Metabolic cycling without cell division cycling in
Nikolai Slavova,b,1, Joanna Macinskasb, Amy Caudyc, and David Botsteinb,1
aDepartments of Biology and Physics, Massachusetts Institute of Technology, Cambridge, MA 02139;bDepartment of Molecular Biology, Princeton University,
Princeton, NJ 08540; andcDonnelly Centre for Cellular and Biomolecular Research, University of Toronto, Toronto, ON, Canada M5S 3E1
Contributed by David Botstein, October 17, 2011 (sent for review September 12, 2011)
Despite rapid progress in characterizing the yeast metabolic cycle,
its connection to the cell division cycle (CDC) has remained unclear.
We discovered that a prototrophic batch culture of budding yeast,
growing in a phosphate-limited ethanol medium, synchronizes
spontaneously and goes through multiple metabolic cycles, whereas
the fraction of cells in the G1/G0 phase of the CDC increases
monotonically from 90 to 99%. This demonstrates that metabolic
cycling does not require cell division cycling and that metabolic
synchrony does not require carbon-source limitation. More than
3,000 genes, including most genes annotated to the CDC, were
expressed periodically in our batch culture, albeit a mere 10% of the
cells divided asynchronously; only a smaller subset of CDC genes
correlated with cell division. These results suggest that the yeast
metabolic cycle reflects a growth cycle during G1/G0 and explains
our previous puzzling observation that genes annotated to the CDC
increase in expression at slow growth.
vision cycle (CDC), consists of four phases (G0/G1, S, G2, and
M) that are easily distinguished by morphological criteria. When
the growth rate of budding yeast is slowed by mutations or
chemicals inhibiting growth, the duration of the G1/G0 phase
increases relative to the durations of the S, G2, and M phases
(1). Recently, we confirmed and quantified this CDC trend (Fig.
1A) in chemostat cultures whose steady-state growth rate was
controlled by limiting natural nutrients (2, 3). The second kind of
cycle, the yeast metabolic cycle (YMC), was first observed more
than four decades ago (4) as periodic oscillations in the oxygen
consumption of continuous, glucose-limited cultures growing in a
chemostat. Like the CDC, the YMC can be divided phenome-
nologically into two phases: the low oxygen consumption phase
(LOC), when the amount of oxygen in the medium is high be-
cause the cells consume little oxygen, and the high oxygen con-
sumption phase (HOC), when the reverse holds (SI Appendix).
We also reported previously (3) similar growth-rate changes in
the relative durations of the phases of the YMC (Fig. 1B). As the
growth rate increases, the relative duration of the LOC decreases
whereas the relative duration of the HOC increases (Fig. 1B),
similarly to the analogous changes in the CDC.
These changes in the relative durations of the phases affect the
composition of asynchronous cultures, because single cells from
asynchronous cultures cycle metabolically (5, 6) and thus the
fraction of cells in a particular phase is proportional to the du-
ration of that phase relative to the entire cycle period. The in-
crease in the relative duration of a phase results in the increase
in the fraction of cells in that phase, and thus an increase in the
population-average expression levels of genes peaking during
that phase. Consider, for example, a ribosomal gene that peaks
in expression during the HOC phase; at slow growth rate, most
cells are in the LOC, expressing the ribosomal gene at low levels
(Fig. 1C) and thus resulting in low population-average levels, and
vice versa at high growth rate. We will refer to this dependence
between the composition of an asynchronous culture and its
wo kinds of periodic behavior have been characterized in
slowly growing yeast cultures. The first, the classical cell di-
population-average gene expression as “ensemble average over
phases” (EAP). In what follows, distinguishing between the
population behavior and the behavior of individual cells is critical,
because we present evidence that the population behavior rep-
resents at least two physiologically nonidentical subpopulations.
Metabolically synchronized populations, manifested by oscil-
lations in oxygen consumption, have been observed in batch
yeast cultures grown on trehalose media (7), as well as in yeast
cultures starved for a carbon source (glucose or ethanol) and
subsequently fed continuously with either glucose- or ethanol-
limited media (8). DNA microarray studies of continuous glu-
cose-limited cultures showed that very many genes are expressed
periodically during the YMC (9, 10). Biosynthetic genes peak
during the HOC phase, whereas autophagy and vacuolar genes
peak during the LOC. In a continuous YMC culture, a fraction of
the culture divides synchronously during each YMC period (4, 10,
11), indicating a coupling between the YMC and the CDC. How-
ever, the mechanism of this coupling remains unclear. Previous
work has shown that during each YMC period, the YMC culture
has at least two distinct subpopulations: dividing and nondividing.
