In Vitro Selection of External Guide Sequences for Directing RNase P-mediated Inhibition of Viral Gene Expression* 210

External guide sequences (EGSs) are small RNA molecules that bind to a target mRNA, form a complex resembling the structure of a tRNA, and render the mRNA susceptible to hydrolysis by RNase P, a tRNA processing enzyme. An in vitro selection procedure was used to select EGSs that direct human RNase P to cleave the mRNA encoding thymidine kinase (TK) of herpes simplex virus 1. One of the selected EGSs, TK17, was at least 35 times more active in directing RNase P in cleaving TK mRNAin vitro than the EGS derived from a natural tRNA sequence. TK17, when in complex with the TK mRNA sequence, resembles a portion of tRNA structure and exhibits an enhanced binding affinity to the target mRNA. Moreover, a reduction of 95 and 50% in the TK expression was found in herpes simplex virus 1-infected cells that expressed the selected EGS and the EGS derived from the natural tRNA sequence, respectively. Our study provides direct evidence that EGS molecules isolated by the selection procedure are effective in tissue culture. These results also demonstrate the potential for using the selection procedure as a general approach for the generation of highly effective EGSs for gene-targeting application.

Antisense technology has been shown to be a promising gene-targeting approach for use in basic research and clinical therapeutic applications. The gene-targeting agents used can be a conventional antisense oligonucleotide, an antisense catalytic molecule (ribozyme or DNA enzyme), or an antisense molecule with an additional (guide) sequence that targets the mRNA for degradation by endogenous RNases such as RNase L and RNase P (1)(2)(3)(4)(5)(6). Antisense molecules with guide sequences have several unique features as gene-targeting agents. Targeting with these molecules results in irreversible cleavage and the cleavage can be in a catalytic fashion. Moreover, this targeting approach uses the cellular endogenous RNases (e.g. RNase P) for degradation of the target mRNA and, therefore, assures the stability and efficiency of the targeting enzymes in the cellular environment.
Ribonuclease P (RNase P) is a ribonucleoprotein complex found in all organisms examined. It is one of the highly active enzymes in cells and is responsible for the maturation of 5Ј termini of all tRNAs, which account for approximately 2% of total cellular RNA (7,8). This enzyme catalyzes a hydrolysis reaction to remove the leader sequence of precursor tRNA (9). Human RNase P has at least nine polypeptides and a RNA subunit (H1 RNA) (7,10). One of the unique features of RNase P is its ability to recognize the structures, rather than the sequences, of the substrates, which allows the enzyme to hydrolyze different natural substrates in vivo or in vitro. Accordingly, any complex of two RNA molecules that resembles a tRNA molecule can be recognized and cleaved by RNase P (Fig.  1, A and B) (11,12). One of the RNA molecules is called the external guide sequence (EGS). 1 In principle, an mRNA sequence can be targeted for RNase P cleavage by using EGSs to hybridize with the target RNA and direct RNase P to the site of cleavage. The EGSs used to direct human RNase P for targeted cleavage resemble three-quarters of a tRNA molecule and consist of two sequence elements: a targeting sequence complementary to the mRNA sequence and a guide sequence, which is a portion of the natural tRNA sequence and is required for RNase P recognition ( Fig. 1B) (11,12). Subsequent studies have shown that expression of EGSs in human cells can reduce the expression of both cellular and viral genes (11)(12)(13)(14)(15). For example, we have previously shown that EGSs efficiently direct human RNase P to cleave the mRNA sequence encoding the thymidine kinase (TK) of herpes simplex virus 1 (HSV-1) in vitro (16). A reduction of 50 -70% in the TK mRNA and protein expression was observed in HSV-1-infected cells expressing the EGSs.
Targeted cleavage of mRNA by RNase P using EGSs provides a unique approach to inactivate any RNA of known sequence expressed in vivo. Further studies aimed at increasing the targeting activity of the EGSs are needed to develop the EGS-based technology as a general tool for use in gene-targeting applications. Although little is known about the rate-limiting step of EGS-targeting reactions in cultured cells, we believe that binding of the EGSs to the target mRNA as well as the efficiency of cleavage are important for the efficacy of the * This work was supported in part by a Chancellor's special initiative award (University of California, Berkeley) and by National Institutes of Health Grants AI43250 and DE14145. 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. EGSs. Indeed, recent studies on ribozymes and antisense phosphothioate molecules suggest that binding of these molecules to their target RNAs appears to be rate-limiting in vivo (17)(18)(19)(20)(21). In our previous studies, EGS RNAs were used to target a region of TK mRNA that is accessible to modification by dimethyl sulfate in cell culture and is also accessible to EGS binding (16). Moreover, the EGSs were expressed primarily in the nuclei by using the promoter of small nuclear U6 RNA (12,16,22). This design would increase the probability for the constructed EGSs to bind to its target mRNA sequence and co-localize with RNase P, which is localized exclusively in the nuclei. Under such conditions, it is possible that the EGS efficacy in culture cells is dictated by the overall efficiency (V max /K m ) of the EGS-induced RNase P cleavage. If this is the case, increasing the targeting activity of the EGS may lead to a more effective inhibition of the target mRNA expression in cultured cells.
