Biosynthesis of the Monoguanidinated Inositol Moiety of Bluensomycin, a Possible Evolutionary Precursor of Streptomycin*

SUMMARY The general pattern of biosynthesis of the bluensidine (1D - 1 - 0 - carbamoyl-3 - guanidino - 3 - deoxy - scyllo - inositol) moiety of bluensomycin, a monoguanidinated analogue of dihydrostreptomycin, has been studied in extracts of Streptomyces hygroscopicus forma glebosus ATCC 14607 (S. glebosus). Our results are consistent with the following C biosynthetic pathway: A major uncertainty concerns the step (U) at which the carbamoyl group is introduced.

From the Department of Biochemistry, Rice University, Houston, Texas 77001
S. glebosus extracts also have 1 -guanidino -1 -deoxy -scyllo -inositol -4-P phosphohydrolase activity; neither this enzyme nor the corresponding enzyme from streptomycin producers can dephosphorylate the transamidination product presumed to be bluensidine-6-P. Acid hydrolysis of the latter compound gave a compound which, * This research was supported by Grant C-153 from the Robert A. Welch Foundation, Houston, Texas. unlike the unhydrolyzed compound, was converted to lD-lguanidino-3-amino-1,3-dideoxy-scyllo-inositol by enzymes from a streptomycin producing strain. S. glebosus cannot carry out the above conversion since it apparently lacks at least two enzymes which occur in streptomycin producers: guanidinodeoxy-(scyllo)-inositol dehydrogenase and L-alanine : lD-1-guanidino-3-keto-1-deoxy-scyllo-inositol aminotransferase.
It is sbggested that streptomycin producing strains might be descendents of an ancestral strain which, like S. glebosus, produced the monoguanidinated inositol derivative, bluensomycin.
It is further suggested that gene duplication and subsequent evolutionary divergence resulted in biosynthesis of the diguanidinated inositol derivative, streptomycin, which is 10 times more effective than bluensomycin as an antibiotic and inhibitor of protein biosynthesis.
Evolutionary mechanisms for the acquisition of novel biosynthetic capabilities remain among the important unsolved problems in biology.
It is particularly difficult to understand how biosynthetic pathways involving 20 or more specific enzymatic reactions could have arisen when neither intermediates nor the final end products appear to be required for growth.
Such compounds might be termed idiolites, since they are synthesized during the idiophase of the growth cycle, and have a restricted biological distribution.
Our laboratory has been st,udying one such biosynthetic pathway, the biosynthesis of dihydrostreptomycill, an aminocyclitol antibiotic (I, Fig. 1) secreted by certain strains of filamentous soil bacteria of the genus Streptomyces.
Mild acid hydrolysis of these compounds gives streptidine or bluensidine plus the disaccharide, dihydrostreptobiosamine.
Bluensidine has also been called glebidine (4). None of the components of these antibiotics has SO far been found elsewhere in nature.
and H), transamination (Reactions 11 and I), phosphorylation (Reactions E and J), transamidination (Reactions F and K), and dephosphorylation (Reactions G and N). For certain pairs of corresponding reactions in the two sequences it is known that different enzymes are involved, although some have overlapping substrate specificities (7-g).
An attractive hypothesis for later stages in the evolution of the streptidine biosynthetic pathway is that a segment of chromosomal or episomal DNA containing genes involved in one sequence of five reactions was duplicated in an ancestral strain, followed by independent evolution of the respective genes. The question then arises whether descendents of the hypothetical ancestral strain can be found which have not undergone duplicat,ion in this pathway, and which therefore catalyze only one sequence of five reactions.
One of the purposes of the present investigation was to determine the general enzymatic pattern of biosynthesis of the monoguanidinated inositol (bluensidine) moiety of bluensomycin, and compare that pattern with the one proposed for biosynthesis of the diguanidinatcd inositol (streptidine) moiety of dihydrostreptomycin.
