Notch and Wnt Signaling: Mimicry and Manipulation by Gamma Herpesviruses

See allHide authors and affiliations

Science's STKE  16 May 2006:
Vol. 2006, Issue 335, pp. re4
DOI: 10.1126/stke.3352006re4


A small number of fundamental cell signaling pathways are key to the regulation of proliferation and differentiation responses during normal development. Two of these pathways, the Notch and Wnt pathways, have proven to be attractive targets for virus interaction and manipulation. In general, viral gene expression and replication are intimately linked to the differentiation state of the infected cell and, in the case of the gamma herpesviruses, establishment of a lifelong persistent infection in the host is also dependent on the proliferative expansion of an infected B cell population. This review examines the ways in which the gamma herpesviruses Epstein-Barr virus (EBV) and Kaposi’s sarcoma–associated herpesvirus (KSHV) have exploited the Notch and Wnt pathways to advance their own life cycles. The virus-pathway interactions are compared with the mechanisms and outcome of cellular Notch and Wnt signaling.


Epstein-Barr virus (EBV) and Kaposi’s sarcoma–associated herpesvirus (KSHV) are human lymphocryptoviruses and rhadinoviruses, respectively, of the gamma herpesvirus subfamily. EBV was originally identified in samples of Burkitt’s lymphoma and was one of the first viruses to be associated with a human cancer. EBV is associated with B cell malignancies (such as Burkitt’s lymphoma, posttransplant lymphoproliferative disease, and a subset of Hodgkin’s lymphomas) and with epithelial cancers (such as nasopharyngeal carcinoma and a subset of gastric carcinomas) (1). KSHV, like EBV, is an ancient virus but was not discovered until 1994 when viral DNA sequences were identified in samples of AIDS-associated Kaposi’s sarcoma (2). KSHV is associated with the B cell malignancies primary effusion lymphoma (PEL) and Castleman’s disease and the endothelial lesion Kaposi’s sarcoma (3). PEL is an unusual tumor in that the cells are dually infected with both KSHV and EBV in the majority of cases.

The association of these viruses with cancers derives from two features of their life-styles. First, as is typical of herpesviruses, they establish a lifelong infection in their human hosts. Persistent infection by EBV and KSHV depends on establishment of a reservoir of latent infection in B cells. In a latent infection, the virus expresses only a limited number of its genes and does not produce progeny virions, and the infected cells survive. This contrasts with lytic infection, in which the full repertoire of viral gene expression occurs, progeny virus is replicated, and the infected cells die. Second, during latency, each virus expresses proteins that stimulate cell proliferation. It is believed that this property serves to expand the infected B cell population upon primary infection to ensure entry into the memory B cell compartment before induction of the host immunological response. Viral-induced proliferation also serves as a mimic of antigen-induced B cell activation during lifelong persistence. Although the growth stimulation induced by these viruses is normally tightly controlled by the host, loss of immunological competence or the accumulation of secondary genetic errors in the host over time can be reflected in the development of viral-associated malignancies.

Two cell signaling pathways that are targeted by EBV and KSHV are the Notch and Wnt pathways. The viruses use components of these pathways to regulate their own viral gene expression and additionally alter cell gene expression through mimicry or manipulation of downstream pathway responses. The canonical Notch and Wnt pathways are outlined in Figs. 1 and 2; details of the participating proteins and pathway regulation can be found in several recent reviews (47).

Fig. 1.

EBV and KSHV proteins target CSL and mimic NotchIC. (A) The Notch receptor is present on the plasma membrane as a heterodimer. Interaction with ligand on neighboring cells results in sequential proteolytic cleavages that release the intracellular domain, Notch IC. The CSL [CBF1/Su(H)/Lag1] nuclear mediator of Notch signaling binds to the core sequence GTGGGAA and represses downstream promoters through recruitment of a multicomponent corepressor complex. Upon signaling, NotchIC enters the nucleus, binds to CSL, and, in cooperation with Mastermind and coactivators, displaces the corepressor complex and activates gene expression. Known target genes include basic helix-loop-helix proteins such as Hairy and enhancer of Split (HES) and Hairy/enhancer of Split related with YRPW motif (HEY). (B) The EBV latency protein EBNA2 is a transcriptional transactivator that binds CSL. EBNA2 responses are modulated negatively by EBNA3A and positively by EBNA-LP. (C) The KSHV lytic regulatory protein RTA also binds CSL. EBNA2 and RTA displace the corepressor complex in a mechanistically similar manner to NotchIC. Examples of EBNA2- and RTA-responsive genes are shown. Transcriptional responses to EBNA2 and RTA overlap with but differ from those of NotchIC. The differences may partially reflect the ability of EBNA2 and RTA to target promoters through CSL-independent mechanisms.

Fig. 2.

