Indole Channeling by Tryptophan Synthase of Neurospora*

SUMMARY Tryptophan synthase (L-serine hydro-lyase-adding indole) (EC of Neurospora crassa was studied in reaction mixtures containing [2-14C]indoleglycerolphosphate and nonlabeled indole. A detailed kinetic analysis of the passage of isotope from indoleglycerolphosphate to indole and to tryptophan during the synthesis of tryptophan revealed that: (a) indole disappears first and indoleglycerolphosphate is utilized after the indole concentration falls to barely detectable levels; (b) significant amounts of radioactivity appear in indole; and (c) the specific radioactivity of the tryptophan formed is much lower than that of indoleglycerolphosphate but 2- to 5-fold higher than that of indole depending inversely upon the initial indole concentration. The analyses of these observations are interpreted as evidence that indole is an enzyme-bound or “channeled” intermediate in the conversion of indoleglycerolphosphate to tryptophan. The enzyme was separated from reaction mixtures containing an equilibrium mixture of labeled indoleglycerolphosphate and labeled indole by passage of the mixtures through ultrafilters which retained material molecular

A detailed kinetic analysis of the passage of isotope from indoleglycerolphosphate to indole and to tryptophan during the synthesis of tryptophan revealed that: (a) indole disappears first and indoleglycerolphosphate is utilized after the indole concentration falls to barely detectable levels; (b) significant amounts of radioactivity appear in indole; and (c) the specific radioactivity of the tryptophan formed is much lower than that of indoleglycerolphosphate but 2-to 5-fold higher than that of indole depending inversely upon the initial indole concentration.
The analyses of these observations are interpreted as evidence that indole is an enzyme-bound or "channeled" intermediate in the conversion of indoleglycerolphosphate to tryptophan. The enzyme was separated from reaction mixtures containing an equilibrium mixture of labeled indoleglycerolphosphate and labeled indole by passage of the mixtures through ultrafilters which retained material of molecular weights greater than 50,000.
Radioactive indole was found exchangeably bound to the retained enzyme.
From these results, it was estimated that 2 to 3 moles of indole were bound per mole of enzyme. The results are interpreted as evidence that the Neurospora enzyme fits well the conceptual framework of the "surface model" (DAVIS, R. H. (1967)  Press, New York), from a primitive multicomponent complex.
Tryptophan synthase (L-serine hydro-lyase-adding indole) (EC 4.2.1.20) from Neurospora crassa catalyzes the three reactions shown in Fig. 1. III 1962 I>el\Ioss (1) conducted a detailed study of this enzyme, the results of which formed the basis of a useful and often cited model. A diagram depicting this model is also * This work was supported by Grant GM 18083 from the National Institutes of Health. presented in Fig. 1. The salient features of the DeMoss model are as follows.
1. l'yridoxal phosphate is bound to the enzyme at Site II. Serine also binds at this site through formation of a Schiff base with enzyme-bound pyridoxal phosphate. 2. For catalysis of Reaction 1, indoleglgcerolphosphate binds at Site I of the enzymepyridoxal phosphate-serine complex. Glyceraldehyde 3-phosphate is liberated and the indole moiety of indolcglycerolphosphate is transferred to Site II. At Site II a condensation occurs between indole and the reactive carbon skeleton of scrine which apparently involves the prior dehydration of the serine molecule (2). After condensation, tryptophan and free enzyme arc liberated.
3. For catalysis of Reaction 2, indole is bound to Site II of the enzyme-pyridoxal phosphate-serine complex and dehydration and condensation occur as above.
4. For catalysis of Reaction 3f, indolcglycerolphosphate is bound to resolved enzyme' at Sit,e I. Indole and glyceraldehyde 3-phosphate are liberated from the enzyme. In the reverse Reaction 3r, indolc and glyceraldehyde S-phosphate are bound at Site I, and indoleglycerolphosphate and free enzyme are generated.
In this study I~e~Ioss reconsidered the question, originally studied by Yanofsky and Rachmeler (3), concerning the participation of indole as an irnermediate in Reaction 1. Yanofsky and Rachmelcr (3) had concluded that free indole (i.e. material in equilibrium with solvent indole) was not an intermediate in Reaction 1 from two observations. First, Reaction 3f proceeded at a rate far below that required for its participation as a component of Reaction 1 and second, a toluene overlay failed to trap indolc during the course of Reaction 1. Dc1loss (1) came to essentially the same conclusion from experiments in which the enzyme was presented simultaneously with tritiated indoleglycerolphosphate and nonlabeled indole.
