What Is a Seed Train?
A seed train (also called an inoculum train) is the stepwise expansion of cells from a frozen cryovial to a volume and density sufficient to inoculate the production bioreactor. Every mammalian cell culture manufacturing campaign depends on a well-designed seed train to deliver healthy, actively dividing cells at the right density and the right time.
The seed train begins when a vial from the working cell bank (WCB) is thawed, typically containing 5–20 × 106 total cells in 1–2 mL. Cells are expanded through progressively larger vessels—T-flasks, shake flasks, spinner flasks, wave bags, and stirred-tank bioreactors—until the final stage (called N-1) produces enough cells to inoculate the production vessel (N stage) at the target seeding density.
For a 2,000 L production bioreactor seeded at 0.3 × 106 cells/mL, you need 6 × 1011 total viable cells. Reaching that number from a single cryovial requires roughly a 30,000–60,000-fold expansion, which typically takes 5–8 passages over 18–28 days.
Designing the Expansion Sequence
The seed train design starts at the end and works backward. Calculate the total viable cells needed at production inoculation, then divide by the achievable harvest density at each preceding stage to determine the required volume and number of passages.
Two parameters govern the expansion at each stage:
- Split ratio — the ratio of old culture volume to new culture volume (e.g., 1:5 means 20% of the harvest is carried forward). CHO cells typically use 1:3 to 1:10 split ratios depending on the stage.
- Growth duration — the time between seeding and the next passage, typically 3–4 days per stage for mammalian cells with doubling times of 20–30 hours.
A well-designed seed train minimizes the total number of passages (reducing contamination risk and hands-on time) while keeping cells in exponential growth at every transfer. Each passage where cells reach stationary phase or drop below 90% viability introduces a lag phase that compounds through subsequent stages.
The expansion factor per stage depends on the seeding density, harvest density, and whether the vessel volume changes. For a CHO cell line seeded at 0.3 × 106 cells/mL and harvested at 3.0 × 106 cells/mL (a 10-fold increase in density), transferring into a vessel 5× larger gives a net 50-fold expansion per stage.
Seeding Density, Split Ratio, and Passage Timing
Seeding density is the single most important parameter in seed train design because it determines both the lag phase duration and the growth trajectory at each stage. Seed too low and the culture takes an extra day to reach exponential phase; seed too high and you waste cells from the previous stage and may encounter nutrient limitation early.
| Cell Line | Seeding Density (106/mL) | Passage Density (106/mL) | Split Ratio | Passage Interval (days) | Doubling Time (h) |
|---|---|---|---|---|---|
| CHO-K1 / CHO-DG44 | 0.2–0.5 | 2.0–4.0 | 1:5 to 1:10 | 3–4 | 20–28 |
| HEK293 | 0.3–0.5 | 2.0–3.5 | 1:4 to 1:8 | 3–4 | 22–30 |
| Vero | 0.1–0.3 | 1.5–3.0 | 1:3 to 1:6 | 3–5 | 24–36 |
| Hybridoma | 0.2–0.5 | 1.5–2.5 | 1:3 to 1:5 | 2–3 | 14–22 |
| NS0 myeloma | 0.2–0.4 | 1.5–2.5 | 1:3 to 1:6 | 3–4 | 20–30 |
Passage timing should be based on cell density and viability rather than fixed calendar schedules. Kern et al. (2016) demonstrated that optimizing passage timing based on space-time yield (the average cell production rate per unit time) rather than fixed 3-day intervals saved up to 108 hours across a full CHO seed train before reaching a 5,000 L production vessel.
The practical rule: passage when cells are in late exponential phase (density 70–80% of maximum), viability is >93%, and the growth rate has not yet begun to decline. Waiting until stationary phase wastes 12–24 hours and forces cells through a longer lag phase in the next stage.
