CAR-T Generations: From Zeta-Only to Composable Architectures

Side-by-side analysis of CAR-T receptor designs discussed at SDDS 2026. Each generation addresses a specific failure mode of the prior one.

---

Design Evolution

Generation	Architecture	Co-stim	Key Feature	Limitation Addressed	Champion
1st	scFv–CD3ζ	None	Cytolysis	Proof of concept	Eshhar (1993)
2nd (CD28)	scFv–CD28–CD3ζ	CD28	Potent activation + co-stimulation	AICD / anergy of 1st gen	Sadelain (2003)
2nd (4-1BB)	scFv–4-1BB–CD3ζ	4-1BB	Better persistence	Short-lived CD28 CARs	Campana, adapted by June
1XX	scFv–CD28–CD3ζ (1 live ITAM)	CD28	Durable at 10M cells	Over-activation / exhaustion	Sadelain (2024)
HIT	Ab V-regions in TCRαβ at TRAC	Endogenous CD3	10–50× sensitivity	Antigen-low escape	Sadelain (2026)
Pre-TCR	Pre-TCR signaling domain in CAR	Pre-TCR	100–200× expansion	Poor in vivo proliferation	Sadelain / Shi (2026)
Armored	CD19 CAR + IL-18 secretion	CD28 or 4-1BB	Microenvironment remodeling	Immunosuppressive TME	June
Dual-target	Bicistronic OR-gate (EGFRvIII + IL-13Rα2)	Varies	Antigen escape resistance	Single-antigen loss	June
CRISPR-KO	CAR + Regnase-1/EGR2 double KO	Varies	Enhanced persistence + function	T-cell intrinsic brakes	June

---

Key Insight: The Failure Was Never CD28 vs. 4-1BB

The clinical observation — CD28 CARs hit hard but don’t persist; 4-1BB CARs persist but hit slower — was read by the community as an intrinsic property of the co-stimulatory domain. Sadelain’s 1XX work showed this was wrong: the problem was over-activation through three ITAMs. A single live ITAM (the proximal one) restored durability without sacrificing potency. The lesson: it’s the balance between activation and co-stimulation, not the choice of costim domain.

Composability

The designs are not competitors — they are composable:

• 1XX addresses calibration (how hard to signal)
• HIT addresses sensitivity (how little antigen is enough)
• Pre-TCR addresses expansion biology (how many cells from how few)
• Armored addresses microenvironment (what the cell secretes)
• CRISPR-KO addresses intrinsic brakes (what checkpoints to remove)

A future CAR could combine HIT-level sensitivity with 1XX-calibrated signaling, pre-TCR expansion, IL-18 armoring, and Regnase-1/EGR2 double KO — each layer independently validated.

---

Cross-References

• Michel Sadelain — 1XX, HIT, pre-TCR
• Carl June — armored, dual-target, CRISPR-KO
• CAR-T therapy — hub page
• Glioblastoma — dual-target CAR trial
• Lupus — immune reset application