Evidence of a Functional Role for Interaction between ICAM-1 and Nonmuscle α-Actinins in Leukocyte Diapedesis

Abs and reagents

All organic chemicals were obtained from Sigma-Aldrich unless otherwise specified. Restriction enzymes and DNA-modifying enzymes were obtained from Invitrogen Life Technologies. Synthetic oligonucleotides were prepared by Genosys Biotechnologies. The following mAbs were used: mouse mAbs 3D6 and 2D5 to human ICAM-1 were produced in our laboratory, as described previously (21); mouse mAb directed against α-actinin 1 (BM-75.2) was obtained from Sigma-Aldrich. In some experiments, 3D6 was labeled with cy3, according to the manufacturer’s instructions (Amersham Biosciences). For α-actinin 4, initial experiments were performed using a rabbit polyclonal anti-α-actinin 4 Ab (a gift from S. Hirohashi, National Cancer Research Institute, Pathology Division, Tokyo, Japan; Ref. 22); for all experiments described in this study, we used a rabbit polyclonal anti α-actinin 4 Ab (LC17) raised against a keyhole limpet hemocyanin-conjugated synthetic peptide, MVDYHAANQSYQYGPS (containing N-terminal aa 1–16 of α-actinin 4; Covalab). The specificity of this polyclonal Ab was verified by immunoblot and immunofluorescence (see Results). A rabbit antiserum against actin (Sigma-Aldrich) was used for quantification and normalization of Western blots. Texas Red-phalloidin from Molecular Probes was used to label F-actin.

Yeast two-hybrid assays

A cDNA fragment encoding the complete cytoplasmic portion plus 6 aa from the transmembrane domain of human ICAM-1 (residues 1509–1614) was amplified by PCR. The following primers were used: 5′-GAATTCCTCAGCACGTACCTCTATAAC-3′ (downstream) and 5′-CGGGATCCTCAGGGAGGCGTGGCTTGTGTGTT-3′ (upstream). The primers were designed to include an EcoR1 site for the upstream primer and a BamH1 site for the downstream primer. The resulting PCR product was verified by sequencing and subcloned into pAS2–1. A human placenta cDNA library, cloned into the pACT2 vector fused with the GAL4 activation domain, was purchased from BD Clontech. The yeast strain Y190 (BD Clontech) was sequentially transformed with pAS2/ICAM-1 as bait, then with a pACT2 plasmid containing the cDNA library. Transformants were grown on synthetic dextrose medium lacking the amino acids leucine, tryptophan, and histidine in the presence of 25 mM 3-amino-1.2.4-triazole. On day 5, colonies were tested for the activity of the lacZ reporter gene with a β-galactosidase (β-Gal) filter assay. All pAS2-1 subclones were analyzed by DNA sequencing to confirm the correct in frame fusion of GAL4 and ICAM-1 cytoplasmic sequence.

A mating assay was performed to eliminate false positive; clones scored as positive after the screening were cultured on synthetic dextrose medium without tryptophan in the presence of 10 μg/ml cycloheximide. The cycloheximide-resistant Y190 yeast clones were verified in a mating assay with yeast strain expressing the pAS2 plasmid with either ICAM-1, the unrelated protein lamin C (pLAM5′-1 vector), or the nonfused GAL4 DNA-binding domain as baits. Positive clones were further retested in a cotransformation assay with purified plasmid cDNA combined to a range of plasmid controls: pLAM5′-1, in pAS2-1, pACT2, pVA3-1, and pTD1-1 (BD Clontech).

To quantify two-hybrid interactions and compare the relative strength of protein-protein interactions observed, an O-nitrophenyl-β-d-galactopyranoside assay was used (23).

Inserts from positive cDNA clones were sequenced by the T7 Sequenase chain-termination DNA sequencing method using the T7 Sequenase version 2.0 DNA sequencing kit from Amersham Biosciences. The sequences were analyzed by the basic local alignment search tool program, at National Center for Biotechnology Information (〈www.ncbi.nlm.nih.gov/BLAST/〉).

GFP and ICAM-1 mutant constructs

A cDNA encoding full-length ICAM-1 was cloned into the pEGFP-N1 vector (BD Clontech) to obtain a fusion protein expressing GFP at the C terminus of ICAM-1. ICAM-1 cytoplasmic mutants: CHO ICAM-1 R480A, K481A, K483A, K484A, and R486A were generated using the QuickChange Site-Directed Mutagenesis kit (Stratagene). Procedure was performed to mutate each charged amino acid selected on ICAM-1 into the neutral amino acid alanine using the pcDNA3 plasmid (Invitrogen Life Technologies) containing the full-length human ICAM-1 coding sequence.

