Efficient production and secretion of bovine β-lactoglobulin by Lactobacillus casei

Background Lactic acid bacteria (LAB) are attractive tools to deliver therapeutic molecules at the mucosal level. The model LAB Lactococcus lactis has been intensively used to produce and deliver such heterologous proteins. However, compared to recombinant lactococci, lactobacilli offer some advantages such as better survival in the digestive tract and immunomodulatory properties. Here, we compared different strategies to optimize the production of bovine β-lactoglobulin (BLG), a major cow's milk allergen, in the probiotic strain Lactobacillus casei BL23. Results Using a nisin-inducible plasmid system, we first showed that L. casei BL23 strain could efficiently secrete a reporter protein, the staphylococcal nuclease (Nuc), with the lactococcal signal peptide SPUsp45 fused to its N-terminus. The fusion of SPUsp45 failed to drive BLG secretion but led to a 10-fold increase of intracellular BLG production. Secretion was significantly improved when the synthetic propeptide LEISSTCDA (hereafter called LEISS) was added to the N-terminus of the mature moiety of BLG. Secretion rate of LEISS-BLG was 6-fold higher than that of BLG alone while intracellular production reached then about 1 mg/L of culture. The highest yield of secretion was obtained by using Nuc as carrier protein. Insertion of Nuc between LEISS and BLG resulted in a 20-fold increase in BLG secretion, up to 27 μg/L of culture. Furthermore, the lactococcal nisRK regulatory genes were integrated into the BL23 chromosome. The nisRK insertion allowed a decrease of BLG synthesis in uninduced cultures while BLG production increased by 50% after nisin induction. Moreover, modification of the induction protocol led to increase the proportion of soluble BLG to around 74% of the total BLG production. Conclusion BLG production and secretion in L. casei were significantly improved by fusions to a propeptide enhancer and a carrier protein. The resulting recombinant strains will be further tested for their ability to modulate the immune response against BLG via mucosal delivery in a cow's milk allergy model in mice.


Background
Lactic acid bacteria are non-invasive and non-pathogenic Gram-positive bacteria with GRAS (generally regarded as safe) status that are widely used for food-processing and preservation. In addition, some strains were reported to exert probiotic effects [1][2][3][4][5]. Using the Nisin-Controlled Expression (NICE) system, β-lactoglobulin (BLG), a major cow's milk allergen, was successfully produced in Lactococcus lactis [6][7][8]. Administrations of BLG-producing lactococci to mice has been shown to induce a mucosal immune response that could partially prevent mice from a further sensitization to BLG [6,9]. However, L. lactis is rapidly lysed in each compartment of the digestive tract [10] whereas other LAB, such as lactobacilli, exhibit a greater resistance to the gastric environment and a better survival. Moreover, recent works suggest that some lactobacilli have stronger adjuvant properties than L. lactis [11]. Lactobacilli may thus appear as more attractive candidates to deliver therapeutic proteins to the intestinal mucosa. Unfortunately, studies with recombinant lactobacilli are often impaired by the lower levels of antigen production compared to those obtained with L. lactis [12]. This is a major concern since mucosal immune response depends on the amount of antigen delivered by the bacterial vector [13].
We previously described the construction of L. casei strains carrying a chromosomal BLG expression cassette inserted downstream an endogenous constitutive promoter [14]. Such chromosomal insertions led to BLG yields reaching 2 μg/L of culture. In the present work, we adapted lactococcal tools to improve BLG production in L. casei. For this purpose, we tested different expression cassettes coding for BLG in fusion with a carrier protein and/or with a secretion-enhancer propeptide [15]. We quantified and analyzed the structure of the recombinant BLG, using two immunoassays, one specific for BLG in its native conformation and the other specific for reduced and carboxymethylated, i.e. denatured, BLG [16]. We thus succeeded to increase BLG production in L. casei BL23, up to 1 mg/L of culture. As the two-plasmid NICE system appeared to be leaky in L. casei BL23, we also integrated the nisRK genes (necessary for the nisin-inducible expression of BLG) into the bacterial chromosome. This led to a 1.5-fold increase in BLG production. Finally, we obtained higher yields of soluble BLG by using different conditions of nisin-induction.

