The Role of the TRPl Gene in Yeast Tryptophan Biosynthesis*

Transcription of the gene for phosphoribosyl-an-thranilate isomerase (TRPl) from the TRPl promoter is initiated only approximately half as frequently as, for example, from the TRP3 promoter, but TRPl mRNA is approximately twice as stable as TRP3 mRNA. Therefore, the steady state amount of TRPl mRNA in yeast cells, grown without amino acid limitation, is similar to the steady-state amount of TRP3 mRNA. The protein concentration of both enzymes in yeast cells is about the same, but the basal specific enzyme activity in permeabilized cells of the TRPl gene product N-(5‘-phosphoribosyl-l)-anthranilate isomerase is about 2-3 times higher than that of any of the other TRP enzymes. According to the kinetic parameters of the purified isomerase protein, the enzyme is more active than, for the purified TRP3 enzyme indoleglycerol-phosphate synthase. It is the TRPl of Saccharomyces cerevisiae of a rearrangement

The de novo biosynthesis of tryptophan in all prokaryotic and eukaryotic organisms studied so far proceeds through an invariable series of reactions. A set of seven enzymatic activities is necessary to perform the five biosynthetic steps from chorismic acid to tryptophan (Crawford, 1975). In bacteria, the genes that encode the tryptophan biosynthetic enzymes are combined in clusters of one, two, or three transcriptional units on the chromosome (Crawford, 1975). In contrast, in all eukaryotic microorganisms studied so far, the TRP genes are scattered over the genome. On the other hand, the encoded enzymes appear to be more highly organized in eukaryotic than in prokaryotic organisms, resulting in multifunctional proteins. Different patterns of fusion have produced multifunctional enzymes with different combinations of functional domains (Hutter et al., 1986).
In most ascomycetes, four genes encode the seven functional domains of the tryptophan pathway. One of these genes codes for a trifunctional polypeptide, NH2-glutamine amido-transferase-InGP synthase'-PRA isomerase-COOH. Several genes with this arrangement have been cloned and characterized further from different ascomycetes such as Neurospora crassa (Schechtman and Yanofsky, 1983), Aspergillus nidulans (Mullaney et al., 1985), Aspergillus niger (Kos et al., 1985), Cochlwbolus heterostrophus (Turgeon et al., 1986), and Penicillium chrysogenurn (Shchez et al., 1986). On the contrary, the yeast Saccharomyces cereuisiae carries five genes for tryptophan biosynthesis instead of four (Braus et al., 1985). No gene was found coding for the trifunctional polypeptide: the two functions glutamine amidotransferase and InGP synthase are encoded by the TRP3 gene (Aebi et al., 1984), whereas PRA isomerase is encoded by the separate TRPl gene (Tschumper and Carbon, 1980).
The TRPl gene is of special interest because a yeast ARS (autonomously replicating) sequence is located adjacent to the 3'-end of the TRPl gene (Beggs, 1978). Therefore, the TRPl gene is used in many yeast vectors as a selectable marker.
The yeast tryptophan pathway is part of the complex general control regulatory network which couples the derepression of at least 30 structural genes (involved in multiple amino acid biosynthetic pathways) to starvation for any one of a number of different amino acids (Schiirch et al., 1974;Hinnebusch, 1986). Four of the five TRP genes of S. cerevisiae can be derepressed under the general control. The TRPl gene is the only exception; its expression is not regulated by amino acid availability (Miozzari et al., 1978a). It has been demonstrated that binding of the GCN4 activator protein to the promoter regions of the genes regulated by general control causes an increased initiation of transcription (Hill et al., 1986;Hope and Struhl, 1985;Arndt and Fink, 1986). The TRPl promoter does not bind the GCN4 activator protein (Hope and Struhl, 1985) and represents an example of a constitutive, weakly expressed promoter of a structural gene (Kim et al., 1986).