In the present study, we employed an in vitro selection procedure (23)(24)(25)(26) to isolate highly active EGSs from an EGS pool that contained random mutations. One of the selected EGSs exhibited a targeting activity at least 35 times higher than that of the EGS derived from the natural tRNA sequence. When the selected EGS was expressed in cells infected with HSV-1, a reduction of 95% in TK expression was observed. These studies demonstrate the feasibility of developing effective EGSs for gene-targeting applications.

Generation of Substrate and EGS RNAs and RNase P Cleavage in
Vitro-DNA template coding for substrate tk46 was constructed by annealing oligonucleotide OliT7 (5Ј-TAATACGACTCACTATAG-3Ј) with OliTK46 (5Ј-ACCGCGCAGCCTGGTCGAACGCAGACGCGTGTT-GATGGCAGGGGTCTATAGTGAGTCGTATTA-3Ј). The DNA sequence coding for the EGS TK-Ser was synthesized by the polymerase chain reaction (PCR), using DNA that encodes yeast tRNA Ser as the template (27), and was cloned under the control of the T7 RNA polymerase promoter. The 5Ј and 3Ј primer were oligo1 (5Ј-GTTAACGTCGGACA-GACTC-3Ј) and oligo2 (5Ј-AAGCTTTAAACGTCTGCGGCAGGATTTG-3Ј), respectively. DNA sequence coding for TK17 was generated by PCR using the selected substrate construct C17 as the template, and oligonucleotides TK24EGS (5Ј-GGAATTCTAATACGACTCACTATAGGTT-AACGTC-3Ј) and TK23 (5Ј-AAACGTCTGCG-3Ј) as the 5Ј and 3Ј primers, respectively. C-TK-Ser and C-TK17 were derived from TK-Ser and TK17, respectively, and contained point mutations (5Ј-UUC-3Ј 3 AAG) at the three highly conserved positions in the T loop of these EGSs (see Fig. 4, A and C). The DNA sequences coding for EGS C-TK-Ser and C-TK17 were generated by introducing point mutations into the TK-Ser and TK17 sequences, as described previously (16).
Human RNase P was prepared from HeLa cellular extracts as described previously (10,12). EGS RNAs and substrate tk46 were synthesized in vitro by T7 RNA polymerase and further purified on 8% urea-polyacrylamide gels. Subsequently, the EGS RNAs and 32 P-labeled tk46 were incubated with human RNase P. The cleavage reactions were carried out at 37°C in a volume of 10 l in buffer A (50 mM Tris, pH 7.4, 100 mM NH 4 Cl, and 20 mM MgCl 2 ) (12). Cleavage products were separated in denaturing gels and analyzed with a STORM840 PhosphorImager (Amersham Biosciences).
In Vitro Selection-Double-stranded DNA templates were synthesized by mouse Moloney leukemia virus reverse transcriptase (Roche Molecular Biochemicals), from two DNA oligonucleotides, TK21 (5Ј-G-GAATTCTAATACGACTCACTATAGACCCCTGCCATCAACACGCGT-CTGCGTTCGACCAGGCTGCGCGGTTAACGTCG-3Ј) and TK22 (5Ј-AAACGTCTGCGNNNNNNNNNNNNNNNNNNNNNNNNNCCG-ACGTTAA-3Ј). TK21 contains the leader sequence for the substrate (Fig. 1E) and the promoter sequence for T7 RNA polymerase. TK22 contains a randomized sequence indicated as (N) 25 and underlined (Fig.  1E). The general strategy for the design of these oligonucleotides followed that described by Yuan and Altman (6). The RNA substrate was incubated at 37°C with human RNase P in buffer A. The 3Ј proximal cleavage product was separated in and isolated from a denaturing 8% polyacrylamide gel, and was then used as template for reverse transcription to generate cDNA, in the presence of 20 mM primer oligodeoxynucleotide, TK23 (5Ј-AAACGTCTGCG-3Ј), and avian myeloblastosis virus reverse transcriptase (Roche Molecular Biochemicals). The mixture was incubated for 2 h at 42°C. The cDNA was subsequently amplified by PCR with oligodeoxynucleotide primers TK21 and TK23 and was used to generate RNA substrates for the next round of selection. TK21, the 5Ј primer for PCR, allows restoration of the T7 promoter sequence and the leader sequence of the RNA substrates.
