1542 MICROBIOLOGY: H. NIKAIDO PROC. N. A. S.

supernatant fluid, whereas the cobamide coenzyme remains in the ribosomal par- ticles. The 5,6-dimethylbenzimidazolylcobamide coenzyme has been identified in the azotobacter.

We are grateful to David Perlman, the Squibb Institute for Medical Research, New Jersey, for determining vitamin B12 in our preparations and for the gift of cobalt58-labeled hydroxocobala- min. * Sabbatical leave October 1961-1962 at the University of Wisconsin. Aided by grant E-1417 (C6) from the National Institutes of Health and grant C-2826 from the National Science Founda- tion. t Appeared on the program for the Annual Meeting and in the abstracts published in Science as "Metabolism of Inorganic Nitrogen and Its Compounds in Microorganisms" by D. J. D. Nicholas and P. W. Wilson. l Holm-Hansen, 0. G. C., H. Gerloff, and F. Skoog, Physiol. Plantanum, 7, 665 (1954). 2 Shaukat-Ahmed, and H. J. Evans, Biochem. Biophys. Res. Comm., 1, 271 (1959) and Soil Science, 90, 205 (1960). 3 Lowe, R. H., and H. J. Evans, J. Bacteriol., 83, 210 (1962). 4Reisenauer, H. M., Nature, 186, 375 (1960). 5 Hallsworth, E. G., S. B. Wilson, and E. A. N. Greenwood, Nature, 187, 79 (1960). 6 Delwiche, C. C., C. M. Johnson, and H. M. Reisenauer, Plant Physiol., 36, 73 (1961). 7Lowe, R. H., H. J. Evans, and Shaukat-Ahmed, Biochem. Biophys. Res. Comm., 3, 675 (1960). 8 Nicholas, D. J. D., Y. Maruyama, and D. J. Fisher, Biochim. Biophys. Acta, 56, 623 (1962). 9 Hu, A. S. L., R. Epstein, H. 0. Halvorson, and R. M. Bock, Arch. Biochem. Biophys., 91, 210 (1960). 10 Nicholas, D. J. D., and D. J. Fisher, J. Sci. Food Agric., 11, 603 (1960). 11 Robrish, A. R., and A. G. Marr, J. Bacteriol., 83, 158 (1962). 1 Hirozi, K. K., H. 0. Halvorson, and R. M. Bock, Biochim. Biophys. Acta, 49, 212 (1961). 1 Lowry, 0. H., N. J. Rosebrough, A. L. Farr, and R. J. Randall, J. Biol. Chem., 193, 265 (1951). 14 Mejbaum, W., Z. physiol. Chem., 258, 117 (1939). 16 Kliewer, M., and H. J. Evans, Nature, 194, 108 (1962). 16 Alexander, M., and P. W. Wilson, these PROCEEDNGS, 41, 843 (1955).

STUDIES ON THE BIOSYNTHESIS OF CELL WALL POLYSACCHARIDE IN MUTANT STRAINS OF SALMONELLA, II BY HIROSHI NIKAIDO

INSTITUTE FOR PROTEIN RESEARCH, OSAKA UNIVERSITY, JAPAN* Communicated by Herman M. Kalckar, June 26, 1962 In the preceding paper, ' we described the properties of cell wall lipopolysaccharide (LPS)t from the mutant strains of Salmonella enteritidis and S. typhimurium, which cannot synthesize galactose.' LPS from wild type strains of these species contains glucose, galactose, mannose, rhamnose, 3,6-dideoxyhexose, and probably heptose, while LPS from these mutant strains (M mutants) contains only glucose and probably heptose as neutral sugar.1-3 When, however, M mutant cells were grown in the presence of galactose for a short period, they synthesized LPS, the composition of which was very similar to that in the wild type strains.24 This study was initiated in order to investigate the mechanism underlying these Downloaded by guest on September 24, 2021 VOL. 48, 1962 MICROBIOLOGY: H. NIKAIDO 1543

phenomena. The properties of the strains used were already described in the pre- ceding article.' Two hypotheses were considered to explain these observations. In the first and the simpler hypothesis, we assume that several sugars which cannot be found in M mutant LPS-mannose, rhamnose, and 3,6-dideoxyhexose-are all synthesized from galactose in Salmonella. Thus M mutant cells cannot synthesize galactose, and consequently the other sugars; but if they are grown in the presence of galac- tose, the exogenous galactose is converted into uridine diphosphate galactose (UDP- Gal) (for pathway, see Fig. 1) and then into the other sugars, which are now in- corporated into the cell wall polysaccharide. exogenous galactose I galactokinase Gal-i-P

