Efficient expression vectors and host strain for the production of recombinant proteins by Yarrowia lipolytica in process conditions

Background The oleaginous yeast Yarrowia lipolytica is increasingly used as an alternative cell factory for the production of recombinant proteins. Recently, regulated promoters from genes EYK1 and EYD1, encoding an erythrulose kinase and an erythritol dehydrogenase, respectively, have been identified and characterized in this yeast. Hybrid promoters up-regulated by polyols such as erythritol and erythrulose have been developed based on tandem copies of upstream activating sequences from EYK1 (UAS1EYK1) and XPR2 (encoding extracellular protease, UAS1XPR2) promoters. Results The strength of native (pEYD1) and engineered promoters (pEYK1-3AB and pHU8EYK) was compared using the extracellular lipase CalB from Candida antarctica as a model protein and a novel dedicated host strain. This latter is engineered in polyol metabolism and allows targeted chromosomal integration. In process conditions, engineered promoters pEYK1-3AB and pHU8EYK yielded 2.8 and 2.5-fold higher protein productivity, respectively, as compared to the reference pTEF promoter. We also demonstrated the possibility of multicopy integration in the newly developed host strain. In batch bioreactor, the CalB multi-copy strain RIY406 led to a 1.6 fold increased lipase productivity (45,125 U mL−1) within 24 h as compared to the mono-copy strain. Conclusions The expression system described herein appears promising for recombinant extracellular protein production in Y. lipolytica.


Background
The oleaginous yeast Yarrowia lipolytica is increasingly used as an alternative to model yeasts such as Saccharomyces cerevisiae or Komagataella phaffii (Pichia pastoris). for the production of recombinant proteins [1,2]. More than one hundred heterologous proteins have already been produced at high yield, highlighting its potential utilization as a cell factory. Y. lipolytica has been also shown able to produce efficiently chemicals such as citric acid [3], itaconic acid [4], erythritol [5], erythrulose [6] and lipid for biodiesel and biojet fuel [7,8]. In nature, Y. lipolytica can grow in environments rich in lipids and proteins due to its ability to synthesize and secrete hydrolytic enzymes such as proteases or lipases [9,10]. Based on these metabolic features, several engineering tools such as strong promoters and efficient sorting signals, including the prepro sequence of the LIP2 gene encoding an extracellular lipase, have been developed [1,11]. The promoter from the XPR2 gene encoding alkaline extracellular protease was the first developed for recombinant gene expression [12]. Its functional dissection led to the identification of an upstream activating sequence (UAS1 XPR2 ), later used to construct constitutive

