Recombinant Lactobacillus plantarum induces immune responses to cancer testis antigen NY-ESO-1 and maturation of dendritic cells

Abstract

Given their safe use in humans and inherent adjuvanticity, Lactic Acid Bacteria may offer several advantages over other mucosal delivery strategies for cancer vaccines. The objective of this study is to evaluate the immune responses in mice after oral immunization with Lactobacillus (L) plantarum WCFS1 expressing a cell-wall anchored tumor antigen NY-ESO-1. And to investigate the immunostimulatory potency of this new candidate vaccine on human dendritic cells (DCs). L. plantarum displaying NY-ESO-1 induced NY-ESO-1 specific antibodies and T-cell responses in mice. By contrast, L. plantarum displaying conserved proteins such as heat shock protein-27 and galectin-1, did not induce immunity, suggesting that immune tolerance to self-proteins cannot be broken by oral administration of L. plantarum. With respect to immunomodulation, immature DCs incubated with wild type or L. plantarum-NY-ESO-1 upregulated the expression of co-stimulatory molecules and secreted a large amount of interleukin (IL)-12, TNF-α, but not IL-4. Moreover, they upregulated the expression of immunosuppressive factors such as IL-10 and indoleamine 2,3-dioxygenase. Although L. plantarum-matured DCs expressed inhibitory molecules, they stimulated allogeneic T cells in-vitro. Collectively, the data indicate that L. plantarum-NY-ESO-1 can evoke antigen-specific immunity upon oral administration and induce DC maturation, raising the potential of its use in cancer immunotherapies.

Keywords:

• adjuvants
• dendritic cells
• Lactobacillus plantarum
• oral immunization
• tumor antigens

Abbreviations

APC	=	antigen presenting cells
DC	=	dendritic cells
HPV	=	human papilloma virus
HSP	=	heat shock protein
iDC	=	immature DCs
IDO	=	indoleamine 2,3-dioxygenase
Ig	=	immunoglobulin
IL	=	interleukin
LAB	=	lactic acid bacteria
L	=	lactobacillus
mDC	=	mature DC
PD	=	programmed cell death
Th	=	T helper

Introduction

Mucosal vaccines offer great potential since they can be administered via oral or intranasal delivery.^1,2 To induce immune responses, however, a delivery system is needed that protects antigen from degradation, facilitates antigen uptake in the gastrointestinal tract, and triggers immunity rather than tolerance often seen in oral vaccination with soluble antigens.³ Given their food grade status and history of safe use in humans, Lactic Acid Bacteria (LAB) are considered more attractive than other live vaccine vectors such as Salmonella.^1,4 Moreover, several members of LAB such as Lactobacillus (L.) casei, L. plantarum, L. acidophilus, and Lactococcus lactis, exhibited self-adjuvant properties and found to be weakly immunogenic themselves [reviewed in 1]. With respect to the effects of antigen location (cytoplasmic, secreted, or anchored to the cell wall) on immunogenicity, there is no clear consensus on which cloning strategy is preferable.^5,6 This is likely due to several factors such as the expression level of the antigen and the nature of the bacterial strain. Conflicting results are also reported regarding the comparison between intragastric and intranasal routes of administration.^7,8 Regardless of the applied strategy, in most studies mucosal immunization resulted in induction of antigen-specific immunity.

Because mucosal immunization induces not only mucosal IgA, but also systemic T-cell and antibody responses in peripheral lymphoid tissues, this route of vaccination has been used to induce immunity against infections and certain cancer types.^9–16 Indeed, a number of studies of oral vaccines generated from genetically engineered pathogenic or commensal bacteria have been reported. For example, administration of L. casei expressing a cell-wall anchored E7 antigen from human papilloma virus (HPV) 16, which is the main etiological agent of cervical cancer, induced cellular immunity and protected mice from tumor development.¹⁵ Intranasal administration of recombinant lactococci displaying E7 antigen and secreting biologically active IL-12 induced E7-specific immune responses and showed therapeutic effects against HPV-16-induced tumors.¹⁶ Interestingly, E7-expressing L. casei vaccine induced E7-specific immunity and clinical response in patients with cervical intraepithelial neoplasia grade 3,¹⁷ confirming the clinical feasibility of developed vaccines against HPV-related cervical cancer in humans.

