(Received for publication, February 25, 1997, and in revised form, May 23, 1997)
From the Gladstone Institute of Virology and
Immunology, San Francisco, California 94141-9100, the
§ Department of Pathology, ¶ School of Public Health,
Columbia University, New York, New York 10032, and the
Department of Medicine, University of California,
San Francisco, California 94141-9100
The present studies analyzed the biologic
activity of a gene product (vIL-6) encoded by the recently discovered
Kaposi's sarcoma-associated herpesvirus (KSHV) bearing 24.8% amino
acid identity with human interleukin-6 (huIL-6). Based on this
similarity, we hypothesized that this viral homolog might trigger the
JAK/STAT pathway, which typically is engaged by IL-6 and other
cytokines. Activation of receptor-associated Janus tyrosine kinases
(JAKs) results in the subsequent phosphorylation of signal transducers
and activators of transcription (STATs) leading to nuclear entry and
transcriptional regulation of target genes. Treatment of HepG2 cells
with culture medium containing recombinant KSHV-encoded vIL-6 led to
rapid induction of JAK1 phosphorylation and a nuclear DNA-binding
activity found to contain STAT1 and STAT3. An antibody to the IL-6
receptor (IL-6R) subunit effectively neutralized the response to
huIL-6 but failed to block STAT activation by vIL-6. In contrast, an antibody reactive with the gp130 subunit of IL-6R abrogated signaling of both responses. Moreover, a transfected cell line expressing human
gp130 without IL-6R
exhibited a robust response to vIL-6 but not to
huIL-6. These results demonstrate that KSHV encodes a cytokine that
activates specific JAK/STAT signaling via interactions with the gp130
signal transducing subunit independently of the IL-6R
chain. This
activity may have an impact on gp130-mediated signaling in response to
native cytokines and thereby influence disease pathogenesis upon KSHV
infection.
A recently discovered virus belonging to the subfamily of gamma herpesviruses, Kaposi's sarcoma-associated herpesvirus (KSHV),1 is a likely causative factor for the development of Kaposi's sarcoma (KS) (1) as well as other neoplasias associated with KS, including body cavity based lymphoma (BCBL) (also known as pleural effusion lymphomas, or PEL) (2-4) and multicentric Castelman's disease (5). Multiple KSHV genome-containing cell lines have been cultured out of PELs, including BCBL-1, BC-1, and BCP-1 (6-8). Sequencing of the viral genome has revealed co-linearity and a notable degree of sequence homology with herpesvirus saimiri (9, 10). Like herpesvirus saimiri, KSHV contains multiple open reading frames encoding cellular homologs, including a G-protein-coupled receptor, cyclin D1 (11, 12), and bcl-2 (10, 13, 14) genes. The KSHV open reading frame K2 gene (15) encodes a protein (vIL-6) with 24.8% amino acid sequence identity with that of human IL-6 (huIL-6) that can promote survival of an IL-6-dependent cell line (15, 16). vIL-6 is expressed in KSHV-infected hematopoietic cells including PEL and derived cell lines (15) and some forms of multicentric Castelman's disease2 but is rarely found in KS lesions (15).
Human IL-6 may contribute to a variety of B cell neoplastic disorders (17-20). Normally, IL-6 functions in the regulation of the immune system and other tissues. It is known to stimulate B cells (18, 21), and it has the ability to induce growth arrest and differentiation of M1 myeloid precursor cells into macrophages (22-24). Other studies demonstrate an anti-apoptotic function induced by IL-6 in some B cell neoplasms and mediated through apoptotic antagonists such as Bcl-xL (25, 26). IL-6 also plays a pivotal role in hepatic tissue, where it causes the expression of multiple acute phase response (APR) proteins (27).
