Recombinant Limosilactobacillus (Lactobacillus) delivering nanobodies against Clostridium perfringens NetB and alpha toxin confers potential protection from necrotic enteritis

Abstract Necrotic enteritis (NE), caused by Clostridium perfringens, is an intestinal disease with devastating economic losses to the poultry industry. NE is a complex disease and predisposing factors that compromise gut integrity are required to facilitate C. perfringens proliferation and toxin production. NE is also characterized by drastic shifts in gut microbiota; C. perfringens is negatively correlated with Lactobacilli. Vaccines are only partially effective against NE and antibiotics suffer from the concern of resistance development. These strategies address only some aspects of NE pathogenesis. Thus, there is an urgent need for alternative strategies that address multiple aspects of NE biology. Here, we developed Limosilactobacillus (Lactobacillus) reuteri vectors for in situ delivery of nanobodies against NetB and α toxin, two key toxins associated with NE pathophysiology. We generated nanobodies and showed that these nanobodies neutralize NetB and α toxin. We selected L. reuteri vector strains with intrinsic benefits and demonstrated that these strains inhibit C. perfringens and secrete over 130 metabolites, some of which play a key role in maintaining gut health. Recombinant L. reuteri strains efficiently secreted nanobodies and these nanobodies neutralized NetB. The recombinant strains were genetically and phenotypically stable over 480 generations and showed persistent colonization in chickens. A two‐dose in ovo and drinking water administration of recombinant L. reuteri strains protected chickens from NE‐associated mortality. These results provide proof‐of‐concept data for using L. reuteri as a live vector for delivery of nanobodies with broad applicability to other targets and highlight the potential synergistic effects of vector strains and nanobodies for addressing complex diseases such as NE.


| INTRODUCTION
Necrotic enteritis (NE) is a common intestinal disease that causes significant economic losses (~6 billion dollars annually) to the poultry industry worldwide Wade et al., 2015). NE generally manifests as clinical or subclinical forms. Clinical NE is acute and characterized by high mortality (30%-60%) and associated symptoms such as ruffled feathers, wet litter, diarrhea, and passage of undigested feed (Hofacre et al., 2018). In some cases, mortality is the only sign with no other premonitory symptoms. Subclinical NE is chronic and associated with damage to the intestinal mucosa, leading to reduced digestion and absorption of nutrients, decreased weight gain (3%-5%), and elevated feed conversion ratio (6-9 points) (Hofacre et al., 2018). Subclinical NE contributes to the majority of the economic losses associated with NE and is the most prevalent form of NE . NE is caused primarily by Clostridium perfringens type A strains that infiltrate the mucosa of the small intestine and produce toxins such as NetB and α toxin (Al-Sheikhly & Truscott, 1977a, 1977bHofacre et al., 2018;Justin et al., 2002;Keyburn et al., 2008;Kiu et al., 2019;Lee et al., 2012;Sheedy et al., 2004). C. perfringens is a spore-forming, anaerobic, Gram-positive commensal that is ubiquitously found in the gastrointestinal tract of animals and the environment. NE is a complex disease and several factors are known to influence the gut environment of the host and favor the growth of C. perfringens strains Fernandes Da Costa et al., 2016). Mucosal damage caused by Eimeria species, nature of the feed, sudden diet change, high-density bird housing conditions, and extreme environmental temperatures are among the key factors that predispose birds to NE (Fernandes Da Costa et al., 2016). The proliferation of C. perfringens leads to dramatic shifts in microbiota, with decreased abundance of beneficial species such as those belonging to Lactobacillus Yang, Liu, Wang, et al., 2021).
The C. perfringens NetB is a pore-forming toxin and it plays a key role in NE (Keyburn et al., 2008). A nontoxic variant of NetB called W262A is commonly used for immunization and has been shown to partially protect birds from NE (Fernandes Da Costa et al., 2016;Hunter et al., 2019;Jiang et al., 2015;Keyburn et al., 2008;Wilde et al., 2019). The role of α toxin in NE is not clear; however, immunization with α toxin antigen partially protects birds from NE (Abildgaard et al., 2009;Al-Sheikhly & Truscott, 1977a, 1977bFernandes Da Costa et al., 2016;Hunter et al., 2019;Jiang et al., 2015;Keyburn et al., 2006;Kulkarni et al., 2007Kulkarni et al., , 2010Wilde et al., 2019;Zekarias et al., 2008). A combination of NetB and α toxin provides improved protection against NE (Fernandes Da Costa et al., 2016). Nevertheless, vaccine approaches are only partially effective in reducing NE and the efficacy of vaccines depends on several factors such as the host genetics, immune system, and nutrition. To date, the administration of antibiotics has been the only effective treatment for NE; however, there is an increasing demand to reduce the use of antibiotics due to concerns around antimicrobial resistance. In addition, vaccines and antibiotics address only some aspects of NE (C. perfringens and their toxins), necessitating the need for developing safe and effective alternatives that address multiple aspects of NE biology.
Since their discovery in the 1990s, nanobodies (Nbs) have emerged as a promising alternative for disease prevention and treatment in both animal and human health (Amcheslavsky et al., 2021;Del Rio et al., 2019;Dulal et al., 2021;Hussack et al., 2014, Jovcevska & Muyldermans, 2020Riazi et al., 2013;Steidler et al., 2000;Tremblay et al., 2013;Unger et al., 2015;Vanmarsenille et al., 2017;Zhang et al., 2021). Nbs are single-domain antibodies, derived from heavy chain only (lack light chains and the first constant C H 1 domain) antibodies that naturally occur in the serum of camelids (dromedaries, camels, llamas, alpacas, guanacos, vicunas) (Jovcevska & Muyldermans, 2020). The variable domain of these heavy-chain only antibodies is the only domain involved in binding of this special class of antibodies (therefore called VHH, Variable domain of the Heavy chain of Heavy chain only antibodies) and are often referred to as "nanobodies," which is a registered trademark of Ablynx. Compared to traditional antibodies, Nbs possess several unique and favorable properties such as small size, high stability, strong antigenbinding affinity, water solubility, and ease of production in bacteria (Dumoulin et al., 2002;Goldman et al., 2017;Jovcevska & Muyldermans, 2020;Muyldermans et al., 2001Muyldermans et al., , 2009Van Der Linden et al., 1999;Van Der Vaart et al., 2006). Despite all these advantages, like any other protein and peptide biotherapeutics or preventatives, oral delivery of Nbs remains unsuccessful due to their degradation in the acidic and enzyme-rich environment of the stomach (Dhalla et al., 2021;Gleeson et al., 2021). Desired efficacy also demands frequent administration at higher concentrations, which is not economically feasible.
Microbial vectors (MVs) offer an excellent opportunity for oral delivery of bio-therapeutics and preventatives. MVs include bacteria such as Limosilactobacillus (Lactobacillus), Lactococcus, Salmonella, Bacillus, Listeria, and Escherichia coli, which are engineered to deliver target molecules directly to the site of action. Delivery using MVs not only protects the target molecules from the harsh gastrointestinal (GI) environment but also maximizes effectiveness and minimizes offtarget effects (Del Rio et al., 2019). MVs also have the advantage of easy and inexpensive manufacturing with flexible scalability and storage. Although several species of bacteria have been investigated as potential delivery vectors, species belonging to Limosilactobacillus and Lactococcus are among the most used genera for mucosal delivery of biotherapeutics and preventatives (Del Rio et al., 2019; and play a key role in regulating local microbiota, restoring barrier function, preventing inflammation associated with GI diseases, improving growth performance, and protecting against infectious diseases Han et al., 2021;Hu et al., 2021;Klaenhammer et al., 2002;Li et al., 2018;Manes-Lazaro et al., 2017;Ouwehand et al., 2002;Siddique et al., 2021;Vineetha et al., 2017; Y. Wu et al., 2021). Of particular importance, Lactobacilli have been shown to possess strong antagonistic activity against C. perfringens, inhibit toxin production, reduce proinflammatory cytokines, improve intestinal integrity and immune response, correct microbial dysbiosis, restore performance deficiencies associated with subclinical NE, and protect chickens from clinical NE (Dec et al., 2016;Gong et al., 2020;Guo et al., 2017Guo et al., , 2021Khalique et al., 2019;Kizerwetter-Swida & Binek, 2005La Cao et al., 2012;La Ragione et al., 2004;Li et al., 2017;Shojadoost et al., 2022;Xu et al., 2020). Lactobacilli are also well known for surviving the harsh environment of the GI tract (Hai et al., 2021;Mandal et al., 2021;Noohi et al., 2021;Vesa et al., 2000). Furthermore, Lactobacilli are generally associated with mucosa and thus ensure delivery of target molecules directly to the mucosa. The availability of genetic tools for engineering Limosilactobacillus further makes them attractive candidates for in situ delivery of biomolecules.
In previous studies, we showed that two novel Limosilactobacillus reuteri (L. reuteri) isolates, ATCC PTA-126787 (L. reuteri 3630) and ATCC PTA-126788 (L. reuteri 3632) possess favorable safety properties based on the results from in silico, in vitro and in vivo analyses in chickens and Sprague Dawley rats . In the present study, we describe the development of L. reuteri 3630 and L. reuteri 3632 as live vectors for in situ delivery of llama derived Nbs against C. perfringens NetB and α toxin to prevent NE in poultry. This study describes the application of L. reuteri delivered Nbs as a broad strategy to address a complex disease like NE.