It had previously been suggested that DNA replication is restricted
12, 13), but we found that, on a population basis, DNA replication
can coincide with the HOC phase (3). These results left open the
question of the link between the CDC and the YMC.
In this article, we describe growth conditions in which non-
dividing cells in the G1 phase of the CDC go through multiple
metabolic cycles. We also find that many genes whose expression
previously has been associated with the CDC in rich media (14)
oscillate with the YMC in the absence of cell division, whereas
others do not. These results suggest that the YMC is occurring
during the G1 phase of slowly growing cells, and that at least
some of it may be relevant to cells growing more rapidly. Fur-
thermore, the oscillation of genes annotated to the CDC even in
nondividing cells resolved the puzzling observation that many of
these genes increase in expression as the growth rate decreases (2,
3), although the fraction of dividing cells decreases. The discovery
of the G1-phase growth cycle and the associated periodic ex-
pression of genes previously annotated to the CDC in the absence
of cell division can account for the increased expression of such
genes at slow growth rates, and enabled us to characterize each of
the genes expressed periodically with each cycle.
Experiment and Physiology. We discovered that wild-type diploid
prototroph (DBY12007, S288c background) synchronizes spon-
Author contributions: N.S., A.C., and D.B. designed research; N.S., J.M., and A.C. per-
formed research; N.S. contributed new reagents/analytic tools; N.S., A.C., and D.B. ana-
lyzed data; and N.S., A.C., and D.B. wrote the paper.
The authors declare no conflict of interest.
Freely available online through the PNAS open access option.
1To whom correspondence may be addressed. E-mail: firstname.lastname@example.org or botstein@
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
| November 22, 2011
| vol. 108
| no. 47 www.pnas.org/cgi/doi/10.1073/pnas.1116998108
taneously when grown in batch in phosphate-limited medium
contacting 100 mM ethanol as the only source of carbon and
energy. We measured continuously the dissolved oxygen in the
medium (Fig. 2A), which tracks oxygen consumption very closely,
as can be verified by adding a pulse of a carbon source, which
causes the measured level of dissolved oxygen to fall immediately
(15, 16). After inoculation, the dissolved oxygen decreased ex-
ponentially (Fig. 2A), indicating exponential growth and increase
in the oxygen consumption of the culture. Then the dissolved
oxygen in the culture media returned to the level before inocu-
lation (100%), indicating sudden and complete cessation of
respiration. Subsequently, the oxygen level in the media and thus
the consumption of the culture oscillated for many hours, just as
seen in continuous cultures (4, 10). We reproduced this syn-
chronization in several independent cultures (SI Appendix, Fig.
S1). At the indicated time (red arrow in Fig. 2A), 20% of the
culture was diluted with fresh medium, resulting in substantial
decrease in cell density and therefore low oxygen consumption.
After a growth period, the culture resynchronized spontaneously
and was sampled for DNA content, gene expression, cell size,
and cell density at 67 time points (marked with red circles in
the YMC (3). (C) Distributions of the number of mRNAs per cell of a ribosomal gene in an asynchronous culture growing at three growth rates.
Growth-rate (GR) changes in the yeast metabolic cycle and the cell division cycle. Changes in the relative duration of the phases of (A) the CDC and (B)
consumption. The culture was sampled at the positions indicated by red circles. (B) Cell density at each of the sampled points. (C) Distribution of DNA content
obtained by FACS counting 150,000 cells labeled with SYTOX Green. (D) Fraction of cells in the CDC phases as inferred from the DNA content.
Oxygen, biomass, and DNA content data in a metabolically cycling nondividing culture. (A) Dissolved oxygen in the medium reflecting the oxygen
Slavov et al. PNAS
| November 22, 2011
| vol. 108
| no. 47
Fig. 2 C and D shows that despite the periodic oxygen con-
sumption, only about 10% of the cells divided initially, after
which the DNA content remained constant at the level expected
for diploid cells with unreplicated DNA as measured by FACS
(Fig. 2C). Thus, as shown in Fig. 2D, most of the culture was in
the G0/G1 phase of the CDC for the entire interval that we
sampled, during which there were three metabolic cycles and
a slight increase in the distribution of cell sizes (SI Appendix, Fig.
