JOURNAL OF BACTERIOLOGY,

0021-9193/99/$04.000
Feb. 1999, p. 981990 Vol. 181, No. 3
Copyright 1999, American Society for Microbiology. All Rights Reserved.
Cell Density-Dependent Starvation Survival of Rhizobium
leguminosarum bv. phaseoli: Identication of the Role
of an N-Acyl Homoserine Lactone in Adaptation
to Stationary-Phase Survival
STEPHEN H. THORNE AND HUW D. WILLIAMS*
Department of Biology, Imperial College of Science, Technology and Medicine, London SW7 2AZ, United Kingdom
Received 23 March 1998/Accepted 11 November 1998
The cell density dependence of stationary-phase survival of Rhizobium leguminosarum has been investigated.
Following starvation by exhaustion of carbon or nitrogen, but not of phosphorus, the survival of cultures was
dependent on the cell density at entry into stationary phase. High-density cultures survived with little or no loss
of viability over a 20-day period in stationary phase. In contrast, low-density cultures lost viability rapidly but
consisted of a heterogeneous population, a small fraction of which successfully adapted and eventually formed
a stable, surviving population. The threshold density above which the cultures survived successfully in
stationary phase was dependent on the growth conditions and the strain used. We took advantage of the fact
that R. leguminosarum survives poorly following starvation by resuspension in carbon-free medium to demon-
strate that cell density-dependent survival was mediated by a component accumulating in the growth medium.
The effects of this medium component on survival in resuspension assays could be mimicked by an N-acyl
homoserine lactone, N-(3R-hydroxy-7-cis-tetradecanoyl)-L-homoserine lactone, previously demonstrated to
have a role in controlling cell density-dependent phenomena in R. leguminosarum. The Sym plasmids pRP2JI
and pRL1JI were found to be essential for the production of the extracellular factor, which could also be made
in Escherichia coli carrying the cosmid clone pIJ1086 containing a specic region of pRL1JI.
Nutrient starvation is a condition in which bacteria in many
environments, including soil, regularly nd themselves (9, 20,
31, 37). However, little is known about the physiological pro-
cesses involved in the starvation survival of soil bacteria. We
are investigating the starvation survival response of Rhizobium
leguminosarum. We have previously reported on the changes in
the physiology and metabolism of R. leguminosarum bv.
phaseoli in response to nutrient starvation (36). In contrast to
the starvation survival kinetics of enteric bacteria, no reduction
in the viability of R. leguminosarum cultures was observed
during more than 2 months in carbon-starved stationary phase
(36). A further distinguishing feature is the poor survival of R.
leguminosarum following nutrient starvation by resuspension in
minimal medium without carbon, indicating that this bacte-
rium cannot respond to such a rapid deprivation of nutrients
(36).
It has been known for over 30 years that a population effect
is observed when a culture of bacteria is starved (16, 28, 29). In
those early studies the specic death rates of high-density bac-
terial populations were lower than those of low-density popu-
lations. It is also well established that Myxococcus xanthus and
Bacillus subtilis show developmental responses to nutrient de-
privation, i.e., sporulation and fruiting body formation, respec-
tively, that are dependent on population cell density (7, 13, 18,
19). The effect of cell density on survival has not been thor-
oughly addressed in recent studies of the starvation survival of
bacteria (9, 11, 21, 26, 27).
In this study we demonstrate the cell density-dependent
stationary-phase survival of R. leguminosarum following star-
vation by nutrient exhaustion of carbon or nitrogen. We show
that this response is controlled by the accumulation of an
extracellular factor(s) in the growth medium and that the effect
is mimicked by the N-acyl homoserine lactone (AHL) N-(3R-
hydroxy-7-cis-tetradecanoyl)-L-homoserine lactone (7cisHtDHL).
MATERIALS AND METHODS
Strains and plasmids. R. leguminosarum bv. phaseoli 4292 is a rifampin-
resistant strain that carries the pRP2JI Sym plasmid and nodulates beans (23).
This strain was used in our earlier study of starvation survival of R. leguminosa-
rum (36). R. leguminosarum 8401 is a derivative of a biovar phaseoli isolate
(strain 8002), and it is streptomycin resistant and has been cured of the Sym
plasmid pRP2JI. R. leguminosarum 8401/pRL1JI carries the biovar viciae Sym
plasmid. The four pLAFR1-derived cosmids pIJ1527, pIJ1529, pIJ1085, and
pIJ1086 contain DNA cloned from the R. leguminosarum Sym plasmid pRL1JI.
pIJ1085 and pIJ1086 are two overlapping cosmids carrying about 60 kb of DNA
from the nif and nod gene-carrying regions of pRL1JI (8). pIJ1527 and pIJ1529
are poorly characterized, overlapping cosmids carrying about 45 kb of DNA from
a different region of pRL1JI, which does not overlap with pIJ1085 and pIJ1086
(7a). Together the four cosmids carry DNA covering about half of the Sym
plasmid pRL1JI.
Growth and starvation of bacteria. Strains were grown and starved for carbon
(mannitol), nitrogen (NH
4
Cl), and phosphorus (K
2
HPO
4
and KH
2
PO
4
) in
MOPS (morpholinepropanesulfonic acid) minimal medium as described previ-
ously (36). For routine culture maintenance, growth of starter cultures, and
plating for viable-cell counting, rifampin was added at 25 g ml
1
. YEM medium
was also used at pH 7.5 (38) and consisted of (per liter) yeast extract (0.4 g)
(Oxoid), K
2
HPO
4
(0.5 g), NaCl (0.1 g), MgSO
4
7H
2
O (0.2 g), and mannitol (10
g).
Starter cultures were prepared by inoculating a single colony from a plate
stock into 5 ml of medium and incubating overnight at 30C in a shaking
incubator (200 rpm). A 0.5-ml portion of this exponentially growing culture was
then subcultured into a second 5 ml of medium and again incubated overnight to
ensure that cells were fully adapted to exponential growth. About 5 ml of this
starter culture was then inoculated into 50 ml of medium in a 250-ml conical ask
to give an initial optical density at 600 nm of 0.02. This method of inoculation was
* Corresponding author. Mailing address: Department of Biology,
Imperial College of Science, Technology and Medicine, Sir Alexander
Fleming Building, Imperial College Rd., London SW7 2AZ, United
Kingdom. Phone: 44(171) 594 5383. Fax: 44(171) 584 2056. E-mail:
h.d.williams@ic.ac.uk.
