The N Terminus of Andes Virus L Protein Suppresses mRNA and Protein Expression in Mammalian Cells

Abstract

Little is known about the structure and function of the 250-kDa L protein of hantaviruses, although it plays a central role in virus genome transcription and replication. When attempting to study Andes virus (ANDV) L protein in mammalian cells, we encountered difficulties. Even in a strong overexpression system, ANDV L protein could not be detected by immunoblotting. Deletion analysis revealed that the 534 N-terminal amino acid residues determine the low-expression phenotype. Inhibition of translation due to RNA secondary structures around the start codon, rapid proteasomal degradation, and reduced half-life time were excluded. However, ANDV L protein expression could be rescued upon mutation of the catalytic PD-E-K motif and further conserved residues of the putative endonuclease at the N terminus of the protein. In addition, wild-type ANDV L rather than expressible L mutants suppressed the level of L mRNA, as well as reporter mRNAs. Wild-type L protein also reduced the synthesis of cellular proteins in the high-molecular-weight range. Using expressible ANDV L mutants as a tool for localization studies, we show that L protein colocalizes with ANDV N and NSs but not Gc protein. A fraction of L protein also colocalized with the cellular processing (P) body component DCP1a. Overall, these data suggest that ANDV L protein possesses a highly active endonuclease at the N terminus suppressing the level of its own as well as heterologous mRNAs upon recombinant expression in mammalian cells.

INTRODUCTION

Andes virus (ANDV) is a member of the genus Hantavirus within the family Bunyaviridae. The main reservoir host for ANDV is the long-tailed pygmy rice rat (Oligoryzomys longicaudatus). Upon transmission to humans, ANDV causes a respiratory disease called hantavirus cardiopulmonary syndrome (HCPS) that is associated with a case fatality of up to 40%. ANDV is the only hantavirus that is further transmissible from human to human. The virus is endemic in Argentina and Chile, but related HCPS-causing hantaviruses are found throughout the Americas (1, 2).

The hantavirus genome consists of three antisense RNA segments designated L (large), M (medium), and S (small) (3). The L segment encodes the 250-kDa L protein, which presumably possesses several enzymatic functions involved in RNA transcription and replication (4, 5). Expression of the M segment results in the glycoprotein precursor (GPC), which is cleaved posttranslationally into the envelope proteins Gn and Gc. The S segment encodes the nucleoprotein (N), which encapsidates the viral genome. Some hantaviruses, including ANDV, additionally encode on the S segment a nonstructural protein (NSs), which may play a role in modulating the interferon response of the infected cell (6, 7).

The hantavirus L protein has been poorly characterized thus far. A 250-kDa L protein was detected in Tula hantavirus-infected cells, indicating that the majority of the protein is not processed posttranslationally (5). Homologies to polymerase proteins of other RNA viruses revealed a RNA-dependent RNA polymerase (RdRp) domain in the center of the protein (4). Hantavirus mRNAs contain heterologous nucleotides at the 5′ end, suggesting the virus removes cap structures from cellular mRNAs to prime its own mRNA synthesis, a process called cap-snatching (8, 9). The prototype of cap-snatching viruses is influenza A virus (IAV; family Orthomyxoviridae), which employs an RNA endonuclease in the N terminus of PA protein for cleavage (10, 11). Recently, endonucleases were also identified in the N termini of the L proteins of La Crosse virus (LACV; genus Orthobunyavirus, family Bunyaviridae) and lymphocytic choriomeningitis virus (family Arenaviridae) (12, 13). An endonuclease motif is also found in the N terminus of hantavirus L protein, suggesting that it harbors an endonuclease as well (4, 12, 13). In addition, it is assumed that the N protein of hantaviruses is involved in cap-snatching due to its ability to bind cap structures and its association with processing (P) bodies (14, 15). P bodies play a role in cellular mRNA turnover (16, 17), and there is evidence that N protein binds to and protects 5′-terminal nucleotides of capped mRNA from degradation in P bodies (14). These capped RNA fragments may be transferred to L protein, further trimmed in length by the endonuclease and used by the RdRp to initiate viral mRNA synthesis.

Here, we studied the recombinant expression and subcellular localization of ANDV L protein in mammalian cells. Like other researchers (18), we encountered difficulties in expressing hantavirus L protein in these cells. Several mechanisms possibly underlying the low-expression phenotype of L protein were experimentally approached. By mutagenesis, we were able to overcome the low-expression phenotype of L protein and map regions and residues determining it. ANDV L mutants showing a high-expression phenotype were used as a tool to study the localization of the protein in the cell.

MATERIALS AND METHODS

Construction of plasmids for protein expression.

Vero cells in 75-cm² tissue culture flasks were inoculated with ANDV strain Chile-9717869 (kindly provided by Heinz Feldmann, Winnipeg, Canada), Sin Nombre virus (SNV) strain Convict Creek 107, Hantaan virus (HTNV) strain 76-118, or Puumala virus (PUUV) strain Sotkamo. After 8 to 10 days, the supernatant was cleared by low-speed centrifugation, and virus was pelleted from the cleared material by ultracentrifugation. The pellet was resuspended in water, and virus RNA was purified by using TRIzol LS Reagent (Invitrogen). Purified RNA was reverse transcribed by using Superscript III reverse transcriptase (Invitrogen) with virus-specific primers, and the cDNA was amplified with the Phusion High-Fidelity DNA polymerase (New England BioLabs). ANDV genes were cloned into expression vector pCITE-2a (Novagen) containing a T7 RNA polymerase promoter, an internal ribosomal entry site (IRES), and a favorable Kozak sequence (pCITE-ANDV-L, pCITE-ANDV-N, pCITE-ANDV-GPC, and pCITE-ANDV-NSs). For immunoblotting and immunoprecipitation studies, hemagglutinin (HA), or FLAG tags were fused to the C termini of the genes. ANDV, HTNV, PUUV, and SNV L gene fragments (amino acid residues 1 to 534) were cloned into pCITE-2a and fused to a C-terminal HA tag (pCITE-L_N constructs). The sequence of cloned genes corresponded to published sequences (GenBank accession no. for ANDV L, AF291704; ANDV GPC, AF291703; ANDV N, AF291702; HTNV L, X55901; PUUV L, Z66548; and SNV L, AF425256) except the following differences which have been confirmed by direct sequencing of our virus isolates (numbering corresponds to nucleotide and amino acid position in the gene): ANDV L, G541A leading to amino acid exchange V181I; ANDV GPC, C2580T silent mutation and G3405T leading to K1135N; HTNV L, C47A leading to P16H; and SNV L, C124A leading to L42M and G6343A leading to V2115I. The construct for expression of Lassa virus L protein (pCITE-Las-L) has been described previously (19).

