Engineering of a Single Conserved Amino Acid Residue of Herpes Simplex Virus Type 1 Thymidine Kinase Allows a Predominant Shift from Pyrimidine to Purine Nucleoside Phosphorylation*

Studies of herpes simplex virus type 1 (HSV-1) thymidine (dThd) kinase (TK) crystal structures show that purine and pyrimidine bases occupy distinct positions in the active site but approximately the same geometric plane. The presence of a bulky side chain, such as tyrosine at position 167, would not be sterically favorable for pyrimidine or pyrimidine nucleoside analogue binding, whereas purine nucleoside analogues would be less affected because they are located further away from the phenylalanine side chain. Site-directed mutagenesis of the conserved Ala-167 and Ala-168 residues in HSV-1 TK resulted in a wide variety of differential affinities and catalytic activities in the presence of the natural substrate dThd and the purine nucleoside analogue drug ganciclovir (GCV), depending on the nature of the amino acid mutation. A168H- and A167F-mutated HSV-1 TK enzymes turned out to have a virtually complete knock-out of dThd kinase activity (at least ∼4-5 orders of magnitude lower) presumably due to a steric clash between the mutated amino acid and the dThd ring. In contrast, a full preservation of the GCV (and other purine nucleoside analogues) kinase activity was achieved for A168H TK. The enzyme mutants also markedly lost their binding capacity for dThd and showed a substantially diminished feedback inhibition by thymidine 5′-triphosphate. The side chain size at position 168 seems to play a less important role regarding GCV or dThd selectivity than at position 167. Instead, the nitrogen-containing side chains from A168H and A168K seem necessary for efficient ligand discrimination. This explains why A168H-mutated HSV-1 TK fully preserves its GCV kinase activity (Vmax/Km 4-fold higher than wild-type HSV-1 TK), although still showing a severely compromised dThd kinase activity (Vmax/Km 3-4 orders of magnitude lower than wild-type HSV-1 TK).

Studies of herpes simplex virus type 1 (HSV-1) thymidine (dThd) kinase (TK) crystal structures show that purine and pyrimidine bases occupy distinct positions in the active site but approximately the same geometric plane. The presence of a bulky side chain, such as tyrosine at position 167, would not be sterically favorable for pyrimidine or pyrimidine nucleoside analogue binding, whereas purine nucleoside analogues would be less affected because they are located further away from the phenylalanine side chain. Site-directed mutagenesis of the conserved Ala-167 and Ala-168 residues in HSV-1 TK resulted in a wide variety of differential affinities and catalytic activities in the presence of the natural substrate dThd and the purine nucleoside analogue drug ganciclovir (GCV), depending on the nature of the amino acid mutation. A168H-and A167F-mutated HSV-1 TK enzymes turned out to have a virtually complete knock-out of dThd kinase activity (at least ϳ4 -5 orders of magnitude lower) presumably due to a steric clash between the mutated amino acid and the dThd ring. In contrast, a full preservation of the GCV (and other purine nucleoside analogues) kinase activity was achieved for A168H TK. The enzyme mutants also markedly lost their binding capacity for dThd and showed a substantially diminished feedback inhibition by thymidine 5-triphosphate. The side chain size at position 168 seems to play a less important role regarding GCV or dThd selectivity than at position 167. Instead, the nitrogen-containing side chains from A168H and A168K seem necessary for efficient ligand discrimination. This explains why A168H-mutated HSV-1 TK fully preserves its GCV kinase activity (V max /K m 4-fold higher than wild-type HSV-1 TK), although still showing a severely compromised dThd kinase activity (V max /K m 3-4 orders of magnitude lower than wild-type HSV-1 TK).
