Article
the cell divided, or the field of view became obstructed by adjacent
dividing cells, or the cell became dislodged from the glass surface and
we lost track of it. Cells that were followed for 43 or more frames were
considered for analysis. This threshold was chosen on the basis of a
systematic analysis of different values for this threshold. We wanted
to establish an upper bound on the number of nongrowing cells after
the shift to acetate. We did not expect nongrowing cells to be over-
represented in transiently present cells that briefly settled on the glass
bottom and were then washed away. These transiently present cells
become more important for low values of the threshold. On the other
hand, for high values of the threshold we were artificially enriching for
non-growing cells. The intermediate value that we chose established
the most stringent upper bound for the fraction of nongrowing cells
in the population.
We segmented 1,761 cells, after which we set an arbitrary threshold
of a 10% increase in single-cell area to identify cells that showed signifi-
cant growth. In Extended Data Fig. 4, cell traces that crossed this 10%
threshold are marked in blue, and cells that did not are marked in red.
Out of 1,761 segmented single-cell traces, 1,500 (roughly 85.17%) crossed
the chosen 10% threshold, and only 261 (around 14.83%) showed less
than a 10% increase in area over the experiment (Extended Data Fig. 4).
Therefore, using our method we have detected 14.8% non-growing cells.
This number sets an upper bound to the fraction of the non-growing
cell population. It is likely that many of these cells would have showed
substantial growth at later time points, which we were unable to meas-
ure owing to experimental limitations. This suggests that the actual
population of cells that do not resume growth is in reality much smaller
than the roughly 14.8% that we have measured.
Metabolite mass spectrometry
Sample collection and quenching. For metabolite measurements
and^13 C-labelling experiments, we transferred an amount proportional
to 1 ml*OD 600 of the culture broth onto a Durapore filter with a pore size
of 0.45 μm (Millipore) and vacuum-filtered the sample. For metabolite
measurements, the filter with cells was immediately transferred after
filtration into 4 ml of 20 °C acetonitrile/methanol/water (2/2/1) to quench
metabolism and 200 μl of a uniformly^13 C-labelled E. coli metabolite
extract were added as an internal standard^45.^13 C-labelling experiments
were performed immediately after vacuum-filtration on the filter, as
described^46. Specifically, cells on the filter were first washed with fresh,
preheated (37 °C) acetate M9 medium for 10 s, and^13 C-labelling was initi-
ated by changing the washing solution to preheated (37 °C) M9 medium
containing uniformly^13 C-labelled acetate. After each labelling step, the
filter was transferred into 4 ml of 20 °C acetonitrile/methanol/water
(2/2/1) for quenching. To extract metabolic intermediates, the filter was
kept in this solution at −20 °C for 1 h. Then the cell debris was removed
from the extracts by centrifugation (4 °C, 10,000 rpm, 10 min); the super-
natants were transferred into new tubes and dried to complete dryness.
Sample preparation. For liquid chromatography/mass spectrometry
(LC/MS) analysis, dried extracts were resuspended in 100 μl deion-
ized water, of which 10 μl were injected into a Waters Acquity ultraper-
formance liquid chromatography (UPLC) system (Waters) with a Waters
Acquity T3 column coupled to a Thermo TSQ Quantum Ultra triple
quadrupole instrument (Thermo Fisher Scientific) with negative-mode
electrospray ionization. Compound separation was achieved using a
gradient of two mobile phases: A, 10 mM tributylamine, 15 mM ace-
tic acid and 5% (v/v) methanol; and B, 2-propanol^47. Mass isotopomer
distributions of carbon backbones was acquired as described^48. We
carried out peak integration using in-house software (B. Begemann
and N. Zamboni, personal communication).
