The T loop structure is dispensable for substrate recognition by tRNase ZL.

tRNA 3'-processing endoribonucleases (tRNase Z, or 3'-tRNase; EC 3.1.26.11) are enzymes that remove 3'-trailers from pre-tRNAs. An about 12-base-pair stem, a T loop-like structure, and a 3'-trailer were considered to be the minimum requirements for recognition by the long form (tRNase ZL) of tRNase Z; tRNase ZL can recognize and cleave a micro-pre-tRNA or a hooker/target RNA complex that resembles a micro-pre-tRNA. We examined four hook RNAs containing systematically weakened T stems for directing target RNA cleavage by tRNase ZL. As expected, the cleavage efficiency decreased with the decrease in T stem stability, and to our surprise, even the hook RNA that forms no T stem-loop-directed slight cleavage of the target RNA, suggesting that the T stem-loop structure is important but dispensable for substrate recognition by tRNase ZL. To analyze the effect of the T loop on substrate recognition, we compared the cleavage reaction for a micro-pre-tRNA with that for a 12-base-pair double-stranded RNA, which is the same as the micro-pre-tRNA except for the lack of the T loop structure. The observed rate constant value for the double-stranded RNA was comparable with that for the micro-pre-tRNA, whereas the K(d) value for the complex with the double-stranded RNA was much higher than that for the complex with the micro-pre-tRNA. These results suggest that the T loop structure is not indispensable for the recognition, although the interaction between the T loop and the enzyme exists. Cleavage assays for such double-stranded RNA substrates of various lengths suggested that tRNase ZL can recognize and cleave double-stranded RNA substrates that are longer than 5 base pairs and shorter than 20 base pairs. We also showed that double-stranded RNA is not a substrate for the short form of tRNase Z.

Bacterial and archaeal genomes contain a gene for the short form of tRNase Z only, whereas eukaryote genomes encode either only the long form or both the short and long forms (15). The C-terminal half region of tRNase ZL has high similarity to the whole region of tRNase ZS (1). These regions contain a well conserved histidine motif, which has been shown to be essential for the tRNase Z activity in the T. maritima enzyme (5).
The facts that tRNase ZL is ϳ2-fold larger than tRNase ZS and that it has the extra N-terminal region that is dispensable for the RNA cleaving catalytic reaction suggest that the long form might have additional functions. Indeed, tRNase ZL can function as a four-base-recognizing RNA cutter (RNase 65) through a relatively stable complex between the enzyme and a 3Ј-truncated tRNA (Fig. 1B), although little is known about the physiological role and substrate of RNase 65 (16,17).
The long forms of tRNase Z show further versatility in substrate recognition (18 -21). tRNase ZL can recognize and cleave a micro-pre-tRNA, a heptamer/target or hooker/target RNA complex that resembles a micro-pre-tRNA, as well as a pre-tRNA-like complex of 5Ј-half-tRNA and 3Ј-half-tRNA ( Fig. 1, C-F). This property of tRNase ZL suggested that generally, the enzyme can cleave any RNA at any site with the aid of appropriate small guide RNA (sgRNA), and indeed, we have shown this by examining various RNA complexes in vitro. Furthermore, we have demonstrated the efficacy of this specific RNA targeting method in the living cells by introducing sgRNAs either as their expression plasmids or as 2Ј-O-methyl RNAs (22).
In vitro tRNase ZL cleavage assays for various RNA substrates suggested that micro-pre-tRNAs or micro-pre-tRNAlike complexes are the smallest substrates ( Fig. 1, D-F); that is, an about 12-base-pair stem, a T loop-like structure, and a 3Ј-trailer are the minimum requirements for tRNase ZL recognition (19 -23). Although we have shown that several doublestranded RNA substrates are recognized and cleaved by tRNase ZL, this has been attributed to the formation of alternative conformers that look like pre-tRNAs retaining the T loop-like structures (20). Here we further investigated whether or not the T loop structure is really needed for substrate rec-* This work was supported in part by the Science Research Promotion Fund and the Academic Frontier Research Project Grant from the Promotion and Mutual Aid Corporation for Private Schools of Japan. 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. ognition by tRNase ZL and elucidated that the T loop structure is not indispensable. We also showed that tRNase ZL can recognize and cleave double-stranded RNA substrates that are longer than 5 base pairs and shorter than 20 base pairs. In this report, we refer to double-stranded RNA containing a 3Ј-trailer as double-stranded RNA.

