The Small Envelope Protein E Is Not Essential for Murine Coronavirus Replication

The importance of the small envelope (E) protein in the assemblyof coronaviruses has been demonstrated in several studies. Whileits precise function is not clearly defined, E is a pivotalplayer in the morphogenesis of the virion envelope. Expressionof the E protein alone results in its incorporation into vesiclesthat are released from cells, and the coexpression of the Eprotein with the membrane protein M leads to the assembly ofcoronavirus-like particles. We have previously generated E genemutants of mouse hepatitis virus (MHV) that had marked defectsin viral growth and produced virions that were aberrantly assembledin comparison to wild-type virions. We have now been able toobtain a viable MHV mutant in which the entire E gene, as wellas the nonessential upstream genes 4 and 5a, has been deleted.This mutant ( ${Delta}$ E) was obtained by a targeted RNA recombinationmethod that makes use of a powerful host range-based selectionsystem. The ${Delta}$ E mutant produces tiny plaques with an unusual morphologycompared to plaques formed by wild-type MHV. Despite its lowgrowth rate and low infectious titer, the ${Delta}$ E mutant is geneticallystable, showing no detectable phenotypic changes after severalpassages. The properties of this mutant provide further supportfor the importance of E protein in MHV replication, but surprisingly,they also show that E protein is not essential.

INTRODUCTION

Top

Introduction

The coronavirus virion consists of a nucleocapsid core surroundedby an envelope containing three integral membrane proteins thatare common to all members of the genus (19). The nucleocapsidis a helical ribonucleoprotein made up of the positive-strandRNA genome with monomers of a single nucleocapsid (N) proteinbound along its length (20). It has been suggested that thisstructure is further compacted into a shell with spherical,possibly icosahedral, symmetry (31, 32). The viral envelopethat packages the nucleocapsid is derived from the membraneof the intracellular budding site, which is the endoplasmicreticulum-Golgi intermediate compartment or cis-Golgi network(16, 34, 40). Embedded in the envelope membrane are two majorstructural proteins. The first of these is the spike (S) glycoprotein,which forms morphologically characteristic projections on thevirion surface that mediate both attachment to species-specifichost cell receptors and fusion with the plasma membrane. Thesecond is the membrane (M) glycoprotein, the most abundant virioncomponent, which is a triple-spanning integral membrane proteinwith a short ectodomain and a large carboxy-terminal endodomain(33). The carboxy terminus of M has been shown by both biochemical(10, 26, 27, 38) and genetic (18) criteria to interact withthe nucleocapsid, specifically, with the carboxy terminus ofthe N protein (18).

The third critical membrane-bound constituent of the virionis the small envelope (E) protein. E was only belatedly recognizedas a viral structural protein, owing to its size (ca. 10 kDa)and its very low abundance relative to the M, N, and S proteins(13, 21, 45). E proteins are well conserved within each of thethree groups of coronaviruses, but they show very limited homologyacross the different groups. All E proteins have the same generalstructure: a short (7- to 9-amino-acid) hydrophilic amino terminuspreceding a large (21- to 29-amino-acid) hydrophobic region,followed by a large hydrophilic carboxy terminus making up one-halfto two-thirds of the mass of the molecule (21). Earlier worksuggested an external topology for the carboxy-terminal domainof the E protein of transmissible gastroenteritis virus (TGEV)(13). However, more recent studies of the E proteins of mousehepatitis virus (MHV) and infectious bronchitis virus (IBV)have demonstrated that E is an integral membrane protein displayingits carboxy terminus on the cytoplasmic face of the endoplasmicreticulum or Golgi, which corresponds to the interior of assembledvirions (3, 30, 42). Interestingly, it is the carboxy-terminaltail of the IBV E protein, in the absence of the membrane-bounddomain, which specifies Golgi targeting for that molecule (4).The disposition of the amino terminus of the E molecule is lesswell established. A lumenal (or virion-exterior) orientationof this domain has been inferred for the IBV E protein (3),while for the MHV E protein, the amino terminus has been proposedto be buried within the membrane near the cytoplasmic face (23).

The case for a major role for the E protein in coronavirus assemblywas solidified by fundamental observations by Vennema and coworkers(42) and Bos and coworkers (2) that cellular expression of justthe M and E proteins of MHV leads to formation and release ofvirus-like particles (VLPs) that are morphologically identicalto spikeless virions. These results were subsequently generalizedto TGEV (1), bovine coronavirus (BCoV) (1), and IBV (3). Thus,the coronavirus envelope can form itself in the absence of eitherS protein or nucleocapsid, although work with constructed MHVM protein mutants indicates that the nucleocapsid stabilizesvirion structure (7, 18). Further support for the importanceof the E protein came from the construction of clustered charged-to-alaninemutants of the MHV E protein (12). Two of these mutants werepartially temperature sensitive and markedly thermolabile, andelectron microscopic examination of the more impaired mutantrevealed virions with dramatic deviations from the more uniformlyspherical virions of the wild type.

