genesdev.cshlp.orgDownloaded from
`
` on July 17, 2015 - Published by
`
`
`
` Cold Spring Harbor Laboratory Press
`
`Two RNA-binding motifs in the double-stranded RNA-activated protein kinase, DAI Simon R. Green and Michael B. Mathews 1 Cold Spring Harbor Laboratory, Cold Spring Harbor, New York 11724 USA. The protein kinase DAI, the double-stranded RNA-activated inhibitor of translation, is an essential component of the interferon-induced cellular antiviral response. The enzyme is regulated by the binding of activator and inhibitor RNAs. We synthesized DAI in vitro and located its RNA-binding domain within the amino-terminal 171 residues. This domain contains two copies of an RNA-binding motif characterized by a high density of basic amino acids, by the presence of conserved residues, and by a probable ,,-helical structure. Deletion of either of the two motifs prevents the binding of dsRNA, but their relative positions can be exchanged, suggesting that they cooperate to interact with dsRNA. Clustered point mutations within the RNA-binding motifs and duplications of the individual motifs indicate that the first copy of the motif plays the more important role. Mutations that impair binding have similar effects on the binding of double-stranded RNAs of various lengths and of adenovirus VA RNA~, implying that discrimination between activator and inhibitory RNAs takes place subsequent to RNA binding. [Key Words: Kinase; translational control; VA RNA; RNA-binding motif; double-stranded RNA; interferon] Received August 6, 1992; revised version accepted September 21, 1992. Phosphorylation plays an important role in regulating the activities of many of the protein components of the translational machinery (Hershey 1989, 1990, 1991 ). One of the most intensively studied of these regulatory events involves the phosphorylation of eukaryotic initi- ation factor 2 (eIF-2). This factor is comprised of three different subunits (or, B, and ~) of which both the a and J3 subunits are phosphorylated (Hershey 1990, 1991). Al- though the physiological significance of eIF-213 phospho- rylation remains to be determined, it is well established that phosphorylation of the ot-subunit can lead to the rapid cessation of protein synthesis (Hershey 1990, 1991). In one of the first steps of translational initiation, eIF-2 transports the initiator tRNA IMet-tRNA~) in a ter- nary complex with GTP to the 40S ribosomal subunit (for review, see Moldave 1985; Pain 1986; Hershey 1991). The a- and ~3-subunits of eIF-2 are also involved in trans- lational start site selection in yeast (Donahue et al. 1988; Cigan et al. 1989). Before the association of the 40S com- plex with the 60S subunit to form the 80S ribosomal complex, eIF-2 is released in a binary complex with GDP that must be replaced by GTP to permit the binding of a flesh molecule of Met-tRNAi and reentry of the factor into the initiation process (Moldave 1985; Pain 1986; Hershey 1991). This regeneration is catalyzed by the gua- nosine nucleotide exchange factor (GEF or eIF-2B) (Konieczny and Safer 1983; Panniers and Henshaw 1983). If the a-subunit of eIF-2 is phosphorylated at Ser sl, the 1Corresponding author. exchange reaction is blocked by the formation of a non- dissociable complex between GEF and eIF-2. GDP (Proud 1986; Colthurst et al. 1987). Initiation ceases in the absence of free GEF, and because GEF is present in cells at a lower concentration than eIF-2 (Safer 1983), the recycling factor may be completely sequestered when only a fraction (20-50%) of the eIF-2et in the cell is phos- phorylated. Three known protein kinases are capable of phospho- rylating eIF-2a on Set sl and they mediate a variety of control processes. In yeast, GCN2 kinase causes a gene- specific translational derepression rather than a general shutdown of polypeptide synthesis (Dever et al. 1992). Under starvation conditions, this kinase is responsible for the translational induction of the transcriptional ac- tivator GCN4 by phosphorylating eIF-2et, thereby allow- ing ribosomes to bypass the regulatory open reading frames in the 