site-directed mutagenesis of mouse dihydrofolate reductase

THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1988 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 263, No. 25, Issue of September 5, pp. 12Mx)-12508,1988 Printed in U. S. A.

Site-directed Mutagenesis of Mouse Dihydrofolate Reductase MUTANTS WITH INCREASED RESISTANCE TO METHOTREXATE AND TRIMETHOPRIM*

(Received for publication, April 13, 1988)

Joelle ThilletS, Josette Absil, Stuart R. Stone#, and Raymond Pictet From the Unit4 257 de 1’Institut National de la Sante et de la Recherche Medicale, Institut Jacques M o d du Centre National de la Recherche Scientifique, Tour 43-2 Place Jussieu, 75005 Paris, France and §Friedrich Miescher Institute, Basel, Switzerland

Site-directed mutagenesis was used to generate mutants of recombinant mouse dihydrofolate reductase to test the role of some amino acids in the binding of two inhibitors, methotrexate and trimethoprim. Eleven mutations changing eight amino acids at positions all involved in hydrogen bonding or hydrophobic interactions with dihydrofolate or one of the two inhibitors were tested. Nine mutants were obtained by site-directed mutagenesis and two were spontaneous mutants previously obtained by in vivo selection (Grange, T., Kunst, F., Thillet, J., Ribadeau-Dumas, B., Mousseron, S., Hung, A., Jami, J., and Pictet, R. (1984) Nucleic Acids Res. 12,3585-3601). The choice of the mutated positions was based on the knowledge of the active site of chicken dihydrofolate reductase established by x- ray crystallographic studies since the sequences of all known eucaryotic dihydrofolate reductases are greatly conserved. Enzymes were produced in great amounts and purified using a plasmid expressing the mouse cDNA into a dihydrofolate reductase-deficient Esche- richia coli strain. The functional properties of recombinant mouse dihydrofolate reductase purified from bacterial extracts were identical to those of dihydrofolate reductase isolated from eucaryotic cells. The K,,,(NADPH) values for all the mutants except one (Leu- 22 -+ Arg) were only slightly modified, suggesting that the mutations had only minor effects on the ternary conformation of the enzyme. In contrast, all K,,, (H,folab) values were increased, since the mutations were located in the dihydrofolate binding site. The catalytic activity was also modified for five mutants with, respectively, a 6-, lo-, 36-, and 60-fold decrease of V,, for Phe-314 Arg, Ile-7 - Ser, Trp 24 -+ Arg and Leu-22 -+ Arg mutants and a 2-fold increase for Val- 115 4 Pro. All the mutations affected the binding of methotrexate and six, the binding of trimethoprim: Ile-7 + Ser, Leu-22 - Arg, Trp-24 + Arg, Phe-31-+ Arg, Gln-35 -+ Pro and Phe-34 -+ Leu. The relative variation of Ki for methotrexate and trimethoprim were not comparable from one mutant to the next, reflecting the different binding modes of the two inhibitors. The mutations which yielded the greatest increases in Ki are those which involved amino acids making hydrophobic contacts with the inhibitor.

* This work was supported by the Centre National de la Recherche Scientifique, the Institut National de la Santk et de la Recherche Mbdicale, and by grants from the Ministire de 1’Industrie et de la Recherche, the Ligue Francaise Centre le Cancer, and the Association pour la Recherche sur le Cancer. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

$ To whom correspondence should be addressed.

Dihydrofolate reductase (5,6,7,8-tetrahydrofolate: NAD- PH’ oxidoreductase, EC 1.5.1.3.) plays a central metabolic role since it catalyzes the NADPH-dependent reduction of dihydrofolate to tetrahydrofolate, an essential cofactor for the synthesis of thymidylate, purines, and some amino acids. Thus, inhibition of dihydrofolate reductase results in deple- tion of the tetrahydrofolate pool with subsequent death of the cell. This inhibition can be achieved by a series of synthetic antifolates, some of which show a high degree of species selectivity. Trimethoprim, for instance, binds about 3000 times more tightly to the Escherichia coli enzyme than to vertebrate enzymes (I), whereas methotrexate has the same affinity for both types of enzymes.

We investigated the possibility of increasing the resistance of a mammalian (mouse) dihydrofolate reductase to one an- tifolate, methotrexate, with a minimal change in the activity of the protein towards its substrate. Such mutants would be of interest for at least two reasons. First, the modified dihydrofolate reductase could be used as a dominant selective marker for transfection of any eucaryotic cells. Only a few systems can act as dominant selective markers and dihydrofolate reductase could have the advantage that it is amplified easily in the cell together with an associated gene (2). Second, a better knowledge of the specific traits responsible for the differences in substrate and inhibitor binding could help to design new synthetic antifolates of greater selectivity than those already known (3). The effect of a series of mutations has been studied on E. coli enzyme (4-7), which has led to a better understanding of the enzymatic mechanism of the bacterial enzyme. Comparison of x-ray crystallographic anal- yses of eucaryotic and procaryotic dihydrofolate reductases (8, 9) has revealed a common three-dimensional structure, as well as a great conservation of the amino acids implicated in the active site, even though the primary structures are less than 25% homologous. The first accurate x-ray crystal structure determined for eucaryotic dihydrofolate reductase was that of chicken dihydrofolate reductase (8). A preliminary comparison between the crystallographic data for mouse (10) and chicken dihydrofolate reductases indicates that their three-dimensional structures were nearly identical (11). Re- cently, the refined structures of mouse L1210 dihydrofolate reductase complexed with NADPH and either trimethoprim or methotrexate have been published (12). These structures reveal that the intercations between inhibitors and mouse dihydrofolate reductase are comparable to those observed for chicken dihydrofolate reductase except for residue 31, which is not strictly conserved between mouse (Phe) and chicken (Tyr) dihydrofolate reductases.

In a previous set of experiments, we expressed a cDNA coding for the mouse dihydrofolate reductase, both in Bacillus subtilis and E. coli, and used this system to select i n vivo

12500

Mutants of Mouse Dihydrofolate Reductase 12501

mutations of the dihydrofolate reductase protein conferring a methotrexate resistance to B. subtilis (13). We have now used the same system to study the effect of a series of amino acid changes on the binding of inhibitors to mouse dihydrofolate reductase.

EXPERIMENTAL PROCEDURES

Strains-E. coli JMlOl (Alac pro, SupE, thi, F' tra D36, pro AB, lac lq, Z AM15) was used for infection with M13. A streak of JMlOl was maintained at 4 'C for approximately 1 month on a M9 + B1 plate. Overnight cultures were grown in YT medium.

E. coli HBlOl (F', hsdS20, (r-B, m-B), rec A13, ara-14, proA2, lac Y1, galK2, rpsL2O (Sm'), xyl-5, mt-1, supE44,X) was used as a recipient for transformation and purification of plasmids.

