Metabolism and function of polyisoprenol sugar intermediates in membrane-associated reactions

417

BIOCHIMICA ET BIOPHYSICA ACTA BBA 85112

METABOLISM AND FUNCTION OF POLYISOPRENOL SUGAR INTERMEDIATES IN MEMBRANE-ASSOCIATED REACTIONS W. J. LENNARZ and MALKA G. SCHER Department of Physiological Chemistry, The Johns Hopkins University School of Medicine, Baltimore, Md. (U.S.A.) (Received April 19th, 1972)

CONTENTS I. Introduction . . . . . . . . . . . . . . . . . . . . . II. Microbial systems . . . . . . . . . . . . . . . . . . A. Peptidoglycan . . . . . . . . . . . . . . . . . . . B. Lipopolysaccharide . . . . . . . . . . . . . . . . . C. Modification of 0-antigen . . . . . . . . . . . . . . D. Mannan. . . . . . . . . . . . . . . . . . . . . . E. Yeast mannan . . . . . . . . . . . . . . . . . . . F. Capsular polysaccharide . . . . . . . . . . . . . . . G. Mycobacterial mannolipid synthesis. . . . . . . . . . H. Teichoic acid . . . . . . . . . . . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

417 418 418 419 422 424 425 426 427 428

III. Metabolism of polyisoprenol compounds . . . . . . . . . . . IV. Plant and animal systems . . . . . . . . . . . . . . . . . . A. Phaseolus aureus . . . . . . . . . . . . . . . . . . . . B. Mammalian systems . . . . . . . . . . . . . . . . . . V. Concluding comments . . . . . . . . . . . . . . . . . . . Note added in proof . . . . . . . . . . . . . . . . . . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . .

429 433 433 435 438 439

Acknowledgement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

439 439

.

I. INTRODUCTION The discovery i n bacteria that lipophilic c o m p o u n d s c o n t a i n i n g sugar residues are required for the transfer of these sugars to glycans proved to be a critical o n e for the elucidation of the pathways o f biosynthesis of bacterial glycans 1-s. Earlier review articles have included comprehensive discussions of these pathways 6-1°, a n d only the most recent investigations will be presented here. Because the majority o f studies o n the polyisoprenoid intermediates have dealt with bacterial systems, most o f this review will be devoted to t h e m systems. However, recent work i n m a m m a l i a n a n d p l a n t systems suggesting that the polyisoprenoid lipid intermediates m a y participate in glycosylation reactions in higher organisms will also be reviewed. Biochim. Biophys. Acta, 265 (1972) 417-441

418

W.J. LENNARZ, M. G. SCHER

II. MICROBIAL SYSTEMS IIA. Peptidoglycan

Cell wall peptidoglycan synthesis has been studied most extensively in Staphylococcus aureus and Micrococcus lysodeikticus. It was in these organisms that Strominger and his coworkers 1 made the important discovery that a glycolipid formed from endogenous lipid in the particulate preparations served as an obligatory intermediate in peptidoglycan biosynthesis. When isolated, the glycolipid, containing N-acetylD-glucosamine and N-acetyl-D-muramyl pentapeptide, could serve as a substrate for polysaccharide biosynthesis in place of the nucleotide derivatives of these two compounds. After isolation and purification, the structure of the lipid intermediate was definitively established. A pyrophosphate group joins the reducing end of the sugar to the lipid, and the results of mass spectroscopy performed on the lipid moiety released by acid hydrolysis demonstrated that this unique lipid is a polyisoprenoid alcohol containing 11 isoprene units, each unit having a double bond (Fig. 1) s. Subsequent UDP-MurNAcpentopeptide -0--O--RII~

Pi 0 t

0

? \

UMP Mur NAc~ UDP-GI f cNAc(-penl°pepl1-0ide -,P-O-P. O - R~)" ~)_

GIcNAc-MurNAc(- pentapeptide)-O-~-O-IP-O-R / O- O3X~--- GIyCyI-tRNA

-O-OIP-O"OIP-O-II~/''-" GIcNAc-MurNAc

' RNA

O..

4 GIcNAc-MurNAc (-decapeptide)-O-P-O-=P-O-R O- O-

(-aecopeptide)-Acceptor

Acceptor

CH~ / CH~ ~ ,CH3 R=-CHz-CH=C-CHz--~-CH2-CH=C-CH2~CI-~-CH=C,, 9

CH3

Fig. 1. Sequence for the biosynthesis of the cell wall peptidoglycan. The abbreviations used are: GlcNAc, (N-acetyl-D-glucosamine;MurNAc (N-acetyI-D-muramicacid).

investigations established the structure of the acceptor lipid (Reaction 1, Fig. 1) to be undecaprenyl phosphate and elucidated the cyclical nature of its involvement in the biosynthetic pathway. As indicated, the initial step in the biosynthetic pathway Js reversible 11. From UDP-N-acetylmuramyl pentapeptide, initially a cytoplasmic component, an activated lipophilic derivative is synthesized in the membrane 12. This high-energy intermediate remains intact until the final transfer of sugar residues from the lipid to form a glycosidic bond with acceptor polysaccharide. The second step in the reaction sequence produces the disaccharide linked to lipid, and at this stage modifications of the peptide chain occur, such as the sequential transfer of glyeine units from tRNA to the lipophilic intermediate to form a pentaglycine side chain Biochim. Biophys. Acta, 265 (1972) 417-441

POLYISOPRENOLSUGAR INTERMEDIATES

419

(Reaction 3) 13'14. Reaction

4 is the transfer of the disaccharide to endogenous polysaccharide acceptor and concomitant release of undecaprenyl pyrophosphate. It is not yet known if the disaccharide is further polymerized on the lipid before transfer to polysaccharide. Undecaprenyl phosphate must be enzymatically regenerated (Reaction 5) before participating again as acceptor lipid. This reaction is the site of inhibition of peptidoglycan synthesis by bacitracin' 5. The synthesis of peptidoglycan is completed by transpeptidation, resulting in cross-linking of the peptide chains via the pentaglycine bridge' 6.

liB. Lipopolysaccharide

Similar pathways involving undecaprenyl phosphate intermediates operate in the synthesis of the O-antigen portion of the lipopolysaccharide associated with the cell envelope of Salmonella typhimurium 2 and Salmonella newington 3. The O-antigen chain of S. typhimurium consists of branched tetrasaccharide repeating units. The repeating

UDP-Gal +O-i-O-R UMP'~ O" O O t t GaI-O-P-O-P-O-R I I

Rha-Gal-O- P-O-P-O-R

I

I

GDP,=."

an

GO

Man-Rha-Ga1-0- Pl-O-Pl -O-R O(3CDP,~ r CDP Abe I

nMan-R~-GM-O- -0- -O-R O- O, t, o ? o I -t o t Man-Rha-Gal-( Man-Rha-Gol)- O-P-O-P-O-R +1o-I IO-P-O-P-O-R ~n-/ / l

O"

I

Compl. t. LPS

+

/

0"

~ O-Antipn-Leu /

0"

I

O"

LPS

O-~-O-!-O-R

Fig. 2. Sequence for the biosynthesis of the O-antigen chain of the ]ipopolysaccharide of S. typhi-

murium. The abbreviations used arc: Oal, D-galatose; Rha, L-rhamnose; Man, D-mannose; Abe,

abequose (3,6-dideoxy-D-galactose); LPS, lipopolysaccharide; and R, the hydrocarbon portion of undecaprcnol. Biochim. Biophys. Acta, 265 (1972) 417--441

420

W . J . LENNARZ, M. G. SCHER

trisaccharide is mannosylrhamnosylgalactose, and abequose linked to mannose forms the branch. S. newington contains the same repeating trisaccharide but the anomeric configurations are different and there is no branch sugar. Weiner et al. 2 and Wright et al. 3 at about the same time reported the first evidence for the participation of lipid-linked sugar intermediates in the synthesis of the O-antigen of S. typhimurium and S. newington. The structure of the lipid intermediate, determined independently of the work from Strominger's group x, was also found to be an oligosaccharide linked via a pyrophosphate bridge to undecaprenol 4. Since the stereochemistry has not been determined for these two lipids, identical isomers may or may not be involved. The reaction cycle depicted in Fig. 2 is the sequence of reactions for S. typhimurium O-antigen synthesis. The same sequence is followed in S. newington preparations. The initial step is the reversible transfer of galactose 1-phosphate from UDPgalactose to acceptor lipid, undecaprenyl phosphate, with release of UMP 2,3,17 Following formation of galactosylpyrophosphorylundecaprenol, transferases catalyze the sequential transfer of rhamnose and mannose to the monosaccharide lipid intermediate. Kanegasaki and Wright is utilizing Salmonella anatum cell envelope preparations have demonstrated that exogenous galactosylpyrophosphoryl-antigen carrier lipid (galactosylpyrophosphorylundecaprenol) and rhamnosylgalactosyl-

GDP-Mon "~ TDP-Rho UDP-GoI~J

l

Man-Rho-GaI-Man-Rha

0 0 t t -GoI- O - P - O - P - O - R

~- ~-

qrowinQ chai.

®

-O-i-O-R

O-

~-Mon } Mon-Rha-

Man- Rha-Gal -O-P-O-P-O-R I I

GoI-Man-Rha-Gal-

~®

UDP-GoI

o- o-

Man- Rho-Gol - Mon- Rho-Gol - Mon-Rho-GoIo¢

"0- -O-R

®

t itol ~

1®

Mon, Rho ,Gol, G o l a c t i t o l ~

0 !

