Crystal structure of the Escherichia coli transcription termination factor Rho

A structure is reported of the E. coli transcription terminator factor Rho that was crystallized as an impurity present in preparations of an overexpressed bacterial membrane-transport protein.


Introduction
One of the challenges in the crystallography of membrane proteins is their typically low expression level, which necessitates a significant degree of purification to separate the protein of interest from all other cellular proteins. This can consequently lead to the inadvertent purification of contaminant proteins that might otherwise be present at negligible levels when the target protein expresses at high levels. In unfortunate cases, these impurities may crystallize more readily than the target protein, leading to misplaced enthusiasm until the contaminant is recognized. As examples, the multi-drug efflux pump AcrB is a well known crystallization contaminant in membrane-protein preparations owing to its relatively high expression level during recombinant protein expressions with antibiotic selection and its nonspecific binding to Ni-NTA columns (Veesler et al., 2008;Das et al., 2007). Bacterioferritin has also been crystallized as a contaminant in preparations of cytochrome cbb 3 oxidase (Nam et al., 2010). In addition, exogenous proteins such as DNase, lysozyme and various proteases used in target protein purification have also been shown to be crystallization contaminants (Niedzialkowska et al., 2016). Compilations facilitate the identification of crystals of contaminant proteins (Hungler et al., 2016;Simpkin et al., 2018), but the crystallization of 'novel' impurities is still a concern. In this work, we report the crystal structure of a previously unreported contaminating protein, the transcription termination factor Rho from Escherichia coli, which was obtained during the structural analysis of a bacterial ATP-binding cassette (ABC) transporter.
Rho is a hexameric RNA helicase that functions in transcription termination in E. coli. The six subunits together form a ring-like structure, and the structure switches between an open-ring staircase conformation and a closed-ring conformation coupled to the binding and hydrolysis of ATP (Skordalakes & Berger, 2003;Thomsen & Berger, 2009). Here, we present the crystal structure of Rho in an open-ring staircase conformation at 3.30 Å resolution with ATP bound.

Crystallization
Rho crystallized during the crystallization trials of NaAtm1 under optimized conditions based on MemGold (Molecular Dimensions) condition No. 68 at 20 C by hanging-drop vapor diffusion. The final crystallization condition consisted of 100 mM NaCl, 100 mM Tris pH 8.3, 28% polyethylene glycol 2000 monomethyl ether (PEG 2000 MME), 0.2 M nondetergent sulfobetaine 221 (NDSB-221), 20 mM ATP pH 7.5. The crystallization sample was prepared in 1 mM ATP pH 7.5, 5 mM EDTA pH 7.5 in the presence and absence of 5 mM oxidized glutathione (GSSG) pH 7.5, which is a transport ligand for NaAtm1. Thin plate-shaped crystals appeared in about two weeks. The crystals were harvested in cryoprotectant solution consisting of 100 mM NaCl, 100 mM Tris pH 8.3, 28% PEG 2000 MME with PEG 400 at 10%, 15% and 20% before flash-cooling in liquid nitrogen. Crystallization information is summarized in Table 2.

Data collection and processing
Crystals were screened on the GM/CA beamline 23-ID-B at the Advanced Photon Source (APS) and on beamline 12-2 at the Stanford Synchrotron Radiation Laboratory. The final data set was collected on the GM/CA beamline 23-ID-B with an EIGER 16M detector (Dectris) using JBluIce-EPICS (Stepanov et al., 2011), processed and integrated with XDS (Kabsch, 2010) and scaled with AIMLESS (Evans & Murshudov, 2013). The crystals of Rho diffracted to about 3.30 Å resolution in space group C2, with unit-cell parameters a = 161.8, b = 101.9, c = 184.0 Å , = 107.8 . Data-collection and processing statistics are summarized in Table 3.

Structure solution and refinement
The self-rotation function was calculated with the CCP4 program MOLREP ( Table 1 Macromolecule-production information.
The protein was expressed in E. coli from the native promotor without the use of an expression plasmid.   Table 4.

