RNA (2000), 6:79–87+ Cambridge University Press+ Printed in the USA+
`Copyright © 2000 RNA Society+
`
`Expanded CUG repeat RNAs form hairpins that
`activate the double-stranded RNA-dependent
`protein kinase PKR
`
`BIN TIAN,1 ROBERT J. WHITE,2 TIANBING XIA,3 STEPHEN WELLE,4
`DOUGLAS H. TURNER,3 MICHAEL B. MATHEWS,1 and CHARLES A. THORNTON2
`1Department of Biochemistry and Molecular Biology, New Jersey Medical School, University of Medicine
`and Dentistry of New Jersey, Newark, New Jersey 17103, USA
`2Department of Neurology, School of Medicine and Dentistry, University of Rochester, Rochester, New York 14642, USA
`3Department of Chemistry, School of Medicine and Dentistry, University of Rochester, Rochester, New York 14642, USA
`4Department of Medicine, School of Medicine and Dentistry, University of Rochester, Rochester, New York 14642, USA
`
`ABSTRACT
`Myotonic dystrophy is caused by an expanded CTG repeat in the 39 untranslated region of the DM protein kinase
`(DMPK) gene. The expanded repeat triggers the nuclear retention of mutant DMPK transcripts, but the resulting
`underexpression of DMPK probably does not fully account for the severe phenotype. One proposed disease mech-
`anism is that nuclear accumulation of expanded CUG repeats may interfere with nuclear function. Here we show by
`thermal melting and nuclease digestion studies that CUG repeats form highly stable hairpins. Furthermore, CUG
`repeats bind to the dsRNA-binding domain of PKR, the dsRNA-activated protein kinase. The threshold for binding to
`PKR is ;15 CUG repeats, and the affinity increases with longer repeat lengths. Finally, CUG repeats that are patho-
`logically expanded can activate PKR in vitro. These results raise the possibility that the disease mechanism could be,
`in part, a gain of function by mutant DMPK transcripts that involves sequestration or activation of dsRNA binding
`proteins.
`Keywords: CUG repeat; double-stranded RNA; dsRNA-binding motif; muscle; myotonic dystrophy; PKR
`
`INTRODUCTION
`Myotonic dystrophy (dystrophia myotonica, DM) is an
`autosomal-dominant, multisystem disease in which the
`phenotype is highly variable (Harper, 1989)+ The con-
`genital form of DM is characterized by delayed muscle
`maturation, mental retardation, and respiratory dis-
`tress+ The adult-onset form is characterized by muscle
`wasting, neuropsychiatric impairment, heart block, and
`cataracts+ The only disease manifestation in some late-
`onset cases is presenile cataract+
`The genetic basis for DM is an expanded CTG re-
`peat in the 39 untranslated region of the DM protein
`kinase (DMPK) gene (Brook et al+, 1992)+ The number
`of CTG repeats at this locus ranges from 5 to 37 re-
`peats in the normal population, whereas individuals with
`DM have from 50 to more than 3,000 repeats+ The
`severity of disease manifestations correlates with the
`length of the expanded repeat+
`
`Reprint requests to: Charles A+ Thornton, Box 673, URMC, Room
`5-4306, 601 Elmwood Avenue, Rochester, New York 14642, USA;
`e-mail: charles_thornton@urmc+rochester+edu+
`
`79
`
`Transcripts from the mutant DMPK allele are re-
`tained in the nucleus (Taneja et al+, 1995; Davis et al+,
`1997), resulting in a modest reduction of DMPK protein
`in skeletal muscle (Maeda et al+, 1995)+ Partial loss of
`DMPK probably does not, however, provide a unitary
`explanation for the phenotype, because the develop-
`ment and maturation of skeletal muscle proceeds nor-
`mally in mice that are homozygous for a DMPK-null
`allele (Jansen et al+, 1996; Reddy et al+, 1996)+ Al-
`though one strain of DMPK knockout mice developed a
`mild, late-onset myopathy, the phenotype was not clearly
`similar to DM (Reddy et al+, 1996)+ Of note, point mu-
`tations in the DMPK gene have not been found in any
`kindreds with DM+
`One proposed mechanism for the pathogenesis of
`DM is that nuclear accumulation of expanded CUG
`repeats is deleterious+ A similar mechanism may be
`operating in spinocerebellar ataxia type 8 (SCA8), a
`dominantly inherited neurodegenerative disease that is
`also caused by the expansion of an untranslated CTG
`repeat (Koob et al+, 1999)+ As there are no precedents
`in human genetics for transdominant RNA gain-of-
`function, mechanistic models are at an early stage of
`
`Benitec - Exhibit 1009 - page 1
`
`

`
`B.Tianetal.