Notably, cancer vaccines hold great promise in the treatment of certain cancers such as melanoma, and have been the focus of extensive pre-clinical and clinical testing in recent years.¹⁸ Among cancer associated antigens, NY-ESO-1 has emerged as one of the most promising targets in vaccination.¹⁹ The frequent finding of humoral and cellular immune responses against this antigen in cancer patients with NY-ESO-1-expressing tumors makes it one of the most immunogenic human tumor antigens.¹⁹ Many strategies for development of NY-ESO-1-based vaccines have been reported, including recombinant live vector vaccines and protein or peptide vaccines.^20,21 In contrast to the currently used vectors, LAB are generally regarded as safe microorganisms, and some of them are able to stimulate the immune system of the host as adjuvants due to their probiotic and anti-inflammatory properties.^1,22 Except for cervical cancer, little is known about the use of LAB in cancer vaccines. We previously reported that the 37 kDa immunogenic oncofetal antigen (OFA) expressed on the cell surface of L. plantarum can induce antibody response.²³ Encouraged by this finding, in this study we investigated whether L. plantarum expressing NY-ESO-1 can induce specific immunity in mice. As controls for immunogenicity, we included 2 conserved proteins between human and mouse, HSP-27 and galectin-1 (Gal-1). We furthermore evaluated the immunomodulatory properties of recombinant and wild type L. plantarum on human monocytes, immature (i) DCs, and mature (m) DCs derived from the same donors. Immunization analysis demonstrated that surface-displayed NY-ESO-1 induced immune responses and exhibited an adjuvant activity on iDCs, providing a new strategy for the development of cancer vaccines.

Results

Expression of the recombinant proteins

The cancer/testis protein NY-ESO-1 is a suitable model antigen to explore systemic and mucosal immune responses. Full-length NY-ESO-1 protein and control proteins (HSP-27, Gal-1) were expressed on the cell wall of L. plantarum (Fig. 1A). To evaluate gene expression, equal amounts of whole cell protein extracts of L. plantarum carrying the empty vector (pEV), expressing Gal-1, NY-ESO-1, or HSP-27 protein were analyzed by Western blotting using specific monoclonal antibodies (Fig. 1B). Immune-reactive fusion proteins with the appropriate sizes were detected. In all experiments Gal-1 seems to migrate as a double band. This is more likely due to a specific degradation because the same results were obtained with 2 different antibodies. It should be noted that in human cells Gal-1 was detected as a single isoform.²⁴ Using flow cytometry, we further confirmed the display of the expressed proteins on the surface of L. plantarum (Fig. 1C). Indeed, recombinant L. plantarum stained positively with the corresponding antibodies, whereas none of the used monoclonal antibodies stained the cells harboring the pEV vector (Fig. 1C last panel). These results validate our surface expression system to display tumor antigens for mucosal vaccines and/or other therapies.

Figure 1. Characterization of recombinant Lactobacillus. (A) Schematic overview of the expression cassette for C-terminal anchoring of the 3 target proteins, NY-ESO-1, HSP-27 and Gal-1. The expression cassette is translationally fused to the inducible P_sppA promoter. The target genes are fused through a SalI linker to a signal peptide (SP) derived from the L. plantarum. The C-terminal LPxTG anchor sequence derived from the L. plantarum Lp_2578 is translationally fused to the target genes through a MluI linker. Three variants of this linker have been developed [(named, cwa1, cwa2 and cwa3; (21)] which differ in terms of how large a part of the lactobacillal protein preceding the LPxTG motif is included; see Materials & Methods for details. Previously published plasmids that were used as starting points for these constructions contain the same restriction sites. (B) Western blotting. Whole-cell protein extracts prepared from L. plantarum expressing Gal-1, NY-ESO-1, or HSP-27 were prepared, analyzed by 10% SDS-PAGE gels, transferred to nitrocellulose and incubated with protrein-specific antibody as indicated. L. plantarum harboring pEV vector was used as a control. The three immunoblots are triplicates. (C) Flow cytometry analysis of L. plantarum expressing Gal-1 (L.p-Gal-1), NY-ESO-1 (L.p-NY-ESO-1), or HSP-27 (L.p-HSP-27). After induction of protein expression, the live cells were washed with PBS, stained with the corresponding biotin-conjugated monoclonal antibodies (grew histograms) followed by PE-conjugated streptavidin and then analyzed by flow cytometry. Black histograms correspond to cells stained with only PE-conjugated streptavidin (negative control). None of the used biotin conjugated monoclonal antibodies against NY-ESO-1, Gal-1, or HSP-27 stained L. plantarum harboring the pEV vector (L.p-pEV, last panel).