Electron-microscopy sample preparation and data processing
The expression plasmid for the membrane-scaffolding protein MSP1D1 was purchased from Addgene (plasmid No. 20061). The expression and purification of MSP1D1 were carried out using published protocols with minor modifications (Ritchie et al., 2009). Reconstitution of NaAtm1 (and the Rho contaminant) with MSP1D1 was carried out with 1-palmitoyl-2-oleoyl-glycero-3-phosphocholine (POPC) at a 1:2:130 molar ratio of NaAtm1:MSP1D1:POPC. This reconstituted sample was incubated overnight at 4 C. After two hours of incubation, BioBeads were added at 200 mg ml À1 for detergent removal. The sample was then subjected to sizeexclusion chromatography on a Superdex 200 Increase 10/300 column (GE Healthcare). The peak fractions were pooled and concentrated to $8 mg ml À1 . EM grids were prepared using a protein concentration of 4 mg ml À1 in the presence of 5 mM GSSG and 5 mM AMPPNP. 3 ml protein sample was applied onto freshly glowdischarged QuantiFoil Cu R2/2 300 mesh grids and blotted for 4 s with a blot force of 0 and 100% humidity at room temperature using a Vitrobot Mark IV (FEI). Data were collected on a 200 keV Talos Arctica with a Falcon III detector at a magnification of 92 000 and a total dose of 81 e Å À2 at the Caltech CryoEM Facility.
Data processing was performed with cryoSPARC2 (Punjani et al., 2017), following motion correction with full-frame motion and estimation of the contrast transfer function (CTF) with CTFFIND (Rohou & Grigorieff, 2015). Particles were picked using a reconstruction of NaAtm1 as a template and extracted in cryoSPARC2. The initial 2D classification revealed a single class of Rho with $2500 particles. The particles were then 2D-classified again into five classes, with four good classes with a total of $2200 particles, as shown in Fig. 4(b).

