Noise recorded from an air conditioner and a heat pump water heater injured mouse kidneys over five days — but only its low-frequency component, and blocking endothelin signalling prevented it.

Noise is regulated by loudness. Limits are written in decibels, usually A-weighted to approximate human hearing sensitivity — which means the weighting deliberately discounts low frequencies, on the reasoning that what you cannot hear well cannot bother you much.

A study in Environmental Science & Technology from Nagoya University tests that reasoning in mice and finds against it. The low-frequency component of ordinary appliance noise damaged kidneys; the high-frequency component at the same sound pressure did not.

The work was led by a team including Takumi Kagawa of Nagoya University’s Graduate School of Medicine and Masashi Kato of the Department of Occupational and Environmental Health.

Real noise, split in two

The experimental design is the strength of the study. Rather than using a synthetic tone, the researchers recorded noise from an air conditioner’s outdoor unit and from a heat pump water heater — two of the most common sources of persistent low-frequency sound in residential settings — and then separated each recording into a low-frequency version at or below 100 hertz and a high-frequency version above it.

Mice were exposed for 12 hours a day over five days, timed to their natural nighttime activity period.

Comparing the two bands at matched sound pressure is what makes the result interpretable. If a given decibel level produced a given effect, both bands would do the same damage; any difference between them is a frequency effect rather than a loudness effect, and that is precisely the distinction that A-weighted regulation assumes away.

What happened to the kidneys

Full-spectrum noise and low-frequency noise both raised blood markers of kidney dysfunction — creatinine and urea nitrogen. High-frequency noise alone did not.

The tissue matched the blood chemistry. Exposed kidneys showed swollen glomeruli, the microscopic filtering structures, with thickened filtering membranes, and higher levels of endothelin-1.

Glomerular change is the right thing to find if the damage is real rather than incidental. The glomerulus is a tuft of capillaries whose job is pressure-driven filtration, and thickening of the filtration membrane with raised creatinine and urea is a coherent picture of filtration becoming less efficient — not a vague stress response but a specific structural change in the specific structure that does the work.

“The biggest surprise to me was that the low-frequency component caused kidney dysfunction even though it was below the hearing range of mice,” Kagawa said. “In contrast, the higher-frequency component did not cause kidney dysfunction, even when both had the same physical sound pressure level.”

Below the hearing range is the detail that makes this more than a stress study. If the mechanism were simply that noise is distressing, inaudible sound should be the least distressing kind. That the inaudible band did the damage points at something other than the auditory pathway — low-frequency pressure waves penetrate tissue rather than stopping at the ear, which is why infrasound and low-frequency noise have long been suspected of effects that loudness-weighted measures miss.

The pathway, and the drug that blocked it

The mechanistic evidence comes from an intervention. Giving mice ambrisentan, which blocks endothelin signalling, kept kidney markers closer to normal and reduced glomerular damage.

Endothelin-1 is among the most powerful vasoconstrictors the body produces, and the kidney is unusually sensitive to it: it constricts the vessels feeding and draining the glomerulus, which changes the pressure across the filtration membrane. Chronic endothelin activation is an established contributor to kidney disease, and ambrisentan is a licensed drug — used in pulmonary arterial hypertension — so this is a known pathway blocked by a known agent rather than a speculative chain.

A pharmacological block that prevents the damage is what converts a correlation into a mechanism. The noise raised endothelin-1 and injured glomeruli; blocking endothelin signalling prevented the injury; therefore the endothelin rise was doing the damage rather than accompanying it.

“Our findings highlight the importance of considering low-frequency noise, which has often been overlooked, in future human studies on environmental noise,” Kato said.

The qualifications, including one the authors raise themselves

The authors note that the results come from mice and that their relevance to humans is unknown, which is the governing limitation. A mouse kidney is smaller than a human one and a mouse body couples differently to a pressure wave of a given wavelength, so neither the dose nor the anatomy transfers directly.

No sound pressure level is reported, which is a significant gap for a study whose whole argument is about frequency at matched loudness. Whether the exposure corresponded to standing beside an outdoor unit or to hearing one through a wall determines whether this is a finding about a real residential exposure or about an intense laboratory one.

Nor are animal numbers given, and five days is a short exposure — informative about acute injury, silent about whether the kidney adapts, recovers or deteriorates over the months and years that a person actually lives next to a heat pump.

Kato adds a caution that cuts against simple interpretation: “the biological effects of low-frequency sound depend on its frequency, sound level, and duration of exposure.” The same group has previously reported beneficial effects of low-frequency sound, including reduced motion sickness and changes in skin blood flow — so their own position is that this is a dose- and context-dependent agent rather than uniformly harmful.

The disclosure

The competing-interests statement deserves to be quoted rather than summarised. The authors declare that “SHIZUKA Inc. provided the industry-academia collaborative research grant and soundproof boxes for this study,” adding that the support does not affect compliance with the journal’s data and materials sharing policies.

SHIZUKA Inc. is a Japanese soundproofing company. A finding that an overlooked category of noise causes organ damage is, straightforwardly, a finding favourable to a business that sells noise reduction — and the company supplied both funding and the soundproof enclosures the experiment depended on.

That is disclosed, which is how it should work, and the authors’ own framing is notably restrained: their stated conclusion is that human studies should include low-frequency noise, not that anyone should buy anything. A reader should hold the disclosure alongside the result rather than in place of it; the ambrisentan experiment is a mechanistic test whose outcome a funder cannot arrange.

The paper was published on 18 June 2026. Other funding came from the Japan Society for the Promotion of Science, JSPS KAKENHI, the Frontier Next-Generation Researcher Program, the Tokai Higher Education and Research System, MEXT and KEIRIN RACE.

---

Notes

1. “Glomerular Injury Induced by Daily Exposure to the Low-Frequency Component of Environmental Noise via Endothelin Signaling in Mice.” Environmental Science & Technology, 2026. DOI: 10.1021/acs.est.5c12555.
2. Nagoya University. “Environmental rumbles: A risk to kidney health?” EurekAlert!, 6 August 2026. https://www.eurekalert.org/news-releases/1139096.

Photo by Tom Fisk on Pexels.com