S4). During the last sampled metabolic cycle, more than 99% of
the culture was in G1/G0, indicating that the few cells with
replicated DNA (G2/M phase) had divided (Fig. 2D). These
results are fully consistent with the small increase in cell density
during cycling (Fig. 2B). The dividing subpopulation S/G2/M is
unlikely to cause periodic oxygen consumption because its frac-
tion decreases monotonically 10-fold so that by hour 85 it is
nearly zero. However, the oxygen consumption continues to os-
cillate robustly. Indeed, we followed the culture and observed
that it continued consuming oxygen periodically for 12 h after the
last sample (Fig. 2A), long after the S/G2/M fractions had fallen
below 1%, and without a significant increase in cell density.
Global View of Gene Expression. To display global patterns in gene
expression, we hierarchically clustered the gene-expression data
from the culture characterized in Fig. 2 and from a metabolically
synchronized, glucose-limited, continuous culture (10). In our
batch culture, genes that peak in the LOC phase generally have
higher mean expression levels relative to the reference, a
glucose-limited, asynchronous culture growing exponentially at
growth rate μ = 0.25 h−1. The reverse holds for genes peaking in
the HOC phase. A possible reason for this observation is that
a fraction of the batch culture is already fully quiescent, that is,
arrested in an LOC-like state (resulting in the high levels of LOC
genes), and only the remaining fraction of the culture cycled
metabolically, resulting in the periodic gene expression and
consumption of oxygen. The hierarchically clustered mRNA data
show that many periodic genes are in-phase, that is, they peak in
the same phase both in the batch and continuous cultures (Fig.
3A). To characterize such genes, we ordered them by phase using
correlation analysis (Materials and Methods) (Fig. 3B). These
3,000+ genes include the genes with carbon source- and limita-
tion-independent growth-rate response that can be explained by
the EAP mechanism introduced above applied to the growth-
rate (GR) changes in the YMC (Fig. 1B) (3), and have similar
amplitudes of oscillation in the two cultures (SI Appendix, Fig. S5).
Cell Division Cycle. To define better the relationship of the YMC
and the CDC, we explored the expression of genes annotated
to the CDC with two questions in mind: (i) Which genes are
expressed with the division cycle, and thus decrease mono-
tonically in our batch culture, and which genes are expressed
with the growth cycle, and thus oscillate in our batch culture?
(ii) Can we connect the periodic expression of genes with either
the division or the growth cycle to the growth-rate changes in
the expression of these genes in asynchronous cultures? To
YMC cultures. (Left) The black bars at the top display dissolved oxygen and correspond to the expression levels in our batch culture (Fig. 2) relative to
a glucose-limited culture, μ = 0.25 h−1. (Center) The blue bars correspond to the same data but centered to a zero mean for each gene. (Right) The magenta
bars correspond to mean centered data from a continuous glucose-limited culture (10). The clustering is based on noncentered Pearson correlation computed
from all data shown in the panel. The dissolved oxygen in the media of the two cultures is indicated by bars at the top (SI Appendix). (B) The expression levels
of about 90% of the periodic genes (more than 3,000) have the same phase in the batch culture (Fig. 2) and in the continuous culture (10). These genes are
ordered by phase of peak expression (Materials and Methods and SI Appendix).
Global pattern of periodic gene expression in metabolically cycling cultures. (A) Hierarchically clustered, log2-transformed, gene-expression data from
| www.pnas.org/cgi/doi/10.1073/pnas.1116998108Slavov et al.
answer these questions, all 468 genes annotated to the CDC in
yeast (biological process “cell cycle”; Gene Ontology accession
no. GO:0007049; for brevity, we will refer to these genes as
“CDC genes” throughout the article) were plotted, portioned
into three clusters based on their expression profiles in our
batch culture (Fig. 4).
The first cluster (Fig. 4, CDC genes, Top) is composed of
genes whose expression steadily decreases, with minimal evi-
dence of cycling. The cluster includes genes essential for CDC
progression, such as cyclins (CLN1, CLB1, CLB2), CDC20,
which is a component of the anaphase-promoting complex/cyclo-
some, CDC12, which is required for cytokinesis, and other es-
sential classical CDC genes such as MCD1 and CDC5. The genes
in this cluster tend to have positive growth-rate slopes [as defined
previously by Brauer et al. (2)], with a mean of +0.81. This is
entirely consistent with the idea that, at higher growth rates, the
increase in the fraction of cells that are dividing and thus
expressing those genes results in higher population-average ex-
pression levels of these genes, that is, the EAP mechanism in-
troduced above applied to the growth-rate (GR) changes in the
CDC (Fig. 1A).