Production of GST fusion proteins and in vitro translation

Full-length α-actinin 1 was obtained by performing a RT-PCR from HeLa cells and full-length α-actinin 4 was ordered from Incyte Genomics. Both cDNAs were sequenced by Genaxis Biotechnology and subcloned into pGEX 5X3 plasmid (Amersham Biosciences). GST fusion proteins were expressed in BL21 Escherichia coli. After lysis by sonication, GST fusion proteins were purified using glutathione-Sepharose 4B beads according to the manufacturer’s protocol (Amersham Biosciences).

In vitro transcription and translation experiments were performed using T7 TnT Quick coupled transcription/translation system from Promega. α-Actinin polypeptides were synthesized from full-length human α-actinin cDNAs subcloned in pcDNA3-1 plasmid.

Cell culture and transfection experiments

HUVECs were isolated from human umbilical cords by collagenase treatment as described by Jaffe et al. (24), and cultured as previously described (6). In some experiments, HUVECs were stimulated with 200 U/ml TNF-α (Boehringer Mannheim) for 4 h at 37°C.

CHO cells were cultured in DMEM (BioWhittaker) containing 5% FBS (Invitrogen Life Technologies), 2 mM l-glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin and nonessential amino acids (BioWhittaker).

Obtention of stable CHO cell lines expressing ICAM-1 or ICAM-1 cytoplasmic deletion mutant (ΔCyt) has been previously described (6). CHO cell lines expressing ICAM-1-GFP or point mutations on ICAM-1 cytoplasmic portion were obtained by transfecting CHO cells by electroporation. Selection of stable transfectants was performed in DMEM medium containing 1 mg/ml G418 (geneticin; Invitrogen Life Technologies) for selection or 500 μg/ml for routine culture (25). Phenotypic characterization of transfectants was conducted by flow cytometry and immunofluorescence staining on adherent cells with anti-ICAM-1 (3D6) mAb.

Polymorphonuclear neutrophils (PMN) used for transmigration assays were isolated by differential centrifugation on Ficoll-Hypaque gradient (Sigma-Aldrich), followed by dextran sedimentation and hypotonic lysis of erythrocytes.

siRNA transfection of HUVEC

Predesigned annealed siRNAs targeting α-actinin 1 or α-actinin 4 were purchased from Eurogentec, together with a siRNA-negative control. HUVECs were seeded in 6-well dishes at a density of 1.6 × 10⁵ cells/well 24 h before the experiment. Cells were transfected with siRNAs directed to human α-actinin 1 (5′-CACCAUGCAUGCCAUGCAA-3′ at position 852 of α-actinin 1 mRNA, starting from AUG start codon, siRNA Actn1_pos852), or human α-actinin 4 (5′-GCAGCAGCGCAAGACCUUC-3′ at position 150, siRNA Actn4_pos150; Ref. 26) (300 nM each) or a negative control siRNA, known to have no effect on mammalian protein expression. Oligofectamine (Invitrogen Life Technologies) was used to transfect siRNAs according to the manufacturer’s instructions. Transfection was conducted in 1 ml OptiMEM (Invitrogen Life Technologies). Four hours after transfection, cells were supplemented with additional 3 ml of EC growth medium and cultured for another 3 days. Down-regulation of actinin protein was verified by Western blotting, after lysis of the cells. Samples were resolved on a 7.5% SDS-PAGE and blotted. Actinins were detected using anti α-actinin 1 and α-anti-actinin 4-specific Abs. Blots were also probed with an anti-actin Ab to monitor equal loading.

Precipitation of overexpressed proteins

CHO cell lines expressing native ICAM-1 or ICAM-1 mutants were lysed with immunoprecipitation (IP) buffer (PBS containing 0.5% Triton X-100, 0.5% Nonidet-P40, complete protease inhibitors (Roche)) on ice for 30 min. After centrifugation, supernatant (300 μg of protein) was incubated for 3 h at 4°C with glutathione-agarose beads preadsorbed with 50 μg of GST or GST-actinin 1 and 4. Beads were washed three times for 10 min with IP buffer; bound proteins were eluted with SDS-PAGE loading buffer and subjected to 7.5% SDS-PAGE. After blotting onto nitrocellulose, ICAM-1 was detected using mAb 2D5 at 2 μg/ml.

