A Bipartite Model of 2-5A-dependent RNase L*

(Received for publication, March 6, 1997, and in revised form, June 4, 1997)

Beihua Dong and Robert H. Silverman Dagger

From the Department of Cancer Biology, The Lerner Research Institute, The Cleveland Clinic Foundation, Cleveland, Ohio 44195

ABSTRACT
INTRODUCTION
EXPERIMENTAL PROCEDURES
RESULTS
DISCUSSION
FOOTNOTES
ACKNOWLEDGEMENTS
REFERENCES


ABSTRACT

The 2-5A-dependent RNase (RNase L) is a tightly regulated endoribonuclease of higher vertebrates that is catalytically active only after engaging unusual effector molecules consisting of the 2',5'-linked oligoadenylates, p1-3A(2'p5'A)>=2 (2-5A). Progressive truncations from either terminus have provided insight into the structure, function, and regulation of RNase L. We determined that deletion of the N-terminal 335 amino acids of RNase L, about 45% of the enzyme, produced a constitutively active endoribonuclease, thus effectively eliminating the requirement for 2-5A. The truncated nuclease had 6-fold lower catalytic activity against an oligo(rU) substrate than wild type RNase L. However, the two enzymes showed identical RNA cleavage site preferences with an mRNA as substrate. The repressor function required only the last three of a series of nine ankyrin-like repeats present in the N-terminal part of RNase L. In contrast, the entire ankyrin repeat region was necessary and sufficient for 2-5A binding activity. Deletion of a 10-amino acid sequence near the C terminus of RNase L, between residues 710 and 720, eliminated both the catalytic and RNA substrate binding functions of the enzyme. The ability to bind native RNase L in response to 2-5A required amino acid sequences near both termini of the protein. A bipartite model for the structure of RNase L emerged in which the regulatory functions of the molecule are located in the N-terminal half, while the catalytic domain is present in the C-terminal half.


INTRODUCTION

RNase L provides a unique paradigm of enzyme regulation because it is the only enzyme known to require the unusual 2',5'-oligoadenylate(s), p1-3A(2'p5'A)>=2 (2-5A),1 for its catalytic activity (1, 2). The activators of RNase L, 2-5A, are produced from ATP by double-stranded RNA stimulation of the 2-5A-synthetases, a family of enzymes induced by interferon treatment of mammalian cells. Production of double-stranded RNA during some virus infections leads to 2-5A synthesis and activation of RNase L (3, 4). The proposed biological functions of RNase L includes involvement in the antiviral and antiproliferative mechanisms of interferon action and in the more general control of RNA decay (5, 6). The human RNase L is a 741-amino acid protein that forms homodimers upon binding 2-5A (7, 8). Activation of human RNase L requires 2-5A molecules containing at least three 2',5'-adenylyl residues, such as pA(2'p5'A)2, and the binding of 2-5A to RNase L, KD = 40 pM, is highly specific (9, 10). Previously, we demonstrated that the 2-5A binding activity of RNase L is located in the N-terminal half of RNase L, a region containing nine ankyrin-like, macromolecular recognition domains (2, 6). Within the seventh and eighth ankyrin repeats are two P-loop motifs. In murine RNase L, substitutions of both of the conserved P-loop lysine residues with asparagines resulted in a defect in 2-5A binding activity, while deletion of 89 C-terminal amino acids caused a loss of ribonuclease activity (2, 6). The truncated RNase L mutant, RNase LZB1, which binds 2-5A but lacks ribonuclease activity, functions as a dominant negative inhibitor of wild type RNase L (6).

To further study the structure and function of RNase L, we have performed both a systematic shortening of RNase L from either terminus and a biochemical analysis of each mutant protein. The results provide information about the location of the structural and functional domains in RNase L. Moreover, these studies provide insight into the mechanism of regulation of RNase L. An internal repressor of ribonuclease function was identified, deletion of which released the ribonuclease domain from its dependence on 2-5A.


EXPERIMENTAL PROCEDURES

Construction of Truncated Forms of Human RNase L

A full-length coding sequence DNA for human RNase L in plasmid pZC5 (2) was subcloned downstream (3') of the coding sequence for glutathione S-transferase (GST) in expression vector pGEX-4T-3 (Pharmacia Biotech Inc.). The cloning strategy involved creating a BamHI site immediately upstream of the RNase L coding sequence. This was done using an oligonucleotide primer for PCR containing the BamHI site. The PCR product containing the BamHI site at the 5' terminus and the natural NcoI site at the 3' terminus was subcloned into BamHI/NcoI digested pZC5. The complete RNase L coding sequence was subcloned as a BamHI/XhoI fragment into BamHI/XhoI-digested pGEX-4T-3. The deletion mutants of RNase L were constructed with PCR by creating new BamHI restriction sites for the 5'-truncations and new XhoI sites for the 3'-truncations. To generate the N-terminal deletions of RNase L, the 5'-primers contained 4-6 nucleotides and GGATCC for creating BamHI sites followed by 18-20 nucleotides of the RNase L sequences. To generate the C-terminal truncations, 5'-GGCGGCCTCGAGTCA (underlined sequence is the XhoI site followed by a sequence complementary to a stop codon sequence), attached to the 16-24 nucleotides of different RNase L sequences was used as the 3'-primers for PCR. Specific RNase L mutants were constructed with the primers and restriction enzymes indicated in Table I. All mutants were confirmed by DNA sequence analysis.

