BPN14770 Research

BPN14770

How long should
a signal last?

Cells have a surprisingly demanding correspondence habit. A message has to arrive, be understood, and eventually stop. Keeping everything switched on would make for a very busy inbox.

BPN14770, also called zatolmilast, is a small molecule studied for its interaction with an enzyme called PDE4D. Its story starts with the timing of a signal.[6][10]

Start with the messenger, cAMP.

Pencil illustration of an enzyme with moving messenger symbols and a red compoundScrolling advances the messenger shapes, brings the red compound to the conceptual catalytic region, and slows the illustrated conversion sequence. This is conceptual choreography, not a molecular simulation.
Conceptual illustration. Shapes, sites and timing are not structural data.

Before it was called zatolmilast

BPN14770 began in Tetra’s drug-development programme. Tetra and Shionogi announced a collaboration in December 2018, and a detailed account of the compound’s design appeared the following year. Zatolmilast is the later name for the same molecule, which is why both names appear in the research.[6][1][10]

The compound is a synthetic small molecule. Its target, PDE4D, is an enzyme—a protein that speeds up a chemical reaction. PDE4D belongs to a family of four related enzymes named PDE4A, PDE4B, PDE4C and PDE4D. These enzymes break down cAMP, a chemical messenger that carries signals inside cells.[1]

What happens at the enzyme

PDE4D converts cAMP into AMP by opening a ring in the messenger’s chemical structure. A section of the enzyme called UCR2 folds over the working site like a lid. BPN14770 binds where this lid meets the working site and helps hold it closed, reducing access for cAMP.[1][3]

The 2019 experiments measured enzyme activity as the concentration of BPN14770 increased. In two lab-prepared forms that mimic activated PDE4D, some activity remained at the highest inhibition levels measured. This is called partial inhibition: reducing activity without stopping it completely. The researchers also tested other forms of the enzyme and found that the response changed when the UCR2 section was removed.[1]

A three-dimensional view of the binding site

Researchers used X-ray crystallography to map BPN14770 attached to a laboratory-made section of human PDE4D. X-rays scattered by a crystal allow scientists to calculate the positions of its atoms. The resulting structure includes the enzyme’s working region, its UCR2 lid and the bound compound.[3]

That structure is publicly available as 6NJJ in the Protein Data Bank, a collection of molecular structures. It was released in May 2019. The illustration above uses simplified shapes and movement; the database entry contains the experimental coordinates used to build a molecular model.[3]

The mouse experiments

The 2018 paper compared ordinary mice with mice whose PDE4D had been changed at one position. Human PDE4D has phenylalanine at that position, while mouse PDE4D has tyrosine. These are two amino acids, the building blocks of proteins. The modified mice had phenylalanine in that spot, matching the human enzyme.[2]

The researchers measured cAMP signalling in the brain, proteins involved in nerve-cell activity, and performance in memory tasks. They reported increased cAMP signalling and increased levels of BDNF, a protein involved in nerve-cell growth and connections. The researchers reported that BPN14770 was more potent in the modified mice, meaning a smaller amount produced an effect in the memory tasks.[2]

Fragile X syndrome and the human studies

Fragile X syndrome is an inherited condition that affects learning and development. It involves a change in the FMR1 gene, which provides instructions for making a protein called FMRP. A shortage of this protein affects the connections between nerve cells. People with the condition can have learning disabilities and delayed speech or language development.[12]

The study published in 2021 enrolled 30 adult men with fragile X syndrome, aged 18 to 41, and ran for 24 weeks. Participants were assigned to receive both BPN14770 and a placebo, an inactive comparison, in an order chosen by chance. This arrangement is called a crossover study: the same person takes part in both study conditions.[4]

What the 2021 study measured

The main assessments concerned how well participants tolerated the compound and unwanted health events recorded during the study. Additional assessments included computer-based tests of thinking abilities and ratings from caregivers and physicians. These additional measures were the study’s secondary outcomes.[4]

During the BPN14770 periods, participants scored higher, on average, on tests of word reading and vocabulary than during the placebo periods. The paper names these tests Oral Reading Recognition and Picture Vocabulary. A combined score from the two tests was also higher during BPN14770.[4]

Caregivers separately rated language skills and day-to-day functioning. Those ratings also improved during BPN14770 compared with placebo. The computer tests recorded the participants’ own performance, while the caregiver scales recorded observations of their communication and everyday activities.[4]

The larger trials reported in 2026

The later programme included CNS-301 in adults aged 18–45 and CNS-204 in children and teenagers aged 9–17. Shionogi’s May 12, 2026 presentation reported that neither met its primary endpoint, the main result selected to test the trial’s central question. Its reported comparison groups contained 171 adults and 155 younger participants.[5][8][9]

Shionogi reported greater improvements than placebo in adults’ language/communication and daily-function ratings. These reached nominal statistical significance—a statistical cutoff applied without adjusting for the number of comparisons. Shionogi labelled these findings exploratory. The adult and adolescent studies have separate registrations; the adolescent registration and sponsor report differ in their participant counts and the name of the main measurement, as detailed in the FAQ.[5][8][9]

What the brain-wave recordings measured

The original 30-person trial also collected EEG recordings. EEG uses sensors on the scalp to record electrical activity in the brain. Later papers returned to these recordings to examine responses to sounds and patterns of activity at rest.[13][14]

