Register for Mechanobiology 2016!

α-Actinin-dependent cytoskeletal anchorage is important for ICAM-5-mediated neuritic outgrowth
Henrietta Nyman-Huttunen, Li Tian, Lin Ning, Carl G. Gahmberg


Intercellular adhesion molecule-5 (ICAM-5, telencephalin) is a dendrite-expressed membrane glycoprotein of telencephalic neurons in the mammalian brain. By deletion of the cytoplasmic and membrane-spanning domains of ICAM-5, we observed that the membrane distribution of ICAM-5 was determined by the cytoplasmic portion. Therefore we have characterized the intracellular associations of ICAM-5 by using a bacterially expressed glutathione S-transferase (GST) fusion protein encompassing the cytoplasmic part of ICAM-5. One of the main proteins in the neuronal cell line Paju that bound to the ICAM-5 cytodomain was α-actinin. ICAM-5 expressed in transfected Paju cells was found in α-actinin immunoprecipitates, and ICAM-5 colocalized with α-actinin both in Paju cells and in dendritic filopodia and spines of primary hippocampal neurons. We were also able to coprecipitate α-actinin from rat brain homogenate. Binding to α-actinin appeared to be mediated mainly through the N-terminal region of the ICAM-5 cytodomain, as the ICAM-5857-861 cytoplasmic peptide (KKGEY) mediated efficient binding to α-actinin. Surface plasmon resonance analysis showed that the turnover of the interaction was rapid. In a mutant cell line, Paju-ICAM-5-KK/AA, the distribution was altered, which implies the importance of the lysines in the interaction. Furthermore, we found that the ICAM-5/α-actinin interaction is involved in neuritic outgrowth and the ICAM-5857-861 cytoplasmic peptide induced morphological changes in Paju-ICAM-5 cells. In summary, these results show that the interaction between ICAM-5 and α-actinin is mediated through binding of positively charged amino acids near the transmembrane domain of ICAM-5, and this interaction may play an important role in neuronal differentiation.


Cell adhesion receptors, such as integrins, selectins and immunoglobulin superfamily (IgSF) members, play pivotal roles in cell adhesion to target cells and extracellular matrices (Hynes, 1999). In brain, linkages of cell surface proteins, such as the adhesion molecules and neurotransmitter receptors, with cytoskeletal elements play key roles in synapse formation and synaptic plasticity (Walsh and Doherty, 1997; Shapiro and Colman, 1999).

The intercellular adhesion molecules, ICAMs, form a major subgroup within the IgSF and one of their major functions is to mediate leukocyte adhesion through binding to the leukocyte-specific β2-integrins (CD11/CD18) (Larson and Springer, 1990; Gahmberg, 1997; Gahmberg et al., 1997). Intercellular adhesion molecule-5 (ICAM-5; telencephalin) (Yoshihara and Mori, 1994; Yoshihara et al., 1994; Mizuno et al., 1997) is a cell adhesion molecule expressed in the somatodendritic membrane of telencephalic neurons in mammalian brain. It is homologous with the other ICAMs, although more complex (Gahmberg, 1997; Hayflick et al., 1998). ICAM-5 contains nine extracellular Ig-like domains, a short transmembrane (TM) domain and a 64-amino acid cytoplasmic domain. Like the other ICAMs, it binds to leukocyte β2-integrins (CD11a/CD18), and may be an important regulator of the immune response in the central nervous system (Tian et al., 1997; Tian et al., 2000a). ICAM-5 is also involved in hippocampal long-term potentiation, which regulates memory formation and learning (Sakurai et al., 1998). In addition, the molecule may have an important function in synapse formation in the developing brain since the onset of its expression parallels the dendritic development and the initiation of synapses (Mori et al., 1987; Oka et al., 1990), and it has been shown to induce dendritic outgrowth (Tamada et al., 1998; Tian et al., 2000b) and delay the dendritic spine maturation (Matsuno et al., 2006). It also has a tendency to form higher molecular weight forms via homophilic interactions of its extracellular domains (Tian et al., 2000b). These multimeric forms might also be complexes of different proteins including cytoskeletal components as our previous studies implied (Tian et al., 2000b). A C-terminal 17 amino acid sequence of ICAM-5 has recently been found to act as a dendritic sorting signal (Mitsui et al., 2005). However, its cytoskeletal partners have not been characterized.

α-Actinin is an F-actin-binding and cross-linking protein, forming linkages between the actin cytoskeleton and the plasma membrane (Otey and Carpén, 2004). In brain, it is a major actin-binding protein concentrated in dendritic spines (Shirao and Sekino, 2001). Several molecules, such as ICAM-1 (Carpén et al., 1992), ICAM-2 (Heiska et al., 1996), NCAM (Büttner et al., 2003), β1- and β2-integrins (Otey et al., 1990; Otey et al., 1993; Pavalko and LaRoche, 1993), L-selectin (Pavalko et al., 1995) and glutamate receptors (Wyszynski et al., 1997), have been shown to bind to α-actinin. α-Actinin is also associated with signalling molecules such as the mitogen-activated protein kinase kinase kinase and the rho-kinase type protein kinase N (Christerson et al., 1999; Mukai et al., 1997).

In the present study, we show that ICAM-5 associates with α-actinin through a stretch of positively charged cytoplasmic residues close to the membrane in vitro and in vivo. Our results indicate that the interaction between ICAM-5 and α-actinin may be involved in neuritic outgrowth and in maintaining the neuronal morphology.


