The bacterial tryptophan synthase 
complex (EC 4.2.1.20) is a useful system for investigating
protein-protein interaction and the mechanism of subunit assembly (for
reviews see (1, 2, 3, 4, 5, 6) ).
The complex dissociates into two
subunits and one subunit. The monomeric
subunit catalyzes the reaction. The dimeric subunit contains one pyridoxal phosphate at each active site and
catalyzes the pyridoxal phosphate-dependent reaction. The
complex catalyzes the and
reactions and the overall reaction, which is formally
the sum of the and
reactions.
The three-dimensional structure of the tryptophan synthase
complex from Salmonella
typhimurium(7) reveals that the and subunits are arranged in a nearly linear
order. The active sites of the and subunits are 
25
Å apart and are connected by a tunnel that passes through the
/ interaction site and between the two structural domains of
the subunit. The pyridoxal phosphate coenzyme is
located at the interface between the two structural domains (N-domain
and C-domain) of the subunit at one end of the tunnel
and interacts with residues from both domains. The carbonyl group of
pyridoxal phosphate forms an internal aldimine with Lys-87 in the
N-domain. The phosphate group of the coenzyme forms hydrogen bonds with
Gly-232, Gly-233, Gly-234, Ser-235, Asn-236, and Ala-237 in the
C-domain. The pyridine ring nitrogen atom, N1, is close to the
sulfhydryl of Cys-230 and within hydrogen bonding distance of the
hydroxyl of Ser-377, both residues in the C-domain.
Early studies of
the association of the and subunits of
tryptophan synthase using sucrose gradient centrifugation showed that
pyridoxal phosphate partially stabilized the
complex(8) . The
apo- complex (minus pyridoxal
phosphate) was totally dissociated, the
holo- complex (plus pyridoxal
phosphate) was partially dissociated, and the holo-
complex in the presence of L-serine was stable. Pyridoxal phosphate and L-serine
also increased the apparent subunit association constants measured
under assay conditions(8) . The reaction of L-serine
with the complex resulted in a
400-fold increase in affinity of the subunit for the
subunit(9) .
In the present work we ask whether
binding pyridoxal phosphate per se increases subunit affinity
or whether formation of the internal aldimine with subunit lysine 87 is required. To answer this question, we use a
mutant form of the subunit having the lysine 87
replaced by threonine (K87T)(10) . Studies with the K87T
complex demonstrated that lysine 87
serves critical roles in transimination, catalysis, and product
release(11) . The K87T complex is inactive in the reaction but retains activity in
the reaction. The mutant enzyme binds pyridoxal phosphate as the
free aldehyde and forms Schiff base intermediates (external aldimines)
with L-serine (K87T-Ser) or L-tryptophan (K87T-Trp)
at the site. Crystallographic analyses of the external aldimines
formed by the K87T complex with L-serine and L-tryptophan have been carried out at
2Å resolution (
)and have localized the substrate
and product binding sites in the subunit(12) . The crystallographic results show that L-serine and L-tryptophan are bound between the N-
and C-domains and interact with some residues of the N-domain. The
results presented here provide evidence that formation of interdomain
bridges by pyridoxal phosphate-aldimine intermediates stabilizes
subunit association and stabilizes pyridoxal phosphate binding.
EXPERIMENTAL PROCEDURES
Chemicals and Buffers
Indole-3-glycerol
phosphate was synthesized enzymatically and isolated by ion exchange
chromatography on DEAE-Sephadex A25 as described(13) .
Pyridoxal phosphate, L-serine, NAD, and
glyceraldehyde-3-phosphate dehydrogenase were from Sigma. All
experiments utilized Buffer B (50 mM sodium N,N-bis(2-hydroxyethyl)glycine containing 1 mM EDTA
at pH 7.8). Phenylhydrazine hydrochloride was from Eastman.
Enzymes
Wild type (10) and K87T (11) forms of the tryptophan synthase
complex and wild type and
holo- subunits (14) from S. typhimurium were isolated and purified to homogeneity. The apo- subunit was prepared by treatment of the holo- subunit with 10 mM hydroxylammonium
chloride(15) . The K87T subunit was prepared
from K87T complex by heat
denaturation of the subunit as described for the wild type
complex(16, 17) .