Antibody response to oral administration of recombinant L. plantarum

To evaluate the ability of the recombinant L. plantarum to induce an antigen-specific immune response, first BALB/c mice were orally immunized with L. plantarum expressing NY-ESO-1 or harboring the pEV vector. NY-ESO-1 specific serum IgG, IgG1, IgG2a, and mucosal IgA levels in intestinal lavages, were determined by ELISA. Mice vaccinated with L. plantarum expressing NY-ESO-1 developed IgG, IgG1 and IgG2a against NY-ESO-1 protein when compared to mice immunized with L. plantarum harboring the pEV vector (Fig. 2). Furthermore, mice developed IgA antibodies in intestinal lavages. Thus, both systemic (serum IgG) and mucosal (intestinal IgA) NY-ESO-1- specific antibodies were induced by oral administration. By contrast, mice immunized with L. plantarum expressing Gal-1 or HSP-27 did not develop antibody response (Fig. 2). These results may be expected because both proteins are conserved between human and mouse.

Figure 2. Antibody responses to L. plantarum expressing NY-ESO-1. BALB/c mice immunized with either L. plantarum expressing NY-ESO-1 (A) or harboring the pEV vector (B) were analyzed for the presence of NY-ESO-1-specific IgG, IgG1, IgG2a, and IgA by ELISA. Bars represent the median value of each experimental group (n = 6 ).

Oral immunization induces antigen-specific cellular immune response in spleen

In mice the IgG2a B-cell response is associated with a Th1-like immune response, whereas IgG1 with a Th2-like response. The induction of both responses requires T-cell help. Therefore, we have tested the ability of L. plantarum expressing NY-ESO-1 to induce T-cell response. After vaccination, splenocytes were prepared on day 40 from control and immunized mice and T-cell proliferation in response to recombinant NY-ESO-1 was assayed by thymidine incorporation (Fig. 3). Spleenic T cells isolated from L. plantarum-NY-ESO-1 immunized mice proliferated upon in-vitro stimulation with recombinant NY-ESO-1 (p < 0.001). Mice orally immunized with the L. plantarum harboring pEV vector showed a moderate response that was similar to that seen in mice injected with PBS buffer (control). In accordance with the antibody response, spleenic T cells from mice immunized with L. plantarum expressing HSP-27 or Gal-1 protein failed to proliferate in response to recombinant proteins (data not shown).

Figure 3. T-cell response to NY-ESO-1. Spleenic cells from mice immunized with either L. plantarum expressing NY-ESO-1 or L. plantarum harboring the vector (pEV) were prepared and stimulated for 6 d with NY-ESO-1 protein (10 μg/ml). Thereafter, T-cell proliferation was determined by incorporation of [³H]-thymidine following overnight pulsing. The results are expressed as the mean values of stimulation index ±SD for triplicate determination. Stimulation index is calculated by dividing the number of cpm for NY-ESO-1-stimulated cells by the number of cpm in unstimulated cells. Control = mice injected with PBS buffer. The data are representative for 5 separate experiments. ***p < 0.001.

Dendritic cell response to recombinant and wild type L. plantarum

Given the potential use of L. plantarum-based cancer vaccines in humans,¹ we investigated the responses of iDC and mDC to wild type L. plantarum expressing NY-ESO-1. In these experiments, iDC and TNF-α-matured DC (mDC) derived from the same donors were co-cultured with live wild type or L. plantarum expressing NY-ESO-1 for 18 h, culture supernatants were collected and the amount of pro-inflammatory cytokines TNF-α, IL-12, and anti-inflammatory cytokine IL-10 was quantified by ELISA. Control samples were un-stimulated cells. L. plantarum expressing NY-ESO-1 or harboring the pEV vector induced high levels of IL-12 and TNF-α (Table 1). Notably, the level of IL-10 was significantly lower than that of IL-12 (900 ± 150 vs 3100 ± 500 pg/ml). In contrast to iDC, mDC did not respond to L. plantarum, indicating that the signaling pathways activated in iDC are switched off in mDC. Comparable cytokine induction profiles were obtained with UV-killed L. plantarum (data not shown). We also quantified the Th-2 cytokine IL-4. Both recombinant and wild type L. plantarum did not induce IL-4 production (Table 1). Similar to iDCs, monocytes prepared from the same donors responded to L. plantarum by producing IL-12, TNF-α and IL-10 (data not shown).

