ºÚÁϳԹÏÍø researchers have identified ways in which low-frequency noise in our environment coming from sources such as ACs and water pumps can trigger blood vessel damage to kidneys in mice.

Close your eyes and just listen for a few seconds. No matter where you are, countless low and deep rumblings are feeding into your ear, coming from AC units, elevator motors, distant traffic, and ventilation ducts.

These rumbles form low-frequency noise below 100 hertz, about the same pitch as a kick drum. Almost all its sources are human made. So they¡¯re a relatively new environmental effect; Living things weren¡¯t meant to be immersed in this kind of noise.

Previous studies have hinted that people who are surrounded with more environmental noise tend to have worse kidney function. But these studies could not point out which part of the noise was to blame, or the specific ways in which it might affect kidneys.

Now, a team led by Takumi Kagawa and Masashi Kato at the have provided the first direct experimental evidence that the low-frequency component of environmental noise is the culprit for adverse kidney health in mice. Their findings were published in the journal .

Pitch matters more than volume

Kagawa and Kato first started by recording real noise from two ordinary household machines: the outdoor unit of an air conditioner and a heat pump water heater. Using audio software, they then split each recording in two: a low-frequency version containing only sounds at or below 100 hertz, and a high-frequency version containing everything above it.

Mice were played these sounds for 12 hours a day over five days, during nighttime when they are naturally active. The researchers then measured two standard markers of kidney health in the animals’ blood, creatinine and urea nitrogen, which build up when the kidneys stop filtering waste efficiently.

Playing the unedited environmental noise symphony containing all frequencies pushed both markers up, signifying kidney damage. But when only the high-frequency noise was played, nothing happened. This meant the lower frequencies were responsible for pushing up the creatinine and urea nitrogen levels, as the researchers soon verified.

¡°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,¡± he added, implying that the frequency of noise mattered more than its loudness.

Squeezed blood vessels

To find out the modus operandi, the researchers looked inside the kidneys. Kidneys clean the blood using a vast network of glomeruli¡ªtiny knots of blood vessels that act as the organ’s filters. In the exposed mice, these filters were swollen and their delicate filtering membranes had thickened.

Additionally, the damaged kidneys were producing more endothelin-1, a molecule that can bind to receptors to clamp down blood vessels. Under normal conditions, endothelin-1 helps regulate blood flow. But researchers suspected that low-frequency noise can push it into overdrive, causing constricted renal blood vessels

To test this idea, the researchers gave a second group of mice ambrisentan, a drug already prescribed to human patients to treat hypertension, which blocks endothelin’s effects. The treated animals kept their kidney markers closer to normal and showed much less damage to their glomeruli, confirming endothelin signaling as a key pathway.

Because this study was conducted in mice, its relevance to humans remains unknown. However, ¡°our findings highlight the importance of considering low-frequency noise, which has often been overlooked, in future human studies on environmental noise,¡± Kato said.

A double-edged sword

The result is also intriguing because the same researchers have also found some beneficial effects of low-frequency sound, such as reducing motion sickness and blood flow in skin blood vessels.

This suggests that ¡°the biological effects of low-frequency sound depend on its frequency, sound level, and duration of exposure,¡± Kato said. ¡°Moving forward, we hope to investigate both the beneficial and adverse effects of low-frequency sound and scientifically clarify which exposure conditions are harmful and which promote health.¡±

Publication Information:

Takumi Kagawa, Dijie Chen, Nobutaka Ohgami, Keming Tong, Yanjun Gao, Naruhito Iwasaki, Akihito Harusato, Toyonori Tsuzuki, Takumi Hayashi, Yuuki ?Shimizu, Toyoaki Murohara, and Masashi Kato, 2026. Glomerular Injury Induced by Daily Exposure to the Low-Frequency Component of Environmental Noise via Endothelin Signaling in Mice, Environmental Science & Technology.?DOI:

Funding Information:

This work was supported partly by Grants-in-Aids for Scientific Research (B) (22K11731, 23H03147, 25K02875, and 23K27837), Challenging Research (Exploratory) (25K22731, 25K22732, 25K22741), Young Research (24K19748, 26K20515), and Fund for the Promotion of Joint International Research (22KK0145) from the Ministry of Education, Culture, Sports, Science and Technology (MEXT), Grant-in Aid for JSPS Research Fellow (22J22680 and 22KJ1602) from Japan Society for the Promotion of Science, Frontier Next-Generation Researcher Program of the Tokai Higher Education and Research System (JPMJSP2125) from JST SPRING and JKA and its promotion funds from KEIRIN RACE (2023M-407). The funders had no role in the study design, data collection and analysis, decision to publish, or manuscript preparation.

Expert contact:

Masashi Kato
Department of Occupational and Environmental Health,
ºÚÁϳԹÏÍø
Email: kato.masashi.r6@f.mail.nagoya-u.ac.jp

Media contact:

Sumeet Kulkarni 
International Communications Office 
ºÚÁϳԹÏÍø 
Email: icomm_research@t.mail.nagoya-u.ac.jp 

Top Image:

Visual representation of low-frequency environmental noise coming from climate control systems. Credit: Sumeet Kulkarni, ºÚÁϳԹÏÍø.

Researchers at ºÚÁϳԹÏÍø in Japan have created an innovative treatment for periodontitis that targets and eliminates harmful bacteria while preserving beneficial oral microbes essential for health. The findings were published in the

Oral health relies on a diverse and balanced microbial community. Porphyromonas gingivalis (P. gingivalis) is a keystone pathogen that triggers inflammation and disrupts this balance. Such disruption can damage the bone supporting teeth and eventually lead to tooth loss.

“Current treatments for periodontitis, including antibiotics and antimicrobial photodynamic therapy (aPDT), eliminate both pathogenic and beneficial oral bacteria,” said , one of the corresponding authors and a lecturer at ºÚÁϳԹÏÍø. “These approaches often disrupt the entire oral microbial ecosystem and can also release lipopolysaccharide (LPS), an endotoxin that may exacerbate inflammation.”

To address these challenges, researchers from ºÚÁϳԹÏÍø Graduate School of Medicine, including first author Hiroshi Maruyama, Professor , and corresponding authors Sato and , adapted near-infrared photoimmunotherapy, a technique originally developed for cancer treatment. This method uses an antibody-dye compound that binds to target cells and is activated by near-infrared light.

