The cascade of herpesvirus gene expression that results in viral replication begins with the activation of IE genes by the virion-associated protein VP16. VP16 on its own is inefficient at associating with complexes formed on IE promoters and depends upon the cellular factor HCF for its activity. In this respect VP16 mimics the host bZIP protein Luman, which also requires HCF for activating transcription. Our objective is to explore interactions between Luman and HCF and to determine if they play a role in the biology of herpesviruses. The results presented here suggest that Luman may be involved in both the establishment of latent infections and the reactivation of latent virus. By sequestering HCF in the cytoplasm of sensory neurons, Luman may prevent the initiation of the replicative cascade leading to latent infection. Subsequent release of the Luman-HCF complex from the cytoplasm in response to external signals and its translocation to the nucleus may activate IE110 and LAT, key viral genes required for reactivation of the latent virus.
We found that ectopically expressed Luman in cultured epithelial cells, and possibly also in sensory neurons of the trigeminal ganglia, is retained in the ER. This is reminiscent of the sterol enhancer binding proteins (SREBPs) (5
). These leucine-zipper-containing transcription activators are also sequestered in the ER and are released by proteolysis in response to reduced levels of cholesterol and sterols in the cellular environment. Subsequently, SREBPs are translocated to the nucleus, where they coordinately activate the expression of several genes involved in cholesterol and fatty acid biosynthesis. This strategy has probably evolved to allow a rapid response to low levels of substances that are critical for cellular survival.
Like the SREBPs, Luman is anchored to the ER by a hydrophobic domain. We found that deletion of this domain dramatically altered the location of Luman in the cell. Mutants of Luman that either lacked the transmembrane domain (Luman ΔTm) or lacked the domain as well as sequences downstream from it (Luman 1–220) accumulated in the nucleus. SREBP has two hydrophobic Tm domains that allow both the amino and carboxyl portions of the protein to protrude into the cytoplasm. The amino portion of the protein contains the activation domain, basic DNA binding domain, and leucine zipper, while the carboxyl terminus is thought to play a regulatory role. Luman appears to have a single transmembrane domain, suggesting that its proline-rich carboxyl terminus remains in the lumen of the ER. We speculate that as with the SREBPs the amino-terminal portion of Luman, which contains its domains for transcription activation dimerization and interactions with HCF and DNA, is released by proteolysis. The released product is probably unstable since mutants lacking the carboxyl portion of the protein accumulated in the cell at much lower levels than the wild-type protein (results not shown). The instability of the truncated protein may explain why, while we could visualize HCF in the nucleus of neurons of trigeminal ganglia, we were unable to follow the possible translocation of Luman from the cytoplasm to the nucleus in these cells. Instability of the truncated, and presumably active, form of Luman may be a way of ensuring that the activation of the genes that it regulates is transient.
We had previously shown that Luman strongly binds HCF both in vitro and in vivo. Here we demonstrate that Luman retained in the ER sequesters HCF. Since HCF is required for cell cycle progression, one consequence of this might be that cells expressing Luman are arrested in the G0
phase. This may explain why, despite several attempts, we have been unable to obtain cell lines expressing Luman, even when we have used tetracycline (14
)- and Ecdysone (34
)-inducible systems. It is possible that in the absence of inducers these systems allow the synthesis of small amounts of Luman which are enough to cause cell cycle arrest.
We observed that Luman-expressing cells were resistant to productive infection by HSV, at least as assessed by the lack of gC in these cells after infection. Since gC is made late in infection, we do not know at which stage of the replicative cycle the infection was interrupted or the mechanism by which Luman blocked HSV replication. Interaction between Luman and HCF was required for protection against HSV since the Luman (Y81A) mutant that does not bind HCF and does not retain it in the cytoplasm, failed to protect cells. This suggests that cells expressing Luman were protected from HSV replication by a mechanism that relied on retention of HCF in the cytoplasm. A simple explanation for how Luman might protect cells could be that Luman prevents HCF from entering the nucleus, thereby blocking VP16-mediated IE gene expression. However, our observations suggest that the mechanism might be more complex. In transfected cells enough Luman-HCF makes it to the nucleus to activate the IE110-ICP0 promoter in a VP16-independent manner. In addition, our preliminary results (unpublished) suggest that although Luman-expressing HSV-infected cells fail to make detectable levels of the structural proteins gC, gB, and gD, they do synthesize the IE proteins ICP0, ICP4, and ICP27.