Open Access
Microbial Cell Factories *Correspondence: jean-marc.nicaud@inra.fr † Young-Kyoung Park and Marie Vandermies contributed equally to this work 3 Micalis Institute, UMR1319, Team BIMLip: Integrative Metabolism of Microbial Lipids, INRA-AgroParisTech, Domaine de Vilvert, 78352 Jouy-en-Josas, France Full list of author information is available at the end of the article hybrid promoters [13]. These are based on UAS1 XPR2 repeats fused upstream of the minimal LEU2 promoter (mLEU2) [14,15]. The number of UAS1 XPR2 repeats was found to modulate the strength of the promoter, allowing thus a fine-tuning of gene expression. A similar strategy was also applied to design constitutive promoters derived from the TEF1 gene encoding the translation elongation factor 1α [16]. Regulated promoters derived from the LIP2 and POX2 genes encoding an extracellular lipase and an acyl-CoA oxidase, respectively, have been also developed for recombinant gene expression [17]. Although these regulated promoters delivered strong gene expression, their utilization at industrial scale is impeded by the hydrophobic nature of their inducers (i.e. triglycerides and fatty acids).
Recently, we have characterized two genes, namely EYD1 and EYK1, involved in erythritol catabolic pathway [6,18]. Gene EYD1 was found to encode an erythritol dehydrogenase able to convert erythritol to erythrulose, while gene EYK1 was suggested to encode an erythrulose kinase. Induction of EYK1 native promoter (pEYK300) is significantly increased, in a dose-dependent manner, in the presence of erythritol or erythrulose, and drastically reduced in the presence of glycerol and glucose. Moreover, we demonstrated that erythrulose is a better inducer than erythritol [19]. The dissection of the EYK1 promoter highlighted the existence of two UAS, namely UAS1 EYK1 and UAS2 EYK1 . Using a reporter system based on yellow fluorescent protein (YFP) and mutated promoter, UAS1 EYK1 was identified as essential for promoter induction by both erythritol and erythrulose. By contrast, UAS2 EYK1 was found involved in repression by glucose. Synthetic promoters were constructed by addition of multiple copies of UAS1 EYK1 or UAS1 XPR2 upstream of the native pEYK300 promoter. These promoters yielded, respectively, 3.2 and 15.6-fold higher expression levels of YFP-encoding gene than those obtained with wild-type pEYK300 promoter [19]. Carly et al. [6,18] found that the disruption of the EYK1 gene impairs the ability of yeast cells to fully metabolize erythritol, since an eyk1Δ mutant could only convert erythritol into erythrulose. Therefore, with such a mutant, erythritol and/or erythrulose could be used as a free inducer as demonstrated previously [19]. Promoter engineering efforts were pursued with a reporter system based on RedStar fluorescent protein [20]. It concerned the influence of UAS1 EYK1 repeats and core element (EYK1, TEF) on promoter strength in wild type and eyk1Δ mutant.
Based on previous results, three promoters, namely pEYK1-3AB which comprises three repeats of UAS1 EYK1 , pHU8EYK which comprises eight copies UAS1 XPR2 from XPR2 gene and the native promoter of gene EYD1 seemed to be promising to drive the production of recombinant proteins [19,20]. Herein, these promoters were challenged for the production in process conditions of a protein of industrial interest, the lipase CalB from Candida antarctica. For that purpose, the pro-CalB nucleic acid sequence was codon-optimized for Y. lipolytica, fused with the secretion signal from the LIP2 gene, and cloned under the control of promoters pEYD1, pEYK1-3AB, pHU8EYK, and of strong constitutive promoter pTEF used for comparison. The different CalB expression cassettes were then introduced in a novel host strain specifically developed for erythritol/erythrulose-inducible expression systems. For the different constructed strains, CalB gene expression and extracellular activity were monitored in process conditions, during cultures in bioreactor.
Cultures in bioreactor were inoculated at an initial optical density at 600 nm (OD 600 ) of 0.5 with PBS washed cells from a 16-h preculture in YPD medium. Cultures in 2Mag bioREACTOR (Munich, Germany) were performed in triplicate for 48 h in 10 mL of YNBGE medium with agitation set at 800 rpm. Cultures in DASGIP bioreactor (DASbox Mini Bioreactors SR0250ODLS, Eppendorf, Hamburg, Germany) were performed in duplicate for 48 h in 150 mL of YNBG 2 E medium supplemented with 500 µL L −1 antifoam (Tego KS911, Evonik, Essen, Germany). Airflow was set at 1 vvm, agitation was ranged from 800 to 950 rpm to ensure a dissolved oxygen level above 20% and pH was automatically adjusted to 6.8 by addition of H 3 PO 4 8 M or NaOH 12.5 M. Culture in Duetz deepwell plate (24-well plate with pyramidal bottom, Kühner AG, Birsfelden, Switzerland) were performed in 2 mL of YNBG 2 E medium for 48 h; they were inoculated with of 200 μL of a 24 h precultures carried out in 400 µL YPD in 48-well microplates (multiwell cell culture plates, flat bottom, TC-treated, VWR, Radnor, PA, USA) for 24 h.

Strains and plasmids construction
Standard molecular genetic techniques were used in this study [22]. Restriction enzymes were obtained from New England Biolabs (MA, USA). PCR amplifications were performed in an Applied Biosystems 2720 thermal cycler with GoTaq DNA polymerases (Promega, WI, USA) or Q5 High-Fidelity DNA Polymerase (New England Biolabs). PCR fragments were purified with a QIAgen Purification kit (Qiagen, Hilden, Germany) and plasmids DNA were isolated with a QIAprep Spin Miniprep kit (Qiagen).
All strains and plasmids used in this study are listed in Table 1.
For the construction of JMY7126 ( Fig. 1), LYS5 and EYK1 were disrupted using PUT cassettes according to Fickers et al. [23]. For LYS5 disruption, the promoter (P) and terminator (T) regions of the gene were amplified by PCR with primer pairs LYS5-P1/LYS5-P2 and LYS5-T1/ LYS5-T2, respectively. Primers LYS5-P1 and LYS5-T2, contained sequence of NotI restriction site while primers LYS5-P2 and LYS5-T1 contained the sequence of the I-SceI restriction ( Table 2). The corresponding amplicons were purified and used as a template for the second PCR step resulting PT fragment, that was subsequently cloned to pCR4Blunt-TOPO plasmid after purification (Invitrogen, CA, USA). Finally, URA3ex marker from JMP1046 was introduced at the I-SceI site of this plasmid, yielding plasmid JMP3267. The LYS5 PUT cassette obtained by NotI digestion of JMP3267 was used to transform Y. lipolytica strain JMY1212 to yield JMY5207 (Fig. 1). The

