Increased Expression of Genes Linked to FcεRI Signaling and to Cytokine and Chemokine Production in Lyn-Deficient Mast Cells1

Cross-linking the high-affinity IgE receptor, FcεRI, on mast cells activates signaling pathways leading to the release of preformed inflammatory mediators and the production of cytokines and chemokines associated with allergic disorders. Bone marrow-derived mast cells (BMMCs) from Lyn-deficient (Lyn−/−) mice are hyperresponsive to FcεRI cross-linking with multivalent Ag. Previous studies linked the hyperresponsive phenotype in part to increased Fyn kinase activity and reduced SHIP phosphatase activity in the Lyn−/− BMMCs in comparison with wild-type (WT) cells. In this study, we compared gene expression profiles between resting and Ag-activated WT and Lyn−/− BMMCs to identify other factors that may contribute to the hyperresponsiveness of the Lyn−/− cells. Among genes implicated in the positive regulation of FcεRI signaling, mRNA for the tyrosine kinase, Fyn, and for several proteins contributing to calcium regulation are more up-regulated following Ag stimulation in Lyn−/− BMMCs than in WT BMMCs. Conversely, mRNA for the low-affinity IgG receptor (FcγRIIB), implicated in negative regulation of FcεRI-mediated signaling, is more down-regulated in Ag-stimulated Lyn−/− BMMCs than in WT BMMCs. Genes coding for proinflammatory cytokines and chemokines (IL-4, IL-6, IL-13, CSF, CCL1, CCL3, CCL5, CCL7, CCL9, and MIP1β)are all more highly expressed in Ag-stimulated Lyn−/− mast cells than in WT cells. These microarray data identify Lyn as a negative regulator in Ag-stimulated BMMCs of the expression of genes linked to FcεRI signaling and also to the response pathways that lead to allergy and asthma.

10% FCS (HyClone), 2 mM L-glutamine, 0.1 mM nonessential amino acids, 1 mM sodium pyruvate, 100 U/ml penicillin, 100 g/ml streptomycin, 55 M 2-ME, and 1 mM HEPES (complete RPMI medium) and 30% WEHI-3-conditioned medium. Culture reagents were from Invitrogen Life Technologies. Mast cell morphology, granularity, and differentiation were analyzed by toluidine blue-stained cytospin preparations and flow cytometry as described previously (6). By 6 wk, WT and Lyn Ϫ/Ϫ mast cells expressed similar levels of Fc⑀RI and c-kit at their surfaces and were morphologically similar with respect to granule content. Microarray experiments were carried out on 6-wk-old mast cells. For stimulation, WT and Lyn Ϫ/Ϫ BMMCs were sensitized with 1 g/ml anti-DNP IgE overnight (12 h) in complete RPMI medium without WEHI-3 medium. BMMCs were harvested, washed, and stained with trypan blue to verify viability. Viability was Ͼ95% for all experimental conditions. BMMCs were resuspended in complete RPMI medium and activated by the addition of 10 ng/ml DNP-BSA for 2 or 4 h at 37°C.

Isolation and labeling of mast cell RNA
Total RNA was prepared from resting and activated WT and Lyn Ϫ/Ϫ BMMCs using TRIzol reagent (Invitrogen Life Technologies) according to the manufacturer's instructions. Further purification of RNA was done using the RNeasy mini kit clean-up protocol (Qiagen). A total of 10 g of total RNA was converted to cDNA using the Superscript II RNA reverse transcriptase (Invitrogen Life Technologies). Double-strand cDNA was produced with an oligo(dT) primer containing the T7 RNA polymerase site at the 5Ј end. The cDNA was labeled with biotinylated nucleotides directly with the ENZO in vitro labeling kit (Affymetrix) to produce antisense RNA. Labeled cRNA was fragmented and purity was checked using Agilent RNA chips (Agilent Technologies). A total of 15 g of labeled cRNA was hybridized for 16 h to U74A murine Genechips (Affymetrix) containing 12,488 genes and expressed sequence tags (ESTs). A mixture of bacterial RNA (BioB, bioC, bioD, cre) with known concentrations (1.5, 5, 25, and 100 pM, respectively) was added to each chip hybridization mixture. Chips were washed, stained with PE-streptavidin and read with an HP Gene Array Scanner according to the manufacturer's instructions.

