Major Capsid Protein Synthesis from the Genomic RNA of Feline Calicivirus

Caliciviruses are a cause of important diseases in humans and animals. It is crucial to understand the prerequisites of efficient replication of these viruses in order to develop strategies for prevention and treatment of these diseases. It was shown before that all caliciviruses except vesiviruses have established mechanisms to achieve major capsid protein (VP1) translation from the genomic RNA. Here, we show for the first time that a member of the genus Vesivirus also has a translation initiation mechanism by which a precursor protein of the VP1 protein is expressed from the genomic RNA. This finding clearly points at a functional role of the calicivirus VP1 capsid protein in early replication, and we provide experimental data supporting this hypothesis.

ABSTRACT

Caliciviruses have a positive-strand RNA genome with a length of about 7.5 kb that contains 2, 3, or 4 functional open reading frames (ORFs). A subgenomic mRNA (sg-RNA) is transcribed in the infected cell, and both major capsid protein viral protein 1 (VP1) and minor capsid protein VP2 are translated from the sg-RNA. Translation of proteins from the genomic RNA (g-RNA) and from the sg-RNA is mediated by the RNA-linked viral protein VPg (virus protein, genome linked). Most of the calicivirus genera have translation mechanisms leading to VP1 expression from the g-RNA. VP1 is part of the polyprotein for sapoviruses, lagoviruses, and neboviruses, and a termination/reinitiation mechanism was described for noroviruses. Vesiviruses have no known mechanism for the expression of VP1 from the g-RNA, and the Vesivirus genus is the only genus of the Caliciviridae that generates VP1 via a precursor capsid leader protein (LC-VP1). Analyses of feline calicivirus (FCV) g-RNA translation showed a low level of VP1 expression with an initiation downstream of the original start codon of LC-VP1, leading to a smaller, truncated LC-VP1 (tLC-VP1) protein. Deletion and substitution analyses of the region surrounding the LC-VP1 start codon allowed the identification of sequences within the leader protein coding region of FCV that have an impact on VP1 translation frequency from the g-RNA. Introduction of such mutations into the virus showed an impact of strongly reduced tLC-VP1 levels translated from the g-RNA on viral replication.

IMPORTANCE Caliciviruses are a cause of important diseases in humans and animals. It is crucial to understand the prerequisites of efficient replication of these viruses in order to develop strategies for prevention and treatment of these diseases. It was shown before that all caliciviruses except vesiviruses have established mechanisms to achieve major capsid protein (VP1) translation from the genomic RNA. Here, we show for the first time that a member of the genus Vesivirus also has a translation initiation mechanism by which a precursor protein of the VP1 protein is expressed from the genomic RNA. This finding clearly points at a functional role of the calicivirus VP1 capsid protein in early replication, and we provide experimental data supporting this hypothesis.

INTRODUCTION

The members of the family Caliciviridae are small nonenveloped viruses that are causative for gastrointestinal diseases in humans and different diseases in animals (1). The genera classified within the family are Norovirus, Sapovirus, Nebovirus, Lagovirus, and Vesivirus (1). Caliciviruses are positive-strand RNA viruses. Genomic RNA (g-RNA) is nonsegmented, with a length of about 7.5 kb, and it contains two functional open reading frames (ORFs) for members of the genera Lagovirus, Sapovirus, and Nebovirus or three ORFs for the genera Vesivirus and Norovirus (Fig. 1). Only the murine norovirus (MNV) has an additional fourth ORF, encoding virulence factor 1 (VF1), thought to be involved in counteracting the innate immune response of the host cell (2, 3). A subgenomic mRNA (sg-mRNA) coterminal with the 3′-terminal 2.2 kb of the genome is transcribed in the calicivirus-infected cell. g-RNA and the subgenomic RNA (sg-RNA) are polyadenylated at the 3′ end and carry a viral protein, VPg (virus protein, genome linked), covalently linked to the RNA 5′ end, which is responsible for translation initiation at the 5´ end of the viral RNAs (4–8). The nonstructural proteins are expressed from the g-RNA as polyproteins that are processed by the viral proteinase into the mature viral proteins; in the case of feline calicivirus (FCV), a stable proteinase-polymerase (Pol) precursor protein can be detected (9).

FIG 1.

Schematic map of calicivirus genomic and subgenomic RNAs. The basic organization of the genomic and subgenomic RNAs is illustrated (not drawn to scale). Shaded bars represent regions coding for the capsid proteins VP1 and VP2. The white bars symbolize the nonstructural protein coding regions, with the different cleavage products given within the bars. The nontranslated regions (NTR) at the 5′ end and 3′ end are shown as black lines. VPg (virus protein, genome linked) present at the 5′ ends of both the genomic and subgenomic RNAs is symbolized by a black circle at the end of the line representing the 5′ NTR of the RNAs. The encoded proteins are indicated by abbreviations as follows: Hel, helicase; VPg, virus protein, genome linked; Pro, protease; Pol, polymerase; L, leader of the capsid; VP1, major capsid protein; VP2, minor capsid protein. T, TURBS (responsible for termination/reinitiation of translation).

Both capsid proteins are translated from sg-RNA, including the major capsid protein viral protein 1 (VP1) (mediated by the 5′ VPg) and the minor capsid protein VP2, via a termination upstream ribosomal binding site (TURBS)-based termination/reinitiation mechanism (10–13). For all caliciviruses except vesiviruses, it is known that VP1 is translated in smaller quantities from the g-RNA in addition to the major expression from the sg-RNA. Lago-, nebo- and sapoviruses express VP1 from the g-RNA as part of the ORF1-encoded polyprotein and noroviruses via a termination/reinitiation mechanism similar to VP2 synthesis from the sg-RNA (Fig. 1) (14, 15). As a unique feature of the genus Vesivirus, the VP1 protein is translated as a precursor protein, which is cleaved to yield the ca. 14-kDa capsid leader protein (LC) and the ca. 60-kDa major capsid protein VP1 (16–18). The LC protein has a function in viral spread. It causes the activation of caspases and a cytopathic effect (CPE) in cell cultures (19). VP1 is responsible for capsid formation and was found to associate with the replication complex (20), and VP2 is thought to build a channel in the capsid via a portal-like assembly for delivery of the viral genome through the endosomal membrane into the cytoplasm of the host cell (21).

Until now, there was no indication that VP1 is translated from the g-RNA of vesiviruses. In our analyses, we demonstrated that a truncated VP1 precursor protein (tLC-VP1) is translated from the g-RNA via a process depending on sequences in the leader protein coding region. Furthermore, expression of tLC-VP1 protein from the g-RNA has been shown to have an impact on virus replication.

RESULTS

A truncated version of the Leader-VP1 protein is translated from the genomic RNA of FCV.

