HM Medical Clinic

Take Clomid is contraindicated in the presence of cysts in the ovaries, liver and kidney failure, the presence of pituitary tumors or genital organs clomid cost Before using the product you should consult with a specialist and to get acquainted with the instructions approved by the manufacturer.

Cialis ne doit pas être prise à tous. Il est important que cialis en ligne est prescrit par un médecin, bien se familiariser avec les antécédents médicaux du patient. Ich habe Probleme mit schnellen Montage. Lesen Sie Testberichte Nahm wie cialis rezeptfrei 30 Minuten vor dem Sex, ohne Erfolg. Beginn der Arbeiten nach 4 Stunden, links ein Freund ein trauriges Ja, und Schwanz in sich selbst nicht ausstehen, wenn es keinen Wunsch ist.


Microbiology (2002), 148, 1003–1013
Printed in Great Britain Metabolic engineering of lactic acid bacteria,
the combined approach : kinetic modelling,
metabolic control and experimental analysis

Marcel H. N. Hoefnagel,1,2 Marjo J. C. Starrenburg,1,3 Dirk E. Martens,1,2Jeroen Hugenholtz,1,3 Michiel Kleerebezem,1,3 Iris I. Van Swam,1,3Roger Bongers,1,3 Hans V. Westerhoff4 and Jacky L. Snoep4,5 Author for correspondence : Marcel H. N. Hoefnagel. Tel : j31 317 483435. Fax : j31 317 482237.
e-mail : marcel.hoefnagel!
1, 2 Wageningen Centre for Everyone who has ever tried to radically change metabolic fluxes knows that it
Food Sciences1 and Food is often harder to determine which enzymes have to be modified than it is to
and BioprocessEngineering Group,2 actually implement these changes. In the more traditional genetic engineering
Wageningen University, approaches 'bottle-necks' are pinpointed using qualitative, intuitive
PO Box 8129, 6700 approaches, but the alleviation of suspected 'rate-limiting' steps has not often
EV Wageningen,The Netherlands been successful. Here the authors demonstrate that a model of pyruvate
distribution in Lactococcus lactis
based on enzyme kinetics in combination with
3 NIZO Food Research, PO Box 20, 6710 BA, Ede, metabolic control analysis clearly indicates the key control points in the flux to
acetoin and diacetyl, important flavour compounds. The model presented here
4 BioCentrum Amsterdam, (available at http :// showed that the enzymes
Dept of Molecular Cell with the greatest effect on this flux resided outside the acetolactate synthase
branch itself. Experiments confirmed the predictions of the model, i.e.
University, De Boelelaan1087, NL-1081 knocking out lactate dehydrogenase and overexpressing NADH oxidase
increased the flux through the acetolactate synthase branch from 0 to 75 % of
measured product formation rates.
5 Dept of Biochemistry, University ofStellenbosch, Private bagX1, Matieland 7602, Keywords : Metabolic control analysis, in silico modelling, Lactococcus lactis, pyruvate
Stellenbosch, South unable to come with unexpected strategies. Severaltheoretical frameworks, such as biological system Micro-organisms are not naturally optimized for maxi- theory (Savageau, 1991), metabolic control analysis mal production rates of biotechnologically important (MCA) (Kacser & Burns, 1973) and metabolic design compounds. Flavour compounds, such as diacetyl and (Kholodenko et al., 1998), have been developed to acetaldehyde, are often secondary metabolites and are analyse multi-enzyme systems predictively and quan- produced in insignificant amounts with respect to total titatively. In MCA, so-called flux-control coefficients, carbon metabolism (Hugenholtz & Starrenburg, 1992).
which quantify the importance of an enzyme for the Redirecting the carbon flux towards such products magnitude of a flux, are defined as the percentage without disturbing the overall cell physiology is a change in the flux caused by a 1 % modulation of the complex task. Metabolism is a highly branched system enzyme activity. Consequently, MCA points at the connected not only via carbon metabolites but also via enzymes that will have the largest effect on the desired redox [NAD(P)H] and Gibbs free energy (ATP) carriers.
flux upon a very small change in any enzyme activity, An intuitive approach to the optimization of fluxes but it also has the disadvantage that it does not deal with towards the desired end products is not infallible in such the more relevant larger changes. Upon modulation of a complex network and such an approach alone is an enzyme with a high control coefficient, kineticmodelling can be used to integrate the system and to study the effect of a more substantial perturbation in Abbreviations : ALS, acetolactate synthase ; LDH, L-lactate dehydrogen-
ase ; MCA, metabolic control analysis ; NOX, NADH oxidase ; (abbreviations
silico. Because larger increases in flux are important in used in rate reactions and equations are defined in Table 1).
bioengineering, we propose to use an integrated ap- The GenBank accession number for the sequence reported in this paper is proach between kinetic modelling, MCA and exper- imentation to come to a rational strategy for genetic 0002-5221 # 2002 SGM M. H. N. Hoefnagel and others engineering. Here we illustrate this approach for the onies displaying erythromycin resistance and tetracycline optimization of the metabolic flux through the aceto- sensitivity should have been the result of a one-step double- lactate synthase (ALS) branch in lactic acid bacteria.
