A Single Codon in the Nucleocapsid Protein C Terminus Contributes to In Vitro and In Vivo Fitness of Edmonston Measles Virus

Abstract

The major inducible 70-kDa heat shock protein (hsp72) increases measles virus (MV) transcription and genome replication. This stimulatory effect is attributed to hsp72 interaction with two highly conserved hydrophobic domains in the nucleocapsid protein (N) C terminus of Edmonston MV. These domains are known as Box-2 and Box-3. A single amino acid substitution in Box-3 of Edmonston MV (i.e., N522D) disrupts hsp72 binding. The prevalence of the N522D substitution in contemporary wild-type MV isolates suggests that this sequence has been positively selected. The present work determined if the N522D substitution enhances viral fitness and the degree to which any fitness advantage is influenced by hsp72 levels. Both parent Edmonston MV (Ed N) and an N522D substitution mutant (Ed N-522D) exhibited similar growth on Vero and murine neuroblastoma cells and in cotton rat lung, although Ed N-522D virus exhibited an attenuated in vitro response to hsp72 overexpression. In contrast, mixed infections showed a significantly reduced in vitro and in vivo fitness of Ed N-522D virus. Results support the involvement of additional selectional pressures that maintain the circulation of virus containing N-522D despite the cost to viral fitness.

The measles virus (MV) RNA genome is packaged by the viral nucleocapsid protein (N) to form a helical nucleocapsid. The carboxyl-terminal 125 amino acids of the 525-amino-acid N protein (i.e., N_TAIL) is exposed on the surface of the nucleocapsid (8, 9). Sequence variability in the N protein C terminus reflects the fact that this protein domain is intrinsically disordered, imparting structural plasticity that allows it to mediate interactions with a variety of cellular and viral binding partners that include the viral P protein (1, 2, 10), the major inducible 70-kDa heat shock protein (hsp72) (29, 30), the cellular nucleoprotein receptor (12, 13), and interferon-responsive factor 3 (24). Binding events are localized to patches of hydrophobic amino acid side groups whose sequence is well conserved relative to the hypervariable sequence that is otherwise characteristic of N_TAIL (4). These regions of exposed hydrophobicity are referred to as Box-1, Box-2, and Box-3.

Stable complex formation between the viral P protein and N_TAIL involves both high-affinity binding to Box-2 and low-affinity binding to Box-3 (2). By also binding the viral L protein, P tethers the viral L protein to the nucleocapsid template in support of transcription and genome replication, and it is this essential function that is the probable basis for constraints in Box-2 and Box-3 sequence variability. hsp72 also exhibits a high binding affinity for Box-2 and a low binding affinity for Box-3, a fact that would allow hsp72 to destabilize P/N_TAIL complexes (29, 30). It has been postulated that it is necessary to diminish the affinity between P and N_TAIL in order to promote the repetitive cycle of binding and release that underlies polymerase processivity (2). Such a model could explain the increases in MV transcription and genome replication that occur in cells expressing elevated levels of hsp72 (25, 26, 29).

A naturally occurring Box-3 sequence polymorphism has been identified that divides MV isolates into hsp72 binding and nonbinding variants (30). hsp72 binding variants typify MV isolates within the group A genotype, which includes both wild-type and derivative vaccine lineages. Loss of hsp72 binding is associated with an N522D substitution that is observed in 94% of 155 non-genotype A strains that we have examined. The high degree of conservation of hydrophobic and basic residues within Box-3 is well illustrated in several publications (11, 20, 22), and the important role of Box-3 in polymerase template function highlights the potential significance of the prevalence with which the charged side group of aspartic acid is found in place of the noncharged side group of asparagine. When the N522D substitution is introduced into the genomic backbone of Edmonston B MV (Ed MV), a prototype vaccine strain, basal transcription is unaffected, whereas the viral transcriptional response to elevated hsp72 levels is lost (29, 30).

Ed MV has a long history of replication in and adaptation to cell culture conditions. However, the identification of wild-type isolates in genotypes C1 and D6 -8 that contain 522N, independent of mode of isolation or geographic location from which the strain was derived, supports the view that N proteins containing asparagine at position 522 are not simply the result of cell culture adaptation. Direct evidence that an N protein containing 522N is or is not a reflection of tissue culture adaptation must await more extensive passage of wild-type isolates on Vero or CHO cells stably transfected to express SLAM (CD150), the receptor used by wild-type virus. For example, isolates such as IC-B, WTF, and BIL (which contain N-522D) can then be analyzed for retention of the N-522D and pathogenicity in animal models.