5' leader sequence of GCN4 mRNA (Hin- nebusch 1990; Dever et al. 1992). GCN2 kinase is prob- ably activated by the presence of uncharged tRNAs (Wek et al. 1990). The other two eIF-2et kinases are character- istic of higher eukaryotes (for review, see Ochoa 1983; Mathews et al. 1990; Hershey 1991). The heme con- trolled repressor (HCR or HRI) is found mainly in retic- ulocytes. It is activated by a number of stimuli, most notably by the absence of heme or Fe 2 +, and it serves to prevent globin synthesis in the absence of its prosthetic group, heine (Jackson 1991). The third kinase, the dou- ble-stranded RNA (dsRNA)activated inhibitor (DAI, also termed DsI, p68, and P1 kinase), is present in most cell 2478 GENES & DEVELOPMENT 6:2478-2490 (cid:14)9 1992 by Cold Spring Harbor Laboratory ISSN 0890-9369/92 $3.00
`
`Benitec - Exhibit 1008 - page 1
`
`

`
`
`
`genesdev.cshlp.orgDownloaded from
`
` on July 17, 2015 - Published by
`
`
`
` Cold Spring Harbor Laboratory Press
`
`RNA-binding motifs in DAI kinase types. DAI plays an important role in the interferon-in- duced antiviral response, and it has also been associated with cellular differentiation {Petryshyn et al. 1984; Jud- ware and Petryshyn 1991), the inhibition of cell prolifer- ation (Chong et al. 1992), the heat shock response (Dubois et al. 1989; Edery et al. 1989), and possibly tran- scriptional activation (Zinn et al. 1988). The regulation of DAI is poorly understood. The en- zyme is usually present in cells at a low level and in an inactive state. Its cellular concentration is increased by interferon at the transcriptional level (Hovanessian 1989; Samuel 1991), whereas its activity is regulated by both activator and inhibitor RNAs (Mathews and Shenk 1991). Activation of DAI is accompanied by autophos- phorylation of the latent enzyme, an event that appar- ently unmasks the ability of the enzyme to phosphory- late eIF-2oL (Farrell et al. 1977; Levin and London 1978; Sen et al. 1978; Berry et al. 1985; Galabru and Hovanes- sian 1987; Kostura and Mathews 1989). Optimal activa- tion of DAI requires dsRNA that is perfectly duplexed and greater than -85 bp in length, but there is no RNA sequence dependence {Hunter et al. 1975; Minks et al. 1979, 1980; Manche et al. 1992). At relatively high con- centrations, activation is inhibited by short dsRNAs (less than -30 bp), which are not capable of activating DM (Minks et al. 1979, 1980; Manche et al. 1992). Fur- thermore, small, highly structured, single-stranded RNAs such as adenovirus virus-associated (VA) RNAI, the EBERs of Epstein-Barr virus, and TAR RNA encoded by human immunodeficiency virus-1 (HIV-1) are special- ized effectors: They block DAI activation at relatively high concentrations but are incapable of activating the enzyme (for review, see Mathews and Shenk 1991). Par- adoxically, at similar high concentrations, long dsRNA also becomes inhibitory {Hunter et al. 1975; Farrell et al. 1977; Lenz and Baglioni 1978). To understand the complex molecular interactions be- tween DAI and RNA effectors, we undertook a muta- tional analysis of the protein, examining the ability of a variety of deletion, truncation, and substitution mutants to interact with RNA. The RNA-binding domain com- prises two basic regions located in the first 171 amino acids of the DAI sequence. Each of these regions contains a consensus RNA-binding motif and a putative a-helix. Mutations in the second motif are not as debilitating as similar mutations in motif 1, and the duplication of the second motif does not compensate for deletion of the first motif. Positional interchange of the two regions is not deleterious and their spacing can be varied, within limits, indicating that they cooperate to form the RNA- binding site. Clustered point mutations in these two re- gions affect the binding of long and short dsRNAs and of adenovirus VA RNA~ in the same way, implying that inhibitory RNAs bind to the same site as activators. Results Localization of the RNA-binding domain DM is a protein kinase that binds to dsRNA, leading to autophosphorylation and activation of the