Mon-Rb(I -Gol - O-PI-O-R O"

O

O

t

l

• N t t M a n - R h a - G a l -Man-Rha - Gal - Mon-Rha-Gal -Man-Rha-Gal - 0 - P ~ 0 - P-O- R

I® Man- Rha- Gal-Man-Rha-Gal

.......

o- oN

-Man -Rho -Gal - M a n - R h a -

Goloctitol

1®

Man,Rha ,Gal , G a l a c t i t o l

Fig. 3. Determination of direction of chain elongation. The abbreviations used are the same as those identified in Fig. 2. The presence of an asterisk (*) indicates radioactivity derived from UDP[S4C] galactose.

Biochim. Biophys. Acta, 265 (1972) 417-441

POLYISOPRENOL SUGAR INTERMEDIATES

421

pyrophosphoryl-antigen carrier lipid serve as acceptors for the addition of rhamnose and mannose, respectively. Abequose is subsequently added to trisaccharide-lipid intermediate in the ease of S. typhimurium cell envelope preparations. 19 Polymerization of the tetrasaccharide (or trisaccharide with S. newington preparations) constitutes the next step of the biosynthetic cycle. Isolation of an octasaccharide composed of two units of the tetrasaccharide attached to lipid has provided the only direct evidence that polymerization for the O-antigen system (in S. typhimurium) does occur at the level of lipid intermediates 19,2°. Robbins and his coworkers 21 have discovered that elongation of the O-antigen chain occurs by transfer of newly formed trisaccharide units to the reducing end of the growing chains. The methods which they used with a cell-free system from S. anatum are summarized in Fig. 3. The system was exposed to UDP[14C]galactose for a short period of time (Reaction 1, pulse). The O-antigen chain was isolated by treatment with phenol, reduced with borohydride (Reaction 2) and then analyzed for incorporation of radioactivity into the component monosaccharides (Reaction 3). Alternatively, the sample was allowed to undergo further chain elongation in the presence of nonradioactive precursors (Reaction 4, chase) before undergoing the same analytical procedure (Reactions 5, 6). Treatment of the isolated O-antigen with borohydride converted the reducing terminal galactose unit to galactitol, thus enabling one to distinguish the reducing from the non-reducing end of the chain. In the sample exposed to the pulse only, following hydrolysis of the reduced O-antigen chain (Reaction 3), radioactivity was primarily associated with galactitol. That is, newly formed (radioactive) trisaccharide units were present at the reducing end of the chain. Exposure to non-radioactive precursor (chase) resulted in a decreased amount of radioactivity being recovered from the O-antigen as galactitol (Reaction 6) and a greater amount of radioactivity was associated with galactose. These results, then, indicated that the newly formed trisaccharide units must have been added to the growing chain at the reducing end. Similar results were obtained in vivo when cultures were subjected to a pulse of [14C]glucose. It is of general interest that synthesis of fatty acids and proteins also involves elongation at the activated end of the growing chain. Evidence that lipid-linked trisaccharide functions effectively as a substrate for the polymerase enzyme in S. anatum was presented by Kanegasaki and Wright ~8. Significant amounts of radioactivity from [~4C]trisaccharide-antigen carrier lipid were incorporated into polymer with a system in which cell-envelope preparation and aqueous suspensions of the intermediate were added as a mixture prepared by freezing and thawing. They discovered that the manner in which enzyme and lipid intermediate (substrate) were added in an incubation mixture affected the extent of reaction between the two, since freezing and thawing a mixture of cell-envelope preparation and aqueous lipid suspension enhanced the rate of polymer formation, although no stimulation resulted when enzyme or substrate was treated alone. The product of this polymerase reaction obtained after mild acid treatment was excluded from Sephadex G-25 and the amount of excluded material increased with time of incubation. Reducing terminal galactose and internal galactose units of the polymerized product were Biochim. Biophys. Acta, 265 (1972) 417-441

422

W.J. LENNARZ, M. G. SCHER

distinguishable as galactitol and galactose following borohydride reduction and acid hydrolysis. By this method it was determined that 3-5 ~o of the radioactive material (supplied as galactose-labeled trisaccharide-antigen carrier lipid) was galactitol, indicating an average chain length of 20-33 repeat units. The authors also investigated the nature of the interaction between the lipid intermediate and the polymerase catalyzing the above reaction. They attempted to determine whether antigen carrier lipid was associated with a complex of the polymerase and sugar transferases or whether the trisaccharide-lipid intermediate was free to move within the membrane and serve as a substrate for the polymerase. According to the former theory, only one trisaccharide unit would be transferred from the carrier unless the reaction conditions were such that the carrier could be immediately reloaded with another repeat unit. In contrast, if there were free movement of the substrate, trisaccharide-antigen carrier lipid, polymerization would progress even in the absence of carrier reloading. The experimental system was designed to prevent reloading of the carrier by using as an enzyme source a strain of cells which could not form UDPgalactose and by including bacitracin in the reaction mixture. Incubation of preformed [14C]rhamnosylgalactosyl-antigen carrier lipid in the presence of GDPmannose resulted in a decrease in the radioactivity associated with lipid. Oligosaccharides produced during incubation were analyzed on Sephadex G-25 after mild acid hydrolysis. It was found that less radioactivity was associated with disaccharide, and more [~4C]trisaccharide or larger oligosaccharides were produced. The average chain length of the newly formed excluded products was 17 repeating units. It is evident, then, that more than one trisaccharide unit was transferred from the carrier lipid to form the oligosaccharide chain under conditions which prevented reloading of the carrier. However, free movement of the lipid intermediate does not exclude the possibility that such a complex exists. The final step in O-antigen synthesis is transfer of polymerized repeating units to core lipopolysaccharide with release of undecaprenyl pyrophosphate. Undecaprenyl phosphate is regenerated by dephosphorylation of lipid pyrophosphate in a bacitracin-sensitive reaction 22.

IIC. Modification of O-antigen Since the discovery of lipid intermediates in the biosynthesis of peptidoglycan and the O-antigen chain, the participation of similar compounds in other steps of O-antigen biosynthesis as well as in synthesis of a variety of other complex glycans has been established. Utilization of polyprenol intermediates to modify the O-antigen chain of Salmonella strains A, B, D1, and E during lipopolysaccharide biosynthesis has been reported by Wright 23"24 and by Nikaido et al. 2s. Wright z3"24 has studied the mechanism of the modification reaction in which glycosyl groups are added to the O-antigen in lysogenic cells of Salmonella group E (S. anatum). The O-antigen of E group Salmonella is a heteropolysaccharide composed of mannosylrhamnosylgalactose units. In cells lysogenic for bacteriophage e 15 or doubly lysogenic for e ~5 and e 34 the galactosyl units are of the fl-configuration. In the doubly lysogenic cells the galactosyl units are also substituted with a-D-glucosyl groups. Incubation of Biochim. Biophys. Acta, 265 (1972) 417-441

423

POLYISOPRENOL S U G A R INTERMEDIATES

particulate enzyme from doubly lysogenic cells with [32p]UDP[l*C]-glucose results in only [z4C]glucose and not a2p transfer to the particulate preparation. Lipid-linked [14C]glucose comprises 549/0 of the radioactivity transferred; the remainder was characterized as [14C]lipopolysaccharide. Upon further incubation the impure glycolipid contained in the particulate preparation does serve as the [14C]glucose donor for lipopolysaccharide glucosylation. The ratio of glucose to phosphate in the isolated intermediate is 0.9:1, and the sugar moiety is fl-glucose 1-phosphate. The free lipid produced by acid hydrolysis was analyzed by mass spectrometry and the results indicated that the lipid is a mixture of Cs5 and Cs o polyisoprenoid alcohols, with each isoprene unit having one double bond. Thus, without knowledge of the isomeric configuration of the double bonds the lipid moiety is identical in structure to the lipid moieties involved in formation of the peptidoglycan and O-antigen chains. The glucosyl lipid, however, is distinguished from peptidoglycan and O-antigen intermediates by the presence of a phosphodiester bridge rather than a pyrophosphoryl bridge, and is similar in this respect to mannosylphosphorylundecaprenol (see Section liD). The acceptor for [C14]glucose in the glucose transfer reaction is proposed to be the growing O-antigen chain attached to antigen carrier lipid. Experimentally, this hypothesis was tested in vivo with cells containing incomplete core lipopolysaccharide. Since the core material was incapable of accepting O-antigen chains, but transfer of glucose from the carrier to O-antigen (attached to lipid) was nevertheless observed, it seems likely that glucosyl units are transferred from glucosyl lipid stepwise to the growing O-antigen chain still linked to lipid. A hexasaccharide (x = 2 in Fig. 4) would be the shortest possible sequence of repeating units that act as a glucose ac-

UDP

J.

0

o-

/"

'Man-Rho-Gai ) - O - P - O - P - O - R ~X I t

0"

0

0-

GIc-0-P,-0-R

UDP-GIc

'-R-O-!-O" 0GIc~

.°.-..o-'o,

¢,~

0

n

O"

0-

Acceptor LPS

H-P,-o-P,-o'~" O"

O"

.o°-.,o-Go,)x_~--(Mo.-.,°-G°,)- .~s Fig. 4. Schematic presentation of proposed sequence involved in glycosylation of Salmonella e 15, e 34 O-antigen. The abbreviations used here are the same as those identified in Fig. 2, and Glc, D-glucose.