Table I. PCR primers and restriction sites for construction of RNase L mutants


Mutants 5'-Primersa 3'-Primersb Restriction sites

RNase LNDelta 23 5'-CGCGGGATCCGAAGACAATCACTTGCTG-3' 5'-CCATCACTGTCTGTGTCA-3' BamHI, NcoI
RNase LNDelta 56 5'-CGCGGGATCCGAAGGGGGCTGGACACCT-3' 5'-CCATCACTGTCTGTGTCA-3' BamHI, NcoI
RNase LNDelta 90 5'-CGCGGGATCCAATGGGGCCACGCCTTTT-3' 5'-CCATCACTGTCTGTGTCA-3' BamHI, NcoI
RNase LNDelta 123 5'-GAGTGTGGATCCTATGGCTTCACAGCCTTC-3' 5'-GAGTTTGCCAATCATAGGG-3' BamHI, NcoI
RNase LN237 5'-GTGAGGGGATCCAGAGGGAAGACTCCCCTGAT-3' 5'-GAGTTTGCCAATCATAGGG-3' BamHI, SacI
RNase LNDelta 335 5'-CCTGCCGGATCCTGGAAGCCTCAGAGCTCA-3' 5'-CGATGAATGAGGTCCTTAGTTTCC-3' BamHI, DraIII
RNase LCDelta 21 5'-ATTACGCATCTGCTGCTG-3' 5'-GGCGGCCTCGAGTCAGTGGGTTTGGGGGAAATG-3' NcoI, XhoI
RNase LCDelta 31 5'-ATTACGCATCTGCTGCTG-3' 5'-GGCGGCCTCGAGTCATGTGTTCTGTAGTTTTGTGTAGAC-3' NcoI, XhoI
RNase LCDelta 41 5'-ATTACGCATCTGCTGCTG-3' 5'-GGCGGCCTCGAGTCACACCAGATCTGGAAATGTCTTCTG-3' NcoI, XhoI
RNase LCDelta 51 5'-GTGAGGGGATCCAGAGGGAAGACTCCCCTGAT-3' 5'-GGCGGCCTCGAGTCACAGGGAAGGGTCTCCA-3' BglI, XhoI
RNase LCDelta 80 5'-GTGAGGGGATCCAGAGGGAAGACTCCCCTGAT-3' 5'-GGCGGCCTCGAGTCAATCACCCACAGTGTTCTG-3' BglI, XhoI
RNase LCDelta 399 5'-ATTACGCATCTGCTGCTG-3' 5'-GGCGGCCTCGAGTCAGTGTGAGCTCTGAGGCTT-3' NcoI, XhoI
RNase LCDelta 406 5'-ATTACGCATCTGCTGCTG-3' 5'-GGCGGCCTCGAGTCAGTCTTCAGCAGGAGGGTG-3' NcoI, XhoI
RNase LNDelta 23/CDelta 504 5'-CGCGGGATCCGAAGACAATCACTTGCTG-3' 5'-GGCGGCCTCGAGTCATTCTCCCCTCACATTGAC-3' BamHI, XhoI

a Underlined sequences are BamHI sites immediately upstream of RNase L coding sequences which create different N-terminal truncations.
b Underlined sequences are XhoI sites followed by a sequence complementary to a stop codon sequence, indicated in bold print, immediately adjacent to sequences complementary to RNase L coding sequences which create different C-terminal deletions.

Expression and Purification of GST-RNase L and GST-RNase L Mutants

The cDNAs for the wild type and mutant forms of RNase L in plasmid pGEX4-T-3 were transformed into Escherichia coli strain DH5alpha . The transformed bacteria were grown at 30 °C to A595 = 0.5 before being induced with 0.1 mM isopropyl-1-thio-beta -D-galactpyranoside for 3 h. Harvested cell pellets were washed with PBS, resuspended in 3-5 volumes of PBS-C (PBS with 10% glycerol, 1 mM EDTA, 0.1 mM ATP, 5 mM MgCl2, 14 mM 2-mercaptoethanol, 1 µg/ml leupeptin, and 1 mM PMSF) supplemented with 1 µg/ml lysozyme, and incubated at room temperature for 20 min. The suspended cells were sonicated on ice for 20 s four times, Triton X-100 was added to a final concentration of 1% (v/v), and the cell lysates incubated at room temperature for 20 min. The supernatants were collected following centrifugation at 16,700 × g for 20 min at 4 °C. Purification of fusion proteins was performed as described by the manufacturer of the glutathione-Sepharose 4B (Pharmacia) with modifications. Briefly, glutathione-Sepharose 4B (200 µl of a 50% slurry in PBS-C) was added to extract from 200-ml cultures of bacteria at room temperature for 20 min with shaking. After washing the protein-bead complexes three times with 3 ml of PBS-C, the fusion proteins were eluted with 20 mM glutathione in 50 mM Tris-HCl, pH 8.0, containing 1 µg/ml leupeptin, with shaking at room temperature for 20 min. Expression and purity of the protein preparations was determined by SDS-PAGE and Coomassie Blue staining and by Western blots with monoclonal antibody to RNase L (7).

