Swine probiotic strains: a new bacteriocin isolate explained
A swine probiotic candidate that tests resistant to erythromycin and carries three copies of a vancomycin resistance gene looks like a strain to discard. That is the first reading of the data on a newly characterized bacteriocin producing isolate taken from the pig gut. and it is the wrong reading.
Where those genes sit matters more than whether they are present. In this isolate, every resistance and virulence determinant found was on the chromosome. None were on the plasmids. and plasmids are the part of a bacterial genome that moves between strains.
That distinction is the whole argument for doing genomic safety screening properly instead of chasing gene names. A 2026 characterization of this strain ran the genome, the phenotype and the antimicrobial activity side by side. and the results are a useful template for anyone evaluating a probiotic claim in feed.
Why antibiotic reduction needs better candidates, not more claims
Pig production faces pressure on antibiotic use from three directions at once: resistance surveillance, import requirements. and retail specifications. The feed industry response has been a flood of products labelled probiotic, most carrying far less evidence than the label implies.
The standard of proof that matters is narrow. A candidate has to survive the stomach, cope with bile, inhibit a relevant pathogen. and carry no mobile resistance it could hand to something worse. Four tests. Most products have been through one or two.
What swine probiotic strains have to survive before reaching the gut
The isolate was exposed to a pH gradient and a bile salt gradient for 20 hours, in triplicate, against a control at pH 6.5 with no bile.
| Condition | Viable count (CFU/mL) | Relative survival | p value |
|---|---|---|---|
| Control, pH 6.5, no bile | (4.6 ± 0.35) x 10^8 | 100% | reference |
| pH 6.0 | (1.4 ± 0.16) x 10^9 | 304.3% | < 0.0001 |
| pH 8.0 | (4.8 ± 0.62) x 10^8 | 104.3% | > 0.9999 |
| pH 4.0 | (1.7 ± 0.24) x 10^8 | 36.9% | < 0.0001 |
| pH 3.0 | (1.0 ± 0.35) x 10^4 | 0.002% | < 0.0001 |
| pH 2.0 | < 10 | below detection | 0.0014 |
| Bile salts 0.1% | (1.0 ± 0.25) x 10^9 | 217.4% | < 0.0001 |
| Bile salts 0.2% | (4.0 ± 1.00) x 10^7 | 8.7% | 0.0039 |
| Bile salts 0.3% | (4.0 ± 1.10) x 10^7 | 8.7% | < 0.0001 |
| Bile salts 0.4% | (3.3 ± 0.25) x 10^7 | 7.2% | 0.0028 |
| Bile salts 0.5% | (2.8 ± 0.72) x 10^7 | 6.1% | < 0.0001 |
Two things in that table deserve attention, and only one of them is flattering.
The flattering one is the pH 6.0 result, where viable counts rose to 304.3 percent of control. Growth at 0.1 percent bile salts also came in above control at 217.4 percent. Whatever else this strain is, it is not merely tolerating the upper gut. It is growing there.
The unflattering one is pH 3.0. Survival dropped to 0.002 percent, which is a four log kill. At pH 2.0 no colonies were recovered at all.
Read the survival claim against the survival numbers
This is where the marketing version and the data version separate. A strain can be described as viable under simulated gastrointestinal conditions. and every word of that can be true in the sense that some cells were still countable. Viable is not the same as comfortable.
If the target is the small intestine and the strain has to cross a gastric pH that in pigs routinely sits between 2 and 3, a 99.998 percent loss at pH 3.0 is the number that matters. Encapsulation is the obvious answer. and this study did not test it.
What it actually kills
The activity came from the cell free supernatant, neutralized and heat treated, which rules out organic acid and hydrogen peroxide as explanations.
| Indicator organism | Inhibition zone | Activity area | p value |
|---|---|---|---|
| Listeria monocytogenes | 24.52 ± 1.03 mm | 4.34 ± 0.07 cm2 | 0.0005 |
| Enterococcus faecalis | 16.14 ± 0.79 mm | 1.66 ± 0.20 cm2 | 0.0001 |
| Staphylococcus hyicus | 15.24 ± 1.74 mm | 1.44 ± 0.22 cm2 | 0.0010 |
| Streptococcus suis | 12.20 ± 1.47 mm | 0.78 ± 0.27 cm2 | 0.0086 |
| Escherichia coli | no inhibition | none | not applicable |
Three of those four are genuine swine production problems. Streptococcus suis and Staphylococcus hyicus are directly implicated in pig disease. Listeria is a feed and processing concern.
Escherichia coli showed no inhibition at all. That is the expected result for a bacteriocin from a Gram positive producer, and it is the honest boundary of this candidate. No labelling changes that.