The receptor for IL-6 contains two heterologous subunits, IL-6R and
gp130. Both receptors belong to the cytokine receptor superfamily
containing described canonical motifs (28). To initiate signaling by
the IL-6 receptor complex, the ligand must first make a low affinity
contact with the IL-6R
subunit which then recruits the signal
transducing gp130 subunit. Importantly, receptor function depends on
the dimerization of gp130 subunits (29). IL-6 thus belongs to a small
family of cytokines that function via the homodimerization of gp130 or
the heterodimerization of gp130 with LIFR
, another signal
transducing subunit.
A prominent signaling event induced by this dimerization of signal transducing subunits is activation of the JAK-STAT system, a recently described pathway employed by numerous cytokine receptor systems (30) and known to be activated specifically by the IL-6 family of cytokines (31, 32). Ligand-induced receptor assembly causes the activation of associated Janus protein tyrosine kinases (JAKs) resulting in phosphorylation of receptor tyrosines and signal transducers and activators of transcription (STAT) factors that interact physically with specific receptor phosphotyrosines. Finally, phosphorylation of the STAT factors directs their dimerization and relocalization to the nucleus where they influence the transcription of target genes (30, 33) by binding to specific promoter elements. Previous work in the IL-6 system provided evidence that the gp130 subunit can associate variously with the JAK kinases, JAK1, -2, and Tyk2, and that JAK activation leads to STAT1 and STAT3 DNA binding and gene transactivation in a cell context-dependent manner (34-36). Earlier reports suggest that some APR genes contain STAT3 binding motifs (37-39). In addition, STAT3 may be the essential factor in mediating many of the principal functions of IL-6, such as growth arrest and differentiation of M1 cells into macrophages (22-24) and survival of IL-6-responsive cells (40).
In these studies we endeavored to determine whether the JAK-STAT signaling program was induced by the KSHV cytokine homolog vIL-6, whether this response mimicked that of huIL-6, and whether the cell surface receptor engaged by the viral cytokine is related to the classical IL-6 receptor complex.
HepG2 cells were cultured in minimal essential medium (Life Technologies, Inc.), 10% fetal bovine serum (Gemini, Calabasis, CA), 1% 100 × non-essential amino acids and 1% 100 × sodium pyruvate. BCBL-1 cells (a gift of Dr. Don Ganem) were grown in RPMI 1640, 10% fetal bovine serum, and 0.05 M 2-mercaptoethanol.
vIL-6 and the reverse vIL-6 (6-LIv) control were prepared as described
previously. Open reading frame K2 was cloned into the pMET7 expression
vector in forward and reverse orientations and transfected into COS7
cells. COS-derived supernatants were then concentrated 10-fold with a
Centriplus 10 filter. HepG2 cells were treated for 10-15 min with
0.5-1.0 ml COS media containing vIL-6 or reverse vIL-6, 100 ng/ml
human IL-6 (R & D Systems, Minneapolis, MN), or 1 nM human
LIF (R & D Systems) prior to preparation of nuclear extracts. Soluble
IL-6R fragments (sIL-6R
, R & D Systems) were used where indicated
at 50 ng/ml.
BCBL-1 cells were stimulated with 20 ng/ml 12-O-tetradecanoylphorbol-13-acetate (TPA; Sigma), and then cell culture supernatant was harvested. HepG2 cells were then incubated with 1 ml of supernatant for 10-15 min before nuclear extraction.
BAF-130 cells, representing Ba/F3 cells expressing transfected human
gp130 but not IL-6R (41), were kindly provided by Dr. M. Narazaki
(Osaka University Medical School). They were maintained in RPMI 1640 supplemented with 10% fetal bovine serum, 10% WEHI 3B-conditioned
medium as a source of IL-3, and geneticin 0.3 mg/ml (Life Technologies,
Inc.).
10-40 × 106 HepG2 cells or 40-50 × 106 BAF-130 cells were incubated with COS7 supernatants and then washed in calcium- and magnesium-free phosphate-buffered saline. Nuclear extracts were prepared as described previously (42, 43).