Several attempts have been made to increase the catalytic activity of HSV-1 TK for GCV and ACV by mutagenesis of the active site of the enzyme. Random sequence mutagenesis at a segment of the putative nucleoside binding site created new HSV-1 TKs with preferential phosphorylation of GCV and/or ACV. In these studies, six codons (encoding Leu-159, Ile-160, Phe-161, Ala-168, Leu-169, and Leu-170) were targeted (1-3). Thymidine and GCV phosphorylation could not be improved, but ACV phosphorylation could be enhanced for the TK enzymes containing Leu-160 ϩ Leu-161 ϩ Val-168 ϩ Met-169 (4.3-fold) or Leu-161 ϩ Ser-168 ϩ Tyr-169 ϩ Cys-170 (2.0-fold). In a number of cases, the ratios of GCV/dThd or ACV/dThd phosphorylation by the mutant enzymes could be increased by ϳ20 -30-fold (1). The creation of a second generation semi-randomized mutant TK library further optimized the enzyme for GCV/ACV substrate sensitivity. These HSV-1 TK mutants contained, besides other mutations in the substrate active site, predominantly 168F/Y (4). The ratio of catalytic capacity of the mutant TKs with GCV or ACV versus dThd increased up to 11-83-fold or 70 -567-fold, respectively (4). One mutant HSV-1 TK (I160F ϩ F161A ϩ A168F) (designated SR-39) was endowed with a 14-fold decrease in K m for GCV compared with wild-type TK. The SR-26 mutant HSV-1 TK (L159I ϩ I160F ϩ F161L ϩ A168F ϩ L169H) showed a 124-fold decrease in K m with ACV as the substrate. However, the k cat values were also markedly decreased by these mutations. The SR-39 TK mutant enzyme had k cat values that were 27-35-fold less than wild-type TK for dThd, GCV, and ACV, and the SR-26 TK mutant k cat values of the enzyme were 32-54-fold less than the wild type (5). In other studies, mutations at the pyrimidine base-binding amino acid Gln-125 and nearby sites involved in catalysis and substrate binding resulted also in mutant HSV-1 TKs with lower dThd-phosphorylating potential, although retaining pronounced GCV-phosphorylating activity (6).
On the basis of the distinct characteristics of dThd versus GCV binding to the HSV-1 TK and computer-assisted modeling studies, we previously introduced a Tyr instead of the conserved Ala at amino acid position 167 of the enzyme (7). We found that the A167Y mutant TK completely lost its pyrimidine nucleos(t)ide-phosphorylating activity but still showed pronounced GCV/ACV/PCV-phosphorylating potential, although clearly inferior to wild-type HSV-1 TK (7,8). However, further modeling studies led us to hypothesize that introduction of less bulky amino acids than tyrosine at position 167 or more bulky amino acids than wild-type alanine at position 168 in HSV-1 TK would enhance substrate selectivity. Specifically, such changes may still provide sufficient steric hindrance to the methyl group of thymine, although allowing ganciclovir to be kept further away from the functional group of amino acid 167 and, in particular, amino acid 168 to avoid any negative interference with this latter substrate. To validate this proposal, we decided to perform an in-depth site-directed mutagenesis study at amino acid positions 167 and 168 of HSV-1 TK. This study led to better insights into the role of different amino acids in the dThd-and GCV-phosphorylating capacity of the mutated HSV-1 TK enzymes and to the design and construction of mutant HSV-1 TKs with superior GCV/dThd catalytic ratios than the existing mutant HSV-1 TKs. In particular, we aimed to fully preserve the catalytic activity for GCV and related analogues and, at the same time, to abolish as much as possible the dThd catalytic activity. It was found that A167F-and A168Hmutated HSV-1 TKs showed the highest ratio of GCV kinase/dThd kinase activity. In particular, A168H HSV-1 TK proved the most interesting enzyme mutant with fully preserved GCV catalytic substrate activity and heavily compromised dThd kinase activity both at the same time. Such a discriminative activity has never previously been observed by one single amino acid point mutation in HSV-1 TK. The insights obtained have resulted in a new predictive structural modeling approach for generating substrate-specific HSV-1 TKs and are instrumental in optimizing HSV-1 TK enzyme constructs for use in combined gene/chemotherapy of cancer or other diseases.