Expanding on this therapy, the team developed a method using IgY, an antibody from egg yolks of hens immunized against P. gingivalis. Because IgY can be produced in large quantities at low cost, it offers significant clinical potential. The researchers evaluated this new approach, called near-infrared photo-antibacterial targeting therapy (NIR-PAT?), in human cell cultures and mice to eliminate P. gingivalis while preserving a healthy oral environment, comparing its performance to conventional therapies.

Selective elimination of pathogens without damaging healthy tissue

In cell culture experiments, the antibody-dye compound selectively bound to P. gingivalis, leaving other bacteria and healthy human gum cells unaffected. When exposed to near-infrared light, the dye accumulated on P. gingivalis, disrupted its outer membrane, and eliminated the bacterium.

Microscopic analysis showed that treated P. gingivalis developed small membrane holes but retained its overall structure, while aPDT resulted in complete bacterial destruction. Tests confirmed that the new therapy did not harm human gum cells, whereas aPDT caused cellular injury and delayed healing.

Reduction of bone loss and improved oral microbiota in mouse models

In mouse models of periodontitis, the new treatment significantly reduced alveolar bone loss. Saliva analysis indicated that the therapy improved oral microbiota by eliminating pathogenic P. gingivalis while preserving beneficial Streptococcus populations. In comparison, standard antibiotics and aPDT removed both pathogenic and beneficial bacteria, further disrupting the oral microbiota.

“Results demonstrated that, unlike antibiotics or standard light therapy, this approach selectively removes the primary pathogenic species while preserving the remainder of the oral bacterial community,” Sato said.

To achieve clinical application

The researchers explain that periodontitis is caused by several types of bacteria, so focusing on just one may not be enough. They intend to use artificial intelligence (AI) to analyze publicly available oral bacteria data to identify other important species and better understand how they interact. These insights will help develop more precise treatments.

Because periodontitis is connected to health issues such as rheumatoid arthritis and diabetes, the researchers believe that their AI-driven analysis could also help determine which patients are most likely to benefit from targeted treatments.

Publication information

Hiroshi Maruyama, Kazuhide Sato, Kiyoshi Sakaia, Hirotoshi Yasui, Ryu Okada, Li Xinheng, Koji Umeda, Shofiqur Rahman, Van Sa Nguyenf, and Hideharu Hibi, 2026. Near infrared photo-bacterialflora modulation technology realized controlling periodontitis: Modulation of disease-associated dysbiosis in oral microbiota using near infrared photo-antibacterial targeting therapy (NIR-PAT?), Journal of Translational Medicine

Expert contact

Kazuhide Sato
ºÚÁϳԹÏÍø Graduate School of Medicine
Email: sato.kazuhide.k5@f.mail.nagoya-u.ac.jp

Media contact

Naomi Inoue
ºÚÁϳԹÏÍø International Communications Office
Email: icomm_research@t.mail.nagoya-u.ac.jp

Top image

NIR-PAT? uses an antibody linked to a light-sensitive dye to find and bind only to P. gingivalis, the key bacterium that drives gum disease, and destroys it when exposed to near-infrared light. This targeted attack removes the harmful bacterium while leaving good bacteria untouched, helping restore a healthy balance in the oral microbiome.
Credit: Kazuhide Sato et al., Journal of Translational Medicine, 2026. Volume 24, Journal of Translational Medicine, licensed under CC BY-NC-ND 4.0

What enables us to sustain effort toward goals, even when the task becomes increasingly difficult? A recent study by researchers at ºÚÁϳԹÏÍø in Japan revealed the underlying brain mechanism, showing that orexin neurons play a crucial role in driving and regulating motivated behavior. The findings were published in .

Motivational deficits, including loss of motivation, are often seen in mental disorders such as depression, addiction and ADHD. However, the brain mechanisms behind these problems remain largely unclear.

The research team, led by , associate professor, and Kiyofumi Yamada, professor emeritus, at ºÚÁϳԹÏÍø’s , focused on orexin neurons. These neurons regulate essential physiological functions such as sleep, appetite and energy expenditure. Although recent studies suggest that orexin neurons also influence motivation, their exact role has remained unclear.

This study used rats to examine how changes in orexin neuron activity influence motivation to obtain food rewards. While most previous studies used mice, rats offer superior learning abilities and are better suited for complex behavioral experiments. Because of technical challenges in targeting specific neurons in rats, research in this area has been limited.

To address this limitation, the team developed genetically modified “orexin-Cre” rats, allowing precise targeting and manipulation of orexin-producing neurons. This model was used to investigate how these neurons influence motivation.

First, using chemogenetics, the researchers activated the rats’ orexin neurons and had them perform a progressive ratio test where the number of touches required to earn a food reward increased with each trial. The point at which a rat gave up (the breakpoint) measured motivation intensity. Rats with activated orexin neurons showed higher breakpoints, meaning they worked harder for the reward. Conversely, in a model where orexin neurons were selectively degenerated, breakpoints were lower, indicating reduced motivation.

Next, using fiber photometry, the team recorded real-time activity of orexin neurons as the rats anticipated and received their reward. Activity increased before the reward was obtained, decreased once it was received, and remained elevated when an expected reward failed to arrive. Notably, the more effort required, the stronger the orexin neuron activity became. The researchers say this pattern may reflect how the brain links reward expectation to the effort required to pursue them.

To test this causally, the researchers used optogenetics to control orexin neuron activity at the moment a reward was anticipated. When orexin neuron activity was suppressed using an inhibitory protein, the rats’ motivated behavior decreased ¡ª they took longer to complete effort-based tasks, and their breakpoints dropped. By contrast, when the team attempted to boost orexin neuron activity at that moment using an excitatory protein, no further increase in motivated behavior was observed, even though the stimulation reliably activated the neurons.

In other words, suppressing orexin neurons impaired motivation, but artificially exciting them beyond natural levels did not enhance it. The researchers say this asymmetry suggests orexin neurons are necessary for sustaining motivated behavior, though simply raising their activity may not be sufficient to increase it. Further studies are needed to determine what governs this effect, such as the duration or pattern of orexin neuron activity.

Mizoguchi concluded, “Our study demonstrated significant changes in orexin neuron activity depending on expected rewards and the effort required, suggesting a potential mechanism for translating expectations into sustained action.”

Future research will investigate the input and output circuits connected to orexin neurons. A deeper understanding of orexin function may inform new approaches to addressing motivational deficits, including loss of motivation or challenges in sustaining goal-directed behavior.