DNA microarray analysis
All Affymetrix chips were scaled in the Affymetrix MAS 5.0 software to a target fluorescence of 500 and had scaling factors ranging from 2 to 15, indicating good probe preparation and hybridization. Filtering and statistical analyses of microarray data were performed using Genespring software version 6.0 (Silicon Genetics). For all statistical analyses, the experiment interpretation mode was set to the log of ratios and data were normalized per gene to the median of the appropriate unstimulated duplicate samples. For analysis of differential gene expression between stimulated WT and Lyn Ϫ/Ϫ BMMCs, data were normalized so that the baseline mRNA expression corresponded to the median of the stimulated WT BMMCs. To evaluate changes in gene expression, data were filtered in Genespring for genes expressing a 3-fold up or down change in expression when compared to the control samples. A Flag filter was applied to exclude those data in which a gene's expression was absent or marginally present in at least half of the samples based on the Affymetrix algorithm. Finally, the one-way ANOVA filter (p Ͻ 0.05) was used to identify statistically significant differences in gene expression between resting and stimulated samples. For generation of heat maps, lists of regulated genes were clustered with the gene tree method by measuring similarity with smooth correlation. Statistical analyses were performed with two-way ANOVA to identify strain and/or treatment interactions for all eight samples. Statistical analyses were performed with two-way ANOVA to determine significant differences due to the kinase status (ϮLyn), treatment (ϮAg), and/or to the combination of kinase status and treatment for all eight samples. Variance was computed by applying the Cross-Gene Error Model when generating lists of regulated genes. The Benjamini-Hochberg false discovery rate was applied to the two-way ANOVA model as a multiple testing correction to control for occurrences of false positives that arise from performing multiple tests. The numbers of genes with a Benjamini-Hochberg adjusted p value Ͻ0.05 are listed in the Venn diagram.

Quantitative real-time PCR
RNA isolation and first double-strand cDNA synthesis were performed as described above. IL-13, Fyn, and actin primers and probes were designed using Primer Express Software according to guidelines recommended by Applied Biosystems for TaqMan Gold RT-PCR. Primers were used to amplify 100 ng of template cDNA (triplicate samples each of WT Ϯ activation and Lyn Ϫ/Ϫ Ϯ activation) and produced single products when analyzed on 0.8% Tris acetate EDTA (TAE) agarose gels. PCR conditions used to amplify template DNA were as follows: 30 cycles of each 95°C for 48 s, 58°C for 45 s, 73°C for 3 min followed by a final 10-min extension at 72°C using a standard 50 l PCR. The PCR amplification product obtained from each set of primers was purified from the agarose gel using the Qiaex II gel extraction kit (Qiagen). Fragment concentration was determined by measuring A260 absorbance of purified PCR fragments. These PCR fragments were then used to construct standard curves for real-time PCRs. Resultant cDNA used for microarray hybridization experiments served as a template for real-time quantitative PCR and was performed on an ABI 7700 Sequence Detection system (Applied Biosystems) in which a fluorescent reporter dye (FAM) was released and quantified during each specific replication of the template. Each PCR (25 l) contained 100 ng of cDNA and was mixed with 2ϫ TaqMan Universal PCR Master Mix (Applied Biosystems), gene-specific forward