Physiological implications of the substrate specificities of acetohydroxy acid synthases from varied organisms

Acetohydroxy acid synthase (AHAS; EC 4.1.3.18) catalyzes the following two parallel, physiologically important reactions: condensation of two molecules of pyruvate to form acetolactate (AL), in the pathway to valine and leucine, and condensation of pyruvate plus 2-ketobutyrate to form acetohydroxybutyrate (AHB), in the pathway to isoleucine. We have determined the specificity ratio R with regard to these two reactions (where VAHB and VAL are rates of formation of the respective products) as follows: VAHB/VAL = R [2-ketobutyrate]/[pyruvate] for 14 enzymes from 10 procaryotic and eucaryotic organisms. Each organism considered has at least one AHAS of R greater than 20, and some appear to contain but a single biosynthetic AHAS. The implications of this for the design of the pathway are discussed. The selective pressure for high specificity for 2-ketobutyrate versus pyruvate implies that the 2-ketobutyrate concentration is much lower than the pyruvate concentration in all these organisms. It seems important for 2-ketobutyrate levels to be relatively low to avoid a variety of metabolic interferences. These results also reinforce the conclusion that biosynthetic AHAS isozymes of low R (1 to 2) are a special adaptation for heterotrophic growth on certain poor carbon sources. Two catabolic "pH 6 AL-synthesizing enzymes" are shown to be highly specific for AL formation only (R less than 0.1).

In the pathway for the biosynthesis of branched-chain amino acids, the same set of enzymes catalyzes four consecutive reactions leading to two different sets of products, valine and leucine, on the one hand, and isoleucine on the other (40,41) (Fig. 1). Acetohydroxy acid synthase (AHAS, also known as acetolactate synthase; EC 4.1.3.18) catalyzes the first of these reactions, the irreversible decarboxylation of pyruvate and its condensation with either pyruvate or 2-ketobutyrate (15). The relative amounts of the two possible products formed by AHAS thus determines the relative amounts of valine plus leucine and isoleucine synthesized in these pathways.
We have shown for a number of AHASs (3,15) that the ratio of the relative rates of formation, V, of acetohydroxybutyrate (AHB) and acetolactate (AL) by a given enzyme is proportional to the ratio of the concentrations of the substrates 2-ketobutyrate and pyruvate and to a specificity ratio, R (1) Such a fixed specificity is expected if the competition between 2-ketobutyrate and pyruvate occurs on the enzyme subsequent to the irreversible and rate-determining formation of an intermediate from the first pyruvate (15).
Pyruvate is a major intermediate in metabolism, whereas 2-ketobutyrate is a minor one, serving mainly as a precursor to isoleucine. Furthermore, 2-ketobutyrate may be toxic at high intracellular concentrations (7,24). In Escherichia coli grown on glucose, the concentration of pyruvate is nearly 2 orders of magnitude higher than that of 2-ketobutyrate (7,25). In order to synthesize similar quantities of each of the branched-chain amino acids under such circumstances, an * Corresponding author. organism must have at least one AHAS with a high specificity for 2-ketobutyrate, i.e., with an R of 20 to 100. E. coli K-12 and Salmonella typhimurium LT2 each express such isozymes, AHAS III (R = 40) and AHAS II (R = 65), respectively (15). In addition, these bacteria express isozyme AHAS I (R = 2), which probably serves a special function when pyruvate levels are particularly low (3,5,6). We would predict that, if low 2-ketobutyrate-to-pyruvate ratios are a general phenomenon, every organism which normally synthesizes branched-chain amino acids will express an AHAS of high specificity for 2-ketobutyrate.
In this paper, we examine the above prediction, by determining the partitioning ratio R for AHAS enzymes from a number of procaryotic and eucaryotic organisms, both heterotrophic and autotrophic. In addition, we also consider the substrate specificity of two enzymes that are believed to have purely catabolic roles and which have been called "pH 6 AL-synthesizing enzymes" (pH6 ALS) (17,19,35).

MATERIALS AND METHODS
Microorganisms and plasmids. The microorganisms and plasmids used in this study and their sources are listed in Table 1.
Media. The minimal medium for E. coli strains was that described by Vogel and Bonner (44) supplemented with the required amino acids (50 pug/ml each). LB was used as the rich medium. Tetracycline (15 ,ug/ml) and ampicillin (50 pug/ml) were added to the rich medium, when required.