enzyme's ability to phosphorylate eIF-2. Inspection of its predicted amino acid sequence, derived from a cDNA clone (Meurs et al. 1990; Thomis et al. 1992), suggested that the pro- tein can be divided into two regions of roughly equal size. The 270 amino acids of the carboxyl terminus con- tain the 11 subdomains that are essential for kinase ac- tivity and evidently comprise the catalytic domain of a protein kinase (Fig. 1A; Hanks et al. 1988). Because the first 280 amino acids of the protein are free of such cat- alytic domains, we hypothesized that this region is in- volved in regulating kinase activity and contains the RNA-binding domain{s) of the enzyme. The amino-ter- minal half of the protein possesses no RNA-binding mo- tifs typical of small nuclear ribonucleoproteins (snRNPs) (Kenan et al. 1989), but it contains a high density of lysine and arginine residues (Fig. 1B), which have been implicated in both DNA helix and RNA hairpin binding (Lazinski et al. 1989; Steitz 1990). These basic residues are concentrated in three distinct regions within the amino terminus of DM (Fig. 1A). To examine their role in RNA binding, we excised the basic regions individu- ally and assessed the ability of the deleted proteins to interact with RNA ligands. Templates encoding the three deletion mutants, A1, A2, and A3 (Fig. 1A), were transcribed in vitro with bac- teriophage T7 RNA polymerase, and the resultant RNAs were translated in a wheat germ cell-free system to gen- erate 3SS-labeled mutant proteins. These proteins were then compared to the full-length protein for their ability to bind to either dsRNA or adenovirus VA RNA~ coupled to Sepharose beads. Equal volumes of the translation products {Fig. 2A) were exposed to the immobilized ligands, and the resultant complexes containing radiola- beled protein were analyzed by gel electrophoresis and autoradiography (Fig. 2B,C). Neither al nor A2 {deletions of amino acids 1-97 and 104-157, respectively) was able to bind to dsRNA or VA RNA, but a3 (a deletion of amino acids 234-272) retained its ability to bind both types of RNA (Fig. 2B, C; lanes 3-5). Furthermore, a de- letion mutant lacking sequences near the carboxyl ter- minus (A4, a deletion of residues 482-523) and a point mutant in kinase domain II (Lys 296 ~ Arg) also bound to both RNA matrices (Fig. 2B, C; lanes 6 and 2). Back- ground binding to beads lacking an RNA ligand was neg- ligible (Fig. 2B, lane 8). These data suggested that the first two regions of basic amino acids are essential for RNA binding, whereas the third basic region is not required. To vary the spacing between the first two basic re- gions, we also made two internal deletion mutants, A5 and h6. Decreasing the distance between basic regions 1 and 2 by four amino acids (AS) had very little effect on dsRNA binding, whereas removing 19 amino acids from this area (A6) abrogated dsRNA-binding ability (see Fig. 5, lanes 6,7 below). Therefore, the spacing is not abso- lutely critical, but some of the residues between the two basic regions appear to be indispensable. In addition to the full-length wild-type protein, many shorter polypeptides also bound to the dsRNA-Sepha- rose matrix (Fig. 2B, lane 1). The smallest of these poly- peptides, which are probably carboxy-terminal trunca- GENES & DEVELOPMENT 2479
`
`Benitec - Exhibit 1008 - page 2
`
`

`
`
`
`genesdev.cshlp.orgDownloaded from
`
` on July 17, 2015 - Published by
`
`
`
` Cold Spring Harbor Laboratory Press
`
`Green and Mathews A 100 200 l I 18 80 107 159 I I I 1 2 I K296R ~1 ~ ..... 103 158 ~3 I a4 I DAI (amino acids) 30O t 233 268 I ,, ! 