| Library construction
Immunization and RNA preparation were performed by Eurogentec (Belgium). Before use, the obtained RNA was precipitated and 5 µl (~100 ng) of RNA was loaded onto gel to confirm the integrity of 28S and 18S rRNA. The remaining RNA was stored in 70% ethanol, containing 200 mM sodium acetate at −80°C. About 40 µg RNA was transcribed into cDNA using SuperScript III Reverse Transcriptase Kit (Invitrogen) using commercial random hexamer primers (Thermo Fisher). The cDNA was cleaned on Macherey-Nagel NucleoSpin Gel and PCR Clean-up kit (Macherey-Nagel). Immunoglobulin H (both conventional and heavy chain) fragments were amplified using primers annealing at the leader sequence region and the CH2 region as described previously (Dolk et al., 2012;Pardon et al., 2014). Five microliters were loaded onto a 1% Tris-Borate-EDTA (TBE) agarose gel to confirm amplification. The rest of the samples were loaded onto a 1% Tris-Acetate-EDTA (TAE) agarose gel. The 700-bp fragment was excised from the gel and purified. About 80 ng was used as a template for the nested PCR. The amplified fragment was cleaned on Macherey-Nagel NucleoSpin Gel and PCR Clean-up kit and eluted in 120 µl. The eluted DNA was digested first with SfiI and next with BstEII. Restriction digestion was confirmed by agarose gel electrophoresis using 1.5% TBE agarose gel. After the restriction digestion, the samples were loaded onto a 1.5% TAE agarose gel. The 400-bp fragment was excised from the gel and purified on Macherey-Nagel NucleoSpin Gel and PCR Clean-up kit. The 400-bp fragments were ligated into the phagemid pUR8100 vector (QVQ BV) and GANGAIAH ET AL. | 3 of 37 transformed into E. coli TG1 (Nectagen). The transformed E. coli TG1 were titrated using 10-fold dilutions. Five microliters of the dilutions were spotted on Luria-Bertani (LB) agar plates supplemented with 100 µg/ml of ampicillin and 2% glucose.
The number of transformants was calculated from the spotted dilutions of the rescued E. coli TG1 culture. The titer of the library was calculated by counting colonies in the highest dilution and using the formula below: library size = (amount of colonies) × (dilution) × 8 (ml)/0.005 (ml; spotted volume). The transformants were stored in a 2xYT (Sigma-Aldrich) medium supplemented with 20% glycerol, 2% glucose, and 100 µg/ml ampicillin at −80°C. The insert frequency was determined by picking 24 different clones from transformations from each library and performing a colony PCR. Bands of~700 bp indicate a cloned VHH fragment. Bands of~300 bp indicate an empty plasmid.

| Phage production and selection
Phages were produced from the libraries as outlined below (Dolk et al., 2012;Parmley & Smith, 1988;Smith, 1985). E. coli TG1 containing libraries from SNL133 (Days 43 and 78) and SNL134 (Days 43 and 78) were diluted from the glycerol stock up to an OD 600 of 0.05 in 2xYT medium containing 2% glucose and 100 µg/ml ampicillin and grown at 37°C for 2 h to reach an OD 600 of~0.5. Subsequently, about 7 ml of the cultures were infected with helper phage VCS M13 using an MOI (multiplicity of infection) of 100 for 30 min at 37°C. E. coli TG1 were spun down and resuspended into 50 ml fresh 2xYT medium supplemented with both ampicillin (100 µg/ml) and kanamycin (25 µg/ml) and grown overnight at 37°C with shaking. Produced phages were precipitated from the supernatant of the cultures using polyethylene glycol (PEG)-NaCl precipitation. Titers of the produced phages were calculated by serial dilution of the phage sample and infection of E. coli TG1.
Twenty microliters of the precipitated phages (~10 11 phages, which is >1000-fold the diversity of the libraries) were applied to wells coated with α toxin and NetB. In short, for each library, 100 µl antigen was coated on the MaxiSorp plate overnight at two concentrations of 5 and 0.5 µg/ml. As a negative control, one well was incubated with PBS only.
The next day, after removal of non-bound antigen, the plate was washed three times with PBS and blocked with 4% MPBS. At the same time, freshly precipitated phages were pre-blocked with 2% MPBS for 30 min. Preblocked phages were incubated on coated antigen for 2 h.
Upon extensive washing with PBS-Tween and PBS, bound phages were eluted with 0.1 M triethylamine solution and subsequently neutralized with 1 M Tris/HCl, pH 7.5. Eluted phages were serially diluted and then used to infect E. coli TG1 bacteria and spotting on LB agar plates supplemented with 2% glucose and 100 µg/ml ampicillin and incubated overnight at 37°C.
Glycerol stocks were prepared from all outputs rescued by infection of E. coli TG1 and stored at −80°C. Simultaneously, TG1 cultures infected with the output of the selection on 5 µg/ml α toxin or NetB (highest coating) were used for phage production of SNL-133 and SNL134 sublibraries to perform the second round of selection. Overnight grown rescued outputs were diluted 100-fold in 5 ml fresh 2xYT medium supplemented with 2% glucose and 100 µg/ml ampicillin and grown for 2 h until log phase. Subsequently, 1 µl of helper phage VCS M13 was added and incubated at 37°C for 30 min. Cultures were allowed to produce phages overnight at 37°C.
Produced phages were precipitated from the supernatant of the cultures using PEG-NaCl precipitation. One microliter of the precipitated phages was applied to wells coated with α toxin or NetB as indicated below. Antigens were coated on a MaxiSorp plate overnight at three concentrations (5, 0.5, and 0.05 µg/ml) and phages that bind specifically to α toxin or NetB were identified as described in the first round of selection.