Collectively, these results would imply that the E protein isessential for coronavirus replication, and indeed, this hasrecently been reported for TGEV (5, 29) and for equine arterivirus(EAV), a related member of the order Nidovirales (37). Surprisingly,however, when we rigorously tested this assumption for MHV,we were able to isolate mutants in which the E gene was totallydeleted. These viruses, although severely debilitated, werenevertheless viable and infectious. We therefore conclude thatthe E protein is critical, but not essential, for MHV replication.

MATERIALS AND METHODS

Top

Materials and Methods

Cells and viruses. Wild-type MHV-A59 and MHV recombinants were propagated in mouse17 clone 1 (17Cl1) or L2 cells; plaque assays and plaque purificationwere performed with mouse L2 cells. The interspecies chimericvirus fMHV (17) was grown in feline AK-D fetal lung cells.

Plasmid constructs. For the purpose of deleting the E gene in this study, plasmidpLK70 was constructed from the transcription vector pMH54, whichhas been described previously (17). The deletion mutation, whichencompasses genes 4 and 5a as well as the E gene, was derivedfrom a PCR fragment that was generated using primers LK87 andLK10 (Table 1) with pMH54 as the template. After restrictiondigestion with Sse8783I and BssHII, the PCR product was transferredinto pMH54, replacing the Sse8783I-BssHII fragment of that plasmid.The resulting construct, pLK70, was 1,023 bp shorter than thepMH54 parent but was otherwise identical to pMH54.

View this table:
[in this window]
[in a new window]
TABLE 1. Primers used in this study

Targeted RNA recombination. To incorporate the ${Delta}$ E mutation into the MHV genome, targetedRNA recombination was performed using fMHV as the recipientvirus, as described previously (18, 25). In brief, feline AK-Dcell monolayers were infected with fMHV at a multiplicity ofroughly 1 PFU per cell for 4 h at 37°C; the infected cellswere then suspended by treatment with a low concentration oftrypsin. Capped donor RNA transcripts were synthesized withthe T7 mMessage mMachine transcription kit (Ambion) using PacI-truncatedpLK70 or pMH54 template following the manufacturer's instructions.Synthetic donor RNA (5 to 10 µg) was transfected into1.2 x 10⁷ fMHV-infected AK-D cells by two pulses with a GenePulser electroporation apparatus (Bio-Rad) at settings of 960µF and 0.3 kV. The infected and transfected AK-D cellswere then overlaid onto murine 17Cl1 cell monolayers. Progenyvirus released into the supernatant medium was harvested whensyncytia or cytopathic effect was apparent in the murine cells,typically 48 h postinfection at 37°C. Both large- and small-plaquerecombinant candidates were purified through two rounds of plaquetitration on murine L2 monolayers at 37°C.

Genomic analysis of candidate recombinants. Each purified recombinant candidate was used to infect a 10-cm²monolayer of L2 cells at 37°C. Total cellular RNA was extractedeither 24 (for large-plaque candidates) or 72 (for small-plaquecandidates) h postinfection using Ultraspec reagent (Biotecx).Reverse transcription (RT) of RNA was carried out using a randomhexanucleotide primer (Boehringer Mannheim) with avian myeloblastosisvirus reverse transcriptase (Life Sciences) under standard conditions(35). To confirm the presence, absence, or alteration of differentgenomic regions, PCR amplifications of cDNA were performed usingvarious oligonucleotide primer pairs. PCR was carried out withAmpliTaq polymerase (Perkin-Elmer) under standard conditionsof 1 min at 94°C, 1 min at 49°C, and 2 min at 72°C,repeated for 30 cycles. The resulting products were analyzeddirectly by agarose gel electrophoresis or were purified withQuantum-prep columns (Bio-Rad) prior to automated sequencing.

Radiolabeling of viral RNA. Metabolic labeling of virus-specific RNA was carried out underconditions similar to those reported previously (25). Monolayersof 17Cl1 cells in 20-cm² dishes were infected with the ${Delta}$ E mutantAlb291 or the wild-type MHV recombinant Alb240 at a multiplicityof 0.125 PFU per cell, and incubation was continued at 33°C.When infections had progressed to the extent of $~$ 50% syncytiumdevelopment (at $~$ 12 h), the cells were starved for 2 h at 37°Cin Eagle's minimal essential medium containing 5% dialyzed fetalbovine serum and 1/10 of the normal phosphate concentration.Labeling was then carried out for 2 h at 37°C by incubationof the starved cells in phosphate-free medium containing 100µCi of [³³P]orthophosphate (ICN) per ml, 5% dialyzed fetalbovine serum, and 20 µg of actinomycin D (Sigma) per ml.Total cytoplasmic RNA was purified by a gentle NP-40 lysis procedure.RNA samples containing equal amounts of radioactivity were denaturedwith formaldehyde and formamide, analyzed by electrophoresisthrough 1% agarose containing formaldehyde, and visualized byfluorography.