The reduced subunit and complex were obtained by treating the wild type subunit and complex with 10
mM sodium borohydride for 30 min followed by dialysis. The
K87T complex and subunit as isolated contain some tightly bound L-serine
(about 0.03 and 0.1 mol/mol of protomer,
respectively)(11) . Schiff base derivatives of the K87T
complex or subunit
with L-serine (K87T-Ser) were prepared by incubating the
mutant enzyme for 24 h at 23 °C with 40 mML-serine (11) followed by gel filtration on a
PD-10 column (Pharmacia Biotech Inc.) in Buffer B. Protein
concentrations were determined from the specific absorbance at 278 nm
of the complex (E
= 6.0), the holo- subunit (E = 6.5), the apo- subunit (E = 5.8), or of the
subunit (E = 4.4)(17) . Protein
concentrations of the K87T-Ser complexes were determined by the BCA
protein assay reagent (Pierce) using purified subunit
or complex as standard.
Spectroscopic and Analytical Methods
Absorption
spectra were made using a Hewlett-Packard 8452 diode array
spectrophotometer. Time course measurements at single wavelengths were
made using a Cary 118 spectrophotometer. Gel filtration experiments
were performed using a prepacked Superose HR12 10/30 column (Pharmacia)
on a LCC-501 Plus FPLC system (Pharmacia) at a flow rate of 0.5 ml/min
at 23 °C. Pyridoxal phosphate content was determined using a
phenylhydrazine reagent (2% phenylhydrazine hydrochloride (w/v) in 10 N HSO
)(18) . SDS-gel
electrophoresis was carried out on the Phastgel system (Pharmacia)
using 10-15% gradient gels.
Enzyme Assays and Subunit Interchange
Experiments
One unit of activity in any reaction is the
formation of 0.1 µmol of product in 20 min at 37 °C. Activity
in the reaction was measured by a direct spectrophotometric
assay(17) . The activities of the complex in the reaction were measured by spectrophotometric
assays coupled with D-glyceraldehyde-3-phosphate
dehydrogenase(19) . The kinetics of dissociation of the
complex was determined by subunit
interchange experiments that utilize a two-step system(9) . The
complex (70 µM protomer in Buffer B) was mixed with 20 volumes of
subunit (70 µM protomer in
Buffer B) at 25 °C. Aliquots (0.05 ml) were removed at various time
intervals and added to 0.95 ml of enzyme assay mixture at 37 °C.
When inactive (reduced or K87T) complex was mixed with active subunit, increase
in activity in the reaction was determined. When active wild
type complex was mixed with inactive
(reduced or K87T) subunit, decrease in activity in
the reaction was determined. In some cases, 40 mML-serine was premixed with one or both enzymes.
RESULTS
Stability of Enzyme-bound Pyridoxal Phosphate to
Dialysis
Fig. 1shows the effects of dialysis for up to
10 h on the pyridoxal phosphate content of wild type and K87T
tryptophan synthase complex (A) and subunit (B). Pyridoxal
phosphate remains largely bound to the wild type
complex (Fig. 1A, curve1). The pyridoxal phosphate content of the wild
type subunit decreases from 1.0 to 0.8 mol of
pyridoxal phosphate/mol of protomer after 10 h (Fig. 1B, curve 1). In contrast, the pyridoxal
phosphate content of the K87T complex and subunit decreases to about 0.4 and
0.1 mol/mol of , respectively, after 10 h (Fig. 1, A and B, curves2). Thus, formation of
the internal aldimine with lysine 87 in the wild type enzymes
stabilizes the enzymes to loss of pyridoxal phosphate upon dialysis. L-Serine reacts with the K87T subunit and
complex to form an external aldimine
with pyridoxal phosphate. Formation of this external aldimine markedly
stabilizes the K87T complex (Fig. 1A, curve3) and K87T
subunit (Fig. 1B, curve3) to loss of pyridoxal phosphate upon dialysis. The L-serine moiety in the external aldimine may stabilize the
pyridoxal phosphate by interacting with residues in the substrate
binding site of the subunit. This hypothesis is
consistent with our previous finding that formation of the external
aldimine with L-serine by the K87T
complex greatly increases the
ellipticity band of the coenzyme(20) .
Figure 1:
Effect of time of dialysis on the
pyridoxal phosphate content of various forms of tryptophan synthase.
Solutions of the holo- complex (A) and of the holo- subunit (B) at
1 mg/ml were dialyzed against Buffer B at 4 °C. Aliquots were
analyzed at intervals for protein concentration and for pyridoxal
phosphate content. Curve1, wild type; curve2, K87T; curve3, K87T-Ser
complex.
Gel Filtration Analysis of Different Forms of Tryptophan
Synthase
We have monitored the association state of the
and subunits from the wild type, the reduced wild
type, and the K87T tryptophan synthase by high performance gel
filtration on a Superose column with an FPLC system (Fig. 2).
