• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptNIH Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Biochemistry. Author manuscript; available in PMC Jul 14, 2009.
Published in final edited form as:
PMCID: PMC2710129

Utilization of a Soluble Integrin-Alkaline Phosphatase Chimera To Characterize Integrin α8β1 Receptor Interactions with Tenascin: Murine α8β1 Binds to the RGD Site in Tenascin-C Fragments, but Not to Native Tenascin-C


The integrin α8β1 has been reported to bind to fibronectin, vitronectin, and tenascin-C in cell adhesion or neurite outgrowth assays. Here, we describe cDNA cloning of the murine α8 subunit, purification of a recombinant soluble heterodimer consisting of the extracellular domains of the murine α8 and β1 subunits, and development of a sensitive binding assay using a modified form of this heterodimer fused to alkaline phosphatase (AP). In binding assays, the purified α8β1-AP chimera exhibited the same divalent ion requirements for activation and binding specificity as cell surface α8β1: in the presence of Mn2+ it bound to fibronectin and vitronectin in an RGDS-peptide inhibitable manner. Contrary to previous reports, we found no evidence that α8β1, expressed on K562 cells or as an AP chimera, interacts strongly with native tenascin-C. In binding, adhesion, and spreading assays, significant interactions were observed only to short fragments of tenascin-C containing the third fibronectin type III repeat which contains an RGD sequence. Full length tenascin-C and longer fragments containing this repeat did not appear to serve as ligands, implying that the RGD site in native tenascin-C is a cryptic binding site for this integrin, exposed by removal of adjacent domains. Soluble integrin-AP chimeras should be generally useful for identifying and characterizing integrin interactions with ligands.

Integrins are a family of noncovalently linked αβ heterodimers which serve as cellular receptors for a wide variety of ligands including extracellular matrix (ECM)1 glycoproteins, immunoglobulin, and cadherin-class cell adhesion molecules, and disintegrins (1-3). Integrins have been shown to play important roles in development, hemostasis, wound healing, immune responses, and malignant transformation (1). The integrin α8 subunit associates with the integrin β1, but not other integrin β subunits, to form a receptor for several ECM constituents including fibronectin (FN), vitronectin (VN), and osteopontin (4-7, 44). Evidence has also been presented indicating that tenascin-C (TN-C) is a ligand for this integrin using cell adhesion and neurite outgrowth assays (6, 8). Recently, we have shown that this integrin mediates epithelial-mesenchymal interactions essential for normal development of the metanephric kidney, although none of the known ligands for this integrin appear to have appropriate expression patterns to account for the requirement for this integrin during kidney morphogenesis (9). Motivated in part by these observations, we constructed a soluble truncated mouse α8β1 integrin heterodimer as a fusion protein with alkaline phosphatase (α8tβ1-AP) to use as a probe to characterize known receptor–ligand interactions and to help identify novel ligands. While AP chimeras have been used to identify the ligands for receptor tyrosine kinases (10, 11), they have not been previously constructed with heterodimeric receptor molecules such as integrins. In previous work, detection of integrin ligands using purified integrins has been difficult because integrin interactions with ligands are often of comparatively low affinity. Interpretation of cell adhesion assays has often been difficult because many cells express several integrins capable of binding to the same ligand. Confidence in interpretation of results using function-blocking antibodies has been reduced by possibilities of indirect effects mediated by steric hindrance or perturbations of cellular signaling pathways. In the present work, we report successful expression, secretion, and purification of an α8tβ1-AP chimera retaining ligand-binding activity and AP activity. Using the purified recombinant α8tβ1-AP chimera in solid phase binding assays, we show that it provides a sensitive reagent for characterizing receptor–ligand interactions. We have reexamined ligand interactions of this integrin and present evidence that this integrin’s interactions with human and chicken TN-C are mediated through a binding site containing an RGD sequence which appears to be cryptic in native TN-C, but is exposed in certain TN-C fragments.



Human plasma FN, the 120 kDa and 40 kDa FN fragments (FN120, FN40), bovine plasma VN, human TN-C, human thrombospondin-1, and the peptides GRGDSP and GRGESP were purchased from Gibco BRL (Gaithersburg, MD). Chicken TN-C was purified as described (8). Rat collagen types I and IV were purchased from Collaborative Research Inc. (Bedford, MA). Laminin-1/EHS laminin was purified as described (12). Recombinant chicken TN-C fragment fusion proteins were prepared according to published procedures (13). Recombinant human TN-C fragment proteins (14) and native human TN-C (15) were prepared as described. Recombinant mouse entactin/nidogen was a gift from Dr. R. Timple, Max-Planck Institute für Biochemie, Martinsried, Germany. Glutathione-S-transferase (GST)-entactin fragment fusion proteins (16) were a gift from Dr. A. E. Chung, University of Pittsburgh. A synthetic peptide containing the C-terminal residues of the human integrin α8 cytoplasmic domain was synthesized by Dr. C. W. Turck, Howard Hughes Medical Institute, University of California, San Francisco. The mouse integrin β1 cDNA clone (17) was a gift from Dr. S. Tominaga, Jichi Medical School, Tochigi, Japan. The human placental AP fusion vector (10) was a gift from Dr. J. G. Flanagan, Harvard Medical School.