The second and largest cluster of genes annotated to the CDC
(Fig. 4, CDC genes, Middle) is expressed periodically with the
metabolic cycle. Because there are so few dividing cells (less than
genes in this second cluster were expressed in the subpopulation
of cells that cycled metabolically (SI Appendix). This conclusion is
fortified by the observation that the GR response of these genes
is predicted by the EAP mechanism applied to the GR changes in
the YMC, not the CDC (Fig. 1B). The genes that peak during the
HOC tend to have positive GR slopes (mean of +0.79), whereas
the genes peaking during the LOC tend to have negative GR
slopes(meanof−0.76);the difference betweenthedistributionsof
GR slopes is very significant (P < 10−21, rank-sum test).
The third and smallest cluster (Fig. 4, CDC genes, Bottom)
consists of genes that increase in expression during the course of
the oscillations. Consistent with the results from Fig. 2, this
cluster includes genes repressing (negatively regulating) the CDC
and resulting in CDC arrest. These genes have mostly negative
GR slopes that can be explained by the EAP mechanism applied
to the GR changes in the CDC, not the YMC (Fig. 1A). Thus, we
conclude that the genes from the first and third clusters are
expressed by the cells that divided, whereas the genes from the
second cluster are expressed by the cells that cycled metabolically.
Coupling of the Metabolic and Cell Division Cycles. Our goal was to
understand better the relationship between the CDC and the
YMC. Our gene-expression data from a metabolically cycling,
largely nondividing culture of budding yeast indicate that a very
large fraction of CDC genes cycles even when the majority of the
culture is not dividing, whereas a smaller subset of CDC genes
is expressed only in dividing cells. These periodic genes likely
comprise what has been described as a free-running oscillator
driving G1 events (17) and are consistent with previous reports
of periodic expression of CDC genes in the absence of mitotic
cyclins (18). Furthermore, the phase of the metabolic cycle in
which the cycling CDC genes are expressed correlates strongly
with the GR slope (change in gene expression with growth rate)
in asynchronous cultures (2, 3), fortifying the relevance of the
autonomous YMC to single cells from asynchronous cultures.
As we stressed before, the GR response of CDC genes in
asynchronous cultures could not be accounted for simply by the
change in the CDC with growth rate. Taking into account the
to the reference (glucose limitation at μ = 0.25 h−1). (Center) The same data centered to mean zero. (Right) The growth-rate (GR) slopes of the corresponding
genes in asynchronous cultures (2). We previously computed GR slopes by regressing transcript levels of asynchronous cultures against the growth rate (2, 3);
positive GR slopes indicate increasing expression with growth rate, and negative GR slopes indicate decreasing expression.
Genes annotated to the CDC are periodically expressed in a nondividing culture in correlation with the metabolic cycle. (Left) Expression levels relative
Slavov et al.PNAS
| November 22, 2011
| vol. 108
| no. 47
metabolic cycling of CDC genes, however, we can largely explain
their GR responses by extending our model of changes in the
relative durations of the HOC and LOC phases of the YMC with
growth rate (3). A possible explanation for the periodic expres-
sion of CDC genes in nondividing cells is that genes thought to
be involved in the CDC are tied to metabolism and growth.
Importantly, about half of the periodically expressed CDC genes,
such as histone genes, are expressed at very low levels relative to
the reference (even during their peak expression), indicating that
nondividing cells may not be affected strongly by the periodic
expression, especially if the mRNAs are not translated. The
other half of the CDC genes expressed periodically in our cul-
ture, however, are expressed at high levels.
The smaller subset of CDC genes whose expression decreases
during the course of the oscillations is likely to be the genes
ultimately regulating CDC progression. These genes are likely
expressed only in diving cells (S/G2/M phases), and have positive
GR responses (increasing in expression with growth rate in
asynchronous cultures) that can be explained only by the growth-
rate changes in the CDC without considering the YMC: As
growth rate increases, the fraction of budded cells also increases,
and thus genes expressed only in budded cells increase in
population-average expression. This mechanism can account for
both the expression levels in our batch culture and for the GR
response in asynchronous cultures (2, 3).
