CsgG

CsgG is a protein of the outer membrane of Escherichia coli and related bacteria. Nine copies assemble into a channel through which the cell pushes out CsgA, the subunit of curli, the amyloid fibers that hold bacterial biofilms together. In 2016 Oxford Nanopore Technologies introduced a pore developed from CsgG as the pore of its R9 chemistry, and its descendants still read DNA in the company's sequencers.^[1]

Curli were described in 1989 as surface structures of E. coli that bind the host protein fibronectin.^[2] In 2002 biochemical and imaging experiments showed that they are amyloid, the class of fiber associated with Alzheimer's and prion diseases. Unlike amyloid in those diseases, the authors wrote, curli formation is "a productive pathway requiring a specific assembly machinery".^[3] The genes lie in two operons, one holding the fiber proteins CsgB and CsgA and the other csgDEFG. CsgA is the major subunit, and CsgE, CsgF and CsgG are non-structural proteins that help build it.^[4][5]

CsgG is a lipoprotein. It is made as a chain of 277 residues, loses a signal peptide of 15, and its new N-terminal cysteine is lipidated, which anchors it in the membrane.^[6] A stretch of 22 amino acids at the start of CsgA is enough to send other proteins to the channel. Overproducing CsgG makes the bacterium susceptible to the antibiotic erythromycin.^[5]

The structure was published in 2014. Nine units that resemble anticodon-binding domains form a 36-stranded β-barrel that crosses the membrane and opens onto a cage-shaped vestibule in the periplasm. Between the two lies a constriction 0.9 nm wide, made of three stacked rings of phenylalanine, asparagine and tyrosine, and the authors suggested that it guides the unfolded polypeptide through.^[7] The two structures, released together on September 24, 2014, are a non-lipidated, soluble pre-pore form (PDB 4UV2, resolution 2.8 Å) and the membrane-bound form (4UV3, 3.59 Å).^[8] The channel has no gate of its own. The authors called it "ungated" and "non-selective", and the cell's control is a second protein: CsgE assembles into a nonamer that binds the periplasmic face and closes it.^[7]

For comparison, alpha-hemolysin has seven subunits and a narrowest point of 1.4 nm, and MspA has eight, with a constriction no more than 0.6 nm long. CsgG has the narrowest opening of the three, and its barrel has the most strands.

Oxford Nanopore announced in May 2016 that the pore of its R9 chemistry was CsgG, licensed from a laboratory at VIB in Belgium and from University College London, and that sequencing speed would rise from about 70 to about 250 bases a second per pore.^[1] The structure had been solved by the laboratory of Han Remaut at VIB, and the two organizations announced a collaboration in March 2016.^[9]

The natural constriction proved useful twice. A 2020 paper combined CsgG with the 35-residue N-terminal region of CsgF, its partner protein outside the cell. The first 33 residues of CsgF bind inside the barrel and form a second constriction about 25 Å from the first, and a prototype pore with two constrictions improved single-read accuracy by 25 to 70% in runs of identical bases up to nine long.^[10] In 2024 an unfoldase, ClpX, moved protein strands hundreds of residues long through CsgG on a commercial sensor array in steps of two residues (see Nanopore protein sequencing).^[11]

Molecular Reality Corporation builds solid-state nanopores, but its atlas of the field, The Hole World, draws two of its 2,112 entries with the CsgG archetype, a barrel of nine subunits, and one of them is labeled "CsgG + motor" and belongs to Oxford Nanopore. Alpha-hemolysin is drawn 33 times.^[12] The company's quiz game JeoPOREdy! has a clue about Oxford Nanopore that says its "commercial pores are now built from a mutant of the curli transporter CsgG".^[13]

• Biological nanopore
• Protein Data Bank
• History of nanopore sequencing