Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

Inversing the natural hydrogen bonding rule to selectively amplify GC-rich ADAR-edited RNAs

Inversing the natural hydrogen bonding rule to selectively amplify GC-rich ADAR-edited RNAs Published online 30 May 2008 Nucleic Acids Research, 2008, Vol. 36, No. 12 e72 doi:10.1093/nar/gkn295 Inversing the natural hydrogen bonding rule to selectively amplify GC-rich ADAR-edited RNAs 1,2 1,2 1,2 1,2 ` ´ Rodolphe Suspene , Myrtille Renard , Michel Henry , Denise Guetard , 1,2 3 3 ` ` David Puyraimond-Zemmour , Agnes Billecocq , Michele Bouloy , 2,4 1,2 1,2, Frederic Tangy , Jean-Pierre Vartanian and Simon Wain-Hobson * 1 2 3 Unite´ de Re´ trovirologie Mole´ culaire, CNRS URA 3015, Unite´ de Ge´ ne´ tique Mole´ culaire des Bunyaviride´ s and Laboratoire de Ge´ nomique Virale et Vaccination, Institut Pasteur, 28 rue du Dr Roux, 75724 Paris cedex 15 Received January 28, 2008; Revised April 25, 2008; Accepted April 29, 2008 amplify so-called G!A hypermutants of the human ABSTRACT immunodeficiency virus (HIV) (5). They arise from genetic DNA complementarity is expressed by way of editing of nascent viral cDNA by two host cell cytidine three hydrogen bonds for a G:C base pair and two deaminases of the APOBEC3 family (6–11). Deamination for A:T. As a result, careful control of the denatura- of numerous cytidine (C) residues on the viral minus tion temperature of PCR allows selective amplifica- strand yields multiple uracil (U) residues, which are copied tion of AT-rich alleles. Yet for the same reason, as a thymidine (T). With respect to the viral plus strand the converse is not possible, selective amplification as reference, these show up as genomes with numerous of GC-rich alleles. Inosine (I) hydrogen bonds to G!A transitions giving rise to the term G!A hyper- mutants (12,13). Temperature differences as small as cytosine by two hydrogen bonds while diaminopu- 1–28C were enough to allow differential amplification of rine (D) forms three hydrogen bonds with thymine. A rich hypermutants in the presence of as much as 10 fold By substituting dATP by dDTP and dGTP by dITP in a excess of wild type, or reference genomes (14,15). The PCR reaction, DNA is obtained in which the natural method was referred to as differential DNA denaturation hydrogen bonding rule is inversed. When PCR is PCR, or 3D-PCR for short (5). Obviously the converse is performed at limiting denaturation temperatures, not possible, that is selective amplification of GC-rich it is possible to recover GC-rich viral genomes and alleles with respect to a reference clone, because such inverted Alu elements embedded in cellular mRNAs alleles would melt at even higher temperatures. resulting from editing by dsRNA dependent host cell This not a moot point in virology for example, adenosine deaminases. The editing of Alu elements where there are examples of A!G hypermutated RNA in cellular mRNAs was strongly enhanced by type I viral genomes, the paradigm being measles virus (MV). interferon induction indicating a novel link mRNA Such genomes have been identified in autopsy samples metabolism and innate immunity. from cases of MV-associated subacute sclerosing panece- phalitis and inclusion body encephalitis (16). They arise from deamination of numerous adenosine residues INTRODUCTION in the context of double stranded RNA (dsRNA) by host cell adenosine deaminases of the ADAR family [for review It is a truism that a GC base pair has three hydrogen see (17)]. Editing of adenosine yields inosine (I). As I bonds while AT has two. In fact, Watson and Crick did hydrogen bonds essentially as guanosine (G), edited RNA not quite see it that way back in 1953 (1,2). It was Pauling sequences are recovered as G-rich alleles. The extent of and Corey who demonstrated the validity of the third editing may vary from a few bases to up to 50% of hydrogen bond in the GC pair in 1956 (3). The third potential target adenosine residues (18,19). hydrogen bond helps understand why GC-rich DNA Of the two ADAR1 gene transcripts ADAR-1L and melts at higher temperatures compared to AT-rich DNA. -1S, only the former can be induced by interferon a/b and Indeed, when performing PCR on GC-rich segments the g (20). Despite this, the number of examples of ADAR denaturation temperature is sometimes increased to ensure edited RNA viral sequences has remained little more than complete melting (4). a handful, being confined mainly to negative stranded Generally speaking, the denaturation temperature has viruses such as vesicular stomatitis virus, respiratory not been considered as a variable in PCR. Recently, lower denaturation temperatures were exploited to selectively syncytial virus and paramyxovirus (19,21,22) the signal *To whom correspondence should be addressed. Tel: +33 1 45 68 88 21; Fax: +33 1 45 68 88 74; Email: [email protected] 2008 The Author(s) This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/ by-nc/2.0/uk/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. e72 Nucleic Acids Research, 2008, Vol. 36, No. 12 PAGE 2 OF 10 O N A B HN O N N N N N N N N N N MC M O O N N dG:dC dA:dT O N HN N N N N N N N O O N N dD:dT dI:dC E 80.0 y= −0.16x + 79.5 r = 0.95 77.5 C D TCID TCGA 23 0 dF 75.0 dF y = 0.10x + 70.2 72.5 2 r = 0.99 70.0 Td=70.3 72.6°C Td=75.9 79.4°C 0 5 10 15 20 25 G/I-rich #subs. A/D-rich Figure 1. The basis of selective amplification of GC-rich alleles. (A) Base pairing of standard and DNA base pairs as well as those involving inosine (I) and diaminopurine (D). (B) Five of 8 commercially available thermostable polymerases can efficiently incorporate dITP and dDTP into DNA. 2+ dNTP concentrations were 200 mM throughout, [Mg ] = 2.5 mM, Td = 958C. C = negative buffer control, M = markers in bps. The input material was DNA corresponding to the reference sequence (34% GC) in Supplementary Figure 1. The Bioline, Eurobio, Promega, Qiagen and Takara enzymes are all variants of Taq polymerase. Takara I and II refer to two buffers supplied by the manufacturer. At 20 cycles the Bioline enzyme gave better product yield and was therefore used in all subsequent work. (C) SYBR Green melting profiles for TCGA DNA corresponding to the HIV-1 V1V2 region fragments. The reference is marked ‘0’ while ‘23’ denotes the clone differing uniquely by 23 G!A transitions. Midpoint Tds are given below the x-axis. (D) SYBR Green melting profiles for TCID DNA corresponding to the HIV-1 V1V2 region fragments. Midpoint Tds are given below the x-axis. (E) Linear correlations between midpoint Tds and G/I or A/D composition of 7 HIV-1 clones whose sequences are given in Supplementary Figure 1. exception being measles virus in vivo. The genome of of GC-rich alleles. In view of the 3:2 hydrogen bonding the hepatitis D satellite virus may also be edited by rule for GC and AT base pairs, differential denaturation ADAR-1L (23). of target DNA would appear to be out of the question. With the explosion of information on small cellular Yet the beginnings to a solution lie in ADAR editing itself. RNA molecules, it is recognized that many fold up Inosine base pairs with cytidine through two hydrogen into tight rod like structures (24,25). Some micro and bonds rather than the three typical for a GC base pair siRNAs undergo adenosine editing yielding the char- (Figure 1A). acteristic A!G transition when recovered as cloned Modified bases are often encountered in DNA bacteri- DNA (26–32). ophage genomes, usually as a means to avoid host Large numbers of Alu retroelements are found in genes restriction enzymes (43). Invariably modifications involve (33,34). When two are inserted in opposite orientations, cytidine or thymidine, for example 5-hydroxymethyl the inverted Alu RNAs hybridize forming long dsRNA cytidine in phage T4 DNA. There is however, just one duplexes, which are substrates for ADARs (35–39). While example of a modified purine, 2,6-diaminopurine (44), inverted Alus can be found in introns, they are generally or ‘D’. It is found in the cyanophage S-2L DNA genome embedded in the 3 non-coding region of the mRNAs. where it totally substitutes for adenosine and has the Through massive and labour intensive EST studies and singular feature of base pairing with thymine (T) via three bioinformatics comparisons with the human genome it is hydrogen bonds (Figure 1A). As dITP and dDTP are known that hundreds of human mRNAs undergo A!I commercially available, the outlines of a PCR based editing (35,38,39). method allowing selective amplification of GC-rich alleles Given the emerging importance of ADAR editing of becomes clear—a combination of differential denaturation a wide variety of RNAs (40–42), it would be useful to PCR using the modified bases dITP and dDTP. Does have a PCR based method to allow selective amplification it work? Bioline Eurobio Pfu Promega Pwo Qiagen Takara I Takara II Vent Td/°C PAGE 3 OF 10 Nucleic Acids Research, 2008, Vol. 36, No. 12 e72 MATERIALS AND METHODS amplification of PCR to generate sufficient material. Conditions were: 2.5 mM MgCl , 50 mM KCl, 10 mM Viruses Tris-HCl (pH 8.3), 200mM of dATP, dTTP, dCTP and MRC5 and Vero cells were grown in Dulbecco’s modified dGTP, 100mM each primer and 5 U of BioTaq DNA Eagle’s medium containing 5–10% fetal calf serum and polymerase (Bioline) in a final volume of 50ml. The second antibiotics (5 U/ml penicillin and 5 mg/ml streptomycin) reaction converted standard DNA to that containing the in the presence of 5% CO . Cell monolayers in 6-well 2 modified based D and I, referred to as TCID DNA. This plates were infected with live attenuated measles virus is essential because if input material is TCGA DNA, the (Schwarz strain amplified on Vero cells) at a multiplicity Tds of GC-rich alleles are governed by the natural base of infection of 0.1 for Vero cells and 3 for MRC-5. Two pairing rule and so cannot be differentially amplified. The days after infection culture medium was collected and cells conditions were as above except that 200mM each dTTP, were trypsinized. After clarification of cell debris, RNA dCTP, dDTP and dITP, 100mM each primer and 10 U of was extracted. Subconfluent monolayers were infected BioTaq DNA polymerase (Bioline) were used in a final with RVFV clone 13 at a multiplicity of infection of 0.01 volume of 50ml. The denaturation temperature was 958C. pfu per cell and incubated for 3 days at 378C. Differential amplification was performed in the third round by using an Eppendorf gradient Mastercycler S RNA extraction, oligonucleotides and PCR reagents programmed to generate 2–108C gradients in the denatur- and cloning ation temperature. The reaction parameters were per- formed by using, for example, a 88C denaturation gradient Samples including cell lysates and viral supernatants were for 5 min, followed by 35 cycles (a 88C denaturation digested in SDS/proteinase K buffer (0.1 mg/ml, Eurobio) gradient for 30 s, annealing 558C for 30 s and constant at 568C for 2 h. Total nucleic acids were extracted using polymerization temperature equal to the minimum dena- the MasterPure complete DNA and RNA purification kit turation gradient temperature for 1 min) and finally (Epicentre) according to the manufacturer’s procedure. 