Interleukin 32 (IL-32) Contains a Typical α-Helix Bundle Structure That Resembles Focal Adhesion Targeting Region of Focal Adhesion Kinase-1*

Background: IL-32 is involved in several cell processes, most likely through integrin signaling. Results: Modeling of IL-32 revealed a structure similar to FAT and binds to integrins, paxillin, and FAK-1, all members of the integrin-signaling pathway. Conclusion: IL-32 interacts with members of the focal adhesion protein complex. Significance: IL-32 might be a key protein in integrin signaling and downstream processes. IL-32 can be expressed in several isoforms. The amino acid sequences of the major IL-32 isoforms were used to predict the secondary and tertiary protein structure by I-TASSER software. The secondary protein structure revealed coils and α-helixes, but no β sheets. Furthermore, IL-32 contains an RGD motif, which potentially activates procaspase-3 intracellular and or binds to integrins. Mutation of the RGD motif did not result in inhibition of the IL-32β- or IL-32γ-induced cytotoxicity mediated through caspase-3. Although IL-32α interacted with the extracellular part of αVβ3 and αVβ6 integrins, only the αVβ3 binding was inhibited by small RGD peptides. Additionally, IL-32β was able to bind to αVβ3 integrins, whereas this binding was not inhibited by small RGD peptides. In addition to the IL-32/integrin interactions, we observed that IL-32 is also able to interact with intracellular proteins that are involved in integrin and focal adhesion signaling. Modeling of IL-32 revealed a distinct α-helix protein resembling the focal adhesion targeting region of focal adhesion kinase (FAK). Inhibition of FAK resulted in modulation of the IL-32β- or IL-32γ-induced cytotoxicity. Interestingly, IL-32α binds to paxillin without the RGD motif being involved. Finally, FAK inhibited IL-32α/paxillin binding, whereas FAK also could interact with IL-32α, demonstrating that IL-32 is a member of the focal adhesion protein complex. This study demonstrates for the first time that IL-32 binds to the extracellular domain of integrins and to intracellular proteins like paxillin and FAK, suggesting a dual role for IL-32 in integrin signaling.

IL-32 can be expressed in several isoforms. The amino acid sequences of the major IL-32 isoforms were used to predict the secondary and tertiary protein structure by I-TASSER software. The secondary protein structure revealed coils and ␣-helixes, but no ␤ sheets. Furthermore, IL-32 contains an RGD motif, which potentially activates procaspase-3 intracellular and or binds to integrins. Mutation of the RGD motif did not result in inhibition of the IL-32␤-or IL-32␥-induced cytotoxicity mediated through caspase-3. Although IL-32␣ interacted with the extracellular part of ␣V␤3 and ␣V␤6 integrins, only the ␣V␤3 binding was inhibited by small RGD peptides. Additionally, IL-32␤ was able to bind to ␣V␤3 integrins, whereas this binding was not inhibited by small RGD peptides. In addition to the IL-32/integrin interactions, we observed that IL-32 is also able to interact with intracellular proteins that are involved in integrin and focal adhesion signaling. Modeling of IL-32 revealed a distinct ␣-helix protein resembling the focal adhesion targeting region of focal adhesion kinase (FAK). Inhibition of FAK resulted in modulation of the IL-32␤-or IL-32␥-induced cytotoxicity. Interestingly, IL-32␣ binds to paxillin without the RGD motif being involved. Finally, FAK inhibited IL-32␣/paxillin binding, whereas FAK also could interact with IL-32␣, demonstrating that IL-32 is a member of the focal adhesion protein complex. This study demonstrates for the first time that IL-32 binds to the extracellular domain of integrins and to intracellular proteins like paxillin and FAK, suggesting a dual role for IL-32 in integrin signaling.
