Non-pathogenic lactic-acid bacteria (LAB), such as particular species of lactococci and lactobacilli, have been handled for centuries in the fermentation and preservation of food. Sequencing of the entire genome of a number of heterofermentative L. lactis strains (14, 15) has led to the design of a plethora of genetic tools to engineer these gram-positive bacteria into next-generation mucosal delivery tools for bioactive peptides. Moreover, L. lactis strains are specifically important because of their use in the production of probiotic dairy products (16). L. lactis consists of three subspecies: L. lactis subsp. cremoris, L. lactis subsp. hordniae, and L. lactis subsp lactis. The L. lactis subsp. cremoris MG1363 is the international archetype for LAB genetics; it is a plasmid-free and phage-cured derivative of the dairy starter strain NCDO712-lacking extracellular proteases. The removal of the pLP712 plasmid, which encodes the lac operon and proteases necessary for casein degradation, precludes growth in milk thus limiting propagation of this strain outside well-controlled environmental niches (17). Today, there is sufficient knowledge to support the exploitation of GM LAB, and especially the L. lactis subsp. cremoris, to distribute therapeutic proteins to the mucosal surfaces (18). The widespread historical use of the L. lactis strains in the food industry rendered them with an important “generally regarded as safe (GRAS)” status by the Food and Drug Administration (FDA). Moreover, L. lactis strains do not colonize the GIT of humans and animals.

Engineering these bacteria to secrete active therapeutic biologicals can be advantageous for multiple reasons (Figure 1): (1)

The L. lactis-based delivery system can circumvent the use of large amounts of crude proteins that for a large part will be broken down by digestive enzymes. Furthermore, soluble proteins have low immunogenicity and stability when given mucosally (i.e., oral, nasal…). The L. lactis can survive the entire GIT while expressing one or more bioactive proteins. This is not only cost-effective but eliminates the variation of how much is digested during transit before reaching its target. Since systemic exposure to the therapeutic biological is negligible, the chance of side effects will also be significantly lower. Although reduced viability in the human GIT due to acid sensitivity is an inherent feature of L. lactis, this can be limited through proper enteric coating of freeze-dried L. lactis (19). Another modification that can offer robust protection against bile-toxicity and gastric-acid assault is intracellular accumulation of trehalose, a known cryoprotectant, by introducing trehalose synthesizing genes (20).

Lactic-acid bacteria displaying antigens on their cellular surface have been shown to be able to elicit strong immune responses with one of the most detailed examples being mucosal vaccination against the human papilloma virus (HPV) oncoprotein E7. Mice vaccinated with L. lactis expressing the HPV E7 antigen and IL-12 were protected against HPV-16-induced tumors (36). Interestingly, they tested three cellular locations for the oncoprotein expression in L. lactis and found that intracellular production ultimately lead to protein degradation and the cell wall-anchored form of E7 gave the strongest immune response (37, 38). A similar therapy, using attenuated Lactobacillus casei, was evaluated for safety and efficacy in patients with cervical intraepithelial neoplasia grade 3 (CIN3) as a result of HPV. No adverse events were reported, demonstrating the safety profile of these GM LAB, and oral vaccination induced successful anti-neoplasm mucosal immunity (39). The authors hypothesized that the bacterial cell wall, more specifically the proteoglycan compounds, may display adjuvant properties, thus enhancing the immune response (40). It is possible to increase interactions with host cells by co-expression of adhesive factors on the bacteria cell wall as an adjuvant.

While this way of protein expression opens doors to new approaches, it remains difficult to predict which route is better. An important factor is the amount of protein expressed and how relevant that is when comparing secreted proteins to cell wall-anchored proteins. In the case of bioactive proteins, such as cytokines and hormones, extracellular production will be critical in order to get a functional molecule. Ultimately, it will be necessary to study each protein on a case-by-case basis.