cross-over event on both flanking regions of the ldh gene of thelas operon and should have contained the desired ldh : : ery Lactic acid bacteria are used in milk fermentation and replacement mutation. The anticipated genetic organization the major product from this process is lactate. For some of the ldh : : ery mutant candidates was verified by PCR and dairy products, like butter, diacetyl produced by Lacto- Southern analysis ; the resulting ldh : : ery variant of L. lactis coccus lactis is an important flavour component. Di- NZ9000 was designated NZ9010.
acetyl is produced in only small amounts and various Cloning of the lactococcal nox gene and construction of the
groups have tried to optimize its production (Monnet et nox overexpression plasmid. The nox gene of Streptococcus
al., 1994b ; Platteeuw et al., 1995 ; Swindell et al., 1996 ; mutans, derived from pNZ2600 (Lopez de Felipe et al., 1998), Lopez de Felipe & Hugenholtz, 1999). Their experimen- was used as a heterologous probe in Southern blotting tal strategies have involved the overexpression and experiments to clone the NADH-oxidase(NOX)-encoding deletion of genes intuited to be important for the regu- gene of L. lactis MG1363. A 1n8 kb fragment of EcoRI-digested lation of the carbon fluxes in L. lactis. Here, we use a chromosomal DNA of L. lactis MG1363 hybridized with this more rational combination strategy consisting of (i) a probe. The 1n8 kb fragment was cloned into EcoRI-digested pUC19 and colonies containing the hybridizing fragment were detailed kinetic model of the branches around pyruvate selected by colony blotting, again using the nox gene of S.
metabolism, (ii) MCA and (iii) experiments to describe mutans as a probe. Sequence analysis of the 1n8 kb fragment the various mutations and to illustrate the use of revealed that it contained the 3h end of a gene that had high (i) and (ii) in a more successful metabolic engineering similarity with the nox gene of S. mutans, putatively rep- resenting the 3h end of the lactococcal nox gene. Moreover, downstream of the 3h end of this putative nox gene, ORFs were found that shared a high level of similarity (43n8% at the protein sequence level) with the ssb gene of B. subtilis, Construction of an ldh deficient mutant of L. lactis NZ9000.
encoding a putative single-stranded DNA-binding protein To construct an ldh : : ery replacement variant of L. lactis (Meyer & Laine, 1990). Following the ssb gene, the 5h end of NZ9000 (Kuipers et al., 1997), a plasmid was constructed that the previously cloned groES operon of L. lactis was found allowed the direct selection of the desired mutant obtained by (Kim & Batt, 1993).
the occurrence of a double cross-over event. The upstreamflanking region of the ldh gene of the las operon was amplified The 5h end of the lactococcal nox gene was amplified by PCR by PCR using MG1363 genomic DNA as a template and the using L. lactis MG1363 chromosomal DNA as a template and the fully degenerated primer NOX2F (5h-ACNGGNACY- GC-3h) and LDHREV-6 (5h-ATCAGCCATGGTTTTCTTT-
GAYCANGCNGCNGGYATHGC-3h ; NlA C G T, Yl AATTCC-3h) (EcoRI and NcoI sites in bold, respectively).
C T and HlA C T), based on the previously determined Similarly, the downstream flanking region of the ldh gene of N-terminal sequence of the lactococcal NOX protein (Lopez the las operon was amplified by PCR using the primers de Felipe & Hugenholtz, 2001), combined with primer ACC-3h), which was based on the 3h region of the cloned nox sequence, and was extended with a PstI-generating clamp TAC-3h) (NcoI, BamHI and XbaI sites in bold, respectively).
sequence (indicated in bold). The approximately 400 bp Both the up- and downstream fragments, of approximately fragment obtained was cloned into pGEMT (Promega) and 1n1 kb each, were cloned as EcoRI–NcoI and NcoI–XbaI sequenced ; the sequence revealed that the fragment contained fragments, respectively, into similarly digested pUCNCO.
the 5h region of nox. This fragment (the NcoI–PstI fragment The sequence of the cloned fragment was verified by auto- from the pGEMT clone) was used as a probe in the cloning of mated sequence analysis with an ALF DNA sequencer a hybridizing 1n4 kb EcoRI L. lactis MG1363 chromosomal (Pharmacia Biotech). Sequence reactions were performed fragment. Sequence analysis of this fragment revealed that it according to the manufacturer's protocols, using the autoread contained the entire 5h region of the lactococcal nox gene.
sequencing kit and fluorescein-labelled universal M13 primers.
Upstream of nox, an ORF encoding a hypothetical protein Both of the ldh flanking regions were re-isolated as EcoRI– displaying local similarity with a phenylalanine tRNA ligase NcoI (5h) and NcoI–XbaI (3h) fragments and cloned into (78 % identity with the YdjD protein of L. lactis IL1403) was EcoRI–XbaI-digested pUC19 in a three-point ligation. A found. The sequence of the combined EcoRI fragments, BamHI–XbaI fragment of pUC19ERY (Kuipers et al., 1995), containing the entire noxE sequence (93 % identity with the containing the erythromycin-resistance gene, was introduced NoxE protein of L. lactis IL1403), has been deposited in the into the BamHI restriction site of the resulting pUCAB plasmid GenBank database under accession number AY046926.