Anti-GST-mouse α8 antibody was prepared by immunizing rabbits with powdered lyophilized gel slice containing a 130 kDa GST-mouse α8 fusion protein purified by SDS–PAGE from Escherichia coli inclusion bodies. This fusion protein was prepared by expressing the plasmid pGEX-4T3 (Pharmacia LKB Biotechnology Inc. Piscataway, NJ) containing the BamHI–Bsp120I fragment of a PCR-amplified clone (see below) encoding amino acids 1–917 of the mature mouse integrin α8. This clone has a mutation (F to L) at the amino acid number 481. Anti-α8 cytoplasmic domain antibody was prepared as described (4), using the synthetic peptide (28 residues) of the cytoplasmic domain of human integrin α8 (7). These antibodies recognized specifically the human and mouse integrin α8 subunit in Western blots and in immunoprecipitations. The other antibodies were as follows: anti-human integrin αvβ5 mAb P1F6 (Gibco BRL); anti-human integrin α5 mAb B1E5 (18); anti-human integrin β1 mAb AIIB2 (18); anti-mouse integrin β1 mAb 9EG7 (19); anti-rat integrin β1 mAb HA2/11 (20); anti-human placental AP mAb MIA1801 (Medix Biotech, San Carlos, CA); anti c-Myc mAb 9E10 (Santa Cruz Biotechnology, Santa Cruz, CA). IgG fractions were prepared from antisera or ascites using protein A or protein G-Sepharose (Pharmacia LKB Biotechnology).

cDNA Cloning and Plasmid Construction

The Tri-clone system (Invitrogen, San Diego, CA) was used to clone mouse integrin α8 cDNA: mRNA was purified from NIH 3T3 cells; cDNA was synthesized by reverse-transcriptase using random primers; several amplification reactions were performed using different sets of degenerate or gene-specific primers; the amplified fragments were gel-purified using Qiaex (Qiagen, Chatsworth, CA) and cloned into pCRII (TAcloning vector, Invitrogen). The degenerate primers used were sense primer A (nucleotide position 1–20, where position 1 is the 5′ end of coding sequence for mature α8 protein), 5′-TT(CT)AA(CT)(CT)T(ACGT)GA(CT)G(AT)(ACGT)GA(AG)AA-3′; antisense primer H (position 3065–3046), 5′-GTCTTGT(CT)ATTTGTCAG(CT)TG-3′; and antisense primer I (position 3078–3060), 5′-G(CT)(CT)A(ACGT)GC(ACGT)TCAG(GT)(ACGT)GTCTTG-3′. The gene-specific primers used were primer B (position 1–35), C (position 919–938), D (position 1730–1746), E (position 2858–2877), F (position 962-946), and G (position 1821–1797). The plasmid inserts were sequenced using Sequenase 2.0 (Amersham, Arlington Heights, IL). The cDNA clone (pCRII-33) encoding the consensus sequence of mouse α8 (amino acids 1–1021) was generated by ligating the PCR-amplified clones. To express a secreted protein, a synthetic BamHI-AvaI DNA fragment encoding the modified chicken α8 signal peptide fused to the N-terminal portion of the mature mouse α8 subunit (MARRQPPRPLLLLSALLCAPASAFNLDVEKLTVYSG), generated by annealing two complimentary oligonucleotides, was ligated into BamHI-AvaI sites of pCRII-33, resulting in the clone named pCRII-35. A clone encoding a truncated mouse α8 (α8t) was generated by PCR using pCRII-35 as a template, with a gene-specific 5′-primer containing a 5′BamHI site and a 3′-primer encoding the C-terminus of the deduced extracellular domain of mouse α8 plus a (His)6-myc tag and a stop codon (VIWATPNVSHHHHHHGEQKLISEEDL-stop). This amplified fragment was subcloned into the BamH1-EcoRV site of the eucaryotic expression vector pCR3 (Invitrogen). The truncated mouse β1 (β1t) was generated by introducing a stop codon after the end of extracellular domain (amino acid sequence number 728) by PCR using the mouse β1 cDNA clone ST1 (17) as a template with a gene specific 5′-primer containing a BamHI site and a 3′-primer containing an EcoRI site. This was subcloned into the BamHI and EcoRI sites within the multiple cloning site of pCR3. The mouse β1 extracellular domain-AP chimera was generated by isolating a modified pCR3-β1t, in which the stop codon at the end of extracellular domain of β1t was eliminated by PCR. The SnaBI (filled in with Klenow)-XhoI fragment containing the AP from APtag-1 (10) was ligated into the EcoRI (filled in with Klenow) and XhoI sites of the pCR3-β1t plasmid modified to eliminate the stop codon. All of the PCR-generated clones were sequenced to confirm that they do not contain mutations caused by Taq polymerase.

Purification of Soluble Truncated Integrin Heterodimers

COS-7 cells were grown in 15 cm diameter dishes and transiently cotransfected using LipofectAmine Reagent (Gibco BRL) with the two plasmids, encoding extracellular domains of the α and β subunits plus indicated C-terminal tags (α8t and β1t to express α8tβ1t; α8t and β1-AP to express α8tβ1-AP). After treatment with the lipofection mixture for 5 h, the medium was changed to DMEM supplemented with Nutridoma HU (Boehringer, Indianapolis, IN) and antibiotics. The conditioned medium was collected every 2–4 days for 1 week. After addition of MgCl2 (at final concentration of 1 mM), phenylmethanesulfonyl fluoride (PMSF, 1 mM), and sodium azide (0.02%), the conditioned medium was filtered (paper number 1, Whatman, Maidstone, UK) and concentrated 10–20-fold using a YM100 membrane (Amicon, Beverly, MA). The concentrated medium was supplemented with Tris-Cl, pH 8.0 (at final concentration of 20 mM), imidazole (10 mM), and PMSF (0.5 mM), and was then incubated at 4 °C for 2 h in batch with Ni-NTA beads (Qiagen). The beads were transferred into an empty column and washed with buffer A (20 mM Tris-Cl, pH 7.5, 300 mM NaCl, 1 mM MgCl2, 0.02% sodium azide, 20 mM imidazole, 0.5 mM PMSF). After washing, bound proteins were eluted with elution buffer (20 mM Tris-Cl, pH 7.5, 50 mM NaCl, 1 mM MgCl2, 0.02% sodium azide, 100 mM imidazole, 0.5 mM PMSF). To remove imidazole, α8t monomer, and nonspecifically bound contaminants of low molecular weight, the buffer of the eluate was exchanged by repeating concentration and dilution with the elution buffer without imidazole, using a Centricon 100 or Centriplus 100 filter apparatus (Amicon, Beverly, MA). Protein concentration of the purified integrins was determined by both Coomassie Plus Protein Assay Reagent (Pierce, Rockford, IL) and silver staining of proteins after fractionation by SDS–PAGE. The purified heterodimers retained activity for at least 5 months when stored at either 4 °C or −80 °C.

