Mutational Analysis of Cell Cycle Inhibition by Integrin β1C *

Integrin β1C is an alternatively spliced cytoplasmic variant of the β1 subunit that potently inhibits cell cycle progression. In this study, we analyzed the requirements for growth suppression by β1C. A chimera containing the extracellular/transmembrane domain of the Tac subunit of the human interleukin 2 receptor (gp55) fused to the cytoplasmic domain of β1C (residues 732–805) strongly inhibited growth in mouse 10T1/2 cells even at low expression levels, whereas chimeras containing the β1A, β1B, β1D, β3, and β5 cytoplasmic domains had weak and variable effects. The β1C cytoplasmic domain is composed of a membrane proximal region (732–757) common to all β1variants and a COOH-terminal 48-amino acid domain (758–805) unique to β1C. The β1C-specific domain (758–805) was sufficient to block cell growth even when expressed as a soluble cytoplasmic green fluorescent protein fusion protein. These results indicate that growth inhibition by β1C does not require the intact receptor and can function in the absence of membrane targeting. Analysis of deletions within the β1C-specific domain showed that the 18-amino acid sequence 775–792 is both necessary and sufficient for maximal growth inhibition, although the 13 COOH-terminal residues (793–805) also had weak activity. Finally, β1C is known to be induced in endothelial cells in response to tumor necrosis factor and is down-regulated in prostate epithelial cells after transformation. The green fluorescent protein/β1C (758–805) chimera blocked growth in the human endothelial cell line EV304 and in the transformed prostate epithelial cell line DU145, consistent with a role for β1C as a growth inhibitor in vivo.

protein interactions, intracellular ion concentrations, and lipid metabolism. Integrin cytoplasmic domains lack intrinsic enzymatic activity but have been shown to activate a variety of signaling pathways including protein kinases (focal adhesion kinase, integrin-linked kinase, protein kinase C, and mitogenactivated protein kinase), lipid kinases, phospholipases, and small GTP-binding proteins (Ras, Rac, and Cdc 42) and to bind to cytoskeletal proteins including ␣-actinin, talin, filamin, and paxillin.
Four different ␤ 1 cytoplasmic domain isoforms have been identified (␤ 1A , ␤ 1B , ␤ 1C , and ␤ 1D ), and at least one is expressed in nearly every cell type (3)(4)(5)(6). The ␤ 1A isoform is the primary isoform in most cells except for skeletal muscle, where ␤ 1D predominates (4,5). These isoforms are generated by alternative splicing that occurs between the sixth and seventh exons (7). As a result, each of the ␤ 1 cytoplasmic domains contains a common 26-aa 1 membrane proximal sequence and a unique COOH-terminal sequence. The potential significance of these splice variants is highlighted by the fact that many integrinregulated signals are mediated by the ␤ subunit cytoplasmic domain. For example, the ␤ 1A cytoplasmic domain has been shown to interact with the cytoskeletal proteins ␣-actinin, filamin, and talin, and sequences within the unique ␤ 1A domain are required for focal adhesion localization (8 -15). Taken together, these observations suggest that alternative splicing within the ␤ 1 gene may have profound effects on receptor function.
We previously investigated the function of the ␤ 1C isoform in mouse 10T1/2 fibroblasts. We found that expression of ␤ 1C induced a potent inhibition of cell cycle progression, leading to arrest in late G 1 (16). ␤ 1C failed to localize to focal adhesions, suggesting that it has altered interactions with cytoskeletal proteins (16). ␤ 1C also blocked the growth of Chinese hamster ovary cells (17). The functions of ␤ 1C in vivo are unknown, but ␤ 1C expression has been shown to correlate with growth arrest. It is up-regulated in human umbilical endothelial cells in response to treatment with tumor necrosis factor, which blocks the growth of these cells (17), and was found in quiescent prostate epithelial cells but was down-regulated in prostate carcinoma, suggesting that it could function as a tumor suppressor (18).