Initially, 20 nmol of the pool of RNA substrates, which contained the randomized sequence and were synthesized in vitro, were digested by human RNase P in a volume of 2 ml. In subsequent cycles of selection, 5 pmol of substrate were used and digested with the appropriate enzyme in a volume of 20 l. During the first four cycles of selection, substrates were incubated for 120 min at 37°C with 100 units of RNase P. During the final five cycles of selection, incubation time was shortened to 10 min. The amount of human RNase P used was reduced by 200-fold. This strategy allowed enrichment of all sequences that exhibited similar susceptibility to cleavage. After nine cycles of selection, cDNA that contained substrate sequences was cloned into pUC19 and sequence analysis was performed.
Kinetic and Binding Analysis and RNase Structural Mapping-RNase T1, nuclease S1, and RNase V were purchased from Amersham Biosciences or Invitrogen. The procedures for mapping the structure of tk46-EGS complexes were carried out as described previously (6,28). RNases were diluted and incubated at 37°C for 10 min with the tk46-EGS complexes in buffer A that had been supplemented with 10 g of bulk tRNA from Escherichia coli in a final volume of 10 l. Cleavage products were separated in 10% polyacrylamide gels that contained 7 M urea.
RNA substrates were incubated in buffer A (50 mM Tris, pH 7.4, 100 mM NH 4 Cl, and 20 mM MgCl 2 ) at 37°C with human RNase P and cleavage products were separated on 8% denaturing gels and quantitated with a STORM840 PhosphorImager. Assays to determine kinetic parameters were performed under multiple turnover conditions, as described previously (29 -31) (see supplemental data available on-line). In brief, the cleavage of substrates was assayed in buffer at various concentrations of substrates, both above and below the K m for each respective substrate. The amount of substrates are in large excess to that of the enzymes to assure that saturation with the substrate was achieved in the multiple-turnover conditions (see supplemental data). Aliquots were withdrawn from reaction mixtures at regular intervals and analyzed in polyacrylamide-urea gels. Values of K m(apparent) and V max(apparent) were obtained from Lineweaver-Burk double-reciprocal plots.
The procedures to measure the equilibrium dissociation constants (K d ) of complexes of the EGSs and the substrates were modified from Pyle et al. (32). In brief, various concentrations of EGSs were preincubated in buffer B (50 mM Tris, pH 7.5, 100 mM NH 4 Cl, 20 mM MgCl 2 , 3% glycerol, 0.1% xylene cyanol, 0.1% bromphenol blue) for 10 min before mixing with an equal volume of different concentrations of substrate RNA preheated under identical conditions. The samples were incubated for 10 -120 min to allow binding, loaded on a 5% polyacrylamide gel, and run at 10 watts. The electrophoresis running buffer contained 100 mM Tris-Hepes, pH 7.5, and 10 mM MgCl 2 (32). The value of K d was then extrapolated from a graph plotting percentage of product bound versus EGS concentration. The values were the average of three experiments.
Construction of Cell Lines Expressing EGSs-Constructs pTK-Ser, pC-TK-Ser, pTK17, and pC-TK17 were generated by placing the DNA sequences that coded for EGS TK-Ser, C-TK-Ser, TK17, and C-TK17 under the control of the U6 promoter, respectively (16). Human 143tk Ϫ cells were cotransfected with pTK116 (containing the neomycin-resistant gene) and the EGS plasmids, with the aid of a mammalian transfection kit (Stratagene Inc., La Jolla, CA). At 48 h after transfection, neomycin (Invitrogen) was added to the culture medium in a final concentration of 400 g/ml. Cells were subsequently selected under neomycin for 2 weeks and were eventually cloned. The level of EGS expression in individual cell clone was determined by Northern analyses with probes complementary to EGSs. Only those cell clones that expressed similar levels of EGSs were used for subsequent experiments.
Viral Infection and Assays for TK Expression-Approximately 10 6 cells in a T25 flask were either mock-infected or infected with HSV-1 in 1.5 ml of Medium 199 at a multiplicity of infection (m.o.i.) as specified under "Results." At 8 -16 h after infection, cells were harvested and RNA and protein extracts were prepared as described previously (33). The RNA probes used to detect TK mRNA and the transcripts of the ␣47, Us10. Us11 genes were synthesized from pTK129 and pTK141, and RNase protection assays were performed as described previously (34). The protected RNA products were separated in 8% urea-polyacrylamide denaturing gels and quantitated with a STORM840 PhosphorImager.
The denatured, solubilized polypeptides from cell lysates were separated on 9% (v/v) SDS-polyacrylamide gels cross-linked with N,NЉmethylenebisacrylamide (16). The separated polypeptides were transferred electrically to nitrocellulose membranes and reacted with the antibodies against HSV-1 TK (16) or ICP27 (purchased from Goodwin Institute of Cancer Research, Plantation, FL). The membranes were subsequently stained with a chemiluminescent substrate with the aid of a Western chemiluminescent substrate kit (Amersham Biosciences) and quantitated with a STORM840 PhosphorImager. Quantitation was performed in the linear range of RNA and protein detection (e.g. 2-fold changes in RNA and protein samples result in a 2-fold change in signal bracketing the range of experimental values; see supplemental data).