Publication information

Yutao Dong, Sheikh Mizanur Rahaman, Wenjun Zhu, Ayumu Inutsuka, Daisuke Ono, Rinako Tanaka, Tetsuo Matsuzaki, Eiji Shibata, Madoka Isobe, Shuntaro Izawa, Akihiro Yamanaka, Kiyofumi Yamada, and Hiroyuki Mizoguchi, 2026. Reward prediction is encoded by orexin neuron activity during motivated behavior, Proceedings of the National Academy of Sciences of the United States of America,

Funding information

This work was supported by Grant-in-Aid for Scientific Research [22K19749; 23K27360; and 23H02669 (2023)]; SENSHIN Medical Research Foundation; Naito Foundation, Japan; Takeda Science Foundation, Japan; SRF, Japan; Asahi Glass Foundation, Japan; Mishima Kaiun Memorial Foundation, Japan; Kao Health Science Foundation, Japan; and AMED, Japan (JP21wm0425014).

Expert contact

Hiroyuki Mizoguchi
ºÚÁϳԹÏÍø Graduate School of Medicine
Email: mizoguchi.hiroyuki.d5@f.mail.nagoya-u.ac.jp

Media contact

Naomi Inoue
ºÚÁϳԹÏÍø International Communications Office
Email: icomm_research@t.mail.nagoya-u.ac.jp

Protein known for circulating in the blood to fight infection, plays a very different role when produced by cells inside tumors. Scientists link higher levels in tumor tissue to better treatment outcomes and survival in patients. 

An ancient molecule that existed long before the blood circulation system evolved, has been found to support cancer immunotherapy. Researchers at ºÚÁϳԹÏÍø in Japan found that complement C3 protein acts inside tumors to prevent the accumulation of immunosuppressive cells, but only when produced inside the tumor. C3 circulating in the blood did not affect treatment outcomes. Published in , the findings suggest that artificially recreating this effect can help patients with tumors that do not naturally produce enough of this protein.?
?
C3 protein is extremely old in evolutionary terms and present in simple organisms such as sponges and jellyfish. It is produced mainly in the liver and plays a vital role in the body¡¯s immune system, travelling through the blood to protect against infections. However, its role when produced locally in tissues and organs is largely unknown.?
?
¡°Cancer tumors are surrounded by normal cells called fibroblasts. Until now, the role of complement C3 produced by these cancer-associated fibroblasts within tumor tissue was not known¡± said lead author Yuki Miyai, assistant professor at the , ºÚÁϳԹÏÍø.?
?
The researchers discovered that C3 produced in tumor tissue prevents immunosuppressive myeloid cells from infiltrating the tumor microenvironment. The body¡¯s immune defenses then have a better chance of fighting cancer. The results identify C3 as a new factor that regulates the efficacy of cancer immunotherapy ¡ª cancer treatment that helps the immune system recognize and attack cancer cells.??
?
To test whether C3 in the blood was also responsible for immunotherapy effectiveness, the research team used mice to separate the roles of C3 according to its sources. When C3 produced by the liver was decreased by 90%, a drug that helps the immune system attack tumors (anti-PD-1 antibody) worked just as well as it did in mice with normal C3 levels.?
?
However, when C3 production by fibroblasts within the tumor was stopped, the same drug became less effective, even though circulating C3 in the blood changed very little (a 9% decrease).??
?
“What determined the efficacy of the immunotherapy treatment was not the C3 in the blood, but the local C3 produced at the tumor site. When this C3 breaks down, it forms a fragment called iC3b that stops harmful myeloid cells from entering the tumor. As a result, immunotherapy is more likely to work,¡± Miyai explained.?
?
To test if recreating the effect of C3 could help in cancers that do not respond to immunotherapy, the researchers tested a drug that mimics the blocking effect of C3 on myeloid cells. This allowed immunotherapy to work in tumors that had previously resisted treatment and significantly extended survival in mice. The results may help doctors predict which patients immunotherapy can help and offer new options for cancers that do not initially respond to it.?
?
Tumor samples from lung cancer patients were also analyzed. Patients with higher C3 levels in the tissue surrounding cancer cells had better treatment outcomes and survival rates. ºÚÁϳԹÏÍø half responded, compared to none of the patients with lower C3 levels. Again, C3 levels in the blood made no difference.

Next, the research team will conduct experiments to boost local C3 levels and identify the best timing for treatment. The authors believe that understanding how this protein works could shed light on other biological processes, such as how the body heals wounds and manages inflammation.?

Paper information: 

Yuki Miyai, Yukihiro Shiraki, Ryota Ando, Daisuke Sugiyama, Yoshitaka Sato, Katsuhiro Kato, Naoya Asai, Fuyang Cao, Nobuyoshi Nagao, Kana Tanabe, Masahiro Nakatochi, Tetsunari Hase, Toyofumi Fengshi Chen-Yoshikawa, Tomoko Kobayashi, Shintaro Iwama, Nobutoshi Esaki, Shinji Mii, Hiroshi Arima, Hiroshi Kimura, Masahide Takahashi, Yuichi Ando, and Atsushi Enomoto, 2026. Local, but not circulating, complement C3 shapes immune checkpoint blockade efficacy by controlling myeloid cell infiltration, Nature Communications. DOI: ?

Funding information: 

This study was supported by funds from the Ministry of Education, Culture, Sports, Science, and Technology of Japan (23K14591, 25K18851, 22H03350, 22H04923, 20H03528, 22H02848, 22K18390), JSMO/KFCR Young Investigator Research Grant, Aichi Cancer Research Foundation, Japan Agency for Medical Research and Development (JP24gm1210009, JP24ama221333), Naito Foundation, Princess Takamatsu Cancer Research Fund, DAIKO Foundation, and Toyoaki Foundation. 

Expert contact: 

Yuki Miyai 
Graduate School of Medicine 
ºÚÁϳԹÏÍø 
E-mail: miyai.yuki.i3@f.mail.nagoya-u.ac.jp 

Media contact:

Merle Naidoo
International Communications Office
ºÚÁϳԹÏÍø
Email: icomm_research@t.mail.nagoya-u.ac.jp 

Top image:

Tumors with fibroblasts that produce more C3 (right) contain fewer immune-suppressing macrophages (blue arrows) than tumors that produce less C3 (left). Higher C3 was linked to better outcomes with cancer immunotherapy. Credit: Miyai et al., Nature Communications 2026?
?

