Mapping of the Tacaribe Arenavirus Z-Protein Binding Sites on the L Protein Identified both Amino Acids within the Putative Polymerase Domain and a Region at the N Terminus of L That Are Critically Involved in Binding^▿

Cells, virus, and sera.

CV1 cells were grown in Dulbecco's modified Eagle's medium supplemented with 5% fetal calf serum (Invitrogen, Carlsbad, CA). Recombinant vaccinia virus vTF7-3, which expresses the T7 RNA polymerase, was kindly provided by B. Moss (National Institutes of Health, Bethesda, MD) (15). Monospecific sera against the recombinant TacV L were raised in rabbits as indicated before (22). Serum against glutathione S-transferase (GST) was obtained from GE Healthcare.

Plasmids.

Plasmids pL, pgst, and pgstZ, expressing, respectively, the TacV L protein, the glutathione S-transferase (GST) protein, and the GST-tagged Z protein (gstZ), were described previously (18, 22).

Plasmids expressing the TacV L C-terminal deletion mutants L 1-1816 and L 1-1444 were generated by digestion with SmaI at the polylinker sequence together with either SnaBI (at nucleotide [nt] 1620 of the L genome) or PvuII (at nt 2740), respectively. For construction of the plasmid expressing L 1-1270, pL was digested with EcoRI at nt 3266 and at the polylinker sequence. In each case, the DNA fragment containing the truncated L gene was purified and religated, the stop codon being provided by the polylinker sequence of pL. To generate the plasmid expressing mutant L 1-1666, a PCR fragment was obtained from pL using a forward primer complementary to nt 2656 to 2636 of the L genome covering an SspI site and a reverse primer containing (5′ to 3′) an SmaI site followed by a stop codon and a sequence covering nt 2078 to 2093 of the L genome. The amplified fragment was digested with SspI and SmaI and inserted into pL digested with the same enzymes. Plasmids expressing the N-terminally truncated mutants L 156-2210 and L 292-2210 were constructed by the following procedure. PCR fragments were generated from pL using forward primers containing (5′ to 3′) a BamHI site followed by a start codon and a sequence complementary to nt 6604 to 6587 or to nt 6199 to 6188 of the L genome, respectively. In both cases, the reverse primer contained nt 6087 to 6101 of the L genome, including the StuI unique site. The amplification fragments were digested with BamHI and StuI, and each was inserted into pL, replacing the wild-type (wt) BamHI-StuI fragment. For the plasmid expressing mutant L 912-2210, the PCR fragment was generated using a forward primer containing a sequence complementary to nt 4339 to 4322 of the L genome preceded by (5′ to 3′) a BamHI site followed by a start codon and a reverse primer covering nt 3706 to 3725 of the L genome, including an AflII site. The amplification fragment was digested with BamHI and AflII and inserted into pL digested with the same enzymes.

Single-point mutations on pL were performed using the Quick Change PCR mutagenesis kit (Stratagene) with primers containing the mutated sequences. For construction of the plasmid expressing mutant L C2a (see the legend to Fig. 3), pL was excised with BstEII at nt 2404 and 3526 and the restriction fragment (encoding amino acids [aa] 1182 to 1556) was replaced by the corresponding BstEII-BstEII fragment carrying the substituted D1329 codon.

All plasmids were purified by Qiagen tip-100 (Qiagen Inc.,Valencia, CA). The introduced mutations, reversions, and inserted sequences were confirmed by automated dideoxynucleotide DNA sequencing (Macrogen, Korea).

Analysis of protein interactions by coimmunoprecipitation.

Subconfluent monolayers of CV1 cells were infected with 3 to 5 PFU of vTF7-3 per cell and transfected using LipofectAMINE 2000 reagent (Invitrogen) as described before (18). The amount of plasmids added to approximately 4 × 10⁵ cells was 100 ng of either L or each of the TacV L mutants and 200 ng of the plasmids expressing gstZ or GST. Labeling of the cells for 1 h at 5 h posttransfection with the addition of [³⁵S]methionine-cysteine mix (150 μCi/ml; NEG 772, NEN PerkinElmer) and preparation of cell lysates were performed as previously described (18). Aliquots corresponding to about 0.5 × 10⁵ to 1 × 10⁵ cells were immunoprecipitated with sera to L or to GST, and the immunoprecipitated proteins were resolved by two-step sodium dodecyl sulfate-polyacrylamide gel electrophoresis containing 8% and 12% polyacrylamide and visualized by autoradiography as indicated before (18). Quantification of the bands was performed on a PhosphorImager (Molecular Dynamics). Z binding refers to the amount of L, L mutants, and revertants coimmunoprecipitating with gstZ using serum to GST and normalized for protein expression determined by immunoprecipitation with serum to L. Z binding to L was taken as 100%.