Assay of 2-5A Binding Activity

2-5A binding activity was determined by a filter binding method (11, 12). Purified fusion proteins (2 µg/assay) were incubated with about 10,000 cpm of a 32P-labeled and bromine substituted 2-5A analog, p(A2'p)2(br8A2'p)2A3'[32P]pCp (3,000 Ci/mmol) in buffer A (20 mM Tris-HCl, pH 7.5, 10 mM magnesium acetate, 8 mM 2-mercaptoethanol, 90 mM KCl, 0.1 mM ATP, 10 µg/ml leupeptin) on ice for 2 h. The reaction mixtures were transferred to nitrocellulose filters (Millipore; 0.45 µm Type HA). Filters were washed twice with distilled water and dried. The amount of the 32P-labeled 2-5A probe bound to protein on the filter was measured by scintillation counting.

RNase L/RNase L Interaction Assay

Extracts (50 µg) containing GST-RNase L or GST-RNase L mutants were incubated with extract of recombinant, native human RNase L (25 µg) (not a fusion protein) from insect cells (9) in the presence and absence of 0.8 µM pA(2'p5'A)3 in buffer A on ice for 2 h. Subsequently, 5 µl of 20% (v/v) glutathione-Sepharose 4B was added and the mixture was incubated with platform shaking at room temperature for 20 min with gentle vortexing every 5 min, followed by washing three times with 0.3 ml of PBS-C. Analysis of the bound protein was by SDS-PAGE and Western blot analysis probed with a monoclonal antibody to RNase L (7) using the enhanced chemiluminescence (ECL) method (Amersham Corp.).

Ribonuclease Assays with Oligo(U) as Substrate

A chemically synthesized oligouridylic acid, U25 (Midland Certified Reagent Co.) was labeled at its 3' terminus with [5'-32P]pCp (3,000 Ci/mmol) (NEN Life Science Products) with T4 RNA ligase (Life Technologies, Inc.). The U25-[32P]pCp (80-160 nM) was incubated with 400 ng of purified GST-RNase L or GST-mutant RNase L in the presence and absence of 100 nM pA(2'p5'A)3 in a final volume of 25 µl of buffer A at 30 °C for 30 min.

To compare the relative activities of RNase LNDelta 335 to RNase Lwild type, assays were performed with different amounts of the purified proteins. RNase Lwild type (600 nM) was preincubated in the presence or absence of 10 µM pA(2'p5'A)3 in buffer A at 0 °C for 30 min prior to dilution and addition of 80 nM U25-[32P]pCp. RNase LNDelta 335 was preincubated at 0 °C for 30 min without pA(2'p5'A)3 prior to dilution and addition of substrate. The cleavage reactions were in final volumes of 20 µl for 30 min at 30 °C.

Reaction mixtures were heated to 100 °C for 3 min in loading buffer, and RNA and RNA degradation products were separated in 20% polyacrylamide, 8% urea sequencing gels. The amount of intact U25-[32P]pCp remaining after the incubations was determined from autoradiograms of the gels with a Sierra Scientific high resolution CCD camera (Sunnyvale, CA) and the computer program NIH Image 1.6.

Determination of RNA Cleavage Site Preferences

Human PKR cDNA in plasmid pBS (Ref. 13, a gift of B. R. G. Williams, Cleveland Clinic Foundation) was linearized by AflIII digestion (523 nucleotides downstream from start codon) and transcribed in vitro with T7 RNA polymerase (MEGAscript, Ambion) according to the manufacturer's protocol. The PKR RNA fragment was purified by electrophoresis in 8% polyacrylamide, 8 M urea gels followed by elution. The PKR RNA fragment (200 ng) was incubated with purified RNase LNDelta 335 or RNase Lwild type in the presence or absence of 100 nM pA(2'p5'A)3 in 20 µl of buffer A at 30 °C for 10 min. Reactions were stopped by heating at 75 °C for 15 min and the RNA cleavage products precipitated in 0.3 M sodium acetate, 70% ethanol at -20 °C, followed by washing with 70% ethanol. The dried cleavage products were dissolved in 4 µl of diethyl pyrocarbonate-treated water.

A 18-nucleotide primer complementary to PKR nucleotides 263-280 of the coding sequence of PKR cDNA (5'-GGCCTATGTAATTCCCCA-3') was labeled with [gamma -32P]ATP (NEN Life Science Products) and T4 polynucleotide kinase (Life Technologies, Inc.), and purified through Sephadex G-25 cartridges (Boehringer Mannheim). The 32P-labeled primer was mixed with the RNA cleavage products at 70 °C for 10 min and then at 37 °C for 1 h. The reaction mixtures were incubated with dNTPs and reverse transcriptase (superscript RNase H-, Life Technologies, Inc.) at 40 °C for 90 min. The cleavage sites were determined by comparing the migration in 10% polyacrylamide, M urea gels of the primer extension products and the DNA sequencing products using the same primer on PKR cDNA with Sequenase 2.0 (U. S. Biochemical) and deoxyadenosine [5'-alpha -35S]thiotriphosphate (NEN Life Science Products).