In this study, we used ␤ 1C chimeras to define the sequences necessary and sufficient for regulating growth. We identified a critical 18-aa region and showed that ␤ 1C sequences can function apart from the rest of the integrin, even in the absence of membrane targeting. We also found that the ␤ 1C -specific domain blocks DNA synthesis in endothelial and prostate carcinoma cells, suggesting that ␤ 1C protein may inhibit growth in vivo.

MATERIALS AND METHODS
Cell Culture-Mouse fibroblast C3H 10T1/2 cells were grown in Dulbecco's modified Eagle's medium low glucose (Life Technologies) supplemented with 10% fetal bovine serum (Life Technologies) and glutamine/penicillin/streptomycin (Life Technologies). The human prostate epithelial cell line DU145 was grown in Dulbecco's modified Eagle's medium high glucose supplemented with glutamine/penicillin/ streptomycin, nonessential amino acids (Sigma), and 10% fetal bovine serum. The human endothelial cell line EV304 was grown in endothelial growth media (Clonetics) supplemented with 10% fetal bovine serum. C3H 10T1/2 cells were transiently transfected using Lipo-fectAMINE (Life Technologies) as per the manufacturer's instructions. DU145 and EV304 cells were transiently transfected using Effectene (Qiagen) as per the manufacturer's instructions. DNAs for transfection were prepared using Plasmid Maxi Kits (Qiagen).
DNA Constructs-IL2 receptor integrin chimeras were generated by fusing various integrin ␤ subunit cytoplasmic domains to the extracellular/transmembrane domain of the IL2 receptor ␣ subunit (gp55) using the unique membrane proximal HindIII site. Chimeras were expressed under the control of the cytomegalovirus promoter in pCDNA3 expression vector (Invitrogen). Cytoplasmic domains and cytoplasmic domain fragments of the integrin ␤ 1A , ␤ 1C , ␤ 1D , and ␤ 5 subunits were generated by polymerase chain reaction amplification using the pBJ␤ 1A , pBJ␤ 1C , pC␤ 1D , and pBJ␤ 5 cDNAs, respectively. The full-length ␤ 1B cytoplasmic domain was generated by polymerase chain reaction using overlapping primers and pBJ␤ 1A cDNA. All fragments were subcloned as HindIII/ XbaI fragments containing an in-frame NH 2 -terminal HindIII site and a COOH-terminal stop codon followed by a XbaI site. Construction of the IL2 receptor chimeras lacking a cytoplasmic domain (IL2R/TL) or containing the integrin ␤ 3 cytoplasmic domain (IL2R/␤ 3 ) or the fulllength ␤ 1D cytoplasmic domain (IL2R/␤ 1D ) 2 has been described previously (19,20). The GFP/␤ 1C chimera was generated by polymerase chain reaction amplification of the ␤ 1C -specific domain 758 -805 containing an in-frame NH 2 -terminal EcoRI site and a COOH-terminal stop codon followed by a SalI site and subcloned into the EcoRI/SalI sites of the GFP expression vector pEGFP-C1 (CLONTECH). Polymerase chain reaction primers were purchased from Life Technologies. All constructs were verified by DNA sequencing. The protein numbering for ␤ 1C is based on the mature protein such that the first residue of the ␤ 1C -specific domain is 758; other investigators have used a numbering system based on the immature protein such that the ␤ 1C -specific domain begins with 778 (17).
