Recombinant herpes simplex virus type 1 strains with targeted mutations relevant for aciclovir susceptibility

Here, we describe a novel reliable method to assess the significance of individual mutations within the thymidine kinase (TK) gene of herpes simplex virus type 1 (HSV-1) to nucleoside analogue resistance. Eleven defined single nucleotide polymorphisms that occur in the TK gene of clinical HSV-1 isolates and a fluorescence reporter were introduced into the HSV-1 strain 17+ that had been cloned into a bacterial artificial chromosome. The susceptibility of these different strains to aciclovir, penciclovir, brivudin, and foscarnet was determined with a modified cytopathic effect reduction assay. The strains were also tested for their aciclovir susceptibility by measuring the relative fluorescence intensity as an indicator for HSV-1 replication and by quantifying the virus yield. Our data indicate that the amino acid substitutions R41H, R106H, A118V, L139V, K219T, S276R, L298R, S345P, and V348I represent natural polymorphisms of the TK protein, whereas G61A and P84L mediate broad cross-resistance against aciclovir, penciclovir, brivudin, and susceptibility to foscarnet. This method allows the definition of the resistance genotype of otherwise unclear mutations in the TK gene of HSV-1. Thus, it provides a scientific basis for antiviral testing in clinical isolates of patients suffering from serious diseases and will facilitate testing of new antivirals against HSV-1.

en passant mutagenesis 26 , the native UL23 gene of the cloned strain HSV1(17 + )Lox was replaced by a modified gene containing the desired single nucleotide substitution. For this purpose, transfer plasmids were generated which harbor the native gene sequence as well as a kanamycin resistance gene as positive selection marker. Eleven point mutations were generated by site-directed mutagenesis (R41H, G61A, P84L, R106H, A118V, L139V, K219T, S276R, L298R, S345P, and V348I). The UL23-kanamycin cassettes were amplified and the respective PCR products were integrated in the UL23 deletion site that had been generated before. All virus variants were equipped with an EGFP fluorescence cassette whereby EGFP is driven by the elongation factor 1 alpha (EF-1α ) promoter. The EGFP cassette was integrated between the UL55 and UL56 genes allowing transgene expression 25,35 . The integrity of all BACs was confirmed by restriction fragment length polymorphism analysis and sequencing of the respective region (data not shown). Vero cells were used for BAC transfection and first individual transfected cells were observed already one day post transfection. Viral progeny was cultivated for several passages and the integrity of the desired nucleotide substitutions was validated by sequence analysis (data not shown). For each strain replication kinetics were determined and compared to the kinetics of its fluorescent counterpart as well as to the recombinant strain HSV1(17 + )Lox. Only minor differences were observed that were most likely due to small variations in the starting inoculum ( Fig. 1).

Analysis of TK gene expression.
The expression from the TK gene was analysed by PCR from cDNA generated from whole-cell RNA extracts obtained one day post infection (p.i.). The detection of glyceraldehyde 3-phosphate dehydrogenase (GAPDH) exon-specific transcripts, as control, ensured correct cDNA synthesis from mRNA (Fig. 2a). Furthermore, immunoblotting and immunofluorescence microscopy demonstrated that mutated TK proteins of approximately 45 kDa had been expressed in cells infected by each virus strain variant (Fig. 2b,c). The viral glycoprotein D (gD) and beta-actin served as viral and host loading controls, respectively. Susceptibility testing to several antivirals. The generated strains were tested for their susceptibility to ACV, PCV, BVDU, or FOS in comparison to the reference isolate HSV-1 MI and the parental HSV1(17 + )Lox strain in cytopathic effect (cpe) inhibition tests ( Table 1). The TK variants with the amino acid substitutions R41H, R106H, A118V, L139V, K219T, S276R, L298R, S345P, and V348I had lower half-maximum virus inhibition values (effective concentration 50% -EC 50 ) ranging from < 0.28 to < 0.58 μ M ACV, from < 0.88 to ≤ 2.67 μ M PCV, < 0.19 μ M BVDU, and 84.80 ± 7.30 to 211.60 ± 31.60 μ M FOS, respectively. These HSV-1 mutants were, therefore, considered susceptible to ACV, PCV, BVDU, and FOS based on the comparison to the EC 50 values of the susceptible reference strains and the resulting cut-offs. However, the two amino acid substitutions G61A and P84L led to resistance against ACV, PCV, and BVDU with EC 50 The TK-mutated strains R41H, G61A, P84L, R106H, A118V, L139V, K219T, S276R, L298R, S345P, and V348I  (dotted lines) were compared with their EGFP-tagged counterparts (solid lines). The growth kinetic of the HSV1(17 + )Lox strain is depicted separately (dashed line). The data were plotted as a mean and SD of triplicates.  Table 1).