Biophys. Biochim. Acta, 265 (1972) 417-441

424

W.J. LENNARZ, M. G. SCHER

ceptor, since glucosylation requires fl-galactosyl units and these are formed only during polymerization of the repeating units catalyzed by the e 15-specific O-antigen polymerase. Further experiments will be necessary to substantiate this proposal. O-antigen factor 122 present in Salmonella groups A, B, and D1 is determined by branches of the O-antigen which consist of glucosyl residues linked to galactose units of the O-antigen chain. Nikaido et al. 2s discovered that cell envelope preparations from S. typhimurium and S. enteritidis (group B and group D1 respectively) strains possessing O-antigen 122 catalyze the transfer of glucose from UDPglucose to an endogenous acceptor. Results of chromatography on Sephadex G-100 and DEAEcellulose of the radioactive product formed by the S. typhimurium preparations indicate that it is most probably lipopolysaccharide and not merely the O-antigen side chains. Moreover, the lipopolysaccharide is characteristic of strains which bear a modified O-antigen because radioactivity is present in the product as glucose and partial acid hydrolysis releases the oligosaccharide glucosylgalactosylmannosylrhamnose. Under the hydrolytic conditions employed abequose linkages would have been most susceptible to acid hydrolysis and rhamnosyl linkages would have been the second most susceptible bonds. The time course of incorporation of radioactivity from [32p]UDP[3H]glucose into lipid and lipopolysaccharide suggested that the glucosyllipid was an intermediate in the transfer of glucose, and glucosyllipid isolated from an incubation mixture served as a direct donor of glucose for lipopolysaccharide synthesis. The acid lability and chromatographic behavior after catalytic hydrogenation of this lipid intermediate and rhamnosylgalactosyl-antigencarrier lipid were similar. However, phosphate is not transferred to the lipid along with glucose and the ratio of glucose to phosphate in the intermediate is approximately 1. Although the degree of unsaturation was not determined, the glucosyl lipid carrier appeared to be the same as antigen carrier lipid and the structure of the intermediate was therefore proposed to be glucosyl-l-phosphorylpolyisopreno126. Using a Salmonella mutant which could not transfer completed O-antigen chains from antigen carrier lipid to the defective core of the lipopolysaccharide produced by this organism it was demonstrated that [14C]glucose was transferred from UDP[~4C]glucose to the O-antigen side chains attached to antigen carrier lipid 27. Thus the mechanism of glycosylation and the structure of the lipid intermediate are completely analogous to the structure and mechanism proposed by Wright 2a'24 as described above. l i D . Mannan

A phosphodiester-linked lipophilic sugar intermediate is also involved in mannan synthesis catalyzed by a M . lysodeikticus particulate preparation zs. The system utilizes GDPmannose as the hexose source and the acceptor lipid required for the initial transfer of mannose has been isolated, purified and characterized as undecaprenyl phosphate z9 (Fig. 5). The lipid intermediate produced in this reaction has also been purified, and analysis established the structure to be mannosyl-l-phosphorylundecaprenol a°, the undecaprenol residue containing 2 internal trans and 8 internal cis Biochim. Biophys. Acta, 265 (1972) 417-441

POLYISOPRENOL SUGAR INTERMEDIATES

425

Mon-(- Mon )x-- Ml°n .... (Mon)y GDP

R-O-P-O"

-_-

j

,

ooCoo,

0 t

Man

Mon-O-P-0 R-----~

[

o \1 '~-o -~_--oR *Uon-lUQn-(-ao.~ Mo...... (MlOn)y

Mort

~Mlart Fig. 5. Role of mannosyl-l-phosphorylundecaprenol in mannan synthesis in M. lysodeikticus. R is the hydrocarbon portion of undecaprenol. double bonds. The isoprenoid moiety of the intermediate appears to be identical to the intermediates involved in peptidoglycan and O-antigen synthesis, although the sugar is linked to the lipid via a phosphodiester bridge rather than a pyrophosphoryl bond. Clearly, more than one sugar residue is attached to the lipid intermediates of peptidoglycan and O-antigen synthesis, but it has not been possible to detect a lipid intermediate containing mannose oligosaccharides in the mannan biosynthetic system. Moreover, [14C]mannosyl units are transferred only to the non-reducing termini of endogenous mannan. Mannosyl-l-phosphorylundecaprenol thus seems to function only to complete the synthesis of endogenous mannan, in a manner similar to the glucosyl-1-phosphorylundecaprenol intermediates which serve in the formation of branches in lysogenic Salmonella O-antigen. Addition of mannosyl units to the non-reducing termini is analogous to the classical method of synthesizing glycogen and starch by addition of glucosyl units to the non-reducing ends of growing chains 31 and is in contrast to the mechanism of O-antigen synthesis. liE. Yeast mannan Mannan is also a component of yeast cell walls. Cell-free preparations of Saccharomyces carlsbergensis catalyze the synthesis of mannan from G D P mannose 32. No evidence for a lipid intermediate was demonstrable in the S. carlsbergensis system. Using a Saccharomyces cerevisiae particulate preparation which also catalyzes mannan synthesis, Tanner 33 discovered the synthesis of an acid-labile mannose-containing lipid from GDPmannose. The lipid was formed rapidly early in the incubation period during which time mannan synthesis was low, but at later times mannan synthesis increased and there was no further increase in radioactivity in lipid. An exchange reaction between [x4C]GDP and GDPmannose was demonstrated, and addition of G D P to the system containing GDP[14C]mannose decreased the radioactivity found in lipid. Either Mn 2+ or Mg 2+ serve as cofactors for synthesis of the lipid, whereas Mn 2+ is necessary for mannan production. In an experiment with only Mg 2+ present, Biochim. Biophys. Acta, 265 (1972) 417-441

426

w.J. LENNARZ, M. G. SCHER

mannolipid formation was measured. Addition of Mn 2+ to this system caused a decrease in the level of mannose-containing lipid while stimulating the production of mannan. Evidence for participation of this lipid as an intermediate in polysaccharide synthesis was thereby obtained. Subsequently, Tanner et al. 34 reported that dolichol monophosphates from yeast or liver served as acceptor lipids in the transfer of mannose from GDPmannose catalyzed by the same particulate preparation from S. cerevisiae which was capable of synthesizing mannan. The presence of endogenous acceptor lipid accounted for radioactive lipid production in the absence of exogenous dolichol phosphates. It has not definitely been determined if mannosyl monophosphatedolichol is identical to the earlier described lipid intermediate produced in the presence of endogenous acceptor lipid. However, it has been determined that addition of dolichol monophosphate to the particulate enzyme preparation enhances the incorporation of radioactivity into an alcohol-insoluble material, and addition of GDP produces an expected decrease in radioactivity in the mannosyl monophosphatedolichol fraction. Since the nucleoside diphosphate, but not the monophosphate, inhibits production of the mannose-containing lipid, this system appears similar to the M. lysodeikticus mannosyl-l-phosphorylundecaprenol synthetase, and the authors therefore suggest formation of a phosphodiester link analogous to the isoprenoid intermediate in M. lysodeikticus mannan synthesis. In a related study Sentandreu and Lampen 35 have presented evidence that synthesis of S. cerevisiae mannan involves a lipid intermediate. Similarly, their preparation catalyzed transfer of mannose from GDP[~4C]mannose to an endogenous acceptor to form a mannosyl lipid and mannan. The time course of mannose incorporation into alkaline-stable mannolipid and mannan, and the alteration of the level of radioactivity upon unlabeled GDPmannose addition supported the contention that mannolipid of unknown structure is a necessary precursor for mannan synthesis. IIF. Capsular polysaccharide

Biosynthesis of capsular polysaccharide with the participation of lipid intermediates in Klebsiella (Aerobacter) aerogenes has been reported by Troy et al. a6. The polysaccharide is composed of the repeating trisaccharide galactosylmannosylgalactose with glucuronate branches on each mannose residue (Fig. 6). Initially it was observed that [14C]galactose accumulates in the lipid phase as a result of incubation of a cell envelope fraction with UDP[14C]galactose at 12 °C. Galactosylpyrophosphorylundecaprenol is formed in a reversible reaction which constitutes the first step toward synthesis of polysaccharide. The acceptor lipid in this reaction has been definitively characterized, primarily by means of mass spectrometry, as undecaprenyl phosphate. This lipid restores enzymatic activity to a lipid-depleted particulate preparation and it is active as acceptor lipid for M. lysodeikticus mannan synthesis aT. The functional equivalence of the acceptor lipids from M. lysodeikticus and A. aerogenes does not necessarily indicate identical structures since the enzymes may not have absolute specificity for the acceptor. Mannosyl and glucuronyl units are sequentially added to the galactosyl lipid intermediate. Glucuronic acid must be incorporated into the lipid Biochim. Biophys. Acta, 265 (1972) 417-441

POLYISOPRENOL SUGAR INTERMEDIATES

427

0 0 t t GDP-Mon GoI-O-P-O-P-O-R ~/" UMP . I I I -... ('~

uo -oo,J/ o -o-~,-o-~

~ °UDP-GIcUA ~

?