IP of endogenous proteins in ECs

Confluent cultures of HUVECs were treated with TNF-α for 4 h and then lysed with IP buffer. To immunoprecipitate ICAM-1 or α-actinin 1, protein A-Sepharose 6MB coated with rabbit IgG anti-mouse was incubated with anti ICAM-1 mAb (2D5) or anti α-actinin 1 mAb. For α-actinin 4 IP, rabbit polyclonal anti α-actinin 4 LC17 was directly incubated with protein A-Sepharose. ICAM-1, α-actinin 1 or α-actinin 4 were immunoprecipitated from HUVEC lysates, and coprecipitation of associated proteins was revealed by Western blotting with specific Abs against ICAM-1, α-actinin 1, or α-actinin 4.

Flow cytometry analysis

Expression level of ICAM-1 was analyzed on a FACScan (BD Biosciences), as described previously (6).

Immunofluorescence staining and confocal imaging

HUVECs or CHO cells were seeded on glass coverslips precoated with 25 μg/ml human plasma fibronectin (27) and processed as described previously (6). Cells were stained with the following primary Abs: anti ICAM-1 3D6 (5 μg/ml), anti-α-actinin 1 (1/100 dilution of ascites), anti-actinin 4 (4 μg/ml). F-actin was revealed with Texas Red-conjugated phalloidin. Images of HUVECs and CHO cells were acquired using an Olympus inverted microscope (AX70).

Zeiss confocal microscope was used to perform cross-sectioning of cells. To combine the entire Z series into a stack projection and to accomplish cross-sectional reconstruction, we used the Metamorph software (Universal Imaging). The percentage of the proteins that colocalize with each other was measured with the Zeiss software. Statistical analysis with the software determined the Pearson’s correlation (PC), which compares the colocalization of two different types of fluorescence and their intensities; the higher the fluorescence intensities are at the site of colocalization, the higher the PC will be, with a maximum of 1.

Permeability assay

Permeability assays to determine monolayer permeability of CHO cell lines were proceeded as described in our previous study (6).

Transmigration assays

CHO cell monolayers were grown to confluency on fibronectin-coated porous membranes (6.5-mm diameter and 5-μm pore size; Costar) for 3 days at 37°C. CHO cells were then incubated in duplicate with PMN (2 × 10⁵/well); to maintain the viability of human PMN, autologous normal human plasma (1/5 dilution in DMEM/HEPES 20 mM (pH 7.4)) was added with the PMN in the upper compartment in the presence of 100 μM PPACK (Boehringer), a specific inhibitor of thrombin. A concentration of 2 × 10⁻⁸ M fMLP in DMEM/HEPES 20 mM (pH 7.4; Calbiochem-Novabiochem) was added in the lower compartment, to create a chemotactic gradient for PMN. Migrated neutrophils were recovered from the bottom of the well after 3 h at 37°C in 5% CO₂, centrifuged, stained with trypan blue, and counted microscopically.

Statistics

Statistical differences between experimental groups were evaluated using the Student t test.

Identification of cellular proteins interacting with ICAM-1

To identify intracytoplasmic proteins interacting with ICAM-1 cytoplasmic domain, we used a yeast two-hybrid system (Matchmaker Gal4 Two-Hybrid). A bait construct (pAS2–1/ICAM-1) encoding the human cytoplasmic domain of ICAM-1 was designed to screen a human placenta cDNA library (>5 × 10⁶ independent clones). Human placenta is a heavily vascularized tissue, and proteins expressed by ECs including ICAM-1 are well-represented. Using this system, 47 clones were detected as positive candidates, of which 14 were confirmed positive after extensive testing and DNA sequencing. These 14 clones were sequenced and were found to correspond to the following genes: α-actinin 1, α-actinin 4, Bin 1, MAGE-4b, succinate dehydrogenase, and the T cell activation protein. In this study, we focused our attention on α-actinin 1 and α-actinin 4.

All the clones coding for α-actinins corresponded to the C-terminal two-thirds of these proteins (residues 501–914 for α-actinin 1, and 544–930 for α-actinin 4). The specificity of ICAM-1/α-actinin 1 and α-actinin 4 interactions were confirmed by two-hybrid mating assays using purified pACT2 α-actinin 1 and 4 in association with pAS2-1 ICAM-1 as positive control, pTD1-1 encoding large T-Ag, pVA3-1 encoding murine p53, and pLAM5-1 encoding human lamin C as negative controls (Table I⇓).