Enzyme treatment confirmed the mechanism. Trypsin abolished activity in both the crude supernatant and the purified active fraction, while pepsin and proteinase K reduced or removed it. The inhibitory compound is a protein, which is what a bacteriocin should be. Only one HPLC fraction, eluting between 30.9 and 36.5 minutes at an acetonitrile concentration of 46.3 to 52.6 percent, carried activity.
The safety picture is the interesting part
Whole genome sequencing gave a 2,741,738 base pair genome at 38.1 percent GC, with 2,632 coding sequences, 91 tRNAs and 108 pseudogenes. The chromosome is about 2.58 Mb, with one large plasmid at 160.63 Kb.
Average nucleotide identity confirmed the species at above 97.8 percent against the reference strain and below 95.5 percent against Enterococcus faecium, which is the separation that matters given how different the risk reputations of the two species are.
Four bacteriocin like sequences were identified: enterolysin A, mundticin ATO6, enterocin SE-K4 and enterocin P, plus two immunity proteins. Conservation varied widely. Enterocin P was highly conserved, with 47 genomes at 100 percent amino acid identity and an average of 94.5 percent. Mundticin ATO6 was far less conserved, averaging 68 percent identity with only three genomes at 100 percent.
Resistance genes: present on the chromosome, absent from the plasmid
The chromosome carries three copies of vanZ, associated with low level glycopeptide resistance, along with efrA, efrB and efmA, which cover macrolide, fluoroquinolone, tetracycline and chloramphenicol efflux. Also present are aac6′-PRIME, an aminoglycoside acetyltransferase. and msrC, which mediates erythromycin resistance.
No plasmid associated resistance or virulence genes were detected. That is the finding that reframes the opening paragraph. Chromosomal determinants can still be expressed, but they do not transfer horizontally the way a plasmid does. and for a feed additive that difference is decisive.
Phenotype beats genotype
Susceptibility testing showed the strain was susceptible to almost everything it was tested against.
| Antibiotic | MIC | Breakpoint | Call |
|---|---|---|---|
| Erythromycin | 32 ug/mL | 4 ug/mL | Resistant |
| Streptomycin | 128 ug/mL | 128 ug/mL | Susceptible |
| Kanamycin | 1024 ug/mL | 1024 ug/mL | Susceptible |
| Gentamicin | 16 ug/mL | 32 ug/mL | Susceptible |
| Chloramphenicol | 2 ug/mL | 16 ug/mL | Susceptible |
| Clindamycin | 0.5 ug/mL | 4 ug/mL | Susceptible |
| Ampicillin | 1 ug/mL | 2 ug/mL | Susceptible |
| Vancomycin | 1 ug/mL | 4 ug/mL | Susceptible |
| Tetracycline | 0.25 ug/mL | 4 ug/mL | Susceptible |
One resistant call out of nine, against erythromycin. and it matches the msrC gene found on the chromosome. Vancomycin MIC came in at 1 ug/mL despite the three vanZ copies, which is exactly the low level, sub clinical phenotype that vanZ is associated with.
The strain was gamma hemolytic. and negative for catalase, coagulase and gelatinase. It carries three hemolysin related genes and a collagen adhesin precursor, but the adhesin is annotated as a pseudogene, so it is unlikely to be expressed. Two intact prophages were also found in the chromosome, which is worth noting for anyone thinking about long term genomic stability.
What buyers and nutritionists should take from this
Ask where the resistance genes are, not how many. A chromosomal count without a plasmid screen says almost nothing about transfer risk.
Ask for MIC values, not only gene screens. Genotype predicted erythromycin resistance and the MIC confirmed it, but genotype also flagged tetracycline and vancomycin efflux the phenotype did not deliver.
Ask for survival data at pH 2.0 and pH 3.0, not one gastric pH point. A strain that survives pH 3.0 with a 99.998 percent loss is a different product from one that holds up at that pH.
Ask what the spectrum misses. This candidate has no activity against E. coli, so it will not touch post weaning colibacillosis.
Treat genomic safety as necessary, not sufficient. The species is not on the qualified presumption of safety list used in the European Union because species level safety data are limited. Strain level evidence substitutes for a regulatory listing, it does not replace one.
Where the evidence stops
Everything here is in vitro. No pig trial, no colonization study, no dose response, no challenge model, no safety or toxicology work in the target species. A strain that inhibits Streptococcus suis on a plate has not been shown to reduce it in a barn.
Tolerance was measured in triplicate over 20 hours under simulated conditions. A simulated stomach is not a stomach. Feed matrix, transit time and the resident microbiota all change the picture.
The genome predicts what the strain might do; it does not measure what it does. Two intact prophages were identified, and the stability of the bacteriocin loci was not assessed.
One boundary deserves stating plainly. Antibiotic reduction is a management outcome, not a product outcome. No probiotic removes the need for biosecurity, sensible stocking density, and diagnosis before treatment.