The wild type holo- complex elutes
largely as a single peak at 9.75 ml, whereas the isolated wild type
holo- subunit and subunit chromatographed
separately elute at 10.35 and 12.3 ml, respectively (Fig. 2A). The holo- complex has been observed previously as a single species by
ultracentrifugation (21, 22) , by sucrose gradient
centrifugation (8, 22) and by gel
filtration(8, 21, 23) . The wild type
apo- complex is completely separated
into the apo- subunit and subunit in this system
(data not shown) as demonstrated previously for gel filtration of the
apo- complex on Sephadex G100
Superfine (15) in the same buffer.
Figure 2:
Superose gel filtration of various forms
of wild type and mutant (K87T) tryptophan synthase. A, wild
type holo- complex (1 mg), holo
subunit (1 mg), and subunit (0.38 mg) were
injected in separate runs. B, wild type subunit (0.42
mg) was preincubated for 3 h at 23 °C with approximately equimolar
reduced subunit (0.66 mg), K87T subunit (0.6 mg), or K87T subunit plus 50
mML-serine before injection in separate runs. C, wild type subunit was preincubated with 2 eq of
protomer to give the complex; the K87T
complex was injected as in A.
Mixing the subunit
with 2 or more eq of holo- protomers results in a stable
complex that can be observed by sucrose
centrifugation (8, 22) or
ultracentrifugation(21) . The wild type holo- complex elutes at 10.0 ml in our high performance gel filtration
system (Fig. 2C). Gel filtration of the K87T
holo- complex yields two peaks at
positions characteristic of the complex and of
the subunit (Fig. 2C). ()The identity
of these two species was confirmed by assay of activity in the
reaction and by SDS-gel electrophoresis (data not shown). Thus the K87T
holo- complex partially dissociates
under these conditions. Similar results were obtained with a
reconstituted K87T holo- complex
prepared by mixing the K87T holo- subunit and 2 eq of
wild type subunit (Fig. 2B). As controls,
reconstituted complexes were also
prepared by mixing holo- subunit or reduced
subunit with 2 eq of subunit. The reconstituted
reduced complex (Fig. 2B) and holo- complex (data not shown) each yielded a major peak at the
position of complex. Gel filtration
of the reconstituted K87T holo- complex in the presence of L-serine also yielded a major
peak at the position of the complex.
Thus L-serine largely prevents dissociation of the K87T
holo- complex to and subunit under these conditions.
Effects of the K87T Mutation on Association with the
Subunit and on Activation of the Subunit
The
association reaction between the and subunits
can be described as shown in .
Fig. 3shows the enzymatic activity of the wild type
subunit in the reaction obtained upon titration with
increasing amounts of wild type holo- subunit, reduced
subunit, and K87T holo- subunit.
Titrations with the reduced and K87T holo- subunit
were also carried out in the presence of L-serine. It is
noteworthy that the maximum specific activity of the subunit in
the reaction (see S in Table 1) varies significantly when
the subunit is complexed with different species of subunit. Early studies also observed that
complexes containing reduced
subunit from Escherichia coli had 2-fold
greater activity than the wild type complex in the reaction(23) . It is known that the
activity of the subunit is sensitive to the conformational state
of the subunit because ligands that bind to the
active site of the subunit alter the kinetics of
reaction at the active site of the subunit 25 Å
distant(13, 19, 24, 25, 26, 27, 28) .
Reduction of pyridoxal phosphate at the active site of the subunit (reduced ) and substitution of threonine
for lysine 87 (K87T ) probably cause alterations in
the conformation of the subunit that are communicated
to the active site of the subunit and alter the activity of the
subunit.
Figure 3:
Titration of the subunit with
various subunits. The activity of the subunit
(0.88 µM) in the reaction was determined in the
presence of various amounts (0-2.64 µM) of the wild
type subunit (
), the reduced subunit (
), or the K87T subunit
(
). Data in the presence of 40 mML-serine was
determined with various amounts of the reduced subunit (
) or the K87T subunit (
).
Specific activity expressed in units/mg subunit is plotted versus the molar ratio of /. Data were analyzed as
described in the text and fit to models 1-3 (see Table 2).
Derived apparent dissociation constants are shown in Table 2.
