Formulations indistinguishable by DLS, NTA and encapsulation efficiency (EE) can still differ in particle structure — and whether that difference is decision-relevant for your product is the question this page is built to help you answer. It lays out what ATEM’s cryo-EM measures, how precisely, how each readout is interpreted, and the case-study data behind every claim — including where it’s strong and where it stops.
Cryo-EM images individual LNPs vitrified in their native formulation buffer — no dilution, no labels, no models. Every particle is segmented and measured, then classified by a validated machine-learning model and released by a scientist.
Per-particle diameter and the full distribution — not an intensity-weighted average. Area and perimeter are measured directly from the pixel mask; diameter is derived from area.
Every particle assigned to its structural class — Solid Core, Biphasic Split, or Biphasic Dense (Blebbed) — giving the true population composition behind the average.
Circularity relative to a perfect sphere, computed from measured area and perimeter. Cryo-EM is the only method that resolves this directly.
Reliability is established two ways: technical replicates (same sample, repeated) define the method’s own noise floor, and accuracy is benchmarked against expert human operators. The numbers below are why a shift in the data can be read as a real change, not measurement scatter.
Benchmarked against a trained human operator, on data unseen in training. Human-like agreement on morphology class.
Standard deviation on average particle diameter across 10 technical replicates (5,000 particles each). Sub-nanometer reproducibility.
A random-draw study shows statistics stabilize beyond ~2,000 particles — essential for resolving less-frequent morphology classes.
Read how we validated the method
Across an independent 21-formulation DoE and an in-vivo potency study, a consistent structural signature tracked with better function — and, just as informatively, independent Fraunhofer studies showed where that signature does not hold. Here’s the short version of each finding, how we got there, and where to read the full analysis.
Encapsulation stays high below a size threshold; above it, variance rises and EE falls.
In an independent 21-formulation DoE, encapsulation was consistently high (>90%) and low-variance for smaller particles, up to an overall-size threshold of ~55 nm (≈53 nm on the solid-core population); above it, EE variance increased and average EE decreased. The threshold is product-specific — we report the shape of the relationship, not a universal optimal diameter.
More Solid Core, fewer blebbed particles cooccured with better function.
Between two mixing technologies with identical lipids and mRNA, the more homogeneous, Solid-Core-rich material (64% vs 31%) showed significantly higher EE. Both structure and function differed — but the link did not survive within-group validation, so this is a population-level association, not a proven within-type correlation.
Near-spherical particles encapsulate better; falling shape ratio is a warning.
Healthy reference material sits above 92% circularity. In the formulation DoE, lower shape ratios were associated with worse encapsulation, and shape ratio was among the strongest synthesis-controlled structural variables.
The findings rest on validated precision (96.3% classification accuracy, sub-1 nm reproducibility) and human-in-the-loop release. Stress responses are reproducible across replicates, and structural differences are read against the method’s own measured noise floor.
Our data suggests that a structure-function relationship exists. Evidence is based on a limited number of case studies, so we are continuing to investigate. We state this as our current conclusion: structure relates to function — the specific links vary from case to case – and the evidence is actively being extended. We don’t claim a specific, universal link, and whether blebbing helps or harms remains product-specific (see Stage 3).
Summarizing the structural signature, the supporting data, the interpretation rules, and how to read a morphology report.
Follow your program from candidate selection to commercial release. Each step shows how cryo-EM is used and what to watch; each links to an in-depth guide covering how it works, what samples to send, and the readouts that matter.
Your candidates show similar size, EE and PDI — the readouts can’t separate them. Structure can: it distinguishes the best-performing formulation among apparent equivalents.
Where size, EE and PDI converge, cryo-EM resolves the structural differences that still separate candidates — ranking them by the signature associated with better performance and by population homogeneity. It identifies the most suitable formulation by selecting:
A stability-indicating extension of formulation selection: from your downselected candidates, identify the most stable ones as early as possible — ranking robustness before instability shows up later.
Building directly on formulation selection, this applies the same comparative logic to stability. Cryo-EM characterizes each candidate by the rate of structural change under stress, producing a robustness ranking — so you advance the most stable material and catch issues early. Each stress also leaves a distinct structural fingerprint, so the data shows not just that a sample degraded but how:
Establish a structural reference standard from your potency-characterized material — the ground truth your downstream product must continue to match.
Building directly on formulation selection, this applies the same comparative logic to stability. Cryo-EM characterizes each candidate by the rate of structural change under stress, producing a robustness ranking — so you advance the most stable material and catch issues early. Each stress also leaves a distinct structural fingerprint, so the data shows not just that a sample degraded but how:
An honest note on potency & blebbing
ATEM has indications that morphology corresponds to potency, mainly driven by sample integrity. But there is no global standard, no proven correlation within a single sample type, and no definitive answer on blebbing. In our data, blebbing is a stress-induced artifact — yet particles deliberately formulated to bleb may be desirable (Cheng et al.), which should still be verified against adverse stability effects.
A sound structural ground truth for upscaling — acceptance ranges grounded in your own product, and Quality-by-Design data that can support a later IND
Establish a reference standard for the selected product on both drug product and in-process control. This gives you defensible acceptance ranges grounded in your material — and generates structural data that supports Quality-by-Design and a later IND:
Scaling from lab to production can quietly change the product — you need to prove comparability, or pinpoint where it diverged.
Scaling a process introduces new challenges. Cryo-EM answers them in two stages:
Every batch has to match the one you validated — and you need defensible evidence of it for release, not just a number that drifts.
With a reference standard established, you’re positioned to monitor batch-to-batch reproducibility. Three applications:
Once a structural reference standard exists, cryo-EM supports several decisions outside the linear development path — each a comparison against that standard.
Cryo-EM uniquely combines 3D structural resolution, throughput, and native-state analysis, addressing the most relevant limitations of traditional methods.
Quantify how a change of raw-material source shifts your product’s structure against the standard.
Track the rate of change of the morphological profile over time as a stability-indicating readout.
Read how we validated the method
A number that moved isn’t a finding until it clears the method’s own noise floor. These rules sit behind every application above.
Significant if >1 nm diameter, >2% fraction, >0.2% shape ratio. Rare classes carry larger sampling error — demand a bigger effect before trusting a shift there.
Stability is read as how fast structure drifts under stress, not a single snapshot. Compare rates across candidates to rank robustness.
In stress data, rising blebbed fraction is a degradation signature. Synthesis-induced blebs may behave differently — so the good/bad call is product-specific.
A correlation that holds across pooled samples but vanishes within homogeneous subgroups is a process artifact, not a structure–function law. Always test the link within group before believing it.
A single, simple set of requirements covers most LNP studies. If cold-chain or cross-border logistics are a hurdle, ATEM can manage the shipment end-to-end so you can focus on the science.
Requirements may vary by study type. Our scientists confirm the exact specification for your samples before you ship.
Cold-chain handling, customs paperwork and courier coordination are often the slowest part of getting started. We take that off your plate.
We’ve outlined where structural data supports each decision across the LNP lifecycle. To discuss how it applies to your program and scope the right study, reach out to our team.
You are currently viewing a placeholder content from Vimeo. To access the actual content, click the button below. Please note that doing so will share data with third-party providers.
More InformationYou are currently viewing a placeholder content from YouTube. To access the actual content, click the button below. Please note that doing so will share data with third-party providers.
More InformationYou need to load content from reCAPTCHA to submit the form. Please note that doing so will share data with third-party providers.
More Information