`
`and prevents the synthesis of antisense contaminants
`(Mellits et al+, 1990) or hybrid sense/antisense run-off
`transcripts (Triana-Alonso et al+, 1995)+ Similarly, CAG
`repeat transcripts were synthesized in the absence of
`UTP+
`The normalized UV absorption melting curves for six
`(CUG)n transcripts ranging in size from 5 to 69 repeats
`all displayed a single sharp transition with 44–72% hy-
`perchromicity (Fig+ 2)+ The melting temperatures were
`;75 8C for each of the (CUG)ntranscripts, but the tran-
`sitions were steeper, and the hyperchromicities larger,
`for longer repeats+ The melting temperature of the tran-
`script with 69 repeats was independent of RNA con-
`centration over a 100-fold range of concentrations (not
`shown)+ By comparison, CAG repeat RNAs showed a
`broader transition with a lower melting point and a
`smaller hyperchromicity+ As expected, the CUC repeat
`RNA showed no consistent transition+
`The melting profiles of CUG repeat transcripts were
`consistent with a simple, unimolecular secondary struc-
`ture, such as the predicted hairpin, or other, more com-
`plex, multistem/loop folded structures+ To distinguish
`among these, RNAse T1, an endoribonuclease that
`cleaves preferentially after unpaired guanosines, was
`used to map the position of nuclease-sensitive loops+
`
`80
`development+ Timchenko and colleagues (Timchenko
`et al+, 1996; Philips et al+, 1998) have proposed that
`expanded CUG repeats titrate a CUG-binding protein,
`CUGBP1, rendering it unavailable to perform its nor-
`mal function as a splice enhancer+
`To investigate the RNA gain-of-function model, we
`have studied the structural properties and protein in-
`teractions of CUG repeats in vitro+ A panel of triplet
`repeat RNAs was synthesized by in vitro transcription+
`Optical melting and nuclease mapping studies indicate
`that CUG repeats form highly stable hairpins+ Although
`conventional base pairing in the stem of the hairpin is
`interrupted by a periodic U•U mismatch, these tran-
`scripts are able to bind, and to activate PKR, the dsRNA-
`activated protein kinase+ These observations suggest
`that interactions with dsRNA-binding proteins could play
`a role in the nuclear retention or toxicity of mutant DMPK
`transcripts+
`
`RESULTS
`An RNA folding algorithm (Walter et al+, 1994; Mathews
`et al+, 1999) predicts that CUG repeats form stable
`hairpins (Fig+ 1A)+ G•C and C•G base pairs in the stem
`of the hairpin are separated by a periodic U•U mis-
`match+ The most stable structure predicted for a CUG
`repeat of any length is the hairpin with a single loop
`and the longest possible stem, but multistem/loop struc-
`tures (e+g+, Fig+ 1B) are also possible, as their predicted
`free energies are only slightly less favorable+ CAG re-
`peat RNAs are also predicted to form hairpins, whereas
`CUC repeat RNAs are not expected to form any stable
`secondary structure+
`To test these predictions, UV absorbance melting ex-
`periments were conducted using a panel of triplet re-
`peat RNAs synthesized by in vitro transcription+ The
`plasmid templates for synthesizing CUG and CUC re-
`peat transcripts were designed so that ATP could be
`omitted from the transcription reactions+ This design
`minimizes the RNA-dependent polymerase activity of
`the T7 RNA polymerase (Cazenave & Uhlenbeck, 1994),
`
`FIGURE 1. Folding algorithms predict a stable secondary structure
`for CUG repeats+ A: A simple hairpin is the most stable secondary
`+ B: One example of an alternative
`structure predicted for (CUG)20
`multiloop structure+
`
`FIGURE 2. Thermal melting profiles of triplet repeat RNAs+ Absor-
`bance was normalized relative to the value at 80 8C (for CUC re-
`peats) or 97 8C (all other transcripts)+ Transcripts have the following
`formulas: GGGCGG(UGC)nU for CUG repeats, GGGAGG(AGC)n
`for CAG repeats, and 59-GGCGCUGG(CCU)nCCC for CUC repeats+
`
`Benitec - Exhibit 1009 - page 2
`
`

`
`CUGrepeatsbindtodsRBM
`Digests of the 35-, 69-, and 140-repeat transcripts re-
`vealed a nuclease sensitive site near the midpoint of
`each transcript (Fig+ 3)+ Thus, under the renaturation
`and digestion conditions used here, even the longest
`transcript (423 nt) formed a simple hairpin with an ex-
`tended stem, rather than a series of smaller folds+ These
`data confirm observations by Napierala and Krzyzo-
`siak (1997), who used RNase and lead cleavage to
`map hairpin loops in transcripts with up to 49 CUG
`repeats, and extend the findings into the range of re-
`peat sizes that cause DM+ Expanded CUG re-
`peats, however, are resistant to cleavage by RNAse III
`at a concentration that completely degrades authentic