Frequently asked questions
What makes a good swine probiotic candidate?
Four properties, and most products have evidence for one or two. The strain must survive gastric pH, tolerate intestinal bile salts, inhibit a pathogen that actually matters in pigs, and carry no mobile resistance genes. The isolate characterized in 2026 met three of those four convincingly, with a clear caveat on gastric survival at low pH.
How do probiotic bacteria survive stomach acid?
This isolate survived well near neutral pH and poorly under strong acid. At pH 6.0 viable counts reached 304.3 percent of control, but at pH 3.0 survival fell to 0.002 percent and at pH 2.0 no colonies grew at all. Survival under strong acid appears to require formulation rather than intrinsic tolerance.
Does bile tolerance predict probiotic performance?
Partly, and the pattern is not linear here. At 0.1 percent bile salts the strain grew to 217.4 percent of control, but at 0.5 percent bile salts survival dropped to 6.1 percent. The threshold behaviour matters more than a single bile concentration, and most product claims quote only one number. Ask for a full gradient from 0.1 to 0.5 percent instead.
What pathogens does this strain inhibit?
It inhibited Listeria monocytogenes with a 24.52 mm zone, Enterococcus faecalis at 16.14 mm, Staphylococcus hyicus at 15.24 mm and Streptococcus suis at 12.20 mm. Escherichia coli showed no inhibition at all, which is the expected limit of a bacteriocin from a Gram positive producer. Two of those four are directly implicated in pig disease.
How do you know the inhibition is a bacteriocin?
Enzyme treatment confirmed it. Trypsin eliminated activity entirely in both the crude supernatant and the purified active fraction, while pepsin and proteinase K reduced or removed it. The active compound is a protein, which rules out acid or hydrogen peroxide as the cause. Heat treatment had already excluded live cells.
Are resistance genes in a probiotic a safety problem?
It depends on location. In this isolate all resistance determinants, including three vanZ copies and msrC, were chromosomal. No plasmid associated resistance or virulence genes were found. Chromosomal genes are far less likely to transfer horizontally, which is the risk that matters for a feed additive.
What does the resistance phenotype look like?
Nine antibiotics were tested by microdilution. Eight came back susceptible, including tetracycline at an MIC of 0.25 ug/mL against a breakpoint of 4, and vancomycin at 1 ug/mL against a breakpoint of 4. Erythromycin was the single resistant call, at an MIC of 32 ug/mL against a breakpoint of 4, matching the msrC gene found on the chromosome.
Why does the vanZ gene not cause vancomycin resistance here?
Three copies of vanZ were found on the chromosome, and vanZ is associated with low level glycopeptide resistance rather than the high level resistance seen with vanA or vanB operons. The measured vancomycin MIC was 1 ug/mL against a breakpoint of 4, so the genotype did not translate into a resistant phenotype.
Can a probiotic replace antibiotics in pig feed?
Not on this evidence. Everything here was measured in vitro, with no pig trial, no challenge model and no dose response work. A probiotic candidate is a tool inside an antibiotic reduction program, alongside biosecurity, stocking density and diagnosis before treatment, not a substitute for any of them.
Is Enterococcus lactis safe to use in feed?
The species is not currently on the qualified presumption of safety list used in the European Union, because species level safety data are limited. This work provides strain level genomic evidence, which is a starting point rather than a regulatory approval. Commercial use would still need the toxicology and animal work a formal assessment requires.
What genome features support the safety case?
The genome is 2,741,738 base pairs at 38.1 percent GC with 2,632 coding sequences. Average nucleotide identity was above 97.8 percent to the reference strain and below 95.5 percent to Enterococcus faecium. The strain was gamma hemolytic and negative for catalase, coagulase and gelatinase, a combination that argues against a virulent enterococcal profile.
What are the biggest gaps in this research?
No animal data at all. There was no pig trial, no colonization or shedding study, and no toxicology work. Tolerance was measured in simulated conditions over 20 hours in triplicate. The genome also contains two intact prophages, and the stability of the bacteriocin loci over time was not assessed. These are the gaps a full feed registration would need to close.
The honest summary
This is a well characterized candidate, not a finished product. It kills the right Gram positive organisms, the active compound is confirmed as a protein, the genome is clean where it counts, and eight of nine MICs came back susceptible. It also loses 99.998 percent of its cells at pH 3.0, and there is not a single pig in the dataset.
Both halves of that sentence belong together. The value of this work is not that it found a replacement for antibiotics. It is that it shows what a properly screened candidate looks like, and how far short of that bar most of what is currently sold falls.
*This article summarizes published research for informational purposes. It does not constitute veterinary or regulatory advice. Always follow local regulations and the product label when using veterinary medicines.