Characterization of proteins interacting with the ICAM-5 cytoplasmic domain

To identify proteins that associate with the cytoplasmic domain of ICAM-5 in vitro, we made a glutathione S-transferase (GST) fusion protein encompassing the cytoplasmic sequence of ICAM-5 (Fig. 1A). The fusion protein was coupled to glutathione-Sepharose and used as an affinity matrix. To study the interacting molecules, we used the human neural crest-derived Paju cell line (Zhang et al., 1996), which does not express ICAM-5. Paju cell lysates were incubated with the affinity matrix and the bound proteins were eluted. To identify associated cytoskeletal proteins the samples were immunoblotted and analyzed with antisera against different cytoskeletal proteins. α-Actinin was identified as one of the proteins which bound to the GST-ICAM-5 cytoplasmic domain, and it did not bind to GST alone (Fig. 1B,a,b). Talin and filamin were also able to bind to the ICAM-5 cytoplasmic domain (not shown). Antiserum against ezrin or tropomyosin showed no immunoreactivity (not shown).

Since α-actinin from the cell lysate bound strongly to the cytoplasmic domain of ICAM-5 (Fig. 1B,a), we also studied whether purified α-actinin would bind to GST-cytoICAM-5. Purified chicken gizzard α-actinin was incubated with equal amounts of GST-cytoICAM-5 and GST alone. Our results showed that purified α-actinin was able to bind to the full-length GST-cytoICAM-5 but not to GST alone, confirming a direct interaction between the two proteins (Fig. 1B,b).

Mapping of ICAM-5 interaction with α-actinin

To map the binding region for α-actinin in the cytoplasmic domain of ICAM-5, a short ICAM-5 cytoplasmic domain peptide representing amino acids 857-861 (ICAM-5857-861) from the N-terminus was synthesized and linked to thiopropyl-Sepharose through a cysteine residue added to its C-terminus (Fig. 1A). This sequence was chosen because it resembles the α-actinin-binding sites in ICAM-1 and ICAM-2 having clustered positively charged amino acids (Carpén et al., 1992; Heiska et al., 1996). Purified α-actinin efficiently bound to this peptide (Fig. 1C,a). As a negative control, a scrambled version of the same peptide was used. The control peptide showed some background binding, but was much weaker than that with the specific peptide. After this, a series of peptides which had one or both of the lysines substituted either with alanine or arginine were tested (Fig. 1A). ICAM-5-K857/R and ICAM-5-K858/R peptides were still capable of binding to α-actinin although more weakly. However, there was no α-actinin binding to ICAM-5-K858/A and ICAM-5-K857-K858/A-A peptides. Binding of the ICAM-5-K857/A peptide was weak (Fig. 1C,a). This implies that the two lysines are necessary amino acids for efficient binding to α-actinin. Talin and filamin were not able to bind to the ICAM-5857-861 peptide suggesting that they may not compete with α-actinin at this site (Fig. 1D,a,b). Taken together, the results provide evidence for a direct interaction between ICAM-5 and α-actinin and indicate that a major binding site for α-actinin is located within residues 857-861 (Fig. 1C,b).

Surface plasmon resonance (SPR) analysis

To determine the binding kinetics of α-actinin interaction with ICAM-5-derived peptides, an SPR analysis with a Biacore biosensor was performed. The biotinylated peptide GGGKKGEY was captured on streptavidin-coated flow cells, and a concentration series of α-actinin was injected over the surface. The association and dissociation rates could not be evaluated from the obtained sensorgrams, owing to the fact that the interaction was very fast under the current conditions. The interaction reached the steady state within seconds (not shown), and binding of α-actinin to the GGGKKGEY peptide was detected already with 0.5 μM of α-actinin concentration (Fig. 1E). The interaction was saturated at the highest concentrations of α-actinin used (20-40 μM). The obtained data fitted well with Sigmaplot-software's ligand-binding model, in which we assumed a simple one to one interaction. The experiments were done in duplicate, and averaged for calculation of the dissociation equilibrium constant (Kd=22.9±20.5 μM) and maximal binding level (Rmax=235±104 RU).

Co-immunoprecipitation of ICAM-5 with α-actinin

We then tested whether binding between ICAM-5 and α-actinin took place in vivo. α-Actinin was immunoprecipitated from ICAM-5 transfected Paju cell lysates under the conditions that leave the cytoskeletal interactions relatively intact. Under such conditions, the cytoskeletal proteins are released from the insoluble actin filaments by DNase I treatment. ICAM-5 coprecipitated with α-actinin (Fig. 2A). The apparent molecular weight of the α-actinin-associated ICAM-5 was somewhat higher than the major band in the lysate, indicating a more mature form of the molecule. The successful precipitation of α-actinin was ensured by immunoblotting with anti-α-actinin antibody (not shown). ICAM-5 was also precipitated from rat brain homogenates, and α-actinin could be detected in the ICAM-5 immunoprecipitate, thus providing evidence for the in vivo interaction (Fig. 2B,C).

Cellular colocalization of ICAM-5 with α-actinin and F-actin

To examine whether ICAM-5, α-actinin and F-actin colocalize within cells, Paju cells were transfected with the ICAM-5 cDNAs (Fig. 3) and analyzed by confocal microscopy. As expected, no ICAM-5 could be detected in Paju-neo cells (Fig. 4A,a). In ICAM-5-transfected cells the full-length ICAM-5 was mainly localized to the uropods and cell-cell contact sites (Fig. 4D,d), where it colocalized with both α-actinin (Fig. 4F) and F-actin (Fig. 4f) as indicated by the yellow color in the merged image.

Fig. 1.