The STAT-response element from the FcRI gene promoter
was used for EMSA assays. The probe was prepared by end-labeling
oligonucleotides with [
-32P]dATP (Amersham Corp.) and
polynucleotide kinase (Boehringer Mannheim). Binding reactions were
performed using 105 cpm, 3 µg of poly[d(I-C)], binding
buffer, and 10 µg of nuclear extract (52). Probe specificity
competition was done with 50 ng of Fc
RI and IRF-1
oligonucleotides. Sequences of oligonucleotides are as follows:
Fc
RI, GTATTTCCCAGAAAAAGGAC; IRF-1,
GCCTGATTTCCCCGAAATGACGG. Preincubations of nuclear extracts with
antibodies were performed for 45 min on ice in the absence of
poly[d(I-C)] and binding buffer as described previously (42). The
STAT1 antibody was from Transduction Laboratories (Lexington, KY), and
STAT2 and -3 antibodies were obtained from Santa Cruz Biotechnology
(Santa Cruz, CA), JAK1, 4G10, and anti-SHC were from Upstate
Biotechnology (Lake Placid, NY), and MOPC21 (monoclonal
IgG1) and MOPC195 (IgG2b control) were from
Cappel (Durham, NC).
Receptor neutralization studies were performed using monoclonal
antibodies reactive with human IL-6R, gp130, and a polyclonal antibody against IL-6R
(R & D Systems, Minneapolis, MN) or with MOPC21 (monoclonal IgG1) and anti-SHC (polyclonal IgG)
antibodies as controls. 10-40 × 106 HepG2 cells were
incubated for 1 h with various concentrations of antibody and then
subjected to cytokine stimulation for 15 min, and nuclear extracts were
subsequently prepared as described above.
Cell lysates were prepared from 30-40 × 106 cells using a buffer containing 1% Nonidet P-40, 20 mM Tris·HCl, pH 8.0, 150 mM NaCl, 50 mM NaF, 100 mM sodium orthovanadate, 1 mM phenylmethylsulfonyl fluoride, leupeptin (10 mg/ml), aprotinin (10 mg/ml), pepstatin A (1 mg/ml) as described previously (63) and pre-cleared once overnight and twice for 30 min with 5 µg of rabbit preimmune serum (Cappel, Durham, NC) and 100 µl of protein A-agarose (Boehringer Mannheim). JAKs were immunoprecipitated with 5 µg of rabbit polyclonal antibody against JAK1, -2, and -3 (Upstate Biotechnology, Lake Placid, NY) and Tyk2 (a gift of Dr. Oscar Colaminici) for 1 h at 4 °C and then incubated with 30 µl of protein A-agarose for 1 h. Immunoprecipitates were washed 3 times in lysis buffer and boiled in SDS sample buffer for 5 min. Immunoblots were performed with 4G10 and visualized by enhanced chemiluminescence (Amersham Corp.). To assess protein loading, blots were stripped (100 mM 2-mercaptoethanol, 2% sodium dodecyl sulfate, 62.5 mM Tris·HCl, pH 6.7) for 30 min at 55 °C and reblotted with appropriate JAK antibodies.
Luciferase AssayHepG2 cells were plated at 1.8 × 106 cells/well density in 6-well plates 16 h before transfection. Transfections were performed with LipofectAMINE reagent (Life Technologies, Inc.) as per manufacturer's instructions using 1 µg per well of reporter plasmid DNA containing 4 IRF-1 promoter STAT-response elements upstream of a minimal TK promoter and firefly luciferase gene (a gift of Ligand Corp.) (44). 48 h after transfection cells were incubated in media containing 100 ng/ml IL-6, 1 ml of vIL-6, or 1 ml of control reverse vIL-6 (6-LIv) media for 6 h. Cells were washed in phosphate-buffered saline and lysed in 200 µl of Reporter Lysis Buffer (Promega, Madison, WI). Lysates were exposed to luciferase substrates (Analytical Luminescence Laboratory, Ann Arbor, MI) and counted for luminescence units on a Monolight 2000 (Analytical Luminescence Laboratories). Lysates were assayed for protein concentration using the Bio-Rad Protein Assay (Bio-Rad) to control for protein quantity.
