We reasoned that the multiple MT binding domains within ase1p-oligomers may be oriented such that, due to steric effects, simultaneous binding of all domains to the same MT is prevented, whereas binding to two antiparallel MTs is allowed (Figure 4A). Interactions with overlapped MTs may therefore generate strong multiple bonds that compete ase1p away from single MTs. To find support for this localization mechanism, we expressed a monomeric ase1p mutant in ase1Δ cells. GFP-ΔNase1p (ase1p without the putative spectrin domain) could not recover the unbundled-MT phenotype of ase1Δ cells (Loiodice et al., 2005
) and localized dimly along the full length of MTs (Figure 4B and Movie S10). Localization and MT bundling are thus directly dependent on ase1p oligomerization, which itself may be regulated by phosphorylation like recently demonstrated for PRC1 (Zhu et al., 2006). To investigate whether the dynamics of binding between ase1p oligomers and MTs is consistent with the proposed localization mechanism, we performed fluorescence recovery after photobleaching experiments (FRAP) on GFP-ΔNase1p along single MTs (one MT binding domain per ase1p monomer) and GFP-ase1p along bundled MTs (multiple MT binding domains per ase1p oligomer). GFP-ΔNase1p expressed in ase1Δ cells recovered quickly with a time constant of 0.55 ± 0.05 s (n = 9; ±SEM; Figure 4C and Movie S11). For GFP-ase1p expressed in ase1Δ/mto2Δ cells, we measured a significantly longer time constant of 17.0 ± 1.5 s (n = 5; Figure 4C and Movie S12). The relatively long bundled regions in mto2Δ cells allowed for partial bleaching of ase1p stretches (Janson et al., 2005), which helped to distinguish between GFP-ase1p recovery and the generation of new bundled regions. An intensity-based analysis of the fractional coverage of MTs by GFP-ΔNase1p and GFP-ase1p allowed for an estimation of molecular rate constants from the measured time constants (Figure S2). Approximately one out of six available binding sites along MTs were occupied by GFP-ΔNase1p, implying that the molecular off-rate of GFP-ΔNase1p was significantly higher than 1/0.55 s = 1.8 s−1, whereas the off-rate of GFP-ase1p at regions of overlap was close to 1/17.0 s = 0.059 s−1. Brief interactions involving a single MT binding domain may therefore allow ase1p to quickly “scan” MTs for overlapped regions and subsequently bind tightly.
Cell：“分子发动机和刹车”作用机制的研究
Figure 3F). Furthermore, targeting to single MT regions (n = 44) in bundles was very dim and often undetectable (3.4% ± 1.5%, ±SEM, of average intensity along antiparallel MTs), indicating a large affinity of ase1p for overlapped MTs and in particular for antiparallel overlapping MTs. A similar enrichment at bundled regions was observed with GST-tagged ase1-GFP, indicating that this property is independent of the protein tags used (data not shown). To compare these results with ase1p localization in vivo, we quantified the intensity of ase1-GFP along two parallel or two antiparallel MTs using kymographs of cells that coexpress mRFP-tubulin (Figure 2C). In vivo, the intensity of ase1-GFP along two parallel MTs, as judged from two-color kymographs, was on average 9% ± 1% (±SEM; n = 16) of that along two antiparallel MTs in the same bundle (Figure 3F), and ase1p-GFP targeting to single MTs could not be detected. The larger in vivo orientational preference may yield a strong bias toward the formation of antiparallel MT arrays in living cells. This would explain observations in MT nucleation deficient mto2Δ cells, in which only interactions between antiparallel treadmilling MTs generate bundling (Janson et al., 2005). Part of the observed differences between in vivo and in vitro measurements may be related to differences in ase1p-GFP concentration. In support of this, we note that specific targeting was inhibited in vitro by further increasing the ase1p-GFP concentration, which yielded significant localization of ase1p-GFP to single MTs (up to 40% of targeting to overlapped regions; data not shown). Similarly, overexpression of ase1p-GFP in fission yeast yielded localization to single MTs in vivo (data not shown). Furthermore, we note that specific interactions of ase1p with antiparallel MTs, and possible nonspecific interactions with parallel MTs will depend to varying degrees on ionic strength. Nonspecific interactions are likely screened in the cell, increasing the importance of specific interactions.