The TRPl gene product PRA isomerase catalyzes a practically irreversible Amadori rearrangement, the third step in tryptophan biosynthesis. The basal enzyme activity of the PRA isomerase in permeabilized cells was observed to be higher than that of all the other TRP enzymes (Niederberger et al., 1984).
was compared with that of the normally regulated TRP3 gene. The rate of transcription initiation (promoter strength) of both promoters, the half-life of the mRNAs, and the total amount of transcripts were determined and compared. Furthermore, the TRPl gene product PRA isomerase was purified, characterized, and compared to the purified TRP3 gene product InGP synthase (Prantl et al., 1985).

RESULTS
TRPl and TRP3 Promoter Expression-As a first approach to compare the transcription initiation frequencies of the TRPl and TRP3 promoters, hcZ fusions expressed by both TRP promoters were constructed. @Galactosidase is a convenient enzyme for constructing translational fusions because removal of the first 27 amino acid codons does not affect pgalactosidase activity (Bassford et al., 1978;Guarente, 1983). As basic vectors, the pNM480/1/2 plasmids described by Minton (1984) were used. So that the TRPl and TRP3 fusion proteins would be similar, TRP gene fragments carrying the complete promoter, the start codon plus a similar number of amino-terminal codons, were cloned upstream of the promoterless lacZ gene. The resulting TRPl-lacZ and TRP3-lacZ fusion proteins contained 29 and 30 amino-terminal amino acids of PRA isomerase and glutamine amidotransferase, respectively. Both TRP-lacZ fusions were finally cloned into the yeast low copy number CEN4 vector YCp50 (Johnston and Davis, 1984) in order to prevent significant fluctuations in the copy number. The final plasmids were named pME587 and pME588 (Fig. 1).
The derived clones were identified by their blue color on 5bromo-4-chloro-3-indolyl 8-D-galactoside indicator plates with Escherichia coli lac deletion strain MC1061 (Casadaban et al., 1983 as host. Before transformation in yeast, the nucleotide sequences of the TRP-lacZ linkage regions of the final clones pME587 and pME588 were determined as described under "Experimental Procedures." The sequence confirmed that the gene fusions had preserved the reading frames (data not shown).
The level of p-galactosidase directed by the TRPl-lacZ fusion gene was less than half of the level directed by the nonderepressed TRP3-lacZ fusion. The constitutive regulatory mutation gcd2-1 in the yeast strain RH1310 was used as a control. As for the original genes, this mutation caused derepression of the TRP3-lacZ gene, but had no effect on the TRPl -lacZ fusion.
Comparative Analysis of TRPl and TRP3 Transcripts-The results obtained from TRP-lacZ fusions could reflect effects either on transcription or on translation, or on both. In order to distinguish between these possibilities, the TRPl and TRP3 transcripts were further analyzed.
The half-lives of TRPl and TRP3 mRNAs were determined by quantitative hybridization of yeast RNA labeled in vivo against plasmid DNA. Exponentially growing cultures of the  Table 1s) are presented in miniprint at the end of this paper.
Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are included in the microfilm edition of the Journal that is available from Waverly Press. sequences are boxed with different patterns for TRPl, TRP3, and other yeast sequences. The origin of the plasmids is described under "Experimental Procedures." For pME587, the TRPl promoter fragment (BglII (=-850)-XbaI (+85)) from pMA33 was inserted into the multiple cloning site of pUC18. The promoter was recloned as an SmI-HindIII fragment into pNM481, and a BamHI linker was inserted into the S m I site. After recloning the TRPl promoter as a BamHI-HindIII fragment into pNM482, the TRPl-lac2 fusion was inserted as a BamHI-AsuII fragment into YCp50. For pME588, the TRP3 promoter fragment (EcoRI (pBR322 sequence, including the =900-bp 5'-upstream region of TRP3)-ScaI (+88)) from pME503 was inserted into the multiple cloning site of pNM481. The EcoRI-AsuII fragment was cloned into YCp50. The TRP3 promoter was recloned as a ClaI-Hind111 fragment into pNM482. Finally, the TRP3-lac2 fusion was cloned as an EcoRI-AsuII fragment into YCp50. Ac, AccI; A, AsuII; B, BamHI; Bg, BglII; C, ClaI; H , HindIII; RZ, EcoRI; Sc, ScaI; Sm, SmaI; X, XbaI; AA, amino acids.
a Values of at least three independent cultivations, each measured twice; the standard deviation did not exceed 20%.