after the cohesive ends had been filled using the Klenow The water-forming NOX-encoding nox gene of L. lactis was fragment of DNA polymerase I of Escherichia coli. The amplified by PCR using L. lactis MG1363 chromosomal DNA resulting plasmid, in which the orientation of the erythromycin as a template and the primers NOXF (5h-CGTACCAT-
gene was the same as the preceding pyk gene, was designated GGAAATCGTAGTTATCGGTAC-3h) and NOXR (5h-CG-
pUCAeryB. Finally, a SwaI–Ecl136II fragment of pGhost8 (Chopin et al., 1984), containing the tetracycline-resistance PCR fragment obtained was cloned as a NcoI–XbaI fragment gene, was cloned into the XbaI site of pUCAeryB after filling (restriction sites were introduced into the primers and are in the XbaI cohesive ends using Klenow. The resulting indicated in bold) into similarly digested pNZ8048 (de Ruyter plasmid, designated pUCAeryBTc, was transformed into et al., 1996 ; Kuipers et al., 1997). The resulting plasmid was competent L. lactis NZ9000 cells and erythromycin-resistant designated pNZ2610 and contained the lactococcal nox gene colonies were selected ; the resulting transformants were translationally fused to the nisA promoter. Transcription of analysed for tetracycline resistance by replica plating. Col- nox in this construct is dependent on the activity of the nisin- Metabolic engineering of Lactococcus lactis Table 1. Mathematical symbols and abbreviations for equations and kinetic models used
in this study
Equilibrium constant Inhibition constant Affinity constant Predicted enzyme activities Maximal enzyme activities under saturating substrate and activator conditions, and in the absence of inhibitors Acetyl coenzyme A Acetoin dehydrogenase Alcohol dehydrogenase Acetolactate synthase Lactate dehydrogenase Non-enzymic acetolactate decarboxylase Inorganic phosphate Pyruvate dehydrogenase inducible nisA promoter, whereas translation of the nox about 2 and 5n5 h post-inoculation. The flux distribution was transcript depends on the nisA-derived RBS (de Ruyter et al., calculated during this exponential (pseudo steady-state) phase.
Analysis of fermentation products. Glucose, lactate, acetate,
Control strain. L. lactis NZ9000(pNZ8048) was used as a
formate, ethanol, acetoin and 2,3-butanediol were analysed by control during this study.
HPLC, as described previously (Starrenburg & Hugenholtz,1991). In contrast to the wild-type, the carbon recovery for the Fermentation. Cultures were grown at 30 mC in M17 medium
mutant strains was incomplete (between 70 and 85 % re- (Merck) supplemented with 1 % (w v) glucose. Chloram- covery). This may have been due to activity of the pentose phenicol and erythromycin were used at 10 and 5 µg ml−", phosphate pathway, which leads to CO respectively. Nisin was used at 1 ng ml−". A 1 l bioreactor # release. However, this can not be confirmed as this pathway was not monitored (Applikon Dependable Instruments) was inoculated with cells in this study.
from an overnight culture to an initial OD'!! of about 0n1 in 700 ml medium. A pH of 6n5 was maintained by the addition Growth and enzyme assays. The OD'!! was determined and
of 2 M NaOH and the stirrer speed was set at 500 r.p.m. Air corrected for the optical density of the growth medium. When was bubbled through with a flow rate of 520 ml min−". The the OD'!! value was too high (0n75) the sample was diluted batch cultures used showed exponential growth between with medium.
M. H. N. Hoefnagel and others Table 2. Rate equations used in this study
The reaction numbers correspond to those depicted in Fig. 1.
V +0 GLC 1i0 NAD 1i0 ADP 1 01j GLC j PYR 1i01j NAD jNADH1i01j ADP j ATP 1 01j PYR j LAC 1i01jNADHj NAD 1 1i0 PYR 1i0 NAD 1i0 COA 1 01j PYR 1i01j NAD jNADH1i01j COA jACCOA1 01jACCOAj P jACPjCOAj((ACCOAiP) (K iK )j((ACPiCAO) (K iK ))1 01j ACP j AC 1i01j ADP j ATP 1 01j NAD jNADH1i01jACCOAj COA jACALj ACALiCOA 1 01j NAD jNADH1i01jACALjETOH1 V +0 PYR 1i01k ACLAC 1i00 PYR jACLAC1N−"1 01j0 PYR jACLAC1N1 V +0ACLAC1 Metabolic engineering of Lactococcus lactis Table 2 (cont.)
V +0 ACET 1 01j ACET j BUT 1i01jNADHj NAD 1 V +0 ATP 1N 1j0 ATP 1N V +0 NADHiO 1 01jNADHj NAD 1i01j O 1 Fig. 1. Reactions included in the model to
describe the distribution of carbon from
pyruvate in L. lactis. Numbers in circles
indicate the following enzymes or steps : 1,
‘ lumped ' glycolysis ; 2, LDH ; 3, pyruvate
dehydrogenase ; 4, phosphotransacetylase ;
5, acetate kinase ; 6, acetaldehyde dehydro-
genase ; 7, alcohol dehydrogenase ; 8, ALS ;
9, acetolactate decarboxylase ; 10, acetoin
efflux ; 11, acetoin dehydrogenase ; 12, AT-
Pase ; 13, NOX ; 14, non-enzymic acetolactate
decarboxylation ; 15, pyruvate formate lyase,
which is considered not to be active under
aerobic conditions (see Table 3) ; 16, chemical
conversion to diacetyl, not included in the
model. Substrates and products that were
clamped in the model are indicated in italics.
