lobacter
Cell, Vol. 118, 579590, September 3, 2004, Copyright
2004 by Cell Press
Cytokinesis Monitoring during Development: Rapid
Pole-to-Pole Shuttling of a Signaling Protein by
Localized Kinase and Phosphatase in Caulobacter
the ordered sequence of flagellar rotation, pili synthesis,
flagellum shedding, and ultimately stalk formation (Fig-
ure 1). It has been shown that the PleC and DivJ histidine
kinases couple SwaPS development with cell division
through the regulation of a shared single-domain re-
Jean-Yves Matroule, Hubert Lam,
Dylan T. Burnette, and Christine Jacobs-Wagner*
Department of Molecular, Cellular,
and Developmental Biology
Yale University
P.O. Box 208103
sponse regulator, DivK (Ohta et al., 2000). Mutations in
New Haven, Connecticut 06520
pleC block SwaPS development; thus, pleC mutants
remain nonmotile, pililess, and stalkless after cell divi-
sion (Sommer and Newton, 1989). Cold-sensitive mu-
Summary
tants of divJ and divK were isolated as developmental
suppressors of pleC mutants at 37 C and found to be
For successful generation of different cell types by
filamentous at 24 C, indicating that DivJ and DivK are
asymmetric cell division, cell differentiation should be
also involved in cell division (Sommer and Newton,
initiated only after completion of division. Here, we
1991). The level of phosphorylated DivK (DivK-P) is re-
describe a control mechanism by which Caulobacter
duced in a
divJ mutant, consistent with DivJ acting
couples the initiation of a developmental program to
as a kinase of DivK (Wheeler and Shapiro, 1999). The
the completion of cytokinesis. Genetic evidence indi-
biochemical function of PleC is less clear. In pleC mu-
cates that localization of the signaling protein DivK at
tants, the DivK-P level is increased relative to wild-type,
the flagellated pole prevents premature initiation of
suggesting that PleC directly or indirectly promotes
development. Photobleaching and FRET experiments
DivK-P dephosphorylation (Wheeler and Shapiro, 1999).
show that polar localization of DivK is dynamic with
DivJ, PleC, and DivK undergo spatial regulation. In
rapid pole-to-pole shuttling of diffusible DivK gener-
predivisional cells, DivJ and PleC are asymmetrically
ated by the localized activities of PleC phosphatase
localized with DivJ at the stalked pole and PleC at the
and DivJ kinase at opposite poles. This shuttling is
flagellar pole (Wheeler and Shapiro, 1999). As shown in
interrupted upon completion of cytokinesis by the seg-
Figure 1, DivK is bipolarly localized during most of the
regation of PleC and DivJ to different daughter cells,
predivisional stage but is released specifically from the
resulting in disruption of DivK localization at the flagel-
flagellar pole right after completion of cytokinesis, when
lated pole and subsequent initiation of development
the cytoplasm is divided into two physically separated
in the flagellated progeny. Thus, dynamic polar local-
compartments (Jacobs et al., 2001). DivK localization is
ization of a diffusible protein provides a control mech-
regulated by DivJ and PleC. In a kinase-inactive divJ
H338A
anism that monitors cytokinesis to regulate devel-
mutant, DivK localizes normally at the stalked pole,
opment.
where DivJ
H338A
is located, but it fails to localize at the
flagellar pole, indicating that the kinase activity of DivJ
Introduction
promotes localization of DivK at the flagellar pole (Lam
et al., 2003). Conversely, PleC mediates the release of
Asymmetric cell division is an important mechanism for
DivK from the flagellar pole because, in the absence of
the generation of cell diversity from bacteria to mam-
PleC catalytic function, DivK fails to delocalize from the
mals. For correct cell fate specification by this mecha-
flagellar pole at cell division (Lam et al., 2003). The role
nism, initiation of developmental programs relies on the
for the cell cycle-dependent polar localization of these
segregation of cell fate determinants to only one daugh-
signaling proteins has remained largely elusive, and the
ter cell upon cell division. The temporal coupling be-
mechanism by which these proteins sense cytokinesis
tween cell division and initiation of differentiation sug-
to control SwaPS development is unknown.
gests that control mechanisms may exist.
