Activation of Tyk2 and Stat3 Is Required for the Apoptotic Actions of Interferon-β in Primary Pro-B Cells*

The growth-inhibitory effects of type 1 interferons (IFNs) (IFNα/β) are complex, and the role of apoptosis in their antigrowth effects is variable and not well understood. We have examined primary murine interleukin-7-dependent bone marrow-derived pro-B cells, where IFNβ, but not IFNα, induces programmed cell death (PCD). IFNβ-stimulated apoptosis is the same in pro-B cells derived from wild type and Stat1–/– mice. However, in pro-B cells from Tyk2–/– mice, where there is normal activation of Stat1 and Stat2, IFNβ-stimulated PCD is not observed. Loss of B cells in lymphocytic choriomeningitis virus-infected mice has been shown to be mediated through the expression of IFNα/β (1). In wild type mice infected with lymphocytic choriomeningitis virus, there is a greater loss of B cells in the bone marrow and spleen than in Tyk2–/– mice infected with the virus, suggesting that the expression of this kinase plays an in vivo role in IFNα/β-mediated PCD. In contrast to IFNβ-stimulated tyrosine phosphorylation of Stat1 and Stat2, Stat3 tyrosine phosphorylation is defective in Tyk2–/– pro-B cells, suggesting that this Stat family member is required for apoptosis. In support of this hypothesis, inhibition of Stat3 activation in wild type B cells reverses the apoptotic effects of IFNβ. Furthermore, expression of a constitutively active form of Stat3 in Tyk2–/– B cells partially restores IFNβ-stimulated PCD. These results demonstrate an important role of Tyk2-mediated tyrosine phosphorylation of Stat3 in the ability of IFNβ to stimulate apoptosis of primary pro-B cells.

The antiviral, antigrowth, and immunomodulatory activities of interferons (IFNs), 2 appear to be controlled, in part, by a set of cellular genes that are rapidly induced upon the binding of type 1 interferons (IFN␣/␤) or type 2 interferon (IFN␥) to specific cell surface receptors. Type 1 IFNs consist of several subtypes of IFN␣, a single gene encoding IFN␤ as well as IFN and IFN. These cytokines bind to a common receptor composed of two subunits, IFNaR1 and IFNaR2. In most systems, the actions of IFN␣ and IFN␤ are similar; however, there are reports that IFN␤ and IFN␣ can stimulate distinct sets of genes and selectively regulate different biological responses (2).
Interferon-stimulated gene factors mediate gene induction by binding to enhancers found within the promoters of type 1 IFN-induced early response genes (3,4). Src homology 2 domain-containing transcription factors called Stats are required for the activation of genes by IFNs. These transcription factors are covalently modified by tyrosine phosphorylation and subsequently translocate to the nucleus such that they interact with enhancers needed for interferon-stimulated gene expression. The Jak family of tyrosine kinases is also an integral component in these signaling cascades (5). Expression of both Jak1 and Tyk2 is required for type 1 IFN activation of Stats in human cells, whereas in mouse cells, the actions of type 1 IFNs are only partially dependent upon Tyk2 (6,7) in that Stat1 and Stat2, but not Stat3, are tyrosine-phosphorylated in Tyk2-null cells incubated with IFN␣. The biological consequences of type 1 IFN activation of Stat3 in most cells are not clear.
Type 1 IFNs are inhibitors of cell growth and are presently used for the treatment of leukemias and other malignancies, such as melanomas and renal cell carcinomas. Although it is clear that type 1 IFN activation of the Jak/Stat pathway is required for their antiproliferative effects (8), it is not clear what role, if any, this signaling cascade plays in determining whether type 1 IFNs stimulate cell cycle arrest, a slowing of the cell cycle without accumulation in a single phase of the cell cycle, or induction of programmed cell death (PCD). In Daudi cells, a human Burkitt lymphoma B cell line, cells are arrested at the G 0 /G 1 phase of the cell cycle after treatment with IFN␣, whereas in the human leukemic Jurkat cell line, IFN␣ causes a slowing of the cell cycle without inducing cell cycle arrest or apoptosis (9). Daudi cells show a suppression of the DNA binding activity of the E2F transcription factor as well as decreased levels of phosphorylation of the retinoblastoma protein (10,11).