`dsRNAs (Fig+ 4A), confirming that the duplex structure
`is not due to a contaminating authentic dsRNA+ Evi-
`dently, the U•U mismatches in the hairpin stem confer
`resistance to this dsRNA-modifying enzyme+
`These results suggest that the dominant structural
`feature of an expanded CUG repeat is the duplex-like
`stem+ To support this conclusion, we studied the inter-
`actions of CUG repeats with a dsRNA-binding protein+
`PKR was selected for these analyses because its bind-
`ing activity is well characterized+ Furthermore, an in-
`ventory of muscle mRNA using the serial analysis of
`gene expression (SAGE) method (Velculescu et al+,
`1995) indicated that PKR transcripts were more abun-
`dant in skeletal muscle than transcripts for other dsRNA-
`binding proteins (Table 1)+
`Interactions between triplet repeat RNAs and p20,
`the bacterially expressed RNA-binding domain of PKR,
`were shown in protection assays, where relatively high
`concentrations of p20 were able to suppress the diges-
`tion of (CUG)69 by RNAse T1 (Fig+ 4B)+ More detailed
`examination of the interaction was conducted by gel
`mobility shift analysis+ Figure 5A shows that the (CUG)69
`transcript was retarded by p20 at a concentration of
`
`81
`50 mg/mL, whereas transcripts containing 10 or 20 re-
`peats were unaffected+ At this concentration, the (CUG)35
`transcripts were partially shifted to a slower mobility,
`confirming that p20 has a greater affinity for longer
`CUG repeat transcripts+ More detailed analysis indi-
`cated that the threshold length for p20 interactions lies
`between 10 and 15 repeats: (CUG)20 was shifted at
`high concentrations of p20, as was (CUG)15 to a lesser
`extent, but (CUG)10 and (CUG)5 were not discernibly
`retarded (Fig+ 5B)+ The behavior of transcripts contain-
`ing 20–140 CUG repeats toward increasing p20 con-
`centrations demonstrates the progressively increasing
`affinity of p20 for longer repeat transcripts (Fig+ 5C)+
`Furthermore, this experiment shows that increasing con-
`centrations of p20 result in complexes of diminishing
`mobility, which suggests the binding of additional mol-
`ecules of p20 to long CUG repeats+ These findings are
`to be compared with the packing density of ;11 bp for
`p20 binding to regular dsRNA duplexes (Manche et al+,
`1992)+ p20 also binds to CAG repeat RNAs to a limited
`extent, but with lower affinity than the (CUG)69 and
`(CUG)140 transcripts+ RNAs containing 21 or 56 CAG
`repeats bound with an affinity similar to that of (CUG)20
`(data not shown)+ CUC repeat RNA did not bind to p20,
`,
`as expected for an unstructured RNA+ For (CUG)140
`the longest CUG repeat RNA tested, the binding affinity
`is bracketed by that of two naturally occurring viral RNAs
`that are inhibitors of PKR activation (Fig+ 5D)+ The af-
`finity is higher than that of HIV-1 TAR RNA and com-
`parable to adenovirus-2 VA RNAI (Kd 5 3+5 3 1027 M
`(Schmedt et al+, 1995)), suggesting that the binding of
`CUG repeat RNA to p20 may be biologically significant+
`PKR is subject to both positive and negative regula-
`tion by RNA molecules (reviewed by Mathews & Shenk,
`1991; Clemens, 1996)+ Extended perfect duplexes, such
`as those present in reovirus RNA, activate the kinase,
`whereas certain highly structured RNAs that are not
`perfectly duplexed, such as VA RNAI
`, prevent activa-
`tion+ To determine the nature of the interactions with
`CUG repeat RNAs, in vitro kinase assays were con-
`ducted using PKR purified from interferon-treated hu-
`man 293 cells+ PKR was strongly activated by (CUG)140
`and (CUG)69 as evidenced by autophosphorylation
`(Fig+ 6) or by phosphorylation of its substrate, eIF2a
`(data not shown)+ Weak activation occurred with (CUG)35
`+ Thus, PKR
`but not detectably with (CUG)20 or (CUG)15
`was activated by pathologically-expanded CUG re-
`peats but not by repeat lengths in the normal range for
`the DM locus+
`
`DISCUSSION
`
`FIGURE 3. Nuclease mapping of CUG repeat RNAs+ (CUG)35, 69,
`and 140 transcripts were digested with ribonuclease T1 (0 U for
`lane a, 0+01 U for lane b, 0+1 U for lane c, 1 U for lane d, and 10 U
`for lane e), and then loaded on denaturing gels+ Markers indicate the
`size of the original CUG repeat transcripts or RNA standards syn-
`thesized from the polylinker of pBSK+
`
`The optical melting studies indicate that CUG repeats
`form highly stable secondary structures with melting
`points of ;75 8C, irrespective of their lengths+ Based on
`optical melting studies of (CTG)10 and (CTG)30 ssDNA
`hairpins, which showed a similar constancy of melting
`
`Benitec - Exhibit 1009 - page 3
`
`

`
`82
`
`B.Tianetal.