Flow Cytometry-Cell growth in transiently transfected cells was measured by labeling cells with bromodeoxyuridine/fluorodeoxyuridine (BrdUrd; Amersham) for 16 h starting at 32 h after transfection. Labeled cells were then detached with trypsin/EDTA (Life Technologies), washed once with growth media and twice with 1% bovine serum albumin/PBS (nuclease-and protease-free; Calbiochem), and then fixed in cold 70% ethanol on ice for 30 min. Cellular DNA was denatured by treating fixed cells with 2 N HCL/0.5% Triton for 30 min and stopped by incubating cells in 0.1 M Na 2 B 4 O 7 , pH 8.5. Cells were then washed once with 0.5% Tween 20/1% bovine serum albumin/PBS (TBP). IL2 receptor chimera expression was determined by incubating cells with a phycoerythrin-conjugated anti-IL2 receptor antibody (PharMingen), and Br-dUrd incorporation was determined by using a fluorescein isothiocyanate-conjugated anti-BrdUrd antibody (Becton Dickinson). Antibodies were diluted in TBP buffer, and labeling reactions were incubated at 37°C for 30 min. Labeled cells were analyzed using a Becton Dickinson FACScan and CellQuest software. Approximately 100,000 cells were analyzed per construct in each experiment. BrdUrd incorporation was determined for different levels of chimera expression, with expression level gates set such that at least 1,000 cells were included in each gate. BrdUrd incorporation levels were determined, normalized against the IL2R/TL control, and expressed as the percentage of inhibition per mean expression level.
Immunofluorescence-Transiently transfected cells were detached with trypsin/EDTA and replated in growth medium on glass coverslips 24 h after transfection. Cell growth was measured by labeling cells with BrdUrd for 5 h starting at 48 h after transfection. Labeled cells were fixed in 2% formaldehyde (EM Sciences)/PBS for 15 min at room temperature, extracted with 0.2% Triton/PBS for 10 min, treated with 0.1 unit/l DNase I for 30 min at 37°C, and blocked in 10% normal goat serum/PBS (Life Technologies). Primary and secondary antibodies were diluted in 10% normal goat serum/PBS, and labeling reactions were incubated for 30 min at 37°C. In order to measure BrdUrd incorporation in cells expressing IL2 receptor chimeras, cells were probed with an anti-BrdUrd primary antibody (Calbiochem) followed by a fluorescein isothiocyanate-conjugated goat anti-mouse IgG secondary antibody (Jackson ImmunoResearch). Cells were then blocked with mouse IgG (Sigma) before labeling with a phycoerythrin-conjugated anti-IL2 receptor antibody (PharMingen). In order to measure BrdUrd incorporation in cells expressing GFP chimeras, cells were probed with an anti-BrdUrd primary antibody followed by a lissamine rhodamine sulfonyl chloride-conjugated goat anti-mouse IgG secondary antibody (Jackson ImmunoResearch). Coverslips were mounted in Immunofluore mounting media (ICN) and examined using a Leitz diaplan fluorescence microscope. Fluorescent images were obtained using a Bio-Rad 1024 MRC Scanning Confocal Microscope.

Localization of Integrin Cytoplasmic Domain-IL2 Receptor
Chimeras-To map the growth regulatory sequences in ␤ 1C , we generated a chimera containing the extracellular/transmembrane domain of the 55-kDa Tac subunit of the IL2 receptor fused to the cytoplasmic domain of ␤ 1C (IL2R/␤ 1C ). Similar chimeras have been used to investigate the signaling of other integrin ␤ subunit cytoplasmic domains, including ␤ 1A , ␤ 1D , and ␤ 3 (19 -24). These chimeras have been shown to mimic properties of wild type receptors, including focal adhesion localization and the activation of focal adhesion kinase phosphorylation, but in a ligand-independent manner (21). Mouse 10T1/2 fibroblasts were transiently transfected with IL2 receptor chimeras containing the ␤ 1A , ␤ 1B , ␤ 1C , and ␤ 1D cytoplasmic domains and stained for surface expression. A tail-less construct (TL) that did not contain any integrin sequences was used as a control. As shown in Fig. 1, IL2R/␤ 1C (␤ 1C ) exhibited a diffuse staining pattern similar to the staining pattern of the tail-less IL2R chimera (TL). This result is consistent with the diffuse staining pattern observed for the full-length receptor (16). The IL2R/␤ 1B chimera (␤ 1B ) also displayed a diffuse staining pattern similar to the wild type ␤ 1B receptor (25). In contrast to the ␤ 1C and ␤ 