KSHV LANA manipulation of the Wnt pathway stabilizes β-catenin. The Wnt/β-catenin pathway regulates the availability of nuclear β-catenin. Binding of Wnt ligands to the Frizzled receptor and LRP5/6 coreceptors induces a series of phosphorylation events that lead to recruitment of axin to LRP5/6 and axin degradation and recruitment of Dishevelled and Frat1 to the β-catenin destruction complex. These events culminate in the disassociation of the destruction complex and stabilization of β-catenin. In the absence of Wnt signaling, β-catenin is held in a complex with axin, APC, GSK-3, and CKI. GSK-3 phosphorylation of axin and APC strengthens complex association. CKI provides priming activity for GSK-3 phosphorylation of β-catenin; β-catenin phosphorylation leads to its ubiquitination and degradation by the proteasome. Stabilization of β-catenin after Wnt signaling facilitates nuclear entry of β-catenin. In the nucleus, β-catenin binds to TCF/LEF transcription factors in association with coactivators (pygopus/BCL9) and activates gene transcription. In the absence of β-catenin, TCF/LEF associates with corepressors and negatively regulates downstream promoters. KSHV LANA binds to GSK-3 entering the nucleus in S phase and prevents nuclear export, resulting in nuclear accumulation of GSK-3. The reduced cytoplasmic abundance of GSK-3 allows stabilization and nuclear entry of β-catenin. GSK-3 phosphorylation of LANA is required for interaction between the two proteins. LANA also binds ERK1/2, which participates in Ser9 inactivation of GSK-3. Thus, active nuclear GSK-3 is LANA-bound, whereas inactivated GSK-3 is released from LANA, with the paradoxical outcome that LANA-mediated nuclear accumulation of GSK-3 is associated with a reduction in nuclear GSK-3 enzymatic activity.

The Notch receptor and Notch ligands are single-pass transmembrane proteins that influence cell fate decisions, proliferation, and differentiation (Fig. 1). In mammals, there are four Notch receptors, Notch1, 2, 3, and 4, and the ligands Jagged-1 and -2 and Delta-like-1, -3, and -4. Engagement of the extracellular domain of the Notch receptor by ligand induces TACE [tumor necrosis factor (TNF)–α converting enzyme] cleavage followed by a gamma-secretase complex intramembrane cleavage that releases the Notch intracellular domain (NotchIC). NotchIC enters the nucleus and targets the DNA binding protein CSL (CBF1/Suppressor of Hairless/Lag-1, also known as RBP-Jκ, for recombination signal sequence-binding protein–Jκ). In the absence of signaling, DNA-bound CSL is associated with a corepressor complex. Upon Notch signaling, the corepressor complex is displaced by NotchIC and the coactivator Mastermind, up-regulating target genes containing CSL binding sites. Structural analyses of the CSL-Notch-Mastermind complex suggest that a conformational change induced in DNA-bound CSL by the recruitment of NotchIC-Mastermind may contribute to corepressor displacement (8).

The Wnt pathway, which regulates development and proliferation, is complex both in terms of the number of recognized participants and in terms of the mechanism by which the signal is transmitted from the plasma membrane to the nucleus (Fig. 2). In outline, the Wnt family of secreted glycoproteins activates signaling by binding to transmembrane Frizzled family receptors and to an Lrp5/6 (low-density lipoprotein receptor–related protein 5 or 6) coreceptor. The nuclear effector of Wnt signaling is β-catenin, whose availability is regulated by a destruction complex that contains axin, APC (adenomatous polyposis coli), β-catenin, casein kinase I (CKI), and glycogen synthase kinase–3 (GSK-3). In this complex, β-catenin is phosphorylated by CKI, which acts as a priming kinase that enables β-catenin phosphorylation by GSK-3. Phosphorylation of β-catenin targets the protein for degradation. Upon Wnt signaling, a series of phosphorylation events facilitates axin recruitment to Lrp5/6 and recruitment of Dishevelled and Frat, with the resulting dissociation of the destruction complex and the loss of GSK-3–mediated phosphorylation of β-catenin. β-Catenin accumulates and then enters the nucleus, where it binds to the TCF/LEF (T cell factor/lymphoid enhancer factor) family of transcription factors to activate responsive genes. Again, in the absence of signaling, TCF/LEF proteins are associated with corepressors, whereas the TCF/LEF–β-catenin complex is associated with positive cofactors such as pygopus and BCL9 (B cell lymphoma 9).

KSHV and Manipulation of the Wnt Pathway

An initial examination of latently KSHV-infected B cell lines derived from PELs for evidence of alteration in one of the fundamental developmental signaling pathways revealed that β-catenin, the nuclear effector of the Wnt signaling pathway, accumulated to high levels in the cytoplasm (9). Evaluation of the mechanism underlying β-catenin accumulation determined that β-catenin protein turnover was reduced in PEL cells relative to that in virus-negative B cell lines. Stabilization of β-catenin in cancer cells is frequently associated with mutations in the serine or threonine residues (Ser45, Thr41, Ser37, and Ser33) whose priming by CKI and subsequent phosphorylation by GSK-3β results in ubiquitination and β-catenin degradation by the proteasome. However, the sequence of this region of the β-catenin gene in PEL cells proved to be wild type, raising the possibility that stabilization of β-catenin was effected through a KSHV-induced manipulation of the Wnt signaling pathway. KSHV establishes a predominantly latent infection in PEL cells, with only a variable and minor proportion (0.2 to 10%) of the cells spontaneously expressing KSHV lytic cycle proteins. Because β-catenin stabilization was detectable by immunofluorescence in virtually all PEL cells in the culture (10), it seemed likely that the alteration in β-catenin turnover was associated with expression of a KSHV latency protein. The protein in question proved to be LANA (latency-associated nuclear antigen), which is encoded by KSHV open reading frame (ORF) 73. PEL cells transiently transfected with an anti-LANA small interfering RNA expression vector showed concordant loss of LANA and β-catenin.