Radiolabeling and Immunoprecipitation

COS-7 cells were transfected with the plasmids as described above. After 1 day, the cells were metabolically labeled in methionine-and-cysteine-free DMEM supplemented with Nutridoma HU for 2 h with 500 μCi/mL Tran35S-label (ICN, Irvine, CA). After labeling, the medium was replaced with fresh radioisotope-free medium, and the cells were further incubated for 17 h. The conditioned medium was harvested, precleared by incubation with Sepharose CL-4B (Pharmacia LKB Biotechnology), and immunoprecipitated with indicated antibodies and protein A or protein G-Sepharose beads (Pharmacia LKB Biotechnology). After collection, the beads were washed with buffer B (50 mM Tris-Cl, pH 7.5, 150 mM NaCl, 1 mM CaCl2, 1 mM MgCl2, 1% Triton X-100, 0.1% SDS, 0.1% NP-40). To collect the secreted α8t with Ni-NTA beads, the precleared medium was incubated with Ni-NTA beads, and beads were washed with buffer A. For immunoprecipitations or Ni-NTA binding of metabolically labeled integrins, 500 μL of medium was used directly for each sample without prior fractionation.

Solid-Phase Binding Assays

96 Well plates (Maxisorp, Nunc, Rochester, NY) were coated with indicated concentrations of substrate proteins in TBS (25 mM Tris-Cl, pH 7.5, and 100 mM NaCl) at 4 °C overnight, blocked with 1% BSA (RIA grade, Sigma, St. Louis, MO) in TBS, and washed with TBS containing 1 mM MnCl2 (TBS-Mn). Thrombospondin-1 was coated in the presence of 1 mM CaCl2. As negative and positive controls, wells were coated with 1% BSA or 10 μg/mL FN120, respectively. A total of 100 μL of α8tβ1-AP (5 μg/mL in TBS-Mn) was then added to each well, which was then incubated for 2 h at room temperature. After washing wells 5 times with TBS-Mn, 100 μL of AP substrate (12 mM p-nitrophenyl phosphate, 0.5 mM MgCl2, and 1 M diethanolamine, pH 9.8) was added and incubated at room temperature for an appropriate time. Integrin binding was quantified by measuring absorbance at 405 nm. α8tβ1t biotinylated with NHC-LC-biotin (Pierce, Rockford, IL) was also used in the binding assay, and binding was detected as described (21). Where indicated, the plates were coated first with nitrocellulose and then with proteins as described (8).

Cell Adhesion and Spreading Assays

The human erythroleukemia cell line K562, KA8 cells (chicken α8 subunit-transfected K562 cells), cell adhesion, and cell spreading assays have been described (5).


Cloning of the Coding Sequence of the Mouse α8 Integrin by RT-PCR

To obtain clones encoding the murine homologue of the integrin α subunit, degenerate primers were designed using the published sequences of cDNAs encoding the chicken and human α8 proteins (4, 7). Using different sets of primers in RT-PCR reactions with the mRNA from NIH 3T3 cells, several overlapping cDNA clones were obtained. Multiple independently derived PCR clones were sequenced to determine the consensus sequence of the integrin coding region (Figure 1). The deduced amino acid sequence of the integrin coding region was 90% identical to the human and 76% identical to the chicken α8 subunit. In comparisons with other integrin α subunits, the α5 and αv subunits have been shown to be most closely related to α8 (e.g., refs 4 and 7). Comparisons of the murine α subunit sequence described above to those of the published human and murine αv and α5 subunits revealed only 45–48% identity to each. Together, these results indicate that the subunit characterized above represents that of the murine integrin α8 subunit. As previously described for the avian and human homologues (4, 7), the murine α8 subunit contains three potential metal-binding domains and a putative post-translational cleavage site. It also includes 16 potential N-glycosylation sites.

Figure 1
Mouse integrin α8 cDNA nucleotide sequence and deduced amino acid sequence. The consensus sequence of the clones obtained from different RT-PCR reactions is shown, except that the nucleotide sequence number 3046–3059 was derived from the ...

To provide independent evidence that our clones encode the murine integrin α8 subunit, we prepared antibodies to a bacterially expressed GST fusion protein containing the extracellular domain of the murine integrin α subunit described above. This antiserum was used in western blots using extracts of tissues of wild-type mice or mice lacking α8 protein due to a targeted mutation in the α8 gene (9). As predicted, a protein with the expected size of the α8 subunit was detected in extracts from the wild-type, but not mutant animals (ref 9 and data not shown). A second antibody was prepared to a peptide with the same sequence as the deduced cytoplasmic domain of the human α8 subunit (7). We expected that the antibody against the human cytoplasmic domain would recognize its murine counterpart due to the conservation of cytoplasmic domain sequences of integrin α subunits across species. This antibody was also shown to recognize a protein with the expected size of the α8 subunit in a variety of murine tissues from wild-type, but not mutant mice (9). The results using both antibodies provide further, independent evidence that the sequence presented here encodes the murine homologue of the integrin α8 subunit.

