Seed Train and Scale-Up — How a Vial Becomes a Batch

A seed train is the staged expansion that takes a single thawed cryovial through progressively larger vessels until there is enough biomass to inoculate a production run. The term comes from bioprocess engineering and it is now standard vocabulary in cell supply, because the economics of every quote you receive are governed by it.

The two terms that sit above it are defined formally, and it is worth using the formal definitions because they are the ones your supplier’s quality system uses.

Under ICH Q5D:

A cell bank is a collection of appropriate containers whose contents are of uniform composition, stored under defined conditions, each container representing an aliquot of a single pool of cells.
A master cell bank (MCB) is an aliquot of a single pool of cells, generally prepared from a selected cell clone under defined conditions, dispensed into multiple containers and stored under defined conditions. The MCB is used to derive all working cell banks.
A working cell bank (WCB) is prepared from aliquots of a homogeneous suspension of cells obtained from culturing the MCB under defined culture conditions.

And the quantity that ties the structure together:

In vitro cell age is the measure of time between thaw of the MCB vial and harvest of the production vessel, measured by elapsed chronological time, by population doubling level, or by passage level when subcultivated by a defined procedure.

Why the structure is two-tier

The reason is not bureaucratic. It is that cells change with culture, and the two-tier bank exists to bound how much change can accumulate between your characterised reference material and any lot you ship.

Tier	Made from	Typical size	Testing depth	Consumed by
Master cell bank	The selected clone or the primary isolate	Small enough to last the product’s commercial life; every vial is irreplaceable	Deepest. Identity, purity, adventitious agents, karyology, stability — the full characterisation	Only ever used to make working banks, never for routine production
Working cell bank	One MCB vial, expanded	Large. Made repeatedly over the product’s life	Lighter panel, focused on identity and contamination	Routine production. One WCB vial starts one seed train
Production lot	One WCB vial, expanded through the seed train	The batch you buy	Release testing per lot	You

The consequence is a fixed and bounded distance from reference material to product. Every lot you ever receive is: MCB vial → WCB expansion → seed train → harvest. No lot is ever further from the characterised reference than that. When the WCB runs out, a new one is made from a fresh MCB vial, and the distance resets rather than accumulating.

This is precisely the discipline that a research laboratory serially passaging one culture for three years does not have — and it is why supplier lots are more consistent than the buyer’s own cultures, not because the supplier’s cells are better but because the structure bounds the drift.

What limits how far a seed train can go

Three separate ceilings apply, and which one binds depends on the cell type.

Replicative capacity. Normal human diploid cells have a finite in vitro lifespan, the observation established by Hayflick and Moorhead in 1961. For primary cells this is the hard ceiling: there are only so many doublings in the strain, they are consumed by the seed train whether you want them for expansion or for function, and a batch large enough to be commercially interesting may consume most of them.

Functional drift, which arrives earlier. The passage at which cells stop performing is generally earlier than the passage at which they stop dividing. See passage number and population doublings. For a supplier, this is the real ceiling on primary-cell batch size.

Genomic selection. For pluripotent lines, replicative capacity is effectively unlimited and the binding constraint is selection of culture-adapted variants. Gain of 20q11.21 is the best-documented example: it is selected because it confers a survival advantage, and the published consequence is a dramatic negative effect on neuroectodermal differentiation while mesendodermal capacity is retained. A long seed train is exactly the condition under which such a variant sweeps, which is why bank-and-seed-train discipline matters more for pluripotent material than for anything else.

That last point is what makes the published matrix-free hPSC seed-train work informative rather than merely impressive. It reports optimised suspension cultivation over 8 passages (32 days) producing a cumulative yield in the region of 4.7 × 10¹⁵ cells while maintaining pluripotency, differentiation potential and karyotype stability — the qualifier being the point. A scale-up claim without the stability claim is a claim about biomass, not about product. The same work used intermediate high-density cryopreservation of suspension-derived cells followed by direct bioreactor inoculation, which is the seed-train equivalent of adding a bank tier: it lets the process restart from a defined intermediate rather than from the beginning.

The shape of a seed train

The geometry is dictated by inoculation density. Cells will not grow if seeded too sparsely, so each step can only expand by a bounded factor before the next, larger vessel is required. A seed train is therefore a sequence of vessel sizes, each inoculated from the last, with a fixed number of days between steps.

Stage	Adherent route	Suspension route	What governs the step
Thaw	One vial into a small flask	One vial into a shake flask or small vessel	Post-thaw recovery. A poor thaw compresses everything downstream
Early expansion	Sequential flask formats	Shake flasks, then small stirred vessels	Minimum viable seeding density at each step
Intermediate	Multi-layer vessels or roller bottles	Bench-scale stirred-tank bioreactor	Surface area (adherent) or working volume (suspension). Adherent scale-up is where labour cost concentrates
Production	Large surface-area systems or microcarriers in a bioreactor	Production-scale stirred tank	The target batch size
Harvest	Detachment, wash, formulate, fill, freeze	Harvest, wash, formulate, fill, freeze	Harvest and fill are their own risk and cost centre, not an afterthought

Two structural facts follow.

Adherent scale-up is fundamentally different from suspension scale-up. Adherent cells scale with surface area, which means more vessels, more handling, more open operations, and labour that rises close to linearly with batch size. Suspension scales with volume in a single closed vessel, which is why the industry moves anything it can into suspension or onto microcarriers. Microcarrier processes are the hybrid — adherent biology in a suspension vessel — and they introduce their own problem, cell detachment from the carrier, which is enough of a bottleneck to have its own optimisation literature.