E. coli C600 mut L (F-, thi-1, thr-1, leu B6, lac Y1, ton A21, sup E44, A-, tn5, mutL211) was used for transformation with the elon- gation mixture. This strain was a gift from M. Radman (Institut Jacques Monod, Paris) (14).

E. coli D3-157 (F-, guaB22, xyl-7, rpsL125, fol-200, xyl+), a gift from S. Singer (Wellcome Research Laboratories, Research Triangle Park, NC) (15), was used for preparation of crude extracts and further purification of the proteins after transformation with the desired plasmids. This strain lacks bacterial dihydrofolate reductase which avoids contamination of the mutant mouse dihydrofolate reductases. It was maintained by periodic transfer on L agar supplemented with 0.2% glucose and 50 pg/ml thymidine.

Materials-E. coli DNA polymerase (large fragment) was obtained from P. L. Biochemicals, T4 DNA ligase from Bethesda Research Laboratories and T4 polynucleotide kinase from Boehringer. Restric- tion enzymes were from Bethesda Research Laboratories [cx-~'P] dATP (3000 Ci/mmol), [-y3'P]ATP (>5000 Ci/mmol) and [LY-~~S] dATP (400 Ci/mmol) were purchased from Amersham Corp. Dihy- drofolate and trimethhoprim were from Sigma, NADPH was from Boehringer, and methotrexate was a gift of Lederle laboratories.

Site-directed Mutagenesis-The mutagenic oligonucleotides were synthesized on a Biosearch model 8600 apparatus by the phospho- triester method (16) and purified on 20% acrylamide-urea gels. Table I gives their sequences with the corresponding mutated positions. A 920-base pair EcoRI-BglII fragment containing the coding region of the cDNA of mouse dihydrofolate reductase under the control of a modified 8-lactamase promoter (13) was subcloned into the polylinker of phage M13mp8 between the EcoRI and BamHI sites. This fragment was subjected to oligonucleotide site-directed mutagenesis essentially according to Zoller and Smith (17). In some cases, it was possible to check directly for the presence of the mutation by digestion of the replicative form (Table I). In the others, the mutants were selected

TABLE I Oligonucleotides used for directed mutagenesis

Nucleotides underlined were those used to change the codon. Position in

mouse dihy- Created drofolate mutation reductase

Mutagenic oligonucleotide

Ile-7 -Pro-? ATC

~ e r - 7 ccc TCC

Leu-22 -Arg-22 CTA CGG Trp-24 -Arg-24 TGG CGG Phe-31 "+Arg-31 TTC CGC

TTC Phe-34 -Leu-34

CTC Gln-35 ~ A r g - 3 5 CAA CGA Val-115-Pro-115 GTC ccc Thr-136-Val-136 ACA GTA

5"CACGGCGACGGGGCAGTTCA -

CACGGCGACGGAGCAGTTCA -

AGGCCAGGGCCGGTCTCCGTT" -

GCGGAGGCCGCGGTAGGTC -

GAAGTACTTGCGCTCGTTCCTGb -

ATTCTTTGGAGGTACTTGAAC' -

GTCATTCTTCGGAAGTACTTG -

TGCCTCCGGGTATCCAAAC" -

ATGATCCTTgCACAAAGAGT

HpaII restriction site created. XmnI restriction site missing.

e ScaI restriction site missing.

by plaque hybridization (18). Single-strand DNA from each mutant was prepared and chain terminator DNA sequencing was carried out according to Sanger et al. (19) as modified by Biggin et al. (20) using [ Y - ~ ~ S I ~ A T P instead of [Y-~*P]~ATP. For each mutant the entire sequence of the cloned fragment was performed which represented more than 900 base pairs. To facilitate this work, two synthetic primers were used in addition to the M13 universal primer, dividing the fragment to be sequenced into three parts of about 300 base pairs each, easily read on a 6% acrylamide gradient gel. Mutants which exhibited additional mutations, even silent ones, were discarded. After preparation of the replicative form, the mutant cDNAs were digested by EcoRI and Hind111 and transferred from the recombinant M13mp8 into the EcoRIIHindIII cloning sites of pUC8. These recombinant plasmids, and the plasmids containing the two spontaneous mutants (Ser-31 and Pro-%), were used to transform the dihydrofolate reductase-deficient E. coli strain D3-157 (15) for production of the mutant proteins.

Purification of Dihydrofolute Reductase-For preparation of bacterial extracts, dihydrofolate reductase was purified from 1 1 of overnight culture in rich medium (yeast extract, 5 g; bacto-tryptone, 20 g; NaCl, 5 g; MgS04, 5 g; glucose, 2%). Bacteria were harvested in 60-70 ml of 100 mM potassium phosphate buffer, pH 7.9 (standard buffer), and disrupted by sonication using a Vibra Cell sonifier, model 500 W, at a setting of 10 for 6 periods of 30 s, 0.5 pulse per second. All the buffers contained 1 mM dithiothreitol and 1 mM EDTA. After centrifugation at 26,000 X g for 30 min, the supernatant was recen- trifuged at 100,000 X g for 1 h. The cytosol was subjected to ammonium sulfate fractionation and the precipitate at 50% saturation was removed. Ammonium sulfate was then added to the supernatant to achieve 80% saturation by 10% steps. The three fractions obtained in this way were dissolved in a minimum volume of standard buffer. After dialysis, fractions containing the dihydrofolate reductase activity were applied to a 2 X 10 cm column of DEAE-cellulose (Whatman) equilibrated in 50 mM Tris-HC1 buffer. The pH varied from 7 to 8 according to the mutants. The maximum of dihydrofolate reductase activity was most of the time in the elution volume at the pH of equilibration of the DEAE-cellulose column. It was sometimes necessary to apply a saline gradient (50-100 mM KCl) to increase the yield of recovery. Fractions containing the highest specific activity of dihydrofolate reductase were pooled and concentrated to about 2 ml by freeze-drying. The solution from the preceding step was passed through a 2 X 90 cm Ultrogel AcA 54 column equilibrated in 50 mM Tris-HC1, pH 7.5, 100 mM KCl.

For preparation of cellular extract, DXBll cell line (21) transfected with pSV2-DHFR and resistant to 100 p~ methotrexate was used to purify wild-type dihydrofolate reductase from eucaryotic cells. In this cell line (C100) the transfected dihydrofolate reductase cDNA has been amplified and dihydrofolate reductase represents about 3% of the total soluble proteins.' Cells were passaged so that they were grown at least four generations in the absence of methotrexate. About 2 X 10' cells in late log phase were harvested in 150 mM NaCl solution, centrifuged for 5 min at 500 X g, and resuspended in a minimum volume of 50 mM Tris-HC1 buffer, pH 9. The cells were disrupted by three 15-s bursts of sonication. The lysate was centrifuged at 100,000 X g for 1 h. The supernatant was submitted to the same steps of purification as the bacterial extracts except that the ammonium sulfate precipitation was omitted.