Probiotic activities (i.e., having health-promoting properties) have been mostly studied in other LAB, such as those of the Lactobacillus genus (41). In a recent randomized double-blind placebo trial in human subjects, the probiotic properties of three Lactobacillus plantarum strains were studied. All strains influenced the intrinsic repair processes of the small intestinal mucosa on a gene transcriptional level and the strongest effects were demonstrated by Lactobacillus plantarum TIFN101 (42). The ability to adhere to the intestinal mucosa is one of the more important selection criteria for probiotics as adhesion is considered to be a necessity for colonization. The Lactococcus genus has often been overlooked because it is not part of the normal microflora and it does not proliferate or colonize at a specific location. Nonetheless, there are some studies attributing beneficial anti-inflammatory effects to certain subspecies of L. lactis (Figure 1). For example, the L. lactis subsp. lactis NCDO 2118 reduces symptoms of recurrent colitis in the dextran sulfate sodium (DSS)-induced colitis model. Early IL-6 production may enhance mucosal repair and preserved colonic IL-10 production which could be responsible for the marked anti-inflammatory effects (43). Furthermore, oral treatment with NCDO 2118 nearly doubled the frequencies of CD4⁺CD25⁺LAP⁺ Tregs in mesenteric draining lymph nodes and spleen. Another study found that oral administration of milk fermented with L. lactis subsp. cremoris FC protected host animals against influenza virus infection (44). The protective effects against influenza virus were mostly derived from the cell components of L. lactis subsp. cremoris FC and its metabolites, such as exopolysaccharide. Many studies have shown that L. lactis can support barrier function in terms of improved mucus, production of antimicrobial peptides, and secretion of soluble immunoglobulin (Ig) A (45).

One of the first GM L. lactis was designed to treat IBD. IL-10 plays a pivotal role in intestinal homeostasis (46) as IL-10^−/− mice spontaneously develop colitis (47) and single-nucleotide polymorphisms in IL-10 signaling have been associated with IBD in genome-wide association studies (GWAS) (48–50). Treatment with parenteral administration of recombinant IL-10 showed some efficacy in human trials for CD; however, full remission was never obtained (51–53). Some discrepancies remain concerning the required dose, since high doses exert immunostimulatory effects and lower doses lack clinical efficacy. An important unanswered question regarding systemic IL-10 administration is whether the cytokine reaches its target as the serum half-life of IL-10 is less than 3 h and IL-10 has limited mucosal bioavailability (54). Oral delivery of the recombinant protein might provide an attractive alternative; however, it is complicated by the extreme acid sensitivity of IL-10 in addition to the general complications of oral protein delivery (i.e., degradation in the GIT, hepatic first-pass metabolism and high-production costs). Intragastric administration of IL-10 secreting L. lactis (LL-IL-10) circumvents many of these issues, bringing the cytokine synthesis directly to the inflamed mucosal tissues. Pioneering work by Steidler et al. showed that with LL-IL-10 much lower amounts of IL-10 were required to reduce inflammation in DSS-induced chronic colitis, compared with systemic recombinant IL-10. Furthermore, it was possible to prevent disease-onset in IL-10^−/− mice (55). One of the many anti-inflammatory effects of IL-10 is the generation of tolerogenic dendritic cells (tDCs), which regulate intestinal homeostasis by inducing suppressor T cells. LL-IL-10 can modulate immature DCs in vitro to upregulate CD83 and CD86, which in turn will induce suppressive helper T cells (Th). This suppressive effect was 40 times more efficient with LL-IL-10 than exogenous recombinant human IL-10 (56). Even in an in vitro setting, where a potentially hostile GI environment is eliminated, it is still more efficient to deliver IL-10 by the L. lactis delivery system rather than as a soluble protein.

Based on these results, a small safety and tolerability phase-I clinical trial in 10 CD patients was initiated. Participants were treated with L. lactis in which the thyA gene was replaced with the human IL-10 gene (24). This presented the first safety test in humans for this biological containment system. Treatment with LL-IL-10 showed to be harmless, well tolerated and effectively contained in vivo, encouraging further use for human application (57). A decrease in disease activity was also observed in 8 out of 10 patients which, combined with a promising safety profile, encouraged further testing. In a larger phase-II trial (NCT00729872) the safety and environmental containment were confirmed, which were the primary objectives of the study. However, no statistically significant differences were observed between placebo and LL-IL-10 treatment in mucosal healing. The lack of clinical effect may have been due to insufficient exposure to viable bacteria which in turn can be improved by several technological modifications (19, 20, 58).

