Virology 351 (2006) 42–57 www.elsevier.com/locate/yviro

The catalytic properties of the recombinant reverse transcriptase of bovine immunodeficiency virus ⁎ Orna Avidan 1, Ron Bochner, Amnon Hizi ,2

Department of Cell and Developmental Biology, Sackler School of Medicine, Tel-Aviv University, Tel-Aviv, 69978, Israel Received 3 January 2006; returned to author for revision 16 February 2006; accepted 9 March 2006 Available online 2 May 2006

Abstract

Bovine immunodeficiency virus (BIV) is a lentivirus with no proven pathogenesis in infected cattle. Yet, in experimentally infected rabbits, it causes an AIDS-like disease. Consequently, we expressed two recombinant isoforms of BIV reverse transcriptase (RT), which differ in their C- termini, and studied their catalytic properties. Both isoforms prefer Mg+2 over Mn+2 with most DNA polymerase and ribonuclease-H substrates. The processivity of DNA synthesis by the BIV RTs is higher than that of HIV-1 RT, whereas the fidelity of synthesis is even lower than that of the HIV-1 enzyme. The ribonuclease-H cleavage pattern suggests that the spatial distance between the polymerase and ribonuclease-H active sites of the two BIV RT isoforms equals 20 nt, unlike the 17 nt distance observed in almost all other RTs. The longer BIV RT version is somewhat less active than the shorter version, suggesting that the extra 74 residues (with homology to dUTPases) might obstruct efficient catalysis. © 2006 Elsevier Inc. All rights reserved.

Keywords: Bovine immunodeficiency virus; Retrovirus; Reverse transcriptase; Lentiviruses; Human immunodeficiency virus; DNA polymerase; Ribonuclease H; Protease

Introduction 1996). However, there is no conclusive evidence that BIV is the direct cause for the etiology of these syndromes. Such a The bovine immunodeficiency virus (BIV) is a member of demonstration is complicated by the co-infection with bovine the lentivirus subfamily of retroviruses, a group of exogenous, leukemia virus (BLV), another retrovirus belonging to the nononcogenic viruses that cause chronic, multi-system disease HTLV/BLV group (Flaming et al., 1993, 1997). Lentivirus in susceptible hosts (Gonda et al., 1987; Narayan and Clements, infections are typically characterized by the establishment of a 1989). BIV was first isolated from a dairy cow with persistent persistent, lifelong infection and a slow, progressive disease lymphocytosis, lymphoid hyperplasia, central nervous system course in the naturally infected host. There is wide variation in lesions and progressive weakness (Van der Maaten et al., 1972). clinical disease both among and within the different members of Serological studies indicate that BIV is present worldwide the lentivirus subfamily. The pathogenesis of BIV still remains (Cockerell et al., 1992; Forman et al., 1992), seropositivity for unknown in infected cattle and buffaloes, as no virus-specific BIV has been correlated with decreased milk production in diseases have been identified in acutely infected herds. dairy cattle (McNab et al., 1994). An increased incidence of Interestingly, BIV induces in experimentally infected rabbits encephalitis and secondary bacterial infections has been clinical syndromes, including fatal immune dysfunctions reported in herds with high BIV seroprevalence (Snider et al., similar to AIDS induced in humans by human immunodefi- ciency virus (HIV), in monkeys or in cats by the simian or feline immunodeficiency viruses, respectively (SIV and FIV) (Kal- ⁎ Corresponding author. Fax: +972 36407432. vatchev et al., 1995, 1998). E-mail address: [email protected] (A. Hizi). BIV resembles other lentiviruses in certain aspects of its 1 The results presented are in partial fulfillment of a Ph.D. thesis at Tel Aviv University. pathogenesis, ultrastructure, genome organization and infec- 2 A. Hizi is an incumbent of The Gregorio and Dora Shapira Chair for the tious cycle in susceptible cells (Braun et al., 1988; Garvey et al., Research of Malignancies at Tel-Aviv University. 1990; Gonda et al., 1987). This nonprimate lentivirus resembles

0042-6822/$ - see front matter © 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.virol.2006.03.012 O. Avidan et al. / Virology 351 (2006) 42–57 43 human and nonhuman primate immunodeficiency viruses and variability is an outstanding feature of lentiviruses, contrib- was, therefore, named BIV to reflect its genetic relationship uting to the pathogenesis of these viruses by enabling them to and biological similarity to HIV (Gonda et al., 1987). persist in the host and evade anti-retroviral treatments. As Nevertheless, BIV is unable to infect human cells (Gonda et shown for known lentiviruses, isolates of BIV exhibit a al., 1994). In terms of genome organization, BIV is the most striking genomic diversity, located mainly in the viral complex nonprimate lentivirus. In addition to the obligate envelope gene (Meas et al., 2001). However, assessment of retroviral gag, pol and env structural genes, BIV encodes six the extent of variation in the reverse transcriptase (RT) putative nonstructural/accessory genes; vif, tat, rev, vpw, vpy domain of the pol gene revealed that this is one of the most and tmx (Garvey et al., 1990; Gonda, 1994, 1987; Oberste conserved regions of the genome. The RT region of BIV and Gonda, 1992; Oberste et al., 1991, 1993). Genetic genome, like that of HIV, is subjected to purifying selection

Fig. 1. Clustal W pairwise alignment of the amino acids residues encoded by the pol genes of BIV and HIV-1 (of the BIV127 and BH-10 strains, respectively). The upper lines represent the amino acids sequences of BIV PR and RT (and the amino-terminus of integrase). Recombinant BIV PR is 105 residues-long and recombinant RTs are 546 and 620 residues-long. BIV RT1 carboxyl terminus aligns with HIV-1 RT carboxyl terminus, RT2 carboxyl terminus is the last residue in the BIV sequence before the alignment with the amino terminus of HIV-1 IN. Asterisks in the lower lines mark the identical amino acids. The numbers of the residues are indicated on the left of each line. (AT) denotes amino terminus and (CT) denotes carboxyl terminus. 44 O. Avidan et al. / Virology 351 (2006) 42–57 and these two lentiviruses probably diverged from a common gaard et al., 1991; Kirchhoff et al., 1990; Mang et al., 2001; progenitor (Cooper et al., 1999). Martoglio et al., 1997; Takehisa et al., 2001). Therefore, it was Reverse transcriptase is a key enzyme in the life cycle of all surprising to see that the sequence similarity between BIV IN retroviruses (Coffin et al., 1997; De Clercq, 1997; Skalka and and HIV-1 IN starts 74 residues downstream to the putative Goff, 1993). The replication of these viruses starts with the carboxyl terminus of BIV RT and there is no sequence homolog reverse transcription of the single-stranded viral RNA genome in HIV-1 for this protein segment. As we could not predict to double-stranded DNA, which is subsequently integrated into whether this segment is part of the BIV RT, we have expressed the host cell chromosomal DNA. This complex process of two versions of the recombinant BIV RTs. The first one, proviral DNA synthesis, which is highly critical to retroviral life designated RT1, is the 546 residues long BIV RT. The second cycle, is catalyzed by a single enzyme, the viral RT, by BIV RT version is longer and includes at the carboxyl terminus employing its various activities. The plus-strand RNA genome the extra 74 residues segment. Therefore, this RT version, is copied by the RNA-dependent DNA polymerase (RDDP) designated RT2, is 620 residues long and ends at its carboxyl activity of RT, producing RNA–DNA heteroduplexes. Con- terminus with VFMTC-COOH, immediately adjacent to the comitant with this synthesis, the ribonuclease-H (RNase H) FLEN sequence that is the putative amino terminus of the BIV activity of RT specifically hydrolyzes the RNA strands in these IN (see Fig. 1). RNA–DNA hybrids and the remaining DNA strands are copied The genomes encoding the genes for the expression of the into the complementary DNA strands by the DNA-dependent BIV RT isoforms (and protease—see below) were synthesized DNA polymerase (DDDP) activity of RT (generating double- by PCR using as a template the pBIV127 plasmid that contains stranded DNA). Due to its critical role in retroviral replication, the genome of an infectious molecular clone of BIV (Braun et inhibitors of RT are effective in inhibiting viral propagation al., 1988) as described in detail in Materials and methods. All (Coffin et al., 1997). known lentiviral RTs are composed of heterodimeric proteins Due to the potential uniqueness of BIV RT, we have where the longer subunit (for example, the p66 or p68 subunit in expressed this novel lentiviral RT and studied its catalytic RTs of HIV-1 or HIV-2, respectively) is cleaved at their properties. Here, we report on the expression in Escherichia coli carboxyl termini by the viral protease (PR) removing the RNase of the enzymatically active recombinant BIV RT. This RT is H domain and yielding a shorter subunit (as p51 or p54, giving derived from a pBIV127, a plasmid containing the complete and rise to the heterodimeric RT forms, p66/p51 or p68/p54, in HIV- infectious proviral BIV genome sequence (Braun et al., 1988). 1 or HIV-2, respectively). One successful way to create such Because of the relatively low sequence homology with other recombinant RTs is to co-express simultaneously by the same RTs at the carboxyl terminus (see Fig. 1), we have expressed plasmid vector both the RT the PR proteins, assuming that in the two recombinant versions of BIV RT that differ in length at their expressing bacteria the PR will cleave the RT to generate the carboxyl termini. Consequently, a comparative study on the authentic heterodimeric RTs (Coffin et al., 1997; DeVico et al., catalytic and molecular properties of these two RTs was 1989; di Marzo Veronese et al., 1986; Gao et al., 1998; Sevilya conducted, thus allowing the prediction of the most effective et al., 2001; Skalka and Goff, 1993). Therefore, the BIV RT- carboxyl terminus. The recombinant BIV RTs show high expressing vectors, constructed by us, also induce the synthesis catalytic activities, resembling in most cases those of HIV-1 RT. of the 105 residues-long BIV PR (according to the sequence Expressing and investigating the molecular and catalytic shown in Fig. 1). The simultaneous expression of the BIV RT properties of BIV RT can potentially allow a better understand- versions and PR was done in Bl21 E. coli under the control of ing of the closely related HIV-1 RT. Moreover, this knowledge the phage T7 promoter (see Materials and methods). Both RT can allow targeted drug design against lentiviral diseases, versions were designed to contain at their amino termini a six- including acquired immunodeficiency syndrome (AIDS). histidines tag to allow rapid purification of the enzymes. The plasmid designations are pT5m-6His-BIV RT1/PR and pT5m- Results 6His-BIV RT2/PR, for the shorter and the longer BIV RT versions, respectively. The bacterial clones harboring these Plasmids construction for protein expression plasmids were analyzed for the expression of BIV RT using two methods. First we have tested the RDDP activity of the Two recombinant BIV RTs were designed based on the recombinant RTs in bacterial lysates, as done by us earlier (Hizi sequence homology of BIV pol gene with HIV-1 pol gene (Fig. and Hughes, 1988; Perach and Hizi, 1999; Taube et al., 1998b). 1), both share the same amino terminus and differ in their Second, we have employed Western blot analyses with anti-His carboxyl termini. A comparison between the pol genes of both tag antibodies. Both methods clearly indicated an expression of HIV-1 and BIV revealed that the best fit between HIV-1 RT (560 soluble and enzymatically active RT species. amino acids in length) and BIV RT is obtained when the BIV RT protein starts at its amino terminus with the sequence NH2- Purification of BIV RTs VHTEKI and ends at the carboxyl terminus with the sequence ALEQMA-COOH. This gives a BIV RT protein that is 546 The BIV RTs were purified to homogeneity, employing amino acids in length. In most retroviruses, the integrase (IN) the Ni-NTA agarose affinity chromatography, designed to protein is encoded by the genomic sequence that is adjacent selectively bind the six histidines tag-containing proteins, immediately downstream to the RT-coding sequence (Foms- followed by carboxymethyl (CM)-Sepharose ion-exchange O. Avidan et al. / Virology 351 (2006) 42–57 45