In this study, we used I-TASSER software to predict the secondary and tertiary structure of the three major isoforms of IL-32. Next, we investigated the role of the RGD motif in the IL-32␤and IL-32␥-induced cell death by developing several mutants. Furthermore, IL-32 binding assays were conducted to investigate whether IL-32 would bind to integrins and intracellular proteins that are involved in the integrin-signaling pathway. Moreover, a small molecule inhibitor for the integrin-signaling pathway was used to investigate the IL-32-induced cytotoxicity. Finally, we propose a model for the possible IL-32 integrin-signaling cascade.
RGD Mutants-pCDNA3 expression plasmids containing human IL-32␣, IL-32␤, or IL-32␥ sequences were used to mutate the RGD motif into RGE. Primers (forward, 5Ј-CAAG-CTTGCCACCATGTGCTTCCCGAAGGTCCTCTCTGAT-GACATGAAGAAGCTGAAGG-3Ј, and reverse, 5Ј-GTCTA-GATCATTTTGAGGATTGGGGTTCAGAGCACTTCTGG-GGTGTCAGCTCCTCCTTCTC-3Ј) were manufactured by Biolegio (Nijmegen, The Netherlands), which contained the RGD to RGE single nucleotide mutation. Subsequently, the primers were used to construct the mutants by conventional PCR. The RGD deletion mutants were produced by digesting the pCDNA3-IL-32␤/IL-32␥ expression plasmids with Hin-dIII (New England Biolabs, Ipswich, MA) and XbaI (New England Biolabs) to separate the backbone and insert. The inserts were digested with BtgI (New England Biolabs) to remove the GD amino acid sequence until the stop codon. Subsequently, the modified IL-32␤/IL-32␥ inserts were ligated at the HindIII overhang into the previously isolated backbone, which also contained a HindIII overhang, by using T4 DNA ligase (New England Biolabs). Next, the BtgI and XbaI sites were both treated with T4 DNA polymerase (New England Biolabs) together with 100 M dNTPs to create blunt ends to facilitate ligation of both ends with T4 DNA ligase (New England Biolabs). Exactly after the arginine amino acid, a stop codon was present because of the restriction and ligation reaction. Finally, all of the mutants were verified by sequencing.
Overexpression of IL-32 Wild Type and RGD Mutants in HEK293T Cells-Cytotoxicity was investigated in HEK293T cells that were transfected with plasmids (pCDNA3) expressing eGFP, 4 wild type, or RGD mutants of IL-32␣, IL-32␤, and IL-32␥ by using Lipofectamine 2000 (Invitrogen) and cytotoxicity was determined as described earlier.
IL-32/Integrin Binding Assays-MaxiSorp flat-bottomed 96-well plates (Nunc, Roskilde, Denmark) were coated with 1 g/ml recombinant ␣V␤3 integrin (R & D Systems, Minneapolis, MN), 1 g/ml recombinant ␣V␤6 integrin (R & D Systems), 1 g/ml ␣V␤8 integrin (R & D Systems), or 1 g/ml BSA all diluted in PBS and incubated overnight at 4°C. Next, the wells were blocked with 1% BSA (Invitrogen) in PBS at 37°C for 1 h followed by three wash steps with PBS containing 0.05% Tween 20. Subsequently, some wells were preincubated with 10 M cyclo-(RGDfV) (Peptide Institute Inc, Osaka, Japan) or 10% FCS, whereas other were incubated with wash buffer (PBS with 0.05% Tween 20) at room temperature for 1 h. Next, the wells were incubated with different concentrations recombinant IL-32 proteins (R & D Systems) diluted in PBS, with or without 10 M cyclo-(RGDfV) or 10% FCS at room temperature for 1 h. The wells were washed three times with wash buffer and incubated with biotinylated anti-IL-32␣ antibody (R & D Systems) diluted in PBS with a concentration of 0.2 g/ml at room temperature for 1 h. After the incubation, the wells were washed three times with wash buffer and incubated with streptavidin-poly-HRP (Sanquin, Amsterdam, The Netherlands) at room temperature for 30 min. Next, wells were washed six times with wash buffer and incubated with TMB substrate (Biomérieux, Boxtel, The Netherlands) at room temperature for ϳ10 min, and the color reaction was stopped with 2.5 M H 2 SO 4 . The absorbance/optical density was measured with an ELISA reader (Tecan Group Ltd., Männedorf, Switzerland) at 450 nm.