3 =1, (cid:12)9 (cid:12)9 (cid:12)9 (cid:12)9 II I II IV L K---~ R 233 273 400 500 551 f f m ,._=.. .. - I Vl VII VIII IX Xl 524 A A6 [ ,, 84 104 481 157 As I I a9 I I 184 al0 1 4 278 B LS2 LS10 GALGAL I~E]o G v rRl LS3 LS11 GAL GAL r~DrRr~ ~]E F P LS4 LS 13 GAL A A I~ls~-T~l [] LS9 LS5 GAP GAL ^ ^ ^l'k-I ~EIR[~ LS12 LS6 GAL GAL 128 LS7 LS8 G A--'--L GAL r~ FK] C rK1 [] [] [] AI~]L " ,,~v~vv~J, vvv~ 160 Eael 1 I Ddel BsaAI 186 Figure 1. Structure of DAd and mutants. (A) Schematic representation of full-length DAd and mutants. The top line represents the linear DAI protein sequence indicating the highly conserved kinase domains in the carboxy-terminal half of the protein (boxes I-XI) and the three basic regions within its amino-terminal half (thick lines 1, 2, and 3). The position of the single amino acid substitution in mutant K296R is marked. The lower lines depict the deletion mutants (Al-~6) and the sites of truncation (A7-M0). The numbers below each line denote the amino acid residues present in the deleted proteins. (B) LS mutants. The first 186 amino acids of the DAd sequence are represented, with the location of basic regions 1 and 2 indicated by thickening of the line. All the basic amino acids are boxed, and the residues changed by site-directed mutagenesis are shown with the mutant residues positioned above the mutated wild-type residues. Restriction enzyme sites used for the generation of truncated DNA (A7-A9) are also shown. tions resulting from premature termination events, had apparent molecular weights of <25,000. To define the minimum length of the amino-terminal segment of DAI required for RNA binding more precisely, truncated pro- teins were prepared (A7-A10, Fig. 1A) and tested in a similar manner. The truncated protein A9 (amino acids 1-184) bound to both types of RNA affinity matrix (Fig. 3A, lanes 6,9), whereas A7 (residues 1-155) did not bind efficiently {lanes 5,8), indicating that the carboxy-termi- nal boundary of the RNA-binding domain lies between residues 157 and 184. This assignment was confirmed in a second type of binding assay. DAI was immobilized on protein A-Sepharose beads by polyclonal antibody and was then exposed to synthetic 32p-labeled dsRNA of 85 bp. The bound RNA was quantitated by direct radioac- tive counting and visualized by electrophoresis and au- toradiography. Figure 3B shows that full-length DAI and the a9 truncated protein (residues 1-184) bound the la- beled dsRNA with similar efficiencies. Further experi- ments demonstrated that a reduction in length to 171 amino acids (A8) had little or no effect on the ability of DAI to bind dsRNA, whereas truncation at residue 157 (A7) effectively eliminated binding as in the dsRNA- Sepharose binding assay (see Fig. 5, lanes 2,3, below). Thus, we conclude that the KNA-binding domain lies within the first 171 amino acids of DAI, in agreement with the results of Katze et al. (1991) and Patel and Sen (1992). This domain contains both of the basic regions that appear to be essential. Mutational analysis of the RNA-binding domain To examine the two regions more closely, we made a series of clustered point mutations that changed basic amino acids to residues with uncharged side chains (Fig. 1B; Table 1). The mutations were made in such a way that three adjacent amino acids were exchanged in each mutant by linker scanning mutagenesis. Initially, seven such mutants were generated, LS2-LS8, distributed through the two basic regions in A9 (residues 1-184). In addition, we constructed a number of double mutants by combining two LS mutations (e.g., LS3,5, which contains 2480 GENES & DEVELOPMENT
`
`Benitec - Exhibit 1008 - page 3
`
`

`
`
`
`genesdev.cshlp.orgDownloaded from
`
` on July 17, 2015 - Published by
`
`
`
` Cold Spring Harbor Laboratory Press
`
`RNA-binding motifs in DAI kinase Figure 3. RNA-binding activity of truncated DAI. (A) RNA- Sepharose assay. RNA from full-length wild-type construct and the truncations A7 and A9 was translated. Equal volumes of the translation products (5 ~1) were tested in both dsRNA-Sepha- rose and VA RNA-Sepharose-binding assays. Translation prod- ucts and adsorbed proteins were analyzed by electophoresis in 15% polyacrylamide-SDS gels and autoradiography. (B) Immo- bilized DM assay. Equal amounts of radioactive protein of wild- type full-length and A9 DM were adsorbed to polyclonal anti- body and immobilized on protein A-Sepharose. After incuba- tion with 32P-labeled dsRNA {85 bp), the beads were washed and the resultant RNA-protein complexes were analyzed as in A. Figure 2. DAI translation and