Microbiology (2014), 160, 1501–1512
DOI 10.1099/mic.0.078089-0
Protein turnover forms one of the highest
maintenance costs in Lactococcus lactis
Petri-Jaan Lahtvee,1,23 Andrus Seiman,2 Liisa Arike,2,34
Kaarel Adamberg1,2,3 and Raivo Vilu1,2
Correspondence
Petri-Jaan Lahtvee
[email protected]
1
Department of Chemistry, Tallinn University of Technology, Akadeemia tee 15,
12618 Tallinn, Estonia
2
Competence Centre of Food and Fermentation Technologies, Akadeemia tee 15a,
12618 Tallinn, Estonia
3
Department of Food Processing, Tallinn University of Technology, Ehitajate tee 5,
19086 Tallinn, Estonia
Received 10 February 2014
Accepted 9 April 2014
Protein turnover plays an important role in cell metabolism by regulating metabolic fluxes.
Furthermore, the energy costs for protein turnover have been estimated to account for up to a
third of the total energy production during cell replication and hence may represent a major
limiting factor in achieving either higher biomass or production yields. This work aimed to measure
the specific growth rate (m)-dependent abundance and turnover rate of individual proteins,
estimate the ATP cost for protein production and turnover, and compare this with the total energy
balance and other maintenance costs. The lactic acid bacteria model organism Lactococcus lactis
was used to measure protein turnover rates at m50.1 and 0.5 h”1 in chemostat experiments.
Individual turnover rates were measured for ~75 % of the total proteome. On average, protein
turnover increased by sevenfold with a fivefold increase in growth rate, whilst biomass yield
increased by 35 %. The median turnover rates found were higher than the specific growth rate of
the bacterium, which suggests relatively high energy consumption for protein turnover. We found
that protein turnover costs alone account for 38 and 47 % of the total energy produced at m50.1
and 0.5 h”1, respectively, and gene ontology groups Energy metabolism and Translation
dominated synthesis costs at both growth rates studied. These results reflect the complexity of
metabolic changes that occur in response to changes in environmental conditions, and signify the
trade-off between biomass yield and the need to produce ATP for maintenance processes.
INTRODUCTION
Recently, protein turnover rates have been recognized as
an important factor that provides growth advantage both
in terms of higher maximal specific growth rate (m) and
biomass yield between different strains (Hong et al., 2012),
and in laboratory evolution experiments (González-Ramos
et al., 2013). In addition, protein synthesis and degradation
are firmly related and therefore represent an important
control factor for metabolic regulation (Schwanhäusser
et al., 2013). Individual protein turnover rates have been
3Present address: Department of Chemical and Biological Engineering,
Chalmers University of Technology, Kemivägen 10, Gothenburg, Sweden.
4Present address: Department of Medical Biochemistry, University of
Gothenburg, Medicinaregatan 9A, Gothenburg, Sweden.
Abbreviations: ADI, arginine deiminase; DW, dry weight; ESI, electrospray ionization; H/L, heavy/light; GO, gene ontology; LC, liquid
chromatography; MFA, metabolic flux analysis.
Five supplementary tables are available with the online version of this
paper.
078089 G 2014 The Authors
determined in mammalian cells, yeast and bacteria by
measuring the incorporation rate of fluorescent tags
(Khmelinskii et al., 2012), affinity tags (Belle et al., 2006)
or isotopic labels into the proteome (methods reviewed
by Hughes & Krijgsveld, 2012; Trötschel et al., 2013). The
metabolic incorporation of isotopic labels is currently the
most widely used method for cell cultures and conducted
using either labelled ammonium (Helbig et al., 2011; Martin
et al., 2012), carbon (Cargile et al., 2004) or amino acids
(Gerth et al., 2008; Maier et al., 2011; Schwanhäusser et al.,
2011). Protein turnover rates are predominantly measured
during the logarithmic or stationary phase of batch growth
and are assumed to be constant during this phase of growth.
Only a few studies report protein degradation rates determined from a steady physiological state (Helbig et al., 2011;
Pratt et al., 2002). Although overall protein turnover has
been estimated at different temperatures (Trötschel et al.,
2012), no study to date has compared individual protein
degradation rates determined under different environmental
growth conditions (e.g. growing on different substrates or
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Mon, 31 Jul 2017 19:51:01
Printed in Great Britain
1501
P.-J. Lahtvee and others
specific growth rates) to gain insight into the energetic costs
of cellular maintenance. Whilst the growth rate dependence
of mRNA turnover rates has been described (Dressaire et al.,
2013), analogous studies for protein turnover rates are
missing.
Lactococcus lactis has become an important model microorganism due to its importance in the food industry and
its increasing use as a cell factory for the production
of therapeutic recombinant proteins and DNA vaccines
(Bahey-El-Din, 2012; Bermúdez-Humarán et al., 2011,
2013; Buonaguro et al., 2011). The physiology, metabolism
and genetics of L. lactis have been studied intensively,
and a number of genetic modification technologies have
been developed. As L. lactis has lost its ability for oxidative
phosphorylation in the absence of haem, it has a relatively
simple energy metabolism. During anaerobic growth on
glucose energy metabolites are recycled via substrate-level
phosphorylation in glycolysis and additional energy can
be produced via the arginine deiminase (ADI) pathway.
However, not all of the energy produced is used for biomass
formation. The energetic costs for biomonomer transport,
de novo synthesis and polymerization have been estimated
previously (Adamberg et al., 2012; Karr et al., 2012; Russell
& Cook, 1995; Wodke et al., 2013). Whilst the biomass
composition is required to calculate the energetic costs for
biomass formation, it is rarely determined, yet often taken
from the literature. Moreover, dynamic changes in the
biomass composition are usually neglected. Nevertheless, it
is well accepted that in addition to biomonomer transport,
biosynthesis and polymerization, there are additional costs
for cell maintenance. Pirt (1965) defined maintenance
energy as ‘the energy consumed for functions other than the
production of new cell material’. In more recent studies,
maintenance energy has been divided into non-growth
components and physiological or growth-related maintenance (van Bodegom, 2007; Taymaz-Nikerel et al., 2010).
Whilst global-scale analytical and modelling methodologies
are advancing rapidly under the framework of systems
biology, only a few studies have quantified various cellular
maintenance costs (Karr et al., 2012; Wodke et al., 2013). A
number of reviews have discussed various components
of maintenance and its dependence on specific growth
(van Bodegom, 2007; Russell & Cook, 1995); however, a
quantitative understanding of maintenance as a function of
the specific growth rate is still lacking. Recently, energy
generation and consumption have been quantified in the
small bacteria Mycoplasma pneumonia (Wodke et al., 2013)
and Mycoplasma genitalium (Karr et al., 2012). For fastgrowing bacteria, both protein turnover and ion leakage are
considered to be the main energy-consuming maintenance
processes (Russell & Cook, 1995; Taymaz-Nikerel et al.,
2010).
In the current study, protein turnover rates were measured
in L. lactis subsp. lactis IL1403 to understand the role of
protein turnover from the perspective of physiology and
this knowledge was used to elucidate the proportion of
protein resynthesis cost from total ATP expenditures under
1502
steady-state conditions at two specific growth rates. More
specifically, we aimed to determine whether there was a
correlation between protein concentrations and turnover
rates, and which enzymes or pathways were the most costly
for the cell to express. This is the first study that uses this
knowledge to simultaneously increase both protein turnover rates and specific growth rate (i.e. metabolic flux
rates). The ATP costs for proteome turnover were in good
accordance with the energetic calculations, forming 38 and
47 % of the total energy production and 44 and 75 % of
the total maintenance expenditures at m50.1 and 0.5 h21,
respectively. Specific growth rate-dependent measurement
of protein abundance and turnover rates, combined with
mRNA and flux levels, could shed additional light on future
studies of metabolic regulation; however, this approach was
beyond the scope of the current paper.
