The membrane-distal regions of integrin α cytoplasmic domains contribute differently to integrin inside-out activation

Functioning as signal receivers and transmitters, the integrin α/β cytoplasmic tails (CT) are pivotal in integrin activation and signaling. 18 α integrin subunits share a conserved membrane-proximal region but have a highly diverse membrane-distal (MD) region at their CTs. Recent studies demonstrated that the presence of α CTMD region is essential for talin-induced integrin inside-out activation. However, it remains unknown whether the non-conserved α CTMD regions differently regulate the inside-out activation of integrin. Using αIIbβ3, αLβ2, and α5β1 as model integrins and by replacing their α CTMD regions with those of α subunits that pair with β3, β2, and β1 subunits, we analyzed the function of CTMD regions of 17 α subunits in talin-mediated integrin activation. We found that the α CTMD regions play two roles on integrin, which are activation-supportive and activation-regulatory. The regulatory but not the supportive function depends on the sequence identity of α CTMD region. A membrane-proximal tyrosine residue present in the CTMD regions of a subset of α integrins was identified to negatively regulate integrin inside-out activation. Our study provides a useful resource for investigating the function of α integrin CTMD regions.

. Sequence and structure diversity at the membrane -distal region of α integrin cytoplasmic domain. (A) Sequence alignment of α and selected β cytoplasmic tails (CTs) of human integrins. The membraneproximal (MP) and the membrane-distal (MD) regions are indicated. Highly conserved residues are in blue. Tyrosine and methionine residues in the MD region that are conserved within a subset of α integrins are shown in red. Other tyrosine residues of interest that are adjacent to the GFFKR motif in several α subunits are shown in magenta. The lengths of the MD regions are in parentheses. The binding sites of talin and kindlin on integrin β CTs are indicated with dashed lines. (B) Comparison of the reported α integrin CT structures. PDB codes or references are shown below the corresponding structures. The structures are color-coded as indicated on the right. All the structures are superimposed onto the α IIb structure (PDB code 1M8O) based on the GFFKR region and presented separately. Side chains of the GFFKR motif and the conserved tyrosine residues in the MD regions are shown as sticks.  3 to the α IIb chimeras. Binding of the mAbs was measured by flow cytometry with HEK293FT cells co-transfected with the indicated integrin constructs and GFP or GFP-TH. The GFP and integrin double-positive cells were analyzed. β 3 -D723A was used to increase the sensitivity of the assay. Data are presented as the MFI of the mAb binding normalized to the MFI of integrin expression. Data are presented as mean ± s.e.m. (n ≥ 3) and two-tailed Student's t-test was performed to compare the α IIb -chimeras to α IIb -WT under the GFP-TH condition in B and D or under same GFP-TH concentration in C; *P < 0.05, **P < 0.01. Integrin and GFP-TH expression levels were presented in MFI in the lower panel in B. n = 2 for α IIb -α 1 in D.
Scientific REPORTS | (2018) 8:5067 | DOI: 10.1038/s41598-018-23444-w Integrin α IIb -chimeras bearing various α CTMD regions respond differently to talin head (TH) stimulation. The β 3 integrin's partners α IIb and α V subunits share 6 consensus residues at their CTMD regions, but the α V CTMD region is about two times longer than that of the α IIb (Fig. 1A). We have shown that complete deletion of the CTMD regions of α IIb and α V subunits abolished TH-induced α IIb β 3 and α V β 3 activation 13 . Here, we asked whether the CTMD regions of α IIb and α V subunits could be exchangeable and whether they could exert different effect on β 3 integrin inside-out activation. The ligand-mimetic mAb PAC-1 was used to access the α IIb β 3 activation induced by the overexpression of GFP-TH in the presence of the β 3 cytoplasmic mutation β 3 -D723A, which has been shown to greatly enhance the responsiveness of TH-induced α IIb β 3 activation 13 . When the CTMD region of α IIb was replaced by the α V CTMD region, the α IIb -α V /β 3 -D723A chimeric integrin still remained responsive to GFP-TH-induced activation (Fig. 2B). However, the activation of α IIb -α V was significantly decreased compared with the α IIb -WT (Fig. 2B). The reduced activation of α IIb -α V chimera was not due to the differences in GFP-TH expression since the lower activity of α IIb -α V compared with α IIb -WT was consistently seen at various levels of GFP-TH expression (Fig. 2C). As a comparison, we replaced the α IIb CTMD region with those of α 1 and α L integrins that do not heterodimerize with β 3 integrin. Remarkably, the presence of both the α 1 and α L CTMD regions significantly enhanced the GFP-TH-induced activation of α IIb β 3 integrin (Fig. 2B). The increased activation of α IIb -α 1 and α IIb -α L was also obvious in the absence of TH expression (Fig. 2B), indicating that the mismatch of the α IIb CTMD mutant with the β 3 CT renders α IIb β 3 more active than the wild type. This may be due to the destabilization of α IIb -β 3 CT interaction, being consistent with the previous observations that the α CTMD regions contribute to maintaining integrin in the resting state 13,14 . The expression level of α IIb -α L was decreased possibly due to the high integrin activity (Fig. 2B), which is commonly seen among the active integrin mutants 13 . We next asked whether the replacement of α IIb CTMD region affects the TH-induced conformational change of α IIb β 3 integrin. The active conformation-specific mAb 370.3 was used to report the extension of α IIb integrin. Consistent with the PAC-1 binding assay, the α IIb -α V chimera showed decreased while the α IIb -α 1 and α IIb -α L chimeras showed increased binding of mAb 370.3 either in the presence or absence of TH expression (Fig. 2D). This data demonstrates that the CTMD regions of α IIb and α V are not completely interchangeable. They can exert different effect on β 3 integrin activation at least in part through regulating the conformational change of integrin.