L protein mutants were generated via a classical two-step PCR mutagenesis approach as described previously (20) or QuikChange mutagenesis (Stratagene) using pCITE-L or L_N as a template. The resulting fragments were cloned or used for transfection without prior cloning as described previously (20). Codon-modified L gene fragments (nucleotide positions 1 to 135) were chemically synthesized (GeneArt) and exchanged with the corresponding wild-type fragment in pCITE-ANDV-L_N. The presence of a fusion, deletion, or artificial point mutation was ascertained by sequencing.

Cells and transfections.

BSR-T7/5 cells stably expressing T7 RNA polymerase (21) were grown in Glasgow's minimal essential medium (Gibco) supplemented with 5% fetal calf serum (FCS). Every second passage, 1 mg of Geneticin (Gibco) per ml of medium was added to the cells. BHK-21 cells were grown in Dulbecco modified Eagle medium (DMEM; PAA Laboratories) supplemented with 10% FCS. Cells were seeded at a density of 1 × 10⁵ or 5 × 10⁵ cells per well of a 24- or 6-well plate, respectively, at 1 day prior to transfection. Cells were inoculated with modified vaccinia virus Ankara expressing T7 RNA polymerase (MVA-T7) (22) for 1 h before transfection. All transfections were performed with 3 μl of Lipofectamine 2000 (Invitrogen) per μg of DNA in cell culture medium without supplements. Medium was replaced at 4 h after transfection by fresh medium complemented with FCS.

Immunoblot analysis.

BSR-T7/5 or BHK-21 cells in a well of a 24-well plate were infected with MVA-T7 and transfected with 250 to 500 ng of expression construct (plasmids or PCR products). One day after transfection, cells were lysed in 100 μl of 1× NuPAGE LDS sample buffer (Invitrogen) complemented with dithiothreitol (DTT) and then heated to 95°C for 5 min. Proteins were separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to nitrocellulose membrane (Carl Roth). Membranes were blocked in 1× Roti-Block (Carl Roth) for 1 h and incubated with rabbit anti-HA (1:10,000; H6908; Sigma-Aldrich) for 1 h and peroxidase-conjugated secondary antibody (1:50,000; 111-035-003; Jackson ImmunoResearch) in phosphate-buffered saline (PBS)–0.2× Roti-Block at room temperature for 1 h. After washing, protein bands were visualized by chemiluminescence using SuperSignal West Pico substrate (Pierce) and X-ray film (Kodak) or the Fusion SL image acquisition system (Vilber Lourmat).

Confocal immunofluorescence microscopy.

BHK-21 cells were grown on coverslips in 24-well plates, infected with MVA-T7, and transfected with 50 to 100 ng of plasmid DNA per well. Cells were washed with PBS at 10 h after transfection and fixed with 4% paraformaldehyde–4% sucrose in PBS for 15 min. Cells were permeabilized with 0.2% saponin in PBS for 10 min, washed with PBS, blocked with 1% bovine serum albumin (BSA) in PBS at 37°C for 30 min, and incubated with mouse monoclonal anti-HA F-7 (1:100; sc-7392; Santa Cruz), rabbit anti-FLAG (1:2,000; A190-102A; Bethyl), mouse monoclonal anti-FLAG M2 (1:100; F1804; Sigma-Aldrich), rabbit anti-HA (1:250; H6908; Sigma-Aldrich), mouse monoclonal anti-DCP1a (1:100; ab57654; Abcam), and/or rabbit anti-ANDV-N serum (1:500) in 0.2% BSA–0.2% saponin in PBS at 37°C for 1 h. After the cells were washed, they were incubated with 4′,6-diamidino-2-phenylindole (DAPI; 1:10,000; Boehringer Ingelheim), anti-mouse IgG coupled to rhodamine (1:100; 115-025-072; Jackson ImmunoResearch), and anti-rabbit IgG coupled to fluorescein isothiocyanate (1:250; PA1104; SIFIN) in 0.2% BSA–0.2% saponin in PBS at 37°C for 1 h. Cells were washed, air dried, and covered with ProLong Gold antifade reagent (Invitrogen). Confocal images were captured at ×100 magnification using a FluoView 1000 microscope (Olympus) and processed with the ImageJ program (http://rsb.info.nih.gov/ij/).

Quantification of mRNA.

BHK-21 cells in a well of a 24-well plate were infected with MVA-T7 and transfected with 250 ng of pCITE-ANDV-L, 10 ng of pCITE-FF expressing firefly luciferase, and 10 ng of pCMV-RenLuc expressing Renilla luciferase. Total RNA was purified with the RNeasy kit including DNase I treatment (Qiagen) and quantified spectrophotometrically. ANDV L mRNA, firefly luciferase mRNA, and Renilla luciferase mRNA were quantified by real-time reverse transcription-PCR (RT-PCR) on the LightCycler 480 II system (Roche) using 3 μl of purified RNA, the QuantiTect SYBR green RT-PCR Kit (Qiagen), and primer combinations ANDV-L-fwd/ANDV-L-rev (CTTGCACCTGGAACGGTATC and TGTCCAGAAGGGCTATTTGG), Firefly-fwd/Firefly-rev (AGATCGTGGATTACGACGCCAG and TGCGTCGAGTTTTCCGGTAAG), and Renilla-fwd/Renilla-rev (GACAAGATCAAGGCCATCGTCC and GCACCATTTTCTCGCCCTCTTC), respectively. The specificity of the reaction was verified by melting-curve analysis. Relative mRNA levels were calculated using standard curves and normalized with the total RNA amount added to the RT-PCR assay.