and reverse primers (15 M), and dyelabeled oligonucleotide probes (2 M). GAPDH primers and probes were from Applied Biosystems and were provided by Dr. C. Schwab (University of New Mexico, Albuquerque, NM). The forward and reverse primers used here were: for IL-13, 5Ј-CGCAAGGCCCCCACTAC-3Ј and 5Ј-AGTTT TGTTATAAAGTGGGCTACTTCGA-3Ј; for Fyn, 5Ј-GGTTACATTCC CAGCAATTACGT-3Ј and 5Ј-TGCGGCCAAGTTTTCCA-3Ј; for ␤-actin, 5Ј-AGAGGGAAATCGTGCGTGAC-3Ј and 5Ј-CAATAGTGATGA CCTGGCCGT-3Ј. The fluorogenic probes used for hybridization were as follows: for IL-13, 5Ј-(6FAM)-TCTCCAGCCTCCCCGATACCA-(carboxytetramethylrhodamine), [TAMRA]-3Ј; for Fyn, 5Ј-(6FAM)-TCCAG TTGACTCCATCCAGGCAGAAGA (TAMRA)-3Ј; for ␤-actin, 5Ј-(6FAM)-CACTGCCGCATCCTCTTCCTCCC-(TAMRA)-3Ј. PCR parameters were as recommended for the TaqMan Universal PCR master mix kit: 50°C for 2 min, 95°C for 10 min, followed by 40 cycles of 95°C for 15 s and 60°C for 1 min. Triplicate samples of 2-fold serial dilutions (ranging from 1 ϫ 10 4 molecules up to 2 ϫ 10 6 molecules) of IL-13, Fyn, actin, or GAPDH fragments were assayed and used to construct the standard curves. Data were analyzed with ABI Prism 7000 SDS software version 1.0 and were normalized to the average copy number of GAPDH. Data are presented as fold induction with respect to unstimulated WT samples. Experiments were performed twice with triplicate samples (cDNA isolated from three separate cultures of BMMCs). Statistical analysis was performed with the Student unpaired t test using GraphPad Prism software.

Cytokine ELISA
Supernatants were collected from resting and activated WT and Lyn Ϫ/Ϫ BMMCs and analyzed for IL-4 or IL-13 using a two-site sandwich ELISA as described previously (15). Abs specific for IL-4 (11B11, and biotin-BVD6-24G2) were from BD Pharmingen; Abs specific for IL-13 (38213.11, and biotin-BAF413) were from R&D Systems. Capture mAbs were bound to ELISA plates diluted in 0.1 M Na 2 HPO 4 overnight at 4°C. Plates were washed, blocked with 1% BSA in PBS, and incubated overnight at 4°C with samples. Bound cytokines were detected by the addition of biotinylated mAbs followed by streptavidin-HRP (0.625 g/ml final concentration for IL-13; 2.5 g/ml final concentration for IL-4) and colorimetric substrate (ABTS for IL-4; tetramethylbenzidine (TMB; Sigma-Aldrich), for IL-13). OD 405 was determined and cytokines were quantified by comparison to standard curves generated using rIL-4 (BD Pharmingen) and IL-13 (R&D Systems). Supernatants were analyzed with commercial ELISA kits for IL-6 (BD Biosciences) and for IL-2 and TNF-␣ (eBioscience) according to the manufacturer's instructions. Detection limits for each cytokine assay were assigned as the lowest concentration in the linear portion of the standard curve. Measurements were made in duplicate using cells from three independent cultures of BMMCs.  Table I summarize the principal differences in gene expression between WT and Lyn Ϫ/Ϫ BMMCs under resting conditions. The data are strikingly similar between the duplicate samples, indicating that the technical procedures from RNA preparation to data acquisition and analysis are highly reproducible. Expression levels for the mRNAs encoding the receptors CXCR4 and insulin-like growth factor 2 receptor (IGF2R) are higher in Lyn Ϫ/Ϫ BMMCs than in WT BMMCs. In addition, annexin A1 that inhibits PLA 2 activity in vitro and phospholipase A2 group VII are expressed at higher levels in Lyn Ϫ/Ϫ BMMCs. Levels of several other mRNAs are consistently lower in Lyn Ϫ/Ϫ BMMCs compared with WT BMMCs; in general, the down-regulated genes lack known roles in signal transduction pathways.