1B chimeras, the ␤ 1A and ␤ 1D chimeras were localized to focal adhesions, in agreement with previous reports (19,20,24). Effects of IL2R Chimeras on DNA Synthesis-We next tested the effects of the different chimeras on DNA synthesis. Transiently transfected cells were replated on glass coverslips after 24 h and incubated with the thymidine analogue BrdUrd for 5 h starting at 48 h after transfection. Cells were then fixed and double-stained for IL2R expression and BrdUrd incorporation. We have found that high levels of chimera expression induced cell retraction and rounding and inhibited cell adhesion or spreading, 3 consistent with previous reports that these chimeras can function as dominant negative inhibitors for endogenous integrin function (20,23,26). Because the 10T1/2 fibro-blasts are anchorage-dependent, effects on cell adhesion would result in nonspecific growth arrest. Therefore, to rule out this type of effect, only morphologically normal cells were scored for nuclear labeling with anti-BrdUrd. As shown in Fig. 2, the IL2R/␤ 1C chimera blocked BrdUrd incorporation in Ͼ99% of the expressing cells. In contrast, the IL2R/␤ 1A , ␤ 1B , and ␤ 1D chimeras showed varying but relatively modest effects on Br-dUrd labeling. The ␤ 1A and ␤ 1B chimeras caused decreases in DNA synthesis, whereas the ␤ 1D chimera induced a modest increase compared with the IL2R/TL control (Fig. 2). We also examined chimeras containing the integrin ␤ 3 and ␤ 5 cytoplasmic domains, both of which localized to focal adhesions. 3 As shown in Fig. 2, the ␤ 3 chimera induced an increase in cells labeled with BrdUrd, whereas the ␤ 5 chimera partially inhibited labeling. These results indicate that only the ␤ 1C cytoplasmic domain functions as a potent inhibitor, although the other ␤ subunit cytoplasmic domains have detectable effects on growth. These data also indicate that the ␤ 1C cytoplasmic domain is sufficient to induce cell cycle arrest when separate from the rest of the integrin.
Dose-Response Curves-To obtain a more quantitative estimate of the dependence of growth inhibition on expression levels, dose-response curves for the ␤ 1A and ␤ 1C chimeras were generated. Starting at 30 h after transfection, cells were labeled with BrdUrd for 16 h and then detached, fixed, and stained for both BrdUrd incorporation and IL2R chimera expression. Cells were then assayed by flow cytometry. Using this protocol, the maximum growth inhibition was 45-50%; even cells that were detached and put into suspension at the begin- ning of the labeling period were inhibited by only 45%. Presumably, this result reflects the fact that a significant fraction of the cells have passed the restriction point and are already committed to or have entered S phase. The correlation between expression level for IL2R and BrdUrd staining is shown in Fig.  3. We found that both of the integrin chimeras induced growth inhibition, although at markedly different expression levels. The ␤ 1A chimera blocked growth at high expression levels, consistent with the idea that this chimera may function as dominant negative inhibitor for endogenous integrin function. In contrast, the ␤ 1C chimera was a distinctly more potent, maximally inhibiting DNA synthesis at a level of expression just above background.
The results in Fig. 3 using fluorescence-activated cell-sorting analysis show significant inhibition of DNA synthesis at high levels of ␤ 1A expression, in disagreement with Fig. 2, in which DNA synthesis was only slightly decreased in ␤ 1A -expressing cells scored after plating on coverslips. One possible source of this discrepancy is that high expressors might be lost after replating and their influence might be further decreased because rounded cells were not scored. To investigate this question, cells transfected with ␤ 1A or ␤ 1C were left in the original dish or replated into a fresh tissue culture dish and then detached and analyzed for expression of the transfected integrins. Replating caused a substantial loss of the ␤ 1A high expressors from the population, whereas no decrease in ␤ 1Cexpressing cells was observed. 3 This result explains the difference between the two methods and supports the conclusion that growth inhibition by ␤ 1A is due to a dominant negative effect on cell adhesion, whereas inhibition by ␤ 1C is caused by a distinct mechanism.
