NGN2 vs Dual-SMAD vs Direct Conversion — Neuron Protocol Routes

There are three routes to a human neuron in a dish, and they differ from each other far more than their products’ datasheets suggest.

NGN2 induction forces a pluripotent stem cell into a neuronal fate by overexpressing a single transcription factor, producing neurons in days. Dual-SMAD directed differentiation walks a pluripotent stem cell through the developmental stages of cortical neurogenesis using small molecules and growth factors, taking months. Direct conversion reprograms a somatic cell — usually a fibroblast — straight into a neuron without passing through pluripotency.

Which one produced the cells in your vial determines what they can be used for, and it is frequently not stated on the product page.

The three routes in one paragraph each

NGN2 (transcription-factor induction). Forced expression of Neurogenin-2 drives pluripotent cells directly to a neuronal fate, bypassing the progenitor stage. Zhang and colleagues reported that induced neuronal cells are “clearly identifiable already on day 6”, and functional readouts on NGN2 induced neurons are typically taken at two to three weeks. The output is predominantly excitatory glutamatergic and strikingly homogeneous. Delivery is by lentivirus, piggyBac, or a targeted engineered cassette in the parent line, and that engineering step is why the route carries platform licensing baggage that the small-molecule routes do not.

Dual-SMAD (developmental directed differentiation). Inhibition of both SMAD signalling branches pushes pluripotent cells to neuroectoderm, from which cortical progenitors emerge and then generate projection neurons in the stereotypical temporal order seen in development. Shi, Kirwan and Livesey describe an 80-day, three-stage process that recapitulates cortical development and produces all classes of cortical projection neurons that go on to fire action potentials and form functional synaptic networks. It is slow, it is variable between lines, and it is the route that gets you developmental fidelity.

Direct conversion (transdifferentiation). Vierbuchen and colleagues showed that ASCL1, BRN2 and MYT1L convert mouse fibroblasts directly into functional neurons; the same factor combination was subsequently extended to human cells. The cell never becomes pluripotent, which preserves donor age-associated signatures that reprogramming to iPSC erases — the main scientific reason to choose it. Conversion efficiency from human fibroblasts is the route’s chronic weakness.

Route comparison

	NGN2 induction	Dual-SMAD directed differentiation	Direct conversion
Starting material	iPSC or ESC	iPSC or ESC	Somatic cell, usually dermal fibroblast
Mechanism	Forced transcription factor expression, bypassing the progenitor stage	Sequential extrinsic signalling recapitulating development	Forced transcription factor expression, bypassing pluripotency
Time to identifiable neurons	Days. iN cells identifiable at day 6 in the founding report	Weeks to the progenitor stage; neurons emerge over an extended neurogenesis period	Days to weeks depending on factor set
Time to functional maturity	Functional readouts commonly at 2–3 weeks	Long. The Shi protocol is an 80-day process; EB-derived neurons in comparative work were assessed at 3–6 months	Weeks; maturity depends heavily on co-culture support
Purity / homogeneity	High and its main selling point. Near-uniform excitatory identity	Mixed by design. Multiple cortical layers and classes, plus residual progenitors and glia	Low conversion efficiency is the standard limitation; the converted fraction requires enrichment or sorting
Subtype fidelity	Narrow. Predominantly glutamatergic. Other subtypes require additional factors or patterning cues	High. Produces the developmental diversity and layer identity of cortex	Depends entirely on the factor cocktail; subtype control is an active research problem
Developmental information retained	Little. The route skips the progenitor stage, so it does not model neurogenesis	Substantial. This is the route for studying cortical development itself	None of the pluripotent developmental trajectory; retains donor somatic ageing signatures
Electrophysiological maturity	Real but comparatively modest at the timepoints used. In a comparative assessment, roughly 67% of NGN2 iNs met the maturity ranking versus roughly 90% of EB-derived neurons	Higher on the same ranking, at far longer culture times	Variable; typically requires glial co-culture to reach synaptic competence
Network activity on MEA	Reaches synchronised bursting, but astrocyte co-culture is commonly required to get there reliably	Forms synchronised oscillatory networks over weeks in culture	Weakest of the three without substantial support; not the usual choice for network work
Reproducibility	High. The main reason it displaced other routes for screening	Lower. Extrinsic-factor protocols are noted for low reproducibility and yields, and line-to-line variability is substantial	Low
Relative cost per usable neuron	Lowest of the three at scale — short culture, low reagent burden, high yield	Highest. Months of media changes, incubator time and technician attention, plus failure risk across a long run	High per usable neuron because of conversion efficiency, though the starting material is cheap
Licensing burden	Highest. Inducible expression systems and editing tools carry their own patent estates and may require licences from named third parties	Lowest. Small molecules and growth factors are commodity reagents	Moderate. Vector and factor delivery carry their own terms
Genomic modification	Yes. The parent line is engineered	No	Yes, unless a non-integrating delivery method is used