The question addressed in the present work is whether or not the N522D substitution has a cost or benefit in terms of in vitro or in vivo viral fitness. Prevalence of N522D suggests a fitness advantage, whereas the diminished hsp72 responsiveness suggests a fitness disadvantage, in that in vitro hsp72-dependent increases in viral gene expression may promote infectious progeny release (25, 26). We therefore examined in vitro fitness in the context of different hsp72 levels. The recombinant infectious Ed MV clone and the corresponding N-522D variant were used in coinfections of either control murine neuroblastoma or cells that constitutively overexpress hsp72. Murine cells were used due to their permissiveness to Ed MV infection and their lack of basal hsp72 expression, thereby maximizing differences between basal (control) and elevated hsp72 expression levels. Lack of basal (constitutive) hsp72 expression is also characteristic of rodent tissues in vivo (3), strengthening comparisons of in vitro growth characteristics to those observed in rodent models of infection. Coinfection of Vero cells was used to establish the degree to which results are cell line dependent, using hyperthermic treatment to induce elevated hsp72 levels prior to infection. Individual infections with either Ed N or Ed N-522D MV confirmed comparable infection parameters on control cells and the attenuated hsp72 responsiveness of the N-522D variant. Dual infection of cotton rats established the correlation between in vitro and in vivo fitness differences.

MATERIALS AND METHODS

Viral variants.

Infectious virus containing the N-522D variant was generated as previously described (29). In brief, the asparagine at amino acid 522 of Edmonston B MV was changed to aspartic acid by oligonucleotide site-directed mutagenesis of a cassette inserted into the full-length MV genomic cDNA plasmid (p+MV) (19). Infectious virus was rescued from Hep-2 cells transfected with plasmid containing the genomic construct in addition to the three plasmids encoding the Edmonston MV N (pT7MV-N), P (pT7MV-P), and L (pT7MV-L) proteins. Progeny from individual plaques were amplified to generate low-pass viral pools. Sequence of virus from these pools confirmed the presence of the desired C-terminal N gene mutation.

Generation and characterization of transgenic murine neuroblastoma cells.

hsp72 transgenic murine neuroblastoma cells (ATCC CCL-131, also designated Neuro-2a or N2a) were established by stable transfection to constitutively express hsp72. Generation and characterization of these cell lines were performed in a manner identical to that previously described for human astroctyoma cells (ATCC HTB-17, also designated U373) (26). Plasmid DNA contained the human hsp72 gene (hsp70A) driven by a β-actin promoter and was designated pHβhsp70 (28). The plasmid vector alone (pHβ) was used to generate stably transfected control cell lines. The constructs contain a neomycin resistance gene driven by the simian virus 40 promoter. Cells transfected with either the hsp72 expression construct or vector alone were selected with 400 mg/liter G418 (Geneticin; Gibco) and cloned by limiting dilution in 96-well plates. Levels of hsp72 were analyzed by Western blot analysis of whole-cell lysates using monoclonal antibody SPA-810, which specifically recognizes hsp72, and monoclonal antibody SPA-820, which recognizes an epitope shared between hsp72 and hsp73 (hsc70; StressGen Biotechnologies, Victoria, BC, Canada). Southern blot analysis of total cell DNA was performed to establish transgene copy number and to confirm the isolation of genetically distinct clones. A probe was generated by random primer extension of an 898-bp BglII fragment of the hsp70A coding region. The method has been described previously in the characterization of transgenic U373 cells (26). Thermotolerance of hsp72-overexpressing clones was used to establish the functionality of hsp72 expressed from the transgene (26). For this purpose, cells in log-phase growth were exposed to a 45°C hyperthermic treatment for 1.5 to 4.5 h, followed by measurement of the colony-forming ability of serial cell dilutions relative to that of vector-transfected control cells.

Characterization of in vitro viral infection parameters.

Growth curves were established at a culture incubation temperature of 37°C in hsp72-overexpressing murine neuroblastoma cells and control cells. Duplicate flasks were harvested at regular postinfection intervals by scraping the monolayers into the culture supernatant, subjecting the lysate to two freeze-thaw cycles, and then titrating each lysate in duplicate on Vero cells to calculate the 50% tissue culture infective dose (TCID₅₀)/ml. Growth curves were similarly established on control and preconditioned Vero cells (i.e., cells exposed to a 1.5-h, 43°C heat shock, 16 h prior to infection) (29).

Northern blot analysis of total cell RNA was used to quantify levels of viral transcript and genomic RNA as previously described (26). Viral minus- and plus-strand-specific [α-³²P]UTP-labeled riboprobes were prepared from Ed MV N and H cDNAs (21). The blots were first hybridized with riboprobe detecting minus-sense genome, the signal was quantified by phosphorimager analysis, the blots were stripped (Strip-EZ RNA kit; Ambion), and the membranes were subsequently probed for H, with the procedure being repeated for the detection of viral N signal. Equivalent loading was confirmed based upon cellular glyceraldehyde 3-phosphate dehydrogenase (GAPDH) transcript signal from a 359-base riboprobe derived from the mouse GAPDH coding sequence (pTRI-GAPDH-Mouse; Ambion).

Analysis of in vitro fitness.

