Scientists have mapped the first three-dimensional structure of a vault particle linked to bacterial life, uncovering a previously unexplored branch of these mysterious protein assemblies. The discovery could give researchers a simpler way to investigate structures found in human cells. Anthropic’s biology research team used Claude throughout the investigation, from identifying promising proteins to reconstructing their molecular architecture.Vault particles have puzzled biologists for nearly four decades. Despite their enormous size and widespread presence in complex organisms, researchers still cannot explain their biological purpose. The new findings reveal that related proteins can assemble into similar structures in bacteria and bacteriophages. This discovery could help scientists trace their evolutionary history and understand their role in living cells.Mystery across lifeLeonard Rome and Nancy Kedersha first identified vault particles in 1986. These hollow, barrel-shaped structures rank among the largest protein complexes found in human cells. Mammalian vaults contain 78 copies of a major protein arranged into two symmetrical halves.Their biological purpose remains uncertain despite decades of investigation. Researchers have proposed roles in cellular transport, drug resistance, and immune defense. However, experiments have yet to establish any of these functions conclusively.The structures also appear across much of the evolutionary tree. Their presence in organisms ranging from slime molds to humans suggests an important biological role. However, researchers have struggled to investigate them because common laboratory organisms lack vaults.We’re sharing how our biology research team worked with Claude to map the first 3D structure of a vault particle from bacterial life.Vault particles are a longstanding mystery. They are barrel-shaped objects, among the largest protein structures found in our cells. They were…— Anthropic (@AnthropicAI) November 4, 2026Anthropic’s team searched approximately 15 billion predicted protein sequences for distant relatives. The search identified potential vault proteins across 29 major bacterial groups and four groups of archaea. The researchers selected 44 candidates for laboratory testing.Inside the bacterial vaultExperiments revealed that seven of the 15 successfully produced proteins assembled into vault-like particles. Cryogenic electron microscopy, or cryo-EM, confirmed the structures of four candidates. Two came from bacteria, while the other two originated from bacteriophages.The researchers selected a phage-derived protein called L05 for detailed structural analysis. Bacteriophages are viruses that infect bacteria and use their cellular machinery. The L05 protein assembled into a complete structure containing 96 copies of itself.The resulting particle measures approximately 54 nanometers long and 45 nanometers wide. Its combined molecular mass reaches roughly 9.5 megadaltons. Despite its shorter length, the structure encloses approximately 25 percent more volume than mammalian vaults.Researchers also modified the protein to investigate how these structures grow. Adding two repeating segments produced an even larger particle. Its enclosed volume increased by approximately 20 percent.These experiments suggest protein duplication could have helped vaults evolve different sizes. They also revealed how small structural changes can prevent particles from closing properly.Claude accelerates discoveryClaude contributed throughout the research process, including protein searches, DNA design, purification, and microscopy analysis. One major breakthrough involved simplifying the particle isolation process.Traditional purification methods take approximately three days. The team developed a chromatography-based approach that reduced the process to roughly 90 minutes. Claude helped identify suitable equipment and optimize the experimental procedure.The researchers also used Claude to process cryo-EM data. Their first complete structural model emerged within 34 hours. The team estimates conventional analysis could require several weeks.However, the project exposed limitations alongside those gains. One experimental protocol introduced an additive that interfered with DNA delivery. That mistake delayed the research by two weeks. Claude also produced an inaccurate structural map after skipping a standard processing step.The researchers eventually corrected these problems through closer supervision. They also shared laboratory observations with Claude’s development team to improve future performance.Despite the structural breakthrough, the central mystery remains unsolved. Researchers have yet to determine what vault particles actually do inside living organisms. Leonard Rome, one of their original discoverers, described the question as enormously challenging. The new findings provide another route toward answering it, but much of the biology remains unexplored.