How E. Coli find its middle

1. How E. Coli find its middle Journal Club talk by Xianfeng Song Advisor: Sima Setayeshgar

2. Outline • Introduction to E. coli • Regulation of division site placement by Min proteins • Experiments: In vivo (qualitative) observations of Min proteins dynamics • Modeling: • Quantitative description • How and why Min proteins regulate accurate cell division in E. coli • Open questions

3. About E. coli • E. coli is a bacterium commonly found in the intestinal tracts of most vertebrates. • Studied intensively by geneticists because of its small genome size, normal lack of pathogenicity, and ease of growth in the laboratory. • Size: • 0.5 microns in diameter • 1.5 microns in length • Length of cell cycle: ~ 1 hr From: cwx.prenhall.com

4. FtsZ ring E. coli life cycle FtsZ ring: • polymerizes on membrane at division site, provides framework for assembly of other cell division proteins • constricts like a drawstring during cell division, splitting the cell in two; it disassembles after division • accuracy of placement determines division accuracy

5. Accuracy of cell division in E. coli • Division accuracy: .50 +/- .02 • Placement of FtsZ ring: .50 +/- .01

6. Big Picture • Quantitative understanding of the mechanism of cell division is important, since it is • a fundamental cellular process • of practical importance in development of anti-virulence strategies • Eukaryotic (us) and prokaryotic (bacteria) cell division proceeds very differently; focus on simpler bacterial systems. • How does a single bacterial cell establish a “meterstick”? • Localization of cell division apparatus is an example of the generic problem of spatiotemporal localization of proteins inside bacterial cells to coordinate the cell’s functions.

7. Two systems regulate division site placement • Nucleoid occlusion • Min proteins • Includes MinC, MinD, and MinE

8. Function of Min Proteins (from experimental observations) • MinC • Inhibits FtsZ ring formation • Recruited by MinD:ATP onto membrane • MinD • MinD:ATP stick onto membrane • MinD:ADP tends to go into cytoplasm • MinD:ATP recruits MinC and MinE to membrane • MinE • Recruited by MinD:ATP onto membrane • induces ATP hydrolysis (ATPADP) Black: MinC Red: MinD Blue: MinE

9. Without Min proteins, get minicelling phenotype (Min-) If MinC is over-expressed, get filamentous growth, i.e., no division (Sep-) Min protein phenotypes (from experiments)

10. MinD oscillations: Hale et al.(2001) MinD-GFP

11. MinE ring oscillation caps MinD polar region: • MinE ring is membrane bound. • Ring appears near cell center, moves to one pole, back to center, and on to next pole. Hale et al.(2001) MinE-GFP

12. Filamentous cell has “zebra stripe” pattern of oscillations • Wavelength of oscillations is ~10 microns. Raskin and de Boer(1999) Hale et al.(2001) MinD-GFP MinE-GFP

13. Phenomenology of Min oscillations from in vivo observations • MinD polar regions grow as end caps • MinE ring caps MinD polar region • Filamentous cell has “zebra stripe” pattern of oscillations • Oscillation frequency: • [MinE]   frequency  • [MinD]   frequency  • Oscillations require MinD and MinE but not MinC

14. Summary of modeling efforts • Howard et al. (2001) • Simple 1D model • MinE is recruited by cytoplasmic MinD to membrane • MinD polar region fails to reform at poles (does not agree with experiment) • Meinhardt and de Boer (2001) • Huang and Wingreen (2003) • MinE is recruited by membrane-bound MinD:ATP • MinD aggregation on the membrane follows a one-step process • Kruse et al. (2005) • Consider protein diffusion within the membrane • MinD aggregation on the membrane follows a two-step process: first attachment to membrane, then self-assembly into filament

15. Huang and Wingreen (2003) Model

16. Governing equations (from Huang and Wingreen, 2003)

17. Result: MinD/E movie MinE MinD

18. Simulation results: MinD end caps and MinE ring

19. Mechanism for growth of MinD polar regions (according to Huang and Wingreen, 2003) • MinD:ADP ejected from old end cap diffuses in cytoplasm. • Slow MinD:ADP  MinD:ATP conversion implies uniform reappearance of MinD:ATP in the cytoplasm. • Capture of MinD:ATP by old end cap leads to maximum of cytoplasmic MinD:ATP at opposite pole.