Lyn deficiency alters the profile of Ag-induced gene expression
To identify other differentially transcribed genes between WT and Lyn Ϫ/Ϫ BMMCs, we compared the gene expression profiles of eight separate samples representing two separate experiments for each of four conditions: resting and Ag-stimulated from both WT and Lyn Ϫ/Ϫ cells. Two-way ANOVA was used to identify genes whose expression levels were significantly altered based on kinase status (ϮLyn), treatment (ϮAg), or on the interaction of both parameters (kinase status and treatment). Through this analysis, we identified 501 gene products with a Benjamini-Hochberg adjusted p value Ͻ0.05, indicating their expression is statistically different due to the absence of Lyn, the presence of Ag, or the interaction of both parameters. Fig. 2 provides two alternative graphical representations of these results.
The data on 501 gene products are shown in Fig. 2A as a heat map that compares all eight samples, normalized to the median of the two unstimulated WT samples. The striking similarity between the duplicate samples is again evident. Gene expression is moderately altered between unstimulated samples and is greatly altered in Ag-stimulated samples, with almost as many genes showing decreased expression (blue) as increased expression (red).
The Venn diagram (Fig. 2B) displays the numbers of these gene products whose p values are influenced by each parameter independently (kinase status or treatment) and also by the effect that each parameter has on the other. From a total of 145 genes that display a kinase effect, 27 are significant due to the kinase effect only while 2 genes displayed a kinase effect and an interaction effect (Fig. 2B, left circle). Another 470 genes displayed a treatment effect (Fig. 2B, right circle). Of those, 85 genes are affected by both kinase status and treatment but each factor affects gene expression independently of one another. As shown in Fig. 2B, lower circle, there are 31 total genes that have p values Ͻ0.05 for each of the tests: kinase effect, interaction effect, and treatment effect. In addition, 14 genes are significant due to the treatment effect and interaction effect but not the kinase effect. Two genes that are not significant in either the kinase test or treatment test alone had a p value Ͻ0.05 for the interaction effect. Table II organizes a subset of these 501 gene products into functional categories and shows their fold change in expression after Fc⑀RI cross-linking in both WT and Lyn Ϫ/Ϫ cells. A complete list of these genes is available at ͗www.cellpath.unm.edu͘.
Among genes involved in tyrosine kinase-coupled signaling, levels of Fyn mRNA are increased in both WT and Lyn Ϫ/Ϫ BMMCs after Fc⑀RI cross-linking, with the increase being greater  The absence of Lyn alters the gene expression profile of resting mast cells. Two independent cultures of 6-wk-old WT and Lyn Ϫ/Ϫ BMMCs were incubated in complete RPMI without growth factors for 16 h. Cells were harvested, their viability was confirmed by trypan blue exclusion, and total mRNA was isolated. Biotinylated antisense cRNA was hybridized to Affymetrix U74A DNA chips for 16 h. DNA chips were washed and scanned with an Affymetrix scanner. Data were normalized so that the level of baseline mRNA expression corresponded to the median of unstimulated WT BMMCs. The heat map represents 14 genes that were upor down-regulated by a factor of 3 or greater and was generated using a stringent filter scheme that incorporated the ANOVA filter. Each row corresponds to a single gene and each column represents an independent condition. Location of the gene and common name are provided. Changes in gene expression correspond to the color scale shown.