LANA is a 220-kD protein with unique N- and C-terminal domains separated by a central domain that varies in size between different KSHV isolates, represents about half of the polypeptide sequence, and contains three sets of repeats that are rich in aspartate and glutamate, proline and glutamine, and glutamate, respectively. LANA functions in the replication and maintenance of KSHV genomes during latency (11). In addition, LANA has interactions with cell proteins such as pRb and p53 that imply a role for LANA in promoting proliferation and protecting against p53-mediated cell death, and LANA-expressing cells have altered transcriptional profiles. LANA represses transcription when targeted to DNA as a Gal4 fusion protein. It is currently not clear how repression of specific promoters is generated in the face of core histone interactions, although interactions with several transcription factors such as CSL, Sp1, AP-1, and ATF4 have been described.

A substantial proportion of LANA-induced positive alterations in gene expression are likely to be a downstream effect of the manipulation of β-catenin levels by LANA. LANA dysregulates β-catenin through a novel mechanism that centers on an interaction between LANA and GSK-3. This interaction was first detected in a yeast two-hybrid screen performed to identify LANA-interacting proteins and was verified in PEL cells by coimmunoprecipitation of the endogenous proteins. There are two related forms of GSK-3, GSK-3α and GSK-3β, and both isoforms can participate in β-catenin regulation. LANA binds to both GSK-3α and GSK-3β; subsequent work on LANA used GSK-3β. GSK-3 is present in the cytoplasmic destruction complex, where it mediates the phosphorylation of β-catenin that results in β-catenin ubiquitination and degradation. However, GSK-3 also shuttles into the nucleus in a cell cycle–dependent manner and is detected in the nucleus after apoptotic stimuli and during cellular senescence of fibroblast cultures. Western blot analyses revealed that GSK-3β accumulated to high levels in the nucleus of PEL cells during S phase. This finding led to a model of LANA activity in which cell cycle–regulated nuclear accumulation of GSK-3 resulted in cytoplasmic depletion of GSK-3, loss of GSK-3 from the cytoplasmic destruction complex, and consequent accumulation of nonphosphorylated β-catenin, which is then able to enter the nucleus and activate responsive cellular genes.

Nuclear Accumulation of GSK-3

How does binding to LANA result in nuclear accumulation of GSK-3β, and how is this process reversed, at least partially, during the remainder of the cell cycle? Although our understanding of these processes is incomplete, what is known suggests a degree of mimicry of the interactions that occur in the cytoplasmic β-catenin destruction complex. Two separate regions of LANA, one N-terminal and the other at the C terminus, are required for efficient interaction with GSK-3β. The LANA C-terminal interaction region bears some resemblance to the GSK-3 interaction domain (GID) of axin. The functional similarity between the LANA GID and the axin GID was supported by studies using GSK-3 and LANA mutants (12). A GSK-3β Phe291 → Leu (F291L) mutation reduces the interaction with axin to 10% of that of wild-type GSK-3β, and this GSK-3 mutant also shows a 90% loss of binding to LANA. Similarly, conversion of a leucine residue to proline within the LANA GID or incursion into the LANA GID by deletion of the three C-terminal amino acids of the domain caused a loss of GSK-3 binding. Thus, LANA appears to be recapitulating the axin–GSK-3 interaction through the presence of a structurally similar GID. This mimicry of the interaction domain may also contribute to nuclear retention of GSK-3 bound to LANA.

The F291L mutation also reduces GSK-3β binding to Frat ("frequently rearranged in advanced T cell lymphomas"), a protein that has been postulated to serve two functions in canonical Wnt signaling (13). First, as an intermediate brought to the destruction complex upon activation of Wnt signaling, Frat1 interacts with Dishevelled and with the Wnt coreceptor LRP-5/6. In this context, Frat binds to GSK-3 in the destruction complex, where it is proposed to displace GSK-3 from axin, resulting in loss of GSK-3–mediated phosphorylation of β-catenin. The second postulated function of Frat is as a regulator of the export of GSK-3 from the nucleus. Humans have two Frat proteins, Frat1 and Frat2, whereas mice have a third Frat protein (Frat3) and Drosophila and Caenorhabditis elegans lack orthologous proteins. A confounding observation in terms of an essential role for Frat in the Wnt pathway in mammals comes from triple Frat1, Frat2, and Frat3 knockout mice, which are viable with no obvious developmental defects. Frat may therefore play a modulatory rather than an essential role in Wnt signaling in mice, or there may be as yet unrecognized redundant compensatory mechanisms. Nonetheless, the concept of Frat-mediated displacement of GSK-3 from axin provides an interesting model for LANA-mediated nuclear retention of GSK-3.

