose. These same five isolates also had weaker urease activity. Nutritional requirements. These bacteria grew in a defined

TABLE 2. Relative frequencies of the amino acids found in cell wall hydrolysates (ng/g p.mollg Frequency AminoAminoacidConcnacid of cells) of cells (%u c Glycine 2.9 0.0386 18.7 Alanine 2.6 0.0289 14.0 * Glutamic acid 2.1 0.0144 7.0 b Lysine 1.3 0.0086 4.2 Omithine 0.4 0.0030 1.5 Arginine 1.9 0.0109 5.3 Aspartic acid 2.1 0.0154 7.5 Threonine 1.9 0.0156 7.5 Serine 1.7 0.0164 7.9 FIG. 1. Detection of LPS in the cell wall by treatment with Valine 1.9 0.0164 7.9 polymyxin B (300 ppm). (a) P. aeruginosa (positive control) with Isoleucine 0.8 0.0061 2.9 blebs (arrow) indicating LPS. (b) Eubacterium from A. filiculoides. Leucine 1.2 0.0091 4.4 Note the absence of blebs, indicating that the membranous structure Tyrosine 1.7 0.0092 4.4 at arrow is not LPS; therefore, the microorganism appears to be Phenylalanine 1.5 0.0092 4.4 gram positive. A thick peptidoglycan layer is not apparent on these Methionine 0.3 0.0018 0.9 cells. (c) B. subtilis (negative control) without blebs but with a thick Histidine 0.5 0.0030 1.4 peptidoglycan layer (arrow). 428 WALLACE AND GATES APPL. ENVIRON. MICROBIOL.

