The Novel α4B Murine α4 Integrin Protein Splicing Variant Inhibits α4 Protein-dependent Cell Adhesion*

Background: α4 Integrin participates in tumor metastasis and autoimmune diseases. Results: α4B Markedly inhibited α4 integrin-dependent cell adhesion through the cytoplasmic region of α4B. Conclusion: α4B Is a novel endogenous inhibitor of α4 integrin. Significance: This study provides a possible role of α4B as an inhibitor of α4 integrin-mediated metastasis. Integrins affect the motility of multiple cell types to control cell survival, growth, or differentiation, which are mediated by cell-cell and cell-extracellular matrix interactions. We reported previously that the α9 integrin splicing variant, SFα9, promotes WT α9 integrin-dependent adhesion. In this study, we introduced a new murine α4 integrin splicing variant, α4B, which has a novel short cytoplasmic tail. In inflamed tissues, the expression of α4B, as well as WT α4 integrin, was up-regulated. Cells expressing α4B specifically bound to VCAM-1 but not other α4 integrin ligands, such as fibronectin CS1 or osteopontin. The binding of cells expressing WT α4 integrin to α4 integrin ligands is inhibited by coexpression of α4B. Knockdown of α4B in metastatic melanoma cell lines results in a significant increase in lung metastasis. Expression levels of WT α4 integrin are unaltered by α4B, with α4B acting as a regulatory subunit for WT α4 integrin by a dominant-negative effect or inhibiting α4 integrin activation.

Integrin adhesion receptors are large heterodimeric cell surface receptors that mediate the adhesion of cells to the extracellular matrix and other cells (1). They participate in embryonic development and in the maintenance of homeostatic balance (2). They also function in a range of pathological processes, including wound repair, inflammation, leukocyte trafficking, and tumor metastasis (3). These integrin adhesion receptors comprise large extracellular, transmembrane, and small cytoplasmic domains. The extracellular domain is responsible for ligand binding, whereas the cytoplasmic domains play a crucial role in the attachment of cells to extracellular matrix ligands (4). The activation of integrins is accompanied by a conforma-tional shift from a low to a high binding affinity state because of ligand stimulation (5).
The ␣4 integrin (CD49d) is expressed on several types of leukocytes and tumor cells and is involved in autoimmune diseases and tumor metastasis by mediating cell attachment to vascular cell adhesion molecule 1 (VCAM-1), 4 the CS1 domain within alternative splicing forms of fibronectin, the propolypeptide of von Willebrand factor (pp-vWF), and osteopontin (OPN) (6 -9). The ␣4 integrin on CD4 ϩ T cells is required for the development of experimental autoimmune encephalomyelitis (EAE) (10). On tumor cells, ␣4 integrin promotes dissemination to distal organs by increased adhesion to the vascular endothelium and by facilitating the extravasation of tumor cells (6). In vivo blocking studies have demonstrated that autoimmune diseases, including EAE, and tumor metastasis are inhibited by antibodies against ␣4 integrin (11,12). Analyzing the regulatory mechanisms of ␣4 integrin is important for understanding the extravasation of autoimmune diseases and tumor metastasis.
In this study, we attempted to clone a novel murine ␣4 integrin splicing variant (␣4B) that contained a short cytoplasmic tail. This ␣4B variant is translated endogenously and expressed on the cell surface with ␤1 integrin. The ␣4B variant is able to bind to VCAM-1 but is dependent upon KVIL cytoplasmic sequences.

EXPERIMENTAL PROCEDURES
Mice-Mice were kept under specific pathogen-free conditions and were provided food and water ad libitum. All experiments were conducted in accordance with the guidelines of the Institutional Animal Care and Use Committee of Hokkaido University.
