1
Caloric restriction causes symmetric cell division and delays aging in Escherichia coli
Uttara Lele
1
, Prajakta Belsare
2
, Samit Watve
3
, Snehal Bari
2
, Sarika Karande
2
, Milind
Watve
1*,
1 Anujeeva Biosciences Pvt Ltd, Pune, India
2 Dept of Microbiology, Abasaheb Garware College, Pune, India
3 Institute of Bioinformatics and Biotechnology, University of Pune, India
Aging is one of the most intriguing processes of biology and despite decades of
research, many aspects of aging are poorly understood. Aging is known to occur in
bacteria and yeast that divide with morphological asymmetry
1, 2
. Morphologically
symmetrically dividing bacteria such as Escherichia coli were assumed not to age until
they were shown to divide with functional asymmetry leading to aging and death of
some of the cells even in exponentially growing cultures
3
. In asymmetrically dividing E.
coli the newly synthesized components are presumed to occupy one pole so that after
division one of the daughter cells receives newly synthesized components whereas the
other retains the older components
3, 4
. Mathematical models predicted that at the
population level, asymmetric growth should result in higher growth rate
5, 6, 7
and
symmetric growth in higher growth yield
7
. Therefore, arguably symmetric cell division
should be selected in low nutrient environments and asymmetric division in nutrient
rich environments. A further prediction was that lower substrate concentrations should
strengthen repair mechanisms and suppress aging whereas higher substrate
concentrations suppress repair and enhance aging
7
. We show here that E. coli divides
2
more symmetrically under caloric restriction, that both genetic selection and phenotypic
plasticity are important determinants of cell division symmetry and also that the
proportion of cells that stop dividing and therefore are presumably dead is significantly
lower in symmetrically dividing cultures. However, contrary to the prediction,
symmetry was not always accompanied by reduced growth rate. These results
demonstrate that asymmetry of division in E. coli is not hardwired but responsive to the
nutritional environment. This provides a new perspective on why caloric restriction
increases lifespan in organisms ranging from microbes to mammals
8
. Symmetry of
division may be a mechanism spanning across the width of life forms but regulating
aging in different ways in different forms.
The central mechanism of cellular aging is likely to be cell division asymmetry with
respect to the distribution of older and newly synthesized components. In animal systems
asymmetric division is involved in differentiation and self renewing of stem cells and
germline cells
9-12
. The phenomenon of asymmetry leading to aging can also be seen in
unicellular forms such as bacteria and yeast. In budding yeast oxidatively damaged proteins
remain preferentially in the mother cell whereas newly synthesized components occupy the
bud cells
13
so that eventually the rate of reproduction of the mother cell declines until it stops
dividing. In bacteria such as Caulobacter sp. that have a morphological asymmetry and
differentiation, aging similar to budding yeast was demonstrated
2
.
If asymmetric division is the key to cellular aging, unicellular organisms dividing
symmetrically should be immune to aging as long as environmental conditions are
favourable. It was shown however that fission yeast as well as Escherichia coli cells
distribute the older and newly synthesized components to the daughter cells asymmetrically
3,
4, 14
resulting into one of the daughter cells being born older than the other. In fact, the two
cells can be viewed as a parent cell and a daughter cell rather than as two sister cells. This
3
gives rise to a population dynamics that is similar to multicellular organisms that show age
structured populations. Based on a Laslie Matrix model Watve et al
7
simulated symmetric
and asymmetric division and its effects on population growth. Their model assumed that the
efficiency of cellular components declined with age and growth rate of a cell was a function
of the relative proportions of new and old components. In asymmetrically dividing cells
although a proportion of cells accumulated older components and ultimately died, young cells
were being continuously generated and therefore the growth rate of the population remained
high. On the other hand, in a symmetrically dividing culture, all the cells retained a
proportion of older components resulting in slower growth of the entire population. Other
modelling approaches agreed on the growth rate advantage of asymmetric division
5, 6
. In the
Leslie Matrix model, at an optimum rate of repair the growth yield or biomass conversion
efficiency of symmetrically dividing cells was predicted to be higher than asymmetrically
dividing ones. Based on the simulation results Watve et al
7
argued that symmetric division
was a better strategy under low nutrient conditions when biomass conversion efficiency
would be more critical. On the other hand under a nutritionally rich but highly competitive
environment, asymmetrically dividing cells would gain a reproductive advantage. A shift in
cell division strategy may be observable in ecological or evolutionary time. There is a
suggestion that asymmetric division in yeast may not be hardwired but responsive to
environmental conditions
13, 15
, although this possibility has never been rigorously tested. If
the cells have phenotypic plasticity, they may change cell division strategies in response to
the environment in a short duration. Alternatively optimum cell division may evolve by
prolonged selection in a given environment. Such a change should be observable in
evolutionary time.
