Development of a Rubella Virus Vaccine Expression Vector: Use of a Picornavirus Internal Ribosome Entry Site Increases Stability of Expression

Abstract

Rubella virus (RUB) is a small plus-strand RNA virus classified in the Rubivirus genus of the family Togaviridae. Live, attenuated RUB vaccines have been successfully used in vaccination programs for over 25 years, making RUB an attractive vaccine vector. In this study, such a vector was constructed using a recently developed RUB infectious cDNA clone (Robo). Using a standard strategy employed to produce expression and vaccine vectors with other togaviruses, the subgenomic promoter was duplicated to produce a recombinant construct (termed dsRobo) that expressed reporter genes such as chloramphenicol acetyltransferase and green fluorescent protein (GFP) under control of the second subgenomic promoter. However, expression of the reporter genes, as exemplified by GFP expression by dsRobo/GFP virus, was unstable during passaging, apparently due to homologous recombination between the subgenomic promoters leading to deletion of the GFP gene. To improve the stability of the vector, the internal ribosome entry site (IRES) of a picornavirus, encephalomyocarditis virus, was used instead of the second subgenomic promoter to eliminate homology. Construction was initiated by first replacing the subgenomic promoter in the parent Robo infectious clone with the IRES. Surprisingly, viable virus resulted; this virus did not synthesize a subgenomic RNA. The subgenomic promoter was then reintroduced in an orientation such that a single subgenomic RNA was produced, GFP was the initial gene on this RNA, while the RUB structural protein open reading frame was downstream and under control of the IRES element. GFP expression by this vector was significantly improved in comparison to dsRobo/GFP. This strategy should be applicable to increase the stability of other togavirus vectors.

Rubella virus (RUB) is an important pathogen of humans. RUB is a small, quasi-spherical, enveloped, nonsegmented, plus-strand RNA virus that is the sole member of the Rubivirus genus in the Togaviridae family (5). The RUB genome is roughly 10,000 nucleotides (nt) long and is capped and polyadenylated. In infected cells, three viral RNA species are synthesized: the genomic RNA, which also is the mRNA for translation of the nonstructural proteins (whose function is in viral RNA synthesis) from a long open reading frame (ORF) at the 5′ end of the genome; a complementary genome-length RNA of minus polarity which is the template for synthesis of plus-strand RNA species; and a subgenomic (SG) RNA which is initiated internally and contains the sequences of the 3′-terminal one-third of the genome and serves as the mRNA for the translation of the structural proteins (C [capsid protein] and two envelope glycoproteins, E1 and E2) from a second, 3′-proximal ORF. In the other togavirus genus, the alphaviruses, synthesis of the SG RNA is directed by a short (∼25-nt) sequence located immediately upstream from the SG start site known as the SG promoter (20).

Because RUB causes serious birth defects when infection occurs during the first trimester of pregnancy, live, attenuated vaccines were developed and have been used in vaccination programs in developed countries since 1970 (8). The standard vaccination strategy is universal vaccination of children at 15 to 18 months of age. The vaccine induces an immune response in over 95% of recipients and has been among the most successful live, attenuated vaccines developed. Because of their effectiveness and universal acceptance, a vaccine vector based on live, attenuated RUB vaccines would be highly desirable for use in a pediatric setting. Immunization with a RUB vector would result in induction of immunity against both RUB and the heterologous virus whose genes were expressed.

Infectious cDNA clones have been developed for a number of togaviruses including RUB (11). An infectious cDNA clone is a plasmid containing a cDNA copy of a viral genome positioned adjacent to an RNA polymerase promoter such that infectious in vitro transcripts can be synthesized. The infectious cDNA clones of several alphaviruses have been modified to produce vaccine/expression vectors, most notably Sindbis virus (SIN) (1, 9), Semliki Forest virus (6, 19), and Venezuelan equine encephalitis virus (15). The initial alphavirus vectors were engineered by duplicating the SG promoter, resulting in a virus that synthesized two SG RNAs, one from which the native structural protein ORF (SP-ORF) is translated and one from which the foreign gene is translated (alphavirus expression vectors were most recently reviewed in reference 17). Thus, our initial RUB vector was constructed by this strategy using the wild-type Therien strain RUB infectious clone Robo302 (which is based on the low-copy-number plasmid pCL1921 [11]). In the alphavirus vectors, the second SG promoter has been placed both between the ORFs and downstream of the SP-ORF within the 3′ untranslated region, which is 400 to 500 nt long in these viruses. However, since the RUB 3′ untranslated region is relatively short (60 nt) and the 3′ 300 nt (including the 3′ end of the SP-ORF) appear to be necessary for efficient virus replication (2, 3), we placed the additional SG promoter between the ORFs (Fig. 1 and Table 1). Since the RUB SG promoter has not been mapped, we duplicated a region consisting of the 3′-terminal 126 nt of the nonstructural protein ORF (NSP-ORF) and the entire 120-nt noncoding region between the NSP-ORF and the SP-ORF. A multiple cloning site (MCS) containing convenient restriction sites (including unique XbaI, BstBI, HpaI, and NsiI sites) was placed between the SG promoters for insertion of foreign genes. Thus, in this construct the SG RNA transcribed from the upstream SG promoter is translated to produce the foreign gene, while the SG RNA transcribed from the downstream SG promoter is equivalent to the standard SG RNA and is translated to produce the virus structural proteins. The plasmid was termed dsRobo302.