Saccharomyces cerevisiae was grown in the medium described by Zimmerman (48). Porphyridium sp. was grown in artificial seawater (22), and Chlorella emersonii was grown in medium N8 (33). Cloning and transformation procedures. Cleavage with restriction enzymes and ligations were carried out in accordance with the recommendations of the suppliers. The preparation and isolation of DNA, restriction mapping, and transformations with plasmids were carried out by the methods of Maniatis et al. (26). Transformants were selected for the relevant antibiotic resistance and for the Ilv+ phenotype. CU9090(pDK6) and MM294 transformants were selected for antibiotic resistance only.
The pNG047 plasmid was constructed in the course of attempts to clone the gene for the Klebsiella aerogenes pH6 ALS (17,35). Chromosomal DNA from K. aerogenes AA-1 was subjected to partial cleavage by Sau3AI, and DNA fragments 4 to 6 kilobases in length were isolated by agarose gel electrophoresis with the Geneclean kit (45) according to the recommendations of the manufacturer (Bio 101 Inc., La Jolla, Calif.). These fragments were ligated into pBR322 cleaved with BamHI, and MM294 was transformed with the resulting plasmids. Colonies transformed with plasmids that contained inserted fragments were identified as Ampr Tets. Such clones were then screened for a positive Voges-Proskauer test (acetoin production) (8). Plasmid pNG047 was isolated from one of the positive clones. MF2000 was transformed with this plasmid.
Enzymes. The purified isozymes AHAS II from S. typhimurium (32) and AHAS I from E. coli (A. Aulabaugh and J. V. Schloss, unpublished results) were gifts from J. V. Schloss. AHAS III was purified as described previously (2). Crude enzyines were prepared from exponentially growing cultures of E. coli in a minimal medium containing glucose and harvested at a turbidity of about 50 Klett units (filter no. 66). Cells were washed and concentrated to approximately 1010 bacteria per ml in a disruption buffer (0.1 M potassium phosphate buffer, pH 7.6, containing 10 ,ug of flavin adenine dinucleotide (FAD) per ml, 0.5 mM dithiothreitol, and 10 mM EDTA) and lysed by ultrasonic disruption as previously described (14). Unbroken cells and debris were removed by centrifugation, and glycerol was added to the crude extract to 20% (vol/vol). The AHAS activities of similar extracts of host strains MF2000 and CU9090 were negligible under conditions which yield high activity for each of the cloned enzymes. The pH6 ALS from K. aerogenes was prepared from a culture of K. aerogenes AA-1 grown to late stationary phase in minimal medium (36). Crude extracts containing AHAS were prepared from cultures of Porphyridium sp. (43) and C. emersonii [D. Landstein, D. M. Chipman, S. (Malis) Arad, and Z. Barak, unpublished data] as previously described.
Crude yeast AHAS was prepared from exponentially growing cultures (30°C, with vigorous mixing and aeration) of S. cerevisiae YT669. Washed cells were suspended in disruption buffer (see above) and were broken by ultrasonic 172,1990 Organism and strain S. C. Falco (10) University of Texas, Austin Cambridge Culture Collection, Cambridge, United Kingdom disruption (six successive 1-min treatments at 4°C with 1-min intervals between them). Debris was spun down, and glycerol was added to the supematant to a final concentration of 33%. Extracts were stored at -70°C. Enzyme assay. AHAS activity was measured either as AL formation in the presence of pyruvate by the standard colorimetric method (4) or by the gas chromatographic method, which determines the formation of both products (AL and AHB) simultaneously (12,13). Unless otherwise stated, enzyme assays were carried out as previously described in reaction mixtures containing 10 mM MgCl2, 20 ,ug of FAD per ml, and 30 ,ig of thiamine PPi per ml (15). For the pH6 ALS from K. aerogenes and Bacillus subtilis, the reaction was carried out at pH 6 and pH 6.2, respectively (0.1 M phosphate or morpholineethanesulphonate [MES] buffer and 50 mM sodium acetate) in the absence of FAD.

RESULTS
The substrate specificity of AHAS activities from a variety of sources was examined by determining the amount of the two products (AL and AHB) formed in the presence of the two substrates (pyruvate and 2-ketobutyrate) by the gas chromatographic method (12,13,15) (Table 2). The first three lines in Table 2 are summaries of data obtained with purified enzymes from enteric bacteria (15). The remaining data were obtained by using crude extracts of the indicated strains. Dependence on 2-ketobutyrate concentration of the rates of formation of AL (0) and AHB (O) by cloned K. aerogenes AHAS II-like enzyme. A crude extract of E. coli K-12 strain MF2000(pPU137), which encodes ilvGM from K. aerogenes (18), was prepared as described in Materials and Methods. The reaction was carried out in the presence of 20 mM pyruvate and different concentrations of 2-ketobutyrate, at pH 7.6 in 0.1 M phosphate buffer containing FAD, thiamine pyrophosphate, and Mg2+, as described in Materials and Methods. The reaction was stopped with phosphoric acid, and the mixture was analyzed for the diketone derivatives of AL and AHB by the gas chromatographic method (12,13,15). The normalized results of two separate experiments are shown at the bottom. The curves are the theoretical fit of the data to the kinetic equations for AHAS (15). Shown above is the specificity ratio R for each experimental point, with the horizontal line at R = 79.5, the value calculated from the kinetic fit (15).