In order to test these predictions, we reared four strains of E. coli, starting with an
environmental isolate that was able to grow on both nutrient rich and dilute media, on two
4
types of substrates, one with glucose as the sole source of carbon and energy and the other
with a complex substrate - peptone, each in 1 and 0.01 g/dl concentration. In order to keep the
temporal variation in substrate concentration to a minimum, the cultures were transferred by
mid to late exponential phase so that during selection period none of the cultures faced
starvation and stationary phase (we denote the strain selected under high caloric conditions as
H and the one selected under low caloric conditions as L. The current medium condition for
the experiment is denoted by small letters h and l for high and low caloric conditions
respectively, e.g. a strain selected under high concentration but currently being grown in low
concentration is denoted as Hl)
.
After an estimated 500 and 1000 generations of selection
each of the strains was examined for growth rate, growth yield and symmetry of cell division
on both high and low nutrient media.
Consistent with the expectation, the strain selected under low glucose concentration
gave higher yields expressed as total cell protein per unit substrate consumed as compared to
the strain selected under high glucose concentration at any given current glucose
concentration, i.e. growth yield of Ll was higher than that of Hl and Lh was higher than Hh.
Selection in low or high concentration of peptone had similar effect when grown in glucose
(figure 1) showing that the effect of selection was not substrate specific. Peptone being a
complex substrate, growth yield could not be calculated precisely but Lh gave higher
stationary phase absorbance than Hh in glucose as well as peptone media (data not shown).
The growth of any of the strains was slower in dilute media but the effect of selection on
growth rate was complex. For Lh in
peptone, the growth rate substantially declined by 500
generations as the growth yields improved. However, after 1000 generations of selection,
growth rate was observed to be high and comparable to Hh in peptone. The strains selected in
glucose, did not differ from each other in growth rates after 500 or 1000 generations although
growth of both was faster than the ancestral strain.
5
For quantifying functional symmetry in cell division we defined an index of
asymmetry based on the assumption that if the cell division was asymmetric, the daughter
cell receiving older components will take longer to complete the next cell division
5
.
Therefore an index of asymmetry was defined as the ratio of the difference in division time
(sign ignored) of two sister cells as to the average division time of the two. Cell division
observations were made by spreading mid exponential phase broth culture on to a slide
layered with agar having the same nutrient composition as the broth. If the cell division in the
broth culture was asymmetric, some of the cells transferred on the slide would have been old
and therefore fail to grow or grow slowly. Therefore death and aging during the broth phase
could be detected by failure or delay in the development of microcolonies. Subsequent
divisions on the slide could be directly observed for about four generations. For each strain
the mean index of cell division asymmetry was calculated after selection for approximately
500 and 1000 generations. According to the observations of Stewart et al
3
and the Watve et al
model
7
, asymmetry between two daughter cells grows as the cell ages. However in an
exponentially growing culture the majority of cells are in the youngest age class that should
show minimum asymmetry. As a result the frequency distribution of the index of asymmetry
should always be positively skewed. The expected skewness was observed in all populations
(see supplementary material). It can be argued that the cell division asymmetry could be a
stochastic event and even chance asymmetries in cell size or component distribution will lead
to an apparently positively skewed distribution. We tested this by fitting two alternative
models to the distributions. Stochastic asymmetry around a mean of zero asymmetry will give
a half-normal or mod-normal distribution, whereas accumulating asymmetry as in the Watve
et al model
7
would give a negative exponential. Out of the six distributions namely those of
Ll, Lh, Hl, Hh after 500 generations in glucose and the wild type in high and low calorie
glucose (Wh and Wl respectively), normal distribution was rejected for all the six whereas
6
negative exponential was rejected only for one (see supplementary material) indicating that
the asymmetry was accumulating as per the aging model and not stochastic.