Mixtures of Ed N and Ed N-522D viruses were prepared in defined ratios for total multiplicities of infection (MOIs) of 0.01, 0.1, 1.0, and 10.0. Cells were infected in triplicate. Cell culture supernatants were harvested at 60 h for N2a cells and 48 h for Vero cells, and total RNA was extracted (QIAamp viral RNA mini kit; QIAGEN). Viral cDNA was synthesized at 37°C with Moloney murine leukemia virus reverse transcriptase (Invitrogen) using the sense PCR primer (5′-TTGACACTGCATCGGAGTCCA-3′) corresponding to genome nucleotide (nt) positions 1531 to 1551. PCR was performed using the sense and antisense (5′-TCGTGGGAGTGGATGGTTGAT-3′) primer (genome nt 1777 to 1791) with 32 amplification cycles as follows: denaturation at 95°C for 30 s, primer annealing at 56°C for 30 s, and extension at 72°C for 1 min. PCR amplicons were column purified (QIAquick PCR purification kit; QIAGEN), and the concentration was calculated using 260/280-nm light absorbance (ND-1000 spectrophotometer; NanoDrop Technologies). Approximately 250 ng of amplicon was digested with BsmAI (10 U) for 3.5 h at 55°C. The amount of amplicon derived from either Ed N or Ed N-522D was estimated by quantifying the DNA that was uncut (267 bp, representing Ed 3′ N and 5′ P sequence) and cut (a single band representing 134- and 133-bp fragments of the Ed N-522D amplicon). Bands were resolved by 2.0% agarose gel electrophoresis and stained with ethidium bromide, and ethidium bromide staining intensity was quantified using an Alpha Imager (Alpha Innotech).

Analysis of in vivo fitness.

Female specific-pathogen-free cotton rats (inbred strain COTTON/NIco; Harlan), 6 to 8 weeks of age, were infected with Ed N, Ed N-522D, or a 1:1 ratio of Ed N and N-522D. The total dose was 1 × 10⁵ TCID₅₀ in less than 100 μl, administered via the intranasal route to isofluorane-anesthetized animals. Four rats were used in each treatment group, and one rat received saline intranasally as an uninfected control. Rats were euthanized at 4 days postinfection (p.i.) using CO₂ asphyxiation, and the lungs were removed and weighed. The right lung was minced and homogenized, and serial 10-fold dilutions were prepared in 1× MEME for viral titration on Vero cells. Total RNA was isolated from homogenates generated from the left lung, and N_TAIL genomic sequence was amplified using a nested reverse transcriptase PCR (RT-PCR). Viral cDNA was synthesized using the sense primer 5′-GGCAAGAGATGGTAAGGAGG-3′ (genome nt 1201 to 1220). PCR was performed using the sense and antisense (5′-GTTTGCTGAGACCCGAACTG-3′) primer (genome nt 1974 to 1993) with 40 amplification cycles as follows: denaturation at 95°C for 1 min, primer annealing at 55°C for 1 min, and extension at 72°C for 2 min. The product of this reaction is a 739-bp amplicon, and 2.5 μl of the reaction product was used in a PCR using primers and conditions described for the in vitro fitness analysis (above). Restriction analysis of amplicons was also as described above. RT-PCR signal intensity from total RNA extracted from a tissue homogenate is proportionate to the amount of infectious virus that can be recovered from that tissue (3), with RT-PCR and nested RT-PCR yielding comparable results.

RESULTS

N-522D virus exhibits an attenuated response to hsp72 overexpression in murine cells.

hsp72 overexpression in murine neuroblastoma cells was based upon stable transfection with a plasmid construct directing expression of the human hsp70A gene from a β-actin promoter. Two vector-transfected (N2a-V₁ and N2a-V₂) and two hsp72-overexpressing (N2a-HSP₁ and N2a-HSP₂) cell lines were characterized (Fig. 1). Western blot analysis of total cell protein showed that the vector-transfected controls did not express detectable levels of hsp72, whereas high levels of hsp72 were produced in the overexpressing lines (Fig. 1A). Southern blot analyses indicated that N2a-HSP₁ and N2a-HSP₂ were distinct clones, based upon differences in construct copy number: 129 for N2a-HSP₁ and 258 for N2a-HSP₂ (data not shown). All four cell lines exhibited identical growth rates at 37°C. In each of three experimental trials, a 1.5-h exposure to 45°C did not affect colony-forming ability of N2a-HSP₁ and N2a-HSP₂, whereas that of vector-transfected controls was decreased by 34 to 73% relative to control cells not exposed to the 45°C treatment (Fig. 1B). The resistance phenotype exhibited by N2a-HSP₁ and N2a-HSP₂ cells supports the functional nature of the overexpressed hsp72.

FIG. 1.