10 min at the minimum denaturation gradient temperature Total RNA was then reverse transcribed in a final volume to finish elongation. While the magnitude of the denatur- of 20ml of a mixture containing 1 buffer reaction ation gradient can be changed, the constant polymeriza- (Gibco), 300 ng of random hexamers (Pharmacia), tion temperature is always equal to the minimum 500mM each standard dNTP, 10 U of MLV reverse denaturation gradient temperature. The buffer conditions transcriptase (Invitrogen) and 10 U RNAsin (Promega). were 2.5 mM MgCl , 50 mM KCl, 10 mM Tris-HCl (pH Ten percent of the reaction was used for PCR 8.3), 200mM each dTTP, dCTP, dDTP and dITP, 100mM amplification. each primer and 10 U of BioTaq DNA polymerase A fragment of the M gene of MV and of the L gene (Bioline) in a final volume of 50ml. of RVFV clone 13 was amplified by a nested procedure. Increasing the concentration of dITP and dDTP to To increase sensitivity and specificity, a hot-start PCR was 300mM did not increase product yield (not shown). performed for both amplifications. First-round primers Although inosine base pairs essentially as guanosine, it for MV were 5ROUout and 3ROUout, respectively 5 GG can form base pairs with T and A, hence the use of dITP CAGGCYGGYGCCCCAGGYCAGAG and 5 GGRR in PCR is somewhat mutagenic. In an attempt to favorize CCTCTGCGGGGTRTCGRGCGG, and maps to 3522- dC:dITP pairing the concentration of dCTP was increased 3903 on the Schwarz genome. For the second round, from 200 to 300mM while the dTTP was lowered to primers were 5ROUin and 3ROUin, respectively 5 AGA 100mM and the fidelity compared to that resulting from YCCYGGYCYAGGCGACAGGAAGG and 5 GCR amplification using equimolar 200mM dNTPs. As no TTGCRCRCTTGGTTTGCGTTG, where Y=T/C and change in PCR fidelity was found (4.1 10 versus R=A/G. First-round primer for RVFV amplification were 3.9 10 per base), all subsequent amplifications were 5RFout and 3RFout, respectively 5 GTCGCCAATGY performed using equimolar dNTPs. CGAGGAGGCCCAYGA and 5 CTCCAGATCATCT RTCCTRRTGCTTCC, and map to 5872-6255 on the Amplification by 3DI-PCR of cellular mRNA L fragment of RVFV. For the second round, primers were embedded Alu sequences 5RFin and 3RFin, respectively 5 GATGATAGAAG AYGCCAAGAACAAYGC and 5 TGCTTCCTTCTGG Total RNA from infected and uninfected MRC5 cells was TCTCTGTRGRGTTC. extracted (Epicentre). cDNA synthesis was performed by Standard dNTPs were purchased from Sigma and using random priming as described above. 1/10 of the dDTP, dITP, dUTP, 5Me-dCTP were purchased from cDNA reaction was used for PCR amplification with TriLink. DAPI was from Fluka while 7-deazadGTP and primers Alu1 (5 CACGCCTGTAATCCCAGCACTTT the Hoechst bisbenzamide dye (H33258) were from Sigma. GGG) and Alu2 (5 TGTCGCCCAGGCTGGAGTGC PCR products were purified from agarose gels and ligated AGTGG). PCR conditions were 958C for 5 min followed into the TOPO TA cloning vector cloned and sequenced by 35 cycles with 958C for 30 s, 608C for 30 s and 728C as described (5). for 1 min and a final elongation step of 728C for 10 min. First PCR was performed with standard dNTPs (TCGA). PCR protocol 1/50 of the first PCR reaction was used for 3DI-PCR with Hypermutated genomes were identified by a three-step modified dNTPs (TCID) using a Td gradient from 84 protocol. The first reaction involved a standard to 608C for 5 min then 45 cycles with 60–848C for 45 s, e72 Nucleic Acids Research, 2008, Vol. 36, No. 12 PAGE 4 OF 10 608C for 45 s and 60–728C for 1 min. PCR products were molecular clones varied linearly with G/I or A/D content purified and cloned as described above. (Figure 1E). The temperature sensitivity of TCID DNA as a function of G/I content was only 60% that of Amplification of Ig Vk1 sequences from patients 2 and 3 TCGA DNA. We explored a variety of PCR conditions to try and CD14+ B-lymphocytes were purified from two splenecto- manipulate the denaturation sensitivity of TCID DNA. mized patients using a B-cell isolation Kit (Miltenyi Biotec) Despite trying a range of small organic molecules that and DNA extracted (Epicentre). DNA was amplified using bind to AT motifs via the minor groove, i.e. Hoechst primers Ig1 (5 GCGGACATCCAGATGACCCAGTCT) bisbenzmide dye H33258, modified bases such as dUTP, and Ig2 (5 GCGCTGTTGACAGTARTAAGTTGCA). 5-MedCTP and 7-deazadGTP, monovalent (K ) and Amplification conditions were: 958C for 5 min, then 35 2+ divalent cations (Mn ), none had any significant cycles with 958C for 30 s, 608C for 30 s and 728C for 1 min impact on the minimal denaturation temperature/base 1/50 of the PCR product was used for respectively 3D-PCR composition relationship of the seven standards (Figure 2 and 3DI-PCR. For 3D-PCR conditions were, 74–948C for and not shown). In short, while the overall Td can indeed 5 min then 74–948C for 1 min 558C for 30 s and 728C for be manipulated, the denaturation temperature/base com- 1 min for 35 cycles, and for 3DI-PCR conditions were, position relationship of TCID DNA is relatively refrac- 60–758C for 5 min followed by 60–758C for 30 s, 558C for tory to manipulation. 30 s and 60–758C for 1 min with 35 cycles and a final elongation step of 60–758C for 10 min 3D- and 3DI-PCR products were purified and cloned as described above. Recovery of in vitro hyperedited measles virus sequences We sought to validate the method using measles virus RESULTS (MV) samples grown in the interferon sensitive cell line MRC-5. As a control Vero cells were used which are A wide variety of thermostable DNA polymerases were defective for interferon-a and b production (45). The first screened for their ability to amplify DNA using attenuated MV Schwarz strain was used because it is a dTTP, dCTP, dITP and dDTP. Using a standard buffer good inducer of interferon (46). Two days post-infection anda958C denaturation temperature, five of eight supernatant and cell pellets were collected and total RNA thermostable polymerases resulted in reasonable product extracted. Complementary DNA was converted into PCR recovery after 30 cycles using an extended elongation time products, a fraction of which was converted into TCID of 1 min (Figure 1B). All five were commercial variants of PCR products using a 958C denaturation temperature. Taq polymerase. However, product recovery was 3-fold Selective amplification was then applied to the TCID compared to amplification using dGTP and dATP. DNA using a denaturation gradient of 63–728C. As can be The denaturation properties of PCR DNA containing seen from Figure 3A the minimum temperature at which the two modified bases (TCID DNA) were established for MV genomes were amplified from Vero cells was 67.48C. a series of seven 262 bp DNA fragments that differed only By contrast MV specific products were amplified from by up to 23 G!A transitions distributed across the locus the MRC-5 cells down to 658C. TCID products amplified (Supplementary Figure 1). As can be seen from SYBR at the lowest Td were used for molecular cloning into Green melting profiles, midpoint denaturation tempera- TOPO plasmids. Probably in view of the unusual bases, tures (Td) of 70.3 and 72.68C were obtained for TCID transformation of standard bacteria with cloned TCID DNA corresponding to the reference (0) and 23 base products not only gave very low efficiencies (<500-fold variant respectively, as anticipated from the change in lower than TCGA DNA) but also was invariably accom- hydrogen bonding patterns (Figure 1C). As expected for panied by large deletions within the MV sequences. To standard PCR products (i.e. TCGA DNA), the converse overcome this, a fraction of TCID PCR products was prevailed, i.e. the A-rich allele was denatured at a lower temperature, Td=75.98C, than the G-rich allele converted into standard DNA by 10 cycles of PCR using (79.48C, Figure 1D). The midpoint Tds of the seven normal dNTPs and then cloned. As controls, DNA 76 74 75 +100 mM KCl B C +100 µM + 50 µM 74 + 50 mM KCl T5meCID DNA + 10 µM H33258 73 72 TCID DNA TCID DNA TCID DNA 71 70 70 70 0 5 10 15 20 25 0 5 10 15 20 25 0 5 10 15 20 25 I-rich # subs. D-rich I-rich # subs. D-rich I-rich # subs. D-rich Figure 2. The temperature differential of TCID DNA is relatively refractory to manipulation. (A) Hoechst bisbenzmide dye H33258 that binds to AT-rich DNA via the minor groove increases the Td but not the temperature differential. (B) Substituting modified bases such as 5-MedCTP for dCTP, dUTP for dTTP and 7-deazadGTP for dGTP (not shown) also failed to increase the temperature differential of TCID DNA. (C) Increasing + 2+ the ionic strength by the addition of monovalent (K ) and divalent cations (Mn , not shown) failed to increase the temperature differential. Td°C Td°C Td°C PAGE 5 OF 10 Nucleic Acids Research, 2008, Vol. 36, No. 12 e72 Figure 3. 3DI-PCR amplification of ADAR edited measles virus genomes. (A) Agarose gel of TCID DNA amplified from measles infected Vero and MRC-5 cells. The PCR products amplified from the latter between 65 and 66.28C are indicative of genomes enriched in GC. C, negative buffer control; M molecular weight markers. (B) MV sequences derived from amplification at the lowest denaturation temperature (658C). Sequences are aligned to the reference MV sequence, only differences being shown. The monotonous A!G transitions are typical of ADAR editing. Complete sequence sets are given in Supplemental Figure 2. (C) Mutation matrices for the sequence sets. The number of sequences per matrix is, starting top left and going clockwise n=15, 13, 21 and 19. The symmetry of the matrices 958C amplification controls is typical of a viral quasispecies. A slight skew in the Vero/MV 67.48C matrix from AU!GC is understandable given that 3DI-PCR amplifies GC-rich sequences, and represents the GC-rich end of the mutant spectrum. (D) Sequences of two C-rich MV sequences compared to the reference genome. The first encodes A!G and U!C transitions and arises from editing of the viral genome and anti-genome, while the latter U!C transitions indicating editing only of the anti-genome. amplified from reactions using a Td=958C was also protocol was indeed capable of recovering GC-rich alleles. cloned and sequenced. Two cytidine-rich MV sequences compared to the reference As can be seen from Figures 3B and Supplementary genome were identified (Figure 3D). The first encoded Figure 2, the MV genomes selectively amplified from A!G and U!C transitions and probably arose from MRC-5 cells (Td=658C) were littered with A!G transi- editing of the viral genome and anti-genome, while the tions. Indeed, up to 83% of A residues could be edited latter sequence encoding only U!C transitions (mean=70%, range 3–83%). By contrast, those amplified was presumably derived from editing of mRNA or the from MV-infected Vero cells at the lowest possible anti-genome. temperature (Td = 67.48C) were typical of quasispecies To ascertain their frequency, the initial TCID products variation of an RNA virus. MV sequences amplified under were serially diluted and standard and selective PCR standard PCR conditions (Td = 958C, normal dNTPs) performed. The signal from standard PCR titrated out showed balanced mutation matrices (Figure 3C). This 100-fold further than selective PCR indicating that the indicates that the highly edited sequences from the MRC-5 highly edited genomes were present in the sample at 1% cell line must represent a subset, and that the selective PCR (data not shown). A!G hypermutants were also found in e72 Nucleic Acids Research, 2008, Vol. 36, No. 12 PAGE 6 OF 10 viral supernatants from MV-infected MRC-5 cells indicat- restrictive MRC-5 culture compared to the permissive ing that hyperedited genomes can be packaged and raises Vero cell culture, 66.38C compared to 67.28C (Figure 4A). the possibility that editing might continue within the Cloning and sequencing of the PCR products revealed virion (Supplementary Figure 3). extensive A!G editing of viral RNA from the MRC-5 We refer to this novel method as inverse differential culture and nothing more than a quasispecies variation DNA denaturation PCR, or 3DI-PCR, to distinguish from the Vero cells (Figure 4B and C). Although a handful it from 3D-PCR that allows amplification of AT-rich of hyperedited sequences are shown, all 26 clones derived DNA (5). from the MRC-5 infection were distinct and harbored between 63 and 77% of edited adenosine targets. Thus Genetic editing of a segmented RNA virus hyperediting of RVFV RNA in MRC-5 can be just as extensive as for MV in the same cell line. When analysed In order to see if 3DI-PCR could be applied to another viral by standard PCR (Td = 958C, normal dNTPs) the system and hence generate novel findings, we analysed mutation matrices were balanced, indicating that the Rift Valley fever virus (RVFV), a segmented negative highly edited RVFV genomes