Perhaps this approach was still not sufficient to protect the synthesized IL-10 from gastric acidity and proteases. A novel tactic for IL-10 delivery by L. lactis was designed in 2009, where a plasmid, pValac, was constructed for DNA delivery into eukaryotic host cells. This strategy has demonstrated that expression of the protein encoded in the DNA vector was expressed by epithelial cells of the large intestine (59). Delivery of DNA into the target cells allows for appropriate protein expression (i.e., with the necessary PTMs and correct conformational epitopes) and recognition by the host. The L. lactis co-expressing fibronectin binding protein A (FnBPA), which likely acts as an adhesin facilitating plasmid internalization, was slightly more successful and was also able to increase intestinal secretory IgA production, an important first-line of defense mechanism. Nonetheless, this IL-10-encoding DNA plasmid pValac has been cloned into several L. lactis strains and was consistently able to diminish intestinal inflammation in a chemically induced murine model for colitis (60).

Another approach to enhance enteric IL-10 production is based on the host immune evasion strategy of the enteropathogenic Yersinia species. These bacteria can secrete the soluble protein low-calcium response V antigen (LcrV) that stimulates host-IL-10 production which in turn will facilitate bacterial survival through its anti-inflammatory effects, more specifically on IFN-γ (61). Oral administration of LL-LcrV significantly enhanced colonic IL-10 production in a TLR-2-dependent manner and was able to prevent and improve colitis in two different mouse models (62).

Since IL-10 is the key regulator of inflammatory cascades, it has also been studied in other inflammatory diseases outside the GIT. Allergic asthma, the inappropriate immune response to harmless proteins or allergens, is a major health problem with increasing prevalence worldwide (63). It is a hallmark Th2 disorder marked by recruitment of eosinophils, IgE switching, and production of several chemokines that further attract additional leukocytes. Local administration of IL-10 to this respiratory site by means of intranasal administration of LL-IL-10 was able to modulate acute airway inflammation in a murine model (64). Two xylose-inducible expression systems were tested to target IL-10 to the cytoplasm or extracellular medium (secretion). A decrease of Th2 cytokines (i.e., IL-4, IL-5), a decreased presence of local eosinophils, and improved histology of the lung tissue were observed. Interestingly, these effects were most outspoken with the L. lactis strain-producing cytoplasmic IL-10. It is noteworthy that these effects were not mediated by an increase of CD4⁺Foxp3⁺ Tregs. Perhaps increasing L. lactis inoculation to more than two exposures could improve therapeutic outcomes and have an effect on Tregs as well.

Systemic treatment with tumor necrosis factor (TNF)-α antagonists, blocking its pro-inflammatory effects, has become a standardized treatment for IBD. However, 10–30% of patients are primary non-responders and 23–46% become secondary non-responders over time, in part due to immunogenicity (65). Furthermore, there are serious side effects leading to a low compliance associated with this therapy (66). Again, the rationale here is that systemic administration requires a much higher dose in order to obtain sufficient amounts at the target site, which is the inflamed GIT. The L. lactis has been engineered to secrete mono- and bivalent neutralizing anti-TNF-α nanobodies and is able to ameliorate DSS-induced chronic colitis in mice to a comparable degree as the LL-IL-10 (67). Interestingly, LL-anti-mTNF-α also shows some effects on disease severity in established colitis in IL10^−/− mice. The anti-inflammatory effect is limited to the intestine, indicating that there is no generalized immune suppression which would leave patients vulnerable to infections or malignancies.

Interferon (IFN)-γ is first and foremost known as a key pro-inflammatory cytokine produced by T cells and natural killer cells. However, it also exerts anti-inflammatory and immunoregulatory activities making it a complex, though interesting, drug target. Sheikh and colleagues demonstrated that IFN-γ exerts its anti-inflammatory properties through inhibition of IL-23. In germ-free mice colonized with enteric microbiota, inhibition of colonic Il23a correlated with IFN-γ generation. Colonic CD11b⁺ cells seem to be the main source of IL-23 and are an IFN-γ target (68). Microbial production of soluble and functional recombinant IFN-γ was achieved in Escherichia coli (69). While many therapeutic applications require large amounts of recombinant IFN-γ for parenteral use, it has been shown that oral IFN-γ also elicits systemic suppressive effects which cannot be inhibited by circulating antibodies against IFNs (70). The production of mature, biologically active murine IFN-γ by L. lactis allows both purification and therapeutic use to target IFN-γ to the mucosal immune system (71). This formulation may prove to be useful in cases where IFN-γ has therapeutic effects on mucosal afflictions, such as oral submucous fibrosis (72).