Results and Discussion
Nuc is efficiently secreted by L. casei We first investigated the ability of the BL23 strain to produce and secrete the reporter staphylococcal nuclease (Nuc, [15]) under the transcriptional control of the nisininducible promoter P nisA . Two plasmids were used ( Table  1). The plasmid pNZ9520 encodes the nisRK genes that are essential for the regulation of the P nisA promoter [17] and pSEC:Nuc contains a Nuc expression cassette with the P nisA promoter, the ribosome binding site (RBS Usp45 ) and the signal peptide (SP Usp45 ) of the major lactococcal secreted protein Usp45 ( [18], Fig. 1).
Nuc expression was induced with increasing concentrations of nisin up to 50 ng/mL (Fig. 2). Western blot analysis of protein extracts from cell lysates and supernatant fractions showed a dose-response correlation between nisin concentration and Nuc synthesis. The highest yield was obtained with 50 ng nisin/mL. We also detected an increasing release of mature Nuc into the culture medium in response to nisin induction, showing that SP Usp45 is functionally recognized and processed by the secretion machinery of the L. casei BL23 strain. As observed on Fig.  2, the efficient processing of the Nuc precursor is confirmed by the small amounts of SP usp45 -Nuc found in cell lysates.

Improved production and secretion of BLG in L. casei
We further investigated BLG production in L. casei with a construct (on pCYT plasmid, Table 1) targeting intracellular location and constructs (on pSEC plasmids, Table 1) encoding different fusion proteins previously designed for secretion of heterologous proteins in L. lactis [8].
Use of a lactococcal signal peptide Intracellular production of BLG obtained with the pCYT:BLG plasmid was relatively low (78 μg/L of culture, Table 2). The small amount of BLG occasionally detected in the S fraction was certainly due to BLG release in the culture medium from lysed cells. Considering that secretion is an efficient way to increase the overall heterologous protein production in L. lactis [15], we tested whether enhancement of BLG secretion could improve BLG yields in L. casei. As SP Usp45 -Nuc is efficiently secreted (Fig. 2), we first investigated the secretion of the precursor SP Usp45 -BLG. Surprisingly, no secreted BLG was detected in the S fraction of L. casei BL23(pSEC:BLG) ( Table 2). In contrast, we measured a significant enhancement of the insoluble intracellular BLG concentration (around 706 μg/L). Such effect of SP Usp45 has been already observed for production of heterologous proteins in L. lactis [6,19]. We suggest that presence of SP Usp45 sequence improved the translation of the blg gene but that the resulting protein was aggregating because of an inefficient recognition of the hybrid precursor by the secretion machinery. In L. casei, we thus found 92% of the recombinant BLG in the Ci fraction. Therefore, BLG in the Cs fraction represented only 8% of the total BLG production and only 1% of the soluble BLG exhibited a native conformation.

Use of a propeptide enhancer
The propeptide LEISSTCDA (LEISS) was inserted between the SP Usp45 and the mature moiety of the BLG. This insertion has been shown to improve secretion of heterologous proteins produced in L. lactis because of the presence of negatively charged residues [20]. For example, fusion of LEISS to BLG led to a 5-fold increase in BLG secretion in L. lactis [8]. In L. casei, this modification resulted similarly in a 6-fold increase in BLG secretion, reaching 9 μg/mL of culture. Total BLG production also increased from 706 μg/ L to 1022 μg/L of culture, but mostly in the Ci fraction. As observed with SP Usp45 -BLG in the Cs fraction, only 1% of soluble LEISS:BLG protein exhibited a native BLG conformation. However, the proportion of native BLG was sig-nificantly higher in the supernatant than in the cytoplasm since 40% of the BLG protein secreted in the culture medium displayed a native conformation.