A variety of hematopoiesis-derived T and B cell lines, certain melanomas, and other cell lines derived from solid tumors have been reported to undergo apoptosis when incubated with type 1 IFNs (12)(13)(14)(15)(16). The best described system to examine type 1 IFN-stimulated apoptosis of primary cells is murine IL-7-dependent bone marrow-derived B cells. These pro-B cells show both inhibited cell growth as well as apoptosis when incubated with a combination of virally produced IFN␣ and IFN␤ for 24 -48 h. Similar effects of this mixture of cytokines are seen in several IL-7-dependent murine pro-B cell lines (16). Two reports using mice support in vitro studies concerning the ability of type 1 IFN to inhibit the expression of pro-B cells in vivo. IFN␣ treatment of newborn mice inhibits the development of both T and B cell populations, and infection of mice with lymphocytic choriomeningitis virus (LCMV) causes a transient bone marrow aplasia due to the production of type 1 IFNs (1,17). In both of these reports, the actions of either IFN␣ or  infection of mice with LCMV did not occur in mice where the IFNaR1  subunit of the type 1 IFN receptor was deleted. It remains to be determined whether under these same conditions type 1 IFNs inhibit cell  proliferation by inducing apoptosis. IFN ␤-mediated inhibition of cell growth in primary murine pro-B cells does not require the expression of Stat1 but appears to require IFN␣-stimulated expression and translocation of Daxx (18). Daxx expression has been associated with Fas-regulated apoptosis. This protein has been found in promyelocytic leukemia protein (PML) nuclear bodies as well as in the cytoplasm. IFN␣-mediated induction of Daxx does not require the expression of Fas and appears to be blunted in Tyk2 Ϫ/Ϫ cells (18,19).
Previous studies have indicated that in the presence of vanadate, IFN␣ can stimulate PCD in a variety of tissue culture cells, whereas treatment with IFN␣ alone slows cell growth without inducing apoptosis (20). In the human 2fTGH fibrosarcoma cell mutant that does not express the tyrosine kinase Tyk2, the apoptotic actions of IFN␣ and vanadate are lost. If these Tyk2 Ϫ/Ϫ cells are reconstituted with wild type Tyk2, but not kinase-inactive Tyk2, IFN␣ ϩ vanadate-stimulated PCD is restored. This observation was interesting, because expression of the kinase-inactive form of Tyk2 does restore IFN␣-induced tyrosine phosphorylation of Stat1 and Stat2 and the induction of early response genes that are controlled by these transcription factors (21). However, incubation of cells that express kinase-inactive Tyk2 with IFN␣ does not restore tyrosine phosphorylation of Stat3. Given that the kinase-active form of Tyk2 is required for IFN␣ ϩ vanadate-stimulated PCD, it is likely that other events in addition to activation of Stat1-dependent genes must also be needed for this event. These results provided a clue to a potential role of type 1 IFN-stimulated tyrosine phosphorylation of Stat3 as it relates to the biological actions of this cytokine. Stat3 activation was also implicated in a report examining an uncharacterized Daudi cell variant that is insensitive to the antigrowth and antiviral actions of type one interferons (22).
In this study, we confirm that IFN␤-stimulated PCD of primary pro-B cells requires the expression of Tyk2 in vitro, and these results are also seen in mice infected with LCMV. Our studies also implicate Stat3 as a key regulator of IFN␤-stimulated apoptosis of pro-B cells. Although Stat3 has been known for several years to be activated by type 1 IFNs, this is one of the first studies that documents a physiological role of IFN␣/␤-induced tyrosine phosphorylation of this transcription factor in primary cells.

MATERIALS AND METHODS
Cell Culture and Reagents-Murine IL-7-dependent bone marrowderived pro-B cells were isolated from 129Sv mice and the various mice deficient in Stat1, Stat 1, 5a/b, and Tyk2, as described (23). IL-7 was present at all times in the experiments described here. Murine IFN␤ was prepared as described (24). Murine IFN␣ was a generous gift of Dr. Ernest Borden.
Antibodies and Chemical Reagents-Anti-Stat1, anti-Stat2, and anti-Stat3 antisera were used as described previously (25). Anti-phosphospecific Stat1, Stat2, and Stat3 were purchased from Cell Signaling. Stat3 blocking peptide was obtained from Calbiochem and used at a final concentration of 500 M (26).