The enzymes were generally used directly after the AcA 54 column or further concentrated using Centricon M10 membranes or Sartorius collodion tubes. They could be stored at -20 "C and most mutants were stable for at least 1 month under these conditions.

The protein content was determined according to Bradford (22). Purity of the preparation was checked by SDS-PAGE' according to Laemmli (23).

Enzyme Assays-Dihydrofolate reductase was assayed spectropho- tometrically for standard steady-state conditions as follows: 50 p~ dihydrofolate, 100 p M NADPH, 100 mM potassium phosphate, pH 7.9, 100 mM potassium chloride, in a total volume of 1 ml. These concentrations of substrates were found to be saturating for all the enzymes. The reaction was started by addition of enzyme (0.02-0.2

' J. Thillet, J. Absil, and R. Pictet, unpublished results. ' The abbreviations and trivial names used are: SDS-PAGE, so-

dium dodecyl sulfate-polyacrylamide gel electrophoresis; H'folate, dihydrofolate; H4folate, tetrahydrofolate; MTX, methotrexate; TMP, trimethoprim; MES, 4-morpholineethanesulfonic acid; M+T, 2,4- diamino-5,6-dimethyl-5-(4'-methoxyphenyl)-5-triazine.

12502 Mutants of Mouse Dihydrofolate Reductase

pM) and was monitored by the decrease in absorbance at 340 nm during 3-5 min.

Results were calculated using an extinction coefficient of 12,000 I”’ cm” at 340 nm (24). All measurements were made at 30 “C with an Uvikon 810 spectrophotometer fitted with a thermostated cell compartment.

For K,,, (H2folste) values higher than 1 p ~ , the K,,, was determined under standard experimental conditions using dihydrofolate concentrations varying from 1 to 10 pM or from 6 to 60 p ~ , according to the value of K,,,. NADPH was held constant at 100 FM.

For K, (Htfolate) values lower than 1 pM, it was very difficult to obtain accurate values by classical steady-state measurements. Conse- quently, these K,,, were determined by progress-curve experiments according to Stone and Morrison (25). These authors demonstrated that the values of the kinetic parameters did not vary appreciably from those obtained by initial rates when both determinations were possible. NADPH concentration was kept constant at 100 p~ and different progress curves were performed at 5,10, and 15 p~ dihydrofolate. Data were analyzed with a BASIC computer program.

For K, (NADPH) determinations, the NADPH concentrations varied from 10 to 100 pM, and the dihydrofolate concentration was held constant at 100 p ~ .

The reported kinetic values are the average of at least two separate determinations and the variation never exceeded 10%.

The pH dependence of dihydrofolic acid reduction was measured using 50 mM MES, 25 mM Tris, 25 mM ethanolamine buffer containing 100 mM KC1 from pH 5.4 to 9.4 (26). The ionic strength of this buffer is essentially constant over the pH range used (27).

Measurement of Enzyme Concentration-The enzyme concentration was determined by methotrexate titration (28) for wild-type dihydrofolate reductase, Ser-7, Ser-31, Leu-34, Arg-35, Pro-115, and Val-136 mutants. For the other mutants (Arg-22, Arg-24, and Arg- 31) it was not possible to use this technique, since the Ki for methotrexate was too high. The concentration was estimated, assuming that the percentage of active enzyme was the same as for other mutants (60%). The concentration of Pro-35 was not determined since the extent purification was not sufficient to allow an accurate assessment.

Determination of Methotrexate and Trimethoprim Inhibitions Con- stants-When methotrexate gave rise to classical, steady-state inhibition, the K; was determined according to Dixon (29), varying the concentrations of inhibitor and dihydrofolate while NADPH is kept constant at 100 PM.

When slow-binding inhibition was observed, progress curves with different methotrexate concentrations were obtained with the concentrations of dihydrofolate and NADPH held constant at 50 and 100 p ~ , respectively. Assays were started by addition of the enzyme at zero time. Data from progress curves were fitted to equations describing either Mechanism A or Mechanism B as described (30).

In case of tight-binding inhibition, NADPH and dihydrofolate were held constant at 100 and 50 p ~ , respectively, and methotrexate was varied. The enzyme was incubated with inhibitor and the reaction was started with dihydrofolate. Data were analyzed according to Morrison (31).

The effect of trimethoprim was determined according to Dixon (29) by varying the concentrations from M to 5 X M. It was not possible to increase the concentration above 5 X M since trimethoprim is soluble only in 50% ethanol solution beyond that concentration.

Computer Graphics-Coordinates of mouse dihydrofolate reductase from analysis of x-ray diffraction by crystals were examined on an Evans and Sutherland PS 300 using the FRODO program in the Astbury Department of Biophysics (Leeds) with the kind authoriza- tion of A. J. Geddes and C . R. Beddell (Wellcome Research Labora- tories, Beckenham, United Kingdom).

RESULTS

Purification of Enzymes-Table I1 lists the positions of the mouse dihydrofolate reductase which have been mutated and describes some of their characteristics. All the mutations were designed to change amino acids located in the active site, and were expected to modify the binding properties of the enzyme. Therefore, it was necessray to use a purification procedure that was not dependent on ligand affinity columns.

The purification scheme was the same for all the mutants, with slight modifications according to the mutation. Table I11

gives the conditions used and the yields of purification for two mutants. The concentration of ammonium sulfate at which the enzymes precipitate varied from 50-60% to 60-80% saturation. The pH of equilibration of the DEAE-cellulose column had also to be adapted according to the net charge modification produced by the mutation. Whereas the wild- type enzyme and mutant enzymes with neutral substitutions eluted at pH 7.2 (for example Leu-34 in Table 111), mutants with Arg substitutions eluted at pH 8 (Arg-31 in Table 111). The purity of the preparation was checked by SDS-PAGE electrophoresis (Fig. 1). By this criterion, the protein was, in most cases, more than 95% pure. Ser-7, Pro-35, and Pro-115 were unstable (see below) and could only be obtained with a purity of 20%. Only one mutant, Pro-7 could not be purified. A very low activity was found in crude extracts which was lost during the purification.

Stability and Chemical Characterization-Although all the cDNAs coding for the mutated enzymes are under the control of identical regulatory sequences, the amounts of protein in bacteria were variable from one mutant to another and could be estimated at 3-20% of the total soluble proteins (Fig. 2) with the exception of the three mutants (Ser-7, Pro-35, and Pro-115) which could not be quantified (see, for example, Ser- 7 in Fig. 2). To establish the reason for this variability we checked the “in uitro” stability of the enzymes at 37 “C since it could reflect their “in uiuo” stability. The results are shown in Fig. 3. Six of the mutants were more stable than the wild- type enzyme and corresponded to those found in higher amounts in bacteria (Fig. 2). Three mutants, Ser-7, Pro-35 (not shown), and Pro-115 which were less stable than the wild-type enzyme, were also those which are present in bacteria in lower amounts than the wild-type enzyme.