The data obtained in Fig. 3and additional
data (not shown) for titration with the apo- subunit
were analyzed using the PC-MLAB program (Civilized Software, Bethesda,
MD) and fitted to three possible models (Table 1): 1) the
subunit binds two subunits with equal
affinities (K
= K; both bound subunits are
active); 2) the complex is active,
whereas the complex is inactive; 3) the
subunit binds two subunits with different
affinities (both bound subunits are active). Model 3 gave the
best fit (as judged by the sum of squares) for the wild type and
reduced subunits. (
)The affinities of the
first subunit for the holo- subunit are too high
to measure by this method (K = <0.01
µM; Table 1). The dashedline in Fig. 3shows the theoretical curve for stoichiometric binding of
two subunits to the reduced dimer where the K = <0.01 µM for both
subunits. Analysis showed that the second subunit binds to the
reduced subunit and to the wild type holo- subunit with K = 0.07-0.08
µM (Table 1). In marked contrast, the K87T
holo- subunit exhibits a gradual titration curve that
indicates much weaker and equal affinities for both subunits with K = 0.4 µM (Table 1). In the presence of L-serine, the K87T
holo- subunit again exhibits a sharp biphasic
titration curve, with the second K =
0.05 µM (Table 1). This result shows that L-serine greatly strengthens the association of the K87T
holo- subunit with the subunits. Addition of L-serine to the reduced subunit has no
effect. This is to be expected because the reduced subunit does not react with L-serine. Addition of L-serine to the wild type holo- subunit in
the presence of the subunit and indole-3-glycerol phosphate
results in the overall reaction which is about 20 times
faster than the reaction (data not shown). Association of the
and subunits under the conditions of the
reaction is reported to be very tight (K = 0.4 
10
µM)(8) . The apo- subunit
exhibits much weaker and equal affinities for both subunits with K = 0.3 µM (Table 1).
Subunit Interchange Experiments
The rate of
dissociation of the wild type complex has been measured by determining the loss of activity in
the reaction upon addition of a 20-fold excess inactive
reduced subunit(9) . ()Fig. 4, curve1, shows that the
enzymatic activity decreases rapidly to an equilibrium value. Some
residual activity is observed at equilibrium since the wild type
subunit is not completely displaced by the
20-fold excess of reduced subunit. The rate of
dissociation is greatly reduced in the presence of L-serine (Fig. 4, curve2). Thus, reaction of L-serine with the wild type complex tightens association (8) and decreases the rate
of dissociation (9) as found previously.
Figure 4:
Kinetics of dissociation of the wild type
complex. Dissociation was initiated
by addition of excess reduced subunit in the absence (curve1) or presence (curve2) of
40 mML-serine or by addition of the K87T-Ser
subunit complex in the absence (curve3) or presence (curve4) of 40 mM of L-serine. Activity measured at intervals in the
reaction is shown relative to the activity of the
complex alone (see
``Experimental Procedures'').
Our plan to study
the rate of dissociation of the wild type complex upon addition of excess K87T subunit
was precluded by our inability to obtain completely serine-free K87T
subunit. Trial experiments (not shown) indicated that
the small fraction (10%) of the K87T subunit
that bound L-serine (K87T -Ser) was much more
effective than the serine-free fraction (K87T ) in
binding subunit and causing dissociation of the wild type
complex. Consequently we determined
the rate of dissociation of the wild type complex upon addition of the isolated K87T -Ser
complex in the absence of excess L-serine (curve3) and in the presence of excess L-serine (curve4). Addition of K87T -Ser (curve3) results in more rapid dissociation of the
wild type complex than does addition
of the reduced subunit (curve1).
Thus, K87T -Ser binds the subunit more rapidly
than does reduced . Addition of excess L-serine (curve4) tightens the association
of the wild type complex and
decreases the rate of dissociation of the complex.
The rate of dissociation of an
complex containing inactive
subunit can be measured by determining the increase
in activity upon addition of excess wild type subunit. This method has been used previously for the reaction of
reduced complex with excess wild
type subunit (9) as illustrated in Fig. 5, curve1. We have determined activity
in the reaction because the free subunit
has no activity in the reaction, whereas the subunit does have some activity in the reaction that would
give a high blank value. Addition of excess wild type subunit to the K87T complex
results in 40% activity within the mixing time (Fig. 5, curve2) followed by a slower increase in activity.
When the K87T -Ser complex was mixed
with wild type subunit in the presence of 40 mML-serine (Fig. 5, curve3), the
rate of activation was much slower in the first 5 min than in curve2, but full activity was achieved after about 30 min. The
reaction of the K87T -Ser complex
with excess subunit (Fig. 5, curve4) showed a decreased rate of activation similar to that
in curve3. The results indicate that the first
subunit in the K87T complex
dissociates readily in the absence of L-serine but more slowly
in the presence of L-serine. The very slow rate of subunit
interchange with the K87T -Ser
complex (Fig. 5, curve4) implies that the
K87T -Ser dissociates very slowly.