-Lactate dehydrogenase (LDH) activity was determined by (Platteeuw et al., 1995 ; Lopez de Felipe et al., 1998), but they the method of Hillier & Jago (1982) and the NOX and ALS were not performed in an isogenic background and did not activities were determined according to Lopez de Felipe contain precise measurements of catabolic fluxes. Instead, et al. (1998) and Platteeuw et al. (1995), respectively. Protein they consisted merely of end-point determinations of product concentrations were determined according to the Bradford concentrations. This made them unsuitable for a proper method (Bradford, 1976), with BSA as a standard. Preliminary demonstration of the rational engineering procedure we experiments along these lines have been published previously propose here.
M. H. N. Hoefnagel and others Table 3. Kinetic parameters used for the enzymes in reactions 1–15
Unless indicated otherwise, the data are from L. lactis. The numbers of the reactions correspondwith those depicted in Fig. 1 and the rate equations in Table 2.
Obtained by ‘ fitting ' Set in this study Set in this study Obtained by ‘ fitting' Obtained by ‘ fitting' Obtained by ‘ fitting' Obtained by ‘ fitting' This study and in agreement with Hugenholtz & Starrenburg (1992) Thauer et al. (1977) Crow & Pritchard (1977) Snoep et al. (1992a) Snoep et al. (1992b) Azotobacter ; Bresters et al. (1975) Estimated from Snoep et al. (1993) Abbe et al. (1982) Clostridium kluyverii ; Henkin & Abeles (1976) Thauer et al. (1977) S. mutans ; Abbe et al. (1982) E. coli ; Fox & Roseman (1986) Thauer et al. (1977) S. mutans ; Abbe et al. (1982) E. coli ; Shone & Fromm (1981) Thauer et al. (1977) Cachon & Divies (1993) Zymomonas ; Wills et al. (1981) Thauer et al. (1977) Metabolic engineering of Lactococcus lactis Table 3 (cont.)
Snoep et al. (1992a) ; Platteeuw et al. (1995) Snoep et al. (1992a) Guessed ; introduced to represent product sensitivity of the reaction Platteeuw et al. (1995) Swindell et al. (1996) Monnet et al. (1994a) Petit et al. (1989) Gibson et al. (1991) Strecker & Harary (1954) Enterobacter ; Carballo et al. (1991) Obtained by ‘ fitting' Set in this study Set in this study Wild-type ; NOX overexpressed S. faecalis ; Schmidt et al. (1986) Monnet et al. (1994b) * Units for each parameter are : V+, mmol (l internal vol.)−" min−" ; K , mM ; K , mM ; k, min−".
† The kinetic parameters of the glycolytic step (reaction 1) and the ATP consumption (reaction 12) wereestimated by fitting the model to the glucose fluxes of oxygenated and anaerobically [160 and 190 mmol(l internal vol.)−" min−", respectively] grown wild-type L. lactis (strain NZ9000). For this estimation andMCA we used  (Mendes, 1993). The kinetic parameters of the other enzymes were derived fromthe literature. Under aerobic conditions pyruvate formate lyase is considered not to be active (Abbe etal., 1982) ; therefore, reaction 15 was not included in the model.
The kinetic model. A set of ordinary differential equations was
Most of the enzymes were modelled using a reversible used to describe the time dependence of the metabolite Michaelis–Menten equation with noncompeting substrate– concentrations. Table 1 gives definitions of the mathematical product couples. When the specific enzyme has been charac- symbols and abbreviations used in the equations and kinetic terized with respect to the kinetic type, then this type was used.
models in this study. Please note that the flux of glycolysis is (reaction 3) an inhibition by a high NADH : NAD ratio has been described and these kinetics have been included in the equation. For v (reaction 4) the equation was derived from Henkin & Abeles (1976). For v reversible Hill equation, as derived by Hofmeyr & Cornish- Bowden (1997), was used. For v irreversible Hill equation, sensitive to the ATP : ADP ratio, was used. For reaction 9 and 12 kinetics for an irreversible reacion were used. Simple Michaelis–Menten kinetics were used for reaction 10. Reaction 14 represents non-enzymic decarboxylation of acetolactate. The rate equations shown in Table 2 were used.
The concentrations of the following metabolites were assumed to be constant : glucose, 15 mM ; lactate, 1 mM ; inorganic phosphate, 10 mM ; ethanol, 1 mM ; butanediol, 0n01 mM; oxygen, 0n2 mM; and acetate, 1 mM. The sums of M. H. N. Hoefnagel and others (Table 4). This qualifies as a high flux-control coefficient, j[acetyl coenzyme A] were assumed to be constant at the as flux-control coefficients always sum up to 1. However, respective values of 5, 10 and 1 mM.
flux-control coefficients can be negative and the com- Kashket & Wilson (1973) found that in L. lactis the internal plete MCA analysis shows how relevant this is for this cell volume is 1n5 µl (mg dry weight)−". We assumed that branched pathway – the highest flux-control coefficients protein makes up 50 % of the dry weight, as it does in E. coli for the acetolactate branch reside in enzymes outside (Neidhardt & Umbarger, 1996), and that an OD'!! of 1 is this branch! (Namely LDH and NOX ; Table 4). This equal to 0n445 mg dry weight ml−".