Northern blot hybridization. For the analysis of viral RNA by Northern blotting, total cytoplasmicRNA was harvested and purified from 17Cl1 cells infected withMHV recombinants or wild-type MHV. After fractionation in a1% agarose gel containing formaldehyde, the RNA was blottedand then UV cross-linked onto a Nytran Supercharge membrane(Schleicher and Schuell). Alternatively, RNA denatured withformaldehyde and formamide was directly dot blotted onto a membraneby filtration through a vacuum manifold followed by UV cross-linking.A DNA probe corresponding to the 5' 283 nucleotides of the MHV3' untranslated region (UTR), generated with PCR primers CM95and PM112 (Table 1), was used to detect all virus-specific RNAspecies. As a probe for E gene-specific sequences, a DNA fragmentof 233 bp, covering almost the entire E open reading frame (ORF),was generated with the PCR primers PM160 and FF42 (Table 1).The DNA fragments were directly labeled during synthesis with[ ${alpha}$ -³²P]dCTP in PCR using the respective primers (35). Hybridizationwas carried out at 65°C (16 to 20 h) in a buffer containing0.5 M sodium phosphate, pH 7.2, 7% sodium dodecyl sulfate, and1 mM EDTA. After hybridization, the blots were washed at 65°Cthree times in 0.5x SSC (1x SSC is 0.15 M NaCl plus 0.015 Msodium citrate) containing 0.1% sodium dodecyl sulfate (35),and the radioactivity was visualized by fluorography.

RESULTS

Top

Results

Construction of a viable MHV recombinant with the E gene deleted. Targeted RNA recombination has been successfully employed asa method to carry out reverse genetics on the 3'-most 10 kbof the 31-kb MHV genome (15, 17, 24, 25), a region that encompassesall of the structural genes of this coronavirus. This techniqueexploits the recombinational propensity of the viral polymeraseto obtain crossover events between a recipient virus and a transfectedsynthetic donor RNA. Recombinant viruses are then recoveredbased on a counterselectable property of the recipient virus.A particularly strong selection is afforded by the host cellspecies restriction of a chimeric mutant, fMHV, that containsthe ectodomain of the S protein of feline infectious peritonitisvirus (FIPV) in place of its MHV counterpart (17). As a resultof this substitution, fMHV grows in feline cells and has lostthe ability to grow in murine cells. We have recently shownthat the use of fMHV as the recipient virus in targeted RNArecombination allows the recovery of extremely defective MHVmutants based on selection for the reacquisition of the MHVS protein ectodomain and the concomitant ability to grow inmurine cells (18).

To pursue the fundamental question of whether the E proteinis essential for MHV replication, we deleted the E gene froma transcription vector from which donor RNA could be synthesized.Plasmid pLK70 (Fig. 1A) was constructed from the previouslydescribed wild-type pMH54 (17) via PCR-facilitated removal ofthe entirety of genes 4, 5a, and E. A number of prior reportshave established that genes 4 and 5a are nonessential in MHV(8, 11, 28, 43, 44). The 1,023-bp deletion in pLK70 ran fromthe transcription-regulating sequence for gene 4 (TRS4) throughthe TRS for the M gene (TRS6); the canonical TRS4 core sequence(AATCTAAAC) was retained instead of the slightly variant (underlined)TRS6 core sequence (AATCCAAAC) (Fig. 1A).

View larger version (33K):
[in this window]
[in a new window]
FIG.1. Selection of the ${Delta}$ E mutant. (A) Construction of a transcription vector for synthesis of donor RNA. Plasmid pLK70 was derived from pMH54 (17), as detailed in Materials and Methods. In each plasmid schematic, the arrow indicates the T7 promoter, and the solid circle represents the linker between cDNA segments corresponding to the 5' and 3' ends of the MHV genome. The restriction sites shown are those relevant to plasmid construction (Sse8387I, BssHII, and NheI), in vitro transcription (PacI), or mutant analysis (SmlI). The DNA sequences shown are as follows: 1 and 2, the boundaries of the region that was deleted from the wild type (wt); and 3, the newly created junction in the deletion mutant. The underlined nucleotides are the three base changes made in pMH54 that generated an Sse8387I site (17). Also indicated are the TRSs governing synthesis of sgRNA4 and the M mRNA. Shown beneath the DNA sequences are translations of the S, M, and E ORFs (asterisks indicate stop codons). (B) Scheme for generation of the ${Delta}$ E mutant by targeted RNA recombination between the interspecies chimera fMHV (17) and donor RNA transcribed from plasmid pLK70. fMHV contains the ectodomain-encoding region of the FIPV S gene (hatched rectangle) and is able to grow in feline cells but not in murine cells. A single crossover (solid line), within the HE gene, should generate a recombinant that has simultaneously reacquired the MHV S ectodomain and the ability to grow in murine cells and has also incorporated the deletion of the E gene. A potential second crossover (broken line), in the distal portion of the S gene, would produce a recombinant retaining the E gene. At the bottom are shown the mixed progeny of two independent targeted recombination experiments, forming tiny and large plaques on mouse L2 cells. The monolayers were stained with neutral red 75 h postinfection and were photographed 19 h later.

