Date Published: October 01, 2017
Publisher: International Union of Crystallography
Author(s): Simon Boje Hansen, Nick Stub Laursen, Gregers Rom Andersen, Kasper R. Andersen.
Nanobodies are used as crystallization chaperones and here site-specific mercury labelling of nanobodies is shown as a new tool for phasing.
The production of well diffracting protein crystals is a major challenge in macromolecular X-ray crystallography. Large multi-domain proteins and membrane proteins are inherently difficult to crystallize owing to conformational heterogeneity and the lack of suitable surface chemistry that allows the formation of a crystal lattice. Crystallization chaperones are auxiliary proteins that increase the chance of crystallization by reducing conformational flexibility and providing well ordered surfaces to form crystal lattice contacts. Monoclonal antibody Fab fragments derived from IgG are the most widely used chaperones (Uysal et al., 2009 ▸) and may simultaneously provide phase information for structure determination by molecular replacement. Several alternative chaperones have been developed, including DARPins, single-chain variable fragments and nanobodies (Nbs; Pardon et al., 2014 ▸). Nbs are derived from natural llama heavy-chain antibodies that are devoid of a light chain and in which the heavy-chain variable domain (VHH) exclusively mediates the interaction with the antigen (Muyldermans, 2013 ▸). The VHH domain is structurally similar to the IgG VH domain, with three complementary-determining regions (CDRs) that are responsible for antigen binding. In contrast to Fab fragments, which must be expressed in mammalian or insect cells, Nbs are easy to express and manipulate in Escherichia coli. As a result of their favourable characteristics Nbs are also increasingly being used in imaging, where they can be labelled with GFP or fluorescent dyes (Chakravarty et al., 2014 ▸; Rothbauer et al., 2006 ▸). Covalent attachment of fluorescent dyes to Nbs has proven to be effective using NHS ester, isothiocyanate or maleimide functional groups, with maleimide labelling being superior to NHS ester dyes when comparing background staining in cell-permeabilizing imaging (Pleiner et al., 2015 ▸; Röder et al., 2017 ▸). NHS esters and isothiocyanates react readily with N-terminal and lysine amines, while maleimide reacts specifically with cysteine thiols in the pH range 6.5–7.5.
It was our expectation that the crystallization of a monomeric PCMB-derivatized Nb would reveal structures with a Hg atom inserted between the benzoate moiety and the side chain of Cys85. In the two PCMB structures we have modelled a total of 11 Hg atoms; four of these were clearly bridging two cysteines, but the remaining seven did not display density that could be attributed to the benzoate fragment. This suggests a high tendency for benzoate release, possibly by a mechanism reminiscent of the protonolysis of organomercurial compounds catalysed by the MerB enzyme, which can accelerate protonolysis by up to 107 (Parks et al., 2009 ▸). Models of the MerB reaction mechanism suggest that the release of the organic substituent is catalysed through its protonation by an aspartic side chain that acts as a proton shuttle during transfer of the proton from one of the reacting cysteine –SH groups to the C atom bound to Hg (Parks et al., 2009 ▸), here the benzoate C4 atom. Whether this mechanism of benzoate release is relevant in our Nb36-C85 PCMB crystallization experiments cannot be decided upon, especially since we do not have a structure of the intermediate from which the benzoate is released. There are several deposited structures in the PDB containing intact PCMB (residue identifier MBO) in which the CysS–Hg–benzoate substructure is modelled. Intriguingly, there are also two entries, PDB entries 5ec5 (Podobnik et al., 2016 ▸) and 1naq (Arnesano et al., 2003 ▸), in which PCMB has been used in which some of the Hg sites have the benzoate modelled while other sites have only the Hg bound to the cysteine. Hence, the release of benzoate observed in our structures appears to be a general feature of this reagent. Although unintentional, it may have given rise to new crystal forms as our crystals of nonderivatized Nb36 exhibit C2 symmetry in contrast to P21 or P212121 symmetry for the PCMB-derivatized Nb, and in both cases there are Hg-bridged Nb dimers. Methane and ethane are released very slowly from organomercurials by protonolysis (Begley et al., 1986 ▸), suggesting that the use of mercury compounds such as CH3HgCl and CH3CH2HgCl may minimize the observed unintentional liberation of the organic group during crystallogenesis and storage.