Concentrated COS7 cell supernatants expressing
recombinant vIL-6 or the reverse 6-LIv as a negative control were used
to stimulate HepG2 cells, a human hepatoma line that expresses IL-6
receptors (45). Serial immunoprecipitations of JAKs were carried out in HepG2 cells after cytokine stimulation, followed by immunoblotting with
anti-phosphotyrosine antibodies to measure the tyrosine phosphorylation of these kinases as a marker of their activation (Fig.
1A). Phosphorylation of JAK1
occurred in response to both vIL-6 and huIL-6 but not the 6-LIv control
supernatant. Stripping and re-blotting with anti-JAK1 antibodies
demonstrated equivalent levels of JAK1 protein expression in each
sample. However, the phosphorylation status of other JAK kinases
revealed no reproducible activation of either JAK2 or Tyk2 in samples
treated with either huIL-6 or vIL-6 (data not shown).
Subsequently, the induction of STAT factors by the expressed viral
cytokine was assessed. Treatment of HepG2 cells with the recombinant
supernatants resulted in the appearance of three discrete nucleoprotein
complexes in electrophoretic mobility shift assays (EMSA) employing the
STAT-binding probe FcRI (Fig. 1B). This tripartite DNA-binding activity mirrors that found upon stimulation with huIL-6. To demonstrate that these DNA-binding activities were
binding the oligonucleotide probe specifically, cold competition was
performed with a related gamma activation sequence element probe, from
the IRF-1 promoter (Fig. 2A).
Preincubation of extracts with excesses of either Fc
RI
or IRF-1 oligonucleotide sequences caused complete elimination of the 3 bands. Furthermore, incubation of these extracts with
anti-phosphotyrosine antibodies caused a diminution of DNA-binding
activity, providing evidence that this activity requires tyrosine
phosphorylation sites, characteristic of STAT factors (Fig.
2B). Finally, antibody supershift experiments with
antibodies raised against STAT proteins were performed to elucidate
precisely the components involved in these complexes. An antibody
against the N terminus of STAT1 caused a band supershift concomitant
with the loss of the lower two bands in the tripartite band pattern
(Fig. 2B). Additionally, the anti-STAT3 antibody completely
abolished the upper two bands but did not yield a supershifted complex.
This indicates that the lower and upper bands consist of STAT1 and
STAT3 homodimers, respectively, whereas the intermediate band contained
STAT1/STAT3 heterodimers. A similar profile of STAT activation has been
described in response to huIL-6 (36) and other gp130-interacting
cytokines (46-48).
vIL-6 Induces STAT-dependent Transcription
To
examine vIL-6 induced gene expression, a reporter plasmid containing
the firefly luciferase gene downstream of the STAT-responsive IRF-1
element linked to a minimal promoter was transiently expressed in HepG2
cells. A 30-fold increase in luciferase activity in response to huIL-6
relative to the unstimulated condition was observed in these cells
(Fig. 3).
Others have reported similar levels of transcriptional transactivation
in response to IL-6 (33). Likewise, exposure of reporter-expressing
cells to vIL-6 caused a comparable increase in transcriptional
activity. The lower level of transcriptional activity triggered by
vIL-6 relative to that of huIL-6 may be due either to lower specific
activity of vIL-6 compared with huIL-6 or to differences in relative
concentration. These data demonstrate that STAT factors activated
following stimulation with the viral cytokine are capable of
transactivating gene expression and therefore may play a role in
modulating gene expression in cells expressing the appropriate cell
surface receptors.