Specificity of the polyclonal Ab anti-α-actinin 4

Rabbit polyclonal Ab LC17 was raised against the NH2-terminal amino acid sequence of α-actinin 4. The immunoblot in Fig. 1⇓A, using full-length α-actinins produced in vitro, shows that this polyclonal Ab reacted only with the in vitro translation product of α-actinin 4 cDNA (lane 4) but not with α-actinin 1 (lane 3), thus demonstrating its specificity. mAb (BM75.2) against α-actinin 1 was used as a control. As shown in Fig. 1⇓A, this Ab was reactive only with the in vitro translation product of human α-actinin 1 cDNA (lane 1) but not with human α-actinin 4 (lane 2).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

A, Reactivity of polyclonal Ab LC17 raised against α-actinin 4 peptide. In vitro translation products of full-length α-actinin 1 (lanes 1 and 3) and α-actinin 4 (lanes 2 and 4) were immunoblotted with mAb BM75.2 specific to α-actinin 1 (lanes 1 and 2) or with polyclonal Ab LC17 raised against α-actinin 4 (lanes 3 and 4). B, Confocal microscopy in HUVECs showing actin and α-actinin localization. Actin was detected by Texas Red-conjugated phalloidin (panels 1 and 4 and in red in panels 3 and 6). α-Actinin 1 was stained by BM75.2 mAb (panel 2 and in green in panel 3), and α-actinin 4 was detected by polyclonal Ab LC17 (panel 5 and in green in panel 6). Z projections of stacks are shown. α-Actinin 1 is localized at the end of actin stress fibers (arrowheads in panel 3), and α-actinin 4 colocalized with actin F (panel 6). α-Actinin 4 was also detected at low levels in the nucleus (arrowhead in panel 5). C, Localization of ICAM-1 and α-actinins in HUVECs in a confocal section. HUVECs were first treated with 200 U TNF-α for 4 h to increase ICAM-1 expression. ICAM-1 expression was detected using Cy3-labeled mAb anti-ICAM-1 (panels 1 and 4 and in red in panels 3 and 6), and actinins were revealed by double immunofluorescence with either α-actinin 1 stained by BM75.2 mAb (panel 2 and in green in panel 3) or α-actinin 4 detected by LC17 Ab (panel 5 and in green in panel 6). α-Actinin 1 colocalized with ICAM-1 particularly at the edge of the cell (panel 3). α-Actinin 4 also colocalized with ICAM-1 in structures that look like focal adhesions (arrowheads in panel 6). D, Localization of ICAM-1 and α-actinins in CHO cells expressing ICAM-1-GFP. A confocal section is represented. ICAM-1 expression is illustrated on panels 1, 4, 3, and 6. α-Actinin 1 was stained by mAb BM75.2 (panel 5 and in red in panel 6), and α-actinin 4 was detected by polyclonal Ab LC17 (panel 2 and in red in panel 3). In CHO cells, α-actinin 1 showed diffuse cytoplasmic expression (panel 6), while α-actinin 4 was colocalized with ICAM-1 at the edge of the cell membrane (arrowhead in panel 3). Bar, 10 μm.

Localization of α-actinins in EC and CHO cells

The cellular localization of α-actinins in HUVECs was studied by immunofluorescence microscopy. α-Actinin 1 was expressed along actin fibers and at adherens junctions, at the end of stress fibers (Fig. 1⇑B, panels 1–3), as already described by other groups (22, 28), and was colocalized with ICAM-1 at the border of the cell (Fig. 1⇑C, panels 1–3), suggesting an interaction between both proteins in vivo in HUVECs. α-Actinin 4 was expressed in a punctate fashion along actin fibers, and near the cell membrane (Fig. 1⇑B, panels 4–6). Faint α-actinin 4 expression was also observed in the nucleus (Fig. 1⇑B, panel 5, white arrow), as previously reported by Honda et al. (22). ICAM-1 was found to be colocalized with α-actinin 4 at the periphery of the cells (Fig. 1⇑C, panels 4–6, white arrows), suggesting a possible interaction between both proteins.