Proteasome inhibition.

Transfected BHK-21 cells were treated at 4 h posttransfection with 0 to 100 μM of the proteasome inhibitor carbobenzoxy-l-leucyl-l-leucyl-l-leucinal (MG-132; Merck Millipore) diluted in GMEM–10% FCS. One day posttransfection, cells in a well of a 24-well plate were lysed with 100 μl of passive lysis buffer (Promega), and nuclei were pelleted by centrifugation. Proteins in cytoplasmic lysate were separated by SDS-PAGE and detected by immunoblot. Proteasome inhibition was monitored in parallel in a 20S proteasome activity assay: 10 μl of lysate was incubated with 100 μl of substrate solution containing 100 μM N-succinyl-Leu-Leu-Val-Tyr-7-amino-4-methylcoumarin (Suc-LLVY-AMC) for 1 h at 37°C in black microtiter plates. The proteolytic activity was measured by release of the fluorogenic AMC group (excitation 360 nm, emission 460 nm) with a Synergy Reader (BioTek).

Pulse-chase labeling and immunoprecipitation.

BHK-21 cells in a well of a six-well plate were infected with MVA-T7 and transfected with 2.5 μg of pCITE-ANDV-L_N. Cells were first incubated in regular DMEM for 4 h, then in DMEM lacking methionine and cysteine for 1 h, and finally in DMEM containing [³⁵S]methionine and [³⁵S]cysteine (Hartmann Analytic) for 2 h (pulse). The radioactive medium was replaced with regular DMEM, and cells were incubated for an additional 1 and 2 h, respectively (chase). Pulsed and pulse-chased cells were lysed in 500 μl of lysis buffer (50 mM Tris-HCl [pH 7.5], 200 mM NaCl, 1 mM EDTA, 0.5% Triton X-100, 40 μl of complete protease inhibitor cocktail/ml [Roche]). Nuclei were pelleted by centrifugation; 450 μl of supernatant was mixed with 1 μl of anti-HA antibody solution (0.5 to 0.7 mg/ml; H6908; Sigma-Aldrich), and the mixture was incubated at 4°C overnight with gentle agitation. Protein A-Sepharose 4 Fast Flow solution (40 μl, as supplied by GE Healthcare) was added, and the mixture was incubated at 4°C with gentle agitation for 2 h. Sepharose-coupled antibody-protein complexes were precipitated by centrifugation and washed twice with lysis buffer and twice with 50 mM Tris-HCl (pH 7.5). Precipitated proteins or lysate were mixed with 1× NuPAGE LDS sample buffer (Invitrogen) and DTT, heated at 95°C for 5 min, and separated by SDS-PAGE. Radioactive signals were detected with a FLA-7000 phosphorimager (Fuji) and quantified using Multi Gauge software (Fuji). Half-life time was calculated from the signal intensities by fitting an exponential curve to the data using Excel software (Microsoft).

RESULTS

Cloning and expression of ANDV L protein.

Virus RNA was isolated from ANDV-infected cells, converted to cDNA, and amplified by PCR using various combinations of L gene-specific primers. A complete L gene was assembled from several fragments and cloned into an expression vector under the control of T7 RNA polymerase promoter and IRES (pCITE-ANDV-L). A C-terminal HA tag was fused to the L protein to facilitate detection in immunoblots. BSR-T7/5 cells, which constitutively express T7 RNA polymerase, were transfected with pCITE-ANDV-L. To maximize protein expression, the cells were additionally infected with recombinant vaccinia virus expressing T7 RNA polymerase (MVA-T7) before transfection. However, despite several attempts, ANDV L protein could not be detected in immunoblot, while arenavirus L protein (Lassa virus) expressed in parallel was readily detected (Fig. 1B, left). Therefore, we assumed that ANDV L protein might be poorly translated or instable due to specific sequence elements. To identify these elements, ANDV L was truncated at N and C terminus (Fig. 1A). C-terminally truncated ANDV L protein showed a wild-type phenotype, i.e., it was not detectable (Fig. 1B, middle). In contrast, ANDV L protein truncated at the N terminus was clearly detectable (Fig. 1B, right). Removal of the first 12 amino acid residues already led to a measurable increase in expression level, although deletion of 45 or more N-terminal residues resulted in strong signals in immunoblot as observed for the arenavirus L protein. In conclusion, ANDV L protein is poorly expressed after transfection, which appears to be linked to regulatory RNA or protein elements residing in the N-terminal 135 nucleotides or 45 amino acid residues, respectively.

Fig 1.

Expression of L protein in mammalian cells. (A) Schematic representation of constructs for expression of ANDV L proteins with N- and C-terminal deletions. The constructs contain a T7 RNA polymerase promoter (T7p) and IRES to facilitate cap-independent translation. L gene amino acid positions are indicated with the constructs. (B) Immunoblot analysis of wild-type L protein of ANDV (ANDV-L), wild-type L protein of Lassa virus (LASV-L), and deletion mutants of ANDV L protein. BSR-T7/5 cells were infected with MVA-T7 and transfected with expression plasmids or empty vector as a negative control (Mock). L proteins were separated by SDS-PAGE, blotted, and detected with anti-HA antibody via the HA tag.

Potential sequence elements interfering with ANDV L protein translation.