In several instances, eucaryotic proteins synthesized in bacteria keep the translation initiating methionine at their NH2 terminus. To determine whether this was the case for the mouse dihydrofolate reductase or if other modifications had occurred, the sequence of the first 8 amino acids at the amino terminus of purified wild-type dihydrofolate reductase and those of two mutants (Arg-22 and Arg-31) were determined by automatic sequencing using an Applied Biosystems protein sequenator. The sequences found were identical to the NH2 terminus of the previously described mouse dihydrofolate reductase (32), except that, in all three cases, 50% of the first step degradation product was a methionine residue. Moreover, a discrepancy was found between the amount of material quantified by amino acid analysis and the amount of NHz terminus amino acid estimated by sequencing, indicating that about 25% of the NHz terminus was blocked.

Kinetic Studies-As the mouse dihydrofolate reductase synthesized in bacteria could have a modified enzymatic activity, it was essential to verify that the wild-type enzyme isolated from bacteria had the same properties as the enzyme isolated from mammalian cells. To be sure that we compare proteins with the same sequence, we used DXBII (CHO-DHFR-) cells transfected with pSVz-DHFR as source of eucaryotic material. The cDNA inserted in pSV2 was strictly identical to the cDNA inserted in the plasmid used for the production of mouse dihydrofolate reductase in bacteria. The results shown in Table IV demonstrate that the kinetic parameters for the enzymes isolated from both sources were similar.

Table IV also gives the kinetic parameters obtained for the mutant dihydrofolate reductases. Since all the mutations OC-

cur in the dihydrofolate binding site, a preliminary evaluation of the effect of these mutations on other parts of the dihydrofolate reductase molecule is provided by the binding of

Mutants of Mouse Dihydrofolute Reductase 12503 TABLE I1

Positions of mutations made on the cDNA of mouse dihydrofolute reductase Position in mouse

dihydrofolate reductase

Created mutation

Corresponding residue in

E. coli conservation Sequence Interactions with MTX

Ile-7 - Pro-7 1 ~ e r - 7

Leu-22 - Arg-22

Trp-24 - Arg-24

Phe-31 Ser-31 - Arg-31 Phe-34 - Leu-34 Gln-35 - Arg-35 - Pro-35 Val-115 -Pro-115

Thr-136-Val-136

Ile-5

Met-20

Trp-22

Leu-28

Phe-31 Lys-32

Ile-94

Thr-113

Invariant in

Invariant in

Strictly invariant

Or Tyr-31 in vertebrates

Strictly invariant Invariant in

Invariant in

Strictly invariant

vertebrates

vertebrates

vertebrates

vertebrates

Hydrogen bond

Hydrophobic contacts

Hydrophobic weak hydrogen bond via water



Hydrogen bond

Hydrogen bond via water

TABLE 111 Purification of two dihydrofolute reductase mutants from 03-157 E. coli strain

step Fraction Volume Protein Activitf Specific activity Recovery Purification ml mg units units f m g % -fold

DHFR-Arg-31 1 Cell-free extract 69 1170 327 0.28 100 2b Ammonium sulfate (60-70%) 7.5 150 125 0.83 38 2.96 3 DEAE-cellulose, pH 8 26.5 69 76 1.1 60 3.93 4 Ultrogel AcA 54 12 46 75 1.63 99 5.82

DHFR-Leu-34 1 Cell-free extract 70 1120 326 0.29 100 2 b Ammonium sulfate (50-60%) 12 336 234 0.69 71 2.38 3 DEAE-cellulose, pH 7.2 33 18 50 2.7 79 9.3 4 Ultrogel AcA 54 12 5.1 32.4 6.3 65 21.7

One unit of enzyme activity is defined as the amount that produces 1 pmol of H.folate/min at pH 7.9 and 30 "C with H,folate as substrate.

The yield of precipitation including all the fractions is more than 90%, but we used only the fraction with the maximum of specific activity. The yields of the following steps were calculated from the quantity used after ammonium sulfate precipitation.

M r A B C D 92.5- 66.2 - 45.0-

31.0 - - 21.5 -

FIG. 1. Purification of Arg-31 mutant visualized by Coo- massie Blue staining of SDS-PAGE. A, crude extract of D3-157 strain containing pUC8-Arg-31; B, ammonium sulfate precipitation (60-70%); C, elution from DEAE-cellulose at pH 8; D, elution from Ultrogel Aca 54. The arrow indicates the position of dihydrofolate reductase. The bottom of the gel is the anode end.

NADPH. All the mutant enzymes exhibited a value for K,,, (NADPH) that was comparable with the value for the wild- type enzyme, with the exception of Arg-22, which exhibited a

10-fold decrease. These results suggest that the mutations had a minor effect on the ternary conformation. Residue 22 is, in fact, in contact with NADPH by hydrophobic interactions and thus, the decrease in K,,, in this case is probably due to modifications of direct interactions.

In contrast, the values of K, (H,folatl) were significantly increased in all cases, presumably because all the mutations are located at positions involved in the binding of dihydrofolate. The K,,, (Hrf,,late) of Ser-7 was difficult to determine because a marked hysteresis was observed. Baccanari and Joyner (33) have described a similar phenomenon for E. coli dihydrofolate reductase. A slight hysteresis also existed for the mouse wild- type enzyme but could be circumvented by appropriate incu- bation of the enzyme in the presence of one of the substrates before initiating the reaction. In the case of Ser-7 the hysteresis was always present, whatever the conditions of preincu- bation used. One possible explanation for this hysteresis is a slow conversion of an inactive complex to an active one.

Five mutations also affected the catalytic activity of the enzyme (Table IV). Arg-31, Ser-7, Arg-24, and Arg-22 mutants are much slower than the wild-type enzyme with, respectively, 6-, lo-, 36-, and 60-fold lower V,.. . As their K, (H,foiats) values are increased in all cases, their efficiency is also lower than that of the wild-type enzyme. One mutant, Pro-115, exhibited an increased Vmax.

Preliminary results indicate that a number of mutations caused changes in the pH dependence of the dihydrofolate

12504 Mutants of Mouse Dihydrofolute Reductase

FIG. 2. SDS-PAGE of cytosols prepared from D3-157 strain transformed with the different recombinant pUC8-DHFR plasmids. The arrow indicates the position of dihydrofolate reductase.

01 Arg 24 Arg 31

90 +3

01 10 20 30 40 50 60 min

FIG. 3. Inactivation of dihydrofolate reductases at 37 "C. The thermal inactivation of dihydrofolate reductase was checked under standard steady-state conditions. Sets of duplicate tubes containing only the enzyme and the buffer were incubated for 0-60 min at 37 "C. Samples were then mixed with the substrates and incubated at 30 "C to determine their activities.