Both, current substrate concentration and concentration during selection affected the
cell division symmetry (Figure 2a). For wild type strain as well for strains H and L, for 500 or
1000 generations in glucose, the asymmetry index was significantly higher for Hh and Lh as
compared to Ll and Hl
.
However after 1000 generations of selection, the L strains retained
their symmetry in spite of current substrate concentration and the H strains retained
asymmetry even in dilute environments. Thus prolonged selection seems to have resulted into
commitment to symmetric or asymmetric division. In complex media the effect of current
concentration was not significant whereas at both current concentrations strains selected
under high concentration had significantly higher asymmetry (Figure 2b). In both the media
and durations of selection, Hh showed significantly higher indices of asymmetry as compared
to Ll. In multifactorial analysis the effect of selection on asymmetry index was highly
significant and the effect of current concentration was marginally significant where as
duration of selection and type of substrate were non-significant (see supplementary material
for details of statistical analysis). Also selection under high concentrations resulted in
significantly greater variance in asymmetry as compared to selection in low concentration.
The results demonstrate that, both phenotypic plasticity and genetic selection determined the
cell division strategy but prolonged selection resulted into loss of phenotypic plasticity
suggesting that there may be a cost associated with plasticity that exerted a negative selection
effect when plasticity was no more required.
The shift in the cell division symmetry in response to the substrate concentration can
be a passive effect of reduced growth rate. It can be hypothesized that when growth rate is
slow it may be difficult to maintain asymmetry owing to diffusional mixing of old and new
components. As opposed to the model
7
the observed symmetry may only be an effect rather
7
than the cause of slow growth. To test this we used a vitamin B12 auxotroph of E. coli (E.
coli 113-3D ATCC) and subjected it to similar selection under high and low glucose
concentration for 500 generations. Because of its auxotrophic nature its growth rate could be
regulated by the growth factor concentration and made independent of the energy source.
Under growth factor limited conditions the index of asymmetry was significantly higher in
the high concentration selected strain (median=0.19718) as compared low concentration
selected one (median=0.01212) (Mann-Whitney U test; n1=58, n2=53, W=4272.5 and
p<0.0001). This suggests that the concentration of the caloric nutrient influenced the
symmetry of division rather than the growth rate itself or the concentration of the growth
limiting nutrient.
To monitor possible death of cells we scanned between 500 to 700 developing
microcolonies growing on each of the slide cultures that were prepared for cell division
observations. The cells that failed to divide and form microcolonies when the modal
microcolony size exceeded 16 cells were taken as presumably dead during the broth phase of
growth. The proportion of such 'left behind' cells was higher when cells were exposed to
different caloric environment than what they were selected for (i.e. Hl and Lh). When the
selection and current caloric conditions were similar (Hh and Ll), there was a significant
positive correlation between cell division asymmetry and the proportion of left-behind cells
(Figure 3). This is compatible with the theoretical prediction that asymmetric division leads
to aging whereas symmetric division should protect from aging. Further there was a negative
correlation between asymmetry and growth yield (Figure 4). The observations support the
predictions of the Watve et al
7
model that strains adapted to low nutrient concentrations
should adopt symmetric division, show a lower rate of aging and presumable death and have
a higher growth yield. The results differ from the model prediction in that symmetric division
did not obligately reduce growth rate. It is particularly interesting to note that in the strain L
8
on peptone, there was a substantial reduction in growth rate after 500 generations but a rise
again by the 1000
th
generation. This suggests that adopting symmetric cell division may have
reduced the growth rate initially but it may have been compensated later by positive selection
on other mechanisms determining growth rates. We routinely subcultured exponential phase
cultures to avoid starvation or caloric fluctuations during the experimental evolution. This
could have selected for higher growth rates in all the strains irrespective of substrate
concentration.
