Description

Description

Bin/Amphiphysin/Rvs (BAR) domain proteins have emerged as key regulators of membrane curvature. Interestingly many BAR proteins are also steering actin dynamics during many cellular processes (Carman, 2018). The BAR protein bridging integrator 1 (BIN1) has been initially identified as part of the endocytosis and membrane trafficking machinery (Ren, 2006 ). Recently, it has been shown that BIN1 stabilizes and bundles actin filaments. In the same study, the authors reported that BIN1 influences actin dynamics. Mass spectrometry identified interaction interfaces between BIN1 and actin (Drager, 2017). We were interested to visualize the molecular details of the interaction between human BIN1 (isoform 8) and actin filaments. Therefore we collected cryo electron microscopy (cryoEM) data of actin filaments that were assembled in vitro and in the presence of BIN1. Curiously, we could not detect any BIN1 on single actin filaments.

We qualitatively assessed the binding between BIN1 and actin filaments through a co-sedimentation assay. BIN1 was added to G-actin at 2:1 molar ratio and left to polymerize for 30 minutes at room temperature. The reaction was then subjected to ultracentrifugation at 100,000 g for 1.5 h. A wash step was introduced to remove excess/unbound BIN1. The collected supernatant, wash and pellet fractions were analyzed by SDS PAGE (Figure 1A). Our data reaffirm that BIN1 was found in the pellet fraction together with actin, which consistent with a previous report (Drager, 2017). We also tested the presence of BIN1 on bundles with differential centrifugation of the sample; first at 10,000g for 10 min (Figure 1 A lane L) then at 100,000g as described above (Figure 1 A lane H). This experiment confirms the presence of BIN1 on bundles. To ensure that BIN1 is functional, we tested its ability to form tubes in vitro (Gowrisankaran, 2020). We confirmed negatively stained lipid nanotubes in electron micrographs (Figure 1B). Next we evaluated the in vitro formation of actin filaments by negative stain EM and cryoEM (Figure 1C and Figure 1D). The apparent diameter of ~7nm of the observed filaments is within the range of single actin filaments. The images of actin filaments in the presence of BIN1 did not differ much in negative stain or in cryoEM (Figure 1D). Yet, based on the sedimentation assay we concluded that there is an actin-BIN1 complex. For structure analysis, we collected a data set of a BIN1-actin complex without removing BIN1 from the reaction to ensure maximum coverage of actin filaments. We processed the movies and images with the CryoSparc software suite. After filament tracing and particle extraction, we performed unsupervised 2D classification (Figure 1E). Based on the X-ray structures of BIN1 and other N-BAR domains we know that BIN1’s BAR domain spans ~150 Å (Casal, 2006). Therefore we chose an initial box size of 294 Å for particle extraction. Yet, no density other than actin was observed in the 2D classes (Figure 1 E+F). If BIN1 is not bound in a 1:1 stoichiometry, we need to investigate longer stretches of the actin filament to cover larger segments. So, we increased the particle box size (and the binning). However, neither of the 2D classes showed any evidence for the presence of BIN1 (Figure 1F). Last we reconstructed actin filaments from the best 336Å box size 2D classes (180008 particles) to 3.25 Å. This resolution allowed us to unambiguously fit a model of actin. We could identify the acidic residues that were proposed to partake in the BIN1 interaction (Figure 1G, magenta). We could not identify any excess density at these residues that may indicate the presence of an interaction partner. Even, if BIN1 is flexible, we would have expected that the contact interface between BIN1 and actin would have been somewhat inflexible.

Taken together, our data confirms that BIN1 binds to actin. However, this study clarifies some of previous reports. Namely, BIN1 does not bind to single actin filaments. Instead we speculate that BIN1 can be found on actin bundles with various single filament numbers, which our differential centrifugation experiments indicate. This is in line with the cited study, where it was reported that BIN1 bundles actin filaments (Drager, 2017). Interestingly, a recent report shows that another N-BAR protein (ASAP1) can create polar bundles with a low filament number (Chen, 2020). Other BAR domain proteins (PACSIN2) are found on single filaments (Kostan, 2014). Therefore, it remains an open question if BAR domains can differentiate between single actin filaments and actin bundles.