location of the RNA-binding domain. (A) Total translation products. Capped RNA tran- scripts synthesized by T7 polymerase in vitro, were used to program the wheat germ cell-free translation system. The re- sultant translation products were analyzed by electrophoresis in 12.5% polyacrylamide-SDS gels and autoradiography. Lanes 1-6 contain products from the wild-type clone (SRG2AL) and the mutants K296R, A1, A2, a3, and A4, respectively. Lane 7 contains ~4C-labeled molecular weight markers. (B) Binding to dsRNA-Sepharose. The translation products (5 ~1) were incu- bated with dsRNA-Sepharose beads. After washing, the ad- sorbed proteins were analyzed as in A (lanes 1-7). (Lane 8) Con- trol using A4 protein and beads lacking RNA ligand. (C) Binding to VA RNA-Sepharose. As in B, except that the translation products were incubated with VA RNA-Sepharose beads in- stead of dsRNA-Sepharose. both the LS3 and LS5 mutations). The mutant proteins were labeled by in vitro transcription and translation and were subjected to binding analysis using the dsRNA- Sepharose assay. Essentially equal amounts of protein were used in each assay (Fig. 4A). Strikingly, the introduction of the LS4 mutation resulted in a dramatic shift in gel mobility. DM migrates anomalously during electrophoresis in SDS gels, with an apparent molecular weight of 68,000 compared with 62,000 predicted from the cDNA se- quence (Meurs et al. 1990). The wild-type 184 residue A9 protein migrated with the expected mobility (equivalent to -20 kD), but the LS4 substitution reduced its appar- ent molecular mass by -4 kD. This effect was not ob- served with any of the other LS mutations, suggesting that LS4 may disturb a structurally important region of the DAI molecule. Two different algorithms (Chou and Fasman 1974; Gamier et al. 1978) predict that the orig- inal LS4 residues represent the amino terminus of an a-helix lying between residues 58 and 82 in region 1 (see below and Fig. 8B). Although the LS4 mutation, like all of the mutations described here, had been verified by DNA sequence analysis, we confirmed that the aberrant mobility was not the result of the presence of an unde- Table 1. LS mutants Original Mutated Mutant Position residues residues LS2 18-20 Arg Gin Lys Gly Ala Leu LS3 38-40 Asp Arg Arg Gly Ala Leu LS4 58-60 Arg Set Lys Gly Ala Leu LS5 78-80 Glu Lys Lys Gly Ala Leu LS6 111-113 Lys Lys Arg Gly Ala Leu LS7 134-136 Lys Cys Lys Gly Ala Leu LS8 158-160 Ala Lys Leu Gly Ala Leu LS9 66-68 Ma Ala Ala Gly Ala Pro LS10 21-23 Glu Gly Val Gly Ala Leu LS 11 51-53 Glu Phe Pro Gly Ala Leu LS12 108-110 Ile Ala Gln Gly Ala Leu LS13 61-64 Lys Glu Ala Lys Ala Glu Ala Ma GENES & DEVELOPMENT 2481
`
`Benitec - Exhibit 1008 - page 4
`
`

`
`
`
`genesdev.cshlp.orgDownloaded from
`
` on July 17, 2015 - Published by
`
`
`
` Cold Spring Harbor Laboratory Press
`
`Green and Mathews Figure 4. Binding of dsRNA to DAI mutants using the dsRNA- Sepharose assay. (A)Total translation products. Uncapped RNA was synthesized from a number of mutant templates truncated as for A9. Equal amounts of radioactive proteins were analyzed by electrophoresis in 20% polyacrylamide-SDS gels and auto- radiography. (B) Binding of DAI mutants to dsRNA. Equal amounts of labeled translated proteins were reacted with dsRNA-Sepharose beads and analyzed as in A. tected stop codon by truncating the template so as to yield a longer polypeptide of 278 residues (A10; Fig. 1A). Relative to the LS4 A9 polypeptide, the A 10 product ex- hibited the expected increase in apparent size (data not shown), thereby excluding the existence of undetected stop codons in the LS4 construct. However, this ex- tended LS4 protein still ran anomalously slowly in rela- tion to the wild-type A10 protein (data not shown), con- sistent with a structural peculiarity determined by the residues located in the region of the LS4 mutation. The ability of the mutant polypeptides to bind dsRNA varied dramatically (Fig. 4B). In the dsRNA-Sepharose assay, mutants LS3, LS5, and LS8 all appeared to bind dsRNA as efficiently as the wild-type A9 protein; LS6 and LS7 