The KSHV Cytokine Interacts with the gp130 Subunit but Not the IL-6R
The IL-6R subunit contains an extracellular
domain that binds the ligand at low affinity but is not alone
sufficient to spark an intracellular signal. Signal initiation depends
on the ligation of IL-6/IL-6R
with gp130 subunits. In light of the
similarities found between huIL-6 and vIL-6 in terms of JAK/STAT
activation, we asked whether these two molecules also mimic each other
in their interactions with receptor subunits. To this end, HepG2 cells
were treated with increasing concentrations of a monoclonal antibody
against IL-6R
and then stimulated with huIL-6. A loss of signaling
capacity was observed at high antibody concentrations using STAT DNA
binding activity in EMSA as an indicator (Fig. 4A). Unexpectedly, when the same procedure
was carried out using vIL-6 instead of huIL-6, no loss of STAT
activation was detected. Furthermore, when cells preincubated with
antibody were exposed to a 1:3 dilution of vIL-6 supernatant, no loss
of STAT-activated DNA binding activity was observed. Similar studies
employing a polyclonal antibody to neutralize IL-6R
also
demonstrated no inhibition of vIL-6 activity, while substantially
diminishing that of huIL-6 (data not shown). A similar antibody
competition approach was then employed to elucidate the role of the
gp130 subunit in signaling by the KSHV cytokine. In contrast to the anti-IL6-R
antibody, an anti-gp130 monoclonal antibody at high concentrations inhibited both the huIL-6 and vIL-6 signals (Fig. 4B). These findings suggest that the viral cytokine, unlike
huIL-6 itself, utilizes gp130 without necessarily interacting with
IL-6R
; IL-6R
apparently does not facilitate the association of
vIL-6 with gp130 as it does with IL-6. Therefore, by these criteria, it
appears unlikely that the viral cytokine must rely on recruitment of
IL-6R
to elicit its intracellular signal.
As an alternate test of the hypothesis that vIL-6 differs from huIL-6
by selectively engaging gp130, signaling studies were performed using a
transfected IL-3-dependent murine cell line expressing
exclusively human gp130 but not IL-6R. These BAF-130 cells have been
shown previously to respond to huIL-6 only upon the addition of soluble
fragments of the IL-6R
chain (sIL-6R
) to the culture (41).
Likewise, we observed that the induction in these cells of STAT DNA
binding by huIL-6 was undetectable by EMSA with the STAT-binding probe
in the absence of sIL-6R
but was readily evident as a
three-component nucleoprotein complex when sIL-6R
was added (Fig.
5). Strikingly, supernatants containing vIL-6 (but not the 6LI-v
negative control) induced the indistinguishable tripartite
nucleoprotein species in BAF-130 cells regardless of the presence or
absence of sIL-6Ra (Fig. 5). These findings confirm that vIL-6 engages
the IL-6R in a markedly different manner from that of huIL-6 and that
this interaction is functionally independent of the IL-6R
subunit.
BC-1 and BCP-1, cell lines
that are coinfected with KSHV and Epstein-Barr virus or KSHV alone,
respectively, have been demonstrated to produce transcripts encoding
vIL-6 in response to TPA (15). BCBL-1 is a body cavity-based lymphoma
(BCBL) cell line that is KSHV-positive and Epstein-Barr virus-negative,
similar to BCP-1. To assess further the potential biologic activity of
vIL-6 within the viral life cycle, we subjected BCBL-1 cells to TPA
treatment (20 ng/ml for 48 h) and harvested the culture medium;
this treatment of BCBL-1 cells has been shown to induce both the
expression of KSHV genes (including that encoding vIL-6) and the lytic
replication of KSHV, which is otherwise latent and transcriptionally
suppressed in these cells (7, 49). HepG2 cells subsequently were
exposed to the TPA-induced BCBL-1 supernatant for 15 min (Fig.
6). EMSA of nuclear extracts from these
cells using a probe containing a STAT-binding motif demonstrated
induction of a DNA binding complex similar to that seen after
stimulation by huIL-6, with a pattern that suggested activation of
multiple STAT factors. In contrast, exposure of the HepG2 cells to TPA
itself or uninduced BCBL-1 supernatant did not result in a similar
banding pattern. Further studies have indicated that BCP-1, when
induced by TPA, also secretes a similar biologic activity (data not
shown). Therefore, induction of KSHV-infected cells by phorbol ester
results in the secretion of a bioactivity with signaling properties
that resemble those of vIL-6.