Synthetic donor RNAs transcribed from pLK70, or from the control,pMH54, were independently transfected into feline AK-D cellsthat had been infected with fMHV (17) (Fig. 1B), and recombinantswere selected in murine cells. When the titers of these weredetermined on mouse L2 cells, the pLK70-derived recombinantsgenerated a mixed population of very tiny plaques and much larger,wild-type-sized plaques (Fig. 1B). By contrast, only wild-type-sizedplaques were observed with RNA from the parental plasmid, pMH54(not shown). Based on our previous experience with this selectionscheme (18), we expected that the small plaques represented ${Delta}$ E recombinants, produced by a single crossover upstream of theS gene, while the large plaques were expected to be wild-typerecombinants resulting from an additional downstream crossoverevent. (These expectations were subsequently verified.) It islikely that the small plaques outnumbered the large ones becausethe elimination of genes 4 and 5a, in addition to the E gene,in pLK70 provided a minimal (233-nucleotide) region of homologydownstream of the S ectodomain in which a productive secondcrossover could occur (Fig. 1B).

Plaque phenotype of the ${Delta}$ E mutant. Multiple plaques resulting from independent infection-transfectionexperiments with pLK70-derived donor RNA were purified by plaquetitration. The respective phenotypes of the small and largeplaques did not change throughout two rounds of plaque purificationor following additional passages in murine cells. Infectionsbegun from the small-plaque ${Delta}$ E mutants produced the hallmarksyncytia and subsequent cytopathic effect of MHV in cell monolayers,but they did so much more slowly than the wild type, and ${Delta}$ E stocksreached optimal titers of only 1 x 10⁵ to 2 x 10⁵ PFU/ml, atleast 3 orders of magnitude lower than those typically observedfor the wild type. The large-plaque recombinants, on the otherhand, were indistinguishable from wild-type MHV in growth rate,plaque size, and virus titers.

In addition to their drastically reduced size, plaques of the ${Delta}$ E mutants also exhibited an intriguing morphology. Figure 2shows two dilutions of two independent ${Delta}$ E mutants to demonstratethis feature. In contrast to the uniform, circularly symmetricplaques typically formed by wild-type MHV, plaques formed by ${Delta}$ E mutants showed much more pronounced size variations, withsome plaques of pinpoint size and others several times larger.The tiny plaques also had irregular shapes with jagged edges,unlike the usually smooth edges found in wild-type plaques.Notably, these peculiarities were not seen for wild-type plaquesat early stages postinfection, i.e., at a comparably small plaquesize. This suggests that the absence of E protein somehow affectsthe rate or the directionality of cell-to-cell spread of thevirus during infection. Elucidating the basis for this unexpectedobservation will require much further work.

View larger version (106K):
[in this window]
[in a new window]
FIG. 2. Titration of passage 2 stocks obtained from purified tiny plaques from independent infection-transfection experiments. Two dilutions each of the ${Delta}$ E mutants Alb289 and Alb291 are shown (top, 10^-2; bottom, 10^-3) to emphasize the atypical morphology of the tiny plaques. Also shown, for comparison, are plaques of Alb240, a wild-type (wt) recombinant derived from pMH54 and fMHV (17), and mock-infected cells (mock). The titers of the viruses were determined on mouse L2 cell monolayers, which were stained with neutral red 75 h postinfection and photographed 19 h later.

Confirmation of the ${Delta}$ E genotype. For genotypic analysis, we initially chose eight plaque-purifiedsmall-plaque recombinants (designated Alb289 through Alb296),which were collectively derived from four independent infection-transfectionexperiments. As controls, we used four large-plaque recombinants,which were derived from four independent infection-transfectionexperiments, and Alb240, a well-characterized wild-type recombinantthat had previously been obtained from targeted recombinationbetween fMHV and pMH54 RNAs. Total cellular RNA was isolatedfrom monolayers of 17Cl1 cells infected with each of the recombinantsor from mock-infected cells as an additional control. To examinethe site of the ${Delta}$ E deletion in each recombinant, the RNA wasreverse transcribed with a random primer and then PCR was carriedout with primers CK4 and PM147, which are specific for the regionsof the MHV S and M genes, respectively, flanking the deletedgenomic segment (Fig. 3A and Table 1). Each of the four independentlygenerated large-plaque recombinants yielded a single productthat was identical in size to that of the wild-type control,Alb240, and consistent with the predicted size of 1,639 bp (Fig.3A). Direct sequencing of these PCR products showed that theycorresponded exactly to the sequence of the wild type. Therefore,in accord with our earlier experience in isolating the extremelydefective M ${Delta}$ 2 mutant (18), the large-plaque recombinants appearedto be regenerated wild-type viruses resulting from two crossoverevents during targeted recombination (Fig. 1A), and they werenot analyzed further.