corresponds to an early MCA observation for branched Kinetic parameters. The kinetic parameters used for the enzymes pathways, in which the fluxes over the different branches in reactions 1–15 (Fig. 1) and the rate equations in Table 2 can differ by orders of magnitude (Kacser, 1983). With be found in Table 3.
hindsight, this can be understood also intuitively : aminor decrease in any branch carrying a major flux from pyruvate might increase the pyruvate concentrationmore than proportionately provided that the concen- Modelling the pyruvate branches in L. lactis
tration exceeded the apparent Michaelis constant forthat major pathway. This could then result in a more We first constructed a kinetic model describing the flux than proportional increase in flux through the ALS distribution over the pyruvate branches in L. lactis (Fig.
1). Glycolysis was modelled as a single irreversible step(Fig. 1, 1) with product inhibition ; the ATP-consuming MCA suggested that ALS should have a substantial reactions (Fig. 1, 12) were grouped in a single module.
control on the ALS flux, implying that a doubling of the The kinetic parameters of these two modules were activity of the enzyme could at most double the very estimated on the basis of ‘ fits ' to in vivo measurements small flux towards acetoin. However, we were seeking of fluxes (see Methods and Table 3). In contrast to these much more substantial increases. Moreover, the over- unstructured modules, all of the kinetic steps in the expression of enzymes with high control coefficients is pyruvate branches were modelled in detail. In vitro often not successful because the control coefficient tends kinetic data from the literature were used as input. A to decrease strongly as the enzyme is overexpressed (i.e.
detailed description of the model is given in Methods the enzyme stops being limiting). A calculation with the and Table 3. The model prediction of the flux dis- kinetic model (not shown) showed that increasing the tribution in the wild-type strain (white lettering) as amount of ALS by a factor of 20 [the overexpression compared to the experimental data (black lettering) is level found in Platteeuw et al. (1995)] increased the ALS shown in Fig. 2(a). The model predicted an essentially flux to only 0n02% of the glucose flux. Indeed, Platteeuw (97 %) homolactic fermentation, which is in agreement et al. (1995) demonstrated that in anaerobically cultured with experimental observations (Fig. 2), where between L. lactis (MG5267) overexpression of ALS did not 90 and 100 % of the glucose is converted to lactate increase the production of acetoin and 2,3-butanediol.
(Hugenholtz & Starrenburg, 1992 ; Platteeuw et al., Therefore, we decided that overexpression of ALS was 1995). The flux through the acetolactate branch was not going to be a successful strategy to increase the flux predicted to be less than 0n1% of the glucose influx, in through ALS.
agreement with what was observed experimentally (Fig.
2a). Thus, in the wild-type strain only a negligible fluxwas directed towards the branch of interest and this canbe understood on the basis of the known enzyme kinetics LDH knockout
(i.e. the model gives a good description of this con- LDH has the highest (negative) flux-control coefficient dition). What strategy can now be followed to optimize (k2n3 ; Table 4). Thus, a 1 % reduction in LDH activity the flux through this branch? should lead to a 2n3% increase in the flux through the MCA quantifies the importance of each of the enzymes acetolactate branch. After MCA has indicated the best in controlling the flux through the acetolactate branch.
candidate for genetic engineering, modulations sub- Thus, the enzyme with the highest flux-control co- stantially exceeding 1 % can be test-run in the kinetic efficient is the enzyme that will have the biggest effect model. Such a test-run is an important part of the genetic upon a small modulation around its wild-type level, engineering strategy, since MCA is defined at the original suggesting that MCA is a good approach for genetic steady state and the size of the optimal perturbation engineering. But is it? might be limited due to kinetic restrictions in the system.
For example, if the acetolactate dehydrogenase andpyruvate dehydrogenase complex branches have only a MCA versus the intuitive approach
low capacity these branches might not be able to absorb In an intuitive approach ALS would be the enzyme of all of the carbon flow from lactate upon a knockout choice for genetic engineering, as it is the first dedicated mutation in LDH. Such an effect was observed in the enzyme in the branch and it is far from equilibrium. But model calculations, where pyruvate and NADH accumu- what does MCA say? At first sight it might seem to say lated upon a deletion in LDH and the glycolytic flux was the same, i.e. in the wild-type situation the ALS has a severely inhibited (Fig. 2b). Such a major perturbation flux-control coefficient of 1n0 for the flux to acetoin of primary metabolism is bound to lead to a strong Metabolic engineering of Lactococcus lactis Fig. 2. Flux distribution over the pyruvate
branches in L. lactis. Experimental results are
shown in black letters on a white back-
ground, whereas model predictions are
shown in white letters on a black back-
ground. The fluxes are given in mmol (l
internal vol.)−1 min−1 [multiply by 290 to
obtain µmol (mg protein)−1 min−1]. (a) Wild-
type ; (b) Ldh− mutant ; (c) NOX over- expression mutant ; (d) Ldh− and NOX over-expression mutant. As can be seen from thesum of the fluxes, the carbon balance is notcomplete in all strains. In (b, d) pyruvatethat accumulated in the medium accountedfor up to 15 % of the carbon flux.
