| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
(Received for publication, September 27, 1994; and in revised form, December 5, 1994) From the
Although substitution of tyrosine, phenylalanine, tryptophan, or
arginine for leucine 22 in human dihydrofolate reductase greatly slows
hydride transfer, there is little loss in overall activity (k
Many variants of mammalian dihydrofolate reductase (EC 1.5.1.3,
DHFR In much of this work, cDNA for L22R mouse DHFR has been used to
confer MTX resistance on
cells(1, 3, 4, 9, 12, 13, 14, 15) ,
but this variant has low catalytic efficiency(16) . Other
variant DHFRs that have been examined as such selectable markers
include F31W mouse DHFR(5) , L22F hamster DHFR(6) , and
F31S and F34S hDHFR(10) , and other variants of mouse
DHFR(17) . However, detailed kinetic results have not been
published for any of the mouse or hamster DHFR variants, and F34S hDHFR
has low catalytic efficiency(18) . The fact that they provide
only modest protection until amplification has generated many copies of
the cDNA per cell(6, 8, 10, 19, 20, 21, 22) suggests
that they are not ideal as selectable markers. An ideal variant DHFR
for conferring protection on transfected cells must not only have a
very low affinity for MTX compared with the wt enzyme, but must also
provide sufficient tetrahydrofolate for the needs of the cell. That is,
at the usual intracellular level of DHFR, it should provide catalytic
activity comparable with that of wt DHFR under normal conditions. This
would obviate the need for amplification of the mutant cDNA after its
integration into the cell genome. Further, the variant protein must be
sufficiently stable that the concentration of the active form in the
cell can be maintained without an increased rate of expression. Since previous studies (17, 23, 24) have
indicated that variants of mammalian DHFR with substitution of bulky
amino acids for Leu
In
the case of variants with low K
K
Computer simulations of the reaction
course were performed as described previously(25) , with the
use of k In some experiments, NADPD replaced NADPH in
the appropriate syringe. The isotope effect on k
In many cases, the
thermodynamic equilibrium constant, k
where the asterisk indicates the nondissociating conformer. The
thermodynamic equilibrium constant for this isomerization step, K The apparent dissociation rate constant for release of
MTX (k
Since the crystals were isomorphous with
those of wt hDHFR ternary complex with MTXT(35) , hDHFR
coordinates from this structure were used to calculate phases for the
structure determination. Refinement was continued using the restrained
least square program PROLSQ (36, 37) in combination
with the model building program CHAIN(38) . Current refinement
for these structures is nearly complete to the resolution of the data,
and the detailed analysis of these structures will be reported
elsewhere.
Figure 1:
Dependence of activity on pH. Initial
rates observed in 100 µM dihydrofolate, 100 µM NADPH in MATS buffer at the indicated pH at 20 °C. For L22W
and L22R, lines show the fit of data to: activity = a/(1 + 10
In the case of the L22W and L22R variants, the close agreement
between k
Affinity of the unliganded
variant enzymes for MTX is only slightly lower than that of wt (Table 2), except in the case of L22R where K
The
magnitude of the affinity decreases for L22W for inhibitors are
intermediate between those for L22Y and L22F, with the notable
exception of the affinity of L22W for piritrexim, which is decreased by
only a small factor. The affinity of L22R for MTX is decreased by a
factor comparable to that for the other variants, but the affinity of
L22R for the other inhibitors is decreased more than for any other
variant.
Two pieces of evidence
indicate that formation of the ternary complex cannot be a simple
binding reaction. First, k In the direct
measurement of MTX binding to enzymes by stopped-flow fluorimetry, only
one reaction phase, corresponding to the binding step, was observed. An
expected, second phase corresponding to isomerization of the initial
complex, and consequent additional ligand binding, was not observed
even under conditions where it was calculated from the constants in Table 4that additional binding due to the isomerization accounted
for a large fraction of the total binding. However, this second phase
would only be observed under the experimental conditions if k
The crystal structures for the
complexes are compared with the wt hDHFR The active site regions of the three complexes are shown
in Fig. 2Fig. 3Fig. 4. As shown in Fig. 2,
the largest differences between L22R hDHFR
Figure 2:
Crystal structure of L22R hDHFR.