In such experiments, net synthesis of tryptophan was accompanied by net disappearance of indole in equimolar quantities.
Changes in the concentration of indoleglycerolphosphate in the reaction mixture were not detected.
However, significant quantities of radioactivity appeared in both tryptophan and indole. The specific radioactivity of the tryptophan formed was approximately 5-fold higher than t.hat of indole.
From this result, DelIloss concluded that free indole Ivas not an intermediate in the conversion,of indoleglycerolphosphate to tryptophan.
The experiments described in this report are an extension of earlier observations (2) on the inhibition of tryptophan synthase by iudoleacrylic acid. This compound exerts its inhibition by interfering with the binding of serinc at Site II. The inhibitor is thus a probe which magnifies the catalytic contribution of Site I relative to Site II in reactions of the enzyme which might require the function of both Sites I and II (i.e. Reaction 1). In studies described here, the enzyme, in the presence and absence of inhibitor, was presented with saturating quantities of indole glycerolphosphate labeled with 14C in l'osition 2 of the indole moiety.
The reaction mi.\-tures also contained various levels of ~onlabelcd indole.
The kinetics of appearance of isotope in both indole and tryptophan were determined by sampling the reaction mixtures at appropriate intervals during net synthesis of tryptophan.
From these measurements, the argument is developed that indole is a channeled intermediate in the conversion of indoleglycerolphosphate to tryptophan, and further, it is corlfined to a microenvironment at or near the catalytic surfaces of the enzyme.

METHODS
Organisms and E'xtructs-Partially purified tryptophan synthase prepared from extracts of wild type N. cras.sa (strain 74A) was employed in this work. The methods of purification and some properties of the preparation have been described previously (2) mM; and a IO-fold excess (5 units) of tryptophan synthase. This mixture was incubated at 37" for 1 hour. The reaction mixture was partially deproteinized by immersion in a boiling bath for 5 min. Protein that precipitated was removed by centrifugation. The reaction mixture was then thoroughly extracted with toluene to remove unreacted indole. The radioactive indoleglycerolphosphate was isolated from the reaction mixture by ascending chromatography on Whatman No. 3MM filter DaDer bv develop-__ " ment in a solvent system consisting of isopropanol:ammonia: water (100:5:5).
One milliliter of the denroteinized and tolueneextracted reaction mixture was applied to each of six sheets of filter paper (42 X 50 cm). The radioactive indoleglycerolphosphate was located (Rp, 0.09) with a hand-held radiation detector fitted with a 5-mm slit aperture.
The radioactive material was eluted from the paper with hot ammoniacal ethanol, and the resulting solution was concentrated in VUC~LO. This material cochromatographed with authentic indoleglycerolphosphate. In control experiments, it was used as substrate in reaction mixtures containing excess tryptophan synthase, serine, and pyridoxal phosphate.
The amount of tryptophan formed in such reaction mixtures was stoichiometricalii dependent upon the amount of indoleglycerolphosphate added. The sDecific radioactivitv of the tryptophan formed was equal to the*indoleglycerolphosphate supplied.
Unlabeled indoleglycerolphosphate was prepared as previously described (2), except that a chromatographic procedure described by Lester (4) was employed in place of the barium precipitation step. Using this method, indoleglycerolphosphate is well resolved from indoleglycerol and other materials which absorb at 280 nm.
Assay of Enz?/mic Acliuities-The react.ions of tryptophan synthase were measured by methods described previously (1, 2). One unit of activity is defined as the formation of 1 pmole of product or the utilization of 1 pmole of substrate in 1 min at 37". Other J~ethods-Radioactivity was measured with a Packard liquid scintillation spectrometer.
The counting efficiency was about 83c/c and quenching was insignificant,.
Count rates were adjusted to at least 10 times background and in all cases a minimum of 5ooo counts was registered.
Reuge& and Other illuterials-[2-'4C]IIldole was purchased from the International Chemical and Nuclear Corporation. Other reagents and chemicals were obtained from previously mentioned sources (2) or from standard sources of supply. The ultrafilters (Centriflo membrane ultrafilters CF50A) which retain soluble materials of a molecular weight greater than 50,000 were purchased from the Amicon Corporation, Lexington, Mass.

RESULTS
Tryptophan synthase in the presence of ltyridosal phosphate will catalyze the formation of tryptophan from indoleglycerolphosphate and serine. If indole is also present in the reaction misture, the rate of appearance of tryptophan is considerably higher.