Truncated Integrin α8 and β1 Subunits Are Secreted as a Heterodimer

Transient transfection experiments followed by immunoprecipitations confirmed that the expressed mature full-length murine α8 subunit formed cell surface heterodimers with the endogenous β1 subunit in human 293 cells and indicated that this α subunit is proteolytically processed into large and small disulfide-linked fragments during maturation (not shown), similar to maturation of this subunit in other species (4, 7). To generate a secreted, soluble heterodimer containing the extracellular domains of the murine α8 and β1 subunits, we prepared constructs designed to express the extracellular domains of the mouse α8 and mouse β1 subunits by elimination of the transmembrane and cytoplasmic domains (Figure 2, panels A and B). Similar constructs have been previously used to express soluble extracellular domains of other integrin heterodimers (22-27). For purposes of affinity purification and detection, respectively, (His)6 and c-myc epitope tags were added to the C-terminal coding sequence of the α8 extracellular domain to form the truncated α8 construct named α8t. The construct β1t encoded the entire extracellular domain of β1 truncated with an artificial stop codon following amino acid 728. Human placental alkaline phosphatase (AP) was chosen for addition to the C-terminal coding sequence of the truncated β1 subunit to form the chimera named β1-AP because AP has well-characterized and sensitive detection properties (10, 11).

Figure 2
Expression of truncated forms of the mouse integrin α8 and β1 subunits. (A) Schematic representation of the expression constructs. Domain structures of the translated proteins are shown. After cleavage of the signal peptides (SP), the ...

To examine the abilities of the truncated proteins encoded by these constructs to be assembled into heterodimers and secreted, COS-7 cells were transfected with constructs encoding α8t or β1t alone or in combination and grown in [35S]methionine-containing media. Following immunoprecipitation or Ni-affinity chromatography, radiolabeled proteins were separated by SDS–PAGE and visualized by fluorography (Figure 2C). No secreted proteins corresponding in size to the α8t and β1t subunits were seen in media conditioned by mock-transfected cells (not shown). When constructs encoding α8t or β1t were transfected alone, each truncated subunit was secreted into the media (Figure 2C, lanes 1 and 2). When α8t and β1t were cotransfected, both subunits were also secreted into the medium and apparently formed heterodimers as demonstrated by coelution of both subunits from Ni-affinity columns and coimmunoprecipitates with an anti-β1 mAb (Figure 2C, lanes 3 and 4). The differences in the relative intensities of the α8t and β1t bands in lanes 3 and 4 suggest that subunit monomers as well as heterodimers were secreted from the COS-7 cells cotransfected with both α8t and β1t, and this will be more convincingly demonstrated below.

Purification of Soluble Truncated α8β1 Heterodimers

Two combinations of cDNA constructs encoding α8 and β1 subunits (α8t with β1t; α8t with β1-AP) were coexpressed in COS-7 cells, and conditioned media from these cells were used to establish purification schemes for both the α8tβ1t and the α8tβ1-AP heterodimers (see Experimental Procedures and Figure 3). As a control, secreted AP was also expressed alone, not fused to the integrin β1 subunit. Purified proteins were fractionated by SDS–PAGE in reducing or nonreducing conditions to establish purity and to determine whether the secreted α8t subunit was proteolytically cleaved into two disulfide linked polypeptides, similar to the full-length α8 subunit. Molecules were visualized either by silver staining or by antigen-blot employing subunit-specific antibodies or antibodies to AP (Figure 3). In silver-stained gels, only proteins with the expected sizes of α8t, β1t, and β1-AP were visible, suggesting that the purification schemes yielded pure α8tβ1t and α8tβ1-AP heterodimers (Figure 3, lanes 1–4). Antigen blots with anti-α8, anti-β1, or anti-AP confirmed that the highly purified proteins did indeed correspond to α8t, β1t, and β1-AP (Figure 3, lanes 6–13). Antigen blots revealed that the apparent molecular weight of the α8t subunit was shifted from 152 to 137 kDa by addition of a reducing agent (Figure 3, lanes 6–9), strongly suggesting that the secreted α8 subunit had been cleaved into two disulfide linked polypeptides, similar to the normal α8 subunit.

Figure 3
Purification of soluble truncated integrin heterodimers α8tβ1t and α8tβ1-AP. The purified α8tβ1t heterodimer (lanes 1, 3, 6, 8 and 10), α8tβ1-AP heterodimer (lanes 2, 4, 7, 9, 11 and 12), ...

Comparisons of antigen blots using conditioned media, eluates of Ni-affinity chromatography, or highly purified α8tβ1-AP revealed that the COS-7 cell-conditioned medium contained 1.0–1.5 μg/mL of α8t. However, only 30–50% of the α8t was purified as secreted α8tβ1-AP heterodimer while the rest appeared to have been secreted as monomer. The monomers and heterodimers were recovered, respectively, in the flow-through and retained fractions of the YM100 membrane filtration at the last step of the purification. The final yield of heterodimers varied in each transfection, but was typically approximately 100 μg from 10 dishes (15 cm diameter). Once purified, the α8tβ1t and α8tβ1-AP heterodimers were stable during prolonged storage at 4 or −80 °C. The secreted α8t not associated with a β subunit seems most likely to have been monomeric and not aggregated since it passed through a YM100 membrane with a 100 kDa cut off, which was used to concentrate the α8tβ1t and α8tβ1-AP heterodimers.