RESULTS
In Silico Modeling of IL-32-I-TASSER software was used to predict the secondary and tertiary structure of IL-32. IL-32 consists of several isoforms, and we used the amino acid sequences of IL-32␣, IL-32␤, and IL-32␥ for the modeling. Table 1 shows the secondary structure of IL-32␣, IL-32␤, and IL-32␥. Coils (C) and ␣-helixes (H) are aligned with the corresponding amino acid together with the confidence score. Remarkably, no ␤ sheets were predicted but only coils, and ␣-helixes are found in the secondary structure. Furthermore, in the amino acid sequences of IL-32␣, IL-32␤, and IL-32␥, two known motifs are present (bold and underlined), namely a DDM and an RGD sequence. Interestingly, IL-32␣ misses the large ␣-helix located in front of the RGD motif. Fig. 1 shows the tertiary structure of IL-32␣ and IL-32␤ with a confidence score (C-score) of Ϫ1.89 and Ϫ2.19, respectively. Furthermore, the RGD motif is differently located in the IL-32␣ and IL-32␤ models, which could lead to a less accessible RGD motif. Predicting the tertiary structure of IL-32␥ is rather complicated because of the possibility of a transmembrane domain. The C-score of IL-32␥ is Ϫ2.61, which is less reliable than the predicted IL-32␣/IL-32␤ models, whereas models with C-scores higher than Ϫ1.5 are considered to have correct folds. Table 2 shows the intracellular (i, inside), transmembrane (H, transmembrane helix), and extracellular domains (o, outside) of IL-32␥ as predicted by HMMTOP software. In addition, the PR3 cleavage site between threonine and valine is underlined together with an arrow in Table 2, which could be involved in releasing IL-32␥ from the membrane. Fur-thermore, the IL-32␥-specific amino acid sequence is displayed bold.
Mutation of RGD Motif Present in IL-32␤ or IL-32␥ Does Not Prevent Cell Death-Small soluble peptides containing an RGD motif can induce apoptosis through direct activation of procaspase-3, leading to caspase-3-induced apoptosis (46). Interestingly, an RGD motif is present in IL-32 (Table 1), and therefore, we hypothesized that the RGD motif present in IL-32 could activate procaspase-3 and finally result in apoptosis. Surprisingly, the RGD motif present in IL-32␤ or IL-32␥ is not involved in the IL-32␤/␥-induced caspase-3-dependent apoptosis, because mutation of the RGD motif into RGE did not reduce cell death (Fig. 2B). To study whether the lack of the RGD motif instead of RGD to RGE mutation was protective, we deleted the GD until the stop codon of IL-32␤ and IL-32␥. Overexpression of these deletion mutants did not show inhibition of the IL-32␤ or IL-32␥-induced cytotoxicity compared with the wild type isoforms of IL-32␤ and IL-32␥ (Fig. 2C). Remarkably, mutating the RGD motif resulted in enhanced IL-32␥-cytotoxicity (Fig. 2B), whereas deletion of the RGD motif increased the IL-32␤and IL-32␥-induced cytotoxicity (Fig. 2C).