Table 1. Cytokine production by L. plantarum-treated DCs

Cytokines (pg/ml)	IL-4	IL-10	IL-12	TNF-α
iDC	5 ± 2	30 ± 5	6 ± 2	7 ± 3
iDC + LP-pEV	20 ± 5	900 ± 150	3100 ± 500	2400 ± 200
iDC + LP-NY-ESO-1	20 ± 5	1100 ± 200	3259 ± 650	2500 ± 300
mDC	1 ± 2	8 ± 1	10 ± 2	13 ± 1
mDC + LP-pEV	1 ± 1	7 ± 2	9 ± 1	10 ± 2
mDC + LP-NY-ESO-1	2 ± 2	5 ± 3	8 ± 2	14 ± 3

Immature DC (iDC) or TNF-α-matured DCs (mDC) prepared from the same donor were incubated for 18h with live L. plantarum expressing NY-ESO-1 (LP-NY-ESO-1) or harboring the pEV vector (LP-pEV) (10⁷ CFU/ml each). Subsequently, cytokine contents in culture supernatants were measured by ELISA. The results are presented as mean values ± SD for triplicate determination and are representative for at least 18 separate experiments using cells from different donors.

iDC-matured with L. plantarum upregulated the expression of co-stimulatory molecules

To investigate whether L. plantarum-treated iDCs upregulate the expression of co-stimulatory molecules, cells were stimulated for 48 h with either wild type, recombinant L. plantarum or TNF−α and then they were stained with marker-specific antibodies. Data were acquired on a flow cytometer and expressed as mean fluorescence (Table 2). Both recombinant and wild type L. plantarum increased the expression all markers when compared to un-stimulated iDCs. Moreover, the expression of CD80, CD86 and CCR7 was significantly increased as compared to TNFα-matured DCs (p < 0.01). In general, TNF-α-matured DCs showed less pronounced phenotype as the wild type or recombinant L. plantarum-matured DCs.

Table 2. Surface expression of the activation markers on DCs

	Mean fluorescence intensity
Markers	CD40	CD80	CD86	CCR7	HLA-DR
iDC	850	4301	1372	284	35000
iDC + TNF-α	1615	7284	3789	785	143000
iDC + L. plantarum	1522	17000	8953	1274	129000
iDC + L. plantarum-NY-ESO-1	1600	1850	82500	11200	132000

Expression of the indicated markers on iDC was measured by fluorescence-activated cell sorting (FACS) after activation for 48 h with either TNF-α (50 ng/ml), wild type or L. plantarum expressing NY-ESO-1 (10⁷ CFU/ml each). The data are from one experiment and are representative for 3 separate experiments.

L. plantarum- matured DCs upregulated the expression of immunosuppressive factors

Since the programmed death (PD, B7-H1)-ligand-1 and IDO play a major role in immune regulation,²⁵ their expression in DCs exposed to L. plantarum was analyzed. The expression level of PD-L1 was significantly increased (p < 0.01) in L. plantarum-treated iDCs as compared to that of TNFα-treated cells (Fig. 4A). Moreover, L. plantarum-treated iDCs upregulated the expression of IDO, an immunosuppressive enzyme (Fig. 4B). In contrast, TNF-α treated iDCs did not express IDO (Fig. 4C). Similar to the wild type, L. plantarum expressing NY-ESO-1 induced IDO and PD-L1 gene expression (data not shown).

Figure 4. Expression of PD-L1 and IDO by iDC in response to L. plantarum. (A) PD-L1 expression. iDCs were stimulated with TNF-α or live L. plantarum for 48 h and then the expression of PD-L1 was analyzed by flow cytometry using a specific monoclonal antibody. (B) Expression of IDO in response to various L. plantarum concentrations. iDC were stimulated for 18 h and then protein extracts were prepared and analyzed by Western blotting. Lanes 1, 2, and 3 correspond to cells stimulated with 10⁵, 10⁶ and 10⁷ CFU/ml, respectively. (C) As in B, iDCs were stimulated with L. plantarum (10⁷ CFU/ml) or TNF-α (50 ng/ml) and then the expression of IDO was analyzed by Western blotting. Ct = unstimulated cells. MF = mean fluorescence intensity. The data are representative of 4 separate experiments.

L. plantarum-matured DCs induced T-cell proliferation

Having shown that L. plantarum treated iDCs express co-stimulatory molecules and secreted IL-12, next we investigated whether the increased expression of inhibitory factors (e.g., IL-10, IDO, PD-L1) by L. plantarum affects T-cell response. In these experiments, we used purified CD4+ T cells from one donor as responder and DCs from a second donor as stimulator in mixed lymphocyte reaction (MLR) assays (Fig. 5). iDCs induced only weak proliferative responses in CD4+ T cells. As expected, TNF-α matured DCs induced a strong allogeneic response as compared to iDCs. The potency of L. plantarum-matured DCs to induce T-cell proliferation at DC/T cell ratios of 1:40 and 1:20 was comparable to that of TNF-α matured DCs. However, at DC/T ratio of 1:10, L. plantarum-matured DCs were less T-cell stimulator than TNF-α matured DCs (p < 0.01).