TABLE IV Kinetic parameters of dihydrofolate reductase (DHFR+) reaction The kinetic parameters are the average of at least two separate

determinations. V ~ X Km(HZr*hUl K,NNADPH, rnin" 10" M 10-6 M

DHFR+ purified from 400 0.04" 1.4

DHFR+ purified from 360 0.09" 1.5

Ser-7 36 2 2 Arg-22 6 1" 0.14 Arg-24 10 10 2.5 Ser-31 330 4.5 3.3 Arg-31 60 2.5 1.1 Leu-34 380 4.3 2 Arg-35 480 2.3 3.1 Pro-35 NDb 20 1.6 Pro-115 850 2 2.9 Val-136 230 12 2.1

mammalian cells

bacteria

Determined by progress curve measurements. ND, not determined.

reductase reaction. The wild-type enzyme displayed the typical pH profile of the eucaryotic dihydrofolate reductases (Fig. 4) (34-37). The same profile was found for Ser-7, Arg-35, and Pro-115. All other mutants had modified pH profiles (Fig. 4). Arg 22 and Arg 24 exhibited pH profiles with a continuous increase of activity as the pH decreases. The pH profile of the other mutants appeared similar to that of bacterial dihydrofolate reductase with apK. of about 8 (38). Fig. 4 displays the pH profile of Arg-31, which was typical of this class of mutants. Obviously, more detailed kinetic analysis will be required to understand fully the observed effects.

Table V gives the inhibition constants of wild-type and mutants for trimethoprim as well as for methotrexate. The Ki ( m p ) for wild-type enzyme is in a normal range for eucaryotic dihydrofolate reductases. Four of the mutations, Ser-7, Arg-31, Leu-34, and Pro-35, induced such a marked decrease in the binding of trimethoprim that the Ki could not be accurately determined. At concentrations of T M P higher than 5 X IO-' M the concentration of ethanol necessary to solubilize T M P affects the activity of the enzyme. Two mutants, Arg- 22 and Arg-24, also exhibited an increased K ~ ( T M P ) , although to a lesser extent, whereas the others (Ser-31, Arg-35, Pro- 115, and Val-136) had Ki (TMP) values in the same range as the wild-type enzyme.

The slow, tight-binding inhibition of the wild-type enzyme for methotrexate conformed to Mechanism B previously described by Williams et al. (39) (Table V). Similar results have been obtained with other dihydrofolate reductases (40) and the overall inhibition constant was in the picomolar range. In contrast to the effects observed with trimethoprim, all the mutations affected the binding of methotrexate (Table V). They also affected the type of inhibition observed, although this difference could be only due to the inability to make a clear distinction between mechanisms A and B. This distinction is essentially dependent on the steady-state concentration of the intermediate complex E . I (enzyme-inhibitor complex) and if it is kinetically insignificant, a mechanism that might be formally in accord with Mechanism B will then appear to be described by mechanism A (30). It is consequently difficult to conclude if the Ser-31 and Arg-35 mutants have no isomerization step or if it is not visible under the conditions used. Whatever the mechanism, the mutations changed significantly the overall inhibition constant. Slow- binding inhibition could not be measured with the Ser-7 and Val-136 mutants because of some hysteresis in initial rates

Mutants of Mouse Dihydrofolute Reductase 12505

5 6 7 8 pH lo

FIG. 4. pH profiles of log V,, for wild-type dihydrofolate reductase (0), Arg-31 (O), Arg-22 (m, and Arg 24 mutants (A). Assays were performed as indicated under "Experimental Pro- cedures."

studies. The relative increase of the Ki (MTX) versus that of the K ~ ( T M ~ ) were not consistent from one mutant to the next, probably reflecting the different binding modes of the two inhibitors. For example, Arg-22, which exhibited the greatest increase in Ki (MTX) (7.5 x 106-fold) had a Ki (TMp) increased only 100-fold; in contrast, Leu-34 mutation had one of the smallest effects on methotrexate binding (200-fold) and the maximal effect on trimethoprim binding (>lOOO-fold). The nature of the replacement was also important. At positions 31 and 35, for which two mutants have been studied, one mutation had no significant effect on trimethoprim binding (Ser at position 31 and Arg at position 35), whereas the others (Arg-31 and Pro-35) had a marked effect on the binding of both inhibitors.

DISCUSSION

Understanding of the relationship between enzyme structure and activity should be greatly facilitated by the possibility to examine the effect of mutations obtained by site-directed mutagenesis. These studies are also easier if the protein can

be expressed in bacteria in a native form. We have developed an expression vector which allows the production of large amounts of active mouse dihydrofolate reductase in E. coli and in B. subtilis (13). By using the D3-157 E. coli strain which lacks this enzyme, the mouse enzyme can be purified free of bacterial dihydrofolate reductase. We have used this system to study the effect of a series of mutations on the dihydrofolate reductase activity and on the binding of two inhibitors of pharmacological interest. Since primary structures of the higher vertebrate dihydrofolate reductases are well conserved, especially the amino acids involved in the active site, it was reasonable to use the knowledge of chicken dihydrofolate reductase three-dimensional structure to choose the positions to be mutated. The recently published structure of the mouse dihydrofolate reductase confirmed the validity of this assumption (12). The amino acid sequence deduced from the cDNA we have used (41) and from cDNAs isolated from other mouse cell lines are identical (42). However, this sequence differs at two positions from the one obtained by direct sequencing of dihydrofolate reductase isolated from the L1210R cell line (32). This divergence concerns, respectively, Gln-127 and Lys-173 in the cDNA which are replaced by Glu- 127, Asp-173 in L1210 sequence. These differences may be artefactual, due to the different methods of analysis used. Examination of the mouse three-dimensional model has shown that residue 127 is certainly a glutamine since it could form a hydrogen bond with the close Asp-94 which is not possible with a glutamic acid. In any case, these two positions are completely external and probably not involved in the active site.

The kinetic parameters of the wild-type enzyme were care- fully examined because we had established by amino acid sequencing that part of the recombinant dihydrofolate reductases isolated from bacterial extracts had a modified NH2- terminal sequence either blocked or with NH2-Met-Val instead of the native NH2-Val. As mentioned by Ben-Bassat et al. (43) the methionine amino-peptidase is less efficient when the NHa-terminal methionine is followed by a large hydrophobic amino acid. Moreover, in case of overproduction of a foreign protein in bacteria, the activity of the methionine aminopeptidase is not sufficient to ensure complete processing of the foreign protein (43). Both these observations could explain the incomplete processing of mouse dihydrofolate reductase synthesized by the bacteria. However, a comparison of wild-type enzymes isolated from recombinant mammalian cells and bacteria demonstrated that their properties are not significantly different (Table IV). Volz et al. (8) have pointed out that the 3 extra residues at the amino terminus of the chicken enzyme, which have no counterpart in the bacterial enzymes, are disordered and thus are probably not implicated in a particular function. The observed discrepancies between our values and the kinetic parameters cited by Cha et al. (44 and references therein) or by Haber et al. (45) can reasonably be attributed to the different experimental conditions used.