Immunoblot Analysis-Proteins were separated on 8% SDS-polyacrylamide gels and transferred to polyvinylidene difluoride membranes (Millipore). Membranes were immunoblotted with the indicated antibodies. Immunoblots were developed using horseradish peroxidaseconjugated secondary antibodies (Zymed Laboratories Inc.) and ECL (Amersham Biosciences).
Measurement of Apoptosis by Annexin V Staining-Cells were left untreated or incubated with type 1 IFNs (1000 units/ml) for various times at 37°C. Cells were collected, washed with phosphate-buffered saline, and resuspended in 100 l of binding buffer (10 mM HEPES, pH 7.4, 140 mM NaCl, and 2.5 mM CaCl 2 ). Cells were then incubated with 5 l of annexin V-fluorescein isothiocyanate or PE (Pharmingen, Palo Alto, CA) for 10 min at room temperature in the dark, followed by the addition of 400 l of binding buffer. Cells were collected ungated (10,000 events) and analyzed by flow cytometry using a FACScan TM (BD Biosciences). Data were further analyzed using CellQuest TM (BD Biosciences).

Expression of Tyk2 Is Required for IFN␤-mediated Apoptosis of
Murine Pro-B Cells-To examine the role of Tyk2 in IFN␤-stimulated apoptosis, we studied primary bone marrow-derived IL-7-dependent pro-B cells, since they have been shown previously to undergo PCD in the presence of a combination of IFN␣ and IFN␤ (16). Furthermore, in Stat1 Ϫ/Ϫ pro-B cells, the antigrowth effects of type 1 IFNs are intact, whereas in Tyk2 Ϫ/Ϫ cells, the antigrowth effects of IFN␤ are lost (19,23). These reports, however, did not address the issue of type 1 IFNstimulated apoptosis in pro-B cells. To examine whether IFN␤-stimulated PCD in pro-B cells requires the expression of Stat1 and Tyk2, we isolated IL-7-dependent pro-B cells from bone marrow of Tyk2 Ϫ/Ϫ , Stat1 Ϫ/Ϫ , and Stat1,5a,b Ϫ/Ϫ mice. Since the genetic background of each knock-out mouse varied, B cells from wild type littermates were used to confirm that IFN␤-mediated apoptotic responses were intact. IFN␤stimulated apoptosis in pro-B cells isolated from mice with different genetic backgrounds did not significantly differ (data not shown). Cells were incubated with or without IFN␤ (1000 units/ml) or IFN␥ (10 ng/ml) for 36 h. IL-7 was present during all incubations. Apoptosis was assayed by staining with annexin V. (Table 1). Both IFN␤ and IFN␥ stimulated apoptosis in pro-B cells from wild type mice. Induction of PCD by IFN␤ was also observed in cells from mice that do not express Stat1 or Stat1,5a,b. In contrast, IFN␥ was not able to stimulate PCD in cells from Stat1 Ϫ/Ϫ or Stat1,5a,b Ϫ/Ϫ mice. Cells isolated from mice that do not express Tyk2 showed normal induction of apoptosis in the presence of IFN␥, but IFN␤-induced apoptosis was absent. Similar results were observed using terminal dUTP nick-end labeling assays (data not shown). Interestingly, the antiproliferative actions of IFN␤ are present

Interferon-induced apoptosis in murine IL-7-dependent pro-B cells
Cells were incubated with or without IFN␤ (1000 units/ml) or IFN␥ (10 ng/ml) for 36 h prior to staining with annexin V. Annexin V from untreated cells was substracted from cells incubated with IFNs. The background staining ranged from 2 to 10%. Data are presented as mean Ϯ S.D. values of four independent experiments.