Kinetic flux estimation. Flux estimation closely followed ref.^49 , based
on the kinetics of incorporation of a^13 C-acetate isotope. At numer-
ous time points, after cells were rapidly switched from unlabelled to
isotope-labelled acetate, LC/MS analysis was performed. Resulting
plots of unlabelled compound versus time were fitted by an exponential
decay, and the flux was calculated as the decay rate multiplied by the
intracellular metabolite concentration.Proteomic mass spectrometry
Metabolic labelling with^15 N (ref.^50 ) provides relative quantitation of
unlabelled proteins with respect to labelled proteins across growth
conditions of interest. Each experimental sample in a series is mixed
in an equal amount with a known labelled standard sample as refer-
ence, and the relative change of protein expression in the experimental
sample is obtained for each protein.Sample collection. For each culture, 1.8 ml of cell culture at OD 600 =
0.4–0.5 was collected by centrifugation. The cell pellet was
resuspended in 0.2 ml of water and fast frozen on dry ice.Sample preparation. A balanced mixture of the two^15 N-labelled cell
samples (from glycolytic and gluconeogenic growth conditions, with
cells grown on glucose and acetate respectively) was prepared as a uni-
versal reference. We added 100 μg of the labelled reference proteome
to 100 μg of each experimental sample. This balanced preparation
(equal amounts of total protein) enabled the measurement of proteome
mass fraction for each protein. The mixed reference ensured that the
distribution of proteins in the reference was not strongly biased by a
particular growth condition.
Proteins were precipitated by adding 100% (w/v) trichloroacetic
acid (TCA) to a final concentration of 25%. Samples were left to stand
on ice for a minimum of 1 h. The protein precipitates were spun down
by centrifugation at 13,200g for 15 min at 4 °C. The supernatant was
removed and the pellets were washed with cold acetone and dried in
a Speed-Vac concentrator.
The pellets were dissolved in 80 μl 100 mM NH 4 HCO 3 with 5% ace-
tonitrile (ACN). We added 8 μl of 50 mM dithiothreitol (DTT) to reduce
the disulfide bonds before the samples were incubated at 65 °C for
10 min. Cysteine residues were modified by adding 8 μl of 100 mM
iodoacetamide (IAA) followed by incubation at 30 °C for 30 min in the
dark. Proteolytic digestion was carried out by adding 8 μl of 0.1 μg μl−1
trypsin (Sigma-Aldrich) with incubation overnight at 37 °C. The peptide
solutions were cleaned by using PepClean C-18 spin columns (Pierce,
Rockford, IL). After drying in a Speed-Vac concentrator, the peptides
were dissolved into 10 μl sample buffer (5% ACN and 0.1% formic acid).Mass spectrometry. The peptide samples were analysed on an AB
SCIEX TripleTOF 5600 system (AB Sciex) coupled to an Eksigent Na-
noLC Ultra system (Eksigent). The samples (2 μl) were injected using
an autosampler. The samples were first loaded onto a Nano cHiPLC
Trap column (200 μm × 0.5 mm, ChromXP C18-CL, 3 μm, 120 Å; Ek-
sigent) at a flow rate of 2 μl min−1 for 10 min. The peptides were then
separated on a Nano cHiPLC column (75 μm × 15 cm, ChromXP C18-CL,
3 μm, 120 Å; Eksigent) using a 120-min linear gradient of 5–35% ACN in
0.1% formic acid at a flow rate of 300 nl min−1. Settings were: MS1, mass
range m/z 400–1,250 and accumulation time 0.5 s; MS2, mass range m/z
100–1,800, accumulation time 0.05 s, high sensitivity mode, charge
state 2–5, selecting anything over 100 counts per second, maximum
number of candidates per cycle 50, and excluding former targets for
12 s after each occurrence.Protein identification. The raw mass spectrometry data files generated
by the AB SCIEX TripleTOF 5600 system were converted to centroided
mzml files, which were searched using the X!Tandem search engine
(https://thegpm.org) against the E. coli proteome database (Uniprot;
https://www.uniprot.org) to identify proteins. The following param-
eters were used in the X!Tandem searches: parent mass error 50 ppm,
fragment mass error 100 ppm. Ions with charge 1, 5, 6 or 7 were ignored,