Binding of axin and Frat1 to GSK-3 is mutually competitive. Mutagenesis of GSK-3 identified a common binding region for Frat1 and axin wherein most, but not all, mutations had comparable effects on both axin and Frat1 binding. The crystal structures of GSK-3 bound to Frat or to axin-derived peptides described very similar tertiary structures, despite the differences in the primary sequence of the Frat and axin peptides and the fact that the interaction interface comprises an unbroken α helix in the axin peptide and two distinct α helices in the case of the Frat peptide. The mutagenesis experiments indicate that the LANA GID, again despite differences in primary amino acid sequence, is structurally similar to that of axin and Frat. In the presence of the CRM1 nuclear export receptor inhibitor leptomycin B, GSK-3 accumulates in the nucleus, which suggests that GSK-3 is actively exported. The GSK-3 interacting proteins axin and Frat1 each contain nuclear export signals and shuttle between the nucleus and cytoplasm. Frat1 has been implicated as the mediator of GSK-3 export on the basis of observations that overexpression of Frat1 increased GSK-3 cytoplasmic localization in transfection assays and that a peptide derived from the axin GID domain increased nuclear GSK-3 localization. However, a GSK-3 mutant, Glu290 → Gln (E290Q), that is impaired for Frat interaction was still exported, although less efficiently. Because this GSK-3 mutant retains axin binding and because the axin peptide would have blocked binding of both axin and Frat1 to GSK-3, one possibility is that axin may also mediate GSK-3 export. Nuclear accumulation of GSK-3 in the presence of LANA would then be explicable by competition between LANA and either Frat1 or axin for GSK-3 interaction. LANA could compete successfully either because of a higher binding affinity for GSK-3 or by being present at a higher concentration in the nucleus than either Frat1 or axin.

Phosphorylation Regulates the LANA–GSK-3 Interaction

Phosphorylation by GSK-3 affects the affinity of the protein-protein interactions in the β-catenin destruction complex. GSK-3 phosphorylation of axin increases axin stability and affinity for β-catenin, and GSK-3 phosphorylation of APC increases APC binding to the axin interaction domain on β-catenin. GSK-3 binding to LANA not only requires the LANA C-terminal GID domain but is also dependent on GSK-3 phosphorylation of the LANA N terminus. Deletion analyses defined a region between LANA amino acids 219 and 268 that was necessary to detect coprecipitation of LANA with GSK-3. Examination of this region revealed the presence of four consensus GSK-3 phosphorylation sites [(Ser/Thr)xxx(Ser/Thr)p, where the +4 position is primed by phosphorylation by another kinase]. Mutation of all four consensus sites abolished GSK-3 binding to LANA, indicating that the interaction between these two proteins is regulated by phosphorylation (12).

Endogenous LANA from PEL cells was confirmed to be a GSK-3 substrate. Most GSK-3 substrates require priming by another kinase before they can be phosphorylated by GSK-3. Axin is one of the exceptions and, given the presence of an axin-like GID in LANA, it was of interest to examine the requirement for priming. The GSK-3 mutant Arg96 → Ala (R96A) is mutated in the binding pocket for the priming phosphate residue and is unable to phosphorylate primed substrates, although it retains the ability to bind to and phosphorylate axin. The R96A mutant was unable to bind effectively to LANA, and GSK-3 phosphorylation of LANA in vitro required the addition of priming kinases. CKI family members participate in the regulation of Wnt signaling, and CKIα functions as the priming kinase for GSK-3 phosphorylation of β-catenin and APC in the destruction complex (14). CKI and mitogen-activated protein kinase (MAPK) were each found to function as priming kinases for GSK-3 phosphorylation of LANA in vitro.

KSHV can make use of the ability of LANA to promote nuclear accumulation of GSK-3 as a mechanism for dysregulation of β-catenin activity. The functional outcome is illustrated by an increased activity of β-catenin reporters in the presence of LANA, by increased expression of a β-catenin–responsive gene (cyclin D1), and by the observation that manipulation of the Wnt pathway along with the E2F-Rb pathway accounts for a substantial percentage of LANA-induced changes seen in gene array analyses (15). The ability of LANA to induce S-phase entry, a property that is also dependent on the integrity of the LANA GID (9), may be one of the properties that contributes to B cell hyperplasia and lymphoma in a LANA transgenic mouse (16). Although cytoplasmic depletion of GSK-3 is the model for β-catenin accumulation, this mechanism raises the question of the consequences of accumulating nuclear GSK-3. GSK-3 phosphorylation of p53 increases p53 transcriptional activity, whereas phosphorylation of targets such as c-Myc and cyclin D1 results in their increased turnover and degradation. An examination of GSK-3β in LANA-expressing cells found increased Ser9 phosphorylation, suggesting that, despite the positive role of GSK-3 in phosphorylating bound LANA, most of the nuclear GSK-3β was in an inactive phosphorylated form. Thus, paradoxically, increased accumulation of GSK-3 in the nucleus of LANA-expressing cells is associated with an overall decrease in nuclear GSK-3 activity.