Here, we provide evidence that localized PleC phos-
The bacterium Caulobacter crescentus provides a
phatase and DivJ kinase activities at opposite poles
simple experimental system to address this question be-
create a rapid shuttling of DivK between poles. This pole-
cause, in this organism, asymmetric cell division yields a
to-pole shuttling provides a control mechanism that moni-
daughter cell whose developmental fate is controlled
tors cytokinesis to regulate initiation of SwaPS devel-
by cell division (Ohta and Newton, 1996). At each cell
opment.
cycle, Caulobacter divides asymmetrically to produce
a stalked cell and a smaller swarmer cell, which subse-
quently differentiates into a stalked cell before initiating
Results and Discussion
DNA replication (Figure 1). Genetic studies with muta-
tions or drugs that inhibit cell division indicate that the
developmental program of the swarmer progeny depends
DivK Localization at the Flagellar Pole
on the completion of previous cytokinesis (Huguenel
Is Involved in Coupling SwaPS
and Newton, 1982; Terrana and Newton, 1976). This
Development to Cytokinesis
swarmer progeny-specific (SwaPS) program includes
As mentioned above, pleC mutants fail to release DivK
from the flagellar pole (Figure 2A) and to initiate SwaPS
development upon cell division, resulting in nonmotile,
*Correspondence: christine.jacobs-wagner@yale.edu Cell
580
site closer to the stalked pole, to create stalked daughter
cells smaller than their siblings (Figure 2C). Thus, to
ascertain the identity of the poles, we used a strain
(CJW979) in which DivK and DivJ (stalked pole marker)
were fused to the monomeric forms of CFP and YFP
(mCFP and mYFP), respectively. Restoration of cytoki-
nesis by addition of xylose to elongated CJW979 cells
resulted in delocalization of DivK-mCFP specifically
from the pole distal to the stalked DivJ-mYFP pole; this
was true even in the instances when cell division created
a stalked progeny smaller than its sibling (Figure 2C),
suggesting that DivK release is not sensitive to cell
length. We made similar observations with cells treated
with the -lactam antibiotic cephalexin, which at suble-
thal concentration inhibits a later step of cell division
(data not shown). Thus, DivK release from the flagellar
pole is cytokinesis dependent.
If a control mechanism involving DivK localization at
Figure 1. Completion of Cytokinesis Is Required for Initiation of a
the flagellar pole specifies the known dependence of
Developmental Sequence of Events in the Ensuing Cell Cycle of the
SwaPS development on cytokinesis, this dependence
Swarmer Progeny
should be relieved by the D90G mutation in DivK, which
The swarmer progeny-specific (SwaPS) developmental program (blue
prevents DivK from localizing to the flagellar pole. A relief
arrows) includes flagellar rotation, pili synthesis, flagellum shedding,
of this dependence would permit SwaPS development
and ultimately stalk formation. The cell cycle-dependent localization
to occur even when cytokinesis is inhibited, ultimately
of DivK (red) culminates in the complete release of DivK from the
leading to the formation of a stalk at the pole where the
flagellar pole upon completion of cytokinesis (division of the cyto-
plasm into two). SW, swarmer cell; ST, stalked cell.
flagellum was previously located. To test this hypothe-
sis, cytokinesis was inhibited in synchronized wild-type
divK (YB1585) and divK
D90G
(CJW1169) cells by FtsZ
depletion. As previously described (Osley and Newton,
pililess, and stalkless cells with paralyzed, unshed fla-
1977), inhibition of cytokinesis blocked SwaPS develop-
gella. This suggested that a maintained localization of
ment in the wild-type divK background, yielding fila-
DivK at the flagellar pole after cytokinesis inhibited
ments with a stalk at one pole and a flagellum at the other
SwaPS development. To test this hypothesis, we exam-
(Figure 2D). In contrast, divK
D90G
cells initiated SwaPS
ined the localization of a DivK mutant with a D90G muta-
development regardless of cytokinesis inhibition, yield-
tion that suppresses all developmental defects of a
ing filamentous cells with a stalk at each pole (Figure
pleC::Tn5 mutant at 37 C (Sommer and Newton, 1991).
2D). This relief of dependence between cytokinesis and
DivK
D90G
fused to GFP was able to localize at the stalked
SwaPS development argues for the existence of a con-
pole of pleC::Tn5 cells at 37 C but failed to significantly
trol mechanism.
accumulate at the flagellar pole during the cell cycle
Altogether, our data strongly suggest that the cytoki-
(Figure 2A). A similar localization pattern was observed
nesis-dependent release of DivK from the flagellar pole
in a pleC background (data not shown). The D90G muta-
is the molecular event that signals entry into SwaPS
tion does not affect DivK phosphorylation (Hung and
development in response to completion of cytokinesis.
Shapiro, 2002). Therefore, it is the inability of DivK
D90G
to localize to the flagellar pole at 37 C that accounts for
the suppression of pleC developmental defects. Thus,
The PleC Histidine Kinase Primarily Acts
as a Phosphatase of DivK-P In Vivo
localization of DivK at the flagellar pole is involved in
the inhibition of SwaPS development.
To unravel the mechanism by which the cell senses
the completion of cytokinesis to release DivK from the
Next, we showed that DivK release from the flagellar
pole was dependent on the completion of cytokinesis.
flagellar pole, it w