A brain enzyme thought to have a single function builds a sugar chain on itself, is secreted from the cell, and switches off ¡ª reactivating only when the sugar is removed

A chance discovery at ºÚÁϳԹÏÍø in Japan has shown a well-known brain enzyme has a hidden ability: it builds a sugar chain on itself, becomes secreted from the cell and deactivates, then switches on outside the cell once the chain is removed. The finding, published in the , overturns a decades-old assumption about how polysialic acid, a sugar chain critical for brain development and function, is produced and shows a new way an enzyme can regulate its own activity.

The brain¡¯s sugar chains

The human brain is covered in sugar chains, or glycans, molecular structures that coat cells and regulate how they communicate. One of the most important is polysialic acid, a long chain found mainly in the brain.

Polysialic acid keeps brain cells from adhering too tightly to each other and binds to growth factors and neurotrophins to regulate the presentation of their receptors, through which it plays a key role in learning, memory, and neural development. Importantly, these sugar chains change rapidly in response to brain activity. The ability to restore them quickly is thought to be essential for normal brain function.

Until now, scientists believed only two enzymes were responsible for building polysialic acid in the brain: ST8Sia2 and ST8Sia4.

A chance discovery

ST8Sia5 was discovered in 1996 and was known only as a builder of fatty brain molecules called gangliosides. It is expressed almost exclusively in the brain and its ability to produce polysialic acid was unknown until now.

The enzyme exists in three forms, short (S), medium (M), and long (L), that differ only in the length of one structural region. Only the long form, ST8Sia5L, showed this newly discovered activity. Unlike the short and medium forms, ST8Sia5L localizes to a different intracellular compartment, which may allow it to undergo autopolysialylation. The function of the short and medium forms is not yet known.

ºÚÁϳԹÏÍø¡¯s had been testing all six members of the ST8Sia enzyme family.

¡°We found that a third enzyme, ST8Sia5, also builds polysialic acid, but only on itself, and only in its longest form, ST8Sia5L,¡± said first author Fumiya Sakamoto.

Coauthor and Director of iGCORE Professor Chihiro Sato commented: ¡°We were checking each enzyme one by one and found this activity by chance.¡±

Four discoveries that define this mechanism

1. The enzyme builds its own off switch. Unlike most enzyme regulation, where a separate molecule switches an enzyme on or off, ST8Sia5L modifies itself. It builds polysialic acid chains directly onto its own structure, a process called autopolysialylation. No external regulator is required.
2. The sugar chain is the switch. Polysialic acid is not typically known as a regulator of enzyme activity, but here it acts as one. While the chain is attached, the enzyme¡¯s ganglioside-building function is completely suppressed. This is a new role for polysialic acid.
3. Self-modification is linked to secretion. Once coated in polysialic acid, the enzyme is cut free from the cell membrane by metalloprotease enzymes and released into the fluid outside the cell. The sugar coat does not just silence the enzyme; it is also associated with its release from the cell.
4. The enzyme reactivates outside the cell. The researchers showed experimentally that the secreted enzyme, collected from outside the cell, regains its ganglioside-building activity once the polysialic acid chains are removed. This could happen, for example, when sialidase enzymes are released during stress or inflammation. Reactivation does not require the enzyme to re-enter the cell.

A surprise finding for other enzymes too

The ST8Sia family are all sialic acid-building enzymes, but they differ in how long a chain they build. Most add just two or three units. ST8Sia2 and ST8Sia4 were the only ones known to build long chains of polysialic acid. ST8Sia5L has now joined that group, but with one key difference: it only builds the long chain on itself, not on other molecules.

The study also found, for the first time, that ST8Sia2 and ST8Sia4 are also secreted from cells in a polysialic acid-coated form. What this means for those enzymes is not yet known.

Broader implications

One of the most significant conceptual implications of the study is where sugar modification can happen in the body.

¡°It¡¯s been assumed that the process of adding sugar chains to molecules, called glycosylation, takes place inside the cell,¡± said Professor Sato. ¡°This study provides evidence that modification can also happen outside the cell.¡±

The research team hypothesizes that after release, the enzyme may travel to specific sites on cell surfaces and rapidly repair damaged ganglioside structures, without needing to re-enter the cell first. The conventional pathway for ganglioside repair requires the molecule to travel back inside the cell for modification. This proposed ¡°on-site recovery¡± mechanism, if confirmed, would represent a much faster alternative. The hypothesis is currently being investigated.

ST8Sia5L may also play a role in regulating microglia, the brain¡¯s immune cells. The researchers hypothesize that the polysialic acid coat on the secreted enzyme may interact with inhibitory receptor molecules called Siglecs on microglia, helping to keep immune activation in check under normal conditions.

During inflammation or stress, sialidase enzymes could remove this coat, allowing immune responses to proceed and freeing the enzyme at the same time to resume its ganglioside-building activity at cell surfaces.

¡°Polysialic acid abnormalities have also been associated with schizophrenia, but the mechanism behind this link is not yet understood,¡± said Ken Kitajima, coauthor and professor at iGCORE. ¡°The secreted polysialylated enzyme is one candidate for further investigation in this context.¡±

To test these hypotheses in a living system, the team is currently generating mice in which the ST8Sia5 gene has been disabled. The researchers also intend to investigate the unknown function of ST8Sia5S and ST8Sia5M, which both localize to a different compartment within the cell.

Paper information:

Fumiya Sakamoto, Rina Hatanaka, Masaya Hane, Di Wu, Ken Kitajima, Chihiro Sato, 2026. A novel autopolysialylation activity of the ganglioside sialyltransferase ST8Sia5 regulates its secretion and enzyme activity, Journal of Biological Chemistry, 302(7). DOI:

Funding information:

This research was funded by the Japan Agency for Medical Research and Development (AMED) (18ae0101069h0003, 19ae0101069h0004, 20ae0101069h0005, 20gm6410007h0001, 21gm6410007h0002, 22gm6410007h0003, and 23gm6410007h0004 ) and a Grant-in-Aid for Scientific Research from JSPS (Grant numbers 23K21291 and 25K02224). A part of this research was also funded by the CIBoG program, ºÚÁϳԹÏÍø.