Most of the mutants constructed during this study exhibited an increased resistance to denaturation at 37 "C (Fig. 3). From a comparison between the sequences of thermostable and thermolabile proteins, it has been assumed that the thermo- stability is raised by increasing the internal hydrophobicity or by stabilization of helices (46). The majority of the mutations increased the internal hydrophilicity, and three of them (Arg-22, Arg-24, and Val-136) are not in helix structures. An analysis of the effect of the mutations by X-ray crystallographic studies would be required to determine how these residues stabilize the folding of the protein. In contrast, it is not surprising that Pro-35 and Pro-115 mutations produce

12506 Mutants of Mouse Dihydrofolate Reductase

TABLE V Inhibition constants for the interaction of trimethoprim and methotrexate with dihydrofolate reductases

Trimethoprim Methotrexate

Ki Type of inhibition KZ k"" kbn K;*

W nM min" nM Wild-type 3.5 Slow, tight-binding B 0.025 525 100 0.004 Ser-7 >3O0Ob Tight-binding 6 Arg-22 360 Classical 3000 Arg-24 55 Classical 300 Ser-31 8 Slow, tight-binding A 0.18 0.041 4.4 Arg-31 >3O0Ob Classical 190 Leu-34 >3000b Slow, tight-binding B 2.68 7.23 3.53 0.81 Arg-35 3 Slow, tight-binding A 0.81 20.98 0.039 Pro-35 >3O0Ob Classical 175 Pro-115 4 Slow, tight-binding B 0.2 4.72 3.71 0.088 Val-136 3 Tight-binding 1.35

The units of k,. are nM min" for Mechanism A and min" for Mechanism B. The lack of solubility of this compound did not allow accurate determination of K;.

unstable proteins since they may disrupt an a-helix in the case of Pro-35 and a P-sheet in the case of Pro-115.

Volz et al. (8) have determined the structure of chicken dihydrofolate reductase in a ternary complex with NADPH and a phenyltriazine inhibitor (M@T) at 2.9 A resolution. According to their analysis, 12 amino acids are involved in the binding of inhibitor: Ile-7, Glu-30, Val-115, and Thr-136 through hydrogen bonds and Val-8, Ala-9, Leu-22, Tyr-31, Phe-34, Thr-56, Ser-59, and Ile-60 through hydrophobic contacts. Since the M@T-protein interactions in the chicken structure are analogous to those observed for the methotrexate binding to E. coli and Lactobacillus casei dihydrofolate reductases (9), all these positions are potential targets for mutations attempting to modify the binding of methotrexate to the mouse enzyme. Position 24 is also potentially interesting since tryptophan at this position has been found in all dihydrofolate reductases sequenced to date and the corresponding tryptophan in L. casei and E. coli interacts with MTX. The refine- ment of the structure of the mouse dihydrofolate reductase complex with methotrexate has shown that MTX and M@T interactions were analogous and that Trp-24 was hydrogen- bonded to MTX (12). Furthermore, the analysis of this complex indicates that Gln-35 may also be hydrogen-bonded to MTX. The interest of this potential interaction is emphasized by the fact that a mutation (Pro-35) increasing the resistance in vivo to methotrexate was obtained in B. subtilis (13). Five of the amino acids forming hydrogen bonds with methotrexate were substituted-Ile-7, Trp-24, Gln-35, Val-115, and Thr- 136-as well as 3 amino acids involved in hydrophobic interactions: Leu-22, Phe-31, and Phe-34 (see Table 11). Unless otherwise stated, the results will be discussed according to the mouse structure.

The carbonyls of Ile-7 and Val-115 are hydrogen-bonded to the 4-amino group of the M@T or methotrexate. Additionally, hydrophobic contacts also exist between the side chains of these amino acids and the triazine ring of M@T or its attached methyl groups (8). The situation is different for the trimethoprim which is hydrogen-bonded to Ile-7 but not to Val-115. On the assumption that the hydrogen bond is more important than the hydrophobic contacts for the stability of the enzyme- inhibition complex, we have replaced Ile-7 and Val-115 by a proline. The specific activity of the Pro-7 mutant was SO low and the protein so unstable, that it could not be studied. The drastic effect of the Pro-7 mutation can be explained by a disruption of the p-sheet which could dramatically change the conformation of the protein and of its active site, thus low- ering its activity. In view of the results obtained with Pro-7,

Ile-7 was also replaced by a serine. Surprisingly, the Ser-7 mutation also leads to a very unstable protein with low activity, even though the modifications affects only the side chain. The introduction of a proline at position 115 also lowers the stability but increases the Vmax. Interpretation of the increased V,, of Pro-115 requires further studies. Both mutations Ser-7 and Pro-115 increased the K; (MTX) value but Ser-7 caused a greater increase than Pro-115, which has one of the lowest overall K; (MTX) values for the mutant enzymes. This result was unexpected since both hydrogen bonds seemed equivalent with regard to the binding of methotrexate. How- ever, modeling the Pro-115 substitution shows that the proline could be positioned without great changes in the backbone structure. Consequently, the carbonyl could possibly conserve its interaction with methotrexate. The results obtained with trimethoprim (no change for the Ki (TMp) of Pro-115 and a dramatically increased K; (TMP) for Ser-7) can be explained since the Val-115 wild-type residue, in contrast to Ile-7, is not hydrogen bonded to the 4-amino group of trimethoprim.

Threonine 136 is conserved in the structure of all dihydrofolate reductases sequenced to data (8). The hydroxyl group of Thr-136 hydrogen bonds to a carboxylate oxygen of Glu- 30 and to the 2-amino group of the inhibitor via an intervening water molecule either (with M@T) in the chicken liver enzyme (8) or (with MTX and TMP) in the mouse dihydrofolate reductase (12). We mutated Thr-136 into a valine which deletes the hydroxyl group without modifying the steric space of the amino acid. Val-136 mutant has a greatly increased K,,, (Hzfolate) but a normal catalytic activity. It is likely that it is the loss of the hydrogen bond via a water molecule between Thr-136 and the substrate which is responsible for decreased binding of dihydrofolate as reflected by a greatly increased K,,, (Hzfolate). In this case, the loss of binding energy has not led to a loss of catalytic activity, since the latter was not dramatically modified. Similar results were obtained by Chen et al. (6) with the same substitution at the equivalent position (Thr- 113) in the E. coli enzyme. There was an increase of the Ki (MTX) value which could also be explained by the loss of the hydrogen bond. However, the K; (TMP) was not changed in spite of the fact that the mode of interaction of Thr-136 with both inhibitors seems to be the same (12). These results could suggest either that the binding of trimethoprim and methotrexate differ slightly or that new interactions occur between Val-136 and each inhibitor.