View larger version (51K):
[in this window]
[in a new window]
FIG. 3. Analysis of large-plaque and small-plaque progeny from targeted recombination. (A) RNA was isolated from cells infected with plaque-purified recombinants, reverse transcribed with random primer p(N)₆, and amplified with primers CK4 and PM147. The PCR products were analyzed by agarose gel electrophoresis. Large plaques (#1 to 4) were purified from independent infection-transfection experiments. The small-plaque recombinants Alb289 through Alb296 represent four independent sets of mutants in which consecutively numbered odd-even pairs are siblings. Lanes M, DNA fragment size markers. wt, wild type; mock, mock-infected cells. (B) Sequence of the S gene-M gene junction in the ${Delta}$ E mutant Alb291. The junction sequences of Alb289, Alb290, and Alb292 to Alb296 were identical.

By contrast, each of the small-plaque recombinants produceda single PCR product that was much smaller than that of thewild type and was consistent with the size of 616 bp predictedfor the ${Delta}$ E mutant (Fig. 3A). This strongly suggested that thesmall-plaque recombinants lacked the 1,023-bp region encompassinggenes 4, 5a, and E. Direct sequencing of these purified PCRfragments confirmed that all eight of them had the expectedsequence for the entire amplified region. In particular, thejunction between the S and M genes was exactly as had been constructedin the transcription vector pLK70. A representative sequence,that of the junction in Alb291, is shown in Fig. 3B. Becausewe were concerned about the possibility that MHV, in order tosurvive in the absence of the E protein, might require adaptivemutations in the M protein, we sequenced another PCR product,obtained with primers CK4 and PM145 (Table 1), that containedthe entire M gene. This revealed that there were no mutationsin the M genes of the small-plaque recombinants, except thatAlb289 and Alb293 each had a point mutation in the last codonof the M ORF, resulting in the amino acid change T228I. Sincethe other six small-plaque recombinants had the entire wild-typeM gene sequence, this mutation cannot be required for ${Delta}$ E mutantviability. We have presented evidence elsewhere that the T228Imutation likely results from recombination between the MHV genome(g) RNA and subgenomic (sg) RNA7, the mRNA for N protein (18).

Analyses to rule out a heterologous source or location of the E gene. It has previously been established by a number of criteria thatthe interspecies chimeric virus fMHV is incapable of growthin murine cells that lack feline aminopeptidase N, the receptorfor FIPV (17). All of the ${Delta}$ E mutants that we isolated from fMHV-basedtargeted RNA recombination were purified through two roundsof plaque titration on murine cells to ensure their homogeneityand were then propagated in murine cells, in some cases foras many as seven passages. Nevertheless, it remained conceivablethat, if MHV were unable to survive in the absence of E protein,we might then have selected for an fMHV variant that was pseudotypedwith MHV S protein that was somehow provided by a copackagedRNA. We tested for this possibility in two ways. First, exhaustiveattempts were made to cultivate fMHV from passage 3 or 4 ${Delta}$ E mutantstocks by serial passaging in feline AK-D cells. We were neverable to detect signs of infection caused by ${Delta}$ E virus, in sharpcontrast to fMHV controls, which produced extensive syncytiaand cytopathic effect in the same cell line.

As a further test for the possible presence of fMHV in ${Delta}$ E virusstocks, total cellular RNA was extracted from murine 17Cl1 monolayersthat had been infected with passage 4 stocks of three ${Delta}$ E mutants,Alb289, Alb290, and Alb291, or as controls, with wild-type MHVor the wild-type recombinant Alb240. Additional controls wereprovided by RNA isolated from mock-infected 17Cl1 cells andfrom fMHV-infected AK-D cells. Following random-primed RT, PCRwas performed using several specific primer pairs (Table 1).With primer pair LK59-PM147, products of two alternative sizeswere detected in the viral samples (Fig. 4). The upstream primer,LK59, resides in the region of the S gene encoding the S proteinendodomain, which is common to both MHV and fMHV. For fMHV andthe two MHV wild-type controls, we observed a PCR product inagreement with the expected size of 1,438 bp. For all three ${Delta}$ E mutants, a much smaller product was detected, at a mobilityconsistent with the predicted size of 415 bp (Fig. 4), and therewas no trace of the larger product that had been found withthe fMHV or wild-type control. This further validated the previousRT-PCR result (Fig. 3) that had showed the deletion of genes4, 5a, and E in the ${Delta}$ E mutants. It also confirmed the integrityof the RNA and RT product for all viral samples, particularlythose of the ${Delta}$ E mutants. PCRs were then carried out with threeadditional primer pairs. Two of these, LK50-LK67 and LK57-LK60,were specific for sequences mapping entirely within the regionencoding the FIPV S ectodomain. The third set paired an FIPVS-specific primer, LK58, with a downstream primer, LK17, locatedin the M gene. All three primer pairs yielded products onlywith fMHV, and these were consistent with the expected sizesof 578, 545, and 1,574 bp, respectively (Fig. 4). No signalwas obtained with the ${Delta}$ E mutants or with the wild-type or mock-infectedcontrols. Collectively, these results ruled out the presenceof fMHV in the ${Delta}$ E mutant virus stocks.