in Lyn Ϫ/Ϫ cells (Table II). In contrast, mRNA expression for the low-affinity IgG receptor Fc␥RIIB is significantly reduced after Ag stimulation in both WT cells and Lyn Ϫ/Ϫ cells, with the extent of down-regulation being greater in the Lyn Ϫ/Ϫ BMMCs. Recent studies implicate Fyn in Lyn-independent signaling in activated mast cells (2) and Fc␥RIIB in the negative regulation of Fc⑀RI signaling (16,17). Thus, both observations may be relevant to the hyperresponsiveness of Lyn Ϫ/Ϫ BMMCs. Several genes implicated in calcium regulation, including the type I inositol-1,4,5trisphosphate receptor (Ins(1,4,5)P 3 receptor 1), sarco(endo)plasmic reticulum calcium ATPase 2 (SERCA2), and sphingosine kinase-1, are also significantly up-regulated in Ag-stimulated Lyn Ϫ/Ϫ BMMCs compared with control cells and could contribute to cellular hyperresponsiveness.
A consistent trend of higher up-regulation in activated Lyn Ϫ/Ϫ mast cells extends to other genes that may contribute to signaling pathway activities. These include genes involved in G proteincoupled signal transduction and cellular retinoic acid-binding protein II (CRABP2) (Table II). Additionally, expression of mRNA for at least one transcription factor, the NF-ATc isoform, is upregulated in Lyn Ϫ/Ϫ mast cells. In contrast, phospholipid scramblase 2 (PLSCR2) mRNA is only up-regulated 2.1-fold in Lyn Ϫ/Ϫ mast cells compared to 17.1-fold in WT cells.
The most striking differences noted between the two cell types is in the expression of cytokine and chemokine genes known to be up-regulated upon Fc⑀RI aggregation in mast cells and basophils. In general, cytokine and chemokine gene expression is the same in WT and Lyn Ϫ/Ϫ BMMCs under resting conditions. As an exception, the expression of CXCR4 is increased in unstimulated Lyn Ϫ/Ϫ mast cells (Table I). Following Fc⑀RI cross-linking (Table  II)
Previous studies suggest that both IL-2 and TNF-␣ production are increased in Ag-stimulated Lyn Ϫ/Ϫ BMMCs (5). Both Agstimulated WT and Lyn Ϫ/Ϫ cells secrete very little IL-2, however, we note a slight increase in IL-2 production in Lyn Ϫ/Ϫ BMMCs after 4 h of Ag stimulation compared with WT mast cells (Fig.  4D). We confirmed that Lyn Ϫ/Ϫ mast cells secrete more TNF-␣ at both 2 and 4 h of activation than WT cells (Fig. 4E).

Discussion
The cross-linking of IgE-Fc⑀RI complexes with multivalent Ag activates mast cell signaling pathways leading to the release of a wide range of proinflammatory cytokines, chemokines, and other FIGURE 2. Gene expression profiles in resting and Ag-stimulated WT and Lyn Ϫ/Ϫ mast cells. Total RNA was isolated from two independent cultures of 6-wk-old WT and Lyn Ϫ/Ϫ BMMCs without (Ϫ) and with (ϩ) stimulation with 10 ng/ml DNP-BSA for 4 h. Labeled antisense cRNA was hybridized to DNA chips and read with an Affymetrix scanner. Data were normalized to the median baseline mRNA expression of unstimulated WT BMMCs and filtered as described in Materials and Methods. The resulting gene list was analyzed by twoway ANOVA and resulted in 501 statistically significant regulated genes that are depicted in the heat map (A). Changes in gene expression correspond to the color scale shown. The Venn diagram (B) is an alternative graphical representation of the same data. There are 501 genes that received a Benjamini-Hochberg adjusted p value Ͻ0.05 in the two-way ANOVA model. The Venn diagram shows the number of genes that had an adjusted p value Ͻ 0.05 for the kinase effect, treatment effect, and the interaction effect between the two groups. mediators associated with allergic responses. Activated mast cells release IL-4, IL-13, and other Th2 cytokines that enhance the IgE response by stimulating the differentiation of Th2 lymphocytes and promoting class switching to IgE in B cells, leading to increased production of IgE and resensitization of Fc⑀RI on mast cells (reviewed in Ref. 18).