␣4B cDNA Cloning-The cDNAs encoding ␣4B were cloned from B16-BL6 cells using a 3Ј rapid amplification of cDNA ends method. Total RNA isolated from B16-BL6 cells was reversetranscribed using an oligo(dT) primer containing an anchor sequence. Amplification using PCR was carried out using the anchor primer and an ␣4 integrin-specific primer (5Ј-AGC GAT AAC AAA CTC CCC ACT-3Ј). To increase the specificity of the reaction, a nested PCR was conducted with a second ␣4 integrin-specific primer (5Ј-TAG AGG CCA CAT ACC ACC TTG-3Ј or 5Ј-TGG ATC TAG CGA AGA AAA CGA-3Ј). Amplicons were then cloned into the TOPO-TA vector (Invitrogen) and sequenced. The coding region of ␣4B was amplified by PCR and cloned into pcDNA3.1 (Invitrogen) or pBabepuro and pWZL-blast2 using an infusion system (Takara). FLAGtagged deletion mutants of ␣4B (␣4B⌬VIL) cDNAs were amplified using specific primers (5Ј-GGC GCC GGC CGG ATC  CGC CAC CAT GGC TGC GGA AGC GAG GTG-3Ј and  5Ј-ATT CCA CAG GGT CGA CTT ACT TGT CAT CGT CAT  CCT TGT AGT CCT TCC ACA TAA CAC ATG AAA T-3Ј) and then cloned into pBabepuro.
EAE Induction-EAE was induced in SJL/J mice by immunization with an emulsion of 100 g of PLP 139 -151 peptide in a mixture containing 100 l of complete Freund adjuvant (Difco). Each mouse received an intravenous injection of 400 ng of pertussis toxin on days 0 and 2 (13).
Cell Adhesion Assays-The wells of 96-well plates were coated with the various integrin ligands overnight at 4°C, followed by blocking with 0.5% BSA in PBS for 1 h at room temperature. Cells were suspended in DMEM containing 0.25% BSA and 200 l of cell suspension (at a cell density of 2.5 ϫ 10 5 cells/well), seeded into the 96-well plates, and incubated for 1 h at 37°C. The medium was then removed, and all wells were washed twice. Adherent cells were fixed and stained with 0.5% crystal violet in 20% methanol for 30 min. The wells were rinsed three times with water, and adherent cells were lysed with 20% acetic acid. The resulting supernatants from each well were analyzed using a plate reader (Bio-Rad), with the absorbance at 595 nm measured to determine the relative number of adherent cells.
Immunoprecipitation and Western Blot Analysis-NIH3T3 cells expressing ␣4 integrin, ␣4B, or B16 cells were lysed on ice for 30 min in lysis buffer (50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1% Triton X-100, and 1ϫ protease inhibitors (1ϫ Complete Mini protease inhibitor mixture, Roche Molecular Biochemicals)). Lysates were clarified by centrifugation at 16,000 ϫ g for 10 min at 4°C and incubated with protein G-Sepharose beads coated with anti-␣4 integrin antibody at 4°C for 1 h. The beads were washed with the same buffer five times, and precipitated polypeptides were extracted in Laemmli sample buffer. Protein samples were separated by SDS-PAGE under reducing conditions; probed with the antibody against FLAG, HA, ␤1 integrin, ␣4 integrin, or ␣4B; and then positive signals were detected by Plus-ECL (PerkinElmer Life Sciences).
Flow Cytometry-For ␣4 integrin or ␣4B expression, cells were blocked with normal goat serum and incubated with a phycoerythrin-labeled anti-mouse ␣4 integrin antibody. For WT ␣4 integrin-specific expression, cells were incubated with the 19E4 antibody and phycoerythrin-labeled goat anti-rat IgG antibody after blocking. All analyses were conducted on a FACSCalibur flow cytometer (BD Biosciences).
Analysis of mRNA Expression-Total RNA from mouse tissues and from the spinal cords of EAE mice at day 14 were extracted with TRIzol (Invitrogen). Specific primers were used for RT-PCR and quantitative PCR (qPCR) assays to amplify G3PDH (5Ј-ACC ACA GTC CAT GCC ATC AC-3Ј and 5Ј-TCC ACC ACC CTG TTG CTG TA-3Ј), ␣4 integrin (5Ј-AAG GAA GCC AGC GTT CAT ATT-3Ј and 5Ј-TCA TCA TTG CTT TTG CTG TTG-3Ј), ␣4B integrin (5Ј-AAG GAA GCC AGC GTT CAT ATT-3Ј and 5Ј-AAA GGC ATG GTG TCC ATG TAA-3Ј), and ␣9 integrin (5Ј-GTC TGG GAG GAG GCT AAA CC-3Ј and 5Ј-CAC TGA GGT GCT GTG ATG TTG-3Ј). The qPCR assays were conducted on an Mx3005P (Stratagene). Amplified cDNAs was detected using SYBR Green (Invitrogen) and standardized to ROX dye levels. The cDNA concentrations were expressed as the number of cycles to threshold (Ct), and Ct values were normalized to G3PDH cDNA levels in the same samples. The absolute copy numbers of particular transcripts in B16 cells were calculated from standard curves generated with a 10-fold dilution series of a quantified template DNA.