Strain (plasmid)
Genotype References This work disruption of LYS5 was verified by auxotrophy check on YNBD and YNBD-lysine. In that strain, URA3ex marker was rescued by transient expression of Cre recombinase using the replicative plasmid JMP547 as described previously [23]. In the resulting strain JMY7121, EYK1 was disrupted using a PUT cassette obtained from plasmid RIE124 by NotI digestion. The disruption of EYK1 was verified by colony PCR with primers pair preTEYK Fw/ postPEYK Rv and growth on YNBD-lysine and YNBElysine. This yielded to strain JMY7123 that was further transformed with plasmid RIE132 in order to excise the URA3ex marker. This yielded to the final strain JMY7126 (Table 1, Fig. 1). The loss of the replicative plasmid was checked by replica plating on YPD supplemented or not with hygromycin for JME547, or on YNBD and YNBE for RIE132. To restore LYS5 prototrophy, strains were transformed with the expression cassette of plasmid RIE279 obtained by NotI digestion, and gene integration was verified by colony PCR using primer LPR-R and LYS5PR. The erythritol-inducible plasmids were constructed from JMP1046 by replacing the pTEF by inducible promoters (namely pEYK1-3AB, pHU8EYK1, plasmid JMP4123 and JMP4001, respectively) by digestion with ClaI and BamHI and subsequent ligation. Here, pEYK300A3B described in Trassaert et al. [19] was renamed pEYK1-3AB according to Park et al. [20]. The Y. lipolytica LIP2 pre-CalB pro-CalB gene was codon-optimized by Biocatalysts LTD, The auxotrophic strain JMY1212 [27] was used as parental strain. LYS5 and EYK1 genes were successively disrupted with the corresponding purified disruption cassette obtained from JMP3267 and RIP124 by NotI digestion, respectively. URA3ex auxotrophy marker was excised from strains using Cre-Hyg or Cre-EYK1 replicative plasmids (JME547 and RIE132, respectively) as described in [23,33] (Cardiff, UK) and synthesized by Geneart (Regensburg, Germany). The sequence of the optimized synthetic gene (15ACCYPP_1762989_LIP2-CalB-Yl-Opt) is displayed in Additional file 1: Table S1. LIP2pre-CalBpro-CalB gene (CalB) was cloned into the vectors JME1046 (pTEF), JME4266 (pEYK1-3AB) and JME4230 (pHU8EYK) at BamHI/AvrII restriction sites ( Fig. 2 and Table 1). To obtain a pEYD1-CalB construct, promoter pEYK1-3AB from plasmid JMP4365 was exchanged by pEYD1 obtained by PCR on genomic DNA of Y. lipolytica using primer ClaI-pEYD1-Fw and BamHI-pEYD1-Rev and subsequent enzyme digestion with ClaI and BamHI. For the construction of plasmid JME4579, the URA3ex marker was exchanged with the LYS5ex marker by I-SceI digestion and ligation. Gene expression cassettes were obtained by NotI digestion of the corresponding plasmid and used to transform Y. lipolytica strains JMY7126 by the lithium acetate method as described previously [24].

Cell growth monitoring
Cell growth was monitored by optical density at 600 nm (OD 600 ). 1 unit of OD 600 corresponds to 0.29 g of dry cell weight (DCW).