IL-32 Binds to ␣V␤3 and ␣V␤6 Integrins-RGD amino acid sequences that are localized in extracellular matrix proteins interact with integrins on cell surfaces and are important for several processes such as adhesion, migration, proliferation, survival, and apoptosis (35)(36)(37)46). These RGD motifs are also present in the IL-32 isoforms (Table 1) and could potentially bind to integrins. Therefore, three integrins were tested: ␣V␤3, and ␣V␤6 integrin, whereas ␣V␤8 integrin does not bind to IL-32␣. B, IL-32␣ to ␣V␤3 binding can be inhibited by cyclo-(RGDfV), a small peptide expressing an RGD motif. C, binding of IL-32␣ to ␣V␤6 could not be inhibited by small RGD peptides but increased IL-32␣ to ␣V␤6 binding. The values are the means with S.E. One-way ANOVA with Bonferroni's multiple comparison test was used (n ϭ 4; **, p Ͻ 0.01; ***, p Ͻ 0.001).

IL-32 Binds to FAK and Paxillin, Both Members of Focal Adhesion Protein
Complex-Modeling of IL-32 revealed a typical structure of ␣-helixes, which resembles FAT. FAT localizes FAK-1 toward focal adhesions that are formed after integrin/ extracellular matrix engagement. FAT binds to focal adhesion protein paxillin, which results in intracellular signaling (53,54). Fig. 6A shows that IL-32␣ can bind to paxillin, whereas the interaction could not be inhibited by cyclo-(RGDfV), demonstrating that the RGD motif is not involved in the IL-32␣/paxillin engagement. Moreover, the IL-32␣/paxillin interaction was inhibited by recombinant FAK-1, containing the FAT region that binds to paxillin (Fig. 6B). Remarkably, the IL-32␣/ ␣V␤3 interaction was slightly inhibited by FAK, indicating that IL-32␣ also binds to FAK (Fig. 6C). Fig. 6D demonstrates that IL-32␣ and FAK interact with each other.

DISCUSSION
IL-32 does not contain any conversed domains in its amino acid sequence, which complicated the modeling. Nowadays, sophisticated modeling software can predict secondary and tertiary structures based on amino acid sequences. By I-TASSER, the secondary structure of IL-32␣, IL-32␤, and IL-32␥ was pre-dicted and revealed ␣-helixes with short coils but no ␤ sheets. Subsequently, I-TASSER predicted the tertiary structure by comparing the secondary structure of known proteins with the IL-32 predicted secondary structure that resulted in an ␣-helix bundle shape-like protein. IL-32␥ contains a potential transmembrane helix specific for IL-32␥, which complicated the modeling. HMMTOP software predicted a transmembrane helix specific for IL-32␥ and not for IL-32␣ or IL-32␤. Supporting evidence for a transmembrane helix in IL-32␥ is provided by the observation that IL-32␥ secretion is enhanced by TNF␣ or IL-1␤ stimulation in rheumatoid arthritis fibroblast-like synoviocytes without cell death being involved (4). More evidence is provided by the observation that intestinal epithelial cells that were stimulated with TNF␣ and INF␥ showed subcellular colocalization of IL-32 with cell membranes of lipid droplet-like structures (29). Interestingly, PR3, a serine proteinase binds to IL-32 (33) and can cleave IL-32␥ into a more active form (32). Moreover, expression of membrane PR3 is elevated in rheumatoid arthritis patients (55), which might release transmembrane IL-32␥. PR3-released IL-32␥ could be an explanation for the increased IL-32 levels observed in synovial fluid (31), although cell death could be involved.
IL-32 contains two interesting amino acid motifs, namely DDM and RGD. These two motifs are involved in integrin signaling and activation of procaspase-3 (46,48). By modeling, it became clear that the IL-32␣ RGD motif is differently orientated and probably less accessible for binding compared with the IL-32␤ or IL-32␥ RGD motif. In integrin signaling, the RGD motifs expressed by extracellular matrix proteins (37) can bind to DDM motifs present in ␤-integrins (56). Moreover, RGD and DDM motifs are also present in procaspase-3, and small peptides expressing RGD motifs can directly activate procaspase-3, resulting in active caspase-3 leading to apoptosis (46). In addition, RGD-induced cell death, probably by activation of caspase-3, was reported in chondrocytes and synovial cells that were exposed to small RGD peptides (47). Because RGD motifs can bind to DDM motifs in the integrin signaling, it is plausible that the RGD and DDM motifs located in procaspase-3 bind to each other, leading to a self-inhibitory conformational state, whereas small RGD peptides probably interfere with this binding and cause conformational changes leading to autoactivation of procaspase-3 as suggested by Porter (48). However, the RGD motif present in IL-32␤ and IL-32␥ is not directly involved in the IL-32-induced cell death, because deletion of the RGD motif did not show decreased cytotoxicity.