(A) A schematic picture of the ICAM-5 molecule with the amino acid sequence of the cytoplasmic domain and the peptides used in the ICAM-5-α-actinin interaction studies. The short peptides contain an additional C-terminal cysteine for coupling purposes. The peptides used in SPR analysis were biotinylated. PM, plasma membrane. (B) A GST-ICAM-5 cytodomain fusion protein binds α-actinin. (a) A GST-cytoICAM-5 fusion protein immobilized on glutathione Sepharose beads was incubated with a Paju cell lysate, and interacting proteins characterized by western blotting, in this case an anti-α-actinin antibody. (b) Binding of purified α-actinin to GST-cytoICAM-5, visualized by blotting. (C) Binding of purified α-actinin to membrane proximal ICAM-5 cytoplasmic peptides. (a) Purified α-actinin was incubated with the ICAM-5 cytodomain peptides (described in detail in Materials and Methods and in Fig. 1A) affinity matrices and the eluates were analyzed by SDS-PAGE together with purified α-actinin, and immunoblotted with anti-α-actinin antibody. (b) Based on results shown above, the main α-actinin-binding domain in the ICAM-5 cytodomain is located at residues 857-861 (box). (D) Binding of talin and filamin to ICAM-5857-861 peptide. Paju cell lysates were incubated with the ICAM-5 peptide and the scrambled control peptide, and the eluates were run on SDS-PAGE and immunoblotted for talin (a) and filamin (b). No binding was observed. (E) Binding of α-actinin to GGGKKGEY peptide as determined by SPR. Binding levels of α-actinin at steady state of interaction with GGGKKGEY as response units (RU) after subtraction of control peptide sensorgrams from each GGGKKGEY binding sensorgram, respectively.

Fig. 2.

Co-immunoprecipitation of ICAM-5 with α-actinin. (A) α-actinin was immunoprecipitated from Paju-ICAM-5 cell lysates with a specific anti-α-actinin antibody and the precipitates were subjected to SDS-PAGE and immunoblotted with anti-ICAM-5 antibody to detect possible co-immunoprecipitation of ICAM-5. The successful precipitation of α-actinin was ensured by immunoblotting with anti-α-actinin antibody (not shown). (B) ICAM-5 was immunoprecipitated from rat brain homogenates and the precipitates were run on SDS-PAGE and immunoblotted for ICAM-5 and after stripping for α-actinin.

To confirm that this colocalization was dependent on the ICAM-5 cytodomain, cells were also transfected with ICAM-5-TM cDNAs, in which the cytoplasmic domain of ICAM-5 was deleted, and ICAM-5-GPI cDNAs, in which the cytoplasmic domain of ICAM-5 was replaced with a glycophosphatidylinositol anchor (Fig. 3). In both cases, ICAM-5 showed a patchy uniform distribution throughout the cell body (Fig. 4G,J,g,j) and only weak colocalization was observed (Fig. 4I,L,I,l). The patches in Paju-ICAM-5 cells, where α-actinin and ICAM-5 colocalized, were significantly bigger than those in the truncated Paju-ICAM-5-TM (Fig. 4I,i) or Paju-ICAM-5-GPI (Fig. 4J,j) cell lines. In the mutant Paju-ICAM-5-KK/AA cell line (Fig. 4M,m), the colocalization (Fig. 4O,o) was no more seen, indicating the importance of the lysines in binding to α-actinin. It also seemed that the mutant ICAM-5-KK/AA construct caused much more cell death during the transfection compared to the other transfections. The cells always grew very slowly and their morphology was more epithelial cell-like (data not shown).

To study the colocalization in a more natural environment, we also studied the in vitro cultured rat primary hippocampal neurons. The hippocampal neurons were cultured in vitro for 7 (Fig. 5A-F) or 14 days (Fig. 5G-L), which represents the developmental stage of dendritic filopodia and dendritic spine formation, respectively. ICAM-5 and α-actinin were colocalized in the cell soma and dendrites of hippocampal neurons at both stages (Fig. 5C,L). By close observation at high magnification, we found that both molecules showed a punctated-pattern of expression along the dendrites. ICAM-5 colocalized with α-actinin in the filopodia along the apical dendrites at day 7. The two molecules colocalized in nearly 50% of early dendritic spines already at day 7 (Fig. 5F), and in nearly 80% of more mature dendritic spines at day 14 (Fig. 5L). Thus, our results suggest that ICAM-5 is associated with the actin cytoskeleton through binding to α-actinin.

Fig. 3.

Schematic presentation of full-length ICAM-5, C-terminal truncated constructs (ICAM-5-GPI, ICAM-5-TM), and point mutant (ICAM-5-KK/AA) used in the cell transfections. The asterisks indicate the sites of point mutations. EC, extracellular; TM; transmembrane; CP, cytoplasmic part.

ICAM-5-cytoskeletal association is involved in neuritic outgrowth

To study the importance of ICAM-5-cytoskeleton interaction in neuronal differentiation, we activated Paju-ICAM-5, Paju-ICAM-5-TM and Paju-ICAM-5-KK/AA cells with 100 nM phorbol 12,13-dibutyrate (PdBu), and continued the culture for 2-3 days until neurites could be clearly seen. After this, the cells were fixed, stained for ICAM-5, α-actinin and actin, and analyzed by confocal microscopy. The cells containing visible neurites were chosen for immunofluorescent imaging. In Paju-ICAM-5 cells, most of the ICAM-5 expression was localized around the cell body (Fig. 6A,a-d) and at the growth cones of the neurites (I-IV). Paju-ICAM-5-TM (Fig. 6A,e-h) and Paju-ICAM-5-KK/AA cells (Fig. 6A,i-l) showed less colocalization. Truncation (Paju-ICAM-5-TM, average length 36 μm/neurite, 36.2±11.9 μm) (Fig. 6B) or mutation of the ICAM-5 cytoplasmic tail (Paju-ICAM-5-K857-K858/A-A, average length 35 μm/neurite, 35±9.7 μm) (Fig. 6B) resulted in significantly shorter neurites compared with the full-length ICAM-5 (Paju-ICAM-5, average length 73 μM/neurite, 72.7±22.4 μm) (Fig. 6B). The statistical analysis showed that the differences in neuritic length between Paju-ICAM-5 and Paju-ICAM-5-TM or Paju-ICAM-5-KK/AA cells were significant (P<0.001). There was no significant difference in the neuritic length between the Paju-ICAM-5-TM and Paju-ICAM-5-KK/AA cells (P=0.6). These results indicate that the α-actinin-dependent cytoskeletal association is important for ICAM-5-mediated neuronal differentiation.