FIG. 1.

Genomic arrangements of the RUB constructs developed and used in this study. Robo302 (pCL1921 plasmid backbone [11]) and Robo402 (pBR322) contain the standard virus genome with its modular NSP- and SP-ORFs. Robo402 was additionally modified by the addition of an NsiI site immediately following the NSP-ORF to produce Robo402/NsiI. The region of the genome containing the putative SG promoter, nt 6260 to 6506, was duplicated by PCR; two amplicons were produced, the first using primers 106 and K1 (Table 1) and the second using primers K3 and 1. Following digestion of amplicon 1 with BglII and XbaI and amplicon 2 with XbaI and AscI, a three-fragment ligation was performed with Robo302 digested with BglII and AscI. GFP was PCR amplified from SINrep/GFP plasmid (obtained from I. Frolov) with primers that retained the initiation and termination codons of the GFP gene but added flanking XbaI and NsiI sites and cloned into the MCS of the dsRobo302 plasmid using these two enzymes. To make Robo402/IRES, a ∼600-nt amplicon containing the complete encephalomyocarditis virus internal ribosome entry site (IRES) was PCR amplified from pCEN plasmid (obtained from I. Frolov) using primers IR-5 and IR-3 (Table 1). This amplicon was rendered blunt ended with T4 DNA polymerase and then digested with NsiI. A second amplicon containing RUB sequence between the second codon of the SP-ORF and AscI (nt 7313) was PCR amplified using primers IRES-R and 1 (Table 1). This amplicon was digested with Eco47III and AscI. The two amplicons were then combined in a three-fragment ligation with NsiI-AscI-digested Robo402/NsiI. To produce siRobo/GFP, the BglII-NsiI fragment of dsRobo302/GFP was ligated into Robo402/IRES that had been restricted with these two enzymes. An siRobo402 vector containing the dsRobo402 MCS between the SG promoter and the IRES element was created by similar introduction of the BglII-NsiI fragment from dsRobo402 into Robo402/IRES. Production of in vitro transcripts from these plasmids and subsequent transfection of Vero cells were done as described previously (11).

TABLE 1.

PCR primer pairs used in vector construction^a

	Sequenceb	Restriction site(s)c
SG promoter duplication
Amplicon I
106 (U)	AGCTCACCGACCGCTAC (5319–5335)
K1 (D)	GCCTCTAGATTCGGGCACCCTGGGGCTCT (6488–6507)	XbaI
Amplicon II
K3 (U)	GAATCTAGAGGCCTTCGAACGCGTTAACATGCATGTCCTCGCTATCGTGCGCGAA (6260–6280)	XbaI, StuI, BstBI, MluI, HpaI, NsiI/Ppu10I
1 (D)	GAAGCGGATGCGCCAAGG (7323–7340)
Robo402/IRES construction
Amplicon I (EMCV IRES)
IR-5 (U)	CACAATGCATAATTCCGCCCCTCTCCCTC (NA)	NsiI
IR-3 (D)	CATGGTTGTGGCAAGCTTATC (NA)
Amplicon II
IR-R (U)	CGCTAGCGCTTCTACTACCCCCATCACC (6433–6453)	Eco47III
1 (D)	GAAGCGGATGCGCCAAGG (7323–7340)

^aThe sequences of oligonucleotide primers used in two manipulations, duplication of the SG promoter in Robo302, and substitution of the SG promoter in Robo402/NsiI with an IRES from EMCV are given. In both manipulations, two PCR amplicons were generated and the manipulation was done via a three-fragment ligation. The upstream primer (U) sequence is at the 5′ end of the amplicon with respect to the RUB genomic construct; the sequence of RUB nucleotides is thus colinear with the genomic sequence. The downstream primer (D) sequence is at the 3′ end of the amplicon with respect to the RUB genomic construct; the sequence of RUB nucleotides is thus complementary with the genomic sequence. 