Figure 5. Potency of various DC preparations to stimulate allogeneic T cell proliferation. DCs were co-cultured with allogeneic CD4+ T cells at a T cell/DC ratio of 1:40, 1:20 or 1:10. Cells were co-cultured for 5 d at 37°C and T-cell proliferation was determined by incorporation of [³H]-thymidine following overnight pulsing. The results are presented as mean values ±SD for triplicate determination and are representative for 3 separate experiments. p < 0.01.

Discussion

Oral vaccines, depending on their composition and nature, can activate every effector arm of the immune system.¹ The present study demonstrated the expression of NY-ESO-1 antigen at the surface of L. plantarum and confirmed that oral immunization with this recombinant bacterium induced both humoral and T-cell responses in mice. Although L. plantarum matured iDC expressed IL-10, IDO, and PD-L1 as potential inhibitory molecules, they activated T cells more likely via the overexpression of IL-12 and co-stimulatory molecules. IL-12 is a critical factor in switching naïve or memory T cells to Th-1 response against infections and other disease such as cancers.²⁶ It is generally acknowledged that excessive activation of immune effector cells may result in host detrimental conditions, such as in chronic inflammation and autoimmunity. To ensure the proper regulation of effector mechanisms, the production of proinflammatory molecules (e.g., IL-12, TNF-α) is always associated with the upregulation of secreted and membrane-bound inhibitory factors such as IL-10 and PD-L1.^25,27 The ability of L. plantarum WCFS1 strain to induce counter-regulatory cytokines should provide an important regulatory mechanism and may explain their anti-inflammatory and protective effects ^22,28

Most studies in the literature have focused on mucosal vaccination against infections.¹ With respect to cancer immunotherapy, recently Kawana and colleagues demonstrated that oral vaccination with L. casei expressing modified E7 protein can induce clinical response in patients with cervical intraepithelial neoplasia.¹⁷ NY-ESO-1 tumor antigen that is expressed in a variety of human malignancies, but not in normal tissues except for the testis and placenta.¹⁹ Therefore, a vaccine that can induce both mucosal and systemic immunity might be more effective for eradicating NY-ESO-1-expressing tumors than parenteral vaccines. Our strategy of expressing NY-ESO-1 in L. plantarum induced both systemic and mucosal immunity after oral administration. The present designed self-adjuvanted subunit vaccine does not require additional adjuvants to induce immune responses in mice. Consistent with our data, Meng and colleagues showed that oral vaccination with attenuated Salmonella enteric strains encoding T- cell epitopes from NY-ESO-1 can induce specific cytotoxic T-lymphocyte responses in mice.²⁰ The authors have used the Tafi display system to express NY-ESO-1 T-cell epitopes.

Notably, innate immunity plays a key role in the development of protective adaptive immune responses during vaccination. Several adjuvants are used in vaccines to activate innate immune cells, including DCs. Such activation of innate cells by a vaccine is critical to subsequent activation and differentiation of effector T cells. Since DCs are the most potent APCs,¹⁸ several studies have investigated their capacity to response to LAB. For example, monocyte-derived DC stimulated with L. Johnsonii, L. reuteri, or L. gasseri produced IL-12, but not IL-10.²⁹ Bone marrow-derived murine DCs stimulated with L. casei produced a significant amount of TNF-α, IL-6, and IL-12, whereas those stimulated with L. reuteri produced low level of IL-12.³⁰ Likewise, murine DCs stimulated with L. acidophilus produced IL-10, IL-12 and TNF-α.³¹ Human monocyte-derived iDCs stimulated with L. plantarum expressing OspA lipoprotein from B. burgdogferi produced TNF-α and IL-10, but not IL-12. By contrast, iDCs stimulated with wild type L. plantarum did not produce significant levels of TNF-α and IL-10.³² Although the induction levels of the investigated cytokines varied between donors, monocytes and monocyte-derived iDCs prepared from more than 20 healthy individuals responded to L. plantarum WCFS1 strain by secreting TNF-α, IL-12, and IL-10 (Table 1). The variation in strain immune profiles suggest that there could be some underlying strain-dependent genetic differences from the cytokine secretion. Indeed, Meijerink et al. identified gene loci in L. plantarum WCFS1 that modulate the immune response of DCs.³³ The authors used 42 different L. plantarum strains isolated from humans and different food resources. The strains differed considerably in their ability to induce cytokine expression in monocyte-derived iDCs. Consistent with the present study, the levels of IL-12, IL-10, and TNF-α induced by the wild type L. plantarum WCFS1 were around 3500, 250, and 2400 pg/ml, respectively. We found that in contrast to monocytes and iDCs, TNF-α matured DCs are unresponsive to wild type and recombinant L. plantarum WCFS1. Although further work is needed, this observation indicates that the maturation stage of DCs may influence their response to bacterial cell components.