Role for Stat3 Activation in IFN␤-stimulated Apoptosis
but are not as pronounced in Tyk2 Ϫ/Ϫ pro-B cells when compared with wild type pro-B cell counterparts. The difference in the antigrowth effects of IFN␤ in wild type compared with Tyk2 Ϫ/Ϫ cells is approximately the percentage of annexin V-positive cells seen in wild type cells incubated with IFN␤ (Table 2). These results differ from those of Shimoda et al. (19), who observed no antiproliferative effects of type 1 IFNs in bone marrow cells from Tyk2 Ϫ/Ϫ mice. However, they were not isolating the same population of bone marrow cells used in these assays. Previous studies examining the actions of type 1 IFNs on PCD of pro-B cells have used either a combination of IFN␣ and IFN␤ isolated from virally infected cells or recombinant IFN␣ (16,19,23). We wanted to examine whether there were any differential actions of IFN␣ and IFN␤ on apoptosis of pro-B cells. Cells isolated from wild type mice were incubated with or without recombinant murine IFN␣ or recombinant IFN␤ for 48 h, and numbers of cells were counted or cells were stained with annexin V (Table 3). Cells incubated with either IFN␤ or IFN␣ displayed about 50 and 34% fewer cells, respectively, than untreated cells. However, only cells treated with IFN␤ were annexin V-positive. Although this finding was somewhat surprising, since most of the actions of IFN␣ and IFN␤ appear to be similar, there are a number of reports suggesting that the actions of these two IFNs can be selective with regard to the activation of immediate early genes as well as their antiviral and antiproliferative actions (for a review, see Ref. 2).
To confirm that the lack of Tyk2 expression is responsible for resistance to IFN␤-induced PCD, Tyk2 Ϫ/Ϫ cells were reconstituted with murine Tyk2, encoded in the retroviral vector MSCV-IRES-GFP (27). Pools of cells that stably express either GFP or Tyk2-GFP were obtained by FACS sorting, and these cells were incubated without or with IFN␤ for 48 h. The amount of apoptosis was then determined by annexin V staining (Fig. 1). Tyk2-null cells that express only GFP are resistant to IFN␤-stimulated apoptosis, whereas wild type or Tyk2-null pro-B cells reconstituted with Tyk2 showed equivalent degrees of IFN␤-stimulated apoptosis. These results demonstrate that the mechanisms by which IFN␤ and IFN␥ induce PCD are distinct, although they both use Stat1 as a transcription factor to regulate a variety of their biological actions. Furthermore, expression of Tyk2 is required for the apoptotic actions of IFN␤. This result is surprising in light of reports that activation of Stat1 and Stat2, as well as antiviral effects of IFN␣, are similar in wild type and Tyk2Ϫ/Ϫ mice (6, 7).

IFN␤-induced Apoptosis of Pro-B Cells Requires Activation of Stat3-
Previous studies have indicated that there is a selective defect in IFN␣stimulated tyrosine phosphorylation of Stat3 in Tyk2 Ϫ/Ϫ bone marrowderived macrophages, whereas tyrosine phosphorylation of Stat3 is normal in cells incubated with IFN␥, IL-12, IL-6, and IL-10 (6, 7). We wanted to confirm these findings in primary B cells that were used for the apoptosis assays. Cell extracts were prepared from wild type, Stat1 Ϫ/Ϫ , Stat1, 5a/b Ϫ/Ϫ , and Tyk2 Ϫ/Ϫ pro-B cells that had been incubated with or without IFN␤ (1000 units/ml) for 20 min. Immunoblots were probed with phosphotyrosine-specific Stat1 antiserum ( Fig. 2A top panel) and also with Stat1 antiserum to ensure equal loading of protein ( Fig. 2A, bottom). As reported by Shimoda et al. (6) and Karaghiosoff et al. (7), using primary bone marrow-derived macrophages, we detected comparable levels of IFN␤-stimulated tyrosinephosphorylated Stat1 in pro-B cells from wild type and Tyk2 Ϫ/Ϫ mice. We next examined IFN␤-stimulated tyrosine phosphorylation of Stat3 (Fig. 2B, top). Whereas IFN␤-stimulated tyrosine phosphorylation of Stat3 was observed in wild type, Stat1 Ϫ/Ϫ , and Stat1,5a,b Ϫ/Ϫ pro-B cells, we were unable to detect any tyrosine phosphorylation of Stat3 in Tyk2 Ϫ/Ϫ pro-B cells. IFN␤-stimulated tyrosine phosphorylation of Stat3 in the Tyk2-null B cells was examined after a variety of incubation times (10 -60 min) with the cytokine to ensure that there was no change in the kinetics of activation. These findings correlate with an absence of

Role for Stat3 Activation in IFN␤-stimulated Apoptosis