Mapping Inhibitory ␤ 1C Sequences-The ␤ 1C cytoplasmic domain (residues 732-805) is composed of a common region (732-757) shared with the other ␤ 1 splice variants and a unique region (758 -805) specific to ␤ 1C . Hungerford et al. (27) showed that microinjection of a peptide derived from the common ␤ 1 region can block endogenous integrin function and induce programmed cell death in chicken fibroblasts in serum-free media. As shown in Fig. 4, deletion of the common region had no effect on growth inhibition by ␤ 1C , indicating that the ␤ 1C -specific domain was sufficient. We next asked whether the ␤ 1C -specific domain could block growth when expressed as a soluble protein in the absence of membrane targeting. A chimera was generated containing the ␤ 1C -specific domain (758 -805) fused to the GFP. This chimera blocked growth in 93% of the expressing cells (Fig. 4B). Thus, growth inhibition by ␤ 1C occurs even when expressed as a soluble cytoplasmic protein.
In order to further define growth regulatory sequences, we generated a series of IL2R/␤ 1C chimeras containing deletions within the ␤ 1C -specific domain (758 -805). Deletion of the membrane-proximal 17 aa (Fig. 5, construct 2) had no effect on growth inhibition, whereas the deletion of an additional 8 aa (construct 3) led to a substantial increase in DNA synthesis. Expression of the COOH-terminal 13 aa alone (construct 4) had weak but reproducible growth-inhibitory activity. When we screened deletions from the COOH terminus of the ␤ 1C -specific domain, we found that deletion of the COOH-terminal 13 aa failed to increase DNA synthesis (constructs 2-4). This result was surprising, because we had previously reported that the deletion of this COOH-terminal 13 aa region (793-805) abolished growth inhibition by the intact receptor (16). Deletion of the next 10 residues (Fig. 5B, construct 5) almost completely eliminated growth inhibition. Taken together, these results identify an 18-aa region, 775-792, that is the major site of the growth-inhibitory activity. The COOH-terminal 13-aa region also has weak activity; curiously, it is required for growth inhibition by the intact receptor but not by the shorter ␤ 1Cspecific domain expressed as a chimera with the IL2R. This sequence may therefore be important for the proper conformation of the intact receptor but not a shorter peptide. Growth Inhibition in Human Cell Lines-Previous reports suggested that ␤ 1C may function as a cell growth regulator: its expression is induced in human umbilical endothelial cells in response to tumor necrosis factor, which causes growth arrest, and ␤ 1C is down-regulated during the transition from normal, quiescent prostate epithelium to prostate carcinoma (17,18). To test the ability of ␤ 1C to block growth in these cell types, GFP/␤ 1C was transiently expressed in the human endothelial cell line EV304 and in the transformed prostate epithelial cell line DU145. Expressing cells were labeled with BrdUrd to assay cell cycle progression. As shown in Fig. 6, GFP/␤ 1C induced growth arrest in both the EV304 cells and the DU145 cells, blocking BrdUrd labeling by 46.7% and 72.8%, respectively, compared with GFP controls. These results indicate that ␤ 1C can block cell growth in human cell lines relevant to its in vivo expression. DISCUSSION We used chimeras of the ␤ 1C cytoplasmic domain with both the IL2 receptor and GFP to map growth regulatory sequences in integrin ␤ 1C . The data revealed that growth inhibition by ␤ 1C occurred at low expression levels and was mediated entirely by the ␤ 1C -specific domain. This domain functioned in the absence of the ␤ subunit extracellular, transmembrane, and common cytoplasmic sequences, did not require the ␣ subunit, and was effective in the absence of membrane targeting. Additional deletion mapping showed that the primary activity resided within an 18-aa region (775-792). The ␤ 1C cytoplasmic domain also blocked growth when expressed in human endothelial cells and transformed prostate epithelial cells, suggesting that the endogenous protein may function in a similar capacity under physiological conditions. These conclusions are summarized in Fig. 7.