To address the issue of whether VP1 is expressed from the g-RNA, full-length cDNA clones of FCV were analyzed in a modified vaccinia virus Ankara-T7 (MVA-T7) expression system in Crandell Reese feline kidney (CRFK) cells. In this system, RNA molecules resembling FCV g-RNA are generated via T7 polymerase transcription from the FCV cDNA full-length clone (22). This RNA is capped by the vaccinia virus enzymes and serves as mRNA for protein synthesis. The transcript subsequently serves as the template for transcription of g-RNAs and sg-RNAs by the viral polymerase. Proteins expressed in this system were subjected to [³⁵S]methionine labeling and were precipitated with antisera specific for VP1 or the protease-polymerase fusion protein (ProPol). Detection of the ProPol protein of FCV was used to validate transfection efficiency and translation of the polyprotein and its processing.

The 73-kDa precursor LC-VP1 and the 59-kDa major capsid protein VP1 are present in cells transfected with a replication-competent construct (Rep+_cp; Fig. 2). The major amount of VP1 proteins detected for this wild-type (wt) construct results from translation of the sg-RNA, generated in consequence of the initiation of viral replication and transcription. To determine the presence of translation that is specific for g-RNA, replication-deficient mutants were generated through substitution of the motif GDD to GAA in the viral polymerase (Rep-). This replication-deficient mutant showed no precursor protein but, importantly, did show a faint band representing the mature VP1 protein (marked by a cross [×] in Fig. 2). This band partially overlapped a background signal also present in the nontransfected cell control, but the higher relative signal intensity and the reduced migration rate demonstrate that this band most likely contained the VP1-specific signal (Fig. 2). Thus, we provide first evidence for VP1 expression from the g-RNA.

FIG 2.

Transient expression of FCV proteins. (A) Schematic map of the FCV full-length construct and the sg-RNA, with the expected VP1-containing products and kDa values depicted, respectively. VPg is symbolized by a black circle, and T7 represents the start site for T7 polymerase transcription of full-length clones leading to capped RNAs in a Vaccinia MVA-T7 system. (B) Radiographs show results for transient expression of FCV proteins in CRFK cells, with 35S labeling and subsequent immunoprecipitation (IP) with a VP1-specific antibody (left panel) and an antibody specific for the viral polymerase (right panel). “–”, control without plasmid transfection; Rep+, replication competent, Rep−, replication deficient; cp, cleavage site positive; cn, cleavage site negative (E124L substitution of P-1 for the LC/VP1 cleavage site). The calculated molecular weight (MW) of the proteins in kDa is given next to the highlighted proteins. For tLC-VP1, codon 86 was assumed to be the start site of tLC-VP1 translation for MW calculation. The molecular weight of the used marker proteins is given to the right of the radiographs. tLC-VP1 is highlighted with an asterisk (*) for Rep+_cn and VP1 with a cross (×) for Rep−_cp.

For further investigation of VP1 translation from FCV g-RNA, we generated mutants with inactivated cleavage site between the leader and VP1 protein (E124L, cn) for detection of unprocessed leader-VP1 (LC-VP1) precursor proteins. The cleavage site mutant of the replication-competent construct (Rep+_cn) showed, in addition to the expected 73-kDa precursor LC-VP1, another faint VP1-specific band representing a 63-kDa truncated LC-VP1 protein (tLC-VP1; highlighted by an asterisk [*] in Fig. 2). This tLC-VP1 protein could also be detected for the replication-deficient GAA mutant, with deletion of the cleavage site (cn), but no 73-kDa LC-VP1 protein expression was detectable for this construct (Rep-_cn, Fig. 2). These findings led to the conclusion that a VP1 precursor protein was in fact translated from the g-RNA and that its translation initiated downstream of the original leader protein start codon, resulting in a smaller, N-terminally truncated LC-VP1 protein (tLC-VP1). This 63-kDa tLC-VP1 protein was expressed in much smaller amounts than the full-length 73-kDa LC-VP1 protein in replication-competent mutants but was evidently also cleaved by the viral protease, resulting in the mature VP1 protein and a truncated leader protein (tLC). tLC-VP1 translation might start at the second AUG in the LC coding frame, which would fit with the observed molecular weight of the truncated protein of 63 kDa perfectly. Therefore, this M86 codon was assumed to be a potential start site of tLC-VP1 for further investigations.

Detection of LC proteins via a nanoLuciferase reporter system.

To show that tLC-VP1 protein synthesis also occurs in cells infected with FCV, a reporter gene construct was generated. The gene coding for a nanoLuciferase was inserted into the FCV full-length cDNA construct downstream of the above-mentioned second AUG in the LC coding region between codon 88 and 89 (pJT15nLuc, Fig. 3A). This site in the FCV genome was shown previously by Abente et al. (23) to be a possible insertion site allowing the generation of viable virus. This reporter insertion enables detection of LC-fusion proteins, from translation initiation at the first AUG of the LC coding region (LC-nLuc [LC-NanoLuc]) and from initiation at a potential downstream initiation site resulting in tLC-nLuc. Thus, both LC proteins can be analyzed through sensitive detection via nanoLuciferase activity independently of antibody recognition. A viable virus mutant (FCV-LC-nLuc) was recovered from the recombinant full-length construct, but the mutant showed impaired growth properties compared to wt FCV with a 50% tissue culture infective dose (TCID₅₀) difference of 1 to 2 log10 for the final titer (data not shown). As controls for the PAGE analysis, two expression plasmids were generated that coded for LC-nLuc fusion proteins and that contained the LC gene starting at the first AUG (pLC-nLuc) or at the second AUG in frame (ptLC-nLuc), both with insertion of the nLuc gene such as was done for the full-length clone (Fig. 3A). As a further control, a replication-deficient full-length clone (pJT23-nLuc) was generated to visualize g-RNA translation products.

FIG 3.

LC-nLuc protein detection. (A) (Top) Schematic map of FCV RNA, given for FCV wt virus in comparison to a nanoLuciferase reporter virus (FCV LC-nLuc, generated from full-length clone pJT15-nLuc). (Bottom) Schematic maps of reporter constructs for a full-length clone (pJT23-nLuc) and pCI-based control expression plasmids (pLC-nLuc and ptLC-nLuc). (B) In-gel detection of nanoLuciferase activity for samples from infection of CRFK cells with FCV-nLuc virus and transient expression of ptLC-nLuc or pLC-nLuc control plasmid or full-length construct pJT23-nLuc in BHK-21 cells. The molecular weight of the marker proteins is given in kDa to the right.