Fig. 2. SDS/PAGE analysis of the purified recombinant BIV RTs. The purification of BIV RTs was performed as described in Materials and methods and in the text. The purified BIV RT1 and BIV RT2, marked as BIV(1) and BIV(2), respectively, were analyzed by 7.5% SDS/PAGE. The gels were then stained with Coomassie (A) or underwent Western blot analysis with HRP-conjugated anti-6-Histidines rabbit antiserum, purchased from Sigma (performed as described previously (Entin-Meer et al., 2002) (B). The apparent molecular sizes of the PERV RT and the subunits of HIV-1 RT are shown (left), and the molecular sizes of the subunits of the preparations recombinant BIV RTs are marked (right). chromatography (Fig. 2 and Table 1), as described earlier for versions is similar (Fig. 2). Nevertheless, beside the two other RTs (Avidan et al., 2003; Perach and Hizi, 1999; Sevilya subunits, there is an additional polypeptide of about 69 kDa, et al., 2001; Taube et al., 1998b). The pT5m-6His-BIV RT1/PR slightly shorter than the large RT2 subunit. Since these two plasmid induces the synthesis of a heterodimeric protein with polypeptides (as well as the p51 subunit) were recognized by two subunits with molecular sizes of approximately 64 and anti-his tag antibodies, as seen in a Western blot analysis (Fig. 51 kDa at roughly an equimolar ratio. The larger subunit of RT1 2B), it is conceivable that they share the same amino terminus is compatible with the expected full length of the RT1 546 and differ only at their carboxyl termini. On the other hand, residues recombinant protein (Fig. 2). Since a parallel purification of BIV RT2 from the expressing bacteria, harboring purification of RT1, expressed from a plasmid that lacked the a plasmid that lacks the PR gene, resulted in obtaining the same PR genome, yielded only the 64 kDa polypeptide (data not p72 and p69 polypeptides (data not shown). Taken together, it is shown), it is highly likely that the p51 polypeptide is the product evident that the formation of the smaller 51 kDa subunit in both of the p64 subunit, and it was generated by the specific RT versions is due to a specific proteolytic cleavage by the BIV proteolytic activity of BIV-encoded PR. In HIV-1, the PR PR, whereas the generation of the p69 polypeptide of RT2 is removes from the larger RT subunit a segment of 120 residues, likely to be due to either proteolysis of the 72 kDa subunit by generating the p51 subunit (see the cleavage point in Fig. 1). nonspecific bacterial proteases or by premature termination of The size of the shorter BIV1 RT subunit is compatible with a the synthesized protein. similar PR-dependent cleavage that generates a BIV RT subunit A quantitative analysis of the purification steps follows the of about 420 residues in length. This suggests that the cleavage extent of increase of the specific activity of the two novel ver- specificity of the novel BIV PR is similar to that of HIV-1 PR. sions of BIV RT (Table 1). The first affinity purification step on As to the polypeptides synthesized by the pT5m-6His-BIV Ni-NTA has lead to a huge increase in the specific DNA RT2/PR plasmid, a substantial portion of the full length subunit, polymerase activity of both BIV RT versions (of 3600-fold to induced by this plasmid, is about 72 kDa in size that is 23,000-fold for RT2 and RT1, respectively). This increase is due compatible with the expected 620 amino acid-long polypeptide to a combination of both removals of bacterial proteins and, not (Fig. 2). The smaller subunit of RT2, generated in the induced less importantly, a dramatic increase in the total polymerase bacteria by apparently the BIV-encoded PR cleavage, shows the activity by about 150-fold. After the CM-Sepharose ion ex- same length as the small subunit of RT1. This suggests that, as change chromatography, there is a second, although a less pro- expected, the protease cleavage site in the two novel BIV RT nounced, increase in the specific DNA polymerase activity, of

Table 1 The quantitative pattern of BIV RTs purification from bacterial extracts BIV RT1 BIV RT2 Protein DNA polymerase Specific DNA polymerase Protein DNA polymerase Specific DNA polymerase (mg) activity (units) activity (units/μg protein) (mg) activity (units) activity (units/μg protein) Crude bacterial extract 480 4.8 × 103 1×10−2 (1) 210 1.5 × 103 7.2 × 10−3 (1) Ni-NTA agarose column eluate 3 6.8 × 105 227 (2.3 × 104) 9 2.3 × 105 25.7 (3.6 × 103) CM-sepharose column eluate 1.9 9.6 × 106 5065 (5 × 105) 1 1.1 × 106 1057 (1.5 × 105) The DNA polymerase activity is expressed in units. A unit is defined as 1 pmol of [3H]dTTP incorporated into TCA-insoluble material in 30 min at 37 °C in the poly

(rA)n·oligo(dT)12–18-directed reaction, as described in “Material and methods”. The figures in parenthesis are the purification fold, i.e. the factor by which the specific RDDP activity of the purified protein increased relative to that of the crude bacterial extract. 46 O. Avidan et al. / Virology 351 (2006) 42–57 about 20-fold for RT1 and 40-fold increase in the case of RT2 apparent that both BIV RT isoforms prefer Mg+2 over Mn+2 with (Table 1). At this purification step, there is still a mild increase all substrates used to assay both the DNA polymerase and the (5–14-fold) in the total DNA polymerase activity. Taken RNase H activities, except for one substrate, poly(dA)n·oligo together, both versions of the novel BIV RT underwent (dT)12–18. It is also obvious that, for both RT versions, the extensive purifications; with a massive increase in the specific specific preferences for the template-primers, employed for the activity (of more than 100,000-fold). The increase in the total polymerase activities, show similar selections. As for the RDDP RT activity precludes the calculation of the relative recovery of function of almost all studied RTs, the BIV RT isoforms exhibit the BIV RTs during the purification process. We have observed higher activities with poly(rA)n·oligo(dT)12–18 than with poly previously a similar phenomenon of an overall increase in the (rC)n·oligo(dG)12–18. To assay the DDDP activity, we have used total DNA polymerase activity throughout the purification of both synthetic substrates and activated DNA, prepared by a the recombinant RTs of MMTV (Taube et al., 1998b), BLV limited digestion of herring sperm DNAwith pancreatic DNase I (Perach and Hizi, 1999) and porcine endogenous retrovirus to create single-stranded gaps (Hizi et al., 1991b). Here, the (PERV) (Avidan et al., 2003). This phenomenon can be order of preference with the various template-primers is as fol- explained by a removal of bacterial components that strongly lows:poly(dC)n·oligo(dG)12–18 >poly(dA)n·oligo(dT)12–18 >ac- inhibit RT activity. tivated DNA. It is also evident that, for almost every given substrate and divalent cation employed, BIV RT1 exhibits a The relative DNA polymerase activities specific activity somewhat higher than that observed with BIV – RT2 (except for the poly(rC)n·oligo(dG)12 18-directed DNA The enzymatic activities of BIV RT1 and BIV RT2 were polymerase activity, where the activities of the two isoforms are +2 +2 3 analyzed in the presence of 0.8 mM Mn or 8 mM Mg , very similar). The [ H]poly(rA)n·poly(dT)n-directed RNase H allowing also a comparison of the activities and the divalent activity measured is also higher for RT1 relative to RT2 with a cation preferences of the two BIV RT isoforms. These two preference for Mg+2 over Mn+2. Interestingly, RT2 does not concentrations of divalent cations were chosen, since we had show any detectable activity in this specific assay in the presence found in preliminary experiments that they were the optimal of Mn+2. The preference for Mg+2 as the divalent cation, found concentrations for the activity of both BIV RT isoforms (data not in this study, is in line with most retroviral RTs, except for shown). Previous data show that retroviral RTs differ in their mammalian type C retroviruses RTs (Avidan et al., 2003; Hizi divalent cation preferences (Avidan et al., 2003; Hizi and and Joklik, 1977; Kohlstaedt et al., 1992; Taube et al., 1998b). Hughes, 1988). As already reported earlier (Avidan et al., 2003; Perach and Hizi, 1999; Skalka and Goff, 1993; Taube et al., Steady-state kinetic parameters for the DNA polymerase 1998b), we also show here that this preference of the two RT activity isoforms depends on the substrate used for assaying RT activities. The relative specific activities of the BIV RTs with The kinetics for the incorporation of the relevant dNTPs into a variety of substrates used are presented in Table 2. Like all RTs the nascent DNA strand were determined for both BIV RT studied so far, both versions of the new BIV RT have the RDDP isoforms from the double reciprocal plots of the initial velocity and DDDP activities as well as an intrinsic RNase H activity. It is of dNTP incorporation versus substrate concentration (not shown). For RDDP activity, we have tested variable concentra- Table 2 tions of dTTP with poly(rA)n·oligo(dT)12–18 and the variable Specific catalytic activities, substrates and divalent cation preferences of BIV dGTP with poly(rC)n·oligo(dG)12–18. For the DDDP function, RT1 and RT2 dTTP or dGTP were the variable substrates with activated RT Substrate BIV RT1 BIV RT2 gapped double-stranded DNA as the template-primer (with activity Mg+2 Mn+2 Mg+2 Mn+2 dATP and dCTP at a constant concentration). To examine the kinetic parameters under optimal conditions, the divalent cation DDDP dA·dT 185 ± 4 365 ± 0.5 153 ± 3 175 ±4 dC·dG 708 ±16 217±8 551 ±41 137±8 used in each reaction with all the template-primers studied was +2 Activated DNA 144 ±6 65±4 80 ±8 23±3 the preferable one, Mg (see Table 2). The results presented in RDDP rA·dT 7800 ± 200 4120 ± 20 3860 ± 75 2080 ± 30 Table 3 show that BIV RT1 is somewhat more efficient as a rC·dG 552 ± 14 57 ± 0.4 567 ±30 48±2 DNA polymerase relative to RT2, as deduced from the k /K 3 cat m RNase H [ H]rA·dT 91 ±11 72±5 43 ±3 0 values calculated. This higher efficiency of BIV RT1 is espe- The DNA polymerase-specific activities are expressed in pmol of dNTP cially pronounced with the parameters calculated for dTTP in incorporated into DNA in 30 min at 37 °C per μg purified RT. For activated the poly(rA) ·oligo(dT) – -directed assay (an 8.5-fold differ- DNA, the dNTP incorporation was calculated for all four dNTPs, assuming n 12 18 equimolar incorporation. The specific RNase H activity was expressed in pmol ence). In this case, both a lower Km value and a higher turnover 3 3 [ H]AMP released from [ H]poly(rA)n·poly(dT)n in 30 min at 37 °C per μg number (kcat) contribute to the substantially higher value calcu- protein (as described in “Material and methods”). The different catalytic lated for RT1. In all, the values calculated are quite similar to activities were determined in the presence of either 0.8 mM MnCl2 or 8 mM comparable ones calculated for the recombinant RTs of HIV-1, MgCl2. The values presented are averages calculated from four independent EIAV, BLV and MMTV and are lower than those of the RTs of experiments (after subtracting background levels with no RT present in the reaction) ± the standard deviations. The specific activities values marked in bold HIV-2, MLV and PERV RTs (Avidan et al., 2003; Hizi et al., figures are the maximal ones obtained for each RT (for every given substrate) 1991b; Perach and Hizi, 1999; Rubinek et al., 1994; Taube et al., while comparing the Mg+2-dependent versus the Mn+2-dependent reactions. 1998b). O. Avidan et al. / Virology 351 (2006) 42–57 47