Peptide internalization induces morphological changes in Paju-ICAM-5 cells

The cytoplasmic ICAM-5857-861 and ICAM-5-K857-K858/A-A peptides were covalently coupled to activated penetratin 1 to get the peptides internalized into the cells. Penetratin 1 is a 16 amino acid peptide corresponding to the third helix of the homeodomain of Antennapedia protein, which is able to translocate across biological membranes (Prochiantz, 1996; Derossi et al., 1996). The peptide with the sequence AAGEY was used as a control. After the incubation, the cells were washed and visualized. The most dramatic effect occurred with the cytoplasmic peptide: the cell morphology changed in that the cells became more round-shaped compared to the normal Paju cells (Fig. 7A,B). The control peptide had no specific effect on Paju-ICAM-5 or Paju cells (Fig. 7C,D), which appeared similar to the cells without any treatment (Fig. 7E,F). The differences in relative percentage of Paju-ICAM-5 cells with round-shaped morphology between the cells treated with ICAM-5857-861 peptide (95.2%), the ICAM-5-K857-K858/A-A peptide (19.2%) or without treatment (17.3%) were statistically significant (P<0.001) (supplementary material Fig. S1).

Fig. 4.

The distribution of ICAM-5, α-actinin and F-actin in different transfected Paju cell lines. The Paju-neo (A-C,a-c), Paju-ICAM-5 (D-F,d-f), Paju-ICAM-5-TM (G-I,g-i), Paju-ICAM-5-GPI (J-L,j-l), and Paju-ICAM-5-KK/AA cells (M-O,m-o) were double-stained for ICAM-5 (green) and α-actinin (red, left panel) or F-actin (red, right panel) as described in Materials and Methods, and analyzed by confocal microscopy. The regions where the two proteins colocalize appear in yellow. Bars, 10 μm.


In this study, we describe an interaction between the intercellular adhesion molecule-5 (ICAM-5) and α-actinin. α-Actinin from the Paju cell lysate bound efficiently to a GST fusion protein encompassing the whole cytoplasmic part of ICAM-5, as did the purified α-actinin. Furthermore, ICAM-5 was observed to coprecipitate with α-actinin under mild conditions where the cytoskeletal interactions were relatively well preserved.

By using an immobilized peptide representing the amino acids 857-861 (ICAM-5857-861, KKGEY) of the ICAM-5 cytodomain as an affinity matrix, we were able to determine that the main binding site for α-actinin is in the N-terminal region of the ICAM-5 cytodomain. The main binding site for α-actinin, comprising the amino acids KKGEY contains positively charged basic residues. According to our results the basic lysines at positions 857 and 858 are important in binding to α-actinin. Substitution of the lysines either with neutral alanine or positively charged arginine markedly reduced the binding of the peptides to α-actinin. However, the peptides ICAM-5-K857/R and ICAM-5-K858/R, which each had one lysine substituted with arginine showed some binding to purified α-actinin indicating that positive charges are beneficial in the ICAM-5/α-actinin interaction. There was no binding to the ICAM-5-K858/A peptide indicating that the second lysine is most important in binding to α-actinin. SPR analysis confirmed the interaction between the KKGEY peptide and α-actinin.

Fig. 5.

Colocalization of ICAM-5 with α-actinin in cultured rat hippocampal neurons. The 7-day-old (A-F) or 14-day-old neurons (G-L) were stained for ICAM-5 (green) and α-actinin (red), and analyzed by confocal microscopy. Both ICAM-5 and α-actinin showed a punctated expression pattern along the dendrites. The colocalization of the two molecules in apical dendritic filopodia (indicated by arrows, image F) at day 7, and in dendritic spines at day 14 (indicated by arrowheads, image L) are shown under high magnification. Bars, 10 μm.

Fig. 6.

Effect of ICAM-5 contructs on neuritic outgrowth. (A) The Paju-ICAM-5 (a-d, I-IV), Paju-ICAM-5-TM (e-h) and Paju-ICAM-5-KK/AA cells (i-l) were activated by 100 nM PdBu to induce neurite outgrowth. The cells were triple-stained for α-actinin (green), actin (red), ICAM-5 (magenta), and visualized by confocal microscopy. Concentrations of ICAM-5, α-actinin and F-actin at cell-anchorage sites (a-d) and growth cones of neurites (I-IV) in Paju-ICAM-5 cells were seen. Bars, 20 μm. (B) Histogram displaying neurite outgrowth in PdBu-activated Paju-ICAM-5, Paju-ICAM-5-TM and Paju-ICAM-5-KK/AA cells (neurite length presented as micrometers). The differences between the neuritic length in Paju-ICAM-5 cells compared with Paju-ICAM-5-TM and Paju-ICAM-5-KK/AA cells are statistically significant (P<0.001). Standard deviations are shown.

Fig. 7.

Effect of ICAM-5 cytoplasmic peptides on cell morphology. Paju-ICAM-5 (A,C,E) and Paju (B,D,F) cells were incubated with penetratin-coupled ICAM-5857-861 and ICAM-5-K857-K858/A-A peptides at 20 μM concentrations. After 2 hours of incubation the cells were washed and bright-field pictures were taken to visualize the effects of the peptides on cell morphology. Bars, 20 μm. The relative percentage of round-shaped Paju-ICAM-5 cells (treated with ICAM-5857-861 peptide) was 95.2%, and the difference is statistically significant (P<0.001) when compared with the cells treated with ICAM-5-K857-K858/A-A peptide or without treatment (data not shown).

