A garage-built Alzheimer drug shows how far DIY biotech has come

Why it matters

PAC-832 is a founder-led attempt to revive a neglected Alzheimer mechanism with small-molecule chemistry, but the evidence is still preclinical and company-generated.

Douglas Yao (@DouglasYaoDY) published preclinical data on June 27 for PAC-832, an experimental Alzheimer disease drug candidate from Pace Pharmaceuticals that he says he designed and synthesized in a chemistry lab built in his garage. The claim is less that a garage has solved Alzheimer disease than that drug-discovery tooling has advanced to the point where a founder can push a novel CNS molecule from computational design to animal data largely outside the traditional pharmaceutical infrastructure.

Yao, who says on his personal site that he received a Ph.D. in computational biology from Harvard in 2023 and a B.S. in molecular biology from UCLA in 2018, is framing PAC-832 as a return to an older Alzheimer strategy: boosting cholinergic signaling, rather than chasing amyloid plaques alone. In a 16-post thread on X, Yao called PAC-832 the first selective GalR1 antagonist and said Pace Pharmaceuticals is moving the compound through IND-enabling studies.

The company has not reported human data. The evidence Yao released is a technical report hosted on Pace Pharmaceuticals' own site, not a peer-reviewed journal article. That matters because the claim sits in one of the hardest categories in drug development: central nervous system disease, where mouse cognition data has repeatedly failed to translate into durable clinical benefit.

The specific bet is narrower than most Alzheimer pitches. PAC-832 is designed to selectively block galanin receptor 1, or GalR1, a G-protein coupled receptor involved in galanin signaling. The company says the compound antagonized GalR1 with an IC50 of 0.28 uM and showed more than 30-fold selectivity over GalR2 and GalR3 in functional assays described in its PAC-832 technical report. The distinction is the whole strategy: GalR1 signaling has been tied to suppression of acetylcholine release, while GalR2 signaling has been associated in the literature with neuroprotective effects. A nonselective galanin blocker could interfere with both.

Pace Pharmaceuticals' report says PAC-832 reversed galanin-mediated suppression of acetylcholine release in differentiated SH-SY5Y cells and improved performance in two mouse memory tasks after scopolamine challenge. In the Y-maze test, the report says PAC-832 significantly increased alternation rates at 3 mg/kg, 10 mg/kg and 30 mg/kg, while the 1 mg/kg dose did not. In the novel object recognition test, the report says the 10 mg/kg and 30 mg/kg doses produced significant increases in discrimination index. The company compared the effect sizes with donepezil, the acetylcholinesterase inhibitor used as a positive control.

That comparison is useful but limited. Donepezil is an FDA-approved symptomatic treatment for dementia of the Alzheimer type, and cholinesterase inhibitors remain part of standard Alzheimer care. Newer amyloid-directed antibodies such as lecanemab and donanemab have FDA approvals for Alzheimer disease, but their labels include warnings about amyloid-related imaging abnormalities, including brain swelling and bleeding. PAC-832 is not competing with those drugs on the same mechanism. It is a small-molecule attempt to improve cognition by disinhibiting acetylcholine release upstream of the pathway targeted by cholinesterase inhibitors.

Pace Pharmaceuticals' own report flags the central caveat. The mouse model used scopolamine to impair cognition by blocking muscarinic receptors, which means the behavioral data does not directly prove that PAC-832 restored cholinergic signaling in the brain in vivo. The authors wrote that establishing that mechanism will require direct CNS acetylcholine measurements or GalR1-dependent validation, such as work in GalR1-knockout mice. The behavioral tests also were not blinded, according to the methods section.

Yao is unusually explicit about the startup's constraints, and that is what makes the case notable beyond PAC-832 itself. Pace Pharmaceuticals says the work was supported by the authors' personal funds and received no external funding. The report lists Douglas Yao as founder of Pace Pharmaceuticals, which holds the PAC-832 patent, and Dongyuan Yao as a Pace Pharmaceuticals employee. It also says BioCurious provided access to cell culture reagents, a tissue culture room, HPLC-UV equipment and liquid handling robotics. Raw cell-experiment data and analysis code are posted on GitHub, while mouse video, tracking and behavioral-analysis data are on Zenodo.

Pace Pharmaceuticals' public pipeline is still sparse. The company says it has two compound series that have completed discovery: PAC-832 for Alzheimer disease and PAC-336, a selective M4 agonist for schizophrenia. Its site says both are undergoing IND-enabling studies and are expected to enter clinical testing within the next year. That would put PAC-832 at the edge of the real financing and regulatory test: toxicology, manufacturing controls, FDA clearance to dose humans, and the capital to run even a small Phase 1 study.

The sharper question is no longer whether a founder can make a garage-built molecule look credible online. Yao has shown how far the stack has moved: computational design, small-scale synthesis, community lab equipment, outsourced animal work, public code and raw data. The question is whether a self-funded, founder-led CNS program can produce the independent validation investors and regulators will need before treating PAC-832 as more than a pharmacology tool. For now, Pace Pharmaceuticals has a disclosed mechanism, mouse data, posted raw data and a clinical ambition. It does not yet have human safety data, clinical efficacy data, peer review or disclosed institutional backing.