In such a mixture, indole is utilized first and indoleglycerolphosphate is utilized only after the concentration of indole has fallen to barely detectable levels (1). Fig. 2 presents data obtained from 10 of 18 separate reaction mixtures each of which received, initially, a saturating concentration (0.55 mM) of [2-'4C]indoleglycerolphosphate (specific radioactivity -400 nCi per pmole).
These reaction mixtures also received, initially, concentrations of indole varying from 0 to 2.64 mM. Reaction mixtures r through ./ received the inhibitor indoleacrylic acid at a level (0.35 mM) which gives about 5O0/0 inhibition of Reactions 1 and 2 of tryptophan synthase (2). The inhibited reactions received twice the amount of enzyme so as to adjust the rate of appearance of tryptophan to a level approximately equal to that of the uninhibited reactions (d through B). Each reaction mixture was sampled at intervals of 10 min. A sample (1.0 ml) was withdrawn and delivered into a tube immersed in a boiling water bath. The tube was capped, and the reaction was terminated by boiling for 2 min. The sample was then chilled and subjected to the following assay procedures: (a) total indole as Ehrlich-positive material extractable into toluene; (b) radioactivity in indole (see "Jlethods"); (c) tot,al tryptophan in the aqueous layer from a as Ehrlich-positive material extractable into toluene over a tryptophanase digest (1); (d) radioactivity in tryptophan as in b above; (e) total indoleglycerolphosphate in the aqueous layer from c as material absorbing at 290 nm extractable into ethylacetate aft,er oxidation with sodium metaperiodate (1) ; (f) radioactivity in indoleglycerolphosphate as radioactivity in the ethylacetate layer from e.  show the time course of appearance of total radioactivity in indole.
Panels C and D of Fig. 3 show the rates of appearance of radioactivity in tryptophan.
The initial rates c f e&l of these processes were estimated from the first three or four points of each curve.
Ai comparison of these initial rates provides some information 011 the apparelit catalytic fate of indoleglycerolphosphate molecules bound to Site I at early times during the reactions. III Reaction -1 (Fig. 2), for example, isotope appears in tryptophan initially at a rate 170.fold lrighcr than that for solvent indole (Fig. 3). III Reaction B (Fig. 2) which received initially 0.4 rnM indole, however, the rate at which isotope entered solvent indole was considerably higher than that observed in Reaction 11 and the rate at \vhich isotope entered tryptophan was lolver (Fig. 3). The ratio of the two rates was 9.8 or about l'i-fold lower than the ratio noted for Reaction A.
Similarly, in Reactions C, D, and I:' (Fig. 2)) as the initial indole concentration increased, this ratio decreased (Fig,  3). In Reaction E ( Fig. 2), the observed ratio ( Fig. 3)  in such reaction mixtures. The rate at which radioactivity from indoleglycerolphosphate entered tryptophan was markedly reduced. This result was, of course, expected from the well known early utilization of indole reported previously (1). Of equal or even greater interest was the observed trend in the rate of appearance of radioactivity in indole in the reaction mixtures. This rate increased with increasing concentrations of indole. The indole initially present appeared to act with increasing ef fectiveness as a trapping pool for isotope passing from indoleglycerolphosphate to tryptophan. Thus, indolc in this reaction seems to exhibit one of the traditional criteria of an intermediate in an enzymically mediated conversion.
One of the original arguments employed in ruling out, the participation of Reaction 3f (and therefore of indole as an intermcdiat,e) in t,he coIlversion of indoleglycerolphosphatc to tryptophan was the observation that Reaction 3f proceeded at an apparent rate X-fold lower than that required to sustain Reaction 1. Analysis of the rates presented in Panels B and D of Fig. 3 bears on this point.
The reaction mistures described in these panels contained 0.35 m&c indoleacrylic acid, an inhibitor, which for the reasons stated above, would tend to magnify the catalytic activity of Site I relative to Site II in t,he context of the De&loss model. In these tubes the ratios (InGl' + Trp/InGI' + Ind) (where InGP is indolcglycerolphosphate, Trp is L-tryptophan, and Ind is indole) were observed to vary over a 100.fold range with increasillg concentratiorls of initially supplied indole. AS shown in Panel R of Fig. 3, the initial rates of appearance of isotope in indole were measurably higher than those sho\vn in Panel A, while the rates of isotope entry into tryptophan (Panel D) were nearly equal to those shown in Panel C. For each intial concentration of indole, consequently, the ratio (InGE' -Trp/InGl' + Ind) of these rates was somewhat, lower, indicating about a 2.fold higher relative activity of catalysis at Site I in the presence of indoleacrylic acid. 4. Influence of initial concentration of indole on its ability to trap isotope passing from indoleglycerolphosphate to tryptophsn. The ratio of the initial rate of appearance of isotope in tryptophan to the initial rate of appearance of isotope in indole (cf, Fig. 3) is plotted as a function of the initial concentration of indole in the reaction mixture.