To facilitate mutagenesis and expression analysis, further experiments were conducted using a C-terminally truncated version of ANDV L protein (ANDV L_N, 55 kDa, amino acid residues 1 to 534 plus HA tag), which shares the low-expression phenotype with full-length L protein. In the context of ANDV L_N, we first tested whether the 5′ end of the L gene contains sequence elements, codons, or RNA secondary structures which might reduce translation efficacy (23, 24). To this end, the nucleotide sequence of the L gene (positions 1 to 135) was modified without changing the amino acid sequence. Two different L genes were constructed in which 29 and 35%, respectively, of the 5′-terminal nucleotides were exchanged; one was codon optimized for expression in Homo sapiens cells, and the other was randomly codon modified (Fig. 2A). Potential sequence or RNA secondary structure elements should have been inactivated or at least affected by the high number of nucleotide exchanges. However, no change in protein level was observed when ANDV L was expressed from these constructs (Fig. 2B). Thus, RNA elements interfering with translation are unlikely to account for the low expression level.

Fig 2.

Influence of multiple nucleotide exchanges in the 5′ end of L gene on expression of ANDV L protein. (A) The L gene sequence (nucleotide positions 1 to 135) in the context of ANDV L_N was modified without changing the amino acid sequence. In one construct, codons were optimized for Homo sapiens cells (Mut-1, 29% mutations) and in the other construct, each codon was randomly modified (Mut-2, 35% mutations). (B) BSR-T7/5 cells were infected with MVA-T7 and transfected with expression plasmids or empty vector as a negative control (Mock). The expressible L mutant ΔN-45 lacking nucleotides 1 to 135 served as a positive control. L proteins were separated by SDS-PAGE, blotted, and detected with anti-HA antibody via the HA tag.

Mapping of amino acid residues associated with downregulation of protein expression.

To map amino acid residues that might be involved in the low-expression phenotype, clusters of 3 to 5 amino acids were replaced by alanine between positions 2 and 46 of ANDV L_N. Five of the nine alanine cluster mutants were detectable in immunoblot (Fig. 3A), suggesting that amino acid residues within these clusters are involved in the low expression. To narrow down the sequences, single amino acid residues were exchanged within these clusters. Seven mutants were identified by this approach (Y32A, R35A, H36A, D37A, D40A, I43A, and K44A) showing detectable protein expression (Fig. 3B). However, no change in phenotype was observed when introducing single amino acid substitutions between residues 16 to 20 and residues 27 to 31, although the corresponding alanine cluster mutants were expressed at high levels. Probably, changes in the local protein structure inducible only by substituting two or more residues are needed to change the phenotype here. Taken together, individual amino acid residues as well as structural features in the N terminus appear to be involved in the low-expression phenotype of ANDV L protein.

Fig 3.

Mapping of amino acid residues within the N terminus of L protein mediating the low-expression phenotype. BSR-T7/5 cells were infected with MVA-T7 and transfected with expression plasmids or empty vector as a negative control (Mock). L proteins were separated by SDS-PAGE, blotted, and detected with anti-HA antibody via the HA tag. (A) Three to five amino acid residues in a row were replaced by alanine in the context of ANDV L_N. The position of the mutated stretch is indicated above the immunoblot (first–last residue). (B) Individual amino acid residues were substituted by alanine in the stretches from 16 to 20 and from 27 to 46 in the context of ANDV L_N. (C) Expression of SNV, PUUV, and HTNV L_N proteins (wild-type and K44A mutant), corresponding in length to ANDV L_N, was analyzed in immunoblots.

Expression analysis of Sin Nombre (SNV), Puumala (PUUV), and Hantaan virus (HTNV) L proteins.

The residues Y32, R35, H36, D37, D40, I43, and K44 associated with the low expression of ANDV L protein are highly conserved within the hantavirus genus (4). To investigate whether the expression phenotype of SNV, PUUV, and HTNV L proteins is similar to that of ANDV L protein, the N-terminal residues 1 to 534 were cloned into pCITE-2a. The partial genes corresponding in length to ANDV L_N were fused to a HA tag. In addition, constructs for expression of SNV, PUUV, and HTNV L_N proteins with the K44A mutation were generated. Cells were infected with MVA-T7 and transfected with the L protein constructs. None of the wild-type L proteins of SNV, PUUV, and HTNV was detectable in immunoblots, whereas all K44A mutants produced clearly visible signals (Fig. 3C). Thus, the L proteins of ANDV, SNV, PUUV, and HTNV share the low-expression phenotype and expression may be rescued by exchanging the conserved residue K44.

Proteasomal degradation and half-life time of ANDV L protein.

The experimental data obtained thus far may be consistent with a low half-life time of ANDV L protein, for example, due to rapid proteasomal degradation. Other hantavirus proteins have been shown to be subject of enhanced proteasomal degradation (25). Proteins are usually targeted for degradation by conjugation with ubiquitin molecules at lysine residues (26). Since K44 is highly conserved among all hantavirus L proteins and alanine substitution enhances the protein level, we hypothesized that this residue might be a target for ubiquitinylation. If this is true, replacement of K44 by any other side chain should enhance the protein level. Indeed, all K44 mutants generated and tested (K44A, K44F, K44P, K44S, K44Q, K44E, and K44R) were expressed, suggesting that low expression depends on a lysine residue at position 44 (Fig. 4A). Two sets of experiments were conducted to test the hypothesis of enhanced degradation. First, the proteasome activity was inhibited with MG-132. The inhibitor was added at different concentrations (0 to 100 μM) to cells transfected with constructs for expression of wild-type ANDV L_N or the K44A mutant. No increase in L protein level was observed (Fig. 4B), although the proteasome activity was nearly completely inhibited at 100 μM MG-132 (Fig. 4C).

Fig 4.