For example, data was obtained with extracts of E. coli carrying multicopy plasmids encoding the valine-resistant AHAS from the ilvGMEDA operons of the enteric bacteria Serratia marcescens, K. aerogenes, and Edwardsiella tarda (18) (Table 2, lines 4 to 6, respectively). As the host strain was devoid of any ALor AHB-synthesizing activity, it is reasonable to assume in each case that the reaction measured was due to the enzyme encoded by the plasmid. Figure  2 shows an example of the results obtained with an extract from MF2000(pPU137), which expresses the AHAS II-like enzyme of K. aerogenes. It can be seen that R is relatively constant over a range of 2-ketobutyrate concentrations, as previously demonstrated for purified enzymes (15). In each of the cases reported in Table 2, the R value of a given enzyme (equation 1) was practically constant over a wide range of substrate concentrations, as expected for a single enzyme. On the other hand, if two enzymes of different specificities contributed significantly to product formation, one would expect R to decrease systematically with increasing 2-ketobutyrate concentration.
The data which appear in Table 2 on lines 7, 9, and 10 were also derived by using extracts of E. coli carrying a single AHAS on a multicopy plasmid and are characteristic of the single enzyme in question by the above criteria. The enzyme from K. aerogenes that is encoded by plasmid pNG047 (line 7) has properties that are similar, but not identical, to those of AHAS III of E. coli. These properties include pH optimum, Ki,, valine sensitivity (data not shown), and R ( Table   2, line 7 compared with line 2), as well as similar restriction maps (34). The pNG047-encoded enzyme is different from the valine-insensitive K. aerogenes ilvGM product encoded by plasmid pPU137 (18) and from the pH6 ALS (17,35) of this bacterium (see below). The presence of both valinesensitive and -insensitive biosynthetic AHASs in K. aerogenes has been previously suggested by Asada et al. (1).
The yeast enzyme studied (Table 2, line 12) was also encoded by a plasmid bearing the sole AHAS of S. cerevisiae in a host yeast strain free of background AHAS activity (10).
The pH6 ALS activity of K. aerogenes ( Table 2, line 8) was studied in extracts of bacteria grown to the stationary phase to ensure maximal synthesis of this enzyme (17,36).
To minimize the contribution of residual biosynthetic AHAS enzymes to product formation, the reaction was carried out at pH 6 in the absence of FAD (37).
In contrast to the above cases, the AHAS activities in crude extracts from the two algae, Porphyridium sp. and C. emersonii ( Table 2, lines 13 and 14), could in principle be due to more than a single enzyme. However, as the R values were also found to be essentially invariant with substrate concentration, the AHAS activities must be overwhelmingly due to a single enzyme or to two enzymes with similar R values. Other properties of the AHAS activities in these organisms support the idea that each has a single AHAS (43; Landstein et al., unpublished data; D. van Moppes, unpublished data).
The R value for the AHAS activity of Corynebacterium glutamicum (line 9) is based on the data reported by Eggeling et al. (9). These workers have shown that C. glutamicum has a single AHAS, and the calculated constant R over a range of substrate ratios is consistent with this.

DISCUSSION
On the basis of their R values, the enzymes listed in Table  2 can be separated into three major groups: (i) those with a strong preference for reaction with 2-ketobutyrate to form AHB (R = 20 to 80); (ii) those with a preference for reaction with pyruvate to form AL (R < 0.1); and (iii) AHAS I of enteric bacteria with almost equal preference for the two substrates (R = 2). In accord with our expectations, each of the organisms we have considered contains at least one AHAS of group one, with a high specificity for 2-ketobutyrate, i.e., R > 20. The same principle seems also to hold for the bacteria Mycobacterium pellegrino, Streptomyces rimosus, and Pseudomonas aeruginosa. Szentirmai and Horvath (38) have measured the rate of AHB formation as a function of the 2-ketobutyrate concentration in crude extracts of these bacteria, and despite the limitation of their method and the absence of information concerning the multiplicity of AHAS isozymes in these organisms, it is clear from an analysis of their data (not shown) that each of these organisms also has at least one AHAS with an R of at least 10.