identified represent a stranded RNA virus. Currently, there are no reports of minority (Figure 4C). Complete RVFV sequence sets ADAR edited RVFV genomes. RVFV clone 13 is a highly can be found in Supplemental Figure 4. immunogenic, yet attenuated strain that encodes a 549 bp in frame deletion within the NSs gene. As the vestigial Selective amplification of edited Alu elements in mRNA NSs protein has lost its ability to antagonize interferon production, clone 13 is a good inducer of interferon, unlike By comparison of EST sequence libraries and genome virulent strains (47). While clone 13 grew well on Vero cells, sequences several reports have shown that inverted Alu viral titers were 100-fold lower on MRC-5 cells. elements embedded in cellular mRNAs can undergo Clone 13 was cultured on both cell lines for 3 days and ADAR editing (26,35,37,38,48). We decided to see if total cellular RNA recovered. Using primers specific for 3DI-PCR could supplant such powerful, yet brut force a 257 bp fragment from the L gene, 3DI-PCR could approaches. Randomly primed cDNA from MV infected recover RVFV genomes at a lower temperature from the MRC-5 cells was used, as there was prima face evidence Figure 4. Massive adenosine deamination of Rift valley fever virus genomes. RVFV clone 13 is a highly immunogenic yet attenuated strain that encodes a 549 bp in frame deletion within the NSs gene. As the vestigial NSs protein has lost its ability to antagonize interferon production, clone 13 is a good inducer of interferon, unlike virulent strains (47). While clone 13 grows well on Vero cells, viral titres were 100-fold lower on MRC-5 cells. (A) Agarose gel of TCID DNA amplified from RVFV infected Vero and MRC-5 cells. The PCR products amplified from the latter between 66.3 and 66.78C are indicative of genomes enriched in GC. C, negative buffer control; M molecular weight markers. (B) RVFV sequences derived from amplification at the lowest denaturation temperature (66.38C). Sequences are aligned to the reference MV sequence, only differences being shown. The monotonous A!G transitions are typical of ADAR editing. Complete sequence sets are given in Supplementary Figure 4. (C) Mutation matrices for the sequence sets. The number of sequences per matrix is, starting top left and going clockwise n = 6, 10, 11 and 26. The symmetry of the matrices 958C amplification controls is typical of a viral quasispecies. PAGE 7 OF 10 Nucleic Acids Research, 2008, Vol. 36, No. 12 e72 Figure 5. Selective amplification of ADAR-edited Alu elements nested in cellular mRNAs. (A) A versus G base composition (red and black) and U versus C (blue and green) of individual Alu sequences from cellular mRNAs from measles virus (MV) infected MRC-5 cells. Sequences were derived at standard (958C, n=77 sequences) and restrictive temperatures (63.88C, n=108 sequences). A versus G correlations pertain to plus sense 3 3 Alu sequences (%A=0.996%G+59.8; r=0.915, n=108, P=<10 ) and U versus C (%U=0.922%C+38.1; r=0.914, n=108, P< 10 ), the anti-sense. The two gradients indicate a strict interconversion of A to G and U to C. (B) A versus G and U versus C correlations for Alu sequences derived from cellular mRNA from uninfected MRC5 cells under normal (n=58 sequences) and 3DI-PCR (n=67 sequences). (C) Two examples of highly edited Alu elements (B04 and C02) nested within cellular mRNAs. The alignments were those of the best hits derived from blasting the human genome. Only differences are shown to emphasize the effects of ADAR editing. Gaps were introduced into the C02 sequence to allow comparison with the B04 alignment. The number of A!G transitions are shown to the right. of ADAR activity (Figure 3). Alu-specific primers were could be useful when applied to a complex problem, we chosen corresponding to a region that apparently under- took the example of somatic hypermutation of rearranged went little ADAR-editing (35,39,49). PCR products were immunoglobulin variable (V) genes. These loci are subject amplified at a temperature as low as 63.88C. As reflected to somatic hypermutation, initiated by genetic editing of by their A versus G (+strand) and U versus C (strand) ssDNA in transcription bubbles by activation induced base compositions (Figure 5A), there was considerable deaminase, AID (50). The process features an initial phase evidence of G and C enrichment compared to Alu targeting GC base pairs followed by a second targeting AT pairs. Primers were designed to amplify the Vk1 light sequences derived from amplification using a 958C dena- chain DNA from CD14 positive splenic B cells isolated turation temperature. Indeed the majority of Alu from two patients with follicular hyperplasia who had sequences were enriched in either G or C (Figure 5A). undergone splenectomy due to untreatable thrombocyto- Blast analyses showed that the majority of selectively penia (51). As can be seen from Figure 6A and B, 3DI- amplified Alu elements had undergone ADAR-editing PCR failed to amplify highly GC-rich alleles over and (Supplementary Table 1A and B), two of which are shown above that found by PCR using a denaturation tempera- in Figure 4C. In fact ADAR-1L is induced by type I ture of 958C (Supplementary Tables 3A and C, 4A and C). interferons and was detectable by RT-PCR among MV- By contrast its counterpart, 3D-PCR, recovered a series infected MRC-5 mRNAs and not from uninfected cells of AT-rich variants (Figure 6A, Supplementary Tables 3B (not shown). Hence, the majority of edited Alu sequences and 4B). can be ascribed to ADAR-1L. When the same 3DI-PCR When blasted against the human genome, seven protocol was applied to Alu-containing mRNAs from sequences, one of which is shown in Figure 6C, showed uninfected MRC-5 cells, there were no comet tails out to an excess of GC!AT transitions in both strands, most of >40%G or >27%C as noted for MV-infected MRC-5 which were concentrated in the two complementary deter- cells (Figure 5B, Supplementary Table 1A and B). mining regions, CDR1 and CDR2 (Figure 6D, Supple- mentary Figure 5). As the nucleotide context surrounding GC and AT-rich rearranged immunoglobulin V regions the GC!AT transitions is AGCT (Figure 6E), highly Clearly 3DI-PCR is a robust method and complementary analogous to WRC motif (W = A,T; R = A,G where Cis to its sister, 3D-PCR capable of amplifying up AT-rich the edited nucleotide) observed for AID on single stranded alleles. To see if a combination of the two techniques DNA in vitro (52), and identical to that for AID e72 Nucleic Acids Research, 2008, Vol. 36, No. 12 PAGE 8 OF 10 A B %AT Patient 2 %AT Patient 3 60 60 [email protected]°C [email protected]°C [email protected]°C [email protected]°C [email protected]°C [email protected]°C 45 45 40 40 40 45 50 55 60%GC 40 45 50 55 60%GC CDR1 V T I T C R A S Q S I S S Y L N W Y Q Q K P GTC ACC ATC ACT TGC CGG GCA AGT CAG AGC ATT AGC AGC TAT TTA AAT TGG TAT CAG CAG AAA CCA ... ... ... ... ... ... ... .A. ... ..T ... ..T ..T .T. ... ... ... ... ... ... ... ... CDR2 G K A P K L L I Y A A S S L Q S G V P S R F GGG AAA GCC CCT AAG CTC CTG ATC TAT GCT GCA TCC AGT TTG CAA AGT GGG GTC CCA TCA AGG TTC ... ... ... ... ... ... ... ... ... AT. ... ... ... ... ... ... ... ... ... ... ... ... S G S G S G T D F T L T I AGT GGC AGT GGA TCT GGG ACA GAT TTC ACT CTC ACC ATC ... ... ... ... ... ... ... ... ... ... ... .T. ... FR1 CDR1 FR2 CDR2 FR3 + − ns Patient 2 A05 3 0 4 1 0 1 1 5 3 A2 0 4 0 2 1 5 2 4 3 C06 0 4 0 0 0 1 3 2 2 C07 1 1 0 2 0 0 1 2 2 E07 0 4 2 0 1 3 4 5 2 H10 1 0 0 4 0 2 3 3 2 Patient 3 C07 0 2 1 2 1 2 4 4 2 Σ 2 18 6 8 4 15 23 23 normalized data raw data, n=38 −2 −1 C +1 −2 −1 C +1 T 13 4 - 19 T 1.2 0.4 - 1.4 C 3 4 38 6 C 0.4 0.4 1.0 0.6 G 4 24 - 1 0.5 2.2 - 1.4 A 18 6 - 12 A 1.6 0.6 - 0.9 Con. C T A G Figure 6. Identification of GC and AT-rich rearranged immunoglobulin V region sequences from CD14-purified human B cells. (A and B) A+T versus G+C correlations of individual V region sequences derived by PCR (black), 3D-PCR (green) and 3DI-PCR (red) from patients 2 and 3. (C) Alignment of best Blast match to the human genome for the A2 sequence from patient 2. See Supplemental Tables 3 and 4 for complete analyses. (D) The majority of C!T transitions (68%) map to the CDR2 and CDR3 regions even though they comprise only 22.5% of the sequence.  refers to C!T transitions on the sense and antisense strands. Ns/s refer to non-synonymous and synonymous transitions respectively. (E) Sequence context of C!T transitions. The raw data is given for the 38 C!T transitions and normalized to that for the target sequences. For example, the normalized frequency of T at position –2 is the quotient of (12/38)/(32/110), the latter fraction comprises the sense and anti-sense sequences weighted by the fraction of mutations in the sense (+) and anti-sense strands () (Figure 6D). The frequency of G at position +1 is very low due to the low CpG content of the human genome. The quotient of 1.4 for G at position +1 is not robust as it reflects the quotient of two small fractions, notably (1/38)/(2/110). mutational hotspots (50), it is plausible that these bond, and allows selective amplification of AT-rich DNA sequences represent examples of the initial step in the (5). By using modified bases the 3:2 rule can be inversed, complex process of somatic hypermutation. allowing selective amplification of GC-rich alleles (Figure 1A, C and D). A range of commercially available Taq polymerases was able to undertake incorporation DISCUSSION of the two modified bases, although product yields are Differential DNA denaturation PCR exploits the intrinsic somewhat less than when standard dNTPs are used. The stability of GC base pairs arising from a third hydrogen magnitude of the temperature/GC content coefficients PAGE 9 OF 10 Nucleic Acids Research, 2008, Vol. 36, No. 12 e72 for 3D- and 3DI-PCR were not equivalent, the latter being modified nucleotides, PCR can now be extended to allow selective amplification of GC-rich DNA. 60% less than the former (Figure 1D). When applied to measles virus, the prototype for ADAR edited viral genomes, there was no difficulty in SUPPLEMENTARY DATA recovering highly edited genomes from the MRC-5 culture (Figure 3). Not only are the MV genomes more extensively Supplementary Data are available at NAR Online. edited from cultured virus than in vivo, they are more heterogeneous (19). The degree of editing observed here is unprecedented; typically ADAR-edited genomes rarely ACKNOWLEDGEMENTS contained more than 50% of edited adenosines (53). We would like to thank Chantal Combredet for the Among the present sequences sets the upper limits were measles virus cultures. This work was supported by grants 77 and 83% for RVFV and MV respectively. Although from the Pasteur Institute. R.S. was a recipient of a these RNA sequences can form secondary structures as Boehringer-Ingelheim Fonds Fellowship. Funding to pay shown by computer programs such as M-fold, never were the Open Access publication charges for this article was 80% of adenosine residues sequestered in dsRNA. provided by Institut Pasteur. That such genomes were present at frequencies of 1% Conflict of interest statement. None declared. in the MRC-5 culture may help explain why MV A!G hypermutants have not been described before in culture. The finding of numerous A!G hypermutants in culture REFERENCES of RVFV clone 13 is also novel and suggests that similar findings could be obtained with most RNA viruses if 1. Watson,J.D. and Crick,F.H. (1953) Genetical implications of the grown on interferon sensitive cells. structure of deoxyribonucleic acid. Nature, 171, 964–967. 2. Wain-Hobson,S. (2006) The third Bond. Nature, 439, 539. Why would interferon-induced ADAR-1L target 3. Corey,R.B. and Pauling,L. (1956) Specific hydrogen-bond ‘only’ 1% of genomes? The MV sequence sets shown in formation between pyramidines and purines in deoxyribonucleic Figure 3B were obtained at the lowest positive Td, i.e. acids. Arch. Biochem. Biophys., 65, 164–181. 