`
`FIGURE 4. CUG repeat RNA is resistant to RNase III and partially protected by p20 from RNase T1 digestion+ A: Double
`gel-purified (CUG)69 and 145-bp dsRNA were incubated with T1 RNase or RNase III, then analyzed by 8 M urea/5%
`polyacrylamide gel electrophoresis+ Lanes 1 and 6: incubated without RNase; lanes 2 and 7: 0+2 U/
`mL RNase T1; lanes 3
`and 8: 0+02 U/
`mL RNase T1; lanes 4 and 9: 10 units RNase III; lanes 5 and 10, 1 U RNase III+ B: (CUG)100 was incubated
`with p20 then digested with T1 RNase at the concentrations shown (in units/
`mL)+ The p20 concentrations were: 0 mg/mL
`for lanes 1, 2, 6, 10, and 14; 1+2 mg/mL for lanes 3, 7, 11, and 15; 12 mg/mL for lanes 4, 8, 12, and 16; and 120 mg/mL for
`lanes 5, 9, 13, and 17+
`
`temperature, it was proposed that the (CTG)30 oligo-
`mer formed a series of smaller loops (Petruska et al+,
`1998)+ Our nuclease-mapping experiments, however,
`indicate that transcripts with up to 140 CUG repeats
`preferentially form a single-loop hairpin with an ex-
`tended stem+ The same result was obtained with tran-
`scripts that were refolded after chemical (Fig+ 3) or
`thermal (Fig+ 4) denaturation+ The constancy in melting
`temperature for CUG repeats of different lengths is pre-
`sumably due to the relatively small contribution of the
`loop to the total H8 and S8 for folding+ These studies
`confirm the earlier nuclease-mapping studies of non-
`expanded CUG repeats (Napierala & Krzyzosiak, 1997),
`and extend the findings into the range of repeat sizes
`that cause DM and SCA8+ Our data suggest, therefore,
`that the transition from wild-type to disease allele (.50
`repeats at the DM locus, and *90 repeats at the SCA8
`locus) is related to increasing length of the hairpin stem,
`rather than the adoption of a fundamentally different
`structure+
`PKR binding and activation studies support the idea
`that CUG repeats form hairpins+ PKR binds RNA via its
`
`TABLE 1+ Relative abundance of transcripts that encode dsRNA-
`binding proteins in normal skeletal muscle: number of occurrences in
`a database of 110,000 SAGE tags+
`
`46
`PKR
`6
`SON
`3
`ADAR1
`2
`ADAR2
`1
`RNA helicase A
`1
`PACT
`K12h4+8 putative helicase
`1
`NF90/ILF3/MMP4/DRBP76
`0
`0
`TRBP
`0
`2–5A synthetases
`0
`STAUFFEN
`SAGE analysis (Velculescu et al, 1995) was carried out on pooled
`muscle mRNAs from 16 healthy adult subjects+ The total counts for
`ADAR1, ADAR2, 2–5A synthetases, and TRBP include all isoforms
`deposited in GenBank (release #112)+ The tally could overestimate
`the abundance of a particular transcript, in the event that its SAGE
`tag is not unique (i+e+, another transcript shares the same 14-nt se-
`quence at the first NlaIII restriction site upstream from the polyad-
`enylation site)+ Note that transcripts that do not appear once in the
`database could still be significantly expressed in muscle (the 95%
`upper confidence bound for transcripts that were not detected is ;4
`per myonucleus, based on the Poisson distribution and an assump-
`tion of 150,000 transcripts per nucleus)+
`
`Benitec - Exhibit 1009 - page 4
`
`

`
`CUGrepeatsbindtodsRBM
`
`83
`
`FIGURE 5. Mobility shift assays for purified triplet repeat RNAs using p20+ A: Purified (CUG)10, 20, 35, and 69 repeat
`RNAs with same concentration (mg/mL) were incubated (a) with 0+05 mg/mL p20, or (b) without p20, and loaded on native
`gels+ B: (CUG)5,10,15, and 20 repeat RNAs with same concentration (mol/mL) were incubated with different concentrations
`of p20 protein (a: 0 mg/mL, b: 0+01 mg/mL, c: 0+05 mg/mL, d: 0+1 mg/mL, e: 0+5 mg/mL, and f: 1 mg/mL), and loaded on
`native gels+ C: (CUG)20,35,69, and 140 repeat RNAs were mixed together at the same concentration (mg/mL), then
`incubated with various concentrations of p20 protein (a: 0+5 mg/mL, b: 0+25 mg/mL, c: 0+05 mg/mL, d: 0+025 mg/mL, e: 0+005
`mg/mL, f: 0+0025 mg/mL, and g: 0 mg/mL), and loaded on a native gel+ D: (CUG)140, VA RNAI from Adenovirus-2, and TAR