METHODS
Strain and cultivation conditions
Strain/medium. L. lactis IL1403, acquired from INRA, France, was
used in all experiments and grown with the following chemically
defined base medium (g l21): D-glucose, 4.5; L-alanine, 0.0782;
L-arginine, 0.1852; L-asparagine, 0.12; L-aspartate, 0.0723; L-cysteine,
0.0636; L-glutamate, 0.0703; L-glutamine, 0.2; L-glycine, 0.0578;
L-histidine, 0.0597; L-isoleucine, 0.1018; L-leucine, 0.2072; L-lysine,
0.1576; L-methionine, 0.0407; L-phenylalanine, 0.086; L-proline,
0.0917; L-serine, 0.1634; L-threonine, 0.0758; L-tryptophan, 0.0164;
L-tyrosine, 0.0295; L-valine, 0.1072; biotin, 0.305; choline chloride, 9.8;
D-pantothenate, 0.65; folic acid, 1.21; niacinamide, 0.325; pyridoxine
hydrochloride, 0.642; riboflavin, 0.326; thiamine hydrochloride,
0.51; vitamin B12, 0.98; adenine, 0.025; hypoxanthine-Na, 0.025;
MgSO4.7H2O, 0.2; FeSO4.4H2O, 0.0014; CaCl2, 0.05; MnSO4.5H2O,
0.016; ZnSO4.7H2O, 0.005; CoSO4.5H2O, 0.003; CuSO4.2H2O, 0.003;
(NH4)6Mo7O24.4H2O, 0.003; NaCl, 2.9; K2HPO4, 3; KH2PO4, 2.5.
Glucose, amino acids, buffers and minerals were dissolved separately
and mixed together after autoclaving in the above-mentioned order to
avoid precipitation. Vitamins, nucleotides, cysteine and lysine were
added by filter sterilization. Inoculum was prepared from a glycerol
stock culture stored at 280 uC. Stock culture was pre-grown twice
on the base medium and 2 % (v/v) of overnight-grown culture was
inoculated into the bioreactor.
Experiments/switch of medium. The cultivation system (described
in Lahtvee et al., 2011) was run in chemostat mode. Briefly, this
system consisted of a Biobundle 1.25 l reactor (Applikon) controlled
by an ADI 1030 biocontroller (Applikon), together with the
cultivation control program BioXpert (Applikon). Chemostat
experiments were carried out in triplicate at each of two dilution
rates (0.1 and 0.5 h21). All cultivation experiments were performed at
34 uC, pH 6.4, in an anaerobic (N2) environment with an agitation
speed of 300 r.p.m. and a working volume of 300 ml. In all chemostat
experiments, the lysine concentration in the medium was reduced to a
level which did not affected the growth parameters (production
yields), but resulted in a joint limitation of glucose and lysine (both
concentrations were under the limit of detection). Hence, 0.05 and
0.03 g lysine l21 were used in the chemostat experiments at dilution
rates of 0.1 and 0.5 h21, respectively. To measure protein degradation
rates, the rate of medium inflow was kept constant whilst switching
from isotopically unlabelled medium (light) to a medium with
identical composition aside from the use of [13C6-15N2]lysine (heavy)
(Cambridge Isotope Laboratories). Rapid turnover of lysine in the
cultivation environment was required to ensure a rapid switch from
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Mon, 31 Jul 2017 19:51:01
Microbiology 160
Energetic burden of protein turnover
light (unlabelled) lysine to heavy lysine. Samples for proteome analysis
were collected from the steady physiological state before the medium
switch (0 h), and at 0.5, 1.5, 5 and 10 h following the switch to
isotopically labelled heavy medium. Note that the 10 h sample was
collected only in the case of the dilution rate of 0.1 h21. Although WT
L. lactis is not auxotrophic for lysine, 97 % of lysine in the proteome
was labelled in L. lactis IL1403 after roughly nine generations in
separate batch cultivation experiments (data not shown).
Analytics
Biomass concentration, composition and extracellular metabolites.
The biomass concentration was monitored by measuring the OD600
and the dry weight (DW) was determined gravimetrically. The OD600
and DW correlation constant K was 0.31±0.02, and was not found to
be specific growth rate dependent.
Samples of culture medium were centrifuged (14 000 r.p.m., 4 min);
supernatants were collected and stored at 220 uC until analysis.
Levels of glucose, lactate, formate, acetate and ethanol in the culture
medium were measured using liquid chromatography (LC; Alliance
2795 system; Waters) with a HPX-87H column (Bio-Rad) and
isocratic elution of 5 mM H2SO4 at a flow rate of 0.6 ml min21 at
35 uC. A refractive index detector (model 2414; Waters) was used to
both detect and quantify each substance with a detection limit of
0.1 mM. The concentrations of free amino acids were determined
using an AccQ?Tag (Waters) method supplied by the manufacturer.
Briefly, all samples and an amino acid standard mixture (Waters)
were subject to pre-column AccQ?Tag Ultra derivatization and
analysed with the Acquity UPLC (ultra performance liquid chromatography) system (Waters). The compounds within derivatized
samples were separated using reverse-phase chromatography and
detected with a UV detector, and the outflow was fed into an
electrospray ionization (ESI) mass spectrometer (LCT Premiere;
Waters) operated using the following parameters: negative ionization
mode, desolvation gas 700 l h21 and 200 uC, capillary voltage 2500 V,
sample cone 30 V. For each amino acid, we analysed the ratio of heavy
to light (H/L) chromatographic peaks using Empower or MassLynx
software (Waters). To determine the amino acid composition of the
protein content, a biomass sample was hydrolysed with 6 M HCl for
20 h at 120 uC in an N2 environment with an Eldex Hydrolysis/
Derivatization WorkStation (Eldex Laboratories). From this hydrolyte,
amino acids and their H/L protein ratios were determined as described
above. The H/L ratios of intracellular free amino acids were measured
from snap-frozen culture samples which were pelleted by centrifugation (14000 r.p.m., 4 uC), washed twice with 0.85 % NaCl solution
followed by extraction of the intracellular metabolites with a 70 %
ethanol solution at 80 uC. This extraction solution was freeze-dried at
0.14 mbar to complete dryness and the free amino acids were
determined as described above. Due to the leakage of intracellular
metabolites into the extracellular environment, this method cannot be
used to determine the concentration of intracellular amino acids;
however, it should correctly represent the H/L isotope ratios. The
cellular protein content was determined both from an amino acid
analysis of dry biomass and measured directly using the Lowry method
(Lowry et al., 1951), where BSA was used as a standard.