The RGD motif was first found in fibronectin (57), which is an extracellular matrix protein. The fibronectin receptor was discovered through binding with fibronectin in an RGD-dependent manner (58,59), which demonstrated the importance of the RGD motif. This fibronectin receptor is now known as ␣5␤1 integrin (60). Soon after that initial report, it became clear that the RGD motif is the cell adhesion site of many other adhesion proteins such as vitronectin, fibrinogen, von Willebrand factor, thrombospondin, and osteopontin (60) that bind to integrins. Integrins are membrane-bound receptors containing an ␣ and ␤ subunit forming a heterodimer (35). Furthermore, integrins belong to the type I transmembrane glycoproteins containing a short cytoplasmic tail (35), and nowadays, 18 dif- Interactions between IL-32, Integrins, Paxillin, and FAK-1 FEBRUARY (61,62). The crystal structure of ␣V␤3 together with small RGD peptides demonstrated that the RGD motif binds at the major interface between the ␣V and ␤3 subunits and interacts with both (63). Integrins can signal bidirectional, which mean that binding to the extracellular matrix is regulated from the inside of the cell, whereas binding to the extracellular matrix triggers signals that are transferred into the cell (37). Integrin signaling is essential for many important cell processes such as anchorage, proliferation, migration, survival, cytokine production, and differentiation (35,37). Many of these properties are observed for IL-32, and the question arose regarding whether the RGD motif present in IL-32 could bind to integrins. In the present study, we show that IL-32␣ binds to ␣V␤3, ␣V␤6, but not to ␣V␤8 integrins. All three integrins can bind RGD motifs; however, only the ␣V␤3/IL-32␣ interaction could be inhibited by small RGD peptides. It is possible that the ␣V␤6/IL-32␣ interaction is not mediated via the RGD motif or that the small RGD peptides bind to the DDM motif present in IL-32␣, thereby altering the tertiary structure of IL-32␣ and resulting in more free RGD motifs that can bind to ␣V␤6. Furthermore, IL-32␤ was able to bind to ␣V␤3, but the interaction could not be inhibited by small RGD peptides, suggesting that besides the RGD motif, other sites in IL-32␤ can bind to the ␣V␤3 integrin. Previous experiments showed reduced production of cytokines by human peripheral blood mononuclear cells that were stimulated with recombinant IL-32␥ in combination with serum, in contrast to serum-free conditions. Culturing cells often require attachment to a surface, which is facilitated by serum adhesive proteins. In FCS, serum adhesive proteins such as fibronectin and vitronectin are present (64, 65) that contain RGD motifs. Presumably, these serum adhesive proteins inhibit the IL-32/integrin interactions through their RGD motifs, resulting in reduced cytokine production. As expected, FIGURE 7. A proposed model for the role of IL-32 in the integrin/FAK-signaling pathway. A, presence of extracellular IL-32 isoforms released by dying cells that interact with integrins through RGD and non-RGD motifs and thereby activating integrin/FAK-signaling pathways. B, RGD motifs expressed by extracellular matrix that bind to integrins and activate integrin/FAK-signaling pathways. C, cell/cell interactions through membrane-bound IL-32␥, which binds to integrins present on other cells. D, release of membrane-bound IL-32␥ by PR3 cleavage that interacts with integrins and activates integrin/FAK-signaling pathways. Besides the activation of the integrin/FAK-signaling pathway by extracellular IL-32, intracellular IL-32 modulates the integrin/FAK-signaling pathway by interacting with paxillin (PAX) and FAK (A-D).