RNA Binding Assay

Glutathione-Sepharose 4B, 10 µl of a 20% (v/v) suspension, or poly(U)-Sepharose 4B, 10 µl of a 60% (w/v) suspension (~0.5 mg of poly(U)/ml of gel; ~100 uridyl residues/chain) (Pharmacia) were added to extracts of the GST fusion proteins, 50 µg/assay, in PBS-C buffer. Reactions were incubated at room temperature for 20 min with shaking and gentle vortexing every 5 min. The protein-bead complexes were washed three times with 0.3 ml of PBS-C, and the proteins bound to glutathione-Sepharose 4B and poly(U)-Sepharose 4B were eluted in gel loading buffer containing SDS at 100 °C for 5 min and separated by electrophoresis on 7.5% polyacrylamide-SDS gels. The proteins were detected after transfer to nitrocellulose membranes (Schleicher & Schuell) by probing with murine polyclonal antibody to human RNase L (sera of immunized mice used for making monoclonal antibody) (7). The relative amounts of the bound proteins were determined from Western blots by enhanced chemiluminescence (Amersham) after capture of the images with a video camera.


RESULTS

Strategy for Constructing RNase L Mutants

To map and study the functional domains in RNase L, complete or truncated RNase L coding sequence DNAs were expressed as GST fusion proteins from vector pGEX-4T-3 (Pharmacia) in E. coli (Fig. 1). Previously, it was shown that GST fusion proteins of RNase L are fully active in the presence of 2-5A (14). Because ankyrin repeats are structural as well as functional units (15), the approach taken for the design of the N-terminal truncations was to progressively delete the individual ankyrin repeats. The exception was the first mutant RNase LNDelta 23, which lacks the 23 N-terminal amino acids proceeding the first ankyrin repeat. The other N-terminal truncations, mutants RNase LNDelta 56, RNase LNDelta 90, RNase LNDelta 123, RNase LNDelta 237, and RNase LNDelta 335, lack 1, 2, 3, 6, and 9 ankyrin repeats, respectively (Fig. 1). On the other hand, a functional approach was used in the design of the C-terminal truncations. In essence, progressive C-terminal mutations were made and the mutant proteins assayed for loss of various functions (Fig. 1). In addition, one mutant, RNase LNDelta 23/CDelta 504, was truncated from both termini. Assays were performed, which measured and compared the abilities of wild type and mutant forms of RNase L to bind with 2-5A, native RNase L, or RNA and to degrade RNA.


Fig. 1. Functional analysis of RNase L and deletion mutants of RNase L fused to GST. n.d., not determined; *, RNase LNDelta 335 had ribonuclease activity in the absence of added 2-5A; Ank, ankyrin repeats.
[View Larger Version of this Image (24K GIF file)]

Localization of the 2-5A Binding Domain

The portion of RNase L that contains the 2-5A binding function was localized by measuring the abilities of the mutant proteins to bind to a radioactive 2-5A analog in a filter binding assay (11, 12). The 2-5A binding activity of RNase LNDelta 23, lacking 23 N-terminal amino acids, was 82% of that of GST-RNase Lwild type (Fig. 2). However, deletion of amino acid sequences including the first ankyrin repeat resulted in a complete loss of 2-5A binding activity (mutant RNase LNDelta 56). Further truncations from the N terminus also produced mutant proteins that lacked 2-5A binding activity. In contrast, C-terminal truncations of RNase L, removing as much as 406 amino acid residues in RNase LCDelta 406, had little or no effect of 2-5A binding activity. However, RNase LNDelta 23/CDelta 504, lacking 504 C-terminal amino acids and 23 N-terminal residues, did not bind 2-5A. These findings show that the 2-5A binding function of RNase L requires the ankyrin repeat region, but not other portions of the enzyme.


Fig. 2. 2-5A binding activity of RNase L truncation mutants. The ability of GST-RNase L mutants to bind to the 32P-labeled 2-5A analog in filter binding assays was determined. Results, averages of triplicate assays, are expressed as a percentage of the activity obtained with GST-RNase Lwild type.
[View Larger Version of this Image (29K GIF file)]