Table 4. Control coefficients of the different steps/enzymes on the flux through ALS as
predicted by the model for the wild-type and three mutants
The only steps included are those that have a control coefficient of at least 0n1 in any of themutants. NOX++, NOX overexpression.
Acetoin decarboxylase Pyruvate dehydogenase reduction in growth rate and possibly to deleterious about 20 % of the glucose would be converted via the effects in Ldh− strains. Therefore, even though all of the ALS branch if NOX was 40-fold overexpressed, whereas remaining flux was directed into the ALS branch, this in the experimental set-up 13 % was converted to acetoin modulation seemed ill-advised (Fig. 2b).
Experimentally, the overall carbon flux was reduced by11 % but this reduction was not as severe as the model The strongest effect was expected when the LDH had predicted. The major product formed was acetoin knockout mutant was combined with NOX over- (50 % of the measured product formation rates, Fig. 2), expression. The model predicted that, under these which was in agreement with the model prediction.
conditions, 92 % of the pyruvate would be convertedvia the acetolactate branch, but with a much higherglycolytic flux when compared to the situation in which Overexpression of NOX
only LDH was deleted (Fig. 2d). Indeed, in the ex- In addition to LDH, NOX also had a high (positive) perimental situation acetoin was the main product control coefficient (Table 4). The model predicted that (75 %) of the recovered carbon.
M. H. N. Hoefnagel and others optimize. On the basis of the control coefficients one canthen design a genetic engineering strategy, which can In this study we have illustrated the combined use of again be tested in the kinetic model. This latter testing is MCA, kinetic modelling and experimentation to in- important, as one will want to make large changes in crease dramatically the flux through a pathway that is of enzyme concentrations in order to attain biotechno- biotechnological relevance. The predicted increase in logically interesting improvements. If indeed the per- flux was substantial enough to be of great industrial turbation leads to an appreciable improvement of the interest and was confirmed experimentally.
production strain in silico one can then move to theexperimental process of genetic engineering.
The model
All in all, the model we used was not perfect. Therefore, L. lactis is an ideal organism for modelling purposes, as our finding that it was very useful in the engineering it is well-studied, genetically accessible and can be strategy, and that it even came close to predicting its cultured readily. Furthermore, its anabolic reactions can results quantitatively, may reflect the phenomenon that be simulated as being separate from the catabolic metabolic fluxes have limited sensitivities to many reactions with respect to carbon metabolism : when L.
kinetic parameters. More generally, it suggests that the lactis is cultured on rich medium almost all the carbon frequent adage that biotechnology is not yet ready for (95 %) from the free-energy source (e.g. glucose) is the application of mathematical models may be much recovered as external products (e.g. lactate) (Novak too pessimistic.
et al., 1998). This simplifies the modelling processconsiderably. The only link between anabolism and catabolism that was considered in this study was ATPand this could be modelled as a single step ; all ATP- For the genetic engineering of the various L. lactis strains the consuming reactions in anabolism were grouped into a financial support of the EU-GEMOLAB project (BIO4-CT98- single module. Biomass formation from glucose was 0118) is gratefully acknowledged.
assumed to be, overall, a redox-neutral process ; thehomolactic fermentation in the wild-type indicates that catabolism itself is redox neutral.
Abbe, K., Takahashi, S. & Yamada, T. (1982). Involvement of
The accuracy of the model depends largely on the kinetic oxygen-sensitive pyruvate formate-lyase in mixed-acid fermen- data on which it was built. These data were taken from tation by Streptococcus mutans under strictly anaerobic con- the literature and were determined by several different ditions. J Bacteriol 152, 175–182.
groups working with different organisms and they were Bradford, M. M. (1976). A rapid and sensitive method for the
not always determined under physiological conditions.
quantitation of microgram quantities of protein utilizing the Also, sometimes no data were available for L. lactis. In principle of protein–dye binding. Anal Biochem 72, 248–254.
these instances, data either from another streptococcal Bresters, T. W., De Kok, A. & Veeger, C. (1975). The pyruvate-
species or from other bacteria were used (see Table 3).
dehydrogenase complex from Azotobacter vinelandii. 2. Regu- Another limitation of the model is that a constant level lation of the activity. Eur J Biochem 59, 347–353.
of gene expression was assumed, i.e. the enzyme levels Cachon, R. & Divies, C. (1993). Localization of Lactococcus lactis
that were measured in the wild-type strain were assumed ssp. lactis bv. diacetylactis in alginate gel beads affects biomass not to change during the genetic manipulation steps.
density and synthesis of several enzymes involved in lactose and
citrate metabolism. Biotechnol Tech 7, 453–456. wcfs.html and it can be run from Carballo, J., Martin, R., Bernardo, A. & Gonzalez, J. (1991).
Purification, characterization and some properties of diacetyl
this site. As well as the standard simulations, the site (acetoin) reductase from Enterobacter aerogenes. Eur J Biochem also allows the user to set various parameters and to test all sorts of mutants.
Chopin, A., Chopin, M. C., Moillo-Batt, A. & Langella, P. (1984).
Two plasmid-determined restriction and modification systems in
Designing the optimal strain : reducing trial and error
Streptococcus lactis. Plasmid 11, 260–263.