Gal-i-P uridyl Glu transferase A Glu IV< Glu UDPGal UDPGal4- epimerase Gal-Gal-Glu Gal-Glu B UTP<--- UDPG Glu I Gal-Gal-Glu Glucose-i-phosphate

Glu-Gal-Man-Rha-Glu-Gal-Man-Rha-Gal-Gal-Glu L I L Gal-Glu C Glu Tyv-Man-Rha-Glu-Gal-Man-Rha-Gal-Gal-Glu JI* FIG. l.-Suggested pathway for the incorporation of exogenous galactose into LPS of M mutants. M mutants have a genetic defect at UDPGal4-epimerase (X), and accordingly cannot syn- thesize UDPGal from UTP and glucose-l-phosphate. Therefore, the cell wall polysaccharide remains at stage A, namely, poly-glucose. When galactose is added exogenously, it is metabolized by galactokinase and Gal-1-P uridyl transferase, and is transformed into UDPGal. UDPGal transfers its galactose moiety to A, forming B, and other sugars are subsequently transferred, thus finally completing the wild type-like polysaccharide C. The structures of polysaccharides are hypothetical. Monosaccharide sequences covered with brackets in C represent the hypothetical "repeating units." UTP, uridine triphosphate; Glu, glucose; Gal, galactose; Man, mannose; Rha, rhamnose: Tyv, tyvelose. The other more complicated hypothesis$ explains the situation in the following way. It is assumed that LPS or the O-polysaccharide of Salmonella has a "core" or "skeleton" of poly-glucose, and that all the side chains from this "core" must start with galactose. It is also assumed that mannose, rhamnose, and 3,6-dideoxyhexoses Downloaded by guest on September 24, 2021 1544 MICROBIOLOGY: H. NIKAIDO PROC. N. A. S.