Table 3 The extent of product elongation in one cycle of synthesis Steady-state kinetic parameters for DNA polymerase activities of BIV RT1 and depends on parameters that affect binding, single nucleotide BIV RT2 addition, translocation, pausing, etc. Retroviral RTs are known rA·dT rC·dG Activated DNA to be inefficient in performing fully processive polymerization (dTTP) (dGTP) (dTTP) (dGTP) events (Avidan and Hizi, 1998; Avidan et al., 2003). Therefore, BIV RT1 we have tested the processivity of the novel isoforms of the −6 −6 −7 −6 Km(M) 4.8 × 10 2.8 × 10 6.0 × 10 2.2 × 10 recombinant BIV RT by examining their capacity to synthesize −1 kcat(s ) 403 33 9.9 12.8 DNA, using a template of single stranded ΦX174am3 DNA −1 −1 7 7 7 6 kcat/Km(M s ) 8.5 × 10 1.2 × 10 1.7 × 10 5.8 × 10 (Fig. 3). This was done in the presence of Mg+2 or Mn+2 as the BIV RT2 divalent cation in the reaction mixture and in comparison with −5 −6 −7 −6 Km(M) 1.8 × 10 2.9 × 10 8.2 × 10 3.2 × 10 the well-studied RTs of HIV-1 and MLV, each in the presence of −1 kcat(s ) 182 26 8.2 11.2 its preferred divalent cation (Avidan and Hizi, 1998; Avidan et −1 −1 7 6 7 6 32 kcat/Km(M s ) 1.0 × 10 9.0 × 10 1.0 × 10 3.5 × 10 al., 2003). The extension of the 5′-end [ P]-labeled primer was The RDDP and DDDP activities of BIV RT1 and BIV RT2 were assayed with carried out in the absence or presence of an excess of unlabeled the specific template-primers, and variable concentrations of either dTTP or activated DNA, which was added after preincubating the “ ” dGTP, as described in Material and methods section. All reactions were carried enzymes with the labeled template annealed to primer (and prior out for 10 min at 37 °C. The Km and Vmax values were determined from the double reciprocal plots of the various substrate concentrations against the initial to initiating polymerization by adding the four dNTPs). This velocities of DNA synthesis. The plots were generated by linear regression unlabeled DNA serves to trap the enzyme that dissociates from analysis, yielding regression coefficients around 0.99 (not shown). The kcat values the template-primer after the first cycle of DNA synthesis is are the steady state rate coefficients calculated from the Vmax values divided by completed. As this DNA trap prevents further extensions of the the moles of each enzyme present in the reactions. Km values are expressed in M labeled primer, due to re-association of the RT, each end-labeled of the variable dNTP and the k values in s−1 and the k /K values in s−1 M−1. cat cat m polymerized molecule is restricted to a single round of DNA synthesis. As expected, both BIV RT isoforms make longer DNA synthesis under processive and nonprocessive conditions DNA products when multiple rounds of synthesis are permitted when no trap is present. The enzymes used have been calibrated During enzymatic polymerization reactions, the lengths of to have equal DDDP activities with ΦX174am3 DNA (Fig. 3). the nascent polymeric products, formed before the polymerase In the absence of the DNA trap, the distributions of the products molecules dissociate from them, define the processivity features (generated by both RTs) show a significant preference for Mg+2. of the enzyme (Avidan and Hizi, 1998; Von Hippel et al., 1994). The overall primer extension of both BIV RT isoforms in the

Fig. 3. DNA synthesis under processive and nonprocessive conditions exhibited by BIV RT. All reactions were performed with the 15-nucleotide synthetic 5′-end- labeled primer and an excess of the template single-stranded circular ϕX174am3 phage, DNA as shown previously (Avidan and Hizi, 1998). The sequence of the primer is shown in Fig. 6A. The symbols for the DNA synthesis experiments are as follows: (−) DNA extension performed without DNA trap; (+) extension +2 experiments conducted in the presence of 0.6 mg/ml unlabeled trap of activated DNA (Avidan et al., 2003). (Mg ) reactions with 8 mM MgCl2 as the divalent cation +2 donor; (Mn ) reactions performed with 0.8 mM MnCl2. Molecular mass markers were HinfI-cleaved dephosphorylated double-stranded ϕX174am3 DNA fragments (Promega) labeled with γ[32P]ATP at the 5′-ends by polynucleotide kinase (not shown). 48 O. Avidan et al. / Virology 351 (2006) 42–57 presence of Mg+2 is somewhat higher than those of the RTs of the relative overall processivity values) for both recombinant HIV-1 and MLV, as can be also seen from the qualitative results BIV RTs, is significantly higher (by up to 2.3-fold) with Mg+2 (Fig. 3). over Mn+2. In conclusion, it is apparent that both enzymes are The extension products were also quantified and the more processive than the RTs of HIV-1 and MLV, these results elongation was calculated as a percentage of the total amount are in agreement with former studies (Avidan and Hizi, 1998; of primers (both unextended and extended primers—Table 4). Avidan et al., 2002, 2003). The overall processivity of DNA The qualitative and quantitative data indicate that both BIV RT synthesis of RT1 over RT2 is about the same in the presence of isoforms extend (in the presence of Mg+2 as the preferred Mg+2, but with Mn+2 RT1 functions better than RT2. divalent cation) the majority of the primers molecules, when multiple rounds of DNA synthesis are permitted (i.e. without a The RNase H activity trap). BIV RT1 extends in the presence of Mg+2 approximately the same proportion of the primers (about 91%) compared to All RTs have an intrinsic RNase H activity that is BIV RT2 (about 89%). The majority of the products of both RTs fundamental for the process of reverse transcription. Sequence (of about 61%) are longer than 100 nt. As can be seen in Fig. 3 and structural analyses have shown that the RNase H domain is and similar to other RTs (Avidan and Hizi, 1998; Avidan et al., located in the carboxyl terminus of the RTs, whereas the DNA 2002, 2003; Taube et al., 1998a), there are specific pausings polymerase at the amino terminal portion of the molecule. To during DNA synthesis, which are quite similar for both BIV RT confirm the presence of the RNase H activity in the novel BIV isoforms and are common to the two other RTs, analyzed here. RT and to characterize it, two assay methods were employed These pausing sites do not seem to depend on whether Mg+2 or (Gao et al., 1998; Hizi et al., 1991a; Sevilya et al., 2001). In the Mn+2 are present. On the other hand, the length of the products first method, the release of [3H]AMP was followed as a result of 3 depends substantially on the divalent cation used. The products the digestion of [ H]poly(rA)n·poly(dT)n and in the second, the generated with the unfavorable cation are shorter, suggesting pattern of cleavage of labeled RNA with a defined length in an that the translocation (or what we define as “persistence of RNA–DNA heteroduplex has been followed (see Materials and synthesis”) is impaired, even when multiple rounds of synthesis methods). The data obtained indicate that the BIV RT isoforms are allowed. When a DNA trap is present, and only one round of exhibit, like most RTs studied so far, an RNase H activity with a DNA synthesis is permitted, both RTs synthesize less and significant preference for Mg+2 over Mn+2 (Table 2 and Fig. 4). shorter products relative to multiple-round synthesis (Fig. 3 and It is also apparent that the specific RNase H activity of BIV RT1 Table 4). Additionally, the ratio calculated for the overall is significantly higher than that of BIV RT2 with both divalent extension with a trap, over that without a trap (which equals to cations. The pattern of the cleavage of the 5′-end-labeled 267 nt RNA Table 4 was used to characterize the RNase H activity of the RTs of Quantitative analysis of DNA primer extension HIV-1 and HIV-2 (Gao et al., 1998; Sevilya et al., 2001) as well +2 +2 as those of PERV, MLV (Avidan et al., 2003) and mouse Product length Mg Mn (nt) mammary tumor virus (MMTV) (Entin-Meer et al., 2002). All No trap With trap Ratio No trap With trap Ratio these RTs usually perform two cleavages. First, a primary RT1 cleavage takes place at a position 17 nt away from the position – 16 50 15.4 19.5 1.3 33.6 34.0 1 corresponding to the 3′-end of the DNA heteroduplex (thus, 51–100 15.0 13.6 0.9 21.2 7.6 0.36 101–250 42.0 41.3 0.98 20.8 12.7 0.6 producing in the specific assay employed, a 47-nt-long RNA >250 18.9 13.1 0.7 7.9 0.6 0.08 product). Since it is assumed that the DNA polymerase domain Overall extension 91.3 87.5 0.96 83.5 54.9 0.66 of the RT binds the 3′-end of the DNA primer, this primary cleavage suggests that the molecular distance between the DNA RT2 polymerase active site and the RNase H active sites equals to 17 16–50 13.7 20.8 1.5 33.4 11.5 0.34 51–100 14.6 15.9 1.09 21.3 7.0 0.33 nt. In many RTs, there is also a secondary RNase H cleavage at a 101–250 43.1 38.5 0.89 21.6 14.0 0.65 position that corresponds to 8 nt away from the 3′-end of the >250 17.9 9.3 0.52 6.7 3.7 0.55 DNA (producing in this assay 38 nt-long RNA molecules). This Overall extension 89.3 84.5 0.95 83.0 36.2 0.44 probably results from a secondary repositioning of the RT The radioactivity in the DNA bands in all polynucleotide length ranges was molecules in the RNA–DNA heteroduplexes after performing summed up and the values were divided by the sums of all extended and the primary cleavage (Gao et al., 1999; Palaniappan et al., 1996; unextended primers (detected in the scanning of the gels autoradiogram as shown Sevilya et al., 2001; Wisniewski et al., 2000). The primary in Fig. 3). The values given are the extended primers, in each product length range, expressed as percentages of the total amounts (all extended and cleavage performed by HIV-1, HIV-2, MMTV, MLVand PERV unextended primers) of the DNA products. The calculations were conducted RTs produces a similar 47 nt-long product (Avidan et al., 2003; separately for gel lanes of reactions carried out in the absence or presence of an Gao et al., 1999; Sevilya et al., 2001; Wisniewski et al., 2000). excess of the unlabeled DNA trap. The extensions in the presence of the DNA As seen in Fig. 4, both BIV RT isoforms show a pattern of trap divided by the comparable figures (obtained with no trap present) yielded the RNase H primary cleavage, that is different than the one ratio values. The ratios calculated for the overall extensions are the relative processivity values. The values are the means calculated from two independent obtained with HIV-1 RT and hence with all above mentioned experiments (one of which is shown in Fig. 3) and the variations were less than RTs, producing a 50-nt-long RNA product. This result suggests 10%. that in BIV RT the putative spatial distance between the RNase O. Avidan et al. / Virology 351 (2006) 42–57 49