Infection of CRFK cells with FCV_LC-nLuc virus resulted in 3 protein bands specific for nLuc-fusion proteins (Fig. 3B). A comparison to LC-nLuc proteins obtained from the pLC-nLuc control expression plasmid indicated that one band most likely represented LC-nLuc starting at the first AUG of the leader protein coding region (sg-RNA translation). Another nLuc-specific band of rather low intensity in the FCV_LC-nLuc infection sample represented the tLC-nLuc protein, comigrating with the product translated from the ptLC-nLuc control plasmid. Proteins migrating at the same position in the gel were expressed from the control plasmid pLC-nLuc, probably translated via leaky scanning. One band at 25 kDa in the FCV-nLuc extract cannot be easily explained by postulating an alternative translation mechanism and thus might have represented a degradation product. To provide further proof for tLC-nLuc translation from g-RNA, protein expression from the replication-deficient full-length construct pJT23-nLuc was analyzed. As expected, due to the absence of sg-mRNA, no LC-nLuc was detected. However, a clear band was visible for the tLC-nLuc protein. Thus, tLC-nLuc (with translation probably starting at the second AUG in the LC ORF) was translated from the g-RNA and LC-nLuc (with translation starting at the first AUG) and a putative cleavage product thereof resulted from sg-RNA translation. This demonstrated that tLC-VP1 translation occurs in infection and that rather small amounts are translated in comparison to LC-VP1 translation from the sg-RNA (ca. 5%).

tLC-VP1 translation from the g-RNA depends on the leader protein coding region.

To analyze the influence of mutations and deletions on the level of expression of tLC-VP1 from the g-RNA, another cDNA reporter construct was generated based on a replication-inactive GAA full-length clone (pJT23, Rep-). Most of the VP1 coding region within this construct was replaced by the gene coding for a firefly luciferase (FFL) (pJT23FF, Fig. 4A). In this way, a sensitive assay for VP1 translation from the g-RNA via transient expression in BHK-21 cells and subsequent measurement of firefly luciferase activity was established. A pCMV-Renilla luciferase plasmid (pRL) was cotransfected for normalization of transfection efficiency (FFL/RL). Different mutations were inserted into the pJT23FF clone and were analyzed for their effects on tLC-VP1-FFL translation. In addition, samples of selected constructs were used in parallel for RNA isolation. RNA transcribed from the transfected plasmids was detected in a Northern blot assay using a firefly gene-specific ³²P probe (see Fig. 7C). These analyses showed the integrity of the RNA in this assay transcribed from the transfected plasmids via the T7-polymerase delivered by the Vaccinia MVA T7 virus.

FIG 4.

Detection of a region important for tLC-VP1 translation. (A) Schematic map of the pJT23FF-FCV full-length construct with insertion of the firefly luciferase gene (FFL) and GAA substitution of the polymerase motif. Positions of mutations are given below, representing an insertion of a stop codon in the polymerase coding sequence (*Pol, nt 5027 to 5029 [TCA→TAG]) or a deletion in the polymerase coding region (dPol) (nt 5075 to 5305 [Pol stop codon still present]) or in the LC coding region (dL) (nt 5326 to 5472 [AUG; LC nt 5314 to 5316 still present]). (B) Transient expression of pJT23FF constructs and measurement of firefly luciferase activity in BHK-21 cell lysates, normalized to Renilla luciferase activity. Results seen with the pJT23FF construct without changes in the polymerase or LC coding region were set as 100% (wt). Asterisks represent P values for differences with regard to wt (****, P < 0.0001). RLU, relative light units.

FIG 7.

Deletion analysis of the dL1 region. (A) Sequences of the FCV genome coding for the amino terminal part of the LC protein from nt 5314 to 5383 are given for the basic pJT23FF construct (see also Fig. 4A for structure of pJT23FF). The encoded amino acid sequence and the positions referring to LC coding region are given above the nucleotide sequences. The region found to be important for tLC-VP1 translation is highlighted in the pJT23FF sequence. The sequences of the different deletion mutants are given below that sequence. (B) Transient expression of pJT23FF constructs and measurement of firefly luciferase activity in the BHK-21 cell lysates, normalized to Renilla luciferase activity. Results determined for pJT23FF construct without changes in the LC coding region were set as 100% (wt). Asterisks represent P values for differences from wt results (**, P < 0.01; ***, P < 0.001; ****, P < 0.0001). (C) The radiograph shows results of Northern blotting. RNA detection was achieved by using a firefly gene-specific ³²P probe, after vaccinia MVA T7 infection, transfection of the given pJT23FF-wt or mutant DNA construct, and RNA isolation. In vitro-transcribed RNA generated via T7 polymerase from a linearized pJT23FF construct served as a control (ivTk).

For analysis in the firefly assay, a stop codon was inserted into the polymerase coding region to examine whether a termination/reinitiation mechanism was responsible for tLC-VP1 translation and deletions were introduced to check for regions important for the translation mechanism.

Insertion of a stop codon in the polymerase coding region 282 nt upstream of the original stop codon (pJT23FF_*Pol) and deletion of 231 nt of the 3´ end of the polymerase coding region (pJT23FF_dPol), both preserving the original stop codon, had no effect on FFL expression (Fig. 4). Thus, neither translation to the stop/start region (Pol/LC-VP1) nor the sequence upstream of the leader protein coding region is important for tLC-VP1 translation. In contrast, deletion of 147 nt downstream of the leader protein start codon (pJT23FF_dL) had a severe effect on FFL expression, demonstrating that part of the 5′ region of the leader protein coding sequence is essential for the tLC-VP1 translation initiation mechanism.

Requirements for tLC-VP1 translation initiation.

The LC coding region contained three AUG codons at codon positions 1, 86, and 103 in the LC-VP1 coding frame, which could be used for tLC-VP1 translation, and no further AUG codons were present in any frame in this region. For identification of the tLC-VP1 translation initiation site, mutations of these three AUG codons were introduced and analyzed in the firefly luciferase assay. Mutation of the first AUG (M1C) had no effect on tLC-VP1 translation, and mutation of the third AUG (M103A) led only to a minor reduction to 86% of the wt level (Fig. 5A and C). In contrast, only a mutation of the second AUG led to a strongly reduced level of tLC-VP1 expression (M86C, 21%, Fig. 5A and C), indicating that this codon is probably used for translation initiation.

FIG 5.

Importance of AUG start site for tLC-VP1 translation. (A) Parts of the nucleotide sequences of the FCV LC coding region are given for the pJT23FF firefly luciferase insertion construct and for mutants thereof. Sequence changes for the mutants are highlighted, and AUG codons are underlined. The amino acid sequence of LC and the codon positions are given above the wt nucleotide sequence. (B) Kozak AUG sequence context preferences (adapted from Harhay et al. [35]) are shown. (C and D) Transient expression of pJT23FF constructs and measurement of firefly luciferase activity in the cell lysates, normalized to Renilla luciferase activity. Results determined for the pJT23FF construct without changes in the LC coding region (wt) were set as 100%. Asterisks represent P values for differences from wt results (*, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001).