IFN␣/␤-stimulated tyrosine phosphorylation of Stat3 in bone marrowderived macrophages isolated from Tyk2-null mice (7). As mentioned earlier, in 2fTGH cells that do not express Tyk2 or express a kinase-inactive Tyk2, IFN␣-stimulated tyrosine phosphorylation of Stat3 and IFN␣ ϩ vanadate-induced PCD are not observed (20,21). The loss of IFN␤-stimulated tyrosine phosphorylation of Stat3 in Tyk2 Ϫ/Ϫ pro-B cells and its activation in all other B cells where IFN␤ stimulates PCD suggests that Tyk2 may mediate some or all of its apoptotic actions through changes in the phosphorylation state of Stat3. To directly address this issue, we infected primary bone marrow-derived B cells from wild type mice with constructs that express either GFP or constitutively active (CA) or dominant negative (DN) Stat3 (27). CA-Stat3 contains two substitutions of cysteines in its Src homology 2 domain, which results in spontaneous dimerization of the protein (28). Pools of cells that stably express GFP were obtained by FACS, and these cells were incubated without or with IFN␤ for 48 h. The amount of apoptosis was then determined by annexin V staining (Table 4). Pro-B cells that express either GFP or GFP-CA STAT3 showed approximately a 50% increase in annexin V-positive cells following treatment with IFN␤ compared with untreated samples. However, pro-B cells that expressed GFP-DN Stat3 showed only a 20% increase in annexin V-positive staining compared with untreated samples. The basal apoptosis in each pool of cells was approximately the same (5%). To determine the amount of tyrosine-phosphorylated Stat3, each of the pools of infected cells was incubated with IFN␤ for 30 min. Cell lysates were then subjected to Western blotting with phosphotyrosine-specific Stat3 antiserum (Fig. 3, middle). Cells expressing MSCV-GFP showed induction in tyrosine phosphorylation of Stat3 after incubation with IFN␤ (Fig. 3,  compare lanes 1 and 2). This induction is decreased about 50% in cells expressing DN Stat3 (Fig. 3, compare lanes 2 and 4), whereas cells expressing CA Stat3 show a basal level of tyrosine-phosphorylated Stat3 that is further increased by incubation of cells with the cytokine (Fig. 3,  compare lanes 5 and 6). The partial decrease in tyrosine-phosphorylated Stat3 in cells expressing the dominant negative form of the protein correlates with the partial but significant decrease in IFN␤-stimulated apoptosis seen in these cells. The fact that elevated levels of tyrosinephosphorylated Stat3 in cells expressing the constitutively active form of the protein show neither enhanced basal nor IFN␤-stimulated apoptosis indicates that activation of Stat3 on its own is not sufficient to program the apoptotic response. To determine whether expression of the DN-Stat3 was causing a nonspecific effect on IFN␤ signaling, we probed the same blot with antiserum that recognizes tyrosine-phosphorylated Stat2 (Fig. 3, top). IFN␤-stimulated phosphorylation of Stat2 was the same in all three cell lines. To ensure equal protein loading of the samples, we also probed the blot for actin (Fig. 3, bottom), which was comparable for all samples.
As an independent confirmation of the role of Stat3 in the apoptotic actions of IFN␤, we also incubated wild type primary pro-B cells with a cell-permeable peptide that has been shown to block the actions of tyrosine-phosphorylated Stat3 (26). This peptide selectively prevents binding of tyrosine-phosphorylated Stat3 to DNA (26). Pro-B cells were incubated with the Stat3 inhibitor peptide (500 M) for 1 h prior to being incubated with or without IFN␤ for an additional 48 h. Apoptotic cells were then measured by staining with annexin V (Table 5). Treatment of primary pro-B cells with IFN␤ induced 27% more annexin V-positive cells compared with untreated cells. Incubation of cells with the Stat3 inhibitor peptide completely reversed the apoptotic actions of IFN␤. Incubation of cells with only the inhibitor peptide without IFN␤ had no effect on the basal number of annexin V-positive cells, which in this particular experiment was 11% (data not shown). Incubation of cells with a nonspecific peptide at the same concentration as the inhibitor peptide did not alter the apoptotic actions of IFN␤ (data not shown). These results confirm our findings that expression of DN Stat3 prevents IFN␤-stimulated apoptosis of pro-B cells and provides an independent assay for the requirement of IFN␤-stimulated tyrosine phosphorylation of Stat3 to induce PCD in these cells.