The cytoplasmic binding sites for α-actinin in ICAM-1, ICAM-2, ICAM-5, L-selectin, β1- and β2-integrins show some resemblance to each other by containing highly positively charged residues (Table 1). The α-actinin-binding site in ICAM-5 (KKGEY) seems to be most similar to that of ICAM-1 (RKIKK) (Carpén et al., 1992), and both contain two adjacent basic residues. It appears that positively charged basic residues are present in many α-actinin-binding sites of transmembrane proteins (Otey and Carpén, 2004; Tang et al., 2001). The crystal structure of the α-actinin molecule shows that the central rod domain region is very acidic, and this acidic surface has been postulated to constitute a potential binding site for many transmembrane receptors (Tang et al., 2001; Ylänne et al., 2001). Otherwise, the cytoplasmic domains of different ICAMs are poorly conserved, in contrast to integrins, indicating that they have quite distinct functions (Hayflick et al., 1998). Except for ICAM-5, all the other ICAMs (ICAM-1-4) are expressed by blood and endothelial cells (Gahmberg, 1997; Gahmberg et al., 1997).

View this table:
Table 1.

Binding sites for α-actinin in different adhesion proteins

Immunofluorescence studies showed that ICAM-5 was mainly concentrated at the uropods and cell-cell contact sites of transfected Paju cells, whereas deletion of the cytoplasmic tail (ICAM-5-GPI or ICAM-5-TM constructs) caused a more even but patchy distribution of ICAM-5 on the cell membrane. α-Actinin colocalized with the intact ICAM-5, and only weakly with the truncated ICAM-5. In hippocampal neurons, ICAM-5 and α-actinin colocalized in the dendritic filopodia and spines. Strong accumulation of α-actinin in ICAM-5-transfected cells may be explained by the homophilic interaction of ICAM-5 multimers at the cell-cell contact sites (Tian et al., 2000). The different distribution patterns in the transfected Paju cells suggest that α-actinin regulates the localization of ICAM-5 through the cytoplasmic domain of ICAM-5. Furthermore, it is possible that ICAM-5 is involved in the organization of the actin cytoskeleton. ICAM-5 also partially colocalized with F-actin. This provides further evidence of an indirect interaction between ICAM-5 and actin filaments due to α-actinin crosslinking.

In Paju-ICAM-5-KK/AA cells, there was no clear colocalization confirming that the lysines are needed for binding to α-actinin. Interestingly, when we transfected the mutant ICAM-5-KK/AA construct into Paju cells, most of the cells died. The morphology of the transfected cells became epithelial cell-like, and they grew slowly. Furthermore, when we treated the Paju-ICAM-5 cells with the ICAM-5857-861 peptide coupled to penetratin, the morphology of the cells changed dramatically and became rounded. Evidently the peptide competes with intact ICAM-5 and other relevant molecules in Paju cells for binding to α-actinin leading to disruption of the normal cytoskeleton structure.

The ICAM-5/α-actinin colocalization pattern also changed quite rapidly after activation with phorbol ester. When the Paju-ICAM-5 cells started to extend neurites, most of the ICAM-5 and α-actinin was concentrated at the cell-anchorage sites around the soma and at the growth cones of the neurites, which differs from the pattern seen in Paju-ICAM-5-TM and Paju-ICAM-5-KK/AA cells. The length of the neurites in Paju-ICAM-5-TM or Paju-ICAM-5-KK/AA cells was shorter than that of the Paju-ICAM-5 cells. This indicates that the interaction between ICAM-5 and α-actinin plays a role in neuritic outgrowth. As shown previously, ICAM-5 shows homophilic binding between neurons and strongly stimulates dendritic outgrowth (Tian et al., 2000b). It is possible that α-actinin is involved in the regulation of these phenomena. Besides, ICAM-5 may also be involved in synapse formation and plasticity, since F-actin is essential for the development and maintenance of young synapses (Zhang and Benson, 2001), and α-actinin is the major actin-binding protein concentrated in the postsynaptic dendritic spines (Shirao and Sekino, 2001).

Although mutant mice lacking ICAM-5 show no detectable abnormalities in brain anatomy or in synaptic structures, their cognition functions were changed (Nakamura et al., 2001). Interestingly, LTP was enhanced in hippocampal synapses, suggesting that ICAM-5 may regulate the synaptic stability by determining the dynamic range of synaptic efficacy (Nakamura et al., 2001). In synapses, ICAM-5 is expressed around the postsynaptic dendritic spines and, thus, it may serve as a structural constriction in synaptic connections by contributing to morphological and functional changes through crosslinking its counter-receptors with the cytoskeleton (Sakurai et al., 1998). Furthermore, it has been recently discovered that ICAM-5 is a negative regulator of spine formation (Matsuno et al., 2006), and since LTP is associated with structural changes of formed synapses as well as new spine formation (Matsuzaki et al., 2004), ICAM-5 may be an essential molecule in contributing to refinement of functional neural circuits in the telencephalon, which regulates higher brain functions such as memory and learning.

A soluble form of ICAM-5 is found in the cerebrospinal fluid and plasma, and its level increases during inflammation and other conditions in the brain, such as acute encephalitis (Lindsberg et al., 2002) and temporal lobe epilepsy (Rieckmann et al., 1998). Decreased immunoreactivity of ICAM-5 in the brain of patients with Alzheimer's disease has also been reported (Hino et al., 1997). More recent studies have shown that ICAM-5 binds to presenilins 1 and 2, which are part of a large γ-secretase complex implicated in Alzheimer's disease (Annaert et al., 2001). Furthermore, presenilin 1 seems to mediate the turnover of ICAM-5 via an autophagic degradative pathway (Esselens et al., 2004). Cytoskeletal protein aggregations are one of the pathological hallmarks of neurodegenerative disorders. Amyloid beta-protein precursor, actin, tropomyosin, vinculin and α-actinin have all been observed in intraneuronal inclusions (Hirano bodies) in patients of Alzheimer's disease (Galloway et al., 1987; Maciver and Harrington, 1995). Therefore, it would be interesting to study whether ICAM-5 is involved in regulation of γ-secretase protein complex formation via cytoskeletal association, and furthermore, whether this is important for the pathogenesis of Alzheimer's disease.