| Screening and sequence analysis of VHHs
After the second round of phage display selection, glycerol stocks were prepared from all outputs rescued by infection of E. coli TG1 and stored at −80°C in the same way as for the outputs obtained after the first round of phage display selection. Subsequently, all rescued outputs of the second round of selection on both α toxin and NetB were plated out to pick single colonies, which were grown in a 96-well plate (master plate EAT-1 for α toxin and master plate ENB-1 for NetB). These master plates were used to produce periplasmic fractions containing monoclonal VHHs for screening of binders.
The master plates were cultivated at 37°C in 2xYT medium supplemented with 2% glucose and 100 µg/ml ampicillin and stored at −80°C after the addition of glycerol to a final concentration of 20%. For the production of periplasmic fractions, master plates EAT-1 and ENB-1 were duplicated into a deep well plate containing 1 ml 2xYT medium supplemented with 0.1% glucose and 100 µg/ml ampicillin and grown for 3 h at 37°C before adding 1 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) for induction of VHH expression. The VHH expression was conducted overnight at room temperature. Periplasmic fractions were prepared by collecting the bacteria by centrifugation and their resuspension into 120 µl PBS. After freezing, bacteria were thawed and centrifuged to separate the soluble periplasmic fraction containing the VHH from the cell debris (pellet). To test the binding specificity, monoclonal VHHs were tested using 25 µl of the periplasmic fractions exactly as described above.
pMEK222 adds a c-myc (EQKLISEEDL) and His-tag (HHHHHH) at the C-terminus of the VHH. The VHHs were produced as described below. Precultures were prepared by growing the bacteria containing the plasmids containing the selected VHH in 8 ml 2xYT medium supplemented with 2% glucose and 100 µg/ml ampicillin overnight at 37°C. The precultures were diluted into 800 ml fresh 2xYT that was pre-warmed at 37°C and supplemented with 100 µg/ml ampicillin and 0.1% glucose. The bacteria were grown for 2 h at 37°C before induction of the VHH expression with 1 mM IPTG. The VHHs were expressed for 4 h at 37°C and bacteria were harvested by centrifugation. Bacterial pellets were resuspended into 30 ml PBS and frozen at −20°C.
Frozen bacteria were thawed at room temperature and centrifugated to separate cell debris and soluble fraction, which contains the VHH. VHH were purified from the soluble fraction using immobilized metal affinity chromatography resin charged with cobalt (TALON beads). Bound VHHs were eluted with 150 mM imidazole and dialyzed against PBS. The protein concentration was measured using absorption at 280 nm and corrected according to the molar extinction coefficient and the molecular weight of different VHHs.
About 1 µg of the purified VHH was loaded onto sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). The binding of purified VHH was analyzed by ELISA using immobilized α toxin or NetB as described above.
2.7 | Neutralization of α toxin and NetB activity by selected nbs 2.7.1 | Neutralization of α toxin activity The inhibitory capacity of the VHH antibodies directed toward α toxin was determined by measuring the α toxin lecithinase activity. Briefly, fresh egg yolk was centrifuged (10,000g for 20 min at 4°C) and diluted at 1:10 in PBS. The ability of the VHHs to neutralize the α toxin activity was assessed by preincubating a twofold dilution series of the VHHs with a constant amount of α toxin (either 5 µg/ml recombinant α toxin [C-terminal fragment of α toxin used for llama immunization] or 3.33 × 10 −4 U/µl α toxin from Sigma, P7633) for 30 min at 37°C before the addition of 10% egg yolk emulsion. As a control, serum from calves immunized with the recombinant α toxin (as used for llama immunization) was used, starting from a ¼ dilution. After incubation at 37°C for 1 h, the absorbance at 650 nm was determined. α Toxin activity was indicated by the development of turbidity which increases absorbance.
Neutralization of the α toxin hemolytic activity by the VHH antibodies directed toward α toxin was determined by measuring its effect on sheep erythrocytes. Similar to the inhibition of the α toxin lecithinase activity, the ability to neutralize the hemolytic activity was assessed by preincubating a twofold dilution series of the VHH antibodies with a constant amount of α toxin (6.25 × 10 −5 U/µl α toxin from Sigma, P7633) for 30 min at 37°C before the addition of 1% sheep erythrocytes. No recombinant α toxin was used in this test because the recombinant toxin is not hemolytic. As a control, serum from calves immunized with the recombinant α toxin (as used for llama immunization) was used, starting from a ¼ dilution. After incubation at 37°C for 1 h, the plates were centrifuged to pellet intact red blood cells. The supernatants were transferred to a new 96-well plate and the absorbance at 570 nm was determined. α Toxin activity was indicated by an increase in absorbance due to the release of hemoglobin from the erythrocytes.

| Neutralization of NetB activity
Neutralization of NetB hemolytic activity by VHH antibodies directed toward NetB was determined by measuring its effect on chicken erythrocytes. Similar to inhibition of α toxin activity, the ability to neutralize the NetB hemolytic activity was assessed by preincubating a twofold dilution series of the VHH antibodies with a constant amount of NetB toxin (20 µg recombinant NetB in a total volume of 2 µl) for 30 min at 37°C before the addition of 1% chicken erythrocytes. The nontoxic NetB variant W262A was included as well and showed no hemolysis. To understand the relative efficacy of Nb candidates compared to polyclonal antisera, serum from rabbits immunized with the recombinant NetB (wild type NetB, not the same as used for llama immunization) was used, starting from a ¼ dilution.
After incubation at 37°C for 1 h, the plates were centrifuged to pellet intact red blood cells. The supernatants were transferred to a new 96-well plate and the absorbance at 570 nm was determined. NetB activity was indicated by an increase in absorbance due to the release of hemoglobin from the erythrocytes.
2.8 | Homology modeling and bioinformatics analyses of Nb clones and their evaluation for affinity, production, and stability Four lead clones were selected and subjected to further optimization to improve affinity, production, and stability. Three-dimensional (3D) structures of the Nb clones were generated using homology modeling as described previously (Khodabakhsh et al., 2021;Moonens et al., 2014). Based on the homology models and bioinformatics analyses, recombinant proteins were produced for the mutant clones and evaluated for affinity, production, and stability (protease and temperature) ( 2.9 | In silico modeling to identify Nb-binding epitopes on NetB Nb binding epitopes on NetB were predicted by molecular docking method using MOE (Molecular Operating Environment), which helps to visualize, characterize, or evaluate the interaction of proteins with other proteins or ligands (Khodabakhsh et al., 2021). 3D structures of the Nb clones were generated using homology modeling. As a template, the crystal structure of NbFedF9 (PDB code: 4W6Y, 1.57A) was used (Moonens et al., 2014). One hundred models were generated by Modeller software version 9.24. The best model was selected based on discrete optimized protein energy score and the quality of the model was assessed by phi and psi analysis. The published crystal structure of NetB was used for in silico modeling (Savva et al., 2013).

| Bacterial strains and growth conditions
Bacterial strains used in the present study are listed in Table 2. L. reuteri strains were propagated on Lactobacilli de Man Rogosa Sharpe (MRS; BD Difco) medium anaerobically at 37-39°C. L. reuteri strains with truncated pyrE were grown on MRS medium supplemented with 200 µg/ml uracil.
E. coli strains were grown on LB or Brain Heart Infusion (BHI) medium aerobically at 37-39°C with shaking at 200 rpm. Where necessary, the growth media were supplemented with chloramphenicol at a final concentration of 25 µg/ml for E. coli and 15 µg/ml for L. reuteri.

| Antimicrobial activity analysis
L. reuteri strains 3630 and 3632 were evaluated for their ability to inhibit C. perfringens using an agar overlay method. Briefly, L. reuteri strains were streaked in the center of MRS agar plates and incubated overnight under anaerobic conditions at 37°C. The next day, C. perfringens was grown in BYC broth (BHI broth, 37 g/L; yeast extract, 5 g/L; L-cysteine hydrochloride, 0.5 g/L) overnight under anaerobic conditions at 37°C was pelleted by centrifugation at 2000g for 10 min, washed twice in sterile PBS, and resuspended to an OD 600 of 0.5. The culture was then added at 5% inoculum into molten agar cooled to 45°C, mixed well by swirling, and layered onto overnight grown L. reuteri plates. The plates were then incubated at 37°C for 24 h and observed for a zone of clearance around the L. reuteri streaks.  Twenty micrograms of each sample were processed by SDS-PAGE using a 4%-12% Bis-Tris NuPAGE mini-gel (Invitrogen) with the MOPS (3-(N-morpholino)propane sulfonic acid) buffer system.