Fig. 4. The RNase H activity of BIV RT. The scheme in the upper panel describes the principle of the reaction and the lengths of the cleavage products of the BIV RT isoforms, which differ from those of HIV-1 RT (see text). The RNase H activity of BIV RT1 and BIV RT2 was compared with RNase H activity of HIV-1 RT employing the assay described in Materials and methods. Fifty nanograms of each highly purified RT was incubated with the RNase H substrate (the 5′-end-labeled 267 nt RNA and oligomeric DNA) at 37 °C, for the indicated times (0.25, 1 and 2 min). The reactions were performed with either Mg+2 or with Mn+2. The products were resolved by urea-PAGE. The sizes of the cleavage products were determined using an RNA size markers ladder. The 5′-end-labeled RNA marker ladder was prepared by incubating the 5′-end-labeled 267 nt RNA in 10 mM NaOH for 30 s at 50 °C (not shown). Lane C shows the product of a mixture with no RT present.

H and the polymerase active sites is approximately 20 nt. assay there was no detectable activity with this divalent cation Interestingly, both BIV RT isoforms perform a secondary (Table 2 and Table 5). It is also apparent that the qualitative cleavage that produces 38 nt-long RNA molecules that pattern of RNase H cleavage with Mg+2 or Mn+2 is quite similar correspond to a −8 nt cleavage (Fig. 4). This indicates that for the two BIV RT isoforms. BIV RT2 is longer than RT1 by 74 the repositioning of BIV RT1 and RT2, after performing the amino acid residues; the addition is on the carboxyl terminus of primary RNase H cleavage, is similar to that of HIV-1 RT, as the enzyme. Since RNase H domain is located in the carboxyl well as HIV-2 RT and MMTV RT (Entin-Meer et al., 2002; terminus of the RTs, it might be that the extra 74 residues in the Sevilya et al., 2001). Another conclusion is that the RNase H longer RT version interfere with the RT structure needed for activity of both BIV RT isoforms is weaker than that of HIV-1 efficient RNA cleavage. This can explain why RT2 has a lower RT. RNase H activity relative to RT1. The results of the RNase H activity detected in two independent experiments (one of which is shown in Fig. 4), The strand transfer activity were also quantified and the percentages of each cleavage are shown in Table 5. Compatible with the data presented in Table 2 During the process of reverse transcription of the retroviral 3 for the [ H]poly(rA)n·poly(dT)n-directed RNase H activity, BIV RNA genome into double-stranded DNA, there are two RT1 has substantially higher RNase H activity than RT2. RT1 template-switches or strand transfers (Coffin et al., 1997). performs almost equally well with either one of the two divalent Both the DNA synthesis and RNase H functions of RT are cations. However, RT2 shows a significant preference for Mg+2 simultaneously involved in this process (Gao et al., 1998; over Mn+2. Still, RT2 has a substantial activity with Mn+, Peliska and Benkovic, 1992). The RNase H activity cannot 3 despite the fact that in the [ H]poly(rA)n·poly(dT)n-directed cleave the 5′-end of the viral RNA until the minus strand strong 50 O. Avidan et al. / Virology 351 (2006) 42–57

Table 5 Kinetics of RNase H primary and secondary cleavages BIV RT1 BIV RT2 HIV-1 RT Mg+2 Mn+2 Mg+2 Mn+2 15″ 2′ 15″ 2′ 15″ 2′ 15″ 2′ 15″ 2′ 1st cleavage 49 ± 1.1 5.9 ± 0.3 51 ± 3 12 ± 1.1 7.5 ± 0.6 4.6 ± 0.2 5.7 ± 0.6 29 ± 1.4 69 ± 0 0 2nd cleavage 6.5 ± 0.5 78 ± 1.2 5.5 ± 0.4 63 ± 1.8 1.6 ± 0.2 64 ± 2.5 0.2 ± 0 10 ± 1.3 31 ± 0 100 Total cleavage 56 ± 1.6 84 ± 1.5 57 ± 3.4 75 ± 2.9 9.1 ± 0.8 69 ± 2.7 5.9 ± 0.6 39 ± 2.7 100 100 The radioactivity in the RNA bands of both first and second cleavages and the 267 nucleotides-long substrate was summed up and the values were subsequently divided by the total amounts, which are the sums of all cleaved and uncleaved labeled RNA bands (detected by scanning of the gels autoradiogram, one of which is shown in Fig. 4). The values given are the cleavage products, expressed as percentages of the total amounts. The calculations were conducted separately for gel lanesof reactions carried out with either Mg+2 or with Mn+2, at 37 °C, for the indicated times. stop DNA is synthesized and this cleavage is required for the that BIV RT1 is more active than RT2, especially with Mg+2 newly synthesized DNA to be released from the original RNA and that for both isoforms the activity in the presence of Mg+2 is template, so it can re-anneal to the 3′-end of the RNA genome. better than the one in the presence of Mn+2. The amount of the Since both DNA synthesis and RNase H activities are full-length 238 nt DNA synthesized, which represents the obligatory for the strand-transfer process, we tested the transfer activity, was quantified and the transfer activity values efficiency of strand-transfer reactions with the two BIV RT of BIV RT1 are higher (5%) than for BIV RT2 (3%) with the isoforms. Employing the in vitro strand transfer assay preferable cation (Mg+2), and 3% versus 2% respectively with previously performed (Gao et al., 1998; Sevilya et al., 2001, Mn+2. These results are compatible with the fact that RT1 is 2003), we show here that, indeed, the BIV enzymes are capable more active than RT2 in both DNA polymerase and RNase H of catalyzing a strand transfer reaction (Fig. 5). It is apparent functions and that Mg+2 is usually the preferred divalent cation (Table 2, Fig. 4 and Table 5). The transfer activity value calculated here for HIV-1 RT (19.8%) is significantly higher than those of BIV RTs and it is mainly due to the higher RNase H activity of HIV-1 RT (Fig. 4 and Table 5). As in the case of the novel BIV RT, a relatively weak RNase H and subsequently a low strand transfer activity, were observed by us also for two strains of HIV-2 RT (Sevilya et al., 2001, 2003). This lower RNase H may not substantially affect the replication rates of virions with a diminished RNase H activity, as it was found that reducing RNase H activity had a lesser effect on viral replication than reducing the DNA polymerase activity (Julias et al., 2001).

The fidelity of DNA synthesis

The fidelity of DNA synthesis was studied in vitro with many retroviral RTs suggesting a relatively low fidelity of DNA synthesis. Similar to all RTs studied so far, BIV RT lacks a 3′ → 5′ exonuclease proofreading activity (data not shown), thereby permitting a direct kinetic analysis of primer extension. It was shown that the primary parameters for fidelity (namely, the 3′-end misinsertion and the extension of the preformed 3′- end mispaired primers) depend largely on the sequences of Fig. 5. Strand-transfer activity of BIV RT. Strand-transfer reactions containing nucleic acids copied, rather than on whether DNA or RNA 0.4 μg of purified RT were incubated at 37 °C for 60 min as described and the templates are used (Bakhanashvili and Hizi, 1993a, 1993b; reaction products were resolved by urea-PAGE (Sevilya et al., 2001). The strong Rubinek et al., 1997; Taube et al., 1998a). Consequently, we stop DNA transcript, which is nearly a full-length copy of the primary RNA, is have analyzed DNA templates, as representing both DNA and marked by (s) and is 139 nucleotides in length. The full-length transfer product is marked by (t) and is 238 nucleotides in length. The reactions were performed RNA substrates. with either 8 mM Mg+2 or with 0.8 mM Mn+2. Lanes 1–5, the products obtained in the presence of both the primary 267 nt RNA and the secondary 139 nt RNA. The formation of 3′ mispaired DNA Lanes 6–10, the products of control reactions performed without the secondary The fidelity of misinsertion was studied in an assay system RNA (required for the strand transfer). Lanes 1, 6, BIV RT1 in the presence of that measures the standing-start reaction of 3′-end misinsertion. Mg+2. Lanes 2, 7, BIV RT1 in the presence of Mn+2. Lanes 3, 8, BIV RT2 in the presence of Mg+2. Lanes 4, 9, BIV RT2 in the presence of Mn+2. Lanes 5, 10, This was done by following the misincorporation of incorrect HIV-1 RT in the presence of Mg+2. M, Molecular mass markers are HinfI- dNTPs opposite to the terminal template “A” nucleotide—in cleaved ϕX174am3 DNA fragments (mentioned in Fig. 3). comparison with the incorporation of dTTP (see Fig. 6A and O. Avidan et al. / Virology 351 (2006) 42–57 51

Fig. 6. The fidelity of misincorporation and mispair extension of BIV RT. (A) The pattern of site-specific nucleotide misinsertion: The synthetic 50-nucleotide template was annealed to the [32P]-5′-end-labeled 15-nucleotide primer. The primer was extended with equal DNA polymerase activity of either BIV RT1, BIV RT2 or HIV-1 RT, in the presence of 1 mM of a single incorrect dNTP (i.e. C, G or A) or 1 μM of dTTP, as described in Materials and methods. (B) The pattern of extension of preformed 3′-end mispaired DNA: the [32P]-5′-end-labeled 16-nucleotide primers were hybridized to the 50-nucleotide template. Four different duplexes were produced, out of which three were the 3′-terminal preformed mismatches, where N represents the incorrect nucleotide (A, C or G) and the fourth is the matched duplex (where N represents T). The primers were extended with equal DNA polymerase activities of BIV RT1, BIV RT2 or HIV-1 RT, in the presence of either 1 mM dATP (when the mispaired template-primers were elongated) or 1 μM dATP (for the extension of the matched template-primer).