To confirm codon 86 as the start site, stop codons were introduced either upstream (P84* and P85*) or downstream (W88*) of codon 86 (Fig. 5A). Introduction of a stop codon at position 84 had only a minor reduction effect on the translation level (80%), and a stop codon at position 85 led to a reduction to 50%, whereas a stop codon at position 88 led to a nearly complete loss of translation (Fig. 5C). Thus, these results confirmed that the AUG codon (86) is most likely used for translation initiation and that the surrounding sequence has some impact on initiation, as mutant P85* with a change of positions −3 and −2 according to the putative start codon showed a reducing effect on the translation level. This sequence surrounding the potential start codon displayed a poor Kozak sequence context (CCAAUGC [where the characters with bold highlighting represent the most important positions for initiation and the underlined characters represent the potential start codon]; compare Fig. 5B). To check for dependency of tLC-VP1 translation initiation on a Kozak sequence, the two most important positions, −3 and +4, were changed according to the Kozak rules to a good sequence context (ACAAUGG, mutant M86k, Fig. 5A) (24). This change to a good Kozak sequence slightly improved translation initiation (124%, Fig. 5C), further supporting the idea of the importance of the region surrounding the potential start codon for translation initiation efficiency.

To analyze whether an alternative start codon would be usable for translation initiation and to make sure that the protein levels would not differ because of changes in the amino acid sequence, two different substitutions of the second AUG codon, both coding for leucine, were introduced into the pJT23FF reporter construct. Codon 86 was mutated to CUG, representing an alternative start codon (M86L1), and the mutation had no effect on FFL translation. Substitution of this codon to UUA, a 3-nucleotide (nt) substitution (M86L2), led to a slightly reduced level of FFL translation (Fig. 5A and C). The reducing effect on tLC-VP1 translation for M86L2 depended on the nucleotide change and not on an amino acid change in the LC protein, as M86L1 showed no effect on the translation rate, but even the exchange of the entire codon still preserved a high level of translation (M86L2, 80%).

In contrast to the results seen with M86L2, the three-nucleotide substitution of M86C had a high impact on the translation level. Therefore, we checked if the AUG generated by mutation M86C in the −1 frame (see Fig. 5A) led to translation initiation on the −1 frame, which would not lead to expression of the firefly reporter gene. We mutated the nucleotide directly upstream of codon 86 in addition to the UGC substitution of this codon to remove the generated AUG codon (see Fig. 5A). This mutant, M86C2, showed a significantly higher translation level (73%) that was seen with mutant M86C (21%) (Fig. 5C). Thus, it might be concluded that the sequence of codon 86 is of minor importance for the efficiency of tLC-VP1 translation. The reduced recovery of tLC-VP1 from mutant M86C was due to −1 frame translation, an out-of-frame translation with regard to tLC-VP1. These results point to the presence of a scanning mechanism leading to tLC-VP1 translation.

Next, we wanted to analyze whether the presence of a scanning mechanism could lead to initiation at the next AUG downstream of codon 86 when this AUG (codon 86) is mutated. We picked two of the mutants showing high levels of translation despite the changed AUG codon at position 86 and in addition either introduced a stop codon at position 88 (W88*) or substituted the third AUG of the LC coding frame to GCA (M103A, see Fig. 5A). Both mutations of codon 86 still showed a high level of firefly luciferase translation with insertion of a stop codon at position 88. Therefore, translation initiation must occur downstream of codon 88 for these constructs. In contrast, mutation of the third AUG (M103A) led to a severely reduced translation level (see Fig. 5C). Thus, the high translation level for mutants M86C2 and M86L2 depends mainly on scanning to and initiation at the next AUG at position 103.

These results clearly show that if an AUG is present at position 86, translation predominantly starts at this position and the sequence context has some influence on initiation efficiency (M86k, P85*). Besides initiation at position 86, some initiation probably takes place at position 103, leading to the observed reduction of the translation level for mutant M103A (86%, Fig. 5). When codon 86 is altered, initiation mainly takes place at codon M103. These data strongly support the idea that a scanning mechanism is responsible for the observed firefly luciferase translation.

Next, we wanted to know if the ribosomes start scanning further upstream of codon 86. For these analyses, we used a mutant showing significantly reduced tLC-VP1 translation (M86C) to enable us to see enhancement of translation efficiency by introduction of upstream AUG codons. M86C showed a reduced level of translation caused by initiation at an out-of-frame AUG with regard to tLC-FF-VP1. If the introduced insertion of an upstream AUG codon in the firefly luciferase coding frame led to initiation at this codon, expression of the firefly luciferase coding frame would be enhanced compared to the M86C mutant. AUG codons were introduced at position 18, 61, or 79 in the M86C mutant (Fig. 5A). All 3 AUG insertion mutants led to a higher level of expression than was seen with mutant M86C (21% of the wt level). Mutation I18M led to a 51% translation level compared to the pJT23FF wt construct, insertion at position 61 to a 78% level, and insertion at position 79 to the wt level (Fig. 5D). We cannot rule out the possibility that the differences in the expression levels of these three mutants could have been due partly to the sequence context of the introduced AUG codons. Nevertheless, these results show that a scanning process starts further upstream of codon M86 in the leader protein coding region.

Different regions within the leader protein coding sequence are important for tLC-VP1 translation from the g-RNA.

The experiments described before showed that the leader protein coding region is important for tLC-VP1 translation from the g-RNA (dL, Fig. 4) and that the second AUG in the LC coding region is probably used as a start codon reached by ribosomal scanning (Fig. 5). To narrow the range of regions in the leader protein coding sequence that might be important for the tLC-VP1 translation mechanism from the g-RNA, the firefly luciferase system was used to screen the fragment between the first and the second AUG of the LC-ORF via deletion analysis (Fig. 6A). This region codes for LC from the sg-RNA but is the upstream noncoding region with respect to tLC-VP1 translation from the g-RNA. Deletion of nt 5320 to 5370 of the FCV-RNA sequence (pJT23FF_dL1) or of nt 5473 to 5520 (pJT23FF_dL4) led to a severe reduction of FFL translation to about 10% to 20% compared to the wt sequence (Fig. 6B). The region between nt 5422 and nt 5472 (pJT23FF_dL3) had less influence, leading to 43% FFL translation. The other two tested deletions, dL2 (localized between deletions of mutants dL1 and dL3) and dL5 (located directly upstream of the putative tLC-VP1 start codon [M86]), had no significant effect on the translation level (Fig. 6B). Thus, several separate regions within the leader protein coding region are important for tLC-VP1 translation from the g-RNA.