In Vitro Selection of EGS RNAs-
The RNA substrate tk46 used in the selection experiment contains a 5Ј TK mRNA sequence of 46 nucleotides (Figs. 1E and 4A). This sequence has been shown to be accessible to modification by dimethyl sulfate and, presumably, to EGS binding in mammalian cell culture (16). We have previously showed that EGSs derived from ptRNA Ser can direct human RNase P to cleave tk46 in vitro and inhibit HSV-1 TK expression in cultured cells (16). However, the cleavage reaction is inefficient compared with the cleavage of a natural tRNA substrate (i.e. ptRNA Ser ). It is believed that the acceptor stem and D stem and loop sequences within a natural tRNA molecule are involved in tertiary interactions with different parts of the tRNA (e.g. variable and T stem and loop regions) and are important for folding of the tRNA and interactions with RNase P (7,8). The reduced susceptibility of the tk46-EGS complex to be cleaved by RNase P, compared with a natural tRNA, is probably a result of replacing the natural tRNA sequence at the acceptor and D stem regions with the tk46 and its complementary sequence in the complex of tk46 and the EGS. These substitutions may disrupt some of the tertiary interactions that are potentially important in maintaining the proper tRNA-like conformation and in recognition by human RNase P. Restoration of these interactions or introducing additional interactions by changing other parts of the EGS sequence, such as those resembling the variable region, and T stem and loop, may increase the susceptibility of the mRNA-EGS complex to be cleaved by RNase P.
To generate EGSs that are highly active in directing RNase P to cleave tk46, a pool of chimeric, covalently linked tk46-EGS substrates that contain partially randomized sequences was constructed (Fig. 1E) and selected for the ability to be cleaved by human RNase P (Fig. 1D). The chimeric RNA, with its 5Ј region consisting of the tk46 sequence, contains the sequences that base pair with tk46 at the regions resembling the acceptor and D stem of a tRNA and, in addition, a randomized sequence of 25 nucleotides at the positions corresponding to the regions resembling the variable and T stem and loop regions (Fig. 1E). The anticodon region, which is dispensable for EGS activity (6), was not included in the chimeric substrate (Fig. 1, C and E). The pool of tk46-EGS chimeric substrates was synthesized in vitro by T7 RNA polymerase. In each round of selection, the pool of RNAs was digested with human RNase P in buffer A (50 mM Tris, pH 7.4, 100 mM NH 4 Cl, and 20 mM MgCl 2 ) and the 3Ј cleavage products were isolated in denaturing gels. cDNA molecules were then synthesized and amplified from these RNA molecules by reverse transcription followed by PCR and used as the templates for synthesis of EGS RNA molecules for the next round of selection. The 5Ј primer for the PCR reaction contains the tk46 sequence as well as the T7 promoter sequence and, therefore, allows the restoration of these sequences for the next cycle of selection. The stringency of the selection was increased at each cycle by reducing the amount of human RNase P and the time allowed for the cleavage reaction, such that only those substrates that were rapidly cleaved by the enzyme were selected. The cleavage efficiency of the substrate population of each generation was monitored ( Fig. 2A). Moreover, EGSs were also constructed from the selected population of each generation and tested for their activity to direct human RNase P-mediated cleavage (Fig. 2B). These results indicate that the susceptibility of the chimeric substrate as well as the targeting activity of the EGSs increase with each selection cycle (Fig. 2). The selection procedure was repeated nine times, until no apparent enhancement of the cleavage efficiency of the substrate population was observed after a short period of incubation (10 min) ( Fig. 2A).
Sequencing and Kinetic Analyses of the Selected EGSs-Twenty-four sequences coding for the EGSs isolated after nine cycles of selection were cloned and determined (Table I). These EGSs were divided into two sets based on their primary nucleotide sequences. Each sequence in set 1 either had the same sequence or extensive homology to other sequences of the same set. In contrast, the three sequences in set 2 did not exhibit significant sequence homology to each other or to those in set 1.
In our selection procedure, the tk46-EGS chimeric substrates were selected for their susceptibility to be cleaved by human RNase P. To determine whether the selected substrates can be cleaved by RNase P, one of the selected substrates, C17, which is the most abundant selected sequence (Table I), was assayed for cleavage by RNase P. Kinetic analyses indicate that human RNase P cleaved substrate C17 as efficiently as ptRNA Ser (Table II), suggesting that C17 may possess proper tertiary interactions and conformations, which are found in natural tRNA substrate and are required for optimal recognition by human RNase P.