Quantitative PCR was also performed to support the results obtained from the cpe inhibition assay exemplarily for ACV. The recombinant HSV1(17 + )Lox strain was tested with the full range of ACV concentrations indicating a distinct decline of viral genomes under increasing concentrations (Fig. 3a). Comparably low amounts of genomes from sensitive-classified TK variants were obtained at high ACV concentrations of 35.2 μ M (values between ≈10 4 and ≈10 5 ). Samples of the resistant strains, in contrast, exhibited a viral load exceeding the susceptible ones by two log levels ( Fig. 3b) and being comparable to viral loads observed in ACV absence.

Discussion
So far, molecular diagnostics of HSV-1 resistance remains challenging. This is due to the high degree of sequence variations within the UL23 gene, and various time-consuming in vitro experiments are required to determine the contribution of single TK mutations on antiviral resistance [21][22][23]36,37 . However, the missing correlation between enzyme activity levels and their relative impact on the viral fitness and viral replication characteristics remain problematic. Susceptibility testing of HSV-1 strains harboring the respective mutations in TK but with the full spectrum of gene expression could solve this challenge and lead to a more reliable calculation of EC 50 values for various antivirals. We have therefore generated several novel recombinant HSV-1 strains by exchanging single amino acids in the viral TK in the BAC-cloned genome of HSV1(17 + )Lox to test their susceptibility against different antivirals in a systematic and quantitative manner.
The BAC technology enables a fast and precise manipulation of herpesvirus genomes by a wide spectrum of molecular biology methods 24,38 . In particular, the method of en passant mutagenesis enables the direct manipulation of the BACs without leaving any scar sequences behind 26 . By this way, specific targeted point mutations have been introduced into the BAC-cloned genome of HCMV, HSV-1, or varicella-zoster virus [39][40][41] . Considering the multiplicity of nucleotide exchanges that had to be generated in this study, we engineered a transfer vector to substitute the native with the mutated TK genes. This procedure enabled the exchange of nucleotides in close proximity using shorter primers. A disadvantage of this strategy is the required additional step of site-directed mutagenesis and the associated longer duration of handling time. Furthermore, an EGFP cassette was introduced between the HSV-1 genes UL55 and UL56 25,35 . The integration of this reporter facilitated the phenotypic testing by measuring the relative fluorescence intensity as an indicator for HSV-1 replication. After BAC transfection of permissive Vero cells, novel recombinant HSV-1 strains with the single amino acid exchanges R41H, G61A, P84L, R106H, A118V, L139V, K219T, S276R, L298R, S345P, and V348I were generated. These substitutions were selected since they have recently been described as novel polymorphism-or resistance-related substitutions, or because their significance for antiviral resistance could not have been clarified in previous studies 27,28,31,32 . Replication kinetics were determined for the non-fluorescent mutant strains and their fluorescent counterparts as well as the parental HSV1(17 + )Lox strain to exclude any impairment by the EGFP reporter.
