Process Biochemistry 45(2010)1334–1341
Contents lists available at ScienceDirect
Process
Biochemistry
j o u r n a l h o m e p a g e :w w w.e l s e v i e r.c o m /l o c a t e /p r o c b i
o
In?uence of speci?c growth rate over the secretory expression of recombinant potato carboxypeptidase inhibitor in fed-batch cultures of Escherichia coli
Juan-Miguel Puertas a ,?,Jordi Ruiz a ,Mónica Rodríguez de la Vega b ,Julia Lorenzo b ,Glòria Caminal a ,Glòria González a
a
Unitat de Biocatàlisi Aplicada associada al IQAC,Departament d’Enginyeria Química,Escola d’Enginyeria,Universitat Autònoma de Barcelona,Edi?ci Q,08193Bellaterra (Barcelona),Spain b
Institut de Biotecnologia i de Biomedicina,Universitat Autònoma de Barcelona,08193Bellaterra,Spain
a r t i c l e i n f o Article history:
Received 16February 2010
Received in revised form 28April 2010Accepted 30April 2010
Keywords:
Potato carboxypeptidase inhibitor Fed-batch cultivation
Secretory expression in E.coli
a b s t r a c t
A high cell density cultivation protocol was developed for the secretory production of potato carboxypep-tidase inhibitor (PCI)in Escherichia coli .The strain BW25113(pIMAM3)was cultured in fed-batch mode employing minimal media and an exponential feed pro?le where the speci?c growth rate was ?xed by limitation of the fed carbon source (glycerol).Plasmid loss rates were found to be proportional to the speci?c growth rate.Distribution of PCI along the cell compartments and the culture media was also dependent on the ?xed growth rate.When speci?c growth rate was kept at =0.10h ?1,1.4g PCI L ?1were obtained when adding the product present in periplasmic extracts and supernatant fractions,with a 50%of the total expressed protein recovered from the extracellular medium.This constituted a 1.2-fold increase compared to growth at =0.15h ?1,and 2.0-fold compared to =0.25h ?https://www.doczj.com/doc/656850535.html,st,a cell perme-abilization treatment with Triton X-100and glycine was employed to direct most of the product to the culture media,achieving over 81%of extracellular PCI.Overall,our results point out that production yields of secretory proteins in fed-batch cultures of E.coli can be improved by means of process variables,with applications to the production of small disul?de-bridged proteins.Overall,our results point out that control of the speci?c growth rate is a successful strategy to improve the production yields of secretory expression in fed-batch cultures of E.coli ,with applications to the production of small disul?de-bridged proteins.
?2010Elsevier Ltd.All rights reserved.
1.Introduction
Secretory expression of heterologous proteins in Escherichia coli has a number of advantages over more common cytosolic expres-sion.First,secretion of the recombinant product is attractive form a downstream processing stand-point,since no cell-disruption steps are needed and contamination with other proteins is reduced both in the periplasm and culture media [1,2].Secondly,the formation of disul?de bridges is actively catalyzed in the periplasmic space [3,4].Also,for proteins that are toxic to the host,secretion may palliate their detrimental effect over culture growth [2].
Several proteins have been successfully produced in the periplasmic space and culture supernatants of high cell density cul-tures of E.coli [5,6],but since the capacity of the bacterial secretion machinery is limited and there are several factors that affect protein
?Corresponding author.Tel.:+34935814795;fax:+34935812032.
E-mail addresses:juanmiguel.puertas@uab.es ,juanmiguel.puertas@uab.cat
(J.-M.Puertas),jordi.ruiz.franco@uab.cat (J.Ruiz),julia.lorenzo@uab.cat (J.Lorenzo),gloria.caminal@uab.cat (G.Caminal),gloria.gonzalez@uab.cat (G.González).expression and translocation [7,8],achieving high protein yields of protein exported through the inner membrane can be a complex task.In this sense,it has been proven that translational and translo-cation levels have to be properly coupled to reach a state where most of the expressed heterologous protein is secreted [9,10].This can be achieved by manipulations of genetic parameters like the promoter strength [11],the nature of the signal sequence [12,13]or the plasmid copy number [14],but optimization of the culture protocols is also necessary.Previous studies show the in?uence of culture media composition,growth kinetics,induction moment and temperature over secretory protein yields [1,2,5].
Potato carboxypeptidase inhibitor (PCI)is a small protein nat-urally occurring in leafs and stems of Solanum tubesorum [15].Composed by 39residues and three disul?de bridges,it has poten-tial biomedical applications given its proven antitumoral properties [16,17].PCI had previously been produced in E.coli using the pIN-III-ompA-derived plasmid pIMAM3,which allows for the translocation of the protein to the periplasmic space where formation of its disul?de bonds was successfully achieved and the active form could be recovered from culture supernatants [18,19].Excretion of PCI out of the cell envelope is probably favored by its small
1359-5113/$–see front matter ?2010Elsevier Ltd.All rights reserved.doi:10.1016/j.procbio.2010.04.024
J.-M.Puertas et al./Process Biochemistry45(2010)1334–13411335
size and compact structure.A fed-batch procedure had previ-ously been designed for the overexpression of PCI in high cell density cultures in semi-complex media,but relatively low lev-els of biomass(15g DCW L?1)were achieved,and the process was not automated,with feedstock additions not responding to any monitored variable.The aim of this work was to design a robust, automated and repeatable fed-batch process at bench-top level in order to increase the production of biologically active PCI by maximizing both the biomass concentrations and the expression-secretion of the inhibitor.Since it was observed that the speci?c growth rate( )had a major in?uence in the amounts of excreted PCI,a series of fermentations at different?xed growth rates were carried out.The dynamics of the PCI concentration pro?les in the cytosol,periplasmic space and culture media was analyzed in order to identify and overcome the bottlenecks in the secretory produc-tion of this protein.