To analyze the relationship between the EGS sequences from the selected tk46-EGS substrates and the capabilities of these EGSs to direct human RNase P for targeting cleavage, an EGS sequence, TK17, was constructed from the sequence of C17. Substrate tk46 was incubated with TK17 in the presence of human RNase P, and the cleavage products were analyzed in denaturing gels (Fig. 3). Kinetic analyses of the cleavage reactions directed by the selected EGS TK17 as well as by EGS TK-Ser, which was derived from the ptRNA Ser sequence, were performed under multiple-turnover conditions. Under these conditions, the amounts of substrates were in large excess to that of the enzyme to assure that saturation with the substrate was achieved (see supplemental data). The values of K m(apparent) and V max(apparent) as well as the overall cleavage efficiency [V max(apparent) /K m(apparent) ] were determined (Table II). Under the selection conditions (i.e. buffer A), TK17 was extremely efficient in directing human RNase P to cleave tk46 (Fig. 3,  lane 3) and was at least 2 ϫ 10 4 -fold more active than the EGS molecules (i.e. TK-G 0 ) that were derived from the initial randomized tk46-EGS chimeric substrate pool (G 0 ) (Table II). Moreover, in the presence of TK17, RNase P-mediated cleavage of tk46 was at least 35-fold more efficient than that in the presence of TK-Ser (compare lanes 3 and 2) (Table II). These observations indicate that highly active EGS RNAs were successfully selected using the in vitro selection procedure.
Analysis of the Structure and Stability of the TK mRNA-EGS Complex-Because TK17 is among the most abundant sequence found after nine cycles of selection and exhibits highly active targeting activity among the selected EGSs tested (Fig.  3, Table II, and data not shown), this EGS was further characterized and expressed in tissue culture to determine its efficacy in inhibiting TK expression (see below).
In our selection, the mRNA-EGS chimeric molecules were subjected to digestion by RNase P and selected for their susceptibility to cleavage by RNase P. As the selection proceeded, the EGS sequence in the chimeric molecule was selected for its ability to complex with the mRNA to form a structure recognizable by RNase P and for its ability to direct RNase P to cleave the target RNA. Examination of the sequence of C17 and TK17 suggests that the sequence at the 25 randomized positions can fold into a structure resembling a variable region, a T loop of 7 nucleotides, and a T stem of 4 base pairs of a tRNA (Fig. 4C). The secondary structure of TK17 in complex with substrate tk46 was probed using an RNase mapping approach. Three RNases were used: RNase T1 and nuclease S1, which recognize single-stranded regions, and RNase V, which only cleaves at base-paired positions or positions involved in tertiary interactions. Most of these mapping results are consistent with the proposed secondary structure of the tk46-TK17 complex shown in Fig. 4C. The structures of the complex of tk46 with EGS TK-Ser as well as with EGS TK-Ser-D, which is derived from TK-Ser with the deletion of the anticodon domain, were also studied using an RNase mapping approach (Fig. 4, A  and B). These results indicate that the structures of the acceptor stem, T stem and loop, and the 5Ј leader sequence of the tk46-TK17 complex are similar to those of the complexes of tk46 with TK-Ser and with TK-Ser-D. Meanwhile, the structures of the 3Ј sequence of tk46 (5Ј-GGA-3Ј) (equivalent to the D loop region) and the EGS sequence (5Ј-GUG-3Ј) resembling the variable region in the complexes of tk46 with TK17 appear to be different from those in the complexes of tk46 with TK-Ser or with TK-Ser-D. For example, these two sequences in the tk46-TK17 complex were susceptible to digestion of RNase V, suggesting that they are involved in tertiary interactions. In contrast, the same regions in the complexes of tk46 with either TK-Ser or TK-Ser-D are susceptible to attack by RNase T1 and nuclease S1, an indication that they are in a single-stranded conformation. These results suggest that the presence of TK17 may lead to new tertiary interactions within the complex between tk46 and the EGS, which may enhance the rate of cleavage by RNase P.
An increase in the cleavage rate of RNase P may also be caused by additional tertiary interactions that may potentially stabilize the mRNA-EGS complex. If this is the case, it is expected that binding affinities of the selected EGS to the TK mRNA sequence may increase. The binding affinities of EGS TK17 as well as TK-Ser to substrate tk46, measured as the dissociation constant (K d ), were determined by separating substrate-EGS complexes in polyacrylamide gels under nondenaturing conditions (Table II). TK17 exhibited ϳ50 times higher binding affinity to tk46 than TK-Ser. Given the fact that both TK17 and TK17-Ser have the same antisense sequences (7 and 4 nucleotides, respectively) to tk46 (Fig. 4, A and C), these results strongly suggest that the increased binding affinity and the stability of the substrate-EGS complex in the presence of   Fig. 1E. The selected sequences have been divided into two sets based on sequence similarities.

TK17 is probably caused by the additional tertiary interactions introduced by this EGS.
Expression of the Selected EGS in Tissue Culture-To express the EGSs in cultured cells, the DNA sequences coding for TK-Ser and TK17 were subcloned and placed under the control of the small nuclear U6 RNA promoter, which has previously been shown to express EGSs and other RNAs steadily (12,15,16,22). This promoter is transcribed by RNA polymerase III, and its transcripts are highly expressed and primarily localized in the nucleus (35,36).