Materials and methods). The reactions were performed with radioactivity in gel bands relative to the total amounts of Mg+2 that is the preferred divalent cation. BIV RT1 and BIV primers present (both the unextended and the extended ones). RT2 (as well as HIV-1 RT) elongate the 5′-end-labeled primer The apparent Km and Vmax values for each dNTP derived from by one nucleotide in the presence of 1 mM dTTP with no further the double-reciprocal curves of the initial velocities of primer extensions. The highest extent of misincorporation by the two extension versus the substrate concentrations (not shown) are BIV RTs is obtained with dCTP, forming C–A mispair that is presented in Table 6. The frequencies of misinsertion (Fins further elongated to form C–T mispair. This is followed by the values) were calculated from the formula: correct C–G pair (18 nt) and by a C–T mispair (19 nt length). ðV =K ÞW With dGTP, the elongation is slightly lower than with dCTP, F ¼ max m – – ins R forming a G A mispair, which is further elongated to form G T ðVmax=KmÞ (17 nt), G–G (18 nt) and C–T (19 nt) mispairs. With the wrong dATP, there is very little misincorporation by BIV RTs, forming (W) stands for the wrong nucleotide (dATP, dCTP and an A–A mispair that, as expected, is fully elongated to form the dGTP) and (R) denotes the dTTP. Therefore, the quantitative correct A–T pair (17 nt). It is apparent that the general pattern of evaluation of misincorporation is always normalized relative to primer extension obtained with the two BIV RT isoforms is the activity of incorporating the right dNTP. As expected from quite similar to this with HIV-1 RT. the results presented in Table 6, the highest Fins value calculated To quantify the capacity of BIV RTs to form 3′-end mispairs, for both BIV RTs is for dCTP (1/3200), whereas, the formation four sets of primer-extension reactions were carried out and of A–G mispair is moderately lower (Fins = 1/9900 and 1/3800) analyzed (each done with increasing concentrations of a single and the Fins values for dATP incorporation is relatively rare (1/ dNTP), thereby determining the standing-start rate of synthesis 34,700 and 1/18,200). It is also evident that the fidelity of of the correct pair compared to the three possible mispairs. To misinsertion by BIV RT1 is better than that calculated for RT2. obey steady-state kinetic conditions, we have used a range of In comparison, the Fins values previously calculated in the same dNTP concentrations below 1 mM and calculated the assay system for HIV-1 RT were 1/3400–1/9000, 1/32,200–1/ 52 O. Avidan et al. / Virology 351 (2006) 42–57

41,500, and 1/52,200–1/75,000, for the formation of A–C, A– Table 7 ′ G and A–A mispairs, respectively (Bakhanashvili et al., 1996; Kinetic parameters of the extension of 3 -end matched or preformed mismatched primer termini by BIV RT isoforms Rubinek et al., 1997). This indicates that both versions of the μ BIV RT studied here are significantly more error prone than Pair extended Vmax (%) Km ( M) Fext HIV-1 RT. RT1 A·T 9.6 0.008 1 Extension of preformed 3′-end mispaired DNA A·C 4.4 11 1/3000 A·G 5.4 31 1/6900 The misincorporation of a nucleotide by itself is not adequate A·A 5.1 11 1/2600 to create site-specific mutations, unless the mispaired DNA terminus is further extended to fix the wrong sequence. RT2 Therefore, the effectiveness of extending 3′ preformed A·T 6.9 0.015 1 mismatched primers is an essential factor in determining the A·C 3.6 15 1/1900 A·G 2.1 13 1/2800 fidelity of DNA synthesis. The ability of the BIV RT isoforms to A·A 3.4 16 1/2200 ′ – extend preformed 3 -end mispaired 16-residue primers (A A, ′ – – The 5 -end-labeled 16-nucleotide primers were hybridized to a 50-nucleotide A C, A G) was assessed by analyzing the extension of these template (as described for Table 6). The duplexes produced were 3′-terminal primers, in the presence of the next complementary dATP (Fig. paired (A–T) or mispaired (A–C, A–GorA–A) primers (Fig. 6B). Each 6B). Both BIV RTs elongate all 5′-end-labeled mispairs by one template-primer was incubated with either BIV RT1 or BIV RT2 in the presence nucleotide with no substantial preference towards any specific of increasing concentrations of dATP. The products were analyzed as described mispair DNA. The kinetics of the mispair extension was studied in the text. The apparent Km and Vmax values were determined from at least two independent experiments and the variations were usually less than 10%. The by following primer elongation as a function of increasing relative frequency Fext values are the ratio of the rate constants (Vmax/Km) for the concentrations of dATP (Table 7). The ratios of all extended mispair divided by the ratio of the corresponding constants for the paired A–T products were calculated relative to the total amount of the terminus. primers as a function of dATP concentration. The relative extension frequency (Fext) values are defined as Vmax/Km for A–A and A–C are higher for both BIV RT versions (about values (calculated for the preformed mismatches) divided by 1/2200, 1/1900 for RT2 and 1/2600, 1/3000 for RT1), whereas the Vmax/Km value, obtained for the correctly paired A–T the frequencies for A–G are slightly lower (about 1/3000 for terminus. The apparent Vmax values for the extension of all RT2 and 1/7000 for RT1). In all, BIV RT2 is slightly more three mispairs studied by BIV RT1 are similar and somewhat error prone than BIV RT1. The Fext values calculated higher than the comparable values calculated for BIV RT2. As previously for HIV-1 RT in the same assay system were: 1/ expected, the Km values calculated for the extension of the A– 1000, 1/11,400, 1/1200, for A–C, A–G and A–A mispairs, A, A–C and A–G mismatches by the BIV RT isoforms are respectively (Bakhanashvili et al., 1996). much higher than the Km of the correct A–T pair. For RT1, the Km value of the mispair A–G is higher and is the cause for Discussion the relatively lower frequency of the elongation of this preformed mismatch. The calculated frequencies of extension Unlike HIV and SIV, the BIV is a lentivirus of unknown pathogenesis in infected cattle and buffalo and no virus-specific disease has been yet identified in infected herds. Yet, in an Table 6 experimentally infected rabbits, the virus causes syndromes Kinetic parameters of the site-specific misincorporation by BIV RT isoforms similar to AIDS (Kalvatchev et al., 1995, 1998), implying its μ Mispair formed Vmax (%) Km ( M) Fins potential importance in a novel AIDS-related animal model RT1 system. The putative amino acid sequence of BIV RT shows A·T 19.9 0.025 1 high homology to the sequence of the well-studied HIV-1 RT. A·C 6.8 27 1/3200 A·G 4.9 61 1/9900 Thus, a pairwise alignment of the amino acids sequences of A·A 1.7 74 1/34,700 these two RTs (Fig. 1) yields a score of 37 (Thompson et al., 1994). Thirty-eight percent of the residues are identical and, on RT2 top of that, 21% are homologous. Since RT is a key enzyme in A·T 9.9 0.043 1 the life cycle of all retroviruses (Coffin et al., 1997; De Clercq, A·C 4.4 62 1/3200 A·G 3.4 56 1/3800 1997; Skalka and Goff, 1993), investigating of BIV RT in A·A 1.0 79 1/18,200 comparison to other known RTs (especially HIV-1 RT) may The 15-residue 5′-end-labeled primer was hybridized to an excess of 50-residue help in understanding the variance in the pathogenesis of template that is derived from the sequence of nucleotides 565–614 of different lentiviruses. ϕx174am3 DNA (Fig. 6A). In each set of the kinetic experiments, the Two versions of the recombinant BIV RTs were cloned: RT1, template-primer was incubated with either BIV RT1 or BIV RT2 in the presence the 546 residues-long BIV recombinant protein and the longer of increasing concentrations of a single dNTP. The oligonucleotide products RT2, that includes at the carboxyl terminus an additional 74 were analyzed as described in Materials and methods. The apparent K and m residues segment and is altogether 620 residues long (Fig. 1). In Vmax values were determined from at least two independent experiments and the variations were usually less than 10%. The values of relative frequency of the absence of data from virus-synthesized RT and in order to insertion (Fins) were calculated as described in the text. determine the optimal length of the BIV RT protein, we have O. Avidan et al. / Virology 351 (2006) 42–57 53 studied these two BIV RT versions and conducted a A processive event is defined as the ability of polymerase to comparative study of their catalytic and molecular properties. bind a polymeric substrate (DNA or RNA) and perform a The major conclusion is that the recombinant BIV RT1 is at polymerization cycle without intervening dissociation. A total least as good a reverse transcriptase as RT2. In fact, the longer processivity of synthesis of DNA or RNA is accomplished BIV RT version has some significant disadvantages relative to when the entire template is copied as a consequence of only one the shorter version of the RT, which can result from the polymerase-binding event (Avidan and Hizi, 1998; Von Hippel excessive sequence at the carboxyl terminus that might interfere et al., 1994). Retroviral RTs are quite inefficient in performing with the enzyme activity. Hence, we conclude that the short totally processive events (Avidan and Hizi, 1998; Avidan et al., version, which aligned quite well with HIV RTs and SIV RT, is 2002, 2003; Taube et al., 1998a). However, the results shown more likely to be closer to the authentic and biologically here (Fig. 3 and Table 4) indicate that these features of the BIV relevant BIV RT. RT isoforms are significantly higher than those of both HIV-1 The obvious question arising is what is the role of the 74 RT and MLV RT. In the presence of Mg+2, the BIV RT isoforms amino acids long peptide. Since the amino terminus of the synthesize significant amounts of product DNA that, even with putative integrase of BIV aligns quite well with that of HIV-1 an excess of unlabelled trap DNA, are longer than about 250 nt. (Fig. 1), it might be that the 74 residues segment is associated In comparison, in the presence of a DNA trap, HIV-1 RT with another protein and/or another catalytic activity. Indeed, synthesizes only a small amount of product DNA that is longer several lentiviruses, such as FIV and EIAV, express a fourth than 250 nt, while MLV RT generates products of only about up retroviral enzyme (additional to RT, IN and PR), namely a to 100 nt. These differences between BIV RT and the two other dUTPase (Bergman et al., 1995; Olmsted et al., 1989). This RTs may suggest that BIV RT binds better the DNA substrate protein is encoded by a gene located between the RT and IN- than most RTs studied including PERV RT and BLV RT encoding segments, where the BIV-related 74 residue-encoding (Avidan et al., 2002, 2003). Taken together, BIV RT is a highly segment is located. Interestingly, there is some homology processive retroviral RT and belongs together with MMTV RT between the sequence of the 74 residues BIV segment and these and AMV RT to a group of apparently more processive RTs lentiviral dUTPases giving scores of 22 and 24 with the (Taube et al., 1998a). All RTs studied show the same premature sequences of the dUTPases of FIV and EIAV, respectively. pausings, with or without trap DNA (Fig. 3), we concluded that However, the FIV and EIAV dUTPases are about 130-residues the pausings are sequence-dependent. It will be of interest to long, almost twice the length of the BIV segment; thus, the study other sequences than those used here to identify any likelihood that this 74 residues segment can possess such an unique sequence that cannot be copied easily by retroviral RTs. independent enzymatic activity is quite low. Moreover, other DNA polymerases which lack the 3′ → 5′ proofreading retroviruses of the type B and type D family express another exonuclease activity are more error prone, explaining, at least in kind of dUTPase as a transframe protein from the overlap part, why RTs have a lower fidelity than other DNA polymerases between the Gag and PR-encoding genes (Coffin et al., 1997; (Bakhanashvili and Hizi, 1993a; Echols and Goodman, 1991; Hizi et al., 1987; Koppe et al., 1994; Oda et al., 1988; York et Perrino et al., 1989; Taube et al., 1998a). Yet, a comparison of the al., 1992). This dUTPase is significantly larger than the overall fidelity of DNA synthesis exhibited by RTs from lentiviral enzyme with 250–320 residues (depending on the different retroviruses reveals significant differences among viral strain). Even the dUTPases of E. coli and Candida them. The RTs of HIV-1 and HIV-2 were found to have a albicans are about 150 residues long (Hayashi et al., 2001; relatively low fidelity of DNA synthesis compared to other RTs, Jones et al., 2004), and as far as we know, there is no evidence such as AMV, MMTV, EIAVor MLV (Bakhanashvili and Hizi, for an active 74 residues-long dUTPase. Still, all dUTPases 1992; Bakhanashvili et al., 1996; Rubinek et al., 1997; Taube et show a limited homology (with five conserved sequence motifs) al., 1998a; Yu and Goodman, 1992). In general, the fidelity of with the BIV protein segment (data not shown). It is likely, DNA polymerases results from the combination of nucleotide however, that the 74 residues-long BIV-derived segment has a insertion and extension in the presence or absence of biological role, since it is also found in another related bovine proofreading activity. Base substitution mutations during reverse lentivirus, the Jembrana disease virus (Chadwick et al., 1995). It transcription can arise from incorporation of a noncomplemen- should be noted that all retroviral dUTPases are cleaved off the tary nucleotide at the 3′-end of the nascent DNA strand, followed precursor polyproteins by the viral PR and probably function as by an extension of the preformed mispair. The misinsertion independent enzymes. However, in our BIV RT-PR bacterial co- frequencies observed with both recombinant BIV RTs show an expression system, the molecular size of the larger subunit of error proneness that is higher than that observed with HIV-1 RT RT2 is significantly larger than RT1 (Fig. 2), in agreement with (Fig. 6A and Table 6). As for the capacity of BIV RTs to elongate the expected size (except for the p69 version of BIV RT2—see preformed mispairs, it is apparent from Fig. 6B that BIV RTs above). Therefore, it seems that, at least in this system, the 74- exhibit a pattern of extension roughly similar to that observed residues segment is not removed from the RT by the BIV PR. It with HIV-1 RT, except for A–C mispair extension that produces might be that this segment has a novel regulatory role on BIV longer products by HIV-1 RT. Accordingly, the Fext values RT, which is unknown in other studied RTs. To this end, we plan calculated for the BIV RT isoforms (Table 7) indicate that the to assay the RT2 protein for dUTPase activity and also to error proneness of the mispair extension by BIV RTs is express this independent segment and study its potential somewhat lower than that of HIV-1 RT (Bakhanashvili et al., activities. 1996; Rubinek et al., 1997). In all, the apparent low fidelity of 54 O. Avidan et al. / Virology 351 (2006) 42–57