For methods see Fig. 2.
t,akcn from reaction mixtures R through E and additional reaction mixtures which had received increasing initial concentrations of indole (see legend to Fig. 2). In this figure the ratio of the initial rat,e of entry of isotope into trypt,ophan to the rate of entry of isotope into indole is plotted on the ordinate as a function of the initial concentration of indole. The implication of this analysis is that indole initially present in the reaction mixture served as a partially effective trapping pool for isotope passing from indoleglyccrolphosphate to tryptophan and therefore that indole or a compound freely exchangeable with indole is an intermediate in this reaction.
The trapping phenomenon argues favorably for the identity of indolc as an intermediate.
Of equal or even greater significance is the observed efficiency of trapping indicated in Fig. 4. It should be recalled (cf. Fig. 1) that the K, of Site II for indole is -0.1 mM. In usual practice 0.5 mM indole is regardedas a saturating concentration with respect to Reaction 2. As shown in Fig. 4, five times this concentration was required for essentially complete trapping.
The observed lack of efficient trapping argues favorably for the concept that indole arising as a result of catalysis at Site I is channeled to and preferentially used by Site II. These results arc formally similar to those reported by Lue and Kaplan (5) for the carbamyl phosphate synthase-aspartatetranscarbamylasc aggregate. Clearly, these results cannot exclude the possibility that. an as yet unknown compound is an intermediate in the conversion of indoleglyccrolphosphate to tryptophan and that this compound is in equilibrium with indole present in the reaction mixture. 011 the other hand, they offer no compelling reason to invoke the existence of such a compound. An alternative view which appears consistent with these results is the surface model originally presented by Davis in a much cited paper (6). According to this concept indolc generaM as a result of catalysis at Site I would enter a space near the catalyt,ic surface of the enzyme. The relatively high affinity of Site II for iudole and the inability of Site I to produce saturating levels of indole (because of its lower V,,,) practically ensures that indole would behave as an enzyme-bound or channeled intermediate. In this cast an abnormally high local concentration of free indole in the solvent would be expected t,o exchange with indole present in the "channel." If the channeled indole were labeled with radioactivity, the radioactivity would thus be "urlchanneled" and would appear in the free indole in the solvent.
The effect noted in Fig.  4 could thus be conveniently interpreted as a case of essentially complete "unchanneling" at indole concentrations in the solvent of 2.0 to 2.5 rnbr.
If the surface model of Davis (6) is applicable to this system, it seems reasonable to suppose that the enzyme would catalyze an exchange reaction involving the indole moiety of indoleglycerolphosphate and free indole dissolved in the solvent. It was of interest, therefore, to study the question of exchange of the indole moiety of indoleglycerolphosphate with free indole in the presence of the enzyme.
Although this question has been studied before (1, 7), and negative results obtained, a reconsideration seemed indicated in view of the fact that indole appeared to be a concentration-dependent trapping agent for isotope originating in indoleglycerolphosphate. For this purpose a preparation of "resolved enzyme" (1, 2) was employed. For this experiment, which is described in Table I If neither pyridosal phosphate nor swine were added (tube l), tryptophan syntlwsk was not detected and no challgc in the concentrations of either indolrglycerolphosphatc or indole was clctccted. However, eschange was detected at the low but easily measured rate of 2.63 nmoles per hour.
It was of considerable interest and significance that this rate, not detectable by the usual calorimetric procedures, is almost exactly that expected since, as shown in Fig. 1, Reaction 3f proceeds at a rate essentially 100.fold lower t,han the rate of Reaction 2. Such au intcrpretatiou is supported by the observed stimulation of the rate of exchange by the single addition of pyridoxal phosphate (tube 2) and the absence of stimulation by the single addition of serine (tube 3). These single additions have already been shown to elicit similar responses by Reaction 3r (2). The observed exchange provides further support for the existence of indole as an enzyme-bound intermediate resulting from reaction of indoleglycerolphosphste at Site I. A somewhat more detailed analysis of the data presented in Fig. 2 provided further support for the concept of indolc channeling by tryptophan synthase.