TABLE 3. Effect of sugar concentration on generation time and The formation of rods and V forms, often with budlike cell yield of a eubacterial isolate from A. filiculoides in a synthetic outgrowths, during log phase and gradual shortening ofthese basal salts medium containing glucose or fructose rods until cocci predominated in death phase, the presence Yield of lysine in cell wall hydrolysates, and the inability of these Sugar Generation bacteria to hydrolyze cellulose indicate that these eubacteria (concn, mM) time (h) Klett Cells/mla are of the genus Arthrobacter. Of the seven species listed in units the eighth edition of Bergey's Manual of Determinative Glucose (5) 2.4 140 3.1 x 109 Bacteriology (4), these organisms are most similar to the Glucose (15) 2.7 310 7.3 x 109 type species A. globiformis Conn and Dimick, differing only Glucose (50) 3.3 380 9.0 x 109 in that A. globiformis is gelatin hydrolytic. As classification Glucose (100) 3.3 390 9.2 x 109 to within this remains Fructose (5) 5.4 100 2.1 x 109 species genus highly controversial, the Fructose (15) 5.4 300 7.1 x 109 organisms described in this manuscript cannot yet be as- Fructose (50) 5.6 320 7.5 x 109 signed to a species. Fructose (100) 5.2 385 9.1 x 109 a Calculated from the formula, cells/ml = Klett units x 2.5 x 107 - 37, ACKNOWLEDGMENTS determined by regressing viable cell concentration upon Klett readings of the same culture during log phase in 50 mM glucose-basal salts medium (r2 = This work was supported in part by U.S. Department of Agricul- 0.90; P < 0.01). ture grant 5901-0410-9-0304-0 and in part by National Science Foundation Alternative Biological Resources Program grant PCM 83-05202. medium containing NaNO3 (1 g/liter), (NH4)2SO4 (1 g/liter), MgSO4 7H20 (0.1 g/liter), KH2PO4 (3 g/liter), K2HPO4 (7 LITERATURE CITED g/liter), and glucose (15 mM). Generation times and cell 1. Allen, M. M. 1968. Simple conditions for growth of unicellular yields were comparable to those obtained when these orga- blue green algae on plates. J. Phycol. 4:1-4. nisms were cultured in PCB and other 2. American Public Health Association. 1971. Standard methods for complex media. the examination of water and wastewater, 13th ed. American Increasing the glucose concentration to 50 or 100 mM Public Health Association, Inc., Washington, D.C. increased cell yields but not growth rates. Additions of 3. Bottomley, W. B. 1920. The effect of organic matter on the amino acids and B vitamins provided no further improve- growth of various water plants in culture solution. Ann. Bot. ment in growth rate or cell yield. Growth responses of one 39:353-365. isolate to different glucose and fructose concentrations are 4. Buchanan, R. E., and N. E. Gibbons (ed.). 1974. Bergey's given in Table 3. manual of determinative bacteriology, 8th ed. The Williams & Wilkins Co., Baltimore. DISCUSSION 5. Cerny, G. 1976. Method for the distinction of gram negative Although some metabolic variation from Gram positive bacteria. Eur. J. Appi. Microbiol. existed between the 15 Biotechnol. 3:223-225. isolates of Azolla eubacteria tested, we do not consider it to 6. Cerny, G. 1978. Studies on the aminopeptidase test for the be sufficient to indicate the presence of more than one distinction of Gram-negative from Gram-positive bacteria. Eur. species. None of these isolates were starch hydrolytic, J. Appl. Microbiol. Biotechnol. 5:113-122. although we reported activity in our original isolates (8). No 7. Cummins, C. S., and H. Harris. 1955. The chemical composition correlation could be found between such variation and either of the cell wall in some Gram positive bacteria and its possible fern species or method of isolation. value as a taxonomic character. J. Gen. Microbiol. 14:583-599. The preponderance of evidence suggests that these 8. Gates, J. E., R. W. Fisher, and R. A. Candler. 1980. The eubacteria are gram-positive organisms. The absence of occurrence of coryneform bacteria in the leaf cavity of Azolla. detectable levels of LPS or KDO is consistent with this Arch. Microbiol. 127:163-165. 9. Gregersen, T. 1978. Rapid method for distinction of Gram- conclusion. The inconsistency with a gram-positive classifi- negative from Gram-positive bacteria. Eur. J. Appl. Microbiol. cation in the KOH test may be explained by the absence of Biotechnol. 5:123-127. a thick peptidoglycan layer common to most gram-positive 10. Karkhanis, Y. D., J. Y. Zeltner, J. J. Jackson, and D. J. Carlo. bacteria. KOH is believed to penetrate the outer membrane 1978. A new and improved microassay to determine 2-keto-3- and thin peptidoglycan layers typical of gram-negative orga- deoxyoctonate in lipopolysaccharide of Gram negative bacteria. nisms, lyse the cells, and release the viscous nucleic acid (9). Anal. Biochem. 85:595-601. The absence of a thick peptidoglycan layer in a significant 11. Newton, J. W., and A. I. Herman. 1979. Isolation of number of cells would allow for the same result in a cyanobacteria from the aquatic fern, Azolla. Arch. Microbiol. gram-positive species. 120:161-165. 12. Osborn, M. J. 1963. Studies on the Gram-negative. 1. Evidence The aminopeptidase test results also are inconsistent with for the role of 2-keto-3-deoxyoctonate in the lipopolysaccharide a gram-positive interpretation. Although both gram-positive of Salmonella typhimurium. Biochemistry 50:499-506. and gram-negative organisms have this membrane-bound 13. Smibert, R. M., and N. R. Krieg. 1981. General characteriza- enzyme, aminopeptidase activity generally cannot be de- tion, p. 409-443. In P. Gerhardt, R. G. E. Murray, R. N. tected in intact gram-positive organisms because the sub- Costilow, E. W. Nester, Willis A. Wood, Noel R. Krieg, and strate, L-alanine-4-nitroanilide, does not penetrate the thick G. B. Phillips (ed)., Manual of methods for general bacteriol- layer of peptidoglycan (5). Therefore, a positive reaction in ogy. American Society for Microbiology, Washington, D.C. gram-positive species could be explained by the presence of 14. Spackman, D. H., W. H. Stein, and S. More. 1958. Automatic cells with an unusually thin peptidoglycan layer. Cerny (6) recording apparatus for use in the chromatography of amino tested acids. Anal. Chem. 30:1190-1206. 46 species of gram-positive bacteria and only 3 were 15. Valentine, R. C., B. M. Shapiro, and E. R. Stadtman. 1968. found to have aminopeptidase activity. It is noteworthy that Regulation of glutamine synthetase. 12. Electron microscopy of two of the three belonged to the genus Arthrobacter, and all the enzyme from Escherichia coli. Biochemistry 7:2143-2152. three possessed a lysine-alanine interpeptide bridge in their 16. Watanabe, I., C. R. Espinas, N. S. Berja, and B. V. Alinagno. peptidoglycan. 1977. Utilization of the Azolla-Anabaena complex as a nitrogen VOL. 52, 1986 AZOLLA EUBACTERIA 429

fertilizer for rice. Int. Rice Res. Inst. Res. Paper Ser. 2:1-15. 19. Wiegel, J., and L. Quandt. 1982. Determination of the Gram 17. Weissbach, A., and J. Hurwitz. 1959. The formation of 2-keto- type using the reaction between Polymyxin B and lipopolysac- 3-deoxyheptonic acid in extracts of Escherichia coli B. J. Biol. charides of the outer wall of whole bacteria. J. Gen. Microbiol. Chem. 234:705-712. 128:2261-2270. 18. Wiegel, J., and F. Mayer. 1978. Isolation of lipopolysaccharides 20. Work, E. 1971. In J. R. Norris and D. W. Ribbons (ed.), and the effect of Polymyxin B on the outer membrane of Methods in microbiology, vol. Sa, p. 361-418. Academic Press, Corynebacterium autotrophicum. Arch. Microbiol. 118:67-69. Inc., New York.