Proteome analysis. L. lactis steady-state culture was harvested in a
2 ml vial (~1023 gDW), pelleted by centrifugation at 14 000 r.p.m.
for 20 s and flash frozen in liquid nitrogen. Samples were stored at
280 uC until further processing. To lyse the cells, the cell pellets were
washed with PBS, suspended in 100 ml SDS lysis buffer (4 % SDS,
100 mM Tris/HCl, pH 8 and 100 mM DTT) and heated to 95 uC
for 15 min. Cell lysates were cleared by sonication with ultrasound
for a few pulses and centrifuged (10 min at 14 000 r.p.m.) in a tabletop centrifuge. The protein concentration was determined with a
tryptophan fluorescence calibration curve at an excitation wavelength
of 295 nm and emission wavelength of 350 nm. Protein samples (5 mg)
http://mic.sgmjournals.org
were digested with trypsin according to a FASP (filter-aided sample
preparation) protocol (Wiśniewski et al., 2009). Prior to LC-MS-MS
analysis, peptides were purified with C18 StageTips (Rappsilber et al.,
2007). LC-MS-MS analysis was performed on an Agilent 1200
series nanoflow system (Agilent Technologies) connected to a LTQ
Orbitrap mass spectrometer (Thermo Electron) equipped with a nanoelectrospray ion source (Proxeon). Purified peptides were loaded onto a
self-packed fused silica emitter (150 mm60.075 mm; New Objective)
packed with ReproSil-Pur C18-AQ 3 mm particles (Dr Maisch) at a
flow rate of 0.7 ml min21. Peptides were separated with a 240 min
gradient from 2 to 40 % B (A, 0.5 % acetic acid; B, 0.5 % acetic acid/
80 % acetonitrile) using a flow rate of 200 nl min21 and sprayed into an
LTQ Orbitrap mass spectrometer operated with a 180 uC capillary
temperature and 2.2 kV spray voltage. Full mass spectra were acquired
in profile mode with a m/z range of 300–1900 at a resolving power of
60 000 (full width at half maximum). Up to five data-dependent MSMS spectra were acquired in centroid mode in the linear ion trap for
each Fourier transform MS full-scan spectrum (normalized collision
energy 35 %, maximum injection time 150 ms, fill value 56103). Each
fragmented ion was dynamically excluded for 60 s.
Raw data files were analysed with the MaxQuant software package
(version 1.2.7.4). The resulting peak lists were searched using the
Andromeda search engine (built into MaxQuant) against the L. lactis
database (downloaded on 28 June 2012 from http://www.genome.jp/
kegg/genes.html). Andromeda searches were performed with full
tryptic specificity with a maximum of two missed cleavages and a mass
tolerance of 0.5 Da for the fragment ions. Carbamidomethylation of
cysteine was set as a fixed modification, and both methionine
oxidation and protein N-terminal acetylation were set as variable
modifications. The required parameter of false discovery rate was set to
1 % for both the peptide and protein levels, and the minimum required
peptide length was set to 6 aa. In addition, the ‘Match between runs’
option was utilized with an allowed time window of 1.5 min.
The concentration of each protein (molecules gDW21) was calculated
by normalizing the MS intensity of each protein to the total sample
intensity and multiplying this normalized intensity with the measured
protein content in the sample (Arike et al., 2012; Wiśniewski et al.,
2012). To estimate the total abundance of the measured proteome,
the MS intensities of individual proteins were correlated with
measured absolute mRNA levels (data not shown). Based on received
correlation, protein concentrations were estimated for unmeasured
proteins. As a result, ~91 % of the total proteome was quantified. In
the energy calculations, the remaining 9 % was taken into account as a
protein with median molecular mass and median degradation rate.
All proteins quantified, together with their property values used in
the following analysis, are listed in Table S5 (available in the online
Supplementary Material). The MS proteomics data have been
deposited in the ProteomeXchange Consortium (http://proteomecentral.
proteomexchange.org) via the PRIDE partner repository (Vizcaı́no et al.,
2013) and can be retrieved using the dataset identifier PXD000494.
Metabolic flux analysis (MFA). To calculate the carbon and ATP
balances and intracellular fluxes we made use of a curated stoichiometric model of L. lactis (Adamberg et al., 2012). Briefly, this simplified
metabolic model is based on data from the BioCyc database, and takes
into account reactions from the central carbon metabolism (glycolysis,
pentose phosphate pathway and pyruvate metabolism), amino acid
metabolism and biomonomer synthesis pathways. Substrate consumption (glucose, amino acids and nucleotides), product formation
and biomass composition values were used as input to the model. This
model allows the calculation of a unique set of intracellular fluxes
without using optimization.
Calculation of protein degradation rates. From the moment
when the unlabelled lysine substrate is replaced by heavy lysine, the
fraction of unlabelled amino acid begins to decrease because lysine is
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Mon, 31 Jul 2017 19:51:01
1503
P.-J. Lahtvee and others
incorporated continuously into the biomass as the bacteria grow.
However, the fraction of unlabelled lysine is also influenced by the
turnover rate of each protein. The rate of change in the fraction of
unlabelled lysine can be described with the differential equation:
df p ,i
dt
= f a ( t ) ( d + k d ,i ) − f p ,i ( t ) k d ,i − f p ,i ( t ) d
(1)
Parameters fa and fp,i are unlabelled fraction of lysine in its pool and
in protein i at time t. kd,i is the turnover rate of protein i. The specific
growth rate m of bacteria is equal to the dilution rate d at steady state
in continuous culture experiments.
The change in the fraction of labelled lysine in protein i is equal to the
amount of synthesized labelled protein minus the amount of labelled
protein degraded or left in the system through dilution. Under steady
state, the total amount of protein is constant and therefore the rate of
protein synthesis is equal to the sum of the dilution rate and protein
turnover rate. The fraction of metabolic label in the proteins being
synthesized depends on the fraction of label in the metabolic lysine
pool, whilst the fraction of label in degraded or diluted proteins is a
function of the lysine fraction in the protein.
The free lysine pool fa is defined as the amount of lysine available to
bacteria for the synthesis of proteins. This free lysine pool is common
for all proteins in the cell. The fraction of unlabelled lysine in its pool
is a function of time and can be defined as:
f a ( t ) = e − at
(2)
where coefficient a describes the rate at which unlabelled lysine is
replaced by heavy lysine in its free amino acid pool. This rate is
affected by the rate of introducing labelled lysine into the cell and also
by the protein degradation rate, because light lysine can be reused for
the synthesis of new proteins. The higher the protein degradation
rates, the higher coefficient a. This concept is very similar to the
dynamic labelling experiments described by Hong et al. (2012).
The integration of equation (1) gives a solution for fp,i:
f p,i ( t ) =
−( d + k
)t
d ,i
e
⎡e( d + kd ,i − a )t ( d + k ) − a ⎤
d ,i
⎦
d + k d ,i − a ⎣
(3)
Equation (3) contains two unknown parameters a and kd,i that are
calculated in the present study by fitting to experimental data. Note
that parameter a is the same for the entire proteome, whilst kd,i is
specific for every protein. This means that the fitting of experimental
data must be performed for all proteins at once, i.e. the system in
equation (4) must be solved for n proteins:
−( d + kd ,1 )t
⎧
e
⎡e( d + kd ,1 − a )t ( d + k ) − a ⎤
⎪ f p ,1 ( t ) =
d ,1
⎦
d + kd ,1 − a ⎣
⎪
⎪
M
⎪
−
d
+
k
t
(
)
d ,i
⎪
e
⎡e( d + kd ,i − a )t ( d + k ) − a ⎤
⎪ f p ,i ( t ) =
d ,i
⎦
d + k d ,i − a ⎣
⎨
⎪
M
⎪
− ( d + kd ,n ) t
⎪
⎡ e ( d + kd ,n − a ) t ( d + k ) − a ⎤
⎪ f p,n ( t ) = e
d ,n
⎦
⎪
d + kd , n − a ⎣
⎪
⎩
(4)
All proteins with at least three corresponding H/L chromatographic
peak ratios were selected for the fitting procedure. Altogether, 634
1504
and 562 proteins were found to be suitable for the fitting procedure
conducted for experiments at dilution rates of 0.1 and 0.5 h21,
respectively. The number of data points used whilst fitting was 5010
and 3756 for experiments at dilution rates of 0.1 and 0.5 h21,
respectively.
Careful study of equation (3) reveals symmetry between a and
(d+kd,i). This means it is impossible to distinguish between the
situation of a high value of a and a low (d+kd,i) or vice versa for
individual proteins. However, because we find a global fit to the
system in equation (4) using a large dataset, this problem can partially
be overcome. Using the extra data, we tested to see if this situation
existed by performing a sensitivity analysis. The optimal value for
aopt found in the original optimization was fixed at different levels
(0.5aopt, 2aopt, 4aopt), whilst new protein degradation rates kd,i were
found in subsequent fitting procedures. An F test was performed to
compare total residual sum of squares between different models. The
model with the aopt value displayed the smallest sum of squared
residuals at a statistically significant level a,0.001.