Present address: School of Biological Sciences, University of Sur-
rey, Guildford, Surrey GU2 5XH, United Kingdom.
981
982 THORNE AND WILLIAMS J. BACTERIOL.
used throughout this study unless otherwise stated. Growth was measured spec-
trophotometrically as optical density at 600 nm (with 1-cm-path-length cuvettes
in a Shimadzu MPS2000 spectrophotometer) and by viable-cell counting, follow-
ing plating onto solid minimal medium after appropriate dilution in minimal
medium. It was found that although all of the strains had similar growth kinetics
in the liquid MOPS minimal medium (36), 8401/pRL1JI and 8401 did not grow
well on MOPS minimal medium agar. Therefore, the viability of these strains was
determined by diluting the samples in TY medium and plating on TY agar. TY
medium consisted of (per liter) Bacto tryptone (5 g), yeast extract (3 g), and
CaCl
2
6H
2
O (1.3 g). Cells were osmotically stressed by adding NaCl to a nal
concentration of 2.5 M as described previously (36). Escherichia coli 803 strains
were also grown in MOPS minimal medium at 30C.
Determination of affects of spent culture medium. (i) Preparation of aseptic
samples of spent culture medium. To obtain spent medium, cells were removed
from a carbon-starved culture by centrifugation at 4,000 g for 45 min. The
supernatant fraction was then collected and sterilized by ltration through 0.2-
m-pore-size cellulose acetate lters. The aseptic spent medium was then used
to resuspend cells collected by centrifugation (see below). Spent medium was
usually prepared from 6-day-old, high-density (1.5 10
9
CFU ml
1
), carbon-
starved stationary-phase cultures and is referred to as high-density spent medium
(HDSM). For some experiments the spent medium was pretreated before being
added to cells. Heat treatment involved 5.0-ml samples of spent medium being
placed in tubes in a water bath at 100C for 15 min. For protease treatment, spent
medium was treated with pronase attached to agarose beads (1 U ml
1
; Sigma
P4531) for 60 min at 30C. The beads were then removed by ltering the sample
through a 25-mm glass ber lter (pore size, 1 m). To approximately determine
the size of the spent-medium component, samples of stationary-phase spent
medium from carbon-starved R. leguminosarum cultures were ultraltrated in a
stirred cell by using an Amicon YM3 ultraow membrane with a molecular mass
cutoff of 3 kDa.
(ii) Collection and resuspension of cells. Fifty-milliliter cultures of R. legu-
minosarum were harvested by centrifugation at 4,000 g for 20 min. The
supernatant fraction was then removed, and the cells were washed twice by
resuspending them in 50 ml of fresh MOPS minimal medium with no carbon
source (MM-C). This was done in order to make sure that any residual carbon
source or molecules secreted into the medium by the cells were completely
washed from the cells. The bacteria were then pelleted by centrifugation one
nal time, and the pellet was resuspended by vortexing in 5.0 ml of MM-C. A
0.5-ml portion of the resuspended cells was then added to 5.0 ml of MM-C or
spent culture medium. Cultures were then kept at 30C in a shaking incubator
(200 rpm), and survival was monitored for several days by viable-cell counting.
Use of R. leguminosarum autoinducer. A synthetic preparation of the R. legu-
minosarum autoinducer 7cisHtDHL, containing a mixture of L and D stereoiso-
mers that differ in the orientation of the 3-hydroxy group, was kindly provided by
Kendall Gray (12, 34). For use in bioassays, a sample of the autoinducer (in ethyl
acetate) was added to an empty culture vessel and dried down in a stream of
sterile air. The assay culture was then added to the vessel to resuspend the
autoinducer.
RESULTS
Effects of culture cell density on stationary-phase survival of
R. leguminosarum. By varying the mannitol concentration in
MOPS minimal medium, R. leguminosarum bv. phaseoli cul-
tures were grown and allowed to enter stationary phase at cell
densities ranging from1.3 10
9
to 1 10
8
CFU ml
1
. It was
found that the survival of cultures in the 10 to 14 days following
entry into carbon-starved stationary phase was dependent on
the cell density at the time of entry (Fig. 1A). Cultures entering
stationary phase at a density of 1.3 10
9
CFU ml
1
had
108% viability after 14 days of stationary phase, whereas cul-
tures that had entered stationary phase at 1 10
8
CFU ml
1
retained only 15% viability after the same time period. Cul-
tures with intermediate cell densities at entry into stationary
phase showed intermediate survival (Fig. 1A). This experiment
was repeated with YEM medium, in which mannitol was the
major but not the sole carbon source. Changing the concen-
tration of mannitol in YEM allowed the cell density at which
cultures entered stationary phase to be altered. It was found
that stationary-phase survival in YEM was dependent on the
culture density (Fig. 1B). The survival of nitrogen-starved sta-
tionary-phase cultures was also dependent on the culture
cell density, with a higher cell density at the entry into
stationary phase resulting in better survival (Fig. 1C). How-
ever, cell density-dependent survival was not seen in phos-
phorus-starved cultures; phosphorus-starved cultures enter-
ing stationary phase at densities of between 7 10
7
and 8
10
5
CFU ml
1
all survived well during the rst 14 days of
stationary phase (Fig. 1D).
A heterogeneous population enters stationary phase in low-
density cultures. An experiment similar to that of Fig. 1A was
set up, but stationary-phase survival was monitored for a much
longer time period. As expected, the high-density cultures sur-
vived long-term starvation with little loss of viability (Fig. 2)
(36). In the low-density culture, there was an initial loss of
viability over the rst 16 days in stationary phase (up to 20 days
after inoculation), and then the proportion of viable cells sta-
bilized at about 20% of the starting population. There was no
further loss of viability for 15 days (Fig. 2). However, this was
followed by a further reduction of viability from day 35 on-
wards that continued until the end of the experiment at day 60.