Materials and Methods

Cell lines

Paju is a human neural crest-derived cell line (Zhang et al., 1996) that is cultured in Dulbecco's modified Eagles Medium (DMEM) containing 10% fetal calf serum, 1% penicillin, streptomycin and L-Glutamine (BioWhittaker). Paju cells were transfected with pCDM8neo stuffer, and with different ICAM-5 constructs: (1) complete ICAM-5 cDNA (Paju-ICAM-5), (2) a truncated cDNA that replaced the transmembrane (TM) and cytoplasmic domains with a glycophosphatidylinositol (GPI) anchor, site of CD55 [complement decay-accelerating factor (Moran and Caras, 1991); Paju-ICAM-5-GPI], (3) another truncated construct where the cytoplasmic domain was deleted (Paju-ICAM-5-TM), (4) the complete ICAM-5 cDNA with amino acids lysine 857 and lysine 858 mutated to alanines (Paju-ICAM-5-KK/AA). All these cDNA constructs were cloned in the expression vector pEF-BOS (Tian et al., 1997; Moran and Caras, 1991) (Fig. 3). The control cell line, Paju-neo, was transfected with the empty pEF-BOS vector and the pCDM8neo stuffer. Transfections were stably performed with the Lipofectamine® 2000 reagent according to the manufacturer's protocol (Invitrogen). Cell surface protein expression was analyzed by FACS and immunofluorescent staining (data not shown). All the cell lines were cultured as mentioned above, but in the presence of 0.5 mg/ml G418 (Calbiochem-Novabiochem Corporation, CA). All the cDNAs had been verified by sequencing (DNA synthesis and sequencing laboratory, Institute of Biotechnology, University of Helsinki).


The polyclonal rabbit anti-serum against the cytoplasmic part of mouse ICAM-5 (Tian et al., 1997) was kindly provided by Y. Yoshihara, Laboratory for Neurobiology of Synapse, Brain Science Institute/RIKEN, Wako-City, Japan. The polyclonal antibody against rat ICAM-5, 1000J, and the monoclonal antibodies (mAbs) 246K and 127E were from ICOS Corporation, Washington, USA. The mAb against ICAM-5, TL-3, has been described previously (Tian et al., 2000a). The mAbs against α-actinin (MAB1682) and filamin (MAB1678) were purchased from Chemicon International, Temecula, USA. The talin mAb 8d4 and the polyclonal rabbit anti-α-actinin A2543 were from Sigma-Aldrich, St Louis, MO. The mAb AT6/172 against α-actinin was from SeroTec (Oxford, UK), and rabbit anti-mouse immunoglobulins from Dako A/S, Denmark. A mouse negative IgG control was purchased from Silenius, Hawthorn, Australia. Sheep anti-mouse and anti-rabbit IgG horseradish peroxidase conjugates were from Amersham Biosciences, Buckinghamshire, UK. Rhodamine-phalloidin was from Sigma-Aldrich. Goat anti-mouse and anti-rabbit linked to fluorophore Alexa488 were from Molecular Probes. Cy3-conjugated goat anti-rabbit and Cy5-conjugated goat anti-mouse IgG were from Jackson ImmunoResearch Laboratories.

Peptide synthesis and coupling

Synthetic peptides from the cytoplasmic domain of ICAM-5 were made in the peptide synthesis unit of the Division of Biochemistry, Faculty of Biosciences, University of Helsinki, Finland, purified with HPLC, and verified by mass spectrometry. The synthetic peptides include (1) the sequence encompassing amino acids 857-861 (ICAM-5857-861) (KKGEY-C), and (2) the same sequence but with one or both of the lysines substituted either with alanine or arginine: ICAM-5-K857/R (RKGEY-C), ICAM-5-K858/R (KRGEY-C), ICAM-5-K857-K858/A-A (AAGEY-C), ICAM-5-K857/A (AKGEY-C), ICAM-5-K858/A (KAGEY-C) (Fig. 1A). The peptide encompassing amino acids 857-861 was also synthesized in random order as a scrambled control peptide (GKEYK-C). An additional cysteine was added to the C-terminus of each peptide. The ICAM-5 peptides and the scrambled control peptide (2.5 mg) were coupled to 0.5 ml of thiopropyl-Sepharose 6B (Amersham Biosciences) and used for peptide affinity chromatography. The coupling efficiency was 60-90%.

The biotinylated peptide GGGKKGEY (Fig. 1A) was used in surface plasmon resonance (SPR) analysis. The biotinylated control peptide had the sequence PQPQNSSLRTPLR.

Surface plasmon resonance analysis

Binding of α-actinin to control and GGGKKGEY peptides was studied by SPR analysis in a Biacore 2000® biosensor (Biacore AB, Uppsala, Sweden). The flow cells of an SA (streptavidin) biosensor chip (Biacore AB) were immobilized with saturating amounts of biotinylated peptides according to the manufacturer's instructions. The coupling levels in resonance units (RU) were 330 for control and 159 for GGGKKGEY peptides. A wide range of concentrations of α-actinin (0.5-40 μM) were injected for 2 minutes at a flow rate of 5 μl/minute over the peptide-coated flow cells. The dilution series of α-actinin were made in the HBS system buffer of the manufacturer (10 mM Hepes, pH 7.4; 150 mM NaCl; 3 mM EDTA and 0.005% p20). Before dilution, α-actinin was in a suspension of 2 M (NH4)2SO4 containing 20 mM Tris acetate, pH 7.6, 20 mM NaCl, 0.1 mM EDTA, 15 mM β-mercaptoethanol and 1 mM phenylmethylsulfonylfluoride (PMSF). Regeneration of the flow cells was achieved by washing for 20 minutes with HBS at the flow rate of 5 μl/minute. The data were evaluated by subtracting the sensorgram of the flow cell containing the control peptide from the sensorgrams of flow cells containing the GGGKKGEY peptide (Biaevaluation 3.1 software, Biacore AB, Uppsala, Sweden). The binding response of α-actinin to the peptide at the steady state was used to evaluate the Kd of the interaction with Sigmaplot 8.0 one site saturation model (Systat software, Richmond, USA).