FIG 6.

Deletion analysis of the LC coding region. (A) A schematic map of the pJT 23FF-FCV full-length construct and the positions of the deletions in the LC coding region (dL) are given. (B) Transient expression of pJT23FF constructs and measurement of firefly luciferase activity in BHK-21 cell lysates, normalized to Renilla luciferase activity. Results determined for the pJT23FF construct without changes in the LC coding region were set as 100% (wt). Asterisks represent P values for differences from wt results (****, P < 0.0001).

To narrow the range of the sequences that might be important for VP1 translation from the g-RNA in more detail, the most critical region, nt 5320 to nt 5370 (pJT23FF_dL1), was screened by the use of smaller deletions of 15 nt (Fig. 7A). The first two deletions in this region (dL1-1 and dL1-2) showed a severe reduction of FFL translation (Fig. 7B) comparable to the reduction seen for the pJT23FF_dL1 deletion mutant with deletion of 51 nt (11% of the wt level). In contrast, deletion mutants dL1-3 and dL1-4 even showed a slight increase in VP1 translation. Further screening of the region from nt 5317 to nt 5349 with 6-nt deletion mutants (dL1-1a and dL1-1b and dL1-2a to dL1-2c) showed that the entire region is important for efficient translation (Fig. 7B, see also Fig. 12). All of the 6-nt mutations led to a strongly reduced level of translation, with most of the mutants showing a reduction to about 10% of the level seen with the wt leader protein sequence. Only deletion of nucleotides 5332 to 5337 (pJT23FF_dL1-2a) had a milder effect (reduction to 40%, Fig. 7B).

FIG 12.

Schematic map of important regions for tLC-VP1 translation. The upper panel shows a schematic map of the FCV LC coding region. Shaded bars represent regions shown to be important for tLC-VP1 translation in the deletion analysis, with the levels of tLC-VP1 translation compared to wt given in percentages below the bars (results are from the experiments performed as described for Fig. 7, 9, and 10). The position of the 3 AUGs (codon 1, 86, and 103) and the positions and results for translation efficiency for insertion of AUG into the M86C mutant are given above the bar (see Fig. 5C). The lower panel shows the two precursor proteins, LC-VP1 and tLC-VP1, translated from the two RNA species.

RNA molecules were checked by Northern blotting for their integrity for the transfected wt construct and a selection of the analyzed mutants. In comparison to the wt construct, no difference in RNA molecule size or stability was detected for mutant dL1-2b or M86C (Fig. 7C) that might be responsible for the reduced translation level of the mutants (Fig. 5C; see also Fig. 7B). Thus, the strongly reduced translation level seen for mutant dL1-2b could not have been due to partial degradation of the mutant RNA since no difference between the RNAs of the wt and mutant was observed.

The importance of this region for tLC-VP1 translation could result from a nucleotide sequence motif or from a secondary structure effect through presentation of a motif, or the introduced amino acid change could have an effect on protein structure and stability. The third codon of the LC coding region was chosen for further analyses to distinguish between these effects. This codon is located within the region of highest importance for tLC-VP1 translation (Fig. 8), and the different effects can be analyzed at this position. The encoded amino acid can be changed by 1-nucleotide to 3-nucleotide substitutions or can be maintained despite replacement of 3 nucleotides (UCA to AGU; S3S in Fig. 8A) exhibiting a major change on the nucleotide level but no change on the amino acid level. A putative secondary structure proposed for the 5′ sg-RNA (25) can be destroyed by a substitution of this codon located within a stem-loop structure or can be maintained by G-U pairing substituting the G-C and U-A pairing (S3L1, see Fig. 8A and B) (26).

FIG 8.

Substitution analysis of codon nt 5320 to 5322. (A) Sequences of the FCV LC coding region surrounding the first AUG in the LC coding frame are given for the basic pJT23FF construct and mutants thereof. Stem regions within the sequence that were predicted as described previously by Alhatlani et al. (25) for 5´sg-RNA secondary structure are underlined, and the loop nucleotides are marked by italic letters. The region important for tLC-VP1 translation is highlighted in the pJT23FF-wt sequence, as well as the substitutions in the mutant constructs. (B) Putative secondary structure of the analyzed sequence in LC coding region as proposed by Alhatlani et al. (25). The polymerase stop codon and the LC start codon are highlighted in bold. (C) Transient expression of pJT23FF constructs and measurement of firefly luciferase activity in the BHK-21 cell lysates, normalized to Renilla luciferase activity. Results determined for the pJT23FF construct without changes in the LC coding region were set as 100% (wt). Asterisks represent P values for differences from wt results (****, P < 0.0001).

For this purpose, the codon was changed by substitutions of 2 or 3 nt (Fig. 8A). All changes affecting this codon led to similar severe reductions of tLC-VP1 translation efficiency no matter whether the encoded amino acid was changed or not (Fig. 8C). Thus, it could be shown that this region is very important for tLC-VP1 translation and that this effect is not based on an amino acid change but relies on the nucleotide sequence. Furthermore, the possible secondary structure proposed for 5′ sg-RNA (25) is not important for tLC-VP1 translation since a mutation preserving the possible secondary structure through G-U pairing (S3L1, Fig. 8A and B) had an effect on translation efficiency similar to that seen with the other mutations of this codon (Fig. 8C).

Nucleotides 5317 to 5349 were identified as a very important region for tLC_VP1 translation mechanism, but further regions were shown to be important too (Fig. 6). To analyze these additional important sequences as well, further deletion analyses were conducted. Deletions affecting the sequence between nt 5422 and nt 5472 led to no or a rather mild reduction of FFL translation to 50% of the wt level for the 51-nt dL3 deletion mutant and the 12-nt dL3-3 deletion mutant (Fig. 9). Thus, the region consisting of nt 5446 to nt 5457 (dL3-3) is of importance for tLC-VP1 translation whereas the parts flanking this sequence are of minor importance (Fig. 9) (see also Fig. 12). Putative 18S rRNA interacting motifs in the LC coding region are present within the deleted sequence of the dL3 mutant. This is a tobacco etch virus TEV motif (UACUCCC) (27) and a TURBS motif (UGUGGGA) (Fig. 9A). No reduction of tLC_VP1 translation was observed for deletion of the possible TURBS motif (dL3-4) and of the first 2 nucleotides of the possible TEV motif (dL3-2) (Fig. 9B). A reduction by only 50% was observed for deletion of the last 5 nucleotides of the possible TEV motif (dL3-3, Fig. 9B). Thus, these putative 18S rRNA interacting motifs do not appear to be essential for the tLC-VP1 translation mechanism.