The data-fitting procedure was performed using Mathematica 8.0
(Wolfram Research). Fitting was performed in two stages. First,
the most abundant proteins with summarized abundance of at least
50 % of the total proteome were selected. Parameters in the system in
equation (4) were found by fitting with experimental data. This
resulted in turnover rates for a select set of proteins together with
the rate at which unlabelled lysine is replaced by heavy lysine in its
metabolic pool a. In the second stage of the fitting procedure,
the remaining proteins were fitted to equation (3) using a from the
results of the first stage. This dual-fitting approach was chosen
because the algorithm used cannot incorporate weights. This means
that all proteins have equal significance in the fitting process regardless
of their abundance. However, the rate at which labelled lysine is
incorporated into the cell is controlled more by abundant proteins.
Therefore, dropping the less abundant proteins from the first fitting
stage actually improves the precision of the results. A secondary reason
the fitting procedure was split into two steps was to reduce the
calculation time. Remarkably, only 39 and 22 proteins cover at least
50 % of the entire proteome for experiments conducted at dilution
rates of 0.1 and 0.5 h21, respectively.
The quality of the resulting fit was good, especially in the experiment
with a dilution rate of 0.5 h21, with the median coefficient of
determination R2 being 0.9962. For the experiment with a dilution
rate of 0.1 h21, the quality of fit was lower with the median R2 being
just 0.8262 and 0.9413 for the upper quartile. The poorer quality fit at
the lower dilution rate can be associated with slower incorporation of
heavy lysine into the proteins which leads to a lower precision in the
quantification of heavy proteins.
The rate at which unlabelled lysine is replaced by heavy lysine in its
metabolic pool (a) was calculated to be of the same order of magnitude
as the dilution rate, 0.067±0.006 and 0.471±0.055 h21 for experiments conducted at dilution rates of 0.1 and 0.5 h21, respectively.
These values were also validated by measuring directly the incorporation of heavy lysine into its metabolic pool, whilst the extracellular
lysine concentration was under the limit of detection. Measured data
showed no significant difference from calculated values (a50.040 and
0.459 h21 for experiments conducted at dilution rates of 0.1 and
0.5 h21, respectively) and, therefore, the calculated a values were used
in the following calculations.
Some of the calculated turnover rates had disproportionally large
confidence intervals which indicates that their values could not be
calculated precisely using this dataset. As there is no obvious distribution of calculation errors, we applied a non-parametric limit of
four median values to remove proteins with large calculation errors.
This filter affected 160 and 30 proteins for experiments conducted
at dilution rates of 0.1 and 0.5 h21, respectively. As an example, 132
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Mon, 31 Jul 2017 19:51:01
Microbiology 160
Energetic burden of protein turnover
of 160 proteins had protein half-lives ,5 min. After non-parametric
filtering, 474 and 532 proteins with turnover rates remained. In the
next step we found proteins whose turnover rate confidence intervals
intersected the zero value. The turnover rate of these proteins was
considered to be zero.
proteins into peptides with 10 aa and further degradation is carried
out by ATP-independent peptidases. These assumptions provide an
estimated turnover cost of 4.506 molecules of ATP per peptide bond.
RESULTS AND DISCUSSION
Energy calculations. Both the amount of ATP generated during
substrate consumption per gram of dry biomass produced and the
ATP cost for biomass formation were calculated with the aid of a
MFA model that was built in-house (see above and Adamberg et al.,
2012). We defined the difference between the ATP generated and the
ATP consumed for biomass production to be the energy expenditure
for maintenance. Cell maintenance could be compared with measured
protein degradation and resynthesis rates. The ATP consumption
associated with protein polymerization has been estimated to be 4.306
ATP equivalents for each amino acid added to the protein chain
(Stephanopoulos et al., 1998), whilst the cost of protein degradation
by the proteasome has been estimated to be two ATP molecules per
peptide bond (Burton et al., 2001; Menon & Goldberg, 1987). In this
work, we presume that the ATP-dependent proteasome degrades
Physiology of L. lactis
Three replicate chemostat cultivations of L. lactis IL1403
were carried out at each of two different dilution rates (0.1
and 0.5 h21) using chemically defined medium (Fig. 1a) to
study its physiology. After obtaining steady-state culture in
each chemostat cultivation, medium containing unlabelled
lysine was switched with an identical medium containing
heavy [13C6-15N2]lysine at the same flow rate. After the
isotopic label switch, we followed the incorporation of
heavy lysine into the protein content (Fig. 1b). Contrary to
(a)
Biomass
Start of dilution Switch of medium
Sampling
Unlabelled
Labelled
Lys
[15N]Lys
Time
(b)
Nano LC-MS-MS
Extracellular metabolome
Intracellular amino acid H/L ratios
Proteogenic amino acid H/L ratios
Protein abundances
Individual protein H/L ratios
UPLC-MS
(c) [15N]Lys
ƒp,i (t) =
e–(d+kd,i )t (d+kd,i –a)t
(d+kd,i)–a]
[e
d+kd,i –a
[15N]Lys
Lys
[15N]Lys
MFA
fa(t) = e–at
Protein
Lys
Cytoplasm
Environment
Fig. 1. Overview of the experimental plan used in this study. Replicate chemostat cultivations were carried out in triplicate with
L. lactis at each of two specific growth rates (m50.1 and 0.5 h”1) using a chemically defined medium where both glucose and
lysine were jointly limiting the growth. After reaching a steady physiological state, medium containing unlabelled lysine was
switched with medium containing heavy [13C6-15N2]lysine after which the incorporation of heavy amino acid into the biomass
was observed over the time course (a). Each sample collected was subjected to multiple analyses including measurement of
individual protein abundances and the incorporation of heavy lysine into the proteome (b). Protein turnover rates were calculated
based on the data collected (c). In contrast with previous studies, intracellular recycling of amino acids was taken into account in
the calculations by adding intracellular amino acid pools into the model. In addition, both the biomass composition and
extracellular fluxes were used as input for a L. lactis MFA model to analyse the intracellular fluxes together with energy (ATP)
balance calculations. Maintenance energy expenditures in bacteria were estimated and compared with ATP costs for protein
turnover.
http://mic.sgmjournals.org
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Mon, 31 Jul 2017 19:51:01
1505
1506
Y, biomass yield; O, outflow (production) flux; I, inflow (consumption) flux.
*Although present in the medium, glutamate, aspartate and glycine (latter only at m50.1 h21) were produced during the cultivation of L. lactis.
9.31±0.20
5.77±0.04
3.28±0.42
1.17±0.62
5.12±0.36
1.01±0.16
8.38±0.82
3.51±0.22
0.091±0.000
0.140±0.003
0.1
0.5
118.3±6.8
71.78±1.36
Iaa (mmol gDW”1)
OLactate
(mmol gDW”1)
OFormate
(mmol gDW”1)
OAcetate
(mmol gDW”1)
OEthanol
(mmol gDW”1)
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Mon, 31 Jul 2017 19:51:01
YXS [gDW
(g glucose)”1]
Using absolute proteome quantification, the intracellular
concentration of 911 proteins at m50.1 h21 and 964
proteins at m50.5 h21 were estimated to contribute 91 % of
the entire L. lactis proteome (see Methods). Protein turnover
rates, together with confidence limits, were calculated for
471 and 529 proteins at m50.1 and 0.5 h21, respectively. We
used an algorithm within MaxQuant (Cox & Mann, 2008)
to calculate both the H/L ratios and amount of heavy lysine
incorporated into the biomass. The latter quantity was
confirmed by conducting quantitative amino acid measurements of both the unlabelled and labelled lysine concentrations in the biomass. In addition, the amino acid analysis
confirmed that lysine was not used as a substrate for the
production of other proteinogenic amino acids because
labelled nitrogen was not incorporated into other amino
acids. Protein turnover calculations were conducted using
the model presented in Fig. 1(c), which takes into account
amino acid recycling. Typically, protein turnover rates are
calculated based only on the decrease of unlabelled protein
Specific growth
rate m (h”1)
Protein abundances and turnover
Table 1. Physiological data from L. lactis chemostat experiments (mean±SD of three biological replicates)
Comparing the specific growth rate-dependent physiology
of L. lactis, the main differences are biomass yield, production of organic acids and ethanol, and the consumption
and production of amino acids (Table 1, Fig. 2c, d).