Replacing the α L CTMD region with that of α X , α D or α M subunit reduced TH-mediated α L β 2 integrin activation. Integrin β 2 subunit forms heterodimers with α L , α M , α X , and α D subunits. α L has the longest while α M has the shortest CTMD sequence among the four subunits (Fig. 1A). The NMR structures of α L , α M and α X CTs show great structural heterogeneities at their MD regions (Fig. 1B). In addition, we found that deletion of the α L CTMD region abolished, while truncation of the α L CTMD region dampened TH-induced α L β 2 activation 13 , arguing a potential regulatory role of the CTMD region. Similar to the α IIb chimeras, we made the α L chimeras by replacing the α L CTMD region with that of α M , α X or α D subunit. Surprisingly, in the TH-induced ICAM-1 binding assay, all the α L -α M , α L -α X, and α L -α D chimeras showed significantly reduced ICAM-1 binding compared with the WT α L when co-expressed with the β 2 -D709A mutation (Fig. 3A,B). The β 2 -D709A mutation was used to increase the sensitivity of our assay by greatly enhancing TH-induced α L β 2 activation as shown in our previous study 13 . The expression levels of integrin and GFP-TH were comparable among the α L transfections (Fig. 3B). Furthermore, as shown for the α L -α M chimera in the GFP-TH titration assay, the reduced ICAM-1 binding was obvious when the GFP-TH expression reached a certain level and became independent of the expression level of GFP-TH (Fig. 3C). Similar results were obtained with the α L -α D chimera (data not shown). In addition, although all the α L constructs exhibited increased TH-induced ICAM-1 binding with the increase of ICAM-1 concentration, all the α L chimeras consistently showed reduced ICAM-1 binding at all the ICAM-1 concentrations tested (Fig. 3D). These data demonstrated that the reduced activation of α L chimeras was due to the replacement of CTMD region and might attribute to a common feature of α M , α X , and α D CTMD regions.
A conserved tyrosine residue in the CTMD regions of α M , α X , and α D subunits negatively regulates the inside-out activation of β 2 integrin. Sequence alignment revealed a conserved tyrosine residue at the second position of all the CTMD regions of α M , α X , and α D subunits (Fig. 1A). We asked whether this tyrosine contributes to the reduced activation of α L chimeras. We first tested the tyrosine mutation on α L -α X chimera with the co-expression of β 2 -D709A mutant. Compared with the α L -α X , the tyrosine to phenylalanine mutation at the α X CTMD region, α L -α X -Y1117F, significantly increased the TH-induced ICAM-1 binding and restored it to the WT α L level (Fig. 4A). Similarly, the α L -α D -Y1115F mutation also increased the TH-induced ICAM-1 binding although it was not to the WT level and not statistically significant (Fig. 4B). We further tested the tyrosine mutation on the α L -α M chimera. Mutating the tyrosine in the α M CTMD sequence to phenylalanine (Y1121F), glutamic acid (Y1121E), and alanine (Y1121A) all significantly increased the TH-induced ICAM-1 binding of α L -α M chimera (Fig. 4C). Interestingly, the Y1121A mutation exerted a higher level of activation than the Y1121F and Y1121E mutations (Fig. 4C), indicating that both the bulky side chain and the hydroxyl group of tyrosine are important for its negative effect on integrin activation. Sequence alignment also shows a conserved methionine at the fifth position of all the α M , α X , and α D regions (Fig. 1A). We tested whether this methionine residue contributes to the negative effect of CTMD region using the α L -α M chimera. We found that the α L -α M -M1124A mutation did not significantly increase TH-induced ICAM-1 binding (Fig. 4C). The expression levels of integrin and GFP-TH were comparable among the transfections within the same experimental group (Fig. 4A-C). These data clearly demonstrate that the negative effects of the CTMD regions of α M , α X , and α D subunits on TH-induced β 2 integrin activation at least in part attribute to the presence of a conserved tyrosine.
The position of the tyrosine mutation at the CTMD region is critical for its negative regulation of the α L β 2 integrin inside-out activation. Interestingly, there is only one tyrosine in the CTMD region of α M , α X , or α D subunit and no tyrosine in the α L CTMD region (Fig. 1A). To further demonstrate the important regulatory role of a tyrosine residue in the CTMD region, we performed a tyrosine scanning mutagenesis for the α L CTMD region. A tyrosine mutation was placed at the 1 st , 2 nd , 3 rd , 4 th , or 6 th position of the CTMD region after the conserved GFFKR motif (Fig. 5A). As shown in the TH-induced ICAM-1 binding assay, the tyrosine mutation at the 1 st position, α L -N1095Y, significantly enhanced ICAM-1 binding (Fig. 5B). This is probably due to the disturbance of α-β association at the GFFKR motif, which was known to be important in maintaining integrin in the resting state 11 . In contrast, the tyrosine mutation at the 2 nd position, α L -L1096Y, which is equivalent to the native tyrosine of the α X , α D and α M CTMD regions, significantly reduced TH-induced ICAM-1 binding (C) ICAM-1 binding of α L -WT and α L -α M chimera in response to the different levels of GFP-TH expression. α L integrins were co-transfected with β 2 -D709A and the indicated amounts of GFP-TH plasmids into HEK293FT cells. For B and C, data are presented as the ICAM-1 MFI normalized to α L MFI and shown as mean ± s.e.m. (n ≥ 3). Two-tailed Student's t-test was performed to compare the α L -chimeras with α L -WT in the presence of GFP-TH in B or under same GFP-TH concentration in C. *P < 0.05; **P < 0.01; ***P < 0.001. (D) Dose response curves of ICAM-1 binding to α L -WT and α L -chimeras. HEK293FT cells were transfected with the α L integrins plus β 2 -D709A and GFP-TH. Different concentrations of ICAM-1 were used for the binding assay. Data are presented as the percentage of maximum ICAM-1 binding of each experimental repeat and shown as mean ± s.e.m. (n ≥ 3). Two-tailed Student's t-test was performed to compare the α L -chimeras with α L -WT under the same ICAM-1 concentration; *P < 0.05; **P < 0.01. ICAM-1 binding was statistically lower for all α L -chimeras compared to α L -WT at the indicated ICAM-1 concentrations, but the analyses were shown for comparison between the α L -WT and α L -α X chimera.