View larger version (47K):
[in this window]
[in a new window]
FIG. 4. RT-PCR analysis to rule out the presence of fMHV in purified ${Delta}$ E recombinants. The random-primed RT product obtained with RNA isolated from infected cells was amplified with primer pairs either unique to fMHV or common to fMHV, wild-type (wt) MHV, and the ${Delta}$ E mutant. In the genomic schematics at the top, the hatched region indicates the ectodomain-encoding portion of the FIPV S gene; primer positions are not drawn strictly to scale. The PCR products were analyzed by agarose gel electrophoresis. Lanes M, DNA fragment size markers; mock, mock-infected cells.

We next tested for any other potential extraneous source ofthe E gene or for the possible presence of the E gene at anectopic location in the ${Delta}$ E mutant genome. To accomplish this,we performed PCR with three different pairs of primers targetedto various regions of genes 5a and E (Fig. 5). Primer pairsCM82-PM152, CM80-PM152, and PM160-FF42 each gave strong, specificproducts of 155, 371, and 233 bp, respectively, with fMHV andthe wild-type MHV controls. However, none of the ${Delta}$ E mutants produceda detectable signal with these primers. Although a weak nonspecificband of roughly 200 bp can be seen for one of the ${Delta}$ E mutants,Alb289, with the primer pair CM80-PM152, the same band was alsopresent in the mock-infected control. Several other sets ofprimers mapping in genes 5a and E, in different combinations,were also used in additional RT-PCR and similarly produced productswith the fMHV and wild-type MHV controls but not with any ofthe ${Delta}$ E mutants (data not shown). It should be noted that allof these PCRs were carried out with the same ${Delta}$ E mutant cDNA samplesthat yielded strong positive signals with the primer pair LK59-PM147(Fig. 4) and also with an N gene-specific primer pair (datanot shown).

View larger version (54K):
[in this window]
[in a new window]
FIG. 5. RT-PCR analysis to rule out the presence of the E gene elsewhere in the genome. The random-primed RT product obtained with RNA isolated from infected cells was amplified with primer pairs internal to the wild-type (wt) E ORF or including the E ORF and the upstream gene 5a. The PCR products were analyzed by agarose gel electrophoresis. Lanes M, DNA fragment size markers; mock, mock-infected cells.

Direct analysis of intracellular RNA. To obtain a more long-range picture of the MHV genome than hadbeen achieved by RT-PCR, we metabolically labeled virus-specificRNA with [³³P]orthophosphate in cells infected with the ${Delta}$ E mutantAlb291 or with the wild-type recombinant Alb240 (Fig. 6A). Duringinfection, MHV-A59 produces a 3'-nested set of six sgRNAs, eachof which consists of a 70-nucleotide leader sequence joinedat a TRS to all genomic sequence downstream of that TRS (36,41). Deletion of a given region of the genome will thus affectnot only gRNA but also all sgRNAs encompassing that region.The pattern of transcripts for the ${Delta}$ E mutant, therefore, exactlyconformed to that expected to result from the deletion of genes4, 5a, and E and their two associated TRSs. The ${Delta}$ E mutant entirelylacked sgRNAs 4 and 5, and sgRNAs 2 and 3 had mobilities consistentwith having been shortened by 1 kb with respect to their wild-typecounterparts. We would not have expected to be able to detectthe size difference between the ${Delta}$ E mutant gRNA (30 kb) and thewild-type gRNA (31 kb). The gRNA of the ${Delta}$ E mutant appeared tobe much less abundant than that of the wild type; however, thislikely reflected variation in recovery of gRNA in this particularsample, since other independent labeling experiments did notshow this contrast. As expected, there were no detectable differencesin the sizes of mutant and wild-type sgRNAs 6 and 7, which areboth governed by TRSs downstream of the deleted region. Notably,in the ${Delta}$ E mutant the amount of sgRNA6, the mRNA for M protein,was not significantly different from that of its wild-type counterpart.This appears to preclude the possibility that the mutant compensatesfor the absence of E protein by overexpressing the M protein.

View larger version (67K):
[in this window]
[in a new window]
FIG. 6. Analysis of RNA species synthesized by the ${Delta}$ E mutant. (A) Virus-specific RNA was labeled with [³³P]orthophosphate in the presence of actinomycin D in cells infected with wild-type (wt) recombinant Alb240 or ${Delta}$ E mutant Alb291 or in mock-infected (mock) control cells. Purified RNA was analyzed in a formaldehyde-agarose gel, as detailed in Materials and Methods. (B) Northern blot analysis of unlabeled RNA isolated from cells infected with wild-type recombinant Alb240 or ${Delta}$ E mutants Alb291 and Alb293 or from mock-infected control cells. Purified RNA was separated in a formaldehyde-agarose gel and transferred to a nylon filter. Alternatively, RNA was directly dot blotted onto nylon filters; each set of dots corresponds to serial twofold dilutions, starting with 5 µg of total cellular RNA. The RNA was hybridized with a ³²P-labeled DNA probe specific either for the E gene or for the 3' UTR of the MHV genome; bound probe was visualized by fluorography. The Alb240 lane marked with an asterisk is a short exposure of the adjacent, overexposed lane.

