B. subtilis RIK 1285 genotype and E. coli DH-5α were grown at 37°C and 200 rpm in Luria-Bertani (LB) broth (HiMedia, Israel) medium. Competent cells of E. coli DH-5α were prepared as per standard protocol. Competent cells of B. subtilis RIK 1285 were prepared as per the Takara manual (Takara Biosciences, Israel). The human ADH4 sequence was obtained from the addgene database (https://www.addgene.org/vector-database) and optimized for codon biases at GenScript. The synthesized gene sequence and primers used for cloning and sequencing are provided in Tables S1 and S2. Plasmid pBE-S (5.9 kb) and SP DNA mixture were purchased from Takara Bio Inc. Antibiotic ampicillin and kanamycin were supplied by Sigma-Aldrich (Sigma, Israel).

A 1.2 kb polymerase chain reaction (PCR) fragment of ADH4 was cloned into a pBE-S plasmid. The resulted pBE-S:ADH4 plasmid was transformed to competent B. subtilis RIK 1285 by electroporation (1.5 kv, 100 Ω, 25 μF, time constant 1.5 ms). The selected transformant RIK 1285:ADH4 (F1) was tested for its in vitro Ethanol (0.5%) removal efficiency. A short fragment of ADH4 gene of 710 bp was also cloned to use as a control and marked as RIK 1285:ADH4 (S1). Transformants were confirmed by colony PCR and further by sequencing at The Hebrew University of Jerusalem Campus [Applied Biosystems (ABI), 3730xl DNA Analyzer]. Detail about the PCR conditions and transformation procedure has been provided in the supplementary method. Changes in bacterial population were calculated by colony forming units (CFU) counting using a serial dilution method.

Wild-type RIK 1285 and ADH4 transformants were grown in 30 mL of LB medium in the presence of appropriate antibiotics. After overnight incubation at 37°C, the culture was centrifuged at 12,000 rpm for 15 min at 4°C, and proteins were extracted as described in the supplementary section, page 1. The ADH4 expression was confirmed by western blot using polyclonal antibodies. To construct the SP-library, the plasmid RIK 1285:ADH4 (F1) was randomly ligated with the SP-DNA mixture. The plasmid library was transformed into B. subtilis RIK 1285 by electroporation and plated on LB-agar medium supplemented with kanamycin (10 μg/mL). Approximately 100 clones were screened for the in vitro Ethanol removal assay and the selected efficient colony was tested for ADH activity.

A minimal basal medium (MM) containing 64 g/L Na₂HPO₄.7H₂O, 15 g/L KH₂PO₄, 5 g/L NH₄Cl, and 2.5 g/L NaCl with 2 mmol/L MgSO₄ and 0.1 mmol/L CaCl₂ was autoclaved (121°C, 15 min), cooled down at room temperature, and supplemented with 0.5% Ethanol. Approximately 1 × 10⁷ bacterial cells were mixed in the Ethanol-supplemented minimal medium and incubated for 24 h at 37°C with shaking of 200 rpm. After the incubation period, each treatment medium was filtered with a 0.22 μm Millipore filter and analyzed on a gas chromatography (GC)-FID to measure the Ethanol concentration. A standard curve (R² value close to 0.999) covering 0.1% to 0.5% of 95% Ethanol was used to calculate Ethanol concentrations. Detail about the condition has been provided in the supplementary method.

Overnight grown cultures (20 μL) of B. subtilis RIK 1285, RIK 1285:ADH4 (F1), and RIK 1285:ADH4 (SP-64) were inoculated into a 20 mL sterile nutrient broth medium supplemented with the appropriate antibiotics. Cultures were grown for 24 h, and the absorbance of the culture in each treatment was determined at 600 nm in an ultraviolet-visible (UV-Vis) spectrophotometer in triplicate sets using un-inoculated broth as a blank.

ADH activity was assessed in the culture supernatant following the standard protocol of nicotinamide adenine dinucleotide (NAD)⁺ reduction. Briefly, the supernatant (100 μL) B. subtilis RIK 1285, RIK 1285:ADH4 (F1), and RIK 1285:ADH4 (SP-64) were added to the phosphate buffer (pH 8.0), which contained NAD⁺ at a final concentration of 10 mmol/L. The reaction was started by adding Ethanol (0.016 mol/L), and the rate of conversion of NADH was monitored at 340 nm. The ADH activity was calculated by using the molar extinction coefficient of NADH (6.22). Protein concentration was determined by the Bradford method.