The link between the YMC and CDC can be conceptualized
by thinking of two related but distinct cycles: the cycle of cell
growth (i.e., biomass increase) and the cycle of cell division. The
link, then, is an expression of the requirement to couple cell
growth and cell division so that the cell size can be maintained
in a physiologically functional range. At fast growth in rich me-
dia, growth and cell division are always coupled in the same way
and have been characterized as the classical CDC. In contrast,
our studies (2, 3) in a variety of growth conditions and growth
rates have enabled us to observe both the cycle of growth and the
cycle of division and to begin characterizing the link between the
two cycles across growth rates.
One attractive way to think about the relationship of the CDC
and the YMC is to recognize that at slow growth rates, when
exponentially growing cells spend much of their time in G0/G1 (2,
3), the barrier of entry to the CDC likely involves some measures
of cell mass or size and nutritional status in addition to DNA
integrity. Cells that are too small or unable to pass the G1/S
checkpoint for other reasons remain in G0/G1, where they con-
tinue to cycle metabolically and grow. What our work emphasizes
is that this metabolic dependence of the CDC appears to require
the oscillatory behavior of the YMC. Once one accepts this, then
much of the literature on growth rate and metabolic oscillation is
reconcilable quantitatively. Previous work has shown that the
YMCfunctions inindividualcellsfrom unsynchronized(5)aswell
as synchronized populations. Thus, we emerge with a picture of
the YMC and CDC in continuous cultures as cartooned in Fig. 5.
By finding conditions in which the YMC oscillates robustly
without cell division, we have been able to parse out two kinds of
oscillating genes: those that oscillate with the YMC and those
that oscillate only with the CDC. Every slowly growing culture of
yeast should then be seen as a mixed population with respect to
the phases of these cycles.
Why Cycle? The suggestion that the biological role of the YMC is
to restrict DNA replication to the reductive phase of the YMC
(9, 12, 13) is not supported by our previous observation that
DNA replication can coincide with the peak of oxygen con-
sumption in a wild-type strain (3). Here we report again a be-
havior that cannot be accounted for by the suggestion that the
function of the YMC is to protect DNA replication; we found
robust periodic oscillations in oxygen consumption and gene
expression, although less than 1% of cells replicated DNA
without measurable periodicity. Both our data from continuous
YMC cultures (3) and from this paper suggest that the change in
oxygen consumption seen at the level of the culture does not
directly reflect the change in oxygen consumption of the sub-
population of cells committed to division (S/M/G2 phases). Our
data do not indicate directly the rate of respiration in the di-
viding subpopulation in continuous cultures; rather, the data
show that the measured changes in oxygen consumption are due
to the G1 subpopulation (Fig. 5). If a requirement for a reductive
environment for DNA replication is not the primary driver of the
YMC, what is it? A possible explanation could be that at slow
growth, when most genes are expressed at very low average levels
(SI Appendix, Fig. S6), on the order of one mRNA per cell (5),
concentrating the expression of biosynthetic genes and active
metabolism into short pulses provides a more reliable mode
of regulation compared with constant low expression levels
throughout the CDC period. In fact, it is rather unlikely for a cell
to make all subunits of a macromolecular complex simulta-
neously if the probability of making each subunit is low and
uniformly distributed throughout the CDC period. Considering
that the biosynthesis of the subunits also needs energy (ATP)
and precursors (such as amino acids), one can easily see why the
coordination of many biosynthetic reactions can be beneficial
in ensuring timely synthesis and assembly of macromolecular
complexes. Another possible and not mutually exclusive role for
the YMC could be the periodic degradation of mistranslated and
damaged proteins, because autophagy and vacuolar genes peak
during the LOC, right after the biosynthesis phase (HOC). The
accumulation of reserved carbohydrates during the LOC, which
was reported with the discovery of the YMC (4) and confirmed
recently (19), likely imposes a fundamental limit on the duration
of the LOC required for accumulating enough trehalose in glu-
cose-limited conditions, and on the duration of the HOC given
the pool size of reserved carbohydrates. The metabolic syn-
chrony at the level of the culture likely requires cell–cell com-
munication, which we do not yet understand.