Constitutive overexpression of hsp72 in stably transfected murine neuroblastoma cells (N2a). (A) Western blot analysis of total cell protein using an antibody recognizing the constitutively expressed and major inducible 70-kDa heat shock proteins (i.e., hsp73 and hsp72, respectively). hsp72 is expressed in construct-transfected cell lines N2a-HSP1₁ and N2a-HSP₂ but not vector-transfected control cell lines N2a-V₁ and N2a-V₂. (B) Thermotolerance of hsp72-expressing and control N2a cells as a measure of hsp72 function. The hsp72-expressing cell lines exhibited enhanced colony-forming ability relative to vector-transfected cell lines following exposure to 45°C for 1.5 to 4.5 h. (C) Northern blot analysis of Ed N viral transcripts in total RNA from infected hsp72-expressing and control N2a cells. Cells were infected at an MOI of 1.0, and the RNA was harvested at 24 h p.i. Phosphorimage analysis of signal intensities, corrected for variations in the loading control (GAPDH), showed that hsp72-expressing cells support stimulation of viral transcript production to a similar degree (i.e., approximately twofold); the mean increase was statistically significant (P < 0.05, t test).

Cell lines were infected with Edmonston MV (Ed N) at an MOI of 1.0 to assess uniformity in their support of viral gene expression. The two vector-transfected control cell lines supported similar steady-state levels of N transcripts at 24 h p.i., and the two hsp72-overexpressing cell lines supported approximately twofold increases in transcript levels at that same time point (Fig. 1C), an increase that was statistically significantly different from controls (P < 0.05, t test). Results are in agreement with a 1.7-fold increase in Ed N transcripts on preconditioned Vero cells relative to nonconditioned controls following infection at an MOI of 1.0 (29). The preconditioned phenotype is characterized by elevated levels of hsp72 induced by a transient heat shock.

Subsequent experiments used the N2a-V₂ and N2a-HSP₁ cells, since they were representative of vector-transfected or construct-transfected lines in terms of hsp72 expression profiles and support of viral RNA expression. Cells were infected at an MOI of 0.05 and subsequently analyzed at 36 to 84 h p.i., and results of three separate experimental trials were averaged. The low MOI was used in order to establish multistep growth curves that would allow the detection of subtle differences in replication between Ed N and Ed N-522D. Measurement of infectious progeny showed that parent Ed N and Ed N-522D virus have indistinguishable growth curves on vector-transfected neuroblastoma cells between 36 and 84 h p.i., with peak progeny release observed at 60 h p.i. (Fig. 2). The magnitude of progeny release was increased for both viruses on hsp72-overexpressing cells between 72 and 84 h p.i., although the magnitude of increase was statistically significant only for Ed N (P < 0.05, t test).

FIG. 2.

Cell-free infectious viral progeny release by hsp72-expressing N2a-HSP₁ and control N2a-V₂ murine neuroblastoma cells infected with either Ed N or Ed N-522D virus. Cells were infected at an MOI of 0.05, and progeny were measured between 36 and 84 h p.i. Results are an average of three consecutive experimental trials and are represented by the mean titer ± the standard error of the mean. Progeny release was increased on hsp72-expressing cells, although the increase was statistically significant (*) only for cells infected with the Ed N virus (P < 0.05, t test).

Total RNA was harvested at 24 to 84 h p.i. for Northern blot analysis of viral genomic and transcript RNA. The same low MOI (i.e., 0.05) was used in order to correlate RNA expression profiles to the growth curves defined above. Signals were quantified by phosphorimage analysis and corrected for variation in GAPDH signal. Viral genome levels were increased by greater than 10-fold in N2a-HSP₁ relative to N2a-V₂ cells for both Ed N and Ed N-522D virus at 60 to 72 h p.i. (Fig. 3A), the time of peak cell-free infectious viral progeny release. Viral N and H transcript levels were also increased on N2a-HSP₁ cells, but to a greater degree for parent virus than for Ed N-522D. This differential effect was illustrated by an approximately twofold greater transcript/genome ratio for Ed N relative to Ed N-522D at 60 to 72 h p.i., independent of whether the N or H transcript was used to calculate the ratios or whether the postinfection interval was 48, 60, or 72 h p.i. (Fig. 3B). Differences in these ratios were statistically significant (P < 0.05, t test).

FIG. 3.

(A) Northern blot analysis of viral genome and transcript levels in hsp72-expressing (N2a-HSP₁) and control (N2a-V₂) neuroblastoma cells infected with either Ed N or Ed N-522D virus at an MOI of 0.05. hsp72 expression enhanced genome and transcript levels for both Ed N and Ed N-522D virus, although the increase in transcript levels was attenuated for N-522D. The latter resulted in a greater transcript/genome ratio for Ed N relative to Ed N-522D virus on hsp72-expressing cells. (B) Ratios reflect phosphorimage signal intensities corrected for variations in GAPDH signal intensity, calculated independently for the N and H transcript, and represent the mean ± SD for samples harvested at 48, 72, and 84 h p.i. (i.e., time points providing readily quantifiable genome signals). Differences between Ed N and Ed N-522D were statistically significant (*, P < 0.05; t test). Results are representative of three experimental trials.