`RNA from HIV-, all purified by the same method and at same concentration, were incubated with different concentrations of
`p20 protein (a: 0+5 mg/mL, b: 0+1 mg/mL, c: 0+05 mg/mL, d: 0+01 mg/mL, e: 0+005 mg/mL, f: 0 mg/mL), and loaded on a
`native gel+
`
`two dsRNA binding motifs (dsRBMs), both of which are
`required for efficient binding to dsRNA as well as to
`highly structured ssRNAs (Schmedt et al+, 1995)+ The
`interactions between PKR and CUG repeat RNAs can
`be compared to their interactions with regular dsRNAs
`
`FIGURE 6. PKR activation by purified CUG repeat RNAs+ Autoradio-
`gram showing autophosphorylation of PKR after incubation with vary-
`ing concentrations of the indicated RNAs (a: 0+2 mg/mL, b: 2 mg/mL,
`c: 20 mg/mL, d: no RNA)+
`
`of known sizes+ Although p20 can pack onto dsRNA as
`closely as ;11 bp (Manche et al+, 1992; Schmedt et al+,
`1995), the threshold for binding is slightly higher, about
`15–20 bp (Manche et al+, 1992; Bevilacqua & Cech,
`1996)+ Similarly, the minimum length of CUG repeat
`RNA that gives detectable binding to p20 in gel shift
`assays (Fig+ 5) is 45 nt (15 repeats)+ Kinase activation
`appears to require PKR dimerization and intermolecu-
`lar autophosphorylation (reviewed by Robertson et al+,
`1996)+ Accordingly, the minimum dsRNA chain length
`for activation is ;30 bp, and the efficiency of activation
`rises with increasing duplex size up to a maximum at
`;85 bp (Manche et al+, 1992)+ With CUG repeat RNA,
`the shortest polymers that activated PKR were 105 nt
`(35 repeats), and longer versions were increasingly ef-
`fective (Fig+ 6)+ These findings demonstrate a greater
`tolerance by PKR for mismatches in an RNA activator
`than has been observed previously+ CUG hairpins with
`U•U mismatches at every third base pair are able to
`
`Benitec - Exhibit 1009 - page 5
`
`

`
`84
`activate PKR, whereas dsRNAs with guanosine•inosine
`mismatches at every eighth base pair, on average, were
`unable to activate in an earlier study (Minks et al+, 1979)+
`PKR and RNase III are both dsRBM-containing en-
`zymes that are specific for dsRNAs, although not to the
`exclusion of mismatched duplexes (Hunter et al+, 1975;
`Robertson & Dunn, 1975; Minks et al+, 1979)+ Strikingly,
`whereas CUG repeat hairpins bind and activate PKR,
`they are not cleaved by RNase III+ It will be interesting
`to define the determinants responsible for this discrim-
`ination+ One other imperfectly duplexed RNA has been
`shown to bind and activate PKR even though it is in-
`sensitive to RNase III (Robertson et al+, 1996)+ Hepa-
`titis delta RNA is highly structured but, like the CUG
`hairpin, it contains only short stretches of uninterrupted
`Watson–Crick base pairs (Circle et al+, 1997)+
`The high stability of the CUG repeat hairpin suggests
`that expanded CUG repeats may form extended hair-
`pins in vivo+ Indeed, duplex formation may be the trig-
`ger for nuclear retention of mutant DMPK transcripts,
`as nuclear retention has been observed for dsRNA
`viral transcripts (Kumar & Carmichael, 1997), and some
`dsRNA-binding proteins are strongly associated with
`the nuclear matrix (Wreschner et al+, 1985; Schroder
`et al+, 1988)+ Furthermore,
`the lack of adenosine
`residues in CUG repeats may confer resistance to en-
`zymes, such as the dsRNA-activated adenosine de-
`aminases in combination with I-RNAse (Scadden &
`Smith, 1997), that target long duplex RNAs for rapid
`degradation+ This unusual sequence characteristic may
`allow expanded CUG repeats to accumulate to higher
`levels in the nucleus than would be permitted for other
`duplex RNAs+ In accord with this idea, in situ hybrid-
`ization studies reveal that mutant DMPK transcripts ac-
`cumulate in hundreds of foci per DM myonucleus (Davis
`et al+, 1997), which is higher than the frequency of ;8
`transcripts per myonucleus that would be predicted from
`the normal abundance of DMPK mRNA (based on 11
`occurrences of DMPK in a database of 110,000 muscle
`SAGE tags, an assumption that both alleles are tran-