For this purpose the rates of formation of tryptophan in each of the reaction mixtures were comput,ed.
These rates were linear during the 40.min period under examination.
Linear regression analyses were performed for each set of data, and correlation coefficients in excess of 0.98 were obtained in all cases. Increments of tryptophan which occurred during 10 min of reaction were obtained directly from these curves. The concomitant rates of appearance of radioactivity in tryptophan (raw data supplied to data repository) were nonlinear and were best described by a general exponential equation of the form Y = a x b". Nonlinear regression analyses were performed testing the goodness of fit to this equation for each set of data on appearance of radioactivity in trgptophan. Correlation coefficients in excess of 0.96 were obtained.
Increments of radioactivity which occurred during 10 min of reaction were taken directly from t,hesc curves. The average specific radioactivity of indole entering Site II and subsequently entering tryptophan will be referred to as the specific radioactivity of tryptophall-precursor indole and can be estimated during a given 10.min period from the expression: A nCi per ml per 10 min in tryptophan A pmoles tryptophan per ml per 10 min = Average specific radioactivity of tryptophan-precursor indole for that particular 10.min period.
The results of these calculations are presented in Fig. 2 along with the measured values of specific radioactivity for indoleglycerolphosphat,e, indole in the solvent, and tryptophan.
The following trends are apparent in these data.
1. The rate at which the specific radioactivity of tryptophan increased was in all c-ascs inversely proportional to the initial concentration of indolc present in the reaction mixture. 2. The disparity bctwecrr the specific radioactivity of tryptopharl-precursor illdole and of free indole in the solvent dccrcased as the initial concentration of indole in the solvent increased. 3. No obvious dccrcasc in the specific radioactivity of indole-glScerolphosphate was observed during thcl first 40 min of each reaction.
These observations are consistent \\-ith the notion that both indoleglycerolphosphatc and indole are contributing to a pool of channclctl indolc (tryptophall-precursor intlolc) within a microenvironment at or near the catalytic surface of the enzyme. Exchange diffusion can occur bet\veen channeled indole and irldole in the solvent.
Indole molecules channeled within this mi~rociivirolimcIit arc drawn randomly to Site II of the enzyme for subsequent conversion to tryptophan.
These values are substituted in the following pair of simultaneous equations.
(a) Moles InGI' x sl ecific radioactivity of InGI' + moles solvent indole x specific: radioactivity solvent indolc = moles tryptophan increment x calculated specific radioactivity of tryptopharl-precursor indole, and (b) lnd~l,oP) + I%olvent) = molts of tryptophan increment. Solution of these equations yields the molar quantities of indoleglycerolphosphate and indole in the solvent which have contributed indole moieties to the population of tryptophan precursors channeled in the microenvironment which arc available for reaction at Site II during any specified 10.min period of a reaction.
The ratio of these quantities, p, is given by the expression, where Ind(,,~vcntJ = moles of indole contributed by solvent illdole, Illd(I"Cp) = moles of indole contributed by solvent indoleglycerolphosphate.
This ratio is a convenient \vay of expressing the relative contribution of each spccics to tryptophall-precursor indole channeled in the microenvironment.
The relat,ionship of the value of p to the cotlcentration of indolc in the react,ion mixture for the first 40 min of several of the reaction mixtures (B through Ls and G through J, Fig. 2, plus three other reactions supplied initially with higher levels of indole) is shown in Fig. 5. Three conclusions follow from this analysis. The estimation of p is discussed in the text. As defined, p is the ratio of solvent indole to indole gcncrated at Site I which is available at Site II for conversion to tryptophan.
2. The, misillg ratio, ad inferred from the specific radioactivity of the c~hannclctl precursor indole, is not a simple linear function of the concclltratiolls of i~~doleglycerolpl~osphatc and indole in the solvent.
Thcs;c collclusiolls are consistent with the concepts advanced by Davis (6) iti his "surface model" and &end the model of DelIoss (1) to iilclude an enzyme-indolc complex.
The possibility of such a c~omplcs has recently been suggested by Kirschner and Wiskocil (8).
Conscqurntly, attclltion was focused on the problem of obtaining direct espcrimeilt,al evidence for the existence of an enzymc-indole complcs. The experiments detailed in Table II were carric>tl out ill an effort to find indole intimately associated or cschangeably bound to the surface of the enzyme. For these esperimciits naiiomolc quantities of tryptophan syiithase were supplied \vitli nearly saturating levels of radioactive [2-%]in-dolegl~ccrol1)liosl)llate (specific radioactivity, 8.0 &i per pmole). No other substratcs or cofactors were supplied.