Tumor Metastasis-B16-BL6 cells were transfected with 50 nM siRNA (0.1 ml/cm 2 ) and complexed with Lipofectamine 2000 (Invitrogen). Mice were inoculated intravenously into the lateral tail vein with 2 ϫ 10 5 B16 cells that had been transfected with siRNA in 0.2 ml of PBS. Mice were sacrificed 21 days postinoculation. The lungs were removed and weighed immediately, and lung metastatic foci were counted.
Statistical Analysis-Data are presented as means Ϯ S.E. and are representative of at least three independent experiments. The statistical significance of differences between groups was calculated using a two-tailed Student's t test. Differences were considered to be significant when p Ͻ 0.05 (*) or 0.005 (**).

RESULTS
␣4B Is an Alternative Splicing Variant of ␣4 Integrin-We used a 3Ј rapid amplification of cDNA ends method involving mouse melanoma B16 cell cDNAs and identified a novel mouse ␣4 integrin splicing variant that we designated ␣4B (GenBank TM accession number: AB850880). The ␣4B variant consisted of identical extracellular and transmembrane domains as the WT ␣4 integrin and contained the novel short amino acid sequence KVIL (Fig. 1A). Truncation occurred after Lys-100, the last amino acid of exon 27. The ␣4B cDNA that encoded the novel VIL amino acid sequence was from an intron (designated region Ex27b) occurring after exon 27.
Expression of ␣4B mRNAs in Normal and Inflamed Mouse Tissues-The WT ␣4 integrin is expressed in various normal tissues, especially immune tissues such as the spleen and lymph nodes. Using normal mouse tissues, RT-PCRs were conducted to determine the expression levels of ␣4B mRNAs. These transcripts were detected in cells expressing WT ␣4 integrin (Fig.  1B). The WT ␣4 integrin is involved in the development of EAE by CD4 ϩ T cells infiltrating the spinal cord. We found that ␣4B mRNAs, as well as WT ␣4 integrin levels, in the spinal cords of EAE mice were increased (Fig. 1C).
␣4B Inhibits Cell Adhesion Mediated by WT ␣4 Integrin-We assessed the ability of ␣4B to support cell adhesion to ␣4 ligands because ␣4B is expressed on the cell surface. Plasma FN was used as a positive control. NIH3T3 cells transfected with an empty vector (Mock/NIH3T3) bound to FN, but not to other ligands, indicating that ␣4 and ␣9 integrins are not expressed on NIH3T3 cells. In NIH3T3 cells expressing ␣4 integrin (␣4 integrin/NIH3T3), appropriate binding to OPN, pp-vWF, FN-CS1, and VCAM-1 was observed. The ␣4B/NIH3T3 cells specifically bound to the VCAM-1 protein. Divalent cations such as Mn 2ϩ can stimulate integrin interactions with ligand (19). Mn 2ϩ leads to enhanced cell adhesion of ␣4 integrin/ NIH3T3, whereas it does not influence the adhesion manner of ␣4B/NIH3T3 (Fig. 2A).
The cytoplasmic region of integrin has been shown to elicit signals from outside of cells and to regulate integrin-mediated cell functions (20,21). Therefore, we next examined cellular signal transductions involved in cell adhesion after cell attachment using phospho-specific antibodies to evaluate the activation status of signaling molecules. Cell lysates from ␣4 integrin/ NIH3T3, ␣4B/NIH3T3, or ␣4 integrin/␣4B/NIH3T3 cells after attachment to FN or VCAM-1 were analyzed by Western blotting. The expression of ␣4B resulted in weak ERK activation compared with that induced by WT ␣4 integrin. No altered activation of AKT, Paxillin, or FAK was observed (Fig. 2B).