The survival kinetics of low-density cultures suggests that a
heterogeneous population enters stationary phase or is present
immediately after entry, comprising a fraction (20%) that
goes on to survive long-term stationary phase and those cells
that will die during the rst 15 days (Fig. 2). We investigated
whether the population was also heterogeneous with respect to
another property of stationary-phase-adapted cells, resistance
to osmotic stress (36). The osmotic stress resistance of high-
and low-density cultures at 5 and 25 days into stationary phase
was determined (Fig. 3). The high-cell-density, 5-day station-
ary-phase culture was resistant to challenge by 2.5 M NaCl,
while the equivalent low-density culture was clearly sensitive
(Fig. 3A). At 150 min after the exposure to osmotic stress, the
viability of the low-density culture had been reduced by 70%,
but the surviving 30% of the population was resistant to 2.5 M
NaCl and there was no further reduction in viability during the
remaining 250 min of the experiment (Fig. 3A). These data are
consistent with the idea that there is a fraction of the popula-
tion that has failed to successfully adapt to stationary-phase
survival and consequently retains physiological traits of expo-
nentially growing cultures. In support of this idea, the decimal
reduction times (D values) (1) for osmotically stressed, expo-
nential-phase cultures (calculated from the data in Fig. 7 of
reference 36) and for the osmotically sensitive fraction of the
population in low-density cultures (Fig. 3A) were similar, being
approximately 150 and 200 min, respectively. Twenty-ve-
day-old high- and low-density stationary-phase cultures were
equally resistant to osmotic stress (Fig. 3B), conrming that 25
days into stationary phase, the surviving bacteria in the low-
density culture are fully adapted for stationary-phase survival.
It seems likely, but is not proven, that this surviving population
FIG. 1. Effect of culture cell density on the stationary-phase survival of nutrient-starved R. leguminosarum bv. phaseoli 4292 cultures. Cultures were grown in
minimal medium (A, C, and D) or YEM (B) with various limiting amounts of carbon source (A and B), nitrogen (C), or phosphorus (D) so that the cell density upon
nutrient-limited entry into stationary phase was varied. (A) In carbon-limited minimal medium, the mannitol concentrations in grams liter
1
(cell densities entering
stationary phase in CFU milliliter
1
) were 10 (I) (1.3 10
9
), 5 (F) (5 10
8
), 2 () (1.5 10
8
), and 0.3 (}) (1 10
8
). (B) In YEM, the mannitol concentrations
in grams liter
1
(cell densities entering stationary phase) were 10 (I) (9.8 10
9
), 5 (F) (4.7 10
9
), 2 () (1.9 10
9
), and 0.3 (}) (1.1 10
9
). (C) In nitrogen-limited
minimal medium, the NH
4
Cl concentrations in grams liter
1
(cell densities entering stationary phase) were 0.25 (I) (9.5 10
8
), 0.1 (F) (3.1 10
8
), 0.05 () (1.7
10
8
), and 0.025 (}) (1.1 10
8
). (D) In phosphorus-limited minimal medium, the phosphate concentrations in grams liter
1
(cell densities entering stationary phase)
were 0.05 (I) (7 10
7
), 0.025 (F) (8.5 10
6
), 0.01 () (3 10
6
), and 0.005 (}) (9 10
5
). Error bars represent standard deviations.
VOL. 181, 1999 STATIONARY-PHASE SURVIVAL OF R. LEGUMINOSARUM 983
FIG. 2. Growth and stationary-phase survival of R. leguminosarum bv. phaseoli 4292 cultures entering stationary phase at high and low cell densities. Cultures were
grown in minimal medium with limiting amounts of carbon source. I, growth with 10 g of mannitol liter
1
, entering stationary phase at 1 10
9
CFU ml
1
, F, growth
with 0.3 g of mannitol liter
1
, entering stationary phase at 2.6 10
8
CFU ml
1
. Growth was monitored by viable-cell counting. Arrows 1 and 2 correspond to the times
that samples were taken for osmotic stress tests (see Fig. 3). Error bars represent standard deviations.
FIG. 3. Osmotic stress resistance of high- and low-density stationary-phase cultures of R. leguminosarum bv. phaseoli 4292. I, growth in minimal medium with 10 g
of mannitol liter
1
, entering stationary phase at 1 10
9
CFU ml
1
; F, growth in minimal medium with 0.3 g of mannitol liter
1
, entering stationary phase at 2.6
10
8
CFU ml
1
. Cultures were osmotically stressed by the addition of sterile NaCl to a nal concentration of 2.5 M. The stress was applied to cultures 10 days after
inoculation, when they had been in stationary phase for 5 days (arrow 1 in Fig. 2) (A), or 30 days after inoculation, when they had been in stationary phase for 25 days
(arrow 2 in Fig. 2) (B). Survival after the osmotic stress was monitored by viable-cell counting. Error bars represent standard deviations.
984
is the fraction of the low-density culture that was resistant to
osmotic stress after 5 days into stationary phase.
High-density stationary-phase cultures contain an extracel-
lular component that promotes adaptation to starvation sur-
vival. We investigated whether an extracellular compound(s)
accumulates that is able to promote the adaptation to station-
ary-phase survival. We took advantage of the fact that R. legu-
minosarum survives poorly following carbon starvation by re-
suspension in MM-C (36). The effect of lter-sterilized, spent
culture medium from a high-density culture on the survival of
a low-density culture was investigated. A low-density culture,
entering stationary phase at 10
8
CFU ml
1
, was harvested and
resuspended in MM-C containing spent medium from a high-
density stationary-phase culture, and the viability was moni-
tored. The data in Fig. 4 clearly show that the HDSM promotes
survival of the low-density culture. At 96 h after resuspension,
98% of the cells resuspended in HDSM were viable, with only
21% of those resuspended in MM-C retaining viability. HDSM
could also promote survival or adaptation of exponentially
growing cells to stationary phase (data not shown). This indi-
cates that the action of the HDSM component is not limited to
cells in the transition phase between exponential and station-
ary phases. Preliminary characterization indicated that the ac-
tive component(s) of HDSM was small (3 kDa), stable, and
heat resistant (surviving 100C for 15 min), and it was protease
resistant, suggesting that it was not a protein (data not shown).