Many of the organisms we studied express only a single AHAS during active growth. The bacteria B. subtilis (42,47) and C. glutamicum (9), the yeast S. cerevisiae (10), and in all probability, the two algae considered here each seems to have only a single biosynthetic AHAS, with an R value between 20 and 40 ( Table 2). In each of these cases, the single AHAS is also inhibited by valine (data not shown) and is thus similar in its properties to AHAS isozyme III of E. coli.
The above analysis suggests that low concentrations of 2-ketobutyrate relative to pyruvate are indeed the rule. It would seem to be advantageous to almost any prototrophic cell for 2-ketobutyrate levels to be relatively low, as a wide variety of metabolic complications are possible if 2-ketobutyrate levels rise too high (23,24,30). For example, 2ketobutyrate can compete with 2-ketoisovalerate, which differs from it only by an additional methyl group, in the reactions catalyzed by isopropylmalate synthase and ketopantoate hydroxymethyltransferase (on the pathways to leucine and coenzyme A, respectively; Fig. 1). 2-Ketobutyrate can also be converted by transaminases to aminobutyric acid, which might compete with valine in protein synthesis. Studies by LaRossa and others (7,23,24,30) have made it clear that 2-ketobutyrate can be toxic in bacteria.
Pyruvate, on the other hand, is a central metabolite under a wide variety of metabolic regimes, including heterotrophic growth with carbohydrates as carbon sources, as well as photosynthetic growth. Its concentration would also be expected to be high and relatively constant in mitochondria or chloroplasts, in which the enzymes of the branched-chain amino acid biosynthetic pathway are located in eucaryotic cells (21,27,29,31,39). AHASs have apparently evolved with high specificity to 2-ketobutyrate to provide comparable quantities of isoleucine and valine under these conditions of high pyruvate and low 2-ketobutyrate concentrations.
AHAS I, with a low preference for 2-ketobutyrate (R = 2), is an additional isozyme necessary for the adaptation of E. coli or S. typhimurium to growth on acetate or palmitate (3,5,6), poor carbon sources that lead to low endogenous pyruvate concentrations (25). One might expect that there will be other organisms (particularly other enterobacteria) which also express such an enzyme. However, a low internal pyruvate concentration is rather a special situation, with which many organisms do not seem to have to cope.
Finally, some organisms, such as B. subtilis and K. aerogenes, have additional enzymes that clearly belong to a separate class of pH6 ALS on the basis of many properties. These specialized enzymes do not use FAD as a cofactor (19,37), as do the biosynthetic AHASs. They are strictly catabolic enzymes, not expressed in exponentially growing cultures and induced, probably together with acetolactate decarboxylase, during the stationary phase of growth (17,20,36,46,47). Their pH optima are low (6.0 to 6.2), and they are thus well suited to function in a fermentation pathway leading to acetoin and butanediol (19,20,47) (Fig. 1). Presumably, this fermentation pathway prevents further acidification of the growth medium (20). The strong tendency of these enzymes towards AL synthesis (R < 0.1) is clearly appropriate to their physiological role.
We believe that an AHAS with a high preference for condensation of the active acetaldehyde moiety with 2ketobutyrate, rather than with pyruvate, is a major factor in the design of the biosynthetic pathway for the branchedchain amino acids. This preference presumably allows an organism to maintain 2-ketobutyrate at low concentrations. The ability of any enzyme to select one substrate in the presence of a competing substrate that is smaller by a single methyl group has physical-limits (about 3 to 3.5 kcal/mol [12 to 14 kJ/mol]) (11). The selective pressure to maintain low 2ketobutyrate concentrations seems to have been strong enough to push AHAS almost to these limits of selectivity (11,15) in many cases. An alternative to an AHAS of high specificity would be the compartmentalization of 2-ketobutyrate, perhaps by its direct, channeled transfer from the enzyme that produces it (threonine deaminase) to the active site of an AHAS. If a prototrophic organism did not adopt one of these two strategies, it would have to deal with the limitations of enzyme specificity at another stage. If it were to maintain free 2-ketobutyrate levels close to those of pyruvate, it would require versions of all of the potentially sensitive enzymes (e.g., isopropylmalate synthase) with an extraordinary ability to select against 2-ketobutyrate or its undesirable metabolic products. On the other hand, if it were simply to produce an excess of valine precursors over isoleucine precursors, it would have to expend considerable metabolic energy on editing of mischarged aminoacyl tRNAs or on specific excretion of these compounds from cells. We thus suspect that all prototrophic organisms will either have a biosynthetic AHAS of high specificity for 2-ketobutyrate or will have some mechanism for channeling 2-ketobutyrate specifically to an AHAS.