Expression of Constitutively Active Stat3 in Tyk2 Ϫ/Ϫ Cells Restores IFN␤-stimulated Apoptosis-If the inability of IFN␤ to induce tyrosine phosphorylation of Stat3 is responsible for the lack of apoptosis in Tyk2 Ϫ/Ϫ pro-B cells, one would predict that expression of constitutively active Stat3 in these cells would restore this response. We therefore infected Tyk2 Ϫ/Ϫ pro-B cells with either wild type or a constitutively active form of Stat3. GFP-expressing cells were isolated by FACS and incubated with or without IFN␤ for 48 or 72 h prior to staining with annexin V (Fig. 4A). Tyk2 Ϫ/Ϫ cells that express wild type Stat3 show no IFN␤-induced PCD when incubated 48 or 72 h with this cytokine. However, Tyk2 Ϫ/Ϫ cells that express CA-Stat3 displayed IFN␤-stimulated PCD after 72 h. Interestingly, induction of apoptosis in the Tyk2 Ϫ/Ϫ cells expressing CA-Stat3 is delayed compared with wild type pro-B cells incubated with IFN␤, since we observed PCD in wild type cells after 48 h (or less) of incubation with the cytokine. The reasons for this delay are not clear, but it may be due to a Tyk2-dependent, Stat3-independent pathway that is activated by IFN␤, which is needed for the

IFN␤-stimulated apoptosis is decreased in cells that express dominant negative Stat3
Wild type pro-B cells were infected with GFP retroviruses that express either GFP alone or GFP and DN or CA Stat3. Stable pools of GFP-expressing cells were selected and incubated with or without IFN␤ for 48 h prior to staining with annexin V. The percentage of annexin V-positive cells from untreated cells was subtracted from each sample. The levels of apoptosis in untreated cells ranged from 3 to 6% and were not significantly different in those lines expressing Stat3 DN or Stat3 CA. Role for Stat3 Activation in IFN␤-stimulated Apoptosis more accelerated PCD seen in wild type cells. In Tyk2 Ϫ/Ϫ pro-B cells, constitutively active Stat3 is also tyrosine-phosphorylated when the cells are left untreated (Fig. 4B, lane 7). IFN␤ does stimulate tyrosine phosphorylation of CA-Stat3 in Tyk2 Ϫ/Ϫ cells but not to the levels seen in wild type cells that express CA-Stat3 (Fig. 3, lanes 5 and 6). IFN␤ also stimulates tyrosine phosphorylation of wild type Stat3, which is overexpressed in Tyk2 Ϫ/Ϫ cells (Fig. 4B, lanes 5 and 6) but not to the same extent as in Tyk2 Ϫ/Ϫ cells that express CA-Stat3 (Fig. 4B, lanes 7 and 8).

MSCV-GFP
Note that the levels of expression of wild type Stat3 and CA-Stat3 in Tyk2 Ϫ/Ϫ pro-B cells are approximately the same. We have no proof of why CA-Stat3 is tyrosine-phosphorylated in response to IFN␤ in Tyk2 Ϫ/Ϫ cells. However, this Stat3 spontaneously dimerizes, which may make it a better substrate for another tyrosine kinase that has a low affinity for Stat3 but can phosphorylate the protein when it is in a dimerized configuration. Tyk2 Ϫ/Ϫ Mice Are Relatively Resistant to LCMV-induced Depletion of Hematopoietic Cells-LCMV-induced bone marrow aplasia occurs in mice that do not express the IFN␥ receptor but not in those that do not express the IFNaR1 subunit of the type 1 IFN receptor (1). The actions of LCMV thus appear to be a direct effect of type 1 IFN expression.