Two additional EGSs, C-TK-Ser and C-TK17, were also constructed and cloned under the control of the U6 RNA promoter. C-TK-Ser and C-TK17 were derived from TK-Ser and TK17, respectively, and contained point mutations (5Ј-UUC-3Ј 3 AAG) at the three highly conserved positions in the T loop of these EGSs (Fig. 4, A and C). These nucleotides were found in most of the known natural tRNA sequences (37) and are believed to be important for the interactions between the tRNA domains and human RNase P. Previous studies have shown that EGSs carrying these mutations precluded RNase P recognition and exhibited little activity in directing RNase P-mediated cleavage (6,16). Indeed, cleavage of tk46 by human RNase P in the presence of these two control EGSs was barely detected (Fig. 3, lanes 1 and 4) and was at least 2 ϫ 10 4 -fold slower than the cleavage in the presence of TK17 (Table II). C-TK-Ser and C-TK17 contain the same antisense sequence to the TK mRNA sequence as TK-Ser and TK17. Indeed, C-TK-Ser and C-TK17 exhibited similar binding affinities to tk46 as TK-Ser and TK17, respectively, when assayed in vitro (Table II). Therefore, C-TK-Ser and C-TK17 can be used as a control for the antisense effect of the guide sequence.
To construct cell lines that express EGSs, human 143tk Ϫ cells were cotransfected with each of these four EGS DNA constructs and a plasmid containing a neomycin resistance gene (pFL116) (16). These cells were then selected in culture medium that contained neomycin, and cells that exhibited neomycin resistance were cloned. The level of EGS RNA expression in individual cell clones was determined using Northern analysis with a DNA probe that is complementary to TK-Ser (Fig. 5A). The expression of actin mRNA was used as the internal control (Fig. 5B). As expected, the EGS RNAs were exclusively expressed in the nuclei, as they were only detected in the nuclear but not cytoplasmic RNA fractions (16) (data not shown). The constructed cell lines and a control cell line in which cells were transfected with the vector DNA were indistinguishable in terms of cell growth and viability for up to 1 month (data not shown), suggesting that the expression of the EGS RNAs did not result in significant cytotoxicity. Only the cell lines that expressed similar levels of these EGSs (as shown in Fig. 5A) were used for further studies in tissue culture.
Efficacy of the Selected EGS for Inhibition of TK Expression-To determine the efficacy of the EGSs in directing human RNase P for inhibiting TK expression, cells were infected with HSV-1 at an m.o.i. of 0.05-1. Levels of TK mRNA in the infected cells were determined by an RNase protection assay. The levels of the overlapping transcripts coding for viral ␣47, Us10, and Us11 proteins were used as the internal controls for quantitation of TK mRNA expression. Fig. 6 shows the results  [V max(apparent) , K m(apparent) , and V max(apparent) /K m(apparent) ] in the RNase P cleavage of tRNA Ser or tk46 in the presence of different EGSs Multiple-turnover kinetic analyses to determine the values of V max(apparent) and K m(apparent) were carried out in buffer A (50 mM Tris, pH 7.4, 100 mM NH 4 Cl, and 20 mM MgCl 2 ) at 37°C, as described previously (31,50). The amounts of substrates are in large excess to that of the enzyme in order to assure that saturation with the substrate was achieved in the multiple-turnover conditions (see supplemental data). To determine the binding affinity (K d ) between the tk46 substrate and EGSs, binding assays were carried out in the absence of human RNase P in buffer B (50 mM Tris, pH 7.5, 100 mM NH 4 Cl, 20 mM MgCl 2 , 3% glycerol, 0.1% xylene cyanol, 0.1% bromophenol blue), using a protocol modified from Pyle et al. (32). The values shown are the average derived from triplicate experiments.  1, 2, and 4) and 5 nM (lane 3) EGS were incubated with 32 P-labeled TK RNA substrate (10 nM) and either 4 units (lanes 1, 2, and 4)  (which are summarized graphically in Fig. 8) of the RNase protection experiments with both the TK and ␣47 probes. Quantitation was performed in the linear range of RNA detection (e.g. 2-fold changes in the amount of RNA samples result in a 2-fold change in signal bracketing the range of experimental values; see supplemental data). A reduction of approximately 96 Ϯ 3, 50 Ϯ 5% (average of three experiments) in the level of TK mRNA expression was observed in cells that expressed EGS TK17 and TK-Ser, respectively (Fig. 6, lanes 4 and 5). In contrast, cells that expressed C-TK17 and C-TK-Ser RNAs only exhibited a reduction of 10 Ϯ 5% (Figs. 6 (lane 3) and 8). The low level of inhibition found in cells that expressed C-TK17 and C-TK-Ser RNAs was probably caused by an antisense effect. This is because C-TK17 and C-TK-Ser, with the point mutations at the T loop, exhibit little targeting activity but bind to TK mRNA sequence as well as TK-Ser and TK17 (Table II). Thus, these observations