UDP-GoI O\GoI-Mo - G o I / - k G o I - M o n - G o I / - O - P - O - P - O - R

""

4- ~,-

(n-)O~'P-O-P-O-R

OlcUA

J

\

0 Acceptot

O-

GIcUA 0 O ' ' ' "nGol- MonGoI-O-I~-OIP-O-R

~-~._

~Acceptor

Go-Mon-Gal)lt . . . . . .

?

GlcUA - M o n - G o I - O - P - O - p - O - R

O-

O-

(~

+ O-I~-O-Pl-O- R O"

0°

Fig. 6. Proposed sequence for capsular polysaccharidesynthesis. The abbreviations used here are those in Fig. 2. GlcUA designatesa glucuronicacid residue. intermediate before the fourth sugar, another galactose, can be added to the repeating units. An octasaccharide which consists of 2 tetrasaccharide repeating units was obtained after release from the carrier phospholipid, indicating that polymerization occurs at the level of lipid intermediate. Ultimately the radioactive oligosaccharide is transferred to an accepter to form polysaccharide product. The nature of endogenous accepter for the polymerized oligosaccharide remains unknown. A specific phageinduced capsular polysaccharide depolymerase which has endogalactosidase activity was used to characterize the enzymaticallyformed polysaccharide as capsular material. Evidence for participation of a lipid intermediate with chromatographic properties similar to the undecaprenyl pyrophosphate intermediate in synthesis of capsular polysaceharide from another strain of Klebsiella has been reported by Sutherland and Norval, 38 but the structure of the intermediate was not analyzed.

IIG. Mycobacterial mannolipid synthesis The production of phosphatidylinositol oligomannosides and an alkalinestable, acid-labile mannolipid was observed upon incubation of a particulate cell-free preparation of Mycobacterium tuberculosis with GDPmannose 39. Product formation was accompanied by production of GDP, and addition of excess exogenous GDP inhibited synthesis of the mannolipid. The lipid labeled with [14C]mannose was extracted from an incubation mixture with chloroform-methanol and this extract was subjected to fractionation by extraction with organic solvents. The recovered mannolipid was then purified chromatographically on DEAE-ceUulose and Sephadex LH-20. Upon subjection to alkaline deacylation all of the mannolipid was recovered as alkali-stable material, which was further purified by DEAE-cellulose column chromatography prior to analysis. Mannose was the only radioactive sugar released by acid hydrolysis. The ratio of mannose to phosphate in an acid hydrolysate was 1.09:0.97, and the structure of the hydrophilic moiety of theflompound was Biochim. Biophys. Acta, 265 (1972) 417-441

428

W.J. LENNARZ, M. G. SCHER

determined to be mannose 1-phosphate. Analysis by NMR demonstrated the presence of peaks characteristic of isoprenyl units. As a result of mass spectroscopy the structure of the lipid moiety was fixed at 10 isoprene units and the mannolipid, therefore, is mannosyl-l-phosphoryldecaprenol. Although the structure of this compound is known, its function in M. tuberculosis remains to be established. IIH. Teichoic acid

Douglas and Baddiley4° have detected accumulation of a lipid intermediate containing N-acetyl [~4C]glucosamine during biosynthesis of Staphylococcus lactis 13 teichoic acid which consists of N-acetylglucosamine and glycerol linked via phosphodiester bridges. Accumulation of N-acetylglucosamine in the lipid phase was stimulated in the absence of CDPglycerol, and the amino sugar was transferred to teichoic acid upon further incubation of the washed particulate preparation with CDPglycerol. Neither the structure of this lipid nor the structure of a similar Nacetylglucosaminyl lipid which has been detected during synthesis of the S. lactis 2102 wall polymer41,42 has been determined. The acid lability of these pyrophosphatecontaining intermediates suggests the presence of a double bond fl to a phosphate group and, therefore, the authors suggest that the lipid may be of the polyisoprenoid type. If the intermediates in teichoic acid biosynthesis were found to have the same structures as the other polyprenol lipid carriers, this system and the peptidoglycan synthesizing system would compete for the polyprenyl phosphate accepter lipid. Watkinson et al. 43 have investigated the competition between the two systems in S. lactis 13 by measuring teichoic acid synthesis alone and teichoic acid synthesis in the presence of peptidoglycan synthesis. During teichoic acid synthesis 83~ of the CDP[14C]glycerol substrate was incorporated into teichoic acid. The percentage decreased to 65 ~ when the conditions were modified to permit simultaneous peptidoglycan synthesis. Vancomycin and bacitracin are known to exhibit peptidoglycan synthesis and should ultimately create a decreased availability of undecaprenyl phosphate for synthesis of either type of polysaccharide. This decreased availability of accepter lipid can account for the further decrease in the amount of substrate incorporated into teichoic acid observed when the antibiotics were added to a reaction mixture catalyzing synthesis of both polysaccharides. Thus, the observed inhibition of teichoic acid synthesis when peptidoglycan synthesis was also occurring, as well as the inhibition produced by the antibiotics, is consistent with the theory of competition between the two systems. In another report it was demonstrated that chloramphenicol can exert inhibition directly on synthesis of teichoic acids, and this effect was independent of any effect on protein synthesis44. Chloramphenicol effectively inhibited only those enzymatic systems which produced polymers containing glucose in the main chain and which had been shown in unpublished experiments to involve lipid intermediates. These results suggest that the site of chloramphenicol inhibition in these cases, at least, may be at the stage of transfer of glucose from the nucleotide precursor to the lipid. However, Biochim. Biophys. Acta, 265 (1972) 417-441

POLYISOPRENOL SUGAR INTERMEDIATES

429

it is clear that further studies must be made before an Understanding of the mechanism of inhibition is reached.

III. METABOLISMOF POLYISOPRENOLCOMPOUNDS Although it has been repeatedly demonstrated that the activated sugar intermediates function effectively in transferring sugars from a cytoplasmic component (sugar nucleotide) to glycans located within or beyond the cytoplasmic membrane, the actual site and mechanism of synthesis of the lipid moiety has not been studied in detail. The existence of polyprenols in the unphosphorylated state has been confirmed by their isolation from bacteria, and the enzymic phosphorylation of such compounds has also been reported. Systems which catalyze the biosynthesis of polyprenyl pyrophosphate chains of varying length and an enzyme with pyrophosphatase activity have been described. The current state of our knowledge on bacterial metabolism of polyprenols, then, includes three types of reaction: phosphorylation, dephosphorylation and synthesis of polyprenol chains. Bactoprenol is a Css isoprenoid alcohol containing only 10 double bonds. Its occurrence in Lactobacillus casei has been known for some time45. More recently, undecaprenol has been isolated from Lactobacillus plantarum 46. Mass spectroscopy, NMR and infrared spectroscopy clearly defined the structure of this unsaponifiable lipid as a polyisoprenoid alcohol containing 11 isoprene units. The stereochemistry was found to be the same as in the undecaprenyl residue of mannosylphosphorylundecaprenol characterized by Scher et al. a°. This compound is different from bactoprenol since there are no saturated isoprene units. Recently investigations have been directed towards determining the site of synthesis of bactoprenol in L. casei 47. After incubating the bacteria with [2-14C]mevalonate for varying lengths of time, the radioactivity per mg of protein was measured in both mesosome and plasma membrane preparations made from protoplasts. Since the concentration of bactoprenol was found to be the same in both types of membrane preparations regardless of the time of exposure to [14C]mevalonate, the possibility that bactoprenol is synthesized on the mesosome membrane and later transferred to the plasma membrane was disproved. Conversion of the free alcohols (C55) to C55-polyisoprenyl phosphates by an ATP-dependent phosphokinase has been studied in S. aureus by Strominger and coworkers 4s. Using the particulate preparation which catalyzes formation of the sugar-linked intermediates required in peptidoglycan synthesis, they discovered that addition of ATP, which is not required for any reaction in the synthetic cycle, caused an increased synthesis of lipid intermediate and an increase in the formation of [32P]UDP-N-acetylmuramyl pentapeptide (a measurement of the exchange between [a2P]UMP and unlabeled UDP-N-acetylmuramylpentapeptide which occurs in the first step of the cycle). The effect of ATP was thereby determined to be at the first step in the reaction sequence. Addition of ficaprenol, an isomer of the bacterial undecaprenol, to the ATP-stimulated system caused a further increase in synthesis Biochim. Biophys. Acta, 265 (1972) 417-441