2.Materials and methods
All reagents were purchased from Sigma–Aldrich(St.Louis,MO,USA)under otherwise stated.
2.1.Strains and plasmid
E.coli strain MC1061(hsdR2hsdM+hsdS+araD139 (ara-leu)7697 (lac)X74 galE15galK16rpsL(StrR)mcrA mcrB1)and plasmid pIMAM3were used in previous works[18,19].BW25113( (araD-araB)567, lacZ4787(::rrnB-3),lambda?,rph-1, (rhaD-rhaB)568,hsdR514)was obtained from the Coli Genetics Stock Center at Yale.
2.2.Shake-?ask cultivation conditions
For shake-?ask experiments,either LB media or MDE media supplemented with100?g mL?1ampicillin were used.The composition of LB was,per liter: 10g peptone(Difco),5g yeast extract(Difco)and10g NaCl;whereas the com-position of MDE media was,per liter:5g glucose or glycerol,11.9g K2HPO4,2.4g KH2PO4,1.8g NaCl,3.0g(NH4)2SO4,0.11g MgSO4·7H2O,0.01g FeCl3,0.03g thi-amine and0.72mL of trace elements solution.Trace element solution composition was,per liter:1.44g CaCl2·2H2O,42mg AlCl3·6H2O,50mg ZnSO4·7H2O,160mg of CoCl2·6H2O,1.6g CuSO4,10mg H3BO3,1.42g MnCl2·4H2O,10mg NiCl2·H2O and 20mg of Na2MoO4·H2O.
Seed cultures were prepared in50mL culture tubes by inoculating10mL of LB broth with a single colony from a fresh transformation plate,followed by incubation overnight at37?C and200rpm in an orbital shaker.Shake-?ask cultures were typ-ically prepared in500mL shake-?asks using1mL of the seed culture to inoculate 100mL of LB or MDE media,which were then incubated under the same conditions as seed cultures.To induce protein expression,IPTG from a100mM stock was asep-tically added to the desired?nal concentration.After induction,cells were allowed to grow into stationary phase for7–8h.
2.3.Bioreactor cultivation conditions
Fed-batch cultivation experiments were carried out using a2L jar and a stan-dard Biostat B?digital control unit.A?ux of1.5vvm of air was injected trough the fermentor to satisfy the respiratory needs of the cultured strain.Additions of15% (w/v)NH4OH were made to keep pH at a set point of7.00.Temperature was set at37?C.Dissolved oxygen levels were kept at60%of the saturation concentration by means of the stirring speed and/or by mixing the inlet gas with pure oxygen in increasing proportions.
Seed cultures of the strain of interest were grown in LB media as described previously for shake-?ask experiments.Inocula cultures were prepared in500mL shake-?asks by adding5mL of seed culture into95mL of fresh MDE media,then incubating at37?C and200rpm in an orbital shaker.Once these cultures reached OD600nm of1–1.2,80mL of them were added into the fermentation jar containing 720mL of MDF media.MDF contains per liter:20.0g glucose,4.5g yeast extract, 2.0g NaCl,4.1g(NH4)2SO4,13.2g K2HPO4,2.6g KH2PO4,0.5g MgSO4,0.03g FeCl3, 25mL of trace elements solution and100mg ampicillin.Once glucose was depleted, the feeding part of the fermentation was started with the addition of FS feeding solution,which contained per liter:450g glycerol,9.6g MgSO4,0.5g FeCl3,0.5g CaCl2·2H2O,0.3g thiamine,32mL trace elements solution and500mg ampicillin. Addition of the feedstock was done according to an open loop method described before in previous works(20).This exponential feeding protocol allows for the con-trol of the speci?c growth rate of the bacterial culture by limitation of the carbon source(21).No source of phosphate was included in the feedstock in order to avoid the precipitation of Ca3(PO4)2and Fe(PO4);however two punctual additions of a concentrated phosphate solution were done,each equivalent to5g of PO43?.2.4.Biomass and metabolites analyses
Bacterial growth was followed by optical density measurements at600nm using a spectrophotometer(KONTRON Uvicon941plus Spectrophotometer).Optical den-sity was correlated to dry cell weight through a calibration curve constructed by standard methods[21].Evaluation of plasmid stability was accomplished by plat-ing properly diluted amounts of culture samples on plain LB-agar plates and on then LB-agar plates supplemented with100?g mL?1ampicillin.The colony count on plain plates represented the total cells,while the count on the antibiotic contain-ing plates stood for plasmid bearing cells.For the determination of glycerol,glucose, phosphate,ammonium and acetic acid,1mL samples of culture were centrifuged at9000×g for3min in a tabletop centrifuge(Haereus).The supernatant was then ?ltered through a0.22?m syringe?lter(Millipore)and puri?ed by HPLC(Hewlett-Packard1050)on an Aminex HPX-87H column(Biorad),with H2SO415mM as the mobile phase at a?ow rate of0.60mL min?1.Analysis was done with an IR detec-tor(Hewlett-Packard1047)at room temperature.Phosphates and ammonia were determined using commercial colorimetric kits(Hach Lange).