Mapping the RNase L/RNase L Interaction Domains

To map the RNase L/RNase L interaction domains, the ability of the GST-RNase L mutants to bind native RNase L was determined in the presence and absence of 2-5A (Fig. 3). The 2-5A-dependent binding of the 83-kDa native RNase L (not a fusion protein) to GST-RNase Lwild type protein was observed after the complex was immobilized on glutathione-Sepharose (Fig. 3A, lane 3, see arrow). The proteins that were retained were analyzed in Western blots probed with a monoclonal antibody to human RNase L, thus distinguishing the 83.5-kDa native RNase L from the 110-kDa GST-RNase L. In the absence of 2-5A, there was little or no interaction between the two proteins (Fig. 3A, lane 4). The removal of the C-terminal 21-amino acid segment did not impair the 2-5A-dependent binding of the proteins (Fig. 3A, lane 5). However, deletion of 80 or 399 C-terminal amino acid residues results in a loss of the ability to bind native RNase L, as observed by the absence of the 83-kDa RNase L (Fig. 3A, lanes 7 and 9). These results show that the C-terminal 22-80-amino acid portion is necessary for RNase L/RNase L binding activity. The monoclonal antibody used to detect RNase L recognizes an epitope in the C-terminal half of RNase L, between residues 342 and 661; therefore, RNase LCDelta 399 could not be detected on the Western blot with this antibody. However, the RNase LCDelta 399 was clearly visualized by staining the protein in the gel with Coomassie Blue dye (data not shown). Despite repeated attempts, we were unable to obtain clearly interpretable data on the RNase L binding activities of the C-terminal mutants lacking 31, 41, or 51 amino acid residues. This was due to the relatively low levels of expression of these mutant proteins compared with the amount of protein required for the assay. In addition, some of the GST fusion proteins were less stable than others. The N-terminal mutants were also assayed for the ability to bind native RNase L. Removal of the N-terminal 23-amino acid segment had no effect on the binding to native RNase L in response to 2-5A (Fig. 3B, lanes 9 and 10). However, an N-terminal deletion including the first ankyrin repeat resulted in the loss of this function (see RNase LNDelta 56 in Fig. 3B, lane 7). Interestingly, RNase LNDelta 56 is also defective for 2-5A binding (Fig. 2). Further N-terminal truncations in RNase L also resulted in proteins that lacked the ability to bind with native RNase L (Fig. 3B, lanes 1-6). Therefore, amino acid residues near both termini of RNase L were required for the binding to native RNase L. 


Fig. 3. Mapping the dimerization domains in RNase L. Extracts containing wild type or mutants of RNase L fused to GST were incubated with recombinant, native human RNase L from insect cells in the presence and absence of pA(2'p5'A)3. Analysis of the glutathione-Sepharose-bound protein was by SDS-PAGE and Western blot analysis probed with monoclonal antibody to RNase L. Panel A, lanes 1 and 2 contain 1 µg and 0.15 µg of insect cell extract containing human recombinant RNase L (9). The position of native RNase L is indicated (arrows).
[View Larger Version of this Image (27K GIF file)]

The Ribonuclease Domains Maps to a C-terminal Fragment of RNase L

To measure the endoribonuclease activity of the mutant proteins, a radiolabeled, short RNA consisting of 25 uridylate residues linked at the 3' terminus to [32P]pCp, U25-[32P]pCp, was synthesized as substrate. In addition, shorter fragments of oligo(U)-[32P]pCp were present as a result of premature terminations during the chemical synthesis of the U25 (Fig. 4A, lane 1). Analysis of RNA cleavages after incubation with the wild type and mutant forms of RNase L was performed in the presence or absence of 2-5A (Fig. 4A). Incubations with RNase Lwild type produced potent 2-5A-dependent ribonuclease activity (Fig. 4A, lanes 2 and 3). The cleavage products were <= 4 nucleotides in length. Deletion of the N-terminal 23 residues had no effect on 2-5A-dependent RNase activity (lanes 4 and 5). However, RNase LNDelta 56, missing sequence including the first ankyrin repeat, lacked ribonuclease activity in the presence or absence of 2-5A (lanes 6 and 7). RNase LNDelta 90, RNase LNDelta 123, and RNase LNDelta 237 (missing 2, 3, and 6 ankyrin repeats, respectively) also lacked the ability to degrade RNA (Fig. 4A, lanes 8-11, and data not shown). Surprisingly, however, deletion of all nine ankyrin repeats, produced a mutant protein, RNase LNDelta 335, with unregulated ribonuclease activity (Fig. 4A, lanes 12 and 13), i.e. ribonuclease activity was observed with RNase LNDelta 335 even in the absence of the activator, 2-5A.


Fig. 4. Localization of the ribonuclease domain. A, purified GST fused to wild type and mutant forms of RNase L were incubated with U25-[32P]pCp in the presence and absence of pA(2'p5pA)3 and the RNA was separated in sequencing gels. The position of the intact, U25-[32P]pCp, is indicated with an arrow. An autoradiogram of a gel is shown. B, quantitation of the levels of intact RNA, U25-[32P]pCp, remaining after the incubations was measured from the autoradiograms and is expressed as a percentage of the input level of intact RNA as averaged from two separate experiments (except the data from RNase LNDelta 56 was from a single experiment). White bars, no additions; black bars, plus pA(2'p5'A)3.
[View Larger Version of this Image (62K GIF file)]

To further localize the ribonuclease domain, the C-terminal mutant proteins were also analyzed. RNase LCDelta 21, lacking 21 C-terminal residues, displayed wild type 2-5A-dependent RNase activity (Fig. 4A, lanes 14 and 15). However, mutant RNase LCDelta 31, lacking an additional 10 residues was inactive (Fig. 4A, lanes 16 and 17). All of the mutants containing further C-terminal truncations, beyond residue 710, were similarly inactive (Fig. 4A, lanes 18-29). Quantitation of the intact RNA remaining after the incubations confirmed that 2-5A-dependent RNase activity required the N-terminal segment that included the first ankyrin repeat and the 10-amino acid segment located between residues 710 and 720 (Fig. 4B).