Our findings are compatible with the general pattern of bluensidine biosynthesis suggested in Fig. 2 The experiments which have been performed on biosynthesis of the bluensidine moiety of bluensomycin (II, Fig. 1) can be most conveniently described by referring to the reaction schemes of Fig The corresponding blutnsidine moiety of bluensomycin contains one guanidino group and a carbamoyl group instead of two guanidino groups (2,4) ; the location of these two groups was recently established by Barlow and Anderson (5). Fig. 2 also indicates one possible pathway for biosynthesis of the bluensidine moiety; in this scheme the single guanidino group is synthesized by the same enzymes which participate in biosynthesis of the first guanidino group introduced on the inositol ring of dihydrostreptomycin.
How-ever, this guanidino group ends up at position 3 of bluensomycin, rather than position 1 as in the case of dihydrostreptomycin biosynthesis. An alternative scheme (not shown) for bluensidine biosynthesis was also COW sidered.
In the latter scheme, the single guanidino group is synthesized by the same enzymes which participate in biosynthesis of the second guanidino group introduced on the inositol ring of Participation is different from enzyme N. S. glebosus extracts catalyze both of epimers of certain intermediates has not been excluded. Ex-Reactions F and K. Note that in these schemes, corresponding tracts of S. glebosus catalyze Reactions C, D, F, and G. Reaction carbon atoms of the inositol moieties of dihydrostreptomycin (I) E can be inferred from the presence of its product in mycelial and bluensomycin (II) are derived from different carbon atoms of extracts, and slight activity was detected in vitro. The step at glucose-6-P (III).
Abbreviations : DSBA, dihydrostreptobioswhich the carbamoyl group is added has not yet been established.
scheme, enzymes H through K are presumed to interact equally well with inositol derivatives containing a guanidino or a carbamoyl group at the indicated position, just as enzymes L and M are presumed to do in both schemes. The following experiments were designed to determine the general pattern of biosynthesis of the bluensidine moiety of bluensomycin by extracts of S. glebosus. (14) is dephosphorylated to keto-scyllo-inositol (VI) during normal biosynthesis from glucose.
Evidence for Conversion of myo-Inositol (V) to Keto-scylloinositol (VI) by S. glebosus E&acts-Horner andco-workers (la), 13) have presented evidence for the interconversion of myo-inositol, keto-scyllo-inositol, and scyllo-inositol in intact mycelia of streptomycin producers, and we have demonstrated Reaction D in cell-free preparations (7). However, dehydrogenation of myoinositol (Reaction C) by cell-free preparations of Streptomyces has not previously been reported.
In the past we have occasionally observed a very low level of conversion of myo-[WIinositol to aminodeoxg-scyllo-[Wlinositol catalyzed by extracts of streptomycin producers, but this conversion could not be enhanced by the addition of cofactors or amino donors, so the data were not reported.
The possibility therefore existed that Reaction C did not occur with free myo-inositol, but with myoinositol in covalent linkage e.g., in phosphatidylinositol.
Al NADP+ was not nearly so active as NAD+ as hydrogen acceptor in this system. The most active amino donors tested were L-glutamine, aminodeoxy-scyllo-inositol, and streptamine, all of which are known to be able to serve as amino donors in Reaction D of streptomycin producers (7,9). L-Alanine was relatively inactive as an amino donor, an indication that enzyme I was not involved.
The occurrence of Reaction C in S. glebosus was confirmed by employing another version of the coupled dehydrogenase-transaminase assay system described above.
The results are shown in Fig. 3. Both myo-inositol and NdD+ were required for conversion of labeled aminodeosyscgllo-inositol to labeled keto-scyllo-inositol by the sum of Reactions 1 and 2.
Separate Assay for Reaction D in S. glebosus-The occurrence of Reaction D in S. glebosus has already been indicated by the previously described coupled reactions.