Another successful therapeutic cloned into the L. lactis system are the trefoil factors 1, 2, and 3 (TFF1–3), a family of peptides which can promote epithelial wound healing and protect it from further damage, including the mucous epithelia from the GIT. TFFs may be involved in IBD pathogenesis and are a prospective treatment option. Mice treated with LL-mTFF-1/2/3 after DSS-induced colitis showed reduced neutrophil activation and reduced epithelial damage. Purified TFF was administered rectally and was also able to slightly improve these parameters; however, doses up to 1,200-fold higher than that secreted by LL-mTFF were tested and similar efficacy was still not achieved (73). TFFs have also been explored as therapeutics for the treatment of oral mucositis (OM), a painful, debilitating, and common complication associated with cancer treatment with limited treatment options available (74). A phase-II trial that evaluated the efficacy of recombinant human intestinal trefoil factor (rhITF) oral spray for prevention of OM showed the spray was safe, well tolerated, and effective in reducing the frequency and severity of OM in patients with colorectal cancer treated with chemotherapy (75). Human TFF1 was cloned into the L. lactis and clinically formulated into a mouth rinse, coded AG013 (76). Encouraging preclinical data from a hamster model formed the basis for a phase-Ib trial (NCT00938080) with AG013 in patients with locally advanced head and neck cancer receiving induction chemotherapy (77). No AG013 bacteria were detected in blood from any subjects, even those given the highest dose (1.2 × 10¹² CFU/day). In addition, live bacterial load and human TFF1 levels in saliva and oral mucosa were not significantly different between different doses (2.0 × 10¹¹; 6.0 × 10¹¹; and 1.2 × 10¹² CFU/day) (77). Preliminary efficacy analysis showed reduced duration of OM, supporting further study in an ongoing phase-II clinical trial (NCT03234465).

An interesting fact about the abovementioned therapies is that much lower doses of therapeutic biologicals are needed when synthesized by a L. lactis strain than when the biological exposed to the GIT as a purified protein. The L. lactis are able to make close contact with the immune cells of the LP and the intestinal epithelia (IE), which likely means the drug is made available in extremely close proximity to its target cells. Inflammation-induced architectural changes in the mucosa increase this contact, perhaps explaining the widespread success of the L. lactis as a mucosal delivery tool for IBD. Oral formulation of cytokine administration, compared with the conventional parenteral route, in its own may be sufficient motivation to give these bugs a closer look.

Many treatments with GM L. lactis aim to treat the symptoms of excessive inflammation. It is also possible to tackle an underlying pathogenesis of IBD, namely, a breakdown of immune tolerance to microbiota using L. lactis (78). In this case, HSPs are extremely interesting antigens to study since they are stress proteins known to be overexpressed in inflamed tissues in IBD (79–81), as well as linkers of pathogen-associated molecular patterns (PAMPs) expressed by commensal bacteria. T-cell-specific inflammatory immune responses to bacterial and human homolog peptides derived from HSP60/65 were found in mucosal biopsies from patients with pediatric CD (82). Circulating IgA antibodies against mycobacterial HSP65 were also elevated in CD patients (83). Clearly, bacterial and/or self-HSP play a part in the pathogenesis of IBD, which motivated their use in a therapeutic setting. Interestingly, HSP boost and direct potent peripherally induced Tregs toward inflammatory sites to reinstate self-tolerance (84). Oral administration of L. lactis secreting HSP65 is able to completely prevent DSS-induced colitis in an IL-10/TLR-2-dependent manner (85). Again, CD4⁺LAP⁺ and conventional CD4⁺Foxp3⁺ Tregs were induced.