The accumulation of an extracellular component may trigger
the adaptation to stationary-phase survival in high-density cul-
tures, with the failure of low-density cultures to adapt being a
consequence of the compound(s) not reaching a threshold
level. This was demonstrated by growing a culture to carbon-
starved stationary phase while taking samples of spent me-
dium, at cell densities above and below the threshold necessary
for successful stationary-phase survival, and testing their ability
to promote survival in resuspension assays (Fig. 5). The spent
medium from an exponentially growing culture that had
reached a density of 5 10
7
CFU ml
1
(Fig. 5A) was unable
to promote survival of low-density cultures following resuspen-
sion (Fig. 5B). In contrast, spent medium from an exponential-
phase culture at approximately 2 10
9
CFU ml
1
(Fig. 5A),
was able to promote starvation survival following resuspension
(Fig. 5B). Interestingly, spent medium from high-density cul-
tures at 32 days into stationary phase was also able to pro-
mote starvation survival in resuspension assays, indicating that
the spent-medium component is still present and active during
long-term stationary phase (Fig. 5B). Therefore, a molecule(s)
secreted into the medium has a role in the adaptation of R.
leguminosarum cultures to stationary phase.
An AHL promotes cell density-dependent survival. The AHL
7cisHtDHL has been shown to induce a cell density-dependent
response in R. leguminosarum (12, 34). It activates the RhiR
protein to induce expression of the rhiABC operon and to
inhibit growth (12). We tested whether 7cisHtDHL could also
protect low-density cultures during starvation survival. Cells
from a low-density, late-exponential-phase culture were resus-
pended in MM-C with 7cisHtDHL added, and survival was
monitored. Ninety-six percent of the cells remained viable 4
days after resuspension in the 7cisHtDHL-containing medium,
compared to 98% of cells resuspended in HDSM and only 21%
of the cells in MM-C (Fig. 4). 7cisHtDHL is clearly able to
promote the survival of low-density cultures. As 7cisHtDHL
has been demonstrated to be present in spent culture medium
of R. leguminosarum 4292, it is possible that the HDSM com-
ponent promoting stationary-phase survival is an AHL similar
to or the same as 7cisHtDHL, although this is not directly
demonstrated by the present data. The 7cisHtDHL used was a
racemic mixture of L and D stereoisomers that differ in the
orientation of the 3-hydroxy group, although the biological
importance of chirality at the hydroxylated C-3 position has yet
to be established (12, 34). The experiment was repeated with
various concentrations of 7cisHtDHL in order to nd the
threshold value at which it is effective at promoting survival. A
FIG. 4. Effect of HDSM and an AHL on stationary-phase survival of R.
leguminosarum bv. phaseoli 4292. Late-exponential-phase, low-density cultures
(density, 1 10
8
CFU ml
1
) were harvested and resuspended to the same
density in spent medium from high density cultures (density, 1.5 10
9
CFU
ml
1
) of strain 4292 (I), in MM-C (), and in MM-C containing 200 ng of
7cisHtDHL ml
1
(). Survival was monitored by viable-cell counting. Data are
the mean values from four experiments, with error bars representing standard
deviations.
TABLE 1. Role of the Sym plasmids pRP2JI and pRL1JI in
production of or response to the HDSM components conferring
stationary-phase survival and the AHL 7cisHtDHL
a
Resuspension medium
Survival (%)
b
of R. leguminosarum:
4292/pRP2JI 8401/pRL1JI 8401
MM-C 30 28 25
MM-C plus 7cisHtDeHL
(200 ng ml
1
)
96 44 41
HDSM from R. leguminosarum:
8401 86 40 35
8401/pRL1JI 94 99 11
4292/pRP2JI 96 89 23
a
Low-cell-density, late-exponential-phase cells (cell density, 1 10
8
CFU
ml
1
) were harvested and resuspended to 5 10
7
CFU ml
1
in either HDSM
from the indicated strain, MM-C, or MM-C containing 200 ng of 7cisHtDHL
ml
1
, and survival was monitored by viable-cell counting.
b
Percentage of the original population that was viable 4 days after resuspen-
sion.
VOL. 181, 1999 STATIONARY-PHASE SURVIVAL OF R. LEGUMINOSARUM 985
concentration of 200 ng of 7cisHtDHL ml
1
conferred full
protection on low-density cultures in stationary phase (Fig. 6)
while less than 100 ng ml
1
caused only a small difference in
starvation survival. These data support the idea that a quorum-
sensing effect, involving AHLs, is functioning to improve the
stationary-phase survival of R. leguminosarum. It seems rea-
sonable to conclude that AHLs have a role in the survival of
high-density cultures and that the absence of a quorum leads to
loss of viability in low-density cultures.
Role of the Sym plasmids pRL1JI and pRP2JI in cell den-
sity-dependent survival. The R. leguminosarum 4292 strain
used so far in this study contains the biovar phaseoli Sym
plasmid pRP2JI. In order to investigate the role of Sym plas-
mids in cell density-dependent starvation survival, the related
strains R. leguminosarum 8401 (no Sym plasmid) and R. legu-
minosarum 8401/pRL1JI (strain 8401 with the biovar viciae
Sym plasmid pRL1JI) were also used. Growth of 8401/pRL1JI
to stationary phase in carbon-limited minimal medium indi-
cated that changing the Sym plasmid from pRP2JI to pRL1JI
had no effect on the cell density-dependent stationary-phase
survival (Fig. 7). R. leguminosarum 8401/pRL1JI grown to high
density survived subsequent starvation well, while cultures
grown to low density showed a loss of viability in stationary
phase similar to that of R. leguminosarum 4292/pRP2JI (Fig. 1
and 7). R. leguminosarum 8401 cultures (no Sym plasmid)
showed a loss and subsequent stabilization of viability similar
to those of the strains with a Sym plasmid. Also, the R. legu-
minosarum 8401 cultures grown to high-cell-density stationary
phase did not show a signicant difference in survival com-
pared to strains with a Sym plasmid.