We then studied the localization of endogenous α-actinins in CHO cells transfected with ICAM-1-GFP. Both α-actinins were detected in CHO cells, with α-actinin 1 (Fig. 1⇑D, panel 5) and α-actinin 4 (Fig. 1⇑D, panel 2) detected by mAb BM-75.2 and polyclonal Ab LC17, respectively. A very high level of colocalization between ICAM-1 and α-actinin 1 was observed (Fig. 1⇑D, panel 6), with a strong expression of α-actinin 1 at the periphery of the cell and at sites of cell/substrate interaction. The percentage of colocalization was between 60 and 80% as determined by software analysis on confocal slices, with a PC of 0.52. Actinin 4 was expressed in these cells with localization similar to the one of ECs, on actin fibers and near the cell surface (Fig. 1⇑D, panel 2). Some colocalization could be observed with ICAM-1 at the cell surface (Fig. 1⇑D, panel 3, white arrows). The percentage of colocalization for α-actinin 4 and ICAM-1 was from 30 to 40%, with a PC of 0.1. This experiment showed that CHO cells express α-actinins and that the localization of transfected human ICAM-1 is similar to the one found in HUVECs.

Cell density and TNF-α modify the cellular localization of α-actinin 4

α-Actinin 4 was first identified in epithelial cells (22), where it was shown to be localized to the cytoplasm and to the nucleus. In HUVECs, α-actinin 4 was expressed along actin fibers and near the cell surface, with only a very faint labeling inside the nucleus (Fig. 1⇑). Thus, we tested the effect of cell density and TNF-α treatment on α-actinin 4 cellular localization. In either sparse or confluent resting cells, α-actinin 4 was expressed in the cytoplasm or near the cell surface with little, if any, protein expression inside the nucleus. In contrast, after TNF-α treatment, α-actinin 4 was also present inside the nucleus, in addition to its localization in the cytoplasm (Fig. 2⇓). In resting confluent HUVECs, α-actinin 4 was strongly expressed at cell-cell contact sites (Fig. 2⇓C), but after treatment with TNF-α, the majority of α-actinin 4 was redistributed to the cytoplasm and the nucleus (Fig. 2⇓D). In contrast, the distribution of α-actinin 1 was unchanged under these conditions (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

Effect of cell density and TNF-α on subcellular localization of α-actinin 4 in HUVECs. HUVECs at low density (A and B) or at confluency (C and D) were either treated for 6 h with 100 U/ml TNF-α (B and D) or left untreated (A and C). After fixation and permeabilization, α-actinin 4 was detected by double immunofluorescence. Bar, 10 μm.

α-Actinin 1 and α-actinin 4 interact with ICAM-1 both in vitro and in eukaryotic cell

ICAM-1 interaction with α-actinins was studied in vitro using fusion proteins containing α-actinin sequences isolated in the two hybrid experiment (residues 501–914 for α-actinin 1, and 544–930 for α-actinin 4) with GST-tag. These fusion proteins were bound to glutathione-agarose beads. These beads were incubated with a cell lysate from CHO-expressing ICAM-1, the protein complexes were then resolved by SDS-PAGE, and ICAM-1 interacting with GST-actinin was detected by immunoblot. ICAM-1 was found associated with α-actinin 1 (Fig. 3⇓A, lane 3) and with α-actinin 4 (Fig. 3⇓A, lane 4). In contrast, ICAM-1 was not precipitated by GST protein alone (Fig. 3⇓A, lane 1). These results indicated that ICAM-1 interacted with α-actinin 1 and with α-actinin 4 in vitro.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

