When the gut is invaded by Campylobacter jejuni, the intestinal epithelium does what mammalian mucosa typically does on a pathogen alarm. It secretes LL-37 ↗, a 37-residue cathelicidin antimicrobial peptide, into the lumen as a first-pass disinfectant. In a paper published online May 20 in Science Advances ↗, a team of bacterial-pathogenesis researchers tested recombinant LL-37 against a broad panel of clinical isolates of the pathogen, which is one of the most common causes of human gastroenteritis worldwide. The peptide killed 86.3 percent of the isolates it was challenged with, the population that grew normally on standard media but that LL-37 reliably disrupted at the membrane level. That is most of the panel but not all of it.
The minority that survived LL-37 exposure had a single trick in common. They cleaved the peptide.
The protease the resistant Campylobacter isolates use is HtrA, a conserved bacterial serine protease. The authors traced the regulatory loop in two steps. Step one: LL-37 itself, when sensed by the bacterium, activates a transcriptional regulator called NssR. Step two: NssR up-regulates htrA expression, and the additional HtrA the bacterium synthesizes is secreted into the same lumenal compartment as the peptide. Once secreted, HtrA cleaves LL-37 at a specific peptide bond, between residues 20 (isoleucine) and 21 (valine), in a region of the cathelicidin's amphipathic helix that is structurally essential for membrane disruption. The cleaved fragments cannot form the alpha-helical pore that the intact peptide depends on, and the antimicrobial activity collapses. The resistant subset survives.
The framing here is what makes the paper read like more than another resistance story. The bacterium is not blocking the antimicrobial peptide at its target, and it is not pumping the peptide out across its envelope. It is reading the peptide as a chemical signal, switching on a protease in response to that signal, and then cutting the peptide at one of its load-bearing bonds. The resistance is a feedback loop, not a constitutive defense, and it requires the peptide to fire it.
The authors then asked whether the loop could be closed at the substrate end. If HtrA recognizes Ile20-Val21 specifically, swapping those two residues to ones the protease does not see should produce a peptide that the resistant Campylobacter cannot inactivate. They built LL-37 I20M/V21R, a two-residue variant that replaces isoleucine 20 with methionine and valine 21 with arginine, and tested it both in vitro and in a mouse colonization model. The variant retained antimicrobial activity against the original LL-37-susceptible panel and added activity against the resistant subset, because the cleavage step was no longer available to the resistant bacteria. In mice, the variant promoted bacterial clearance from the gut in a way wild-type LL-37 did not.
Two residues, swapped on the substrate side rather than on the protease side, restored the antimicrobial relationship.
A few cautions sit alongside this. The human cathelicidin is not an approved drug anywhere. The clinical evidence base for LL-37 is a small set of topical wound-healing trials (chronic venous leg ulcers and diabetic foot ulcers) and one oral COVID-19 trial, and there is no controlled human evidence for systemic exogenous LL-37 at any dose, by any route. The variant here was tested in mice, not patients, and the relevant question for any future translation is whether LL-37 I20M/V21R retains the same selectivity window as the wild-type peptide, particularly the concentration-dependent host-cell cytotoxicity that constrains LL-37 dosing in general. The authors do not characterize that window in this paper, and they do not report on the variant's activity against other Gram-negative pathogens whose HtrA orthologs may or may not recognize the same bond.
What the paper does establish is a reproducible mechanism. The bacterial side of the host-pathogen exchange is not a passive target. It listens for the antimicrobial peptide, makes more of a specific protease in response, and cuts the peptide at a bond the cathelicidin needs to function. The structural news is that the bond can be moved, and when it moves, the resistance loop stops closing. Whether that observation generalizes from gut Campylobacter to other mucosal pathogens with their own NssR analogs is the question the next experiment will need to answer.