Comparison of the Relative Activities and Substrate Specificities of RNase Lwild type and RNase LNDelta 335

The unregulated mutant, RNase LNDelta 335, caused a complete loss of intact RNA, but the appearance of some incomplete RNA digestion products show that its ribonuclease activity was diminished compared with wild type enzyme (Fig. 4A, compare lane 3 to lanes 12 and 13). To directly compare the relative activities of the two proteins, the effects of protein concentration on RNA cleavage was measured with U25-[32P]pCp as substrate. Reactions with RNase Lwild type were performed by preincubating with 2-5A prior to addition of substrate to avoid an initial lag that occurs when 2-5A and substrate are added at the same time (16). The concentrations of protein required to obtain 50% loss of intact RNA in 30 min at 30 °C was 20 nM RNase Lwild type and 115 nM RNase LNDelta 335 (Fig. 5). Therefore, RNase LNDelta 335 was about 6-fold less active than RNase Lwild type. In the absence of 2-5A, no RNA cleavage was seen with 300 nM RNase Lwild type, whereas 300 nM RNase LNDelta 335 caused an 87% decrease in the level of intact RNA (Fig. 5).


Fig. 5. Relative catalytic activities of RNase Lwild type and RNase LNDelta 335. RNase Lwild type was preincubated and assayed in the presence (open circle ) or absence (triangle ) of pA(2'p5'A)3. RNase LNDelta 335 (square ) was preincubated and assayed in the absence of pA(2'p5'A)3. Reactions were at 30 °C for 30 min with U25-[32P]pCp as substrate.
[View Larger Version of this Image (16K GIF file)]

To compare the RNA cleavage site preferences of RNase Lwild type and RNase LNDelta 335, the proteins were incubated in the presence or absence of 2-5A with a fragment of PKR mRNA (13). The cleavage sites were mapped by primer extension assays in comparison with the products of DNA sequencing reactions of PKR cDNA (Fig. 6). RNase Lwild type and RNase LNDelta 335 cleaved the RNA at precisely the same sites. In both cases, the two nucleotides immediately 5' to the cleavage sites were UA (products 1 and 3), AU (products 2 and 4), and UG (product 5) (Fig. 6). With PKR mRNA as substrate, RNase LNDelta 335 appear to be somewhat less active compared with RNase Lwild type than with oligo(rU) as substrate. RNA cleavage by RNase Lwild type was dependent upon addition of 2-5A, whereas the ribonuclease activity of RNase LNDelta 335 was identical in the presence or absence of 2-5A.


Fig. 6. Cleavage site preferences of RNase Lwild type and RNase LNDelta 335 in PKR mRNA. Lanes 1-6 show primer extension reactions on PKR mRNA fragments produced by incubation with RNase Lwild type (WT) or RNase LNDelta 335 (NDelta 335), in the presence or absence of 2-5A, as indicated. Lanes 1 and 2 were with 60 ng of protein, lanes 3 and 4 with 1 µg of protein, and lanes 5 and 6 with 300 ng of protein. Lanes 7-9 show sequencing reactions of PKR cDNA. The nucleotides complementary to those that were sequenced are indicated above lanes 7-9, and the sequence is indicated to the right. The cleavage sites are indicated with arrows and are numbered 1-5.
[View Larger Version of this Image (47K GIF file)]

Localization of the RNA Binding Domain

To map the RNA binding domain, the affinities of the various GST-mutant proteins for poly(U)-Sepharose and glutathione-Sepharose were compared (Fig. 7). The RNase L mutants truncated only from the N terminus showed similar affinities for poly(U)-Sepharose and glutathione-Sepharose (Fig. 7A, lanes 1-12). Analysis of the C-terminal truncated mutants showed that RNase LCDelta 21 bound poly(U)-Sepharose but with reduced activity (Fig. 7A, lane 14). However, further deletions from the C terminus, beginning with RNase LCDelta 31, result in a loss in poly(U) binding activity (Fig. 7A, lanes 15-28). These results were measured and expressed as percentage of the activities of the mutant proteins compared with the wild type protein (Fig. 7B). The data show that the substrate binding activity and the RNase function both require the amino acid sequence between residues 710 and 720. 


Fig. 7. Mapping the RNA binding activity of RNase L. A, proteins bound to glutathione-Sepharose (G) or poly(U)-Sepharose (U) were determined by probing Western blots with polyclonal antibody to human RNase L reactive to all of the RNase L mutants. The positions of the molecular size markers (in kDa) are indicated. B, percentage of RNA binding activity is expressed as [(GST-mutant RNase L bound to poly(U)-Sepharose/GST-mutant RNase L bound to glutathione-Sepharose)/(GST-RNase Lwild type bound to poly(U)-Sepharose/GST-RNase Lwild type bound to glutathione-Sepharose)] × 100. Results were averages of two separate experiments.
[View Larger Version of this Image (58K GIF file)]


