Linkers in the structural biology of protein–protein interactions

Artificial linkers fusing two partners in structural studies range from 2 to 31 amino acids, with common lengths of 5 to 11 amino acids, and the optimal length varies case by case with the system under study. In the linked complexes reviewed for the LIM domain, PXR-LBD, peptide-bound TCR-MHC, g3p-TolA, HIV-1 protease and transthyretin dimer cases, over 60% of the linker amino acids are Gly.

Linkers rich in Gly, Ser and Thr offer rotational freedom of the polypeptide backbone so adjacent domains can move relative to one another, enhanced solubility, and resistance to proteolysis. Gly has a low preference to form an alpha-helix and lacks a side chain, which maximizes backbone conformational freedom, while Ser and Thr are polar residues that prefer to interact with the solvent rather than with the fused proteins.

Natural interdomain linkers are enriched in Pro, Arg, Phe, Thr, Glu and Gln and act as rigid spacers, and proline-rich sequences form relatively rigid extended structures that prevent unfavorable interactions between domains. Flexible Gly-rich regions also occur as natural linkers that form loops connecting domains in multidomain proteins.

The natural Gly-rich linkers of some proteins have functional roles beyond joining domains. The Gly-rich linker joining the paired domain and homeodomain of PAX6 makes minor groove contacts with DNA in the human PAX6 paired domain-DNA crystal structure, and the Gly-rich linker of HTLV-1 transmembrane glycoprotein gp21 is involved in membrane fusion function according to Ala and Pro scanning mutagenesis.

In an Arc-linker-Arc single-chain fusion of the dimeric Arc repressor, a poly-Gly linker gives maximum conformational freedom but fails to give optimal stability. Maximum stability is obtained with a linker containing 11 Ala and 5 Gly residues, or 7 Ser and 9 Gly residues, respectively, in the randomized region of the linker.

Replacing the two-residue loop between helix 1 and helix 2 of the Rop homodimer with (Gly)1-10 linkers of 1 to 10 Gly residues preserves a highly helical dimer structure and wild-type RNA-binding activity in all 10 mutants. Stability of Rop toward thermal and chemical denaturation is inversely correlated with loop length, so a longer Gly loop progressively decreases stability.

Flexible polypeptide linkers of 5, 8 or 10 Gly residues placed between an epitope tag and the tagged protein increase epitope sensitivity and accessibility without interfering with folding and function of the tagged protein.

Replacing the native Type I’ beta-turn loop of staphylococcal nuclease with a five-residue turn sequence from concanavalin A leaves nuclease activity unchanged, and the crystal structure shows a well-folded hybrid in which the introduced turn keeps the conformation it has in concanavalin A.

Covalently linking an antigenic peptide to the N-terminus of the MHC class II beta1 chain through the 16-amino acid linker GGGGSLVPRGSGGGGS yields stable alpha-beta heterodimers, whereas alpha and beta chains expressed separately and mixed form high molecular weight aggregates. The linker does not affect recognition of the peptide-MHC class II complex by specific T-cell hybridomas, and its thrombin cleavage site (LVPRGS) flanked by Gly-rich residues allows the linker to be cleaved.

A six-amino acid GSGSGS linker (3xGS) tethering an ovalbumin peptide to the N-terminus of the beta1 chain of mouse MHC class II I-Ad is sufficient to enhance the stability of the alpha-beta heterodimers, and many crystallization conditions then give diffraction-quality crystals. GSGSGS also links a hemagglutinin peptide to I-Ad, a CLIP peptide to I-Ab, and a myelin basic protein peptide (124-135) to I-Au in crystallized complexes, and the linker electron density is not observed in the I-Ad-ovalbumin structure (PDB 2IAD).

A six-amino acid linker is sufficient to connect the C-terminus of an influenza virus peptide to the N-terminus of the human class II MHC protein HLA-DR1, based on the crystal structure of the unlinked HLA-DR1-peptide complex (PDB 1DLH).

LMO proteins are insoluble and unstable when expressed in bacteria, and fusing them to Ldb1 with a GGSGGHMGSGG or a GGSGGSGGSGG linker rectifies this. The Ldb1 LIM interaction domain (residues 300-340) is unstructured in solution and adopts an extended conformation along the face of the LMO2 or LMO4 LIM domain upon binding.