Stereoscopic view of the active site region of the crystal structure of
L22R hDHFR
Figure 3:
Crystal structure of L22F hDHFR.
Stereoscopic view of the active site region of the crystal structure of
L22F hDHFR
Figure 4:
Crystal structure of L22Y hDHFR.
Stereoscopic view of the active site region of the crystal structure of
L22Y hDHFR
As noted (Table 5), in
the structures of L22Y hDHFR
Since there
appears to be a full isotope effect on k
The additional apparent pK
There is a moderate decrease in the affinity of the L22Y apoenzyme
for NADPH and NADP (Table 2). This may be related to loss of the
binding interactions of CD1 of Leu The
mutation-dependent increases in binary K
The decrease in the affinity of MTX for
the variant apoenzymes is quite small compared with that for
wt apoenzyme: it is only the affinity of MTX for the enzyme The lack of cooperativity in the ternary
MTX complexes of the variants with substitutions for Leu The decrease in affinity of MTX for
hDHFR in the ternary complex can be due to a decrease in k
If the interactions
with MTX of the residues at position 22 are first compared, there seems
to be little difference in the number of close contacts in the L22Y and
wt complexes, but there is virtually no interaction between the
Arg The molecular basis of the
greatly increased k It seems probable that contributions from these and
other small structural effects of the mutations together produce the
overall increase in k
The atomic
coordinates and structure factors have been deposited in the Protein
Data Bank, Brookhaven National Laboratory, Upton, NY.
Volume 270,
Number 10,
Issue of March 10, 1995 pp. 5057-5064
©1995 by The American Society for Biochemistry and Molecular Biology, Inc.
KINETICS, CRYSTALLOGRAPHY, AND POTENTIAL AS SELECTABLE MARKERS (*)
ABSTRACT
INTRODUCTION
EXPERIMENTAL PROCEDURES
RESULTS
DISCUSSION
FOOTNOTES
ACKNOWLEDGEMENTS
REFERENCES
) at pH 7.65 (except for the arginine 22
variant), but K
for dihydrofolate and
NADPH are increased significantly. The greatest effect, decreased
binding of methotrexate to the enzyme-NADPH complex by 740- to
28,000-fold due to a large increase in the rate of methotrexate
dissociation, makes these variants suitable to act as selectable
markers. Affinities for four other inhibitors are also greatly
decreased. Binding of methotrexate to apoenzyme is decreased much less
(decreases as much as 120-fold), binding of tetrahydrofolate is
decreased as much as 23-fold, and binding of dihydrofolate is decreased
little or increased. Crystal structures of ternary complexes of three
of the variants show that the mutations cause little perturbation of
the protein backbone, of side chains of other active site residues, or
of bound inhibitor. The largest structural deviations occur in the
ternary complex of the arginine variant at residues 21-27 and in
the orientation of the methotrexate. Tyrosine 22 and arginine 22
relieve short contacts to methotrexate and NADPH by occupying low
probability conformations, but this is unnecessary for phenylalanine 22
in the piritrexim complex.
) are known that have a low affinity for the inhibitor
methotrexate (MTX). Vectors bearing cDNA for such variants have been
used to confer resistance to MTX on cells in culture (1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11) and in
intact
animals(4, 11, 12, 13, 14) .
When cells are transformed or transfected with vectors carrying the
appropriate cDNA, the resulting MTX resistance can be used to select
transformed cells from the total population by exposure to
MTX(6, 8, 9, 10, 15) . 
may be good selectable markers, we
present here detailed analysis of four such variants of hDHFR that we
have produced by site-directed mutagenesis and expression in Escherichia coli. All of these variants exhibit good stability
and markedly decreased affinity for MTX. Moreover, only the Arg variant suffers from a large decrease in catalytic efficiency at
physiological pH. X-ray crystal structures have revealed the molecular
changes that are present in three of the variants.
Materials
Unless otherwise noted, material used
were as described previously(25, 26, 27) .
Methotrexate and aminopterin were from Sigma.
10-Deaza-10-ethylaminopterin, trimetrexate, and piritrexim were gifts
from Ciba-Geigy, Warner Lambert-Parke Davis, and BurroughsWellcome,
respectively. The pDS5/hDHFR plasmid was a gift from Dr. D.