We have suggested (7) that Reaction D is the sum of Reactions 3 and 4. This transaminase can therefore be assayed by the partial or half-reaction L-Glutamine + pyridoxal-P-enzyme + ol-ketoglutaramate (3) + pyridoxamine-P-enzyme Pyridoxamine-P-enzyme + keto-scyllo-inositol = (4) pyridoxal-P-enzyme + aminodeoxy-scyllo-inositol depicted in Reaction 4. Operationally this assay is carried out as indicated in Reaction 2. The results of such an experiment are given in Fig. 4. A role for L-glutamine rather than L-alanine as the physiological amino donor was suggested by the data of Table  I. These results confirm the presence of Reaction D in S. glebosus. As in the case of st.reptomycin producers, high concentrations of pyruvate can serve as amino acceptor in the reverse reaction (7).  Fig. 3, except that 5 ~1 of 28 mM nonlabeled keto-scylloinositol replaced myo-inositol plus NAD+ in the complete incubation mixture.
This experiment confirmed the Reaction 3 component of the coupled reactions employed in Table I and Fig. 3.
As expected, amidinotransferase activity of harvested mycelia varied markedly with the age of culture (cf. 19), composition of growth medium, pH, temperature, and state of inoculum.
The next question concerned the amidino acceptor specificity of S. glebosus amidinotransferase.
The L-arginine : inosamine-P amidinotransferase occurring in streptomycin producing strains of Streptomyces has a substrate specificity which can be depicted in part as shown in Fig. 5 (18). C er t ain of these substrates were tested with S. glebosus extracts, with the results shown in Fig. 6. The following inosamine derivatives, prepared by nonspecific chemical phosphorylation (11) , were found to serve as amidino acceptors with L-[gUanni&o-i%]arginine as donor: Compound XVa (Fig. 6A); Compound XVb (Fig. 6B); Compound XVc (Fig. 6C) ; and the 2-deoxy derivative of Compound XVc (Fig.  6D). Fig. 6A corresponds to Reaction F of Fig. 2 Detection of Physiological Amidino Acceptors in S. glebosus-A search for physiological amidino acceptors (X-NH2) in extracts of S. glebosus was next undertaken, employing Reaction 8 as an assay.

L-[guanidino-14C]Arginine
In these experiments the supernatant solutions from sonicated mycelia of S. glebosus were used as a source of both amidino acceptors and amidinotransferase activity. When such extracts were incubated with labeled arginine as amidino donor, a single peak containing radioactive products was obtained after paper chro- from streptomycin producing strains (9, 18). It is likely that R can also be -0-C(=O)NH2. matographic separation, as shown in Fig. 7. In contrast, it will be recalled that two distinct peaks were obtained with extracts of streptomycin producing strains, corresponding to Compounds IX and XIV of Fig. 2 (11). The presence of myo-inositol in the growth medium increased the concentrations of amidino acceptors in S. glebosus mycelia.
Furthermore, when the radiochemical enzymatic assay for amidino acceptors (Reaction 8) was con- After incubation at 35" for 130 min, 10 ~1 were spotted and separated on ammoniacal phenol paper chromatograms.
The amidinated reaction products produced in a scaled up incubation mixture could be separated from other labeled compounds in a single peak on a Dowex 50(H+) column, as shown in Fig. 84. The isolated radioactive compounds migrated similarly to Compound IX during paper chromatography with a nurnber of solvents and during high voltage paper electrophoresis at pH 3.6 and pH 10.4. These compounds could be dephosphorylated by incubation with Escherichia coli alkaline phosphatase to give compounds with mobilities similar to Compound X. Evidently the labeled peak contained compounds possessing one guanidino group and one phosphate ester group.
The occurrence of enzyme E in S. glebosus could be inferred from the fact that the amidino acceptors are phosphorylated.
Weak enzyme E activity was occasionally detected in S. glebosus extracts, but enzyme J was not detected in the same extracts.
A hint that the labeled peak of Fig. 8A contained a compound, XGP, different from Compound IX came from its slightly higher RF value on ammoniacal phenol paper chromatograms; its dephosphorylated derivative also had a slightly higher R, than Compound X. The question then arose whether the postulated XGP were a carbamoylated derivative of Compound IX. Such a compound would be difficult to distinguish from Compound IX by the methods described above, but it might react differently from Compound IX when incubated with extracts containing enzymes G, II, and I of Fig. 2.