Effect of C-terminal deletions of the L protein in the interaction with Z.

We started mapping the region(s) of TacV L involved in the interaction with Z by constructing mutants with a series of truncations in the C-terminal region. Mutants L 1-1816, L 1-1666, L 1-1444, and L 1-1270 had deletions of 394, 544, 766, and 940 aa, respectively, from the C-terminal end of L (Fig. 1A). To test the ability of the truncated proteins to interact with Z, cells were cotransfected with plasmids expressing gstZ and either L or each of the L mutant proteins. The cells were then labeled and lysed, and aliquots of the cell lysate were immunoprecipitated with monospecific serum against either GST or L. As a control, a similar procedure was performed with extracts from cells coexpressing GST and L. When the proteins immunoprecipitated with serum to L were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, the results showed that each of the L mutants was synthesized at levels similar to wt L, each mutant exhibiting an appropriately increased mobility with the increasing size of the truncation (Fig. 1B, lower panel). Analysis of the proteins immunoprecipitated with serum to GST (upper panel) confirmed previous findings indicating that the interaction with L is mediated through the Z portion of the fusion protein (18) as L coimmunoprecipitated with gstZ but not with GST (lanes 1 and 6). The results also showed that deletion of the last 394 aa had almost no effect on binding as coimmunoprecipitation of L 1-1816 was more than 90% that of wt L. Even deletion of the last 544 aa in L 1-1666 gave nearly wt levels of binding (Fig. 1B, lanes 1, 2, and 3, and C). When deletion was extended to residue 1444, within domain III of L and approximately 50 aa downstream from motif E (see panel A), the resulting mutant protein (L 1-1444) retained only 12% the binding ability of L, whereas a further truncation to residue 1270, some 10 aa upstream from motif B, completely abolished the capacity of L to interact with Z (Fig. 1B, lanes 4 and 5, and C). It is worth noting that a similar pattern of interaction was observed when cell lysates were immunoprecipitated with serum to L (lower panel).

We also tested whether L proteins with deletions at the C-terminal end retained polymerase activity. For this, we used a previously described transcription and replication reconstituted system consisting of cotransfection of CV1 cells with plasmids expressing L, N, and a TacV RNA analog expressing the reporter gene chloramphenicol acetyltransferase (CAT) (22). We did not detect CAT activity when in the reconstituted system L was replaced by each of the C-terminally truncated L proteins (not shown). Thus, although the last C-terminal 544 aa of L were almost dispensable for its binding ability, they were essential for polymerase activity.

Identification of amino acids critical for binding Z within domain III of L.

The above results indicating that C-terminal truncations extending within domain III of L impaired its ability for interacting with Z (Fig. 1), might suggest that residues within this domain are critically involved in binding. Alternatively, extensive C-terminal deletions might have caused conformational changes that altered the actual Z-binding sites at another region(s) of L. In favor of the first alternative was the observation that substitution of sequence motifs in domain III of TacV L by the homologous sequence motifs of the phylogenetically close Rift Valley fever virus L (33) abolished the capacity to interact with Z, although the TacV L mutants generated were full length (M. Wilda, unpublished).This prompted us to investigate an eventual role of domain III in L-Z interaction.