Stüber of Hoffmann La Roche.Construction and Expression of Mutants of
hDHFR
Mutations were introduced into the cDNA and verified by
sequencing as described previously(28, 29) . After
expression in M15(pREP4) E. coli, the enzyme was purified as
described previously(29) .Purity and Concentration of Enzyme
Polyacrylamide
gel electrophoresis of each of the purified variant enzymes showed a
single band after staining with Coomassie Blue. The concentration of
the binding sites of the purified enzyme was determined by MTX
titration of protein fluorescence at 330 nm (excitation at 280) with
monitoring on an SLM Aminco 8000 spectrofluorimeter.Steady State Kinetics
Unless otherwise indicated,
hDHFR activity was assayed in the presence of 100 µM dihydrofolate, 100 µM NADPH in MATS buffer (25 mM MES, 25 mM acetate, 50 mM Tris, 100 mM sodium chloride, and 0.02% sodium azide) pH 7.65, at 20 °C as
described (29) using a Beckman DU-7500 spectrophotometer.
values for
dihydrofolate, the steady state parameters k and K were determined by the use of an Applied
Photophysics stopped-flow spectrophotometer. Solutions in both syringes
contained 0.29 M 2-mercaptoethanol in MATS buffer, pH 7.65 at
20 °C. The resulting velocities were fit to the Michaelis-Menten
equation using the Kaleidagraph program (Abelbeck Software). K for NADPH was determined similarly. For the L22F
variant, the dependence of activity on the concentration of the
substrates was determined in the same way, but data were fit to the
equation for substrate inhibition(18) .Determination of Enzyme Stability
The enzyme was
incubated at 37 °C in a solution consisting of 250 nM enzyme, 1 mM NADPH, 100 mM potassium phosphate,
pH 7.4. Samples were removed at intervals and assayed for enzymic
activity under standard conditions.K
K
for Inhibitors for various inhibitors was determined as described
previously(30) . K
is
obtained from the steady state activity at various inhibitor
concentrations and since K/K =
1 + [dihydrofolate]/K, K for dihydrofolate must be known. In the case of
the L22F variant, an ``operational'' K was approximated by fitting activity at various dihydrofolate
concentrations up to 10 µM to the Michaelis-Menten
equation, without taking account of the substrate inhibition occurring.
It should be noted that 10 µM is the concentration of
dihydrofolate at which K values were measured and
is also the concentration at which activity is a maximum.Equilibrium Dissociation Constants
The equilibrium
dissociation constants (K
) for the binary
complexes of dihydrofolate and MTX were determined by titration of
protein fluorescence at 330 nm with ligand in MATS buffer, pH 7.65 at
20 °C (30) . Excitation was at 280 nm. Determination of K for binding of MTX to the enzyme-NADP complex
was performed similarly. For determination of K for the binary complex of tetrahydrofolate, 0.29 M 2-mercaptoethanol was included in the solution. for the binary complex with NADPH was
determined from the transfer energy fluorescence at 435
nm(31) . K for binding of MTX to the
enzyme-NADPH complex was determined by measuring MTX quenching of this
transfer energy.Determination of the Rate Constant for Chemical
Transformation
The chemical transformation rate constant (k
) was determined by a modification of the
method used previously(32) . Final concentrations were: 2.5
µM enzyme, 100 µM NADPH, 0.29 M 2-mercaptoethanol, and 0.2-1.5 µM dihydrofolate
in MATS buffer at pH 7.65. In a parallel series of experiments, final
concentrations were: 0.2-1.5 µM NADPH, 100
µM dihydrofolate, and 2.5 µM enzyme.
Independent values of k were obtained in the
two series of experiments. The data were fit to a second order binding
reaction plus the chemical transformation step by a modification of the
CRICF program(33) .
, k
, and k values obtained from the data analysis and
experimental concentrations. A few data sets that did not agree well
with simulation were discarded, and the data fitting routine was
repeated without them. (
k) was obtained by calculating the
ratio of k in the presence of NADPH to k in the presence of NADPD.Determination of Methotrexate Binding Rates
The
rate constants for MTX binding to the enzyme
NADPH complex (k) and release from the
enzymeNADPHMTX complex (k) were
determined by stopped-flow fluorometric methods described
previously(30, 33) ./k, is much higher than
the corresponding values of K. This can be
accounted for by assuming isomerization of the initial ternary complex
to a nondissociating conformer(31) , according to the equation:
(k/k
),
can be calculated by the equation: K = (k/k K)
- 1.