Purified α8tβ1-AP Binds to FN and VN in Solid Phase Binding Assays

To determine whether purified α8tβ1t and α8tβ1-AP interact specifically with ECM ligands, we performed solid phase binding assays with highly purified heterodimers on purified ECM substrates. The binding properties of the purified heterodimers were directly compared to the ligand specificity of cell surface expressed α8β1 heterodimers as assessed using cell adhesion assays. For this purpose, adhesion assays were carried out in parallel with K562 cells and KA8 cells. K562 cells express the integrin α5β1; KA8 is a derivative of K562 cells, transfected with the chicken α8 subunit, that expresses both the α5β1 and α8β1 heterodimers. In addition, both KA8 and K562 cells have been shown to express low levels of αv-containing integrins (e.g., ref 5).

In solid phase binding assays, both purified α8tβ1t and α8tβ1-AP bound to FN, FN120, and VN, but not to FN40 and several other ECM molecules, including laminin-1, collagen I, collagen IV, recombinant entactin, and thrombospondin. Figure 4A shows the data obtained using α8tβ1-AP in the presence of Mn2+; similar data were obtained using α8tβ1t (not shown). Increasing of coating concentrations of substrate proteins resulted in similar data (not shown). Binding was observed in the presence but not absence of Mn2+ (not shown). Binding was inhibited by an RGDS-containing peptide and by a function-blocking mAb specific for the β1 subunit, indicating that binding was mediated by the ligand-recognition domain of the integrin heterodimer (Figure 4B). A virtually identical binding specificity was observed for the integrin α8β1 expressed on the cell surface of KA8 cells (Figure 4C). As K562 cells express α5β1 as the major integrin and very low amount of αvβ5 or other αv-integrins (5), the adhesion to FN and FN120 has been saturated in this assay and there is no detectable difference between KA8 and K562 cell adhesion to FN and FN120. These cells have been shown previously to bind to full-length FN and FN120 via the integrins α5β1 and α8β1 in an RGDS-peptide inhibitable manner (5). Some K562 cells adhere to VN in our assay conditions, reflecting the presence of low levels of αvβ5 and known VN receptor on these cells (5). KA8 cells which express also α8β1 exhibit enhanced attachment to VN, consistent with our evidence that purified α8tβ1-AP binds VN. KA8 cell adhesion to VN is mediated by α8β1 plus αv-containing integrins and is also sensitive to RGDS peptide competition.2 We observed no significant α8β1-mediated adhesion of KA8 cells to any of the substrates to which the purified α8tβ1-AP chimera did not bind and vice versa.

Figure 4
Comparisons of the binding specificities of the purified α8tβ1-AP chimera and cellular α8β1 to different purified ECM proteins. (A, B) Solid phase binding assays were performed with α8tβ1-AP in the presence ...

To summarize, we conclude that the truncated α8tβ1t and α8tβ1-AP heterodimers show similar ligand-binding properties to those of normal α8β1 expressed on the surface of K562 cells. This shows that neither subunit truncation nor fusion of AP to the C-terminus of the β1 subunit perturbs ligand–receptor interactions. Direct comparison revealed that the detection sensitivity and reproducibility of the detection assays were much higher with α8tβ1-AP than with α8tβ1t (not shown), presumably because α8tβ1t had to be chemically modified by biotinylation prior to use in solid phase binding assays. This may lead to variations in the degree of biotinylation and binding activity among different preparations (see also Discussion). Consistent with previous observations with other monomeric integrin α subunits, purified α8t monomer did not bind detectably to FN120 (not shown).

α8β1-Mediated Cell Adhesion and Spreading on Tenascin-C: Binding to Fragments Containing the Third FN-Type III Repeat

In contrast to previous reports (6, 8), using native chicken or human TN-C as a substrate, we did not observe significant α8β1-dependent interactions mediated by either KA8 cells or the purified α8β1 integrin chimera (Figure 4, panels A and C). Similar results were seen using various human and chicken TN-C preparations (including the same preparation as used in ref 8) coated either directly onto plastic or onto nitrocellulose-coated plastic. However, the same TN-C preparations actively promoted neurite outgrowth of embryonic chicken sensory neurons as described (8 and data not shown), demonstrating that the TN-C preparations employed here were biologically active. Since interactions of α8β1 are inhibited by RGDS-containing peptides and both human and chicken TN-C contain a single RGD site within the third FN type III-like repeat, purified fragments of chicken and human TN-C were assayed for possible interactions with this integrin. These fragments are illustrated in Figure 5. Results of binding assays using the purified integrin chimera and cell adhesion assays using α8β1-expressing KA8 cells are presented in Figure 6. In these assays, α8tβ1-AP was shown to bind strongly to fragments containing the third FN-type III repeat of either chicken or human TN-C the TNfn3, weakly to a fragment containing the third through fifth FN-type III repeats of human TN-C the TNfn3–5, but not detectably to native human TN-C or other TN-C fragments (Figure 6A and data not shown). Increasing the concentration (up to 100 μg/mL) of α8tβ1-AP in the assays did not result in specific binding. The binding of α8tβ1-AP to chicken TNfn3, human TNfn3, and human TNfn3–5 was inhibited by an RGDS peptide and by an anti-β1 mAb (Figure 6B). Similar results were obtained in assays of KA8 cell adhesion (Figure 6, panels C and D). Results in Figure 6C show that, in the presence of Mn2+, KA8 cells, but not K562 cells, adhered to chicken TNfn3 and human TNfn3 in a dose-dependent manner but adhered only weakly to human TNfn3–5. Neither cell line exhibited significant adhesion to other TN-C fragments and full-length TN-C (Figure 4C and data not shown). Neither Mg2+ nor Ca2+ promoted adhesion to TNfn3 in the absence of Mn2+ (not shown). Because α5β1 does not recognize TN-C, and probably because very low amount of αv integrins are not sufficient to support the cell adhesion, K562 cell adhesion was not detected in this assay. Consistent with α8tβ1-AP-mediated interactions, KA8 cell adhesion to chicken TNfn3 or human TNfn3 was inhibited by an RGDS-containing peptide and by a function-blocking anti-β1 mAb, but not by anti-α5 or anti-αvβ5 integrin mAbs (Figure 6D). In addition, KA8 cells, but not K562 cells, were observed to spread as effectively on human TNfn3 or chicken TNfn3 as on FN (Figure 7). We thus conclude that human and chicken TN-C have an RGD site in the third FN-type III repeat, which functions as a binding site for the integrin α8β1. However, this site does not serve as an efficient binding site for this integrin in native TN-C. Instead it appears to be cryptic, only exposed in small fragments containing the third FN-type III repeat. In a search for possible additional cryptic binding sites in other ECM proteins, we examined interactions of α8tβ1-AP and KA8 cells with recombinant entactin fragments, one of which contains an RGD site. We did not detect significant interactions with any of these fragments (not shown).