To corroborate the conclusion of the RT-PCR analysis in Fig.5, we used Northern blotting to search for the presence of Egene-related sequences in RNA from ${Delta}$ E mutant-infected cells.RNA isolated from cells infected with two ${Delta}$ E mutants (Alb291and Alb293) and the wild-type recombinant (Alb240) was electrophoreticallyseparated, blotted onto nylon, and hybridized with a labeledprobe encompassing almost the entire E ORF (Fig. 6B). As expected,for wild-type RNA the probe detected sgRNA5, the mRNA for Eprotein, and all species larger than sgRNA5. However, even afterprolonged exposure of the blot, no E gene-specific hybridizationsignal was obtained with RNA from either of the ${Delta}$ E mutants. Asa control, a duplicate blot was hybridized with a labeled probespecific for the 3' UTR. As expected, this probe hybridizedto all viral RNAs in both wild-type and ${Delta}$ E mutant samples. Toguard against the possibility that potential hybridizing materialwas lost in the Northern blots owing to inefficient transferof the largest of the MHV RNA species, we also carried out thesame hybridizations in dot blots and obtained the same results(Fig. 6B, bottom). Overexposed blots using the E gene-specificprobe failed to reveal any hybridizing material in the ${Delta}$ E mutantRNA samples, while the same samples showed an extent of hybridizationto the 3' UTR probe comparable to that exhibited by wild-typeRNA. These results strongly support our conclusion that in the ${Delta}$ E mutants, the E gene has been deleted from its normal locusand is neither present elsewhere in the genome nor providedby an exogenous source.

DISCUSSION

Top

Discussion

Coronaviruses, arteriviruses, and toroviruses have been groupedtogether in the order Nidovirales principally on the basis ofsimilarities in their RNA synthesis mechanisms, polymerase genes,and genome organizations (9). Despite considerable differencesamong their surface glycoproteins and nucleocapsids, all ofthese viruses have in common envelopes with a triple-spanningM protein as a major constituent. In addition, all coronavirusesand arteriviruses encode an E protein, and no naturally occurringstrains or variants have been found that lack E. By contrast,no homolog for E has yet been identified in the torovirus genome.

Three prior reports have used reverse genetics to explicitlyabrogate nidovirus E gene expression. For the arterivirus EAV,Snijder and coworkers (37) eliminated expression of the E geneby creating a point mutation in its start codon. The frugalgenomic organization of EAV prevents complete elimination ofthe E ORF because the latter overlaps extensively with the ORFfor an essential surface glycoprotein, G_S. In addition, theE start codon shares 2 nucleotides with the stop codon of theupstream polymerase gene, and it is very close to the TRS thatgoverns transcription of the mRNA from which both E and G_S aretranslated. The EAV E knockout mutant RNA was found to be unableto produce infectious progeny virus, whereas there was no unintendedeffect on expression of either polymerase or G_S in initiallytransfected cells. The authors did note, however, that in aquantitative transfection-infectious center assay, the E knockoutmutant had an apparent reversion frequency at least 10-foldgreater than those of other EAV start codon point mutants (37).It is conceivable that this higher apparent reversion rate wasgenerated by a low level of replication of the E knockout mutantthat occasionally led to breakaway plaque formation by an arisingrevertant. For the coronavirus TGEV, Curtis and coworkers (5)constructed a virus containing the gene for green fluorescentprotein in place of the nonessential gene 3a. This was accomplishedthrough manipulation of a set of six cloned cDNAs spanning theTGEV genome that were then ligated in vitro to form a templatefor the synthesis of infectious RNA. The authors next deletedthe nonessential gene 3b and the first 10 nucleotides of thedownstream E ORF, and they were not able to obtain infectiousmaterial upon passage of supernatant from cells transfectedwith this mutant RNA. Similarly, Ortego and coworkers (29) deletedthe entirety of the TGEV genes 3a, 3b, and E from a full-lengthinfectious cDNA and found that this resulted in complete lossof ability to recover infectious virus. Each of these TGEV Egene knockout mutants was demonstrated to function as a (noninfectious)RNA replicon that could be packaged in the presence of E proteinexogenously provided by an alphavirus vector (5, 29) or by acell line constitutively expressing E protein (29). In the lattercase, TGEV particles produced by such complementation appearedto follow the same assembly pathway and exhibited the same morphologyas the wild-type virus.

The discrepancy between our successful isolation of a viableE gene deletion mutant of MHV and the apparent lethality ofE gene knockouts in EAV and TGEV may indicate basic differencesin assembly dynamics among various nidoviruses. Alternatively,it may reflect the relative robustness of MHV growth and plaqueformation in tissue culture. With a more defective MHV mutant,the truncated M protein mutant M ${Delta}$ 2, we have previously notedthat in one cell line the kinetics of viral infection were sufficientlyslow that monolayers were always overtaken by the growth ofuninfected cells (18). It is possible that a similar phenomenonmay obscure detection of ${Delta}$ E mutants of some nidoviruses.