The goal of the present studies was to explore the signaling capabilities of a putative cytokine encoded by KSHV. Previous reports have shown that this protein, deemed vIL-6, supports the survival of B9 plasmacytoma cells and can substitute for huIL-6 in this capacity (15, 16, 50). In view of this IL-6-like activity, we sought to elucidate in more detail the functional relationship between huIL-6 and vIL-6 by assessing the JAK-STAT intracellular signaling pathway using HepG2 cells. Indeed, vIL-6 was found to trigger the phosphorylation of JAK1, a result in agreement with the literature surrounding IL-6. Several groups have shown that IL-6 is responsible for the activation of various JAKs in a pattern that is cell context-dependent (34, 35, 51), and in HepG2 cells both IL-6 and vIL-6 selectively activated JAK1.
JAK1 has been suggested to be the primary kinase mediating STAT activation after IL-6 stimulation, and the present findings suggest that JAK1 is also integral to mediating downstream STAT activation events by KSHV vIL-6. Both cytokines induced a tripartite pattern of DNA binding complexes as detected by EMSA; these nucleoproteins were found to represent STAT1 and STAT3 in homo- and heterodimeric formations. The same pattern of induction has been observed previously in the IL-6 system as detected by both EMSA analysis and phosphotyrosine immunoblot analysis (35, 51, 53, 54). There is reported evidence that the pattern of STAT activation, like that of the JAKs, differs in a cell context-dependent manner and that in many situations STAT1 is induced by huIL-6 to a lesser extent than is STAT3 (36). In fact, EMSA experiments utilizing extracts from the TF-1 erythroleukemia line revealed that stimulation with huIL-6, LIF, or vIL-6 induced exclusively homodimers of STAT3 (data not shown). Despite such variation among different cell types, huIL-6 and vIL-6 exert indistinguishable effects upon a given cell type.
Target gene transactivation is the key biologic effect of STAT induction by cytokines. Reporter systems directly assaying transcriptional activity have shown this function for many of the known STAT factors (33). Using a reporter construct containing a STAT-response element from the promoter region of the IRF-1 gene, we detected a level of STAT-dependent transcriptional activity triggered by vIL-6 that was comparable to that induced by huIL-6. This activity directly implies downstream effector function which would result from viral cytokine stimulation of target cells. Specifically, STAT3 has been documented as an integral factor in the APR necessary for the transcriptional activation of APR genes in hepatic cells (37-39). Additionally, STAT3 function has been associated with differentiation of myeloid precursor cells into macrophages (23) and with cell survival (40). The function of STAT1 in IL-6 signaling may be partially clarified by studies of STAT1 gene-disrupted mice. These mice, while incompetent in their ability to respond to interferon stimulation, were able to respond to a host of other cytokines that usually trigger STAT1 activation, including IL-6 (55, 56). In vitro studies have also shown that STAT1 and STAT3 have largely overlapping response element specificity (33, 57). Therefore, it seems likely that STAT3 determines the specificity of the STAT transcriptional effects in response to both IL-6 and vIL-6. Further work is needed to define more precisely the pertinent gene targets.