suggest that the significant reduction of TK mRNA expression in cells that expressed TK17 and TK-Ser was caused by the RNase P-mediated cleavage of the target mRNA directed by these EGSs. No products of the cleavage of TK mRNA were detected in our RNase protection assays presumably because these RNAs, which lacked either a cap structure or a poly(A) sequence, were rapidly degraded by intracellular RNases. It is expected that the level of TK protein should decrease in EGS-expressing cells because of the decreased level of TK mRNA. Protein extracts were isolated from cells either mockinfected or infected with HSV-1. Viral proteins were separated electrophoretically in SDS-polyacrylamide gels and electrically transferred to two identical membranes. One of these membranes was stained with an anti-TK antibody (anti-TK) (Fig.  7B), and the other was stained with a monoclonal antibody against viral ICP27 protein (anti-ICP27) (38,39) (Fig. 7A). The latter is used to detect the expression of HSV-1 immediateearly protein ICP27, which serves as an internal control for the quantitation of TK protein expression. Quantitation was performed in the linear range of protein detection (e.g. 2-fold changes in the amount of protein samples result in a 2-fold change in signal bracketing the range of experimental values; see supplemental data). The results of three independent experiments are summarized in Fig. 8: a reduction of 95 Ϯ 4 and 45 Ϯ 5% in the level of TK protein was observed in cells that expressed TK17 and TK-Ser EGS RNAs, respectively. In contrast, a reduction of only 5-10% was seen in cells that expressed C-TK17 and TK-Ser RNAs. The low level of reduction in the expression level of TK protein observed in cells that expressed these two control EGSs was presumably attributed to the antisense effect of the EGSs.

DISCUSSION
The EGS-based technology represents an attractive approach for gene inactivation because it utilizes endogenous RNase P to generate highly efficient and specific cleavage of the target RNA. In particular, RNase P is among the most ubiquitous and active enzymes found in nature, as it is responsible for processing of all tRNA molecules (7,8). Recent studies have shown that expression of EGSs in human cells can reduce the expression of both cellular and viral genes (11)(12)(13)(14)(15). Moreover, RNase P-mediated cleavage directed by EGSs is highly specific and does not generate "irrelevant cleavage," which is usually observed with RNase H-mediated cleavage induced by conventional DNA-based oligonucleotides (7,14). Thus, EGS molecules represent promising general gene-targeting agents that can be used in both basic research and clinical applications.
Further studies on increasing the efficacy of the EGSs in inhibiting gene expression are needed to develop the EGS technology for practical gene-targeting applications. Little is known about the rate-limiting step of EGS-targeting reaction in cultured cells. Meanwhile, there is currently no comprehensive guideline about how to construct a highly active EGS molecule. In the present study, EGS RNAs were targeted to an accessible region of TK mRNA and were expressed by the small nuclear U6 RNA promoter. This design would increase the probability for the EGS RNAs to bind to their target mRNA sequence and co-localize with human RNase P, which is exclusively localized in the nuclei (7,8). Under the described settings, we hypothesized that the efficacy of EGS technology in cultured cells is dictated by the catalytic efficiency (V max /K m ) of RNase P-medi- ated cleavage directed by the EGS. If this is the case, increasing the activity of EGS in directing RNase P cleavage may lead to more effective inhibition of the target mRNA expression in vivo.
Novel EGS RNAs that exhibited higher targeting activity (V max(apparent) /K m(apparent) ) than that derived from a natural tRNA sequence were isolated from a pool of EGS RNA sequences that contained random mutations. EGS RNAs that exhibited better substrate binding and directed more efficient cleavage by RNase P were selected. EGS TK17 was highly active in vitro and inhibited TK expression in cultured cells by more than 95%. Indeed, this EGS was more effective in cul- FIG. 5. The expression of EGSs in cultured cells. Northern analyses were carried out using nuclear RNA fractions isolated from parental 143tk Ϫ cells (P, lanes 6 and 12) and cell lines that expressed TK-Ser (lanes 1 and 2 and lanes 7 and 8), C-TK-Ser (lanes 3 and 9), TK17 (lanes 4 and 10), and C-TK17 ribozymes (lanes 5 and 11). 30-g (lanes 2-6 and 8 -12) and 60-g RNA samples (2x, lanes 1 and 7) were either separated on 0.8% (B) or 2.5% (A) agarose gels that contained formaldehyde, transferred to a nitrocellulose membrane, and hybridized to a 32 Pradiolabeled probe that contained the DNA sequence coding for TK-Ser (A) or actin DNA sequence (B). The actin mRNA expression (B) was used as the internal control.