Testing the stability of L protein. (A) Lysine on position 44, which might be a target for ubiquitinylation and a mediator of proteasomal degradation, was substituted by residues with related or different chemical properties. BHK-21 cells were infected with MVA-T7 and transfected with expression plasmids or empty vector as a negative control (Mock). L proteins were separated by SDS-PAGE, blotted, and detected with anti-HA antibody via the HA tag. (B) Cells were transfected with wild-type L_N (ANDV-L_N) or the K44A mutant and treated with the proteasome inhibitor MG-132. L protein in cytoplasmic lysate was detected in immunoblot via the HA tag. (C) Control experiment demonstrating inhibition of the proteasome in the above experiment. Proteasome activity in cytoplasmic lysate was measured using a fluorogenic substrate. (D) Pulse-chase labeling of ANDV L_N and K44A mutant. BHK-21 cells were transfected with the corresponding constructs, metabolically labeled with [³⁵S]methionine and [³⁵S]cysteine, and harvested 0, 1, and 2 h after the labeling. L proteins were immunoprecipitated via the HA tag, separated by SDS-PAGE, and detected by autoradiography. (E) The half-life time of ANDV L_N and K44A mutant was calculated from autoradiography data of three independent pulse-chase experiments.

Second, the half-life time of both proteins was determined in a pulse-chase experiment. Cells were transfected with constructs for expression of wild-type ANDV L_N or the K44A mutant and metabolically pulse-labeled with [³⁵S]methionine and [³⁵S]cysteine. The ³⁵S-labeled ANDV L_N proteins were immunoprecipitated at 0, 1, and 2 h after the pulse and separated by PAGE, and their levels were measured by autoradiography. Three independent experiments were performed. Weak and strong signals for wild-type and mutant proteins, respectively, were visible after radioactive labeling (Fig. 4D), facilitating calculation of the half-life times for both proteins. However, no difference in half-life time between wild-type L_N protein and the K44A mutant were observed; it is ∼1 h for both (Fig. 4E). In conclusion, inhibition of proteasomal degradation and half-life time measurement indicates that the low steady-state level of wild-type ANDV L protein is not due to enhanced protein degradation or instability.

Influence of mutations in the putative endonuclease of ANDV L protein on the expression level.

Having largely excluded RNA sequence elements repressing translation or enhanced protein degradation, we assumed the low-expression phenotype is mediated at posttranscriptional level. However, the existence of RNA-destabilizing elements within the 5′ end of L gene is hardly compatible with the observation that substitution of individual amino acid residues (Fig. 3) rather than multiple synonymous nucleotide changes (Fig. 2) can rescue ANDV L protein expression. A solution is to assume that ANDV L protein is both target and effector. To test this, decreasing amounts of wild-type ANDV L_N construct were cotransfected with a constant amount of expression construct for the K44A mutant. Cotransfection of wild-type L_N construct clearly affected the expression of the K44A mutant; even when both constructs were transfected in a 1:100 ratio, the level of the K44A mutant was significantly reduced (Fig. 5A). This dominant-negative trans effect suggests that ANDV L protein plays an active role in downregulating protein expression and is not just a target. The activity would have to be located in the N terminus of L protein.

Fig 5.

Relevance of the putative endonuclease in the N terminus of L protein for the low-expression phenotype. BSR-T7/5 cells were infected with MVA-T7 and transfected with expression plasmids or empty vector as a negative control (Mock). L proteins were separated by SDS-PAGE, blotted, and detected with anti-HA antibody via the HA tag. (A) The expressible L_N K44Amutant was coexpressed with the nonexpressible wild-type ANDV L_N. Different ratios of the corresponding expression plasmids were transfected. (B) Expression analysis of ANDV L_N mutants with mutations of the putative catalytic site of the endonuclease. (C) Expression analysis of ANDV L_N mutants with substitutions of additional conserved and charged residues in the endonuclease domain.

Recently, structural and functional studies revealed an endonuclease of the PD-(D/E)-K superfamily at the N terminus of L proteins of arena- and orthobunyaviruses (12, 13, 27). Primary and secondary structure alignment suggests that the N terminus of hantavirus L protein also harbors an endonuclease. The PD-(D/E)-K motif of hantavirus endonuclease probably corresponds to the conserved residues P96, D97, E110, and K124 (see the alignment in Fig. 11). These residues were exchanged to alanine in the background of the ANDV L_N construct and expression of the mutants was tested by transfection. All catalytic site mutants were clearly expressed as tested in immunoblot (Fig. 5B). To identify further residues important for the low-expression phenotype, 19 conserved E, D, Q, N, K, and R residues within the putative endonuclease domain (residues 1 to 220) were replaced by alanine. Four mutants (N50A, N98A, K127A, and N167A) were found to be expressed, although to a lower level than K44A (Fig. 5C). In conclusion, ANDV L protein expression was enhanced upon mutation of the PD-E-K motif and further conserved residues within the putative endonuclease, suggesting a role for this enzymatic activity in downregulating L protein expression.

Fig 11.

Sequence comparison of the putative endonuclease of hantaviruses with known endonucleases of orthobunyaviruses and orthomyxoviruses. L protein sequences of ANDV, SNV, HTNV, and PUUV were aligned with the L protein of LACV and the PA protein of IAV using CLUSTAL W (41) and MacVector (MacVector). Amino acid residues are colored according to chemical type. The N termini of LACV L protein and IAV PA harbor endonucleases (11, 13), and only those parts of the hantavirus L protein showing significant homology to these enzymes are shown (residues 1 to 51 and residues 95 to 167). The secondary structure (below the sequence alignment) of ANDV L protein was predicted using Jpred (42), while the secondary structure data for LACV and IAV were obtained from the PDB structures 2XI5 and 2W69, respectively. Residues, which have been tested in our study, are indicated with a gray triangle and those, which are important for the low-expression phenotype, are indicated with a black triangle. Residues implicated in the catalytic process of LACV and IAV endonucleases are marked with a filled red triangle; those implicated in RNA binding by an open red triangle. The putative catalytic and RNA-binding residues in hantaviruses—H36, D97, E110, K124, and K127—correspond to H34, D79, D92, K94, and K108 in LACV (13) and H41, D108, E119, K134, and K137 in IAV (11), respectively. The hantavirus sites perfectly match with the putative homologous residues of IAV in the alignment. The match to LACV is not as perfect as to IAV, as the position of the catalytic lysine of LACV (K94, marked with a red triangle) maps to T112 rather than K124 in hantaviruses. Thus, the arrangement of the catalytic site in hantaviruses might be more related to orthomyxoviruses than to orthobunyaviruses.