Preliminary analyses suggest that the inactivation of GSK-3β in LANA-expressing cells is linked to the recruitment of extracellular signal-regulated kinase 1 (ERK1) to LANA (17). This has an interesting parallel in a recent report on β-catenin up-regulation by the hepatitis virus X protein (HBX) (18). A substantial proportion of hepatocellular carcinomas show abnormal β-catenin accumulation, and nuclear β-catenin is particularly prevalent in hepatocellular carcinoma associated with hepatitis B and hepatitis C infection. The ERK and p90 ribosomal S6 kinase (p90/RSK) inhibitor PD98059 and dominant negative ERK both blocked HBX-induced up-regulation of β-catenin transcriptional activity. ERK-associated p90/RSK was recruited to GSK-3β through an ERK docking motif, Phe-Lys-Phe-Pro, in the C terminus of GSK-3β. A pathway was therefore proposed whereby HBX activates ERK, which targets p90/RSK to GSK-3 and phosphorylates GSK-3β at Thr43. This phosphorylation then facilitates the p90/RSK-inactivating phosphorylation of GSK-3 Ser9 and leads to β-catenin stabilization. In the case of KSHV LANA, inactivation of nuclear GSK-3 may increase the activity of those factors such as c-Myc and cyclin D1 that are normally negatively regulated by GSK-3 phosphorylation and further contribute to LANA-mediated growth dysregulation.

β-Catenin Stabilization by EBV

β-catenin accumulation is observed in EBV-infected B cells with type III latency expression and in the epithelial cells of the EBV-associated malignancy nasopharyngeal carcinoma. EBV-infected B cells express either all latency genes (type III latency) or a more restricted latency profile, depending on their origin. In type III latency, the EBV EBNA1 (EB nuclear antigen 1), EBNA2, EBNA3A, 3B, and 3C, EBNA-LP, LMP1 (latent membrane protein 1), and LMP-2A and -2B proteins are expressed. The membrane proteins LMP1 and LMP2A activate the phosphatidylinositol 3-kinase (PI3K)–Akt signaling pathway. Increased Ser9-phosphorylated GSK-3β was observed in EBV-positive B cell lines, but inhibition of PI3K signaling did not prevent cytoplasmic accumulation of β-catenin or inactivation of GSK-3β, indicating that the underlying mechanism of β-catenin dysregulation in these cells does not derive from PI3K activation (19). This is consistent with observations that neither the inhibition of GSK-3 activity by Ser9 phosphorylation induced by insulin or insulin-like growth factor 1 nor the introduction of constitutively active Akt is sufficient to produce β-catenin stabilization in most cell types (5). In the case of EBV-infected B cells, LMP1 has been identified as activating a separate β-catenin stabilizing pathway. LMP1 was found to repress expression of the transcripts for the human homolog of Drosophila seven in absentia (Siah-1) (20). Siah-1 is an E3 ubiquitin ligase that binds to APC and promotes degradation of β-catenin in a GSK-3–independent manner. The β-catenin accumulation in nasopharyngeal carcinoma tissue was associated with the presence of phospho-inactivated GSK-3 and may represent a tissue in which cross talk between PI3K-Akt and β-catenin activation occurs (21).

EBV and CSL Interactions

EBV infection of B cells in vitro leads to the outgrowth of immortalized lymphoblastoid cell lines. EBNA2 is one of the first viral genes expressed after infection and functions as a transcriptional transactivator to control the complete latency III program. EBNA2 is essential for EBV immortalization of B cells. The first indication that EBV interfaced with the Notch pathway came from the recognition that EBNA2 was targeted to the promoters of EBNA2-responsive viral genes through an interaction with CSL, the nuclear effector of the Notch pathway (22). Subsequent studies revealed that both EBNA2 and NotchIC bound to the transcriptional repression domain of CSL (Fig. 1B). Binding of the CSL corepressor complex and either the EBNA2 coactivator complex or the NotchIC coactivator complex is competitive, such that there is a conversion from transcriptional repression of the CSL-bound promoter to transcriptional activation in the presence of activated NotchIC or EBNA2.

Considerable Notch literature indicates that relatively small (haploinsufficiency) dosage effects influence the outcome of Notch signaling (6). EBNA2 is not the only EBV-encoded protein to interact with CSL, and the importance of signal modulation for functional outcome is acutely apparent from studies with EBV (Fig. 1B). The EBV EBNA3A, 3B, and 3C family of proteins also bind to CSL. Binding is competitive with that of EBNA2 and leads to down-regulation of EBNA2-mediated promoter activation. Binding of the EBNA3 proteins to CSL also inhibits CSL binding, thus modulating EBNA2 function both through access to CSL and through CSL promoter targeting. EBNA3A is also necessary for the continued growth of in vitro EBV immortalized B cells. Experiments mapping functional domains of EBNA3A in a conditional expression system found that regions of EBNA3A that mediated repression of EBNA2 activity were essential for continued proliferation of these B cell cultures (23). The EBNA-LP protein acts as a coactivator of EBNA2-mediated transactivation. EBNA-LP localizes to promyelocytic leukemia nuclear bodies (PML NBs), and recently EBNA-LP was found to interact with the nuclear protein Sp100 through the PML NB targeting domain of Sp100 (24). EBNA-LP displaced Sp100 from PML nuclear bodies, and this function was sufficient to coactivate EBNA2 up-regulation of specific gene targets. It is not clear whether this EBNA-LP function is virus-specific or whether there may be cooperativity between Notch signaling and proteins of the PML NB. One example suggesting there may be some interplay comes from studies of the PML-RAR (PML–retinoic acid receptor) fusion protein. PML-RAR, which is generated by a chromosomal translocation in acute myelogenous leukemia, also disrupts PML NB structure. Profiling of PML-RAR–expressing cells identified Jagged1 as up-regulated at both the transcriptional and protein levels and found increased Notch signaling in reporter assays (25).