DNA synthesis by BIV RT, shown in this study, can explain why 2042) were subcloned from the BIV proviral clone (Braun et al., BIV isolates, like those of other known lentiviruses, exhibit a 1988) into the pT5M plasmid designated pT5m-6His-BIV RT1/ striking genomic diversity (Meas et al., 2001). PR. First, the RT1-encoding gene (to which NcoI and NotI sites The RNase H domain is located in the carboxyl terminus of as well as a stop codon were added) was subcloned by ligating the RTs. BIV RT1 and BIV RT2 differ in their carboxyl the NcoI–NotI cleaved PCR product into the NcoI–NotI cleaved terminus, RT2 has an addition of 74 residues. The RNase H plasmid. Then, the PR-encoding gene was PCR-synthesized activity of RT1 is higher than that of RT2 (Fig. 4, Table 2 and with additional NotI and HindIII restriction sites, the appropriate Table 5). It might be that the extra sequence at the carboxyl initiation and stop codons, as well as at its 5′-end segment with a terminus of RT2 interferes with the enzyme structure required Shine–Dalgarno sequence (for the mRNA ribosome-binding for an efficient RNA cleavage; still the cleavage pattern by both site), which is located between the NotI sequence and the ATG isoforms is indistinguishable. Maybe the most interesting result initiation codon. This allows a simultaneous expression of both regarding the novel BIV RT is that it produces an RNase H RT and PR proteins from the single polycistronoic mRNA. The cleavage product (20 nt) that is significantly longer than the plasmid was created by ligating the NotI–HindIII cleaved PR- segment generated by HIV-1 RT and for that matter by all encoding gene into the NotI–HindIII linearized plasmid. The retroviral RTs studied so far. This finding suggests that the BIV RT2-encoding segment (nucleotides 2043–3909) has an molecular spatial distance between the DNA polymerase and internal NcoI site. Therefore, for the construction of the BIV RT2 the RNase H active sites in the novel BIV RT is longer by 3 nt expression plasmid, the PCR-synthesized DNA segment (to than all other retroviral RTs. It should be noted, however, that in which NcoI and NotI sites as well as a stop codon were added) the case of the RT from the LTR-retrotransposon Ty3, the was cleaved with KpnI (located internally in the RT1/RT2 hydrolysis profile suggests an increased spatial distance overlapping sequence at position 2105 of the provirus) and NotI between the two catalytic domains (Rausch et al., 2000). On and introduced into the KpnI–NotI cleaved pT5m-6His-BIV the other hand, our very recent study with the RT of another RT1/PR plasmid, thus replacing the BIV RT1-encoding gene LTR-retrotransposon, Tf1, shows an RNase H cleavage pattern with the RT2-encoding segment. The BIV RT2-expressing similar to that generated by HIV-1 RT (unpublished results). plasmid was designated pT5m-6His-BIV RT2/PR. In all, we hope that this study of a novel BIV RT will lead to a All subcloned BIV-related DNA segments were sequenced better understanding of structure–function relationships in to verify the authenticity of the expressed proteins. Each of lentiviral RTs. The fact that the 74 residues segment at the these constructed plasmids co-expresses both the BIV PR and carboxyl terminus of the constructed RT2 does not add to the the RT1 or RT2. Therefore, the large RT subunit homodimers, activities RT studied and, if anything, interferes to a certain synthesized after the induction by isopropyl β-tiogalactopyr- extent with most activities, leaves open the function of the anoside (IPTG) in the E. coli strain BL21(DE3)pLysS, were dUTPase-like sequence in the biology of BIV. Consequently, proteolytically processed in the expressing cells, leading to the we plan to study this issue and to test whether this sequence can accumulation of heterodimeric RT molecules that are composed function as an independent protein or, as an independent protein each of the larger and smaller subunits (Boyer et al., 1994; Gao or as an integral part of RT and thus has a new role in regulating et al., 1998; Le Grice and Gruninger-Leitch, 1990; Sevilya et RT activities. al., 2001). To allow a fast affinity purification of the recombinant RTs, six-histidine tags were added to the amino Materials and methods termini of both BIV RT versions. Consequently, after generating the heterodimeric RT molecules, both the larger and smaller Plasmids construction for protein expression subunits, of each version, have this tag on the amino termini.

We have expressed and studied two recombinant forms of The relative DNA polymerase activities BIV RT, designated BIV RT1 and BIV RT2. The DNA segments that encode the two RT isoforms were generated by The DNA polymerase activities of BIV RT1 and RT2 were PCR using the plasmid pBIV127, which contains an infectious analyzed and compared in the presence of 0.8 mM Mn+2 or proviral clone of BIV DNA (Braun et al., 1988) that was 8mMMg+2 as the divalent cation. The DDDP reactions were generously provided by the “NIH AIDS Research and performed with the synthetic template-primers, poly(dA)n·oligo Reference Reagent Program”. Both BIV RT versions were (dT)12–18 and poly(dC)n·oligo(dG)12–18, or with herring sperm designed to start with the same amino termini, whereas the activated gapped DNA, which was prepared as described carboxyl terminus of RT2 is 74 amino acids longer than that of previously (Hizi et al., 1991b). The RDDP reactions were RT1. These RT versions were expressed in E. coli by plasmids performed with the synthetic template-primers, poly(rA)n·oligo similar to those used previously for the expression of the RTs of (dT)12–18 and poly(rC)n·oligo(dG)12–18. In addition, the appro- HIV-1 and HIV-2 (Gao et al., 1998; Sevilya et al., 2001). All priate dNTPs were added to the reaction mixtures along with synthetic BIV proteins-encoding genes were generated by PCR 25 mM Tris–HCl, 40 mM KCl and either 8 mM MgCl2 or with the high fidelity Dynazyme polymerase (purchased from 0.8 mM MnCl2, final pH 7.5. Unless otherwise stated, the final Finnzames) using the appropriate synthetic oligonucleotides. concentration of the synthetic template-primers was 5 μg/ml and The BIV RT1-encoding segment (nucleotides 2043–3681) of the activated DNA 30 μg/ml, while the concentration of each and the BIV PR-encoding DNA segment (nucleotides 1727– dNTP used was 25 μM. The specific catalytic activity of the O. Avidan et al. / Virology 351 (2006) 42–57 55