A 2024 paper examined a response called N1, a brief change in electrical activity after a sound. Researchers compared the size of this response with the amount of BPN14770 measured in blood samples. In 12 participants who received BPN14770 first, they reported a relationship between the blood concentration and the N1 response at the end of that study period. The relationship did not meet the paper’s statistical cutoff when the analysis used the change from each person’s starting recording.[13]

A 2025 paper examined resting recordings, taken while participants were not performing a test. It measured the peak alpha frequency, the strongest frequency within one of the brain’s electrical rhythms. Seventeen participants had usable recordings for all three conditions: before treatment, during BPN14770 and during placebo. The analysis combined measurements from different scalp regions to identify patterns associated with each condition.[14]

The 2025 paper reported a higher combined alpha-frequency measurement during BPN14770 than before treatment. Its direct comparison of the whole-head measurement during BPN14770 and placebo did not meet the paper’s statistical cutoff. Both EEG papers analysed recordings from the original trial rather than enrolling new groups of participants.[14][13]

Questions about BPN14770

Is BPN14770 a peptide?

BPN14770 is a synthetic small molecule. Unlike peptides, it is not built from a chain of amino acids—the building blocks also found in proteins. Zatolmilast is another name for this same compound.[3][10]

What does “allosteric” describe here?

Binding helps hold PDE4D in a form that restricts access to its working site. BPN14770 sits where that site meets UCR2, the part of the enzyme that acts like a lid. The crystal structure shows how closely those features sit together.[1][3]

Can I see the actual molecular structure?

The experimental structure is listed as 6NJJ in the Protein Data Bank, a public collection of molecular structures. Researchers used X-rays to map BPN14770 attached to a laboratory-made section of human PDE4D, including its working site and the UCR2 lid. The drawing on this page simplifies those features so the movement is easier to follow; the database record gives the measured positions of the atoms.[3]

Why are there two counts for the younger participants?

The official study registration for CNS-204 lists 163 participants overall, while the company’s reported comparison groups total 155. The sources reviewed do not explain the eight-person difference or establish how many participants contributed to each analysis. There is a second discrepancy: the registration names a combined vocabulary and reading score as the main measure, while the company’s May slide names ratings of communication and everyday activities. The reviewed public records do not explain these differences.[5][9]

Has it been approved for fragile X syndrome?

The US Food and Drug Administration (FDA) lists an orphan-drug designation for zatolmilast in fragile X syndrome. This is a status intended to encourage development for rare conditions. The record shows no FDA approval for this use; orphan designation alone does not authorize a drug to be marketed.[7]

Is it listed in Health Canada’s drug database?

Our October 30, 2026 searches found no product or drug-ingredient match under either zatolmilast or BPN14770 in Health Canada’s Drug Product Database. The database is updated nightly.[11]

Where are the human-study results?

The 30-person study is published in Nature Medicine and linked as reference 4. The larger trials’ sponsor-reported results are on slide 31 of Shionogi’s May 2026 presentation, reference 5. The adult and adolescent registrations are linked separately as references 8 and 9, with each study’s design and listed measurements.[4][5][8][9]

References

  1. Gurney ME et al. PDE4D allosteric-inhibitor design and synthesis. Journal of Medicinal Chemistry, 2019. PMID 31013090. Read the paper.
  2. Zhang C et al. BPN14770 in wild-type and humanized mice. Neuropsychopharmacology, 2018. PMID 30131563. Read the paper.
  3. RCSB Protein Data Bank. PDE4D catalytic domain and UCR2 with BPN14770, structure 6NJJ, released May 8, 2019. Explore the structure.
  4. Berry-Kravis EM et al. PDE4D inhibition in adults with fragile X syndrome: randomized phase 2 trial. Nature Medicine, 2021. PMID 33927413. Read the study record.
  5. Shionogi. Fiscal 2025 financial-results presentation, May 12, 2026, slide 31. Sponsor-reported CNS-301 and CNS-204 results. View the results slide.
  6. Shionogi and Tetra Discovery Partners. BPN14770 development collaboration, December 19, 2018. Read the announcement.
  7. US FDA. Zatolmilast orphan designation and approval record. Check the record; what orphan designation means.
  8. ClinicalTrials.gov. CNS-301 adult study, NCT05358886. View the registration.
  9. ClinicalTrials.gov. CNS-204 adolescent study, NCT05163808. View the registration.
  10. Shionogi. Zatolmilast (BPN14770) development update, April 11, 2024. Read the identity and programme update.
  11. Health Canada. Drug Product Database; product and ingredient searches for zatolmilast and BPN14770, October 30, 2026. Search the database; database scope and updates.
  12. US National Library of Medicine. Fragile X syndrome, MedlinePlus Genetics. About the condition.
  13. Norris JE et al. Auditory N1 response and BPN14770 blood concentration in fragile X syndrome. Molecular Autism, 2024. PMID 39488698. Read the study record.
  14. Norris JE et al. Analysis of resting EEG measurements from the BPN14770 trial. Translational Psychiatry, 2025. PMID 40877251. Read the study record.

Educational research content. The illustration represents a mechanism conceptually and does not predict a clinical outcome.

Back to BPN14770