Although many therapies using GM LAB that aim to deliver anti-inflammatory molecules to the intestine showed promising results in murine models for IBD, it is not guaranteed they will also be successful in clinical trials. One technical hurdle might be applicable to all anti-inflammatory cytokines; they must be able to reach the immune cells located in the LP to exert their effects. The DSS-induced colitis model is characterized by severe IE damage. Perhaps in patients, with less severe destruction and exposed mucosal immune system compared with the murine model therapies such as the LL-IL-10 will not be as efficient. Recent work has shown that high proteolytic activity is found in the intestine of patients with CD and UC. This enhanced proteolytic activity is mostly the result of infiltrating immune cells as well as proteases involved in apoptosis (86). Additionally, there is genetic evidence supporting a role for proteases and protease inhibitors in IBD (87). These observations lead researchers to compare the efficacy of orally administered L. lactis secreting the conventional anti-inflammatory cytokines IL-10 and TGF-β with novel strains secreting serine protease inhibitors Elafin and Secretory Leukocyte Protease Inhibitor (SLPI). Interestingly, the GM L. lactis secreting Elafin was most successful as it displayed the most significant reduction of inflammation (88). Moreover, endogenous protease inhibitors are also released by the IE, where these therapeutics are delivered, indeed supporting their superior effects compared with IL-10 and TGF-β. Protein quantity, and therefore drug quantity, plays an important role in dose-dependent therapeutics. To address this issue, the authors developed a mutated L. lactis strain with an inactivated HtrA protease (htrAΔ). This inactivation led to increased protein production and secretion with only minor effects on bacterial growth. LL-htrAΔ secreting Elafin was even more successful in reducing intestinal inflammation, showing that perhaps other therapeutics cloned into a wild-type L. lactis might also be more therapeutically effective in a protease-deficient strain.

Another contributing factor in the pathogenesis of IBD is oxidative stress signaling which leads to the production of reactive oxygen and nitrogen species that have debilitating effects on the mucosal layer and partake in disease initiation (89). While the human body has natural anti-oxidative capacities, these cannot handle the excessive oxidant load leading to oxidative stress. Some endogenous intracellular antioxidant enzymes, such as superoxide dismutases (SODs), catalase (CAT), and gluthathione peroxidase (GPX), could be used therapeutically to decrease the level of gastrointestinal oxidative stress. In fact, when GM Lactobacillus casei modified to produce SOD or CAT were given to mice prior to the induction of trinitrobenzenesulfonic acid (TNBS)-induced CD, they recovered faster and showed lower intestinal inflammation than controls (90). The anti-inflammatory activity of a Lactobacillus gasseri strain-producing manganese SOD was shown to be associated with a reduction in the severity of colitis in IL-10-deficient mice. Another ingenious method for SOD delivery was discovered by evaluating the beneficial effects of fermented milk products on murine colitis (91) and on human gut homeostasis (92). Host antimicrobial actions result in lysis of the L. lactis subsp. lactis CNCM I-1631 (L. lactis I-1631) bacteria and subsequent release of cytoplasmic SOD that scavenges extracytoplasmic reactive oxygen species and results in colitis attenuation (93).

Considering the abovementioned data, along with others reviewed elsewhere (18, 94, 95), we believe that this clearly underlines the strong potential for L. lactis as a tool to deliver active therapeutics to the mucosa to elicit robust local immune effects.