The mixture was incubated briefly at 37". Tinder these conditions the ellzyme will rollvert some of the indoleglycerolphosphate to indole. After about 10 mill the net conversion stops presumably for the reason that equilibrium of the reversible reaction (31) is achieved (1). Samples of the reaction mixture (4.0 ml) were then passed through an ultrafiltcr which retained molecules in excess of 50,000 molecular weight.
The material retained by the ultrafilter was resuspended in 4.0 ml of 50 mM potassium phosphate buffer (pII 7.8) which was 0.5 mM in nonradioactive indole. Q Samples of 4.0 ml of a mixture containing tryptophan synthase (3.74 units per ml; 2.5 mg per ml); inactive protein (2.5 mg per ml); or no protein.
b Samples of 4.0 ml of a mixture containing tryptophan synthase (3.45 units per ml; 2.7 mg per ml); inactive protein (2.9 mg per ml); or no protein.
In all cases the specific radioactivity of the indole moiety of the compounds present in these mixtures was 8.0 &i per rmole.
The resuspended material was incubated 30 min at 30". The mixture was then passed through the ultrafilter again and indole present in the filtrate \vas examined for radioactivity.
The radioactivity associated with indole in this second filtrate was assumed to represent indole that was exchangeably bound to macromolecular material retained by the ultrafilter in the initial passage of the reaction mixture through the filter.
The amount of indolc thus indentified was computed from the known specific radioactivity of t,he originally supplied indoleglycerolphosphate. The experiment was controlled in the following ways. Catalytically inactive protein was prepared from a tryptophanless mutant (td 120) of N. crassu. This mutant has been shown (9) to have no detectable protein catalytically related to tryptophan synthase.
Crude extracts of this mutant were subjected to the same protocol of purification (2, 10) as that employed to prepare catalytically active enzyme. Protein of the appropriate fraction was thus obtained and tested for catalytic activity.
Active tryptophau synthase preparations used in this study (2, 10) are found to consist of about 5Oa/, tryptophan synthase and are routinely contaminated with a serine deaminase activity (2). The extracts from td 120 xere found to contain the appropriate level of serine deaminase but no tryptophan synthase under conditions in which >iooO of wild type levels could have been detected.
The catalytically inactive protein was incubated in a mixture of radioactive indoleglycerolphosphate and radioactive indole of identical specific radioactivity (8.0 PCi per pmole) and subjected to the protocol described for the catalytically active protein.
III a separate set of controls, protein was omitted from the reaction mixture, and the protocol was repeated to obtain the filter background.
The results reported in Table II show that protein mixtures containing catalytically active tryptophan synthase have the ability to bind indole exchangeably to an cxtrnt considerably above that which cau be ascribed either to background in the assay procedures or to a generalized or nonspecific binding of indole to protein.
In similar crperiments not reported here, mixtures like those described above were passed over Sephadex G-25 columns (2.5 x 40 cm). In these experiments almost no radioactivity was detected in the excluded volume containing the protein and enzymic activity.
These erperiments did show a slight but reproducible skewing of the leadin, cr edge of the indole peak. Interpretation of these results is complicated by the observed affinity-of the gel for indole.
These results may indicate that the presence of an exchangeable species is not absolutely required for removal of radioactivity from the indole-enzyme complex Zndole as Intermediate-Free indole is not detected by colorimetric m&hods during the course of Reaction 1 of tryptophan synthase (3,11). This fact coupled with the observation that Reaction 3f proceeds, as usually measured, at a rate about >io that required to sustain the rate of over-all conversion of in doleglycerolphosphate to tryptophan led Yanofsky and Rachmeler (3) to conclude that indole is not an obligate intermediate in the conversion of indoleglycerolphosphate to tryptophan.
DeMoss (1) re-examined this question by supplying the enzyme with tritium-labeled indoleglycerolphosphate, nonlabeled indole, and other substrates and cofactors necessary for tryptophan production.
In such a reaction mixture, indole was utilized before indoleglycerolphosphate for trpptophan formation.
The tryptophan formed contained appreciable amounts of radioactivity.
Indole, too, became radioactive. The specific radioactivity of the tryptophan, however, was 4-fold higher than that of indole.
DeMoss' interpretation was that small amounts of indoleglycerolphosphate were going to tryptophan without equilibration with free indolc and that radioactivity which appeared in indole did so as a result of the reversibility of Reaction 3. Thus, he concluded, also, that free indole is not an intermediate in t,he conversion of indoleglycerolphosphate to tryptophan.