We next examined whether ␣4B modulates the function of WT ␣4 integrin. For this experiment, ␣4 integrin/␣4B/NIH3T3 cells were used for a cell adhesion assay. This assay revealed that ␣4 integrin/␣4B/NIH3T3 cells exhibited decreased cell adhesion to ␣4 ligands compared with ␣4 integrin/NIH3T3 (Fig.  2C). The negative effect of ␣4B is partially overcome by Mn 2ϩ treatment. To confirm the possibility that ␣4B down-regulates expression of WT ␣4 integrin on the cell surface, we conducted flow cytometry to determine the surface expression level of the WT ␣4 integrin on ␣4 integrin/␣4B/NIH3T3 cells. We found that three commercially available antibodies against ␣4 integrin cross-reacted with WT ␣4 integrin and ␣4B (Fig. 3A). We attempted to generate novel antibodies specific for ␣4 integrin and successfully established the 19E4 antibody, which recognized the WT ␣4 integrin (Fig. 3B). Using the 19E4 antibody in flow cytometry, we found that the surface expression level of WT ␣4 integrin was unaltered by ␣4B (Fig. 3C). Thus, the decreased cell adhesion of ␣4 integrin/␣4B/NIH3T3 was not due to down-regulation of WT ␣4 integrin expression on the cell surface.
The Inhibitory Effect of ␣4B Is Due to a Dominant-negative Effect for VCAM-1 Adhesion but Not for OPN, pp-vWF, or FN-CS1-We next asked whether ␣4B exerts dominant-negative effect. For this experiment, we established CHO cells coexpressing WT ␣4 integrin and a differential expression level of ␣4B (Fig. 5A). We found that cells having the lowest ␣4B expression (clone 10L) did not inhibit cell adhesion to VCAM-1, whereas all three cells exhibited reduced cell adhesion to OPN, pp-vWF, and FN-CS1 (Fig. 5B). This result suggests that inhibition of ␣4 integrin-dependent cell adhesion by ␣4B is caused by a dominant-negative effect for VCAM-1 binding but not for OPN, pp-vWF, or FN-CS1.
␣4B Knockdown in B16 Melanoma Cells Promotes Lung Metastasis-Expression of endogenous ␣4B in B16 melanoma cells was confirmed by RT-PCR (Fig. 6A). We also found that ␣9 integrin is expressed on B16 cells. Endogenous ␣4B protein was evaluated by first determining that, of the three monoclonal antibodies against ␣4 integrin, 5X2 was most efficient for WT ␣4 integrin and ␣4B immunoprecipitation (Fig. 6B). 5X2 was then used to immunoprecipitate lysates of B16 cells. Two bands (150 kDa and 140 kDa) of WT ␣4 integrin or a band (130 kDa) of ␣4B were detected in B16 cells by an antibody against ␣4 integrin (C-20) or ␣4B, respectively (Fig. 6C). To determine the role of ␣4B expression during B16 lung metastasis, we knocked down endogenous ␣4B in B16 cells using specific siRNAs (si-␣4B) on the basis of the unique 3Ј sequence of ␣4B derived from exon 27b. Results from the qPCR assays showed that si-␣4B treatment of B16 melanoma cells resulted in more than a 70% reduction in ␣4B mRNA levels. No effect on WT ␣4 integrin mRNA expression was observed (Fig. 6D). Western blotting confirmed that the expression of ␣4B protein was reduced substantially by si-␣4B treatment. No effect on cell surface expression of WT ␣4 integrin was observed by flow cytometry (Fig. 6E). Thus, si-␣4B treatment specifically reduced the expression of ␣4B on B16 melanoma cells. We transfected si-␣4B or control siRNAs into B16 melanoma cells and then injected these cells intravenously into C57BL/6 mice. A significant increase in lung metastasis and lung weight was observed in the group of mice inoculated with B16 melanoma cells transfected with si-␣4B (Fig. 6F). Collectively, these findings demonstrate that ␣4B acts as a novel endogenous inhibitor protein of WT ␣4 integrin function.