When each route is the right answer

Choose NGN2 when the neuron is a means, not the object of study. Screening, toxicology, device development, biohybrid and neuromorphic work, assay development, and any application where you need many neurons, quickly, that behave the same way in March and September. The homogeneity that makes NGN2 a poor developmental model makes it an excellent assay substrate. If your endpoint is a dose-response curve or an electrode recording rather than a developmental phenotype, this is almost certainly your route.

Choose dual-SMAD when developmental fidelity is the point. Neurodevelopmental disease modelling, cortical layer specification, progenitor biology, migration, anything where the process of becoming a neuron carries the signal you are looking for. Also choose it when you need subtype diversity that NGN2 does not natively produce, or when you specifically need the progenitor stage to exist so you can perturb it. Budget for the timeline honestly: an 80-day protocol is a quarter of a year per attempt, and attempts fail.

Choose direct conversion when donor age matters. Reprogramming a somatic cell to iPSC resets age-associated epigenetic and cellular signatures. For late-onset neurodegeneration, that reset can erase the phenotype you are trying to study. Direct conversion preserves it. The cost is efficiency, and you should plan for enrichment and for smaller numbers.

Reading the route from a product page

Vendors rarely name the route outright. Four reasonably reliable tells:

Speed claims and “assay-ready in days” language point to transcription-factor induction.
Naming NGN2, opti-ox, or an inducible cassette in the parent line is explicit — some vendors state it plainly.
Marketing that emphasises purity percentages above roughly 90 percent for a single excitatory identity points to induction, because developmental protocols do not usually produce that.
Marketing that emphasises subtype diversity, layer markers, or the presence of progenitors points to a developmental route.

Ask directly. The route is not confidential, and it determines whether you are buying a uniform assay substrate or a developmental model. Two products both labelled “iPSC-derived cortical neurons” at similar prices can be these two different things.

What the route does not tell you

It does not tell you the neurons will mature on your bench. All three routes produce cells that need weeks of culture on your equipment before an MEA experiment is meaningful, and every route benefits from — and often requires — astrocyte co-culture for reliable network bursting. See MEA-ready.

It does not tell you the purity you will get. Purity claims are only meaningful against named markers and a named method, at a named timepoint. Ask for all three.

It does not settle the cost question. NGN2 is cheapest per usable neuron at scale, but the licensing burden it carries can dominate the reagent saving for a commercial programme. The small-molecule routes are reagent-expensive and licence-cheap. Which is actually cheaper depends on whether you are running one experiment or building a product, which is a question about your licensing exposure rather than about biology.

It does not remove line-to-line variability. Even within a route, different parent lines differentiate with different efficiency and on different timelines. If your experiment spans lines, that is a variable you have to control rather than assume away.

Hybrid and adjacent routes

The three-way split is a simplification and the field has been eroding it deliberately. Protocols now combine NGN2 overexpression with dual-SMAD and Wnt inhibition, using patterning signals to give the induced neurons a regional identity they would not otherwise acquire — getting some of the speed of induction with some of the fidelity of patterning. Fully defined NGN2 protocols using clonal targeted-engineered lines and defined reagents have been published specifically to address the reproducibility complaints levelled at earlier lentiviral approaches.

There is also a fourth route in commercial practice that is not really a differentiation protocol at all: buying expandable neural stem cells or progenitors and running the final differentiation step yourself. That moves you onto the developmental route for the last stage only, with much of the timeline already paid for by the supplier, and it changes the economics for anyone needing large or repeated neuronal yields. It is the correct answer more often than its market share suggests.

The short version

If you need many identical neurons quickly and cheaply, NGN2. If you need to study how a neuron becomes one, dual-SMAD. If donor age carries your phenotype, direct conversion. And in all three cases, the vial is the beginning of the timeline rather than the end of it.

Differentiation protocol routes: NGN2, dual-SMAD and direct conversion