`scriptionally active, and an estimate of 150,000 tran-
`scripts per myonucleus; S+ Welle, C+ Thornton, unpubl+)
`Long duplex RNAs elicit potent biologic effects+ In
`plants and invertebrates, dsRNAs provoke posttran-
`scriptional gene silencing in a gene-specific, homology-
`dependent fashion+ A different effect is observed in
`mammalian cells, where long duplex RNAs trigger a
`potent cellular stress response, independent of their
`sequence (reviewed by Haines et al+, 1991; Jacobs &
`Langland, 1996)+ Natural activators of the dsRNA re-
`sponse include viral dsRNAs (genomes, replication in-
`termediates, or products of bidirectional transcription)+
`Many downstream effects of the dsRNA response, such
`as inhibition of translation initiation, accelerated mRNA
`and rRNA degradation, and apoptosis, can be under-
`stood as an attempt by host cells to prevent or slow
`viral replication+ It is uncertain whether this response
`
`B.Tianetal.
`has functions unrelated to its antiviral effects, or whether
`it is activated by any endogenous transcripts+ Although
`the conformation of expanded CUG repeats in vivo re-
`mains uncertain, we have recently observed that ex-
`pression of an untranslated, expanded CTG repeat in
`transgenic mice can reproduce many of the muscle
`manifestations of DM, supporting the idea that the patho-
`genic effect of the DM mutation is mediated by mutant
`transcripts (C+ Thornton, unpubl+)
`Our results suggest that dsRNA-binding proteins are
`candidates for involvement in the nuclear retention or
`toxicity of expanded CUG repeats+ PKR is an interest-
`ing candidate because: (1) it is significantly expressed
`in skeletal muscle (Table 1); (2) it may play a role in the
`regulation of muscle differentiation (Kronfeld-Kinar et al+,
`1999); (3) a significant fraction of PKR is localized in
`the nucleus (Jimenez-Garcia et al+, 1993; Jeffrey et al+,
`1995); and (4) the length threshold for PKR activation
`by CUG repeats coincides with the threshold for caus-
`ing DM by expanded CTG repeats (Fig+ 6)+ Further-
`more, activation of PKR inhibits protein synthesis,
`through its ability to phosphorylate the translation-
`initiation factor eIF2a (reviewed by Clemens, 1996;
`Mathews, 1996)+ It is noteworthy that the rate of mus-
`cle protein synthesis in vivo, derived from measure-
`ments of 13C-leucine incorporation into muscle protein,
`is reduced in DM patients compared to healthy sub-
`jects (Griggs et al+, 1990) or patients with other forms of
`muscular dystrophy (R+ Tawil, pers+ comm+)+
`Alternatively, the toxicity of expanded CUG repeats
`may involve sequestration, rather than activation, of
`dsRNA-binding proteins+ Instability of the DM mutation
`in somatic cells can produce very large expansions of
`the CTG repeat+ The expanded CUG repeats in mus-
`cle, heart, and brain range up to 30-fold longer than the
`longest transcript studied here, because they are gen-
`erated from alleles with up to 13 kb of CTG repeats
`(Thornton et al+, 1994)+ The extreme length of the ex-
`panded repeat and its focal accumulation in myonuclei
`(Taneja et al+, 1995) may generate local concentrations
`that are high enough to act as a sink for binding pro-
`teins+ The affinity of expanded CUG repeats for dsRNA-
`binding proteins appears adequate to support such
`interactions, as the binding affinity for PKR is similar to
`, an adenovirus-2 transcript that inhibits
`that of VA RNAI
`PKR activation when it accumulates to high levels late
`in infection (reviewed by Mathews & Shenk, 1991)+ In
`view of the sequence nonselectivity that characterizes
`dsRNA interactions with dsRNA-binding proteins, and
`the presence in nuclear extracts of several proteins
`that bind preferentially to expanded CUG repeats (B+
`Tian & M+B+ Mathews, data not shown), it is possible
`that several different proteins are susceptible to this
`effect+ Efforts to identify other proteins that bind to ex-
`panded CUG repeats and to develop model systems to