430

W.J. LENNARZ, M. G. SCHER

of lipid intermediate. In the absence of the sugar nucleotides required for peptidoglycan or lipid intermediate synthesis, formation of a 3Zp-labeled lipid with the chromatographic mobility of polyisoprenoid alcohol phosphate was detected if the added ATP was labeled with 32p, and an increased amount of this 32p-labeled lipid was formed with added ficaprenol. Thus it was concluded that the stimulatory effects of ATP and ficaprenol were due to the presence of a polyprenol kinase. The specificity of the kinase was not definitely determined although C5o and Css ficaprenols and C35 and C4o betulaprenols (polyprenols isolated from plant sources) were found to be active, while higher prenologs, C8o-C1 os, were inactive. The ability of the product of the ATP-dependent phosphokinase to function as the lipid acceptor in formation of lipid intermediate was determined as follows. A particulate enzyme preparation containing kinase activity as well as the capacity to form lipid-linked sugar intermediates was incubated with [3Zp]ATP and exogenous isoprenoid alcohols in the presence or absence of sugar nucleotides. In the absence of sugar nucleotides, the formation of [32p]polyisoprenol phosphate was observed. However, in the presence of sugar nucleotides synthesis of 3Zp-labeled lipid intermediate was observed, and formation of this intermediate was accompanied by a decrease in the amount of radioactivity associated with polyisoprenol phosphate. Ficaprenol phosphate itself was also active when supplied directly as a substrate for the synthesis of lipid intermediate. The kinase activity is found only in the particulate fraction, and this membrane-associated enzyme has the interesting property of being soluble in butanol. A phospholipid component necessary for kinase activity was separated from the protein component. The phospholipid consists of a mixture of phosphatidylglycerol and cardiolipin either of which can be used alone to restore enzyme activity49. Following a 630-fold purification of the protein, which was still associated with lipid, the molecular weight was estimated by gel electrophoresis as 17000. The amino acid content of the protein moiety was determined 5°. Non-polar amino acids comprised 57.8 ~ of the amino acid residues, which is the highest non-polar amino acid content of a protein yet reported. Generation of the polyprenol chain has been studied by Christenson et al.51 in S. newington. Both the soluble and the particulate fractions catalyze the synthesis of polyprenols from A3-isopentenyl pyrophosphate and farnesyl pyrophosphate. After acid hydrolysis and chromatography the products of the soluble enzyme were identified as polyprenols shorter in length than 55 carbons, whereas the product of the particulate fraction was chromatographically identical to antigen carrier lipid. The relationship between these two different enzyme activities remains unclear. Identification of the product of the particulate preparation as functional antigen cartier lipid was accomplished using parallel incubation mixtures, only one of which contained the sugar nucleotides appropriate for synthesizing the disaccharide pyrophosphoryl-antigen carrier lipid involved in O-antigen biosynthesis. Both mixtures contained the substrates A3-[1-1~C]isopentenyl pyrophosphate and [1-3H]farnesyl pyrophosphate. After the enzymatic reaction, the mixtures were extracted with butanol and both extracts were applied to DEAE-cellulose columns. Both elution profiles contained the same 14C- and 3H-containing peaks except that the reaction mixture Biochim. Biophys. Acta, 265 (1972) 417-441

POLYISOPRENOLSUGAR INTERMEDIATES

431

containing sugar nucleotides produced an additional 14C, 3H-containing peak that eluted in the position expected for disaccharide pyrophosphoryl-antigen carrier lipid. Thus, evidence is provided for synthesis of biologically active polyisoprenol. If this antigen carrier lipid should be identical to undecaprenol from the mannan system which contains two internal trans double bonds 3°, and if these two trans double bonds are located at the end of the molecule distal to the pyrophosphate group, it is likely that the biosynthesis proceeded by successive condensations of A 3-isopentenyl pyrophosphate with trans, trans-farnesyl pyrophosphate in which each new isoprene unit has the cis configuration (cf. Fig. 7). The work of Bloch and his coworkers 52 indicated that geranylgeranyl pyrophosphate synthesis catalyzed by a soluble enzyme from M. lysodeikticus was the result of addition of isopentenyl units to a growing intermediate. The substrates of this system were isopentenyl pyrophosphate and farnesyl pyrophosphate. Dimethylallyl or geranyl pyrophosphate also served as substrates but had a slower reaction rate than farnesyl pyrophosphate. Since the enzyme was not purified to homogeneity the lack of absolute specificity for substrate could be due to catalysis by more than one enzyme. If the production of geranylgeranyl pyrophosphate from all of the substrates tested was attributable to only one protein, then a possible synthetic mechanism would have been the addition of isopentenyl residues to a growing enzyme-bound allyl intermediate. In a second report from Bloch's laboratory s3 another soluble M. lysodeikticus preparation was found to catalyze the synthesis of longer chain polyprenol pyrophosphates. The number of isopentenyl units (from [1-14C]isopentenyl pyrephosphate) added to [1-aH]farnesyl pyrophosphate was estimated by comparison of the ~4C:3H ratios to the reference value for geranylgeraniol (C2o) produced by geranylgeranyl pyrophosphate synthetase. Such comparisons indicated that the major polyprenyl pyrophosphates produced by this preparation have a chain length of approximately 7 to 8 isoprenoid units as a result of condensation of approximately 4 to 5 isopentenyl units with farnesyl pyrophosphate. The result of chromatography also indicated that terpene alcohols, predominantly C35 and C4o, were produced in the enzymatic reaction. In contrast to the product of geranylgeranyl pyrophosphate synthetase, the long-chain polyprenyl pyrophosphates are resistant to alkaline phosphatase. The resistance may be due to the extremely hydrophobic nature of the reaction product or to the strong binding between enzyme and product. Another system which catalyzes the synthesis of long-chain isoprenoid compounds (C55 and lower isoprenologs) from isopentenyl pyrophosphate and farnesyl pyrophosphate was reported by Kurokawa et al. 54. The polyprenyl compounds produced by their curde cell free extract (77000 × g supernatant fraction) of M. lysodeikticus range in length from C2o to Css with a predominant chain length of C4o. These compounds were compared to the long-chain compounds produced by the purifiedlong-chainprenylpyrophosphatesynthetaseofBloch'sgroup 53. The products of both enzymes were extractable with chloroform-methanol or n-butanol, were not hydrolyzed by alkaline phosphatase, and were eluted in the void volume from Sephadex G-25 columns. However, only the product of the crude enzyme system was Biochim. Biophys. Acta, 265 (1972) 417-441

432

W . J . LENNARZ, M. G. SCHER

hydrolyzed by yeast and rabbit liver homogenates (which had been reported to catalyze dephosphorylation), and comparison by thin-layer chromatography of the products of both enzymic systems before dephosphorylation suggested that the product of the crude system was not pyrophosphate, but monophosphate. Therefore, there may exist in the crude enzyme preparation an enzyme which hydrolyzes the proposed initial products, polyprenyl pyrophosphates, to potyprenyl phosphates. Such a polyprenyl pyrophosphatase is required to regenerate undecaprenyl phosphate during synthesis of O-antigen and peptidoglycan, and Siewert and Strominger 15 have focused upon the action of this enzyme in their study of the inhibitory effect of bacitracin on peptidoglycan synthesis in M. lysodeikticus or S. aureus preparations. In reaction mixtures containing disaccharide (pentapeptide)pyrophosphorylundecaprenol doubly labeled with ~¢C and 32p, the addition of bacitracin caused a decrease in release of inorganic phosphate and an increase in the amount of a 32p-containing lipid of chromatographic mobility different from the substrate. The effect of bacitracin on cell wall synthesis was thus attributed to inhibition of a specific polyprenyl pyrophosphatase (see Fig. 7). The mechanism by which bacitracin inhibits this reaction has been clarified by Stone and Strominger 55. They discovered that EDTA, which alone has no effect on dephosphorylation, restores enzymatic activity to a bacitracin-inhibited system. The restorative effect of EDTA is overcome by adding Mg 2÷ in slight excess of the amount of EDTA present. The effect of EDTA 3 Isoprenyl POP

1

Fornesyl POP j.,~----8 Isoprenyl POP

CH CH~ Ci-h s CH~ I I I / C = C H - C H 2 -(CH2-C= CH-CHi) 2- (CHi-C =CH-CH2)7-CHi-C=CH-CH20POP (;H3

traas

/ / /

/

c.,

I

cis

cis

~ Blocked by bacitrocin CH3 ATP CH3 I ,/ = R-C=CH-CHtOP ~ R-C=CH-CHtOH

/

l

/ Rk CInCH-CH 20 POP- 9 lycose

7"""

\

R-C= CH-CH2OPO.QlycoseX

Fig. 7. Undecaprenol metabolism in the eubacteria. XDP-glycose, XDP, X M P represent a sugar nucleotide, nucleoside diphosphate and nucleoside monophosphate, respectively. R is substituted for the portion of the undecaprenol chain not indicated in the figure.

Biochim. Biophys. Acta, 265 (1972) 417-441

POLYISOPRENOL SUGAR INTERMEDIATES

433

in abolishing inhibition was lessened if bacitracin, metal ions and substrate were incubated together before EDTA addition. In addition, formation of a complex was suggested by the change bacitracin produces in the elution profile on Sephadex G-25 of a2p-labeled undeeaprenyl pyrophosphate in buffer containing Mg 2+ and deoxycholate. On the basis of these and other observations the authors concluded that a complex forms between bacitracin, metal cation, and lipid substrate. Presumably once the complex has formed, EDTA has only a minimal effect. In the proposed model Mg 2+ acts as a bridge between 2 oxygen atoms located in adjacent phosphate groups of undecaprenyl pyrophosphate and 2 peptide-linked nitrogens of the bacitracin molecule. The relationship between the enzymic reactions pertaining to undecaprenol metabolism and utilization of undecaprenols in the bacteria is summerized schematically in Fig. 7. The phospholipid which serves as acceptor lipid in the formation of activated sugar intermediates assumes a central position in this scheme. It can be converted to either a phosphodiester or a pyrophosphate-linked sugar intermediate. Formation of acceptor lipid may result either by phosphorylation of the alcohol catalyzed by an ATP-dependent phosphokinase, or by dephosphorylation of undecaprenyl pyrophosphate which has been generated from farnesyl and isopentenyl pyrophosphate precursors. The phospholipid can also be directly recycled in the polysaccharide-synthesizingsystems after release from the phosphodiester-linked sugar intermediates, or indirectly recycled following dephosphorylation of the lipid pyrophosphate released from pyrophosphate-linked sugar intermediates. Many details in this scheme, particularly with regard to synthesis of undecaprenyl phosphate, remain to be clarified. IV. PLANT AND ANIMALSYSTEMS I V A . Phaseolus aureus

