The bacterial strains and plasmids used in this work are listed in Table 118192021. Escherichia coli DH5α was grown on Luria-Bertani medium and incubated at 37°C with vigorous shaking. L. lactis MG1363 was grown in M17 medium containing 0.5% glucose (GM17). Bacteria were selected by addition of antibiotics as follows (concentrations in micrograms per milliliter): for E. coli, erythromycin (100) and chloramphenicol (10); for L. lactis, erythromycin (5) and chloramphenicol (10).

DNA manipulations were performed as described 22 with the following modifications: for plasmid DNA extraction from L. lactis, TES (25% sucrose, 1 mM EDTA, 50 mM Tris-HCl pH 8) containing lysozyme (10 mg/ml) was added for 10 min at 37°C to prepare protoplasts. Enzymes were used as recommended by suppliers. Electroporation of L. lactis was performed as described 23. L. lactis transformants were plated on GM17 agar plates containing the required antibiotic and were counted after 2-day incubation at 30°C.

The eukaryotic region of pValac was obtained from the pVAX1 vector (Table 1). A 860 bp DNA fragment was generated by PCR using a polymerase with proof reading activity (Platinum pfx high fidelity polymerase, Invitrogen, Sao Paulo, Brazil) and the oligonucleotides CMVBglFwd (5' GG

AGATCT

ATCGAT

The gfp ORF was cloned into pValac in order to evaluate its functionality. The 726 bp gfp ORF, obtained from pEGFP-N1 plasmid (Table 1), was digested with XbaI and BamHI restriction enzymes. The gfp ORF fragment obtained was purified (S.N.A.P. gel purification kit, Invitrogen) and then inserted into the pValac MCS using the same restriction enzymes resulting in pValac:gfp (4468 bp). The integrity of the gfp ORF was confirmed by sequencing as described above.

The pValac:gfp plasmid was assayed for GFP expression by transfection into Porcine Kidney cell line (PK15 cells). Fifty to 80% confluent PK15 cells were cultured in Dulbecco modified Eagle medium, 10% fetal calf serum, 2 mM L-glutamine (BioWhittaker, Cambrex Bio Science, Verviers, Belgium), 100 U penicillin and 100 g streptomycin. PK15 cells were transfected with 1.6 μg of pValac:gfp, pEGFP-N1 (positive control) or pIL253 (negative control) previously complexed with Lipofectamine 2000 (Invitrogen). pIL253 was used as a negative control due to the fate that it is an empty lactococcal plasmid; being more suitable for our next step (see below). The GFP-producing cells were visualized 48 hours after transfection with an epifluorescent microscope (Nikon Eclipse TE200 equipped with a digital still camera Nikon DXM1200). Transfection assays were performed in triplicate.

To demonstrate the efficacy of pValac as a delivery vector, LL-inlA+ strains were transformed with pValac:gfp (LL-inlA+ pValac:gfp+) (Table 1). In vitro invasion assays of bacteria into human cells were performed using the human colon carcinoma cell line Caco-2 as previously described 25 with some modifications 18. Briefly, eukaryotic cells were cultured in P6 wells plates containing 1 × 10⁶ cells per dish in RPMI supplemented with 2 mM L-glutamine and 10% fetal calf serum. LL-inlA+ pValac:gfp+ or L. lactis pIL253 pValac:gfp+ (LL-pIL253 pValac:gfp+) (negative control) (Table 1) (OD600 = 0.9–1.0) were added to mammalian cells so that the multiplicity of infection (MOI) was about 10³ bacteria/cell. After three hours of internalization, cells were treated for two hours with gentamicin (20 mg/ml) to kill extracellular bacteria. Fluorescent cell quantification was evaluated at 24 and 48 hours after gentamicin treatment by flow cytometry on Fluorescent Activated Cell Sorter (FACS, Becton Dickinson, France). The GFP-producing cells were visualized with an epifluorescent microscope (Nikon Eclipse TE200 equipped with a digital still camera Nikon DXM1200). Internalization and FACS assays were performed in triplicate.

In this work, which is part of an ongoing project geared to implement safer strategies for DNA deliver and expression into eukaryotic cells, we reinforce the use of Lactococcus lactis as DNA delivery vehicle 1826. To improve our delivery DNA strategy, we constructed a new expression plasmid, the pValac.

pValac is depicted in Figure 1A. It harbours the eukaryotic region containing the CytoMegaloVirus promoter (pCMV), a multiple cloning site (MCS), and the polyadenylation signal of Bovine Growth Hormone (BGH polyA) needed for a gene expression by eukaryotic host cells. Its prokaryotic region contains the RepA/RepC replication origin for both E. coli and L. lactis and a chloramphenicol resistance gene (Cm) for bacteria selection. The MCS (Figure 1B), inserted between the eukaryotic promoter pCMV and the BGH polyA, carries some potential restriction enzymes that can be used to clone a gene of interest and the T7 primer binding site for its sequencing.

The pValac:gfp ORF was used for transfection assays into PK15 cells. Forty-eight hours after transfection with pValac:gfp and pEGFP-N1 (positive control), we observed comparable GFP expression in these epithelial cells (Figure 2A and 2B, respectively). No GFP expression was observed after transfection with pIL253 (Figure 2C). This result demonstrates that pValac is functional and it could be used for our further experiment.

Internalization of LL-inlA+ pValac:gfp+ into Caco-2 cells led to GFP expression in approximately 1% of the cells 48 hours after cell invasion (Figure 3, Panels A1 and B1). A very low percentage of GFP expression was detected when the non-invasive control strain LL-pIL253 pValac:gfp+ was used (Figure 3, Panels A2 and B2). A MOI of 10² bacteria/cell is generally used for pathogens like L. monocytogenes 25 due to its virulence factors that helps bacteria to escape from vacuoles 27. Here we used a higher MOI (10³bacteria/cell) for L. lactis, a suitable multiplicity required for an efficient internalization for these bacteria 18. We thus demonstrate that invasive LL-inlA+ pValac:gfp+ were able to invade Caco-2 cells and to deliver a functional expression cassette (pCMV:gfp) into epithelial cells.

It is worth to note that concerning expression data, it is not surprisingly that we had comparable levels of approximately 1% using pValac:gfp or pVE3890 18 since both plasmids contain the same eukaryotic genetics components, the pCMV promoter and the BGH polyA. Nevertheless, we observed that pValac is more suitable for cloning and transformation in lactococci. It is easily comprehensible that working with a small plasmid is advantageous to assemble these molecular techniques 28293031 than working with a big size plasmid.

The hypothesis for DNA delivery and expression is based on the infection of host cells by bacterial carriers: following internalization, invasive L. lactis is probably taken up in the vacuoles and target for degradation, thereby releasing pValac:gfp. Then, by an unknown mechanism, the plasmid escapes the vacuoles and reaches the nucleus where the gene (gfp ORF in our case) could be translated by the host cell 323334. Questions arise if while non-recombinant lactic acid bacteria are generally regarded as safe they still be viewed as such when invasive. L. lactis inlA+ survival rate was measured and showed a decrease from 4,5 log CFU/ml for 24 hours after internalisation to 2 log CFU/ml after 60 hours (data not shown). In fact, it was already suggested that L. lactis vaccine vectors engineered to access the cytoplasmic antigen presenting pathway are incapable of further growth in this environment 3536. This result suggests that, in vitro, these bacteria could still be regarded as safe when engineered to be invasive.