Research architecture / Updated September 12, 2026
The science
behind the connection.
Not just which bacteria are present.
What they release. Where it goes. What it does.
Fasolati’s science spine asks whether a well-characterized oral bacterial effector measurement can add useful information beyond a periodontal examination, bacterial burden, and conventional risk measures.
Here for the big picture?
Walk through the pathway explorer and see how the guided Life Board handles uncertainty.
Here to test the idea?
Inspect a claim and its revisions in the evidence ledger, then examine the experiments that could change the hypothesis.
Promising mechanisms.
Specific boundaries.
Outer membrane vesicles are a subset of bacterial extracellular vesicles, or bEVs. Vesicle identity requires more than a particle’s size, and an enriched preparation can contain non-vesicular material. MISEV2023 guidance.
This map distinguishes the preparation tested, the reported response, and the claim it cannot support. Open a row to inspect the experimental setting and its limits.
6 of 6 mechanisms
01Endothelial barrierP. gingivalis OMVs / gingipains → barrier effectsCells + zebrafish
What was observed
Wild-type and gingipain-deficient OMV comparisons supported a gingipain contribution to endothelial-barrier disruption, with PECAM-1 loss associated with the response. The zebrafish experiments used direct systemic injection, not oral exposure. Farrugia et al., 2020.
What that does not establish
This does not calibrate an oral-rinse measurement to vascular leak, heart attack, or stroke risk in a person.
Research use
A candidate barrier-response benchmark, with preparation, cargo, and exposure controls.
Direct link to this mechanism →02Hepatic insulin signalingP. gingivalis OMVs → liver and glycogen signalingMice + cells
What was observed
Seyama and colleagues reported liver translocation and reduced insulin-responsive hepatic glycogen synthesis in mice. In HepG2 cells, OMVs attenuated insulin-stimulated Akt/GSK-3β signaling in a gingipain-dependent manner. Seyama et al., 2020.
What that does not establish
This is not proof that a person’s glucose excursions are caused by oral vesicles, and it does not establish the same effect in muscle or adipose tissue.
Research use
A focused metabolic mechanism to test at exposure levels relevant to the intended human question.
Direct link to this mechanism →03Neuroinflammatory responsesP. gingivalis OMV exposure → brain-related phenotypesMouse experiment
What was observed
Aged male mice given OMVs by oral gavage at 4 mg/kg every other day for eight weeks showed neuroinflammatory, tau-phosphorylation, and memory-task effects. Gong et al., 2022.
What that does not establish
The experimental dose is not established as ordinary human exposure. This is not evidence for a personal Alzheimer’s risk meter or proof of human disease causation.
Research use
Retain the animal finding with its dose and route, rather than collapsing it into a mouth-to-brain prediction.
Direct link to this mechanism →04Innate immune activationPeriodontal-bacterial OMVs → receptor responsesMechanistic assays
What was observed
Vesicles from different periodontal organisms produced different TLR2 and TLR4 response profiles under the tested conditions. Cecil et al., 2016.
What that does not establish
A receptor response does not identify every active ligand. Comparing species individually does not establish synergy when their vesicles are combined.
Research use
A panel of functional readouts with ligand-attribution and interference controls, not one universal inflammation score.
Direct link to this mechanism →05Vesicular RNAA. actinomycetemcomitans OMV RNA → host-cell responsesCells + mouse delivery
What was observed
RNA-associated inflammatory effects were studied in U937 macrophage-like cells. The mouse brain-delivery experiment used intracardiac administration, bypassing the oral barrier. Han et al., 2019.
What that does not establish
Delivery by injection does not establish natural oral escape, human brain transit, or Alzheimer’s causation.
Research use
Separate cargo biology from transport biology, and require the relevant route to be tested.
Direct link to this mechanism →06Fusobacterium and tissue interactionsF. nucleatum EVs / FomA → adhesion and colonizationHuman tissue + models
What was observed
In five paired colorectal-cancer and adjacent-normal tissue samples, Zheng and colleagues found greater FomA signal in tumor-derived EV-enriched fractions, interpreted as F. nucleatum EV enrichment. Separate cell and mouse experiments supported FomA-transfer effects on adhesion, colonization, and tumor progression. Zheng et al., 2024.
What that does not establish
FomA is not a vesicle-exclusive marker. Human tissue enrichment and model mechanisms are separate observations; they do not establish oral origin in an individual or validate an oral-rinse cancer predictor. Read the marker-specificity discussion
Research use
A specific tissue-interaction hypothesis. Do not substitute whole-bacterium FadA studies for vesicle-specific FomA evidence.
Direct link to this mechanism →No mechanisms match this search. Try an organism, tissue, or experimental setting, or clear the search to see all six.
Download the cited map (JSON)An experimental effect is not a diagnosis. Evidence about whole bacteria or an isolated molecule is not automatically evidence about an intact vesicle carrying that molecule.
Human evidence exists.
Oral-source validation is a separate question.
Circulating bacterial vesicles
Tulkens and colleagues characterized bacterial-EV-associated plasma fractions in 49 people, including groups with intestinal-barrier dysfunction. The study provides a measurement precedent, not a human oral-to-blood source trace. Tulkens et al.
Ou and colleagues used single-particle LPS/LTA analysis in human cohorts alongside mouse source-discrimination experiments. The paper also discusses residual lipoprotein contamination and unresolved transport questions. Ou et al.