Peptide affinity chromatography

5 μg of chicken gizzard α-actinin (Sigma-Aldrich) or Paju cell lysates (for talin and filamin-binding assay) containing 2-3 mg protein were preincubated with blocked thiopropyl-Sepharose (Amersham Biosciences) without peptide for 1 hour at 4°C under constant agitation. The incubation was further continued with peptide-coupled Sepharose (described in Peptide synthesis and coupling) for 2 hours at 4°C. The beads were washed three times with the lysis buffer (50 mM Tris-HCl, pH 7.4, 60 mM NaCl, 1% TX-100, 10 mM EDTA, 50 mM NaF, 2 mM phenylmethylsulfonylfluoride, 10 μg/ml of both aprotinin and leupeptin, 0.1 mM quercetin and 5 mM iodoacetamide) and bound proteins were eluted with the Laemmli sample buffer, separated in 4-15% or 8-16% SDS-PAGE gradient gels (Bio-Rad Laboratories, CA) and detected by immunoblotting. α-Actinin was visualized with MAB1682, talin with mAb 8d4, and filamin with MAB1678.

Peptide coupling with penetratin and peptide internalization

An equimolar amount of the ICAM-5857-861 (KKGEY-C) peptide or the mutated peptide ICAM-5-K857-K858/A-A (AAGEY-C) was added to the activated penetratin (Qbiogene), and the mixture was incubated for 2 hours at 37°C. After this the coupled peptides were stored at -70°C.

Paju-ICAM-5 and Paju cells (20×104/well) were grown in 24-well plates for 2 days and incubated for 2 hours at 37°C with the penetratin-coupled peptides prediluted in culture medium at a final concentration of 20 μM, and washed with media. After this the bright-field images were taken under 10× or 20× magnification (Olympus IX71 inverted research microscope). The relative amount of cells with changed morphology was quantitated. The cells were counted from two random fields, and the number of rounded cells was compared with the total amount of cells, and averaged. 70-80 cells were counted for each treatment.

Glutathione S-transferase (GST) fusion protein constructs

A plasmid construct encoding the ICAM-5 cytoplasmic domain was generated by PCR (GeneAmp PCR system 9700, Applied Biosystems, Foster City, CA) using appropriate 5′ (ATGGATCCCAGTCCACCGCCT) and 3′ (ATGAATTCTCACGCCGATGTCAG) primers. The represented amino acid sequence was 852-912 for the whole cytoplasmic part (GST-cytoICAM-5). Human ICAM-5 cDNA in the pEF-BOS vector was used as the template. All PCR reactions were performed using HotStarTaq DNA polymerase (Qiagen, CA). The PCR products were cut with BamHI and EcoRI restriction enzymes (New England Biolabs, Beverly, MA) and ligated into the pGEX-2TK vector (Amersham Biosciences, Uppsala, Sweden). All PCR-derived clones were verified by sequencing (DNA synthesis and sequencing laboratory, Institute of Biotechnology, University of Helsinki).

GST fusion protein interaction assay

The GST-ICAM-5 fusion protein and GST were produced in the E. coli strain BL21 (DE3) (Stratagene, La Jolla, CA) after induction for 4 hours at 37°C with IPTG (isopropylthiogalactopyranoside, Sigma-Aldrich). The bacteria were sonicated in 1% TX-100 (Sigma-Aldrich), PBS (phosphate-buffered saline: 0.14 M NaCl, 10 mM Na-phosphate, pH 7.4), and after centrifugation (13,000 rpm, 45 minutes) the supernatants were allowed to bind to glutathione-Sepharose 4B (Amersham Biosciences). Aliquots of this matrix and of the negative control GST, containing comparable amounts of the fusion proteins, were used for binding experiments.

Paju cells were lysed at 0°C for 15 minutes in the lysis buffer containing 50 mM Tris-HCl, pH 7.4, 60 mM NaCl, 1% TX-100, 10 mM EDTA, 50 mM NaF, 2 mM PMSF, 10 μg/ml of both aprotinin and leupeptin, 0.1 mM quercetin and 5 mM iodoacetamide. The lysates were centrifuged in an Eppendorf centrifuge at 4000 rpm for 5 minutes at 4°C. The protein concentration of the lysates was estimated by the Biuret method (Stoscheck, 1990). Lysates containing 2-3 mg protein or purified chicken gizzard α-actinin (5 μg) were rotated with the ICAM-5 GST fusion protein and GST affinity matrices at 4°C and washed three times each with 1 ml of the lysis buffer. The incubation times were overnight for the lysates and 2 hours for purified α-actinin. Bound proteins were eluted with the Laemmli sample buffer, separated in 8% SDS-PAGE or 4-15% gradient gels (Bio-Rad Laboratories, CA) and detected by immunoblotting. α-Actinin was visualized with MAB1682.


Coimmunoprecipitation of ICAM-5 with α-actinin was done by the solid-phase immunoisolation technique, as described (Stefanova et al., 1993). Briefly, for indirect solid-phase immunoisolation, rabbit anti-mouse immunoglobulin was coated on microtiter plates (U-shaped, Nunc A/S, Denmark) and allowed to incubate at 37°C for 2 hours. After washes with PBS, the mAb AT6/172 against α-actinin was added to the wells and incubated at 4°C overnight, excess protein binding sites were then blocked with 5% BSA. Paju-ICAM-5 cell lysates were treated with DNaseI (Roche Diagnostics, Indianapolis, IN) for 50 minutes at 22°C prior to applying to coprecipitations. The cell lysates were applied to the wells and incubated for 4 hours at 4°C. The wells were washed three times with the same lysis buffer as used in peptide affinity chromatography. The proteins eluted with the Laemmli sample buffer were separated using 8% SDS-PAGE gels and visualized by immunoblotting. ICAM-5 was detected with the polyclonal anti-serum against the cytoplasmic part of ICAM-5, and α-actinin with MAB1682.