The discovery of lipid (mung bean) intermediates in polysaccharide synthesis has not been confined to bacterial systems. Since polyisoprenols are present in higher plants s6'57, analogous reactions might be expected in plant systems. Kauss 58 has observed enzymatic transfer of mannose from GDP[~4C]mannose to a radioactive lipid (derived from [aH]mevalonic acid) in a particulate preparation from mung bean shoots. Radioactive mannan is also formed in this system. Reversibility of the reaction was demonstrated by the addition of unlabeled GDP which effected a decrease in the [~4C]mannose associated with glycolipid. Moreover, the addition of [x4C]GDP to an enzyme incubation mixture containing endogenous acceptor lipid and unlabeled GDPmannose resulted in production of a4C-labeled GDPmannose. Similarly when the lipid product labeled with [~4C]mannose was incubated with the enzyme and GDP, GDP[~4C]mannose was produced. Addition of unlabeled GDPmannose caused a decrease in the level of radioactivity of lipid, and a decrease in the rate of incorporation of 14C into plant mannan. The time course of synthesis of lipid and mannan is consistent with the possibility that the lipid is an intermediate in mannan Biochim. Biophys. Acta, 265 (1972) 417-441

434

W.J. LENNARZ, M. G. SCHER

synthesis, but since only minimal incorporation of mannose from the mannosecontaining lipid into polysaccharide was detected, the function of the lipid as an intermediate, as well as its structure, remains to be conclusively established. Villimez and Clark 59 have also reported preliminary evidence that a mannolipid functions as an intermediate with a Phaseolus aureus particulate preparation which catalyzed mannose transfer from GDP[~4C]mannose to polysaccharide. The intermediate and the polysaccharide were present in a trichloroacetic acid precipitate which was extracted with phenol and washed with chloroform-methanol to separate the intermediate from the polysaccharide. A low level of incorporation of radioactivity from the isolated intermediate into the polysaccharide was detected, and rate and isotope dilution studies gave further support to the theory of a lipid intermediate. Subsequently these authors discovered that two components, a mannolipid and a glycoprotein had been included in their measurements of the "intermediate''6°. These components could be separated by subjecting the initial trichloroacetic acid precipitate to a butanol extraction followed by a phenol extraction. The radioactive component in the phenol extract was identified as glycoprotein by its susceptibility to pronase digestion which produced low molecular weight fragments. Mild acid hydrolysis of these fragments followed by paper chromatography produced a single radioactive peak which had the mobility of cellotetrose; mannose was the only radioactive sugar found in the glycoprotein. The isolated mannolipid was subjected to mild acid treatment and mannose was the only radioactive sugar released under conditions which would not have hydrolyzed a glycosidic linkage between two hexose residues. Milder conditions of hydrolysis produced mannose-l-phosphate, and it was therefore concluded that mannose was linked to lipid via a phosphate group. The resistance of the compound to alkali further suggested that this linkage was via a phosphate rather than a pyrophosphate bridge. Preliminary data on the mass spectrum of the lipophilic portion of the molecule are consistent with a long chain polyisoprenoid structure. These properties of the mannolipid along with its chromatographic behavior suggest a polyisoprenol phosphate derivative of mannose analogous in structure to the lipid intermediate involved in M. lysodeikticus mannan synthesis. Further evidence that this mannolipid intermediate is of the polyisoprenol type has been reported. Alam et al. 6~, using the same enzyme preparation, reported that addition of exogenous dolichol phosphates and betulaprenol phosphates caused a stimulation in the incorporation of [~4C]mannose from GDP[~4C]mannose into lipid6L Dolichol is a mixture of polyprenols containing 15 to 20 isoprene units the hydroxy terminal one of which is saturated, and betulaprenols are C3o-C45 polyprenols isolated from plant sources. Alam and Hemming62 prepared [3H]betulaprenol phosphate and showed the transfer of [a4C]mannose from GDP[a4C]mannose to this compound. The mannolipid produced with exogenous betulaprenol phosphate also resembles mannosyl-phosphorylundecaprenol in its chromatographic properties. acid lability and alkaline stability (unpublished results) 62.

Biochim. Biophys. Acta, 265 (1972) 417-441

POLYISOPRENOL SUGAR INTERMEDIATES

435

IVB. Mammalian systems

In 1969, Caccam and coworkers 63 detected the formation of a mannosecontaining lipid in a rabbit liver smooth membrane fraction which catalyzes synthesis of a glycoprotein from GDPmannose. The ability to transfer mannose from the sugar nucleotide to a "tightly bound" lipid acceptor was found in those tissues (oviduct and liver) which are known to secrete mannose-containing glycoproteins. In vitro studies on the rate of incorporation of mannose showed that transfer of [14C]mannose from GDP[14C]mannose to lipid was more rapid than transfer to protein. Unlabeled GDP-mannose added to the reaction mixture caused a decrease in [~4C]mannolipid and an increase in [14C]mannose-labeled glycoprotein, thus suggesting the possibility of a lipid intermediate. The lipid was partially purified on silicic acid and DEAEcellulose columns. Mild acid hydrolysis released [14C]mannose and the hexose to phosphate ratio ranged from 1 : 1 to 2: 1. The lipid was resistant to alkaline hydrolysis and it had the chromatographic properties of mannosyl-l-phosphorylundecaprenol. Further information on the structure and the function of the lipid in glycoprotein synthesis in this system has not been published. There is also preliminary evidence for formation of a mannose- and phosphate-containing lipid in a rat brain microsomal system 64. The lipid resembles other isoprenoid intermediates in its alkaline stability and acid lability. Details of the structure and function of this compound also remain to be established. Incorporation of amino sugars from nucleotide precursors into endogenous lipid is catalyzed by rabbit liver microsomal preparations 65. In a reaction with UDPN-acetyl-[14C]hexosamine (a mixture of UDP-N-acetylglucosamine and UDP-Nacetylgalactosamine), two trichloroacetic acid-insoluble products containing the radioactive sugars were formed. Approximately 50 % of the trichloroacetic acid-insoluble product was extractable with organic solvents and the remaining 50 % was hexosaminecontaining protein. The initial rate of incorporation of N-acetylhexosamine into lipid was always greater than its rate of incorporation into protein, and in a pulse-chase experiment utilizing UDP-N-acetyl-[14C]glucosamine radioactivity incorporated into lipid decreased with the time of incubation in the presence of unlabeled substrate, whereas radioactivity in the protein continued to rise. These observations are consistent with the possibility that a glycolipid is involved in synthesis of the glycoprotein. UDP or UMP added to a reaction mixture caused 60--80% inhibition of the incorporation of N-acetyl-[14C]hexosamine into lipid and glycoprotein. The linkage between the sugar and lipid was susceptible to cleavage by mild acid treatment or snake venom phosphodiesterase. Subsequently 66, the transfer of equimolar amounts of 3zp and N-acetyl-[14C]glucosamine from [32p]UDP-N-acetyl-[~4C]glucosamine to endogenous lipid in the microsomal preparation was reported from the same laboratory. UMP was probably released as the nucleotide product. The linkage between the sugar and lipid was subject to cleavage by mild acid treatment or snake venom phosphodiesterase. Analysis of the water-soluble components released by mild acid hydrolysis of the partially purified lipid containing ~4C and 32p indicated that Nacetylglucosamine, inorganic phosphate, N-acetylglucosamine 1-phosphate, and Biochim. Biophys. Acta, 265 (1972) 417-441

436

W.J. LENNARZ, M. G. SCHER

inorganic pyrophosphate were among the products. The presence of inorganic pyrophosphate suggests that the sugar is probably linked via a pyrophosphate bridge to the lipid moiety, which has not yet been characterized. The nature of the lipid moiety of mammalian glycolipid intermediates was studied indirectly by Alam et al. 61. They observed transfer of mannose from GDPmannose to endogenous lipid in a pig liver preparation under conditions of incubation similar to those described by Caccam eta[. 63. Exogenous lipids including the phosphates of pig liver dolichols, betulaprenols, solanesol (an all-trans C4s alcohol), and ficaprenol produced an increase in the amount of mannose transferred to lipid, suggesting that polyprenol phosphates serve as acceptor lipids in a glycosyl transfer reaction. Further experiments from the same laboratory were oriented toward a direct determination of the nature of the acceptor lipid involved in the pig liver reaction 67. They synthesized high specific activity [3H]dolichol phosphate which was incubated with GDP[14C]mannose. Chromatographic assay of the lipid products of the reaction showed the presence of a compound containing both 14C and 3H that had the expected mobility for a mannose-containing dolichol phosphate. Treatment of this compound with dilute acid resulted in release of [14C]mannose from the lipid and production of a 3H-containing lipid with the mobility of dolichol phosphate. Similarly it was found that a chicken liver enzyme preparation also utilized exogenous dolichol phosphates as acceptor lipid. Moreover, the utilization of endogenous dolichol phosphates by the chicken liver system was demonstrated by the synthesis of [14C]mannolipid having the expected chromatographic properties and containing 3H derived from mevalonic acid which had been inoculated into chick embryos. Participation of dolichol in the biosynthesis of another glycolipid, characterized as dolichol monophosphate glucose has been reported by Behrens and Leloir 6s. Glycolipid formation was catalyzed by rat liver microsomes and the transfer of [14C]glucose into lipid was enhanced by the addition of crude acceptor lipid. Glucose was released during acid hydrolysis of the partically purified glycolipid. When analyzed by infrared spectroscopy the partially purified acceptor lipid had some of the properties associated with dolichol, but the nature of the lipid moiety was determined by the functional and chromatographic similarity of phosphorylated acceptor lipid and phosphorylated pure dolichol. The reaction may not be absolutely specific for the acceptor lipid, since Jankowski and Chojnacki 69 have reported transfer of glucose from UDPglucose to ficaprenol phosphate as well as to an unsaponified lipid from pig liver which they believed to be dolichol monophosphate. Phytol phosphate, on the other hand, was inactive. This enzymatic activity was demonstrated in an enzyme preparation similar to that utilized by Behrens and Leloir 6s. Unpublished results obtained with [a2p]UDPglucose69 suggest that the mechanism of forming glucosylphosphorylficaprenol is similar to formation of the glucolipid with the longer chain alcohol as reported by Behrens and Leloir 6s. Behrens and Leloir 6s also observed that isolated dolichol monophosphate glucose (that had been enzymatically produced) could serve as a glucosyl donor upon reincubation with the enzyme preparation. During the second incubation the level Biochim. Biophys. Acta, 265 (1972) 417-441