EXPERIMENTAL PROCEDURES
Preparation of Various tRNase Z Proteins-Pig tRNase ZL was intensively fractionated from liver through several column chromatography procedures and further purified by glycerol gradient ultracentrifugation as described before (3,24). The recombinant human tRNase ZL and tRNase ZS were overexpressed from the pTYB11 expression plasmids (New England Biolabs) in Escherichia coli and purified with chitin beads as described previously (3). The recombinant tRNase ZS from T. maritima were produced from the pQE expression plasmids (Qiagen) in E. coli and purified with nickel-agarose as described (5).
In Vitro RNA Cleavage Assays-The in vitro RNA cleavage assays for the fluorescein-labeled MPTR1 were performed with tRNase ZL in a mixture (6 l) containing 10 mM Tris-HCl (pH 7.5), 1.5 mM dithiothreitol, and 3.3 mM MgCl 2 (15). The in vitro cleavage assays for the fluorescein-labeled target RNAs were carried out in the presence of the unlabeled sgRNAs using various tRNase Zs under the same conditions as above (15). After resolution of the reaction products on a 10 or 20% FIG. 1. Substrates of mammalian tRNase ZL contain the T loop structure. A, a secondary structure of human pre-tRNA Arg . B, a secondary structure of the complex of a 3Ј-truncated human pre-tRNA Arg (red) with a target RNA. C, a secondary structure of the complex of a 5Ј-half tRNA Arg (red) with a target 3Ј-half tRNA. D, a secondary structure of human micro-pre-tRNA Arg . E, a secondary structure of the complex of an RNA heptamer (red) with a target RNA. F, a secondary structure of the complex of a hook RNA (red) with a target RNA. The T loop structures are shaded. Arrows denote the cleavage sites by tRNase ZL.
polyacrylamide-8 M urea gel, the gel was analyzed with a Typhoon 9210 (Amersham Biosciences).
Likewise, the dissociation constants of the tRNase ZL/substrate complexes were measured by the gel-shift assays (24). The fluoresceinlabeled micro-pre-tRNA MPTR1 or double-stranded RNA T6L1/T-6 (1.6 pmol) was incubated with various amounts of pig tRNase ZL in the above buffer at 4°C for 10 min and analyzed in the same way as above.

The T Stem/Loop Structure Is Important but Dispensable for Substrate Recognition by tRNase ZL-We have shown
how important the T stem-loop structure is for substrate recognition by tRNase ZL by analyzing heptamer-directed cleavages of the 3Ј-half-tRNAs Arg containing systematically weakened T stems (19). In this analysis, we have observed that the cleavage efficiencies decrease with the decrease in the T stem stability, and we have not been able to detect any cleavage of the 3Ј-half-tRNA Arg with no T stem. To corroborate these observations, we examined four hook RNAs, T1H5, T1H4, T1H3, T1H0, containing systematically weakened T stems for directing target RNA cleavage by tRNase ZL (Fig.  2A). This series of the T stems was the same as that used in the heptamer-directed RNA cleavage assays.
As expected, the target RNA T-1 was most efficiently cleaved by tRNase ZL in the presence of T1H5, and the cleavage efficiency decreased with the decrease in the T stem stability (Fig.  2B). The cleavage took place after a nucleotide corresponding to the discriminator. To our surprise, even T1H0, which had no T stem, directed slight cleavage of the target RNA. Furthermore, tRNase ZL cut the target slightly in the presence of the RNA heptamer T1Hep corresponding to the 5Ј-portion of the acceptor stem. These results suggest that the T stem-loop structure is important but dispensable for substrate recognition by tRNase ZL. The discrepancy between this and previous results may be because of the difference in the assay conditions; this analysis was performed in the presence of 3.3 mM MgCl 2 instead of 3.3 mM spermidine. We found that generally, spermidine is much less effective in supporting tRNase ZL cleavage reactions than Mg 2ϩ (Supplemental Fig. 1).