Although lactate production (mmol gDW21) decreased
with an increase in specific growth rate, it constantly
comprised 82±1 % of the total carbon production at both
specific growth rates studied (Fig. 2c, d). In addition, the
total protein content within cells remained constant at 51 %
(data not shown). All physiological data were in accordance with previous specific growth rate-dependent studies
(Dressaire et al., 2008; Mehmeti et al., 2012; Lahtvee et al.,
2011). Based on MFA calculations, we estimated the ATP
requirements for biomass synthesis and total ATP generation during growth (Table S1). It was found that with the
increase of specific growth rate and biomass yield, total ATP
generation decreased from 123.8±1.6 to 75.9±1.0 mmol
gDW21. The ATP cost for non-productive expenditures
per biomass was approximately twofold higher at the lower
specific growth rate, forming 95.1±3.2 and 47.7±0.9 mmol
gDW21 at m50.1 and 0.5 h21, respectively (Fig. 2a, b).
Oaa
(mmol gDW”1)*
the typical SILAC (stable isotope labelling by amino acids
in cell culture) approach of using both labelled lysine and
arginine (Ong et al., 2002), only lysine was used as the
sole labelling source throughout the experiment. Protein
turnover rates were calculated based on the increase of the
H/L protein ratios. In contrast with previous studies that
measured protein turnover, the intracellular recycling of
amino acids between the protein content and a free amino
acid pool was taken into account in a turnover calculation
involving individual proteins (Fig. 1c). Neglecting the
intercellular recycling of unlabelled lysine which is released
during protein degradation may lead to an underestimation
of the rate of protein degradation (i.e. overestimation of the
length of protein half-lives).
2.85±0.18
0.62±0.02
P.-J. Lahtvee and others
Microbiology 160
Energetic burden of protein turnover
80
60
40
20
0
Carbon (mmol gDW–1)
(c) 400
350
300
250
200
150
100
50
0
ATP (mmol gDW–1)
100
100% (b)
100%
70
90%
90%
80%
80%
60
70%
70%
50
60%
60%
40
50%
50%
40%
40%
30
30%
30%
20
20%
20%
10
10%
10%
0%
0%
0
Production Consumption
Production Consumption
100%
100% (d)
250
90%
90%
80%
80%
200
70%
70%
60%
60%
150
50%
50%
40%
40%
100
30%
30%
20%
20%
50
10%
10%
0%
0
0%
Consumption Production
Consumption Production
Carbon (mmol gDW–1)
ATP (mmol gDW–1)
(a) 120
Non-productive expenditures
Biomonomer synthesis
Biomonomer polymerization
Amino acid transport
Acetate pathway
Glycolysis
OBiomass
OFormate, acetate, ethanol
OLactate
Oaa
IHypoxanthine
Iaa
IGlucose
Fig. 2. (a, b) ATP and (c, d) carbon consumption (mmol gDW–1) and production distribution (%) in L. lactis chemostat
cultivations at m50.1 (a, c) and 0.5 (b, d) h”1. O, outflow (production) flux; I, inflow (consumption) flux.
content in the cells (Helbig et al., 2011), thus neglecting the
unlabelled amino acids which are liberated from degraded
proteins. As additional unlabelled amino acids enter into the
intracellular amino acid pools, they can influence considerably the rate of incorporation of heavy amino acids into
proteins. Therefore, we measured the H/L ratio of the free
intracellular pool of lysine at the same time points used to
collect samples for proteome analysis. The observed increase
in the free lysine H/L ratio was compared with theoretical
values obtained from the model used to calculate protein
turnover (refer to Methods for more information).
Protein turnover rates for individual proteins were found
to range from 7 h21 to close to zero (lowest turnover rates
which could be measured adequately within the current
experimental timespan were 0.028 and 0.055 h21 for
chemostat experiments at m50.1 and 0.5 h21, respectively;
Table S2). The median rates of protein turnover were
determined to be 0.122 and 0.913 h21 (corresponding to
340.5 and 45.5 min) for m50.1 and 0.5 h21, respectively
(Fig. 3a). These values are higher than the specific bacterial
growth rate and indicate a relatively high energetic cost
for protein turnover. We found an acceptable correlation
(R250.44) between individual protein turnover rates at
m50.1 and 0.5 h21, meaning that the majority of protein
turnover rates increased with the increase in specific
growth rate (Fig. 3b, Table S2). No significant correlation
was detected between protein turnover rates and protein
abundance or length.
Furthermore, protein turnover rates were grouped according to the gene ontology (GO; according to Bolotin et al.,
2001), and group medians and distributions were compared (Fig. 3c). Although intergroup variations were found
http://mic.sgmjournals.org
to be high, we observed a number of trends. Groups such
as Energy metabolism, Translation, Regulatory functions,
etc., had protein turnover rates close to the median values at
both specific growth rates studied. Lower protein turnover
rates belonged to groups such as Cell envelope, Cellular
processes, and Transport and binding. The higher protein
turnover rates were found for Biosynthesis of cofactors,
prosthetic groups and carriers, Fatty acid and phospholipid
metabolism, and Purines, pyrimidines, nucleosides and
nucleotides (Fig. 3c). An interesting exception was fructose
bisphosphate aldolase (FbaA), which is a glycolytic enzyme
within the group of Energy metabolism. The proteins within
Energy metabolism typically had median protein half-lives;
however, FbaA had more than twofold higher turnover rates
than the group median (turnover rates 2.8 and 1.8 h21
at m50.1 and 0.5 h21, respectively). The main substrate
for FbaA, fructose bisphosphate, is a well-known allosteric
regulator of the L. lactis central carbon metabolism (Neves
et al., 2005; Oh et al., 2011; Thompson, 1978) and high
turnover rates of this enzyme may indicate the need for
rapid regulation of the fructose bisphosphate pool at the
(post) translational level.
Energetic burden of protein synthesis and
turnover
The dataset gathered in this study forms one of the largest
datasets of protein turnover rates published so far, covering
between 70 and 87 % of cellular protein mass at m50.1 and
0.5 h21, respectively. Individual protein abundances and
turnover rates were used to calculate the energy cost for
proteome synthesis and turnover (degradation and resynthesis). For proteins where the turnover rate could not be
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Mon, 31 Jul 2017 19:51:01
1507
P.-J. Lahtvee and others
70
(b)
Median Kdeg
0.91 h–1
60
Counts
50
40
Median Kdeg
0.12 h–1
30
20
10
0
0.01
0.1
1
Protein turnover rate at µ=0.5 h–1
(a)
10
10
n = 294
R2 = 0.44
1
0.1
0.01
0.1
1
10
Protein turnover rate at µ=0.1 h–1
Protein turnover rate (h–1)
Protein turnover rate (h–1)
(c) 10
1
0.1
Unknown, n=82/139
Transport and binding
proteins, n=23/40
Translation, n=62/87
Transcription, n=4/11
Replication, n=7/10
Purines, pyrimidines, ...
n=25/33
Regulatory functions,
n=8/18
Other categories, n=8/16
Cellular processes,
n=7/18
Central intermediary
metabolism, n=4/3
Energy metabolism,
n=59/57
Fatty acid and phospholipid
metabolism, n=12/15
Cell envelope, n=17/24
Amino acid biosynthesis,
n=18/24
Biosynthesis of cofactors,
prosthetic groups and
carriers, n=9/14
0.01
Fig. 3. L. lactis protein turnover rates measured in chemostat experiments at m50.1 (grey bars) and 0.5 (blue bars) h”1: (a)
distribution of individual protein turnover rates, (b) correlation of protein turnover rates and (c) distribution of protein turnover
rates inside GO groups. Horizontal lines on the box-plots represent the group mean, crosses indicate group median value and
error bars indicate 1.5 interquartile range.
calculated, the median protein turnover rate was used
instead. These proteins formed 30 and 13 % of the total
protein abundance at m50.1 and 0.5 h21, respectively.
