Skip to main content
Back to timeline
bioRxivSource publication:

A designed binder jams FtsA's longitudinal interface: late divisome proteins stop arriving, yet E. coli still pinches slowly

Synopsis

How FtsA recruits late divisome proteins at the Z-ring in E. coli has lacked a handle for intervention. Using BindCraft, researchers designed L125, a 125-residue protein binder predicted to block the longitudinal contact between successive FtsA subunits; when expressed, cells stop dividing, late proteins FtsK, FtsB, FtsI and FtsN no longer localize to midcell while early components remain. Unexpectedly, these cells still form broad, shallow constrictions that never complete, and this constriction requires ZipA but not FtsI activity.

Source-provided article image: A de novo FtsA binder blocks late divisome recruitment, unmasking a hidden constriction mechanism
Fig. 1 · bioRxiv · Page 28

Interpretation

A de novo designed protein binder can block bacterial division inside cells and directly tie FtsA's longitudinal oligomerization interface to recruitment of late divisome proteins. Prior work pulled in two directions on FtsA oligomerization and divisome activation: reduced FtsA self-interaction was proposed to promote late-divisome recruitment, while recent studies associated the active divisome with antiparallel FtsA filaments, and no tool could perturb that interface at a defined site in living cells. L125 localizes to midcell with a cell-cycle pattern resembling FtsA-GFP, though its peak midcell excess fraction is lower, 34% versus 53%; after L125 expression FtsK, FtsB, FtsI and FtsN are not recruited, whereas FtsX, ZipA and ZapA remain at midcell.

L125 halts division at concentrations below the FtsA stoichiometry, indicating interference with a higher-order FtsA assembly rather than simple occupancy of a single binding site. With L125-mNG fluorescence calibrated per cell, the division rate stays nearly constant for [L125-mNG]/[FtsA] below 0.15, then falls approximately exponentially with a characteristic concentration scale of 0.25, and no divisions are seen above a ratio of 0.8. This concentration dependence was measured by time-lapse imaging in mother-machine microfluidic devices, with 2127 cells analyzed in the L125-mNG strain; the authors note that a threshold-like response is not readily explained by a one-site competition model between L125 and FtsK.

Cells that have lost late divisome localization still form broad, shallow constrictions, exposing a preseptal cell-wall synthesis pathway not previously known to produce visible constrictions. When core divisome function was inhibited with an FtsI/PBP3 inhibitor, excess peptidoglycan synthesis near midcell through PBP1A/PBP1B and ZipA/FtsN was already known, but those studies did not report visible constrictions. Constrictions appear only about 50–75 minutes after induction, the midcell Ypet-FtsN signal decays exponentially with a characteristic time of 37 minutes, and the phase signal rises only after most FtsN has left midcell; no midcell constrictions are detectable in ΔzipA ftsA* cells, while constrictions persist under 0.3 µg/ml aztreonam.

Perspective

The conclusions rest on time-lapse imaging in E. coli mother-machine microfluidic devices and static imaging of fixed cells, with IPTG-induced expression of L125 or L125-mNG and growth at 28°C in M9 glucose-casamino acid medium; genetic tests of the constriction phenotype are limited to ftsA*(R286W) and ΔzipA ftsA* backgrounds, and the drug test uses 0.3 µg/ml aztreonam. The authors propose that future structure-prediction and design models may make such interfaces more tractable.

The authors explicitly leave two interpretations of L125 open: blocking FtsA filament assembly and thus antiparallel filament formation, or sterically hindering binding of a late divisome protein to FtsA, and they note the latter sits poorly with the observed threshold-like concentration dependence. The exact function of preseptal peptidoglycan synthesis remains to be elucidated, and why broad constrictions escaped earlier detection is explained only by their slow progression and by being hard to distinguish in static images. The authors also note that AlphaFold2, used during BindCraft optimization, was trained before the antiparallel FtsA filament structure became available, and that none of the three hotspot-targeted designs selected for testing showed activity in cells.