The importance of signal intensity and the modulation of signaling through integration with other signaling pathways is also illustrated by the response of cells to enforced EBNA2 expression in the absence of the other EBV latency proteins. Conditional expression of EBNA2 in Burkitt’s lymphoma B cells results in growth arrest (26), as does disturbing the EBNA2-CSL balance by conditional overexpression of EBNA3A. In the context of the EBV-infected lymphoblastoid cell, EBNA2 is required for proliferation; loss of EBNA2 transcriptional function through blockage of the EBNA2-CSL interaction with a cell-permeable peptide results in cessation of growth and a concurrent transcriptional induction of the cyclin and associated kinase (cyclin/CDK) inhibitor p21WAF1 (27). On the other hand, exogenous expression of EBNA2 in various epithelial cell lines leads to cessation of cell growth associated with induction of p21WAF1 (28). p21WAF1 may be central to the apparently contradictory effects of EBNA2 expressed in the presence or absence of other EBV modulatory proteins. The p21WAF1 promoter contains a CSL binding site, and studies on Notch1 have highlighted the importance of integrative signaling involving both transcriptional induction of p21WAF1 (29) and proteosomal degradation of p21WAF1 and the CDK inhibitor p27KIP1 (30) in the differentiation versus proliferation response.

A key unresolved question is the extent to which EBNA2-mediated gene regulation mimics Notch signaling. In reporter assays, NotchIC and EBNA2 both up-regulate expression of a number of promoters containing CSL binding sites; NotchIC and EBNA2 are also partially exchangeable in some functional assays. In conditionally EBNA2-regulated lymphoblastoid cell lines, Notch IC briefly supported cell proliferation in the presence of independently regulated LMP1, but there was a deficit in c-Myc expression and growth was not sustained in one study, although cells expressing higher levels of NotchIC survived in a second study. Gene array analyses in lymphoblastoid B cells conditionally expressing EBNA2 identified ~81 cellular genes whose expression was down-regulated within 24 hours of loss of EBNA2 activity (31). There was negligible overlap between the genes identified as EBNA2-regulated and those identified as Notch-regulated in a study in which T cell receptor activation of genes was compared in mouse thymocytes that did or did not contain activated NotchIC. This may reflect tissue-specific differences or differences in functional outcome. A proteomic comparison of cells conditionally expressing either EBNA2 or c-Myc found that 11 of 20 confirmed EBNA2 cellular targets were also c-Myc targets (32), which indicates that a substantial percentage of EBNA2 up-regulated cell gene expression is mediated through c-Myc and that this may be one component of the differences between EBNA2 and NotchIC responses. CSL independent effects have also been described for both NotchIC and EBNA2. Recent studies on microRNAs highlight another mechanism that could differentially affect gene regulation by EBNA2 and NotchIC. The worm Notch homolog has recently been shown to up-regulate expression of a microRNA, mir-61, that promotes secondary cell fate decisions (33). EBV and KSHV generate their own sets of microRNAs. The functions of these microRNAs have yet to be elucidated, but they may affect virus interaction with the Notch and Wnt signaling pathways.

KSHV and the Notch Pathway

In contrast to EBV, KSHV interaction with the CSL mediator of the Notch pathway is critical for productive lytic infection rather than during viral latency. The KSHV replication and transcription activator (RTA) protein is necessary and sufficient for reactivation of KSHV from latency. The RTA homolog in the rodent gamma herpesvirus MHV 68 functions similarly in the induction of MHV 68 lytic gene expression and replication to the extent that KSHV RTA can substitute for MHV 68 RTA. Interestingly, EBV RTA cannot substitute for MHV 68 RTA, which suggests that these two homologs use different mechanisms for targeting DNA (34). KSHV RTA binds directly to recognition sequences in the KSHV genome but also targets responsive promoters indirectly through interactions with cellular DNA binding proteins. One of these targeting partners is the CSL protein, and RTA targeted genes are poorly responsive in fibroblasts from CSL null mice (35). RTA binds to two regions of CSL, one of which is the same repression domain targeted by NotchIC and EBV EBNA2. This recapitulates the recurrent theme of displacement of corepressor complexes from CSL and introduction of a transcriptional activation domain, in this case that of RTA (Fig. 1C).