Quantification of CalB gene expression
RNA extraction and cDNA synthesis were performed as described previously [17] on 2.2 × 10 7 cells of strains JMY7536, JMY7539, JMY7544, and JMY7548 after 24 h of culture in 2Mag bioREACTOR. qPCR was performed with primer pairs CalB-internal-Fw/CalB-internal-R2 and ACT-F/ACT-R (Table 2) for CalB and actin genes, respectively. CalB gene expression levels were standardized relative to the expression level of the actin gene [17]. Experiments were performed in triplicate.

Lipase activity
Lipase activity in culture supernatants was determined by monitoring the hydrolysis of para-nitrophenyl butyrate (p-NPB), according to Fickers et al. [25]. p-NPB dissolved in acetonitrile (20% v/v) was added to a final concentration of 1 mM into 100 mM phosphate buffer, pH 7.2, containing 100 mM NaCl. The resulting solution was sonicated for 2 min on ice. The reaction was initiated by addition of 20 µL of culture supernatant to 1 mL of p-NPB solution. The release of para-nitrophenol (P-NP) was monitored for 3 min at 405 nm (A 405 ), considering the molar extinction coefficient of P-NP (εPNP) equal to 0.0148 µM −1 cm −1 . When necessary, supernatant samples were diluted to obtain initial velocities below OD 405 of 0.3 U min −1 . All lipase activity assays were performed at least in duplicate from two independent cultures. One unit of lipase activity was defined as the amount of enzyme releasing 1 µmol p-NPB per minute at 25 °C and pH 7.2 (U mL −1 ). Specific lipase activity was defined as lipase activity per gram of dry cell weight (U gDCW −1 ). Lipase volumetric production rate was defined as lipase activity per hour of culture (U mL −1 h −1 ), while lipase specific production rate was defined as unit of lipase activity per gram of DCW and per hour (U gDCW −1 h −1 ).

Protein electrophoresis
Proteins were subjected to sodium dodecyl sulfate (SDS)polyacrylamide gel electrophoresis (PAGE) on a Novex ™ 12% Tris-Glycine Mini Gel (Thermo Fisher Scientific), according to Laemmli [26]. Four milliliter of prestained Protein Marker IV (AppliChem GmbH, Darmstadt, Germany) were used as molecular weight standards, and a supernatant sample contained 40 U of lipase in a volume 5 µL.

Carbon source concentration
Polyol concentrations in culture supernatants were determined by HPLC (Agilent 1200 series, Agilent Technologies, Santa Clara, CA, USA). Compounds were eluted from an Aminex HPX-87H column (300 × 7.8 mm, Bio-Rad, Hercules, CA, USA) at 65 °C, using a mobile phase consisting of 5 mM H 2 SO 4 solution at a flow rate of 0.5 mL/min. Glycerol and erythritol were detected using refractive index and erythrulose was detected using UV absorption at 210 nm. Specific glycerol uptake rate was defined as gram of glycerol per gram of DCW and hour (g gDCW −1 h −1 ).

Construction of the host strain JMY7126 for erythritol-inducible based expression system
Yarrowia lipolytica strain JMY1212 was previously developed to target integration of an expression cassette at a zeta docking platform located at the LEU2 locus [27]. expression related to the integration locus could be avoided, which is a prerequisite for this study. Moreover, such a feature leads to precisely genetically characterized recombinant strains, which is of importance when developing an industrial chassis. Strain JMY1212 is also disrupted for gene XPR2 encoding AEP alkaline protease and for LIP2, LIP7 and LIP8 encoding the main extracellular lipases in Y. lipolytica [10,28]. Despite strain JMY1212 has been used successfully for heterologous gene expression [27,[29][30][31], it suffers from several drawbacks: (i) it possesses only a single auxotrophy based on uracil metabolism (ura3), impairing thus multiple genome editions; (ii) it is able to metabolize erythritol and erythrulose, preventing thus their utilization as free inducer. To adapt the potentialities of JMY1212 to the requirements of a versatile, erythritol/erythrulose-inducible expression system, the strain was further genetically engineered. Firstly, gene LYS5 encoding saccharopine dehydrogenase [32] was disrupted with the URA3ex cassette. The resulting lysine auxotroph strain was named JMY5207 (Fig. 1).
In order to use erythritol as a free inducer, it is requested to dispose of a host strain unable to metabolize this polyol. We demonstrated in a previous study that erythrulose, the first intermediate of the erythritol catabolic pathway [18] is a better inducer than erythritol [19]. Therefore, the second step of JMY1212 strain improvement was to disrupt gene EYK1. Practically, the gene was disrupted in strain JMY7121, an auxothroph derivative of JMY5207 (Fig. 1) using a PUT cassette obtained from RIP124 by NotI digestion. This yielded finally to strain JMY7126, after URA3ex marker excision in strain JMY7123. With such a strain, we obtained a useful system to generate autonomously erythrulose from low-cost erythritol, and allowing the use of these polyols as free inducers. Moreover, in that strain, a Cre-EYK1 replicative vector can be used for transient expression of Cre recombinase [33], which faster delivers transformants without the drawbacks of using Cre-Hyg (hygromycin resistance gene) vector [23]. In summary, JMY7126 contains three auxothrophies (Ura-, Lys-, Eyk-), is compatible with Cre-EYK1 (RIE132) marker rescue, and with the set of erythritol-inducible promoters deriving from pEYK1 and pEYD1 (Trassaert and Vandermies [19,20], unpublished observations).