Presence of Reaction G Activity in S. giebosus-Contrary to our expectations (9), extracts of S. glebosus actively dephosphorylated Compound IX, as shown in Fig. 9A. However, neither extracts of X. glebosus nor S. bikiniensis mycclia were able to dephosphorylate all of the components present in the isolated peak of Fig. 8A, as shown in Fig. 9B for S. glebosus. The isolated peak of Fig. 8A apparently contained Compound 1X, which could be dephosphorylated by enzyme G, plus a component (XGP) of a closely related structure, possibly bluensidine-6-I', which is not a good substrate for enzyme G (Fig. 9B). The radioactive preparation of Fig. 8A was enriched in t)he XGP component by incubation with extracts of S. glebosus or S. bikiniensis to dephosphorylate the Compound IX component, followed by chromatography on a Dowex-50 column as before, as shown in Fig. 8B. The resulting XGP component was resistant to dephosphorylation by enzyme G, as shown in Fig. 9C. Alternatively, enrichment of the XGP component could be obtained by performing the initial transamidination reaction with S. glebosus and labeled arginine in the presence of Mg2+ and absence of EDTA to allow any Compound IX formed to be dephosphorylated by enzyme G present in the same extract.
Further Characterization of Unknown Physiological Transamidination Product (XGP)-When the labeled preparation (XGP) of Fig. 8B was dephosphorylated with E. coli alkaline phosphatase, the resulting compound (XG) could not serve as a substrate for enzymes H plus I from a streptomycin producer, as shown in Fig. IOA. These results demonstrated that XG is not Compound X. However, hydrolysis of XGP with 6 s HCl at 100" for 28 hours gave a compound which did serve as a substrate for enzymes 11 plus I, as shown in Fig. 10B. This treatment with 6 N HCl removed the phosphate group and would hydrolyze any carbamoyl ester present (4). The reaction product (IGN) of Fig. 10B appears to be Compound XII.
These results are consistent with XGl' being bluensidine-6-P, but this has not yet been rigorously established, of course.
At the outset of this investigation, little was known of the enzymatic steps involved in biosynthesis of bluensomycin, although suggestions had been made on the basis of preliminary findings (11). Our experimental results are consistent with the general scheme show-n in Fig. 2 for biosynthesis of the monoguanidinated iriositol (bluensidine) moiety of bluensomycin (II). of lD-l-guanidino-3-amino-l,3-dideoxy-scyllo-inositol (IGN) from XG previously hydrolyzed with 6 N IICl.
The step at which carbamoylation occurs is not yet knolvn. ISxtracts of S. glebosus have been shown in this paper to catalyze Reactions C, D, F, and G; these reactions are apparently catalyzed by enzymes similar to those involved in biosynthesis of the first guanidino group of the diguanidinated inositol (streptidine) moiety of dihydrostreptomycin (I), as depicted in Fig. 2. Since enzymes catalyzing Reactions II and I were not detected in S. glebosus, a possible alternate pathway for biosynthesis of the bluensidine moiety of bluensomycin, involving participation of enzymes utilized in biosynthesis of the second guanidino group of the streptidine moiety of dihydrostreptomycin, appears to be ruled out. The latter scheme required that the substrate spccificities of these enzymes, e.g., enzyme I, could be satisfied by either a guanidino or carbamoyl group at position 1. It is important to note that both of the above schemes for biosynthesis of the bluensidine moiety of bluensomycin predict a different labeling pattern in the end product, starting from specifically-labeled glucose-6-P (III), from that observed for biosynthesis of the streptidine moiety of streptomycin. III studies in z&o of streptomycin biosynthesis, Bruce et al. (20) found that position 5 of streptomycin was derived from C-l of glucose as shown in Fig. 2. In our scheme, position 5 of bluensomycin would be derived from C-3 of glucose.
The biosynthetic scheme of Fig. 2 is consistent with the recent stereochemical assignments for the bluensidine rnoiety by Barlow and Anderson (5). Their assignments, arrived at by physicochemical means (5), are compatible with the substrate spccificities of (a) enzyme F, assuming that R can also be a carbamoyl ester (Fig. 5); and (6) enzymes L and M, assuming that their substrates must be phosphorylated and at the position indicated in Fig. 2.