Expert contact:

Chihiro Sato
Institute for Glyco-core Research (iGCORE)
ºÚÁϳԹÏÍø
Email: chi@agr.nagoya-u.ac.jp

Media contact:

Merle Naidoo
International Communications Office
ºÚÁϳԹÏÍø
Email: icomm_research@t.mail.nagoya-u.ac.jp

Top image:

The three enzymes shown here build polysialic acid (orange), a long sugar chain important for brain development and function. ST8Sia5L (left) builds the chain only on itself, a newly discovered activity. ST8Sia2 (center) and ST8Sia4 (right) build it on other molecules. Credit: Sakamoto et al., 2026

DNA is composed of long chains that act as the blueprint for living organisms. In genetic engineering, scientists cut DNA at specific sites and join the resulting fragments to other DNA sequences, enabling applications such as advanced crop breeding, genetic disease treatment, and the generation of animal models for drug discovery.

Assembling short DNA fragments requires overhanging sequences, known as sticky ends, to facilitate efficient binding. However, generating sticky ends requires precise cutting at targeted sites, which remains challenging with current technologies.

A Japanese research group has developed a silver nanoparticle-based technology to precisely cut and join DNA at targeted sites, achieving two to five times higher DNA assembly efficiency than conventional restriction enzyme methods. These findings were published in the journal .

Traditional long-chain DNA assembly uses restriction enzymes to cut DNA and T4 DNA ligase to reconnect the fragments. However,  restriction enzymes cut only at specific sequences and generate sticky ends that are often too short, thereby limiting joining efficiency.

To address this limitation, a research team led by Professor and Assistant Professor at ºÚÁϳԹÏÍø, in collaboration with Professor Natsuhisa Oka at , studied DNA cleavage at targeted sites using chemical reactions instead of restriction enzymes.

The researchers examined a reaction reported between 1990 and 1992, in which silver ions cleave 3′-thiol-modified DNA at specific sites. They assessed its potential to generate suitable sticky ends. Results showed that although silver ions efficiently cleave DNA, they also bind nonspecifically, leading to precipitation. This resulted in a low DNA recovery rate of about 14%, which is insufficient for practical use.

The team then employed silver nanoparticles instead, hypothesizing that these could be removed after the reaction through centrifugation, thereby potentially increasing DNA recovery.

Experiments showed that DNA-cleaving efficiency reached about 50% at 70¡ãC and nearly 100% at 95¡ãC within two hours. However, these high temperatures pose a risk of damaging long-chain DNA.

To address this, the team coated the nanoparticles with polyethylene glycol (PEG), a water-soluble polymer, to enhance stability and dispersion. This modification increased cleaving efficiency from 36% without PEG to 92% with PEG at 37¡ãC over 31 hours. “In the end, we optimized the conditions to a practical level and, under ambient temperatures, achieved PEG-modified cleaving efficiency above 91% at 50¡ãC within just one to two hours,” stated Inagaki, the study’s first author.

An additional benefit of this process was the removal of unwanted DNA fragments bound to nanoparticle surfaces, leaving only the desired fragments with sticky ends in solution. This purification process increased the final DNA recovery rate from 14% to 98%.

The use of silver nanoparticles also enabled the generation of DNA fragments with 8-base sticky ends, a process that is challenging with conventional restriction enzymes. By employing T4 DNA ligase to join these fragments, the team achieved about double the joining efficiency of traditional methods. With an 18-base overhang, joining efficiency reached 44%, compared to only 8% with a conventional 4-base overhang, representing a fivefold improvement.

To evaluate the practical application of this approach, the researchers assembled a DNA fragment encoding green fluorescent protein (GFP) and introduced it into human HeLa cells. They successfully confirmed GFP expression, indicating accurate assembly.

Inagaki commented, “We believe this technology will be useful for synthesizing genomic DNA, with many possible applications in areas such as mRNA library establishment for cancer vaccines and gene therapy, as well as the development of artificial protein drugs and genome crops.”

He also explained the next step: “We have shown that two DNA fragments can be joined. Now, we need to confirm whether multiple fragments can be joined at the same time¡ªa key step for building genome-scale DNA.”

Paper information

Masahito Inagaki, Mikiya Kase, Haruka Hiraoka, Natsuhisa Oka, Fumitaka Hashiya, Naoko Abe, Yasuaki Kimura, Hiroshi Abe (2026). Silver Nanoparticle Induced Site-Specific Strand Cleavage of Chemically Modified Oligonucleotides for Long-Chain DNA Assembly, Nucleic Acids Research,

Funding information

This work was supported by the Japan Science and Technology Agency (JST) (JPMJCR18S1, JPMJCR23N1, JP25H00427, JP24H00737, JP22H02219, JP22K21346 International Leading Research), and Japan Agency for Medical Research and Development (AMED) [JP22gm0010008 (LEAP), JP25ak0101289, JP223fa827 (SCADA), JP243fa827032 (SCADA), JP23bm1223009, JP24ek0109697, JP25ama221315, JP25km0405209, JP25ama221230; JP23fk0210133) and Tanaka Kikinzoku Memorial Foundation [Precious Metals Research Grants 2021 Silver Award to M.I.].Funding to pay the Open Access publication charges for this article was provided by the Japan Science and Technology Agency.

Expert contact:

Hiroshi Abe
ºÚÁϳԹÏÍø Graduate School of Science
Email: h-abe@chem.nagoya-u.ac.jp

Masahito Inagaki
ºÚÁϳԹÏÍø Graduate School of Science
Email: inagaki.masahito.e6@f.mail.nagoya-u.ac.jp

Media contact:

Naomi Inoue
ºÚÁϳԹÏÍø International Communications Office
Email: icomm_research@t.mail.nagoya-u.ac.jp

ºÚÁϳԹÏÍø scientists develop a machine learning tool to uncover how ALS progresses in patients and offer clues to understand why some patients decline faster than others 

ALS (Amyotrophic Lateral Sclerosis) is a fatal neurodegenerative disease that gradually affects a person’s ability to move, speak, and breathe. It advances differently in every patient. Now, researchers at ºÚÁϳԹÏÍø have developed an AI tool that uses data from patient follow-up studies to estimate the speed of disease progression and identify patterns of muscle decline. The study was published in .??