Vessel Selection at Each Stage
Each seed train stage uses the smallest vessel that holds the required culture volume while maintaining adequate gas exchange and mixing. Over-sizing vessels wastes media and incubator space; under-sizing forces extra passages.
| Stage | Vessel Type | Total Volume | Working Volume | Mixing | Gas Exchange |
|---|---|---|---|---|---|
| S1 | T-75 flask | ~60 mL | 15–20 mL | Static | Vented cap |
| S2 | T-225 flask | ~270 mL | 45–75 mL | Static | Vented cap |
| S3 | Shake flask (125 mL–3 L) | 125–3,000 mL | 30–1,000 mL | Orbital shaker | Vented cap, 130 rpm |
| S4 | Wave bag (2–10 L) | 2–10 L | 1–5 L | Rocking motion | CO2/air overlay |
| S5 | STR bioreactor (10–50 L) | 10–50 L | 7–40 L | Impeller | Sparged air/O2, pH, DO |
| N-1 | STR bioreactor (50–500 L) | 50–500 L | 35–400 L | Impeller | Full process control |
| N | Production STR (200–2,000 L) | 200–2,000 L | 150–1,600 L | Impeller | Full process control |
The transition from static or shaker vessels to controlled bioreactors is the most critical step in the seed train. Cells moving from a shake flask (uncontrolled pH, limited O2) into a bioreactor with pH and DO control may experience a transient growth lag. To minimize this, ensure media is pre-equilibrated to 37 °C, pH 7.0–7.2, and >80% DO before adding cells.
Seed Train Expansion Planner
Auto-select optimal vessels at each stage. Calculates growth duration, media volumes, and timeline for CHO, HEK293, Vero, and more.
Worked Example: 2,000 L CHO Seed Train
Working backward from the production bioreactor, this example shows how to calculate the number of stages and vessels needed for a CHO-K1 mAb process with a 2,000 L production volume.
Worked Example — Seed Train for 2,000 L CHO mAb Fed-Batch
Given:
- Production vessel working volume: 1,600 L (80% of 2,000 L)
- Target seeding density: 0.3 × 106 cells/mL
- Total cells needed: 0.3 × 106 × 1,600,000 mL = 4.8 × 1011 cells
- CHO-K1 doubling time: ~24 h; passage density: 3.0 × 106 cells/mL
- WCB vial: 10 × 106 total cells in 1 mL
Stage-by-stage calculation (working backward):
N-1: 4.8 × 1011 cells ÷ 3.0 × 106 cells/mL = 160 L harvest → 200 L STR (160 L WV)
Seeded at 0.3 × 106/mL → need 4.8 × 1010 cells in → 4 days growth
S5: 4.8 × 1010 cells ÷ 3.0 × 106/mL = 16 L harvest → 20 L STR (16 L WV)
Seeded at 0.3 × 106/mL → need 4.8 × 109 cells in → 4 days
S4: 4.8 × 109 cells ÷ 3.0 × 106/mL = 1.6 L harvest → 3 L shake flask (1.6 L WV)
Seeded at 0.3 × 106/mL → need 4.8 × 108 cells in → 4 days
S3: 4.8 × 108 cells ÷ 3.0 × 106/mL = 160 mL harvest → 500 mL shake flask (160 mL WV)
Seeded at 0.3 × 106/mL → need 4.8 × 107 cells in → 3 days
S2: 4.8 × 107 cells ÷ 3.0 × 106/mL = 16 mL harvest → T-75 flask (16 mL WV)
Seeded at 0.3 × 106/mL → need 4.8 × 106 cells in → 3 days
S1 (thaw): Thaw 1 vial (10 × 106 cells) into T-75 → seed at ~0.6 × 106/mL in 16 mL → 3 days recovery
Total: 7 stages, ~25 days from thaw to production inoculation. Media consumption: ~1,820 L across all stages.