In addition to measuring signal transduction events after cytokine
stimulation we examined the interactions of vIL-6 with human cell
surface receptors. Binding assays performed with radiolabeled huIL-6
and HepG2 cells demonstrated that the KSHV vIL-6 was able modestly
(15-25%) to inhibit radiolabeled ligand binding (data not shown),
indicating that vIL-6 employs at least some of the human IL-6 receptor
components to initiate its signal. Subsequent dissection of the huIL-6R
with neutralizing antibodies found that gp130, but not IL-6R, is a
critical component of the receptor complex engaged by the viral
cytokine. In fact, the vIL-6 molecule apparently contains a conserved
Trp residue homologous to part of the site 3 gp130-binding site of IL-6
(58, 59). These findings do not exclude the possibility that IL-6R
may contact the viral cytokine under normal circumstances, but such an
interaction may not be essential in the initiation of the signaling
program. This hypothesis is strongly supported by the marked functional
responses we observed in BAF-130 cells treated with vIL-6, but not with huIL-6, in the absence of IL-6R
. These findings differ from a recent
report providing evidence with neutralizing antibodies that IL-6R
may play a role in the vIL-6 signal in murine cells (16). This
distinction raises the possibility that the viral cytokine may differ
in the way it interacts with the human and murine receptors and that
vIL-6, as the product of a virus that is tropic for human cells, has a
higher affinity for human gp130 allowing the two to interact without
huIL-6R
. Alternatively, vIL-6 may react with human and murine
IL-6R
chains with similar affinity, but at a contact site that
differs from huIL-6 and that is not blocked by antibodies that
neutralize the huIL-6/huIL-6R
interaction. In any event, it is clear
from our experiments that huIL-6R
chains are dispensable for the
functional interaction of vIL-6 with human gp130 leading to effective
signal transduction, a feature that substantially distinguishes the
viral cytokine from native human IL-6.
The promiscuity of the gp130 subunit within multiple receptor complexes
makes conceivable the notion that the viral cytokine could interact
with a number of other receptor subunits, although vIL-6 is more
similar in its primary sequence to IL-6 than it is to IL-11, OSM, CNTF,
LIF, or CT-1. LIFR, CNTFR
, OSMR
, and IL-11R
are all known
to be involved in receptor complexes containing gp130 (21).
Interestingly, at least one report has indicated that an antibody to
the gp130 subunit is sufficient to cause dimerization and subsequent
signal transduction (60), leaving such homodimerization as a possible
mechanism of vIL-6 action. These possibilities are beyond the scope of
the present studies but are grounds for further examination. In any
event, the engagement of gp130 by vIL-6 is the likely molecular basis
of intracellular signaling initiation by this viral cytokine.
KSHV is preferentially found in peripheral B lymphocytes (61, 62), a feature that is consistent with its classification as a gamma herpesvirus (9). Furthermore, transcription from the KSHV genome in latently infected cell lines is tightly regulated (49). The observation that the KSHV-containing BC-1 and BCP-1 cell lines express vIL-6 transcripts when induced by phorbol ester suggests that this protein may play a role in the maintenance of replicating virus, although vIL-6 is also expressed in latently infected cells (15). Indeed, we observed that phorbol ester induced secretion of protein with IL-6-like signaling activity from KSHV-infected cells, as demonstrated by the profile of STAT activation. The recent findings that STAT3 plays a role in anti-apoptosis of IL-6-stimulated cells (40) and that IL-6 prevents B cell apoptosis (25, 26) argue that vIL-6 may function in protection from apoptotic responses of these cells to viral infection. Additionally, this cytokine may act synergistically with other KSHV products, such as bcl-2 and cyclinD1 (11, 12) homologs, to prolong the survival of infected cells.
In situ immunostaining to detect the viral cytokine demonstrated cells in KS lesions generally do not express vIL-6, in marked contrast to the more prevalent expression in cells derived from a PEL (15). In light of the evidence, it may be more likely that vIL-6 exerts the bulk of its effects on hematopoietic cells rather than on the endothelial cells involved in KS lesion formation. It remains to be seen, however, whether hematopoietic cells are the only cell lineage affected by vIL-6 expression and precisely which in vivo events invoke this expression. Nonetheless, the stimulation of a signaling program resembling that of the gp130-utilizing family of cytokines may play a pivotal role in the life cycle of KSHV and the pathogenesis of associated diseases and may therefore provide a useful therapeutic target.
We thank Dr. S. L. Gaffen and K. D. Liu for helpful discussions and M. Ha for technical assistance. We appreciate the generous gifts of reagents from Dr. D. Ganem and Dr. H. M. Seidel (Ligand Corp.). We acknowledge the excellent assistance of John Carroll, Amy Corder, Jessica Diamond, and Brian Clark in the preparation of this manuscript.