Indeed, if the kinetic parameters for the system under Crow, V. L. & Pritchard, G. G. (1977). Fructose-1,6-diphosphate-
study are available then the construction of a kinetic activated -lactate dehydrogenase from Streptococcus lactis: model is relatively simple and quick. Even without all of kinetic properties and factors affecting activation. J Bacteriol 131,
the kinetic parameters available, one can sometimesmake simplifications that allow for the construction of a de Ruyter, P. G., Kuipers, O. P. & de Vos, W. M. (1996). Controlled
model that can be useful. For instance, here we have gene expression systems for Lactococcus lactis with the food-
grade inducer nisin. Appl Environ Microbiol 62, 3662–3667.
focused on the flux distribution over the pyruvatebranches and have grouped all of the reactions outside Fox, D. K. & Roseman, S. (1986). Isolation and characterization of
these branches, thereby avoiding modelling all of the homogeneous acetate kinase from Salmonella typhimurium and
Escherichia coli. J Biol Chem 261, 13487–13497.
glycolytic steps in detail.
Gibson, T. D., Parker, S. M. & Woodward, J. R. (1991). Purification
Once a kinetic model has been constructed it is simple to and characterization of diacetyl reductase from chicken liver and calculate the control coefficients on the flux (or on the Streptococcus lactis and enzymic determination of diacetyl and concentration of a metabolite) that one wishes to diketones. Enzyme Microb Technol 13, 171–178.
Metabolic engineering of Lactococcus lactis Henkin, J. & Abeles, R. H. (1976). Evidence against an acyl-enzyme
Neidhardt, F. C. & Umbarger, H. E. (1996). Chemical composition
intermediate in the reaction catalyzed by clostridial phospho- of Escherichia coli. In Escherichia coli and Salmonella, Cellular transacetylase. Biochemistry 15, 3472–3479.
and Molecular Biology, 2nd edn, pp. 13–16. Edited by F. C.
Neidhardt and others. Washington, DC : American Society for Hillier, A. J. & Jago, G. R. (1982). -Lactate dehydrogenase, FDP-
activated, from Streptococcus cremoris. Methods Enzymol 89 Pt
, 362–367.
Novak, L., Cocaign-Bousquet, M., Lindley, N. D. & Loubiere, P.
Cometabolism sugar–amino acids in Lactococcus lactis.
Hofmeyr, J. H. & Cornish-Bowden, A. (1997). The reversible Hill
Lait 78, 17–22.
equation : how to incorporate cooperative enzymes into metabolicmodels. Comput Appl Biosci Petit, C., Vilchez, F. & Marczak, R. (1989). Influence of citrate on
the diacetyl and acetoin production by fully grown cells of Hugenholtz, J. & Starrenburg, M. J. C. (1992). Diacetyl production
Streptococcus lactis ssp. diacetylactis. Curr Microbiol 19,
by different strains of Lactococcus lactis ssp. lactis var. di- acetylactis and Leuconostoc spp. Appl Microbiol Biotechnol 38,
Platteeuw, C., Hugenholtz, J., Starrenburg, M. J. C., Van Alen-
Boerrigter, I. & De Vos, W. M. (1995). Metabolic engineering of
Kacser, H. (1983). The control of enzyme systems in vivo : elasticity
Lactococcus lactis : influence of the overproduction of alpha- analysis of the steady state. Biochem Soc Trans 11, 35–40.
acetolactate synthase in strains deficient in lactate dehydrogenase Kacser, H. & Burns, J. A. (1973). The control of flux. Symp Soc Exp
as a function of culture conditions. Appl Environ Microbiol 61,
Savageau, M. A. (1991). Biochemical systems theory : operational
Kashket, E. R. & Wilson, T. H. (1973). Proton-coupled accumu-
differences among variant representations and their significance.
lation of galactoside in Streptococcus lactis 7962. Proc Natl Acad J Theor Biol 151, 509–530.
Sci U S A 70, 2866–2869.
Schmidt, H. L., Stocklein, W., Danzer, J., Kirch, P. & Limbach, B.
Kholodenko, B. N., Cascante, M., Hoek, J. B., Westerhoff, H. V. &
(1986). Isolation and properties of an H
Schwaber, J. (1998). Metabolic design : how to engineer a living
oxidase from Streptococcus faecalis. Eur J Biochem 156, 149–155.
cell to desired metabolite concentrations and fluxes. Biotechnol Shone, C. C. & Fromm, H. J. (1981). Steady-state and pre-steady-
Bioeng 59, 239–247.
state kinetics of coenzyme A linked aldehyde dehydrogenase from Kim, S. G. & Batt, C. A. (1993). Cloning and sequencing of the
Escherichia coli. Biochemistry 20, 7494–7501.
Lactococcus lactis subsp. lactis groESL operon. Gene 127,
Snoep, J. L., Teixeira De Mattos, M. J., Starrenburg, M. J. C. &
Hugenholtz, J. (1992a). Isolation, characterization, and physio-
Kuipers, O. P., Beerthuyzen, M. M., de Ruyter, P. G., Luesink,
logical role of the pyruvate dehydrogenase complex and alpha- E. J. & de Vos, W. M. (1995). Autoregulation of nisin biosynthesis
acetolactate synthase of Lactococcus lactis ssp. lactis bv.
in Lactococcus lactis by signal transduction. J Biol Chem 270,
diacetylactis. J Bacteriol 174, 4838–4841.