Among the regulatory factors, IDO has emerged as one of the most used inhibitory molecule by the immune system.³⁴ IDO-competent DCs exert regulatory effects on T cells that are mediated by tryptophan depletion and by the production of metabolic byproducts collectively known as kynurenines.³⁴ As shown in this study, L. plantarum activated iDCs upregulated the expression of IDO. This observation is interesting given the anti-inflammatory properties of certain probiotic bacteria. In response to L. plantarum, DCs also overexpressed PD-L1. The engagement of PD1/PDL-1 induces negative signals to receptor signaling pathways, leading to inhibition of T-cell proliferation.²⁷ Despite the expression of these 2 inhibitory molecules, L. plantarum-matured DCs activated T cell proliferation in-vitro. However, at high DC number some inhibition effect was evident as compared to TNF-α matured DCs. Thus, depending on the number of DCs within inflammatory sites and/or lymph nodes, IL-12 might override IDO, IL10, and PD-L1-driven blockade of T-cell proliferation.

In conclusion, the present study demonstrates that genetically engineered L. plantarum expressing NY-ESO-1 can induce humoral and T-cell responses and should therefore represent a promising strategy for oral immunization against tumor expressing NY-ESO-1. Moreover, L. plantarum induction of DC maturation along with the secretion of IL-12, a major cytokine that triggers Th-1 response, should support the use of L. plantarum WCFS1 strain as a delivery vector.

Materials and Methods

Reagents

Recombinant NY-ESO-1, Gal-1 and HSP-27 proteins were purchased from ProSpec-Tany TechnologyGene Ltd (East Brunswick, New Jersey), Nordic BioSite (Täby, Sweden), and R& D systems (Minneapolis, Minnesota), respectively. Recombinant proteins were detected using the following monoclonal antibodies: NY-ESO-1 (Santa Cruz Biotechnology, Dallas, Texas), biotin-conjugated Gal-1 (R&D systems), and anti-HSP-27 (R&D systems). Antibodies against NY-ESO-1 and HSP-27 were biotin-conjugated using Sulfo-NHS-biotin (Santa Cruz Biotechnology) as described by the manufacturer's instructions. Anti-mouse IgG and IgA were purchased from Sigma (St. Louis, Missouri) and goat anti-mouse IgG1 and IgG2a were purchased from AbD Serotec (Kidlington, UK). Other antibodies were purchased from Dako (Glostrup, Denmark) or BD Biosciences (San Jose, California).

Animals

Female BALB/c and C57BL/6 mice (5 weeks old) were purchased from Harlan (Boxmeer, The Netherlands) and housed under specific pathogen-free conditions. Mice were acclimated to the new environment for at least one week after arrival before being used for the experiments. They were on average 5 to 10 weeks of age at the onset of use. Animal studies were carried out in agreement with the protocols approved by the Animal Care Committee at the Norwegian Radium Hospital, Institute for Cancer Research.