Scientific REPORTS | (2018) 8:5067 | DOI:10.1038/s41598-018-23444-w (Fig. 5B). This is consistent with the above data. Remarkably, the tyrosine mutations at the 3 rd and 4 th positions, α L -K1097Y and α L -E1098Y, also significantly reduced ICAM-1 binding (Fig. 5B). However, the tyrosine mutation at the 6 th position, α L -M1100Y, had no significant effect on ICAM-1 binding (Fig. 5B). Among the single tyrosine mutations, the tyrosine mutation at the 3 rd position, α L -K1097Y, had the most negative effect (Fig. 5B). We next asked whether the presence of multiple tyrosine mutations at the CTMD region has a synergistic effect on its negative regulation of integrin activation. A triple tyrosine repeat was introduced into the 2 nd to 4 th positions of α L CTMD region (Fig. 5A). The α L -YYY mutation significantly reduced ICAM-1 binding compared with the wild type, but had no significant difference with the single α L -K1097Y mutation (Fig. 5B). Furthermore, the negative The relatively conserved tyrosine was mutated to phenylalanine in all α L -chimeras (A-C), and to glutamic acid or alanine in the α L -α M chimera (C). A conserved methionine was also mutated to alanine in the α L -α M chimera (C). Binding of ICAM-1 was measured by flow cytometry with HEK293FT cells co-transfected with the indicated α L constructs plus β 2 -D709A and GFP or GFP-TH. The GFP and integrin double-positive cells were analyzed. Data are presented as the MFI of the ICAM-1 binding normalized to integrin expression and shown as mean ± s.e.m. (n ≥ 3). Twotailed Student's t-test was performed to compare the α L -chimeras or their mutants to α L -WT under the GFP-TH condition, or as indicated (*P < 0.05; **P < 0.01; ***P < 0.001; n.s., not significant). Integrin and GFP-TH expression levels were presented in the lower panel.
Scientific REPORTS | (2018) 8:5067 | DOI:10.1038/s41598-018-23444-w effect of the tyrosine mutation did not depend on the ICAM-1 concentration (Fig. 5C). These data demonstrate that the presence and the position but not the number of tyrosine mutations at the CTMD region are important in the negative regulation of α L β 2 activation. The α L -K1097Y mutation coincidently formed an Yxxɸ motif (x is any amino acid, ɸ is hydrophobic residue), which has been found recently in a subset of α integrins to play a role in the regulation of integrin endocytosis 18 . It has been shown that the presence but not the position of the Yxxɸ motif is important for its function in integrin endocytosis 18 . To test whether the negative effect of α L -K1097Y mutation was due to the formation of Yxxɸ motif, we generated another tyrosine mutation, α L -G1103Y, which formed an Yxxɸ motif (Fig. 5A). In contrast to the α L -K1097Y mutation, the α L -G1103Y mutation showed no difference with the α L -WT in ICAM-1 binding (Fig. 5D), suggesting that the negative effect of α L -K1097Y mutation is not due to the presence of an Yxxɸ motif.
In our TH-induced ICAM-1 binding assay, all the α L constructs were co-expressed with the β 2 -D709A mutation. To demonstrate that the differences we observed among the α L mutations in the integrin activation assay are not due to the presence of the β 2 -D709A mutation, we did the same assay in the presence of β 2 -WT for several representative tyrosine mutations. The results show that all the selected α L tyrosine mutations reduced TH-induced ICAM-1 binding although not as significant as in the presence of the β 2 -D709A mutation (Fig. 5E).
The α CTMD regions contribute to integrin activation by regulating the conformational change of integrin. TH-induced integrin activation is coupled with the large-scale conformational changes of integrin extracellular domain 5,10 . We used two mAbs, KIM127 and m24, which report β 2 integrin extension and headpiece opening, respectively 19 , to test whether the mutagenesis of α L CTMD region affect TH-induced α L β 2 conformational change. Consistent with the ICAM-1 binding assay, the α L -α M , α L -α X, and α L -α D chimeras all significantly reduced TH-induced binding of both m24 and KIM127 mAbs when compared with the α L -WT (Fig. 6A). Similarly, the α L tyrosine mutations, α L -L1096Y and α L -YYY, also decreased the TH-induced m24 or KIM127 binding to α L β 2 (Fig. 6B). These data demonstrate that the α CTMD region contribute to integrin inside-out activation through regulating the large-scale conformational changes.
Having found that introducing a tyrosine residue into the specific position of α L CTMD region negatively regulates α L β 2 ligand binding and conformational change, the next question is whether the native tyrosine residue present in the α M , α X, or α D CTMD region plays a role in regulating the activation of these integrins. To answer this question, we performed the TH-induced activation assay for α M β 2 integrin by detecting the β 2 integrin headpiece opening using m24. Consistent with the α L β 2 , the presence of β 2 -D709A mutation greatly enhanced TH-induced binding of m24 to α M β 2 (Fig. 6C). When replacing the conserved tyrosine at the 2 nd position of α M CTMD region (Y1121) with either leucine or alanine, the TH-induced binding of m24 was further increased significantly in the presence of β 2 -D709A (Fig. 6C). The same effect was found when the α M -Y1121L or α M -Y1121A was co-expressed with the β 2 -WT (Fig. 6D). Interestingly, the enhanced activation by α M -Y1121A is more obvious than α M -Y1121L when they were co-expressed with β 2 -WT (Fig. 6D), again suggesting that the bulky side chain of tyrosine is important for the negative effect on integrin activation. The expression level of integrin or GFP-TH is comparable among the transfections (Fig. 6C,D). These data, in addition to the α L tyrosine mutations that exerted the opposite effect on α L β 2 integrin activation, clearly demonstrate that a specific tyrosine residue present in a subset of α integrin CTMD regions negatively regulates β 2 integrin inside-out activation.