Brains of adult rats were homogenized in TNF buffer [50 mM Tris-HCl, pH 8.0, 1% Tx-100, 0.5% NP-40, 140 mM NaCl, 1 mM EDTA, NaF, and Na3VO4, 10 μg/ml aprotinin and leupeptin, and 1 mM 4-(2-Aminoethyl)benzenesulphonyl fluoride (AEBSF)]. The homogenates were cleared by centrifugation, and further preabsorbed with protein G Sepharose (Amersham Biosciences) for 2 hours at 4°C in rotation. After this, 10 μg of the mAb 127E against rat ICAM-5 or mIgG were added, and allowed to incubate overnight at 4°C. After the incubation, protein G Sepharose was added to the brain lysates and the incubation was continued for 2-4 hours at 4°C. Protein G Sepharose was washed three times with TNF buffer and the bound proteins were eluted with Laemmli Sample Buffer, and separated by 4-15% gradient gels (Bio-Rad Laboratories). The eluted proteins were visualized by immunoblotting.


30 μl of eluted fractions from the ICAM-5 peptide affinity chromatographies, peptide blocking assays, GST fusion protein interaction assays, or coimmunoprecipitation assays were separated by SDS-PAGE and blotted onto nitrocellulose membranes. The membranes were blocked using 5% milk powder in TCN (50 mM Tris-HCl, pH 8.0, 80 mM NaCl, 2 mM CaCl2) containing 0.2% NP-40. The membranes were reacted with primary and secondary antibodies at appropriate concentrations in 3% milk powder-TCN/0.2% NP-40 for 1 hour, respectively. The blots were washed extensively with washing buffer (TCN/0.2% NP-40 or 0.05% TX-100). The bound antibodies were detected by enhanced chemiluminescence (Amersham, Buckinghamshire, UK).

Immunofluorescent staining

Hippocampus was dissected from the brains of 19-day-old rat embryos and treated with 0.5 mg/ml papain for 10 minutes (Worthington Chemical Corp.) in HBSS (Gibco BRL). The neurons were washed with HBSS and 0.1×106 cells/cover slip were plated on poly-L-lysine (Sigma-Aldrich)-coated 8-mm coverslips. After plating, the cells were cultured in Neurobasal medium (Gibco BRL) with 2% B27 supplement (Gibco BRL), 25 μM L-glutamic acid (Sigma-Aldrich), and 1% penicillin, streptomycin, and L-glutamine for 1-2 weeks before the costainings. The cells were permeabilized with 0.2% TX-100 and blocked with 3% BSA. ICAM-5 was detected with anti-rat mAb 127E and 246K, and α-actinin with the rabbit anti-α-actinin antibody A2543.

Paju, Paju-ICAM-5, truncated Paju-ICAM-5 (TM and GPI), and Paju-ICAM-5-KK/AA cells (1×104/well) were grown on coverslips in serum-containing medium for 3 days in 24-well plates. After paraformaldehyde fixation the cells were permeabilized with 0.1% TX-100 in PBS, blocked with 3% BSA, and double-stained for ICAM-5 (mAb TL-3) and α-actinin (polyclonal rabbit anti-α-actinin antiserum) or F-actin (rhodamine-phalloidin). For ICAM-5 and α-actinin staining, the coverslips were incubated with Alexa488-conjugated goat anti-mouse IgG and Cy3-conjugated goat anti-rabbit IgG.

Neurite outgrowth assay

Paju-ICAM-5, Paju-ICAM-5-TM and Paju-ICAM-5-KK/AA cells were grown in serum-containing medium in 24-well plates (1×104/well) on cover slips over night. Then the cells were activated with 100 nM phorbol 12,13-dibutyrate (PdBu), and cultured for 2-3 days before visualization with the microscope. After paraformaldehyde fixation, the cells were permeabilized with 0.1% TX-100 in PB, blocked with 3% BSA, and triple-stained for ICAM-5, α-actinin and F-actin. The secondary antibodies were Cy5-conjugated goat anti-mouse and Alexa488-conjugated goat anti-rabbit IgG. The fluorescent images were taken with an inverted confocal microscope (Leica TCS SP2 AOBS) under 63× magnification. The length of the neurites was measured in 2-3 random fields and averaged. Cells with neurite length of at least twice the diameter of the cell body were quantitated. 30 neurons were counted for Paju-ICAM-5 and Paju-ICAM-5-TM cells, and 20 neurons for Paju-ICAM-5-KK/AA cells.

Statistical analysis

For all the quantization, statistical analysis of variance (ANOVA) was used to compare the different groups in the neurite outgrowth assay and peptide internalization experiment.


We thank Heikki Rauvala for providing rat hippocampal neurons, Y. Yoshihara for polyclonal antisera to ICAM-5, Patrick Kilgannon for rat ICAM-5 mAbs, Carmela Kantor-Aaltonen for the different cytoplasmic ICAM-5 peptides, Hilkka Lankinen and Tomas Strandin for the SPR analysis work, Outi Nikkilä, Leena Kuoppasalmi and Maria Aatonen for technical assistance, and Yvonne Heinilä for secretarial help. This study was supported by the Sigrid Jusélius Foundation, the Academy of Finland, the Finnish Cultural Foundation, Magnus Ehrnrooth Foundation, and the Finnish Cancer Society.


  • Accepted May 10, 2006.


View Abstract