T A B L E 2 Bacterial strains and plasmids used in this study
Each gel lane was excised longitudinally into 20 equally sized bands and processed by in-gel digestion with trypsin (Promega) using a ProGest robot (Digilab) with the protocol outlined below.
Briefly, the excised gel lane was washed with 25 mM ammonium bicarbonate followed by acetonitrile, reduced with 10 mM dithiothreitol at 60°C followed by alkylation with 50 mM iodoacetamide at room temperature, digested with trypsin at 37°C for 4 h and finally quenched with formic acid and analyzed directly without further processing.

| Construction of pyrE-based integration vectors
All the plasmids and primers used in this study are listed in Tables 2 and 3, respectively. The vector pCG2440 was used to generate the integration vector and contains pE194 origin of replication (Horinouchi & Weisblum, 1982), p15A origin of replication for replication in E. coli, and chloramphenicol resistance marker for selection in both E. coli and L. reuteri. EAT-1G4 and ENB-1D11_R56H sequences were codon-optimized to L. reuteri and synthesized from ATUM. Promoters and secretion signals were selected based on the global proteomics data and evaluated for their ability to express and secrete Nbs, respectively. Based on this analysis, the native promoter, secretion signal, and terminator from cwlS were chosen for further engineering to deliver Nbs. The expression cassette containing the cwlS promoter, cwlS secretion signal, codon-optimized EAT-1F2_R27H, and cwlS terminator was synthesized as a single fragment from GenScript Inc. pCG2440 backbone, 310 bases of 5′ flanking region containing 112 bp bases of pyrE upstream region and 198 bases of pyrE 5′ coding region, the expression cassette containing EAT-1F2_R27H and 1200 bases downstream of pyrE CDS were PCR amplified using P1, P2, P3, and P4 primer pairs, respectively (Table 3), and NEB 2X Phusion master mix following the manufacturer's instructions. The PCR products were treated with DpnI (Thermo Fisher Scientific) to digest the vector backbone following the manufacturer's instructions. The digested products were PCR purified using a Qiagen PCR Purification kit (Qiagen) and assembled to generate pVR01 using Gibson Assembly kit (NEB) following the manufacturer's instructions. The assembled product was transformed into chemically competent E. coli cells provided in the kit and plated on BHI agar containing 25 µg/ml of chloramphenicol to select for transformants. The transformants were confirmed for the presence of all the fragments by PCR using NEB Phusion 2X master mix (NEB) and plasmid DNA was isolated from select clones using Qiagen Midi Prep Plasmid Isolation kit (Qiagen) following the manufacturer's instructions. Similarly, pVR02 and pVR03 were generated by replacing the EAT-1F2_R27H sequence in pVR01 with codon-optimized EAT-1G4 and ENB-1D11_R56H sequences using P5-P8 primer pairs and molecular biology procedures described above (Table 3).  | 9 of 37 chamber at 39°C for at least 4 h. Following incubation, the cells were plated on MRS (BD Difco) with 15 µg/ml of chloramphenicol and incubated for 2-5 days in an anaerobic chamber at 39°C. Transformants were selected, and streak purified three times before proceeding to the next steps.

| Selection of single crossovers and their resolution into double crossovers
To utilize PyrE as a counterselection marker for selecting double crossover integrants, it is imperative that we select single crossover integrants at 3′ end with functional uracil pathway (Sakaguchi et al., 2013). Transformants with a functional PyrE will not grow in the presence of 5-fluoroorotic acid (5-FOA) (Sakaguchi et al., 2013). The 2.17 | Construction of pyrE-based correction vectors and their integration into L. reuteri genome In this step, the truncated pyrE was corrected to its wild-type state using pyrE correction vectors. PyrE was reconstituted by PCR amplification and assembly of pVR02 backbone and 444 bp of the 3′ end of pyrE gene using primer pairs P9 and P10, respectively, generating pVR004, using the molecular biology procedures described above (Table 3). Similarly, pVR05 was generated by reconstituting pyrE using pVR03 as backbone and primer pairs P11-P12 ( Following the withdrawal period, the resuspended experimental product was added to the waterer before chickens were allowed to drink. Immediately after water containing the L. reuteri was consumed, fresh drinking water, without L. reuteri, was added to the waterer for ad libitum consumption.

| Challenge with Eimeria maxima and C. perfringens
The NE model used in the present study includes oral administration of E. maxima followed by serial oral administration of C. perfringens. | 11 of 37 died postchallenge phase between 18 (after challenge with C. perfringens) and 28 days of age were necropsied, and the cause of death was listed as NE-related or non-NE-related mortality.

| Statistical analyses
The in vitro and colonization data presented in this study were analyzed using a mixed model analysis of variance with experimental days or chicks as a random effect followed by Tukey's test for pairwise multiple comparisons where applicable. A p value of 0.05 was considered statistically significant.   and 78) were of good size with more than 10 7 clones per library, which was sufficient for efficient panning selections. Analysis of insert frequency showed that the frequency was close to 100% for all four libraries. Based on the ELISA results, several clones were picked from master plates EAT-1 and ENB-1 for sequence identification. Figure A2 shows the sequence alignment of the clones that were picked from the selection outputs on α toxin. Two families are shown (KEREF and KQREL) within the sequences. There is a diversity of around seven different VHH sequences. Figure A3 shows the sequence alignment of the clones that were picked from the selection outputs on NetB. There is a diversity of around 6 different VHH sequences, which were derived from two families (KEREF and KQREL) within the sequences. and ENB-1D11 from master plate ENB-1 were cloned into an expression vector, and VHHs were produced and purified. Table B2 describes the calculation of the VHH concentrations based on absorption (A280) and the correction factor (CF; the extinction factor calculated from VHH sequence) of the VHHs selected on α toxin. Table B3 describes the calculation of the VHH concentrations based on absorption (A280) and the CF of the VHHs selected on NetB. Figure 2a shows SDS-PAGE analysis of the purity of VHH purified via Immobilized Metal Ion Affinity Chromatography (IMAC). All VHHs appeared to be pure. Clone ENB-1B8 did not produce any protein.
The purified VHHs were tested for binding to α toxin and NetB to determine the apparent binding affinity. As shown in Figure 2b, 3.2 | Selected Nb candidates neutralize NetB and α toxin activity