Although myo-inositol dehydrogenases have been studied in a number of organisms (15,21,22), the enzymatic experiments described in Table I and Fig. 3 have provided the first evidence for Reaction C obtained in cell-free extracts of Streptomyces. It was necessary to couple Reaction C with Reaction D to demon-strate this dehytlrogellation.
It is still not known why this particular myoinositol dehydrogenation reaction has been so difficult to demom stratc in vitro in Slre$omyces which produce the streptomycin family of antibiotic&s.
myo-Inositol dehydrogenascs from other sources can readily br assayed in the reverse direction i.e., reduction of keto-sc?lZ2o-illositol by NAI)II (15,21), but this reverse assay has not so far proved useful with Streptonryces extracts. Further studies will be necessary to determine whcthcr this myoinositol dchytlrogenasc, as an early enzyme in a biosyntlictic pathway, has an important regulatory function; regulatory enzymes often prcscnt assay or stability problems.
The details of its linkage with elec%ron transport enzymes shoul(1 also prove of interest. Although the stage in bluensomycin biosynthesis at which carbamoylatiou occurs is not yet known, despite our efforts in that direction, the unknown transamidination product (XGP) obtained on incubation of S. glebosus extracts with labeled arginine (Figs. 7 to 10) has many of the properties expected for bluensidine-6-P.
The substrate specificity of enzyme F (Fig. 5) would probably permit a carbamoyl group at the indicated position. Any intermediates which escaped carbamoylation would probably be enzymatically convert4 to Compound X, in view of the prescncc of enzyme G activity in S. glebosus (Fig. 9). No kinase activity with Compound X as acceptor has htcn detected in S. glebosus or S. bikiniensis.
The mechanism of the marked crrhancemcnt of transamidination reactions by carbamoyl-P observed in vitro with nondialyzcd mycelial cstracts (Fig. 7) remains unknown. IWiancement might result from: (a) carbamoylation of an inosaminc-1' derivative to give a substrate with a lower K, or higher I',,,,, with amidinotransferasc; (b) carbamoylation of orrrithine, a strong inhibitor of amidinotransferase, to form noninhibitory citrulline; or (c) an allosteric activation of amidiriotransferase. This phenomenon will be csamined further.
It is too early in our investigation to draw dcfiriitive conclusions concerning the evolutionary relationships between the respective enzymatic pathways for biosynthesis of bluensomycin and dihydrostrcptomycin.
IIowever, two possibilities will be briefly considered as a framework for future experiments.
1. One possibility is that blucnsomycin producers resemble an ancestral strain, or contain an ancestral cpisome, which has not untlergonc gene duplication in the guariidirloc:yc:litol biosynthetic pathway.
1Sluensomycin producers can add only one guanidino group to the inositol ring, because they lack gents coding for enzymes II and I. If it turns out that enzymes 11; and ,J arc coded by different genes, gent J would be missing. l<luensomytin producers presumably require an additional plc which codes for a carbamoylation enzyme. In this scenario, dihydrostreptomycin producing strains arc desccndcnts of the above ancestral strain, or contain episomes, which have undergone duplication and subsequent independent mutation of genes coding for Reactions C and D, and possibly E, F, and G. Tha nature of any selection pressure is not known since the physiologic*al functions of these idiolites have not been established (9). However, dihydrostrcptomycin is approximately 10 times more effective than bluensomycin as an antibiotic and inhibitor of protein synthesis (6), and therefore might represent a later evolutionary product.
2. Another possibility is that bluensomycin producers are derived from dihydrostrcptomycin producers.
In this scenario, addition of a gene coding for a carbamoylation enzyme resulted in synthesis of bluensidine-6-I', which cannot be dephosphorylated by enzyme G. Bluensidine-6-P reacts in the presence of enzyme L and subsequent enzymes to form bluensomycin rather than dihydrostreptomycin.
There would be no further need for enzymes H through K, since their substrates would no longer be formed.
Genes coding for these latter enzymes could then be lost, further mutated to serve new functions, or be repressed. Again, the selection pressures are not known.