Based on the absolute concentration of proteins (molecules
gDW21), protein length and the amount of ATP equivalents
required for polymerization of an amino acid into a protein
chain, we calculated the degradation, synthesis and resynthesis costs for all proteins (Table S2). At both specific
growth rates studied, protein synthesis costs for biomass
duplication were in the range of 18.2±0.4 mmol ATP
gDW21. The additional ATP cost for protein turnover
(degradation and resynthesis) was 47.5 and 35.5 mmol
ATP gDW21 at m50.1 and 0.5 h21, respectively. This forms
44 and 75 % of the total maintenance costs (95.1±3.2
and 47.7±0.9 mmol gDW21 at m50.1 and 0.5 h21), and is
a major ATP sink that may prevent bacteria from achieving
1508
a higher specific growth rate or yield. In total, protein
synthesis (including synthesis and turnover costs) is also the
most expensive cellular expenditure, comprising 49 and
69 % of the total energy produced (ATP) at m50.1 and
0.5 h21, respectively.
Based on the abundance and turnover rates of individual
proteins, we calculated the energy demand for the production of individual proteins and cellular protein groups.
Altogether, the 15 most expensive proteins comprised 50 %
of the total energy spent for protein synthesis at m50.1 h21
(Table S2). At m50.5 h21, 24 proteins were responsible for
half of the total protein synthesis costs. Energetically, the
most expensive GO groups at the lower specific growth rate
(0.1 h21) were Energy metabolism (28 % of total protein
synthesis costs), Transport and binding proteins (15 %),
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Mon, 31 Jul 2017 19:51:01
Microbiology 160
Energetic burden of protein turnover
Whilst energy generation in the ADI and acetate generation
pathways was more efficient at lower specific growth rates,
glycolysis was found to be the only pathway whose efficiency
increased at the higher specific growth rate studied. These
results are also in accordance with physiological observations where both the ADI and acetate generation pathways
are downregulated at higher specific growth rates.
Other categories
Amino acid biosynthesis
Cellular processes
Regulatory functions
µ=0.5 h–1
Replication
µ=0.1 h–1
Purines, pyrimidines, ...
Translation
Transport and binding proteins
Energy metabolism
In this study, we demonstrate that protein turnover
(degradation and resynthesis) cost is one of the highest
maintenance expenditures in the L. lactis metabolism. This
finding is in accordance with previous studies with yeast
where it was demonstrated that differences in the rate of
protein degradation may account for observed physiological
differences between different strains and mutants (Canelas
et al., 2010; González-Ramos et al., 2013; Hong et al., 2012).
In addition to protein turnover, mRNA turnover, futile
cycles, protein translocation, etc., are sources of ATP
expenditure that could be categorized under maintenance
costs. Based on median mRNA turnover rates at m50.1 and
0.5 h21 taken from Dressaire et al. (2013) and the mRNA
polymerization cost of 0.4 ATP per phosphate bond (Karr
et al., 2012; Stouthamer, 1973), we estimated mRNA
turnover to account for ~2 % of the total maintenance
expenditure at both specific growth rates studied. Only
futile cycles and ion leakage or pH homeostasis could
account for the majority of the remaining loss of ATP,
which formed 56 and 26 % of the total energy expenditure
at m50.1 and 0.5 h21, respectively. Although difficult to
measure directly, those ATP costs increased by more than
fourfold with decreasing specific growth rate, which is in
accordance with an observed fivefold increase in the length
of the cell cycle. As this cost also depends on the length of
cell cycle, it can help to explain the increased ATP loss at
lower specific growth rates. It has also been shown in both L.
lactis and yeast that a decrease in specific growth rate is
accompanied with a decrease in intracellular pH, which
could be explained by increased energy demand for ion
homeostasis at lower specific growth rates (O’Sullivan &
Condon, 1999; Orij et al., 2012).
0
5
10
15
20
Total protein cost (mmol ATP gDW–1)
Fig. 4. Combined energy (ATP) expenditure for synthesis and
turnover of L. lactis metabolic protein groups at two specific
growth rates. Group ‘Other categories’ includes the ATP cost for
unmeasured proteins assumed to have a median turnover rate.
and Translation (14 %). These three groups were the most
energetically expensive also at the higher specific growth
rate studied (0.5 h21), albeit in a different order: Translation
(24 %), Energy metabolism (20 %), and Transport and
binding proteins (8 %) (Fig. 4). At both specific growth
rates, these groups were followed by Purines, pyrimidines,
nucleosides and nucleotides, Replication, Regulatory functions, Cellular processes and Amino acid biosynthesis, with
each group forming ,6 % of total energetic burden for
protein synthesis (Table S3).
Efficiency of energy generation pathways
We also evaluated the efficiency of energy generation
pathways (EATP) at both specific growth rates studied.
EATP is defined as the ratio of the total ATP produced from
ADP in the given pathway to the ATP expenditure for the
synthesis of enzymes of the pathway (Adamberg et al., 2012;
Valgepea et al., 2013). Even taking into account the energetic
cost of protein turnover, glycolysis, acetate production and
the ADI pathway all showed positive efficiencies (Table 2).
Table 2. Efficiency of various energy generation pathways for L. lactis at two specific growth rates (m50.1 and 0.5 h”1)
Each efficiency represents the ratio between the amount of energy generated in the pathway and the ATP cost for the synthesis of the pathway.
Proteins included within each pathway can be found in Table S4. ‘One-time synthesis’ represents the ATP cost for one-time synthesis of the
pathway enzymes, ‘Total protein synthesis’ also takes into account the protein turnover (degradation and resynthesis) costs.
Efficiency [mmol ATP (mmol ATP)”1]
Pathway
One-time synthesis of biomass
0.1 h”1
ADI pathway
Acetate synthesis
Glycolysis
328.4
46.7
76.1
Total proteome synthesis
0.5 h”1
0.1 h”1
0.5 h”1
0.0*
13.0
53.2
127.4
12.4
10.5
0.0*
5.2
13.9
*Energy was not generated under the corresponding conditions.
http://mic.sgmjournals.org
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Mon, 31 Jul 2017 19:51:01
1509
P.-J. Lahtvee and others
Increased protein degradation rates with increasing specific
growth rate are also in accordance with recently published
mRNA degradation rate data for L. lactis (Dressaire et al.,
2013). Although we could not detect a significant correlation between individual mRNA and protein degradation
rates at m50.1 or 0.5 h21, the median trend for both
datasets was similar. A lack of correlation between
individual protein and mRNA degradation rates has been
observed earlier for mammalian cells (Schwanhäusser et al.,
2011).