are present only in the side chains. Thus, M mutants cannot initiate the side chain containing various sugars, because of the lack of galactose. In the presence of galactose, however, they can make UDPGal, and add galactose onto the poly-glu- cose "core," thus initiating the formation of side chains. Incorporation of C14-Galactose with Intact Cells.-If the first hypothesis is correct, exogenously given C '4-galactose must be converted into various other sugars in the LPS by the intact M mutant cells. Galactose-1-C14 was therefore added to a grow- ing culture of 11-1-M cells at the concentration of 0.005%, and after 20 minutes' incubation, the cells were harvested, the 90% ethanol-insoluble fraction, containing LPS, was hydrolyzed with 2 N H2S04, and the hydrolyzate was chromatographed on paper. Spots of glucose, galactose, mannose, rhamnose, tyvelose, and ribose (probably from RNA) were found, but only the spot of galactose was radioactive. This result completely rules out the first hypothesis. Accumulation of Nucleoside Diphosphate Sugar Compounds in the Cell.-If the second hypothesis is correct, various sugars other than glucose and galactose cannot attach to the "core" simply because of the lack of attachment sites. Thus it would be expected that these sugars would accumulate in the soluble fraction, possibly in the form of the direct precursors before their transfer to the core. In fact, when the ethanol-soluble fraction of M mutant cells was treated with dilute acid, and was then chromatographed on paper, all the missing sugars in M mutant polysacchar- ide-except galactose-were demonstrated. It appeared most probable that the transfer of sugars to the core was done from nucleoside diphosphate sugar deriva- tives. Therefore, we looked for these compounds in the 80% ethanol-soluble frac- tion of M mutants grown without the addition of galactose. The ethanol extract was applied to a column of Dowex-1 (Cl') resin, and the elution was made with a HCl-NaCl system.5 The nucleotides containing 3,6-dideoxyhexoses were detected by the thiobarbituric acid reaction,6 and that containing rhamnose was detected by the cysteine-sulfuric acid reaction.7 The nucleotides were adsorbed to and then eluted from charcoal, and subsequently purified by paper chromatography. As a result of this study, nucleotides containing 3,6-dideoxyhexoses were isolated and were identified as cytidine diphosphate tyvelose, and cytidine diphosphate abe- quose.8 These were obtained from M mutants of S. enteritidis and of S. typhi- murium, respectively; the wild type strains of these bacteria contained in their LPS tyvelose and abequose respectively. It is interesting to note that these were the first cytidine diphosphate sugar compounds to be found in nature, together with the similar compounds isolated from A. vinelandii by Suzuki's group.9 The sugars in the latter compounds appear tohave a rather unusualstructure, and to be definitely different from 3,6-dideoxyhexoses.9 The nucleotide containing rhamnose seemed to be identical with thymidine diphosphate rhamnose.10 The nucleotide content of various strains is shown in Table 1, and it is quite clear that M mutant cells ac- cumulate large amounts of these nucleotides. These results support, or at least are consistent with, the second hypothesis. Several other pieces of evidence can be found which support the second hypothe- sis. Westphal and his co-workers demonstrated that usually smooth forms of Salmonella contain many sugars in their LPS, but that when they mutate to rough forms, various sugars are all lost at the same time, leaving only glucose and galac- tose.11 12 Furthermore, they showed'2 that when colitose was stripped off by mild Downloaded by guest on September 24, 2021 VOL. 48, 1962 MICROBIOLOGY: H. NIKAIDO 1545

TABLE 1 THE ACCUMULATION OF SOME NUCLEOSIDE DIPHOSPHATE SUGARS IN VARIOUS STRAINS -Nucleotide Accumulated- (pmoles/10 g dry wt. cells) CDP- Exp. Strain Description dideoxyhexose TDP-rhamnose 1 No. 11 Wild type (0.1)* (0.2)* 2 11-1-M M mutant 8.0 3.2 3 11-1-M " 5.8 ... t 4 7-M-1 " 2.6 ... t 5 7-M-1 " 2.9 1.7 The cells were grown in nutrient broth (without the addition of galactose) at 370C with aeration by shak- ing, and were harvested close to the end of the exponential phase of growth. They were extracted with 80% ethanol at 751C for 3 min, and the ethanol-soluble fraction was fractionated by column chromatog- raphy.s The amount of nucleotides was determined by using the thiobarbituric acid reaction6 and the cysteine-sulfuric acid reaction7 on appropriate fractions eluted from Dowex-i columns. * In the case of wild type extract, the amounts of nucleotides were too small to be detected as peaks of UV-absorbing substance. Therefore, these assays were carried out on fractions where these nucleotides were expected to be present. t Not determined.