The slow rate of subunit interchange with the K87T
complex (Fig. 5, curve2) may result from the 3% contamination of the
enzyme with L-serine (see ``Experimental
Procedures'').
Figure 5:
Kinetics of dissociation of the reduced
complex and K87T
complex. Excess wild type
subunit was added to initiate dissociation of the
reduced complex (curve1), of the K87T complex in the absence (curve2) or presence (curve3) of 40 mML-serine, or of
the K87T-Ser complex (curve4). Activity measured at intervals in the
reaction is shown relative to the activity of the wild type
complex (see ``Experimental
Procedures'').
DISCUSSION
Our studies of a mutant form of tryptophan synthase, which is
unable to form the internal aldimine between subunit
Lys-87 and pyridoxal phosphate demonstrate that the internal aldimine
is very important for stabilizing cofactor binding and subunit
association.
Stabilization of Cofactor Binding
Pyridoxal phosphate is
removed by dialysis more slowly from the wild type subunit and complex than from
the K87T subunit and K87T
complex which cannot form the
internal aldimine (Fig. 1). Thus, formation of the internal
aldimine with Lys-87 in the wild type enzymes stabilizes the
subunit and complex to loss of pyridoxal phosphate upon dialysis. Our finding
that formation of the external aldimine with L-serine markedly
stabilizes the K87T subunit and
complex (Fig. 1) to loss of
pyridoxal phosphate upon dialysis suggests that L-serine
stabilizes the bound pyridoxal phosphate by interacting with residues
in the substrate binding site of the subunit and
forming a bridge between the N- and C-domain (see below).Binding of
pyridoxal phosphate and pyridoxal phosphate analogues to the wild type
subunit and complex has been studied
extensively(29, 30, 31, 32) .
Pyridoxal phosphate is bound cooperatively to the apo- subunit and noncooperatively to the
apo- complex. The slow rate of
removal of pyridoxal phosphate from the wild type
complex by dialysis probably results
from the very slow reversal of the final isomerization step in
pyridoxal phosphate binding(32) . The rate of removal of
pyridoxal phosphate from the subunit and
complex by dialysis has not been
compared previously. However, we have reported that the pyridoxal
phosphate oxime is much more readily removed from the subunit than from the complex(15) . We have suggested that interaction of the
subunit with the subunit may stabilize the
interaction between the N- and C-domains of the subunit and thus slow removal of pyridoxal phosphate and
pyridoxal phosphate derivatives that are bound between the two
domains(2) .
Stabilization of Subunit Association
-Early
studies using sucrose gradient centrifugation showed that association
of the and subunits of tryptophan synthase was
weak in the absence of ligands, intermediate in the presence of
pyridoxal phosphate, and strong in the presence of pyridoxal phosphate
and L-serine (8) (Table 2). Our gel filtration
results (Fig. 2) and activity measurements ( Fig. 3and Table 1) demonstrate weak association between the subunit
and a mutant form of the subunit (K87T) that binds
pyridoxal phosphate as the free aldehyde and not as an internal
aldimine. Thus the holo-K87T complex
resembles the apo- complex in having
weak subunit association (Table 2). The reaction of L-serine with the K87T complex increases the affinity as shown by gel filtration (Fig. 2C) and activity measurements ( Fig. 3and Table 1). The K87T -Ser complex
thus resembles the holo- complex and
reduced complex in having
intermediate subunit association (Table 2).Subunit
interchange experiments show that the rate of dissociation of the
holo- complex by reduced subunit is greatly decreased in the presence of L-serine (Fig. 4, curve 2 versus1). Thus, reaction of L-serine with the wild type complex decreases the rate of dissociation (9) and
tightens association (strong association in Table 2)(8) .
Dissociation of the holo- complex by
K87T -Ser is also greatly reduced in the presence of L-serine (Fig. 4, curve 4 versus3).
The similarity of the results with reduced and with
K87T -L-Ser in Fig. 4supports the
classification of these two enzymes in the same group (intermediate
association) in Table 2. The finding that the K87T
complex dissociates readily in the
absence of L-serine but more slowly in the presence of L-serine (Fig. 5) implies that the external aldimine
with L-serine increases association from weak to intermediate (Table 2).
The dissociation constants for various forms of the
complex ( Fig. 3and Table 1) are apparent dissociation constants since they are
measured in the presence of a substrate (indole-3glycerol phosphate)
that may alter subunit affinity. Our new methods of analysis of the
titration curves shows that most forms of the subunit
have higher affinity for the first subunit than for the second
subunit. This result is consistent with other evidence that the
wild type subunit binds the subunit with
negative cooperativity(9, 3
ABSTRACT