POLYISOPRENOL SUGAR INTERMEDIATES

437

of radioactivity in the lipid phase decreased and the radioactive product appeared to be a glucoprotein. In a later paper 7° they reported that the microsomal preparation also catalyzed the transfer of N-acetylglucosamine and mannose from UDPN-acetylglucosamine and GDPmannose, respectively, to lipid, but the lipid products of these reactions did not act as substrates for glycoprotein synthesis under the same reaction conditions used with dolichol monophosphate glucose as substrate. The concentrations of detergents, such as Triton X-100 or sodium deoxycholate, that were optimal for the transfer of glucose from lipid to protein were determined. Neither UDP nor UDPglucose affected the reaction, but Mn 2÷ was an inhibitor. In an attempt to determine whether dolichol monophosphate glucose serves as an intermediate in glucosylation of collagen or ceramide, enzyme preparations which had been reported to catalyze glucose transfer from UDPglucose to protein 71 and to ceramide 72 were utilized, and the effect of dolichol monophosphate addition on glucose incorporation was studied 7°. Under conditions optimal for collagen synthesis, that is, with a mammalian skin preparation as enzyme source and gelatin serving as the protein acceptor, addition of dolichol monophosphate stimulated glucolipid formation, but there was no stimulatory effect on protein synthesis. Moreover, comparison of the products of glucose transfer from UDPglucose to protein catalyzed by the skin enzyme, and from dolichol monophosphate glucose to endogenous protein catalyzed by liver microsomes, demonstrated that the acceptor proteins in the two systems were different, since the electrophoretic properties of the hydrolyzed products containing [14C]glucose were distinctly different. The possibility that dolichol monophosphate glucose is involved in glucosylation of ceramide was investigated using a rat brain preparation which catalyzes the transfer of glucose from UDPglucose to ceramide 72. Glucose transfer from UDPglucose to exogenous dolichol monophosphate was observed in this preparation, but in the presence of EDTA, which inhibits formation of dolichol monophosphate glucose, ceramide glucose synthesis continued. Thus the glucolipid does not participate in glucosylation of ceramide or collagen, and the nature of the glucoprotein which is glucosylated by dolichol monophosphate glucose remains unknown. Baynes and Heath T3 have reported that mannose from GDPmannose is transferred to an endogenous lipid in a mouse myeloma tumor preparation. The reaction is inhibited by EDTA, and addition of GDP to the reaction results in a loss of [~4C]mannose from the lipid. Mannose is released from the mannolipid as a result of mild acid hydrolysis, and the mannolipid behaves like an acidic lipid on thin-layer chromatography and DEAE-cellulose. Preliminary evidence suggests that a compound with a structure tentatively established as retinol monophosphate galactose may serve as an intermediate in glycoprotein biosynthesis in mouse mastocytoma particulate preparations 74. With UDP[~4C]galactose serving as the galactosyl donor in the presence of endogenous lipid or exogenous retinol the resulting product has the expected behavior of a lipid-linked phosphodiester upon chromatography on DEAE-Sephadex. Also consistent with the proposed structure is the fact that galactose 1-phosphate, identified Biochim. Biophys. Acta, 265 (1972) 417-441

438

w.J. LENNARZ, M. G. SCHER

electrophoretically, was released from the lipid by hydrogenolysis. Incubation of labeled retinol monophosphate galactose with the particulate preparation resulted in a decrease of radioactivity in the chloroform-methanol phase and an increased radioactivity in chloroform-methanol insoluble material which has not yet been identified. The particulate preparation also catalyzed transfer of mannose from GDP-mannose to endogenous lipid and the product behaved similarly to the galactosyl compound upon chromatography on DEAE-Sephadex. Another system for mannosyl transfer to a polar derivative of retinol has been reported by DeLuca et al. 75. Incorporation of [14C]mannose from GDP[14C]mannose into lipid and protein in the presence of a rat liver preparation was enhanced by the addition of exogenous retinol and by the addition of ATP. As a result of investigations into the nature of the lipid moiety of the mannolipid, similarities in the chromatographic mobility of the unknown lipid and retinoic acid have been found. However, despite these similarities other evidence appears to exclude the possibility that the polar metabolite of retinol is retinoic acid. The function of the mannose-containing retinol derivative in formation of the mannoprotein has not been demonstrated, but the two studies involving retinol suggest that vitamin A or a metabolite of it (retinol phosphate?) might serve as a lipophilic hexose carrier in glycoprotein synthesis. The role of vitamin K in glycoprotein (prothrombin) biosynthesis has been investigated, and the data indicate that administration of vitamin K1 to vitamin Kdeficient rats stimulates the incorporation in vivo of mannose and glucosamine into protein without affecting amino acid incorporation into the protein moiety76. Moreover, unpublished data 76 suggest enzymic formation of a glycolipid from vitamin K1 and GDPmannose. An understanding of synthesis of the glycolipid may lead to clarification of the nature of the vitamin K requirement for formation of prothrombin.

v. CONCLUDINGCOMMENTS Although lipid-linked activated sugars were first discovered in bacterial systems, recent reports suggest the occurrence of similar compounds in plant and mammalian systems. However, much work must be done before the role of these compounds in mammalian glycan or glycoprotein synthesis is understood. If further studies do establish a role for these compounds in the synthesis of mammalian glycoprotein, this would provide yet another example of the unity of biochemical mechanisms. It is clear that the function of the carrier lipids in bacteria is to convert water-soluble activated sugars to hydrophobic molecules that can glycosylate macromolecules within the hydrophobic environment of the cytoplasmic membrane. Since the glycosylation of mammalian glycoproteins is believed to occur within the hydrophobic environment of the Golgi apparatus, it is not difficult to envision an analogous mechanism for the participation of lipids in mammalian glycoprotein synthesis.

Biochim. Biophys. Acta, 265 (1972) 417-441

POLYISOPRENOL SUGAR INTERMEDIATES

439

NOTE ADDED IN PROOF (Received June 14th, 1972) In publications which have appeared since the literature search for this review was completed, additional evidence that lipid intermediates are involved in teichoic acid synthesis has been presented, although the lipid still has not been characterized in detail. Experiments which showed competition between the peptidoglycan and teichoic acid synthesizing systems in S. lactis 13 (ref. 43) have been extended to include B. licheniformis and B. subtilis (R. G. Anderson, H. Hussey and J. Baddiley, Biochem. J., 127 (1972) 11). As a result of these more recent investigations, Anderson et al. suggest that undecaprenol phosphate, acting as a common lipid in synthesis of various wall components, can control the rate of bacterial wall synthesis, and that it is not freely mobile between enzyme complexes in the membrane. Participation of two glycolipids in teichoic acid synthesis has been demonstrated with a B. licheniformis particulate preparation (I. C. Hancock and J. Baddiley, Biochem. J., 127 (1972) 27). Glucose, but not phosphate is transferred from UDPglucose to form a lipid which behaves like glucosylphosphorylundecaprenol, releasing glucose 1-phosphate during hydrogenolysis. Upon incubation with CDPglycerol, glycerol and phosphate are transferred to the glucosyl lipid and polymer is also formed. Similary, in S. lactis 13, a particulate preparation catalyzes transfer of both N-acetylglucosamine and phosphate from UDP-N-acetylglucosamine to a phospholipid (H. Hussey and J. Baddiley, Biochem. J., 127 (1972) 39). Glycerol and phosphate are subsequently transferred from CDPglycerol to the N-acetylglucosamine-containing lipid with concomitant formation of polymeric material. With regard to the biosynthesis of the polysoprenols, Durr and Habbal have reported the de novo synthesis of undecaprenol by a cell-free preparation from L. plantarum (I. F. Durr and M. Z. Habbal, Biochem. J., 127 (1972) 345). Radioactive lipid was formed from [14C]mevaionate, and after purification a 1*C-containing fraction was found to co-chromatograph with authentic bactoprenol, but analysis showed that this fraction was composed primarily of undecaprenol.