To test the stress tolerance, vegetative and spores (1 × 10⁹) of B. subtilis RIK 1285, RIK 1285:ADH4 (F1), and RIK 1285:ADH4 (SP-64) were boiled at 80°C for 20 min. Similarly, acid tolerance was tested by growing the culture in simulated gastric fluid (SGF) of acidic pH 2.0 for 1 h. The bacterial colony was counted by serial dilution plating method. Additionally, to test the colonization ability, spores of RIK 1285, RIK 1285:ADH4 (F1), and RIK 1285:ADH4 (SP-64) were injected into 4 weeks C57BL/6J male mice (n = 3), fed on a normal diet (ND) by the oral gavage. Feces were homogenized in phosphate-buffered saline (PBS) buffer and serial dilution was cultured on an LB-agar plate with appropriate antibiotics and incubated at 37°C for 16 h. The level of colonization was calculated and expressed as the number of CFU per g of feces.

After 10 weeks on the diet, mice were fasted for 12 h, and sacrificed in random order by isoflurane overdose. Before the sacrifice, each mouse’s body weight in tested groups of ND and HFD was recorded. The collected epididymal adipose tissue and liver tissue were immediately stored at –80°C. Similarly, intestinal tissues were collected and frozen for further experimental purposes. The feces from each group (ND & HFD) were stored for microbiota analysis. Blood samples from each mouse were collected and centrifuged at 8,000 g for 10 min, at 4°C, to separate the plasma from the remaining components. The serum was used for aspartate aminotransferase (AST) and alanine aminotransferase (ALT) enzymatic assays. Additionally, total cholesterol (TC), triglycerides (TGs), and bilirubin content were recorded according to the standard protocol. Finally, the glucose content in blood was checked from the tail vein using a glucometer.

Formalin-fixed pancreases from each treatment group were cut into small (4 μm) sections and processed for immunohistochemistry. Slides were de-paraffinized in xylene and sequentially rehydrated using decreasing Ethanol concentrations down to water. Antigen retrieval required microwaving slides with 10 mmol/L citrate buffer for 30 min. Nonspecific binding was blocked by treating sections with Biocare blocking reagent (Biocare Medical, Concord, CA, USA) for 30 min at room temperature, followed by incubation with the primary antibody diluted in blocking buffer. The following primary antibodies and dilutions were used: phospho p-S6 protein (Cell Signalling Technology Inc., USA, 1:300); insulin (Dako Products, USA, 1:300); glucagon (Santa Cruz Biotechnology, USA, 1:300). The 100 uL of primary antibodies mix was loaded onto the slide, closed with parafilm, and kept overnight at 4°C. Slides were washed twice in PBS, incubated for 30 min with a secondary antibody, washed twice with PBS, stained with diaminobenzidine, and counterstained with hematoxylin. Images were captured by Nikon A1R confocal microscope and analyzed using National Institutes of Health (NIH) ImageJ software (v1.31, http://rsb.info.nih.gov/ij/index.html).

To assess the colonization, feces of each treatment were suspended to a concentration of 1 g/mL in PBS buffer and gently homogenized with a vortex for 1 min. The collected fecal suspension was heated at 80°C for 20 min to kill vegetative cells. Serial dilutions were plated on a solid LB-agar plate with the appropriate antibiotics to count the viable spore population. Plates were incubated at 37°C overnight and spores count was represented as CFU per g of feces. To check the inoculated bacterial presence in the collected feces, another set of feces from each treatment was homogenized, and the plasmid was extracted following the standard protocol (Qiagen, Germany). Finally, 50 ng of extracted plasmid was used for ADH4 amplification with a suitable primer set to confirm the presence of injected bacteria.

Fecal samples from each treatment group were collected in sterile tubes and immediately stored at –80°C. A fast DNA stool kit (Qiagen, Germany) was used to isolate DNA following the manufacturer’s protocol. DNA concentration and purity were determined with a NanoDrop™ 2000 UV-Vis spectrophotometer (Thermo Fisher Scientific, USA). Five ng per reaction was used to generate small subunit (SSU) ribosomal RNA (rRNA) amplicons of V4 variable regions using the bacterial primer set CS1_515F (5’-ACACTGACGACATGGTTCTACA_GTGCCAGCMGCCGCGGTAA) and CS2_806R (5’-TACGGTAGCAGAGACTTGGTCT_GGACTACHVGGGTWTCTAAT) [42–44]. The protocol was set as follows: denaturation at 95°C for 3 min, followed by 30 cycles of denaturation at 95°C for 30 s, annealing at 55°C for 30 s, extension at 72°C for 30 s, and final extension at 72°C for 5 min. Library sequencing was done from both ends on an Illumina platform at the Research Resources Centre of the University of Illinois at Chicago following the standard protocols [45] (http://www.earthmicrobiome.org/emp-standard-protocols/16s/), and fastq files were generated using the illumine software package bcl2fastq.