| Neutralization of α toxin activity
The inhibitory capacity of the VHH antibodies towards α toxin lecithinase activity was determined using both commercial and recombinant (used for llama immunization) α toxins. As a control, antisera from calves immunized with recombinant α toxin was used.
The control serum was able to neutralize lecithinase activity of both commercial and recombinant α toxins. An eightfold dilution of the antiserum (corresponding to 3.12% serum) was able to completely neutralize the α toxin lecithinase activity of the recombinant α toxin (Figure 3a), whereas only the highest concentration of antiserum (corresponding to 25% serum) was able to completely neutralize lecithinase activity of commercial α toxin (Figure 3b). Considerable difference in inhibitory capacity was seen between the VHH antibodies. VHH EAT-1F3 did not affect the lecithinase activity of either of the α toxins (Figure 3a,b, yellow). The neutralizing capacity of EAT-1A2 and EAT-1C8 was very similar and was the same for both the recombinant and commercial α toxins (Figure 3a,b). The maximal inhibitory capacity was preserved until a 32-fold dilution (0.16 µM VHH) of the VHHs (Figure 3a,b). However, both EAT-1A2 and EAT-1C8 were unable to completely neutralize lecithinase activity, resulting in 40%-50% residual lecithinase activity (Figure 3a,b). F I G U R E 3 Neutralization of α toxin lecithinase and NetB hemolytic activity by VHH clones. (a, b) A twofold dilution series of VHH antibodies, starting from 5 µM concentration was preincubated with either (a) recombinant α toxin or (b) commercial α toxin, after which egg yolk solution was added. Serum derived from calves immunized with recombinant α toxin was used as a control (control serum). Egg yolk solution incubated with α toxin without VHH, or serum was used to calculate 100% activity. (c) A twofold dilution series of VHH, starting from 5 µM concentration, was preincubated with recombinant NetB (in a total volume of 2 µl), after which 1% chicken erythrocytes was added. Serum derived from rabbits immunized with recombinant NetB was used as a control (control serum). The optical density of 100% hemolysis (mean OD 570 = 0.37, indicated by solid line) was obtained by diluting the chicken erythrocytes in distilled water. As a control, chicken erythrocytes incubated with NetB, but without VHH or serum was used (mean OD 570 = 0.36, indicated by dotted line). This resulted in 100% hemolysis. A phosphate-buffered saline control (1% chicken erythrocytes with no NetB or Nbs) resulted in a mean OD 570 of 0.03. Representative results are shown from three independent experiments. VHH, Variable domain of the Heavy chain of Heavy chain The two other VHHs, EAT-1F2 and EAT-1G4 showed a difference in neutralizing capacity towards recombinant and commercial α toxins.
EAT-1F2 had a high neutralizing capacity towards recombinant α toxin but was unable to completely neutralize commercial α toxin, resulting in ±25% residual lecithinase activity (Figure 3a,b, red). In contrast to EAT-1F2, EAT-1G4 was able to neutralize 100% of the lecithinase activity of the commercial α toxin but was less capable of neutralizing the recombinant α toxin (Figure 3a,b, green).
The inhibitory capacity of the VHH antibodies towards the α toxin hemolytic activity was determined using the commercial α toxin. The recombinant α toxin, which was used to immunize llamas, showed no hemolytic activity. As a control, antisera from calves immunized with recombinant α toxin was used. Up to a 16-fold dilution of the control serum (corresponding to 1.56% serum) was able to completely inhibit α toxin hemolysis. On the contrary, none of the VHHs affected the hemolytic activity of α toxin ( Figure A4). Because the control serum contains polyclonal antibodies and VHHs are monoclonal, the combined effect of all five VHHs towards α toxin was determined (1 µM of each VHH in the highest dilution, corresponding to 5 µM VHHs in total). Combining the VHHs did not affect α toxin hemolysis ( Figure A4).

| Neutralization of NetB activity
The inhibitory capacity of the VHH antibodies towards NetB activity was determined using the recombinant NetB. The NetB variant W262A, which was used for llama immunization, was not hemolytic.
As a control, antisera from rabbits immunized with recombinant NetB was used. The control serum was able to neutralize the hemolytic activity of NetB. VHH antibodies ENB-1F4 and ENB-1F10 did not affect NetB hemolysis (Figure 3c, red and green). ENB-1B9 had intermediate inhibitory capacity, while ENB-1D11 and ENB-1A4 were able to neutralize NetB hemolysis up to a four-to eightfold dilution (1.25-0.625 µM VHHs) (Figure 3c).

| Optimization of candidate VHH clones improves stability and production while retaining affinity
The lead clones (EAT-1G4 and EAT-1F2 for α toxin; ENB-1A4 and ENB-1D11 for NetB) were further optimized for improved affinity, production, and stability using a combination of homology modeling, bioinformatics analysis, and in vitro testing. Figure 4a shows the homology model for ENB-1D11 (as an example), with the critical amino acids highlighted in red for trypsin susceptibility and CDRs, highlighted in yellow. As shown in Figure 4b, the affinity of IF2 mutants was not affected. While the affinity of 1G4 Y103W was only slightly affected, 1G4 W47L lost affinity (Figure 4b). The affinity of the 1D11 mutant was not affected; however, the 1A4 mutant lost affinity dramatically (Figure 4c). While EAT-1F2 was susceptible to trypsin (Figure 4f), EAT-1F2_R27H and EAT-1F2_T28P clones showed improved trypsin resistance (Table 1). EAT-1G4 was slightly susceptible to trypsin (Table 1). As EAT-1G4_W47L and EAT-1G4_Y103W clones lost affinity, they were not tested for trypsin susceptibility. While ENB-1A4 was slightly susceptible to trypsin, ENB-1A4_R57H showed improved trypsin resistance (Table 1). ENB-1D11 and ENB-1D11_R56H had similar trypsin resistance (Figure 4g; Table 1). Thermostability testing showed that there were no dramatic changes in the thermostability of all the tested clones (Table 1). Interestingly, the production levels were remarkably increased for most of the mutants (Table 1). Based on these results, EAT-1G4 and ENB-1D11_R56H were selected as the lead candidates against α toxin and NetB, respectively, for further engineering into L. reuteri.

| In silico modeling identifies putative Nb-interacting epitopes on NetB
NetB is a pore-forming toxin and active NetB contains seven monomers, which assemble into a ring-like structure upon contact with cholesterol on the eukaryotic membrane. Each monomer has three domains-β sandwich, rim, and stem; the stem domain is believed to interact with cholesterol. Structural modeling was performed to identify potential Nb interacting epitopes on NetB. As shown in Figure A5, ENB-1D11_R56H, and ENB-1A4 seem to interact with epitopes in the rim domain, potentially preventing interaction of NetB with cholesterol and subsequent oligomerization, which is required for NetB toxicity.
3.5 | L. reuteri strains 3630 and 3632 inhibit C. perfringens growth L. reuteri 3630 and 3632 vector strains were evaluated for inhibitory activity against C. perfringens strain JP1011, a hypervirulent strain isolated from a clinical case of NE and positive for both NetB and α toxin. As shown in Figure A6, both L. reuteri strains inhibited C. perfringens growth as evident from the clearance zone around the L. reuteri streaks.
3.6 | Global metabolomics analyses of L. reuteri 3630 and 3632 culture supernatants identifies potential metabolites with health benefits An untargeted global metabolomics analysis was performed to identify potential metabolites secreted by L. reuteri strains. In total, 433 and 436 known metabolites were detected in the culture supernatants of L. reuteri 3630 and 3632, respectively. Of these metabolites, compared to media control, 130 metabolites were secreted 1.5-fold or higher in the culture supernatant of at least one strain (Table B4). Several tryptophan metabolites such as indolelactate, indolepropionate, indole-3-acetamide, indole-2-one, and kynurenate were highly enriched in the supernatants of both L. reuteri strains GANGAIAH ET AL.

| Global proteomics analyses identify potential native secretion signals and promoters
Global proteomics analyses were performed on L. reuteri 3632 cell pellet and supernatant to identify potential native secretion signals and promoters for engineering. This analysis identified 21731 matching spectra, 7263 unique peptides, and 607 proteins in the culture supernatant. proteins were identified in the cell pellet. Table B6 lists the top 50 proteins with the highest number of matching spectra in the cell pellet. The identified proteins from the culture supernatant were ranked based on the number of matching spectra and seven secretion signals from the highly secreted proteins were selected for further analyses (highlighted in bold in Table B5). Similarly, the proteins from the cell pellet were ranked based on the number of matching spectra, and six promoters from the proteins with the highest number of matching spectra were selected for further testing (highlighted in bold in Table B6). To integrate expression cassette into L. reuteri genome, we used pyrE as a counterselection marker as described previously (Sakaguchi et al., 2013). In L. reuteri 3630 and 3632 genomes, pyrE (642 bp) is located in an operon with other genes in the order of pyrB-pyrC-pyrDB-pyrF-pyrE ( Figure A7). We designed integration vectors pVE02 and and Sanger sequencing ( Figure 5d). As expected, PCR amplification yielded 3200-bp product for NE01 and NE06, 2800-bp product for NE08 and NE12, and 2200-bp product for the parent strains (Figure 5d).