Construction, Expression, and Purification of Wild-Type and
Mutant HSV-1 TK-Mutant HSV-1 TK enzymes were derived from the TK sequence cloned in pGEX-5X-1 (Amersham Biosciences AB, Uppsala, Sweden) (7). Site-directed mutagenesis was performed using the QuikChange site-directed mutagenesis kit (Stratagene; Westburg, Leusden, The Netherlands) as described previously (9). The two complementary oligonucleotide primers (Invitrogen) that were used contained the desired mutation at amino acid position 167 or 168 of HSV-1 TK ( Table  1). The mutant DNA was transformed into competent Escherichia coli XL-1 blue. Plasmid preparations from ampicillinresistant colonies were checked by sequencing of the TK gene on an ABI Prism 3100 sequencer (Applied Biosystems, Foster City, CA) using the ABI Prism Big Dye Terminator cycle-sequencing ready reaction Kit (Applied Biosystems) and transfected in E. coli BL21(DE3)pLysS. Transfected bacteria were grown overnight in 2YT (yeast/tryptone) medium containing ampicillin (100 g/ml) and chloramphenicol (40 g/ml) and then diluted in fresh medium. After further growth of the bacteria at 27°C (for 5 h), isopropyl-␤-D-thiogalactopyranoside (Sigma) was added to a final concentration of 0.1 mM to induce the production of the GST-TK fusion proteins. After 16 h of further growth at 27°C, the cells were pelleted and resuspended in lysis buffer (50 mM Tris, pH 7.5, 1 mM dithiothreitol, 5 mM EDTA, 10% glycerol, 1% Triton X-100, 0.1 mM phenylmethylsulfonyl fluoride, and 0.15 mg/ml lysosyme). Bacterial suspensions were passed through a SLM Aminco French pressure cell

Mutant Ala-167 and Ala-168 HSV-1 TK
press (Beun de Ronde, La Abcoude, The Netherlands) and ultracentrifuged (11,000 rounds/min, 4°C, 30 min). GST-TK was purified from the supernatant using glutathione-Sepharose 4B (Amersham Biosciences) as described by the supplier. The protein content of the purified fractions was assessed with the Bradford reagent (Sigma) using bovine serum as the concentration standard.
Radiolabeled  Modeling Studies-The Protein Data Bank-coded HSV-1 TK structures 1E2J (10) and 1K12 (11) complexed, respectively, with dThd and GCV were overlaid using the COOT modelbuilding tool for molecular graphics (12). The same software was used to generate the different enzyme mutants, the carbon-␣ chain was kept identical to the model structures, and the side chains were fitted by hand in the active site. The pictures were drawn by PYMOL (DeLano Scientific, San Carlos, CA).

RESULTS
Eleven different amino acid mutations have been introduced in the catalytic substrate-active site of HSV-1 TK at positions 167 and 168, and the mutated enzymes were evaluated for their potential to phosphorylate dThd and GCV. The kinetic K m and V max constants were determined for each of the mutant enzymes and depicted in Tables 2 and 3. Figs. 2 and 3 depict the phosphorylating capacity (V max /K m ) of the individual mutant Ala-167 and Ala-168 TK enzymes for dThd and GCV. Among the TKs mutated at codon 167, replacing alanine by other amino acids with relatively small functional groups, such as serine, cysteine, or threonine, did not markedly affect the kinetic properties of the enzyme with respect to dThd and GCV phosphorylation. In contrast, bulky side-chains at position 167, such as the aromatic amino acids Tyr, Phe, His, and Trp dramatically compromised the dThd kinase activity, yet retained pronounced GCV-phosphorylating capacity. The mutant A167F TK had the highest GCV-phosphorylating capacity (ϳ8fold higher than the previously reported A167Y TK) (9) and a deleted dThd kinase activity. The higher V max /K m value of A167F TK was mainly because of a substantially lower K m for

Mutation
Sense primer sequence Mutant Ala-167 and Ala-168 HSV-1 TK JULY 14, 2006 • VOLUME 281 • NUMBER 28 GCV compared with the A167Y TK mutant enzyme ( Table 2). Also the charged mutants A167E and A167K TK, and surprisingly also the mutant A167I TK, had neither measurable dThd kinase nor GCV kinase activity. When the ratio of the GCV/ dThd-phosphorylating potential of all 167 mutated TK enzymes were compared, A167F-mutated TK proved to be the most successful enzyme ( Table 2, Fig. 2). Similar studies were performed for a broad variety of Ala-168 mutant TK enzymes (Table 3). Similar to the Ala-167 mutants, the amino acid changes with less-bulky side chains showed vir-tually no compromised GCV kinase activity but a more pronounced attenuated dThd kinase activity. The aromatic amino acids resulted in an even better GCV-phosphorylating activity than the corresponding Ala-167-mutated TKs, except for A168W. Mutant A168C TK had pronounced GCV kinase but a ϳ40-fold compromised dThd kinase activity. The charged amino acid substitutions resulted in a better GCV phosphorylation than the corresponding Ala-167 mutants, but in both (A168E and A168K) TK mutants, dThd kinase was virtually abolished (Table 3). Interestingly, the mutant HSV-1 A168H    Table 1) were expressed as the percentage of wild-type HSV-1 TK enzyme.  Table 2) were expressed as the percentage of wild-type HSV-1 TK enzyme.