Kristie and others found that, in contrast to other cell types, in neurons of the trigeminal ganglia HCF is located largely in the cytoplasm. It is translocated to the nucleus after death of the animal and in response to other stimuli that lead to reactivation of HSV in the mouse model (24
). These authors suggested that translocation of HCF to the nucleus may lead to the activation of IE gene expression and reactivation. Since latently infected neurons would not be expected to contain VP16, activation of IE genes by HCF would likely be by a VP16-independent mechanism. Kristie et al. suggest that this may involve GA binding protein (GABP). The IE110 promoter has binding sites for GABP, and T. M. Kristie (unpublished observations) has indicated that activation by GABP may require HCF.
Our observations of sections of bovine trigeminal ganglia support the results of our experiments with transfected cells in culture and may explain two of Kristie's observations. First, these results provide a mechanism for the sequestering of HCF in the neuronal cytoplasm. We found Luman and HCF in neuronal cytoplasm in a pattern similar to that observed by Kristie for HCF. In addition, in sections of ganglia the ER marker Calnexin had the same punctate pattern, suggesting that as in transfected cells Luman and HCF were associated with the ER. In transfected cells we showed that the sequestering of HCF by Luman in the ER correlated with resistance to HSV replication. Similar association of the two proteins in neurons could also suppress replication and lead to latency, if the initiation of viral replication at this site relied on an HCF-dependent process. Second, our observations suggest a mechanism for the initiation of the replicative cascade once both HCF and Luman are translocated to the nucleus. We showed that Luman could efficiently activate the promoters of IE110 and LAT, two genes that appear to be critical for reactivation. Our observations do not rule out the possibility that other neuronal factors, such as GABP (24
), also mediate activation. They simply suggest that Luman may also play a role in reactivation from latency. Recently, however, Davido and Leib (8
) showed that the region of the IE110 promoter that lies between −420 and −70 and includes the GABP-responsive sites is dispensable for reactivation from latency. The Luman-responsive CREs lie downstream from this dispensable region. Those authors discounted the role of the CRE because it did not bind CREB and was nonfunctional in undifferentiated neuroblastoma and pheochromocytoma cells (PC12). However, our preliminary data suggest that the IE110 CRE is relatively specific for Luman, and we have been unable to find Luman in undifferentiated PC12 cells. It is possible that Luman is expressed only after differentiation of these cells into neurons, and we are exploring this possibility. Interestingly, Su et al. (49
) have shown that HSV causes a long-term “quiescent” infection in differentiated PC12 cells. We have not examined differentiated PC12 cells for Luman expression, but it is tempting to speculate that Luman synthesis in response to differentiation causes these cells to become resistant to virus infection.
We hypothesize that, as with SREBP, the retention of Luman in the ER provides a means of responding rapidly to conditions that might be detrimental to neurons. Release of the complex from the ER and translocation to the nucleus leads to the coordinate expression of genes that can correct these conditions. Furthermore, rapid degradation of Luman released from the ER ensures that its activation of downstream genes was transient. By requiring HCF for the initiation of its replicative cascade, HSV exploits this pathway both for the establishment of latency and reactivation from it. In the absence of free nuclear HCF in differentiated neurons of the trigeminal ganglia, the virus is unable to replicate and becomes latent. Release of Luman-HCF and translocation to the nucleus activates IE expression in the absence of viral proteins and leads to reactivation.