acid hydrolysis from the LPS of certain smooth forms of Salmonella (originally containing colitose, galactose, and glucose), a substance was obtained which had the serological specificity of LPS from rough forms. These results are best ex- plained by assuming a central poly-glucose-galactose "core,"", 12 having side chains made of the various sugars. It has been found" that rough forms from various species of Salmonella show similar serological specificity (or at least high cross- reactivity), in contrast to the variety in the serological specificity of smooth forms; this suggests further that this core structure may be common to all Salmonella. In this connection, it appears interesting to examine whether M mutants from vari- ous serotypes of Salmonella all have the LPS of the same structure, and also to examine whether the LPS of the wild type contain the same type of glucose-glucose linkages as found in M mutant polysaccharides. The wild type polysaccharides, in the cases which were analyzed extensively, appeared to consist exclusively of repeating units of several monosaccharides,13 14 which fact was frequently taken to be contradictory to the existence of such a "core" structure. Robbins and Uchida, 3 however, showed that even in the anti- gen 3,15 where no glucose was found in the "repeating units," small amounts of glucose were present, and these were released at the later stages of acid hydrolysis. This fact again is consistent with the hypothesis of a poly-glucose core. The failure by many workers to find a glucosyl-glucose oligosaccharide in the partial hydrolyzate of wild type LPS, can be easily ascribed to the relatively small amount of "core" present: In this case of antigen 3,15, glucose only accounts for less than 10% of the total polysaccharide.11 The hypothesis of a small poly-glucose "core" with rather long side chains is also consistent with the low yield of poly-glucose lipo- polysaccharide from M mutant cells. Robbins and Uchida"3 also showed that the LPS which they studied contained much more galactose than could be accounted for by side chain "repeating units," which would indicate the involvement of much galactose in the "core" structure. The Incorporation of the Galactose Moiety from UDPGal in a Cell-free System. According to the hypothesis discussed above, one of the earliest steps leading to the completion of complex wild type polysaccharide must be the attachment of the galactose moiety onto a poly-glucose "core" (cf. Fig. 1). M mutants appear to be especially suitable for the study of this reaction in a cell-free system, since they con- Downloaded by guest on September 24, 2021 1546 MICROBIOLOGY: H. NIKAIDO PRoc. N. A. S.

tain polyglucose LPS, postulated acceptor for galactose. Also, since they cannot convert UDPGal into UDPG, the incorporation of galactose can be studied sepa- rately from the incorporation of glucose. In order to demonstrate this reaction in vitro, a sonic extract from a UDPGal- 4-epimeraseless mutant, Salmonella enteritidis 1 1-1-M, was incubated with UDPGal- C14. The crude sonic extract contained much lipoid-poly-glucose complex, and this was precipitated by the addition of specific antiserum. It was expected that even if galactose was attached onto poly-glucose, it would not cover the entire surface of poly-glucose, and the newly formed galactosyl poly-glucose should still be precipi- table by anti-poly-glucose serum. In the standard procedure, the reaction mixture was incubated for 60 minutes at 370C, and then 0.2 ml of anti-1 1-1-M rabbit serum§ (containing 0.05% merthiolate) was added. Incubation was further continued for 60 minutes, and then the tubes were set aside at 40C for at least 24 hours in order to bring the antigen-antibody reac- tion to completion. The tubes were centrifuged at 4,000 rpm for 60 minutes, the precipitates were washed 5 times with ice-cold 0.9% NaCl, and the final precipitate was supended in dilute ammonia, plated and counted. A typical result is shown in Table 2. TABLE 2 THE CELL-FREE INCORPORATION OF GALACTOSE-C14 INTO THE SEROLOGICALLY SPECIFIC FRACTION Tube System CPM incorporated 1 Complete* 1655 2 Complete, but with boiled enzyme 2 3 Complete-antiserum added at time 0 128 * UDPgal-C"4, 2 X 104 cpm, or 34 mpmoles; enzyme, 0.2 ml containing 4.0 mg protein; MgCl2, 0.3 Mmole: sodium phosphate buffer, pH 7.5, 15 jsmoles; final volume, 0.3 ml. UDPGal-CI4 was prepared enzymatically from galactose-1-C14 by the combined action of galactokinase and galactose-l-phosphate uridyl transferase, and was purified by charcoal treatment and paper chromatography. Contamination with UDPG (nonradioactive) was less than 3%. The enzyme was prepared by sonicating the 11-1-M cells (grown without galactose) in 0.05 M potassium phos- phate buffer, pH 7.4, for 6 min with a Kubota 10 kc sonic oscillator. The extract was centrifuged for 8 min at 10,000 rpm and the supernatant was dialyzed for 5 hr against water before use. It is seen that 8.3% of the input radioactivity was incorporated into a fraction precipitated with specific antiserum. Since poly-glucose is contained in the crude enzyme, a large amount of antigen-antibody precipitate was obtained in all three tubes, but in tubes 2 and 3 the radioactivity was very low. (The small incorpora- tion in tube 3 may indicate that the stopping of the reaction by this procedure does not come to completion instantaneously.) These controls clearly rule out the pos- sibility of coprecipitation of UDPGal-C14. After acid hydrolysis of the product the incorporated radioactivity could be recovered in the form of unaltered galac- tose. Although the antiserum used was prepared against the whole cells of 11-1-M, the cells were heated at 100°C for 150 minutes before injection, and thus the anti- serum is unlikely to contain much antibody directed against substances other than the LPS. Thus it appears highly probable that here we are actually dealing with the incorporation of galactose-C14 into the poly-glucose "core." In fact, the ex- periment of Table 3 shows that the product is nondialyzable, is resistant to ribonu- clease and trypsin, and is extracted into the aqueous phase by the phenol procedure.7 This suggests the polysaccharide nature of the reaction product. It is important that the product still remained nondialyzable after the drastic treatment with phe- nol, since it indicates that galactose is most probably covalently linked to the poly- Downloaded by guest on September 24, 2021 VOL. 48, 1962 MICROBIOLOGY: H. NIKAIDO 1547