) from both isomers of the
DHFRNADPHMTX complex was determined by competition with
10-deaza-10-ethylaminopterin as described
previously(26, 30) .Crystallization and Structure Analysis
Methods
Crystals of each variant complex were grown with a molar
excess of NADPH and inhibitor together with the enzyme in 0.1 M phosphate buffer, pH 8.0, and 62% ammonium sulfate. Hanging drop
experiments were carried out at room temperature. All crystals grew in
a rhombohedral system, space group R3 with hexagonal indexing,
isomorphous with the wt hDHFR binary (34) and ternary (35) complexes. L22Y hDHFRMTXNADPH: a = 87.496, c = 76.733 Å; resolution,
1.90 Å; R = 19.1%, 12027 reflections from
8.0-1.90 Å. L22F hDHFRPTXNADPH: a = 87.002, c = 76.728 Å; resolution
2.0 Å; R = 16.4%, 10046 reflections from
8.0-2.0 Å. L22R hDHFRMTXNADPH: a = 86.058, c = 77.677 Å; resolution,
2.2 Å; current refinement, R = 21.0%, 6383
reflections from 8.0-2.3 Å. Data were collected on a Raxis
IIc imaging plate system.
Expression and Purification of Variants
DHFR
constituted about 7% of the soluble protein in the extracts of E.
coli. The purified enzymes were homogeneous as judged by gel
electrophoresis following staining with Coomassie Blue. The
concentration of the binding sites as determined by titration with MTX
is in good agreement (>90%) with the concentration of the enzyme
determined both by UV absorbance at 280 nm (27) and
colorimetric protein assay(39) . Tightly bound folate and
degradation products were completely removed by preparative isoelectric
focusing as judged by a ratio of absorbance at 280 nm to that at 320 nm
> 20. All of the variants have reasonable stability under the
conditions used. L22Y and L22F have half-lives comparable to that of wt
hDHFR. L22W is slightly less stable than the wt enzyme, while L22R has
a half-life 2.6 times that of wt hDHFR.Steady State Kinetics
Values of K for NADPH and dihydrofolate and k are
given in Table 1. Initial velocity for the L22F variant decreased
as the dihydrofolate concentration was increased above 10
µM. Substrate inhibition has been observed with other
hDHFR variants(18) , the data were fit to the equation used
previously, and the two values of k and K were calculated as described
previously(18) . By analogy with other variants which exhibit
substrate inhibition(18) , at low dihydrofolate concentrations,
the predominant pathway is probably through ENADPHH
folate 
ENADPH
folate ENADP E ENADPH,
whereas at high dihydrofolate concentrations the pathway going through ENADPHHfolate ENADPHfolate ENADP ENADPHfolate EHfolate predominates. k values for the other variants are all lower than the value for wt
hDHFR, and almost zero for the L22R variant at pH 7.65. K for both substrates is increased for all of the
variants, but the range of K values is smaller
than that for k values. Unlike wt hDHFR, none of
these variants exhibit transient state hysteresis since k is the major rate-limiting step and is common
to both pathways.
Effect of pH on Activity
Despite its low activity
at pH 7.65, the L22R variant has activity comparable to wt and the
other variants at low pH (Fig. 1), and, in fact, all of the
variants display greater pH dependence of activity than does wt. Data
for the L22R and L22W variants fit well to an equation containing a
single pK
(4.9 and 6.0, respectively). The L22F
and L22Y variants require an equation with two pK values: 4.8 and 7.1 for L22Y and 5.2 and 7.7 for L22F. The
pK values obtained from these data probably do
not correspond to true ionization constants, but to the pH at which the
rates of two rate-limiting steps become equal(40) .
) and for L22Y and L22F lines
show the fit of data to: activity = a/(1 +
10
) + b/(1 +
10
). The broken line shows the pH dependence for wt hDHFR.