Ed N-522D exhibits reduced fitness relative to Ed N on control and hsp72-overexpressing cells.

We estimated the relative in vitro fitness of Ed N-522D compared to Ed N on hsp72-overexpressing N2a-HSP₁ and vector-transfected control N2a-V₂ cells by direct competition between the two strains. Progeny were quantified by RT-PCR and distinguished by restriction fragment length polymorphism of the amplicons. Precedents for using PCR-based fitness analyses have been established in other viral systems, including Sindbis virus (6), foot-and-mouth disease virus (5), and eastern equine encephalitis virus (27). Total RNA was isolated from cell-free infectious viral progeny. The genomic region encoding the C terminus of the N protein and 73 nt of the downstream noncoding sequence (nt 1491 to 1758 of the N gene) was amplified by RT-PCR. The AAT codon (nt 1671 to 1673) of Ed N encodes asparagine (amino acid 522), whereas the corresponding GAC codon of Ed N-522D encodes an aspartic acid and also introduces a unique BsmAI site into the cDNA. As a result, amplicons of Ed N and Ed N-522D can be distinguished by their differential susceptibilities to BsmAI, where digestion reduces the 267-bp Ed N-522D amplicon into two fragments of roughly equivalent lengths (134 and 133 bp). Gel electrophoresis and ethidium bromide staining intensities of the 267- and 133/134-bp fragments can therefore be used to establish the proportion of Ed N and Ed N-522D genomic RNA that is released from infected cells.

The relationship between amplicon signal intensities and amount of infectious virus was defined by mixing various ratios of Ed N and Ed N-522D used as the inoculum, extracting total RNA, and performing RT-PCR and analysis of restriction fragment length polymorphism as described above. Ethidium bromide signal intensities of the 267- and 133/134-bp amplicons were quantified by densitometry and plotted against the N-522D/N virus ratio that was used for the RT-PCRs. Linear regression analysis showed a very high degree of correlation between the N-522D/N virus and RT-PCR signal ratios (R² = 0.99) (Fig. 4). This linear relationship between input viral ratio and signal ratio within the range of ratios examined validates use of this approach in establishing the relative fitness of Ed N-522D virus.

FIG. 4.

Linear relationship between the ratio of infectious viral variants Ed N and Ed N-522D and virus-specific RT-PCR amplicon yield. Individual pools of titrated Ed N and Ed N-522D virus were combined at an N-522D/N ratio of 1:1, 1:2, 1:5, or 1:10. Total RNA was extracted, and a 267-nt genomic sequence spanning the N-P junction was amplified by RT-PCR. (A) The proportion of Ed N and Ed N-522D represented in the amplicons was based upon BsmAI restriction fragment length polymorphism. The 267-bp amplicon derived from Ed N virus is not cleaved, whereas the corresponding amplicon from Ed N-522D is cleaved to yield 134- and 133-bp fragments; these fragments resolve as a single band following 2% agarose gel electrophoresis. Reactions lacking RT were used as a negative control (-) for the RT-PCR. (B) Linear regression analysis of the ratio of N-522D and N signal intensities (i.e., ethidium bromide staining intensities of virus-specific amplicons) expressed as a function of the ratio of N-522D and N virus used in the assay. Mean signal ± SD was based upon three separate analyses. The line describing the relationship between the ratio of infectious viral variants and the ratio of virus-specific RT-PCR amplicons (y = 0.55x + 0.42) exhibits an excellent fit to the experimental data (R² = 0.99).

Both N2a-HSP₁ and N2a-V₂ cells infected with a 1:1 ratio of N and N-522D virus at a total MOI of 0.1 produced only the Ed N virus (Fig. 5). Negative controls included culture supernatants of cells challenged with Ed N virus exposed to 6 × 10⁴ μJ of short-wave UV light. Use of UV-inactivated virus did not result in an RT-PCR signal (data not shown). The relative intensity of the N-522D band in our competitions was too low to accurately quantitate fitness. Nevertheless, we could use the relative intensities of the digested product as a qualitative estimate of relative fitness. Results support a reduced fitness of Ed N-522D virus relative to Ed N. Increasing the total MOI in the mixed infections was used to determine if trans-complementation would allow rescue of Ed N-522D. When the MOIs of the N2a-HSP₁ and N2a-V₂ infections were increased to 1.0 and 10.0 while maintaining a 1:1 ratio of N and N-522D virus, there was a progressive increase in the relative frequency of Ed N-522D virus (Fig. 5). For N2a-V₂ cells, the ratio of N-522D/N amplicon signal increased from 0 at an MOI of 0.1 to 0.4 at an MOI of 1.0 and to 0.5 at an MOI of 10. Similar data were obtained in hsp72-overexpressing cells, where the N-522D/N signal ratio at increasing multiplicities of infection was not statistically significantly different from those observed on N2a-V₂ cells.

FIG. 5.