Prior to analysis, forward and reverse primer sequences were removed from the raw fastq files with Cutadapt v.3.265 [46]. Primer-trimmed sequences were quality-filtered and merged into the error-corrected amplicon sequence variants (ASVs) using DADA2 v.1.18 [47]. The ASVs were length-filtered to keep sequences at a 251–255 bp length, to meet the expected size for the V4 variable region of the SSU rRNA gene. Chimera sequences were removed using the removeBimeraDenovo function of the DADA2 package. These high-quality sequences were annotated to the SILVA SSU rRNA reference alignment (release138) using the Ribosomal Database Project (RDP) classifier [48] as implemented in Mothur v.1.44.2 software [49] with a minimum confidence threshold of 50%. ASVs that were classified as ‘chloroplast’ and ‘mitochondria’ or did not match any taxonomic phyla level were removed from the final sequence dataset. The remaining 1,247,299 (38,978 sequences per sample ± 12,927 sequences per sample) high-quality sequences were used for downstream analysis. Alpha diversity indexes were calculated based on the number of unique ASVs (richness), Shannon’s, and Faith’s phylogenetic diversity (PD). Estimates of alpha diversity indexes were obtained, based on 9,466 randomly selected sequences per sample, and tested for significant statistical differences using a two-sided Wilcoxon test. Prior to beta-diversity analysis, the final dataset was normalized by cumulative sum scaling (CSS) [50]. An unconstrained principal coordinates analysis (PCoA) based on Bray-Curtis, weighted, and unweighted UniFrac dissimilarities were performed to quantify the major variance components of beta-diversity. The diet effect on the microbial community dissimilarity was estimated with a permutational analysis of variance (PERMANOVA) with 1,000 permutations. Statistical analysis was conducted in R v.4.0.4 using the phyloseq v1.34 and vegan v2.5-7 R packages [51, 52].

All statistically significant differences between each treatment were evaluated using variance analysis [one-way analysis of variance (ANOVA)] followed by the t-test using the significant level of P < 0.05 for all analyses. The analysis was performed with the JMP 7.0.2 and JMP Pro software suites (SAS Institute, USA).

To develop the probiotic strain featuring enhanced ADH secretion, a full-length ADH4 gene (1.2 kb) and a short fragment of 710 bp (as a control) were amplified (Figure 1A) and cloned into a pBE-S vector under a constitutive promoter. The constructed plasmid pBE-S:ADH4 was transformed into B. subtilis by electroporation and positive transformants were selected on LB-agar plates supplemented with kanamycin (10 μg/mL). The obtained transformants were screened by colony PCR and further confirmed by sequencing (Figure S1). Additionally, the selected transformant RIK 1285:ADH4 (F1) was evaluated for its in vitro Ethanol removal ability, and it showed significant (P < 0.05) Ethanol removal (60%) when compared to the wild-type strain (Figure 1B). Furthermore, CFU counting showed an increase in bacterial population, following incubation in LB-medium supplemented with Ethanol (0.5%, Figure 1C). The crude lysates of the overnight-grown wild-type culture, selected transformant RIK 1285:ADH4 (F1), and RIK 1285:ADH4 (S1) were examined for protein expression by western blotting. We observed a monomeric 21 kDa protein expression in full-length transformant RIK 1285:ADH4 (F1) (Figure 1D).

An SP library driven by aprE promoter was generated and further transformed into B. subtilis, yielding the recombinant strains from B. subtilis RIK 1285:ADH4 (SP-1) to RIK 1285:ADH4 (SP-100). The resulting transformants were tested for their in vitro Ethanol removal capability. The incorporation of SPs significantly enhanced the ability of the transformants to reduce Ethanol concentration. While some of the SP transformants like SP-1, SP-7, SP-13, SP-15, SP-27, SP-31, SP-33, SP-37, SP-55, SP-72, SP-74, etc. showed 35% to 40% of Ethanol removal, the SP-43, SP-46, SP-50, SP-67, and SP-70 transformants were able to consume up to 60% of added Ethanol. Finally, transformants SP-63 and SP-64 showed the highest Ethanol removal in the range of 70% to 75%, among all transformants (Figure S2). Based on these observations, the B. subtilis transformant SP-64 was selected for downstream analysis.