by adding 10% FCS, the interaction between IL-32␣ and ␣V␤3 integrins was significantly inhibited. When RGD motif-containing proteins bind to integrins, a signal is transduced through the cell membrane, which activates intracellular kinases such as FAK. FAK is targeted to focal adhesion formations, which are formed by extracellular matrix/integrin clustering, guided via the FAT region. Interestingly, inhibition of FAK tyrosine 397 phosphorylation regulates the IL-32␤/IL-32␥-induced cytotoxicity. A low concentration of the FAK inhibitor resulted in increased cytotoxicity, which reflects other reports describing increased cell death when FAK is inhibited (41,42). On the contrary, a high concentration of the FAK inhibitor significantly reduced the IL-32␤/-32␥-induced cytotoxicity. This dual effect is rather difficult to understand; however, others reported successful inhibition of FAK tyrosine 397 phosphorylation by a small inhibitor but failed to induce apoptosis (43). On the other hand, the FERM domain of FAK can bind to the catalytic domain of FAK, resulting in autoinhibition (66). Nevertheless, the IL-32␤/IL-32␥-induced cytotoxicity includes FAK activity.
Several reports show that IL-32 can synergize with bacterial fragments for the production of proinflammatory cytokines (3,15). Could there be a link between bacteria and FAK activation? The answer is yes, it is reported that FLS stimulated with protein I/II, a modulin from Streptococci, which binds to ␣5␤1 integrins, induces IL-6 and CXCL8 production via the ERK1/2 and JNK pathways that requires FAK activation (67). Interestingly, the important tyrosine 397 of FAK is not involved in this increased IL-6 and CXCL8 production, indicating that other tyrosines present in FAK are important too. Integrin ␣5␤1 can also bind to RGD motifs (68), whether the RGD motif present in IL-32 can bind to ␣5␤1 remains unknown, and it would be interesting to demonstrate that extracellular IL-32 can interact with integrins and thereby inducing proinflammatory cytokines. Furthermore, IL-32 can synergize with TLR-2 and NOD2 fragments by enhancing both receptors (3). Interestingly, FAK and TLR pathways are interconnected via MyD88 (69,70), and IL-32 fulfills a key role in this process as shown by IL-32 silencing or overexpression (4,30).
Modeling of IL-32 revealed a typical ␣-helix bundle protein that resembles FAT (54). The FRNK domain contains the FAT region that guides FAK to focal adhesions and thereby connects the integrin-signaling pathway. FRNK can be transcribed by using an alternative promoter present in the FAK gene (71), which inhibits FAK signaling (72)(73)(74). Possibly, IL-32 that resembles FAT acts like an intracellular inhibitor of FAK signaling as observed for FRNK, which explains IL-32-induced cell death. IL-32 can also induce proinflammatory cytokines involved in rheumatoid arthritis (3)(4)(5)(6), which indicates that IL-32 can enhance integrin signaling. Perhaps membranebound IL-32␥ signals via cell-cell contact using its RGD motif to interact with integrins and thereby inducing FAK activation. Another possibility is that membrane-bound IL-32␥ is processed by PR3, because PR3 binds (33) and processes/activates (32) IL-32␥, resulting in integrin binding and FAK activation. A different option is that IL-32 modulates FAK activation, as a member of the focal adhesion protein complex, because we demonstrated that IL-32 interacts with both paxillin and FAK.
The involvement of IL-32 in the integrin-signaling pathway as proposed in Fig. 7 would explain the many functions reported for IL-32 such as differentiation (26 -28), inflammation (3-7, 15, 16), and apoptosis (25) that stimulates our imagination to think of the many possibilities of IL-32.