658C. While not shown here, we know that MV sequences 4. Smith,S.M., Markham,R.B. and Jeang,K.T. (1996) Conditional reduction of human immunodeficiency virus type 1 replication taken from the Td=66.28C sample were less extensively by a gain-of-herpes simplex virus 1 thymidine kinase function. substituted suggesting that there is a large range in the Proc. Natl Acad. Sci. USA, 93, 7955–7960. degree of editing, probably reflecting varying levels of 5. Suspene,R., Henry,M., Guillot,S., Wain-Hobson,S. and ADAR-1L expression in individual cells. If larger Vartanian,J.P. (2005) Recovery of APOBEC3-edited human immunodeficiency virus G!A hypermutants by differential DNA segments were analysed the proportion of lightly edited denaturation PCR. J. Gen. Virol., 86, 125–129. sequences would increase. Hence the true number of 6. Harris,R.S., Bishop,K.N., Sheehy,A.M., Craig,H.M., edited MV genomes is probably >1%. As the genomic Petersen-Mahrt,S.K., Watt,I.N., Neuberger,M.S. and Malim,M.H. mutation rate for MV [1.4 substitutions per cycle (54)] (2003) DNA deamination mediates innate immunity to retroviral is close to the error threshold for RNA viruses, a little infection. Cell, 113, 803–809. 7. Lecossier,D., Bouchonnet,F., Clavel,F. and Hance,A.J. (2003) adenosine deamination should be sufficient to kill the Hypermutation of HIV-1 DNA in the absence of the Vif protein. virus (55). Science, 300, 1112. While the fate of ADAR-edited mRNAs is debated, 8. Mangeat,B., Turelli,P., Caron,G., Friedli,M., Perrin,L. and it does appear that it is linked to mRNA turnover (53). Trono,D. (2003) Broad antiretroviral defence by human APOBEC3G through lethal editing of nascent reverse transcripts. The finding that ADAR-editing of cellular mRNAs Nature, 424, 99–103. encoding inverted Alu elements is increased upon inter- 9. Mariani,R., Chen,D., Schrofelbauer,B., Navarro,F., Konig,R., feron induction shows that these dsRNA structures are Bollman,B., Munk,C., Nymark-McMahon,H. and Landau,N.R. relatively unprotected by protein (Figure 5). If interferon (2003) Species-specific exclusion of APOBEC3G from HIV-1 virions by Vif. Cell, 114, 21–31. can impinge on the metabolism of several hundreds of 10. Suspe` ne,R., Sommer,P., Henry,M., Ferris,S., Guetard,D., Pochet,S., mRNAs, then perhaps it might contribute to IFN-induced Chester,A., Navaratnam,N., Wain-Hobson,S. and Vartanian,J.P. cell death. (2004) APOBEC3G is a single-stranded DNA cytidine deaminase A combination of both PCR methods can be applied and functions independently of HIV reverse transcriptase. Nucleic Acids Res., 32, 2421–2429. to complex sets of sequences as highlighted by the 11. Wiegand,H.L., Doehle,B.P., Bogerd,H.P. and Cullen,B.R. (2004) edited human immunoglobulin genes (Figure 6). They A second human antiretroviral factor, APOBEC3F, is suppressed could improve the resolution of metagenomic analyses of by the HIV-1 and HIV-2 Vif proteins. EMBO J., 23, 2451–2458. bacterial genomes that vary greatly in GC content. As 12. Pathak,V.K. and Temin,H.M. (1990) Broad spectrum of in vivo forward mutations, hypermutations, and mutational hotspots in a they are PCR based they can identify low frequency retroviral shuttle vector after a single replication cycle: substitu- components that might otherwise escaped identification. tions, frameshifts, and hypermutations. Proc. Natl Acad. Sci. USA, 3DI-PCR is robust and simple to perform, dDTP and 87, 6019–6023. dITP being commercially available reagents. It is a trifle 13. Vartanian,J.P., Meyerhans,A., Asjo,B. and Wain-Hobson,S. (1991) longer in that extra PCR steps are necessary to perform Selection, recombination, and G!A hypermutation of human immunodeficiency virus type 1 genomes. J. Virol., 65, 1779–1788. the selective amplification as well as to obtain reason- 14. Suspe` ne,R., Gue´ tard,D., Henry,M., Sommer,P., Wain-Hobson,S. able cloning efficiencies. The PCR denaturation tempera- and Vartanian,J.P. (2005) Extensive editing of both hepatitis B virus ture has hitherto remained a constant, understandably DNA strands by APOBEC3 cytidine deaminases in vitro and so as the aim was to denature all DNA. With the use of in vivo. Proc. Natl Acad. Sci. USA, 102, 8321–8326. e72 Nucleic Acids Research, 2008, Vol. 36, No. 12 PAGE 10 OF 10 15. Mahieux,R., Suspe` ne,R., Delebecque,F., Henry,M., Schwartz,O., 35. Athanasiadis,A., Rich,A. and Maas,S. (2004) Widespread A-to-I Wain-Hobson,S. and Vartanian,J.P. (2005) Extensive editing of a RNA editing of Alu-containing mRNAs in the human transcrip- small fraction of human T-cell leukemia virus type 1 genomes by tome. PLoS Biol., 2, e391. four APOBEC3 cytidine deaminases. J. Gen. Virol., 86, 2489–2494. 36. Blow,M., Futreal,P.A., Wooster,R. and Stratton,M.R. (2004) A 16. Schmid,A., Spielhofer,P., Cattaneo,R., Baczko,K., ter Meulen,V. survey of RNA editing in human brain. Genome Res., 14, and Billeter,M.A. (1992) Subacute sclerosing panencephalitis 2379–2387. is typically characterized by alterations in the fusion protein 37. Eisenberg,E., Nemzer,S., Kinar,Y., Sorek,R., Rechavi,G. and cytoplasmic domain of the persisting measles virus. Virology, 188, Levanon,E.Y. (2005) Is abundant A-to-I RNA editing primate- 910–915. specific? Trends Genet., 21, 77–81. 17. Valente,L. and Nishikura,K. (2005) ADAR gene family and A-to-I 38. Kim,D.D., Kim,T.T., Walsh,T., Kobayashi,Y., Matise,T.C., RNA editing: diverse roles in posttranscriptional gene regulation. Buyske,S. and Gabriel,A. (2004) Widespread RNA editing of Prog. Nucleic Acid Res. Mol. Biol., 79, 299–338. embedded alu elements in the human transcriptome. Genome Res., 18. Cattaneo,R., Schmid,A., Eschle,D., Baczko,K., ter Meulen,V. and 14, 1719–1725. Billeter,M.A. (1988) Biased hypermutation and other genetic 39. Levanon,E.Y., Eisenberg,E., Yelin,R., Nemzer,S., Hallegger,M., changes in defective measles viruses in human brain infections. Cell, Shemesh,R., Fligelman,Z.Y., Shoshan,A., Pollock,S.R., Sztybel,D. 55, 255–265. et al. (2004) Systematic identification of abundant A-to-I editing 19. Bass,B.L., Weintraub,H., Cattaneo,R. and Billeter,M.A. (1989) sites in the human transcriptome. Nat. Biotechnol., 22, 1001–1005. Biased hypermutation of viral RNA genomes could be due to 40. DeCerbo,J. and Carmichael,G.G. (2005) Retention and repression: unwinding/modification of double-stranded RNA. Cell, 56, 331. fates of hyperedited RNAs in the nucleus. Curr. Opin. Cell. Biol., 20. Samuel,C.E. (2001) Antiviral actions of interferons. Clin. Microbiol. 17, 302–308. Rev., 14, 778–809. 41. Jepson,J.E. and Reenan,R.A. (2007) RNA editing in regulating gene 21. O’Hara,P.J., Nichol,S.T., Horodyski,F.M. and Holland,J.J. (1984) expression in the brain. Biochim. Biophys. Acta. On line 3 Dec 2007. Vesicular stomatitis virus defective interfering particles can contain 42. Wang,Q., Zhang,Z., Blackwell,K. and Carmichael,G.G. (2005) extensive genomic sequence rearrangements and base substitutions. Vigilins bind to promiscuously A-to-I-edited RNAs and are Cell, 36, 915–924. involved in the formation of heterochromatin. Curr. Biol., 15, 22. Rueda,P., Garcia-Barreno,B. and Melero,J.A. (1994) Loss of 384–391. conserved cysteine residues in the attachment (G) glycoprotein of 43. Gommers-Ampt,J.H. and Borst,P. (1995) Hypermodified bases in two human respiratory syncytial virus escape mutants that contain DNA. FASEB J., 9, 1034–1042. multiple A-G substitutions (hypermutations). Virology, 198, 44. Kirnos,M.D., Khudyakov,I.Y., Alexandrushkina,N.I. and 653–662. Vanyushin,B.F. (1977) 2-aminoadenine is an adenine substituting 23. Chang,J., Gudima,S.O. and Taylor,J.M. (2005) Evolution of for a base in S-2L cyanophage DNA. Nature, 270, 369–370. hepatitis delta virus RNA genome following long-term replication 45. Emeny,J.M. and Morgan,M.J. (1979) Regulation of the interferon in cell culture. J. Virol., 79, 13310–13316. system: evidence that Vero cells have a genetic defect in interferon 24. Birney,E., Stamatoyannopoulos,J.A., Dutta,A., Guigo,R., production. J. Gen. Virol., 43, 247–252. Gingeras,T.R., Margulies,E.H., Weng,Z., Snyder,M., 46. Combredet,C., Labrousse,V., Mollet,L., Lorin,C., Delebecque,F., Dermitzakis,E.T., Thurman,R.E. et al. (2007) Identification and Hurtrel,B., McClure,H., Feinberg,M.B., Brahic,M. and Tangy,F. analysis of functional elements in 1% of the human genome by the (2003) A molecularly cloned Schwarz strain of measles virus vaccine ENCODE pilot project. Nature, 447, 799–816. induces strong immune responses in macaques and transgenic mice. 25. Washietl,S., Pedersen,J.S., Korbel,J.O., Stocsits,C., Gruber,A.R., J. Virol., 77, 11546–11554. Hackermuller,J., Hertel,J., Lindemeyer,M., Reiche,K., Tanzer,A. 47. Billecocq,A., Spiegel,M., Vialat,P., Kohl,A., Weber,F., Bouloy,M. et al. (2007) Structured RNAs in the ENCODE selected regions of and Haller,O. (2004) NSs protein of Rift Valley fever virus blocks the human genome. Genome Res., 17, 852–864. interferon production by inhibiting host gene transcription. 26. Blow,M.J., Grocock,R.J., van Dongen,S., Enright,A.J., Dicks,E., J. Virol., 78, 9798–9806. Futreal,P.A., Wooster,R. and Stratton,M.R. (2006) RNA editing of 48. Levanon,E.Y., Hallegger,M., Kinar,Y., Shemesh,R., human microRNAs. Genome Biol., 7, R27. Djinovic-Carugo,K., Rechavi,G., Jantsch,M.F. and Eisenberg,E. 27. Kawahara,Y., Zinshteyn,B., Chendrimada,T.P., Shiekhattar,R. and (2005) Evolutionarily conserved human targets of adenosine to Nishikura,K. (2007) RNA editing of the microRNA-151 precursor inosine RNA editing. Nucleic Acids Res., 33, 1162–1168. blocks cleavage by the Dicer-TRBP complex. EMBO Rep., 8, 49. Nishikura,K. (2006) Editor meets silencer: crosstalk between RNA 763–769. editing and RNA interference. Nat. Rev. Mol. Cell. Biol., 7, 28. Kawahara,Y., Zinshteyn,B., Sethupathy,P., Iizasa,H., 919–931. Hatzigeorgiou,A.G. and Nishikura,K. (2007) Redirection of silen- 50. Di Noia,J.M. and Neuberger,M.S. (2007) Molecular mechanisms of cing targets by adenosine-to-inosine editing of miRNAs. Science, antibody somatic hypermutation. Annu. Rev. Biochem., 76, 1–22. 315, 1137–1140. 51. Cheynier,R., Henrichwark,S., Hadida,F., Pelletier,E., 29. Knight,S.W. and Bass,B.L. (2002) The role of RNA editing by Oksenhendler,E., Autran,B. and Wain-Hobson,S. (1994) HIV and ADARs in RNAi. Mol. Cell, 10, 809–817. T cell expansion in splenic white pulps is accompanied by 30. Luciano,D.J., Mirsky,H., Vendetti,N.J. and Maas,S. (2004) RNA infiltration of HIV-specific cytotoxic T lymphocytes. Cell, 78, editing of a miRNA precursor. RNA, 10, 1174–1177. 373–387. 31. Scadden,A.D. and Smith,C.W. (2001) RNAi is antagonized by 52. Pham,P., Bransteitter,R., Petruska,J. and Goodman,M.F. (2003) A!I hyper-editing. EMBO Rep., 2, 1107–1111. Processive AID-catalysed cytosine deamination on single-stranded 32. Yang,W., Chendrimada,T.P., Wang,Q., Higuchi,M., Seeburg,P.H., DNA simulates somatic hypermutation. Nature, 424, 103–107. Shiekhattar,R. and Nishikura,K. (2006) Modulation of microRNA 53. Scadden,A.D. (2005) The RISC subunit Tudor-SN binds to hyper- processing and expression through RNA editing by ADAR edited double-stranded RNA and promotes its cleavage. deaminases. Nat. Struct. Mol. Biol., 13, 13–21. Nat. Struct. Mol. Biol., 12, 489–496. 33. Pace,J.K. II and Feschotte,C. (2007) The evolutionary history of 54. Schrag,S.J., Rota,P.A. and Bellini,W.J. (1999) Spontaneous muta- human DNA transposons: evidence for intense activity in the tion rate of measles virus: direct estimation based on mutations primate lineage. Genome Res., 17, 422–432. conferring monoclonal antibody resistance. J. Virol., 73, 51–54. 34. Shen,M.R., Batzer,M.A. and Deininger,P.L. (1991) Evolution of the 55. Biebricher,C.K. and Eigen,M. (2005) The error threshold. master Alu gene(s). J. Mol. Evol., 33, 311–320. Virus Res., 107, 117–127. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Nucleic Acids Research Oxford University Press

Inversing the natural hydrogen bonding rule to selectively amplify GC-rich ADAR-edited RNAs

Download PDF
10 pages

Loading next page...