Influence of ANDV L protein on the mRNA level.

Since the putative endonuclease is supposed to act on mRNA level (consistent with the finding that the low L protein level is not due to increased protein degradation), the influence of various mutations in this domain on the L mRNA level was studied. In addition, the influence of L protein on the level of heterologous mRNAs was measured, since the dominant-negative trans effect suggests that any mRNA might be a target. To this end, two different reporter constructs were cotransfected with the L protein expression constructs: (i) a construct expressing firefly luciferase (FF) under the control of a T7 RNA polymerase promoter and IRES (pCITE-FF) and (ii) a construct for expression of Renilla luciferase (RenLuc) under the control of a polymerase II promoter (pCMV-RenLuc). To verify that the use of the truncated version of L protein (L_N) in the above experiments did not lead to artificial results, selected mutations that enhanced the L protein level were introduced into the full-length L protein (mutants Y32V, R35H, H36R, D37A, D40E, I43A, K44A, N50A, P96A, D97E, N98A, E110A, K124A, K127A, and N167A; the more conservative substitutions at positions 32, 35, 36, 40, and 97 enhanced expression in the context of ANDV L_N [data not shown] just as the alanine substitutions had done [Fig. 3B and 5B]). Cells were infected with MVA-T7 and transfected with L protein expression constructs, as well as with pCITE-FF and pCMV-RenLuc. The L protein level was analyzed in immunoblots, whereas the levels of L mRNA, FF mRNA, and RenLuc mRNA were measured by quantitative real-time RT-PCR.

The expression data obtained with the full-length L proteins confirmed the results with the truncated version of L protein. All mutants were expressed, although at different levels like the analogous L_N mutants (Fig. 6A). Corresponding with the protein data, the L mRNA levels of the mutants were higher than the mRNA level of wild-type L protein (Fig. 6B, upper diagram). Furthermore, there was a rough correlation between L protein and L mRNA levels. Consistent with the dominant-negative trans effect, the L protein level also correlated with the FF mRNA and RenLuc mRNA levels (Fig. 6B, middle and lower diagram). The higher the expression level of a specific mutant, the higher the level of the heterologous mRNAs. These experiments are consistent with the hypothesis that wild-type ANDV L protein downregulates the level of its own mRNA, as well as the level of heterologous mRNAs. The mutational analysis strongly suggests that this effect is mediated by an endonuclease activity in the N terminus of the L protein.

Fig 6.

Expression analysis of full-length L protein mutants and influence of mutations on mRNA levels. Mutations, which enhanced the expression level in the context of ANDV L_N, were tested in the context of full-length L protein. BHK-21 cells were infected with MVA-T7 and transfected with L protein expression constructs, pCITE-FF, and pCMV-RenLuc. Three wells were transfected in parallel. (A) Cells in one well were lysed, L proteins were separated by SDS-PAGE, blotted, and detected with anti-HA antibody via the HA tag. (B) RNA was extracted from cells in the second and third well and the levels of L mRNA, FF mRNA, and RenLuc mRNA were measured by quantitative real-time RT-PCR independently for each well. The relative mRNA levels are shown with or without normalization with the total RNA amount added to the RT-PCR assay (dark and light gray bars, respectively). Mean and range for the two measurements are shown. The corresponding level of the L protein mutants is shown above the diagrams. The immunoblot bands from panel A were rearranged according to signal intensity.

Influence of ANDV L protein on cellular protein level.

Since L protein reduces the level of mRNAs expressed from reporter constructs, we assumed that it also reduces the cellular mRNA and thus protein levels. Cells were transfected with constructs for expression of wild-type ANDV L_N or the K44A mutant and metabolically pulse labeled with [³⁵S]methionine and [³⁵S]cysteine. The ³⁵S-labeled cytoplasmic proteins were extracted at 0, 1, and 2 h after the pulse, corresponding to 7, 8, and 9 h, posttransfection, respectively, separated by SDS-PAGE, and their level was measured by autoradiography. At all time points, the steady-state levels of high-molecular-weight proteins was ca. 50% higher in cells expressing the K44A mutant than in cells expressing ANDV L_N (Fig. 7). No difference was seen in the low-molecular-weight range. The signal intensities in the high-molecular-weight range did not decrease over time, indicating that the reduced steady-state level in ANDV L_N-expressing cells was not due to reduced half-life time of the proteins. We conclude from these data that the expression of ANDV L_N reduces the cellular protein level specifically in the high-molecular-weight range and that this effect is not mediated by enhanced protein degradation.

Fig 7.

Influence of L protein on the cellular protein level. BHK-21 cells were infected with MVA-T7, transfected with constructs for expression of wild-type L_N (ANDV-L_N) or the K44A mutant, metabolically labeled with [³⁵S]methionine and [³⁵S]cysteine, and harvested 0, 1, and 2 h after the labeling. Cytoplasmic lysate was separated by SDS-PAGE, and proteins were detected by autoradiography (top). The intensities of autoradiography signals in regions “a” and “b” indicated to the left of the blot were measured and are shown in the diagram (bottom).

Localization studies with ANDV L, N, Gc, and NSs proteins.

The availability of L protein mutants with single amino acid exchanges in the N terminus, which can be readily expressed and visualized in mammalian cells, enabled us to study the subcellular localization of L protein in the context of other virus and cellular proteins. Cells were infected with MVA-T7 and transfected with various combinations of expression plasmids for ANDV L, N, GPC, and NSs proteins containing HA or FLAG tags. The localization of the proteins was examined by confocal immunofluorescence microscopy using antibodies directed against the tag sequences.

The FLAG-tagged full-length K44A L mutant (L-K44A) was coexpressed with HA-tagged ANDV N and GPC protein, respectively. As the latter is cleaved posttranslationally into Gn and Gc, only Gc carrying the C-terminal HA tag is detected. L-K44A and N proteins are distributed diffusely and in granular and punctate structures in the cytoplasm (Fig. 8A). Both proteins colocalize to a large extent (Fig. 8B). No colocalization with Gc protein was observed (Fig. 8C).