Construction of expression vectors and CalB production strains
In most of recombinant protein production processes, protein synthesis occurs in a growth phase decoupled manner. For that purpose, promoters based on LIP2 and POX2 genes, strongly induced by oleic acid, have been developed (pLIP2 and pPOX2) [17,34]. The main drawback of these systems is the utilization of waterinsoluble inducers, which is not convenient at large scale due to the lower mixing efficiency of industrial bioreactors. To overcome this problem, we recently developed a set of strong and tightly regulated promoters derived from EYK1 and EYD1 genes and induced by hydrophilic substances such as erythritol and erythrulose (Trassaert and Vandermies [19,20], unpublished observations). Some of these promoters, namely pEYK1-3AB, pHU8EYK, and pEYD1, were selected for this study based on previous experiments (Trassaert and Vandermies [19,20], unpublished observations) (Fig. 2).
The industrially relevant lipase CalB from C. antarctica was used as a model protein to assess the ability of these promoters to drive protein production in process conditions (i.e. in bioreactor). For that purpose, the CalB gene sequence, together with its pro-region was codon-optimized and fused with the signal peptide of LIP2 (pre-region). With the resulting construct (preLIP2-proCalB-CalB, here after CalB, Additional file 1) different expression vectors were obtained, namely JME3739 (pTEF-CalB), JME4365 (pEYK1-3AB-CalB), JME4243 (pHU8EYK-CalB) and JME4590 (pEYD1-CalB) and JME3739 (pTEF-CalB) used for comparison. They were used to transform strain JMY7126, after NotI digestion and purification of the expression cassette. The resulting mono-copy strains, respectively JMY7536, JMY7539, JMY7544, and JMY7548, harbor a single copy of CalB expression cassette integrated at their zeta-docking platform. As stated above, the docking system prevents variability caused by random genomic integration, allowing here to compare the three selected erythritol/erythrulose-inducible promoters with the constitutive promoter pTEF used as a reference.