Figure 5
Schematic representation of TN-C and recombinant fragments used in the experiments for Figure 6. The primary structure of the human TN-C monomer is shown at the top of the figure. TN-C is a hexamer, and the monomer is composed of an N-terminal domain, ...
Figure 6
Mapping of the α8β1 binding site in TN-C: the third FN type III repeat containing RGD site is accessible only in short fragments. (A, B) Solid phase binding assays were performed with α8tβ1-AP in the presence of 1 mM MnCl ...
Figure 7
Cell spreading assay. In the presence of 1 mM MnCl2, K562 cells (A, C, E, G) and KA8 cells (B, D, F, H) were plated onto different substrata: (A, B) BSA; (C, D) FN (coated at 5 μg/ mL); (E, F) hu TNfn3 (5 μg/mL); (G, H) ch TNfn3 (20 μ ...


In this report, we described the efficient production and utilization of a soluble functional integrin-AP chimera for biochemical characterizations of integrin-ligand binding properties. The α8tβ1-AP chimera was found to mimic the cellular α8β1 in its “activation” and ligand-binding specificities. As previously reported for other integrins (22-27), the extracellular domains of α8 and β1 can be secreted as a functional heterodimer. Neither subunit association nor ligand binding require the presence of the cytoplasmic or transmembrane domains. In the previous reports using stably transfected cell lines, the amounts of integrins secreted was low (22, 24), and it was not practical to purify large amounts of the secreted integrins using antibody columns. Furthermore, to detect the integrins in biochemical assays, the proteins had to be further modified by either radiolabeling (22-27) or biotinylation (21), which can affect the ligand binding properties of the integrin (e.g., ref 21). We have overcome these difficulties by transient expression and affinity purification utilizing a (His)6-tag and Ni-affinity chromatography. With this approach, we were able to purify hundreds of micrograms of the secreted integrin; the AP-tagged integrin subunit allowed sensitive and quantitative detection of interactions with ligands in solid phase binding assays (this paper), in Far Western blotting, and in tissue staining (9, 44).

We have previously shown that integrin α8β1, when expressed on the surface of human K562 cells, can be activated to bind FN by either Mn2+ or an activating anti human β1 mAb (5). In the present paper, we have observed that this integrin, expressed on the surface of K562 cells, can also mediate binding to chicken or human TNfn3 fragments after activation by Mn2+ (Figure 6C). We have also found that the truncated heterodimer is initially inactive, but can be activated by Mn2+ to bind to FN, VN, and TNfn3 fragments. Unfortunately, we could not determine whether activating anti-β1 mAb is effective on the soluble, truncated integrin because the activating antibodies recognize only the human β1 subunit (expressed in K562 cells), but not the murine β1 subunit present in the secreted heterodimer. Nevertheless, we conclude that the binding specificity of the soluble integrin heterodimer closely reflects the binding properties of cell surface-expressed α8β1. This is further supported by observations that ligand binding by both soluble and cell-surface-associated α8β1 can be prevented by RGDS-containing peptides.

Although α8β1 is not constitutively activated as a soluble heterodimer or in the cells used for these experiments, it almost certainly is activated by physiological stimuli. In previous work, the activation state of an integrin heterodimer has been shown to be determined by the cell-type in which it is expressed and by intracellular signaling pathways (e.g., refs 28-32). Similar to other integrins, activation of cellular α8β1 is likely to require the presence of the transmembrane and cytoplasmic domain of each integrin subunit.

We characterized carefully interactions of the integrin α8β1 with human and chicken TN-C. Using both solid phase binding assays and cell adhesion assays, our data have confirmed previous experiments showing that this integrin binds to the TN-C domain TNfn3 (6). Bacterial expression proteins containing either human or chicken TNfn3 were effectively bound by the α8β1 integrin. Remarkably, binding was reduced when TNfn3 was included in a larger segment. TNfn1–5 supported no detectable binding, nor did native TN-C or any other fragments containing TNfn2. Joshi et al. (34) noted that the RGD segment is very close to the N-terminus of TNfn3 in the three-dimensional structure and suggested that the RGD is partially buried when TNfn2 is connected to TNfn3. They found that endothelial cells adhered strongly (probably through αvβ3) to TNfn3 and TNfn3–5, which have the RGD exposed at the N-terminus, but only weakly to TNfn1–3, TNfn1–5, and larger fragments containing TNfn2. Our results with α8β1 are very similar, except that we also observed reduced binding to TNfn3–5 relative to TNfn3. The reduced binding to TNfn3–5 suggests that the α8β1 integrin interacts with the C-terminal part of TNfn3 as well as the RGD sequence.