 
/lp/oxford-university-press/inversing-the-natural-hydrogen-bonding-rule-to-selectively-amplify-gc-lCAZ6j9227

References (55)

Publisher
Oxford University Press
Copyright
© 2008 The Author(s)
ISSN
0305-1048
eISSN
1362-4962
DOI
10.1093/nar/gkn295
pmid
18515351
Publisher site
See Article on Publisher Site

Abstract

Published online 30 May 2008 Nucleic Acids Research, 2008, Vol. 36, No. 12 e72 doi:10.1093/nar/gkn295 Inversing the natural hydrogen bonding rule to selectively amplify GC-rich ADAR-edited RNAs 1,2 1,2 1,2 1,2 ` ´ Rodolphe Suspene , Myrtille Renard , Michel Henry , Denise Guetard , 1,2 3 3 ` ` David Puyraimond-Zemmour , Agnes Billecocq , Michele Bouloy , 2,4 1,2 1,2, Frederic Tangy , Jean-Pierre Vartanian and Simon Wain-Hobson * 1 2 3 Unite´ de Re´ trovirologie Mole´ culaire, CNRS URA 3015, Unite´ de Ge´ ne´ tique Mole´ culaire des Bunyaviride´ s and Laboratoire de Ge´ nomique Virale et Vaccination, Institut Pasteur, 28 rue du Dr Roux, 75724 Paris cedex 15 Received January 28, 2008; Revised April 25, 2008; Accepted April 29, 2008 amplify so-called G!A hypermutants of the human ABSTRACT immunodeficiency virus (HIV) (5). They arise from genetic DNA complementarity is expressed by way of editing of nascent viral cDNA by two host cell cytidine three hydrogen bonds for a G:C base pair and two deaminases of the APOBEC3 family (6–11). Deamination for A:T. As a result, careful control of the denatura- of numerous cytidine (C) residues on the viral minus tion temperature of PCR allows selective amplifica- strand yields multiple uracil (U) residues, which are copied tion of AT-rich alleles. Yet for the same reason, as a thymidine (T). With respect to the viral plus strand the converse is not possible, selective amplification as reference, these show up as genomes with numerous of GC-rich alleles. Inosine (I) hydrogen bonds to G!A transitions giving rise to the term G!A hyper- mutants (12,13). Temperature differences as small as cytosine by two hydrogen bonds while diaminopu- 1–28C were enough to allow differential amplification of rine (D) forms three hydrogen bonds with thymine. A rich hypermutants in the presence of as much as 10 fold By substituting dATP by dDTP and dGTP by dITP in a excess of wild type, or reference genomes (14,15). The PCR reaction, DNA is obtained in which the natural method was referred to as differential DNA denaturation hydrogen bonding rule is inversed. When PCR is PCR, or 3D-PCR for short (5). Obviously the converse is performed at limiting denaturation temperatures, not possible, that is selective amplification of GC-rich it is possible to recover GC-rich viral genomes and alleles with respect to a reference clone, because such inverted Alu elements embedded in cellular mRNAs alleles would melt at even higher temperatures. resulting from editing by dsRNA dependent host cell This not a moot point in virology for example, adenosine deaminases. The editing of Alu elements where there are examples of A!G hypermutated RNA in cellular mRNAs was strongly enhanced by type I viral genomes, the paradigm being measles virus (MV). interferon induction indicating a novel link mRNA Such genomes have been identified in autopsy samples metabolism and innate immunity. from cases of MV-associated subacute sclerosing panece- phalitis and inclusion body encephalitis (16). They arise from deamination of numerous adenosine residues INTRODUCTION in the context of double stranded RNA (dsRNA) by host cell adenosine deaminases of the ADAR family [for review It is a truism that a GC base pair has three hydrogen see (17)]. Editing of adenosine yields inosine (I). As I bonds while AT has two. In fact, Watson and Crick did hydrogen bonds essentially as guanosine (G), edited RNA not quite see it that way back in 1953 (1,2). It was Pauling sequences are recovered as G-rich alleles. The extent of and Corey who demonstrated the validity of the third editing may vary from a few bases to up to 50% of hydrogen bond in the GC pair in 1956 (3). The third potential target adenosine residues (18,19). hydrogen bond helps understand why GC-rich DNA Of the two ADAR1 gene transcripts ADAR-1L and melts at higher temperatures compared to AT-rich DNA. -1S, only the former can be induced by interferon a/b and Indeed, when performing PCR on GC-rich segments the g (20). Despite this, the number of examples of ADAR denaturation temperature is sometimes increased to ensure edited RNA viral sequences has remained little more than complete melting (4). a handful, being confined mainly to negative stranded Generally speaking, the denaturation temperature has viruses such as vesicular stomatitis virus, respiratory not been considered as a variable in PCR. Recently, lower denaturation temperatures were exploited to selectively syncytial virus and paramyxovirus (19,21,22) the signal *To whom correspondence should be addressed. Tel: +33 1 45 68 88 21; Fax: +33 1 45 68 88 74; Email: [email protected] 2008 The Author(s) This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/ by-nc/2.0/uk/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. e72 Nucleic Acids Research, 2008, Vol. 36, No. 12 PAGE 2 OF 10 O N A B HN O N N N N N N N N N N MC M O O N N dG:dC dA:dT O N HN N N N N N N N O O N N dD:dT dI:dC E 80.0 y= −0.16x + 79.5 r = 0.95 77.5 C D TCID TCGA 23 0 dF 75.0 dF y = 0.10x + 70.2 72.5 2 r = 0.99 70.0 Td=70.3 72.6°C Td=75.9 79.4°C 0 5 10 15 20 25 G/I-rich #subs. A/D-rich Figure 1. The basis of selective amplification of GC-rich alleles. (A) Base pairing of standard and DNA base pairs as well as those involving inosine (I) and diaminopurine (D). (B) Five of 8 commercially available thermostable polymerases can efficiently incorporate dITP and dDTP into DNA. 2+ dNTP concentrations were 200 mM throughout, [Mg ] = 2.5 mM, Td = 958C. C = negative buffer control, M = markers in bps. The input material was DNA corresponding to the reference sequence (34% GC) in Supplementary Figure 1. The Bioline, Eurobio, Promega, Qiagen and Takara enzymes are all variants of Taq polymerase. Takara I and II refer to two buffers supplied by the manufacturer. At 20 cycles the Bioline enzyme gave better product yield and was therefore used in all subsequent work. (C) SYBR Green melting profiles for TCGA DNA corresponding to the HIV-1 V1V2 region fragments. The reference is marked ‘0’ while ‘23’ denotes the clone differing uniquely by 23 G!A transitions. Midpoint Tds are given below the x-axis. (D) SYBR Green melting profiles for TCID DNA corresponding to the HIV-1 V1V2 region fragments. Midpoint Tds are given below the x-axis. (E) Linear correlations between midpoint Tds and G/I or A/D composition of 7 HIV-1 clones whose sequences are given in Supplementary Figure 1. exception being measles virus in vivo. The genome of of GC-rich alleles. In view of the 3:2 hydrogen bonding the hepatitis D satellite virus may also be edited by rule for GC and AT base pairs, differential denaturation ADAR-1L (23). of target DNA would appear to be out of the question. With the explosion of information on small cellular Yet the beginnings to a solution lie in ADAR editing itself. RNA molecules, it is recognized that many fold up Inosine base pairs with cytidine through two hydrogen into tight rod like structures (24,25). Some micro and bonds rather than the three typical for a GC base pair siRNAs undergo adenosine editing yielding the char- (Figure 1A). acteristic A!G transition when recovered as cloned Modified bases are often encountered in DNA bacteri- DNA (26–32). ophage genomes, usually as a means to avoid host Large numbers of Alu retroelements are found in genes restriction enzymes (43). Invariably modifications involve (33,34). When two are inserted in opposite orientations, cytidine or thymidine, for example 5-hydroxymethyl the inverted Alu RNAs hybridize forming long dsRNA cytidine in phage T4 DNA. There is however, just one duplexes, which are substrates for ADARs (35–39). While example of a modified purine, 2,6-diaminopurine (44), inverted Alus can be found in introns, they are generally or ‘D’. It is found in the cyanophage S-2L DNA genome embedded in the 3 non-coding region of the mRNAs. where it totally substitutes for adenosine and has the Through massive and labour intensive EST studies and singular feature of base pairing with thymine (T) via three bioinformatics comparisons with the human genome it is hydrogen bonds (Figure 1A). As dITP and dDTP are known that hundreds of human mRNAs undergo A!I commercially available, the outlines of a PCR based editing (35,38,39). method allowing selective amplification of GC-rich alleles Given the emerging importance of ADAR editing of becomes clear—a combination of differential denaturation a wide variety of RNAs (40–42), it would be useful to PCR using the modified bases dITP and dDTP. Does have a PCR based method to allow selective amplification it work? Bioline Eurobio Pfu Promega Pwo Qiagen Takara I Takara II Vent Td/°C PAGE 3 OF 10 Nucleic Acids Research, 2008, Vol. 36, No. 12 e72 MATERIALS AND METHODS amplification of PCR to generate sufficient material. Conditions were: 2.5 mM MgCl , 50 mM KCl, 10 mM Viruses Tris-HCl (pH 8.3), 200mM of dATP, dTTP, dCTP and MRC5 and Vero cells were grown in Dulbecco’s modified dGTP, 100mM each primer and 5 U of BioTaq DNA Eagle’s medium containing 5–10% fetal calf serum and polymerase (Bioline) in a final volume of 50ml. The second antibiotics (5 U/ml penicillin and 5 mg/ml streptomycin) reaction converted standard DNA to that containing the in the presence of 5% CO . Cell monolayers in 6-well 2 modified based D and I, referred to as TCID DNA. This plates were infected with live attenuated measles virus is essential because if input material is TCGA DNA, the (Schwarz strain amplified on Vero cells) at a multiplicity Tds of GC-rich alleles are governed by the natural base of infection of 0.1 for Vero cells and 3 for MRC-5. Two pairing rule and so cannot be differentially amplified. The days after infection culture medium was collected and cells conditions were as above except that 200mM each dTTP, were trypsinized. After clarification of cell debris, RNA dCTP, dDTP and dITP, 100mM each primer and 10 U of was extracted. Subconfluent monolayers were infected BioTaq DNA polymerase (Bioline) were used in a final with RVFV clone 13 at a multiplicity of infection of 0.01 volume of 50ml. The denaturation temperature was 958C. pfu per cell and incubated for 3 days at 378C. Differential amplification was performed in the third round by using an Eppendorf gradient Mastercycler S RNA extraction, oligonucleotides and PCR reagents programmed to generate 2–108C gradients in the denatur- and cloning ation temperature. The reaction parameters were per- formed by using, for example, a 88C denaturation gradient Samples including cell lysates and viral supernatants were for 5 min, followed by 35 cycles (a 88C denaturation digested in SDS/proteinase K buffer (0.1 mg/ml, Eurobio) gradient for 30 s, annealing 558C for 30 s and constant at 568C for 2 h. Total nucleic acids were extracted using polymerization temperature equal to the minimum dena- the MasterPure complete DNA and RNA purification kit turation gradient temperature for 1 min) and finally (Epicentre) according to the manufacturer’s procedure. 