Other integrins had some effects on cell cycle progression. The ␤ 1A chimera decreased DNA synthesis at high levels, consistent with reports that it functions as a dominant negative inhibitor for endogenous integrin function in cell adhesion (20,23,26). Expression of ␤ 1B also decreased DNA synthesis, consistent with published data indicating that this isoform can function as a dominant negative inhibitor (28). We observed some stimulation of DNA synthesis by ␤ 1D , whereas Belkin and Retta (24) reported that the ␤ 1D chimera blocks cell growth when expressed in C2C12 myoblasts, suggesting that effects of ␤ 1D may be cell type-specific.
Two previous studies reported the effect of COOH-terminal deletions within the full-length ␤ 1C receptor on DNA synthesis (16,17). We found that the COOH-terminal 13-aa region was required for growth inhibition by the full-length receptor in 10T1/2 cells, whereas, using a different protocol, Fornaro et al. (17) reported that a receptor lacking this region blocked growth in Chinese hamster ovary cells. The results reported here may help resolve this discrepancy by showing that the last 13 aa are not required for growth inhibition by a shorter IL2 receptor chimera that contains only the ␤ 1C -specific sequences. Thus, this region may be required for maintaining proper conformation in some contexts, but it does not appear to be essential for signaling.
Our findings and those of Fornaro et al. (17) both indicate that the 775-782 region is critical; however, we found that this region was not sufficient for growth inhibition but also required residues 783-792. By contrast, Fornaro et al. (17) observed growth inhibition in a construct lacking residues COOH-terminal to 782. This difference could be due to the difference in cell type or the distinct protocol used in their studies, in which expressing cells were analyzed after being captured by panning on anti-␤1 antibodies. Alternatively, those studies used fulllength ␤ 1C constructs that contained the common, membraneproximal region. Dominant negative effects leading to growth inhibition could therefore complicate the interpretation of  17) is also indicated. The protein sequences of four Alu-containing proteins that are homologous to ␤ 1C are shown below: an alternatively spliced variant of the neurofibromatosis 2 tumor suppressor gene (NF2), an alternatively spliced variant of the c-myb transcription factor (c-myb), the transformation-related protein (TRP), and the neuronal thread protein AD7c-NTP (NTP). The predicted protein sequence of an Alu-Sx element in the antisense orientation is also shown. those results.
The ␤ 1C -specific exon is formed by an Alu sequence in the antisense orientation located between the sixth and seventh exon of the human ␤1 gene (3,29). Alu sequences are members of the short interspersed nucleotide elements family of repetitive elements, and approximately 700,000 Alu elements are thought to be present per haploid human genome; thus, an average of more than 1 Alu element is predicted to be present in every gene (30). Many genes express Alu-containing sequences, and a few Alu-containing proteins have been reported previously (31). The function or significance of these elements is not known. We have mapped the ␤ 1C growth inhibitory domain to a region of ␤ 1C that overlaps a sequence that is conserved in other Alu-containing proteins including transformation-related protein and neuronal thread protein and alternatively spliced isoforms of neurofibromatosis 2 and c-myb (Fig. 7). The function of these proteins and their roles in growth control are not known. In addition to this conserved region, the ␤ 1C growth inhibitory domain also contains a TSR sequence that is not found in the other Alu-containing proteins (Fig. 7).
The design of cell growth inhibitors is one of the major strategies in the fight against cancer and other human diseases. We have identified a short peptide sequence that is sufficient to induce growth arrest in endothelial cells (EV304), in transformed prostate epithelial cells (DU145), and in the K562 transformed hematopoietic cell line (data not shown) as well as in rodent fibroblast cell lines. These results suggest that the identification of ␤ 1C -binding proteins and the characterization of ␤ 1C -mediated downstream signaling pathways could aid in the development of novel angiogenesis or tumor growth inhibitors.