| Assembly of expression cassettes and construction of engineered L. reuteri strains
NE01 and NE08 were generated using L. reuteri 3630 and NE06 and NE12 were generated using L. reuteri 3632. We reasoned that using two strains, one delivering NetB-specific Nb and another delivering α toxinspecific Nb, allows for efficient production and secretion of Nbs by reducing the burden on the expression and secretion machinery.
3.9 | Engineered L. reuteri strains secrete nbs into the culture supernatant The culture supernatants of NE01 and NE06 were evaluated for expression and subsequent secretion of Nbs using western blot. As shown in Figure 6a, expectedly, the anti-VHH antibody is specifically bound to a protein size of around 25 kDa, which is the expected size of VHHs with anchors. The control VHH was bound to a protein size of around 14 kDa and showed duplet bands (Figure 6a). It should be noted that the L. reuteri secreted Nbs run higher due to the presence of small N-and Cterminal anchors from cwlS used for efficient secretion. The secreted Nbs were intact, and no degradation was observed for any of the tested strains ( Figure 6a). The bands corresponding to Nbs were excised and identified by mass spectrometry, which confirmed that the majority of the spectra matched the respective Nbs (Figure 6b).

| Secreted Nbs neutralize NetB activity
Nbs precipitated from an overnight culture of NE06 were evaluated for their ability to neutralize NetB purified from a clinical isolate of C. perfringens. As shown in Figure 7a, the secreted Nbs neutralized hemolytic activity of NetB and this inhibition was maintained up to 32-fold dilution. The neutralization of NetB activity by Nbs purified from L. reuteri was comparable to that of E. coli until at least 16-fold dilution (Figure 7a). The anti-NetB Nb ENB-1D11_R56H (VHH3) was also evaluated for its specific binding to NetB from culture supernatants from clinical isolates grown at different growth phases. As shown in Figure 7b, ENB-ID11_R56H is specifically bound to a band that corresponds to the size of NetB (38 kDa). There was no difference in the amount of NetB produced between mid-log versus overnight cultures as detected using anti-NetB Nb (Figure 7b). It should be noted that the NetB positive control runs slightly higher due to the presence of His tag and linker sequence at the N terminus.
3.11 | Engineered L. reuteri strains are genetically and phenotypically stable The engineered L. reuteri strains were tested for their genetic stability after 30 passages (approx. 480 generations). As expected, PCR amplification of the expression cassette using primers that bind to flanking regions outside the expression cassette yielded a product of 3200 bp with NE01 and NE06 (Figure 8a). Sanger F I G U R E 6 Western blot analysis showing the ammonium precipitated Nbs in the culture supernatant of NE01 and NE06. (a) 1, LiCor protein ladder; 2, ENB-1D11_R56H control (5 µg, runs as a duplet due to fragmentation); 3, NE01; 4, NE08; 5, NE06; 6, NE12. Please note that the L. reuteri secreted Nbs run higher due to the presence of small N-and C-terminal anchors used for efficient secretion. This is a representative blot from three independent experiments. (b) Confirmation of the secreted Nbs in the culture supernatant of NE01 and NE06 by mass spectrometry. The highlighted areas match the nanobody sequence. The residues highlighted in green were identified with a modification (please refer to the methods for details on the modifications included in the database) F I G U R E 7 Neutralization of NetB activity by L. reuterisecreted Nbs. (a) A twofold dilution series of precipitated VHH antibodies (5 µM) was preincubated with recombinant NetB, after which 1% chicken erythrocytes was added. The optical density of 100% hemolysis was obtained by diluting the chicken erythrocytes in distilled water. As a control, chicken erythrocytes were incubated with NetB, but without Nbs was used. This resulted in 100% hemolysis (OD 570 = 0.54). A NetB positive control (NetB in PBS) resulted in a mean OD 570 of 0.52 and a PBS negative control (PBS with no NetB and Nbs) yielded a mean OD 570 of 0.05. As the initial amounts of L. reuteri and E. coli purified Nbs used for the assay were different, normalized OD 570 values are shown. (b) Western blot analysis binding of anti-NetB Nb to NetB in the culture supernatant from different C. perfringens clinical isolates. 1, Ladder; 2, NetB positive control (5 µg); 3, C. perfringens JP1011 overnight culture supernatant (10 µl); 4, C. perfringens JP1011 overnight culture supernatant, 10× concentrated (10 µl); 5, C. perfringens JP1011 mid-log culture supernatant (10 µl); 6, C. perfringens JP1011 midleg culture supernatant, 10× concentrated (10 µl); 7, C. perfringens CP1-1 overnight culture supernatant (10 µl 3.12 | Engineered L. reuteri strains show similar colonization levels to that of their parent strains Along with their respective parent strains, NE01 and NE06 were evaluated for colonization in chicken after in ovo administration. As shown in Figure 9, the mean CFUs/g of cecal contents ± SD for L. reuteri 3630 and 3632 were 8.36E+06 (±4.14E+06) and 1.07E+07 (±4.49E+06), respectively. Similarly, the mean CFUs/g of cecal contents ± SD for NE01 and NE06 were 1.10E+07 (±1.42E+06) and 4.97E+06 (±1.39E+06) CFUs, respectively ( Figure 9). The mean CFUs of all the four strains were found to be not significantly different from each other (p > 0.55) (Figure 9).
These data suggest that NE01 and NE06 exhibit similar colonization levels to their respective parent strains.
3.13 | Engineered L. reuteri strains partially protect against NE mortality Many strategies have been investigated for experimentally inducing NE; the majority of these strategies use some sort of predisposing factors . A dual challenge model that includes the E. maxima challenge followed by the C. perfringens challenge is a well-accepted model in the NE field . In the present study, a dual challenge model with a primary focus on NE mortality was used for evaluating engineered L. reuteri candidates. As shown in Figure 10, while Group 1 (no challenge control) had no NEassociated mortalities, Group 2 (challenge control) showed 15% NEassociated mortality (Figure 10). Treatment with NE01 and NE06 administered via in ovo at 7.63 × 10 5 CFUs/dose and drinking water at 1.17 × 10 8 CFUs/dose (Group 3) significantly reduced NEassociated mortality to 7.8% ( Figure 10; p < 0.05 compared to Group 2). Treatment with NE01 and NE06 administered via in ovo at 7.36 × 10 6 CFUs/dose and drinking water at 1.43 × 10 8 CFUs/dose (Group 4) significantly reduced NE-associated mortality to 8.33% ( Figure 10; p < 0.05 compared to Group 2). Treatment with NE01 and NE06 administered via spray at 4.35 × 10 7 CFUs/dose and drinking water at 1.35 × 10 8 CFUs/dose (Group 5) numerically reduced NE-associated mortality to 10.0% (Figure 10; p = 0.15 compared to Group 2). Preventative fraction is a ratio used in epidemiological studies to assess the impact of vaccination on a disease. The preventative fractions for Group 3, Group 4, and Group 5 were 48%, 44%, and 33%, respectively ( Figure 10). These data suggest that a two-dose administration of NE01 and NE06 partially protects chickens from NE.