Use of a carrier protein
Finally, we evaluated the effect of Nuc as carrier protein.
Nuc protein was shown to improve immunogenicity of BLG and of a major BLG epitope after oral or subcutaneous administrations to mice [7,9]. Moreover, in L. lactis, insertion of Nuc between LEISS and the mature moiety of BLG resulted in a 2-fold increase in total BLG production but in a 4-fold decrease in secretion when compared with LEISS-BLG [8]. In contrast, Nuc insertion between LEISS and BLG led to a 3-fold higher secretion in L. casei. Secre- Cm r , ori(pWV01), carries the nisin-inducible promoter P nisA pCYT:BLG Cm r , ori(pWV01), with a DNA fragment encoding the BLG mature moiety expressed under P nisA transcriptional control [6] pSEC:BLG Cm r , ori(pWV01), with a DNA fragment encoding the precursor SP Usp45 -BLG expressed under P nisA transcriptional control [6] pSEC:LEISS-BLG Cm r , ori(pWV01), with a DNA fragment encoding the precursor SP Usp45 -LEISS-BLG expressed under P nisA transcriptional control [8] pSEC:LEISS-Nuc-BLG Cm r , ori(pWV01), with a DNA fragment encoding the precursor SP Usp45 -LEISS-Nuc-BLG expressed under P nisA transcriptional control [8] pSEC:Nuc Cm r , ori(pWV01), with a DNA fragment encoding the precursor SP Usp45 -Nuc expressed under under P nisA transcriptional control [18] pNZ9520 Em r , nisRK cloned in pIL253 [17] pMEC10 Em r , integration plasmid containing the nisRK genes and the int-attP cassette for integration to the tRNA Ser locus. [21] Dose-dependent nisin induction of staphylococcal nuclease production and secretion

Cell lysates Supernatants
tion efficiency was evaluated to 6.5% of the total BLG production. As shown on the corresponding Western blot analysis (Fig. 3), only the mature form of LEISS-Nuc-BLG protein was secreted. Moreover, 46% of LEISS-Nuc-BLG protein exhibited a native conformation, showing thus that presence of Nuc does not influence the BLG folding since similar proportion of the native form was measured with LEISS-BLG. In the Cs fraction, proportion of the native form was even improved compared to LEISS-BLG since 8% of the fusion protein exhibited the BLG native form. Nevertheless, most of the LEISS-Nuc-BLG synthesized was located in the Ci fraction and intracellular LEISS-Nuc-BLG concentration was about 2.5-fold lower than intracellular LEISS-BLG concentration. As revealed by Western blot analysis (Fig. 3), accumulation of precursor SP Usp45 -LEISS-Nuc-BLG in cell lysates resulted also in Expression cassettes for production and secretion of staphylococcal nuclease and/or bovine β-lactoglobulin  Taken together, modifications used to improve BLG production in L. lactis are also effective in L. casei. Although BLG production in L. casei remains 2 to 10-fold lower than that obtained in L. lactis [8], we succeeded to obtain a substantial increase in BLG production and secretion. In both L. casei and L. lactis, the highest yield was obtained with the LEISS-Nuc-BLG form. While in L. lactis the optimal secretion was obtained with the LEISS-BLG form, the highest level of secretion in L. casei was reached with the LEISS-Nuc-BLG construct. Use of the two-plasmid NICE system resulted also in a 500-fold increase in BLG production compared to the level obtained with recombinant L. casei strains possessing one copy of the blg gene into their chromosome [14]. However, Western blot analysis of Nuc production (Fig. 2, lane 0) or LEISS-Nuc-BLG production (Fig. 3, first lane) revealed that the NICE system is relatively leaky in the absence of nisin induction. BLG production was then further optimized by investigating the conditions of nisin induction.

Integration of nisRK genes into the L. casei chromosome
It has already been observed that the two-plasmid NICE system is less tightly regulated in Lactobacillus strains than in L. lactis strains [21,22]. In order to strengthen the regu-lation and to prevent the toxicity due to high expression levels of the nisin regulatory genes, the nisRK genes were integrated into the BL23 chromosome, as described by Pavan et al ([21], see Materials and Methods). Competent cells of the resulting strain, BL23(int:nisRK), were transformed with the pSEC:LEISS-Nuc-BLG and intracellular BLG productions with or without nisin induction were analyzed by Western blot. As observed on Fig. 3, no LEISS-Nuc-BLG synthesis was detected in the absence of nisin, suggesting an improved regulation of the NICE system. Compared to BL23 co-transformed with pSEC:LEISS-Nuc-BLG and pNZ9520 plasmids, a 50% increase in BLG concentration was measured in the Ci fraction but neither concentration of soluble LEISS-Nuc-BLG, nor proportion of the native form in S and Cs fractions, were significantly improved ( Table 2).