Two questions, one tool 

ALS patients differ in two main ways: how fast their disease advances, and the order in which functions become impaired. Until now, existing AI-based research tools generally did not clearly separate these two differences in individuals. The research team developed DiSPAH, a machine learning system that addresses both at once by analyzing patient data collected during routine medical visits.??
?
Two datasets of patients with limb-onset ALS were used, a form of the disease where symptoms begin in the arms or legs rather than in the muscles controlling speech and swallowing (bulbar-onset ALS). The first provided data from 264 ALS patients and was used to train the model. The second, larger dataset of 2,565 patients was used to validate the results.

What the AI found 

The system identified six distinct patterns of disease progression among the patients. Some patients showed slow decline in motor function, with little effect on speech or breathing, while others experienced rapid deterioration.??
?
¡°Subtle differences between patients also emerged. For example, in some patients gross motor functions such as walking declined before fine motor skills such as writing or buttoning a shirt, while in others the opposite was true,¡± said Yuichiro Yada, coauthor and associate professor at ºÚÁϳԹÏÍø¡¯s . ¡°These six patterns were identified in one patient dataset and largely reproduced in a second, larger dataset, suggesting that they? capture common progression patterns in limb-onset ALS.¡±??
?
Importantly, speed and decline pattern were found to be independent of each other. A patient could follow a severe pattern at a slow speed, or a milder one at a fast speed. Previous tools could not measure both dimensions at once.

Prediction from day one 

One of the most important findings was that DiSPAH could, to some extent, predict a patient¡¯s progression speed and broad progression pattern from information available at the first medical visit. This consisted of basic functional assessments and the presence of certain gene mutations. 
 
These early predictions have important potential for patient care. Doctors could use them to plan treatment, prepare patients and families, and design better clinical trials by grouping participants according to how their disease advances. 

A genetic clue

The researchers also found that patients with a mutation in a gene called C9orf72 had faster disease progression. When they analyzed data from motor neurons grown in the laboratory from patients’ own stem cells, the results showed that faster ALS progression may be linked to problems in how cells produce and manage proteins, as well as signs of cellular stress. 
 
This points to a possible biological explanation for why some patients decline faster than others and gives scientists a new target for future research into ALS treatments.

Better tools for patients 

ALS currently has no cure. While a few drugs exist, they offer modest benefit. Better tools for prediction and monitoring are essential for the development of new therapies.?
?
Yada noted that DiSPAH is a prototype that needs further validation and refinement: ¡°It¡¯s a promising first step and better than anything that existed before for this specific purpose, but it¡¯s not reliable enough yet to use to make decisions about individual patients.¡±?
?
The researchers aim to extend the tool to all ALS patient types, improve its reliability, and ultimately apply it to other chronic diseases such as Alzheimer’s and Parkinson’s disease.??

Paper information: 

Yuichiro Yada and Naoki Honda, 2026. Decomposing heterogeneity in disease progression speeds and pathways, npj Digital Medicine. DOI:

Funding information:  

This work was partly supported by JSPS Grant-in-Aid for Early-Career Scientists (JP23K16994), Japan Agency for Medical Research and Development (AMED) Multidisciplinary Frontier Brain and Neuroscience Discoveries (Brain/MINDS 2.0)(JP24wm0625416 and JP25wm0625322), JST Moonshot R&D¨CMILLENNIA Program (JPMJMS2024) and JST CREST (JPMJCR25Q2). 

Expert contact: 

Yuichiro Yada?
Research Institute of Environmental Medicine?
ºÚÁϳԹÏÍø?
yada.yuichiro.k4@f.mail.nagoya-u.ac.jp?

Media contact:?

Merle Naidoo   
International Communications Office   
ºÚÁϳԹÏÍø   
Email: icomm_research@t.mail.nagoya-u.ac.jp

Top image:

DiSPAH is an AI tool that uses data from patient follow-up studies to estimate the speed of disease progression and identify patterns of muscle decline. Credit: Kano Okada, ºÚÁϳԹÏÍø 

A research team in Japan has developed an efficient and minimally invasive cancer detection device that uses high-performance zinc oxide nanowires to selectively capture extracellular vesicles (EVs) from bodily fluids.

Using this device, researchers successfully captured cancer-related EVs from the blood serum of ovarian cancer patients. The EVs’ surface membrane proteins and microRNAs remained intact, indicating the potential for sensitive disease analysis. These findings were published in the journal .

A liquid biopsy is a procedure that collects disease-related information from bodily fluids, such as blood and urine. Unlike traditional tissue biopsies, it places less physical burden on patients.

EVs are nanoscale vesicles that carry diverse molecular contents such as microRNA and messenger RNA, and display membrane proteins that indicate their cell of origin. EVs reflect disease states and serve as promising diagnostic indicators for liquid biopsy.

Accurate and efficient isolation of EVs from complex biological fluids is essential for identifying disease-associated molecules, but conventional techniques are time-consuming, require large sample volumes, and lack specificity.

A team led by , a professor at ºÚÁϳԹÏÍø’s , previously achieved efficient EV capture using zinc oxide nanowires they developed.

They are now collaborating with Yasuhide Inokuma, a professor at Hokkaido University, and researchers from the Institute of Science Tokyo, Kyoto University, and the National Institutes for Quantum Science and Technology to develop antibody-conjugated nanowire technology for the selective capture of cancer-derived EVs.

The initial challenge was attaching antibodies to nanowires. Conventional adhesives bind both target and non-specific proteins and require lengthy attachment times.

The team used the synthetic polymer polyketone to create six N-hydroxysuccinimide-functionalized polyketone (pKNHS) variants with different chain lengths. Of these, pKNHS 4.2 showed optimal stability for adsorption onto zinc oxide nanowires and effective antibody immobilization, enabling single-step antibody modification.

Evaluation of the new technology in cultured cell experiments

Researchers evaluated the capture efficiency of antibody-conjugated nanowires for cultured breast cancer cells using pKNHS 4.2. While antibody-free nanowires captured about 65% of CD9-positive EVs, CD9 antibody-conjugated nanowires achieved 90% efficiency. These results demonstrate the technology¡¯s effectiveness in selectively recovering target molecules.

Further experiments showed that nanowires modified with antibodies for ovarian cancer markers CLDN3, FOLR1, and TROP2 enabled the selective recovery of EVs from ovarian cancer cells.

Analysis of serum from cancer and non-cancer patients

Researchers isolated EVs using CLDN3, FOLR1, and TROP2 antibody-modified nanowires from the serum of six patients with high-grade serous ovarian carcinoma, an aggressive ovarian cancer subtype, and six non-cancer individuals. Analysis of microRNAs in EVs revealed distinct profiles between the patient and non-cancer groups.