Mutagenesis of the α IIb and α L CTMD regions does not affect the TH binding to the integrin β 3 and β 2 CTs. Structural and functional studies have demonstrated that TH binds to the integrin β CT to induce integrin activation 10 . To test whether the α CTMD region contributes to integrin activation by affecting the TH association with the β CT, we did co-immunoprecipitation assay for the GFP-TH and integrin β subunit co-expressed with the α CTMD mutants. For the α IIb β 3 transfectants, the cell lysates were precipitated with anti-GFP antibody and the associated β 3 integrins were detected by the anti-β 3 antibody. The β 3 WT or β 3 -D723A was robustly detected in the anti-GFP pull-down only when the GFP-TH but not GFP was present, demonstrating the specific interaction between TH and β 3 subunit (Figs 7A and S1). The expression levels of GFP-TH and β 3 integrin were comparable among the transfectants according to the Western blot of whole cell lysate (Figs 7A and S1). No obvious differences for the TH-bound β 3 -D723A were observed among the α IIb WT and α IIb chimeras, suggesting that swapping the α IIb CTMD region with that of α V , α 1 , or α L does not affect the association of TH with β 3 subunit.
We next performed the same assay for the α L CTMD mutations. To simultaneously detect both the α L and β 2 subunits in the anti-GFP pull-down assay, we did the cell surface biotinylation before lysing the cells for co-immunoprecipitation. The presence of α L and β 2 bands was confirmed by anti-α L pull-down using the anti-α L specific mAb TS2/4. As shown in Fig. 7B, two bands that correspond to the α L and β 2 subunits were readily detected in the anti-GFP pull-down only when both the α L β 2 and GFP-TH were co-expressed. No α L and β 2 bands were detected in the anti-GFP pull-down in the transfections of GFP-TH alone or α L β 2 plus GFP (Figs 7B and S1). The expression levels of integrin and GFP-TH were comparable among the transfections detected by flow cytometry (data not shown). To compare the association of GFP-TH and the α L β 2 constructs, we quantified the Western blots by normalizing the α L and β 2 signals to the GFP-TH signals (Fig. 7C). Compared with the α L WT, only the α L -α D chimera shows an obvious increase in the association of integrin and GFP-TH (Fig. 7C). This is in contrast to the decrease in TH-induced α L β 2 activation as shown above. No significant differences were found among the α L WT, α L -α X , α L -α M , and the α L tyrosine mutations, α L -L1096Y, α L -K1097Y, and α L -M1100Y (Fig. 7C). These data indicate that the regulatory function of α CTMD region on integrin activation should not be due to the effect on the TH and β CT association.
Integrin α 5 -chimeras bearing various α CTMD regions respond differently to TH stimulation when paired with the same integrin β 1 subunit. Twelve integrin α subunits share the integrin β 1 subunit, making the largest β 1 integrin subfamily (Fig. 1A). The major fibronectin receptor, integrin α 5 β 1 , has been relatively well studied structurally and functionally. Like α IIb β 3 and α L β 2 , the α 5 β 1 integrin can be activated by the overexpression of TH 20 . We have demonstrated the important role of the CTMD regions of α IIb , α V , and α L Scientific REPORTS | (2018) 8:5067 | DOI:10.1038/s41598-018-23444-w subunits in TH-induced integrin activation 13 . Here, we extended our study to α 5 β 1 integrin and asked whether the α 5 CTMD region follows the same rule. Using the α 5 β 1 -deficient CHO-B2 cell line 21 , we found that complete deletion of the α 5 CTMD region abolished GFP-TH-induced binding of fibronectin fragment Fn9-10 to α 5 β 1 integrin Data are presented as the MFI of bound m24 or KIM127 normalized to the MFI of integrin expression, and shown as mean ± s.e.m. (n ≥ 3). Two-tailed Student's t-test was performed to compare the α L mutants to α L -WT in A and B, or to compare the α M mutants to α M -WT, or as indicated under the GFP-TH condition in C and D (*P < 0.05; **P < 0.01; ***P < 0.001). Integrin and GFP-TH expression levels were presented in MFI in lower panel of C and D. One representative experiment was shown for the KIM127 binding in B.
( Fig. 8A), demonstrating the requirement of α 5 CTMD region in α 5 β 1 inside-out activation. The next question is whether the diverse α CTMD regions of the β 1 integrin family also differently regulate β 1 integrin activation. To answer this question, we compared the function of all the α CTMD regions of the β 1 integrin family in the context of α 5 subunit. Eleven α 5 chimeras were generated by replacing the α 5 CTMD region with that of α subunits as indicated in Figs 2A and 8B. The TH-induced fibronectin-binding assay for α 5 β 1 was performed using the α 5 β 1 -knockout HEK293FT cells. The activating β 1 -K732E mutation, located at the transmembrane domain 22 , was used to enhance the sensitivity of the assay. As shown in Fig. 8B, the β 1 -K732E significantly increased TH-induced fibronectin binding compared with the β 1 WT. Among the α 5 chimeras, two groups were identified: one group has no but the other has significant effect on TH-induced fibronectin binding compared with the α 5 WT (Fig. 8B). Remarkably, all the α 5 chimeras, including α 5 -α 4 , α 5 -α 9 , α 5 -α 3 , and α 5 -α 6 , which have a tyrosine residue adjacent to the GFFKR motif (Fig. 1A), showed comparable fibronectin binding with the α 5 WT that also contains a tyrosine at the CTMD region (Figs 1A and 8B). In contrast, all the α 5 chimeras, including α 5 -α 1 , α 5 -α 7 , α 5 -α 8 , and α 5 -α 10 , α 5 -α 11 , and α 5 -α V , which lack the tyrosine residue adjacent to the GFFKR motif (Fig. 1A), rendered α 5 β 1 more active than the WT (Fig. 8B). However, an exception is the α 5 -α 2 chimera, which has an equivalent tyrosine at the α 2 CTMD region but significantly increased α 5 β 1 activation (Figs 1A and 8B). The similar results were obtained when the selected α 5 chimeras were co-expressed with the β 1 WT in the TH-induced fibronectin binding assay (Fig. 8C). The α 5 -α 2 and α 5 -α 10 significantly increased α 5 β 1 activation while the α 5 -α 4 had no obvious effect (Fig. 8C). By contrast, the enhanced