10 min at the minimum denaturation gradient temperature Total RNA was then reverse transcribed in a final volume to finish elongation. While the magnitude of the denatur- of 20ml of a mixture containing 1 buffer reaction ation gradient can be changed, the constant polymeriza- (Gibco), 300 ng of random hexamers (Pharmacia), tion temperature is always equal to the minimum 500mM each standard dNTP, 10 U of MLV reverse denaturation gradient temperature. The buffer conditions transcriptase (Invitrogen) and 10 U RNAsin (Promega). were 2.5 mM MgCl , 50 mM KCl, 10 mM Tris-HCl (pH Ten percent of the reaction was used for PCR 8.3), 200mM each dTTP, dCTP, dDTP and dITP, 100mM amplification. each primer and 10 U of BioTaq DNA polymerase A fragment of the M gene of MV and of the L gene (Bioline) in a final volume of 50ml. of RVFV clone 13 was amplified by a nested procedure. Increasing the concentration of dITP and dDTP to To increase sensitivity and specificity, a hot-start PCR was 300mM did not increase product yield (not shown). performed for both amplifications. First-round primers Although inosine base pairs essentially as guanosine, it for MV were 5ROUout and 3ROUout, respectively 5 GG can form base pairs with T and A, hence the use of dITP CAGGCYGGYGCCCCAGGYCAGAG and 5 GGRR in PCR is somewhat mutagenic. In an attempt to favorize CCTCTGCGGGGTRTCGRGCGG, and maps to 3522- dC:dITP pairing the concentration of dCTP was increased 3903 on the Schwarz genome. For the second round, from 200 to 300mM while the dTTP was lowered to primers were 5ROUin and 3ROUin, respectively 5 AGA 100mM and the fidelity compared to that resulting from YCCYGGYCYAGGCGACAGGAAGG and 5 GCR amplification using equimolar 200mM dNTPs. As no TTGCRCRCTTGGTTTGCGTTG, where Y=T/C and change in PCR fidelity was found (4.1 10 versus R=A/G. First-round primer for RVFV amplification were 3.9 10 per base), all subsequent amplifications were 5RFout and 3RFout, respectively 5 GTCGCCAATGY performed using equimolar dNTPs. CGAGGAGGCCCAYGA and 5 CTCCAGATCATCT RTCCTRRTGCTTCC, and map to 5872-6255 on the Amplification by 3DI-PCR of cellular mRNA L fragment of RVFV. For the second round, primers were embedded Alu sequences 5RFin and 3RFin, respectively 5 GATGATAGAAG AYGCCAAGAACAAYGC and 5 TGCTTCCTTCTGG Total RNA from infected and uninfected MRC5 cells was TCTCTGTRGRGTTC. extracted (Epicentre). cDNA synthesis was performed by Standard dNTPs were purchased from Sigma and using random priming as described above. 1/10 of the dDTP, dITP, dUTP, 5Me-dCTP were purchased from cDNA reaction was used for PCR amplification with TriLink. DAPI was from Fluka while 7-deazadGTP and primers Alu1 (5 CACGCCTGTAATCCCAGCACTTT the Hoechst bisbenzamide dye (H33258) were from Sigma. GGG) and Alu2 (5 TGTCGCCCAGGCTGGAGTGC PCR products were purified from agarose gels and ligated AGTGG). PCR conditions were 958C for 5 min followed into the TOPO TA cloning vector cloned and sequenced by 35 cycles with 958C for 30 s, 608C for 30 s and 728C as described (5). for 1 min and a final elongation step of 728C for 10 min. First PCR was performed with standard dNTPs (TCGA). PCR protocol 1/50 of the first PCR reaction was used for 3DI-PCR with Hypermutated genomes were identified by a three-step modified dNTPs (TCID) using a Td gradient from 84 protocol. The first reaction involved a standard to 608C for 5 min then 45 cycles with 60–848C for 45 s, e72 Nucleic Acids Research, 2008, Vol. 36, No. 12 PAGE 4 OF 10 608C for 45 s and 60–728C for 1 min. PCR products were molecular clones varied linearly with G/I or A/D content purified and cloned as described above. (Figure 1E). The temperature sensitivity of TCID DNA as a function of G/I content was only 60% that of Amplification of Ig Vk1 sequences from patients 2 and 3 TCGA DNA. We explored a variety of PCR conditions to try and CD14+ B-lymphocytes were purified from two splenecto- manipulate the denaturation sensitivity of TCID DNA. mized patients using a B-cell isolation Kit (Miltenyi Biotec) Despite trying a range of small organic molecules that and DNA extracted (Epicentre). DNA was amplified using bind to AT motifs via the minor groove, i.e. Hoechst primers Ig1 (5 GCGGACATCCAGATGACCCAGTCT) bisbenzmide dye H33258, modified bases such as dUTP, and Ig2 (5 GCGCTGTTGACAGTARTAAGTTGCA). 5-MedCTP and 7-deazadGTP, monovalent (K ) and Amplification conditions were: 958C for 5 min, then 35 2+ divalent cations (Mn ), none had any significant cycles with 958C for 30 s, 608C for 30 s and 728C for 1 min impact on the minimal denaturation temperature/base 1/50 of the PCR product was used for respectively 3D-PCR composition relationship of the seven standards (Figure 2 and 3DI-PCR. For 3D-PCR conditions were, 74–948C for and not shown). In short, while the overall Td can indeed 5 min then 74–948C for 1 min 558C for 30 s and 728C for be manipulated, the denaturation temperature/base com- 1 min for 35 cycles, and for 3DI-PCR conditions were, position relationship of TCID DNA is relatively refrac- 60–758C for 5 min followed by 60–758C for 30 s, 558C for tory to manipulation. 30 s and 60–758C for 1 min with 35 cycles and a final elongation step of 60–758C for 10 min 3D- and 3DI-PCR products were purified and cloned as described above. Recovery of in vitro hyperedited measles virus sequences We sought to validate the method using measles virus RESULTS (MV) samples grown in the interferon sensitive cell line MRC-5. As a control Vero cells were used which are A wide variety of thermostable DNA polymerases were defective for interferon-a and b production (45). The first screened for their ability to amplify DNA using attenuated MV Schwarz strain was used because it is a dTTP, dCTP, dITP and dDTP. Using a standard buffer good inducer of interferon (46). Two days post-infection anda958C denaturation temperature, five of eight supernatant and cell pellets were collected and total RNA thermostable polymerases resulted in reasonable product extracted. Complementary DNA was converted into PCR recovery after 30 cycles using an extended elongation time products, a fraction of which was converted into TCID of 1 min (Figure 1B). All five were commercial variants of PCR products using a 958C denaturation temperature. Taq polymerase. However, product recovery was 3-fold Selective amplification was then applied to the TCID compared to amplification using dGTP and dATP. DNA using a denaturation gradient of 63–728C. As can be The denaturation properties of PCR DNA containing seen from Figure 3A the minimum temperature at which the two modified bases (TCID DNA) were established for MV genomes were amplified from Vero cells was 67.48C. a series of seven 262 bp DNA fragments that differed only By contrast MV specific products were amplified from by up to 23 G!A transitions distributed across the locus the MRC-5 cells down to 658C. TCID products amplified (Supplementary Figure 1). As can be seen from SYBR at the lowest Td were used for molecular cloning into Green melting profiles, midpoint denaturation tempera- TOPO plasmids. Probably in view of the unusual bases, tures (Td) of 70.3 and 72.68C were obtained for TCID transformation of standard bacteria with cloned TCID DNA corresponding to the reference (0) and 23 base products not only gave very low efficiencies (<500-fold variant respectively, as anticipated from the change in lower than TCGA DNA) but also was invariably accom- hydrogen bonding patterns (Figure 1C). As expected for panied by large deletions within the MV sequences. To standard PCR products (i.e. TCGA DNA), the converse overcome this, a fraction of TCID PCR products was prevailed, i.e. the A-rich allele was denatured at a lower temperature, Td=75.98C, than the G-rich allele converted into standard DNA by 10 cycles of PCR using (79.48C, Figure 1D). The midpoint Tds of the seven normal dNTPs and then cloned. As controls, DNA 76 74 75 +100 mM KCl B C +100 µM + 50 µM 74 + 50 mM KCl T5meCID DNA + 10 µM H33258 73 72 TCID DNA TCID DNA TCID DNA 71 70 70 70 0 5 10 15 20 25 0 5 10 15 20 25 0 5 10 15 20 25 I-rich # subs. D-rich I-rich # subs. D-rich I-rich # subs. D-rich Figure 2. The temperature differential of TCID DNA is relatively refractory to manipulation. (A) Hoechst bisbenzmide dye H33258 that binds to AT-rich DNA via the minor groove increases the Td but not the temperature differential. (B) Substituting modified bases such as 5-MedCTP for dCTP, dUTP for dTTP and 7-deazadGTP for dGTP (not shown) also failed to increase the temperature differential of TCID DNA. (C) Increasing + 2+ the ionic strength by the addition of monovalent (K ) and divalent cations (Mn , not shown) failed to increase the temperature differential. Td°C Td°C Td°C PAGE 5 OF 10 Nucleic Acids Research, 2008, Vol. 36, No. 12 e72 Figure 3. 3DI-PCR amplification of ADAR edited measles virus genomes. (A) Agarose gel of TCID DNA amplified from measles infected Vero and MRC-5 cells. The PCR products amplified from the latter between 65 and 66.28C are indicative of genomes enriched in GC. C, negative buffer control; M molecular weight markers. (B) MV sequences derived from amplification at the lowest denaturation temperature (658C). Sequences are aligned to the reference MV sequence, only differences being shown. The monotonous A!G transitions are typical of ADAR editing. Complete sequence sets are given in Supplemental Figure 2. (C) Mutation matrices for the sequence sets. The number of sequences per matrix is, starting top left and going clockwise n=15, 13, 21 and 19. The symmetry of the matrices 958C amplification controls is typical of a viral quasispecies. A slight skew in the Vero/MV 67.48C matrix from AU!GC is understandable given that 3DI-PCR amplifies GC-rich sequences, and represents the GC-rich end of the mutant spectrum. (D) Sequences of two C-rich MV sequences compared to the reference genome. The first encodes A!G and U!C transitions and arises from editing of the viral genome and anti-genome, while the latter U!C transitions indicating editing only of the anti-genome. amplified from reactions using a Td=958C was also protocol was indeed capable of recovering GC-rich alleles. cloned and sequenced. Two cytidine-rich MV sequences compared to the reference As can be seen from Figures 3B and Supplementary genome were identified (Figure 3D). The first encoded Figure 2, the MV genomes selectively amplified from A!G and U!C transitions and probably arose from MRC-5 cells (Td=658C) were littered with A!G transi- editing of the viral genome and anti-genome, while the tions. Indeed, up to 83% of A residues could be edited latter sequence encoding only U!C transitions (mean=70%, range 3–83%). By contrast, those amplified was presumably derived from editing of mRNA or the from MV-infected Vero cells at the lowest possible anti-genome. temperature (Td = 67.48C) were typical of quasispecies To ascertain their frequency, the initial TCID products variation of an RNA virus. MV sequences amplified under were serially diluted and standard and selective PCR standard PCR conditions (Td = 958C, normal dNTPs) performed. The signal from standard PCR titrated out showed balanced mutation matrices (Figure 3C). This 100-fold further than selective PCR indicating that the indicates that the highly edited sequences from the MRC-5 highly edited genomes were present in the sample at 1% cell line must represent a subset, and that the selective PCR (data not shown). A!G hypermutants were also found in e72 Nucleic Acids Research, 2008, Vol. 36, No. 12 PAGE 6 OF 10 viral supernatants from MV-infected MRC-5 cells indicat- restrictive MRC-5 culture compared to the permissive ing that hyperedited genomes can be packaged and raises Vero cell culture, 66.38C compared to 67.28C (Figure 4A). the possibility that editing might continue within the Cloning and sequencing of the PCR products revealed virion (Supplementary Figure 3). extensive A!G editing of viral RNA from the MRC-5 We refer to this novel method as inverse differential culture and nothing more than a quasispecies variation DNA denaturation PCR, or 3DI-PCR, to distinguish from the Vero cells (Figure 4B and C). Although a handful it from 3D-PCR that allows amplification of AT-rich of hyperedited sequences are shown, all 26 clones derived DNA (5). from the MRC-5 infection were distinct and harbored between 63 and 77% of edited adenosine targets. Thus Genetic editing of a segmented RNA virus hyperediting of RVFV RNA in MRC-5 can be just as extensive as for MV in the same cell line. When analysed In order to see if 3DI-PCR could be applied to another viral by standard PCR (Td = 958C, normal dNTPs) the system and hence generate novel findings, we analysed mutation matrices were balanced, indicating