ppUL83, ppUL57, and ppUL32) was evaluated at an moi of 0.5, all failed to accumulate in HF cell lysates examined up to 192 hpi (data not shown), a pattern that was consistent with inefficient growth and slow spread of this virus. Even at high moi, accumulation of γ proteins, ppUL32, ppUL83, and ppUL44 were delayed in RC303ΔAcc-infected HF cells compared with a Towne control (Fig. 4C). Thus, RC303ΔAcc exhibited a very strong moi dependence for accumulation of IE2579aa, a regulatory protein that is likely to be essential for productive replication. This very early block to expression of IE2579aa is consistent with our prediction that autoregulation of the ie1/ie2 promoter-enhancer via the cellular transcription factor NF-κB is an important IE1491aa function (16, 17). Other activities of IE1491aa may also be disrupted in RC303ΔAcc, but not manifest as strong an impact on viral replication as this autoregulatory function.
RC303ΔAcc is a severely replication defective CMV mutant whose phenotype reveals a role for IE1491aa in productive replication as was predicted from transient assays (16, 17). In culture, IE1491aa is required for successful replication at mois of <0.01, but this requirement may be overcome when sufficient numbers of viral particles are used to initiate infection. The moi-dependent expression of IE2579aa by RC303ΔAcc provides further evidence of the importance of IE1491aa in regulating levels of ie2 expression and replication under conditions where virion transactivators are likely to be limiting. The behavior of ie1-deficient CMV parallels the behavior of herpes simplex virus ICP0 (Vmw110) mutants (36) and predicts a very important role for IE1491aa in the naturally infected human host (37). It is formally possible that IE1491aa fulfills a function analogous to ICP0 (38). These predictions can be tested by inoculation of human tissue implants in SCID-hu mice (39) or by evaluating latency and reactivation properties in granulocyte-macrophage progenitors (40).
Many studies (26, 27, 32, 41) consistently suggested that CMV, like other herpesviruses, carries virion tegument functions responsible for transactivation of α genes. IE1491aa is not a virion protein (unpublished data) and cannot account for the ability of inactivated viral particles to transactivate the ie1/ie2 promoter-enhancer (32). The moi-dependent growth properties of IE1491aa-deficient virus support a role for two tegument proteins ppUL82 (pp71) (27) and ppUL69 (26), in transactivation of the ie1/ie2 promoter-enhancer, although experiments testing the ability of these gene products or UV-inactivated virions will be needed to address this issue more directly. It appears that CMV may encode two different classes of proteins capable of transactivating α gene expression: (i) one packaged into the virion and (ii) one, IE1491aa, expressed immediately upon entering cells. In the absence of IE1491aa, virion transactivators alone may not be able to initiate replication at low mois. Whether IE1491aa can compensate for virion transactivators remains to be determined when mutants disrupting UL82 or UL69 become available. In transient assays, a putative minor ie2 protein, IE2425aa (42), may transactivate expression of the ie1/ie2 promoter-enhancer, but this would not account for the properties of RC303ΔAcc described here. A range of factors, particularly those found in serum (43), that increase levels of cellular transcription factors also stimulate expression from the ie1/ie2 promoter-enhancer during cell growth or differentiation. All of our experiments were carried out at high serum concentrations in actively growing HF cells that may have reduced the requirement for IE1491aa. Studies using resting, serum-starved HF cells are under way. Our comparison of mutant viral growth on complementing and noncomplementing cells (Fig. 3) showed that complementing cells provided a slight advantage to replication even at high mois and suggested that IE1491aa may play additional roles subsequent to α gene activation. Finally, both IE1491aa and IE2579aa are able to block adenovirus E1a-induced apoptosis (25). We did not observe any dramatic alteration in the phenotype of cells infected with RC303ΔAcc that would be consistent with any increased tendency toward apoptosis. An appreciation of how a balance is struck between all of these factors will require considerable additional work in different cell types as well as in different growth and differentiation states.
While the highly passaged, candidate vaccine strain, Towne, is attenuated and fails to reactivate in the immunosuppressed transplant recipient (44), strain Toledo, a virus that has undergone a limited number of passages in cell culture, maintains a level of virulence that can be observed after inoculation of susceptible individuals (45). These virulence differences may be attributed to the loss of a large genomic segment in highly passaged strains (2). The strategy reported here would allow the derivation of avirulent, replication defective mutant virus from any strain of CMV. The mutant virus we have described places 13 kbp of unique sequence from strain Toledo into a Towne background. Although replication defective, this recombinant contains a large array of viral genes that are not present in Towne. As a vaccine, this would be expected to induce a broader ranging immune response but be completely safe even in immunocompromised individuals.
Accumulating evidence suggests that hematopoietic progenitors in bone marrow are important sites of viral latency (40, 46). Cultured granulocyte-macrophage progenitors can support latent infection with either the Towne or Toledo strains and infection is characterized by restricted, latency-specific gene expression (34, 47). Productive viral replication can be induced when these cells are cocultivated with permissive fibroblasts for an extended period of time (40), suggesting that cellular growth or differentiation state may play an important role in the balance between latency and reactivation. By removing ie1 exon 4, RC303ΔAcc disrupts the antisense CMV-latency specific transcript expressed in granulocyte-macrophage progenitors. We are currently evaluating the ability of the ie1 mutant to latently infect this cell type and to reactivate following cocultivation with ihfie1.3 cells. Recombinant CMV lacking IE1491aa may be an appropriate avirulent vector to introduce genes into hematopoietic progenitors without a risk of reactivation or dissemination.
We thank Kirsten Lofgren and Tai-An Cha for providing sequence information on the cosmid ends, Danushka Formankova for immunoblot analysis, and Maria Kirichenko for performing plaque assays. We particularly appreciate the contribution of Jiake Xu who purified the glutathione S-transferase fusion protein from pON2307 and raised ie1 exon-4 specific rabbit antiserum. We acknowledge Mark Penfold, Mark Prichard, Barry Slobedman, Dirk Dittmer, Jiake Xu, and Cynthia Bolovan for helpful comments on the manuscript. Some of the experiments reported here were carried out by E.S.M. at Aviron, a company to which he serves as a scientific advisor. This work was supported by U.S. Public Health Service Grant R01 AI33852.
1. Chee, M.S., Bankier, A.T., Beck, S., Bohni, R., Brown, C.M., Cerny, R., Horsnell, T., Hutchison, C.A.I., Kouzarides, T., Martignetti, J.A., Preddie, E., Satchwell, S.C., Tomlinson, P., Weston, K.M. & Barrell, B.G. (1990) Curr. Top. Microbiol. Immunol. 154, 125–170.
2. Cha, T.A., Tom, E., Kemble, G.W., Duke, G.M., Mocarski, E.S. & Spaete, R.R. (1996) J. Virol. 70, 78–83.
3. Alford, C.A. & Britt, W.J. (1995) in Fields Virology, eds. Fields, B.N., Knipe, D.M. & Howley, P.M. (Lippincott-Raven, New York), pp. 2493–2534.
4. Stenberg, R.M., Witte, P.R. & Stinski, M.F. (1985) J. Virol. 56, 665–675.
5. Stenberg, R.M., Thomsen, D.R. & Stinski, M.F. (1984) J. Virol. 49, 190–199.
6. Stenberg, R.M., Depto, A.S., Fortney, J. & Nelson, J.A. (1989) J. Virol. 63, 2699–2708.