^bNucleotides in the primers containing RUB sequences are underlined; those in the genome (numbered from the 5′ end) to which the nucleotides in the primer are colinear or complementary are given in parentheses. NA, not applicable. 

^cRestriction site sequences in the primer used for cloning and the corresponding name are in bold. In the case of primer K3, used to create the MCS in dsRobo302, several restriction sites are present; they are alternately shown in bold and all italics. 

To test expression, the reporter genes chloramphenicol acetyltransferase (CAT) and green fluorescent protein (GFP) were introduced into dsRobo302. When in vitro transcripts from dsRobo302, dsRobo302/CAT, and dsRobo302/GFP were used to transfect Vero cells, virus was recovered. CAT expression was detected by both immunoprecipitation followed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (Fig. 2) and CAT enzyme assay using lysates from infected cells (18). In the experiment shown in Fig. 2, CAT expression during a similar radiolabeling window from a corresponding SIN vector, pTE5′2J/CAT, was also assayed. Due to the growth differences and degree of cytopathic effect induced in infected cells, radiolabeling was done at 25 h posttransfection for the SIN vector and 41 h with the dsRobo vector. Predictably, expression was greater with the SIN vector, but expression with the dsRobo vector was readily detectable. When an enzyme assay (18) was used to quantitate the difference in expression using lysates prepared at the same times posttransfection, CAT expression by the SIN vector was approximately 7.5 times greater than CAT expression by the dsRobo vector (data not shown). When lysates were prepared 4 and 6 days after transfection with dsRobo302/CAT, CAT expression increased 1.4- and 1.8-fold in comparison with the 2-day lysate.

FIG. 2.

Immunoprecipitation of CAT expressed by dsRobo and SIN vectors. Vero cells were mock transfected (MOCK) or transfected with dsRobo302/CAT transcripts or transcripts from a double-subgenomic SIN vector expressing CAT, pTE5′2J/CAT (dsSIN/CAT). The cells were metabolically radiolabeled with [³⁵S]methionine (1,000 Ci/mmol; Amersham) for 1.0 h at 25 (dsSIN/CAT) or 41 (MOCK and dsRobo/CAT) h posttransfection, followed by lysis with radioimmunoprecipitation buffer, immunoprecipitation using an anti-CAT monoclonal antibody (5′Prime-3′Prime, Inc.), and SDS-PAGE as described previously (4). The molecular weight standards (MW) (from top to bottom) are 200, 97, 68, 43, and 29 kDa; CAT is marked.

GFP expression by the dsRobo vector was detected by examining living Vero cell cultures infected with dsRobo/GFP virus under a microscope with epifluorescence capability (data not shown) and by immunoprecipitation (Fig. 3), and the percentage of cells in an infected culture expressing GFP was determined by flow cytometry (Fig. 4). Properties of the dsRobo and dsRobo/GFP viruses were analyzed in greater detail. The majority of plaques formed by P0 (passage 0) dsRobo and dsRobo/GFP virus (virus produced by transfected cultures) were smaller than Robo302 virus plaques; however, ∼1% were similar in size to Robo302 virus plaques. P0 dsRobo and dsRobo/GFP virus titers were roughly 5 × 10⁵ PFU/ml, in comparison to average P0 Robo302 virus titers of 5 × 10⁶ PFU/ml. When intracellular RNA from infected cells was analyzed, as expected, the genomic RNAs of both dsRobo and dsRobo/GFP virus were larger than Robo302 virus genomic RNA, and both produced an additional, longer SG RNA not found in cells infected with Robo302 virus (Fig. 5A). The intensities of the two SG RNA bands relative to the genomic RNA were similar to each other and to the SG/genomic RNA ratio in Robo302 virus-infected cells, indicating that both SG promoters retained the functional efficiency found in standard virus.

FIG. 3.

Immunoprecipitation of GFP expressed by dsRobo (A) and siRobo (B) vectors. Vero cells were mock infected (Mock), infected with Robo402 virus (R402), Robo402/IRES virus (402/IRES), or a passaged stock of dsRobo/GFP or siRobo/GFP virus. In these multiple passages, P0 is virus recovered from transfection which was subsequently passaged at a low MOI (∼0.1 PFU/cell) to produce P1, P2, etc. For this experiment, the MOI for each virus stock was adjusted to ∼1 PFU/cell. Three days postinfection, cells were metabolically radiolabeled with [³⁵S]methionine (1,000 Ci/mmol; Amersham) for 1.5 h followed by lysis with radioimmunoprecipitation buffer, immunoprecipitation using an anti-GFP polyclonal immunoglobulin G (Clontech), and SDS-PAGE (4). In each panel, the three molecular weight standards (MW) are (from top to bottom) 68, 43, and 29 kDa; GFP is marked.