N-1 Perfusion and Seed Train Intensification
N-1 perfusion is the most impactful intensification strategy for mammalian seed trains because it compresses the final expansion step from a low-density batch into a high-density perfusion culture. Instead of inoculating the production bioreactor at 0.3–0.5 × 106 cells/mL, N-1 perfusion grows cells to 40–100 × 106 cells/mL and transfers a concentrated bolus that seeds the production vessel at 3–10 × 106 cells/mL.
The benefits are substantial:
- Fewer stages — eliminate 2–3 intermediate expansion steps (skip the shake flask and small bioreactor stages entirely)
- Shorter timeline — 35–60% reduction in seed train duration (Seth et al. 2013 reported 60–70% time savings)
- Higher production inoculation density — shortens the lag phase in the production bioreactor by 1–3 days
- Reduced contamination risk — fewer open manipulations and aseptic transfers
| Parameter | Conventional Batch | N-1 Perfusion |
|---|---|---|
| Number of stages | 6–8 | 3–4 |
| Seed train duration | 18–28 days | 10–14 days |
| N-1 harvest density | 2–4 × 106/mL | 40–100 × 106/mL |
| Production inoculation density | 0.3–0.5 × 106/mL | 3–10 × 106/mL |
| Open manipulations | 5–7 | 2–3 |
| Cell retention device | None | ATF / TFF / acoustic settler |
| Media consumption | ~1,800 L | ~2,500 L (perfusion media) |
Cell retention devices for N-1 perfusion include alternating tangential flow (ATF) filters, tangential flow filtration (TFF) with hollow-fiber modules, and acoustic settlers. ATF is the most widely used at scales up to 500 L, with a cell-specific perfusion rate (CSPR) of 20–50 pL/cell/day typical for CHO cells.
Perfusion & Continuous Process Calculator
Calculate CSPR, VVD, bleed rates, and steady-state VCD for perfusion cultures. Compare batch vs. fed-batch vs. perfusion.
High-Density Cell Banking
High-density cell banking (HDCB) eliminates the early, labor-intensive stages of the seed train by cryopreserving cells at 50–260 × 106 cells/mL instead of the conventional 10–20 × 106 cells/mL. Müller et al. (2022) demonstrated cryopreservation at 260 × 106 cells/mL in standard cryovials, with post-thaw viability >85% and successful direct inoculation of an N-1 perfusion bioreactor.
With an HDCB at 100 × 106 cells/mL in a 5 mL cryobag, a single thaw provides 5 × 108 viable cells—enough to directly inoculate a 2 L bioreactor at 0.25 × 106 cells/mL. This skips all T-flask and shake flask stages.
Key considerations for high-density banking:
- Controlled-rate freezing is critical—use −1 °C/min with 7.5–10% DMSO
- Post-thaw recovery may be lower (70–90% viability) than standard-density banks; allow 24–48 h recovery before assessing
- Volume per unit matters: 5 mL cryobags (e.g., CellSeal, Cryostore) hold more cells than 1–2 mL vials
- HDCB is compatible with both batch and perfusion N-1 strategies
Common Seed Train Problems and Fixes
Seed train failures propagate forward through every subsequent stage, so diagnosing and fixing issues early is critical. The most common problems fall into three categories: poor post-thaw recovery, inconsistent growth between passages, and viability drops at scale transitions.