that the Rift Valley fever virus (RVFV), a segmented negative highly edited RVFV genomes identified represent a stranded RNA virus. Currently, there are no reports of minority (Figure 4C). Complete RVFV sequence sets ADAR edited RVFV genomes. RVFV clone 13 is a highly can be found in Supplemental Figure 4. immunogenic, yet attenuated strain that encodes a 549 bp in frame deletion within the NSs gene. As the vestigial Selective amplification of edited Alu elements in mRNA NSs protein has lost its ability to antagonize interferon production, clone 13 is a good inducer of interferon, unlike By comparison of EST sequence libraries and genome virulent strains (47). While clone 13 grew well on Vero cells, sequences several reports have shown that inverted Alu viral titers were 100-fold lower on MRC-5 cells. elements embedded in cellular mRNAs can undergo Clone 13 was cultured on both cell lines for 3 days and ADAR editing (26,35,37,38,48). We decided to see if total cellular RNA recovered. Using primers specific for 3DI-PCR could supplant such powerful, yet brut force a 257 bp fragment from the L gene, 3DI-PCR could approaches. Randomly primed cDNA from MV infected recover RVFV genomes at a lower temperature from the MRC-5 cells was used, as there was prima face evidence Figure 4. Massive adenosine deamination of Rift valley fever virus genomes. RVFV clone 13 is a highly immunogenic yet attenuated strain that encodes a 549 bp in frame deletion within the NSs gene. As the vestigial NSs protein has lost its ability to antagonize interferon production, clone 13 is a good inducer of interferon, unlike virulent strains (47). While clone 13 grows well on Vero cells, viral titres were 100-fold lower on MRC-5 cells. (A) Agarose gel of TCID DNA amplified from RVFV infected Vero and MRC-5 cells. The PCR products amplified from the latter between 66.3 and 66.78C are indicative of genomes enriched in GC. C, negative buffer control; M molecular weight markers. (B) RVFV sequences derived from amplification at the lowest denaturation temperature (66.38C). Sequences are aligned to the reference MV sequence, only differences being shown. The monotonous A!G transitions are typical of ADAR editing. Complete sequence sets are given in Supplementary Figure 4. (C) Mutation matrices for the sequence sets. The number of sequences per matrix is, starting top left and going clockwise n = 6, 10, 11 and 26. The symmetry of the matrices 958C amplification controls is typical of a viral quasispecies. PAGE 7 OF 10 Nucleic Acids Research, 2008, Vol. 36, No. 12 e72 Figure 5. Selective amplification of ADAR-edited Alu elements nested in cellular mRNAs. (A) A versus G base composition (red and black) and U versus C (blue and green) of individual Alu sequences from cellular mRNAs from measles virus (MV) infected MRC-5 cells. Sequences were derived at standard (958C, n=77 sequences) and restrictive temperatures (63.88C, n=108 sequences). A versus G correlations pertain to plus sense 3 3 Alu sequences (%A=0.996%G+59.8; r=0.915, n=108, P=<10 ) and U versus C (%U=0.922%C+38.1; r=0.914, n=108, P< 10 ), the anti-sense. The two gradients indicate a strict interconversion of A to G and U to C. (B) A versus G and U versus C correlations for Alu sequences derived from cellular mRNA from uninfected MRC5 cells under normal (n=58 sequences) and 3DI-PCR (n=67 sequences). (C) Two examples of highly edited Alu elements (B04 and C02) nested within cellular mRNAs. The alignments were those of the best hits derived from blasting the human genome. Only differences are shown to emphasize the effects of ADAR editing. Gaps were introduced into the C02 sequence to allow comparison with the B04 alignment. The number of A!G transitions are shown to the right. of ADAR activity (Figure 3). Alu-specific primers were could be useful when applied to a complex problem, we chosen corresponding to a region that apparently under- took the example of somatic hypermutation of rearranged went little ADAR-editing (35,39,49). PCR products were immunoglobulin variable (V) genes. These loci are subject amplified at a temperature as low as 63.88C. As reflected to somatic hypermutation, initiated by genetic editing of by their A versus G (+strand) and U versus C (strand) ssDNA in transcription bubbles by activation induced base compositions (Figure 5A), there was considerable deaminase, AID (50). The process features an initial phase evidence of G and C enrichment compared to Alu targeting GC base pairs followed by a second targeting AT pairs. Primers were designed to amplify the Vk1 light sequences derived from amplification using a 958C dena- chain DNA from CD14 positive splenic B cells isolated turation temperature. Indeed the majority of Alu from two patients with follicular hyperplasia who had sequences were enriched in either G or C (Figure 5A). undergone splenectomy due to untreatable thrombocyto- Blast analyses showed that the majority of selectively penia (51). As can be seen from Figure 6A and B, 3DI- amplified Alu elements had undergone ADAR-editing PCR failed to amplify highly GC-rich alleles over and (Supplementary Table 1A and B), two of which are shown above that found by PCR using a denaturation tempera- in Figure 4C. In fact ADAR-1L is induced by type I ture of 958C (Supplementary Tables 3A and C, 4A and C). interferons and was detectable by RT-PCR among MV- By contrast its counterpart, 3D-PCR, recovered a series infected MRC-5 mRNAs and not from uninfected cells of AT-rich variants (Figure 6A, Supplementary Tables 3B (not shown). Hence, the majority of edited Alu sequences and 4B). can be ascribed to ADAR-1L. When the same 3DI-PCR When blasted against the human genome, seven protocol was applied to Alu-containing mRNAs from sequences, one of which is shown in Figure 6C, showed uninfected MRC-5 cells, there were no comet tails out to an excess of GC!AT transitions in both strands, most of >40%G or >27%C as noted for MV-infected MRC-5 which were concentrated in the two complementary deter- cells (Figure 5B, Supplementary Table 1A and B). mining regions, CDR1 and CDR2 (Figure 6D, Supple- mentary Figure 5). As the nucleotide context surrounding GC and AT-rich rearranged immunoglobulin V regions the GC!AT transitions is AGCT (Figure 6E), highly Clearly 3DI-PCR is a robust method and complementary analogous to WRC motif (W = A,T; R = A,G where Cis to its sister, 3D-PCR capable of amplifying up AT-rich the edited nucleotide) observed for AID on single stranded alleles. To see if a combination of the two techniques DNA in vitro (52), and identical to that for AID e72 Nucleic Acids Research, 2008, Vol. 36, No. 12 PAGE 8 OF 10 A B %AT Patient 2 %AT Patient 3 60 60 [email protected]°C [email protected]°C [email protected]°C [email protected]°C [email protected]°C [email protected]°C 45 45 40 40 40 45 50 55 60%GC 40 45 50 55 60%GC CDR1 V T I T C R A S Q S I S S Y L N W Y Q Q K P GTC ACC ATC ACT TGC CGG GCA AGT CAG AGC ATT AGC AGC TAT TTA AAT TGG TAT CAG CAG AAA CCA ... ... ... ... ... ... ... .A. ... ..T ... ..T ..T .T. ... ... ... ... ... ... ... ... CDR2 G K A P K L L I Y A A S S L Q S G V P S R F GGG AAA GCC CCT AAG CTC CTG ATC TAT GCT GCA TCC AGT TTG CAA AGT GGG GTC CCA TCA AGG TTC ... ... ... ... ... ... ... ... ... AT. ... ... ... ... ... ... ... ... ... ... ... ... S G S G S G T D F T L T I AGT GGC AGT GGA TCT GGG ACA GAT TTC ACT CTC ACC ATC ... ... ... ... ... ... ... ... ... ... ... .T. ... FR1 CDR1 FR2 CDR2 FR3 + − ns Patient 2 A05 3 0 4 1 0 1 1 5 3 A2 0 4 0 2 1 5 2 4 3 C06 0 4 0 0 0 1 3 2 2 C07 1 1 0 2 0 0 1 2 2 E07 0 4 2 0 1 3 4 5 2 H10 1 0 0 4 0 2 3 3 2 Patient 3 C07 0 2 1 2 1 2 4 4 2 Σ 2 18 6 8 4 15 23 23 normalized data raw data, n=38 −2 −1 C +1 −2 −1 C +1 T 13 4 - 19 T 1.2 0.4 - 1.4 C 3 4 38 6 C 0.4 0.4 1.0 0.6 G 4 24 - 1 0.5 2.2 - 1.4 A 18 6 - 12 A 1.6 0.6 - 0.9 Con. C T A G Figure 6. Identification of GC and AT-rich rearranged immunoglobulin V region sequences from CD14-purified human B cells. (A and B) A+T versus G+C correlations of individual V region sequences derived by PCR (black), 3D-PCR (green) and 3DI-PCR (red) from patients 2 and 3. (C) Alignment of best Blast match to the human genome for the A2 sequence from patient 2. See Supplemental Tables 3 and 4 for complete analyses. (D) The majority of C!T transitions (68%) map to the CDR2 and CDR3 regions even though they comprise only 22.5% of the sequence.  refers to C!T transitions on the sense and antisense strands. Ns/s refer to non-synonymous and synonymous transitions respectively. (E) Sequence context of C!T transitions. The raw data is given for the 38 C!T transitions and normalized to that for the target sequences. For example, the normalized frequency of T at position –2 is the quotient of (12/38)/(32/110), the latter fraction comprises the sense and anti-sense sequences weighted by the fraction of mutations in the sense (+) and anti-sense strands () (Figure 6D). The frequency of G at position +1 is very low due to the low CpG content of the human genome. The quotient of 1.4 for G at position +1 is not robust as it reflects the quotient of two small fractions, notably (1/38)/(2/110). mutational hotspots (50), it is plausible that these bond, and allows selective amplification of AT-rich DNA sequences represent examples of the initial step in the (5). By using modified bases the 3:2 rule can be inversed, complex process of somatic hypermutation. allowing selective amplification of GC-rich alleles (Figure 1A, C and D). A range of commercially available Taq polymerases was able to undertake incorporation DISCUSSION of the two modified bases, although product yields are Differential DNA denaturation PCR exploits the intrinsic somewhat less than when standard dNTPs are used. The stability of GC base pairs arising from a third hydrogen magnitude of the temperature/GC content coefficients PAGE 9 OF 10 Nucleic Acids Research, 2008, Vol. 36, No. 12 e72 for 3D- and 3DI-PCR were not equivalent, the latter being modified nucleotides, PCR can now be extended to allow selective amplification of GC-rich DNA. 60% less than the former (Figure 1D). When applied to measles virus, the prototype for ADAR edited viral genomes, there was no difficulty in SUPPLEMENTARY DATA recovering highly edited genomes from the MRC-5 culture (Figure 3). Not only are the MV genomes more extensively Supplementary Data are available at NAR Online. edited from cultured virus than in vivo, they are more heterogeneous (19). The degree of editing observed here is unprecedented; typically ADAR-edited genomes rarely ACKNOWLEDGEMENTS contained more than 50% of edited adenosines (53). We would like to thank Chantal Combredet for the Among the present sequences sets the upper limits were measles virus cultures. This work was supported by grants 77 and 83% for RVFV and MV respectively. Although from the Pasteur Institute. R.S. was a recipient of a these RNA sequences can form secondary structures as Boehringer-Ingelheim Fonds Fellowship. Funding to pay shown by computer programs such as M-fold, never were the Open Access publication charges for this article was 80% of adenosine residues sequestered in dsRNA. provided by Institut Pasteur. That such genomes were present at frequencies of 1% Conflict of interest statement. None declared. in the MRC-5 culture may help explain why MV A!G hypermutants have not been described before in culture. The finding of numerous A!G hypermutants in culture REFERENCES of RVFV clone 13 is also novel and suggests that similar findings could be obtained with most RNA viruses if 1. Watson,J.D. and Crick,F.H. (1953) Genetical implications of the grown on interferon sensitive cells. structure of deoxyribonucleic acid. Nature, 171, 964–967. 2. Wain-Hobson,S. (2006) The third Bond. Nature, 439, 539. Why would interferon-induced ADAR-1L target 3. Corey,R.B. and Pauling,L. (1956) Specific hydrogen-bond ‘only’ 1% of genomes? The MV sequence sets shown in formation between pyramidines and purines in deoxyribonucleic Figure 3B were obtained at the lowest positive Td, i.e. acids. Arch. Biochem. Biophys., 65, 164–181. 