The Sym plasmids pRL1JI and pRP2JI play an essential
role in the production of the extracellular stationary-phase
survival factor. We examined whether Sym plasmids have a
role in the production of or response to the spent-medium
component. Cells from low-density, late-exponential-phase
cultures were resuspended in HDSM, MM-C, or MM-C con-
taining 200 ng of 7cisHtDHL per ml. Both cells and HDSM
from three strains of R. leguminosarum (4292/pRP2JI, 8401/
pRL1JI, and 8401) were used, and the results are summarized
in Table 1. As expected, none of the strains could survive well
in MM-C. Also, as shown in Fig. 4, R. leguminosarum 4292/
pRP2JI cells could survive in MM-C if 7cisHtDHL was added.
However, neither R. leguminosarum 8401/pRL1JI nor R. legu-
minosarum 8401 showed signicantly increased survival in the
presence of 7cisHtDHL. This leads to two important conclu-
sions. First, the ability of R. leguminosarum 4292 to respond to
7cisHtDHL is encoded on the Sym plasmid pRP2JI, as strain
8401 is cured of this plasmid and can no longer respond to
7cisHtDHL. Second, the fact that 7cisHtDHL does not support
survival of R. leguminosarum 8401/pRL1JI suggests that its cell
density-dependent starvation survival must be mediated by a
second spent-medium component. Consistent with this conclu-
FIG. 5. Effect of spent medium from different stages of the growth curve on stationary-phase survival of R. leguminosarum 4292 in resuspension assays. (A) R.
leguminosarum 4292 was grown in minimal medium (5 g of mannitol liter
1
), entering stationary phase at a high cell density (3 10
9
CFU ml
1
). During growth,
samples were taken at 24 h (5 10
7
CFU ml
1
), 96 h (2 10
9
CFU ml
1
), and 30 days (3 10
9
CFU ml
1
) for the preparation of spent medium. (B) Cells obtained
from a low-density (10
8
CFU ml
1
), late-exponential-phase culture, grown in minimal medium with 0.3 g of mannitol liter
1
, were resuspended to 10
8
CFU ml
1
in MM-C () or in spent-medium samples taken at the different time points as described for panel A, and the viability was monitored. Spent medium was from 24 h
(F), 96 h (I), and 30 days (}) after inoculation. Error bars represent standard deviations.
986 THORNE AND WILLIAMS J. BACTERIOL.
sion, we found that when cells from each of the three strains
were resuspended in HDSM from strain 8401, which contains
no Sym plasmid but is known to produce 7cisHtDHL (12, 34),
only strain 4292/pRP2JI showed improved survival (Table 1).
However, the cell density-dependent survival of R. legumino-
sarum 8401/pRL1JI does require a spent-medium component.
Although this strain is unable to produce 7cisHtDHL (12), it
can produce a spent-medium component, which we refer to as
SMC8401, capable of promoting starvation survival of itself
and of R. leguminosarum 4292/pRP2JI. The strain with pRP2JI
could respond to both 7cisHtDHL and SMC8401. Addition-
ally, HDSM from R. leguminosarum 4292/pRP2JI promotes
survival of R. leguminosarum 4292/pRP2JI and 8401/pRL1JI,
but not 8401. This implies that 4292/pRP2JI produces a mol-
ecule (which we call SMC4292) that is able to promote survival
of strain 8401/pRL1JI and that is chemically distinct from
7cisHtDHL. Whether SMC8401 and SMC4292 are chemically
identical remains to be established, but it is clear from a recent
study that Rhizobium strains can produce a signicant number
of different AHLs (30).
Which region of the Sym plasmid is essential for the pro-
duction of the spent-medium component? To locate which
region of the Sym plasmid pRL1JI is required for production
of AHL8401, we used four cosmid clones in E. coli, each
containing a different region of pRL1JI (see Materials and
Methods). HDSM was obtained from cultures of E. coli carry-
ing one of each of these clones, and its effect was compared to
those of HDSM from R. leguminosarum 4292/pRP2JI and of
MM-C. HDSM from E. coli strains carrying pIJ1085, pIJ1527,
and pIJ1529 proved unable to support starvation survival of R.
leguminosarum 4292/pRP2JI cultures (Fig. 8). In contrast, 81%
of the cells resuspended in spent medium from E. coli/pIJ1086
were still viable 4 days after resuspension (Fig. 8). This nding
indicates that the pRL1JI-derived DNA in cosmid pIJ1086
contains genes that have a role in the production of the starva-
tion survival factor produced by pRL1JI. Interestingly, pIJ1086
contains the rhiABCR genes. RhiR activates the rhiABC genes
in the presence of 7cisHtDHL (3, 8).
DISCUSSION
In this paper we have demonstrated the cell density-depen-
dent starvation survival of R. leguminosarum. This phenome-
non was found to occur in both carbon- and nitrogen-starved
cultures but not phosphorus-starved cultures. It is possible that
the accumulation of large amounts of phosphorus storage com-
pounds, such as polyphosphates, during exponential growth (4,
5) allows the cells to survive phosphorus starvation more ef-
fectively and disguises any cell density effects.
Quantication of an absolute threshold cell density, above
which cells are able to survive in stationary phase, was not
possible, as the density threshold was dependent on both the
growth conditions (Fig. 1) and the strains starved (compare
survival of strains 4292 and 8401/pRL1JI in Fig. 1A and 7,
respectively). R. leguminosarum 4292/pRP2JI cultures survived
carbon starvation when they entered stationary phase at den-
sities greater than 1.5 10
9
CFU ml
1
. However, in YEM
medium, cultures needed to enter stationary phase at densities
at or above 10
10
CFU ml
1
to survive subsequent starvation.
YEM cultures have a signicantly higher growth rate than
minimal medium cultures, and the growth rate prior to starva-
tion will affect the physiology of the cell in numerous ways: the
numbers of ribosomes, DNA and protein levels, and number of
active replication forks are all altered (2, 15). Consequently,
the growth rate may affect the adaptation to starvation survival
and the long-term survival outcome. A link between growth
rate and starvation survival is suggested by the nding that
bacterial populations grown for many generations in chemostat
culture are taken over by mutants which have faster doubling
times but which are less able to survive starvation (22).