For integrins α8β1 and αvβ3, the binding site on TNfn3 appears to be cryptic, in that it is largely or completely masked in larger fragments and native TN-C. The crypticbinding site might be exposed if the TN-C were proteolytically cut in TNfn2, or alternatively if the tenascin arms were mechanically bent. A sharp bend at the interface of TNfn2 and TNfn3, and possibly also at TNfn3 and TNfn4, could expose the full cell adhesion site on TNfn3. There is no evidence for proteolysis in this region of TN-C, so mechanical bending may be a more attractive hypothesis for exposing the cryptic adhesion site. This is still only a speculation, since it is not known that the site is actually exposed in tissues.

Schnapp et al. (6) previously reported that α8-transfected 293 cells adhered and spread on human TN-C. The absorbance value for adhesion to native TN-C was only 25% of that to chicken TNfn3, a quantitative difference that was not discussed but is suggestive of weak adhesion. Varnum-Finney et al. (8) reported adhesion of KA8 cells to chicken TN-C purified from sterna, using a centrifugation assay that can detect very weak, transient cell adhesion. We could not demonstrate binding of purified integrin or cell adhesion in our present assays or her centrifugation assay using two preparations of human TN-C. The adhesion reported by Varnum-Finney et al. (8) may have been a very weak adhesion peculiar to chicken TN-C and the centrifugation assay. One possible explanation for the discrepancies between our results and those of these other workers is that a low level of proteolysis of TN-C, either during purification, storage, or assay, can expose the binding site present in the third FN-type III repeat.

What is the significance of the existence of a cryptic-binding site on TN-C for α8β1? As one possibility, this site may be exposed in physiological conditions by proteolysis or interactions with other proteins. TN-C has been shown to be expressed at many sites of tissue remodeling, both during development and in response to injury (e.g., refs 35-38), so fragments of TN-C that can be recognized by cells expressing α8β1 may well be generated during tissue remodeling. Activation of cryptic-binding sites appears likely to be of general significance in understanding interactions of cells with ECM constituents. Many ECM proteins have been found to contain such cryptic-binding sites for particular integrins, including fibrinogen and laminin. Laminin-1, for example, has been shown to contain cryptic-binding sites for both α1β1 and αvβ3 which are activated by heat or proteolysis (e.g., ref 39).

The importance of interactions of the integrin α8β1 with FN, VN, and TN-C in vivo remains uncertain. α8 mutant homozygotes have severe defects in kidney formation which result in early postnatal lethality (9). In contrast, the VN and TN-C genes have been mutated, but animals lacking each protein develop normally (41, 42). With the α8tβ1-AP chimera, it may be possible to identify the unidentified ligand(s) crucial for kidney morphogenesis and map histochemically regions of potential α8β1 function in vivo (9). There is additional evidence that both TN-C and α8 are expressed in smooth muscle cells and some neurons (7, 35, 42) and that TN-C expression is especially upregulated at these sites after injury (e.g., ref 43). Since various proteases are also upregulated and ECM components and tissue morphology change rapidly and dynamically after injury, the interaction of α8β1 with the RGD site of TN-C may modulate the tissue remodeling.

In summary, findings in this paper have characterized biochemical interactions of the integrin α8β1 with known ligands. The integrin-AP chimeras are easy to prepare, do not require further modification, and appear to be more sensitive and specific as probes of ligand–receptor interactions than biotinylated or iodinated integrins. Integrin-AP chimeras will be generally applicable for characterizing interactions of integrins with known ligands and for identifying additional receptor–ligand interactions with important physiological or developmental roles.


We thank Drs. R. Timpl, A. E. Chung, D. L. Mendrick, D. Vestweber, C. H. Damsky, S. Tominaga, and J. G. Flanagan for kindly providing reagents used in this study. We thank Dr. C. W. Turck for synthesizing the peptide used as an antigen.


This research was supported in part by National Institutes of Health Grant P01-16033 (L.F.R.). L.F.R. is an investigator of the Howard Hughes Medical Institute.

1Abbreviations: AP, alkaline phosphatase; RGDS, arginine, glycine, aspartic acid, serine; ECM, extracellular matrix; TN-C, tenascin-C; FN, fibronectin; VN, vitronectin; GST, glutathione-S-transferase; RT-PCR, reverse transcriptase-polymerase chain reaction; PMSF, phenylmethanesulfonyl fluoride; SDS–PAGE, sodium dodecyl sulfate–polyacrylamide gel electrophoresis; TNfn3, third fibronectin-type III repeat of tenascin-C; FN120, 120 kDa fragment of FN; TSP, thrombospondin-1; Col I, Collagen type I; Col IV, collagen type IV; LN-1, laminin-1; EGF, epidermal growth factor.

2U. Müller, unpublished experiments.

The sequence data are available from Genbank under accession number AF041409.