2.5.Cell fractionation and culture supernatant preparation
Before cell fractionation,OD600nm of each culture was determined to determine the broth volume from were2.4mg DCW could be isolated.The sample volume was centrifuged at3000×g,4?C for10min,and periplasmic extracts were obtained by osmotic shock as described elsewhere[22].The resulting pellet containing spheroplasts was resuspended in300?L PBS buffer and sonicated to release sol-uble cytoplasmic proteins using a Vibracell?model VC50(Sonics&Materials).Cell lysates were centrifuged at12,000×g,4?C for15min to recover the soluble cytosolic contents.Culture supernatants were separated from the bacterial pellet by cen-trifugation at5000×g,4?C for10min,followed by?ltration with a syringe-driven
0.22?m?lter device(Millipore).Clear supernatants(40mL)were then loaded on
a SepPak C18(1g)Reverse Phase column(Waters),previously equilibrated with acetonitrile and rinsed with ultrapure water.Columns were washed with4mL10% acetonitrile and4mL ultrapure water previous to elution with4mL of30%iso-propanol.When needed,supernatants were concentrated using Amicon or Minicon centrifugal?lter devices(Millipore).
2.6.Protein electrophoresis and estimation of protein content
Total protein in samples content was assessed in triplicate using a commercial kit for the Bradford assay(Biorad)with Bovine Serum Albumin(BSA)as a standard. Separation and visualization of proteins over electrophoresis gels was carried out using12%Bis-Tris gels from the Novex system(Invitrogen)using MES-SDS as run-ning buffer.Upon staining with colloidal Coomassie[23],the bands of interest were quanti?ed by gel densitometry(Kodak Digital Science);this type of quanti?cation was mainly used for the estimation of pre-PCI in the cytosol.
2.7.Reverse phase HPLC quanti?cation
Reverse phase liquid chromatography was employed to separate and quantify active PCI in periplasmic and supernatant fractions.An Ultimate300HPLC system (Dionex)and a C18cartridge(Waters)were employed,using a sample volume of 100?L containing1%TFA.Elution of PCI was done over a gradient of acetonitrile (pH=1.00)from20%to80%.Using standards of puri?ed protein,a calibration curve was constructed to estimate the concentration of active PCI in the injected samples.
2.8.Enzymatic assay
The inhibitory activity of PCI samples was determined using a commercial kit (Sigma–Aldrich).This kit allows to measure the activity of carboxypeptidase A(CPA), as well as the screening of inhibitors of this enzyme,as described elsewhere[24]. Non-induced culture samples of BW25113or MC1061(pIMAM3)were used as blanks.
3.Results
3.1.Shake-?ask preliminary experiments
https://www.doczj.com/doc/656850535.html,parison of expression between the strains MC1061and BW25113
As previously mentioned,the expression system MC1061 (pIMAM3)had successfully been used for the production of PCI in both shake-?ask and high-density cultures in complex and semi-complex media[19].However,MC1061is a leucine auxotroph and hence its culture in de?ned media requires the addition of this amino acid.In small scale cultures this did not constitute a rel-evant inconvenient,but it was found to be a major handicap for high-density cultivation,since the amino acid needs can be several
1336J.-M.Puertas et al./Process Biochemistry45(2010)
1334–1341
Fig.1.Effect of IPTG concentration on extracellular PCI yields in shake-?ask cultures of BW25113(pIMAM3).
folds higher and moreover leucine exhibits a low solubility in aque-ous media.Addition of other amino acids source commonly used in culture media(such as hydrolyzed casein extracts,peptone or yeast extract)is incompatible with the fed-batch cultivation pro-tocol used in our research group which is carried out in minimal medium for several reasons:lower cost,higher repeatability of the results,and possibility of controlling the speci?c growth rate by limitation of the availability of a single carbon source.
To overcome this problem,pIMAM3was transformed into the prototrophic strain BW25113.We then compared the expression of PCI using both strains in three different culture media:LB,de?ned medium with glycerol as the carbon source(MDE,glycerol),and de?ned medium with glucose as carbon source(MDE,glucose). After5h of induction,culture supernatants were puri?ed and ana-lyzed.Final yields of extracellular PCI were determined and are presented in Table1.
Overall,BW25113exhibits a more desirable behavior for the expression of PCI.It can also be observed that higher concentra-tions of extracellular product were obtained in MDE when glycerol was employed as the carbon source.Increased secretion in de?ned media compared to complex media has already been described elsewhere[7].
3.1.2.Effect of inducer concentration over cell growth and
glycerol consumption
Concentration of inducer is known to be another important variable in the production of recombinant proteins[25–27].A series of experiments was conducted to study the expression of PCI in BW25113(pIMAM3)under different concentrations of IPTG. Results are depicted in Fig.1,which shows that concentrations over 250?M IPTG do not increase notably the PCI yield.Also,the effect of the induction over cell metabolism was studied by conducting parallel growth experiments with and without addition of IPTG. Induction of PCI expression resulted in a decrease of the speci?c growth rate and?nal biomass value,with a growth arrest hap-pening after approximately two duplications post-induction.It was also determined that the glycerol/biomass yield during the induc-tion phase was around20%higher than that of the non-induced growth(S/X=2.8g glycerol g?1DCW versus S/X=3.4g glycerol g?1 DCW).These parameters are necessary for the control of the growth rate in the following fed-batch cultures according to the protocol described in Section2.
3.2.Bioreactor experiments
3.2.1.Development of a fed-batch protocol for the high cell
density cultivation of BW25113(pIMAM3)
In previous studies of our research group a minimal balanced media was designed and successfully used for the growth of E. coli to high cell densities.An exponential feed pro?le was used to control the speci?c growth rate by limitation of the carbon source while maintaining appropriate levels of other main nutri-ents like nitrogen and phosphorous[20,21,25].Thus,the?rst step to scale up the production of PCI from shake-?ask to bioreactor scale was carrying out a non-induced fed-batch culture of BW25113 (pIMAM3)according to this protocol,in order to validate its ef?-cacy for this particular expression system.It is well known that control of growth rates is crucial to avoid accumulation of undesir-able metabolites,mainly acetic acid[28,29],and also has a proven impact on the production yield of recombinant proteins[30].