As domain III of TacV L is more than 500 aa in length, containing long stretches of sequences that are virtually completely conserved among all arenaviruses, we faced the challenge of selecting residues for site-directed mutagenesis. We started modifying the invariant aspartate (D) 1188, around which is centered sequence motif A (Fig. 2A) (16, 23, 25, 26). We first replaced D1188 with G (mutant L A1) with the idea that the introduction of a residue that might affect the local environment in the region would help to identify sequences playing some role in binding. When lysates from cells coexpressing L A1 and gstZ were immunoprecipitated with serum to GST and the proteins were resolved by gel electrophoresis, it was found that coimmunoprecipitation of L A1 was severely affected, indicating that replacement of D1188 by G disrupted binding (Fig. 2B, compare lanes 1 and 2, and C). Low binding cannot be ascribed to instability of the protein as L A1 was full length and small differences in expression were normalized in order to estimate Z binding (see Materials and Methods). Moreover, as a control for the mutagenesis procedure, two independent L A1 clones were changed back to the wt amino acid and when the revertants (designated L A1Ra and L A1Rb) were tested for their binding ability to Z, each gave wt levels of binding (lanes 3 and 4). For a better understanding of the role of D1188 in binding, we engineered two other mutants in which either Q or A was substituted for D1188 (mutants L A2 and L A3, respectively). Each of these substitutions, however, had almost no adverse effect on binding (Fig. 2B, lanes 5 and 6, and C). Within motif A, we then selected histidine 1189 for site-directed mutagenesis as this amino acid is invariant in the L proteins of arenaviruses and changed this residue to five different amino acids (Fig. 2). When either G or T was substituted for H1189 (mutants L A4 and L A5, respectively) and the proteins were analyzed for their binding ability, each of these gave very low levels of interaction (Fig. 2B, compare lanes 9 and 11 with lane 8, and C). With the idea of exerting little effect in protein conformation (10, 12), we then replaced H1189 with A (mutant L A6). This change, however, also drastically impaired complex formation (Fig. 2B, compare lanes 14 and 15, and C). Finally, substitution of either K or R for H1189 resulted in five- to fourfold decrease in binding (mutants L A7 and L A8, Fig. 2B, lanes 17 and 21, and C). To control that low binding of the mutants was the consequence of amino acid substitutions, mutants L A4, L A5, L A6, L A7, and L A8 were changed back to the wt amino acid, and when the revertants were tested for their binding activity, each of them gave wt levels of binding (Fig. 2B, lanes 10, 12, 16, 18, and 22).

On the bases of sequence alignments, Muller et al. (23) transferred the structural information from human immunodeficiency virus reverse transcriptase to the putative polymerase domain of the negative-stranded RNA virus RdRps localizing the invariant D residues of motifs A and C within the palm region and close to each other. Considering that in TacV L the substitution of G for the invariant D1188 of motif A severely affected interaction with Z (Fig. 2), we expected to identify residues important for binding within motif C.

Motif C exhibits the SDD sequence conserved in segmented negative-stranded virus RdRps (Fig. 2A). The first D residue of the triplet (D1329) is strictly conserved in all classes of polymerases, whereas the second D is somewhat flexible as, i.e., it is replaced by N in nonsegmented negative-stranded virus RdRps (23). To identify residues within motif C playing some role in binding, we replaced each residue of the SDD triplet (positions 1328, 1329, and 1330) with A and the mutants generated (named L C1, L C2, and L C3, respectively) were analyzed for their binding ability. As seen in Fig. 3, substitution of A for S1328 (mutant L C1) or D1330 (mutant L C3) gave nearly wt levels of binding (Fig. 3A, compare lanes 2 and 7 with lane 1), whereas change of the invariant D1329 to A (mutant L C2) disrupted the interaction (compare lanes 1 and 3 in Fig. 3A). To control for whether or not the experimental approach used for the mutagenesis might have introduced systematic unnoticed mutations in L C2 that could affect its binding ability, we used a second experimental procedure to perform the substitution of A for D1329 (see Materials and Methods). The resulting mutant (named L C2a) also failed to interact with Z (compare lanes 1 and 5). Furthermore, both L C2 and L C2a were changed back to the wt amino acid and each of the revertants exhibited wt levels of binding (lanes 4 and 6). These controls confirmed that substitution of A for the invariant D1329 is responsible for the lack of binding ability of L.

We also tested whether mutations in motifs A and C affected polymerase activity. Using the TacV reconstituted system, we did not detect CAT expression above background levels when in the assay L was replaced by each of the L mutants depicted in Fig. 2 and 3 (not shown). These results confirmed and extended published data indicating that mutations in the invariant D of motif A or in the SDD triplet in motif C of LCMV-L generated nonfunctional proteins (28). A novel result is that H1189 in motif A is also strictly required for polymerase activity.