Rate Constant for the Chemical Transformation
The
rate constant for the chemical transformation step, k, is also given in Table 1. For all four
variants, k is greatly decreased compared with
wt, the decrease being much more than for k. and k indicate
that the chemical transformation step is rate-limiting at steady state,
and this is confirmed by the high values of k which are not significantly lower than those for k. For L22Y, k is
lower than k, and k < k, so that the steady state rate
is limited partly by the chemical transformation step and partly by
some subsequent step, presumably product release. However, for the L22F
variant, k is quite similar to k so that the chemical transformation is
rate-limiting, and a good approximation for k,
the overall rate constant for the faster pathway, is given by k. This permits an approximation of K for dihydrofolate(18) . For the L22R
variant, k, like k,
increases markedly as the pH is lowered, reaching 4.9 ± 0.8
s
at pH 5.0. k at pH 5.0
(2.6 ± 0.2) is not significantly different from k (2.9 ± 0.3), so that the chemical
transformation is rate-limiting.Equilibrium Dissociation Constants
(K
The K) values for both
substrates are greater than those of the wt enzyme (Table 2).
However, since the value of K depends on other
rate constants besides k and k for the binding of the substrate, the equilibrium dissociation
constants, K, for binary substrate or product
complexes were determined (Table 2). The binary K for dihydrofolate is lower than the wt value for three of the
variants, but is more than doubled for the L22F variant. In contrast to
the increased affinity for dihydrofolate, tetrahydrofolate is bound
less tightly by all the variant hDHFRs, with K increasing 5- to 16-fold. NADPH and NADP are bound somewhat less
tightly by the variant DHFRs than by wt.
is 120-fold higher than K for wt. However,
the affinity of MTX for NADP complexes of all the variant hDHFRs is
much lower than for the NADP complex of wt hDHFR (by factors of 20 to
300), and the affinity of MTX for the ternary NADPH complexes is still
less (K 740- to 28,000-fold higher than for wt).Inhibition Constants (K
In all cases,
the affinity of the variant hDHFRs for the inhibitors, as measured by K), is decreased in comparison with wt hDHFR, the
extent of the decrease varying considerably (factors of 33 to 960,000)
with the variant and the inhibitor (Table 3). The decrease in the
affinity of L22Y for aminopterin is less (K increased 240-fold above wt K) than for the
other inhibitors, for which decreases are similar. L22F also has the
smallest decrease in affinity for aminopterin, with affinity for
trimetrexate and piritrexim decreasing less than for MTX and
10-deaza-10-ethylaminopterin. Substrate inhibition complicates
determination of K for L22F, since K is dependent on K. At the
concentration of dihydrofolate present during the determination of K, both pathways would be operational in an
unknown ratio. An operational K, corresponding to
the combined activity of the two pathways, determined as described
under ``Experimental Procedures,'' was therefore used. The
rather small discrepancy between K and K or k/k for MTX binding
to L22F (0.546, 1.86, and 0.602 nM, respectively) suggests
that the error in K is not very large.
Rate Constants for MTX Binding to and Release from
Variant hDHFRs
Values of k and k, determined in binding experiments with
enzymeNADPH, are shown in Table 4. k values for the formation of complexes with MTX fall within a
fairly narrow range and are only slightly less than for the wt complex.
Values of k obtained by this method for MTX
dissociation from its ternary complexes are comparable for the four
variants and about 2500 times that for wt enzyme (Table 4).
Consequently, the dissociation constants obtained by this method for
the initial ternary complex are also quite similar for the four
variants, but much higher than that for wt.
/k 
K; second, k k, the rate constant for the
release of MTX from the ternary complex obtained in competition
experiments (Table 4). Binding appears to be followed by
isomerization of the initial ternary complex to a nondissociating
conformer ()(33) . Values of K, the equilibrium constant for the
isomerization, calculated as indicated under ``Experimental
Procedures,'' are shown in Table 4. < k[MTX].