bound at a reduced level; and LS2 and LS4 did not appear to bind at all. Thus, two of the single muta- tions in the first basic region (LS2 and LS4) completely abrogated dsRNA binding, whereas two others (LS3 and LSS) had no effect. None of the single mutations in the second basic region (LS6, LS7 and LS8) had a dramatic effect on dsRNA binding, despite the observation that a deletion of this region eliminates RNA binding (Fig. 2A). However, when the mutations were combined to form double mutants, all nine of those tested displayed a sig- nificantly reduced ability to bind dsRNA (Fig. 4B). This was the case even with mutations in region 2 that had little effect on dsRNA binding individually (LS6, 7, LS6,8 and LS7,8; Fig. 4B, lanes 14,15,17), suggesting that they act synergistically to impair dsRNA binding. To extend and quantitate these results, we turned to the alternative binding assay using immobilized DAI. This assay permits estimation of the dsRNA bound by direct radioactive counting {Fig. 5A), as well as visual- ization of dsRNA and DAI by autoradiography (Fig. 5B). The A9 truncated form of DAI (residues 1-184) and its mutant derivatives were labeled with [3ss-]methionine, bound to antibody-Sepharose beads, and reacted with 32p-labeled dsRNA. For the single mutants, LS2-LS8, the results agreed closely with those derived from the dsRNA-Sepharose-binding assay, although the immobi- lized DAI assay appeared to be more sensitive to small differences in binding efficiency. Thus, LS3, LS5, and LS8 bound 85, 60, and 116% as much dsRNA as wild type and LS6 and LS7 bound only -10-15% as much dsRNA as the wild-type polypeptide. Binding was insig- nificant in LS2 and LS4. Another mutant in the first basic region, LS13, which contains two Lys -o Ala sub- stitutions separated by two unaltered residues {Fig. 1B; Table 1), also failed to bind detectable quantities of dsRNA (Fig. 5, lane 10). These data show that some, but not all, of the basic residues in the amino-terminus are critical for binding and confirm that both basic regions are important, although mutants in region 1 are more severely impaired. As expected from previous results {Fig. 4B), in general double mutants bound less dsRNA than the more se- verely impaired of the single mutants that constituted them (e.g., c.f. lane 12 with lanes 11 and 17). This was true even for those mutations (LS3, LSS, and LS8) that exerted minimal effects on their own, with one notable exception. Although the LS8 mutation caused a syner- gistic effect when combined with other region 2 muta- tions close to its own location (in LS6,8 and LS7,8), it exerted only a mart~inal effect when combined with more distant mutat~ns in region 1 {LS3,8 and LS5,8). This observation raises the possibility that the two bind- ing regions function semiautonomously. Definition of an RNA-bmding motif Up to this point we have focused on the basic residues located in the amino-terminal third of the DAI molecule. Most mutations of such residues were deleterious to dsRNA binding, but it was not possible to decide on the basis of these experiments alone whether it was the charge change or the structural consequence of the mu- tation that was important. To address this question, we made further mutations altering nonbasic amino acids in the proximity of basic residues that are required for dsRNA binding. Mutants LS9-12 were constructed to alter residues near LS13, LS2, LS4, and LS6, respectively. The introduction of the LS10 mutation reduced the ability of DAI to bind dsRNA only slightly, despite the fact that alterations in the preceding 3 amino acids in LS2 destroyed all ability to bind dsRNA (Fig. 5, lanes 8,33). On the other hand, the LS12 mutation reduced dsRNA-binding ability to 10% of the wild-type level, as seen for the adjacent mutation of basic residues, LS6 {Fig. 5, lanes 21,35). The LS9 and LSll mutations impaired binding even more severely (Fig. 5, lane 32; see Fig. 7, lane 17, below). Therefore, changes in nonbasic residues within the amino terminus of DAI can also affect its dsRNA-binding ability, and the basic regions do not function simply as regions of positive charge. Consistent with this conclusion, inspection of the se- quence of this and other RNA-binding proteins sug- 2482 GENES & DEVELOPMENT