The experiments described in this report suggest that indole may be regarded as an enzyme-bound or "channeled" intermediate in the conversion of indoleglycerolphosphate to tryptophan. This suggestion is supported by the following evidence : 1. As the concentration of indole in the solvent was increased, its ability to trap radioactivity passing from indoleglycerolphosphate to tryptophan increased (Fig. 4).
2. At concentrations of indole on the order of 2.5 mM, the compound is completely effective as a trapping pool (Fig. 4).
3. The demonstration of bound indole 011 preparations of enzyme separated from reaction mixtures iu which Reaction 3f was occurring.
4. The demonstration that, when Reaction 1 was conducted in the presence of labeled indoleglycerolphosphate and nonlabeled indole, the specific radioactivity of tryptophan precursor available at Site II for condensation with activated serine was very different from either indoleglycerolphosphate or indole in the solvent.

This interpretation
is consistent with those of Lue and Kaplan (5) and of Gaertner ef al. (12) in that it invokes a channeled intermediate.
In this case, however, the argument is applied to a single enzyme rather than to a supramolecular aggregate. The evolutionary implications of this interpretation are discussed more cstensivrly below, Nature of Indole-Enzyme Con$er-That indole is exchangeably bound t,o the enzyme is indicated by the opcratioiis employed in the csperiment of Table II. Simply resuspcnding the retained cuzyme in a solution containing a large excess of non radioactive indole was sufficient to convert the radioactivity to a form which would pass readily through the filter.
It is possible to estimate the amount of indole thus associated with the en zyme, by making a fe\v apparently reasoiiablc assumptious. For Esperimctit 1 of Table 11, each sample contained 14.9 units of rnzyme.
The data of Meyer et al. (10) indicate that S. crassa tryptoplian syntliase has a molecular \vciglit of 150,000 and a masimal specific activity of 3.34 units per mg. Thus, each sample contained 29.4 nmoles of e~~zynic.
The indole associated with this amount of enzyme over and above that of the controls was 83.5 nmoles. Thus, 2.8 moles of indole were bound per mole of enzyme. For Experiment 2 similar calculation revealed 2.2 moles of indole per mole of enzyme. Experi ments now in progress (see below) in this laboratory indicate that the hreurospora enzyme exists as a dimer composed of subunits of 75,000 molecular weight.
Thus, it would appear that each subunit has the ability to bind at least a molecule of indole under the conditioiis of the experiment of Table II. Jfechanisnl of Jetion-&1 model that attempts to accommodate previous work and the evidence presented in this report is depicted in Fig. 6. The K, for iudoleglycerolphosphate in Reaction 1 is nearly the same as the K, for indole in Reaction 2 but the K, for indole in Reaction 3r is about 30-fold higher than either of these values (cf. Fig. 1). Thus, the affinity of Kite I1 for indolc is greater than that of Site I for iudolc.
The V,,, for Site I, as judged by Reaction 1, is about onehalf that for Site II as judged by Reaction 2. Thus, the conditions for the "surface model" of Davis (6) are met. Indoleglycerolphosphate bound at Site I is converted to indole at a rate insufficient to saturate the potential maximal rate at which Site II can convert indole to tryptophan.
In the case of Reaction I, indole generated at Site I is transferred or channeled immediately to Site II in a manner that prevents its escape to the solvent.
Data presented here indicate that the Site I to Site 11 transfer of indole is about 200.fold more probable than its escape to the solvent. The presence of indole in the solvent at less than saturating concentrations with respect to Site II interrupts this process in a way that provides insight into the mechanism of the reaction. Solvent indole partially engages Site II thereby reducing the probability of transfer of indole from Site I. Indole in the process of being transferred (chanrlcled) is apparently able to exchange with the solvent indole (cf. Fig. 3). The presence of indole in the solvent, reduces considerably the rate at which net conversion of indoleglycerolphosphate t.o tryptophan can occur in t.he absence of solvent indole (cf. Fig. 2). This may be esplained by assuming that the total availability of indole 110~ exceeds the ability of Site II to convert it to tryptophan.
Thus the "channel" would be filled to capacity causing a high local concentration of the bound intermediate in the vicinity of Site I and consequent reduced availability of space at the site for continued turnover of indoleglycerolphosphate.