The T Loop Is Dispensable-We were curious about which portion of the T stem/loop is important for the tRNase ZL activity. A previous study has shown that several doublestranded RNA substrates are recognizable by tRNase ZL (20). In this case, however, formation of alternative conformers that look like pre-tRNAs retaining the T loop-like structures appears to be required. On the other hand, it has been shown that micro-pre-tRNAs lacking distal base pairings of the T stem are not substrates (23). These observations suggest that the T stem may be more important for substrate recognition by tRNase ZL than the T loop.
To address this issue more precisely, we compared the cleavage reaction for a micro-pre-tRNA with that for a 12-base-pair double-stranded RNA, which is the same as the micro-pre-tRNA except for the lack of the T loop structure (Fig. 3A). The micro-pre-tRNA MPTR1 was synthesized with T7 RNA polymerase and 5Ј-end-labeled with fluorescein. Both strands of the double-stranded RNA were separately transcribed by T7 RNA polymerase, and only the target RNA strand T-6 was 5Ј-endlabeled with fluorescein and annealed with the unlabeled sgRNA strand T6L1. Time course assays showed that there is no significant difference in cleavage efficiency between both substrates (Fig. 3B). Although pig tRNase ZL was used to compare these data with previous data, which have shown that the cleavage efficiency of the micro-pre-tRNA is comparable with that of a natural full-length pre-tRNA (23), similar results were also obtained using recombinant human tRNase ZL (data not shown). As expected, each cleavage occurred primarily after a nucleotide corresponding to the discriminator (Fig. 3B). The reason why a simple double-stranded RNA was not cleaved by tRNase ZL in the previous study (20) would be because spermidine was used instead of Mg 2ϩ as mentioned above.
We tried to determine the kinetic parameters K m and k cat for the reaction of the double-stranded RNA, but reliable values were not obtained due to large standard deviations. Thus, we measured the observed rate constant k obs for each reaction under single turnover conditions instead of K m and k cat . First, we determined the dissociation constant of the double-stranded RNA by gel-shift analysis. Using this estimated K d value (50 nM), we calculated an effective concentration of the double-stranded RNA substrate. The k obs value (0.013 min Ϫ1 ) for the double-stranded RNA T6L1/T-6 was comparable with the value (0.023 min Ϫ1 ) for the micro-pre-tRNA MPTR1 (Table I). Taken together, present and previous results suggest that the T loop structure is not indispensable for substrate recognition by tRNase ZL.
In addition, we measured dissociation constants of the tRNase ZL/substrate complexes by gel-shift assays. The K d value (420 nM) for the complex with the double-stranded RNA was 130-fold higher than that (3.2 nM) for the complex with the micro-pre-tRNA (Table I), indicating that the interaction between the T loop and the enzyme exists and helps make the enzyme/substrate complex.
tRNase ZL Can Recognize and Cleave Double-stranded RNA Substrates That Are Longer than 5 Base Pairs and Shorter than 20 Base Pairs-To determine the minimum and maximum lengths of double-stranded RNA substrates that are recognizable and cleavable by tRNase ZL, we examined doublestranded RNA substrates of various lengths for tRNase ZL cleavage. The first series of the substrates was composed of the target T-2 and one of the eight sgRNAs T2L5-T2L12, which are complementary to T-2 (Fig. 4A). These substrates were assayed, and cleavage efficiencies were quantitated and plotted against the length of double-stranded RNA (Fig. 4, B and C). tRNase ZL cleaved the target T-2 with the aid of each sgRNA with the exception of T2L5, and the double-stranded RNA complexes with the longer sgRNAs were better substrates. In each case, the primary cleavage site was located after a nucleotide corresponding to the discriminator (Fig. 4B).