The vector backbone is self-excised by the integrated Cre recombinase that cuts and recombines at the integrated loxP sites but leaves a 34 bp scar in the form of one loxP site downstream of the UL23 gene 24,42,43 . Thus, the promoter sequences of the TK gene were not interrupted and the remaining loxP site, therefore, should not interfere with gene expression. Transcription analysis, immunoblotting, and immunofluorescence microscopy   confirmed the expression of the different TK proteins among the novel recombinant HSV-1 strains. The similar HSV-1 replication characteristics in combination with TK protein expression provided a reliable, stable genetic background to directly compare susceptibility of the different TK mutants to ACV, PCV, or BVDU. An established cpe inhibition assay was converted to a quantitative assay by combining it with a commercially available cell proliferation assay. This assay that has been shown to reliably test phenotypic resistance of HSV in several studies was used as reference method to evaluate the resistance phenotype of the mutants in comparison to the susceptible HSV-1 isolates MI and (17 + ) 20,27,28,31 . Additionally, the results were confirmed for ACV by calculating the EC 50 values on the basis of relative fluorescence intensity related to the viral replication capacity. These direct measurements represent a suitable and fast system for future phenotypic susceptibility testing of recombinant HSV-1 strains. Previously described novel or unclear amino acid substitutions such as R106H 27 , A118V 27 , L139V 27 , K219T 28 , S276R 28 , L298R 27 , S345P [28][29][30] , and V348I 27,32-34 were shown here to be natural gene polymorphisms resulting in susceptibility to the antivirals used in the clinics. With one exception (K219T), these substitutions are located in non-conserved TK regions. The substitution K219T 28 localised within a conserved domain did not increase the EC 50 values for ACV, PCV, or BVDU. The viral genome copies at high ACV concentrations raised only by approximately one log when compared to the sensitive strains. Thus, these data show that individual substitutions clustered within conserved TK gene regions do not necessarily result in a resistance phenotype 3 . The presumed natural polymorphism R41H 31,32,44 clearly was not associated with any resistance in this study. These results are in accordance with previous findings 45 . Interestingly, based on a mass spectrometry assay using as functional TK activity assay, R41H has been suggested to be a resistance-related amino acid exchange 30 . Thus, the results of individual functional HSV-1 TK assays should be interpreted with caution and confirmed by the generation of replication-competent recombinant viruses as shown in this study. By contrast, the amino acid substitutions G61A and P84L 46 caused resistance against ACV, PCV, and BVDU with high EC 50 values far above the cut-offs. Quantitative PCR from samples obtained two days p.i. revealed increased amounts of virus genomes exceeding those of susceptible strains by approximately 1.5 to 2 log (Fig. 3b). The observed differences of the viral load in the HSV-1 strains classified as bearing TK natural polymorphism may reflect different TK enzyme activities as shown by Sauerbrei et al. 45 but may also come from experimental variations 15 . The G61A substitution is located within the glycine-rich P-loop of the ATP-binding site 13 , whereas P84L is located in the highly conserved region of codons 83 to 88 3,18 . Concerning G61A, this substitution has originally been described in an ACV-susceptible clinical HSV-1 strain and, therefore, regarded as natural polymorphism 31 . However, because of its localization, there was great doubt about these findings. Using a functional TK assay 45 , G61A was also classified as resistance-related substitution. In detail, after expression of recombinant TK protein harboring the G61A substitution, no TK enzyme activity could be detected on the basis of an enzyme linked immunosorbent assay using bromodeoxyuridine as TK substrate (unpublished data). Finally, our results confirm previous findings showing that TK mutations in clinical HSV strains resulting in resistance to ACV are also associated with cross-resistance to other nucleoside analogues such as PCV and BVDU 27 . By contrast, all strains were susceptible to FOS. Independently of the viral TK, this pyrophosphate analogue inhibits the viral DNA polymerase 47 that is not affected in the recombinant HSV-1 strains of this study.