When comparing microRNAs in EVs captured with the three antibodies, researchers identified 126 microRNAs common to all, indicating signals shared by ovarian cancer. They also found microRNAs unique to each antibody: 40 for CLDN3, 37 for FOLR1, and 45 for TROP2. These findings suggest that EVs with different membrane proteins have distinct microRNA profiles.

Significance and future perspectives

“In this study, we developed a nanowire microfluidic device capable of selectively capturing cancer-associated EVs with high efficiency, while suppressing nonspecific adsorption through simple chemical modification,” said Yasui, a corresponding author of the study. “We also demonstrated that this approach maintains both EV membrane proteins and internal microRNAs intact, showing strong potential for highly sensitive analysis of cancer states.”

, an assistant professor and corresponding author, said: “We plan to compare and evaluate this technology against existing clinical methods and expand its application to capture more specific EV subpopulations. In the long run, we aim to apply this technology to non-invasive liquid biopsies and early diagnosis across a variety of cancer types.”

Paper information

Kunanon Chattrairat, Akira Yokoi, Yumehiro Manabe, Yuki Ide, Jiahui Shen, Takeshi Hasegawa, Mikiko Iida, Taiga Ajiri, Zetao Zhu, Ryosuke Uekusa, Masami Kitagawa, Yoshinobu Baba, Hiroaki Kajiyama, Yasuhide Inokuma, and Takao Yasui, 2026. Discrete polyketones enable antibody click conjugation for selective exosome profiling. Device, 101153.
DOI:

Funding and other support

This work was supported by the Japan Science and Technology Agency CREST (JPMJCR2576), JST FOREST (JPMJFR211H and JPMJFR204J), the New Energy and Industrial Technology Development Organization (JPNP20004), the JSPS Grant-in-Aid for Scientific Research (A) (24H00792), the JSPS Grant-in-Aid for Scientific Research (B) (24K02586), the Moonshot Research and Development Program (22zf0127004s0902 and JP22zf0127009) from the Japan Agency for Medical Research and Development, the Asahi Glass Foundation Continuation Grants for Outstanding Projects, the Noguchi Institute NJ202308, the Cooperative Research Program of the ¡°Network Joint Research Center for Materials and Devices,” and the World Premier International Research Initiative, MEXT, Japan ¡ªInstitute for Chemical Reaction Design and Discovery (facility use)

Expert contact:

Kunanon Chattrairat
ºÚÁϳԹÏÍø Graduate School of Engineering
Email: kunanon.chat@chembio.nagoya-u.ac.jp

Takao Yasui
ºÚÁϳԹÏÍø Graduate School of Engineering
Email: yasui@chembio.nagoya-u.ac.jp

Media contact:

Naomi Inoue
ºÚÁϳԹÏÍø International Communications Office
Email: icomm_research@t.mail.nagoya-u.ac.jp

Top image:

Scanning electron microscope (SEM) image of zinc oxide nanowires
Credit: Kunanon Chattrairat (Yasui Lab., ºÚÁϳԹÏÍø)

Researchers have identified the neural circuit through which the brain¡¯s circadian clock controls the timing of torpor, a natural state of reduced body temperature and metabolism. The discovery provides new insights into how mammals regulate energy use and may inform future approaches in medicine and long-duration spaceflight. ?

 
You have gone without food for days, and the temperature drops to near freezing. What do you do? For some animals, the answer is influenced by the brain¡¯s circadian clock. Hummingbirds, bats, and mice are among the animals that can enter torpor, which reduces body temperature and metabolism. Scientists suspected that the brain¡¯s circadian clock controls the timing of torpor, but until now the exact mechanism was not known.  
 
Researchers at ºÚÁϳԹÏÍø in Japan have identified the specific neural circuit responsible for this survival strategy. They have shown that the brain¡¯s circadian clock, a small cluster of neurons located in the hypothalamus at the base of the brain, sends silencing signals through this circuit to a nearby temperature-regulating region, suppressing torpor during the day. The findings were published in .

Torpor from midnight to dawn 

“The brain’s preoptic area (POA) controls body temperature and has an important role in initiating torpor,¡± said senior author and lecturer Daisuke Ono from the at ºÚÁϳԹÏÍø. ¡°During the day, the brain¡¯s circadian clock suppresses torpor, which occurs between midnight and dawn in mice.¡± 
 
Using light-based tools (optogenetics) to switch specific neurons on or off, the researchers showed that activating the circadian clock-POA pathway suppressed torpor. When the circadian clock was disrupted, mice either entered torpor at irregular, unpredictable times or showed a marked reduction in torpor. ?  

Additionally, the specific clock cell type responsible for sending these signals was identified. Neurons that produce a protein called arginine vasopressin (AVP neurons) in the circadian clock inhibit neurons in the POA. Mice with impaired inhibitory signaling from AVP neurons to the POA showed abnormal torpor timing, demonstrating that this pathway plays a key role in determining when torpor occurs. ? 

The research team also discovered that the POA becomes more active at night. ¡°The clock does not actively trigger torpor. Instead, it reduces its inhibitory influence at night, allowing neural circuits involved in thermoregulation and energy balance to promote torpor when environmental conditions are favorable. The three systems work in tandem to create the right conditions,¡± Ono explained. 

Implications for medicine and space travel 

A clearer understanding of how the brain times metabolic shutdown may inform a technique that uses controlled cooling to limit tissue damage after injury or surgery (induced hypothermia). The findings may also be relevant to extended spaceflight, where controlled reduction of metabolism could protect the body. ? 

Although humans do not naturally enter torpor, understanding the neural mechanisms that regulate metabolic suppression in mammals could provide clues for developing controlled hypometabolic states in the future.  

Rare accounts of people surviving extreme cold exposure with dangerously low body temperatures hint at this possibility. Understanding the brain circuits that control these states in mammals may one day bring researchers closer to inducing suspended animation in humans, a state long imagined for deep space travel. 

Paper information 

Sheikh Mizanur Rahaman, Shota Miyazaki, Chang-Ting Tsai, Akihiro Yamanaka, Chi Jung Hung, Michihiro Mieda, Takahiro J. Nakamura, Hiroshi Yamaguchi, and Daisuke Ono. 2026. GABAergic projections from the suprachiasmatic nucleus to the preoptic area regulate the timing of torpor in mice, Nature Communications. DOI: ?