When fully adapted to stationary-phase survival, R. legu-
minosarum is cross-protected against several environmental
stresses (36). However, the osmotic stress resistance of station-
ary-phase cultures (Fig. 3) indicated that at 5 days into station-
ary phase, only a small proportion of cells in low-density cul-
tures was protected against osmotic stress and consequently
fully adapted to survival. Therefore, it is likely that at the onset
of starvation, low-density cultures consist of a heterogeneous
population. Approximately 20% of the cells either are able to
successfully adapt themselves to stationary phase or possess
the ability to subsequently adapt by utilizing metabolites
leaked into the growth medium by dying cells. The reason for
the further loss of viability of low-density cultures after 30 days
in stationary phase (Fig. 2) is unclear.
We clearly demonstrate that an extracellular compound that
is able to promote stationary-phase survival accumulates in
high-density cultures. Resuspension assays in the presence of
spent medium from high-density cultures (HDSM) demon-
strated that the compound active in promoting survival accu-
mulates to effective concentrations at high but not low cell
densities. Evidence is provided for the role of AHLs in promoting
the starvation survival of R. leguminosarum. 7cisHtDHL, a previ-
FIG. 6. Concentration of 7cisHtDHL needed to produce a stationary-phase
response. Cells from late-exponential-phase, low-density cultures (density, 10
8
CFU ml
1
) were resuspended in MM-C or in MM-C containing various con-
centrations of 7cisHtDHL. The percent survival of the cultures was determined
4 days after resuspension by viable-cell counting. Data are the mean values from
four experiments, with error bars representing standard deviations.
VOL. 181, 1999 STATIONARY-PHASE SURVIVAL OF R. LEGUMINOSARUM 987
ously identied autoinducer from R. leguminosarum (12, 34), pro-
motes survival of R. leguminosarum 4292/pRP2JI in resuspension
assays (Fig. 4 and 6). The ability of strain 4292/pRP2JI to respond
to the starvation survival-promoting function of 7cisHtDHL is
encoded by pRP2JI, since strain 8401, which is cured of this
plasmid, was unable to respond to 7cisHtDHL. R. leguminosarum
8401 can produce 7cisHtDHL but does not respond to it, while R.
leguminosarum 8401/pRL1JI does not produce 7cisHtDHL (12).
Consistent with this idea is that following resuspension of the
three strains in HDSM from strain 8401, only strain 4292/pRP2JI
showed improved starvation survival (Table 1). 7cisHtDHL pro-
motes transcription of an rhiA-lacZ gene fusion and growth inhi-
bition as long as the RhiR protein, encoded by the Sym plasmid
pRL1JI but not by pRP2JI, is present (6, 12). Therefore, the
7cisHtDHL-promoted starvation survival of strain 4292/pRP2JI
cannot involve the transcriptional regulator RhiR but must
require an unidentied regulator encoded by pRP2JI. This
suggests that the cell density-dependent starvation survival re-
sponse and the cell density-dependent induction of the rhiABC
operon in R. leguminosarum, although both mediated by
7cisHtDHL, represent separate regulatory pathways. Also, the
amount of 7cisHtDHL required to promote starvation survival
(Fig. 6) was substantially higher (200 ng ml
1
) than that
required to induce rhiA-lacZ fusions (10 ng ml
1
[12]). Gray
et al. (12) showed that 7cisHtDHL added to exponentially
growing cultures of R. leguminosarum 8401/pRL1JI caused ces-
sation of growth. However, we did not nd 7cisHtDHL to be
able to promote starvation survival of 8401/pRL1JI in our
experiments, perhaps suggesting that we are not observing the
same phenomenon as Gray et al. (12).
Previous studies have shown that 7cisHtDHL is produced by
strains of R. leguminosarum bv. phaseoli (12, 39). It seems
possible, although it is not proven, that 7cisHtDHL is the
compound that accumulates in high-density R. leguminosarum
4292/pRP2JI cultures and promotes adaptation to and survival
in subsequent carbon-starved stationary phase (Fig. 1). How-
ever, our data (Table 1) indicate that R. leguminosarum 4292/
pRP2JI produces another survival-promoting compound in ad-
dition to 7cisHtDHL and that strain 8401/pRL1JI produces a
compound distinct from 7cisHtDHL. It is clear from recent
work that some Rhizobium species produce a large number of
AHLs (30).
The widespread role of cell density-mediated control of bac-
terial physiology (quorum sensing) is now recognized (10, 32).
How widespread is AHL-mediated cell density-dependent sta-
tionary-phase survival? In E. coli and other gram negative
bacteria, the development of stationary-phase properties and
subsequent stationary-phase survival is coordinately regulated
by the action of the stationary-phase sigma factor RpoS (ref-
erence 40 and references therein). The cellular level of RpoS
increases signicantly upon entry into stationary phase, and it
is controlled at the level of protein stability (40). Huisman and
Kolter (17) proposed a mechanism by which E. coli cells sense
starvation via a homoserine lactone (HSL)-dependent signal-
ling pathway which leads directly to increased cellular levels of
RpoS. They demonstrated that HSL, or a metabolite derived
from it, induced rpoS expression and increased cellular levels
of RpoS. They postulated that HSL constitutes an intracellular
signal that accumulates in starved cells regardless of cell den-
sity and that at high cell densities this signal molecule can be
acylated to allow diffusion across membranes, forming an ex-
tracellular, cell density-dependent signal. However, in recent
studies AHL synthesis from intracellular homoserine lactone
pools has been demonstrated not to be valid in studies in which
LuxI-type AHL synthases have been expressed in E. coli.
Rather AHL are synthesized from S-adenosylmethionine and
acyl carrier protein conjugates (14, 25, 33). No further direct
support for these ideas has been forthcoming from experi-
FIG. 7. Effect of the Sym plasmid pRL1JI on stationary-phase survival at high and low culture cell densities. Cultures of R. leguminosarum 8401/pRL1JI ( and
I) and R. leguminosarum 8401 (E and F) were grown under carbon-starved conditions at either a high (open symbols) (5 g of mannitol liter
1
, entering stationary phase
at 2 10
8
CFU ml
1
) or low (closed symbols (0.3 g of mannitol liter
1
, entering stationary phase at 1 10
7
CFU ml
1
) cell density. Growth was monitored by
viable-cell counting, and data are the mean values from four experiments, with error bars representing standard deviations.