Periodontal care and systemic markers
In people with diabetes and periodontitis, a Cochrane review found HbA1c was a mean 0.43 percentage points lower after periodontal treatment than with no active treatment or usual care at three to four months (95% CI: 0.28 to 0.59 points lower; moderate-certainty evidence). This is not evidence that OMVs mediated the effect. Simpson et al., 2022.
A separate meta-analysis reported lower CRP at six months after periodontal treatment, with substantial heterogeneity. Its smaller analysis at twelve months or longer did not detect an effect; longer-term evidence remained limited. This does not establish that a Fasolati formulation changes inflammation through a vesicle mechanism. Luthra et al., 2023.
A historical clinical counterweight: the GAIN trial
The gingipain-inhibitor GAIN trial enrolled 643 participants and missed both co-primary endpoints in the overall cohort. A sponsor-reported salivary-Pg-positive subgroup signal on cognition did not establish benefit on the functional co-primary. Cortexyme’s 2022 filing described liver-safety concerns and a full IND hold on atuzaginstat dated January 25, 2022. Read the historical filing.
This is a historical trial example, not a statement about the current status of every gingipain-inhibitor program. It does not disprove all oral-systemic mechanisms, but it shows why mechanistic plausibility cannot stand in for successful human treatment. The subgroup does not validate OMV-based stratification.
Within the literature reviewed for this spine, no verified human calibration from a Fasolati oral-rinse vesicle measurement to an individual systemic outcome was located. That is a bounded review finding, not proof that no relevant unpublished or unlocated study exists.
Measure the thing
you say you measure.
A local sample, a transported exposure, and a host response are different quantities. The proposed architecture keeps four questions distinct until data justify a particular connection.
- 01
Source ecology
Which organisms and local conditions are present? A periodontal examination and bacterial burden provide context, not a direct measure of vesicle activity.
- 02
Effector phenotype
What biologically active material is released? Test characterized vesicle-enriched and depleted fractions rather than assuming the effect belongs to vesicles.
- 03
Delivered exposure
What reaches a relevant compartment and remains active? Demonstrate the carrier and source instead of relabeling a local rinse signal as systemic load.
- 04
Host response
What changes in the recipient? Measure the response independently, then test whether the additional information improves a defined decision.
Why a particle count is not an OMV count
Nanoparticle tracking analysis detects particles; it does not by itself identify them as bacterial vesicles. Host vesicles, non-vesicular particles, and co-isolated material require appropriate separation and characterization. MISEV2023.
The proposed assay work must establish identity, recovery, precision, specificity, interference, stability, and a quantifiable range. Neither a bacterial sequence in a mixed fraction nor an inflammatory reporter signal resolves all those questions.
Why fewer particles might not mean less activity
The development program must compare abundance, cargo, and functional activity separately. A change in sample dilution, analytical recovery, active material per particle, or non-vesicular activity could change the interpretation.
For oral-rinse intervention studies, residual formulation must be tested for interference with the assay or reporter cells. An apparent reduction in activity is not enough if the rinse itself suppresses the readout.
Why AI cannot fill a missing causal link
The evidence engine should distinguish a measured result, a validated association, an experimental mechanism, and an untested hypothesis. A fluent explanation cannot turn one category into another.
SciAgent’s proposed role is to organize evidence and explain uncertainty, not to invent missing measurements or issue autonomous treatment orders. Claims about operational safeguards require verification in the implemented system.
What result would
change our minds?
Does characterized vesicle-associated activity add useful information beyond periodontal examination and bacterial count?
These are proposed development gates, not completed Fasolati studies. The sequence is designed to find the weak link before turning it into a customer claim.
Identity before interpretation
Define the analyte, validate the assay in the actual sample matrix, and establish repeatability. Stop personal interpretation if analytical variation overwhelms the signal.
Fraction before mechanism
Compare enriched and depleted fractions, then use cargo-specific perturbations and appropriate rescue controls. If the activity is predominantly non-vesicular, follow that result rather than protect the OMV label.
Source before systemic attribution
Test oral attribution against gut and other competing sources. Model exposure separately from the recipient’s susceptibility.
Incremental value before a score
In a prospective study, test whether the measurement adds information beyond ordinary clinical measures. If it does not, do not add a test merely because it is technologically interesting.
Intervention before efficacy
Evaluate the defined finished formulation against an appropriate comparator. Pre/post changes are temporal associations unless the study design supports a causal conclusion.
Utility before routine use
Test whether acting on the information helps people more than the alternatives, including ordinary clinical care without the extra test.
The customer gets clarity.
The evidence keeps its limits.
The proposed Life Board should answer three questions: what changed, how sure the measurement is, and what next step is justified. It should not average discordant oral and systemic results into a reassuring composite score.
On the surface
The observation, date, measurement status, and an intelligible next step. Unvalidated measurements remain labeled as research, with no individualized organ-risk interpretation.
One layer deeper
The original source, experimental setting, carrier and cargo attribution, dose and route, limitations, and what would be required to support a stronger claim.
Established care does not need to “move your score” before it is worthwhile. Product benefit, assay validity, predictive validity, and clinical utility are separate questions that must earn separate answers.
This page describes a research program and product direction. It is not a diagnostic service, a treatment protocol, or a substitute for clinical evaluation.
Help test the next link.
Fasolati welcomes conversations about vesicle characterization, oral-source attribution, functional assays, and prospective clinical validation. The aim is a measurement that deserves to influence a decision.
Discuss a research collaborationReview scope: selected primary vesicle studies, human bEV measurement studies, clinical intervention reviews, and measurement guidance. This is an evidence-informed research architecture, not an exhaustive systematic review.