TK proved to be endowed with the highest discriminating capacity between GCV and dThd phosphorylation. Its K m for GCV was identical to the K m of wild-type HSV-1 TK for GCV, and its V max was ϳ 4-fold higher than wild-type TK, whereas its dThd kinase-phosphorylating potential was Ͼ3-4 orders of magnitude lower than wild-type TK (Table 3, Fig. 3). Whereas several mutant Ala-167 HSV-1 TKs were completely devoid of dThd kinase activity (below the limit of detection) (Table 1), the mutant Ala-168 HSV-1 TKs usually had detectable dThd kinase activities, although in a number of cases (i.e. A168H, A168E, A168K), this residual TK activity proved at least 3-4 orders of magnitude lower than wild-type HSV-1 TK activity ( Table 3).
The K m and V max values for each mutant enzyme were obtained by measuring the direct conversion of GCV or dThd to their corresponding phosphate derivatives. To reveal whether the mutant TK enzymes that show the highest ratio of GCV versus dThd phosphorylating potential (i.e. A168H TK and A167F TK) still keep the potential of non-functionally binding dThd, we have assessed the competition of dThd and the dThd analogue BVDU at a range of concentrations for the mutant HSV-1 TKs in the presence of radiolabeled GCV. Clearly, dThd (and BVDU) had substantially lost the capacity to bind to the mutant A168H and A167F HSV-1 TK enzymes, because they poorly inhibit mutant TK-catalyzed GCV phosphorylation (Fig. 4). The feedback inhibition of GCV phosphorylation by dTTP was also much less pronounced than with wild-type TK (Fig. 4).
Given the differences in GCV and ACV phosphorylation found for previously reported HSV-1 TK mutants by Black et al. (4), we performed phosphorylation studies with GCV but also with ACV and PCV and found identical kinetic properties for wild type and the mutant A168H enzyme for each individual purine nucleoside analogue (data not shown). These data confirm that introduction of the A168H mutation into HSV-1 TK does not affect phosphorylation of any of the guanine nucleoside analogues but results in heavily compromised dThd kinase activity.

DISCUSSION
The K m value of wild-type HSV-1 TK for GCV, but not for dThd, proved 2.5-3.5-fold lower under our experimental conditions than previously reported by other investigators. Field et al. (13) report a K m ϭ 66 M for GCV against native HSV-1 (Patton strain) TK from HSV-1-infected HeLa cell cultures. Kokoris and Black (5) found 47.6 M for recombinant HSV-1 TK (derived from the pET23d; HSV TK-Dummy vector) purified on a 3Ј-aminothymidine affinity column, and Bohman et al. (14) report a K m ϭ 45 M for GCV against purified native HSV-1 (F strain) TK derived from HSV-1 TK gene-transfected mammary carcinoma FM3A TK Ϫ /HSV-1 TK ϩ cell cultures. The somewhat lower K m of 18 M for GCV against purified recombinant HSV-1 (Lyons strain) TK found in our study could not be explained by the presence of GST covalently linked to the HSV-1 TK, because it was ascertained that removal of GST from the wild-type and A168H HSV-1 TK fusion protein did not change the K m value for GCV (a K m of 18 and 13 M for the wild-type and A168H GST-HSV-1 TK fusion protein versus a K m of 21 and 18 M for the wild-type and A168H HSV-1 TK). However, differences in K m for GCV may perhaps be explained by the different virus strains from which the TKs were derived. A wide variety of catalytic activities and affinities could be observed for the mutant enzymes in the presence of different amino acids at positions 167 and 168. Most of the increased or decreased catalytic activities for dThd and GCV could be explained in a structural context. Mutations at amino acid residue 167 show an abolishment of the dThd kinase activity in the presence of bulky side chains (Fig. 5). Indeed, the 5-methyl group of dThd or any other substituents at this position would clash with large amino acid side chain residues. It is noticeable that the GCV activity varies among the bulky residues. The A167F mutation in TK presents the highest GCV activity. A167Y TK activity is lower than the latter, despite the side chains being of similar bulk. We suggest that tyrosine at 167 makes a hydrogen bond via its hydroxyl side chain to the purine ring and freezes GCV in a conformation not optimal for the phosphorylation. In conclusion, hydrophobic and bulky side chains at position 167 seem to be the best residues to select for GCV activity, apart from residues such as tryptophan that are probably too bulky to fit in the binding pocket. Adding any charged or polar residues does not improve GCV binding.