Snoep, J. L., Westphal, A. H., Benen, J.-A. E., Teixeira De Mattos,
Kuipers, O. P., de Ruyter, P. G. G. A., Kleerebezem, M. & De Vos,
M. J., Neijssel, O. M. & De Kok, A. (1992b). Isolation and
W. M. (1997). Controlled overproduction of proteins by lactic
characterization of the pyruvate dehydrogenase complex of acid bacteria. Trends Biotechnol 15, 135–140.
anaerobically grown Enterococcus faecalis NCTC 775. Eur J
203, 245–250.
Lopez de Felipe, F. & Hugenholtz, J. (1999). Pyruvate flux
distribution in NADH-oxidase-overproducing Lactococcus lactis
Snoep, J. L., De Graef, M. R., Westphal, A. H., De Kok, A., Teixeira
strain as a function of culture conditions. FEMS Microbiol Lett De Mattos, M. J. & Neijssel, O. M. (1993). Differences in sensitivity
to NADH of purified pyruvate dehydrogenase complexes of
Enterococcus faecalis, Lactococcus lactis and Escherichia coli : Lopez de Felipe, F. & Hugenholtz, J. (2001). Purification and
implications for their activity in vivo. FEMS Microbiol Lett 114,
characterization of the water forming NADH oxidase of Lacto- coccus lactis. Int Dairy J 11, 37–44.
Starrenburg, M. J. C. & Hugenholtz, J. (1991). Citrate fermen-
Lopez de Felipe, F., Kleerebezem, M., De Vos, W. M. &
tation by Lactococcus and Leuconostoc spp. Appl Environ Hugenholtz, J. (1998). Cofactor engineering : a novel approach to
Microbiol 57, 3535–3540.
metabolic engineering in Lactococcus lactis by controlled ex- Strecker, H. J. & Harary, I. (1954). Bacterial butylene glycol
pression of NADH oxidase. J Bacteriol 180, 3804–3808.
dehydrogenase and diacetyl reductase. J Biol Chem 211, 263–270.
Mendes, P. (1993).  : a software package for modelling the
Swindell, S. R., Benson, K. H., Griffin, H. G., Renault, P., Ehrlich,
dynamics, steady states and control of biochemical and other S. D. & Gasson, M. J. (1996). Genetic manipulation of the pathway
systems. Comput Appl Biosci 9, 563–571.
for diacetyl metabolism in Lactococcus lactis. Appl Environ Meyer, R. R. & Laine, P. S. (1990). The single-stranded DNA-
Microbiol 62, 2641–2643.
binding protein of Escherichia coli. Microbiol Rev 54, 342–380.
Thauer, R. K., Jungermann, K. & Decker, K. (1977). Energy
conservation in chemotrophic anaerobic bacteria. Bacteriol Rev
Monnet, C., Phalip, V., Schmitt, P. & Divies, C. (1994a). Com-
parison of alpha-acetolactate synthase and alpha-acetolactatedecarboxylase in Lactococcus spp. and Leuconostoc spp. Bio- Wills, C., Kratofil, P., Londo, D. & Martin, T. (1981). Charac-
technol Lett 16, 257–262.
terization of the two alcohol dehydrogenases of Zymomonas
. Arch Biochem Biophys 210, 775–785.
Monnet, C., Schmitt, P. & Divies, C. (1994b). Diacetyl production
in milk by an alpha-acetolactic acid accumulating strain of
Lactococcus lactis ssp. lactis biovar diacetylactis. J Dairy Sci 77,
Received 5 September 2001 ; revised 5 November 2001 ; accepted 20 November 2001.


The role of pirfenidone in the treatment of idiopathic pulmonary fibrosis

Cottin Respiratory Research 2013, 14(Suppl 1):S5 The role of pirfenidone in the treatment ofidiopathic pulmonary fibrosis From AIR: Advancing IPF Research. Working together to translate IPF research into practiceBerlin, Germany. 4-5 November 2011 Idiopathic pulmonary fibrosis (IPF) is a progressive disease, with a median survival time of 2–5 years. The search foreffective treatment has involved numerous clinical trials of investigational agents without significant success.However, in 2011, pirfenidone was the first drug to be approved for the treatment of IPF in Europe. Four keyclinical trials supported the efficacy and tolerability of pirfenidone.In the two recently published Phase III CAPACITY trials evaluating pirfenidone (studies 004 and 006), patients withmild-to-moderate IPF were treated with pirfenidone or placebo. Study 004 and pooled analysis of primaryendpoint data from both studies showed that pirfenidone significantly reduced decline in percent-predicted forcedvital capacity (FVC) compared with placebo (p<0.005). Evidence of beneficial effects of pirfenidone treatment wasalso observed with regard to several secondary endpoints. Pirfenidone was generally well tolerated, with the mostcommon side effects being gastrointestinal and photosensitivity. Data from the RECAP extension phase of theCAPACITY studies, where patients were treated with pirfenidone for up to three years, further support themanageable tolerability profile of pirfenidone. The efficacy data, coupled with long-term safety data, provide furtherevidence of a clinically-meaningful treatment effect with pirfenidone in patients with IPF.