Relative fitness of Ed N-522D compared to Ed N based upon coinfection of hsp72-expressing N2a-HSP₁ and control N2a-V₂ cells. The inocula contained a 1:1 ratio of each virus for a combined MOI of 0.1, 1.0, or 10.0. The proportion of Ed N and Ed N-522D in the viral progeny was based upon BsmAI restriction fragment length polymorphisms of amplicons derived from RT-PCR of cell-free viral genomic RNA. Yield of amplicon was based upon ethidium bromide staining intensity of BsmAI-digested products resolved by gel electrophoresis (A). The ratio of the 133-/134- and the 267-bp products was used as a correlate of the relative fitness of Ed N-522D compared to Ed N (B). Results reflect the means of three experimental trials ± SD. Differences in ratios calculated from infection of N2a-HSP₁ and N2a-V₂ cells were not statistically significant. Controls for the infection were individual inoculations with Ed N or Ed N-522D virus, and reactions lacking RT were used as a negative control (-) for the RT-PCR.

Diminished recovery of Ed N-522D virus in cell-free progeny of mixed infections was also partially overcome by increasing the ratio of N-522D to N virus used in the infections (Fig. 6A). At an MOI of 1.0 and a 1:1 ratio of the two viruses, the Ed N-522D/Ed N signal ratio on N2a-V₂ cells was approximately 0.4 (as described above). The proportion of N-522D virus in the progeny increased to approximately 1:1 when the ratio of N-522D and N virus in the inoculum was 20:1. However, further increasing the ratio of N-522D to N virus in the inoculum to 50:1 or 100:1 did not enhance the recovery of N-522D in the viral progeny. Similar results were observed following infection of hsp72-overexpressing cells.

FIG. 6.

(A) Reduced fitness of Ed N-522D relative to Ed N was also illustrated by coinfecting hsp72-expressing N2a-HSP₁ and control N2a-V₂ cells with each variant at a total MOI of 1.0, using progressively increased proportions of Ed N-522D (i.e., from a 1:1 ratio with Ed N to a 100:1 ratio). The proportion of each viral variant in cell-free progeny was based upon BsmAI restriction fragment length polymorphisms of RT-PCR amplicons derived from genomic RNA, quantifying ethidium bromide staining intensities of products resolved on 2% agarose gels. Increasing the relative amount of Ed N-522D/N in the inoculum to 20:1 resulted in increased recovery of Ed N-522D in the progeny, although further increasing Ed N-522D in the inoculum was without effect. Results are representative of two experimental trials. Controls for the infection were individual inoculations with Ed N or Ed N-522D virus, and reactions lacking RT were used as a negative control (-) for the RT-PCR. (B) Recovery of Ed N-522D in the progeny of mixed infections was also enhanced by transfecting cells with plasmid supporting the expression of N-522D. Cells were infected with a 1:1 ratio of Ed N-522D and Ed N (combined MOI of 1.0) at 2 h posttransfection. Negative controls included nontransfected cells (-) and cells transfected with empty plasmid vector (V). Western blot analysis of uninfected cell lysates processed in parallel showed expression of the N-522D protein. Infections were performed in triplicate, and the Ed N-522D/N RT-PCR signal ratio is expressed as the mean ± SD.

As a corollary to increasing the ratio of Ed N-522D to Ed N in the inoculum, we increased expression of the N-522D protein via plasmid transfection while maintaining a 1:1 ratio of Ed N-522D and Ed N for infections performed at an MOI of 1.0. The T7-mediated plasmid expression of N-522D was performed as previously described by our group (29), and cells were infected at 2 h posttransfection. Nontransfected cells and cells transfected with an empty expression vector served as negative controls. All transfections/infections were performed in triplicate. Mixed infections in nontransfected or vector transfected cells yielded an N-522D/N signal ratio of approximately 0.4 (Fig. 6B), consistent with results presented in Fig. 6A and 5. Infection of cells transfected to express N-522D revealed a dramatic increase in the Ed N-522D/Ed N signal ratio, reproducing the effects of increasing the ratio of Ed N-522D and Ed N virus used in the viral challenge. Western blot analysis confirmed expression of the N-522D protein in transfected but not in uninfected cells that were processed in parallel (Fig. 6B). Collectively, these results support the contribution of either the N protein or the associated transcript in mediating fitness differences between the Ed N-522D and Ed N viruses.

Ed N-522D exhibits reduced fitness relative to Ed N on preconditioned and nonconditioned (control) Vero cells.

Our analysis of in vitro fitness of Ed N-522D was extended to include Vero cells in order to determine the extent to which our results were influenced by cell type. Multistep growth curves were first established on preconditioned and nonconditioned control Vero cells. Results were consistent with those established on the neuroblastoma cells. The Ed N and Ed N-522D viruses exhibited comparable growth kinetics on control Vero cells infected at an MOI of 0.01, yielding a mean peak progeny release of 3.16 × 10⁴ TCID₅₀/ml at 60 h p.i. (Fig. 7A). The growth kinetics on preconditioned cells was also similar between the two viruses, although the mean peak progeny release was increased 5-fold for Ed N and only 1.5-fold for Ed N-522D. Only the former was statistically significant.