Improved production of soluble BLG in L. casei
Recombinant L. casei are currently tested as delivery vehicle of BLG to the gastrointestinal tract of mice. Such experiments require high amounts of lactobacilli. Considering that addition of nisin at early growth phase resulted in a significant growth inhibition, different protocols allowing overnight production of bacterial biomass and nisin induction of L. casei cultures at higher cell density were tested. Consequently, bacterial pellets from overnight cultures were resuspended in fresh medium in order to remove lactic acid and other bacterial metabolites (proto-LEISS-Nuc-BLG production and secretion by L. casei BL23 and BL23 (int:nisRK) strains Interestingly, 74% of intracellular BLG obtained with protocol B remained soluble when only 10% of intracellular BLG obtained with the previously used protocol A were located in the Cs fraction (Table 3). Although total BLG yield with protocol B was nearly 2-fold lower, soluble BLG concentration was about 4-fold higher than that obtained with protocol A. The higher proportion of soluble BLG was certainly due to a lower growth rate at the time of induction. As already observed in E. coli [23][24][25], the folding capacity of the cell could become limiting at high expression rates. A slower rate of heterologous gene expression could then reduce inclusion bodies formation. However, it should be noted that, although proportion of soluble BLG is drastically enhanced, less than 1% of soluble BLG was detected in its native conformation. Since administration of L. lactis producing almost exclusively denatured BLG could partially prevent mice from sensitization [6,9], conformation of the recombinant BLG seemed not to be critical to induce a protective immune response. On the other hand, considering the processing of lactobacilli and their intracellular content by the host immune system, higher concentrations of soluble antigen could be advantageous for the development of a specific immune response by favoring antigen presentation to the mucosal immune system.

Conclusion
We observed that fusion of the LEISS propeptide and Nuc, as initially described in L. lactis, improved both BLG production and secretion in L. casei. Furthermore, integration of nisRK genes into the L. casei BL23 chromosome allowed to strengthen nisin-dependent production of BLG and led to higher yields of recombinant protein. As previously mentioned in L. lactis, all these modifications did not improve the proportion of soluble BLG in the intracellular fraction [8]. Production of soluble BLG was nevertheless improved by performing nisin induction on L. casei cultures at higher cell density.
These recombinant strains were primarily designed to evaluate the potential advantages of using probiotic lactobacilli for the mucosal delivery of an antigen in a mouse model of allergy. This raised different questions such as whether we need in situ production and secretion of the antigen to induce or modulate an immune response or what is the most effective way of administration. In this regard, we are currently testing the BLG-producing L. casei for prophylactic and therapeutic treatments on mice, via delivery of recombinant lactobacilli by oral and intranasal administrations.

Bacterial strains and culture conditions
The bacterial strains used in this study are listed in Table  1. L. casei strains, derived from strain BL23 (ATCC 393 cured of plasmid pLZ15, [26]) were grown at 37°C in De Man-Rogosa-Sharpe broth ( [27], Difco, BD, Le Pont de Claix, France). Plates were incubated in anaerobic jars for 2 days at 37°C in an Anaerocult A system (Merck). When required, erythromycin (Merck, Darmstadt, Germany) was added to the media at 5 μg/mL to select L. casei transformants. For promoter induction, nisin (Sigma, St Louis, MO, USA) was added at the required concentration (2.5-50 ng/mL).

Plasmids and strain constructs
The plasmids used in this study are listed in Table 1. L. casei was transformed by electroporation using a genepulser apparatus (Bio-rad Laboratories, Richmond, California) as previously described [28]. The pMEC10 plasmid was kindly provided by Dr. P. Hols. This plasmid, unable to replicate in L. casei, was introduced in electrocompetent cells to integrate the nisRK genes into the genome of the BL23 strain [21]. Erythromycin-resistant integrants were analyzed by PCR and tested for functional