Glucose was chosen for the discontinuous stage of the fermen-tation because of its reduced cost and faster growth kinetics of the strain over this substrate.This phase of the fermentation lasted around13.5h,when glucose was depleted,which was evidenced by a sudden rise of the dissolved oxygen concentration and also an increase of the pH to the maximum tolerated by the control loop (pH=7.05).The biomass achieved at this point was6.6g DCW L?1 (OD600nm=21).The feeding was then started,programming the parameters of the control software to maintain a speci?c growth rate of =0.25h?1.Glycerol,instead of glucose,was employed in the feedstock for the fed-batch since growth over the?rst substrate yielded higher amounts of extracellular PCI in shake-?ask cultures (see Table1);and it is in the fed-batch stage where induction takes place.
Fig.2depicts the evolution of biomass,glucose,glycerol,ammo-nium,phosphate,acetic acid and dissolved oxygen along the fermentation.Levels of this last metabolite were maximal at the end of the batch stage(1.25g L?1),but were always under critical levels.Concentrations of ammonium and phosphate followed the expected trend throughout the process,never reaching inhibitory or limiting values.The fed-batch stage ended after8.5h,when it was not possible to keep the dissolved oxygen concentration over10%of saturation even supplying1.5vvm of pure oxygen. The point of oxygen exhaustion will be considered as the end-point for all the fed-batch cultures.A?nal biomass concentration of58.2g DCW L?1(OD600nm≈180)was reached.Plasmid stability under these conditions was satisfactory,with over90%of the cells exhibiting ampicillin resistance at the end of the process.
3.2.2.Production of PCI in fed-batch cultures
Once the ef?cacy of the fed-batch protocol for the high cell den-sity cultivation of BW25113(pIMAM3)was tested and the growth limit was determined,a series of induced cultures were performed
Table1
Speci?c growth rates and extracellular PCI yields obtained in shake-?ask cultivation of BW25113(pIMAM3)and MC1061(pIMAM3)in LB and de?ned media.
Strain LB MDE-glycerol MDE-glucose
max(h?1)P/X(mg PCI g?1DCW) max(h?1)P/X(mg PCI g?1DCW) max(h?1)P/X(mg PCI g?1DCW)
BW251130.64±0.05 2.4±0.070.41±0.03 2.9±0.050.53±0.06 1.6±0.08
MC10610.62±0.06 1.7±0.040.48±0.05 2.0±0.060.52±0.04 1.4±0.05
J.-M.Puertas et al./Process Biochemistry45(2010)1334–1341
1337
Fig.2.Time-course of fed-batch culture of BW25113(pIMAM3)without induction.Non-induced culture at?xed speci?c growth rate of =0.25h?1.Concentrations of biomass concentration( ),glucose( ),glycerol( ),ammonium(?),phosphate( ),acetate( )and dissolved oxygen(?)concentrations are shown.A continuous vertical line is used to distinguish the batch and fed-batch stages.Punctual additions of phosphate are depicted with an arrow.
in which the only variable parameter was the speci?c growth rate ?xed by the feed pro?le.The choice of the induction moment was done based on the number of duplications post-induction observed on shake-?ask cultures as a rough estimate of the effect of PCI expression over cell growth.Bearing in mind a previous induc-tion strategy for the expression of a recombinant aldolase under the control of a mild promoter[21],it was reasoned that an early induction might have prevented the system from reaching the max-imum biomass determined in the non-induced culture,and a late induction would have shortened the induction time,limiting pro-tein expression.Following this rationale,it was chosen to induce at a biomass concentration of around15g DCW L?1(OD600nm=45)so that after two duplications a?nal biomass of nearly60g DCW L?1 could still be reached,as for the case of the non-induced culture. Fig.3A–C presents the results of these fermentations in terms of cell concentration,speci?c growth rate and active PCI concentration in the culture media.
It can be observed that growth patterns in all three fermenta-tions followed the expected kinetics(i.e.the speci?c growth rate was kept at the?xed value with an average7%deviation from the set point);reaching a maximum biomass very similar to that of the non-induced culture.Also,concentrations of carbon source, ammonium,phosphate and acetic acid were kept on the desired levels(data not shown).Particularly,the concentration of glyc-erol remained close to zero during the fed-batch phase of all three processes,validating the strategy of growth control by substrate limitation.
However,the amount of extracellular recombinant product depended highly on the?xed growth rate,i.e.growth at =0.1h?1 resulted in2.3-and12.1-fold increases in the concentration of extracellular PCI,respect to the fermentations carried out at =0.15h?1and =0.25h?1,respectively.In order to investigate the causes of these observations,we compared plasmid stability pro?les and distribution of protein along the different cellular com-partments for the three cases.
3.2.3.Effect of the?xed speci?c growth rate on plasmid loss rates
Fig.4presents the plots of plasmid stability over the induction phase of the fermentations,and also that of the non-induced cul-ture.It can be seen that plasmid loss rates are proportional to the ?xed growth rate.It is known that in high cell density cultures ampicillin can be readily degraded,with the consequent loss of selective pressure[31,32].This would promote an increased ten-dency for the cells to lose pIMAM3.Dependence of plasmid loss rates with growth rates has already been reported in several studies [33–35].It is important to recall that the end of the fermentations is imposed by the oxygen-transfer limitation,and in the case of the growth at =0.25h?1this boundary is reached earlier,and with a smaller proportion of PCI-productive biomass,resulting in a decrease in both the induction time and the process productivity.
3.2.