DNA polymerases activities was calculated and is expressed in Primer elongation was measured as a function of increasing pmol of dNTP incorporated into DNA in 30 min at 37 °C per μg concentrations of dATP, as the only dNTP present (0–1mM purified RT (Table 2). range for the mispaired A–A, A–CorA–G termini, or a 0– 1 mM range for the correct A–T terminus). DNA synthesis under processive and nonprocessive conditions All fidelity reactions were incubated at 37 °C for either 2 min (for the correct incorporation or the extension of the accurate The reactions were performed as described by us before template-primer) or for 5 min (for misincorporation or extension (Avidan and Hizi, 1998; Avidan et al., 2002, 2003). The enzymes of preformed mismatched DNA). Kinetic reactions were used have been calibrated to have equal DDDP activities with performed with an about 10-fold molar excess of template- ϕX174am3 as DNA template and the reactions were performed primer over enzyme to ensure steady-state kinetics at the linear in the presence of 8 mM Mg+2 or 0.8 mM Mn+2. The reaction range. The reaction products were analyzed by electrophoresis products were analyzed by electrophoresis through 8% poly- through sequencing gels and the band intensities were acrylamid-urea gels, as described earlier (Bakhanashvili and quantified. Hizi, 1992, 1993a, 1993b). The relative amounts of the labeled extension products were determined from the autoradiograms Acknowledgments and the percentage of the total amounts (unextended and extended) of the extended primers were calculated. We would like to thank Dr. S. Loya for critically reading the manuscript. We also thank The NIH AIDS Research & The RNase H activity Reference Reagent Program for supplying the plasmid pBIV127. This activity was assayed by two methods. In the first, the 3 3 release of [ H]AMP from [ H]poly(rA)n annealed to poly(dT)n References (into the trichloroacetic acid-soluble fraction) was measured, as previously described (Hizi et al., 1991a). In the second method, Avidan, O., Hizi, A., 1998. The processivity of DNA synthesis exhibited by ′ drug-resistant variants of human immunodeficiency virus type-1 reverse we have followed the pattern of cleavage of the labeled 5 -end transcriptase. Nucleic Acids Res. 26, 1713–1717. 267 nt RNA, synthesized from the pBLRA30 plasmid (as Avidan, O., Meer, M.E., Oz, I., Hizi, A., 2002. The processivity and fidelity of described by Gao et al., 1998) and annealed to 30 nucleotides- DNA synthesis exhibited by the reverse transcriptase of bovine leukemia long oligomeric DNA with a complementary sequence: 5′-TGG virus. Eur. J. Biochem. 269, 859–867. GTT CCC TAG TTA GCC AGA GAG CTC CCA-3′ (see Avidan, O., Loya, S., Tonjes, R.R., Sevilya, Z., Hizi, A., 2003. Expression and characterization of a recombinant novel reverse transcriptase of a porcine scheme in Fig. 4). The reaction mixtures contained approxi- endogenous retrovirus. Virology 307, 341–357. 32 mately 0.1 pmol of the [ P] 5′-end-labeled RNA transcript, Bakhanashvili, M., Hizi, A., 1992. Fidelity of the reverse transcriptase of human annealed to 0.4 pmol of the 30mer synthetic oligonucleotide immunodeficiency virus type 2. FEBS Lett. 306, 151–156. DNA. The 3′-end of the DNA oligonucleotide is complemen- Bakhanashvili, M., Hizi, A., 1993a. Fidelity of DNA synthesis exhibited in vitro tary to the region starting 30 nt away from the 5′-end of the by the reverse transcriptase of the lentivirus equine infectious anemia virus. Biochemistry 32, 7559–7567. RNA transcript. The reactions contained 50 ng of each purified Bakhanashvili, M., Hizi, A., 1993b. The fidelity of the reverse transcriptases of RT and were performed, analyzed and calculated as described human immunodeficiency viruses and murine leukemia virus, exhibited by (Sevilya et al., 2001), except for the longer oligomeric DNA the mispair extension frequencies, is sequence dependent and enzyme related. FEBS Lett. 319, 201–205. used and the usage of either MnCl2 or MgCl2 (0.8 mM or 8 mM, respectively). Bakhanashvili, M., Avidan, O., Hizi, A., 1996. Mutational studies of human immunodeficiency virus type 1 reverse transcriptase: the involvement of residues 183 and 184 in the fidelity of DNA synthesis. FEBS Lett. 391, Fidelity of DNA synthesis 257–262. Bergman, A.C., Bjornberg, O., Nord, J., Rosengren, A.M., Nyman, P.O., 1995. The reactions were performed by monitoring the standing- dUTPase from the retrovirus equine infectious anemia virus: high-level start misinsertion and mispair extension properties of the RTs, expression in Escherichia coli and purification. Protein Expression Purif. 6, 379–387. using the same method and calculations as described before Boyer, P.L., Tantillo, C., Jacobo-Molina, A., Nanni, R.G., Ding, J., Arnold, E., (Bakhanashvili and Hizi, 1993a; Taube et al., 1998a). Site- Hughes, S.H., 1994. Sensitivity of wild-type human immunodeficiency specific nucleotide misinsertion was assayed by following the virus type 1 reverse transcriptase to dideoxynucleotides depends on template incorporation of dNTP opposite to the A residue at position 23 length; the sensitivity of drug-resistant mutants does not. Proc. Natl. Acad. – of a 50-nucleotide synthetic template (with a sequence derived Sci. U.S.A. 91, 4882 4886. Braun, M.J., Lahn, S., Boyd, A.L., Kost, T.A., Nagashima, K., Gonda, M.A., from nucleotides 565–614 of ϕX174am3 DNA), see Fig. 6A. 32 1988. Molecular cloning of biologically active proviruses of bovine The [ P] 5′-end-labeled 15 residues primer was extended in the immunodeficiency-like virus. Virology 167, 515–523. presence of increasing concentrations of either 0–1 mM dTTP Chadwick, B.J., Coelen, R.J., Wilcox, G.E., Sammels, L.M., Kertayadnya, G., or 0–1 mM of each incorrect dNTP (dATP, dCTP or dGTP). For 1995. Nucleotide sequence analysis of Jembrana disease virus: a bovine 32 lentivirus associated with an acute disease syndrome. J. Gen. Virol. 76, determining the mispair extension, four sets of 16-residue [ P] – ′ 1637 1650. 5 -end-labeled primers were annealed to the same template used Cockerell, G.L., Jensen, W.A., Rovnak, J., Ennis, W.H., Gonda, M.A., 1992. in the misinsertion assays, producing duplexes with a 3′-termini Seroprevalence of bovine immunodeficiency-like virus and bovine leukemia paired (A–T) or mismatched (A–C, A–GorA–A) primers. virus in a dairy cattle herd. Vet. Microbiol. 31, 109–116. 56 O. Avidan et al. / Virology 351 (2006) 42–57