FIG 9.

Deletion analysis of the dL3 region. (A) Sequences of the FCV LC coding region between nt 5420 and 5475 are given for the wt pJT23FF construct and deletion mutants thereof. Putative 18S rRNA interacting motifs are underlined in the pJT23FF sequence, and the region found to be of importance for tLC-FF-VP1 translation is highlighted in the pJT23FF sequence. (B) Transient expression of pJT23FF constructs and measurement of firefly luciferase activity in the BHK-21 cell lysates, normalized to Renilla luciferase activity. Results determined for pJT23FF construct without changes in the LC coding region were set as 100% (wt). Asterisks represent P values for differences from wt results (**, P < 0.01; ***, P < 0.001).

To determine the sequences within the region consisting of nt 5473 to nt 5520 (pJT23FF_dL4, 19%, Fig. 6A), which are important for tLC-VP1 expression, smaller deletions of 12 nt were introduced into this region (Fig. 10A). Most of the deletions showed just a mild effect on the translation level, with a reduction to 60% (pJT23FF_dL4-3) or 70% (pJT23FF_dL4-1). Deletion of nt 5485 to nt 5496 showed the highest reduction at 25% of the wt level (dL4-2) (Fig. 10B; see also Fig. 12). In further deletions of 6 nt within the dL4-2 region, no prominent reduction of FFL translation was observed; thus, only larger deletions in this region showed an effect on tLC-VP1 translation (Fig. 10B).

FIG 10.

Deletion analysis of the dL4 region. (A) Sequences of the FCV LC coding region between nt 5470 and 5523 are given for the basic pJT23FF construct and deletion mutants thereof. The region found to be of importance for tLC-FF-VP1 translation is highlighted in the pJT23FF sequence. (B) Transient expression of pJT23FF constructs and measurement of firefly luciferase activity in the BHK-21 cell lysates, normalized to Renilla luciferase activity. Results for pJT23FF basic construct without changes in the LC coding region were set as 100% (wt). Asterisks represent P values for differences from wt results (**, P < 0.01; ****, P < 0.0001).

Effects of selected mutations on viral replication.

To analyze the effect of a reduced level of tLC-VP1 translation from the genomic RNA on viral fitness, different mutations were selected. These mutations showed a reduced tLC-VP1 translation level in the firefly luciferase assay but exhibited minor changes of the leader protein amino acid sequence to avoid effects on LC function influencing viral fitness as previously shown by Abente et al. (19). Therefore, one mutation of the second AUG in the LC coding frame (construct M86C), 3 (dL1-1b, dL1-2a, and dL1-2b) of the 6-nt deletion mutants with mutations in the leader protein coding 5′ region and mutation S3S preserving the LC amino acid sequence but severely reducing the tLC translation level were introduced into the full-length FCV wild-type cDNA clone (pJT15) (see Table 1). The Vaccinia MVA-T7 system was used to generate virus mutants (22). Constructs with the mutations M86C, dL1-2b, and S3S gave viable virus, but no virus could be recovered for the mutations dL1-1b and dL1-2a in multiple experiments (Table 1). The virus with the dL1-2b deletion did not induce a cytopathic effect (CPE) but could be detected only by PCR analysis. There is no correlation between the reduction of tLC-VP1 translation and the ability to recover virus. Thus, most likely, the alterations of the leader protein sequence have an additional impact on virus viability (19). Passaging and sequencing of the obtained viruses showed that the M86C, S3S, and dL1-2b mutations were stable and that no reversion or second site mutation occurred in the leader protein and VP1 coding region.

TABLE 1.

Effects of given LC mutations on virus recovery in the context of the results obtained from the firefly luciferase assay for tLC-VP1 translation

Mutationa	tLC-VP1-FFL expression (%)	CPE	RT-PCR
wt	100	+	+
M86C	21	+	+
dL1-1b (Δaa5C6A)	10	−	−
dL1-2a (Δaa7N8V)	40	−	−
dL1-2b (Δaa9L10K)	13	−	+
S3S (uca→agu)	9	+	+

^aaa, amino acid.

Growth curves of the recovered virus mutants compared to the wild-type virus were recorded starting with a very low multiplicity of infection (MOI) of 0.0003 to enable detection of differences in replication levels for this very-fast-growing virus. The M86C virus showed no significant difference in growth, despite the difference in tLC-VP1 translation efficiency as determined in the firefly luciferase assay (only 21% of the wt level, Fig. 11). The S3S mutant had no change in the LC amino acid sequence but showed a significantly reduced level of tLC-VP1 translation from the g-RNA in the firefly luciferase assay (9%). The S3S virus exhibited reduced growth efficiency (Fig. 11) in the course of infection but reached a maximum titer similar to that of the wt virus. Therefore, it is likely that the growth deficiency results from the reduced tLC-VP1 translation leading to a nearly complete absence of VP1 during the early stage of replication. Low levels of VP1 seem to be sufficient to reach good replication level (21% for M86C), but the even further reduced level of tLC-VP1 translation from the g-RNA for S3S seemed to undercut a critical level. As the amino acid sequence and thus the function of LC were not affected by this mutation, the effect on replication was most likely due to the very low level of tLC-VP1 translation from the g-RNA. At later stages, expression of the VP1 protein from the sg-RNA (mRNA) would occur for all viruses, including the wt and mutants, in large amounts.

FIG 11.

Growth curves of viruses with mutations in the leader protein coding region. The growth curves represent viruses recovered from the full-length constructs containing the mutations represented in Fig. 4 and 8 for M86C and S3S, respectively. The curves show the results of at least 4 independent experiments in titer (TCID₅₀) at a given time point (hpi, hours postinfection). Infection was done at a MOI of 0.0003. Asterisks represent P values for S3S mutant differences from wt results (*, P < 0.05; ****, P < 0.0001); for virus mutant M86C, no significant difference from the wt results was seen with respect to growth. Titers of virus dilutions used for growth curve infections were checked in several reproductions; thus, the virus amounts introduced into the growth curve analysis were very similar for the wt and mutant strains.