FIG. 6. Expression of TK mRNA, as detected by an RNase protection assay. RNase protection assays were performed as described previously (16). At 8 h after infection, total RNA was isolated either from parental human 143tk Ϫ cells (P; lanes 1, 2, and 6) 2-6). 40-g (lanes 1-5) and 80-g RNA samples (2x, lane 6) were used in the analyses. The protected products corresponding to TK mRNA (TK mRNA) and the overlapping transcripts of ␣47, Us10, and Us11 mRNA (␣47 mRNA) were approximately 90 and 180 nucleotides long, respectively. RNA probes were used in great excess of the detected RNA species. tured cells than TK-Ser, the EGS derived from tRNA Ser , which reduced TK expression by ϳ50%. These results strongly support our hypothesis that increasing the activity of EGS in directing RNase P cleavage may lead to improved efficacy in inhibiting gene expression in cultured cells. The difference between the in vivo efficacies of TK17 and TK-Ser (e.g. 95% versus 50%) appeared to be more limited than that of the in vitro cleavage efficiencies (more than 35-fold difference). One of the possible explanations is that ϳ5-10% of the target mRNA may not be accessible to EGS binding or RNase P cleavage, possibly because of its rapid transport to the cytoplasm.
Several lines of evidences suggest that the increased targeting activity of EGS TK17 is possibly caused by the enhanced stability of the mRNA-EGS complexes. First, TK17 binds to substrate tk46 at least 50-fold better than TK-Ser (Table II). Second, nuclease mapping studies suggest that the 3Ј region of tk46 downstream from the targeting region as well as the variable region of TK17 are probably involved in additional tertiary interactions that are not found in the complexes between tk46 and TK-Ser. Previous studies on tRNA molecules indicated that tertiary interactions between variable region and D loop are important for maintaining the tRNA conformation and RNase P cleavage (7,8). Given the fact that, in the tk46-EGS complex, the 3Ј region of the tk46 can be considered equivalent to the D loop in a tRNA (Fig. 4C), it is conceivable that these additional interactions stabilize the tk46-EGS complex and result in an enhanced binding affinity and increased targeting activity of the EGS.
In vitro selection (23, 25, 40 -42) has been widely used to generate highly active ribozymes and functional RNA molecules that have increased activity (43)(44)(45)(46)(47)(48). For example, this procedure has been used to generate novel RNA molecules that can serve as the substrates for RNase P and its catalytic RNA subunits (6,31,49). In vitro selection was also used to generate EGS molecules that direct human RNase P to cleave the mRNA encoding chloramphenicol acetyltransferase (6). However, whether these selected EGSs exhibit higher efficacies in targeting chloramphenicol acetyltransferase mRNA in tissue culture has not yet been extensively studied. In this study, we provide direct evidence that EGSs selected in vitro are highly effective in directing human RNase P to cleave a target mRNA in cultured cells. Moreover, our results suggest that improvement of the in vitro targeting efficiencies of EGSs should lead to increased efficacies of the EGS approach in tissue culture. Thus, our study provides a direction for the engineering and generation of highly active and effective EGS molecules by carrying out selection procedures and manipulation of the EGS domain to interact with the mRNA substrates.
The increased targeting activity of TK17 appears to be independent of the primary nucleotide sequence of the targeted mRNA. When an EGS was derived from TK17 to target the IE1 mRNA sequence of human cytomegalovirus that is different from the TK sequence, the constructed EGS also exhibited better binding affinity as well as higher activity in directing RNase P-mediated cleavage than that derived from TK-Ser (data not shown). These results suggest that the domain of the selected TK17 sequence enhances binding of the EGS to the mRNA substrate possibly by interacting with the structural features (e.g. the 2Ј hydroxyl groups) of the mRNA sequence rather than the bases of the nucleotides. More importantly, these observations suggest that the domains of the selected EGS molecules can be generally used to construct highly active EGSs to target any mRNA sequence. Further studies of these selected EGSs and their interactions with the mRNA substrates should provide insight into the mechanism of how RNase P cleaves the mRNA substrate in the presence of the EGSs and develop guidelines for constructing effective genetargeting EGSs.  1 and 2, 6, 7 and 8, and 12) or from cell lines expressing EGS TK17 (TK17, lanes 5 and 11), TK-Ser (TK-Ser, lanes 4 and 10), and C-TK17 (C-TK17, lanes 3 and 9). The cells were either mock-infected (lanes 1 and 7) or infected with HSV-1 (m.o.i. ϭ 0.5) (lanes 2-6 and 8 -12). 20-g (lanes 1-5 and 7-11) and 40-g protein samples (2x, lanes 6 and 12) were used in the analyses. Protein samples were separated in two identical SDS-polyacrylamide gels and transferred electrically to two identical membranes. One membrane was allowed to react with a monoclonal antibody (Anti-ICP27) against HSV-1 immediate-early protein ICP27 (A), whereas the other was stained with the polyclonal antibody (Anti-TK) against HSV-1(F) TK protein (B). Both antibodies were used in great excess of the detected antigens.