Previously, we showed that BMMCs from Lyn Ϫ/Ϫ mice divide faster than WT cells in response to growth-promoting cytokines (IL-3 and stem cell factor) and undergo less apoptosis when cytokine is withdrawn (6). We and others (2,5,7,8,19) have also shown that Lyn Ϫ/Ϫ BMMCs are slow to initiate responses to Fc⑀RI cross-linking but are deficient in the termination of signaling, resulting in prolonged biochemical responses (receptor phosphorylation, the activation of AKT, phospholipase C␥, ERK, and other signaling enzymes) and exaggerated physiological responses (calcium mobilization, secretion, cytokine production, and integrin activation). Resolving previous controversy (reviewed in Refs. 6 and 7), these results now provide a consensus view of Lyn as both a kinetic accelerator and negative regulator of Fc⑀RI signaling in mast cells. Previous work has also provided partial insight into the mechanism of the hyperresponsiveness of Lyn Ϫ/Ϫ BMMCs. Specifically, in-creased basal and Ag-induced Fyn activity is thought to contribute to signal initiation in Lyn-deficient cells (7,8), while the loss of Ag-induced SHIP activation provides at least a partial explanation for the failure of signal termination in Lyn Ϫ/Ϫ cells (7).
Previous groups have used microarray analyses to explore transcriptional profiles induced by Fc⑀RI cross-linking in normal and secretion-impaired rodent and human mast cells and basophils (20 -26). Here, microarray analysis was used to discover new properties of both unstimulated and Ag-stimulated Lyn-deficient BMMCs that might provide further insight into their hyperresponsive phenotype.
Relatively few genes showed strongly different expression levels between unstimulated WT and Lyn Ϫ/Ϫ BMMCs and none could be clearly linked to the enhanced early responses (calcium mobilization, integrin activation, degranulation, and others) to Fc⑀RI cross-linking. The chemokine receptor, CXCR4, is a possible exception. CXCR4 was strongly up-regulated in Lyn Ϫ/Ϫ BMMCs. Ligation of G protein-coupled receptors can prime mast cells for increased Fc⑀RI-mediated degranulation (27).
In contrast, expression of 501 genes was increased or decreased at least 3-fold after 4 h of Fc⑀RI cross-linking. Of these a Data were normalized to the median baseline mRNA expression of either unstimulated WT mast cells or unstimulated Lyn Ϫ/Ϫ cells. The genes listed were selected from a list of 501 genes that resulted from analysis by two-way ANOVA.
genes, more were up-regulated than down-regulated. In general, the extent of up-regulation was greater in Lyn Ϫ/Ϫ than in WT BMMCs. Thus, there is strong potential for a transcriptional component to the enhanced late responses (cytokine and chemokine production and others) to Fc⑀RI cross-linking in the Lyn-deficient cells.
Among genes for signaling proteins, Fc⑀RI cross-linking induced a greater up-regulation of the tyrosine kinase, Fyn, and of a series of genes implicated in calcium regulation in Lyn-deficient than in WT BMMCs. In combination with earlier evidence for increased Fyn activity, perhaps linked in part to delayed activation of Csk-binding protein (Cbp), in Lyn Ϫ/Ϫ BMMCs (7,8), these data support the hypothesis that increased signaling through the recently discovered Fyn-mediated pathway can contribute to the persistent hyperresponsiveness of Lyn Ϫ/Ϫ BMMCs. Conversely, Fc⑀RI cross-linking induced a greater down-regulation of mRNA for the IgG receptor, Fc␥RIIB, in Lyn-deficient than in WT BMMCs. Fc␥RIIB is well-recognized as a negative regulator of Fc⑀RI signaling when the two receptors are co-crosslinked. The mechanism involves the recruitment of the inositol phosphatase, SHIP, to the membrane via its association with phosphorylated ITIMs present in Fc␥RIIB (16,17). Recent evidence that SHIP-deficient BMMCs degranulate spontaneously and are hyperresponsive to Fc⑀RI cross-linking suggests that SHIP also plays a constitutive role in the down-regulation of signaling (28). In this case, reduced levels of Fc␥RIIB in Lyn-deficient BMMCs may help to maintain the reduced levels and activity of membraneassociated SHIP and the elevated levels of membrane phosphatidylinositol 3,4,5-trisphosphate (PI (3,4,5)P 3 ) characteristic of these cells (7).