ICAM-1 interactions with nonmuscle α-actinins. A, In vitro pull-down experiment. Fusion proteins GST-actinin 1 and 4 were generated with sequences identified in two-hybrid screening. GST-actinin 1 (lane 3) or GST-actinin 4 (lane 4) were linked to glutathione-agarose beads and incubated with an extract of a CHO cell line expressing ICAM-1. Proteins attached to agarose beads were subjected to SDS-PAGE and transferred to nitrocellulose membranes. ICAM-1 was detected with a mAb (2D5) against ICAM-1. Glutathione agarose-bound GST was used as negative control (lane 1), and a lysate of a CHO cell line expressing ICAM-1 was used as positive control (lane 2). Lanes 5–7 are GST load controls corresponding, respectively, to GST, GST-actinin 1, and GST-actinin 4, revealed with a GST mAb. B–D, Biochemical evidence that ICAM-1 and nonmuscle actinins interact in ECs forming a complex which can be isolated by both anti-ICAM-1 and anti-actinin Abs. B, Abs against ICAM-1 (lane 2, positive control); actinin 1 (lane 4); actinin 4 (lane 5) were bound to protein A-Sepharose beads and incubated with EC extracts (HUVEC). Immunoprecipitated proteins were analyzed by SDS-PAGE and Western blot with mAb 2D5 specific for ICAM-1, together with an EC extract (lane 1). In lane 3, PECAM-1 mAb 5F410 (Ref. 6 ) was used as negative control. C, Abs against actinin 1 (lane 2, positive control); ICAM-1 (lane 4) bound to protein A-Sepharose beads were incubated with EC extracts (HUVEC). Immunoprecipitated proteins were analyzed by SDS-PAGE and Western blot with mAb BM75.2 directed against actinin 1, together with an EC extract (lane 1). Lane 3, PECAM-1 mAb 5F410 was used as negative control. D, Abs against actinin 4 (lane 2, positive control); ICAM-1 (lane 4) bound to protein A-Sepharose beads were incubated with EC extracts (HUVEC). Immunoprecipitated proteins were analyzed by SDS-PAGE and Western blot with a rabbit antiserum against actinin 4, together with an EC extract (lane 1). Lane 3, PECAM-1 mAb 5F410 was used as negative control.

To provide evidence that ICAM-1 and actinins associate in vivo, IP experiments were performed. Extracts of HUVECs were prepared from cells induced by TNF-α for 6 h to increase ICAM-1 expression. As shown in Fig. 3⇑B, Abs against α-actinin 1 (lane 4) or α-actinin 4 (lane 5) precipitated ICAM-1. The mAb 2D5 was also able to coprecipitate α-actinin 1 and α-actinin 4, respectively (Fig. 3⇑, C and D). These results demonstrated that ICAM-1 interacts in vivo with endogenous α-actinin 1 and 4 in eukaryotic cells.

Identification of amino acids on ICAM-1 cytoplasmic domain involved in ICAM-1/nonmuscular α-actinin interactions: generation of ICAM-1 mutants

To establish the functional role of these interactions, we first identified amino acids involved in ICAM-1/α-actinin interactions. A 5-aa sequence (480-RKIKK-484) has been previously shown to be involved in ICAM-1 interaction with α-actinin (11) (Fig. 4⇓A). Thus, we generated point mutations by switching to alanine the 4 positive-charged aa of this short sequence (one arginine and three lysines). In addition, we mutated arginine 486, another charged amino acid located near the short peptide 480–484 (Fig. 4⇓A). The influence of each mutation was assessed using a quantitative β-Gal assay in a yeast two-hybrid system (Matchmaker Gal4 Two-Hybrid). This method allowed us to quantify interactions and to compare the relative strength of protein-protein interactions. In this study, we compared the interaction level between ICAM-1, considered as positive control, and each mutant (ICAM-1 R480A, K481A, K483A, K484A, and R486A) with both α-actinin 1 and 4. As shown in Fig. 4⇓B, among the 5 aa mutated on ICAM-1, only mutations on arginine 480, 486, and lysine 481 reduced at least by 2-fold ICAM-1 interaction with α-actinin 1 and 4, while mutations on arginine 483 and 484 were without effect. These observations indicated that the binding of both α-actinins on ICAM-1 depended upon the same amino acids on ICAM-1 cytoplasmic tail.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

A, ICAM-1 cytoplasmic sequence 478–505. Underlined, the sequence identified by Carpen et al. (11 ) to be relevant in the interaction with α-actinin. The arrows indicate the five-point mutations that were generated by site-directed mutagenesis: Two arginines in positions 480 and 486 and three lysines in positions 481, 483, and 484 were mutated into alanine. B, Percentage of interaction of ICAM-1 or each five mutant with nonmuscle α-actinin 1 and 4. Quantitative β-Gal assay was performed to compare the relative strength of protein-protein interactions observed. White histograms correspond to ICAM-1 interaction with α-actinin 1 and the gray ones with α-actinin 4. As positive control, we considered ICAM-1 interaction with each α-actinin as 100%. Data represent mean ± SD of three separate experiments. Student t test: ∗ (p < 0.0001).