-1 frameshift mutations in the TRPl-lac2 and TRP3-lac2 fusions also resulted in a very low, barely detectable &galactosidase activity (data not shown). plasmid-carrying strain RH962 (pME582) with the plasmidencoded complete TRPl and TRP3 genes were radioactively labeled with ['Hluracil for 15 min, followed by a chase with cold uracil. Total RNA was isolated from samples at the time points indicated (Fig. 2) and hybridized to filter-bound, saturating amounts of TRPl and TRP3 DNAs, and the specifically bound radioactivity was determined. Hybridization to filters carrying pBR322 DNA was used as a control for nonspecific hybridization. The half-life of the mRNAs was calculated from the slope of semilog plots.
The half-life of the TRP3 mRNA (length = 1.75 kb) was 11 min. For the smaller TRPl transcript (length = 0.8-1.0 kb), the half-life was 19 min, indicating a higher stability of the TRPl mRNA than that of the larger TRP3 mRNA.
The steady-state amounts of TRPl and TRP3 mRNAs were determined by quantitative Northern hybridization. In order to obtain radioactive probes of the same specific radioactivity and to avoid uncertainties arising from the different lengths of the TRPl and TRP3 transcripts (Fig. 3), two internal fragments of the structural genes of almost the same size were used as probes: the 661-bp HinfI-XbaI fragment of the TRP3 gene (Zalkin et al., 1984) and the 651-bp XbaI-PstI fragment of the TRPl gene (Tschumper and Carbon, 1980). Both fragments were isolated from plasmid pME581 simultaneously, labeled by nick translation, and cohybridized against poly(A)+ RNA bound to a nitrocellulose membrane. The specific transcripts were revealed by autoradiography and cut out from the nitrocellulose filters for quantitbtive determination of radioactivity. Results are shown in Fig. 3. The relative values of the basal mRNA levels of both TRPl and TRP3 genes were within the same range for the chromosomally (strain X2180-1A; lane 1 ) as well as for the plasmid-  Table I ) and the gene product InGP synthase (Fig. 4). This served as a control for equivalent poly(A)+ enrichment of both RNA preparations. As expected, no increase in the amount of TRPl mRNA could be found under the genetic derepression signal.
According to the data presented, the similar total amounts of TRPl and TRP3 transcripts present in yeast cells can be explained in the following way: the lower initiation of transcription of TRPl than that of TRP3 is compensated by the higher stability of the corresponding RNA, resulting in the same total amount of mRNA for the two genes under nonderepressed conditions.
Characteristics of the TRPl Gene Product PRA Isomerase-The data on TRPl promoter and mRNA do not explain the higher specific enzyme activity observed for the TRPI gene product PRA isomerase in frozen and thawed, detergenttreated cells as compared to other TRP gene products (e.g. InGP synthase) (Fig. 4). Assuming equal numbers of PRA isomerase and InGP synthase molecules present in the cells, there are two possible explanations. 1) The high specific PRA isomerase activity in the enzyme assay is a special feature of this particular in vitro assay and does not correlate to the corresponding situation in vivo. 2) The PRA isomerase molecule is a more active enzyme with a higher specific activity than that of the other TRP enzymes.