658C. While not shown here, we know that MV sequences 4. Smith,S.M., Markham,R.B. and Jeang,K.T. (1996) Conditional reduction of human immunodeficiency virus type 1 replication taken from the Td=66.28C sample were less extensively by a gain-of-herpes simplex virus 1 thymidine kinase function. substituted suggesting that there is a large range in the Proc. Natl Acad. Sci. USA, 93, 7955–7960. degree of editing, probably reflecting varying levels of 5. Suspene,R., Henry,M., Guillot,S., Wain-Hobson,S. and ADAR-1L expression in individual cells. If larger Vartanian,J.P. (2005) Recovery of APOBEC3-edited human immunodeficiency virus G!A hypermutants by differential DNA segments were analysed the proportion of lightly edited denaturation PCR. J. Gen. Virol., 86, 125–129. sequences would increase. Hence the true number of 6. Harris,R.S., Bishop,K.N., Sheehy,A.M., Craig,H.M., edited MV genomes is probably >1%. As the genomic Petersen-Mahrt,S.K., Watt,I.N., Neuberger,M.S. and Malim,M.H. mutation rate for MV [1.4 substitutions per cycle (54)] (2003) DNA deamination mediates innate immunity to retroviral is close to the error threshold for RNA viruses, a little infection. Cell, 113, 803–809. 7. Lecossier,D., Bouchonnet,F., Clavel,F. and Hance,A.J. (2003) adenosine deamination should be sufficient to kill the Hypermutation of HIV-1 DNA in the absence of the Vif protein. virus (55). Science, 300, 1112. While the fate of ADAR-edited mRNAs is debated, 8. Mangeat,B., Turelli,P., Caron,G., Friedli,M., Perrin,L. and it does appear that it is linked to mRNA turnover (53). Trono,D. (2003) Broad antiretroviral defence by human APOBEC3G through lethal editing of nascent reverse transcripts. The finding that ADAR-editing of cellular mRNAs Nature, 424, 99–103. encoding inverted Alu elements is increased upon inter- 9. Mariani,R., Chen,D., Schrofelbauer,B., Navarro,F., Konig,R., feron induction shows that these dsRNA structures are Bollman,B., Munk,C., Nymark-McMahon,H. and Landau,N.R. relatively unprotected by protein (Figure 5). If interferon (2003) Species-specific exclusion of APOBEC3G from HIV-1 virions by Vif. Cell, 114, 21–31. can impinge on the metabolism of several hundreds of 10. Suspe` ne,R., Sommer,P., Henry,M., Ferris,S., Guetard,D., Pochet,S., mRNAs, then perhaps it might contribute to IFN-induced Chester,A., Navaratnam,N., Wain-Hobson,S. and Vartanian,J.P. cell death. (2004) APOBEC3G is a single-stranded DNA cytidine deaminase A combination of both PCR methods can be applied and functions independently of HIV reverse transcriptase. Nucleic Acids Res., 32, 2421–2429. to complex sets of sequences as highlighted by the 11. Wiegand,H.L., Doehle,B.P., Bogerd,H.P. and Cullen,B.R. (2004) edited human immunoglobulin genes (Figure 6). They A second human antiretroviral factor, APOBEC3F, is suppressed could improve the resolution of metagenomic analyses of by the HIV-1 and HIV-2 Vif proteins. EMBO J., 23, 2451–2458. bacterial genomes that vary greatly in GC content. As 12. Pathak,V.K. and Temin,H.M. (1990) Broad spectrum of in vivo forward mutations, hypermutations, and mutational hotspots in a they are PCR based they can identify low frequency retroviral shuttle vector after a single replication cycle: substitu- components that might otherwise escaped identification. tions, frameshifts, and hypermutations. Proc. Natl Acad. Sci. USA, 3DI-PCR is robust and simple to perform, dDTP and 87, 6019–6023. dITP being commercially available reagents. It is a trifle 13. Vartanian,J.P., Meyerhans,A., Asjo,B. and Wain-Hobson,S. (1991) longer in that extra PCR steps are necessary to perform Selection, recombination, and G!A hypermutation of human immunodeficiency virus type 1 genomes. J. Virol., 65, 1779–1788. the selective amplification as well as to obtain reason- 14. Suspe` ne,R., Gue´ tard,D., Henry,M., Sommer,P., Wain-Hobson,S. able cloning efficiencies. The PCR denaturation tempera- and Vartanian,J.P. (2005) Extensive editing of both hepatitis B virus ture has hitherto remained a constant, understandably DNA strands by APOBEC3 cytidine deaminases in vitro and so as the aim was to denature all DNA. With the use of in vivo. Proc. Natl Acad. Sci. USA, 102, 8321–8326. e72 Nucleic Acids Research, 2008, Vol. 36, No. 12 PAGE 10 OF 10 15. Mahieux,R., Suspe` ne,R., Delebecque,F., Henry,M., Schwartz,O., 35. Athanasiadis,A., Rich,A. and Maas,S. (2004) Widespread A-to-I Wain-Hobson,S. and Vartanian,J.P. (2005) Extensive editing of a RNA editing of Alu-containing mRNAs in the human transcrip- small fraction of human T-cell leukemia virus type 1 genomes by tome. PLoS Biol., 2, e391. four APOBEC3 cytidine deaminases. J. Gen. Virol., 86, 2489–2494. 36. Blow,M., Futreal,P.A., Wooster,R. and Stratton,M.R. (2004) A 16. Schmid,A., Spielhofer,P., Cattaneo,R., Baczko,K., ter Meulen,V. survey of RNA editing in human brain. Genome Res., 14, and Billeter,M.A. (1992) Subacute sclerosing panencephalitis 2379–2387. is typically characterized by alterations in the fusion protein 37. Eisenberg,E., Nemzer,S., Kinar,Y., Sorek,R., Rechavi,G. and cytoplasmic domain of the persisting measles virus. Virology, 188, Levanon,E.Y. (2005) Is abundant A-to-I RNA editing primate- 910–915. specific? Trends Genet., 21, 77–81. 17. Valente,L. and Nishikura,K. (2005) ADAR gene family and A-to-I 38. Kim,D.D., Kim,T.T., Walsh,T., Kobayashi,Y., Matise,T.C., RNA editing: diverse roles in posttranscriptional gene regulation. Buyske,S. and Gabriel,A. (2004) Widespread RNA editing of Prog. Nucleic Acid Res. Mol. Biol., 79, 299–338. embedded alu elements in the human transcriptome. Genome Res., 18. Cattaneo,R., Schmid,A., Eschle,D., Baczko,K., ter Meulen,V. and 14, 1719–1725. Billeter,M.A. (1988) Biased hypermutation and other genetic 39. Levanon,E.Y., Eisenberg,E., Yelin,R., Nemzer,S., Hallegger,M., changes in defective measles viruses in human brain infections. Cell, Shemesh,R., Fligelman,Z.Y., Shoshan,A., Pollock,S.R., Sztybel,D. 55, 255–265. et al. (2004) Systematic identification of abundant A-to-I editing 19. Bass,B.L., Weintraub,H., Cattaneo,R. and Billeter,M.A. (1989) sites in the human transcriptome. Nat. Biotechnol., 22, 1001–1005. Biased hypermutation of viral RNA genomes could be due to 40. DeCerbo,J. and Carmichael,G.G. (2005) Retention and repression: unwinding/modification of double-stranded RNA. Cell, 56, 331. fates of hyperedited RNAs in the nucleus. Curr. Opin. Cell. Biol., 20. Samuel,C.E. (2001) Antiviral actions of interferons. Clin. Microbiol. 17, 302–308. Rev., 14, 778–809. 41. Jepson,J.E. and Reenan,R.A. (2007) RNA editing in regulating gene 21. O’Hara,P.J., Nichol,S.T., Horodyski,F.M. and Holland,J.J. (1984) expression in the brain. Biochim. Biophys. Acta. On line 3 Dec 2007. Vesicular stomatitis virus defective interfering particles can contain 42. Wang,Q., Zhang,Z., Blackwell,K. and Carmichael,G.G. (2005) extensive genomic sequence rearrangements and base substitutions. Vigilins bind to promiscuously A-to-I-edited RNAs and are Cell, 36, 915–924. involved in the formation of heterochromatin. Curr. Biol., 15, 22. Rueda,P., Garcia-Barreno,B. and Melero,J.A. (1994) Loss of 384–391. conserved cysteine residues in the attachment (G) glycoprotein of 43. Gommers-Ampt,J.H. and Borst,P. (1995) Hypermodified bases in two human respiratory syncytial virus escape mutants that contain DNA. FASEB J., 9, 1034–1042. multiple A-G substitutions (hypermutations). Virology, 198, 44. Kirnos,M.D., Khudyakov,I.Y., Alexandrushkina,N.I. and 653–662. Vanyushin,B.F. (1977) 2-aminoadenine is an adenine substituting 23. Chang,J., Gudima,S.O. and Taylor,J.M. (2005) Evolution of for a base in S-2L cyanophage DNA. Nature, 270, 369–370. hepatitis delta virus RNA genome following long-term replication 45. Emeny,J.M. and Morgan,M.J. (1979) Regulation of the interferon in cell culture. J. Virol., 79, 13310–13316. system: evidence that Vero cells have a genetic defect in interferon 24. Birney,E., Stamatoyannopoulos,J.A., Dutta,A., Guigo,R., production. J. Gen. Virol., 43, 247–252. Gingeras,T.R., Margulies,E.H., Weng,Z., Snyder,M., 46. Combredet,C., Labrousse,V., Mollet,L., Lorin,C., Delebecque,F., Dermitzakis,E.T., Thurman,R.E. et al. (2007) Identification and Hurtrel,B., McClure,H., Feinberg,M.B., Brahic,M. and Tangy,F. analysis of functional elements in 1% of the human genome by the (2003) A molecularly cloned Schwarz strain of measles virus vaccine ENCODE pilot project. Nature, 447, 799–816. induces strong immune responses in macaques and transgenic mice. 25. Washietl,S., Pedersen,J.S., Korbel,J.O., Stocsits,C., Gruber,A.R., J. Virol., 77, 11546–11554. Hackermuller,J., Hertel,J., Lindemeyer,M., Reiche,K., Tanzer,A. 47. Billecocq,A., Spiegel,M., Vialat,P., Kohl,A., Weber,F., Bouloy,M. et al. (2007) Structured RNAs in the ENCODE selected regions of and Haller,O. (2004) NSs protein of Rift Valley fever virus blocks the human genome. Genome Res., 17, 852–864. interferon production by inhibiting host gene transcription. 26. Blow,M.J., Grocock,R.J., van Dongen,S., Enright,A.J., Dicks,E., J. Virol., 78, 9798–9806. Futreal,P.A., Wooster,R. and Stratton,M.R. (2006) RNA editing of 48. Levanon,E.Y., Hallegger,M., Kinar,Y., Shemesh,R., human microRNAs. Genome Biol., 7, R27. Djinovic-Carugo,K., Rechavi,G., Jantsch,M.F. and Eisenberg,E. 27. Kawahara,Y., Zinshteyn,B., Chendrimada,T.P., Shiekhattar,R. and (2005) Evolutionarily conserved human targets of adenosine to Nishikura,K. (2007) RNA editing of the microRNA-151 precursor inosine RNA editing. Nucleic Acids Res., 33, 1162–1168. blocks cleavage by the Dicer-TRBP complex. EMBO Rep., 8, 49. Nishikura,K. (2006) Editor meets silencer: crosstalk between RNA 763–769. editing and RNA interference. Nat. Rev. Mol. Cell. Biol., 7, 28. Kawahara,Y., Zinshteyn,B., Sethupathy,P., Iizasa,H., 919–931. Hatzigeorgiou,A.G. and Nishikura,K. (2007) Redirection of silen- 50. Di Noia,J.M. and Neuberger,M.S. (2007) Molecular mechanisms of cing targets by adenosine-to-inosine editing of miRNAs. Science, antibody somatic hypermutation. Annu. Rev. Biochem., 76, 1–22. 315, 1137–1140. 51. Cheynier,R., Henrichwark,S., Hadida,F., Pelletier,E., 29. Knight,S.W. and Bass,B.L. (2002) The role of RNA editing by Oksenhendler,E., Autran,B. and Wain-Hobson,S. (1994) HIV and ADARs in RNAi. Mol. Cell, 10, 809–817. T cell expansion in splenic white pulps is accompanied by 30. Luciano,D.J., Mirsky,H., Vendetti,N.J. and Maas,S. (2004) RNA infiltration of HIV-specific cytotoxic T lymphocytes. Cell, 78, editing of a miRNA precursor. RNA, 10, 1174–1177. 373–387. 31. Scadden,A.D. and Smith,C.W. (2001) RNAi is antagonized by 52. Pham,P., Bransteitter,R., Petruska,J. and Goodman,M.F. (2003) A!I hyper-editing. EMBO Rep., 2, 1107–1111. Processive AID-catalysed cytosine deamination on single-stranded 32. Yang,W., Chendrimada,T.P., Wang,Q., Higuchi,M., Seeburg,P.H., DNA simulates somatic hypermutation. Nature, 424, 103–107. Shiekhattar,R. and Nishikura,K. (2006) Modulation of microRNA 53. Scadden,A.D. (2005) The RISC subunit Tudor-SN binds to hyper- processing and expression through RNA editing by ADAR edited double-stranded RNA and promotes its cleavage. deaminases. Nat. Struct. Mol. Biol., 13, 13–21. Nat. Struct. Mol. Biol., 12, 489–496. 33. Pace,J.K. II and Feschotte,C. (2007) The evolutionary history of 54. Schrag,S.J., Rota,P.A. and Bellini,W.J. (1999) Spontaneous muta- human DNA transposons: evidence for intense activity in the tion rate of measles virus: direct estimation based on mutations primate lineage. Genome Res., 17, 422–432. conferring monoclonal antibody resistance. J. Virol., 73, 51–54. 34. Shen,M.R., Batzer,M.A. and Deininger,P.L. (1991) Evolution of the 55. Biebricher,C.K. and Eigen,M. (2005) The error threshold. master Alu gene(s). J. Mol. Evol., 33, 311–320. Virus Res., 107, 117–127.

Journal

Nucleic Acids Research – Oxford University Press

Published: Jul 30, 2008

There are no references for this article.