In the absence of indole in the solvent, Site II is able to remove indole from the channel at an apparent rate 2.fold higher than its rate of production by Site I. In this case the channel is effectively empty and Site I is maximally available for binding and turnover of indoleglycerolphosphate. The apparent inability of Site I and therefore Reaction 3f to produce indole at a rate sufficient to accommodate the observed rate of Reaction 1 may be accounted for in the same terms. Reaction 3f may be detected only in the absence of serine. In the absence of serine Site II may bind indole but obviously could not produce tryptophan.
Thus, Site II could not remove indole from the channel.
Escape of indole to the solvent would then occur at a rate much lower than its transfer to Sit.e II and subsequent conversion to tryptophan.
In the case of Reaction 3f this rate probably limits the conversion of indoleglycerolphosphate to indolc and glyceraldehyde 3.phosphate for the reason that the channel remains at or near capacity and, as proposed above, sufficient space at Site I is not available for unrestricted turnover of indoleglycerolphosphate.
In this corn nection it should be recalled that indolc inhibits Reaction 3f as judged by net indole appearance (1) but not as judged by isotopic exchange (cj. Table I). The binding studies described here (Table II) were performed with mistures catalyzing Reaction 3f and revealed the presence of 2 to 3 moles of indole bound per mole of enzyme and seem to offer support for the mechanism proposed.
Positioning of Sites I and II-The Neurospora enzyme, in contrast to that of Escherichia coli, appears not to be dissociable into distinct polypeptides coded by separate genes. In the B. co2i system these proteins are referred to as the cx and /3~ subunits and in this sense they may be regarded as composing a small multienzyme complex in the native configuration (i.e. a& complex).
The molecular weight of this comples is 148,000 (CY subunit, 29,000; /I2 subunit, 45,000 x 2). The native enzyme is readily dissociable into cy and /32 subunits and the catalytic capabilities of the separate subunits have been thoroughly studied and extensively described.
The cx subunit alone will catalyze Reaction 3. The flp subunit will catalyze Reaction 2. The native complex, a&, is required for catalysis of Reaction 1. Of particular interest is the fact that the maximal rates of Reactions 2 and 3 as catalyzed by the native complex are much higher than those observed with the individual subunits. The fiz subunit alone will catalyze the deamination of serine but this activity is almost completely absent in the CY& complex (cf. Ref. 13 and literature cited therein).
Kirschner and Wiskocil (8) in studies of steady state kinetics and fast reaction studies on the E. coli enzyme have suggested that in the native configuration (c&2) more efficient conformations of the active sites are either induced or stabilized.
In the Neurospora system it seems unnecessary to invoke this level of conforrnational flexibility. Studies presently in progress (in collaboration with .J. A. DeMoss) on clectrophoretically homogeneous preparations of the Neurospora enzyme indicate a molecular weight of 150,000. The relatively mild conditions which cause the h'. coli enzyme to dissociate apparently do not in the Neurospora system. The Neurospora enzyme is devoid of scrine-deaminating activity (2, 10). Treatment of the enzyme with sodium dotlccyl sulfate (2y0 sodium dodecyl sulfate at 100" for 15 min followed by CO min at 37") and polyacrylamide gel electrophorcsis by the procedure of Wcbcr and Osborn (14) reveals the presence of a single band with a relative mobility suggesting a molecular weight of 75,000. These data would indicate that the Neurospora enzyme is a dimer consisting of two subunits, each a single polypeptide analogous to an cy chain covalently linked to a fl chain. Experiments are now in progress to test this assertion critically.
The selective advantages and evolutionary significance of multicnzyme complexes and channeling of intermediates have been pointed out by Davis (6). The evidence for channeling of indole in tryptophan synthase of Neurospora presented in this report may provide some insight regarding evolution of the enzyme.
llonner's theory (15) of the evolution of gene-enzyme systems is based in large part on several features of the tryptophan synthases from several organisms.
It states that nonidentical catalytically active polypeptidc chains which interact to form complex enzymic activities (e.g. the CY and /3 chains of the E. coli system) rnay in the course of evolution be converted into a single polypeptide under the control of a single genetic unit (e.g. the Ncurospora system). I~onner pointed out a selective advantage that a single gene-single enzyme system would have over a multicomponent system. Once such a multicomponent system became established, coordination of the rate of formation of each of the components would be required to avoid wasteful synthesis of one of them. The most effective means of coordination would be to combine the involved genetic loci into a single unit codin g for a single polypeptide.
Another advantage accruing to such a unified system might be reduction or elimination of the conformational flexibility which is apparent.ly a characteristic of interacting components (8, 13). Thus the catalytically active sites of the unified system would be