It is well known that caloric restriction leads to longevity in widely differing
organisms including yeast, C. elegans, Drosophila, and mammals
8
. We demonstrate here that
a similar phenomenon occurs in E. coli and operates by modulating the symmetry of cell
division. Caloric restriction causing symmetric cell division could be a more general
phenomenon not restricted to E. coli. In low calorie environments fission yeast showed
synchronous cell division cycles
16
. Although Cheng et al
16
did not specifically quantify
symmetry of division, maintenance of long term synchrony is not possible without cell
division symmetry. It is possible therefore that caloric restriction resulted in symmetric
division in yeast in their experiments. Further evidence for an association between caloric
restriction, symmetry and aging in yeast is that mutation in Sir2, a gene necessary for
asymmetric segregation in yeast
13
, resulted in a longer chronological life span under caloric
restriction
17
. It appears that Sir2 activation, leading to asymmetric segregation, reduces the
chronological life span in yeast
17, 18
although its role in reproductive life span extension is
debated
18, 19
.
In unicellular organisms asymmetric division may have evolved to increase the
growth rate of the population by continually generating young cells, where as in multicellular
organisms asymmetric division seems to rejuvenate stem cells
9-12
. This creates an apparent
contradiction. Symmetric division seems to delay aging in unicellular organisms as we show
9
here whereas asymmetric division is crucial for maintenance of stem cells and thereby
delaying aging in complex organisms. If caloric restriction induced symmetric division in all
types of organisms, its effects should be opposite in unicellular and multicellular organisms.
However, there is another possible effect of caloric restriction that the model predicted which
may explain its almost universal effect on longevity. Symmetric cell division should be
accompanied by the optimization of repair rates according to the Watve et al
7
model. Our
experiments could not test this prediction directly. However, in the model, increase in the
growth yield accompanying symmetric cell division critically depended on optimization of
repair rates. Therefore higher growth yield can be taken as an indirect evidence for higher
repair rates accompanying symmetric cell division. Yeast grown under caloric restriction had
lower mutation rates
16
suggesting that the triangular relationship between caloric restriction,
cell division symmetry and repair rates could be more generally true and responsible for
longevity in widely differing organisms.
Our experiment involving division symmetry in a growth factor limited auxotroph
suggests that the available caloric nutrient, rather than the actual utilization of the nutrient,
influenced cell division symmetry and aging. This is an interesting parallel to the finding in
Drosophila that perception of food rather than actual intake of food influenced longevity
20
.
Owing to the parallels as well as important differences in the aging processes of bacteria and
higher organisms, it looks possible now that the inclusion of E. coli among the model
organisms for aging research may throw light on some of the yet undiscovered aspects of
aging.
Acknowledgements: We thank Dr. Sharayu Paranjape for advice on statistical analysis,
Supriya Kamble, Neelesh Dahanukar, Kruttika Satbhai, Neelesh Modak, Rohit Shinde for
help in organizing experiments and data recording.
10
References:
1.
budding life span of Saccharomyces cerevisiae. Mech. Ageing. Dev. 12. 47-52 (1980).
2.
Ackermann, M., Sterns, S. and Jenal, U. Senescence in a bacterium with asymmetric
division. Science. 300. 1920 (2003).
3.
Stewart, E. J., Madden, R., Gregory, P. and Taddei, F. Aging and death in an
organism that reproduces by morphologically symmetric division. PLoS Biol. 3. 295
300 (2005)
4.
Linder, A.B., Madden, R., Demarez, A., Stewart, E.J. and Taddei, F. Asymmetric
segregation of proteins aggregates associated with cellular aging and rejuvination.
Proc. Nat. Acad. Sci. 105(8). 3076-3081 (2008).
5.
Ackermann, M., Lin, C., Bergstrom, C. T. and Doebeli, M. On the evolutionary origin
of aging. Aging cell. 6. 235-244 (2007).
6.
Johnson, L. R. and Mangel, M. Life history and the evolution of aging in bacteria and
other single-celled organisms. Mech. Aging. Dev. 127. 786-7993 (2006).
7.
Watve, M. G., Parab, S., Jogdand, P. and Keni, S. Aging may be a conditional
strategic choice and not an inevitable outcome for bacteria. Proc. Natl. Acad. Sci.
103(40). 14831-14835 (2006).
8.
Kirkwood, T. B. L. and Shanley, D. P. Food restriction, evolution and aging. Mech.
Aging. Dev. 129(9). 1011- 1016 (2005).
9.
