Co-Inhibitory Checkpoints & T-Cell Exhaustion
PD-1 is not the only brake on a T cell. A whole family of co-inhibitory receptors runs in parallel, and each shuts the cell down by a different mechanism. CTLA-4 does not merely outcompete CD28 for CD80 and CD86 — it strips them off the antigen-presenting cell by trans-endocytosis, and recruits PP2A. LAG-3 reads stable-peptide MHC class II and FGL1. TIM-3 has four ligands — galectin-9, CEACAM1, HMGB1 and phosphatidylserine — and no classical ITIM, working instead through Bat3 release. BTLA answers HVEM, VISTA answers VSIG-3, and CD200R answers CD200. Nearly all of them converge on SHP-1 and SHP-2, and PD-1's real substrate is CD28 rather than the TCR. Sustained signalling drives an NR4A1–BLIMP1–EOMES programme that displaces TCF1, and effector output — IL-2, IFN-γ, TNF, granule release — falls away. Click any protein for the matching Assay Genie In Vivo antibody or ELISA kit.
Six receptors, six mechanisms. The co-inhibitory receptors are often drawn as interchangeable brakes. They are not. CTLA-4 binds CD80 and CD86 with far higher avidity than CD28 and then removes them from the antigen-presenting cell by trans-endocytosis, so its effect outlasts the contact and extends to neighbouring cells — this is cell-extrinsic suppression, and it is why CTLA-4 blockade behaves so differently from PD-1 blockade in vivo. LAG-3 has two ligand routes: stable-peptide MHC class II, which it binds more tightly than CD4 does, and the hepatocyte-derived FGL1. TIM-3 is the outlier — no classical ITIM in its tail, four structurally unrelated ligands (galectin-9, CEACAM1, HMGB1, phosphatidylserine), and a mechanism that runs through release of Bat3 from the cytoplasmic tail rather than direct phosphatase docking. BTLA–HVEM, VISTA–VSIG-3 and CD200–CD200R add three further axes that are routinely left out of checkpoint diagrams.
Where they converge, and the CD28 surprise. Most of these receptors recruit SHP-1 or SHP-2 through ITIM and ITSM motifs, and the phosphatases strip phosphate from the proximal machinery — LCK and ZAP-70 — collapsing calcium flux and NFAT activation. CTLA-4 works differently again, recruiting PP2A as well. The detail worth knowing is that PD-1's dominant substrate is not the TCR at all but CD28: the brake is applied to co-stimulation rather than to antigen recognition, which reframes why co-stimulation agonists and checkpoint blockade combine as well as they do. PD-1 and TIGIT appear here as entry nodes only — each has its own dedicated interactive map in this library.
Exhaustion is a programme, not just receptor load. Chronic antigen drives sustained NFAT signalling, which induces NR4A1 and a BLIMP1–EOMES transcriptional state while displacing TCF1. The TCF1-high population is the stem-like progenitor that responds to checkpoint blockade; the BLIMP1/EOMES-high terminal population largely does not. Effector output — IL-2 first, then IFN-γ and TNF, then granzyme B and perforin — is lost in a characteristic order, which is why measuring a single cytokine can miss the depth of exhaustion entirely.
The In Vivo tie-in. Sixteen nodes on this map are covered by the In Vivo functional-grade range, which is what makes it a working plan rather than a picture: low- and ultra-low-endotoxin blocking antibodies against CTLA-4, LAG-3, TIM-3, CD200, PD-1, PD-L1, PD-L2, TIGIT and CD155, agonist and blocking reagents for CD28, CD80 and CD86, anti-CD3 for stimulation, and neutralising antibodies against IL-2, IFN-γ and TNF to read the output arm. Single-agent blockade of any one receptor is usually partial; the point of the map is to show which combinations are mechanistically non-redundant. For research use only; not for use in diagnostic or therapeutic procedures.
Every protein node links to a product — In Vivo antibody, ELISA kit or research antibody.