4.Effect of the?xed speci?c growth rate on protein expression,secretion and excretion
As it previously shown,levels of excreted PCI varied signi?-cantly with the growth rate at which production was carried out. In order to study this correlation,the intracellular contents of PCI were assessed;both as presecretory or immature protein in the cytosol and in active form in the periplasm.Fig.5presents the?nal localization of PCI in the cytosolic extracts,periplasmic extracts and culture supernatants for the production fermentations.Sur-prisingly,a signi?cant amount of the recombinant product was found in the periplasmic space,a fact that contrasts with previous studies where the protein was exclusively found in the extracellu-lar milieu[19].Secretion of PCI to the periplasmic space and culture media represents a76%of the total protein at =0.25h?1,87%of the total protein at =0.15h?1and96%at =0.10h?1,as illus-trated in Fig.6.Regarding the proportion of excreted PCI,even more striking differences were found:extracellular PCI constituted6.9% of the total protein at =0.25h?1,32.2%at =0.15h?1and50.0% at =0.10h?1.
In order to point out the effect of the speci?c growth rate over the distribution of the recombinant product,Fig.7shows the time evo-lution of the PCI amount(mg)in the different cell compartments.It must be noted that for a given time point of the fermentation,the quantities of pre-PCI in the cytosol and PCI in the periplasm depend on the total biomass at that instant,since these concentrations are cell-bound(in contrast to the concentrations of extracellular PCI). To be able to compare the total PCI produced in each compartment, we calculated the quantity of pre-PCI in the cytosol and PCI in the periplasm using the concentrations in the corresponding extracts and the protocol described in Section2(i.e.knowing that300?L of cytosolic and periplasmic extracts are prepared from2.4mg DCW).
1338J.-M.Puertas et al./Process Biochemistry 45(2010)
1334–1341
Fig.3.Time-course of induced fed-batch cultures of BW25113(pIMAM3)at dif-ferent ?xed speci?c growth rates.(A) =0.25h ?1,(B) =0.15h ?1,(C) =0.10h ?1.Biomass (?)and PCI ( )concentrations in culture supernatants are shown.A solid vertical line is used to distinguish the batch and fed-batch stages;whereas the induction moment is depicted with a dashed vertical line.
Regarding extracellular PCI,it was considered that the effective vol-ume of clear supernatant after centrifugation is different than that of the culture broth,since biomass occupies a signi?cant space,especially at high cell densities.In this sense,a calibration curve correlating recoverable media volume and cell concentration was constructed (data not shown).Also,for further calculations involv-ing biomass concentrations,only the plasmid bearing cells were
considered.
Fig.4.Plasmid loss pro?les for the fed-batch cultures of BW25113(pIMAM3). =0.25h ?1( ), =0.15h ?1(?)and =0.10h ?1( );non-induced culture at ?xed speci?c growth rate of =0.25h ?1(?).Note :For the last case ,the data corresponds to the number of duplications as if it had been induced at the same biomass concentration than the induced experiments .
A quick-glance comparison of the protein distributions shows how levels of pre-PCI accumulation in the cytoplasm are pro-portional to the speci?c growth velocity,i.e.proportions of inactive presecretory protein are highest at =0.25h ?1.Also,for =0.25h ?1,levels of protein in the periplasm increase up to end of the process,whereas for =0.15h ?1and =0.10h ?1,there is a decline in the PCI content of the periplasmic space 7.5and 12h post-induction,respectively,as a consequence of enhanced excretion of the protein out of the cell.This effect had already been reported for the high-level expression of cholera toxin
B in the periplasm of E.coli [36].All these observations seem to point out that the disparity in the amounts of secreted PCI are related to the relative rates at which the protein is produced in the cytosol,secreted to the periplasmic space,and excreted to the culture
media.
Fig.5.Analysis of ?nal cell fractions.Soluble cytoplasmic fractions (ec),periplas-micfractions (ep)and supernatants (sn)were separated by SDS-PAGE and stained by Colloidal Coomassie blue.(Lanes 3–5)fractions corresponding to =0.25h ?1;(lanes 7–9)fractions corresponding to =0.15h ?1;lanes 11–13,fractions corresponding to =0.10h ?1and (lanes 1and 15)molecular weight marker.Notice the molecular weight difference between the pre-secretory and mature PCI in the ec and ep -sn samples,respectively.
J.-M.Puertas et al./Process Biochemistry45(2010)1334–1341
1339
Fig.6.Final distribution of recombinant PCI.Relative amounts of PCI in the soluble cytoplasmic extracts(black),periplasmic extracts(light gray)and culture super-natants(dark gray)for the end-point of the production processes at =0.25h?1, =0.15h?1and =0.10h?1.
A further look at the protein pro?les for the fermentations at =0.15h?1and =0.10h?1,shows the presence of two well-differentiated zones:a?rst zone going from the start of the induction until7.5–12h after,and a second region from7.5to12h post-induction until the end of the culture.In the?rst stage,expres-sion of PCI(adding up all cell compartments)is faster than it is in the second one;whereas in the second secretion gains importance. These areas will be named Production-Dominant zone(PD zone) and Secretion-Dominant zone(SD zone),respectively.In an effort to further understand,at least qualitatively,the dynamics of pro-tein expression and traf?cking for each zone,protein and biomass pro?les were used to determine a series of speci?c rates of PCI expression,translocation(from the cytosol to the periplasm)and secretion(from the periplasmic space to the culture broth)for each of the two de?ned zones.As stated before,for these calculations corrected values of biomass were used,taking into account only plasmid bearing cells.