ACKNOWLEDGEMENT It is a pleasure to acknowledge support by the National Institutes of Health (5RO1A10688 07) and the National Science Foundation (GB 31121) for studies on mannan synthesis in M. lysodeikticus. We are also indebted to the National Science Foundation for financial aid in preparation of this review.

REFERENCES 1 J. S. Anderson, M. Matsuhashi, M. A. Haskin and J. L. Strominger,Proc. NatlAcad. Sci. U.S., 53 (1965) 881. 2 I. M. Weiner, T. Higuchi, L. Rothfield, M. Saltmarsh-Andrew,M. J. Osborn and B. L. Horecker, Proc. Natl Acad. Sci. U.S., 54 (1965) 228. Biochim. Biophys. Acta, 265 (1972) 417-441

440

W . J . LENNARZ, M. G. SCHER

3 A. Wright, M. Dankert and P. W. Robbins, Proc. Natl Acad. Sci. U.S., 54 (1965) 235. 4 A. Wright, M. Dankert, P. Fennessey and P. W. Robbins, Proc. Natl Acad. Sci. U.S., 57 (1967) 1798. 5 Y. Higashi, J. L. Strominger and C. C. Sweeley, Proc. Natl Acad. Sci. U.S., 57 (1967) 1878. 6 M. J. Osborn, Ann. Rev. Biochem., 38 (1969) 501. 7 E. C. Heath, Ann. Rev. Biochem., 40 (1971) 29. 8 A. Wright and S. Kanegasaki, Physiol. Rev., 51 (1971) 748. 9 W. J. Lennarz, Ann. Rev. Biochem., 39 (1970) 359. 10 W. J. Lennarz, in S. J. Wakil, Lipid Metabolism, Academic Press, New York, 1970, p. 155. 11 M. G. Heydanek, Jr., W. G. Struve and F. C. Neuhaus, Biochemistry, 8 (1969) 1214. 12 J. S. Anderson, M. Matsuhashi, M. A. Haskin and J. L. Strominger, .L BioL Chem., 242 (1967) 3180. 13 M. Matsuhashi, C. P. Dietrick and J. L. Strominger, Proc. NatlAcad. Sci. U.S., 54 (1965) 587. 14 T. Kamiryo and M. Matsuhashi, Biochem. Biophys. Res. Commun., 36 (1969) 215. 15 G. Siewert, and J. L. Strominger, Proc. Natl Acad. Sci. U.S., 57 (1967) 767. 16 J. Ghuysen, Bacteriol. Rev., 32 (1968) 425. 17 M. Dankert, A. Wright, W. S. Kelley and P. W. Robbins, Arch. Biochem. Biophys., 116 (1966) 425. 18 S. Kanegasaki and A. Wright, Proc. Natl Acad. Sci. U.S., 67 (1970) 951. 19 M. J. Osborn and I. M. Weiner, J. Biol. Chem., 243 (1968) 2631. 20 J. L. Kent and M. J. Osborn, Biochemistry, 7 (1968) 4419. 21 P. W. Robbins, D. Bray, M. Dankert and A. Wright, Science, 158 (1967) 1536. 22 P. W. Robbins and M. J. Osborn, unpublished observations, (cited in ref. 6). 23 A. Wright, Fed. Proc., 28 (1969) 658. 24 A. Wright, J. Bacteriol., 105 (1971) 927. 25 H. Nikaido, K. Nikaido, T. Nakae and P. H. M/ikel~i, J. Biol. Chem., 246 (1971) 3902. 26 K. Nikaido and H. Nikaido, J. Biol. Chem., 246 (1971) 3912. 27 M. Takeshita and P. H. M~ikel~i, J, Biol. Chem., 246 (1971) 3920. 28 M. Scher and W. J. Lennarz, J. Biol. Chem., 244 (1969) 2777. 29 M. Lahav, T. H. Chiu and W. J. Lennarz, J. Biol. Chem., 244 (1969) 5890. 30 M. Scher, W. J. Lennarz and C. C. Sweeley, Proc. Natl Acad. Sci. U.S., 59 (1968) 1313. 31 L. F. Leloir, M. A. R. de Fekete and C. E. Cardini, J. Biol. Chem., 236 (1961) 636. 32 N. H. Behrens and E. Cabib, J. Biol. Chem., 243 (1968) 502. 33 W. Tanner, Biochem. Biophys. Res. Commun., 35 (1969) 144. 34 W. Tanner, P. Jung and N. H. Behrens, FEBS Lett. 16 (1971) 245. 35 R. Sentandreu and J. O. Lampen, FEBS Lett., 14 (1971) 109. 36 F. A. Troy, F. E. Frerman and E. C. Heath, J. Biol. Chem., 246 (1971) 118. 37 F. E. Frerman, E. C. Heath, M. Lahav, T. H. Chiu and W. J. Lennarz, unpublished observations (cited in ref. 36). 38 1. W. Sutherland and M. Norval, Biochem. J., 120 (1970) 567. 39 K. Takayama and D. S. Goldman, J. Biol. Chem., 245 (1970) 6251. 40 L. J. Douglas and J. Baddiley, FEBS Lett. 1 (1968) 114. 41 D. Brooks and J. Baddiley, Biochem. J., 113 (1969) 635. 42 D. Brooks and J. Baddiley, Biochem. J., 115 (1969) 307. 43 R. J. Watkinson, H. Hussey and J. Baddiley, Nature New Biol., 229 (1971) 57. 44 M. Stow, B. J. Starkey, S. C. Hancock, and J. Baddiley, Nature New Biol., 229 (1971) 56. 45 K. J. I. Thorne and E. Kodicek, Biochem. J., 99 (1966) 123. 46 D. P. Gough, A1 L. Kirby, J. B. Richards and F. W. Hemming, Biochem. J., 118 (1970) 167. 47 K. J. I. Thorne and D. C. Barker, Biochem. J., 122 (1971) 45 p. 48 Y. Higashi, G. Siewert and J. L. Strominger, J. Biol. Chem., 245 (1970) 3683. 49 Y. Higashi and J. L. Strominger, J. Biol. Chem., 245 (1970) 3691. 50 H. Sandermann, Jr and J. L. Strominger, Proc. Natl Acad. Sci. U.S., 68 (1971) 2441. 5l J. G. Christenson, S. K. Gross and P. W. Robbins, J. Biol. Chem., 244 (1969) 5436. 52 A. A. Kandutsch, H. Paulus, E. Levin and K. Bloch, J. Biol. Chem., 239 (1964) 2507. 53 C. M. Allen, W. Ailworth, A. Macrae and K. Bloch, J. Biol. Chem., 242 (1967) 1895. 54 T. Kurokawa, K. Ogura and S. Seto, Biochem. Biophys. Res. Commun., 45 (1971) 251. 55 K. J. Stone and J. L. Strominger, Proc. Natl Acad. Sci. U.S., 68 (1971) 3223. 56 A. R. Wellburn and F. W. Hemming, Phytochemistry, 5 (1966) 969. 57 A. R. Wellburn and F. W. Hemming, Nature, 212 (1966) 1364. Biochim. Biophys. Acta, 265 (1972) 417-441

POLYISOPRENOL S U G A R INTERMEDIATES 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76

441

H. Kauss, FEBS Lett., 5 (1969) 81. C. L. Villimez and A. F. Clark, Biochem. Biophys. Res. Commun., 36 (1969) 57. C. L. Villimez, Biochem. Biophys. Res. Commun., 40 (1970) 636. S. S. Alam, R. M. Barr, J. B. Richards and F. W. Hemming, Biochem. J., 121 (1971) 19 P. S. S. Alam and F. W. Hemming, FEBS Lett. 19 (1971) 60. J. F. Caccam, J. J. Jackson and C. H. Eylar, Biochem. Biophys. Res. Commun., 35 (1969) 505. M. Zatz and S. H. Barondes, Biochem. Biophys. Res. Commun., 36 (1969) 511. M. Tetas, H. Chao and J. Molnar, Arch. Biochem. Biophys., 138 (1970) 135. J. Molnar, H. Chao and Y. Ikehara, Biochim. Biophys. Acta., 239 (1971) 401. J. B. Richards, P. J. Evans and F. W. Hemming, Biochem. J., 124 (1971) 957. N. H. Behrens and L. F. Leloir, Proc. Natl Acad. Sei, U.S., 66 (1970) 153. W. Jankowski and T. Chojnacki, Biochim. Biophys. Acta, 260 (1972) 93. N. H. Behrens, A. J. Parodi, L. F. Leloir and C. R. Krisman, Arch. Biochem. Biophys., 43 (1971) 375. H. B. Bosmann and E. H. Eylar, Biochem. Biophys. Res. Commun., 30 (1968) 89. S. Basu, B. Kaufman and S. Roseman, J. BioL Chem., 243 (1968) 5802. J. W. Baynes and E. C. Heath, Fed. Proc., 31 (2) (1972) Abstr. 1239. T. Helting and P. A. Peterson, Biochem. Biophys. Res. Commun., 46 (1972) 429. L. DeLuca, G. Rosso and G. Wolf, Biochem. Biophys. Res. Commun., 41 (1970) 615. H. V. Johnson, J. Martinovic and B. C. Johnson, Biochem. Biophys. Res. Commun., 43 (1971) 1040.

Biochim. Biophys. Acta, 265 (1972) 417-441