mRNA Capping

Capping in three enzymatic steps

Capping is the very first post-transcriptional modification. It occurs while the mRNA is still tethered to RNA polymerase II — the enzymes that build the cap ride along on the phosphorylated CTD of Pol II so they cap the first transcript they see, not later ones.

1. Triphosphatase removes the γ phosphate. The nascent 5'-pppN end is hydrolyzed to 5'-ppN. In humans, RNGTT (RNA guanylyltransferase and 5'-phosphatase) carries this activity in its N-terminal domain.
2. Guanylyltransferase transfers GMP from GTP. The enzyme first forms a covalent enzyme-GMP intermediate (lysine-bound), then attacks the 5'-diphosphate, joining GMP via a 5'-5' triphosphate bridge: GpppN.
3. N7-methyltransferase methylates the new guanine. RNMT, with its activator RAM, transfers a methyl group from S-adenosylmethionine (SAM) onto the N7 nitrogen, completing m7GpppN — the cap0.

Higher eukaryotes go further. CMTR1 methylates the 2'-OH of the first transcribed ribose (cap1); CMTR2 methylates the second ribose (cap2). These additional methylations are critical for self/non-self discrimination by IFIT1 and RIG-I in the cytoplasm.

Capping pathway and cap structure

  Pol II nascent transcript:  pppN-N-N-N-...
       │
       ▼  RNA triphosphatase  (γ-PO₄ removed)
       │
       ppN-N-N-...
       │
       ▼  guanylyltransferase  + GTP
       │
       GpppN-N-N-...        (5'-5' triphosphate bridge!)
       │
       ▼  N7-methyltransferase  + SAM
       │
       m7GpppN-N-N-...       (cap0)
       │
       ▼  CMTR1 (2'-O-methylates 1st ribose)
       │
       m7GpppNm-N-N-...      (cap1, mammalian standard)
       │
       ▼  CMTR2 (2'-O-methylates 2nd ribose)
       │
       m7GpppNm-Nm-N-...     (cap2)


  Cap structure (head-on view):

       7-methylguanosine
              │
        5'-5'│triphosphate
              │
       first transcribed N (often A or G)
              │
              ▼
       rest of mRNA  →  AAA(A)n  3'

Cap-dependent vs cap-independent translation

	Cap-dependent (canonical)	IRES-driven	m6A-driven	Cap-snatching (viral)	Vaccinia-style (viral)	uORF reinitiation
5' end requirement	m7GpppN cap	Structured RNA >100 nt	m6A in 5' UTR	Stolen capped fragment	m7GpppN (made by virus)	m7GpppN (host)
Initiation factor	eIF4E + eIF4G + eIF4A + eIF3	eIF4G fragment + eIF3 (varies)	YTHDF1/2/3 / eIF3	eIF4E (cap is real)	eIF4E (cap is real)	eIF4E (rebinds)
Used in stress / apoptosis	Suppressed	Active (XIAP, BiP, c-myc)	Active (HSP70 under heat)	Viral takeover	Viral takeover	ATF4, GCN4
40S recruitment mode	5' end scanning	Direct internal binding	Direct via YTHDF	Standard scanning of cap	Standard scanning	Re-scan after stop
Drug targets	eIF4E (4EGI-1), eIF4A (silvestrol)	HCV IRES (miravirsen)	METTL3 (STM2457)	PA endonuclease (baloxavir)	Vaccinia D1	—
Cellular example	~95% of mRNAs at rest	BiP, XIAP, p53, c-myc	HSP70 under heat shock	—	—	ATF4 stress response
Viral exemplar	—	Polio, FMDV, HCV	Some flaviviruses	Influenza, hantavirus	Vaccinia, reovirus	—

Real-world relevance

• mRNA vaccines. BioNTech/Pfizer and Moderna COVID-19 vaccines use cap1 analogs (CleanCap AG). Cap1 alone, not cap0, evades IFIT1 sensing and prevents premature interferon response — the difference between a tolerable shot and a pyrogenic one. Karikó and Weissman (Nobel 2023) showed that pseudouridine plus cap1 makes synthetic mRNA non-immunogenic enough for therapy.
• Cap-snatching antivirals. Baloxavir marboxil (Xofluza, 2018) inhibits influenza PA endonuclease, the cap-snatching subunit. A single oral dose shortens flu by about a day. Resistance via PA I38T arose almost immediately and is now monitored.
• eIF4E in cancer. eIF4E is overexpressed in many tumors; it preferentially translates 5'-UTR-structured mRNAs encoding cyclin D1, c-Myc, VEGF, and survivin. 4EGI-1 and ribavirin disrupt eIF4E-eIF4G binding and have been tested clinically.
• 4E-BP signaling. mTORC1 phosphorylates 4E-BP1, releasing it from eIF4E and switching on cap-dependent translation. Rapamycin and its analogs (everolimus, temsirolimus) inhibit this and dampen translation in tumors and immune cells.
• Decapping in disease. Loss-of-function mutations in DCPS (cap m7GpppN scavenger decapping) cause Al-Raqad syndrome with intellectual disability — capping pathway integrity matters for neurons.
• P-bodies and stress granules. Decapping factors concentrate in cytoplasmic P-bodies; stalled preinitiation complexes accumulate in stress granules. The dynamics of these condensates regulate which transcripts get translated under stress.

Cap variants and unusual structures

• Cap0, cap1, cap2. Differ in 2'-O-methylation status. Mammals require cap1 to evade RIG-I/IFIT1.
• m2,2,7G trimethylguanosine cap. Found on snRNAs (U1, U2, U4, U5) and some snoRNAs. Made by TGS1.
• NAD caps. NAD⁺ can prime transcription, leaving an NAD-cap; removed by DXO and Nudt12.
• FAD and CoA caps. Discovered in bacteria; possibly more widespread.
• Cap analogs in vitro. ARCA forces correct orientation; CleanCap AG produces cap1 in one step.

Pitfalls and design traps

• Reverse-orientation cap analogs. Older mCAP could go either direction during T7 transcription; 30–50% of transcripts ended up with reversed caps. ARCA fixed this with a 3'-O-methyl that blocks reverse incorporation.
• Uncapped 5'-triphosphate. Unreacted starting material is recognized by RIG-I as viral; therapeutic IVT products must be CIP-treated or HPLC-purified.
• Cap0 immunogenicity. Older mRNA platforms used cap0 and triggered interferon. Cap1 (CleanCap or vaccinia 2'-O-MTase) is essential for in vivo use.
• Decapping during prep. Some IVT kits contaminate with phosphatase. Validate by LC-MS.
• NAD-cap miscounting. Standard RNA-seq is biased toward 5'-monophosphate ends and misses NAD caps. CapZyme-seq detects them.
• eIF4E oncogenicity. Overexpressing eIF4E in mice drives lymphoma (Ruggero, Nat Med 2004) — translation initiation as a primary oncogene.