Our results show that the MHV ${Delta}$ E mutant produces viable, independentlyreplicating virus, despite lacking the entire E ORF. As yet,we have not been able to formally demonstrate the absence ofE protein in the ${Delta}$ E mutant by immunological criteria. Antibodiesraised against the E protein of either the Utrecht laboratorystrain of MHV-A59 (30) or MHV-JHM (45), in our hands, fail todetect E protein in cells infected with our laboratory strainof MHV-A59. It has been shown that this loss of recognition,in the case of anti-E (MHV-A59; Utrecht), is caused by a mere2-amino-acid-residue difference between the Utrecht and AlbanyMHV-A59 E proteins (8); in the case of anti-E (MHV-JHM), similarstrain-specific E protein amino acid differences may pertain.Nevertheless, by much more sensitive analyses using RT-PCR,RNA metabolic labeling, and Northern blotting, we have establishedthat the E gene has been deleted from its normal position inthe genome of the ${Delta}$ E mutant, that it is not present at some ectopicgenomic locus, and that it is not provided by some epigeneticmechanism in infected cells.

Multiple independent isolates of ${Delta}$ E gave rise to small plaqueshaving the same atypical morphology, the basis for which iscurrently unclear. Stocks of all ${Delta}$ E isolates had low titers,suggesting that virion assembly in the mutant occurs with ordersof magnitude lower efficiency than in the wild type. We alsofound that the ${Delta}$ E mutant was stable upon repeated passaging ofviral stocks, unlike extremely defective mutants of M protein(18) and N protein (unpublished results), for which wild-type-likerevertants overrun higher-passage stocks. This likely indicatesthat there are no single second-site point mutations that cancompensate for the loss of the entire E gene. It must be notedthat the deletion in the ${Delta}$ E mutant also removed the nonessentialgenes 4 and 5a. It has previously been shown that the deletionof genes 4 and 5a produced no detectable change in viral plaquesize or morphology compared to the wild type and had only aminimal effect on viral growth kinetics in tissue culture (8).Thus, we believe that the observed phenotype of the ${Delta}$ E mutantis due almost entirely to the loss of the E protein. Moreover,the tissue culture phenotype of the ${Delta}$ 45a mutant (8) may alsobe attributable to alteration of E protein expression, sincegene 5a is thought to function as an internal ribosome entrysite for translation of E (14, 39).

The dispensability of the E protein, at least for MHV, leadsus to conclude that E protein greatly enhances virion envelopeformation but is not essential for this process to proceed.Results from some previous studies suggested that the E proteinacts independently of other viral components. It has been foundthat the expression of the MHV or IBV E protein, in the absenceof other viral proteins, results in vesicles that are exportedfrom cells (3, 22). Expression of the MHV E protein alone hasalso been shown to induce the formation of clusters of convolutedmembranous structures highly similar to those seen in coronavirus-infectedcells (6, 30). This may mean that the principal role of E isto induce membrane curvature in the budding compartment andthat M-M monomer interactions drive the remainder of virionmorphogenesis. By contrast, other experimental results impliedthat E must specifically act in concert with M. In particular,it was found that, although both TGEV and BCoV VLPs are efficientlygenerated by coexpression of homologous constituents, VLPs cannotbe produced by expression of the M protein of TGEV with theE protein of BCoV. Conversely, VLPs cannot be produced by expressionof the M protein of BCoV with the E protein of TGEV (1). Thisseems to indicate that virion envelope maturation is promotedby a direct interaction between M and E instead of, or in additionto, the effect of E on intracellular membranes.

One of the most pressing questions raised by the existence ofthe ${Delta}$ E mutant is what do its virions look like? We are eagerto learn the answer to this, but we have been thwarted so farby the low titers of the ${Delta}$ E mutant. Similarly, we are curiouswhether the ${Delta}$ E mutant buds at the same intracellular site asthe wild type or whether it exploits some alternative pathway.Preliminary results from immunofluorescence experiments suggestthat there is no significant difference in the intracellularlocalization of M protein in cells infected with the ${Delta}$ E mutantand in those infected with the wild type (data not shown), buthigher-resolution analyses will be required to resolve thisquestion. We expect that further characterization of ${Delta}$ E viruseswill help to shed light on the functions of the E protein inwild-type MHV assembly.

ACKNOWLEDGMENTS

We are grateful to Kelley Hurst for expert technical assistance.We thank Matthew Shudt, Heather Berry, Jolene Wilson, and TimMoran of the Molecular Genetics Core Facility of the WadsworthCenter for oligonucleotide synthesis and for DNA sequencing,and we deeply regret the untimely loss of our friend and colleagueTim Moran.

This work was supported in part by Public Health Service grantAI 39544 from the National Institutes of Health.

FOOTNOTES

* Corresponding author. Mailing address: David Axelrod Institute, Wadsworth Center, NYSDOH, New Scotland Ave., P.O. Box 22002, Albany, NY 12201-2002. Phone: (518) 474-1283. Fax: (518) 473-1326. E-mail: masters{at}wadsworth.org.

REFERENCES

Top