FIG. 3. tRNase ZL cleavage reactions for substrates with and without the T loop.
A, secondary structures of the micro-pre-tRNA MPTR1 and the sgRNA/target RNA complex T6L1/T-6. Arrows indicate cleavage sites by tRNase ZL. B, time courses for the cleavage reactions. The fluorescein-labeled MPTR1 (1.6 pmol) was incubated with pig tRNase ZL (2 pmol) at 52°C for 2 or 10 min, and the fluorescein-labeled T-6 (1.8 pmol) was reacted with pig tRNase ZL (2 pmol) in the absence or presence of T6L1 (4.7 pmol) at 52°C for the indicated time period. The cleavage products were separated on a 10% polyacrylamide-8 M urea gel. The substrate and the primary 5Ј-cleavage product (5Ј P) are indicated by a bar and an arrow, respectively. L denotes an alkaline ladder of the fluorescein-labeled substrate. Although a part of the ladder may not be viewed clearly, each band was visible from the original picture. The substrate band of the precise length among the 3Ј-heterogeneous transcript bands from the intrinsic property of T7 RNA polymerase was identified by comparison with a size standard.
Examination of another series of substrates composed of 10-, 11-, 12-, 13-, 15-, and 16-base-pair RNAs showed that the 12-base-pair RNA is the best substrate (Supplemental Fig. 2). Taken together, these results show that tRNase ZL can recognize and cleave double-stranded RNA substrates that are longer than 5 base pairs and shorter than 20 base pairs and that the enzyme is the most active for 12-16-base-pair RNAs.
Simple Complementary RNA Molecules Can Be Used for Targeted RNA Cleavage by tRNase ZL-Because tRNase ZL can recognize and cleave long RNA targets of more than 300 nt in the presence of appropriate sgRNAs that form pre-tRNA-like complexes with the targets (20), we investigated whether a relatively long target can be cleaved with the aid of simple complementary RNA. A 100-nt RNA target, T-5, was assayed in the presence of the 11-nt sgRNAs T5L1, T5L2, and T5L3 for cleavage by tRNase ZL (Fig. 7A). The target was cleaved at the expected site in the presence of T5L1 or T5L3 (Fig. 7B). T5L2 guided cleavages at one expected site and at another unexpected site. The additional cleavage is most likely due to the presence of a partially complementary sequence to T5L2 in T-5. Indeed, the upstream sequence of the cleavage site was 82% complementary (Fig. 7A). These results suggest that simple complementary RNA can be used for targeted RNA cleavage by tRNase ZL, albeit with relatively low specificity.
Double-stranded RNA Is Not a Substrate for the Short Form of tRNase Z-We examined whether the short form of tRNase Z can also recognize and cleave double-stranded RNA. The complex of the target T-2 with the sgRNA T2L12 (Fig. 8A) was tested for cleavage by tRNase ZSs from human and T. maritima. This 12-base-pair RNA was not cleaved at all by either tRNase ZS, whereas pig and human tRNase ZLs processed the substrate (Fig. 8B). The specific cleavages of the 8-base-pair RNA T2L8/T-2 and the 16-base-pair RNA T3L16/T-3 by T. maritima tRNase ZS were also not detected (Supplemental Fig. 3). These results suggest that double-stranded RNAs are not substrates for tRNase ZSs and that the N-terminal half of tRNase ZL may be required for recognition of double-stranded RNA. Consistently, tRNase ZSs cannot cleave the micro-pre-tRNA containing the T loop (15). DISCUSSION Interaction of tRNase ZL with a Double-stranded RNA Substrate-Previously, we have shown that micro-pre-tRNAs or micro-pre-tRNA-like complexes, which consist of the T stemloop and the acceptor stem plus a 3Ј-trailer, are the smallest substrates for tRNase ZL (19 -23). In this report, however, we found out that the T loop structure is important but not indispensable. Therefore, the genuine minimum substrate of tRNase ZL is considered to be composed of the T stem and the acceptor stem plus a 3Ј-trailer.