saccharide, and rules out all possibility of a loose association of low molecular weight products-such as oligogalactosides-with the antigen-antibody precipitates. Further work is in progress in order to define the newly formed linkage with more precise chemical terms. Several preliminary attempts were made to examine the properties of this in- TABLE 3 CHARACTERIZATION OF THE INCORPORATION PRODUCT The reaction mixture was as follows: UDPGal-CI4, 1 X 105 cpm, or 0.17 Mmole; ATP, 0.5 Amole; 11-1-M "lipo- polysaccharide," 1.0 mg; MgCl2, 1.5 gmole; enzyme, 0.5 ml, containing 10 mg protein; sodium phosphate buffer, pH 7.5, 70 pmoles; final volume 1.1 ml. The mixture was incubated for 60 min at 371C, then 1.0 ml of anti- 11-1-M serum conlaining merthiolate was added. Further incubation and washings were done exactly as described for "standard" procedure. The washed precipitate was suspended in 5.5 ml distilled water, and 1.0 ml aliquots were further subjected to the treatment outlined below and then plated and counted. Further treatment Cpm in the product None 894 Dialysis* 746 1 mg trypsin at pH 7.7, 370C, 1 hr, then dialysis* 622 0.1 mg ribonuclease at pH 7.4, 370C, 1 hr, then dialysis* 666 Extracted with 45% phenol at 681C.4 Combined aqueous phase dialyzed* 628 * Dialysis was done against distilled water for 5 hr, with constant mechanical stirring. corporation system. The reaction had a broad pH optimum between 7 and 8. The amount of the incorporation was dependent on, and roughly proportional to, the amount of enzyme used. The reaction was markedly stimulated with 10-3 M Mg++ or Mn++. If either one of these cations was added, dialysis of the enzyme did not result in a gross decrease of activity, showing that no dialyzable cofactor is necessary aside from these cations. The addition of nonradioactive galactose or galactose-1-phosphate did not result in dilution of the radioactivity. This dem- onstrates that the transfer is probably done directly from UDPGal, and also that we are not dealing with the incorporation of free galactose by the remaining intact cells. More recently, Osborn and co-workers independently confirmed these results using an M mutant strain of S. typhimurium. The incorporation product was re- covered by a nonspecific trichloroacetic acid precipitation, but it was shown to be- have in the same way as LPS during phenol extraction and acetone precipitation. 15 Furthermore, they have shown that the enzymatic activity was almost exclusively located in the particulate fraction in the extract, suggesting that the activity is probably associated with the fragments of cell wall.15 Summary.-Mutants of Salmonella unable to synthesize galactose (M mutants) can make "normal" wild type-like cell wall polysaccharide only in the presence of galactose.I To explain these results, a hypothesis was put forward which as- sumes the existence of a poly-glucose "core" in Salmonella lipopolysaccharide, and also assumes that all the side chains attached to this "core" must be initiated by galactose. The nonconversion of C14-galactose into other sugars by the intact cells of M mutants and the accumulation of nucleoside diphosphate sugars (including two new nucleotides, cytidine diphosphate tyvelose and cytidine diphosphate abe- quose), were considered to be consistent with this hypothesis. Finally, the enzy- matic incorporation of C14-galactose from UDPGal into the serologically active frac- tion was demonstrated, using cell-free extracts of M mutants. The incorporation product was nondialyzable, resistant to ribonuclease or trypsin, and was extracted into the aqueous phase by treatment with hot 45% phenol. Although more Downloaded by guest on September 24, 2021 1548 MICROBIOLOGY: H. NIKAIDO PROC. N. A. S.