This condition is likely to occur only at high MTX concentration, where
the amplitude of the second phase would be very small.Crystal Structures
Crystallization of three
variants (L22Y, L22F, L22R) was carried out in the presence of various
inhibitors and NADPH. The three complexes for which structures are
reported here are the first ones for each variant for which
well-diffracting crystals were obtained. All three crystals are ternary
complexes with NADPH also present.MTTNADPH
complex(35) , by least squares superposition of all backbone
atoms (Profit program, G. D. Smith, Hauptman-Woodward Medical Research
Institute Library). The root mean square deviation in backbone position
in the variant complex from those of the wt complex with which it was
paired, were as follows: L22Y hDHFRMTXNADPH, 0.15 Å;
L22F hDHFRPTXNADPH, 0.20 Å; L22R hDHFR
MTXNADPH, 0.25 Å. As these values indicate, there is little
perturbation of the protein structure apart from the known change in
residue 22.MTXNADPH and the
wt ternary complex is observed for the backbone and side chain
positions of residues 21-27, which have an average backbone
deviation of 0.89 Å. This is the largest difference in individual
backbone positions when the variants are compared with the wt. There is
little perturbation of the backbone or side chain structure in the
vicinity of the active site of variants L22F and L22Y ( Fig. 3and Fig. 4). However, the unusual binding geometry
for PTX in the complex with the L22F variant results in a large shift
in Phe
to relieve the steric constraints of interactions
with the methoxy substituents. Similar changes were noted in the
ternary complex of avian DHFR and trimethoprim(41) . The
largest shifts in the backbone of the L22F variant are near residue 63
in a flexible loop region not shown in Fig. 3. The largest
shifts in the backbone in L22Y are near residue 44, also in a flexible
loop region not shown in Fig. 4.
MTXNADPH (solid lines). The
corresponding region of the crystal structure of wt
hDHFRMTXTNADPH (dotted lines) (35) is
shown for comparison.
PTXNADPH (solid lines). The
corresponding region of the crystal structure of wt
hDHFRMTXTNADPH (dotted lines) (35) is
shown for comparison.
MTXNADPH (solid lines). The
corresponding region of the crystal structure of wt
hDHFRMTXTNADPH (dotted lines) (35) is
shown for comparison.
MTXNADPH and L22R
hDHFRMTXNADPH, the side chains at position 22 adopt low
probability conformations in order to avoid unfavorable intermolecular
contacts with the inhibitor and cofactor. In the case of L22F
hDHFR PTXNADPH, the change in binding orientation of PTX
compared with that of MTX permits Phe to adapt a high
probability conformation in the active site.
Effect of Mutations on the Chemical Transformation and
Catalytic Activity
Although many variants of hDHFR have very
large decreases in k, k is often little affected (as L22F, L22Y, and L22W) because
product dissociation is much slower than the chemical transformation
for wt hDHFR. In the case of L22R, the decrease in k at pH 7.65 is so great (k 3 
10
that of wt) that k is also
very low (4 10
that of wt). for
all these variants, the rate constant for hydride transfer is
decreased. One possible mechanism for these effects on k
is a change in distance between the hydride
donor and acceptor atoms, i.e. C4 of the nicotinamide ring of
bound NADPH and C6 of bound dihydrofolate, or C7 of bound folate. It
has been calculated that an increase in the C-C distance from its
optimum of 2.6 Å by 0.1 or 0.3 Å increases the activation
energy for hydride transfer by 0.7 and 5 kcal/mol(42) ,
respectively.pH Dependence of k
The variants all show far greater changes in activity with
changing pH than the wt enzyme does (Fig. 1). Since the
measurement of activity was made at high concentrations of
dihydrofolate and NADPH, the pH effects are unlikely to be due to
changes in substrate binding. Instead, the effects are most likely due
to pH-induced changes in the rate of hydride transfer which increases
as the pH is lowered( for Variant
DHFRs), in Hfolate release which
decreases as the pH is lowered(43) , or in NADP release, or in
combinations of these. If an increase in k with
decreased pH is the major effect, then it appears that the
pK
governing this effect of pH is considerably
lower than for wt hDHFR, for which it is 6.2. This in turn
suggests that the pK for N5 of dihydrofolate bound
in the ternary substrate complex of the variants is significantly lower
than its pK of 6.3 in the wt hDHFR ternary complex (44) . This changed pK in part accounts
for the lowered k at pH 7.65 for the variant
hDHFR. seen for L22Y
and L22F is perhaps due to the effect of pH on Hfolate
dissociation, but the high activity of all four variants at pH < 5,
compared with wt, suggests that k for
Hfolate is higher for the variants than for wt. This is
consistent with the 5- to 23-fold higher K values
for Hfolate binary complexes of the variant hDHFRs than for
the wt complex (Table 2).Effects of Mutations on Binding of Substrates and
Products
As in the case of mutations causing substitution of
smaller residues for Phe(29) or
Phe
(18) , the mutations replacing Leu did not result in as great a decrease in affinity for
dihydrofolate as for Hfolate (Table 2). In fact, as
in the case of substitutions for Phe, the affinity of the
apoenzyme for dihydrofolate was actually greater than that of wt in the
case of L22Y, L22W, and L22R. Whether the greater sensitivity of
Hfolate binding to changes in the binding pocket is due to
the puckered reduced pteridine ring of Hfolate, to the
space occupied by its additional hydrogens, or to some other mechanism,
remains unclear. However, the result is that K for
dihydrofolate is not increased as much as would otherwise be the case. with C5 of the
nicotinamide ring of NADPH in the wt hDHFRNADPHMTXT
complex. In the corresponding complex of the L22Y variant, little
interaction occurs between Tyr and the nicotinamide ring,
the shortest interatomic distance being 4.22 Å. The large
(40-fold) increase in K for the NADPH complex of
the L22W variant may perhaps reflect steric interference with binding
of the nicotinamide ring of NADPH by the tryptophan side chain. values
for Hfolate are larger than those for NADP (Table 2).