In conclusion, based on the BAC-cloned HSV1(17 + )Lox parental strain, a system was developed to reliably quantify single amino acid exchanges in the TK protein of HSV-1 as being related to natural gene polymorphisms or to the resistance phenotype. Furthermore, this system allows the introduction of mutations alone or in combination elsewhere in the viral genome, thereby, facilitating molecular diagnostics as well as the testing of antivirals with different modes of action. The BAC system might be useful to generate HSV-1 recombinants containing the UL23 gene from clinical isolates, thus allowing for antiviral testing of the complete spectra of the viral population including minority mutants. The present results add important insights to the recently published database of non-synonymous mutations of TK gene of HSV-1 whose association to resistance or natural gene polymorphism has been clarified by phenotypic or/and functional assays 3 . Moreover, the generation of specifically mutated BAC-derived viruses should also allow to define the precise resistance phenotype in more complex genotypic alterations such as several exchanges within the TK protein or in combination with polymerase variants and will allow more specific therapeutic decisions. Plasmids. The transfer plasmids for the UL23 gene-kanamycin cassette were constructed by amplifying the native gene from the BAC HSV1(17 + )Lox 25,35 followed by cloning of the SalI and XmaJI digested amplicon into the plasmid pCeu2 48 . The kanamycin resistance gene was amplified from pEPkan-S 26 and ligated into the unique BglII-site resulting in the plasmid pCeu2-UL23(kan r ). Based on this plasmid, nucleotide substitutions were introduced by the GeneArt site-directed mutagenesis system kit (Thermo Fisher Scientific, Waltham, MA, USA). The cloning strategy is depicted in the Supplementary Figure,

Construction of recombinant TK-mutated BACs by en passant mutagenesis.
The BAC HSV1(17 + )Lox, derived from the HSV-1 strain 17 + , carries the BAC vector pBeloBAC11 between the UL22 and UL23 gene 25 . The vector is flanked by two loxP-sites and comprises the eukaryotic Cre recombinase expression cassette which results autonomously in vector self-excision after transfection into eukaryotic cells 25,42,43 . The BAC was stably propagated in E. coli strain GS1783 to enable en passant mutagenesis 26,49 . The reporter gene cassette for the EGFP was amplified from pORFepEGFP-in (plasmid vector harboring the EF-1α promoter, the open reading frame of EGFP interrupted by an aphAI site and an SV40 poly(A) signal) and inserted between the UL55 and UL56 genes resulting in pHSV1(17 + )Lox-pEF-1α GFP. In both, the parental pHSV1(17 + )Lox and the fluorescence-tagged pHSV1(17 + )Lox pEF-1α GFP BACs, the UL23 genes were deleted resulting in pHSV1Δ 23 and pHSV1Δ 23-GFP, respectively. Subsequent insertion of the modified UL23 genes by the transfer plasmids provided traceless substitutions of the parental with the mutated UL23 genes. The strategy of BAC mutagenesis is depicted in detail in the Supplementary figure and the oligonucleotides used for en passant mutagenesis are provided in the Supplementary table. Virtual cloning and sequence analyses were performed with the Vector NTI Advance 11.1 software (Thermo Fisher Scientific).
Reconstitution of HSV-1 from BACs and one-step replication kinetics. The BAC DNA was purified using the Plasmid Maxi Kit (Qiagen, Hilden, Germany) and transfected into Vero cells by Lipofectamine 2000 (Thermo Fisher Scientific) according to the manufacturer's instructions. The obtained virus strains were cultivated over several passages to obtain high-titred stocks (≈ 10 8 plaque forming units/ml) and to guarantee vector self-excision by the Cre recombinase 42,43,50 . The integrity of the desired mutation was confirmed by sequencing (data not shown). One-step growth curves were performed to analyse the replication properties by plotting the viral yields against the time. Vero cells were infected in triplicates with a multiplicity of infection (MOI) of 5 and incubated for 48 hours. Portions of the culture supernatants were taken at several time points and titrated on Vero cells. Plaques of non-fluorescent strains were counted after crystal violet staining, whereas fluorescence-tagged strains were assessed with a fluorescence microscope (Olympus, Hamburg, Germany).

Analysis of TK gene expression by RT-PCR, immunoblot, and fluorescence microscopy.