analysis is required, in view of the above-mentioned results and the specificity of the immunological method, it is most probable that this is the demonstration of the actual incorporation of galactose into a poly-glucose "core," one of the early steps postulated in the biosynthesis of complex cell wall lipopolysaccharides. It is hoped that this system will further contribute to the elucidation of the mechanism of the biosynthesis of these complex and specific polysaccharides.

It is a pleasure to express the author's gratitude to many people who gave him such friendly help during the course of this work. It was Dr. T. Fukasawa who pointed out the significance of the whole problem to the author in the early days of this study, and the author's indebtedness to him can hardly be exaggerated. Drs. B. A. D. Stocker, 0. Westphal, 0. Luderitz, and W. J. Whelan also afforded him great help through their enlightening discussions, communication of unpub- lished results, and gifts of various compounds. The author owes very much to Drs. H. M. Kalckar and K. Kurahashi for their encouragement, to Dr. A. Rapin for her help in the preparation of the manuscript, and to Miss K. Jokura for her excellent technical assistance. This study was supported in part by a research grant from the National Institutes of Health (A4600, to Dr. K. Kurahashi). * Present address: Biochemical Research Laboratory, Harvard Medical School and the Massa- chusetts General Hospital, Boston. t Abbreviations used: LPS, lipopolysaccharide; UDPGal, uridine diphosphate galactose; Ul)PG, uridine diphosphate glucose; Gal-1-P, galactose-l-phosphate. t This hypothesis was originally suggested to us by Dr. B. A. D. Stocker. § This was kindly prepared by Dr. M. Nakano, Department of Bacteriology, Keio University. I l-1-M cells were grown without the addition of galactose, and were treated at 1000C for 150 minutes before injection. 1 Nikaido, H., these PROCEEDINGS, 48, 1337 (1962). 2 Nikaido, H., Biochim. Biophys. Acta, 48, 460 (1961). 3Fukasawa, T., and H. Nikaido, Virology, 11, 508 (1960). 4Fukasawa, T., and H. Nikaido, Biochim. Biophys. Acta, 48, 470 (1961). 5Cabib, E., L. F. Leloir, and C. E. Cardini, J. Biol. Chem., 203, 1055 (1953). 6 Cynkin, M. A., and G. Ashwell, Nature, 186, 155 (1960). 7Dische, Z., and L. B. Shettles, J. Biol. Chem., 175, 595 (1948). 8Nikaido, H., and K. Jokura, Biochem. Biophys. Res. Communs., 6, 304 (1961). 9 Okuda, S., N. Suzuki, and S. Suzuki, presentation at the 34th general meeting of the Japanese Biochemical Society, November, 1961. 10 Okazaki, R., Biochim. Biophys. Acta, 44, 478 (1960). 11 Luderitz, O., F. Kauffmann, H. Stierlin, and 0. Westphal, Zbl. f. Bakt., I. Orig., 179, 180 (1960); Kauffmann, F., L. Kruger, 0. Luderitz, and 0. Westphal, Zbl. f. Bakt., I. Orig., 182, 57 (1961). 12 Westphal, O., and 0. Luderitz, Pathologia et Microbiologia, 24, 870 (1961); Luderitz, O., I. Beckmann, and 0. Westphal, Biochem. Z., in press. 13 Robbins, P. W., and T. Uchida, Biochemistry, 1, 323 (1962). 14Staub, A.-M., Ann. Inst. Pasteur, 98, 814 (1960). 15 Osborn, M. J., S. M. Rosen, L. Rothfield, and B. L. Horecker, Science, 136, 328 (1962). Downloaded by guest on September 24, 2021