The degree of overlap between NotchIC and KSHV RTA responsiveness has been examined using a tetracycline-inducible system established in BCBL1 PEL cells (36). In this setting, NotchIC induced expression of 24 KSHV genes, including the viral interleukin-6 homolog (vIL-6), but was incapable of inducing the full spectrum of viral lytic gene expression and replication. One of the viral genes not induced by NotchIC was the one encoding RTA itself. This may be related to the observation that KSHV LANA can bind to the CSL binding sites in the RTA promoter and interfere with RTA promoter activation (37). The inability of NotchIC to induce RTA expression places KSHV reactivation under the control of the virus and prevents activation of the Notch pathway from being sufficient to trigger the full lytic cycle. This is biologically relevant because Notch1, 2, and 4 and their activated forms are expressed in KS tumor cells. The situation is also reminiscent of EBV where Notch is unable to activate a key latency gene, LMP-1, and therefore cannot fully substitute for EBV EBNA2. Although the majority of the infected cells in KSHV-associated lesions express only KSHV latency genes, there is consistent detection of a small percentage of tumor cells showing expression of lytic KSHV proteins. Two of the lytic proteins expressed in this setting are vIL-6 and vGPCR (the viral homolog of a G protein–coupled receptor), and both of these genes are now recognized to have NotchIC-responsive promoters (36, 38). Thus, the induction of a subset of KSHV genes by NotchIC may contribute to KSHV-associated pathogenesis.

Why the Notch and Wnt Pathways?

Why do EBV and KSHV mimic or manipulate Notch and Wnt pathway signaling? Dysregulated activation of Wnt and Notch signaling is associated with human cancers. The oncogenic potential of Notch in human disease was first recognized for T cell acute lymphocytic leukemia (T-ALL), and it was recently reported that a substantial proportion of these tumors contain activating mutations in the Notch1 gene (39). Dysregulation of the Wnt/β-catenin pathway was initially detailed in association with colon cancer, where truncating mutations of APC, mutations in the GSK-3 phosphorylation sites of β-catenin, and mutations in the scaffold protein axin 2 are prevalent. EBV and KSHV are also associated with cancers. However, evolutionarily successful viruses persist because they successfully coexist with their human hosts, and induction of cancers by EBV and KSHV is seen as an inadvertent and relatively infrequent accident. Lifelong persistence of an EBV or KSHV infection is maintained in B cells. In vivo latency has been subject to greater analysis in the case of EBV. EBV is transmitted in saliva, and a model has evolved in which EBV infection of naïve B cells in oro- and nasopharyngeal tonsillar tissues drives these cells to become activated and proliferate (40). It is these cells that express the full (type III) EBV latency program that includes the Epstein-Barr nuclear antigens EBNA2, EBNA3A, 3B, and 3C that interact with the CSL mediator of Notch signaling and the LMP1 protein that has been implicated in up-regulation of β-catenin, the nuclear effector of Wnt signaling. The proliferating EBV-infected blasts enter a germinal center reaction from which they emerge either as EBV-positive resting memory B cells with little or no viral protein expression or as infected plasma cells whose more differentiated state provides an environment favorable for EBV lytic replication and release of progeny virions. Aspects of the role of Notch and Wnt in the hematopoietic compartment may therefore be relevant.

The role of Notch signaling in hematopoiesis after the early stages of embryonic development is a subject of debate. Gain-of-function studies involving expression of constitutively active Notch1 and coculture experiments with cells expressing Notch ligands have found a role for Notch signaling in hematopoietic stem cell self-renewal. However, conditional loss of function of CSL in mice produced T cell deficiencies but normal B cell development, whereas conditional loss of Notch2 resulted in a specific defect in the generation of marginal zone B cells. An alternative approach to examining Notch signaling in the hematopoietic compartment involved the use of transgenic Notch reporter mice (41). This study found that Notch signaling was active in hematopoietic stem cells in vivo and was down-regulated as these cells differentiated. A cross between the Notch reporter mice and TOPGAL transgenic mice (in which expression of the lacZ gene is regulated by a promoter containing three TCF/LEF-binding sequences) revealed that, in a high proportion of cells in the hematopoietic stem cell niche, both Notch and Wnt pathways were simultaneously active, suggesting that the activities of these two pathways are integrated. Examination of the effect of Notch signaling on Wnt responsiveness, using hematopoietic stem cells treated with a dominant negative CSL protein and purified Wnt3A and stem cell factor (SCF), found that blocking Notch signaling did not affect cell cycling or viability but rather led to accelerated differentiation. This finding suggested that the Wnt signal is dominant for the proliferative response, whereas Notch signaling functions primarily to maintain these cells in an undifferentiated state. In a study on long-term repopulation of mice with mouse or human hematopoietic stem cells (42), inhibition of GSK-3 led to up-regulation of Wnt and Notch signaling and an initial activation and then a down-regulation of Hedgehog signaling in the stem cell population. The authors proposed that modulation of signaling through these three pathways regulates stem cell output by increasing the number of progenitors produced rather than promoting self-renewal or expansion of the stem cell pool.