The importance of the T loop structure for recognition is indicated by the much lower K d value for the substrate with the T loop than for the substrate without the T loop. From the difference between these K d values (Table I), we estimated the free energy ⌬G for the enzyme/T loop interaction to be Ϫ2.7 kcal/mol. We guess that a couple of weak chemical bonds would be involved in this interaction.
The double-stranded RNA substrates cleavable by tRNase ZL ranged from 6 to 18 base pairs, and the enzyme was the most active for 12-16-base-pair RNAs. The double-stranded RNA substrates less than 6 base pairs would be too small to stably bind to the enzyme, and those longer than 18 base pairs would be too long to fit the T stem end into the enzyme pocket. be able to mimic micro-pre-tRNA most well especially at the end of the T stem. The reason why the 11-base-pair RNA substrate containing the 100-nt target strand was cleaved by tRNase ZL may be because the single strand RNA portion is floppy enough to avoid steric hindrance in the enzyme pocket, whereas the double-stranded RNA portion is too rigid to dodge the hindrance.
Cleavage Site Selection in Double-stranded RNA Substrates-There exists a general rule for cleavage site selection by tRNase ZL with respect to full-length pre-tRNAs (11). Cleavage of pre-tRNAs containing a total of N base pairs (N Ͻ 12) in the acceptor stem and the T stem occurs after 12 Ϫ N and 13 Ϫ N nucleotides 3Ј to the discriminator nucleotide, whereas cleavage of pre-tRNAs with a total of N base pairs (N Ն 12) occurs after the discriminator. For example, the pre-tRNA R-ASD2 containing a total of 10 base pairs was cleaved after 2 and 3 nucleotides 3Ј to the discriminator, and cleavage of the pre-tRNA R-ASA2 containing a total of 14 base pairs occurred after the discriminator (11). This rule needs to be modified with respect to micro-pre-tRNAs containing a total of N base pairs (N Ͻ 12) (23). Cleavage occurs at multiple sites fluctuating between the discriminator nucleotide and 12 Ϫ N or 13 Ϫ N nucleotides 3Ј to the discriminator.
We have proposed a model to explain this difference, taking into account the presence of "bottom-half" tRNA (23). Pre-tRNAs would be prevented from moving along the "top-half" tRNA binding domain on tRNase ZL due to the bottom-half tRNA constraints, whereas micro-pre-tRNAs would be able to slide along this domain. The cleavage site would change depending on the fashion of enzyme/substrate interaction. When the distal ac-ceptor stem is located in the vicinity of the catalytic domain, the cleavage would occur after the discriminator nucleotide. On the other hand, when the T loop is arranged at its original binding domain, the cleavage would occur after 12 Ϫ N and 13 Ϫ N nucleotides 3Ј to the discriminator.
The present study showed that cleavages of the doublestranded RNA substrates occur primarily after the discriminator. This observation can be interpreted in the context of the above model; that is, the distal acceptor stem of doublestranded RNA would be preferentially arranged near the catalytic domain due to the lack of the T loop regardless of the N number.
Cleavages of the RNA complexes T1H4/T-1 and T1H3/T-1, which contain the T loop but lack one or two proximal T stem base pairings, occur after the discriminator (Fig. 2). This is consistent with the previous observation that micro-pre-tRNAs containing proximal base pair disruptions behave like a substrate with 12 intact base pairs and that tRNase ZL can select the cleavage site in a fashion similar to that of their original micro-pre-tRNA (23).
An Extra Physiological Function of tRNase ZL?-The human genome contains both tRNase ZS and tRNase ZL genes, raising an interesting question of whether the short and long forms play different roles in the cells. One may be for nuclear tRNA processing, and the other may be for mitochondrial processing. Alternatively, tRNase ZL, which has the extra N-terminal region that is not essential for the RNA cleaving catalytic reaction, might have an additional role for some RNA metabolism.