Results and discussion
Crystals of Rho were unexpectedly obtained during studies of the bacterial ABC transporter NaAtm1 from N. aromaticivorans. NaAtm1 was recombinantly expressed in E. coli BL21-Gold (DE3) cells with a C-terminal 6ÂHis tag following a previously established protocol (Lee et al., 2014). After solubilization of the E. coli cells in DDM and C12E8, protein purification proceeded by Ni-NTA affinity purification followed by size-exclusion chromatography (Fig. 1a). SDS-PAGE gels indicated a high degree of purity, although in subsequent analysis of overloaded gels a small amount of another protein was present at a molecular weight of $40 kDa (Fig. 1b).
The crystals obtained during the crystallization optimization belonged to space group C2, with unit-cell parameters a = 161.8, b = 101.9, c = 184.0 Å , = 107.8 . The asymmetric unit volume was of sufficient size to accommodate an NaAtm1 dimer (molecular weight of 133 kDa). Analysis of the selfrotation function calculated from the diffraction data revealed a noncrystallographic symmetry (NCS) sixfold axis offset $3-4 from the crystallographic a axis. Interaction of the perpendicular twofold and sixfold axes generates a set of noncrystallographic twofold rotation operations separated by 60 in the plane perpendicular to the NCS sixfold axis (Fig. 1c). Given the unit-cell dimensions, this apparent NCS was incompatible with dimeric NaAtm1, which raised the possibility that a contaminant had crystallized. With an estimated molecular weight of $40 kDa based on the observation of a   faint impurity band in the gel and a Matthews coefficient analysis, we performed molecular replacements with known crystallization contaminants (Hungler et al., 2016;Simpkin et al., 2018), which all failed to yield a molecular-replacement solution.
The identification of Rho (molecular weight 47 kDa) was established by a mass-spectrometric analysis of the peptides prepared by trypsin digestion of the protein in the SDS-PAGE bands. Using this information, we were able to obtain a molecular-replacement solution using Rho in the AMPPNPbound state (PDB entry 1pvo; Skordalakes & Berger, 2003) as the input model. The molecular-replacement results established that in these crystals Rho adopts a six-membered broken-staircase conformation. The structure was refined to an R work and R free of 25.2% and 29.6%, respectively ( Table 4).
The electron-density map also revealed ATP to be bound in all six subunits (Fig. 2a).
The individual Rho subunits are structurally similar overall, except for the terminal subunit at the 'break' in the staircase, where the first 50 residues at the N-terminus exhibited a shift of 2-8 Å relative to the other monomers (Fig. 2b). The r.m.s.d. between the present structure and the molecular-replacement input model (PDB entry 1pvo) is 2.5 Å using the C positions for the superposition and 3.3 Å when using all atoms of the six subunits in the hexamer. The r.m.s.d.s between individual subunits in the present structure and PDB entry 1pvo are 0.59-0.73 Å , reflecting their similar tertiary structures. The relationship between adjacent subunits in the broken staircase of the present structure may be approximated by a screw operation with a rotation around the screw axis of 60. subunit and a corresonding translation along the axis of À7.7 Å . In comparison, the corresponding values for PDB entry 1pvo are 57.9 and À8.2 Å , respectively. Reflecting the larger rotation angle per subunit, the present structure exhibits a more closed ring in comparison to the original brokenstaircase structure (Figs. 3a and 3b). At the level of subunitsubunit interactions, however, the differences are subtle and we could not identify the specific interactions responsible for these differences in hexamer structure. Of note, the nucleotidebinding site is at the interface between subunits and small changes in subunit-subunit interactions may reflect the presence of different nucleotides: ATP in this structure and either no nucleotide (apo), ATPS or AMPPNP in other open-ring structures (Skordalakes et al., 2005;Skordalakes & Berger, 2003).
How did Rho end up in our crystallization conditions? Our hypothesis is that during the overexpression of proteins the E. coli transcription and translation machineries are highly expressed for the production of mRNAs and recombinant proteins, respectively. Rho, as the termination factor, would plausibly be overexpressed as part of the transcriptiontermination process; thus, it is likely that Rho is a general contaminant and does not arise specifically from the overexpression of NaAtm1. The fact that Rho eluted from the Ni-NTA column along with His-tagged NaAtm1 suggests that there is nonspecific binding to the Ni 2+ resin by the histidine residues distributed throughout the whole protein (Fig. 3c). The open-ring conformation that Rho adopts in solution (Thomsen et al., 2016) with a molecular weight of 282 kDa for the hexamer apparently has a comparable hydrodynamic research communications radius to NaAtm1, with a dimer molecular weight of 133 kDa in addition to the detergent micelle. Given the low abundance, the presence of Rho in the SEC fractions was only detected in hindsight. Also in hindsight, Rho was not present in the original NaAtm1 purification, which included a membraneisolation step in which Rho was presumably removed (Lee et al., 2014); in the present work the membrane-isolation step was omitted and Rho subsequently copurified with NaAtm1.
We have also observed Rho in single-particle cryoEM studies of NaAtm1 reconstituted in membrane-scaffolding protein (MSP) nanodiscs. The NaAtm1 nanodisc sample was prepared by incubating detergent-purified NaAtm1 with MSPs and lipids, and was further purified by size-exclusion chromatography. Similar to the purification in detergent, the peak fractions were collected for single-particle cryoEM sample preparation (Fig. 4a). The 2D classification reported one class of Rho in the broken hexameric state (Fig. 4b), again suggesting that Rho has a similar hydrodynamic radius and plausibly a similar molecular weight to our reconstituted NaAtm1 in nanodiscs.
In a structural analysis of a prokaryotic chloride channel, a single peptide of Rho was identified in a mass-spectrometric analysis of the gel band (Abeyrathne & Grigorieff, 2017), representing the first time, to our knowledge, that Rho has been identified as a possible contaminant during membraneprotein expression. As demonstrated in this report, Rho can research communications Acta Cryst. (2020). F76, 398-405 Fan & Rees Transcription termination factor Rho 403 Figure 3 Overall architecture of Rho. Overall structural representations of (a) ATP-bound Rho (this study) and (b) the AMPPNP-bound structure of Rho (PDB entry 1pvo; Skordalakes & Berger, 2003). (c) Distribution of surface histidine residues (blue) in the ATP-bound structure of Rho (this work). The spacings between the surface-exposed histidines are several nanometres and are comparable to the loading density of His-tagged proteins bound to Ni-NTA beads (Hayworth & Hermanson, 2014), which presumably allows multiple binding sites to Ni-NTA and contributes to the observed affinity of Rho for Ni-NTA resin.
crystallize even in the presence of a large excess of other proteins, and thus it should be added to the list of known contaminant proteins in crystallography.