Enzyme Assays for PRA Isomerase Activity in Permeabilized Celk-Four of the TRP enzymes showed a similar basal specific activity at 30 "C (ranging from 1 to 2 milliunits/mg of protein) and could be derepressed by amino acid limitation to approximately 3 milliunits/mg of protein. The only excep- Cells of S. cerevisiae stains X2180-1A and RH558-1 (gcd2-1) were permeabilized with Triton X-100, and the specific enzyme activity of all chromosomally encoded TRP enzymes was determined. Anthranilate synthase (EG) and anthranilate phosphoribosyltransferase (D) activities were determined by measuring the change in concentration of anthranilate; PRA isomerase (F), InGP synthase (C), and tryptophan synthase (AB) activities were determined by measuring the change in concentration of InGP as described under "Experimental Procedures." The given values are means of two independent cultivations, and each one was measured twice (the standard deviation did not exceed 20%). bThe product CDRP was transformed quantitatively to InGP InGP was measured optically as described under "Experimental Procedures." This stop assay was carried out at 25 "C. The values in parentheses correspond to the values shown in Fig. 4 (30 T). 'The decrease in enzymatically synthesized PRA was measured fluorometrically as described under "Experimental Procedures." This kinetic assay was carried out in a constant temperature cuvette (25 "C) and directly measured. tion was the TRPl gene product PRA isomerase with a nonderepressible, high basal enzyme activity of 4.5 milliunits/ mg of protein at 30 "C ( Fig. 4; 3.0 milliunits/mg of protein at 25 "C; Table 11). In the standard enzyme assay, the highly unstable substrate PRA was prepared nonenzymatically immediately before use from anthranilate and ribose 5-phosphate (Creighton, 1968). The enzyme assay takes advantage of the fact that two relevant reactions are practically irreversible: 1) PRA to CDRP conversion catalyzed by the PRA isomerase itself; and 2) the ring closure to the indole nucleus: CDRP to InGP, CO,, and H 2 0 conversion catalyzed by the TRP3 gene product InGP synthase. By adding enough InGP synthase, the PRA isomerase activity is rate-limiting and can be determined by measuring the increase in the product InGP.
An alternative assay of PRA isomerase activity is the measurement of the decrease of the educt PRA. In the assay developed by Hommel and Kirschner? anthranilate and an excess of 5-phosphoribosyl 1-pyrophosphate were completely converted to PRA by the purified TRP4 gene product anthranilate phosphoribosyltransferase. The spontaneous decomposition of PRA to anthranilate was balanced by recycling the anthranilate to PRA with anthranilate phosphoribosyltransferase and excess 5-phosphoribosyl 1-pyrophosphate. After zero adjustment of the fluorescence and addition of PRA isomerase, the decrease in PRA concentration was followed fluorometrically. This second PRA isomerase test resulted in values of about 3 milliunits/mg of protein at 25 "C, which are comparable to the data of the standard enzyme assay (Table 11).
Purification and M, Determination of the PRA Isornerase-In order to ascertain whether the TRPl gene product PRA isomerase catalyzes its reaction more effectively than the other TRP enzymes, the protein was purified from strain RH218 (YARpl). This transformant expresses PRA isomerase at about 100-fold higher levels than the wild-type strain (see Miniprint for detailed description of purification of PRA isomerase). A M, value of approximately 23,000 was calculated from the elution profile of the purified polypeptide from gel filtration of a Superose 12 column. The same value was obtained by analytical ultracentrifugation of pure enzyme4 in good correlation with the values derived from sodium dodecyl sulfate-polyacrylamide gel electrophoresis and the known nucleotide sequence (Tschumper and Carbon, 1980). These data strongly suggested that the PRA isomerase is a monomer.
Kinetic Studies of the Purified PRA Isomerase-In order to determine V,,, and KM, the relevant V, values were measured by steady-state kinetic experiments. Different limiting concentrations of PRA were enzymatically synthesized (as described above and under "Experimental Procedures"). The initial substrate concentrations ( [ S o ] ) in the range of 0.5-5 X KM were calculated from the spectrophotometrically determined initial anthranilate concentrations (ESlO = 2.98 m"' cm"; = 7.17 mM" cm"). The decrease of PRA was measured fluorometrically to determine the initial velocities (Vi).