Characterization of rCHO cell lines expressing ICAM-1 mutants

The direct binding of ICAM-1 to α-actinins may affect its function. To test this hypothesis, we developed CHO cell lines stably expressing ICAM-1 mutants, and tested these cell lines using the reconstituted mammalian cell expression model already engineered in our laboratory (6). We first characterized the expression level and localization of these mutant proteins. Expression levels of ICAM-1 mutants in the five recombinant cell lines were quantified by FACScan analysis and compared with the CHO ICAM-1 cell line. As shown in Fig. 5⇓A, the selected CHO cell lines expressed each ICAM-1 mutant at a level comparable to CHO ICAM-1. This expression level has already been defined to be similar to TNF-α-stimulated HUVECs, a well-known model of ECs usually used in transmigration assay (6). Cellular distribution of ICAM-1 mutants in these cell lines was analyzed by immunofluorescence staining, followed by confocal analysis. Results in Fig. 5⇓B clearly show that the cellular localization of the various mutant ICAM-1 proteins was not altered when compared with the wild-type (WT) protein.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5.

Characterization of CHO cell lines expressing ICAM-1 mutants. A, FACScan analysis of ICAM-1 mutant expression in comparison to ICAM-1 used as positive control and CHO WT as negative control. Experiments were conducted as described in Materials and Methods. B, Localization of ICAM-1 investigated by immunofluorescence staining and confocal imaging on CHO cells expressing ICAM-1 and mutants. Cells were allowed to spread on fibronectin matrix (5 μg/ml) before fixation and staining with monoclonal anti-ICAM-1 (20 μg/ml). Cells were then imaged using confocal microscopy. Upper panels, Z-projections of confocal slices; lower panels, reconstruction in the z-axis to provide representative cross-sectional image. Bar, 10 μm. C, Interaction of ICAM-1 mutants with nonmuscle α-actinins in vitro. Fusion proteins GST-actinin 1 and 4 were generated with full-length sequences. Equal amount of GST-actinin 1 or 4 were linked to glutathione-agarose beads and incubated with 300 μg of CHO cell lines extracts expressing ICAM-1 or ICAM-1 mutants. Detection of ICAM-1 interaction with GST-actinin was analyzed by immunoblotting using an ICAM-1-specific mAb (2D5). (−), Negative control with lysates from CHO cells expressing ICAM-1 incubated with GST protein alone. (+), Positive control, GST-actinins interaction with lysates from CHO cells expressing ICAM-1. The level of interaction between GST-actinins with the different ICAM-1 mutants was quantified on immunoblots using Image J software. GST-actinin interaction with native ICAM-1 was considered as 100%.

We then used these cell lines to test the interaction of the different ICAM-1 mutants with α-actinins. Full-length α-actinins fused with a GST-tag were bound on glutathione-agarose beads. Beads were then incubated with lysate from each rICAM-1 mutant CHO cell line, and ICAM-1 mutant interaction with GST-actinin was analyzed by immunoblot with mAb 2D5. The quantification of immunoblot signals that represents levels of interaction between GST-actinin with the different ICAM-1 mutants is shown in Fig. 5⇑C. This showed that the pattern of interaction of the different ICAM-1 mutants expressed on recombinant cells with α-actinins obtained in this assay was similar to the one obtained in the two-hybrid test represented in Fig. 4⇑B. Despite the fact that results were less significant for α-actinin 1 than for α-actinin 4, again we found that arginine 480, arginine 486, and lysine 481 were essential for ICAM-1 interaction with both α-actinins. Thus, we confirmed, using full-length α-actinins and ICAM-1 mutants from rCHO cell lines, that arginine 480, 486, and lysine 481 are involved in ICAM-1/α-actinin interactions. Based on these results, these rCHO cell lines were used to investigate the role of ICAM-1/actinins interactions in transmigration.

fg-dependent transmigration requires ICAM-1/nonmuscle α-actinin interactions

All the CHO cell lines have been tested for their ability to form a confluent monolayer. The integrity of the monolayers and their function as barriers were studied by a permeability assay using HRP (29). Three days after seeding, resting and TNF-α-stimulated HUVECs presented permeability levels of 19.4 ± 0.1% and 21.6 ± 0.1% of controls (Fig. 6⇓). All the rCHO cell lines expressing ICAM-1, ICAM-1ΔCyt, and mutants (R480A, K481A, K483A, K484A, and R486A) formed monolayers that were nearly impermeable to HRP, with 12.0 ± 3.9%, 9.1 ± 0.8%, 9.8 ± 1.9%, 10.1 ± 5.6%, 16.4 ± 1.9%, 5.7 ± 0.6%, and 12.2 ± 3.3% of controls, respectively; these values are similar to those obtained with WT CHO monolayers (11.9 ± 1.1%) (Fig. 6⇓). These results showed that CHO cells are able to form correct, continuous monolayers with a high cohesion of cell-cell contacts, and a permeability level similar to that observed with HUVECs.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 6.