effect of α 5 -α 7 was not detectable when the β 1 WT was used (Fig. 8C), indicating the low sensitivity of the assay. Finally, we correlated the TH-induced α 5 β 1 activation with the large-scale conformational change of β 1 ectodomain using the active conformation dependent mAb 9EG7. The presence of β 1 -K732E significantly enhanced the TH-induced 9EG7 binding. Consistent with the fibronectin binding assay, the α 5 -α 2 , α 5 -α 7 , and α 5 -α 10 significantly increased 9EG7 binding, but the α 5 -α 9 had no such Figure 7. Swapping or mutating the residues in the CTMD region of integrin α IIb or α L does not dramatically affect the TH binding to the β integrin CT. (A) TH binding to β 3 integrin CT in the presence of α IIb -chimeras. To detect the interaction between GFP-TH and integrin β 3 CT, GFP-TH was immunoprecipitated using anti-GFP antibody from the lysates of HEK293FT cells transfected with GFP (control) or GFP-TH, plus indicated α IIb and β 3 integrin constructs. The associated integrin was detected by immunoblot with anti-β 3 antibody. Expression level of β 3 , GFP or GFP-TH was accessed by immunoblots using whole cell lysates (WCL). β-actin was blotted as a loading control. The asterisk indicates a non-specific band. The membranes were cut and blotted separately. (B) TH binding to β 2 CT in the presence of α L -chimeras or α L tyrosine mutants. HEK293FT cells were cotransfected with the indicated α L constructs plus β 2 -D709A and GFP or GFP-TH. The transfected cells were biotinylated prior to the anti-GFP immunoprecipitation as described in A or anti-α L immunoprecipitation as a control. The GFP-TH associated integrin was detected by blotting with the IRDye ® 800CW Streptavidin.
The immunoprecipitated GFP or GFP-TH were detected by anti-GFP antibody. The results were from the same gel. The membrane was cut and blotted separately. (C) Quantitation of the blotting results of B. Integrin α L or β 2 signals were first normalized to the corresponding GFP or GFP-TH signals, and then presented as a percentage of the normalized α L -WT or β 2 signal to the α L -WT control under the GFP-TH condition. Data are mean ± s.e.m. (n ≥ 3) except for the β 2 signal with α L -M1100Y (n = 2). Please also see Fig. S1 for panels A and B.
Scientific REPORTS | (2018) 8:5067 | DOI:10.1038/s41598-018-23444-w effect compared with the α 5 WT (Fig. 8D). These data suggest that the α CTMD regions of the β 1 family could contribute differently to β 1 integrin inside-out activation. They may also follow the rule of the negative regulation by a tyrosine residue.

Discussion
Compared with the extensive structural and functional studies of the relatively conserved β integrin CT that serve as docking sites for many signaling molecules, little is known about the role of α integrin CT especially the non-conserved MD regions. Since many α integrin subunits share the same β subunit, it is tempting to speculate that on one side the α CTMD regions may be interchangeable; on the other side, they may provide the specificity for integrin function. One of the difficulties in studying the α integrin CTMD regions is the sequence diversity among 18 α subunits. Another challenge is the lack of well-established activation assays for many integrin members, which limits the functional studies of the CTMD regions for many α integrins. Our approach in the current study provides a useful tool to thoroughly examine the potential functions of the CTMD regions of almost all integrin α subunits. By putting the different α CTMD regions in the context of the α IIb , α L , or α 5 subunit, this approach made it possible to compare the function of different α CTMD regions.
The potential function of several individual α integrin CTMD regions in integrin activation had been indicated in previous studies more than 20 years ago. They showed that deletion of the α CTMD region diminished cell adhesion or migration mediated by α 1 , α 2 , α 4 , α V , and α 6 integrins 23-30 and dampened PMA-induced activation of α L β 2 integrin 31 . Direct evidence for a role of α CTMD region in integrin inside-out activation was provided by the observation that complete deletion of the CTMD region of α IIb , α V , or α L integrin abolished Binding of the fibronectin type III domains 9-10 fragment (Fn9-10) was measured by flow cytometry with CHO-B2 cells transfected with the indicated α 5 β 1 constructs plus GFP or GFP-TH. (B) Fibronectin (Fn) binding of the α 5 chimeras co-expressed with β 1 -K732E. To make the α 5 chimeras, the α 5 CTMD region was replaced with those of α integrins that can pair with β 1 subunit. (C) Fn binding of selected α 5 chimeras co-expressed with β 1 -WT. (D) mAb 9EG7 binding of selected α 5 -chimeras co-expressed with β 1 -K732E. For B-D, Fn or 9EG7 binding was measured with HEK293FT-α 5 β 1 -KO cells transfected with the indicated α 5 β 1 constructs plus GFP or GFP-TH. β 1 -K732E was used to increase the sensitivity of the assay. The GFP and integrin double-positive cells were analyzed. Data are presented as the MFI of the ligand or mAb normalized to integrin expression, and shown as mean ± s.e.m. (n ≥ 3). Two-tailed Student's t-tests were performed between α 5 -chimeras and α 5 -WT, or as indicated under the same conditions. *P < 0.05; **P < 0.01; ***P < 0.001; n.s., not significant.
Scientific REPORTS | (2018) 8:5067 | DOI:10.1038/s41598-018-23444-w talin and kindlin-mediated integrin ligand binding and conformational changes 13,14 . It was suggested that the presence but not the sequence of specific residues was required for the α CTMD region to support talin-induced integrin activation 14 . In the current study, we found that replacing the α CTMD region of α IIb , α L , or α 5 integrin with those of other α integrins still maintained the capability of integrin inside-out activation mediated by the overexpression of TH, suggesting that the α CTMD region can be interchangeable for this common activation supportive function. However, we observed significant variations of the activation levels among the same α integrins carrying different CTMD regions, indicating that the α CTMD region also plays a regulatory role in integrin inside-out activation.