Mutations at residue 168 have a lower discriminating power, regarding GCV and dThd activities as substrates, compared with mutations at residue 167. Indeed, mutated residue side chains would be oriented in a way that is less disruptive for GCV and dThd binding (Fig. 6). Most of the mutations seem to affect the K m for dThd (from 3-to 4000-fold) but not the V max (Table  2); therefore, only the apparent binding affinity of dThd is affected rather than its rate of phosphorylation. Additionally, the same mutations only slightly affect GCV binding, meaning that any putative active site modifications are situated around dThd and not GCV. Mutations Lys, His, and Trp significantly increase the dThd K m , at least 1000 -4000-fold. In the case of Lys and His, it is interesting to note that both residues have nitrogen on their side chain, which appears somehow to be involved in interference with dThd binding and not GCV binding presumably because the latter has no oxygen to make a bond with the mutated residue.
Profound structural insights in the substrate active site of HSV-1 TK are required to enable rational development of novel enzyme constructs that recognize new nucleoside analogues as a substrate or improve the substrate affinity and/or catalytic activity of existing substrates. Because combined gene/chemotherapy of cancer makes use of the HSV-1 TK gene as a suicide gene to be used in combination with anti-herpetic drugs such as GCV or ACV, our findings may have a direct application in this field. Indeed, mutant HSV-1 TKs with substantially high K m values for the natural substrate dThd can have a distinct advantage in preferentially phosphorylating purine nucleoside analogues such as GCV, ACV, or PCV because of reduced competition with dThd for the active site and a less pronounced feedback inhibition by dTTP, which also binds to the dThd site. Because GCV and ACV have markedly higher K m values for wild-type HSV-1 TK than dThd (ϳ100-and 1,000-fold, respectively), knocking out of the pyrimidine nucleoside phosphorylation site may become crucial to guaranteeing an efficient phosphorylation of the purine nucleosides in the in vivo (tumor) environment. In fact, our preliminary investigations reveal that uncloned human osteosarcoma (OST TK Ϫ ) tumor cell cultures transduced by the mutant A168H HSV-1 TK gene construct became markedly more sensitive to the cytostatic activity of GCV, PCV, and ACV than the parent cell cultures. Moreover, the addition of 50 M thymidine or BVDU did not reverse the cytostatic activity of these test compounds against the OST TK Ϫ /HSV-1 A168H TK tumor cells under conditions where GCV, PCV, and ACV toxicity could be reversed by dThd in osteosarcoma tumor cell cultures transduced by the wild-type HSV-1 TK gene (data not shown). These preliminary findings validate the beneficial kinetic properties of the mutant A168H HSV-1 TK gene construct. The novel mutant enzyme constructs reported in this study may also prove their value in imaging reporter gene expression in vivo through non-invasive positron emission tomography (PET) scan detection of fluorinated ( 18 F) nucleoside analogues (in particular F-GCV and F-PCV). Our findings that the fluorinated PCV is equally efficiently converted to its monophosphate derivative as the parent  unsubstituted PCV against A168H TK 3 opens interesting perspectives for the mutated A168H (or A167F) HSV-1 TK construct to be used for such an application. Also, other indications, including the prevention of graft versus host disease using a suicide gene expressed in T-lymphocytes (15,16), may make use of engineered enzyme constructs as described above.