On the basis of these data, a KM value for PRA of -5 ,LM was determined. With the Vmax value and the determined protein concentration of the pure enzyme ( [ e o ] ) , the turnover number (catalytic constant kcat = Vma./[eo]) was calculated to be kcat = 60/s. This leads to kc,/KM = 1.25 X lo7 , L M -~ s-l, which is only 50-fold smaller than the diffusion-controlled maximum (Fersht, 1985).
Under "Discussion," these values will be compared with the corresponding values of the TRP3 gene product InGP synthase.

DISCUSSION
The TRPl gene is expressed from a constitutive, unregulated promoter. The TRPl promoter is less efficient than the TRP3 promoter in initiating mRNA synthesis as measured in lac2 fusions under the same growth conditions. On the other hand, the TRPl transcript (0.8-1.0 kb) is about twice as stable as the TRP3 transcript (1.75 kb).
A simple explanation for these differences is a reciprocal relationship between mRNA stability and length as proposed by Santiago et al. (1986). These authors suggest that mRNA length and at least one additional factor strongly influence U. Hommel and K. Kirschner, personal communication.
mRNA stability in yeast. Longer mRNAs presumably present a larger target for an initial random endonucleolytic cut followed by rapid degradation.
The total amount of mRNA at the steady state, as determined by quantitative Northern hybridization, is similar for both genes. Assuming that the messenger stability reflects functional activity, the experimental data can be explained as the compensation between different promoter strengths (TRP3 promoter approximately twice as strong as TRPl promoter) and different mRNA stabilities (TRP2 transcript twice as stable as TRP3 transcript) for the two genes. Therefore, a different rate of translation initiation of both genes seems unlikely.
In S. cerevisiue, the concentration of the small PRA isomerase protein (Mr = 23,000) is very low. The PRA isomerase was purified from a 100-fold overproducing strain by total 15,000-fold enrichment relative to the chromosomally encoded gene. Therefore, the enzyme comprises not more than 0.007% of total cytoplasmic protein in a wild-type yeast cell. The corresponding fraction of InGP synthase/anthranilate synthase (Mr = 122,000) comes to 0.05% (Prantl et al., 1985).
Considering the different molecular weights of both enzymes, PRA isomerase and InGP synthase have comparable numbers of protein molecules per yeast cell. These data are in good agreement with the previous assumption that the translation efficiency for both genes is similar.
During hydroxylapatite chromatography, the PRA isomerase was eluted as a single component. The enzyme of M, = 23,000 was shown to be a monomer under physiological salt concentrations.* The basal enzyme activity of the TRPl gene product PRA isomerase is 2-3 times higher than that of the other TRP gene products. The comparatively high enzyme activity was found independently in two assays, in which product formation or substrate decrease was measured. The high basal enzyme level of the PRA isomerase cannot be explained by a higher transcription or translation rate of the TRPl gene. Table 111 compares the kinetic parameters of the purified yeast PRA isomerase with those of the purified yeast InGP synthase (Prantl et al., 1985). The lower Michaelis-Menten constant (KM) as a parameter for the affinity for its substrate, the higher catalytic constant (kt) as a parameter for the turnover of the substrate, as well as the higher value of kcat/ KM characterize PRA isomerase as being more efficient than InGP synthase.

TABLE I11 Steady-state kinetics of PRA isomerase and InCP s y n t h e in yeast
In E. coli, InGP synthase and PRA isomerase are fused in a bifunctional enzyme (without the glutamine amidotransferase domain). The kinetic constants of the purified E. coli enzyme agree well with the data found for the yeast enzymes (Kirschner et al., 1980;Kirschner et al., 1987). It has been shown that the two active sites of the bifunctional E. coli enzyme are independent and nonoverlapping. Neither channeling of the intermediate CDRP nor cooperative interactions between the two active sites seem to occur (Bisswanger et al., 1979;Cohn et al., 1979;Kirschner et al., 1980). Both functional domains are structured as &fold a/B barrels (Priestle et al., 1987) as reported for about a dozen different enzymes (Muirhead, 1983;Lindqvist and Branden, 1985). The two active sites do not face each other, making any channeling of the substrate between active sites virtually impossible (Priestle et al., 1987). Examination of a multiple sequence alignment for the known PRA isomerase and InGP synthase enzymes from various organisms (including S. cerevisiae) is consistent with the notion that all PRA isomerases and InGP synthases have the same topological fold of a/@ barrels (Priestle et al., 1987). Thus, in the case of InGP synthase and PRA isomerase, gene fusion seems to affect the respective catalytic efficiencies of either enzyme only to a limited extent.