To elucidate their possible differential roles, we have recently investigated whether the long and short forms have different preferences for RNA substrates and have found that indeed tRNase ZL recognizes and cleaves a wider range of substrates than tRNase ZS and that this versatility is attributed to the N-terminal half domain of tRNase ZL (15). It has been shown that in HeLa cells, tRNase ZL physically interacts with the ␥-tubulin complex (25), suggesting that tRNase ZL exists in cytoplasm as well as nuclei where the tRNA 3Ј-processing occurs. This ubiquity of tRNase ZL would extend its niches in cellular RNA metabolism. In this report, we demonstrated that tRNase ZL shows further versatility in substrate recognition; the enzyme can recognize and cleave doublestranded RNA substrates. Taken together, these observations support the notion that tRNase ZL may play extra roles in other cellular RNA metabolism.
Although at present we do not have enough information to specify which other RNA metabolism tRNase ZL is involved in, tRNase ZL may work in the four different fashions as follows: (i) RNase 65 with the aid of 3Ј-truncated tRNAs (16,17), (ii) RNA cutters that recognize pre-tRNA-like complexes of targets with microRNAs (20), (iii) 4-nt RNA cutters in the presence of hook RNAs (21), and (iv) double-stranded RNA cutters with complementary sgRNAs. A huge number of microRNAs, the length of which is ϳ21-23 nt, are found in mammalian cells (26), and other small non-coding RNAs might also exist. These small RNA molecules could work as hook RNAs or complementary sgRNAs.
Although human tRNase ZL is the product of the candidate prostate cancer gene ELAC2 (1, 3), we have no information on how the mutations in the tRNase ZL gene can cause human prostate cancer. The tRNase ZL variants containing amino acid substitutions from such mutations show only trivial differences in both tRNA 3Ј-processing activity and RNase 65 activity (27). These trivial changes might be sufficient to alter the whole gene expression pattern in prostate cells to the cancerous state. Alternatively, activities of the other RNA cutters from tRNase ZL might be affected significantly by these mutations.
Targeted Cleavage of Undesirable RNA by tRNase ZL with the Aid of 11-nt Complementary sgRNA-Mammalian tRNase ZL can cleave any RNA at any site under the direction of sgRNA in vitro (18 -21). sgRNAs can be as short as heptamers, which are much smaller than small interfering RNAs of ϳ21 nt (19). Together with such flexibility in substrate recognition, the ubiquity and the constitutive expression of tRNase ZL have suggested that this enzyme can be utilized for specific cleavage of cellular RNAs by introducing appropriate sgRNAs into living cells. Indeed, the expression of exogenous and endogenous genes has been down-regulated by appropriate sgRNAs that were introduced into mammalian culture cells as expression plasmids or synthetic 2Ј-O-methyl RNAs (22). The discovery that the 11-nt complementary sgRNA can direct tRNase ZL cleavage of the long RNA target may lead us to examine the efficacy of simple complementary sgRNA on down-regulation of cellular gene expressions. Although this type of sgRNA would also work in the cells, the present data suggest that its guiding efficiency and substrate specificity would be lower than those of sgRNA that can form complexes containing a T loop-like structure with target RNAs. Thus, we would prefer the latter type of sgRNA for destroying undesirable cellular RNAs. FIG. 8. Tests of the short forms of tRNase Z for double-stranded RNA cleavage activity. A, a structure of the double-stranded RNA T2L12/T-2. An arrow indicates a cleavage site by tRNase ZL. B, the fluorescein-labeled T-2 (1 pmol) was incubated with pig or human tRNase ZL (2 pmol) or with T. maritima or human tRNase ZS (2 pmol) in the absence or presence of 4.8 pmol of T2L12 at 52°C for 4 min. The cleavage products were analyzed on a 20% polyacrylamide-8 M urea gel. The target and the primary 5Ј-cleavage product (5Ј P) are indicated by a bar and an arrow, respectively. L denotes the alkaline ladder of the fluorescein-labeled T-2. The substrate band of the precise length among the 3Ј-heterogeneous transcript bands from the intrinsic property of T7 RNA polymerase was identified by comparison with a size standard.