If the mutations similarly affect dissociation of the ternary product
complexes, ENADPHfolate, then the
preferred mechanistic pathway will probably be through ENADP, with Hfolate dissociating first, as
in the case of mutants with substitutions for
Phe(18) . Substrate inhibition will occur if k for dihydrofolate association with ENADP is greater than k for NADP
dissociation from this complex and if the subsequent dissociation of
dihydrofolate is rate-limiting. This is the case for L22F hDHFR, where
dihydrofolate inhibition occurs, but not for the other variants. The
higher K values for the ENADP
complexes of the L22W and L22R variants suggests that a higher k for NADP is at least partly responsible for
the absence of substrate inhibition for these variants.Effect of Mutations on Inhibitor Binding
As in the
case of other variants that we have examined in some
detail(18, 29) , variants with substitutions for
Leu exhibit much greater decreases in affinity for
inhibitors than for dihydrofolate or Hfolate ( Table 2and Table 3). Large decreases in affinity occur not
only for MTX but for four other inhibitors that are of clinical
interest (Table 3).NADPH
complex that is greatly decreased by the mutations. In the case of wt
hDHFR, the affinity of MTX is 480-fold greater for the
enzymeNADPH complex than for the apoenzyme, but in the case of
the variant enzymes, complex formation with NADPH has little effect on
MTX affinity (Table 2). These results are in contrast to those
for F31G hDHFR(29) . The affinity of MTX for the apoenzyme of
the latter is decreased 37-fold compared with wt, but because of
greater binding to the ENADPH complex, MTX affinity for
the latter is only decreased 130-fold compared with wt. For the F34A
variant, the contrast with the present results is even greater, since K for the binary MTX complex is increased
8,000-fold compared with wt, and for the ternary complex
18,000-fold(18) . may perhaps be due in part to slightly greater separation between
the nicotinamide and pteridine rings. The distance of C4 of the
nicotinamide ring from C6 of the pteridine ring is 3.93 and 4.27
Å, respectively, for wt and L22Y hDHFR. Similarly, the C4-C7
distances are 4.22 and 4.52 Å, respectively. (For numbering of
MTX atoms, see (45) .), an increase in k, or a
decrease in K (). For all the
variant hDHFRs, k is decreased a little, k is greatly increased (2,300 to 4,100-fold)
with the result that k/k is greatly increased in all cases. For all the variants, the
occurrence of isomerization strengthens binding, especially in the case
of L22Y and L22W, and only in the case of L22R is K less than for wt.Relation of Changes in MTX Binding to Structure
It
is of interest to examine the crystallographic structures with a view
to providing molecular interpretations of the decreased affinity of MTX
for the variant hDHFRs. The most valid comparisons are of the wt
hDHFRNADPHMTXT structure with that of L22Y
hDHFRNADPHMTX, and with that of L22R
hDHFRNADPHMTX. Decreased affinity is due primarily to an
increase in k, and this might be due to
decreased affinity between MTX and active site side chains, or to
increased diffusion of MTX from the active site. side chain and bound MTX. In the wt complex, the
closest interactions are between C7, C6, and N8 of MTX with CD1, CD2,
and CD2 of Leu, respectively, with separations of 4.41,
3.60, and 4.07 Å, respectively. In the case of the L22Y complex,
the closest contacts are between the 10-methyl group of MTX and CD2 and
CE2 of Tyr, and between C7 of MTX and CD2 of
Tyr, with separations of 4.27, 4.58, and 3.99 Å,
respectively. By contrast, the closest approach of atoms of the
Arg side chain and MTX is for CG of Arg and the 10-methyl
of MTX (5.37 Å). Thus binding energy is lost at this residue for
the L22R variant and explains some of the loss of affinity of MTX for
this variant, but not for the L22Y variant. There is no clear evidence
for decreased interaction of side chains of other residues, such as
Phe and Phe. for MTX dissociation from
complexes of the variants might also be sought in structural factors
that facilitate the diffusion of the inhibitor out of the binding site.