We have examined the effects of LCMV infection on expression of nucleated cells and B220 ϩ B cells in the bone marrow and spleen in wild type and Tyk2 Ϫ/Ϫ mice. There were no significant differences in the numbers of nucleated cells in the bone marrow and spleen of uninfected wild type and Tyk2 Ϫ/Ϫ mice (Table 6). Wild type and Tyk2 Ϫ/Ϫ mice (3 mice/group) were given an intravenous dose of 2 ϫ 10 6 plaque-forming units of LCMV-Clone 13. Three days postinfection, bone marrow and spleen cells were harvested and analyzed for total nucleated cells ( Table  7). There was an ϳ84% decrease in the number of nucleated cells in the bone marrow of wild type mice infected with LCMV compared with uninfected mice. Interestingly, in LCMV-infected Tyk2 Ϫ/Ϫ mice, there was only a 49% loss of total nucleated cells compared with uninfected mice. The numbers of nucleated splenic cells from LCMV-infected Tyk2 Ϫ/Ϫ mice showed no significant changes compared with uninfected Tyk2 Ϫ/Ϫ mice, whereas wild type mice infected with the virus showed ϳ50% fewer cells compared with uninfected mice.
We also examined the numbers of B220 ϩ cells (reflective of B lineage) in bone marrow and spleen of wild type and Tyk2 Ϫ/Ϫ mice before and after LCMV infection (Fig. 5). The number of B220 ϩ cells was not altered in the spleens of LCMV-infected Tyk2 Ϫ/Ϫ mice, whereas wild type-infected mice showed a 40% decrease in this population of cells. Similar results were seen in the bone marrow, where there was a 25% loss of B220 ϩ cells in infected Tyk2 Ϫ/Ϫ mice and a 66% loss of cells in wild type infected mice. The fact that there is a decrease in B220 ϩ cells in the bone marrow of Tyk2 Ϫ/Ϫ -infected mice may be a reflection of the antiproliferative effects of type 1 IFNs (see Table 2) on actively dividing cells, which would not be present in the spleen. The decrease in B220 ϩ cells as well as nucleated cells are consistent with the previously published decreases in LCMV-infected wild type mice (1). The fact that Tyk2 Ϫ/Ϫ mice show a significant resistance to LCMV-induced lymphoid hypoplasia, particularly in B lineage cells, is consistent with our in vitro data using primary IL-7-dependent B cells.

DISCUSSION
The mechanisms by which type 1 IFNs inhibit cell growth are variable and complex. Growth inhibition may or may not result in cell death, and many of the reported studies used transformed cells from different origins, which may account for these variable effects. In this report, we have examined IFN␤-induced apoptosis of bone marrow-derived   murine pro-B cells, one of the few types of primary cells known to undergo PCD when only exposed to type 1 IFNs (16). Our results indicate that treatment of wild type pro-B cells with IFN␤ or IFN␣ inhibits their growth. However, only IFN␤ stimulates these cells to undergo apoptosis. Selective actions of IFN␤ and IFN␣ have been previously observed (2). However, the presumed signaling events that are differentially regulated by IFN␤ compared with IFN␣ that may allow for the selective actions of these two related cytokines is unclear.