FIG. 4.

Percentage of cells in cultures infected with dsRobo/GFP and siRobo/GFP viruses expressing GFP. Vero cells were infected at an MOI of 1 PFU/cell with dsRobo/GFP or siRobo/GFP stocks produced by multiple low-MOI passages (virus recovered from transfected cells, designated P0, was passaged in Vero cells to produce P1, P2, etc.). Three to four days postinfection, when 100% of the cells are infected with Robo302 virus under these conditions (10), to determine the percentage of cells expressing GFP, the infected cultures were trypsinized, and the cells were resuspended in medium and subjected to fluorescence-activated cell sorting analysis using a Becton Dickinson FACS Calibur flow cytometer (equipped with a 388-nm, 16-mW argon laser) with CellQuest software (Becton Dickinson); 20,000 events were used to determine each percentage.

FIG. 5.

Virus-specific RNAs produced by Robo constructs. Vero cells were mock infected (Mock) or infected at an MOI of ∼1 PFU/cell with Therien strain RUB (WT [wild type]), Robo302 or Robo402 virus (R302 or R402), or stocks of dsRobo, dsRobo/GFP, Robo402/IRES (402/IRES), or siRobo/GFP viruses passaged one (P1), three (P3), or five (P5) times in Vero cells (MOI of ∼0.1 PFU/cell at each passage) (in panel B, the dsRobo/GFP virus [ds/GFP] was P1). Three days postinfection, total cell RNA was extracted and subjected to agarose gel electrophoresis and virus-specific RNA species were detected by Northern hybridization using a probe complementary to the RUB SP-ORF ([³²P]CTP-labeled negative-polarity RNA transcripts synthesized from pRUB-SP-ORF [7]). The amount of radioactivity present in RNA bands on autoradiographs was quantitated by densitometry with a Fujix BAS1000 Bio Imaging analyzer (Fuji Photo Film, Tokyo, Japan), using software provided by the manufacturer. G, genomic RNA; 28S, the 28S cell rRNA which causes a background blob; SG1, the standard SG RNA; SG2, SG RNAs engineered for expression of foreign genes. In the Robo402/IRES lanes, a faint band of unknown identity is marked with an arrowhead.

When P0 dsRobo/GFP virus was passaged (multiplicity of infection [MOI] of 0.1 PFU/cell, with harvest at 5 to 6 days postinfection), the level of GFP expression diminished and was undetectable by radioimmunoprecipitation by P3 (Fig. 3A). The percentage of cells in infected cultures expressing GFP declined precipitously through P3, and GFP-positive cells were not detectable in the fourth and subsequent passages (Fig. 4). Thus, GFP expression was unstable, as has been encountered with double-subgenomic alphavirus vectors (13, 14). Analysis of intracellular viral RNA revealed that by P3, the dsRobo/GFP and similarly passaged dsRobo viruses synthesized no detectable second SG RNA, and the genomic RNA of these viruses was the same size as that of Robo302 virus (Fig. 5A). This suggests that GFP expression had been lost due to homologous recombination between the two SG promoters, which would restore the genomic RNA to the size of standard virus. Concomitant with loss of GFP expression, P2 and later-passage dsRobo and dsRobo/GFP virus produced plaques similar in size to Robo302 virus plaques.

To eliminate the possibility of homologous recombination in the RUB vector, we next investigated whether an IRES element could be incorporated into our RUB expression vector in place of the second SG promoter. Construction of this vector, described in the legend to Fig. 1, was initiated by replacing the SG promoter with the IRES in Robo402 (a pBR322 derivative of Robo302). Surprisingly, transcripts from this construct, Robo402/IRES, gave rise to viable virus which formed plaques on Vero cells. The average P0 titer of Robo402/IRES virus was 8.5 × 10⁵ PFU/ml; the titer rose to 2.4 × 10⁶ PFU/ml at P3 and 6.0 × 10⁷ PFU/ml at P5. As shown in Fig. 5B, the predominant virus-specific RNA species in Robo402/IRES virus-infected cells was the genomic RNA. A faint band of with a size slightly larger than that of the standard SG RNA was present. The ratio of the intensity of this band relative to the genomic RNA was 0.08 in P1 and declined to 0.006 and 0.003 in P3 and P5, respectively (in comparison, the SG/genomic intensity ratio was 1.2 in Robo402-infected cells). Therefore, although the identity of this band was not determined (for example, it could have been due to adventitious use of the IRES as an SG promoter), it is doubtful that it plays a significant role in Robo402/IRES virus replication.