Like Thr-136, Trp-24 is a strictly invariant residue. Its side chain is also hydrogen-bonded to the carboxylate group of Glu-30 and to methotrexate but not to trimethoprim. The

Mutants of Mouse Dihydrofolate Reductase 12507

replacement of Trp-24 by an arginine residue is certainly less conservative than the replacement of Thr-136 by a valine, which probably explains its more drastic effect on inhibitor binding. The Trp-24 + Arg mutation should delete the hydrogen bond, thus weakening the binding of methotrexate with the resultant observed large increase of Ki. Although the hydrogen bond is absent in the corresponding mouse dihydrofolate reductase-trimethoprim complex, there is also a small but real effect on trimethoprim binding. The mutation also had a dramatic effect on the binding of substrate and on the catalytic activity. The K,,, (H,f&b) value was increased to the same extent as for Val-136 mutant, but the Vmax of Arg-24 was decreased about 60-fold, whereas the Vmnx for Val-136 remained quite normal. This indicates that, for Arg-24, in contrast to Val-136, the loss of substrate binding energy was associated with a diminished catalytic activity.

Pro-35 was obtained as a spontaneous mutation with increased resistance to methotrexate (13). Gln-35 in the chicken enzyme is not expected to be directly involved in contacts with inhibitors, but in the mouse, it could be hydrogen-bonded to the a-carboxylate group of the glutamyl moiety of methotrexate (12). A proline residue at this position in an a-helix could then change the local conformation of the protein and it was consequently difficult to attibute the loss of binding of methotrexate only to the loss of an hydrogen bond. Indeed, the moderate effect of an arginine mutation at this position, with a normal Ki (TMP) and a Ki (MTX) which was the least modified among all the mutants, suggests that Gln-35 is not greatly implicated in the binding of methotrexate.

The other mutations studied concerned residues involved in hydrophobic contacts with the inhibitors.

Leu-22 is normally involved only in hydrophobic contacts with substrate or inhibitors (8). Two spontaneous mutations have been described at that position, both conferring a mark- edly increased resistance to methotrexate. Mutations Leu + Phe in Chinese hamster cells (47) and Leu + Arg in mouse cells (48) were obtained by selection in the presence of increasing amounts of methotrexate. In view of the major functional modifications observed and the decreased binding of inhibitors, Leu-22 residue seems to play a critical role in the function of dihydrofolate reductase. Accordingly, we constructed a Leu + Arg substitution. The presence of a charged residue at this position caused a decrease in K , (NADPH). The corresponding wild-type residue in L. casei (Leu-19) is one of the residues involved in hydrophobic contacts with the nicotinamide ring of NADPH (49). In accordance with the similarity between the structures of the two enzymes, the residue Leu-22 in mouse dihydrofolate reductase should also be in contact with NADPH. However, the deletion of these hydrophobic contacts has resulted in an increased rather than a decreased affinity for NADPH. This substitution also con- siderably reduced the catalytic activity of the enzyme and the binding of the two inhibitors. The Ki (MTX) value was the highest obtained for all the mutants. The arginine would certainly protrude in the active site more than the wild-type leucine residue and could hinder the positioning of inhibitors either by its steric effect or by allowing the penetration of water molecules in the active site.

Positions 22 and 24 are situated on the same side of the active site which seems critical for the catalytic activity and for the binding of methotrexate. Both mutations Arg-22 and Arg-24 induced the greatest changes in Vmax and Ki values even though the wild-type residues at these two positions are not involved in similar interactions with substrate or inhibitors.

The phenyl ring of the Phe-34 is involved in many hydro-

phobic contacts either with trimethoprim or methotrexate (8- 50). In mouse dihydrofolate reductase its orientation toward both inhibitors is identical (12). Our results show that a leucine at this position changed only slightly the Ki (MTX)

value, whereas that for Ki (TMP) was greatly increased. The hydrophobic interactions of the phenyl ring of Phe-34 with trimethoprim could be more important for the stability of this complex than for that of methotrexate complex. Such differences are not, however, apparent from looking at these interactions in the three-dimensional structure.

We have changed Phe-31 because two observations suggest that this position may play a specific role in the interaction with the inhibitors. First, we obtained a mutant by in vivo selection in B. subtilis, situated at this position (Ser-31) which exhibited a slightly increased resistance to methotrexate (13). Second, position 31 seems to play a more important role in the binding of trimethoprim to eucaryotic dihydrofolate reductases than the corresponding residue (Leu-28) in E. coli (50). It is the only position examined which is not strictly invariant between chicken (Tyr) and mouse (Phe) dihydrofolate reductases. In chicken dihydrofolate reductase, Tyr-31 side chain is forced to rotate upon binding of trimethoprim, and this change in orientation allows the hydroxyl group of Tyr-31 to form a hydrogen bond with Trp-24 via a water molecule (11). In the mouse dihydrofolate reductase, Phe-31 is in a completely different orientation. The side chain does not rotate upon binding of inhibitors, and its phenyl ring is nearly parallel to the phenyl ring of Phe-34. Its position is identical in complexes with trimethoprim or methotrexate (12). The replacement of Phe-31 by an arginine modified the Ki values for both inhibitors, but these effects are certainly linked to the nature of the replacing residue since the serine mutation did not lead to comparable changes. The arginine could act by weakening the hydrophobic environment of the active site and allowing the penetration of water molecules.

Molecular modeling showed that, for all the substitutions described in this paper, it was possible to put the new lateral chain of the amino acid in a hypothetical position without any changes in the backbone alignment. This was possible even with Arg-22, Arg-31, and Arg-24, which could enter in the active site. The observed increases of Ki values indicated that the binding of methotrexate is easier to decrease than that of trimethoprim. It could also be noticed that, in contrast to which could be expected, the mutations which yielded the greatest increases of Ki (MTX) are those which involved amino acids making hydrophobic contacts with inhibitors and not hydrogen bonds.

The present work demonstrates that this is possible to obtain mutants of mouse dihydrofolate reductase with increased resistance to inhibitors, but always in association with a modified enzymatic activity. This could be done either by i n vivo selection in bacteria, and this is faster than in eucaryotic cells, or by site-directed mutagenesis. Among the 10 mutants studied, six displayed an interesting inhibition pat- tern and could help to elucidate the differential binding of trimethoprim and methotrexate. X-ray crystallographic studies will now be necessary to establish direct relations between individual mutations and their new properties.

Acknowledgments-We thank C. Dubucs for oligonucleotide synthesis, Y. Blouquit, and Dr. J. Hofsteenge for protein sequencing, M. Reboud for helpful discussions during this work, Dr. N. Ransholdt for reviewing the manuscript for English, and C. Sagot for typing the manuscript.