DISCUSSION

Mapping the Structural and Functional Domains of RNase L

Here we have studied the structure, function, and regulation of RNase L through a series of truncations from either terminus (Fig. 1). The 2-5A binding activity of the protein is localized in the N-terminal half of RNase L. For example, RNase LCDelta 406, which lacks 406 of the 741 amino acid residues of RNase L, binds 2-5A nearly as well as the complete, wild type enzyme (Fig. 2). Previously, we showed that the 2-5A binding domain of the murine RNase L was expressed in a polypeptide consisting of residues 1-342 (2). Furthermore, substitution of the P-loop motif lysines, at residues 240 and 274 in a truncated murine RNase L resulted in a defect in 2-5A binding activity. In this study, we show that the N-terminal 23 amino acid residues do not contribute to 2-5A binding activity. However, the first ankyrin repeat, residues 24-56, is required. Although RNase LCDelta 399 contains the complete ankyrin repeat region, a presumptive protein-protein interaction domain with 2-5A binding activity, it fails to bind with native RNase L in the presence or absence of 2-5A (Figs. 1 and 3). Results show that both termini are necessary for RNase L binding activity. For instance, RNase LCDelta 80 also lacks the ability to bind native RNase L. Therefore, while the ankyrin repeats may contribute to RNase L/RNase L binding, they are insufficient for this function.

The C-terminal 31 residues of RNase L are critical for the catalytic function of the enzyme. RNase LCDelta 31, lacking residues 711-741, has neither ribonuclease nor substrate binding activities. In contrast, RNase LCDelta 21 has full activity in the presence of 2-5A. These findings clearly show that residues 711-720, EYRKHFPQTH, are essential for the RNA binding and ribonuclease activities of the enzyme. However, it is likely that portions of the catalytic domain are located further upstream (toward the N terminus) or that deletion of the C-terminal segment results in conformational alterations that impair function. Interestingly, the C-terminal region of the yeast Ire1p protein, a putative ribonuclease that functions in the splicing of the HAC1 mRNA during the unfolded protein response, has significant homology with the residues 587-706 of RNase L (17, 18). The region of homology with Ire1p suggests that amino acid residues that are N-terminal to residue 706 are also important for the catalytic function of RNase L.

Regulation of RNase L Activity

Results of this study show that the ankyrin repeat region of RNase L functions as a potent repressor. Binding of 2-5A could cause a conformational change in the enzyme that releases the inhibitory effect of the ankyrin repeats leading to both an unmasking of the ribonuclease and interaction domains near the C terminus. Previous studies involving multiple biochemical and biophysical methods established that the binding of 2-5A to monomers of RNase L induces the formation of RNase L homodimers (7, 8). However, additional studies will be required to determine if dimerization is actually required for ribonuclease activity. Dimer formation could be a mechanism for maintaining the enzyme in an active conformation by preventing the ankyrin clamp from reforming. The failure of RNase LNDelta 56 to bind to either 2-5A or to native RNase L was consistent with the requirement of 2-5A for dimerization (7, 8). Presumably, the protein-protein interaction domain in RNase LNDelta 56 cannot be unmasked because the protein lacks the capacity to bind with 2-5A. It is unknown if the ankyrin repeats directly interact with the catalytic domain. However, the failure of the complete ankyrin repeat region in RNase LCDelta 80 and RNase LCDelta 399 to bind with native RNase L in the presence of 2-5A argues against a direct interaction (Fig. 3).

The repressor function requires only three, or possibly fewer, of the nine ankyrin repeats. RNase LNDelta 335, lacking all nine ankyrin repeats was a constitutive ribonuclease whereas RNase LNDelta 237, lacking ankyrin repeats 1-6, was in a repressed state. Therefore, ankyrin repeat domains 7-9 by themselves are capable of causing repression. RNase Lwild type and RNase LNDelta 335 produced identical RNA cleavage sites, providing further evidence that the RNA binding and catalytic sites are localized in the C-terminal half of the enzyme (Fig. 6). Cleavages occurred 3' of UA, UG, and AU dinucleotide sequences in the PKR mRNA fragment. The first two of these RNA cleavages site preferences have been previously reported for RNase L, while cleavages 3' of AU have not been reported (16, 19, 20). 2-5A binding to a complete RNase L releases the inhibition by the ankyrin repeat regions of the protein. However, because mutants RNase LNDelta 56, RNase LNDelta 90, RNase LNDelta 123, and RNase LNDelta 237 lack the ability to bind 2-5A, addition of 2-5A fails to release of the repression (Fig. 1). To date, we have not observed a trans inhibition of ribonuclease activity by mixing RNase LCDelta 399 with RNase LNDelta 335.2 However, further studies are required to determine if the ankyrin repeats must be present in the same polypeptide chain as the catalytic domain to silence the ribonuclease activity.

Bipartite Structure of RNase L

The results of this study show that the 2-5A binding and repression functions of RNase L map to different regions than the RNA binding and ribonuclease domains. Additional features in the C-terminal half are a high cysteine content segment and limited homology to protein kinases, both regions are of unknown function (1, 2, 17). A binary model of RNase L emerges from these studies in which the regulatory functions map to the ankyrin repeat region in the N-terminal half, while the ribonuclease portion localizes to the C-terminal part of the enzyme (Fig. 1).