McTigue et al.(45) have suggested that diffusion of a
pteridine out of the binding site requires the phenyl side chain of
Phe to rotate about the C
-C
bond from the
position in the ternary complex, so that the pteridine ring can exit
between Phe and Leu. This rotation of
Phe should be unhindered in the variants, and indeed in
the crystal structure of the L22F complex (Fig. 3) the
Phe side chain is seen in its alternative position. It
should be noted, however, that the Phe and Tyr side chains are considerably closer to Phe than is
Leu. The activation energy for MTX dissociation may also
be lowered somewhat by the adoption of lower energy conformations by
Tyr and Arg when the pteridine binding site
is emptied..Suitability of Variants for Conferring Resistance to MTX
on Cells
The variants with substitutions for Leu appear eminently suitable for making cells MTX-resistant
according to many criteria: they have good stability in vitro;
three of them have inhibition constants (K)
considerably higher than those of the F31S and F31G variants of hDHFR (K 0.24 and 0.35 nM, respectively), both
of which have been shown to confer significant protection to cells in
culture(10, 11, 46) ; and they have catalytic
efficiencies (k/K) that
compare favorably with that of F34S hDHFR (18) which has also
been reported to confer resistance on cells(10) . k/K values for L22Y, L22F,
L22W, and L22R are 12, 15, 10, and 0.03 s µM, respectively, and for F34S is
0.017 s µM. Our
initial results with the L22Y variant (46) indicate that
transfection of the cDNA into mouse fibroblasts on a retroviral vector
does indeed afford considerable protection from MTX.
)folate,
7,8-dihydrofolate; Hfolate,
(6S)-5,6,7,8-tetrahydrofolate; aminopterin,
4-amino-4-deoxyfolic acid; MTX, methotrexate,
4-amino-4-deoxy-10-methylfolic acid; MTXT,
4-amino-4-deoxypteroyl-
-(1H-tetrazol-5-yl)-L--aminobutyric
acid; PTX, piritrexim,
2,4-diamino-6-(2,5-dimethoxybenzyl-5-methylpyrido[2,3-d]pyrimidine;
trimetrexate,
2,4-diamino-5-methyl-6-[(3,4,5-trimethoxyanilino)methyl]quinazoline;
MES, 2-(N-morpholino) ethanesulfonic acid; NADPD,
(4R)-[H]NADPH; L22X, variant
DHFR in which the leucine at position 22 has been replaced with the
amino acid X; k, rate constant for
chemical transformation step.)
We thank J. Clay McCastlain for producing and cloning
three of the mutant cDNAs.
©1995 by The American Society for Biochemistry and Molecular Biology, Inc.
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  E. Fossati, J. P. Volpato, L. Poulin, V. Guerrero, D.-A. Dugas, and J. N. Pelletier 2-Tier Bacterial and In Vitro Selection of Active and Methotrexate-Resistant Variants of Human Dihydrofolate Reductase J Biomol Screen, July 1, 2008; 13(6): 504 - 514. [Abstract] [PDF]
|
|
|
|
 |
|
 N. Skacel, L. G. Menon, P. J. Mishra, R. Peters, D. Banerjee, J. R. Bertino, and E. E. Abali Identification of Amino Acids Required for the Functional Up-regulation of Human Dihydrofolate Reductase Protein in Response to Antifolate Treatment J. Biol. Chem., June 17, 2005; 280(24): 22721 - 22731. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