Permeability analysis of CHO monolayers expressing ICAM-1, ICAM-1ΔCyt, or ICAM-1 mutants in comparison with resting or TNF-α-stimulated HUVECs. HRP permeability was referred as percentage of control (+), corresponding to fibronectin-coated Transwell without cells. Data are expressed as mean ± SD of three separate experiments.

The implication of ICAM-1 and its cytoplasmic tail in fg-dependent transmigration have already been defined in our laboratory (6). To assess whether ICAM-1 cytoplasmic interaction with α-actinins was involved in this process, we performed an in vitro transmigration assay of neutrophils across the different CHO cell lines monolayers. As shown in Fig. 7⇓, when ICAM-1/α-actinin interactions were altered, the number of neutrophils transmigrating across CHO monolayers was reduced to a level similar to CHO WT or CHO ICAM-1ΔCyt (14.22 ± 3.32% for ICAM-1 R 480A, 14.92 ± 5.25% for ICAM-1 K481A, and 16.2 ± 3.54% for ICAM-1 R486A vs 13.33 ± 2.17% for CHO WT and 14.94 ± 1.97% for ICAM-1ΔCyt). To confirm that these results were due to an inhibition of ICAM-1/α-actinin interactions, we used an ICAM-1 K484A mutant, which does not disturb these interactions. The level of transmigrated PMN across CHO ICAM-1 R484A monolayer was similar to CHO ICAM-1 (38.31 ± 6.04% vs 42.33 ± 3.76%). These experiments indicated that ICAM-1/α-actinin interactions were implicated in fg-dependent PMN transmigration.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 7.

Functional role of ICAM-1/nonmuscle α-actinin interactions. Effect of ICAM-1 mutations on PMN transmigration across CHO monolayers. Transmigration assay was performed as described in Materials and Methods. Data represent mean ± SD of three separate experiments in duplicate. Student t test: ∗ (p < 0.0001). For Student t test, CHO ICAM-1 was compared with CHO WT and each CHO ICAM-1 mutant to CHO ICAM-1.

α-Actinin 4 depletion affects neutrophil diapedesis

To test which isoform of α-actinin was involved in diapedesis, we used siRNA to specifically silence expression of either α-actinin 1 or α-actinin 4 protein in HUVEC and then assayed these cells for leukocyte diapedesis. siRNA Actn4_pos150 has been shown to be specific for α-actinin 4 by Yan et al. (26), while siRNA Actn1_pos852 has been designed in our laboratory.

Transfection of siRNAs directed against α-actinin 1 or α-actinin 4 into HUVECs resulted in a strong and specific silencing of protein expression 3 days after transfection (Fig. 8⇓). siRNA Actn1_pos852 resulted in 70% inhibition of α-actinin 1 expression (Fig. 8⇓A, lane 2) without affecting α-actinin 4 expression (Fig. 8⇓A, lane 3). Similarly, siRNA Actn4_pos150 specifically inhibited by 90% α-actinin 4 expression (Fig. 8⇓A, lane 6), but had no effect on α-actinin 1 expression (Fig. 8⇓A, lane 5). The negative control siRNA had no effect on both actinins (Fig. 8⇓A, lanes 1 and 4).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 8.

RNA interference specifically depletes α-actinin 1 or α-actinin 4 and inhibits neutrophil diapedesis. A, HUVECs were transfected with α-actinin 1 (lanes 2 and 5), α-actinin 4 (lanes 3 and 6) or control siRNA (lanes 1 and 4) oligonucleotides and analyzed 3 days after transfection by Western blot for expression of α-actinin 1 (lanes 1–3) or α-actinin 4 (lanes 4–6). Blots were also probed with an anti-actin Ab to monitor equal loading. B, After transfection (3 days) with siRNA control, anti-actinin 1 (Actn1–852), or anti-actinin 4 (Actn4–150), HUVECs were assayed for their ability to support neutrophil diapedesis. The means and SDs were calculated from quadruplicate wells and compared using Student’s t test (∗, p < 0.001).

As shown on Fig. 8⇑B, reduced α-actinin 4 expression was accompanied by a reduction of >40% of extravasation, while only a small and nonsignificant decrease was observed after α-actinin 1 silencing.

These results indicate that α-actinin 4 interaction with ICAM-1 is essential for diapedesis.