Obtaining antigen-specific tolerance in a therapeutic protocol would be desirable for many diseases which cannot always be prevented. Huibregtse and colleagues showed that it is possible to induce antigen-specific peripheral tolerance by oral administration of L. lactis secreting ovalbumin (LL-OVA) in OVA-immunized transgenic mice with OVA-specific CD4⁺ T-cell receptors (TCR) (96). Interestingly, LL-OVA was able to induce APC-mediated OVA-specific T-cell proliferation at much lower levels than purified OVA. The precise mechanism by which L lactis enhances tolerogenic signals remains unclear. However, LL-OVA clearly induced a splenic regulatory CD4⁺CD25⁻Foxp3⁺/CTLA-4⁺ population, likely iTregs. This confirms the suggestions that L. lactis alters DC functions skewing them toward Treg inducers (56). L. lactis may directly modulate antigen processing and presentation as well as the expression of co-stimulatory molecules on DCs. Most of the luminal OVA after LL-OVA feeding was found in the cecum and colon, and most of the mucosal OVA was found in the terminal ileum. Currently, it is undetermined which intestinal site is most important for tolerance induction. DCs residing throughout the GIT are able to directly sample luminal antigens through the IE. Due to the association with the IE, L. lactis may allow more efficient antigen uptake than is available through oral administration of OVA. Finally, antigen-specific IL-10 production was only observed in mice treated with LL-OVA (compared with an empty vector control or purified antigen) and OVA-specific suppression was dependent on TGF-β, a hallmark characteristic of Th3 cells. Again the L. lactis empty vector control itself had some effects, being able to significantly reduce the delayed-type hypersensitivity response to OVA as well as moderately decrease OVA-specific CD4⁺ T-cell proliferation. Collectively, these findings highlight how different Tregs (iTregs and Th3) can overlap in functionality and phenotype and hint at the complexity of the regulatory pathways involved in mucosal (oral) tolerance. Moreover, these promising results were obtained in a TCR-transgenic mouse and are not guaranteed to be replicable in a host with a normal, broad TCR repertoire. Nonetheless, if this tool can induce Tregs that can confer bystander suppression to T cells reactive to unknown or multiple antigens, this approach would be extremely valuable (97).

Several immune-mediated diseases (such as type-I allergies) are triggered by well-defined antigens. Therefore, tolerance protocols aiming at targeting these food- or inhaled airborne antigens are clearly warranted. It was shown that early feeding or intranasal administration of high antigen doses could induce tolerance in mice and that tolerized mice had more IL-10- and TGF-β-producing T cells in their Peyer patches (1, 98). A valuable asset of L. lactis as bacterial delivery vehicles for vaccines is their potential to elicit antigen-specific secretory IgA responses at mucosal surfaces. Intranasal or oral inoculation of mice with L. lactis engineered to produce β-lactoglobulin (LL-BLG), a major allergen in cow’s milk, induced specific anti-BLG fecal IgA antibodies (28). Furthermore, pretreatment with LL-BLG in the presence of IL-12-producing L. lactis (LL-IL-12) prevented a Th2-type immune response after systemic sensitization with BLG by developing a strong Th1 response that correlated with the amount of recombinant BLG produced (99, 100). Delivery of a deamidated gliadin epitope, an immunodominant epitope in celiac disease, by L. lactis to transgenic humanized non-obese diabetic (NOD) AB°DQ8 mice was also able to induce antigen-specific tolerance mediated by Foxp3⁺ Tregs that function in an IL-10 and TGF-β-dependent mechanism (35). While some probiotic LAB, Lactobacillus reuteri and Lactobacillus casei, can prime DCs to drive the development of IL-10-producing Tregs, supplying IL-10 to the intestine by delivery using L. lactis is also an option (101). Indeed, oral administration of LL-IL-10 diminished anaphylaxis significantly in an animal model of food allergy. Preventative treatment with LL-IL-10 inhibits antigen-specific serum IgE and IgG₁ production and increases antigen-specific GI IgA levels (102). Interestingly, some of the immune effects can be attributed to the L. lactis since the wild-type control also reduces antigen-specific antibodies and moderately increases IL-10 secreting cells in the Peyer’s patches.

With respect to inhaled allergens, the modulation of allergic immune responses to the major dust mite allergen Der p2 by recombinant L. lactis has recently been described (103).

Type 1 diabetes is a chronic autoimmune disease characterized by immune-mediated destruction of the pancreatic insulin-producing beta cells by autoreactive CD4⁺ and CD8⁺ T cells (104). The eventual total loss of insulin production causes patients to become reliable on exogenous insulin to manage their glycemia levels. The prevalence of T1D is estimated to be 20 million patients worldwide, with an alarming increase in incidence rate in children younger than 5 years old (105). Treatment with exogenous insulin is successful in bringing the glycemia to normal levels, both in fasting and postprandial settings. However, vascular complications, both of the macrovascular and microvascular blood vessels, are responsible for the morbidity and mortality of T1D (106). These life-threatening complications and rise in incidence emphasize the need for a cure.