Growth comparison studies of selected transformant RIK 1285:SP-64 with wild-type RIK 1285 showed no significant plasmid effect on the host growth rate under the evaluated conditions (Figure 1E). The NADH reduction assay was used to measure ADH activity, having found the highest ADH enzymatic activity in SP-64 transformants (470.68 U/mg protein, Figure 1F). Time-dependent ADH4 secretion showed the ADH4 secretion in the culture supernatant by the SP-64 (Figure S3). SP sequence analysis identified the phoB SP in the selected transformant SP-64.

As B. subtilis is the spore-forming bacteria, heat and acid resistance of RIK 1285:ADH4 (F1), and RIK 1285:ADH4 (SP-64) were evaluated (Figure S4). After confirming the viability of these bacterial strains in simulated gastric conditions, we did an animal trial study to test the colonization of engineered bacterial strains in mouse model, and also the dosing interval when supplementing as spores. This was done in order to evaluate the effectiveness of the GI tract colonization following gavage supplementation. Administrated bacterial spores in feces were detected after 1 day to 3 days with a gradual decrease (Figure S5).

Based on higher ADH activity and in vitro Ethanol removal efficacy, we selected ADH4-secreting SP-64 strain for in vivo study. After a 10-week diet protocol, mice in the HFD group were shown to gain more weight than that in ND groups. Mice inoculated with RIK 1285:ADH4 (SP-64) resulted in the lowest (P < 0.05) body and liver weight gain among HFD groups (Figure 2A, 2B). The changes in the body weight throughout the 10 weeks have been provided in Figure S6. Additionally, the adipose tissue mass in the ND group was substantially lower than that in the HFD-treated group (Figure 2C). Moreover, the HFD group treated with SP-64 spores showed significantly (P < 0.05) reduced adipose tissue and caecum weight gain when compared to the other HFD treatments (Figure 2D). Finally, the fasting glucose levels of SP-64-treated mice were similar to those in the ND group and significantly lower among the tested HFD groups (Figure 2E).

The analysis revealed a significant (P < 0.05) decrease in the cholesterol level, liver damage markers, and ALT and AST enzymatic activities in the SP-64-treated group. On the other hand, high ALT and AST levels were found in the untreated HFD group (Figure S7A, S7B). Nevertheless, all bacterial treatments [i.e. RIK 1285, RIK 1285:ADH4 (F1), and RIK 1285:ADH4 (SP-64)] were able to lower ALT and AST levels. Additionally, all bacterial treated groups showed a lower level of bilirubin, cholesterol, and TG levels in the HFD group (Figure S7C–E).

After we observed a significant reduction in the glucose blood level in the ADH4-secreting bacteria-treated mice, the pancreas beta cells integrity was analyzed and compared between the treatments. Pancreas immunohistochemistry staining showed that HFD induced higher phospho p-S6 level. S6 is an effector of mammalian target of rapamycin (mTOR) that impaired insulin signals indicating insulin resistance (Figure 3A). In the HFD-treated mouse, insulin-containing beta cells density was lower within the beta cell islet. HFD-treated mouse with ADH4-secreting probiotic bacteria-maintained islet high density of insulin expressing cells and lower level of phosphor p-S6 staining intensity even below ND treated mouse control level (Figure 3B).

B. subtilis administrated spores’ population decreased following oral administration, and it was lowest on the 3rd day. CFU counting showed that the spores’ population decreased over time (Figure 4A). To further support the presence of the probiotic bacteria, PCR with the ADH4 gene showed the expected amplification in RIK 1285:ADH4 (F1) and RIK 1285:ADH4 (SP-64) spores-inoculated group, showing its presence in feces (Figure 4B). There was no amplification in inoculated RIK 1285 spore from the ND and HFD treated groups.

Gut microbiota imparts various roles in homeostasis, metabolic abnormalities, and inflammation. Therefore, we evaluated the microbiota changes following oral administration of tested bacterial spores at different time points, using SSU rRNA genes high-throughput sequencing techniques. High-throughput sequencing of SSU rRNA genes yielded a total of 1,247,299 high-quality sequences (range 9,466–67,795; median 34,936 sequences per sample), revealing profound changes in microbial community diversity and composition associated with mouse diets and bacterial treatments. Rarefaction curve-based alpha diversity analysis also indicates that mouse age and diet significantly affected microbial community species richness (Figure 5A). Overall, more diverse prokaryotic communities were observed at the initial stages when compared to later stages (Figure 5A). A variation in gut-microbiome diversity was observed, featuring a decrease in PD in the first week, followed by progress toward a phylogenetic dispersion at 6 weeks, under both ND and HFD-treated mice (Figure 5B). The microbial community dispersion (weighted UniFrac distances) was higher after 6 weeks when compared to week 1 and week 10 (Figure 5C). Fecal microbiota of the HFD groups clustered differently from ND, illustrating a dominant effect of diet. To determine the influence of diet and probiotic treatment on microbial communities, unconstrained PCoA ordination and PERMANOVA analyses based on the weighted UniFrac-distance matrices were performed. These analyses underlined the important role of mouse physiological age on the distribution of microbial communities. Accordingly, mouse age explained almost 50% of the variation in the prokaryotic community composition (R² = 0.5, P < 0.0001, Figure 5C).