To complete construction of the vector, the SG promoter followed by the GFP gene was introduced into Robo402/IRES to produce siRobo402/GFP. Virus produced from this construct should synthesize a single SG RNA; in this SG RNA, the GFP gene is 5′ proximal and is followed by the IRES and the SP-ORF (Fig. 1). Transcripts from siRobo402/GFP gave rise to virus following transfection of Vero cells. P0 titers of siRobo/GFP virus were 3 × 10⁴ PFU/ml but rose to 4 × 10⁶ PFU/ml at P3 and 1.2 × 10⁷ PFU/ml at P5. GFP expression by siRobo/GFP virus was relatively stable through P5 as assayed by both immunoprecipitation (Fig. 3B) and flow cytometry of infected cultures (Fig. 4). P0 siRobo/GFP virus formed small opaque plaques, and this was the majority plaque morphology through P5, when ∼10% of the plaques had Robo402 virus morphology. Analysis of intracellular virus-specific RNA revealed that the presence of the genomic RNA and an SG RNA larger than the standard SG RNA in P1 siRobo/GFP-infected cells, as expected. However, by P3, a band of intermediate size between the siRobo/GFP SG RNA and the standard SG RNA was present, and by P5 a band of similar in size to the standard SG RNA appeared. Concomitantly, a shorter genomic RNA band appeared with a size similar to the size of the standard genomic RNA. Thus, deletion events occurred during passage of siRobo/GFP virus, but at a much lower rate compared to dsRobo viruses (particularly in Fig. 4, it can be seen that GFP expression by siRobo/GFP virus declined to some extent in the passages during which these deletion events occurred).

Thus, we have successfully constructed RUB vectors which could be useful as both vaccine and expression vectors, and this is the first report of the use of RUB as a recombinant vector. The siRobo vector exhibited greater stability of foreign expression and the strategy of using an IRES to increase stability is also applicable to alphavirus vectors. Both dsRobo and siRobo vectors with MCSs have been developed. Since RUB replicates in a variety of vertebrate cell types and in most of these replication is to low titers and without accompanying cytopathogenicity (unlike the Vero cells used in this study), the niche for a RUB expression vector would be for low-level expression without a drastic effect on the cell, which has been a problem with some of the highly cytopathic alphavirus vectors (17). While the expression experiments in this study used reporter genes, we have successfully expressed a truncated form of the immunogenic E proteins of Japanese encephalitis virus in both dsRobo and siRobo as prototype vaccine candidates (data not shown). Live, attenuated RUB vaccines have been universally accepted as effective and safe in childhood immunization programs. Thus, a RUB-based vaccine would be best used in a pediatric setting to target systemic pathogens against which universal immunization was desired, such as human immunodeficiency virus, respiratory syncytial virus, or one of the hepatitis agents; a cocktail of RUB-based vaccines targeting different pathogens could be used to induce immunity simultaneously against each pathogen targeted in the cocktail. To the end of developing a RUB vector acceptable as a vaccine vector, we recently constructed an infectious cDNA clone based on the RA27/3 vaccine strain (12), the vaccine used in the United States and Europe.

While the focus of this study was vector development, the results did add to our understanding of RUB replication strategy. First, the ability of dsRobo virus to synthesize two SG RNAs with equal efficiency demonstrates that the RUB SG promoter is somewhere within the duplicated region, i.e., 170 nt upstream from the SG RNA start site. The dsRobo302/GFP construct will be of use in mapping the precise boundaries of the SG promoter. Second, we unexpectedly discovered that RUB was viable with an IRES element in place of its SG promoter. A number of other virus families of vertebrates (the caliciviruses, astroviruses, and hepatitis E virus) and plants have evolved a modular expression strategy similar to that used by togaviruses in which the nonstructural and structural proteins are translated from two different ORFs. In all of these viruses, expression of the 3′-proximal structural protein ORF is driven by an SG promoter. Our results show that an IRES can function in this capacity as well in the absence of an SG RNA. Interestingly, it was recently shown that in the insect picorna-like virus family, which have a 3′-proximal structural protein ORF and synthesize no SG RNA, expression of this ORF is driven by an IRES element (16). However, while the IRES element was retained during multiple passaging of Robo402/IRES virus, passaging of the siRoboRUB/GFP virus which contained both the SG promoter and IRES resulted in deletions within the genome and reappearance of an SG RNA similar in size to the standard SG RNA, indicating that SG RNA synthesis was preferred.