Assembly of talin-centered supramolecular machinery at the early stage of cell adhesion
Many fundamental cell functions rely on talin-induced integrin binding to extracellular matrix (ECM). Although talin head domain is known to activate integrin, we discovered the irreplaceable role of talin rod in triggering efficient integrin clustering, leading to stronger integrin-ECM interaction.
The attachment of cells to the extracellular matrix (ECM) is a fundamental cellular process crucial for the development and responses of multicellular organisms. Its dysfunction has been linked to many human disorders such as cancer and thrombosis. This process is critically controlled by integrins, a class of heterodimeric cell surface receptors discovered about 30 years ago.
It has been a long journey to investigate the active players that contribute to the integrin activation, which is crucial for strong cell attachment to ECM. Talin, a large cytoskeletal protein, has now been widely accepted as the major activator, using its N terminal head to bind and activate integrin upon release from autoinhibited status. The talin-mediated integrin activation was also reported to be supported by another integrin regulator called Kindlin. However, it remains puzzling why talin head alone is apparently less potent in inducing integrin activation, compared to the cell models with fully activated integrins. On the other hand, kindlin, although also binds to integrin cytoplasmic tail, does not show interaction with talin. Thus, how talin and kindlin act cooperatively has been highly elusive.
Starting from the classic CHO cell model, with the optimized flow cytometry panels, we figured out that when talin rod is not fully deleted, integrin activation can be further elevated compared to talin head only. Based on the previously solved crystal structure of talin autoinhibition part (F3-R9 interaction), we simultaneously mutated three critical interface residues to generate a constitutively activated full length talin (also called talin-M3, the activation status of this mutant was validated through various methods such as gel filtration and integrin pull down). To our surprise, this talin-M3 has much higher potency than talin head alone in activating integrins!
We then wondered, what does the high potency mean? And what leads to this high potency? Firstly, through monomeric vs multivalent ligand testing, our data indicated that talin-M3 significantly induced clustering of integrins to strongly bind to multivalent ligands. To note, this integrin clustering is considered as microclusters that benefit multivalent ligand binding, a step before the post-ligand-binding-induced macroclustering. Secondly, the talin C terminal DD domain contributes to the microclustering significantly, indicating that talin dimer plays an important role in promoting integrin clustering. Furthermore, we found a kindlin-binding protein, paxillin, also binds to talin mainly through the rod part. Interestingly, despite the contribution from talin DD, linkage of only four extra rod subdomains that exhibit strong binding to paxillin could greatly enhance the potency of talin head to induce integrin-ligand interaction. Moreover, with coexpression of talin and paxillin, no matter if talin is activated, we saw enhanced integrin clustering! This evidence prompted us to consider paxillin as an important bridge between talin rod and kindlin.
It has been a great challenge to study the talin-paxillin-kindlin machinery because of the dynamic assembly of the complex. But through detailed biochemical and cell-based analysis, we utilized both CoIP method and NMR strategy to show the existence of the complex and how our point mutations can efficiently disrupt the complex. We found that paxillin seems to bind multiple talin subdomains via its flexible N-terminal LD containing region. Through extensive mapping studies, we determined talin rod R1-R4 region as the core to bind the paxillin N-terminus. It is a bit hard to imagine how paxillin uses its flexible N terminus to manipulate talin function. However, using NMR approach, we determined the critical role of two helices in the N-terminus, LD1 and LD2, that cooperate to bind to talin R1-R4. Very excitingly, the AlphaFold prediction aligned well with our assumption!
Thanks to the great contribution from our collaborator Dr. Moser and Dr. Bromberger, our story is strengthened with well-rounded functional data. With their knock-in mouse fibroblast model, we can see at endogenous levels how full length talin and its partners (paxillin and kindlin) play essential roles in cell attachment and affect the downstream focal adhesion assembly. Very importantly, small point mutation on talinR2, which hasn't been reported for any other important binding, led to significant cell adhesion and spreading defects, validating the importance of paxillin in talin-mediated integrin activation and adhesion.
This manuscript is completed, but the exploration is still going on. Although our SPR data and some of our spin down results indicate a 1:1 binding between talin and paxillin, it remains unclear how exactly this 1:1 binding occurs at structural level. Additionally, although we verified the importance of talin dimerization, which is also well supported by literature, how full length talin forms dimer structure is still unknown. Whether kindlin utilizes additional pathways to cooperate with talin also remains to be further examined. Nevertheless, while hundreds of proteins are recruited to focal adhesions, we determined talin-paxillin-kindlin as critical machinery to trigger focal adhesion assembly but how this assembly process occurs is unclear. All these questions remain to be answered in the future. Given the critical role of integrin activation in various physiological and pathological conditions, we believe understanding of the talin-centered supramolecular machinery will provide valuable guidance in the development of future diagnostics and therapeutics for integrin-dependent diseases.