DISCUSSION

Caliciviruses produce the first viral proteins from the g-RNA as a polyprotein, which is subsequently cleaved by the viral protease into the mature nonstructural proteins. These proteins orchestrate not only replication of the genome but also transcription of the sg-RNA, which serves as a template for translation of the two structural proteins building the viral capsid. Different calicivirus genera have evolved different mechanisms to express major capsid protein VP1, in addition to the nonstructural proteins, from the g-RNA. Lago-, nebo-, and sapoviruses express VP1 from the g-RNA as part of the ORF1-encoded polyprotein and noroviruses via a TURBS-based termination/reinitiation mechanism (14, 15). For vesiviruses, VP1 is not part of the ORF1 polyprotein. It is expressed from a separate ORF from the sg-RNA as a precursor protein with a leader of the capsid (LC) protein, which is cleaved by the viral protease to obtain the mature VP1 protein. We show here for the first time that vesiviruses (FCV) also express VP1 directly from the g-RNA. The precursor protein tLC-VP1 found to be translated from the g-RNA is smaller than the one obtained from the sg-RNA (LC-VP1). Hence, translation initiation has to occur downstream of the regular start codon for LC-VP1 expression. The protein analyses conducted revealed that this initiation site has to be located at or close to the second AUG codon in frame (M86), since nLuc fusion proteins from infection and transient expression systems exhibited migration rates very similar to those seen with engineered tLC control proteins in the SDS-PAGE (Fig. 3B).

Standard eukaryotic translation initiation is dependent on an AUG start codon embedded in a favorable Kozak sequence context. Scanning from the 5′ end of the RNA is not feasible for tLC-VP1 translation initiation since the putative start site is located 5,569 nt downstream of the g-RNA 5´ end. Thus, its translation initiation must be achieved via an alternative mechanism. Our analyses defined codon M86 (AUG2) of the LC coding region as the start codon for tLC-VP1 translation and revealed that the Kozak context of AUG2 had a limited influence on translation initiation (Fig. 5). Most substitutions of the M86 codon led to no significant reduction or just a minor reduction of tLC-VP1 translation, except for mutant M86C. The prominent effect of substitution of the AUG to UGC on tLC-VP1 translation was caused by an AUG codon in a –1 frame created by this substitution. Our data show that this mutation led to an additional translation event starting at the new AUG codon, generating an aberrant peptide and thus leading to a lower initiation rate for the tLC-VP1 protein. Our analyses showed that the AUG at position 86 was at least predominantly used as an initiation codon and that, when this codon was altered, initiation mainly took place at the next AUG located downstream at position M103. Translation of tLC-VP1 was also able to start upstream of codon 86 when an AUG codon was introduced there (Fig. 5C; see also Fig. 12), but this was possible only downstream of codon 1 as mutation of this AUG codon had no influence on translation efficiency (M1C, Fig. 5C). Thus, ribosomal scanning took place starting at or very closely downstream of the 5′ end of the LC coding region for tLC-VP1 expression and translation initiation occurred at the next downstream AUG. A main issue is how the recruiting of ribosomes to initiate scanning is achieved. Deletion analyses showed that leader protein coding sequences upstream of codon 86 are important for the translation mechanism of the precursor tLC-VP1 from the g-RNA. The highest impact on tLC-VP1 translation was found for nucleotides 4 to 36 of the LC coding region (nt 5317 to 5349 of g-RNA) and for 6-nt deletions in this region, and even different substitutions of nucleotides 7 to 9 of the LC coding region (nt 5320 to 5322 of g-RNA) led to severe reductions of tLC-VP1 translation (Fig. 7 and 8). In contrast, the nucleotides upstream of this region, encompassing the 3´ end of the polymerase coding region and the LC start codon, had no impact on the downstream initiation for tLC-VP1 translation (Fig. 4 and 5). Two further regions within the leader protein coding sequence were found to be important for tLC-VP1 expression, since deletion of nt 5446 to 5457 (dL3-3, Fig. 9) led to a reduction to 50% of the original level and of nt 5485 to 5496 to a reduction to 25% (dL4-2, Fig. 10) but smaller deletions in the latter region had no impact. Thus, several rather short sequence elements are important for the translation initiation mechanism (Fig. 12). These are possible sites for interaction with other molecules and might include structural features leading to recruiting of ribosomes.

We used several experimental approaches to get more information on the underlying translation initiation mechanism. In contrast to noroviruses (14, 15), the vesivirus VP1 is not expressed via a TURBS-driven termination/reinitiation mechanism, because premature termination of ORF1 polyprotein translation had no effect on tLC-VP1 expression. Possible 18S rRNA interacting motifs were found in the leader protein coding region, namely, TURBS motif 1 (UGUGGGA, nt 5465 to 5471) (10) and the UACUCCC motif (nt 5444 to 5450) found to be located in a pseudoknot structure in tobacco etch virus (TEV) and to be essential for cap-independent translation through hybridization with 18S rRNA (27). But the sequences could be deleted without an severe impact on the tLC-VP1 translation level for the TURBS motif (Fig. 9, dL3-4) and with a reduction by only 50% for the TEV motif (Fig. 9, dL3-2 and dL3-3). Mutation of the TEV motif by 2-nt substitutions, impairing the putative interaction with the 18S rRNA or preserving it via G-U pairing, had both no effect on tLC-VP1 expression (data not shown). Thus, these motifs have no essential function for the analyzed translation mechanism. The most reasonable explanation of our results is that an internal ribosome entry site (IRES)-like element is located within the leader protein coding region (between codon 1 and codon 86) that drives rather inefficient translation of the VP1 precursor protein tLC-VP1 from the g-RNA. Despite some indicative data supporting the idea of the presence of an IRES element as determined using bicistronic reporter plasmids, we were not able to clearly demonstrate IRES activity (data not shown). Translation initiation mediated by this putative IRES element might be too weak to be shown in this assay, or additional FCV sequences might be needed to achieve translation initiation.

The LC protein has different functions in the viral life cycle. It causes a cytopathic effect; leads to downregulation of survivin and XIAP, thus inducing apoptosis; and is probably important for viral spread (19, 28). Abente et al. (19) showed that the N-terminal region of the leader of capsid protein is important for recovery of viable virus. Thus, mutations affecting the LC protein sequence might have an impact on LC function and, in consequence, on virus viability. Introduction of deletions into the LC N-terminal region, showing an effect on VP1 translation from the g-RNA in the firefly luciferase assay, prevented virus recovery or generated a virus showing no CPE. This might have been due to functional changes of the LC protein through deletion of the given amino acids or to the effect on tLC-VP1 translation from the g-RNA or to a combination of the two. Mutations M86C (AUG→UGC) and S3S (UCA→ACU) allowed recovery of viable virus. The amino acid change of M86C and reduction of tLC-VP1 translation (21% tLC-FF-VP1 translation) had no effect on virus growth properties (Fig. 11). The substitution for S3S did not change the LC amino acid sequence but reduced tLC-VP1 translation to 10% of the wt expression level. This virus showed impaired viral growth compared to the wt virus in a growth curve starting with a very low MOI of 0.0003 but grew to similar titers after 40 h. Thus, a severe reduction of tLC-VP1 expression inhibits early stages of virus replication. A less prominent reduction of tLC-VP1 translation as seen for mutant M86C did not impair virus growth; at least, such an effect was not visible in our replication analysis in cell culture.