With respect to our previous study 41  infected (MOI 0.1) and mock-infected Vero cells were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and blotted on polyvinylidene fluoride membranes (Millipore, Darmstadt, Germany), and viral and host protein expression were analysed with primary antibodies against HSV-1 TK (PAb, sc-28037, Santa Cruz, CA, USA), against gD (MAb, sc-21719, Santa Cruz), and against beta-actin (MAb, 4970, Cell Signaling Technologies, Frankfurt, Germany), and secondary alkaline phosphatase-conjugated antibodies (Jackson ImmunoResearch, Suffolk, UK). Chemiluminescent signals were obtained with the SuperSignal West Pico Chemiluminescent Substrate (Thermo Fisher Scientific) and visualised with a LAS-3000 CCD camera system (Fujifilm, Düsseldorf, Germany).
Immunofluorescence microscopy was performed with infected cells on cover slips as described previously 41 . The viral TK was stained with the primary anti-TK antibody (sc-28037) and the Alexa-568 coupled secondary antibody (A-11057, Life Technologies, Darmstadt, Germany), and plaques were imaged with the IX80 inverted fluorescence microscope (Olympus).
Susceptibility testing of viral strains. A modified reduction assay for cpe based on the red tetrazolium dye was used to test the susceptibility of fluorescent-tagged recombinant HSV-1 strains against the antivirals ACV (GlaxoSmithKline, Uxbridge, UK), PCV (GlaxoSmithKline, Uxbridge, UK), BVDU ((E)-5-(2-bromovinyl)-2' deoxyuridine, Berlin-Chemie AG, Berlin), or FOS (trisodium phophonoformate, AstraZeneca, Wilmslow, UK) as reported before 16 . In short, Wi 38 cells were seeded at a density of 1 × 10 5 cells ml −1 in 96-well flat bottom microtitre plates and cultured for two days. The cells were infected with a MOI of 0.01 and the antiviral compounds were added at a final half log dilution over a range of 0.275 to 35.2 μ M ACV, 0.25 to 31.6 μ M PCV, 0.19 to 24.0 μM BVDU, and 13.3 to 844.8 μ M FOS. After incubation for 5 days, the virus-induced cpe were assessed by measuring cell proliferation with the Cell Counting Kit-8 (Dojindo Laboratories, Kumamoto, Japan). The EC 50 values were calculated by linear regression analysis using the software SigmaStat, version 1.01 (Jandel Corporation, San Rafael, CA, USA). All experiments were done in independent triplicates and the mean EC 50 and standard deviation (SD) were calculated for each antiviral compound. Resistance to ACV, PCV, or BVDU was defined if the mean EC 50 including SD of the viral strains tested were measured at least five times higher the mean value of the included susceptible control strains HSV-1 MI and HSV1(17 + )Lox 51 . For resistance to FOS, EC 50 values of ≥ 330.0 μ M were considered 52 . In comparison, the recombinant EGFP-tagged HSV-1 strains were tested against a range of ACV by measuring the RFU one day p.i. under same conditions. The EC 50 s were calculated directly as mean and SD from RFU values.
The quantitative real-time PCR (qPCR) was performed from whole cell DNA extracts applying the DNeasy Blood and Tissue Kit and the QuantiTect Probe Kit (Qiagen) as described previously 41 . In brief, Vero cells were infected with a MOI of 0.01. The susceptible recombinant strain HSV1(17 + )Lox was treated with a range of ACV concentrations between 0.275 and 35.2 μ M whereas all other recombinant strains were tested in absence or in presence of 35.2 μ M ACV. The cells were incubated 24 hours prior to DNA extraction from whole cell lysates. Viral genomes were quantified from 3.3% of the eluted volume by a TaqMan probe and oligonucleotides targeting UL30 53 . Purified BAC DNA of HSV1(17 + )Lox with known size and concentration was 10-fold diluted and applied for standard curves by plotting the cycle thresholds (Ct) against the logarithm of the starting amount. The DNA standard and samples were measured in triplicates and the amount of genomes were computed as mean and SD from one representative experiment.