We found that replacing the α IIb CTMD region with that of α V integrin markedly reduced TH-induced activation of α IIb β 3 . By contrast, the α 1 and α L CTMD regions rendered α IIb β 3 more active than WT, consistent with the previous observation that replacing the α IIb CT with those of α 2 , α 5 , α 6A , or α 6B that do not natively pair with β 3 subunit enhanced α IIb β 3 activation despite they share the same GFFKR motif 32 . Thus, the native pair between the α CTMD region and β CT is important to maintain the resting state of integrin. Although the α V CTMD region shares six consensus residues with α IIb CTMD region, which is half of the length of α IIb CTMD region (Fig. 1A), it exerts different effect on α IIb β 3 activation. A predicted β turn structure formed by the PPQEE motif of α V CTMD region was suggested to regulate the conformation and ligand binding of α V β 3 26 . A similar motif PPLEE was also found in the α IIb CTMD region (Fig. 1A). Peptides containing this motif of α IIb or α V CTMD region could inhibit the activation of α IIb β 3 or α V β 3 33 , indicating that it could not be the reason for the different regulation by the α IIb and α V CTMD regions. Although the structure of the α IIb CT has been determined, it shows large conformational variations (Fig. 1B). It is not known if the structural flexibility of α IIb CT is functionally relevant. The α IIb CTMD region has a unique tandem acidic residue motif, EEDDEEGE, which is conserved among the α IIb from different species and not seen in the α V CTMD region (Fig. 1A and data not shown). These negatively charged residues might regulate the conformation of α IIb CT through repulsive interactions with the acidic phospholipid head groups at the cytosolic face of cell membrane, or through ionic interactions with the positively charged residues at the membrane-proximal region of α IIb CT as suggested by a NMR study 34 . The membrane-permeable peptides containing the α IIb CTMD region were shown to block α IIb β 3 activation in platelets 15,35 , and to inhibit the association of talin with α IIb β 3 in thrombin-activated platelets 36 . In addition, the TH domain contains several positively charged surface residues that have been shown to be important for its integrin activating function through interacting with the cell membrane [37][38][39] . It is possible that the negatively charged α IIb CTMD region may also affect the orientation of TH domain when TH encounters with β 3 CT, but a direct interaction between TH and α IIb remains to be confirmed. Being critical in hemostasis and thrombosis, the activation of α IIb β 3 is strictly regulated in platelets 40 . Our data demonstrated that the unique feature of α IIb CTMD region renders α IIb β 3 more susceptible to the signals of inside-out activation, which is in line with the high extent of activation required for α IIb β 3 function.
The lymphocyte-specific α L β 2 is another integrin whose activation is highly regulated by the inside-out signals 5,41 . The α L CT has the second longest MD region that folds into α-helical conformation as shown in an NMR structure (Fig. 1A,B). Our recent study showed that deletion of the α L CTMD region completely abolished TH-induced ICAM-1 binding to α L β 2 integrin 13 . Here we found that replacing the α L CTMD region with that of α X, α D , or α M all significantly reduced ICAM-1 binding mediated by TH. This was not determined by the differences in the length but by a common tyrosine residue seen in the α X, α D , and α M CTMD regions. Remarkably, we found that the position of the tyrosine at the α CTMD region is important to exert the negative effect on integrin inside-out activation. The bulky side chain and the hydroxyl group of tyrosine are critical for this regulatory function. Interestingly, the same negative effect was observed when introducing the tyrosine mutations into the equivalent position of the α IIb CTMD region 13 . One caveat of our study is that the negative effect of the tyrosine was found in the context of α L integrin in which no native tyrosine is present in the CTMD region. However, the physiological relevance of our discovery was built on the same observations on the α M β 2 integrin, in which mutating the native tyrosine residue greatly enhanced the inside-out activation of α M β 2 , suggesting that the tyrosine residue negatively regulates the activity of α M β 2 . The tyrosine residue is also conserved among the α M , α X , or α D integrins from different species (data not shown). Indeed, in contrast to the α L β 2 integrin that only binds selectively to ICAMs, the α M β 2 , α X β 2 , and α D β 2 integrins all exhibit multiligand-binding properties [42][43][44] . The negative regulation by a tyrosine residue in their α CTMD regions may exert a restraint to avoid hyperactivity (or to balance the activation) of these β 2 integrins, in accordance with their less-selective ligand binding functions.
Several studies have provided evidence demonstrating that the activation of α L β 2 and α M β 2 are differently regulated. Different chemokines and chemoattractants were shown to stimulate inside-out activation of α L β 2 and α M β 2 , which was suggested to be mediated by distinct pathways via the α CTs 45,46 . Given the β 2 subunit shares the common integrin activation pathway mediated by talin and kindlin, other signaling molecules may be involved in the different regulation of α L β 2 and α M β 2 probably through direct interaction with the α CT. An example is the Rap-1 interacting effector molecule RapL that specifically binds to the α L CTMD region to support Rap-1-mediated α L β 2 activation 47 . It was known that the α L -K1097 is involved in RapL binding 48 . Our results show that the α L -K1097Y mutation rendered α L β 2 the least active among the tyrosine mutations tested. However, RapL is predominantly expressed in immune cells 49 . It should not account for the decreased activation by the α L -K1097Y mutation since integrin activation was measured in HEK293FT cells lacking RapL expression. A Ser phosphorylation was found in the CTMD regions of both α L and α M subunits 50,51 . However, mutations of the Ser residue only blocked the conformational changes involved in α M β 2 but not α L β 2 activation, indicating different regulation by the α M CT. Here, we identified a tyrosine residue at the α M CTMD region that is also involved in the specific regulation of α M β 2 inside-out activation.