The LIM domains of Lhx3 and Lhx4 tend to aggregate and are largely insoluble when expressed in bacteria, and linking an interaction partner such as Isl1 or Ldb1 to the LIM domains prevents the aggregation and yields soluble tethered complexes. NMR backbone assignment shows the Gly-rich linker GGSGGHMGSGG in such a linked LIM-homeodomain complex exists as a random coil.

The C-terminus of the PXR ligand-binding domain (PXR-LBD) lies 18 Å from the N-terminus of the SRC-1 peptide, which requires at least five amino acids to link them. Fusing the SRC-1 peptide to the PXR-LBD C-terminus with the linkers SGGGSSHS, SGGSGGSSHS and SGGSGGSGGSSHS gives most crystals with the 10-amino acid Gly-rich linker, whereas a dual expression system had produced unstable proteins and an SRC-1 to PXR stoichiometry of less than 1:1.

A TCR-peptide-MHC class II ternary complex can be stabilized for crystallization either by linking the antigen peptide to the N-terminus of the MHC beta chain with a Gly-rich linker such as GGGGSLVPRGSGGGGS, or by linking the peptide to the N-terminus of the TCR variable beta chain with the eight-amino acid linker GGSGGGGG. Structures obtained both ways show no differences in the binding regions of MHC and TCR.

Linking the C-terminus of the measles virus phosphoprotein fragment P457-507 to the N-terminus of nucleocapsid protein fragment N486-505 with the Gly- and Ser-rich linker GSGSGSGS gives a crystallizable complex of two proteins that otherwise bind with weak affinity in a fast-associating reaction. The angle between the N protein helix and the final P protein helix indicates that the interaction in the crystal may be intermolecular between adjacent molecules rather than intramolecular.

If a linker joining two binding partners is not of optimal length, the linked partner engages an adjacent molecule to retain its natural interaction, so the linker length determines whether the interaction is intramolecular or intermolecular. This is seen for P457-507 and N486-505 of measles virus linked by GSGSGSGS, and for calmodulin and the neurogranin IQ motif peptide when linked.

Fusing the C-terminus of E. coli FtsY to the N-terminus of Ffh through a 31-residue Gly- and Ser-rich linker gives a single-chain SRP receptor construct (scSRP) that behaves similarly to unlinked SRP and FtsY. scSRP binds ribosomes efficiently without altering GTPase activity, which allowed its cryo-EM structure bound to the ribosome to be described.

The weak interaction between the N1 domain of phage g3p and the D3 domain of TolA is stabilized for crystallization by fusing the C-terminus of g3p N1 to the N-terminus of TolA D3 through the 18-residue Gly-rich linker GGGSEGGGSEGGGSEGGG. A long linker imposes no constraints because g3p already contains a long natural flexible linker between its N1 and N2 domains.

Linking the alpha3 helix of yeast histone H3 to Asf1 through the alanine-rich linker AAGAATAA traps their weak interaction as a stable complex for crystallography, and it is the only reviewed case in which an alanine-rich linker is used.

Linking the C-terminus of calmodulin to the N-terminus of the calcineurin calmodulin-binding domain (CBD) peptide with the five-Gly linker GGGGG facilitates co-crystallization of a CBD that is largely disordered without calmodulin. The linked structure (PDB 2F2P) is the same as the structure of calmodulin-CBD peptide co-crystallized without a linker (PDB 2R28), so the flexible Gly linker does not restrain the interaction.

A covalently linked HIV-1 protease dimer, built with the five-amino acid linker GGSSG joining the C-terminus of one monomer to the N-terminus of the other, shows enzymatic properties very similar to those of the natural dimer. The linked dimer is more stable than the natural dimer at pH 7.0, where the natural dimer dissociates.

A transthyretin (TTR) dimer tethered with the linker GSGGGTGGGSG is highly stable and structurally and functionally similar to wild-type TTR. Urea cannot denature the tethered dimer whereas guanidine hydrochloride can, which suggests the tethered complex is kinetically rather than thermodynamically stable, and tethering two subunits dramatically reduces tetramer dissociation.

A Gly-Thr (GT) dipeptide linker is sufficient to covalently link the S1 and S2 domains of the ligand-binding core of ionotropic glutamate receptors and to mimic its binding affinity for various agonists.

Interaction between linked partners in a chimera should be confirmed to occur under the same optimized conditions as for the unlinked partners, using in vitro methods such as immunoprecipitation or ITC/SPR. Flexible Gly-rich linkers in general do not alter the properties of the proteins they join and permit the natural interaction to be retained.