20. Model result I: Frequency of oscillations ~ [MinE]/[MinD] • Relation: • [MinE]   frequency , • [MinD]   frequency . • Minimum oscillation period 25s. • No oscillations for [MinE] too high, or for [MinD] too low. (from Huang and Wingreen) (4 micron cell)

21. Model result II: “Zebra stripe” oscillations in long cells • Stripes form with wavelength of ~10 microns

22. Oscillations allow E. Coli to divide accurately • The oscillations result in a minimum MinD concentration on the membrane at the middle. • MinC dynamics simply follows MinD dynamics. • MinC inhibits FtsZ ring formation.

23. Selection of “intrinsic” length scale by cell: Red curve corresponds to a normal cell Linear stability analysis around homogeneous solution: 1/kmax ~ cell dimension below which there are no oscillations

24. Turing instability:   diffusive, symmetry-breaking instability  characterized by intrinsic wavelength (limit of large system size): "meter-stick“ Dimensional analysis L=(D/σ)^(1/2) [D]=[L^/T], [σ]=[1/T] Selection of “intrinsic” length scale by reaction-diffusion mechanism: Turing pattern formation

25. Recent Developments • “Recent” experiment indicate helical morphology of MinD polymers on membrane • Recent modeling efforts (mainly focus on cleaning some details, no breakthrough) • Effect of fluctuating protein numbers (Howard, et al, 2003), • Inclusion of membrane diffusion and more reactions (Kruse, et al, 2005) • Min-protein oscillations in round bacteria (Huang and Wingreen, 2004) From Hu et al. (2002), Shih et al. (2003)

26. Open questions (from the community) • Role of helical polymerization of MinD on the membrane • MinE ring reverses direction temporarily: stochastic effect?

27. Open questions from us • Where does the precision of division come from? • Why is it 4%, not 10% or 20%?

28. Conclusions • There are now established quantitative models of division site placement in E. coli which • do not rely on prescribed (and potentially unknown) topological protein markers • achieved through spontaneous dynamic instability • E. coli’s oscillating Min system is NOT a universal solution; for example, • B. subtillus: MinCD-DivIVA system, no oscillations, capable of symmetric/asymmetric division • C. crescentus: no Min system, capable of symmetric/asymmetric division Physicists can contribute to the quantitative understanding of fundamental biological processes!

29. Thank you! And thanks to Ned Wingreen (Princeton U.) for sharing some material for this talk with us.

30. Evidence from in vitro studies A. Phospholipid vesicles B. MinD:ATP binds to vesicles and deforms them into tubes • MinD:ATP polymerizes on vesicles • Diffraction pattern indicates well-ordered lattice of MinD:ATP E. MinE induces hydrolysis of MinD:ATP and disassembly of tubes Hu et al. (2002)

31. Min proteins in spherical cells:Neisseria gonorrhoeae Szeto et al. (2001) Wild type MinDNg-

32. Min-protein oscillations in nearly round cells

33. In E. coli,Min oscillations “target” MinD to poles

34. Why does E. coli need an oscillator? In B. subtilis, minicelling is prevented by MinCD homologs, but polar regions are static. Marston et al. (1998)

35. How E. coli find its middle Proteins are too small to see the caps’ curvature Subtilis have local proteins fixed at two ends, but E. coli does not have

36. Without Min proteins, get minicelling phenotype (Min-) If MinC is over-expressed, get filamentous growth, i.e., no division (Sep-) Min protein phenotypes (from experiments)

37. Predictions of model (Huang and Wingreen, 2003) • Delay in MinD:ATP recovery is essential (verified by some experiments). • Rate of hydrolysis of MinD:ATP by MinE sets oscillation frequency. • Diffusion length of MinD before rebinding to membrane sets spatial wavelength.