Cloning

All plasmids used in the present study are derivatives of pSIP401, a vector constructed for inducible gene expression in L. plantarum.³⁵ The parental vectors used for cloning were pSIP401 derivatives that had previously been modified for C-terminal cell wall anchoring of the heterologously expressed protein of interest.²³ The NY-ESO-1 sequence previously optimized for secretion³⁶ was codon-optimized for expression in L. plantarum WCFS1 and ordered from Genscript (Piscataway, NJ, USA) in a pUC57-vector (pUC57-NY-ESO). The NY-ESO-1 gene fragment was amplified using pUC57-NY-ESO as template with the primer pair NYESOF (TGCTTCATCAGTCGACCAAGCGGAAGGCCGA) and NYESOcwa3R (CTGGTTGGCT ACGCGTTCGCCGTTGAC) (restriction sites in italics). The resulting fragment was In-Fusion cloned (In-Fusion HD cloning kit, Clontech Laboratories, Mountain View, CA, USA) into the SalI and MluI restriction sites of pLp_0373sOFAcwa3 (4), yielding the plasmid pLp_0373sESOcwa3. The expression vector, pLp_0373ESOcwa3, was directly transformed into electrocompetent L. plantarum WCFS1,³⁷ following the protocol of Holo and colleagues.³⁸ The code “0373” indicates that secretion of NY-ESO-1 is driven by an N-terminal signal peptide (SP) derived from the L. plantarum protein Lp_0373.³⁹ The cell wall anchor referred to as cwa3 originates from L. plantarum protein Lp_2578 and comprises 128 residues (see ref. Twenty-three for more details). The Gal-1 sequence (NP_002296.1) was codon-optimized (Genscript) and cloned into a pUC57-vector pUC57-Gal-1). To allow for cloning directly into the pSIP-secretion vectors, the synthesized gene deviated for the wild-type gene as follows, the start codon in gal-1 was removed and 2 6 bp sequences corresponding to SalI and EcoRI restriction sites were introduced at sites corresponding to the extreme N-terminus and C-terminus of the amino acid sequence, respectively. The pUC57-Gal-1 vector was digested with SalI and EcoRI and ligated into the SalI/EcoRI-digested pLp_3050sNucA,⁴⁰ yielding pLp_3050sGal-1. The Gal-1 anchoring construct was made by amplifying the 3050sGal-1 fragment (Gal-1 fused to the Lp_3050 signal peptide) from pLp_3050sGal-1 using the primer pair 3050F (GGAGTATGATTCATATGAAAAAATTTAACT-TTAAAACCATGTT) and Galcwa2R (GTTCAGTGACAC-GCGTATCAAACGCCACACACTTAATC) and subsequently in-fusion cloned into the NdeI/MluI digested pLp_0373sOFAcwa2 plasmid,²³ yielding pLp_3050sGal-1cwa2. The cell wall anchor referred to as cwa2 is similar to cwa3, described above, but the anchor sequence consist of 194 residues. The hsp-27 sequence (accession no. NP_001531.1) was amplified from cDNA with the primer pair HspF (GTCGACACCGAGCGC-CGC) and HspR (GAATTCTTACTTGGCGGCAGTCTCATCGGAT) and directly sub-cloned into the PCR-Blunt II TOPO vector (Invitrogen, Carlsbad, CA). The hsp-27 gene was then re-amplified from the TOPO vector using the primer pair 3050sHsp27F (GGCCTCCAAGGTCGACACCGAGC GCC-GC) and Hspcwa2R (GTTCAGTGACACGCGTCTTGGCG-GCAGTCTCATCGGA) and In-Fusion cloned into the Sal/MluI-digested pLp_3050sGal-1cwa2, yielding pLp_3050sHsp-27cwa2. Both pLp_3050sGal-1cwa2 and pLp_3050sHsp-27cwa2 were transformed into E. coli TOP10 (Invitrogen) for generation of plasmid DNA subsequently used for electroporation into L. plantarum WCFS1. All PCR generated inserts were verified by DNA sequencing.

Culture conditions

Frozen stocks of recombinant L. plantarum strains were used to inoculate Man-Rogosa-Sharpe (MRS) agar and broth medium supplemented with erythromycin (5μg/ml) at 30°C in an anaerobic cabinet. Bacteria are usually collected during the log phase corresponding to approximately 10⁹ CFU/ml. To induce gene expression, the cells were grown overnight at 30°C in MRS medium containing erythromycin (5 μg/ml). The overnight cultures were diluted to 1/100 in MRS medium and cultured for 2 h at 30°C. Gene expression of the recombinant proteins was then induced by adding SppIP peptide pheromone to a final concentration of 25 ng/ml. Bacteria were subsequently collected when their OD₆₀₀ corresponded to 5 × 10⁸ cell/ml (log phase). Cells were harvested by centrifugation at 12000 g for 10 min at 4°C, washed twice in sterile phosphate-buffered saline (PBS) and re-suspended in PBS at a concentration of approximately 10⁹ CFU/ml. Gene expression was always verified by Western blotting.

Analysis of gene expression by immunoblotting and flow cytometry

To analyze the expression of the cloned proteins, subsequent to gene induction, L. plantarum cultures were collected by centrifugation at 12000 g for 10 min. The pellets were washed twice with PBS buffer and total proteins were prepared with boiling SDS-extraction buffer (phosphate buffered saline, 5% SDS). Proteins were separated on 10% SDS-PAGE, electro-blotted onto nitrocellulose membranes, blocked for 60 min at room temperature in 5% non-fat milk Tris-buffer-saline (TBS) and then incubated overnight at 4°C with mouse monoclonal antibodies against human NY-ESO-1, HSP-27, or Gal-1. After washing, the membranes were incubated with HRP-conjugated anti-mouse IgG antibodies (Dako, Glostrup, Denmark) and reactive bands were visualized by an ECL kit. For flow cytometry, recombinant L. plantarum (5 × 10⁵ cells/well/200 μl staining buffer) were incubated for 30 min at 4°C with biotin-conjugated monoclonal antibodies against Gal-1, NY-ESO-1 or HSP-27. Thereafter, the cells were incubated with PE-streptavidin for 30 min, washed and then analyzed with FACSCanto II flow cytometer (BD Biosciences, San Jose, CA, USA). Data were analyzed with FlowJo software.