1. Hynes RO. Cell. 1992;69:11–25. [PubMed]
2. Sonnenberg A. Curr Top Microbiol Immunol. 1993;184:7–35. [PubMed]
3. Wolfsberg TG, Primakoff P, Myles DG, White JM. J Cell Biol. 1995;131(2):275–278. [PMC free article] [PubMed]
4. Bossy B, Bossy-Wetzel E, Reichardt LF. EMBO J. 1991;10:2375–2385. [PMC free article] [PubMed]
5. Müller U, Bossy B, Venstrom K, Reichardt LF. Mol Biol Cell. 1995;6(4):433–448. [PMC free article] [PubMed]
6. Schnapp LM, Hatch N, Ramos DM, Klimanskaya IV, Sheppard D, Pytela R. J Biol Chem. 1995;270(39):23196–23202. [PubMed]
7. Schnapp LM, Breuss JM, Ramos DM, Sheppard D, Pytela R. J Cell Sci. 1995;108(Part 2):537–544. [PubMed]
8. Varnum-Finney B, Venstrom K, Müller U, Kypta R, Backus C, Chiquet M, Reichardt LF. Neuron. 1995;14(6):1213–1222. [PMC free article] [PubMed]
9. Müller U, Wang D, Denda S, Meneses JJ, Pedersen RA, Reichardt LF. Cell. 1997;88:603–613. [PMC free article] [PubMed]
10. Flanagan JG, Leder P. Cell. 1990;63:185–194. [PubMed]
11. Cheng H-J, Flanagan JG. Cell. 1994;79:157–168. [PubMed]
12. Timple R, Rohde H, Robey PG, Rennard SI, Foidart J-M, Martin GR. J Biol Chem. 1979;254:9933–9937. [PubMed]
13. Phillips GR, Edelman GM, Crossin KL. Cell Adhes Commun. 1995;3(3):257–271. [PubMed]
14. Aukhil I, Joshi P, Yan Y, Erickson HP. J Biol Chem. 1993;268(4):2542–2553. [PubMed]
15. Aukhil I, Slemp CC, Lightner VA, Nishimura K, Briscoe G, Erickson HP. Matrix. 1990;10:98–111. [PubMed]
16. Hsieh JC, Wu C, Chung AE. Biochem Biophys Res Commun. 1994;199(3):1509–1517. [PubMed]
17. Tominaga S. FEBS Lett. 1988;238(2):315–319. [PubMed]
18. Hall DE, Reichardt LF, Crowley E, Holley B, Moezzi H, Sonnenberg A, Damsky CH. J Cell Biol. 1990;110:2175–2184. [PMC free article] [PubMed]
19. Lenter M, Uhlig H, Hamann A, Jeno P, Imhof B, Vestweber D. Proc Natl Acad Sci U S A. 1993;90(19):9051–9055. [PMC free article] [PubMed]
20. Mendrick DL, Kelly DM. Lab Invest. 1993;69(6):690–702. [PubMed]
21. DeFreitas MF, Yoshida CK, Frazier WA, Mendrick DL, Kypta RM, Reichardt LF. Neuron. 1995;15(2):333–343. [PubMed]
22. Berman PW, Nakamura GR, Riddle L, Chiu H, Fisher K, Champe M, Gray AM, Ward P, Fong S. J Cell Biochem. 1993;52(2):183–195. [PubMed]
23. Briesewitz R, Epstein MR, Marcantonio EE. J Biol Chem. 1993;268(4):2989–2996. [PubMed]
24. Dana N, Fathallah DM, Arnaout A. Proc Natl Acad Sci U S A. 1991;88:3106–3110. [PMC free article] [PubMed]
25. Nishimura SL, Sheppard D, Pytela R. J Biol Chem. 1994;269(46):28708–28715. [PubMed]
26. Weinacker A, Chen A, Agrez M, Cone RI, Nishimura S, Wayner E, Pytela R, Sheppard D. J Biol Chem. 1994;269(9):6940–6948. [PubMed]
27. Bennett JS, Kolodziej MA, Vilaire G, Poncz M. J Biol Chem. 1993;268(5):3580–3585. [PubMed]
28. O’Toole TE, Mandelman D, Forsyth J, Shattil SJ, Plow EF, Ginsberg MH. Science. 1991;254:845–847. [PubMed]
29. O’Toole TE, Katagiri Y, Faull RJ, Peter K, Tamura R, Quaranta V, Loftus JC, Shattil SJ, Ginsberg MH. J Cell Biol. 1994;124:1047–1059. [PMC free article] [PubMed]
30. O’Toole TE, Ylanne J, Culley BM. J Biol Chem. 1995;270:8553–8558. [PubMed]
31. Filardo EJ, Cheresh DA. J Biol Chem. 1994;269:4641–4647. [PubMed]
32. Hughes PE, O’Toole TE, Ylanne J, Shattil SJ, Ginsberg MH. J Biol Chem. 1995;270:12411–12417. [PubMed]
33. Prieto AL, Edelman GM, Crossin KL. Proc Natl Acad Sci U S A. 1993;90:10154–10158. [PMC free article] [PubMed]
34. Joshi P, Chung C-Y, Aukhil I, Erickson HP. J Cell Sci. 1993;106:389–400. [PubMed]
35. Erickson HP, Bourdon MA. Annu Rev Cell Biol. 1989;5:71–92. [PubMed]
36. Chiquet-Ehrismann R. Semin Cancer Biol. 1993;4:301–310. [PubMed]
37. Mackie EJ, Halfter W, Liverani D. J Cell Biol. 1988;107:2757–2767. [PMC free article] [PubMed]
38. Laywell ED, Dorries U, Bartsch U, Faissner A, Schachner M, Steindler DA. Proc Natl Acad Sci U S A. 1992;89:2634–2638. [PMC free article] [PubMed]
39. Colognato-Pyke H, O’Rear JJ, Yamada Y, Carbonetto S, Cheng Y-S, Yurchenko PD. J Biol Chem. 1995;270:9398–9406. [PubMed]
40. Saga Y, Yagi T, Ikawa Y, Sakakura T, Aizawa S. Genes Dev. 1992;6:1821–1831. [PubMed]
41. Zheng X, Saunders TL, Camper SA, Samuelson LC, Ginsburg D. Proc Natl Acad Sci U S A. 1995;92:12426–12430. [PMC free article] [PubMed]
42. Einheber S, Schnapp LM, Salzer JL, Cappiello ZB, Milner TA. J Comput Neurol. 1996;370:105–134. [PubMed]
43. Hedin U, Holm J, Hansson GK. Am J Pathol. 1991;139:649–656. [PMC free article] [PubMed]
44. Denda S, Reichardt LF, Müller U. Mol Biol Cell. 1998 in press.
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...