`study gain of function by these transcripts are currently
`underway+
`
`Benitec - Exhibit 1009 - page 6
`
`

`
`CUGrepeatsbindtodsRBM
`
`MATERIALS AND METHODS
`
`Plasmid templates for synthesis
`of triplet repeat RNAs
`
`Plasmid pBCE was constructed by using oligonucleotides to
`insert BsmI and BamHI restriction sites bracketed by two
`BsmF1 sites at the BamHI-EcoRI site of pBSKII+ Oligonucle-
`otides were used to insert a modified T7 RNA polymerase
`promoter (C substituted for A at the 14 position) at the BsmF1-
`HindIII site of pBCE, creating plasmid pCNA+ Derivatives of
`pCNA with 5, 10, 15, 20, 25, and 35 CTG repeats were made
`by cycles of (CTG)5 linker addition at the BsmF1 restriction
`site+ A PCR product with a larger CTG repeat was obtained
`by amplification of the expanded repeat from a patient with
`mild DM as described (Thornton et al+, 1994)+ The BsmI and
`DpnII fragment containing the expanded repeat was cloned
`at the BsmI-BamHI site of pBCE+ The resulting plasmid,
`, contains an insert with 74 uninterrupted CTG re-
`pBCE74
`peats+ BsmF1 digestion of this plasmid releases a CTG re-
`peat fragment+ Nonpalindromic overhangs orient this fragment
`for ligation into concatamers of uninterrupted CTG repeats,
`or directional ligation into pCNA+ This fragment and its dimer
`were inserted at the BsmF1 site of pCNA to create plasmids
`+ These plasmids contain 69 and 140
`pCNA69 and pCNA140
`uninterrupted CTG repeats, respectively, fused directly to the
`T7 promoter+ pCNA100 is a subclone of pCNA140 that under-
`went a deletion within the repeat tract+
`Plasmid pTQA52 with 52 CTC repeats was made by using
`oligonucleotides to insert an MscI site fused to a modified T7
`promoter (GGCGCT in the 11 to 16 positions) at the SacI-
`PstI site in pBSKII+ Oligonucleotides 59-(CCT)9CC and 59-
`(GGA)9GG were used to synthesize expanded CTC repeat
`fragments by slipped strand replication in a thermal cycler as
`described (Ordway & Detloff, 1961)+ These PCR products
`were ligated into the MscI site of pTQA+ By a similar strategy,
`plasmids pASU29 and pASU56 were constructed with 29 and
`56 uninterrupted CAG repeats fused to the wild-type T7
`promoter+
`The stability of triplet repeat plasmids was optimized by
`orienting the CTG repeat on the leading strand of DNA rep-
`lication, cloning in strain HB101, growing cultures at 30 8C,
`and harvesting cultures before they reached stationary phase
`(Bowater et al+, 1996)+ Despite these measures, plasmids
`with more than 50 CTG repeats showed minor instability, in
`that subclones often differed from the parent plasmid by 61
`to 5 repeats+ All triplet repeat inserts were sequenced by
`dideoxy cycle sequencing+
`
`RNA synthesis and purification
`Plasmids in the pCNA, pASU, pTQA series were linearized at
`the final triplet repeat by digestion with BsmF1, BbsI, or
`Eco47III, respectively+ Linearized plasmids were extracted
`with phenol/chloroform and precipitated in ethanol+ RNA for
`gel shift assays, PKR activation experiments, and nuclease
`digests in Figure 4 were synthesized using T7 RNA polymer-
`ase as described (Mellits et al+, 1990), except that the nucle-
`otide concentrations were modified+ Nucleotides used for
`synthesis of CUG and CUC repeat RNA were 6 mM GTP and
`CTP, 12 mM UTP, and 1 mCi/mL [a-32P]UTP+ For reactions
`
`85
`
`labeled with CTP, 6 mM GTP and UTP, 12 mM CTP, and
`1 mCi/mL [a-32P]CTP were used instead+ ATP was omitted
`from transcription reactions for CUG and CUC repeats, and
`UTP was omitted from transcription reactions for CAG re-
`peats+ Transcription reactions were treated with 20 mg/mL
`DNaseI and 1 mM CaCl2 for 30 min at 37 8C, extracted with
`phenol/chloroform, and precipitated with ethanol+ The tran-
`scripts were purified sequentially on denaturing, followed by
`nondenaturing polyacrylamide gels as described (Mellits et al+,
`1990)+
`RNA for T1 digests in Figure 3 was synthesized with T7