One could speculate about slightly different but overlapping scenarios for EBV and KSHV latency. In the case of EBV infection of naïve B cells, EBNA2 stimulates c-myc activation and may also provide an antidifferentiation function that complements the growth-proliferative functions provided by LMP1 mimicry of constitutively activated CD40 signaling (43) and the survival signals provided by LMP2A mimicry of constitutively activated B cell receptor signaling (44). Less is known about the details of KSHV infection of B cells because of the inability to establish KSHV-infected B cell lines in vitro. The KSHV-encoded K15 protein is expressed in latently infected cells and activates some of the same signaling pathways as EBV LMP1 and LMP2A (3). The activation of β-catenin and inactivation of GSK-3 induced by KSHV LANA may not only provide the equivalent of Wnt signaling, including up-regulation of c-Myc, but may also support activation of the Notch pathway through cross talk.

EBV and KSHV also infect cell types other than B cells. EBV infects epithelial cells and KSHV infects endothelial cells, and each virus is associated with malignancies in the respective tissues. The outcome of Notch signaling, proliferation versus differentiation, is cell type–specific. EBNA2 is not expressed in EBV-infected epithelial cells. Because EBNA2 is the regulator of type III latency gene expression in B cells, this allows a completely different EBV gene expression pattern to be used in epithelial cells and renders moot the issue of EBNA2-mediated tissue-specific Notch-like responses. The alternatively spliced EBV BamHI-A rightward transcripts (BARTs) are abundantly expressed in EBV-positive epithelial cells. Two of the ORFs within these transcripts encode proteins that negatively modulate Notch responses when exogenously overexpressed, suggesting that EBV may negatively regulate Notch signaling in epithelial cells. However, one caveat is that although the BART RNAs are readily detected, it has been difficult to demonstrate the presence of BART-encoded proteins in EBV-infected cells. LMP2A is expressed in infected epithelial cells. The activation of the PI3K-Akt pathway mediated by LMP2A in epithelial cells has been linked to β-catenin stabilization.

In endothelial cells, the use of CSL as a targeting protein for the KSHV lytic cycle is consistent with the virus tapping into the Notch pathway and possibly augmenting the natural outcome of Notch signaling in this setting. Notch4 and Jagged1 promote differentiation in endothelial cells. The KSHV lytic cycle takes place in nonproliferating cells. Transcription of KSHV-encoded genes such as K8 (also known as RAP, for replication-associated protein) and RTA is induced by activation of the AP-1 pathway (45) that also activates the Notch4 promoter in endothelial cells. The KSHV K8/RAP and RTA proteins reinforce the arrested state by increasing C/EBPα (CAAT/enhancer-binding protein α) and p21 expression and protein stability to further inhibit G1 to S phase transition (46).


Stabilization of β-catenin by KSHV LANA involves the novel mechanism of intracellular redistribution of GSK-3, such that GSK-3 accumulates in the nucleus and is depleted in the cytoplasm. This redistribution is likely a consequence of normal S-phase entry of GSK-3 into the nucleus, coupled to diminished nuclear export resulting from LANA binding to GSK-3 through an axin and Frat-like GSK-3 binding domain, thus masking the binding site for GSK-3 nuclear export partners. Viruses are talented at multitasking, and LANA uses the GSK-3 interaction to further manipulate nuclear GSK-3 activity. Interaction between LANA and GSK-3 requires GSK-3 phosphorylation of LANA, and hence LANA recruits enzymatically active GSK-3. However, LANA also binds to ERK1/2, which suggests that LANA can facilitate ERK-p90/RSK inactivation of GSK-3 by Ser9 phosphorylation. In this model, LANA-mediated nuclear accumulation of GSK-3 is paradoxically associated with reduced levels of nuclear GSK-3 activity. The importance of the Wnt pathway in gamma herpesvirus infections is reinforced by the observation that β-catenin activity is also up-regulated in EBV-infected B cells and epithelial cells. LMP2A activation of the PI3K-Akt pathway has been implicated in β-catenin stabilization in epithelial cells, and LMP1 has been suggested to affect the alternative pathway of Siah degradation of β-catenin in B cells. Overall, the interaction of KSHV and EBV with the Wnt pathway can be considered to be one of pathway manipulation—in other words, constitutive activation of the nuclear effector of the Wnt pathway predominantly by inactivation of the negative regulator GSK-3.

In contrast, EBV and KSHV interaction with the Notch signaling pathway is more akin to viral mimicry. EBV EBNA2 targets the CSL nuclear effector of Notch signaling in a mechanistically identical fashion of corepressor displacement and coactivator engagement, with the key difference being that Notch signaling is transient whereas EBV EBNA2 is continually present. An important unresolved issue is exactly how much overlap there is in the changes in downstream gene expression elicited by EBNA2 and NotchIC. This issue relates to the relative importance of EBNA2-mediated c-myc activation versus activation of common target genes in secondary lymphoid tissue, the site of naïve B cell infection by EBV, and whether the virus is driving a recapitulation of stem cell behavior in B cells that would not normally be subjected to Notch signaling cues. The targeting of KSHV RTA to CSL differs from that of EBV EBNA2 in that RTA regulates the KSHV lytic cycle, which takes place in growth-arrested cells. On the other hand, the interaction with CSL itself is mechanistically comparable to that of EBNA2 and thus again is a mimic of NotchIC.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
  46. 46.
  47. 47.
View Abstract

Navigate This Article