Coffin, J.M., Hughes, S.H., Varmus, H.E., 1997. Retroviruses. Cold Spring pro gene products and the ribosomal frameshift site by protein sequencing. Harbor Laboratory Press, Cold Spring Harbor, NY. Proc. Natl. Acad. Sci. U.S.A. 84, 7041–7045. Cooper, C.R., Hanson, L.A., Diehl, W.J., Pharr, G.T., Coats, K.S., 1999. Natural Hizi, A., Tal, R., Hughes, S.H., 1991a. Mutational analysis of the DNA selection of the Pol gene of bovine immunodeficiency virus. Virology 255, polymerase and ribonuclease H activities of human immunodeficiency virus 294–301. type 2 reverse transcriptase expressed in Escherichia coli. Virology 180, De Clercq, E., 1997. In search of a selective antiviral chemotherapy. Clin. 339–346. Microbiol. Rev. 10, 674–693. Hizi, A., Tal, R., Shaharabany, M., Loya, S., 1991b. Catalytic properties of the DeVico, A.L., Copeland, T.D., Veronese, F.D., Oroszlan, S., Gallo, R.C., reverse transcriptases of human immunodeficiency viruses type 1 and type 2. Sarngadharan, M.G., 1989. Purification and partial characterization of J. Biol. Chem. 266, 6230–6239. human immunodeficiency virus type 2 reverse transcriptase. AIDS Res. Jones, T., Federspiel, N.A., Chibana, H., Dungan, J., Kalman, S., Magee, B.B., Hum. Retroviruses 5, 51–60. Newport, G., Thorstenson, Y.R., Agabian, N., Magee, P.T., Davis, R.W., di Marzo Veronese, F., Copeland, T.D., DeVico, A.L., Rahman, R., Oroszlan, S., Scherer, S., 2004. The diploid genome sequence of Candida albicans. Proc. Gallo, R.C., Sarngadharan, M.G., 1986. Characterization of highly Natl. Acad. Sci. U.S.A. 101, 7329–7334. immunogenic p66/p51 as the reverse transcriptase of HTLV-III/LAV. Julias, J.G., Ferris, A.L., Boyer, P.L., Hughes, S.H., 2001. Replication of Science 231, 1289–1291. phenotypically mixed human immunodeficiency virus type 1 virions Echols, H., Goodman, M.F., 1991. Fidelity mechanisms in DNA replication. containing catalytically active and catalytically inactive reverse transcrip- Annu. Rev. Biochem. 60, 477–511. tase. J. Virol. 75, 6537–6546. Entin-Meer, M., Sevilya, Z., Hizi, A., 2002. The role of phenylalanine-119 of the Kalvatchev, Z., Walder, R., Barrios, M., Garzaro, D., 1995. Acquired immune reverse transcriptase of mouse mammary tumour virus in DNA synthesis, dysfunction in rabbits experimentally infected with an infectious molecular ribose selection and drug resistance. Biochem. J. 367, 381–391. clone of the bovine immunodeficiency virus (BIV127). Viral Immunol. 8, Flaming, K., van der Maaten, M., Whetstone, C., Carpenter, S., Frank, D., Roth, 159–164. J., 1993. Effect of bovine immunodeficiency-like virus infection on immune Kalvatchev, Z., Walder, R., Perez, F., Garzaro, D., Barrios, M., 1998. Infection function in experimentally infected cattle. Vet. Immunol. Immunopathol. 36, of rabbits with R29 strain of bovine immunodeficiency virus: virulence, 91–105. immunosuppression, and progressive mesenteric lymphadenopathy. Viral Flaming, K.P., Frank, D.E., Carpenter, S., Roth, J.A., 1997. Longitudinal studies Immunol. 11, 159–166. of immune function in cattle experimentally infected with bovine Kirchhoff, F., Jentsch, K.D., Bachmann, B., Stuke, A., Laloux, C., Luke, W., immunodeficiency-like virus and/or bovine leukemia virus. Vet. Immunol. Stahl-Hennig, C., Schneider, J., Nieselt, K., Eigen, M., et al., 1990. A novel Immunopathol. 56, 27–38. proviral clone of HIV-2: biological and phylogenetic relationship to other Fomsgaard, A., Hirsch, V.M., Allan, J.S., Johnson, P.R., 1991. A highly primate immunodeficiency viruses. Virology 177, 305–311. divergent proviral DNA clone of SIV from a distinct species of African Kohlstaedt, L.A., Wang, J., Friedman, J.M., Rice, P.A., Steitz, T.A., 1992. green monkey. Virology 182, 397–402. Crystal structure at 3.5 A resolution of HIV-1 reverse transcriptase Forman, A.J., Gibson, C.A., Rodwell, B.J., 1992. Serological evidence for the complexed with an inhibitor. Science 256, 1783–1790. presence of bovine lentivirus infection in cattle in Australia. Aust. Vet. J. 69, Koppe, B., Menendez-Arias, L., Oroszlan, S., 1994. Expression and purification 337. of the mouse mammary tumor virus gag-pro transframe protein p30 and Gao, H.Q., Boyer, P.L., Arnold, E., Hughes, S.H., 1998. Effects of mutations in characterization of its dUTPase activity. J. Virol. 68, 2313–2319. the polymerase domain on the polymerase, RNase H and strand transfer Le Grice, S.F., Gruninger-Leitch, F., 1990. Rapid purification of homodimer and activities of human immunodeficiency virus type 1 reverse transcriptase. heterodimer HIV-1 reverse transcriptase by metal chelate affinity chroma- J. Mol. Biol. 277, 559–572. tography. Eur. J. Biochem. 187, 307–314. Gao, H.Q., Sarafianos, S.G., Arnold, E., Hughes, S.H., 1999. Similarities and Mang, R., Maas, J., Chen, X., Goudsmit, J., van Der Kuyl, A.C., 2001. differences in the RNase H activities of human immunodeficiency virus type Identification of a novel type C porcine endogenous retrovirus: evidence that 1 reverse transcriptase and Moloney murine leukemia virus reverse copy number of endogenous retroviruses increases during host inbreeding. transcriptase. J. Mol. Biol. 294, 1097–1113. J. Gen. Virol. 82, 1829–1834. Garvey, K.J., Oberste, M.S., Elser, J.E., Braun, M.J., Gonda, M.A., 1990. Martoglio, B., Graf, R., Dobberstein, B., 1997. Signal peptide fragments of Nucleotide sequence and genome organization of biologically active preprolactin and HIV-1 p-gp160 interact with calmodulin. EMBO J. 16, proviruses of the bovine immunodeficiency-like virus. Virology 175, 6636–6645. 391–409. McNab, W.B., Jacobs, R.M., Smith, H.E., 1994. A serological survey for bovine Gonda, M.A., 1994. Molecular biology and virus–host interactions of immunodeficiency-like virus in Ontario dairy cattle and associations lentiviruses. Ann. N. Y. Acad. Sci. 724, 22–42. between test results, production records and management practices. Can. Gonda, M.A., Braun, M.J., Carter, S.G., Kost, T.A., Jr., J.W., Arthur, L.O., Van J. Vet. Res. 58, 36–41. der Maaten, M.J., 1987. Characterization and molecular cloning of a bovine Meas, S., Ohashi, K., Sugimoto, C., Onuma, M., 2001. Phylogenetic relation- lentivirus related to human immunodeficiency virus. Nature 330, 388–391. ships of bovine immunodeficiency virus in cattle and buffaloes based on Gonda, M.A., Luther, D.G., Fong, S.E., Tobin, G.J., 1994. Bovine immuno- surface envelope gene sequences. Brief report. Arch. Virol. 146, 1037–1045. deficiency virus: molecular biology and virus–host interactions. Virus Res. Narayan, O., Clements, J.E., 1989. Biology and pathogenesis of lentiviruses. J. 32, 155–181. Gen. Virol. 70, 1617–1639. Hayashi, T., Makino, K., Ohnishi, M., Kurokawa, K., Ishii, K., Yokoyama, K., Oberste, M.S., Gonda, M.A., 1992. Conservation of amino-acid sequence motifs Han, C.G., Ohtsubo, E., Nakayama, K., Murata, T., Tanaka, M., Tobe, T., in lentivirus Vif proteins. Virus Genes. 6, 95–102. Iida, T., Takami, H., Honda, T., Sasakawa, C., Ogasawara, N., Yasunaga, T., Oberste, M.S., Greenwood, J.D., Gonda, M.A., 1991. Analysis of the Kuhara, S., Shiba, T., Hattori, M., Shinagawa, H., 2001. Complete genome transcription pattern and mapping of the putative rev and env splice sequence of enterohemorrhagic Escherichia coli O157:H7 and genomic junctions of bovine immunodeficiency-like virus. J. Virol. 65, 3932–3937. comparison with a laboratory strain K-12. DNA Res. 8, 11–22. Oberste, M.S., Williamson, J.C., Greenwood, J.D., Nagashima, K., Copeland, Hizi, A., Hughes, S.H., 1988. Expression in Escherichia coli of a Moloney T.D., Gonda, M.A., 1993. Characterization of bovine immunodeficiency murine leukemia virus reverse transcriptase whose structure closely virus rev cDNAs and identification and subcellular localization of the resembles the viral enzyme. Gene. 66, 319–323. Rev protein. J. Virol. 67, 6395–6405. Hizi, A., Joklik, W.K., 1977. RNA-dependent DNA polymerase of avian Oda, T., Ikeda, S., Watanabe, S., Hatsushika, M., Akiyama, K., Mitsunobu, F., sarcoma virus B77: I. Isolation and partial characterization of the alpha, 1988. Molecular cloning, complete nucleotide sequence, and gene structure beta2, and alphabeta forms of the enzyme. J. Biol. Chem. 252, 2281–2289. of the provirus genome of a retrovirus produced in a human lymphoblastoid Hizi, A., Henderson, L.E., Copeland, T.D., Sowder, R.C., Hixson, C.V., cell line. Virology 167, 468–476. Oroszlan, S., 1987. Characterization of mouse mammary tumor virus gag- Olmsted, R.A., Barnes, A.K., Yamamoto, J.K., Hirsch, V.M., Purcell, R.H., O. Avidan et al. / Virology 351 (2006) 42–57 57

Johnson, P.R., 1989. Molecular cloning of feline immunodeficiency virus. Snider III, T.G., Luther, D.G., Jenny, B.F., Hoyt, P.G., Battles, J.K., Ennis, W.H., Proc. Natl. Acad. Sci. U.S.A. 86, 2448–2452. Balady, J., Blas-Machado, U., Lemarchand, T.X., Gonda, M.A., 1996. Palaniappan, C., Fuentes, G.M., Rodriguez-Rodriguez, L., Fay, P.J., Bambara, Encephalitis, lymphoid tissue depletion and secondary diseases associated R.A., 1996. Helix structure and ends of RNA/DNA hybrids direct the with bovine immunodeficiency virus in a dairy herd. Comp. Immunol. cleavage specificity of HIV-1 reverse transcriptase RNase H. J. Biol. Chem. Microbiol. Infect. Dis. 19, 117–131. 271, 2063–2070. Takehisa, J., Harada, Y., Ndembi, N., Mboudjeka, I., Taniguchi, Y., Ngansop, C., Peliska, J.A., Benkovic, S.J., 1992. Mechanism of DNA strand transfer reactions Kuate, S., Zekeng, L., Ibuki, K., Shimada, T., Bikandou, B., Yamaguchi- catalyzed by HIV-1 reverse transcriptase. Science 258, 1112–1118. Kabata, Y., Miura, T., Ikeda, M., Ichimura, H., Kaptue, L., Hayami, M., Perach, M., Hizi, A., 1999. Catalytic features of the recombinant reverse 2001. Natural infection of wild-born mandrills (Mandrillus sphinx) with two transcriptase of bovine leukemia virus expressed in bacteria. Virology 259, different types of simian immunodeficiency virus. AIDS Res. Hum. 176–189. Retroviruses. 17, 1143–1154. Perrino, F.W., Preston, B.D., Sandell, L.L., Loeb, L.A., 1989. Extension of Taube, R., Avidan, O., Bakhanashvili, M., Hizi, A., 1998a. DNA synthesis mismatched 3′ termini of DNA is a major determinant of the infidelity of exhibited by the reverse transcriptase of mouse mammary tumor virus: human immunodeficiency virus type 1 reverse transcriptase. Proc. Natl. processivity and fidelity of misinsertion and mispair extension. Eur. J. Acad. Sci. U.S.A. 86, 8343–8347. Biochem. 258, 1032–1039. Rausch, J.W., Grice, M.K., Henrietta, M., Nymark, M., Miller, J.T., Le Grice, Taube, R., Loya, S., Avidan, O., Perach, M., Hizi, A., 1998b. Reverse S.F., 2000. Interaction of p55 reverse transcriptase from the Saccharo- transcriptase of mouse mammary tumour virus: expression in bacteria, myces cerevisiae retrotransposon Ty3 with conformationally distinct purification and biochemical characterization. Biochem. J. 332, nucleic acid duplexes. J. Biol. Chem. 275, 13879–13887. 807–808. Rubinek, T., Loya, S., Shaharabany, M., Hughes, S.H., Clark, P.K., Hizi, A., Thompson, J.D., Higgins, D.G., Gibson, T.J., 1994. CLUSTAL W: improving 1994. The catalytic properties of the reverse transcriptase of the lentivirus the sensitivity of progressive multiple sequence alignment through sequence equine infectious anemia virus. Eur. J. Biochem. 219, 977–983. weighting, position-specific gap penalties and weight matrix choice. Nucleic Rubinek, T., Bakhanashvili, M., Taube, R., Avidan, O., Hizi, A., 1997. The Acids Res. 22, 4673–4680. fidelity of 3′ misinsertion and mispair extension during DNA synthesis Van der Maaten, M.J., Malmquist, W.A., Cheville, N.F., 1972. Susceptibility of exhibited by two drug-resistant mutants of the reverse transcriptase of calves to bovine syncytial virus given by different inoculation routes. Am. J. human immunodeficiency virus type 1 with Leu74->Val and Glu89->Gly. Vet. Res. 33, 1157–1160. Eur. J. Biochem. 247, 238–247. Von Hippel, P.H., Fairfield, F.R., Dolejsi, M.K., 1994. On the processivity of Sevilya, Z., Loya, S., Hughes, S.H., Hizi, A., 2001. The ribonuclease H activity polymerases. Ann. N. Y. Acad. Sci. 726, 118–131. of the reverse transcriptases of human immunodeficiency viruses type 1 and Wisniewski, M., Balakrishnan, M., Palaniappan, C., Fay, P.J., Bambara, R.A., type 2 is affected by the thumb subdomain of the small protein subunits. J. 2000. Unique progressive cleavage mechanism of HIV reverse transcriptase Mol. Biol. 311, 957–971. RNase H. Proc. Natl. Acad. Sci. U.S.A. 97, 11978–11983. Sevilya, Z., Loya, S., Adir, N., Hizi, A., 2003. The ribonuclease H activity of the York, D.F., Vigne, R., Verwoerd, D.W., Querat, G., 1992. Nucleotide sequence reverse transcriptase of human immunodeficiency viruses type 1 and type 2 of the jaagsiekte retrovirus, an exogenous and endogenous type D and B is modulated by residue 294 of the small subunit. Nucleic Acids Res. 31, retrovirus of sheep and goats. J. Virol. 66, 4930–4939. 1481–1487. Yu, H., Goodman, M.F., 1992. Comparison of HIV-1 and avian myeloblastosis Skalka, A.M., Goff, S.P., 1993. Reverse Transcriptase. Cold Spring Harbor virus reverse transcriptase fidelity on RNA and DNA templates. J. Biol. Laboratory Press, Cold Spring Harbor, NY. Chem. 267, 10888–10896.