REFERENCES 1. Baccanari, D. P., Daluge, S., and King, R. W. (1982) Biochemistry

2 1,5068-5075

12508 Mutants of Mouse Dihydrofolate Reductase 2. Lau, Y. F., Lin, C. C., and Kan, Y. W. (1984) Mol. Cell. Biol. 4 , 25. Stone, S. R., and Morrison, J. F. (1982) Biochemistry 21,3757-

3. Roth, B. (1986) Fed. Proc. 4 5 , 2765-2772 4. Villafranca, J. E., Howell, E. E., Voet, D. H., Stobel, M. S., Ogden,

26. Stone, S. R., and Morrison, J. F. (1983) Biochim. Biophys. Acta

R. C., Abelson, J. N., and Kraut, J. (1983) Science 2 2 2 , 782- 27. Ellis, K. J., and Morrison, J. F. (1982) Methods Enzyml. 8 7 , 788 405-426

5. Howell, E. E., Villafranca, J. E., Warren, M. S., Oatley, S. J., and 28. Sheldon, R., and Brenner, S. (1976) Mol. Gen. Genet. 147,91-97 Kraut, J. (1986) Science 231,1123-1128 29. Dixon, M. (1953) Biochem. J. 55,170-171

6. Chen, J. T., Mayer, R. J., Fierke, C. A., and Benkovic, S. J. (1985) 30. Stone, S. R., Montgomery, J. A., and Morrison, J. F. (1984) J. Cell. Biochem. 29,73-82 Biochem. Pharmacol. 33,175-179

7. Mayer, R. J., Chen, J. T., Taira, K., Fierke, C. A., and Benkovic, 31. Morrison, J. F. (1969) Biochim. Biophys. Acta 1 8 5 , 269-286 S. J. (1986) Proc. Natl. Acad. Sci. U. S. A. 8 3 , 7718-7720 32. Stone, D., Paterson, S. J., Raper, J. H., and Phillips, A. W. (1979)

8. Volz, K. W., Matthews, D. A., Alden, R. A., Freer, S. T., Hansch, J. Biol. Chem. 254,480-488 C., Kaufman, B. T., and Kraut, J. (1982) J. Biol. Chem. 2 5 7 , 33. Baccanari, D. P., andJoyner, S. S. (1981) Biochemistry20.1710- 2528-2536 1716

9. Bolin, J. T., Filman, D. J., Matthews, D. A., Hamlin, R. C., and 34. Delcamp, T. J., Susten, S. S., Blankenship, D. T., and Freisheim, Kraut, J. (1982) J. Biol. Chem. 2 5 7 , 13650-13662 J. H. (1983) Biochemistry 22,633-639

10. Stammers, D. K., Champness, J. N., Dann, J. G., and Beddell, C. 35. Matthews, C. K., and Huennekens, F. M. (1963) J. BWL Chem. R. (1983) in Chemistry and Biology of Pteridines (Blaiz, J. A., 238,3436-3442 ed) pp. 567-571, Walter de Gruyter, Berlin 36. Peterson, D. L., Gleisner, J. M., and Blakley, R. L. (1975) Bio-

Volz, K. W., Kaufman, B. T., Beddell, C. R., Champness, J. N., 37. Smith, S. L., Patrick, P., Stone, D., Phillips, A. W., and Burchall, Stammers, D. K., and Kraut, J. (1985) J. Biol. Chem. 2 6 0 , J. J. (1979) J. Biol. Chem. 254,11475-11484

1469-1475 3765

745,247-258

11. Matthews, D. A., Bolin, J. T., Burridge, J. M., Filman, D. J., chemistry 14,5261-5267

381-391 38. Stone, S. R., and Morrison, J. F. (1984) Biochemistry 2 3 , 2753- 12. Stammers, D. K., Champness, J. N., Beddell, C. R., Dann, J. G., 2758

Eliopoulos, E., Geddes, A. J., Ogg, D., and North, A. C. T. 39. Williams, J. W., Morrison, J. F., and Duggleby, R. G. (1979)

13. Grange, T., Kunst, F., Thillet, J., Ribadeau-Dumas, B., Mousse- 40. Stone, S. R., and Morrison, J. F. (1986) Biochim. Biophys. Acta ron, S., Hung, A., Jami, J., and Pictet, R. (1984) Nucleic Acids 869,275-285 Res. 12,3585-3601 41. Chang, A. C. Y., Nunberg, J. H., Kaufman, R. J., Erlich, H. A.,

14. Doutriaux, M. P., Wagner, R., and Radman, M. (1986) P m . Natl. Schimke, R. T., and Cohen, S. N. (1978) Nature 275,617-624 Acad. Sci. U. S. A. 83,2576-2578 42. Nunberg, J. H., Kaufman, R. J., Chang, A. C. Y., Cohen, S. N.,

15. Singer, S., Ferone, R., Walton, L., and Elwell, L. (1985) J. and Schimke, R. T. (1980) Cell 19,355-364 Bocteriol. 1 6 4 , 470-472 43. Ben-Bassat, A., Bauer, K., Chang, S. Y., Myambo, K., Boosman,

16. Hirose, T., Crea, R., and Ikatura, K. (1978) Tetrahedron. Lett. A., and Chang, S. (1987) J. Bacteriol. 169,751-757

17. Zoller, M. J., and Smith, M. (1982) Nucleic Acids Res. 10,6487- H., and Naguib, F. N. M. (1981) Biochern. P h r m o l . 3 0 ,

18. Norrander, J., Kempe, T., and Messing, J. (1983) Gene (Amst.) 45. Haber, D. A., Beverley, S. M., Kiely, M. L., and Schimke, R. T.

19. Sanger, F., Nicklen, S., and Coulson, A. R. (1977) Proc. Natl. 46. Imanaka, T., Shibazaki, M., and Takagi, M. (1986) Nature 324,

20. Biggin, M. D., Gibson, T. J., and Hong, G. J. (1983) Proc. Natl. 47. Melera, P. W., Davide, J. P., Hession, C. A., and Scotto, K. W.

21. Urlaub, G., and Chasin, L. A. (1980) Proc. Natl. Acad. Sci. U. S. 48. Simonsen, C. C., and Levinson, A. D. (1983) Proc. Natl. Acad.

22. Bradford, M. M. (1976) Anal. Biochem. 72,248-254 49. Filman, D. J., Bolin, J. T., Matthews, D. A., and Kraut, J. (1982) 23. Laemmli, U. K. (1970) Nature 227,680-685 J. Biol. Chem. 257,13663-13672 24. Hillcoat, B. L., Nixon, P. F., and Blakley, R. L. (1967) A d . 50. Matthews, D. A., Bolin, J. T., Burridge, J. M., Filman, D. J.,

Biochem. 2 1 , 178-189 Volz, K. W., and Kraut, J. (1985) J. Biol. Chem. 260,392-399

(1987) FEBS Lett. 2 1 8 , 178-184 Biochemistry 18,2567-2573

28,2449-2452 44. Cha, S., Kim, S. Y. R., Kornstein, S. G., Kantoff, P. W., Kim, K.

6500 1507-1515

26,101-106 (1981) J. Biol. Chem. 256,9501-9510

Acad. Sci. U. S. A. 74,5463-5467 695-697

Acad. Sci. U. S. A. 80,3963-3965 (1984) Mol. Cell. Biol. 4 , 38-48

A. 77,4216-4220 Sci. U. S. A. 8 0 , 2495-2499

site-directed mutagenesis of mouse dihydrofolate reductase

Documents