Lin, H. and Spradling, A. C. A novel group of pumilio mutations affects the
asymmetry division of germline stem cells in the Drosophila ovary. Development.
124. 2463-2476 (1997).
10.
Watt, F. M. and Hogan, B. L. M. Out of eden: stem cells and their niches. Science.
287. 1427-1430 (2000).
11
11.
Takano, H., Ema, H., Sudo, K. and Nakauchi, H. Asymmetry division and lineage
commitment at the level of hematopoietic stem cells: inference from differentiation in
daughter cell and granddaughter cell pairs. J. Exp. Med. 192. 1281- 1288 (2004).
12.
Knobich, J. A. Asymmtric cell division during animal development. Nat. Rev. 2. 11-
20 (2001).
13.
Aguilaniu, H., Gustafsson, L., Rigoulet, M. and Nyström, T. Asymmetric inheritance
of oxidative damaged proteins during cytokinesis. Science. 299. 1751-1753 (2003).
14.
Barker, M. G. and Walmsley, R. M. Replicative ageing in the fission yeast
Schizosaccharomyces pombe. Yeast. 15. 1511-1518 (1999).
15.
Nyström T. A bacterial kind of aging. PLoS Gen. 12(3) e224. 2355-2357 (2007).
16.
Cheng, Z., Odstrcil, E.A., Tu, B.P. and MacKnight, S. L. Restriction of DNA
replication to the reductive phase of the metabolic cycle protects genome integrity.
Science 316. 1916-1919 (2007).
17.
Fabrizio, P. et al. Sir2 blocks extreme life-span extention. Cell. 123(4). 655-667
(2005).
18.
Kaeberlein, M., Burtner, C. R. and Kennedy, B. K. Recent developments in yeast
aging. Plos. Gen. 3(5) e84. 655-660 (2007)
19.
Kaeberlein, M. and Powers, R. W. Sir2 and calorie restriction in yeast: A skeptical
perspective. Age. Res. Rev. 6. 128- 140 (2007).
20.
Libert, S., et al. Regulation of Drosophila life span by olfaction and food derived
odours. Science 315. 1133-1137 (2007).
12
Footnotes to figures:
Figure 1: The effect of selection in high and low caloric environment on the growth rates
(hour
-1)
and growth yields (mg protein per mg glucose consumed). Strains were selected in
glucose or peptone media but determination of growth rates and growth yields was made in
glucose alone since growth yield calculation in complex substrate cannot be precise. Strains L
showed significantly higher yields as opposed to the strains H (paired t test, n=8, t= 7.2 p,
two tailed = 0.000177). There was no significant difference in growth rates (paired t test, n=8,
t=1.6979, p, two tailed= 0.1333). The correlation between growth rate and growth yield was
negative as expected but not significant (r = -0.3992, p > 0.05).
Figure legends:
Black triangles: The wild type environmental isolate
Blue Squares: Strains L after 500 generations.
Blue Circles: Strains L after 1000 generations.
Orange Squares: Strains H after 500 generations.
Orange Circles: Strains H after 1000 generations.
Figure 2: The median index of cell division asymmetry as influenced by current and
selection substrate concentration. (Colour codes and shapes same as in figure 1). Asterisks on
the right indicate significance for differences due to current substrate concentration for a
given strain in pair-wise comparisons using Mann-Whitney test, those above or below the
points indicate significance of difference due to selection on a given current concentration.
(see supplementary material for detailed statistics) a) Selection and growth on glucose
medium: The current glucose concentration affected symmetry in the wild type and in the 500
generation selected strain but after 1000 generations of selection the cells appeared to be
committed to symmetric or asymmetric growth and did not respond to current concentrations.
b) Selection and growth on complex medium: Trends on complex medium were similar to
13
that of glucose but there was less of asymmetry overall and the effect of current concentration
was not significant.
0.6083, p<0.05) (Colour codes as in Figure 1)
Figure 4: Cell division asymmetry and growth yield. (r = -0.6266, p < 0.05) For all the
experimentally evolved strains substrate concentrations during selection and observation
were the same. (Colour codes as in Figure 1)
Figure 1
Figure 2
a
b
0.01
1
Substrate concentration mg/dl
Figure 3
Figure 4