To our knowledge this is the first study where extensive
specific growth rate-dependent proteome turnover data
from a steady physiological state were acquired in an
attempt to understand the ATP requirements for protein
turnover of both individual proteins and various protein
groups together with a comparison of the total ATP
generation and consumption profiles. The main finding
is that protein degradation rates are specific growth rate
dependent. The proteome content was observed to be
relatively constant when the intracellular metabolic fluxes
increased by a mean of 3.3-fold with an increase in specific
growth rate from 0.1 to 0.5 h21. This indicates that the
catalytic activities of enzymes increased (as demonstrated
previously by Adamberg et al., 2012 and Valgepea et al.,
2013), which could result in higher protein degradation
rates at higher specific growth rates. In addition, under
energy-rich conditions (at higher specific growth rate
values), an increase in the protein turnover rate improves
the robustness of cells to potential perturbations in food
supply because cells are able to regulate their protein content
more rapidly compared with slow-growth conditions.
The latter indicates that protein turnover is a dynamic
parameter, which may be dependent on the availability of
free energy. Knowledge regarding the mechanisms by which
micro-organisms regulate their protein stability may provide
an opportunity to design strains with both reduced turnover
and higher production yields.
This work shows that protein turnover rates display similar
trends to specific growth and metabolic flux rates, and
therefore can be influenced by a change in environmental
conditions. In addition, the observed increase in biomass
yield with an increase of specific growth rate for L. lactis is
mainly caused by decreased ATP demand for maintenance
where the ATP cost for protein turnover is one of the most
important factors.
ACKNOWLEDGEMENTS
The authors would like to thank Kadri Aller for her help in
experimental preparation, Kaspar Valgepea for critical reading of
the manuscript and the PRIDE Team for depositing the proteome
data. Anonymous reviewers are acknowledged for their valuable
feedback. Financial support was provided by the European Regional
Development Fund project EU29994, SA Archimedes through the
project 3.2.0701.11-0018 and the Ministry of Education, Estonia,
through grant SF0140090s08.
REFERENCES
Adamberg, K., Seiman, A. & Vilu, R. (2012). Increased biomass yield
of Lactococcus lactis by reduced overconsumption of amino acids and
increased catalytic activities of enzymes. PLoS ONE 7, e48223.
Arike, L., Valgepea, K., Peil, L., Nahku, R., Adamberg, K. & Vilu, R.
(2012). Comparison and applications of label-free absolute proteome
quantification methods on Escherichia coli. J Proteomics 75, 5437–
5448.
Bahey-El-Din, M. (2012). Lactococcus lactis-based vaccines from
laboratory bench to human use: an overview. Vaccine 30, 685–690.
Belle, A., Tanay, A., Bitincka, L., Shamir, R. & O’Shea, E. K. (2006).
Quantification of protein half-lives in the budding yeast proteome.
Proc Natl Acad Sci U S A 103, 13004–13009.
Bermúdez-Humarán, L. G., Kharrat, P., Chatel, J.-M. & Langella, P.
(2011). Lactococci and lactobacilli as mucosal delivery vectors for
therapeutic proteins and DNA vaccines. Microb Cell Fact 10 (Suppl 1),
S4.
Bermúdez-Humarán, L. G., Aubry, C., Motta, J.-P., Deraison, C.,
Steidler, L., Vergnolle, N., Chatel, J.-M. & Langella, P. (2013).
CONCLUSION
Engineering lactococci and lactobacilli for human health. Curr Opin
Microbiol 16, 278–283.
Knowledge of the rate of protein turnover plays an
important role in the interdisciplinary field of systems
biology. Large-scale quantitative datasets are emerging that
include information about the abundance and degradation
rates of individual proteins, their associated mRNA levels,
together with metabolite concentrations and pool sizes, and
kinetic parameters for individual enzymes. This knowledge
will help us better understand how cells actually regulate
their metabolism and adjust their associated fitness
mechanisms. Genome-scale modelling at both the singlecell and population levels which aims to describe cellular
growth aids in our quest to construct efficient microbial cell
factories. Specific growth rate-dependent data of protein
degradation rates together with energetic burden calculations demonstrate the complexity of both cellular regulatory
mechanisms and how they choose their fitness under conditions where the intracellular energy demand fluctuates.
Bolotin, A., Wincker, P., Mauger, S., Jaillon, O., Malarme, K.,
Weissenbach, J., Ehrlich, S. D. & Sorokin, A. (2001). The complete
1510
genome sequence of the lactic acid bacterium Lactococcus lactis ssp.
lactis IL1403. Genome Res 11, 731–753.
Buonaguro, L., Wang, E., Tornesello, M. L., Buonaguro, F. M. &
Marincola, F. M. (2011). Systems biology applied to vaccine and
immunotherapy development. BMC Syst Biol 5, 146.
Burton, R. E., Siddiqui, S. M., Kim, Y. I., Baker, T. A. & Sauer, R. T.
(2001). Effects of protein stability and structure on substrate processing
by the ClpXP unfolding and degradation machine. EMBO J 20, 3092–
3100.
Canelas, A. B., Harrison, N., Fazio, A., Zhang, J., Pitkänen, J.-P., van den
Brink, J., Bakker, B. M., Bogner, L., Bouwman, J. & other authors (2010).
Integrated multilaboratory systems biology reveals differences in protein
metabolism between two reference yeast strains. Nat Commun 1, 145.
Cargile, B. J., Bundy, J. L., Grunden, A. M. & Stephenson, J. L., Jr
(2004). Synthesis/degradation ratio mass spectrometry for measuring
relative dynamic protein turnover. Anal Chem 76, 86–97.
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Mon, 31 Jul 2017 19:51:01
Microbiology 160
Energetic burden of protein turnover
Cox, J. & Mann, M. (2008). MaxQuant enables high peptide
identification rates, individualized p.p.b.-range mass accuracies and
proteome-wide protein quantification. Nat Biotechnol 26, 1367–1372.
Dressaire, C., Redon, E., Milhem, H., Besse, P., Loubière, P. &
Cocaign-Bousquet, M. (2008). Growth rate regulated genes and their
O’Sullivan, E. & Condon, S. (1999). Relationship between
acid tolerance, cytoplasmic pH, and ATP and H+-ATPase levels in
chemostat cultures of Lactococcus lactis. Appl Environ Microbiol 65,
2287–2293.
Oh, E., Lu, M., Park, C., Park, C., Oh, H. B., Lee, S. Y. & Lee, J. (2011).
wide involvement in the Lactococcus lactis stress responses. BMC
Genomics 9, 343.
Dynamic modeling of lactic acid fermentation metabolism with
Lactococcus lactis. J Microbiol Biotechnol 21, 162–169.
Dressaire, C., Picard, F., Redon, E., Loubière, P., Queinnec, I., Girbal,
L. & Cocaign-Bousquet, M. (2013). Role of mRNA stability during
Ong, S.-E., Blagoev, B., Kratchmarova, I., Kristensen, D. B., Steen,
H., Pandey, A. & Mann, M. (2002). Stable isotope labeling by amino
bacterial adaptation. PLoS ONE 8, e59059.
acids in cell culture, SILAC, as a simple and accurate approach to
expression proteomics. Mol Cell Proteomics 1, 376–386.
Gerth, U., Kock, H., Kusters, I., Michalik, S., Switzer, R. L. & Hecker,
M. (2008). Clp-dependent proteolysis down-regulates central meta-
bolic pathways in glucose-starved Bacillus subtilis. J Bacteriol 190,
321–331.
González-Ramos, D., van den Broek, M., van Maris, A. J., Pronk, J. T. &
Daran, J.-M. G. (2013). Genome-scale analyses of butanol tolerance in
Saccharomyces cerevisiae reveal an essential role of protein degradation.
Biotechnol Biofuels 6, 48.
Helbig, A. O., Daran-Lapujade, P., van Maris, A. J., de Hulster, E. A.,
de Ridder, D., Pronk, J. T., Heck, A. J. & Slijper, M. (2011). The
diversity of protein turnover and abundance under nitrogen-limited
steady-state conditions in Saccharomyces cerevisiae. Mol Biosyst 7,
3316–3326.