Deletion of the first 292 residues at the N terminus of L disrupts the interaction with Z.

Although we have identified residues within domain III of L that are critical for interaction with Z, we did not detect binding with mutant L 912-2210, which expressed the entire domain III plus the C-terminal region of L (Fig. 4A, B, lanes 1 and 2, and C). To define additional sequences that may be required for interaction, we engineered other N-terminal deletion mutants, of which L 156-2210 and L 292-2210 were stable (Fig. 4A and B, lanes 5 and 6, lower panel) and could be used to map the N terminus of L for binding. The results revealed that truncation of the first 155 aa did not significantly alter binding, whereas deletion of an additional 136 aa (mutant L 292-2210) disrupted the interaction (Fig. 4B, compare lanes 5 and 6 with lane 4, and C). These results led to identification of another region between residues 156 and 292 of L that is essential for binding Z.

It is noteworthy that, like the L mutants with C-terminal deletions, none of the N-terminally truncated mutants exhibited polymerase activity when analyzed in the TacV reconstituted system (not shown). These results were not surprising since, at least in Sendai virus, L polymerase deletion of short sequences from the N or C termini abolish the functional activity of the protein (6, 30).

To the best of our knowledge, this is the first report identifying Z-protein binding sites on an arenavirus L polymerase. We have shown that the C-terminal region of L, including domain IV, seemed to be dispensable for the interaction and defined two noncontiguous regions that are critical for binding. The importance of domain III in binding was revealed by point mutations in sequence motifs A and C; each of these centered around an aspartic acid invariant in all polymerases. Within motif A, we changed both the invariant D1188 and H1189, the latter strictly conserved in arenavirus L proteins. Of the three substitutions for D1188, only change of D1188 to G impaired binding, whereas both the changes of D1188 to Q, with an uncharged polar side chain, and A, with a short hydrophobic side chain, were tolerated. On the contrary, substitution of either G, T, or A for H1189 reduced binding to 1% to 5% the wt level, and even the change of H1189 to K or R, with positively charged side chains, gave a five- to fourfold decrease in binding. Taken together, these results suggested that H1189 within motif A is critically important for the interaction with Z. We also performed Ala substitutions for each residue of the SDD sequence of motif C, finding that only the invariant first D of the triplet (D1329) appeared to be critical as its replacement with A completely abolished interaction, whereas the change of either S1328 or D1330 to A gave nearly wt levels of binding. Since motifs A and C have been proposed to be part of the catalytic site of the RdRps (23, 25, 26, 29, 31), it is a relevant finding that each of these motifs includes at least one amino acid critically involved in binding the inhibitory Z protein. In all probability, other residues in domain III play some role in binding as well. We are attempting their identification by means of Ala substitutions in conserved motifs within this domain.

Deletion analysis at the N terminus of L led to the definition of another region essential for binding Z between residues 156 and 292. Sequence alignments of this region between arenavirus L proteins revealed a low degree of homology (22%) and 10 invariant amino acids, which represent potential targets for mutagenesis and binding analysis.

Our finding that modifications in either domain III or in the N terminus of L impaired or ablated the interaction with Z could be interpreted to mean that both regions are required in concert for binding. Another possible explanation could be that one of the regions may contribute indirectly to the interaction: i.e., if integrity of the N terminus of L is essential for the overall conformation of the protein. In any case, it should be considered that the L proteins of a number of negative-stranded RNA viruses, including the arenavirus LCMV, may act as a functional oligomer (see references 6 and the references therein and 28) and Z protein might require the oligomeric form of L for binding. It is noteworthy that the L-L interaction site on the L proteins of several nonsegmented negative-stranded RNA viruses maps at the N terminus of the proteins (see reference 6 and references therein). In addition, for L oligomerization (6, 28), as for L-Z binding (18), coexpression of the proteins is required. Taking these data together, it is tempting to speculate that modifications in the N terminus of L might prevent L oligomerization and in turn Z binding, as well. Efforts are under way to introduce modifications in the N-terminal region of L to map in detail the Z-binding site and to define the eventual involvement of this region in L oligomerization.