| DISCUSSION
In this study, we developed L. reuteri as a live vector for in situ delivery of Nbs against NetB and α toxin from C. perfringens. We generated several Nb candidates that successfully neutralized NetB and α toxin from phage display libraries derived from immunized llamas. We showed that vector strains L. reuteri 3630 and 3632 inhibit C. perfringens in vitro, and secrete over 130 metabolites, some of which have been shown to play a key role in maintaining intestinal integrity, regulating microbiota, and reducing inflammation. We also demonstrated that engineered L. reuteri strains can efficiently produce and secrete Nbs and that these Nbs neutralize NetB. The engineered L. reuteri strains were genetically and phenotypically stable and showed persistent colonization in vivo similar to their parent strains. More importantly, a two-dose in ovo and drinking water administration of engineered L. reuteri candidates reduced NE mortality in chickens.
NE is a complex disease, and successful control of NE requires a multifactorial approach (Williams, 2005). C. perfringens is the primary causative agent of NE (Parish, 1961). Several factors such as Eimeria infection, dietary factors, immunosuppression, and Fusarium mycotoxins have been identified to predispose birds to NE (Williams, 2005 F I G U R E 9 Colonization of NE01 and NE06 in chickens. L. reuteri strains were administered via in ovo to 18-day embryonated chicken eggs and the birds were killed 7 days after hatching and CFUs were quantified from cecal contents. The strains were marked with rifampicin resistance to selectively isolate NE01, NE06, L. reuteri 3630, and L. reuteri 3632 from the rest of the microbiota from cecal contents. For each group, the data represent the mean ± SD of the results of five chicks GANGAIAH ET AL.
| 19 of 37 . We selected L. reuteri strains with intrinsic benefits as delivery vectors. More specifically, L. reuteri 3630 and 3632 inhibited the growth of pathogenic isolates of C. perfringens in vitro, suggesting that the backbones have the potential to inhibit C.
perfringens proliferation in vivo and prevent resulting dysbiosis. Using global metabolomics analysis, we showed that L. reuteri 3630 and 3632 secrete several metabolites with potential microbiota modulation and anti-inflammatory activities. Tryptophan metabolites were among one of the highly secreted metabolites by L. reuteri 3630 and 3632; tryptophan metabolites have been previously shown to play a key role in regulating the immune system and local microbiota, maintaining intestinal integrity, and reducing inflammation, most likely via activation of aryl hydrocarbon receptor pathway (Galligan, 2018;Gao et al., 2018;Negatu et al., 2020;Scott et al., 2020). These data suggest that vector backbones potentially work synergistically with Nbs to address other aspects of NE biology such as C. perfringens, dysbiosis, and inflammation ( Figure 11).
NetB and α toxin, produced by C. perfringens, are believed to be the main toxins involved in the pathogenesis of NE (Keyburn et al., 2008;Kiu et al., 2019). Alpha toxin is the most toxic enzyme produced by C. perfringens type A strains and hydrolyzes two major constituents of the eukaryotic membrane (phosphatidylcholine [lecithin] and sphingomyelin) causing membrane disruption and cell lysis (Nagahama et al., 1998;Urbina et al., 2009). We developed several Nbs that neutralize α toxin activity. Different Nbs had different inhibitory capacities towards the lecithinase activity of recombinant and commercial α toxins. This differential inhibitory capacity of the VHHs might have different explanations. First, the recombinant α toxin has a C-terminal His tag, whereas the commercial α toxin is purified from C. perfringens and has no tags. Although an effect of the His tag is unlikely, it cannot be excluded. Next, it should be noted that the recombinant α toxin is derived from an intestinal isolate, whereas the origin of the commercial α toxin is unknown. A difference in α toxin derived from enteric C. perfringens isolates and gas gangrene isolates has been previously reported, with higher trypsin resistance for α toxin of the enteric isolates (Uzal et al., 2014). As the origin of the commercial α toxin is unknown, it cannot be excluded that it has a slightly different activity.
NetB is a heptameric beta pore-forming toxin that forms single channels in planar phospholipid bilayers (Savva et al., 2013). NetB activity is primarily influenced by cholesterol, which enhances the oligomerization of NetB and plays an important role in pore formation (Savva et al., 2013). NetB has high hemolytic activity towards avian red blood cells (Yan et al., 2013). In the present study, we developed several Nbs that neutralize the hemolytic activity of NetB.
The protective range for the top two Nbs ENB-1D11 and ENB-1A4 for neutralizing 20 µg of NetB ranged from 5 µg to 0.312 ng. The polyclonal antisera generated against recombinant NetB in rabbits showed a broader protection range of 25% serum to 0.019% serum.
The differential inhibitory activity of VHH and polyclonal antisera is likely due to the presence of multiple neutralizing antibodies in the antisera compared to the monoclonal nature of VHHs. NetB epitope mapping by in silico analysis showed that ENB-1D11_R56H and ENB-1A4 seem to interact with epitopes in the rim domain and prevent interaction with cholesterol and subsequent oligomerization, which is key for NetB toxic activity (35).
The pyrE gene was used as a counterselection marker to engineer L. reuteri strains to deliver Nbs against NetB and α toxin (Sakaguchi et al., 2013). pyrE encodes for orotate phosphoribosyl transferase, which converts orotic acid to orotidine 5′-monophosphate (OMP).
PyrE also metabolizes 5-FOA, an analog of orotic acid, into 5-fluoroorotidine monophosphate (5-FOMP); the accumulation of 5-FOMP is toxic and leads to cell death. Double crossover integrants were selected from single crossover integrants by plating on MRS containing 5-FOA and uracil. 5-FOA is toxic to single crossover integrants as they still contain intact pyrE, whereas double crossover integrants grow on 5-FOA. In addition, pyrE deletion makes the strain uracil auxotroph; the truncated pyrE mutants were unable to grow in the absence of exogenous uracil. Given that pyrE truncated strains are uracil auxotrophs, pyrE truncation can also serve as a safety and biological containment strategy to reduce the survival of engineered strains in the environment (Heap et al., 2014). It should be noted that F I G U R E 10 Efficacy of NE01 and NE06 on reduction of NE-associated mortality. Efficacy of NE01 and NE06 was evaluated using a dual challenge model using Eimeria maxima and C. perfringens challenge as described in Section 2. Chickens that died postchallenge phase between 18 (after challenge with C. perfringens) and 28 days of age were necropsied, cause of death was listed as NE-related or non-NE-related mortality, and % NE mortality was calculated. a p < 0.05; b p < 0.05; c p = 0.15. CFU, colony-forming unit; NE, necrotic enteritis all the final engineered L. reuteri strains generated using PyrE counterselection contained no antibiotic resistance markers.
Western blot analysis confirmed that the engineered strains were showed that NetB-specific Nb specifically binds to a protein around 30 kDa, which corresponds to the size of NetB. These data suggest that the engineered strains can efficiently secrete Nbs and that these Nbs are functional.
Genetic and phenotypic stability is of paramount importance for developing a microbial strain as an in situ delivery vector. Using PCR and Sanger sequencing, we showed that our engineered L. reuteri Safety is of utmost importance in developing MVs for the delivery of biomolecules. Species belonging to Lactobacillus have been used in fermented foods for decades (Bourdichon et al., 2012). Lactobacilli are also a normal part of the GI microbiota of all vertebrates, including humans, monkeys, chickens, turkeys, doves, pigs, dogs, lambs, cattle, and rodents (Valeur et al., 2004). Several L. reuteri strains have been notified to FDA as "Generally Regarded As Safe" for use in specific foods or as new dietary ingredients . In addition, L. reuteri species are also considered safe for use in, or as a source of food for, human and animal consumption by EFSA, which has granted QPS status for L. reuteri species (Hazards et al., 2020). Several clinical studies have been conducted with L. reuteri species in humans and other species, including immunocompromised patients with no or very few adverse events (Indrio et al., 2008;Mu et al., 2018;Wolf et al., 1998). Previous in silico, in vitro, and in vivo analyses in chickens showed that the vector strains L. reuteri 3630 and 3632 have a favorable safety profile . A 28-day subchronic toxicity study showed that rats can tolerate high doses of L. reuteri 3630 and 3632 and no adverse events were observed when male and female Sprague Dawley rats were administered with 1.6E+10 CFUs of L. reuteri 3630/kg body weight/day plus 5.7E+10 CFUs of L. reuteri 3632/kg body weight/day for 28 days.
The backbone strain L. reuteri 3632 has also been tested via feed administration for 50 days at 1 × 10 7 CFUs/g of feed in swine with no adverse events. No adverse effects on hatchability or on chicks after hatching were observed in chickens when administered in ovo to 18-day old embryos at a dose of 1 × 10 5 , 1 × 10 6 , or 1 × 10 7 CFUs/ embryo (hatchability between 80% and 85%-all groups had similar hatchability). A two-day spray (Day 0, 1 × 10 7 CFUs/bird) and drinking water (Day 13, 1 × 10 8 CFUs/bird) administration to chickens also showed no adverse events. Drinking water administration of the backbone strain L. reuteri 3632 at 1 × 10 8 CFUs/bird/day every day for 21 days showed no toxicity or adverse events (Table C1). Together, these data suggest that the vector strains have a favorable safety profile and are suitable as delivery vectors.
In vivo efficacy depends on several factors, including colonization and persistence of the vector strains and expression and stability of the secreted Nbs in the gut. Following single in ovo administration, both vector strains showed persistent colonization and this colonization remained unchanged until Day 7 after hatching (study end).
F I G U R E 11 Schematic model showing the possible mechanisms of action of L. reuteri vector strains and secreted Nbs to address different aspects of NE biology. Nbs, nanobodies; NE, necrotic enteritis In a follow-up study, two-dose administration of NE01 and NE06 on Day 0 via spray and on Day 13 via drinking water resulted in persistent colonization of strains until at least 28 days after the spray administration with approximately 5 × 10 3 CFUs/g of cecal contents recovered on Day 28, suggesting that the strains remain colonized for the entire duration of NE phase. In addition, the promoter used for delivering Nbs comes from the cwlS gene, which is an essential gene involved in cell separation, a key process that concludes the process of cell division (Fukushima et al., 2006); thus, it is highly likely that this promoter will be active in vivo during replication and drive the production of Nbs. Trypsin is one of the primary proteolytic enzymes in the small intestine, which has the potential to degrade L. reuteri delivered Nbs; however, our in vitro trypsin digestion data showed that our lead Nb candidates are resistant to supraphysiological levels of trypsin and hence can withstand physiological levels of trypsin present in the small intestine. A small intestinal environment has a pH of around 5.7-6.5 (M. Mabelebele et al., 2002); we have shown that Nbs purified from overnight L. reuteri cultures, which consistently reach a final pH of 3.5-4.5, remain active and functional, suggesting that our Nbs can withstand exposure to low pH.
In nature, NE occurs in two forms-acute clinical form and chronic subclinical form Hofacre et al., 2018). Clinical NE is characterized by extensive necrosis of the small intestinal mucosa and high mortality Hofacre et al., 2018). Subclinical NE is characterized by damage to the intestinal mucosa resulting in decreased digestion and absorption, reduced weight gain, and increased feed conversion ratio Hofacre et al., 2018). While clinical NE models use mortality as the primary study parameter, subclinical models use lesion score as the primary study parameter. In this study, we showed that oral administration of NE01 and NE06 delivering Nbs via in ovo/spray and drinking water partially reduces NE mortality.
We have also shown that administration of NE01 and NE06 in drinking water every day for 21 days reduces both NE mortality and lesion score ( writing-review and editing (equal). Shrinivasrao P. Mane: investigation (equal); methodology (equal); writing-review and editing (equal). Enid T.