FOOTNOTES

*   This investigation was supported by United States Public Health Service Grant CA 44059 (to R. H. S.) from the Department of Health and Human Services, National Cancer Institute.The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Dagger    To whom all correspondence should be addressed: Dept. of Cancer Biology, NN1-06, The Cleveland Clinic Foundation, 9500 Euclid Ave., Cleveland, OH 44195. Tel.: 216-445-9650; Fax: 216-445-6269.
1   The abbreviations used are: 2-5A, p1-3A(2'p5'A)>=2; PCR, polymerase chain reaction; PBS, phosphate-buffered saline; PAGE, polyacrylamide gel electrophoresis; PKR, dsRNA-dependent protein kinase.
2   B. Dong and R. H. Silverman, unpublished data.

ACKNOWLEDGEMENTS

We thank Bryan Williams (Cleveland) for comments and for the gift of PKR cDNA, Paul F. Torrence (Bethesda) for the generous gift of the bromine-substituted 2-5A analog, Guiying Li (Cleveland) for synthesizing p(A2'p5'A)3, to Aimin Zhou (Cleveland) for radiolabeling the 2-5A and RNA.


REFERENCES

  1. Silverman, R. H. (1997) in Ribonucleases: Structure and Function (D'Alessio, G., and Riordan, J. F., eds), pp. 516-552, Academic Press, Inc., New York
  2. Zhou, A., Hassel, B. A., and Silverman, R. H. (1993) Cell 72, 753-765 [Medline] [Order article via Infotrieve]
  3. Williams, B. R. G., Golgher, R. R., Brown, R. E., Gilbert, C. S., and Kerr, I. M. (1979) Nature 282, 582-586 [Medline] [Order article via Infotrieve]
  4. Silverman, R. H., Cayley, P. J., Knight, M., Gilbert, C. S., and Kerr, I. M. (1982) Eur. J. Biochem. 124, 131-138 [Abstract]
  5. Chebath, J., Benech, P., Revel, M., and Vigneron, M. (1987) Nature 330, 587-588 [CrossRef][Medline] [Order article via Infotrieve]
  6. Hassel, B. A., Zhou, A., Sotomayor, C., Maran, A., and Silverman, R. H. (1993) EMBO J. 12, 3297-3304 [Abstract]
  7. Dong, B., and Silverman, R. H. (1995) J. Biol. Chem. 270, 4133-4137 [Abstract/Free Full Text]
  8. Cole, J. L., Carroll, S. S., and Kuo, L. C. (1996) J. Biol. Chem. 271, 3979-3981 [Abstract/Free Full Text]
  9. Dong, B., Xu, L., Zhou, A., Hassel, B. A., Lee, X., Torrence, P. F., and Silverman, R. H. (1994) J. Biol. Chem. 269, 14153-14158 [Abstract/Free Full Text]
  10. Silverman, R. H., Jung, D. D., Nolan-Sorden, N. L., Dieffenbach, C. W., Kedar, V. P., and SenGupta, D. N. (1988) J. Biol. Chem. 263, 7336-7341 [Abstract/Free Full Text]
  11. Knight, M., Cayley, P. J., Silverman, R. H., Wreschner, D. H., Gilbert, C. S., Brown, R. E., and Kerr, I. M. (1980) Nature 288, 189-192 [Medline] [Order article via Infotrieve]
  12. Nolan-Sorden, N. L., Lesiak, K., Bayard, B., Torrence, P. F., and Silverman, R. H. (1990) Anal. Biochem. 184, 298-304 [Medline] [Order article via Infotrieve]
  13. Meurs, E., Chong, K., Galabru, J., Thomas, N. S. B., Kerr, I. M., Williams, B. R. G., and Hovanessian, A. G. (1990) Cell 62, 379-390 [Medline] [Order article via Infotrieve]
  14. Sobol, R. W., Henderson, E. E., Kon, N., Shao, J., Hitzges, P., Mordechai, E., Reichenbach, N. L., Charubala, R., Schirmeister, H., Pfleiderer, W., and Suhadolnik, R. J. (1995) J. Biol. Chem. 270, 5963-5978 [Abstract/Free Full Text]
  15. Gorina, S., and Pavletich, N. P. (1996) Science 274, 1001-1005 [Abstract/Free Full Text]
  16. Carroll, S. S., Chen, E., Viscount, T., Geib, J., Sardana, M. K., Gehman, J., and Kuo, L. C. (1996) J. Biol. Chem. 271, 4988-4992 [Abstract/Free Full Text]
  17. Bork, P., and Sanders, C. (1993) FEBS Lett. 334, 149-152 [CrossRef][Medline] [Order article via Infotrieve]
  18. Sidrauski, C., Cox, J. S., and Walter, P. (1996) Cell 87, 405-413 [Medline] [Order article via Infotrieve]
  19. Wreschner, D. H., McCauley, J. W., Skehel, J. J., and Kerr, I. M. (1981) Nature 289, 414-417 [Medline] [Order article via Infotrieve]
  20. Floyd-Smith, G, Slattery, E., and Lengyel, P. (1981) Science 212, 1030-1032 [Medline] [Order article via Infotrieve]

©1997 by The American Society for Biochemistry and Molecular Biology, Inc.