VP1 may have a function in viral replication in the early viral life cycle as it interacts with the viral polymerase NS6/7 (29) and is already found to associate with the replication complex 15 min postinfection (20). It is tempting to speculate that some tLC-VP1 expression is needed for efficient initiation of viral replication in the early viral life cycle in the host. The difference between the effects of mutation M86C and mutation S3S on replication showed that the amount of VP1 needed for efficient replication is very low; a reduction to 21% of the wt level (M86C) showed no effect on replication, but a reduction to 9% seems to have undercut a threshold and thus showed some effect on replication.

Our findings close a gap in our knowledge on caliciviruses since we can now state that (i) all caliciviruses express VP1 from the g-RNA and (ii) that this expression is important for viral replication. Our work provides for the first time data supporting the conclusion that VP1 expressed from the g-RNA has a functional role for viral replication in an infection system and therefore contributes significantly to our view of calicivirus replication in general.

MATERIALS AND METHODS

Cells and viruses.

BHK-21 cells were kindly provided by T. Rümenapf, Veterinary University Vienna, Austria. Crandell Reese feline kidney (CRFK) cells (ATCC CCL94) were used for all infection experiments performed with FCV. Cells were grown in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with nonessential amino acids and 10% fetal calf serum (FCS). Vaccinia virus MVA-T7 was kindly provided by B. Moss (NIH, Bethesda, MD) and G. Sutter (Ludwig-Maximilians-Universität, Munich, Germany) (30, 31). FCV vaccine strain 2024 was kindly provided by K. Danner, Hoechst Roussel Vet GmbH.

Construction of recombinant plasmids.

Restriction and subcloning were done according to standard procedures (32). Restriction and modifying enzymes were purchased from New England Biolabs (Schwalbach, Germany), Thermo Fisher Scientific (Karlsruhe, Germany), and Fermentas GmbH (Sankt Leon-Rot, Germany). FCV full-length constructs (pJT15) were generated on the basis of pIK12 (22). In the pJT23_FFL constructs, the firefly luciferase substitutes for codons 15 to 449 of VP1 (nt 5729 to 7044). For the nLuc reporter plasmids, the NanoLuc (nLuc) luciferase gene (kindly provided by T. Hoenen, Friedrich-Loeffler-Institute, Isle of Riems) was inserted between codons 88 and 89 of the LC protein coding sequence via insertion of restriction sites for FseI and ApaI. pLC-nLuc constructs were generated by subcloning the LC and nanoLuciferase genes into the pCI vector (Promega, Heidelberg, Germany). Point mutations and deletions were introduced by standard PCR-based site-directed mutagenesis methods using thermostable Pfu polymerase (Promega, Heidelberg, Germany) and synthetic primers purchased from Metabion (Munich, Germany). The cloned PCR products were all verified by nucleotide sequencing with a BigDye Terminator cycle sequencing kit (PE Applied Biosystems, Weiterstadt, Germany). Further details of the cloning procedure and the sequences of the primers are available from us on request.

Expression, detection, and quantification of proteins and detection of RNA in the mammalian cell system.

Transient expression of plasmids in BHK-21 cells using vaccinia virus MVA-T7, metabolic labeling with [³⁵S]methionine and [³⁵S]cysteine (Hartmann Analytic, Göttingen, Germany), preparation of cell extracts, and recovery of immunoprecipitates were done as described previously (11) with antisera for VP1 (22) and for the polymerase (raised against bacterially expressed polymerase peptides). Double precipitation was used to ensure quantitative recovery of the proteins as tested before. The precipitates were combined and were separated by 10% PAGE and exposed on BioMax X-ray films (Kodak, Stuttgart, Germany).

For firefly luciferase assays, BHK cells were infected with vaccinia virus MVA-T7, transfected with the given pJT23FF constructs and a CMV-Renilla luciferase construct for normalization, harvested after 6 h with PLB lysis buffer from a Dual-Luciferase reporter assay system (Promega, Heidelberg, Germany), and analyzed with the kit substrates as described in the manual using a Berthold technologies Tristar2 LB942 multimode reader. The data presented here represent averages of results from at least three independent experiments. Statistical analysis was done in the form of a t test using GraphPad Prism software (Statcon GmbH, Witzenhausen, Germany).

For nanoLuciferase activities in the SDS-PAGE, a Nano-Glo in-gel detection system (Promega, Heidelberg, Germany) was used as given in the manual. The nanoLuciferase activity was analyzed with a Bio-Rad ChemiDoc XRS+ system and Image Lab software (Bio-Rad, Munich, Germany). For RNA detection, BHK cells were transfected as described above for the luciferase assay. A 5-μg volume of total RNA isolated via TRIzol (Fisher Scientific GmbH, Schwerte, Germany) was analyzed by agarose gel electrophoresis and transfer to a Duralon membrane (and hybridization with a DNA probe as described before [33]). A 1-kb fragment from the firefly gene labeled with [α-³²P]dCTP (Hartmann Analytic GmbH, Braunschweig, Germany) by nick translation (nick translation kit; Amersham Bioscience) was used as a probe.

Recovery and analysis of FCV mutants.

Recovery of infectious FCV from cloned cDNA and recording of growth curves were done as described before (22). Briefly, CRFK cells were seeded to 80% confluence in 3.5-cm-diameter dishes, infected with vaccinia virus MVA-T7 for 1 h, and transfected with pJT15 (FCV full-length cDNA clone) or mutants thereof. The cells were incubated with the transfection mixture Lipofectamine 2000 (Fisher Scientific GmbH, Schwerte, Germany) for 4 h and incubated for another 16 h in DMEM with 10% FCS. After this incubation time, a pronounced cytopathic effect (CPE) was visible that was at least mostly due to MVA-T7 vaccinia virus, since it was also detectable in mock-transfected cells (data not shown). Cells, together with supernatants, were subjected to three freeze-thawing cycles and passed through a 0.1-mm-pore-size sterile filter (Millex-VV; Millipore Products, Bedford, MA) to eliminate the vaccinia virus (34). A 100-to-400-μl volume of filtrate from each transfection reaction mixture was used to infect CRFK cells. Recovery of infectious virus was monitored by detection of characteristic FCV CPE, and the stability of the introduced mutation was checked by FCV-specific reverse transcription-PCR (RT-PCR) with primers FCV_5070f (5′-GGGGTACCTCACGTGGGCAGCAGCTTTGG-3′) and FCV_5770r (5′-AGCTGCAGTTGACATTTGGGC-3′) and by subsequent sequence analyses for up to 5 passages. Growth curves were recorded after infection of CRFK cells with a multiplicity of infection of about 0.0003 and were conducted with viruses from the third passage.