The primary findings of this report indicate that IFN␤-stimulated tyrosine phosphorylation of Stat3 together with the expression of Tyk2 are essential signaling components required for IFN␤-induced apoptosis of pro-B cells. Although active Stat3 has been generally regarded as contributing to oncogenesis by promoting cell proliferation, there are a number of circumstances where this transcription factor appears to inhibit cell growth and promote differentiation (29 -31). The mechanisms by which activated Stat3 can function as either a promoter or an inhibitor of cell growth are not well understood but may be related to the context of whether activated Stat3 exerts its effects on primary or transformed cells. The physiological role of Stat3 activation by a variety of cytokines, including IL-6, IL-10, and granulocyte colony-stimulating factor, is also well established, but the biological effects mediated by type 1 IFN activation of this transcription factor have been unclear. Our findings are consistent with the observation that in wild type cells, expression of dominant negative Stat3 or treatment of cells with a Stat3 inhibitor blocks IFN␤-induced apoptosis. Furthermore, Tyk2 Ϫ/Ϫ cells that express constitutively active Stat3 display IFN␤-stimulated PCD (Fig. 4A). Interestingly, tyrosine phosphorylation of Stat3 without other signals generated from IFN␤ binding to its receptor is not sufficient to drive apoptosis in wild type or Tyk2 Ϫ/Ϫ cells (Figs. 3 and 4A and Table  4). In Tyk2 Ϫ/Ϫ cells that express CA-Stat3, IFN␤-stimulated tyrosine phosphorylation of this protein is observed, but it is much less pronounced than when CA-Stat3 is expressed in wild type cells (Fig. 4B). These findings suggest that a dimerized form of Stat3 can be tyrosinephosphorylated, albeit less efficiently by another kinase (possibly Jak1), in the absence of Tyk2. The observation that CA-Stat3 is tyrosine-phosphorylated in untreated wild type or Tyk2 Ϫ/Ϫ pro-B cells suggests that other kinases may be responsible for basal phosphorylation of the protein. Other Jaks or members of the Src family are possible candidates, since they have been reported to tyrosine-phosphorylate Stat3 (32). We also observed that IFN␤-stimulated apoptosis of Tyk2 Ϫ/Ϫ cells that expressed CA-Stat3 displayed a delayed onset of PCD compared with wild type cells incubated with IFN␤ (Fig. 4A). This observation suggests that another signal(s) that requires Tyk2 expression, in addition to tyrosine phosphorylation of Stat3, is required for optimal IFN␤-stimulated PCD.
Other signals that work in conjunction with tyrosine-phosphorylated Stat3 to drive the apoptotic response by IFN␤ do not include tyrosinephosphorylated Stat1 or Stat5a,b, since Stat1 Ϫ/Ϫ cells or Stat1,5a,b Ϫ/Ϫ cells show the same sensitivity to IFN␤-induced PCD as wild type cells (Table 1). Furthermore, IFN␤ treatment of Tyk2 Ϫ/Ϫ pro-B cells induces activation of Stat1 and Stat2, suggesting that those IFN␤-activated early response genes regulated by Stat1 and Stat2 are not sufficient to induce PCD. Taken together, it appears that IFN␤ activation of Stat3 is necessary but not sufficient for optimal IFN␤-induced PCD, and it is likely that there is another set of events involved in the apoptotic response. It remains to be determined what Stat3-dependent cellular genes are regulated by exposure of pro-B cells to IFN␤ and how regulation of such genes impinge on the apoptotic response.
We have extended our results that demonstrate a role for Tyk2 in IFN␤-stimulated apoptosis of pro-B cells to an in vivo model where mice were infected with high concentrations of LCMV. In this model, LCMV induces a transient pancytopenia. Mice that lack expression of the IFNaR1 subunit of the type 1 receptor are resistant to LCMV-induced pancytopenia, suggesting a role for type 1 IFNs in the actions of LCMV (1). Compared with wild type littermates, in Tyk2 Ϫ/Ϫ mice, there was a higher number of nucleated and B220 ϩ cells in bone marrow and spleen following LCMV infection (Tables 6 and 7 and Fig. 5). We have not proven that the pancytopenia observed in LCMV-infected mice is directly related to type 1 IFN activation of Stat3-or IFN␤stimulated apoptosis. However, there was a decrease in the B220ϩ B cell population in both bone marrow and spleen in Tyk2-null mice infected with LCMV, but this decrease in cell number was less than that observed in wild type mice (see Fig. 5). This is consistent with the in vitro finding that there is an antiproliferative effect of type 1 IFNs in Tyk2 Ϫ/Ϫ pro-B cells, but it is less pronounced than in wild type cells, since these cells also undergo PCD (see Table 2). Now that we have established a role of both Tyk2 and Stat3 in IFN␤induced PCD of primary B cells in vitro and the role of Tyk2 in vivo, we can identify the downstream targets of Tyk2 and Stat3. Preliminary results indicate that the apoptotic actions of IFN␤ in pro-B cells are mediated by a mitochondria-dependent, caspase-independent pathway (data not shown). Proteins such as apoptosis-inducing factor or HtAr2, which can induce apoptosis in the absence of caspase activation through mitochondria-dependent events, are two such proteins that need to be evaluated to see if they contribute to the apoptotic actions of IFN␤. We are also determining whether there are changes in a variety of mitochondrial functions that are regulated by Tyk2 activation of Stat3. Such information will provide us with targets for the apoptotic actions of IFN␤.