Comparison of erythritol-inducible promoters CalB gene expression and protein production in function of the erythritol-inducible promoters used
As a first characterization, CalB expression were determined after 24 h (i.e. at the end of the exponential growth phase) for strains JMY7536, JMY7539, JMY7544 and JMY7548 grown in YNBGE medium in 2Mag bioREAC-TOR. The culture medium employed here appears more industrially relevant, as compared to defined media previously used for CalB production [30] and more generally for recombinant protein production [35]. Rather than glucose, glycerol was selected as a main carbon source since it is a cheap by-product of the biodiesel industry that has been demonstrated suitable for recombinant protein production [36][37][38][39][40][41][42][43]. Yeast extract and soytone were added at a low concentration to the culture medium to enhance cell growth and protein production. Soytone, a peptone from soy origin, efficiently replaces casein tryptone in bioprocesses mandatorily devoid of components of animal origin [44][45][46][47]. Moreover, erythritol is a perfectly affordable inducer, especially when it could be obtained in a bioprocess from glycerol with a Y. lipolytica metabolically engineered strain [5] and used in combination with a eyk1Δ derivatives that are no longer able to consume it [18,19].
As shown in Fig. 3a, the highest expression were obtained for pEYK1-3AB and pHU8EYK. Although, the CalB expression level obtained with the two promoters were not significantly different, they were, respectively, 2.5 and 2.7-fold higher than the one obtained with pTEF considered as a strong constitutive promoter. By contrast, pEYD1 led to a similar expression level than the one obtained with pTEF.
As a further characterization, biomass and extracellular lipase activity were determined in the same experimental conditions after 48 h of growth, and the specific activities were calculated. As shown in Fig. 3b, strain JMY7539 (pEYK1-3AB) and JMY7544 (pHU8EYK) yielded to the highest lipase specific activities (2041 ± 78 and 1852 ± 487 U mgDCW −1 , respectively). For those two strains, the specific enzymatic activities were, respectively, 2.8-fold and 2.5-fold higher than that obtained with strain JMY7536 (pTEF; 733 ± 88 U mgDCW −1 ). By contrast, specific lipase activity of strain JMY7548 (pEYD1) was 1.3-fold lower than that of strain JMY7536. The enzymatic productivities obtained here with strain JMY7539 (pEYK1-3AB) and JMY7544 (pHU8EYK) were 1.7 and 1.6-fold higher than the lipase productivity obtained with strain JMY1105 (pLIP2-LIP2) in 20-L batch fermentation [48]. The results obtained with strains JMY7539 (pEYK1-3AB) and JMY7544 (pHU8EYK) were, however, somewhat unexpected. Indeed, in previous experiments performed with fluorescent reporter system, pHU4EYK [19] and pHU8EYK (Trassaert and Vandermies, unpublished observations), bearing respectively four and eight copies of UAS1 XPR2 , significantly higher fluorescence level were obtained (4.9 and 9.8 fold, respectively) as compared to that obtained for pEYK1-3AB. However, these experiments were performed in microplate cultures that are known to be not representative of process conditions, notably in terms of oxygen transfer. Based on the results obtained for CalB gene expression and lipase specific activity, strain JMY7539 (pEYK1-3AB-CalB) was selected for further characterizations in batch bioreactor.

Culture of strain RIY368 in DASGIP bioreactor
To challenge the pEYK1-3AB-based expression system to process conditions, the strain RIY368, a prototroph derivative of strain JMY7539 was grown for 48 h in YNBG 2 E medium in DASGIP bioreactor, with pH and pO 2 regulation. Samples were collected over time, and biomass together with lipase activity were determined. The exponential growth phase lasted for 12 h with specific growth rate of 0.29 ± 0.00 h −1 and final biomass of 6.96 ± 0.04 gDCW L −1 (Fig. 4, Table 3). Within the first 24 h, glycerol, the main carbon source, had been entirely consumed, and inducer (i.e. erythritol) assimilated by the cells (Additional file 2: Fig. S1). Lipase activity reached its highest titer (28,024 ± 743 U mL −1 ) after 24 h of culture  (Fig. 4). It decreased slightly then after until the end of the culture (until 20150 ± 1060 U mL −1 ). Analysis of culture supernatant by SDS-PAGE clearly highlighted that CalB is the only secreted protein in those conditions (Fig. 4b). During the enzyme production phase (between 3.5 and 24 h of culture), the lipase volumetric productivity was of 1357 ± 34 U mL −1 h −1 ( Table 3). Culture of strain RIY394 (pTEF-CalB prototroph) in the same experimental conditions yielded to a 6.2 fold lower lipase activity after 24 h (data not shown). Here, we demonstrated that a high CalB production level could be achieved in bioreactor by a combined strategy of codon optimization, and suitable inducible promoter and host strain use. In a previous bioreactor study, the native CalB sequence under the control of promoter POX2 had been cloned in the Y. lipolytica strain JMY1212 (parent strain of JMY7126). The resulting lipase activity was about 5 U mL −1 after 102 h of cultivation [30]. Other studies in bioreactor using K. phaffii as a host strain for CalB production under the control of methanol-inducible promoter pAOX1 yielded 83 U mL −1 in 68 h, without codon optimization [49], and 6100 U mL −1 in 110 h with codon optimization [50]. Here, about 28,000 U mL −1 of lipase CalB were obtained in 24 h. To our knowledge, this process is the most efficient identified for CalB production in yeast, and it can be safely hypothesized that such efficiency will be reflected in the production of other recombinant proteins.