Among the cytokines expressed after 4 h of Ag stimulation in BMMCs, our data show that levels of mRNA and protein for IL-4, IL-6, and for IL-13 are all significantly higher in Lyn Ϫ/Ϫ than in WT BMMCs. Both IL-4 and IL-13 proteins are elevated in the lungs of asthmatic patients, and are thought to be central regulators of this disease. In mice, recent studies suggest that IL-13 may be more directly involved in mediating allergic responses than IL-4 (29 -32). Thus, Ag-exposed Lyn-deficient BMMCs clearly develop a cytokine profile consistent with a predisposition to allergy and asthma.
We failed to see greater increases in TNF-␣ and IL-2 mRNA levels in Lyn Ϫ/Ϫ cells than in WT mast cells. However, ELISAs showed that Lyn Ϫ/Ϫ BMMCs produce higher amounts of TNF-␣ and IL-2 protein than WT cells. These results are consistent with previously published data (5). The discrepancies between mRNA levels and TNF-␣ and IL-2 protein production may be attributed to mRNA instability. We also did not observe robust production of IL-2 in either WT or Lyn Ϫ/Ϫ BMMCs as was reported by Kawakami et al. (5). These different results very likely reflect differences in the time that mast cells were stimulated: our measurements were made after 4 h of activation, while the previous group made measurements after 20 h. Likewise, Nishizumi and Yamamoto (19) reported that Lyn deficiency does not affect the production of cytokines (IL-4, IL-5, IL-6, TNF-␣, TNF-␤) when BMMCs are stimulated with Ag for 2.5 h. We, too, found rather little differences in IL-4 or IL-6 production when WT and Lyn Ϫ/Ϫ BMMCs were stimulated for 2 h, even though the differences were substantial after 4 h (Fig. 4).
Among the chemokines, mRNAs coding for CCL1, CCL3 (MIP1␣), CCL4 (MIP1␤), CCL5 (RANTES), CCL7 (MCP-3), and CCL9 (MIP1␥) are all up-regulated in Ag-stimulated Lyn Ϫ/Ϫ BMMCs compared with WT BMMCs. The greater induction of both RANTES and its receptor CCR1 may contribute to the hyperresponsiveness of Lyn Ϫ/Ϫ BMMCs via a potential autocrine signaling mechanism. Previous studies have demonstrated a central role for chemokines in mediating multiple aspects of the asthmatic response. Chemokines induce B cell Ab class switching (33). In addition, IL-13 is a potent inducer of chemokines (eotaxins) in airway epithelial cells (31,34) and current models suggest that coordinated interactions between IL-13 and chemokines are importantly involved in the pathogenesis of asthma (35). Increased expression of cytokine and chemokine mRNA is likely the consequence of increased activity of transcription factors. Our data show that mRNA coding for at least one transcription factor, the cytoplasmic NF-AT (NF-ATC, also known as NFATC1 and NFAT2) is induced 3-fold more in Lyn Ϫ/Ϫ BMMCs than in WT BMMCs (Table II). In addition, Ag-stimulated Lyn Ϫ/Ϫ BMMCs express slightly higher levels of NF-B (Lyn Ϫ/Ϫ , 1.6fold; WT, 0.8-fold), and slightly lower levels of c-Jun/activator protein-1 (AP-1) (Lyn Ϫ/Ϫ , 2-fold; WT, 4-fold). Studies of the phosphorylation and/or activation of these transcription factors are needed to know if increased levels are linked to increased activities of transcription factors.