FIG. 7.

(A) Infectious viral progeny release for parental Ed N compared to the Ed N-522D variant in control (C) and preconditioned (PC) Vero cells following infection at an MOI of 0.01. Results are an average of two separate experimental trials and expressed as the mean ± standard error of the mean (SEM). Preconditioning of Vero cells was achieved by exposing cells in the log phase of growth to 43°C for 1.5 h, a treatment elevating cytoplasmic hsp72 for 24 to 48 h posttreatment. Cells were infected at 16 h posttreatment. Peak infectious progeny release was statistically significantly elevated for Ed N but not Ed N-522D virus, representing increases of 5- and 1.5-fold, respectively. (B) Relative fitness of Ed N-522D compared to Ed N based upon coinfection of preconditioned and control Vero cells. The inocula contained a 1:1 ratio of each virus for a combined MOI of 0.01, or 1.0. The proportion of Ed N and Ed N-522D in the viral progeny was based upon BsmAI restriction fragment length polymorphisms of amplicons derived from RT-PCR of cell-free viral genomic RNA. Yield of amplicon was based upon ethidium bromide staining intensity of BsmAI-digested products resolved by gel electrophoresis.

Relative fitness was based upon infections performed at MOIs of 0.01, 0.1, and 1.0, using a 1:1 ratio of each virus in the inoculum. Viral progeny were analyzed at 48 h p.i. As observed in the neuroblastoma cells, Ed N-522D showed a reduced fitness relative to Ed N independently of hsp72 levels, and complementation operated at high MOIs (Fig. 7B). These results provide further support for the reduced fitness of Ed N-522D relative to Ed N, where the influence of cell lineage is on the magnitude of fitness differences.

Ed N-522D exhibits reduced fitness in airway infection of cotton rats.

The cotton rat is a well-established model for MV respiratory tract infection and has been used to show the differential effects of vaccine and wild-type viruses on immune suppression (14, 15). Growth curves for Ed N MV following intranasal inoculation have been established, showing that viral inoculum is only recovered through the first day postchallenge, after which the pulmonary titer of infectious virus increases to a peak at 4 days p.i. (14, 15). Intranasal inoculation of cotton rats with 1 × 10⁵ TCID₅₀ Ed N, Ed N-522D, or a 1:1 ratio of Ed N and N-522D virus resulted in the recovery of comparable amounts of virus from lung at 4 days p.i. (14, 15). The mean pulmonary titer (±standard deviation [SD]) for each inoculation group of four animals was (3.5 ± 1.1) ×10² TCID₅₀/g of lung for Ed N, (7.6 ± 4.9) ×10² TCID₅₀/g for Ed N-522D, and (7.8 ± 8.6) ×10² TCID₅₀/g for mixed infections. These titers were not statistically significantly different from one another. RT-PCR and restriction fragment length polymorphism of amplicons from total lung RNA showed that Ed N was the predominant virus recovered from mixed infections. The ratio of the N and N-522D amplicons was an average of 4.25:1 for the four animals receiving the mixed inoculum (Fig. 8).

FIG. 8.

Nested RT-PCR amplification of viral sequence from total lung RNA of cotton rats infected with 1 × 10⁵ TCID₅₀ of Ed N, Ed-N522D, or a 1:1 ratio of Ed N and N-522D. Samples were processed at 4 days p.i. Amplicons were digested with BsmAI to determine the proportion of the 267-bp (Ed N) versus 133/134-bp (Ed N-522D) product on 2% agarose gels stained with ethidium bromide. Negative controls included total lung RNA from uninfected rats (U) and infected rats in which the RT step was omitted (-). Ed-N is the predominant virus identified as a consequence of mixed infections.

DISCUSSION

Our results show that the codon for amino acid 522 of the MV N protein plays an important role in both in vitro and in vivo viral fitness. Reduced fitness of N-522D relative to the parent Ed N virus on hsp72-overexpressing cells would be predicted based upon results showing that an hsp72-interactive Box-3 sequence (i.e., 522N) promotes cell-free infectious viral progeny release when hsp72 levels are increased in the target cell. hsp72 levels do not have a significant stimulatory effect on progeny release for virus with a non-hsp72-interactive Box-3 sequence (i.e., 522D), despite the fact that levels of genomic RNA are enhanced within the infected cells. The latter indicates that genome replication is not rate limiting to progeny release in the neuroblastoma cells. Instead, a viral transcriptional product is more likely rate limiting, with the transcriptional response to hsp72 supplementation being selectively attenuated for the N522D virus.