Additional genome edition Construction and screening of multi-copy CalB expressing strains
Another goal of this study was to endow the recipient strain with the possibility of additional genome editions. For that purpose an additional auxotrophy based on lysine metabolism (lys5) was introduced in strain JMY7126. Indeed, for some recombinant proteins, the process productivity could be enhanced by co-expression of specific chaperone as demonstrated for K. phaffii (see [51][52][53]) or by cloning an additional copy of the expression cassette [54]. To highlight this possibility of additional genome editions, strain JMY7539  (pEYK1-3AB-CalB) was transformed with the expression cassette of plasmid JMP4579 (LYS5ex-pEYK1-3AB-CalB, rescued beforehand by NotI digestion). Since the second expression cassette was integrated randomly in the yeast genome, six independent transformants were tested for their lipase activity. They were cultivated in YNBG 2 E medium in Duetz-System deepwell microplates, alongside with strain RIY368 (mono-copy). After 48 h of culture, supernatants were screened for lipase activity and it was found that the lipase productivity spanned over 3.2-fold of intensity (data not shown). The transformant presenting the highest specific lipase activity was named RIY406 and used for further experiments.

Culture of strain RIY406 in DASGIP bioreactor
Strain RIY406 was grown in DASGIP bioreactors under the same conditions as previously adopted for culture of mono-copy strain RIY368, in order to compare cell growth, carbon source uptake and lipase production. Cell growth kinetics of strain RIY406 was found similar to that of RIY368 (see Fig. 4 and Table 3), demonstrating that this additional recombinant gene expression did not affect cell growth capacity. Similarly to what was observed for strain RIY368, the exponential growth phase lasted for 12 h, with a specific growth rate and a final biomass of 0.27 ± 0.01 h −1 and 6.51 ± 0.02 gDCW L −1 , respectively. Growth and lipase production were sustained by similar glycerol consumption in strains RIY406 and RIY368 (Additional file 2). From these data, it can be concluded that the additional metabolic load resulting from the expression of a second recombinant gene does not alter host strain metabolism. As shown in Fig. 4 and Table 3, lipase activity of RIY406 reached its highest level (45,125 ± 2144 U mL −1 ) after 24 h of culture, again after glycerol entire consumption and erythritol entire assimilation (Additional file 2). At the maximal value, RIY406 lipase activity was 1.6-fold higher than the one of RIY368. This result is consistent with the ratios of volumetric and specific production rates, which are both of 1.6 ( Table 3). In conclusion, the addition of a second expression cassette based on lysine auxotrophy properly increases protein production without negative effect on the host strain. Such results confirm the opportunity of co-expression of two genes, based on two selection markers (URA3 and LYS5), under the control of an erythritol/ erythrulose-inducible promoter in strain JMY7126.

Conclusions
An efficient expression system relies on four mains properties: it depends on the vector used to express the gene of interest, on the host strain used to produce the compound of interest, and on the production and downstream processing steps. The design of a given expression system arises from its use and intended versatility. Here, we developed an expression system suitable for recombinant protein production in the yeast Y. lipolytica. It is based on erythritol/erythrulose-inducible promoters and a dedicated host strain, JMY7126, which enables multi-copy integration of expression cassettes and prevents erythritol complete metabolization. As a proof of concept, lipase CalB was expressed in mono-copy in strain JMY7126 under the control of three types of inducible promoters. The most appropriate promoter (namely pEYK1-3AB) for the production of this given protein was selected for demonstration of multi-copy expression. Under these conditions, about 45,000 U mL −1 of lipase CalB were obtained in 24 h in batch bioreactor, which represents to date the most efficient process identified for CalB production in yeast. Process development shall expand the potentialities of the proposed expression system even further, and the combination may greatly improve the production of other recombinant proteins in Y. lipolytica.
Additional file 1. Nucleic and amino acid sequence of codon optimized CalB. Bold sequences correspond to the pre sequence of the extracellular lipase Lip2p encoded by the LIP2 gene, and underlined sequences correspond to the pro CalB targeting sequence.
Additional file 2. Carbon source consumption of strains RIY368 (monocopy) and RIY406 (multi-copy). Cells were grown for 48 h at 28 °C in YNBG 2 E medium, in DASGIP bioreactors. Displayed means and standard deviations are the result of duplicate experiments.