Fig 8.

Intracellular localization of ANDV L, N, and Gc protein. MVA-T7-infected BHK-21 cells were transfected with constructs for expression of L-K44A-FLAG, N, and/or GPC-HA. The FLAG tag was detected with mouse monoclonal anti-FLAG M2, the HA tag with rabbit anti-HA, and N protein with rabbit anti-ANDV-N serum. Nuclei were stained with DAPI. Images were captured by confocal microscopy. Scale bars, 10 μm. (A) Localization of N, Gc, and L-K44A expressed alone. (B) Localization of coexpressed N and L-K44A. (C) Localization of coexpressed Gc and L-K44A.

The HA-tagged NSs protein preferentially localizes in the nucleus when expressed alone (Fig. 9A). In contrast, coexpression with L-K44A relocalizes NSs to the cytoplasm and a high level of colocalization with L protein was observed (Fig. 9B). Cytoplasmic colocalization of NSs protein was also seen with N-terminal (residues 1 to 534) and C-terminal (residues 962 to 2153) fragments of L protein, although the relocalization of NSs to the cytoplasm was more pronounced when coexpressed with the C-terminal L fragment (Fig. 9C and D). No colocalization was observed between NSs and N protein; they mainly localized in nucleus and cytoplasm, respectively (Fig. 9E).

Fig 9.

Intracellular localization of ANDV L, NSs, and N protein. MVA-T7-infected BHK-21 cells were transfected with constructs for expression of L-K44A-FLAG, L_N-K44A-FLAG (residues 1 to 534), L_C-FLAG (residues 962 to 2153), NSs-HA, and/or N protein. The FLAG-tag was detected with mouse monoclonal anti-FLAG M2 and the HA tag with rabbit anti-HA (A to D); N protein was detected with rabbit anti-ANDV-N serum and the HA tag with mouse monoclonal anti-HA (E). Nuclei were stained with DAPI. Images were captured by confocal microscopy. Scale bars, 10 μm. (A) Localization of NSs expressed alone. (B) Localization of coexpressed NSs and L-K44A. (C) Localization of coexpressed NSs and N terminus of L-K44A. (D) Localization of coexpressed NSs and C terminus of L protein. (E) Localization of coexpressed NSs and N protein.

It has previously been demonstrated that hantavirus N protein colocalizes with P bodies in the cytoplasm of infected cells (14). P bodies are compartments for storage and degradation of cellular mRNAs. To visualize P bodies, a monoclonal antibody against the P body component DCP1a (decapping enzyme) was used (16, 17). A potential colocalization of L protein with DCP1a was studied using two different mutants, L-K44A and L-R35H. Both showed the typical granular to diffuse cytoplasmic distribution, whereas DCP1a was localized to a few discrete cytoplasmic dots (Fig. 10A and B). Although most of the L protein does not colocalize with the DCP1a dots, some of the dense granular L protein structures overlap with these dots. This moderate colocalization could be reproduced with the N-terminal (Fig. 10C), but not C-terminal fragment of L protein (Fig. 10D), suggesting that the N terminus containing the putative endonuclease determines the colocalization with DCP1a. The negative result obtained with the C terminus also provides evidence for the specificity of the L protein-DCP1a colocalization. In agreement with previous data (14), a fraction of ANDV-N protein colocalizes with DCP1a (Fig. 10E). Taken together, ANDV L protein colocalizes with N but not Gc protein and relocates NSs protein from the nucleus to the cytoplasm where it colocalizes with NSs. In addition, a certain fraction of L protein colocalizes with the P body component DCP1a.

Fig 10.

Intracellular localization of ANDV L and N protein in relation to P body component DCP1a. MVA-T7-infected BHK-21 cells were transfected with constructs for expression of L-R35H-HA, L-K44A-HA, L_N-K44A-FLAG (residues 1 to 534), L_C-FLAG (residues 962 to 2153), or N protein. The HA tag was detected with rabbit anti-HA, the FLAG tag with rabbit anti-FLAG, N protein with rabbit anti-ANDV-N serum, and P bodies with monoclonal mouse anti-DCP1a. Nuclei were stained with DAPI. Images were captured by confocal microscopy. Scale bars, 10 μm. (A) Localization of L-K44A and DCP1a. (B) Localization of L-R35H and DCP1a. (C) Localization of N terminus of L-K44A and DCP1a. (D) Localization of C terminus of L and DCP1a. (E) Localization of N protein and DCP1a.

DISCUSSION

In this study, we demonstrate that expression of recombinant ANDV L protein in mammalian cells is severely impaired and that this phenotype is determined by the 534 N-terminal amino acid residues. Evidence is presented that the low expression level is not due to translational inhibition, codon usage, rapid proteasomal degradation, or reduced half-life time. High-level expression can be rescued by mutation of catalytic site and other conserved residues of the putative endonuclease in the N terminus of ANDV L protein. These mutations also increased the levels of L mRNA, as well as coexpressed heterologous mRNAs. Overall, these data are consistent with the hypothesis that wild-type ANDV L protein possesses an endonuclease activity, which downregulates the level of its own and other mRNAs. Using expressible ANDV L mutants as a tool for localization studies, we show that L protein colocalizes with N and NSs protein, and at least partially with the P body component DCP1a.