`
`Benitec - Exhibit 1008 - page 5
`
`

`
`
`
`genesdev.cshlp.orgDownloaded from
`
` on July 17, 2015 - Published by
`
`
`
` Cold Spring Harbor Laboratory Press
`
`RNA-binding motifs in DAI kinase Figure 5. Quantitation of dsRNA binding to DAI mutants using the immobilized DAI assay. 32p-Labeled dsRNA (85 bp) was adsorbed to 3SS-labeled proteins immobilized on antibody-Sepharose beads. (A) The amount of bound dsRNA was determined by direct counting of radioactivity and is expressed relative to that adsorbed by the wild-type A9 protein. (B) RNA-protein complexes were analyzed by electrophoresis in 20% polyacrylamide-SDS gels and autoradiography. All mutant proteins were truncated as for a9, except for A8 and A7 (lanes 3,4). The controls in lanes 4 and 33 contained an unprogramed wheat germ extract and a mutant protein that is not recognized by the polyclonal antibody, respectively. gested the existence of a possible RNA-binding motif. Many of the deleterious mutations fall in highly con- served regions of the consensus sequence (see Discus- sion). Because the motif is present once in each of the basic regions, it seemed possible that the duplication of one region would restore binding activity to a mutant from which the other region was deleted. To test this idea, we constructed mutants sub 1 : 1 and sub 2 : 2 con- taining tandem repeats of regions 1 and 2, respectively (Fig. 6B). The region 1 duplication bound dsRNA nearly as efficiently as wild-type DAd, whereas the region 2 du- plication was severely defective for dsRNA binding (Fig. 6A), indicating that the two regions are not equivalent. Surprisingly, the mutant sub 2 : 1, in which the order of the two regions is reversed compared with the wild-type molecule, bound dsRNA with undiminished efficiency (Fig. 6). Similar data were obtained with VA RNA (not shown). These observations support the view that tan- dem copies of the motif are required for RNA binding but that motif 2 is less effective than motif 1. Binding of short RNA duplexes and VA RNA DM interacts with dsRNA in a length-dependent fash- ion: Duplexes of t--85 bp bind DAd and activate the en- zyme as efficiently as very long natural dsRNAs; short duplexes of ~<30 bp bind and activate the enzyme very weakly, and their most prominent effect is to inhibit activation by long dsRNAs; duplexes of intermediate size exhibit intermediate properties and also form elec- trophoretically distinct complexes with DAd (Manche et al. 1992). Such qualitative and quantitative differences might be reflected in differential effects of DM muta- tions on the binding of dsRNAs of varying sizes. For example, differential effects would be expected if short duplexes bound to a separate inhibitory site (Galabru et al. 1989). On the other hand, if there is a single site for binding both long and short dsRNA, residues at the ex- tremities of the dsRNA-binding site might be dispens- able for the binding of the smaller ligand. To determine whether DAd mutant proteins were capable of distin- guishing between dsRNAs of different sizes, a panel of 17 mutants immobilized on antibody-Sepharose beads was challenged with 32P-labeled dsRNA of 40 or 55 bp, as well as the 85-bp duplex used in previous experiments. The shorter dsRNAs bound less efficiently as expected (Manche et al. 1992), but the relative dsRNA binding was quantitatively unchanged among the DAd mutants (Fig. 7A). Thus, the relative binding efficiency of these mutants is independent of the length of the dsRNA, and DM mutations that discriminate between longer and shorter RNA duplexes were not identified. Adenovirus VA RNA~ is a small highly structured sin- gle-stranded RNA that binds to DAd and inhibits the binding of dsRNA (Kostura and Mathews 1989), as well as the activation of DAd by dsRNA (Kitajewski et al. GENES & DEVELOPMENT 2483
`
`Benitec - Exhibit 1008 - page 6
`
`

`
`
`
`genesdev.cshlp.orgDownloaded from
