Many live attenuated or inactivated vaccines against a variety of pathogens such as Salmonella require booster immunizations to attain the expected protection. Since vaccine efficacy may increase with the use of adjuvants, research on adjuvant performance is necessary. In this context, the positive effect of EDA, when administered with non-live Salmonella antigenic preparations, was studied in a vaccination-challenge mouse model involving two Salmonella Enteritidis rough mutants (SEΔwaaL and SEΔgal) differing in the LPS-Core composition. None of these mutants produced LPS O-PS antigen, which may help to distinguish animals vaccinated with these mutants from those infected by field strains. The absence of either (O-PS)-to-LPS assembly enzymes (SEΔwaaL) or LPS-Core galactose synthesis enzymes (SEΔgal) led to an intact (in SEΔwaaL) or a defective (in SEΔgal) LPS-Core, according to results on SDS-PAGE and susceptibility to SEPTs bacteriophages, as expected from the genetic design of these mutants and previous findings on similar mutants [20, 22, 40, 41]. Functional mutants in the gal operon, which includes gal E, gal T, gal K and gal M genes, all involved in the synthesis of the LPS-Core galactose , have been obtained by inactivation of the galE gene , but these single-gene mutants may revert to a smooth-LPS by incorporating exogenous precursors of galactose into the biosynthetic pathway, both in vitro and in vivo[21, 43, 44]. Thus, the mutant production strategy applied in this work, based on the use of a complete deletion of gal operon to produce the SEΔgal mutant, ensured the rough phenotype of this mutant through the blockage of galactose synthesis from endogenous or exogenous sources .
When attempting the design of non-live vaccines, HS extracts (enriched in outer membrane components) and formalin-inactivated bacteria (bacterins that retain all the external and internal bacterial antigens) were used as antigens [19, 37]. Differences in vaccine efficacy between studies using Salmonella Enteritidis HS extracts in mice [19, 45] could be explained by differences in bacterial genetic makeup, extract preparation and enrichment methods or immunization vehicles. The HS preparations obtained from the Salmonella Enteritidis rough mutants reported here did not protect mice against a virulent challenge, whereas mice immunized with HS-SEwt were protected (100% mouse survival and 80% uninfected spleens), demonstrating the essential role of the LPS O-PS in HS-driven protection. The physicochemical characterization of antigenic preparations confirmed that bacterins had a broader protein spectrum compared to HS extracts. This difference could be related to differences in preparation methods (boiling, ultracentrifugation and dialysis for HS extracts and not bacterins). Bacterins lacking LPS O-PS may be useful vaccines against heterologous Salmonella species and serovars, since the rough phenotype has an enhanced immunogenicity of minor antigens, mainly porins and lipoproteins conserved in Salmonella serotypes . Interestingly, B-SEΔwaaL induced partial protection in mice whereas B-SEΔgal did not confer protection, indicating that a complete LPS-Core could play an essential role.
Once verified that the protection obtained with HS or bacterins from both rough mutants was below that of the B-SEwt, the potential of EDA as immunopotentiator in non-live bacterial vaccines was assessed. Adjuvants such as Freund’s complete or aluminium hydroxide do not appear to improve the immune response against Salmonella[46, 47], in contrast with polymeric carriers used to adsorb or encapsulate bacterial extracts [19, 48]. Alternatively, adjuvants interacting directly with TLRs have successfully immunopotentiated Salmonella Enteritidis sub-cellular fractions. Examples are polyinosinic:polycytidylic acid [poly(I:C)] with TLR-3 , CpG sequences with TLR-9  and bacterial LPS with TLR-4 [46, 50]. Since EDA activates TLR4 favouring viral antigen presentation [6–8, 39], mice were immunized in the presence of recombinant EDA (produced in E. coli) or MEDA (in transformed plant chloroplast), initially as a simple physical mix with HS extracts (EDA) or bacterins (EDA and MEDA) from Salmonella rough mutants.
The effect of EDA in the efficacy of immunizations depended on the type of antigenic preparation (found in bacterins and not HS extracts) and the mutant from which the bacterins were prepared (B-SEΔwaaL or B-SEΔgal). No significant adjuvant effects were observed with EDA or MEDA treatments when B-SEΔwaaL was administered alone (which already conferred partial protection), but both EDA or MEDA significantly enhanced protection when using B-SEΔgal. This suggests that the size and/or composition of LPS-Core may have affected the affinity for TLR-4 and/or may have regulated the intracellular fate of the antigen in dendritic cells, as demonstrated for LPS O-PS antigen . Possibly, the complete (but not the incomplete) LPS-Core antigen competes with EDA for TLR-4 recognition, so that EDA is not free to interact with this receptor. Alternatively, EDA and MEDA may have a higher affinity for their surface receptors when exposed in absence of the external LPS-Core (i.e. B-SEΔgal).
In search of strategies that would help to enhance the binding of EDA to the antigen, the novel recombinant EDA fused to streptavidin molecule (EDAvidin) allowed a significant binding to biotinylated bacterins. Most likely, biotinylated bacterins decorated with EDAvidin enhanced the targeting of LPS defective bacterins to TLR4 expressing cells, modulating the entry of the antigen and/or its intracellular fate and/or the persistence in dendritic cells  to favour the enhancement of the efficacy of these antigenic preparations. This is in line with the significantly improved protection conferred by BEDA-SEΔwaaL and BEDA-SEΔgal complexes compared to bacterins alone, reaching levels similar to those obtained with the live rough mutants and, in the case of BEDA-SEΔwaaL, the levels conferred by the bacterin B-SEwt positive control. Irrespective of the decreased binding of SEΔwaaL to EDAvidin (apparently related to a lower level of biotinylation according to flow cytometry results using CFSE), the protection conferred by each individual bacterin including SEΔwaaL increased significantly in the presence vs. absence of EDAvidin. Altogether, these findings demonstrate that EDA in the form of EDAvidin-biotin complexes improves the efficacy of non-live vaccines. Like in previous work , increased IgG + IgM levels or a Th1 biased response (according to the IgG2a/IgG1 balance) could not be correlated with the protection conferred by both BEDA preparations (BEDA-SEΔwaaL and BEDA-SEΔgal), even though the immune response must have been in both cases sufficiently enhanced to confer significant protection.
Most studies in mice designed to assess Salmonella vaccine efficacy use a lethal challenge model. Here, we have used a sub-lethal dose challenge model  to preserve animal welfare, yielding information in line with that obtained with the lethal challenge model, since e.g. here live SEΔwaaL performed better than SEΔgal, like in previous lethal challenge reports with similar mutants . At the same time, this model allowed the detection of increased protection in mice when EDA or MEDA were administered mixed with B-SEΔgal, and also allowed both the selection of bacterins and not HS from both mutants as Salmonella antigen candidates and the detection of enhanced protection with EDAvidin bound to biotinylated B-SEΔwaaL.
Although additional work should be done in different natural hosts to determine the true innocuousness and efficacy of BEDA preparations, it is clear that EDA (as EDAvidin) improves the efficacy of rough Salmonella bacterins (as biotinylated bacterins) in the mouse model. The association between EDAvidin and B-SEΔwaaL bacterin may be considered safe and effective for use as a non-live vaccine, conferring a high protection against virulent infection. Employing this BEDA immunization strategy with O-PS deficient mutants may also help to distinguish (by conventional anti-O-PS or new anti-EDA serological tests) between vaccinated animals and asymptomatically infected carriers, reservoirs of zoonotic infections. Moreover, the use of non-live vaccines avoids the presence of genetically modified microorganisms in farm animals and their subsequent release to environment or food-chain, having an added value for consumers and veterinary use.