The linkage between increased Fc⑀RI signaling and increased gene transcription in Lyn Ϫ/Ϫ BMMCs is not known with certainty. However, we note that biochemical studies published previously linked reduced SHIP activation (discussed above) to increases in AKT and Ras/MAPK pathway activities (7,28). Ag-stimulated SHIP Ϫ/Ϫ BMMCs also produce increased levels of various proinflammatory cytokines including IL-4, IL-5, IL-6, IL-13, and TNF-␣ (36). Furthermore, SHIP negatively regulates IL-6 production in mast cells by inhibiting NF-B activity (36). In turn, Kitaura et al. (37), linked increased AKT activity to increased cytokine (IL-2 and TNF-␣) gene promoter activities via the activation of NF-B, NF-AT, and AP-1 in Lyn Ϫ/Ϫ BMMCs. Additionally, Monticelli et al. (38) recently described roles for NF-AT proteins in the regulation of IL-13 gene transcription in BMMCs. Thus, the increased levels of mRNA for multiple cytokines and chemokines in Lyn Ϫ/Ϫ BMMCs may result directly from their increased capacity to signal from AKT to the activation of transcription factors.
The 33-fold up-regulation of sphingosine kinase 1 in Ag-stimulated Lyn Ϫ/Ϫ BMMCs may also contribute to increased chemokine production. Sphingosine kinase is activated upon Fc⑀RI crosslinking in RBL-2H3 cells and mast cells and is linked to calcium mobilization (39 -41). Additionally, the product of sphingosine kinase, sphingosine-1-phosphate (S1P), acts as a ligand for G-protein coupled chemokine receptors (42). S1P levels are increased in bronchoalveolar lavage fluid from lungs of asthmatics after challenge with allergen (43) and can lead to a heightened production of chemokines in RBL-2H3 mast cells (44).
Our results in Lyn Ϫ/Ϫ BMMCs differ from results obtained in a recent study of gene expression in Lyn-deficient DT40 chicken B cells, where the absence of Lyn led to down-regulation of numerous genes encoding proteins involved in BCR signaling, proliferation, control of transcription, immunity/inflammation response, and cytoskeletal organization (45). One major difference between Lyn-deficient DT40 cells and BMMCs is that the Lyn Ϫ/Ϫ DT40 cells have no other Src kinase family members, whereas the Src kinase, Fyn, increases in both activity (7) and transcript levels (Fig. 3, Table II) in Ag-stimulated Lyn Ϫ/Ϫ BMMCs.
In conclusion, the Fc⑀RI-mediated activation of Lyn Ϫ/Ϫ BMMCs results in greater increases in mRNAs encoding proteins in the Fc⑀RI signaling pathway, greater decreases in mR-NAs encoding a negative regulator of signaling and greater increases in mRNAs encoding Th2 cytokines and chemokines and key transcription factors than occur in control cells. All of these differences are likely to contribute to the hyperresponsiveness of Lyn Ϫ/Ϫ mast cells and to the greater predisposition of Lyn Ϫ/Ϫ mice to the allergic phenotype as indicated by the higher numbers of skin and peritoneal mast cells, higher serum IgE levels, increased levels of circulating histamine, and increased in vivo expression of surface Fc⑀RI on the mast cells of Lyn Ϫ/Ϫ mice in comparison with WT littermates (6,9,10,12,19). Recently, Beavit et al. (46) confirmed that Lyn Ϫ/Ϫ mice develop severe, persistent asthma suggesting a role for Lyn as an important negative regulator of Th2 immune responses. Overall, our analysis suggests a key role for Lyn in setting the thresholds for mast cell signaling and response pathways, thus determining predisposition to allergic responses.