| Problem | Likely Cause | Fix |
|---|---|---|
| Low post-thaw viability (<70%) | Slow thaw, DMSO toxicity, old cell bank | Thaw rapidly in 37 °C water bath (<2 min), dilute into pre-warmed media within 1 min, centrifuge to remove DMSO within 5 min |
| Extended lag phase (>48 h) | Seeding density too low, cold media, pH shock | Increase seeding density to 0.5 × 106/mL, pre-warm and pH-adjust media, add 10% conditioned media for first passage |
| Viability drop at bioreactor transition | Shear stress, DO/pH shock, impeller speed too high | Start impeller at lower RPM for first 6 h, pre-equilibrate media, use gentle transfer (peristaltic pump, not pouring) |
| Declining growth rate over passages | Nutrient depletion, passage too late, media aging | Passage 12 h earlier (mid-exponential), use fresh media, check glucose >2 g/L and glutamine >1 mM at passage |
| Clumping in suspension culture | Cell line tendency, high density, low agitation | Add 0.1% Pluronic F-68, increase orbital speed by 10 rpm, passage before 3 × 106/mL |
| Mycoplasma contamination | Contaminated cell bank or media | Discard culture, test WCB by PCR, revert to MCB if WCB is positive. Do not attempt rescue. |
Cell Bank Sizing Calculator
Plan MCB/WCB campaigns per ICH Q5D. QC vial allocation, stability testing schedule, and product lifecycle planning.
Frequently Asked Questions
How many passages does a typical mammalian seed train require?
A conventional CHO seed train requires 5–8 passages from cryovial to production bioreactor, spanning 18–28 days depending on target volume and seeding density. Intensified seed trains using N-1 perfusion or high-density cell banks can reduce this to 3–4 passages and 10–14 days.
What is the optimal seeding density for CHO cell culture expansion?
CHO cells are typically seeded at 0.2–0.5 × 106 viable cells/mL for routine passage, with a target harvest density of 2–4 × 106 cells/mL before the next split. For N-1 bioreactor inoculation, seeding densities of 0.3–0.5 × 106 cells/mL are standard, while N-1 perfusion stages can reach 40–100 × 106 cells/mL.
What viability threshold should trigger a passage in the seed train?
Cells should be passaged while viability remains above 90%, ideally at 93–97%. Viability below 90% at any seed train stage indicates suboptimal culture conditions and can compromise downstream productivity. If viability drops below 85%, the seed train should be restarted from a fresh vial.
What is N-1 perfusion and how does it intensify the seed train?
N-1 perfusion uses a perfusion bioreactor in the stage immediately before the production vessel. By growing cells to 40–100 × 106 cells/mL, it enables inoculation of the production vessel at 3–10 × 106 cells/mL instead of the conventional 0.3–0.5 × 106 cells/mL, eliminating 2–3 intermediate stages and reducing seed train duration by 35–60%.
How do you calculate the number of cells needed to inoculate a production bioreactor?
Multiply the target seeding density by the working volume: total cells = seeding density (cells/mL) × working volume (mL). For a 2,000 L bioreactor at 0.3 × 106 cells/mL, you need 6 × 1011 total cells. Work backward through each stage with a 1:5 split ratio to determine the starting vessel size and number of passages.
Related Tools
- Seed Train Expansion Planner — Auto-select vessels and calculate timelines from cryovial to production bioreactor
- Cell Bank Sizing Calculator — Plan MCB/WCB campaigns with QC vial allocation and stability schedules
- Scale-Up Calculator — Compare five scale-up criteria (P/V, tip speed, Re, kLa, mixing time) side-by-side
References
- Kern S, Platas-Barradas O, Pörtner R, Frahm B. Model-based strategy for cell culture seed train layout verified at lab scale. Cytotechnology. 2016;68(4):1019–1032. doi:10.1007/s10616-015-9858-9
- Hernández Rodríguez T, Pörtner R, Frahm B. Seed train optimization for suspension cell culture. BMC Proceedings. 2013;7(Suppl 6):P9. doi:10.1186/1753-6561-7-S6-P9
- Seth G, Hamilton RW, Stapp TR, et al. Development of a new bioprocess scheme using frozen seed train intermediates to initiate CHO cell culture manufacturing campaigns. Biotechnol Bioeng. 2013;110(5):1376–1385. doi:10.1002/bit.24808
- Müller J, Ott V, Eibl D, Eibl R. Seed train intensification using an ultra-high cell density cell banking process. Processes. 2022;10(5):911. doi:10.3390/pr10050911