988 THORNE AND WILLIAMS J. BACTERIOL.
ments on E. coli. However, recent work with Pseudomonas
aeruginosa has shown that the expression of rpoS is abolished in
P. aeruginosa lasR mutants, which are unable to respond to
either of the two AHLs [N-(3-oxododecanoyl)-L-homoserine
lactone and N-butyryl homoserine lactone] produced by this
bacterium (24). This provided direct evidence linking cell den-
sity-dependent signalling systems with the regulation of the
stationary-phase sigma factor RpoS. Further experiments
showed that the regulator RhlR, when activated by N-butyryl
homoserine lactone, functions as a transcriptional activator of
RpoS (24). The full signicance of these ndings for the sta-
tionary-phase or starvation survival of P. aeruginosa is not
clear. At present it is not known whether an RpoS homologue
is involved in stationary-phase survival of R. leguminosarum.
Recently it has been shown that an extracellular signal mole-
cule(s) is involved in the carbon starvation response of Vibrio
sp. strain S14 (35). When an extract of stationary-phase medium
was added to exponential-phase Vibrio sp. strain S14, it re-
sulted in the up-regulation of carbon starvation-induced pro-
teins, a process that was inhibited by halogenated furanones,
compounds that are putative antagonists of AHLs (35).
ACKNOWLEDGMENTS
This work was funded by a grant from The Royal Society. S.H.T. was
supported by a BBSRC research studentship.
We are very grateful to Kendall Gray for providing us with samples
of R. leguminosarum autoinducer and to Allan Downie for strains.
REFERENCES
1. Barkley, W. E., and J. H. Richardson. 1992. Laboratory safety, p. 715734. In
P. Gerhardt, R. G. E. Murray, W. A. Woods, and N. R. Krieg (ed.), Methods
for general and molecular bacteriology. American Society for Microbiology,
Washington, D.C.
2. Bremer, H., and P. P. Dennis. 1987. Modulation of chemical composition
and other parameters of the cell by growth rate, p. 15271542. In F. C.
Neidhardt, J. L. Ingraham, K. B. Low, B. Magasanik, M. Schaehter, and
H. E. Umbarger (ed.), Escherichia coli and Salmonella typhimurium: cellular
and molecular biology. American Society for Microbiology, Washington,
D.C.
3. Cubo, M. T., A. Economou, G. Murphy, A. W. B. Johnston, and J. A. Downie.
1992. Molecular characterization and regulation of the rhizosphere-ex-
pressed genes rhiABCR that can inuence nodulation by Rhizobium legu-
minosarum biovar viciae. J. Bacteriol. 174:40264035.
4. Dawes, E. A. 1976. Endogenous metabolism and the survival of starved
prokaryotes. SGM Symp. 26:1953.
5. Dawes, E. A., and P. J. Senior. 1973. The role and regulation of energy
reserve polymers in micro-organisms. Adv. Microb. Physiol. 10:135266.
6. Dibb, N. J., J. A. Downie, and N. J. Brewin. 1984. Identication of a rhizo-
sphere protein encoded by the symbiotic plasmid of Rhizobium leguminosa-
rum. J. Bacteriol. 158:621627.
7. Downard, J., and D. Toal. 1995. Branched-chain fatty acids: the case for a
novel form of cell-cell signalling during Myxococcus xanthus development.
Mol. Microbiol. 16:171175.
7a.Downie, J. A. Personal communication.
8. Downie, J. A., M. Qing-Sheng, C. D. Knight, G. Hombrecher, and A. W. B.
Johnston. 1983. Cloning of the symbiotic region of Rhizobium leguminosa-
rum: the nodulation genes are between the nitrogenase and a nifA-like gene.
EMBO J. 2:947952.
9. Foster, J. W., and M. P. Spector. 1995. How Salmonella survive against the
odds. Annu. Rev. Microbiol. 49:145174.
10. Fuqua, W. C., S. C. Winans, and E. P. Greenberg. 1994. Quorum sensing in
bacteria: the LuxR-LuxI family of cell-density-responsive transcriptional reg-
ulators. J. Bacteriol. 176:269275.
11. Givskov, M., L. Eberl, S. Moller, L. K. Poulsen, and S. Molin. 1994. Re-
sponses to nutrient starvation in Pseudomonas putida KT2442: analysis of
general cross-protection, cell shape, and macromolecular content. J. Bacte-
riol. 176:714.
12. Gray, K. M., J. P. Pearson, J. A. Downie, B. E. A. Boboye, and E. P.
Greenberg. 1996. Cell-to-cell signalling in the symbiotic nitrogen-xing bac-
terium Rhizobium leguminosarum: autoinduction of stationary-phase and
rhizosphere-expressed genes. J. Bacteriol. 178:372376.
13. Grossman, A. D., and R. Losick. 1988. Extracellular control of spore forma-
tion in Bacillus subtilis. Proc. Natl. Acad. Sci. USA 85:43694373.
14. Hanzelka, B. L., and E. P. Greenberg. 1996. Quorum sensing in Vibrio
scheri: evidence that S-adenosylmethionine is the amino acid substrate for
autoinducer synthesis. J. Bacteriol. 178:52915294.
15. Harder, W., and L. Dijkhuizen. 1983. Physiological responses to nutrient
limitation. Annu. Rev. Microbiol. 39:123.
16. Harrison, A. P. 1960. The response of Bacterium lactis aerogenes when held
at growth temperatures in the absence of nutriment: an analysis of survival
curves. Proc. R. Soc. London Ser. B 152:418428.
17. Huisman, G. W., and R. Kolter. 1994. Sensing starvation: a homoserine
lactone-dependent signalling pathway in Escherichia coli. Science 265:537
539.
18. Kaplan, H. B., and L. Plamann. 1996. A Myxococcus xanthus cell density-
sensing system required for multicellular development. FEMS Microbiol.