Funding information: 

This work was supported by the HIROSE foundation, LOTTE Foundation, Foundation of Kinoshita Memorial Enterprise, Astellas Foundation for Research on Metabolic Disorders, UBE Foundation, JST FOREST Program (JPMJFR211A), and JSPS KAKENHI (25H02445, 24K02060, 24H02006, 23H04939, 21H02526, 25KF0138, 21H00422, 24KJ0102 and 25K18507). 

Expert contact:

Daisuke Ono 
Research Institute of Environmental Medicine 
ºÚÁϳԹÏÍø 
dai-ono@riem.nagoya-u.ac.jp 

Media contact: 

Merle Naidoo   
International Communications Office   
ºÚÁϳԹÏÍø   
Email: icomm_research@t.mail.nagoya-u.ac.jp

Top image:

When facing freezing temperatures and food deprivation, mice enter a state of low metabolism known as ¡°torpor¡± from midnight until dawn. Researchers at ºÚÁϳԹÏÍø have now identified the specific brain circuit that controls this timing, running from the brain’s biological clock to its temperature-regulating region. Credit: Daisuke Ono, ºÚÁϳԹÏÍø 

A gene-encoded blueprint tells growing neurons which brain regions to connect with ? 

How complex neural circuits are genetically designed and wired is a fundamental question in neuroscience. Scientists have shown for the first time that genes encode a ¡°wiring map¡± that guides neurons to connect with the correct brain regions. The findings, based on machine learning analysis of mouse brain data, were published in , and offer new avenues for research into brain development and disease.

Mapping connections between brain regions with data

The research team, led by scientists from ºÚÁϳԹÏÍø in Japan, aimed to understand the wiring rules that guide nerve fibers during brain development. These long, thin fibers, called axons, extend from neurons and send signals to other neurons.?
?
The researchers developed an analysis method called SPERRFY that combines two datasets. One dataset maps which brain regions are connected to each other, and the other tracks the activity levels of 763 genes in all 213 brain regions in mice.?
?
¡°Some genes are highly active in certain brain regions and less active in others. These differences create distinct patterns of gene activity throughout the brain,¡± said Naoki Honda, senior author and professor from ºÚÁϳԹÏÍø’s . ¡°When hundreds of patterns overlap, they give each brain region a unique molecular identity. These identities are what SPERRFY was designed to decode.¡±?
?
By feeding both datasets into a machine learning algorithm, SPERRFY identified these patterns of gene activity, called gene expression gradients, that predict which brain regions are likely to connect. For each pair of connected brain regions, SPERRFY paired the gene activity profile of the source region (where the nerve fiber originates) with the profile of the target region it connects to.???
?
From these gene expression gradients, the researchers produced a brain wiring map that tells each brain region where it is relative to every other region. Overlapping patterns of gene activity reconstructed the brain’s connection patterns with a prediction-performance score of 0.88 on a 0-to-1 scale, where 1.0 indicates perfect prediction. By comparison, predictions based only on the physical distance between brain regions scored about 0.70.

Additionally, the researchers discovered that the brain’s wiring map operates on two levels. Broad gene activity patterns determine the overall organization between brain regions, while more detailed patterns regulate the specific connections within them.

Testing a 60-year-old theory on the whole brain 

The findings build on the chemoaffinity theory proposed by Nobel laureate Roger Sperry in 1963. He suggested that neurons find their connection partners by following molecular concentration gradients ¡ª chemical signals that vary in strength throughout the brain. These gradients act like a GPS system for growing nerve fibers.??
??
¡°The chemoaffinity theory was well established for simple circuits such as the visual and olfactory systems. But until now, the complexity of whole-brain connectivity made it difficult to test whether the same principle operates across the brain,¡± said Jigen Koike, first author and former PhD student at Hiroshima University, who also conducted research as a special research student at ºÚÁϳԹÏÍø’s Graduate School of Medicine.?
?
This complexity made it extremely difficult to test Sperry¡¯s theory across the entire brain without computational tools. Using machine learning, the researchers developed the tools to do this for the first time. Their findings support the idea that this long-standing principle is not limited to simple sensory circuits, but also helps explain how connections are organized across the whole brain.??

Future research 

By comparing the activity of 763 genes against the wiring map, SPERRFY also identified specific genes with activity patterns that closely matched, including genes known to guide nerve growth. This supports the validity of the method and provides a starting point for research on the molecular mechanisms of brain wiring. 
 
The researchers note that their method can be applied to any species for which maps of the brain¡¯s neural circuits and gene expression data are available, such as humans, marmosets, and fruit flies. As these datasets expand, the method could help determine if the same molecular wiring principles are shared across species and how they have evolved. SPERRFY could also assist scientists in understanding how disruptions in brain wiring contribute to neurodevelopmental disorders.   

Paper information:

Jigen Koike, Ken Nakae, Riichiro Hira, Yuichiro Yada, Naoki Honda, 2026. A data-driven framework linking the connectome to spatial gene expression gradients inspired by chemoaffinity theory. Proceedings of the National Academy of Sciences, 123(10). DOI:

Funding information: 

This work was supported by JST, the establishment of university fellowships toward the creation of science technology innovation (grant number JPMJFS2129), JST SPRING (grant number JPMJSP2132), JSPS KAKENHI (grant number JP22H05163), Moonshot R&D¨CMILLENNIA Program (grant number JPMJMS2024-9), Agency for Medical Research and Development (AMED) Multidisciplinary Frontier Brain and Neuroscience Discoveries (Brain/MINDS 2.0) (grant number JP25wm0625322 and JP25wm0625210), and the Cooperative Study Program of Exploratory Research Center on Life and Living Systems (ExCELLS: program number 19¨C102).  

Expert contact: 

Honda Naoki   
Graduate School of Medicine   
ºÚÁϳԹÏÍø   
Email: honda.naoki.t1@f.mail.nagoya-u.ac.jp 

Media contact:  

Merle Naidoo   
International Communications Office   
ºÚÁϳԹÏÍø   
Email: icomm_research@t.mail.nagoya-u.ac.jp  

Top image:

A 3D visualization of the 13 major regions in the mouse brain. Black dots mark the centers of the 213 subdivisions used by SPERRFY to analyze relationships between brain connectivity and gene activity patterns. Credit: Koike et al., PNAS, 2026. CC BY 4.0