DATA AVAILABILITY STATEMENT
All data generated or analyzed during this study are included in this published article. F I G U R E A1 Binding of the periplasmic fractions of the master plate EAT-1 to α toxin (a) and binding of the periplasmic fractions of the master plate ENB-1 to NetB (b). The binding strength is indicated by absorbance at 490 nm GANGAIAH ET AL.
| 27 of 37 F I G U R E A2 Alignment of the sequences of the VHHs picked from master plate EAT-1. Conserved residues are highlighted F I G U R E A3 Alignment of the sequences of the VHHs picked from master plate ENB-1. Conserved residues are highlighted F I G U R E A4 Neutralization of nanobodies α toxin hemolysis by VHH antibodies. A twofold dilution series of VHH antibodies, starting from 5 µM concentration, was preincubated with a commercial α toxin, after which 1% sheep erythrocytes was added. Serum derived from calves immunized with recombinant α toxin was used as a control (control serum, same immunogen as used for llama immunization). The optical density of 100% hemolysis was obtained by diluting the sheep erythrocytes in distilled water. As a control, sheep erythrocytes were incubated with the α toxin, but without VHH antibodies or serum was used. This resulted in 100% hemolysis. The data are representative of three independent experiments. VHH, Variable domain of the Heavy chain of Heavy chain F I G U R E A5 Putative ENB-ID11_R56H and ENB-IA4 interacting epitopes on NetB predicted based on in silico structural modeling. (a) Nanobody sequences; the regions of differences are highlighted in brown and red. (b) Homology models of VHH3 (blue) and VHH4 (purple). (c,d) Nanobody interacting epitopes on NetB (β sandwich, rim, and stem domains; shown in red) F I G U R E A6 Inhibitory activity of L. reuteri 3632 (a) and L. reuteri 3630 (b) against C. perfringens in an agar overlay assay. The data are representative of three independent experiments F I G U R E A7 Genomic organization of pyrE locus in L. reuteri 3630 and 3632 genomes. Note that both L. reuteri 3630 and 3632 have identical genomic organization; the genomic organization of only L. reuteri 3632 is shown here

APPENDIX C
Selection of final L. reuteri candidates delivering nanobodies against NetB and α toxin.
Colonization of engineered L. reuteri candidates.

Materials and methods
Colonization of engineered L. reuteri candidates delivering different Nb clones against NetB and α toxin was determined as described in the materials and methods section.

Results
Along with their respective parent strains, NE01 and NE06 were evaluated for colonization in chicken after in ovo administration.
Screening of engineered L. reuteri candidates for efficacy.

Materials and methods
Efficacy study was conducted at Colorado Quality Research and the study was reviewed and approved by Elanco Animal Health Inc.
Animal Care and Use Committee (IACUC# 1205). Two hundred and forty Cobb 500 commercial male broiler chicks were allocated to six groups with each containing 40 birds per group (10 birds per cage).
Chicks from each replicate were housed in the same cage and replicates of the same group were housed on the same rack. The birds were housed and cared for according to the Guide for the Care and Use of Agricultural Animals in Research and Teaching. The birds were provided with ad libitum feed and drinking water. The feed ration included a commercial-type broiler diet formulated to meet or exceed requirements stipulated by the National Research Council (Council, 1994).
The L. reuteri experimental products were prepared exactly as described in the "Materials and methods" section. L. reuteri candidates were administered in drinking water every day for 21 days at a dose of 1.0 × 10 8 CFUs/bird (Table C1). Each day, vials containing the experimental product were resuspended with the appropriate volume of non-chlorinated distilled water. Following resuspension, an aliquot was used for titration. Full-day water requirement was divided into four equal parts. Each part was provided to one cage of the same treatment group. A similar procedure was applied to all the treatment groups from T2 to T6. Medicated water (containing L. reuteri experimental products) was provided to the respective treatment group birds for 24 h as ad libitum.
The NE model used in the present study includes oral administration of a mixture of E. maxima and E. acervulina followed by serial oral administration of C. perfringens. At Day 14 of age, the birds in Groups 2, 3, 4, 5, and 6 were challenged with 10,000 sporulated oocysts of E. maxima-E. acervulina/bird. The challenge inoculum was administered via oral gavage using a 10 ml syringe fitted with an 18gauge gavage needle.
Fresh C. perfringens (strain CL-15, Type A, α and β2 toxins) challenge inoculum was prepared every day from a stock culture in fluid thioglycollate broth overnight at 35°C under anaerobic conditions by Microbial Research Inc. C. perfringens was administered by F I G U R E C1 Colonization of engineered L. reuteri candidates. L. reuteri strains were administered via in ovo to 18-day embryonated chicken eggs and the birds were killed 7 days after hatching and CFUs were quantified from cecal contents. The strains were marked with rifampicin resistance to selectively isolate NE01, NE06, L. reuteri 3630, and L. reuteri 3632 from the rest of the microbiota from cecal contents. For each group, the data represent the mean ± SD of the results of five chicks