Lett. 139:8995.
19. Kim, S. K., and D. Kaiser. 1992. Control of cell density and pattern by
intercellular signalling in Myxococcus development. Annu. Rev. Microbiol.
46:117139.
20. Kjelleberg, S., M. Hermansson, and P. Marden. 1987. The transient phase
between growth and nongrowth of heterotrophic bacteria, with emphasis on
the marine-environment. Annu. Rev. Microbiol. 41:2549.
21. Kolter, R., D. A. Siegele, and A. Tormo. 1993. The stationary phase of the
bacterial life-cycle. Annu. Rev. Microbiol. 47:855874.
22. Korona, R. 1996. Genetic divergence and tness convergence under uniform
selection in experimental populations of bacteria. Genetics 143:637644.
23. Lamb, J. W., G. Hombrecher, and A. W. B. Johnston. 1982. Plasmid-deter-
mined nodulation and nitrogen-xation abilities in Rhizobium phaseoli. Mol.
Gen. Genet. 186:449452.
24. Lati, A., M. Foglino, K. Tanaka, P. Williams, and A. Lazdunski. 1996. A
hierarchical quorum-sensing cascade in Pseudomonas aeruginosa links the
transcriptional activators LasR and RhlR (VsmR) to expression of the sta-
tionary-phase sigma factor RpoS. Mol. Microbiol. 21:11371146.
25. More, M. I., D. Finger, J. L. Stryker, C. Fuqua, A. Eberhard, and S. C.
Winans. 1996. Enzymatic synthesis of a quorum-sensing autoinducer through
use of dened substrates. Science 272:16551658.
FIG. 8. Identication of regions of the Sym plasmid pRL1JI required for the
production of the spent-medium component. Survival of R. leguminosarum 4292
cells harvested from late-exponential-phase, low-cell-density cultures was mon-
itored after resuspension in R. leguminosarum HDSM (I), MM-C (), or spent
medium from late-exponential-phase E. coli cultures carrying cosmids containing
different regions of the Sym plasmid pRL1JI (, pIJ1085; {, pIJ1086; ,
pIJ1527; , pIJ1529). For all experiments, the data are the means from four
experiments and the standard deviation was never more than 6%.
VOL. 181, 1999 STATIONARY-PHASE SURVIVAL OF R. LEGUMINOSARUM 989
26. Nystrom, T., N. H. Albertson, K. Flardh, and Kjelleberg, S. 1990. Physio-
logical and molecular adaptation to starvation and recovery from starvation
by the marine Vibrio sp. S14. FEMS Microbiol. Ecol. 74:129140.
27. Ostling, J., L. Holmquist, K. Flardh, B. Svenblad, A. Jouper-Jaan, and S.
Kjelleberg. 1993. Starvation and recovery of Vibrio, p. 103127. In S. Kjel-
leberg (ed.), Starvation in bacteria. Plenum Press, New York, N.Y.
28. Postgate, J. R. 1976. Death in macrobes and microbes. SGM Symp 26:118.
29. Postgate, J. R., and J. R. Hunter. 1963. The survival of starved bacteria.
J. Appl. Bacteriol. 26:295306.
30. Rosemeyer, V., J. Michels, C. Verrath, and J. Vanderleyden. 1998. luxI- and
luxR- homologous genes of Rhizobium etli CNPAF512 contribute to synthesis
of autoinducer molecules and nodulation of Phaseolus vulgaris. J. Bacteriol.
180:815821.
31. Roszak, D. B., and R. R. Colwell. 1987. Survival strategies of bacteria in the
natural environment. Microbiol. Rev. 51:365379.
32. Salmond, G. P. C., B. W. Bycroft, G. S. A. B. Stewart, and P. Williams. 1995.
The bacterial enigma: cracking the code of cell-cell communication. Mol.
Microbiol. 16:615624.
33. Schaefer, A. L., D. L. Val, B. L. Hanzelka, J. E. Cronan, Jr., and E. P.
Greenberg. 1996. Generation of cell-to-cell signals in quorum sensing: acyl
homoserine lactone synthase activity of a puried Vibrio scheri LuxI pro-
tein. Proc. Natl. Acad. Sci. USA 93:95059509.
34. Schripsema, J., K. E. E. de Rudder, T. B. van Vliet, P. P. Lankhorst, E. de
Vroom, J. W. Kijne, and A. A. N. van Brussel. 1996. Bacteriocin small of
Rhizobium leguminosarum belongs to the class of N-acyl-L-homoserine lac-
tone molecules, known as autoinducers and as quorum-sensing cotranscrip-
tion factors. J. Bacteriol. 178:366371.
35. Srinivasan, S., J. Ostling, T. Charlton, R. De Nys, T. Takayama, and S.
Kjelleberg. 1998. An extracellular signal molecule(s) involved in the carbon
starvation response of marine Vibrio sp. strain S14. J. Bacteriol. 180:201209.
36. Thorne, S. H., and H. D. Williams. 1997. Adaptation to nutrient starvation
in Rhizobium leguminosarum bv. phaseoli: analysis of survival, stress resis-
tance, and changes in macromolecular synthesis during entry to and exit
from stationary phase. J. Bacteriol. 179:68946901.
37. van Elsas, J. D., and L. S. van Overbeek. 1993. Bacterial responses to soil
stimuli, p. 5579. In S. Kjelleberg (ed.), Starvation in bacteria. Plenum Press,
New York, N.Y.
38. Vincent, J. M. 1970. A manual for the practical study of root nodule bacteria.
Blackwell, Oxford, England.
39. Wijffelman, C. A., E. Pees, A. A. N. van Brussel, and P. J. J. Hooykaas. 1983.
Repression of small bacteriocin excretion in Rhizobium leguminosarum and
Rhizobium trifolii by transmissible plasmids. Mol. Gen. Genet. 192:171176.
40. Zgurskaya, H. I., M. Keyhan, and A. Matin. 1997. The
s
level in starving
Escherichia coli cells increases solely as a result of its increased stability,
despite decreased synthesis. Mol. Microbiol. 24:643651.
990 THORNE AND WILLIAMS J. BACTERIOL.