Tumours hijack nerve cells to send signals to the brain that disarm nearby immune cells. Plus, China has awarded its first ‘practical PhDs’ and how remote sensors are giving researchers new ways to study cities.
Hello Nature readers, would you like to get this Briefing in your inbox free every day? Sign up here.
Lung cancer cells (shown) in mice can connect with nearby neurons to send a ‘shutdown’ signal to the brain that suppresses tumour-killing immune cells.Credit: Steve Gschmeissner/SPL
Tumours lure and then hijack nearby sensory neurons to boost their own growth. The cancer cells use these neurons to send a signal to the brain that subdues the activity of immune cells around the tumour, which allows it to grow unchecked. When researchers deactivated these neurons in mice with lung cancer, they saw “a huge, dramatic reduction” in tumour growth — more than 50% — says cancer immunologist and study co-author Chengcheng Jin.
Nature | 5 min read
Reference: Nature paper
Snakes don’t have genes to make the ‘hunger hormone’ ghrelin, which could explain how they can go for months between meals. In most other vertebrates, ghrelin stimulates the appetite, and plays a role in breaking down fat to use as energy. Researchers found that other reptiles such as chameleons also lack ghrelin genes, which might help these animals preserve their fat stores for longer without needing to eat. Studying how these animals process food without ghrelin could help researchers understand how the hormone works in humans, says genomicist Todd Castoe.
Science | 3 min read
Reference: Royal Society Open Biology paper
An open-source artificial intelligence model called OpenScholar can outperform some major large language models (LLMs) at reviewing scientific literature, and gets the citations correct more often. OpenScholar combines an LLM with a database of 45 million open-access articles and links the information it sources directly back to the literature to stop the system from ‘hallucinating’ citations. The model is limited by the scope of its database, say the team who developed it. But running it is much cheaper than running a major model such as GPT-5, says computer scientist Hannaneh Hajishirzi.
... continue reading