Regulation and Predictions. The majority of the genes cycling in-
phase both in our batch culture and in the continuous glucose
culture have universal growth-rate response (3) and are regu-
lated by the target of rapamycin (TOR) and protein kinase A
(20). Therefore, we expect periodic changes in the activity of at
least one of these master regulators. This expectation is bol-
stered by the report that TOR regulates the acetylation of his-
tones of ribosomal proteins (21) and by the recent finding that
for many HOC genes (including the ribosomal proteins), the
time of peak expression coincides with the acetylation of their
model for the composition of a metabolically synchronized culture consist-
ing of three synchronized subpopulations depicted by a representative cell.
Note that we do not have direct data for the oxygen consumption during
the S/G2/M phases and that not all cells need to have the same number of
metabolic cycles before entering S phase.
Composition of a continuous metabolically synchronized culture. A
| www.pnas.org/cgi/doi/10.1073/pnas.1116998108Slavov et al.
histones (16). Similarly, most targets of MSN2/4 transcription Download full-text
factors are expressed periodically and peak in the LOC. Thus, we
expect that MSN2/4 will be localized in the nucleus during the
LOC and in the cytoplasm during the HOC. This prediction can
be tested directly in real time in live cells growing in flow cells. In
fact, the finding of fluctuating brief localization of MSN2/4 to the
nucleus may reflect remaining elements of metabolic cycling even
in cells growing by aerobic glycolysis in nutrient-rich media (22).
Metabolic Cycling, Nutrient Limitation, and Slow Growth. Our ob-
servation of metabolic cycling in a phosphate-limited batch cul-
ture fortifies our suggestion (3, 5) that carbon-source limitation
is not required for metabolic cycling. We presented evidence
(based on gene–gene correlations) for metabolic cycling in single
cells limited on phosphate (5) and found signs (based on the GR
response) for metabolic cycling in cultures limited on ammo-
nium, ethanol, phosphate, glucose, and sulfur (2, 3). Now we
report an example of metabolic cycling observed directly in time
in non-carbon source-limited cultures. The evidence for the
YMC in cultures whose growth is limited on a variety of natural
nutrients strongly suggests that the cycling is likely related to the
slow growth rather than the nature of the nutrient limitation. A
related inference based on our data is that metabolic cycling
does not require continuous influx of fresh media and fluctua-
tions in nutrient concentrations that may arise in continuous
cultures, such as oscillation in the glucose concentration in the
media of glucose-limited YMC cultures. In our experimental
system of a culture cycling in batch, all essential nutrients are
likely to decrease as the cells cycle metabolically but not to in-
crease, because the only source of nutrients could be the lysing of
starved cells, which is very unlikely to occur at a significant rate
a few hours after the exponential growth of a prototrophic cul-
ture (23, 24).
Materials and Methods
Culture Conditions and Gene Expression. The culture was grown in 500-mL
fermenter vessels(Sixfors;Infors)containing500mL ofculture volume, stirred
at 400 rpm, and aerated with humidified and filtered air. We used the
phosphate-limited medium that we described previously (3). Gene expression
was measured with 8 × 15k Agilent microarrays as described previously (3).
See http://genomics-pubs.princeton.edu/YMC_in_Ethanol_Batch to down-
load and explore the data interactively.
Ordering Genes by Phase of Peak Expression. Among the many possible
approaches for identifying the phase of peak expression of each gene, we
chose correlation analysis with the following algorithmic steps:
• For each experiment (three cycles), the data for oxygen consumption
and for the ith gene were interpolated at 150 equally spaced time
points resulting in two vectors, o and gi, for oxygen and gene expres-
sion, respectively: o ∈ ℝ150and gi∈ ℝ150.
• A vector of correlations (r ∈ ℝ40) was computed by sliding the two
interpolated vectors (o and gi) relative to each other, one element at
a time for the first 40 elements.
• The index (position) of the largest element of r was selected to repre-
sent the phase (ϕ) of peak expression of the ith gene: ϕ = argmaxjr(j),
j = 1...40.
• The genes were ordered by sorting the phases ϕ of all genes for which
the maximum correlation between o and giexceeded a threshold.
See SI Appendix for more details.
ACKNOWLEDGMENTS. We thank Benjamin Tu, Steven McKnight, Sanford
Silverman, Alexander van Oudenaarden, Max Staller, Sergey Kryazhimskiy,
and Juan Alvarez for feedback on the manuscript. Research was funded by
grants from the National Institutes of Health (GM046406) and the National
Institute of General Medical Sciences (GM071508).
1. Hartwell LH, Unger MW (1977) Unequal division in Saccharomyces cerevisiae and its
implications for the control of cell division. J Cell Biol 75:422–435.