Hong, K.-K., Hou, J., Shoaie, S., Nielsen, J. & Bordel, S. (2012).
Dynamic 13C-labeling experiments prove important differences in
protein turnover rate between two Saccharomyces cerevisiae strains.
FEMS Yeast Res 12, 741–747.
Orij, R., Urbanus, M. L., Vizeacoumar, F. J., Giaever, G., Boone, C.,
Nislow, C., Brul, S. & Smits, G. J. (2012). Genome-wide analysis of
intracellular pH reveals quantitative control of cell division rate by
pHc in Saccharomyces cerevisiae. Genome Biol 13, R80.
Pirt, S. J. (1965). The maintenance energy of bacteria in growing
cultures. Proc R Soc Lond B Biol Sci 163, 224–231.
Pratt, J. M., Petty, J., Riba-Garcia, I., Robertson, D. H., Gaskell, S. J.,
Oliver, S. G. & Beynon, R. J. (2002). Dynamics of protein turnover, a
missing dimension in proteomics. Mol Cell Proteomics 1, 579–591.
Rappsilber, J., Mann, M. & Ishihama, Y. (2007). Protocol for micro-
purification, enrichment, pre-fractionation and storage of peptides
for proteomics using StageTips. Nat Protoc 2, 1896–1906.
Russell, J. B. & Cook, G. M. (1995). Energetics of bacterial growth:
balance of anabolic and catabolic reactions. Microbiol Rev 59, 48–62.
Hughes, C. & Krijgsveld, J. (2012). Developments in quantitative
Schwanhäusser, B., Busse, D., Li, N., Dittmar, G., Schuchhardt, J.,
Wolf, J., Chen, W. & Selbach, M. (2011). Global quantification of
mass spectrometry for the analysis of proteome dynamics. Trends
Biotechnol 30, 668–676.
Schwanhäusser, B., Wolf, J., Selbach, M. & Busse, D. (2013).
mammalian gene expression control. Nature 473, 337–342.
Karr, J. R., Sanghvi, J. C., Macklin, D. N., Gutschow, M. V., Jacobs,
J. M., Bolival, B., Jr, Assad-Garcia, N., Glass, J. I. & Covert, M. W.
(2012). A whole-cell computational model predicts phenotype from
Synthesis and degradation jointly determine the responsiveness of the
cellular proteome. BioEssays 35, 597–601.
genotype. Cell 150, 389–401.
Metabolic Engineering: Principles and Methodologies. New York:
Academic Press.
Khmelinskii, A., Keller, P. J., Bartosik, A., Meurer, M., Barry, J. D.,
Mardin, B. R., Kaufmann, A., Trautmann, S., Wachsmuth, M. & other
authors (2012). Tandem fluorescent protein timers for in vivo
analysis of protein dynamics. Nat Biotechnol 30, 708–714.
Lahtvee, P.-J., Adamberg, K., Arike, L., Nahku, R., Aller, K. & Vilu, R.
(2011). Multi-omics approach to study the growth efficiency and
amino acid metabolism in Lactococcus lactis at various specific growth
rates. Microb Cell Fact 10, 12.
Lowry, O. H., Rosebrough, N. J., Farr, A. L. & Randall, R. J. (1951).
Protein measurement with the Folin phenol reagent. J Biol Chem 193,
265–275.
Maier, T., Schmidt, A., Güell, M., Kühner, S., Gavin, A.-C., Aebersold,
R. & Serrano, L. (2011). Quantification of mRNA and protein and
integration with protein turnover in a bacterium. Mol Syst Biol 7, 511.
Martin, S. F., Munagapati, V. S., Salvo-Chirnside, E., Kerr, L. E. & Le
Bihan, T. (2012). Proteome turnover in the green alga Ostreococcus
tauri by time course 15N metabolic labeling mass spectrometry.
J Proteome Res 11, 476–486.
Mehmeti, I., Faergestad, E. M., Bekker, M., Snipen, L., Nes, I. F. & Holo,
H. (2012). Growth rate-dependent control in Enterococcus faecalis:
Stephanopoulos, G., Aristidou, A. A. & Nielsen, J. H. (1998).
Stouthamer, A. H. (1973). A theoretical study on the amount of
ATP required for synthesis of microbial cell material. Antonie van
Leeuwenhoek 39, 545–565.
Taymaz-Nikerel, H., Borujeni, A. E., Verheijen, P. J. T., Heijnen, J. J. &
van Gulik, W. M. (2010). Genome-derived minimal metabolic models
for Escherichia coli MG1655 with estimated in vivo respiratory ATP
stoichiometry. Biotechnol Bioeng 107, 369–381.
Thompson, J. (1978). In vivo regulation of glycolysis and characterization of sugar: phosphotransferase systems in Streptococcus lactis.
J Bacteriol 136, 465–476.
Trötschel, C., Albaum, S. P., Wolff, D., Schröder, S., Goesmann,
A., Nattkemper, T. W. & Poetsch, A. (2012). Protein turnover
quantification in a multilabeling approach: from data calculation to
evaluation. Mol Cell Proteomics 11, 512–526.
Trötschel, C., Albaum, S. P. & Poetsch, A. (2013). Proteome turnover
in bacteria: current status for Corynebacterium glutamicum and related
bacteria. Microb Biotechnol 6, 708–719.
Valgepea, K., Adamberg, K., Seiman, A. & Vilu, R. (2013). Escherichia
effects on the transcriptome and proteome, and strong regulation of
lactate dehydrogenase. Appl Environ Microbiol 78, 170–176.
coli achieves faster growth by increasing catalytic and translation rates
of proteins. Mol Biosyst 9, 2344–2358.
Menon, A. S. & Goldberg, A. L. (1987). Binding of nucleotides to the
van Bodegom, P. (2007). Microbial maintenance: a critical review on
its quantification. Microb Ecol 53, 513–523.
ATP-dependent protease La from Escherichia coli. J Biol Chem 262,
14921–14928.
Neves, A. R., Pool, W. A., Kok, J., Kuipers, O. P. & Santos, H. (2005).
Vizcaı́no, J. A., Côté, R. G., Csordas, A., Dianes, J. A., Fabregat, A.,
Foster, J. M., Griss, J., Alpi, E., Birim, M. & other authors (2013). The
Overview on sugar metabolism and its control in Lactococcus lactis –
the input from in vivo NMR. FEMS Microbiol Rev 29, 531–554.
PRoteomics IDEntifications (PRIDE) database and associated tools:
status in 2013. Nucleic Acids Res 41 (D1), D1063–D1069.
http://mic.sgmjournals.org
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Mon, 31 Jul 2017 19:51:01
1511
P.-J. Lahtvee and others
Wiśniewski, J. R., Zougman, A., Nagaraj, N. & Mann, M. (2009).
Universal sample preparation method for proteome analysis. Nat
Methods 6, 359–362.
Wiśniewski, J. R., Ostasiewicz, P., Duś, K., Zielińska, D. F., Gnad, F. &
Mann, M. (2012). Extensive quantitative remodeling of the proteome
between normal colon tissue and adenocarcinoma. Mol Syst Biol 8, 611.
1512
Wodke, J. A., Puchałka, J., Lluch-Senar, M., Marcos, J., Yus, E.,
Godinho, M., Gutiérrez-Gallego, R., dos Santos, V. A., Serrano,
L. & other authors (2013). Dissecting the energy metabolism in
Mycoplasma pneumoniae through genome-scale metabolic modeling.
Mol Syst Biol 9, 653.
Edited by: J. Kok
Downloaded from www.microbiologyresearch.org by
IP: 88.99.165.207
On: Mon, 31 Jul 2017 19:51:01
Microbiology 160