Values q p,q t and q s were calculated(in mg PCI g?1DCW h?1)for each time point of the fermentations,and then their mean values in each zone were calculated.The results are registered in Table2. In the Production-Dominant zone(PD zone),there are two trends that apply for all three speci?c growth velocities:First,q p is pro-portional to ,a fact already reported for the production of other proteins[29,32],and secondly,the secretion rates are around one order of magnitude lower than translocation rates,which explains why most of the active PCI is retained in the periplasmic space. For =0.25h?1,q t/q p=0.51,a ratio that indicates that the protein is being expressed in the cytosol at a higher rate than it can be exported by the SecYEG translocon,which can clarify the observed accumulation of the presecretory protein.In the same zone,for =0.10h?1,the rates of production and translocation are simi-lar,with q t/q p=0.92,and hence no signi?cant buildup of pre-PCI is detected in the cytosolic extracts.For =0.15h?1,q t/q p=0.77, slightly lower than the lowest growth rate.The same underlying principle can be applied to illustrate the higher levels of periplas-mic retention in the fermentation at =0.25h?1(and conversely the lower levels of extracellular product):In this set-up,the secre-tion to translocation rate is q e/q t=0.069,much lower than that of =0.10h?1,q e/q t=0.35.
In the Secretion-Dominant zone(SD zone),the trends are quite different.For the process at =0.25h?1this stage is not
reached Fig.7.Dynamics of recombinant PCI distribution.Time-course evolution of recom-binant PCI amounts in the soluble cytoplasmic extracts(black),periplasmic extracts (light gray)and culture supernatants(dark gray)for the production processes at(A) =0.25h?1,(B) =0.15h?1and(C) =0.10h?1.
because the oxygen exhaustion limit is met before(and a signif-icant proportion of the cells are plasmid-free);but for the case of =0.15h?1and =0.10h?1,signi?cant changes in the values of the expression,translocation and secretion rates can be observed,all of them being on the same order of magnitude.The increase of q s
1340J.-M.Puertas et al./Process Biochemistry45(2010)1334–1341
Table2
Mean protein expression,translocation and secretion rates by speci?c growth rate in the PD and SD zones of the fed-batch cultures.
Speci?c rates PD zone SD zone
=0.25h?1 =0.15hh?1 =0.10hh?1 =0.25hh?1 =0.15hh?1 =0.10hh?1
q p(mg PCI g?1DCW h?1)16.7±1.614.8±1.59.4±1.0– 5.9±1.2 1.9±0.6
q t(mg PCI g?1DCW h?1)8.5±1.311.5±2.18.7±1.7– 4.7±1.1 1.8±0.3
q e(mg PCI g?1DCW h?1)0.7±0.10.8±0.1 3.1±0.2– 4.0±0.1 3.1±0.2
compared to q p and q t re?ects the drop of the protein content in the periplasm,related to a higher level of PCI present in the extracellu-lar fraction.The increase in the rate at which the protein leaks from the periplasmic space is probably caused by the osmotic pressure buildup caused by retention of the recombinant product,and also by the weakening of the outer membrane due to the prolonged time that cells spend in the reactor subjected to physiological,chemical and mechanical stress[6,37].
Even though these rates are approximate and more time points would be necessary to establish quantitative conclusions,they sup-port the hypothesis that control of speci?c growth rate allows to optimize the levels of recombinant protein secretion by achieving
a proper coupling between the expression and translocation rates.
3.2.5.Considerations on the repeatability of the process
In the experiments presented up to now,PCI in periplasmic frac-tions and culture supernatants had been quanti?ed by means of enzymatic inhibition of CPA using a commercial kit,as described in Section2.It was desired to verify the validity of these results by quantifying PCI by Reverse Phase Chromatography(RP-HPLC).For these purpose the fermentations at =0.10h?1and =0.15h?1 were repeated and quanti?ed by the both methods.As it can be seen in Table3,total amounts of PCI quanti?ed by RP-HPLC and enzymatic assays agree with a mean2.6%variation.Also,the reproducibility of the process was validated,since very similar results were obtained compared to the original experiments:for =0.15h?1,a mean deviation of6.8%was observed,whereas for =0.10h?1this value was estimated in8.4%.
3.2.6.Enhancement of extracellular PCI yields by chemical permeabilization of the cells
Previous research has shown the synergistic action of the deter-gent Triton X-100and the amino acid glycine can promote the excretion of periplasmic proteins as they increase the permeabil-ity of the outer bacterial membrane[38,39].Since still for the best case scenario( =0.10h?1,Fig.7C)half of the expressed PCI was ultimately found in the periplasmic extracts,it was thought that a treatment with Triton X-100and glycine could be used to recover more active PCI from the culture media.
First,the optimal levels
of both reagents were determined in shake-?ask cultures.Triton X-100and glycine from sterile stocks were added into the culture media after4h of induction,once cells reached the growth plateau,in order to avoid interference of the chemical treatment with bacterial growth.Highest extracellular PCI concentration were obtained when cells were treated with0.5% Triton X-100and2.0%(data not shown).
The treatment was then applied in a bioreactor culture.The fed-batch process at =0.10h?1was repeated,but this time0.5%Triton X-100and2.0%glycine were added4h before the end of the culture (i.e.12h post-induction).The permeabilization treatment resulted in12%lower?nal biomass concentrations;but it allowed achieving over900mg L?1extracellular product.
Fig.8.Effect of cell permeabilization with Triton X-100+glycine over?nal PCI distribution.Relative amounts of PCI in the soluble cytoplasmic extracts(black), periplasmic extracts(light gray)and culture supernatants(dark gray)for the end-point production processes at =0.10h?1with and without cell permeabilization.
Table3
Analyses of repeatability of the processes at =0.15h?1and =0.10h?1.Sub-indexes A and B stand for the two different replicas.EA:enzymatic assay and RP-HPLC:reverse phase liquid chromatography.