DISCUSSION
Cytoplasmic domains of integrin are highly conserved during evolution, whereas extracellular domains are not as highly conserved. This would suggest that integrin signaling is indispensable for the maintenance of cellular functions (4). Alternative splicing variants of several integrins in extracellular or cytoplasmic regions have been identified (22). Among them, at least six integrin subunits (␣3, ␣6, ␣7, ␤1, ␤3, and ␤5) have alternative splicing variants in their cytoplasmic domains (23).
Recently, we reported that an alternative splicing variant in the extracellular domain of ␣9 integrin (SF␣9) promotes WT ␣9 integrin cell adhesion by inside-out signaling (24). Both ␣4 and ␣9 integrin, which are in the same integrin family because of structural and functional similarity, are involved in autoimmune diseases (25)(26)(27). We demonstrated the expression and function of a novel murine ␣4 integrin splicing variant, ␣4B, in this study. It contains a unique cytoplasmic amino acid sequence, KVIL. The VIL residues of the KVIL cytoplasmic sequence are derived from exon27b. We evaluated expression patterns of WT ␣4 integrin and ␣4B in mouse tissues and B16 melanoma cells by PCR. We found that ␣4B is expressed in all tested cells expressing the WT ␣4 integrin. The ␣4B variant is expressed on cell surfaces with the ␤1 integrin. These results suggest that ␣4B modulates the function of WT ␣4 integrin by coexpressing on the cell surface.
We analyzed the functional properties of the ␣4B variant and showed that cells expressing ␣4B adhere to VCAM-1. This adhesion requires the presence of the VIL cytoplasmic amino acids, suggesting that a novel cytoplasmic sequence in ␣4B is critical for exertion of its function. Although the amino acid sequences of the extracellular regions in WT ␣4 integrin and ␣4B are identical, the way in which they adhere to ␣4 ligands differs. Integrin conformations and functions are regulated by the cytoplasmic region (28 -30), with varying cytoplasmic sequences in each integrin inducing changes in the structure and manner of binding. This issue was clarified by generating an antibody (clone 19E4) that specifically recognizes ␣4 integrin but not ␣4B.
We also compared the activation of integrin-mediated signaling molecules (ERK, AKT, FAK, and Paxillin) after binding to VCAM-1 for WT ␣4 integrin and ␣4B. We found that the activation of AKT, Paxillin, or FAK was similar. These are major molecules that undergo phosphorylation in response to adhesion via integrins. Thus, integrin-mediated signaling of ␣4B is similar to that for the WT ␣4 integrin. However, activation of ERK is reduced markedly in cells expressing ␣4B. These results indicate that there is a particular molecule responsible for ERK activation. A candidate molecule is caveolin-1, which interacts with some integrins through the transmembrane domain. This functional link between integrin and caveolin 1 activates Ras-ERK signaling (31). We observed that B16 melanoma cells in which ␣4B was knocked down exhibited increased caveolin 1 expression levels (data not shown), suggesting that ␣4B may have an inhibitory function during caveo-lin 1 expression. It is well known that ␤1 integrin binds to FAK and Paxillin (32,33). Both molecules can directly activate AKT (34,35). Therefore, activation of AKT, FAK, and Paxillin ( Fig.  2B) may be elicited by ␤1 integrin associating with ␣4B as a heterodimer subunit. Thus, ␣4B␤1 integrin possesses the unique ability to modulate ERK activation.