The vast majority of retrieved bacterial SSU rRNA gene sequences were affiliated with the phyla Proteobacteria, Bacteroidota, Actinobacteria, and Firmicutes, which comprised up to 100% of bacterial sequences (Figure 6A). At the phylum level, Firmicutes were enriched at week 1 and week 10 (75% on average) under HFD, whereas at 6 weeks Bacteroidota and Proteobacteria were dominant. Interestingly, SP-64 inoculation significantly increased Firmicutes and decreased Bacteroidota compared to other treatments. Under ND, they maintained a constant presence. Phylum Actinobacteriota and Verrucomicrobiota were not observed among bacterial-treated groups under HFD. Bacteria treatments significantly reduced the Bacteroidota family.

An HFD substantially affects the microbial taxonomic composition. A significant taxonomic shift was observed after 6 weeks with an increase in proinflammatory taxa of γ-Proteobacteria, Bacilli, and Bacteroidia (Figure 6B). Bacilli population was found to be higher after 10 weeks, among the bacteria-treated groups under HFD. HFD negatively affected the relative abundance of Bacilli families and stimulated the propagation of proinflammatory members of the Staphylococcaceae family. Nonetheless, B. subtilis treatments partially restored the overall quantities of Bacilli families (Figure 6C). While the Staphylococcaceae family increases their relative abundance, Erysipelotrichaceae was negatively affected by an HFD. These opposite responses of the Bacilli families resulted in an overall increase in the relative abundance of Bacilli due to prolonged exposure to an HFD (Figure 6C). However, the highest population of Bacillaceae was recorded in the SP-64 treated group.

A total of eight genera were observed as the most prevalent in the microbial community. The cumulative abundance of these taxa gradually increased with time. The quantity of these genera in the spore-unamended treatment increased from 34.6% ± 1.1% to 50% ± 3.9% under an ND and from 38.1% ± 1.3% to 43% ± 0.2% under an HFD (Figure 6D). A similar taxonomic succession was observed in the spores-amended treatments. Nevertheless, RIK 1285:ADH4 (F1) was associated with the highest increase in cumulative abundance from 38.1% ± 1.3% to 79.8% ± 0.4% (Figure 6D). While the dominant genera composition was stable in mice on an ND, we detected a substantial variation in mouse microbiomes under HFD. The relative abundance of the Faecalibacterium genus gradually increased from 0.8% to 31.2% ± 10.5% under an ND. When comparing the taxonomic profile of the 8 groups at the genus level, we found that with the exception of Bacillus and Lachnospiraceae, the relative abundance of other genera (> 1% relative abundance in at least one sample) decreased in the entire sample after 10 weeks. This variation pattern of the genera reveals that Bacillus was a more effective colonizer among all groups, and it was the major driver of the community composition. However, some genera showed distinct patterns regarding the variation of their relative abundance among different treatment groups. Bacteroides species were positively associated with an HFD. Under ND conditions, Bacteroides abundance remained low (3.4% ± 2.4%) after 10 weeks. On the other hand, under HFD conditions, Bacteroides contributed to almost 20% of the microbial community (Figure 6D). Supplementation of an HFD combined with Bacilli spores, suppressed the development of Bacteroides populations. Among the Bacteroides under an HFD, SP-64 inoculation decreased the population of Flavobacteriaceae and Bacteroidaceae (Figure 6E).

An HFD negatively affected the relative abundance of the probiotic genus Bifidobacterium, taxonomically affiliated with the Actinobacteria phylum. Under ND conditions, the Bifidobacterium relative abundance rose in an age-dependent manner from 3.4% ± 1.5% at week 0 to 16.6% ± 5.1% by the end of the experiment (Figure 6F). In contrast, after 1 week of feeding, an HFD resulted in a decline of the Bifidobacterium to undetectable levels. Nevertheless, feeding mice with Bacilli spores resulted in the recovery of the Bifidobacterium abundances. Furthermore, after 6 weeks on an HFD supplemented with Bacilli spores, Bifidobacterium was detected in all spores-amended treatments, but absent in spores-unamended treatments (Figure 6F). These results suggest that the improvement in HFD- induced metabolic disorders by SP-64 supplementation was associated with significant gut microbiota composition changes.