Induction time(h)EA RP-HPLC
PCI A(mg L?1)PCI B(mg L?1)Deviation A–B(%)PCI B(mg L?1)Deviation EA-HPLC(%) =0.15h?1
250559.552 5.6
3262273 4.1268 1.9
4412430 4.3439 2.1
56455948.25920.3
79561008 5.31135 2.2
9109612049.41168 3.0
=0.10h?1
1.5127135 6.1129 1.6
3.54514187.6443 1.8
5.510179229.8989 2.8
7.5113112147.11108 2.1
9.51253140611.51313 4.7
11.5145915848.21512 3.6
J.-M.Puertas et al./Process Biochemistry45(2010)1334–13411341
Analysis of the?nal distribution of PCI on the different cell com-partments was also done(Fig.8).As expected,cell permeabilization signi?cantly favored the passage of periplasmic PCI towards the culture media,increasing the percentage of extracellular PCI from 50.0%to81.3%of the total recombinant product.
4.Conclusions
The developed fed-batch protocol allowed obtaining a total of1.4g L?1of active product when the growth rate was?xed at =0.10h?1.A signi?cant effect of this last parameter over the secretory protein yields was observed and explained in terms of lower plasmid stability at higher growth rates,but also due to different distribution of the recombinant protein in the cell compartments.It was determined that the limiting step in the pro-duction of PCI was the excretion from the periplasm,where most of the product was retained.Levels of extracellular protein were inversely proportional to the growth rate;whereas accumulation in the cytosol followed the opposite trend.For growth at =0.10h?1, 90%of the protein was secreted to the periplasm and supernatants, compared to78%at =0.15h?1and73%at =0.25h?1.Since it was determined that the limiting step in the production of PCI was the excretion from the periplasm,a chemical treatment was used to permeabilize the cells and release the protein from the outer membrane.By means of this strategy,the extracellular propor-tion of PCI was incremented to81.3%of total protein.The results here presented seem to point out that signi?cant improvement in secreted protein yields can be achieved by ef?cient bioprocess development.
Acknowledgements
The authors wish to thank Dr.F.X.Avilès for providing scienti?c support to this work,and Dr.Manuel Mansur for critical reading of the manuscript.The authors are cooperative members of the Xarxa de Referència en Biotecnologia(XRB,Generalitat de Catalunya). This work was granted by Spanish Ministry of Science and Innova-tion(MICINN),project CTQ2008-00578and DURSI2005SGR00698 (Generalitat de Catalunya.)J.M.P.and J.R.are recipients of pre-doctoral grants also from MICINN.M.R.D.V.and J.L.acknowledge support from Proyecto Genoma Espa?na.
References
[1]Shokri A,Sandén AM,Larsson G.Cell and process design for targeting of recom-
binant protein into the culture medium of Escherichia coli.Appl Microbiol Biotechnol2003;60(February(6)):654–64.
[2]Wan EW,Baneyx F.TolAIII co-overexpression facilitates the recovery of
periplasmic recombinant proteins into the growth medium of Escherichia coli.
Protein Express Purif1998;14:13–22.
[3]Mujacic M,Baneyx F.Recombinant protein folding and misfolding in Escherichia
coli.Nat Biotechnol2004;22(November(11)):1399–408.
[4]Miot M,Betton JM.Protein quality control in the bacterial periplasm.Microb
Cell Fact2004;3(May(1)):4.
[5]Choi JH,Lee SY.Secretory and extracellular production of recombinant proteins
using Escherichia coli.Appl Microbiol Biotechnol2004;64:625–35.
[6]Mergulh?o FJ,Summers DK,Monteiro GA.Recombinant protein secretion in
Escherichia coli.Biotechnol Adv2005;23(May(3)):177–202.
[7]Mergulh?o FJ,Monteiro GA.Analysis of factors affecting the periplasmic pro-
duction of recombinant proteins in Escherichia coli.J Microbiol Biotechnol 2007;17(August(8)):1236–41.
[8]Georgiou G,Segatori L.Preparative expression of secreted proteins in bacteria:
status report and future prospects.Curr Opin Biotechnol2005;16(5):538–45.
[9]Simmons LC,Yansura DG.Translational level is a critical factor for the secretion
of heterologous proteins in Escherichia coli.Nat Biotechnol1996;14:629–34.
[10]de Marco A.Strategies for successful recombinant expression of disul?de bond-
dependent proteins in Escherichia coli.Microb Cell Fact2009;May(8):26. [11]Mergulh?o FJ,Monteiro GA,Larsson G,Bostrom M,Farewell A,Nystr?m T,
Cabral JM,Taipa MA.Evaluation of inducible promoters on the secretion of
a ZZ proinsulin fusion protein in Escherichia coli.Biotechnol Appl Biochem
2003;38:87–93.[12]Kajava AV,Zolov SN,Kalinin AE,Nesmeyanova MA.The net charge of the
?rst18residues of the mature sequence affects protein translocation across the cytoplasmic membrane of gram-negative bacteria.J Bacteriol2000;182: 2163–9.
[13]Hegde RS,Bernstein HD.The surprising complexity of signal sequences.Trends
Biochem Sci2006;31(10):563–71.
[14]Mergulh?o F,Monteiro G,Larsson G,Sanden A,Farewell A,Nystrom T,Cabral JM,
Taipa M.Medium and copy number effects on the secretion of human proinsulin in Escherichia coli using the universal stress promoters uspA and uspB.Appl Microbiol Biotechnol2003;61:495–501.
[15]Ryan CA,Hass GM,Kuhn RW.Puri?cation and properties of a carboxypeptidase
inhibitor from potatoes.J Biol Chem1974;249(17):5495–9.