2. Brauer MJ, et al. (2008) Coordination of growth rate, cell cycle, stress response, and
metabolic activity in yeast. Mol Biol Cell 19:352–367.
3. Slavov N, Botstein D (2011) Coupling among growth rate response, metabolic cycle,
and cell division cycle in yeast. Mol Biol Cell 22:1997–2009.
4. Küenzi MT, Fiechter A (1969) Changes in carbohydrate composition and trehalase-
activity during the budding cycle of Saccharomyces cerevisiae. Arch Mikrobiol 64:
5. Silverman SJ, et al. (2010) Metabolic cycling in single yeast cells from unsynchronized
steady-state populations limited on glucose or phosphate. Proc Natl Acad Sci USA 107:
6. Wyart M, Botstein D, Wingreen NS (2010) Evaluating gene expression dynamics using
pairwise RNA FISH data. PLoS Comput Biol 6:e1000979.
7. Jules M, François J, Parrou JL (2005) Autonomous oscillations in Saccharomyces cer-
evisiae during batch cultures on trehalose. FEBS J 272:1490–1500.
8. Keulers M, Suzuki T, Satroutdinov AD, Kuriyama H (1996) Autonomous metabolic
oscillation in continuous culture of Saccharomyces cerevisiae grown on ethanol. FEMS
Microbiol Lett 142:253–258.
9. Klevecz RR, Bolen J, Forrest G, Murray DB (2004) A genomewide oscillation in tran-
scription gates DNA replication and cell cycle. Proc Natl Acad Sci USA 101:1200–1205.
10. Tu BP, Kudlicki A, Rowicka M, McKnight SL (2005) Logic of the yeast metabolic cycle:
Temporal compartmentalization of cellular processes. Science 310:1152–1158.
11. Robertson JB, Stowers CC, Boczko E, Johnson CH (2008) Real-time luminescence
monitoring of cell-cycle and respiratory oscillations in yeast. Proc Natl Acad Sci USA
12. Chen Z, Odstrcil EA, Tu BP, McKnight SL (2007) Restriction of DNA replication to the
reductive phase of themetabolic cycle protects genome integrity. Science 316:1916–1919.
13. Chen Z, McKnight SL (2007) A conserved DNA damage response pathway responsible for
coupling the cell division cycle to the circadian and metabolic cycles. Cell Cycle 6:2906–2912.
14. Spellman PT, et al. (1998) Comprehensive identification of cell cycle-regulated genes
of the yeast Saccharomyces cerevisiae by microarray hybridization. Mol Biol Cell 9:
15. Ronen M, Botstein D (2006) Transcriptional response of steady-state yeast cultures to
transient perturbations in carbon source. Proc Natl Acad Sci USA 103:389–394.
16. Cai L, Sutter BM, Li B, Tu BP (2011) Acetyl-CoA induces cell growth and proliferation
by promoting the acetylation of histones at growth genes. Mol Cell 42:426–437.
17. Haase SB, Reed SI (1999) Evidence that a free-running oscillator drives G1 events in
the budding yeast cell cycle. Nature 401:394–397.
18. Orlando DA, et al. (2008) Global control of cell-cycle transcription by coupled CDK and
network oscillators. Nature 453:944–947.
19. Shi L, Sutter BM, Ye X, Tu BP (2010) Trehalose is a key determinant of the quiescent
metabolic state that fuels cell cycle progression upon return to growth. Mol Biol Cell
20. Zaman S, Lippman SI, Schneper L, Slonim N, Broach JR (2009) Glucose regulates
transcription in yeast through a network of signaling pathways. Mol Syst Biol
21. Rohde JR, Cardenas ME (2003) The Tor pathway regulates gene expression by linking
nutrient sensing to histone acetylation. Mol Cell Biol 23:629–635.
22. Cai L, Dalal CK, Elowitz MB (2008) Frequency-modulated nuclear localization bursts
coordinate gene regulation. Nature 455:485–490.
23. Boer VM, Amini S, Botstein D (2008) Influence of genotype and nutrition on survival
and metabolism of starving yeast. Proc Natl Acad Sci USA 105:6930–6935.
24. Klosinska MM, Crutchfield CA, Bradley PH, Rabinowitz JD, Broach JR (2011) Yeast cells
can access distinct quiescent states. Genes Dev 25:336–349.
Slavov et al. PNAS
| November 22, 2011
| vol. 108
| no. 47