`RNA polymerase using unlabeled nucleotides, treated with
`shrimp alkaline phosphatase,
`labeled at the 59 end with
`[g232P]ATP using T4 polynucleotide kinase, and then puri-
`fied on 40 cm 5 or 8% polyacrylamide/6 M urea gels+ RNA
`was eluted from excised gel bands overnight at 4 8C in 0+2%
`SDS, 0+5 M ammonium acetate, pH 6, and then precipitated
`and washed in ethanol+ For optical melting experiments, a
`fragment containing the triplet repeat and T7 promoter was
`first isolated from each plasmid+ RNA was synthesized using
`T7 RNA polymerase, treated with DNAse I for 30 min at
`37 8C, extracted with phenol/chloroform, passed through a
`Sepharose cartridge to remove unincorporated nucleotides,
`precipitated with ammonium acetate and ethanol, and then
`purified on 18 cm 6% polyacrylamide/6 M urea gels with UV
`shadowing+ Purity of these transcripts was verified by end
`labeling with [g232P]ATP, followed by analysis on sequenc-
`ing gels+
`
`Melting curves
`
`Ethanol-precipitated RNA was resuspended in water and then
`, 150 mM
`adjusted to give a final concentration of 5 mM MgCl2
`KCl, 10 mM PIPES, pH 7+0, and 0+5 mM EDTA+ Absorbance
`versus temperature-melting curves were measured at 280 nm
`with a heating rate of 0+5 8C/min from 10 to 97 8C on a Gilford
`250 spectrometer controlled with a Gilford 2527 thermopro-
`grammer+ Melting temperatures were calculated as described
`(Serra et al+, 1994)+
`
`Nuclease digests
`For Figure 3, 59 end-labeled RNA was dissolved in 10 mM
`Tris, pH 7+0, at a concentration of 1,500 cpm/
`mL+ Nuclease
`digests were prepared on ice as 5-mL reactions containing
`3 mL RNA and various concentrations of RNAse T1 in
`12+5 mM Tris, pH 7+5, 5 mM MgCl2
`, and 75 mM KCl+ Di-
`gests were incubated at 37 8C for 45 min, and then loaded
`directly on 40-cm 5% polyacrylamide denaturing gels along
`with internally labeled RNA size standards+ Gels were dried
`and subjected to autoradiography+ For Figure 4, digestion
`with RNase III (kindly provided by H+D+ Robertson, Cornell
`Medical School) was conducted in 30 mM Tris, pH 8+0,
`100 mM NaCl, 5 mM MgCl2
`, 0+1 mM EDTA, and 0+02 mM
`DTT; and digestion with RNase T1 was conducted in 10 mM
`Tris, pH 7+4, 150 mM NaCl, and 1 mM EDTA+ Incubation
`was for 45 min at 37 8C+ For p20 protection analysis, RNA
`was first incubated with p20 on ice for 20 min in the buffer
`used for mobility shift assays, and then incubated with RNase
`T1 for 45 min at 37 8C+
`
`Benitec - Exhibit 1009 - page 7
`
`

`
`86
`
`Frequency of transcripts that encode
`dsRNA-binding proteins
`
`mRNA was isolated from normal adult human skeletal mus-
`cle biopsies, pooled, and analyzed by the SAGE technique
`(Velculescu et al+, 1995) as described (Welle et al+, 1999)+
`Half of the tags (54,791) came from muscle of young men
`(21– 42 years old, n 5 8) and half (54,802) from muscle of
`older men (66–77 years old, n 5 8)+ The number of occur-
`rences for transcripts that encode each dsRNA binding pro-
`tein was counted in the database of 110,000 SAGE tags+
`
`Mobility shift assay
`
`The Escherichia coli-expressed dsRNA binding domain of
`PKR, p20, was purified to apparent homogeneity and used in
`gel shift assays as described (Schmedt et al+, 1995)+ p20 was
`incubated with in vitro-transcribed RNA and the other com-
`ponents indicated for 20 min at 4 8C and resolved on a 5%,
`40 mM Tris-glycine polyacrylamide gel run at 4 8C+
`
`Kinase assay
`
`PKR from interferon-treated human 293 cells was partially
`purified to the Mono S stage as described (Kostura & Mathews,
`1989)+ Kinase reactions (10 mL) were carried out with 0+5 mL
`PKR (about 5 ng) and 2+5 mCi of [g32P]ATP (ICN Radiochem-
`icals) as described (Manche et al+, 1992)+ Reactions were
`resolved on 10% polyacrylamide/SDS gels, fixed, dried, and
`subjected to autoradiography+
`
`ACKNOWLEDGMENTS
`
`This work was supported by the Muscular Dystrophy Asso-
`ciation, grant AI34552 from the National Institutes of Health
`to MBM, a predoctoral fellowship 9810005T fro