`
` on July 17, 2015 - Published by
`
`
`
` Cold Spring Harbor Laboratory Press
`
`Green and Mathews Discussion DAI is a cellular protein kinase, subject to both positive and negative control by RNA effectors. Several functions have been attributed to DAI, the most studied of which is its role in the interferon-induced host antiviral re- sponse. Activation of the kinase by virus-derived dsRNA shuts down protein synthesis and limits virus multipli- cation. As a counter-measure, some viruses synthesize short, highly structured single-stranded RNAs, such as VA RNA, which block the activation of DAI. To explore the molecular basis for these interactions we undertook a mutational analysis to define the RNA-binding do- main(s) in DAI. Two regions of DAI interact with RNA Noting that the amino terminus of the protein is rich in basic residues, we hypothesized that the positively charged side chains would form ionic bonds with the phosphodiester backbone of RNA. The basic amino acids are concentrated in three regions within the amino ter- minus of the protein. Deletion of the third basic region had no effect on RNA binding, but deletion of either the Figure 6. Interchange and duplication of RNA-binding motifs. {A) a2P-Labeled dsRNA was adsorbed to immobilized DAI mu- tant proteins. The amount of bound dsRNA is shown relative to that adsorbed by the wild-type DAI, truncated at amino acid residue 221. (B) Schematic representation of the proteins used in A. Numbers correspond to amino acid residues in the original DAI sequence. 1986; O'Malley et al. 1986}. It is not clear whether this is the result of direct competition between VA RNA~ and dsRNA for a single binding site on DAI (Kostura and Mathews 1989} or whether there are separate binding sites for the activator and inhibitor {Galabru et al. 1989}. To address this question, we tested the same 17 DAI mutants for their ability to bind VA RNA~. The immo- bilized DAI assay gave a nonspecific background of VA RNA~ binding presumably owing to interaction with a wheat germ protein capable of binding VA RNA~; there- fore, we used the VA RNA-Sepharose assay for this study. The ability of the DAI mutants to bind to VA RNA I {Fig. 7B) matched the pattern obtained for dsRNA binding to immobilized DAI (Fig. 7AJ. As also seen with the truncation mutant A7 (Fig. 3A, lane 8), however, DAI mutants that are unable to bind dsRNA often interact to a small but detectable extent with VA RNA-Sepharose. Because Sepharose beads lacking RNA ligands do not bind any detectable DAI {Fig. 2B, lane 8), this observation implies that DAI mutations that abrogate dsRNA bind- mg may still permit a weak interaction with VA RNAt. Further investigation of this phenomenon is under way. Figure 7. Binding of short dsRNAs and VA RNA. (A) Binding of dsRNA to immobilized DAI. RNAs of 40, 55, and 85 bp were exposed to mutant DAI proteins truncated as for Ag. RNA- protein complexes were analyzed by electrophoresis in 20% polyacrylamide-SDS gels and by autoradiography. The gels were exposed for 12, 8, and 4 hr, respectively, to allow for the differing affinities of DAI for the three RNAs. (B) Binding of DAI to VA RNA-Sepharose. Equal radioactive counts of translated DAI mutant A9 proteins were incubated with VA RNA-Seph- arose beads. The resultant protein-RNA complexes were ana- lyzed by electrophoresis in 20% polyacrylamide-SDS gels and autoradiography. 2484 GENES & DEVELOPMENT
`
`Benitec - Exhibit 1008 - page 7
`
`

`
`
`
`genesdev.cshlp.orgDownloaded from
`
` on July 17, 2015 - Published by
`
`
`
` Cold Spring Harbor Laboratory Press
`
`RNA-binding motifs in DAI kinase first or second basic region rendered DAI unable to bind RNA. In agreement with recent results from other labo- ratories (Katze et al. 1991; Chong et al. 1992; Feng et al. 1992; McCormack et al. 1992; Patel and Sen 1992), a truncated protein, comprising the first 171 amino acids and containing regions 1 and 2 in their entirety, was capable of binding RNA. Removal of an additional 14 amino acids, encroaching on basic region 2, eliminated dsRNA binding, suggesting