Immunization

After gene induction, recombinant L. plantarum cells were harvested, washed twice with PBS and resuspended in PBS and used for immunization. Administration was performed intragastrically using a ball-tipped syringe for oral gavage inoculation (6 BALB/c mice/group). Each mouse received approximately 5 × 10⁸ CFU recombinant L. plantarum. The control group received L. plantarum harboring the pEV, a control vector without target gene. Mice received 3 to 4 immunizations. One week after the last immunization, mice were bled and sera were analyzed by ELISA for the presence of antibodies against the recombinant proteins. At the end of the experiments, mice were killed and blood, intestinal lavages and spleens were collected for further analysis. With respect to T-cell response, prepared splenocytes were incubated with recombinant proteins for 5 days, pulsed with [³H]-thymidine and then DNA-associated reactivity was quantified by using a gamma counter. Some C75BL/6 mice were also immunized with L. plantarum expressing NY-ESO-1, HSP-27, or Gal-1 protein.

ELISA

Sera and intestinal lavages from orally inoculated mice were tested by indirect ELISA for the presence of total IgG, IgG1, IgG2a or IgA to recombinant NY-ESO-1, HSP-27, or Gal-1 protein. Purified recombinant proteins were coated at 0.2 μg/well/100 μl coating buffer, blocked and ELISA was performed using serum 1/100 and intestinal lavage (1/20). Subsequent to incubation, plates were washed and further incubated with alkaline phosphatase-conjugated goat-anti mouse IgG or IgA. For the detection of IgG1 and IgG2a antibodies, goat anti-mouse IgG1 or IgG2a antibodies conjugated to alkaline phosphatase were used. Absorbance values were read at 405 nm on an ELISA plate reader.

Preparation of monocytes-derived DCs

Peripheral blood mononuclear cells (PBMCs) were isolated from blood donor buffy coats by density-gradient centrifugation with Lymphoprep (Nycomed, Oslo, Norway). Monocytes were isolated by plastic adherence. Immature monocyte-derived dendritic cells (iDCs) were generated by culturing monocytes in complete RPMI 1640 medium supplemented with granulocyte macrophage-colony-stimulating factor (GM-CSF, 25 ng/ml; R&D systems) and interleukin 4 (IL-4, 50 ng/ml; R&D systems) for 5 d To generate mature (m) monocyte-derived DCs, iDCs were stimulated with tumor necrosis factor α (TNF-α 50 ng/ml, R&D systems) for additional 2 d In parallel, iDCs were stimulated with live L. plantarum (10⁶ or 10⁷ CFU/ml) for 2 d in order to investigate the adjuvant activity of the bacteria.

Cytokine induction

Monocytes, iDCs or mDCs derived from the same donors were resuspended in RPMI medium supplemented with 10% FCS and antibiotics and then seeded in 12-well tissue culture plates at 10⁶ cells/ml. Cells were stimulated with live wild type or L. plantarum expressing NY-ESO-1 (10⁷ CFU/ml) for 18 h. Negative control cultures contained unstimulated cells. In some experiments, the cells were also stimulated with UV-killed L. plantarum. Thereafter, culture supernatants were collected and TNF-α, IL-12, IL-10, and IL-4 were quantified by ELISA (Quantikine, R&D Systems). For phenotypic analysis, iDCs were stimulated with the test molecules for 48 h. All cultures were incubated at 37°C in a humidified atmosphere containing 5% CO₂.

Phenotypic analysis

The phenotype of DCs was analyzed by direct immunofluorescence staining of cell surface antigens using fluorochrome-conjugate mAbs for CD40, CD80, CD83, CD86, CCR7, B7-H1 (PD-L-1), and appropriate control Ab isotypes. After staining, the cells were analyzed on FACSCantoII flow cytometer. The data were processed using FlowJo software and expressed as mean fluorescence intensity.

Allogeneic stimulation of CD4+ T cells with monocyte-derived DCs

Allogeneic CD4+ T cells were isolated from PBMCs by direct magnetic labeling with anti-CD4+ Dynabeads according to the manufacturer's protocol (Dynal Biotechn, Oslo, Norway). Thereafter, T cells (10⁵/250 μl medium) were cocultured with allogeneic iDCs, TNF-α-matured DCs or L. plantarum-matured mDCs at DC/T cell ratios 1:10, 1:20, and 1:40. Co-cultures were incubated for 5 d and then pulsed with 1 μCi/well of [³H]-thymidine for 18 h before harvesting onto glass filters.

Statistical analysis

All data are presented as means ± standard deviation. Statistical analysis was performed using GraphPad Prism, version 5.00 for Windows.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Funding

This work was supported in part by grants from the Gene Therapy Program at the Oslo University Hospital to M. Sioud.