The reduced fitness of Ed N-522D on control neuroblastoma cells or in cotton rats cannot be explained by individual differences in the magnitude or kinetics of cell-free infectious viral progeny release. Individual growth curves for Ed N and N-522D were virtually identical in vitro, and there was no significant difference in the titer of virus recovered from the lungs of infected cotton rats at 4 days p.i. Relevance of the in vivo model to MV infection of the respiratory tract and agreement between results of in vivo and in vitro fitness analyses suggest that the reduced fitness of Ed N-522D relative to Ed N is not conditional upon the specific infection system. Results of in vitro fitness experiments on hsp72-overexpressing (i.e., heat shocked) and control Vero cells further support this view.

The basis for fitness differences between Ed N and N-522D viruses may lie, at least in part, in the amino acid sequence of the N protein. This conclusion is based upon the enhanced recovery of Ed N-522D virus when coinfections with Ed N are performed at progressively higher MOIs, conditions that should support trans-complementation of N proteins. Further support for a role of the protein in mediating fitness differences is the result of using plasmid-based expression of N-522D to enhance recovery of Ed N-522D virus in mixed infections containing a 1:1 ratio of the viruses, mimicking the effect of increasing the ratio of Ed N-522D to Ed N in the inoculum. However, the overall results are inconsistent with a simple model in which the level of complementation is determined by the level of coinfection and where both proteins are equally shared by both strains (16). Thus, we cannot rule out additional effects, for instance, at the genomic level. Actual fitness differences may be greater than that calculated from direct measurement of amplicon signal intensities. Linear regression analysis showed that ratios of RT-PCR amplicon signal intensities change at a slower rate than does the ratio of the two viruses when viral inocula are analyzed (Fig. 4). We must therefore entertain the possibility that nucleotide differences between Ed N and N-522D also contribute to fitness differences independent of their coding capacities. The Ed N and N-522D viruses differ by two nucleotides within a single codon (nt 1671 to 1673); Ed N has AAT, whereas Ed N-522D has GAC. The latter represents a codon found in circulating wild-type viruses, particularly in genotype C2 (20). This codon was originally selected for the generation of a 522D virus because it was considered more genetically stable than use of a single nucleotide substitution to encode aspartic acid (i.e., GAT). Although we consider the possibility remote, fitness differences may exist between viruses that employ GAC versus GAT to encode aspartic acid.

The cost in fitness of the N522D substitution supports the involvement of additional selectional pressures that would assure its prevalence in wild-type MV isolates. Immunological pressure could directly or indirectly favor change in the composition of this viral motif. A direct role would be one in which 522N is part of a viral epitope mediating cell-mediated viral clearance. So far, however, amino acid 522 has not been described as part of either a CD4 or CD8 T-cell epitope, although one CD4 T-cell epitope (amino acids 483 to 502) has been mapped in close proximity (7). An indirect role may be one related to the hsp72 responsiveness of the Box-3 motif. An hsp72-interactive motif (i.e., 522N) could mediate increased viral gene expression in the face of elevated cellular levels of hsp72 and this, in turn, could augment primary or memory T-cell responses leading to viral clearance. MV infections in humans should occur in tissue environments of elevated hsp72 levels, consistent with the fact that fever is a consistent feature of MV infection (7) and that fever is a well-known inducer of hsp72 (23), supporting the relevance of the model. Further support comes from studies of Ed MV infection of mouse brain. Transient hyperthermic induction of hsp72 in tissues of neonatal BALB/c mice and subsequent intracranial inoculation with Ed MV resulted in accelerated clearance relative to control mice (3). The enhanced clearance is associated with increased MV-specific blastogenic responsiveness of splenic lymphocytes. Similar studies with N-522D virus have not been performed, although the model predicts that N-522D virus should resist hsp72-dependent clearance.

These findings underscore the fact that one can significantly underestimate the functional significance of MV sequence variants if infection parameters of variant viruses are examined in isolation. In the present case, acknowledging the apparent paradox between a sequence that reduces fitness and prevalence of that sequence in circulating wild-type virus points to the importance of additional selection pressures that drive viral evolution. Wild-type viruses encoding N-522D might have adapted to such pressure by acquiring compensatory changes in the genomic nucleotide sequence. Alternatively, the N522D substitution may be involved in epistatic interactions and confer advantage only in the presence of specific additional mutations. We have examined the relative fitness of Ed N to two different wild-type strains, WTF and BIL MV. Both of these strains contain the N protein 522D and use the identical codon used to create the N522D substitution in Ed N. Fitness was measured in Vero cells expressing CD150 (SLAM), the receptor of wild-type virus. Results of these experiments showed equivalent recovery of N-522D and N-522N for both WTF and BIL relative to Ed MV on the Vero-SLAM cells (unpublished observation). These results support the existence of sequence changes that offset the reduced fitness otherwise imparted by N-522D. We cannot at this time say which sequence(s) might be responsible. Sequence differences are present throughout the genome even for closely related members of the Edmonston MV vaccine lineage (17, 18). However, this observation provides an important context from which to view sequence differences in the genome of vaccine versus wild-type isolates.