Seed train length compounds risk. Each step is a handling event, a contamination opportunity, and a passage. A contamination at step four discards everything from step one onwards and restarts the calendar. This is the main reason lead times for large batches are long in a way that looks disproportionate to the growth arithmetic: the quoted time includes the fact that the train must be started well before you need the product, and the schedule has no slack in it.

What the terms mean commercially

This is the translation layer that determines whether a quote makes sense to you.

Term you will hear	What it means commercially
“We’ll bank it for you”	A service: expand your material, freeze a defined number of vials, test to an agreed panel, store. Priced by vial count, test panel and storage duration — the panel is usually the largest lever and the one buyers scope loosest
“MCB/WCB structure”	The supplier can produce comparable lots over years. Its absence means each production is closer to a fresh start
“Campaign manufacture”	Production runs on a scheduled campaign rather than continuously. If you miss the campaign window you wait for the next one, and this dominates lead time far more often than growth rate does
“Seed train is already running”	Your delivery date is real. If it has not started, add the train length to any date you are given
“Bridging lot”	Material produced to link an old bank or process to a new one, so data spanning the change remains comparable. Someone pays for it. Establish who at contract, not at the point it is needed
“Scale-up” vs “scale-out”	Scale-up is a bigger vessel; scale-out is more of the same vessel. Scale-out preserves the validated conditions and costs linear labour; scale-up changes the physical environment — mixing, shear, gas transfer — and requires re-validation. A supplier quoting a large batch should say which they mean
“Minimum order quantity”	Usually a seed-train artefact rather than a commercial preference. The train produces a batch of a given size; selling you a fraction leaves the rest to be discarded or held
“Lot” or “batch”	One production run from one seed train. The unit across which comparability cannot be assumed

The buy-versus-expand decision

Every buyer of expandable cells eventually faces this. The structure of it is simple even though the numbers are project-specific.

Expanding in house is attractive when the cells are genuinely expandable, the doubling time is short, your requirement is repeated over a long period, you have incubator capacity and trained hands, and passage-related drift is tolerable in your assay.

Buying each batch is attractive when the cells are post-mitotic (most iPSC-derived neurons and cardiomyocytes cannot be expanded at all, which settles the question), when consistency between experiments matters more than unit cost, when your labour is the scarce resource, or when the licence does not permit you to expand and distribute the material anyway.

That last point catches people. Expanding purchased material for your own use is normally permitted; expanding it to supply a collaborator or to build a product is a distribution question governed by the licence, not a technical one. Check before you build a process on it.

The hidden cost of in-house expansion is not medium and plastic. It is that you have constructed a seed train without a bank structure, so your cells drift, and the drift shows up as unexplained variance months later. If you are going to expand, bank early and bank deeply first — you are building a working cell bank whether or not you call it one, and it is far cheaper to do it deliberately.

Questions to ask a supplier quoting a batch

Question	Why it matters
Is there an MCB/WCB structure behind this product, or is each production independent?	Determines whether lot-to-lot consistency has a structural basis or is an aspiration
How many passages, or how many population doublings, from WCB vial to the vial I receive?	This is in vitro cell age. It is the number that predicts drift, and it is rarely volunteered
How many WCB vials remain, and when will a new WCB be made?	A WCB changeover is the most likely point at which product behaviour shifts. Knowing the date lets you plan a bridge
Will you notify me of a WCB changeover or a process change?	For research-grade material you are usually owed nothing. Ask anyway; it costs nothing and it is the top source of unexplained variance
Has the seed train for my order started?	Converts a quoted lead time into a real one
Adherent or suspension, and has this scale been run before?	A first run at a new scale is a development project. Price and schedule it as one
What is the campaign schedule?	Usually the binding constraint on delivery
If I need a bridging lot when you change banks, who pays?	Settle at contract
For a custom bank: what test panel, how many vials, what storage term, and who owns the bank?	The four variables that set the price, and the ownership question is the one most often left ambiguous

Boundary cases

A “master cell bank” made without characterisation is a freezer box. The term implies a testing dossier. Ask what was tested and at what passage, or you have bought vocabulary.

Post-mitotic products have no seed train in the usual sense. A differentiated neuron does not divide, so the expansion happens upstream at the pluripotent stage and the differentiation is the terminal batch operation. The relevant questions move to the parent line’s passage history and the differentiation batch size.

Progenitor products change the economics precisely because they are expandable. Buying a neural progenitor rather than a terminal neuron transfers the seed train to you — with its labour, its drift risk and its cost advantage. Whether that is a good trade depends on your volumes and your tolerance for variability.

Two-tier is a minimum, not a maximum. Some programmes run a pre-master or a research bank above the MCB, and some run an intermediate bank inside a long seed train, as in the intermediate-cryopreservation approach cited here. More tiers means more control and more storage cost.

The one-line summary

A seed train is the staged expansion from one vial to one batch; the master and working cell bank structure above it exists to bound how far any lot can drift from characterised reference material. Ask how many doublings separate the working bank from your vial, how many working bank vials remain, and whether the train for your order has started — those three answers predict consistency, timing and risk better than anything on the datasheet.

Seed train, master and working cell banks: the terms and what they cost you