It remains unknown whether the activation of β 1 integrin family members are all subjected to inside-out regulation. Recent structural studies demonstrated that the conformational activation of α 5 β 1 integrin could be modulated by many components including the transmembrane and cytoplasmic domains [52][53][54][55] . We found that similar to α IIb , α V , and α L integrins the α 5 CTMD region is also required for TH-induced inside-out activation, rationalizing the use of α 5 β 1 as a model integrin to study the function of α CTMD regions. Our studies on all the α CTMD regions of the β 1 subfamily suggest a potential regulatory function of these regions in β 1 integrin inside-out activation. Remarkably, a tyrosine residue at the α CTMD region seems to play a similar negative role as seen in α M integrin in regulating α 5 β 1 inside-out activation. All the α CTMD sequences lacking a tyrosine proximal to the GFFKR motif promoted α 5 β 1 inside-out activation, while the α CTMD sequences having the tyrosine had no such effect, with the exception of α 2 CTMD region. Notably, the α 7 CTMD region contains a tyrosine at the 11 th position distal from the GFFKR motif (Fig. 1A), but the α 5 -α 7 exhibits a high level of integrin activation, consistent with the hypothesis that the membrane-proximal location of the tyrosine is important to exert an inhibitory effect. Interestingly, the α 10 CTMD region that is abundant in negatively charged residues as seen in α IIb CTMD region exerted the most dramatic effect on α 5 β 1 activation. Consistently, the α V CTMD region increased α 5 β 1 activation due to the lack of the tyrosine residue when compared with α 5 WT, but it decreased α IIb β 3 activation due to the lack of the cluster of acidic residues when compared with α IIb WT. We found that α 2 CTMD region rendered α 5 β 1 more active despite having the consensus tyrosine, suggesting that other residues of the α 2 CTMD region may counteract with the negative effect of tyrosine in regulating integrin activation.
Our study raised the question of the mechanism by which the α CTMD regions contribute to integrin inside-out activation. Previous and our current data demonstrated that the integrin α CTMD region is involved in the associations at the α/β transmembrane and CT domains, required to maintain integrin in the resting state 13 . In line with this function, our current data suggest two functional aspects of α CTMD region in integrin inside-out activation, i.e. the activation-supportive function and the activation-regulatory function. Structure analysis suggested that when binding to the β 3 CT, the talin 1 head domain might encounter steric hindrance with the α IIb CTMD residues immediately following the GFFKR motif 13,14,[56][57][58] . Such interactions may disrupt the integrin α/β association at the cytoplasmic as well as transmembrane domains, leading to an active ectodomain conformation capable of high-affinity ligand binding. This non-specific interaction is required for the activation-supportive function of α CTMD region, which could be independent of the amino acid sequences. In addition, a minimal length of two amino acids of the α CTMD region could support the integrin inside-out activation although at a reduced level 13 . This model is consistent with our observation that the α CTMD regions are interchangeable for the activation-supportive function that is not sensitive to the diversities of sequence and length.
In contrast, the activation-regulatory function of α CTMD region is dependent on certain amino acids. Several potential mechanisms are involved in this regulation. We found that the differences of the α CTMD regions in regulating integrin activation were not due to the effect of talin binding to the β CT since truncation or swapping the α CTMD region did not reduce the amount of TH bound to the β CT 13 . The specific amino acids of the α CTMD region may directly affect the conformational change of α cytoplasmic as well as transmembrane domains induced by the binding of integrin activators such as talin and kindlin. Recent studies provided evidence suggesting the conformational change of α IIb transmembrane and cytoplasmic domains in the context of full-length integrin on the cell surface 59 . In addition, different levels of integrin activation had been observed when introducing mutations into the transmembrane or cytoplasmic domains, which correlate with the different levels of ectodomain conformational changes (unpublished data). Certain amino acids such as the acidic residue clusters present in α IIb and α 10 CTMD regions and the tyrosine present in a subset of α CTMD regions may either facilitate or restrain the conformational change of α subunit as demonstrated by the active conformation-specific mAbs. The regulation of receptor activity by a tyrosine residue at the cytoplasmic domains has been seen in many cell surface receptors 60 . It was proposed that the tyrosine residue could be buried in the cell membrane to restrain the movement of the cytoplasmic domain in the resting state. Such restraints could be released upon the tyrosine phosphorylation. This mechanism may also be applied to the specific tyrosine of α integrin CTMD regions. Structures of α 4 , α M , and α X CT indicate that the tyrosine could be buried in the cell membrane (Fig. 1B). The tyrosine may restrain the piston like movement of α CT or the conformational plasticity of the GFFKR region as suggested by structural studies 59,[61][62][63] . This negative effect depends on the position of tyrosine as shown by our current data. Whether the tyrosine can be phosphorylated to release its negative effect on integrin activation clearly requires further investigation on individual integrins. Another potential mechanism by which the α integrin CTMD regions regulate the levels of integrin activation is through their interacting proteins. Several α integrin CT binding proteins have been identified to function as either negative or positive regulators for integrin activation 8,64 . Some of the regulators such as SHARPIN, MDGI, and filamin interact with a subset of α integrins while the others bind to specific α integrins, such as Nischarin for α 5 and CIB1 for α IIb 54,62,[65][66][67] . Interestingly, most of the current α CT binding proteins interact with the membrane-proximal region containing the conserved GFFKR motif 8 . More novel integrin activation regulators interacting with the α CTMD region are yet to be identified.
There are accumulating data showing that the diverse α integrin CTMD regions can specify the cellular function of integrins. Novel functions of the α CTMD regions have been identified in recent years. For example, a subset of α integrin CTMD regions was found to regulate integrin internalization through interacting with the endocytic clathrin adaptor AP2 18 . Specific interaction between integrin α 5 CTMD region with phosphodiesterase-4D5 (PDE4D5) was found to regulate endothelial inflammatory signaling 68 . It is tempting to speculate that more novel functions of α integrin CTMD regions are yet to be identified. Our large-scale analysis of the function of α integrin CTMD regions provokes new hypotheses that need to be tested on individual integrins. In addition, our approach provides a valuable tool and resource to study the integrin signaling events that are specified by the α CTMD regions.