A significant reduction of ␣4-dependent cell adhesion was observed in cells expressing both WT ␣4 integrin and ␣4B, irrespective of the change in expression of WT ␣4 integrin. There are several possible mechanisms for how ␣4B inhibits ␣4-dependent cell adhesion. Initially, we hypothesized that ␣4B has a predominantly negative effect on WT ␣4 integrins. This hypothesis is consistent with the fact that ␣4B has no known binding site for integrin binding proteins in its short cytoplasmic tail. However, the adhesion assay result for ␣4B in Fig. 2A, which indicates that ␣4B does not have a binding ability against OPN, pp-vWF, and FN-CS1, does not account for these dominant-negative effects, whereas ␣4B binds to VCAM-1, suggesting that ␣4B exerts a dominant-negative effect. This is consistent with the adhesion assay result using cells coexpressing WT ␣4 integrins and the differential level of ␣4B in Fig. 5, which shows that the lowest ␣4B-expressing cells (clone 10L) exhibit comparable binding against VCAM-1 but reduced binding against OPN, pp-vWF, and FN-CS1. Thus, the inhibi-FIGURE 6. Enhancement of lung metastasis in B16 melanoma cells. A, RT-PCR analysis of the endogenous expression of WT ␣4 integrin, ␣4B, and ␣9 integrin in B16 cells. B, lysates from ␣4 integrin/␣4B/NIH3T3 cells were immunoprecipitated (IP) by anti-␣4 integrin R1-2, PS/2, or 5X2. Precipitated proteins were then separated by SDS-PAGE and blotted with antibody against HA or FLAG. C, endogenous expression of ␣4B in B16 cells. Lysates from B16 cells were immunoprecipitated by 5X2. Precipitated proteins were blotted with antibody against WT␣4 (Santa Cruz Biotechnology) or ␣4B. D, WT ␣4 integrin and ␣4B mRNA levels in control (si-cont) or ␣4B siRNA-transfected B16 cells 2 days post-transduction were quantified using qPCR assays. E, surface expression of WT ␣4 integrin using antibody 19E4 on cells transfected with control or ␣4B siRNAs. F, B16 melanoma cells treated with control or ␣4B siRNA were injected intravenously into C57BL/6 mice via the tail vain. 21 days after injection, the lungs were removed, and the number of metastases were counted. The weight of each lung was also determined. *, p Ͻ 0.05; **, p Ͻ 0.005. Data are mean Ϯ S.E. of one representative of three independent experiments. tory mechanism of ␣4B may be dominant-negative for VCAM-1 binding but not for OPN, pp-vWF, and FN-CS1. The most likely possibility of negative effects on OPN, pp-vWF, and FN-CS1 by ␣4B is inhibition of ␣4 integrin activation. The ␣4 integrin/NIH3T3 cells expressing ␣4B⌬VIL can bind to ␣4 ligands (Fig. 4D), suggesting that the cytoplasmic region of ␣4B elicits negative signals. However, the mechanism responsible for this remains to be elucidated.
WT ␣4 integrin and ␣4B mRNA levels were increased in the spinal cords of EAE mice (Fig. 1C). This result suggests that cells expressing both molecules infiltrate into the spinal cord. It seems inconsistent to infiltrate cells expressing ␣4B in the spinal cords of EAE because ␣4B functions as an ␣4 integrin inhibitor. We found that Mn 2ϩ treatment overcomes the negative effect of ␣4B (Fig. 4E), indicating that the function of ␣4B is reduced in the situation of ␣4 integrin activation states such as EAE. Thus, EAE develops readily, despite the presence of ␣4B.
Three different alternative splicing variants of ␤1 integrin have been reported. ␤1A is the wild type, whereas ␤1B-D are splicing variants (23,36,37). ␤1B and ␤1C are known to have negative functions. ␤1B and ␤1C compete with ␤1A for the formation of the ␣ chain or ligand binding, followed by interference with integrin signaling in cells. The expression levels of ␤1D integrin are similar to that for ␤1A localized in focal adhesions, whereas ␤1B and ␤1C remain diffuse on the surface and do not localize in focal adhesions. Thus, it seems that the negative effects of ␤1B and ␤1C may be due to localization and/or dominant-negative effects. The negative effects of ␣4B seem to be caused by two different ways, a dominant-negative effect or inhibition of WT ␣4 integrin activation, suggesting a different mechanism compared with that for ␤1B and ␤1C. Thus, ␣4B is a novel endogenous inhibitor of ␣4 integrin through a unique mechanism.
In conclusion, our results indicate that the alternative splicing variant ␣4B inhibits ␣4 integrin-dependent cell adhesion via a dominant-negative effect or inhibition of WT ␣4 integrin activation. This is a novel inhibitory manner of regulating ␣4 integrin-dependent cell adhesion. Taken together with our previous report of SF␣9, integrin splicing variants of the ␣4 and ␣9 integrin family might be important for the regulation of wildtype integrin functions. In this study, we introduce the expression and function of murine ␣4 integrin variant ␣4B. However, human ␣4 integrin variants have not yet been identified. To identify and clear the function of human ␣4 integrin variants deserves further investigation.