One question that arises is whether the TRPl gene of S. cerevisiue resulted from a rearrangement event. Due to the long evolutionary periods involved in the formation of a certain gene arrangement, it is impossible to prove its mode of generation in any specific organism experimentally. For eukaryotic cells, a common hypothesis is that multifunctional proteins encoded by single genes are formed concomitantly with higher development . The formation of multifunctional enzymes may confer evolutionary advantages to an organism, such as simultaneous regulation of gene expression, equimolar synthesis of all enzymatic activities, and channeling of interme~ates (reduction of intermediate concentration). By contrast, the results obtained for the TRPl gene of 5' . cerevisiae support the concept of a late detachment and rearrangement of a promoterless part of a gene (coding for the PRA isomerase domain) from an originally trifunctional general control regulated gene (including the glutamine amidotransferase and InGP synthase functions). The evidences are summarized as follows. l) The TRPl promoter is unusual among the TRP promoters; it is the only TRP promoter that is not regulated by the general control system. Transcription from the TRPl promoter is weaker than from the regulated TRP3 promoter. This situation is compensated by higher stability of TRPl mRNA. The largest TRPl transcripts have a 200-bp untranslated leader that presumably impairs translation of the transcript (Kim et al., 1986). 2) The TRPl gene product PRA isomerase is a very active enzyme providing the cell with a 2-3 times higher basal enzyme activity than any of the other TRP enzymes. 3) Among the lower fungi, separation of the genes of InGP synthase and PRA isomerase is uniquely found only in yeasts (Braus et ai., 1985). All other ascomycetes analyzed so far carry a single gene coding for both InGP synthase and PRA isomerase. Comparison of the connector regions and the COOH and NHz termini of known fused and separate InGP synthase and PRA isomerase proteins (Fig. 5) reveals a remarkable feature: the NH, terminus of the yeast PRA isomerase contains additional amino acids in comparison to prokaryotic monofunctional PRA isomerases that correspond in length to the natural connector region in the proteins of the ascomycetes A. nidulam and N. crassa.
If the hypothesis of a rearrangement of the TRP1 gene is correct, why then are the PRA isomerase and the InGP synthase fused in so many organisms if there is no evident advantage for this arrangement? An answer to this question can only be speculative. Two models were proposed for the evolution of enzymes in multistep pathways. Jensen (1976) proposed that primitive enzymes possessed a broad substrate specificity and were active in several metabolic pathways. EvoIution could then have been achieved by gene duplication and subsequent mutations leading to specialization in the substrate specificity of the encoded proteins. This model was supported by Parsot (1986Parsot ( , 1987, who has shown sequence homologies between the tryptophan synthase p chain (last step in tryptophan biosynthesis), threonine synthase, threonine dehydratase, and D-serine dehydratase and has postulated a common ancestor for these enzymes. Horowitz (1945) earlier proposed an alternative hypothesis that biochemical pathways could evolve in a stepwise manner by duplication and evolution of new functions in the reverse direction compared to the direction of synthesis (retrograde evolution). It has been suggested that the eight-stranded cy/@ barrel is particularly suited for the evolution of new functions since function seems to be determined to a large extent by surface loop modification at the carboxyl termini of the B strands of the basic structure (Lindqvist and Branden, 1985). Although the cy subunit of the tryptophan synthase (second last step in tryptophan biosynthesis), the InGP synthase (third last step), as well as the PRA isomerase (fourth last step) show only a low degree of homology (Priestle et al., 1987), all three enzymes have the topological fold of an 8-fold a/j3 barrel (Crawford and .