The integrin α IIb -chimeras were generated by the overlap PCR to replace the cDNA of α IIb CTMD region with the cDNA of the CTMD region of α V , α 1 , or α L integrin. The chimeric full-length α IIb cDNA was cloned into the pEF1/V5-HisA vector using the 5′ EcoRV and 3′ XbaI restriction sites. A stop codon was added right before the 3′ XbaI site. The integrin α L -chimeras were constructed using the 5′ Bsp1407I site right before the cDNA of KVGFFKR motif and the 3′ XbaI site preceded by a stop codon. The extra Bsp1407I and XbaI sites in the WT α L vector were silenced by site-directed mutagenesis. A set of sense and antisense overlapping primers were designed to encode the sequence of KVGFFKR followed by the CTMD region of α D , α M , or α X integrin. The 5′ Bsp1407I and 3′ XbaI sites as well as a stop codon were included in the primers. The chimeric cDNA fragments were obtained by mixing the primers for PCR amplification and subcloned into the α L construct using the Bsp1407I and XbaI sites. The full-length cDNA of human α M integrin was cloned into the pcDNA3.1 vector using the 5′ KpnI and 3′ XbaI sites. Mutations at the α M CTMD region were introduced by PCR. The integrin α 5 -chimeras was constructed using a 5′ HindIII site preceding the GFFKR-coding sequence and a 3′ MluI site preceded by a stop codon. A set of sense and antisense overlapping primers were designed to encode the sequence of GFFKR followed by the CTMD region of α V , α 1 , α 2 , α 3 , α 4 , α 6 , α 7 , α 8 , α 9 , α 10 , or α 11 integrin. The chimeric cDNA fragments were generated by PCR and subcloned into the WT α 5 vector. All the DNA constructs were validated by DNA sequencing.
Human ICAM-1-Fc was expressed as the secreted form in HEK293FT cells via transient transfection using polyethylenimine (PEI). The transfected cells were cultured for 10 days before the culture supernatant was collected. The concentration of ICAM-1-Fc in the supernatant was determined by ELISA using the anti-human ICAM-1 mAb (SinoBiological, Inc.) and the peroxidase-conjugated anti-human IgG1 (Fc specific) (Jackson ImmunoResearch Laboratories, Inc.). The culture supernatant was used for the ICAM-1 binding assay. Human fibronectin type III domains 9 th −10 th fragment (Fn9-10) was expressed in E. coli and purified as described before 77 . Human fibronectin (Fn) was purchased from Sigma. Fn9-10 and Fn were conjugated with Alexa Fluor-647.
Conformation-specific antibody binding. Integrin conformational changes detected by the conformation-specific mAbs on the cell surface was as described previously 13,70 . In brief, HEK293FT transfectants were first incubated with the biotinylated conformation-specific mAb in Ca/Mg at 25 °C for 30 mins, washed and then incubated with the detecting reagents: Alexa-Fluor-647-labeled streptavidin plus R-PE-labeled AP3 for α IIb β 3 or PE-labeled TS2/4 for α L β 2 . For α 5 β 1 integrins, HEK293FT-α 5 β 1 -KO transfectants were incubated with rat 9EG7 mAb in Ca/Mg at 25 °C for 30 mins, washed and then incubated with PE-labeled MAR-4 and Alexa-Fluor-647-labeled goat anti-rat IgG (cross-absorbed) (Abcam). Integrin and GFP double-positive cells were analyzed Scientific REPORTS | (2018) 8:5067 | DOI:10.1038/s41598-018-23444-w for calculating the MFI of mAb binding. The binding of conformation-specific mAb is presented as their MFI normalized to the MFI of integrin expression.
For m24 binding with α M β 2 integrin, HEK293FT transfectants were incubated with m24 or mouse anti-α M mAb (2LPM19c) separately, washed and incubated with Alexa-Fluor-647-labeled anti-mouse IgG. GFP-positive cells were analyzed for calculating the MFI of bound m24 or 2LPM19c. The m24 binding was presented as the m24 MFI normalized to the 2LPM19c MFI reporting α M β 2 expression.

Co-immunoprecipitation.
To detect the TH-binding to integrin β 3 CT, the α IIb β 3 constructs were co-transfected with GFP-TH or GFP into HEK293FT cells. The cells were lysed to perform co-immunoprecipitation with rabbit anti-GFP antibody. The GFP-TH associated β 3 subunit was detected by immunoblot using rabbit anti-β 3 antibody H-96 and HRP-labeled anti-rabbit IgG antibody. The amount of GFP or GFP-TH in pull-down samples was detected by rabbit anti-GFP antibody and HRP-labeled anti-rabbit IgG antibody. Total expression levels of integrin β 3 and GFP-TH in the whole cell lysates were detected by immunoblot. β-actin was blotted with anti-β-actin mAb as a loading control.
To detect the TH binding to integrin β 2 CT, the α L constructs were co-transfected with β 2 -D709A and GFP-TH into HEK293FT cells. GFP-TH only or α L β 2 plus GFP transfectants were used as negative controls. The cells were washed with PBS at pH 8.0 and the cell surface molecules were biotinylated using 2 mM biotin reagent (EZ-Link ™ Sulfo-NHS-Biotin, Thermo Scientific) in PBS at 25 °C for 30 min. The cells were washed and lysed for co-immunoprecipitation with anti-GFP antibody. As a positive control for integrin α L and β 2 bands, α L β 2 was immunoprecipitated with anti-α L mAb TS2/4. Integrin α L and β 2 subunits that associated with GFP-TH were quantitatively detected by blotting with IRDye ® 800CW Streptavidin (LI-COR). The immunoprecipitated GFP or GFP-TH were detected by Western blot using anti-GFP antibody. Integrin α L and β 2 signals were normalized to the precipitated GFP or GFP-TH, and shown as a percentage of the normalized α L -WT signal. The cell surface expression of α L β 2 integrins and GFP-TH were detected by flow cytometry.
Statistical Analysis. Data are expressed as mean ± s.e.m from at least three independent experiments (n ≥ 3 in each group) unless specified. Statistical analyses were performed with GraphPad Prism using parametric Student's t-test (two-tailed).