[16]Blanco-Aparicio C,Molina MA,Fernández-Salas E,Frazier ML,Mas JM,Querol
E,Avilés FX,de Llorens R.Potato carboxypeptidase inhibitor,a T-knot protein, is an epidermal growth factor antagonist that inhibits tumor cell growth.J Biol Chem1998;273(20):12370–7.
[17]Sitjà-Arnau M,Molina MA,Blanco-Aparicio C,Ferrer-Soler L,Lorenzo J,Avilés
FX,Querol E,de Llorens R.Mechanism of action of potato carboxypeptidase inhibitor(PCI)as an EGF blocker.Cancer Lett2005;226(2):169–84.
[18]Molina MA,Aviles FX,Querol E.Expression of a synthetic gene encoding
potato carboxypeptidase inhibitor using a bacterial secretion vector.Gene 1992;116:129–38.
[19]Marino-Buslje C,Molina MA,Canals F,Avilés FX,Querol E.Overproduction
of a recombinant carboxypeptidase inhibitor by optimization of fermentation conditions.Appl Microbiol Biotechnol1994;41:632–7.
[20]Pinsach J,de Mas C,Lopez J.A simple feedback control of Escherichia coli
growth for recombinant aldolase production in fed-batch mode.Biochem Eng J2006;29(3):325–42.
[21]Durany O,de Mas C,López-Santín J.Fed-batch production of recombinant
fuculose-1-phosphate aldolase in E.coli.Process Biochem2005;40:707–16. [22]Betton J-M,Boscus D,Missiakas D,Raina S,Hofnung M.Probing the structural
role of an alpha beta loop of maltose-binding protein by mutagenesis:heat-shock induction by loop variants of the maltose-binding protein that form periplasmic inclusion bodies.J Mol Biol1996;262(2):140–50.
[23]Candiano G,Bruschi M,Musante L,Santucci L,Ghiggeri GM,Carnemolla
B,Orecchia P,Zardi L,Righetti PG.Blue silver:a very sensitive colloidal Coomassie G-250staining for proteome analysis.Electrophoresis2004;25(9): 1327–33.
[24]Puertas JM,Betton JM.Engineering an ef?cient secretion of leech carboxypep-
tidase inhibitor in Escherichia coli.Microb Cell Fact2009;8(October):57. [25]Pinsach J,de Mas C,Lopez-Santin J.Induction strategies in fed-batch cultures for
recombinant protein production in Escherichia coli:Application to rhamnulose 1-phosphate aldolase.Biochem Eng J2008;41(2):187–92.
[26]Donovan RS,Robinson CW,Click BR.Optimizing inducer and culture condi-
tions for expression of foreign proteins under the control of the lac promoter (review).J Ind Microbiol1996;16(3):145–54.
[27]Kweon DH,Nam SH,Park KM,Seo JH.Overproduction of Phytolacca insularis
protein in batch and fed-batch culture of recombinant Escherichia coli.Process Biochem2001;36(6):537–42.
[28]Shiloach J,Fass R.Growing E.coli to high cell density—a historical perspective
on method development.Biotechnol Adv2005;23:345–57.
[29]Vidal L,Durany O,Suau T,Ferrer P,Benaiges MD,Caminal G.High-level
production of recombinant His-tagged rhamnulose1-phosphate aldolase in Escherichia coli.J Chem Technol Biotechnol2003;78:1171–9.
[30]Eiteman MA,Altman E.Overcoming acetate in Escherichia coli recombinant
protein fermentations.Trends Biotechnol2006;24(11):530–6.
[31]Korpimaki T,Kurittu J,Karp M.Surprisingly fast disappearance of?-lactam
selection pressure in cultivation as detected with novel biosensing approaches.
J Microbiol Meth2003;53(1):37–42.
[32]Vidal L,Ferrer P,Alvaro G,Benaiges MD,Caminal G.In?uence of induction and
operation mode on recombinant rhamnulose1-phosphate aldolase production by Escherichia coli using the T5promoter.J Biotechnol2005;118(1):75–87. [33]Khalilzadeh R,Shojaosadati SA,Maghsoudi N,Mohammadian-Mosaabadi J,
Mohammadi MR,Bahrami A,Maleksabet N,Nassiri-Khalilli MA,Ebrahimi M, Naderimanesh H.Process development for production of recombinant human interferon-gamma expressed in Escherichia coli.J Ind Microbiol Biotechnol 2004;31(2):63–9.
[34]Riesenberg D,Guthke R.High-cell density cultivation of microorganisms.Appl
Microbiol Biotechnol1999;51:422–30.
[35]Ryder DF,Dibiasio D.An operational strategy for unstable recombinant DNA
cultures.Biotechnol Bioeng1984;26(8):942–7.
[36]Slos P,Speck D,Accart N,Kolbe HV,Schubnel D,Bouchon B,Bischoff R,Kieny MP.
Recombinant cholera toxin B subunit in Escherichia coli:high-level secretion, puri?cation,and characterization.Protein Express Purif1994;5(5):518–26. [37]Shokri A,Sandén AM,Larsson G.Growth rate-dependent changes in Escherichia
coli membrane structure and protein leakage.Appl Microbiol Biotechnol 2002;58(3):386–92.
[38]Yang J,Moyana T,MacKenzie S,Xia Q,Xiang J.One hundred seventy-fold
increase in excretion of an FV fragment-tumor necrosis factor alpha fusion protein(sFV/TNF-alpha)from Escherichia coli caused by the synergistic effects of glycine and Triton X-100.Appl Environ Microbiol1998;64(8):2869–74. [39]Li Z,Gu Z,Wang M,Du G,Wu J,Chen J.Delayed supplementation
of glycine enhances extracellular secretion of the recombinant alpha-cyclodextrin glycosyltransferase in Escherichia coli.Appl Microbiol Biotechnol 2010;85(3):553–61.