Although our study has been mainly conducted with ANDV L protein, the experiments with SNV, PUUV, and HTNV suggest that hantavirus L proteins are generally difficult to express. Three studies have investigated recombinant L protein expression thus far (5, 18, 28). In agreement with our findings, either no L protein could be detected, or the immunoblot and fluorescence signals were much less intense than the signals obtained here with L protein mutants. The inability to express sufficient quantities of hantavirus L protein and the negative effect on coexpressed proteins also have consequences for the development of hantavirus replicon and reverse genetics systems. Compared to the achievements with orthobunyaviruses (29), phleboviruses (30), nairoviruses (31), arenaviruses (32), and orthomyxoviruses (33), only minor progress has been made with the hantaviruses. Replicon systems—two for HTNV and one for ANDV—have been described (18, 28, 34); a reverse genetics system is still not published. At least one group reported problems with reliability and reproducibility of minigenome rescue despite thorough efforts to optimize the system (18). These authors could not at all or only at very low level detect L protein expression and assumed this was a major obstacle to successful establishment of a reliable replicon system as well as rescue of recombinant hantavirus. Our findings may help designing experimental strategies to overcome this methodological hurdle.

X-ray crystallographic and functional studies revealed the presence of an endonuclease in the N terminus of arena- and orthobunyavirus L proteins (12, 13, 27), as well as the N terminus of IAV PA protein (11). The N terminus of hantavirus L protein shows significant homology to these endonucleases (Fig. 11) and host cell mRNA-derived nucleotides are found at the 5′ end of hantavirus mRNA (8, 9). Both suggest that the N terminus of hantavirus L protein also harbors a cap-snatching endonuclease. Here, we provide further evidence for this hypothesis. Although we cannot directly prove that downregulation of L protein expression is mediated by the endonuclease activity, the observations leave little room for alternative explanations. Several other possible mechanisms have largely been excluded. The key finding supporting the existence of an endonuclease activity is the increase in steady-state levels of L protein and various mRNAs upon mutation of the predicted catalytic residues of the endonuclease, that is, D97, E110, and K124.

Accepting the hypothesis that the endonuclease activity of L protein underlies our observations, one may infer some features of this enzyme. First, in addition to the catalytic site residues, the mutational analysis identified further residues that seem to be essential for enzymatic activity: Y32, R35, H36, D37, D40, I43, K44, N50, P96, N98, K127, and N167 (Fig. 11). Several of them are located in a conserved region between amino acid position 30 and 45, which is predicted to form an alpha helix corresponding to helix 2 of LACV and helix 3 of IAV. Residue H36 in the center of this helix is completely conserved among bunya- and orthomyxoviruses, and the available endonuclease structures indicate that it contributes to the active site. The putative homologs in LACV and IAV of residue K127 are assumed to contribute to RNA binding (11, 13) (Fig. 11).

Second, the nuclease appears to be quite active. As the wild-type endonuclease downregulates its own expression to undetectable levels (a strong negative-feedback loop), it is only a very low quantity of L protein molecules mediating the effects.

Third, the endonuclease cleaves mRNAs irrespective of sequence and cytoplasmic or nuclear origin. This would predict that cellular mRNAs are degraded by L protein as well, which would eventually downregulate host cell protein synthesis as demonstrated here by metabolic labeling. The endonuclease may also play a role in regulating virus protein levels. However, in the context of a virus infection the properties of the L protein might be different due to sequestration or interaction with other virus proteins or RNA. Just coexpression of ANDV N, G, or NSs proteins alone or in combination did not change the low-expression phenotype (data not shown).

Fourth, the properties of the hantavirus endonuclease are different from that of arena- and orthobunyaviruses. L proteins of the latter viruses can be readily expressed in mammalian cells (20, 35) and mutations in the arenavirus endonuclease do not enhance the expression level of L protein or coexpressed proteins (27). Thus, the endonuclease activity of hantaviruses appears to be less tightly regulated than that of arena- and orthobunyaviruses.

In contrast to the latter viruses, there are striking parallels between the IAV endonuclease encoded by the PA gene and hantavirus L protein. A repressive effect due to expression of the PA gene on the steady-state level of other coexpressed proteins, as well as PA itself—which corresponds to the observations reported here—had been described several years ago, but mistakenly ascribed to a reduced half-life time of the proteins (36, 37); an option we have excluded for the hantavirus L protein. Only recently, it has been shown that this repressive effect is mediated by the so-called PA-X protein, which is expressed via ribosomal frameshifting from the PA gene (38). PA-X comprises the N-terminal endonuclease domain of PA fused to the X peptide encoded in the +1 frame. Although PA-X is expressed only at a very low level from the PA gene, it efficiently downregulates the level of coexpressed protein and corresponding mRNA. As shown here for ANDV, mutation of catalytic site residues of the influenza virus endonuclease abolished this effect. Because the repressive effect is mediated by PA-X rather than full-length PA, it is tempting to speculate that in hantaviruses the effect might also be due to expression of a truncated L protein comprising only the endonuclease domain. However, we could neither identify within the hantavirus L gene a frameshifting motif as described for PA-X nor identify a conserved +1 ORF corresponding to the X ORF. Nevertheless, this does not exclude the possibility that a truncated L protein, expressed at a low level via any other mechanism, is responsible for the effects described here.

The expressible hantavirus L protein mutants, although predictably endonuclease deficient, may be a useful tool to study protein localization or interaction. However, some influence of the mutations on both aspects may not be fully excluded. Like L proteins of arenaviruses and other bunyavirus genera (35, 39, 40), the hantavirus L protein is localized in granular to punctate cytoplasmic structures. In agreement with previous studies (5), it sometimes accumulated in the perinuclear area. Evidence for expression of NSs protein during ANDV infection has been provided only recently, and its function is still unknown (7). Our data suggest that this small protein might interact with L protein, as indicated by its relocalization from nucleus to cytoplasm upon the coexpression of L protein. A small fraction of L protein also colocalizes with P bodies, mediated by the N-terminal part of the protein harboring the putative endonuclease. This finding is consistent with the colocalization of L and N protein, since N protein also colocalizes with P bodies as shown here for ANDV and previously for SNV (14). This supports the proposed mechanism for hantavirus cap-snatching involving the protection by N protein of 5′-terminal nucleotides of capped mRNA in P bodies and their transfer to L protein for endonucleolytic cleavage and initiation of transcription (14). The colocalization of L protein with P bodies in the absence of N protein indicates that both virus proteins target this structure independent of each other.