Biochemical Society Transactions

BioScience2005

Crystallographic and single-particle analyses of native- and nucleotide-bound forms of the cystic fibrosis transmembrane conductance regulator (CFTR) protein

N.H. Awayn, M.F. Rosenberg, A.B. Kamis, L.A. Aleksandrov, J.R. Riordan, R.C. Ford

Abstract

Cystic fibrosis, one of the major human inherited diseases, is caused by defects in the CFTR (cystic fibrosis transmembrane conductance regulator), a cell-membrane protein. CFTR acts as a chloride channel which can be opened by ATP. Low-resolution structural studies of purified recombinant human CFTR are described in the present paper. Localization of the C-terminal decahistidine tag in CFTR was achieved by Ni2+-nitriloacetate nanogold labelling, followed by electron microscopy and single-particle analysis. The presence of the gold label appears to improve the single-particle-alignment procedure. Projection structures of CFTR from two-dimensional crystals analysed by electron crystallography displayed two alternative conformational states in the presence of nucleotide and nanogold, but only one form of the protein was observed in the quiescent (nucleotide-free) state.

  • ATP-binding cassette (ABC)
  • cystic fibrosis transmembrane conductance regulator (CFTR)
  • crystallization
  • membrane protein
  • single-particle analysis
  • three-dimensional structure

The ABC (ATP-binding cassette) proteins are a superfamily of active transporter membrane proteins and are composed of approx. 50 functionally diverse classes of prokaryotic and eukaryotic transmembrane proteins [1]. Despite the differences in the type of substances that are transported by each protein, most members of the ABC proteins have a similar pattern of two NBDs (nucleotide-binding domains) and two TMDs (transmembrane domains) [2]. The CFTR (cystic fibrosis transmembrane conductance regulator), also termed ABCC7, is a unique member of the ABC superfamily of transporter proteins, which works as an ion channel and has an extra domain called the regulatory domain (R) located between NBD1 and TMD2 [3]. Channel activity is affected by phosphorylation (by protein kinases A and C) and by ATP. CFTR regulates secretion and reabsorption of ions at epithelial surfaces [4]. Defects in the channel function and/or folding and processing lead to cystic fibrosis [5].

Structural data for many ABC transporter proteins suspected to be responsible for human diseases are mostly lacking. Structural studies of ABC proteins were successful in determining the high-resolution structures of MsbA (bacterial lipid transporter) [6,7] and BtuCD (bacterial vitamin B12 transporter) [8] and the low-resolution structures of YvcC (Bacillus subtilis multidrug protein) [9], Pgp (P-glycoprotein) and MRP1 (multidrug resistance protein 1) [11]. Recently, Rosenberg et al. [12] purified wild-type human CFTR and determined its low-resolution three-dimensional structure in the presence of p[NH]ppA (adenosine 5′-[β,γ-imido]triphosphate); and Lewis et al. [13] determined the high-resolution structure of the first NBD1 (nucleotide-binding domain 1) from mouse CFTR. Here, we describe our structural analysis of CFTR protein under different conditions using SPA (single-particle analysis) and electron crystallography methods. Expression, purification and functional assays, including ATPase assays, nucleotide-binding assays and channel gating activity measurement of the CFTR protein, have been described recently [12].

SPA of CFTR under different conditions

The single-particle technique provides a rapid three-dimensional reconstruction methodology for proteins even with very low levels of protein. SPA has been used to reconstruct the three-dimensional structure of several ABC proteins [9,10,14,15]: two reports [10,14] showed data that were interpreted as monomeric particles, whereas the other two reports [9,15] reported particles that were identified as dimeric aggregates. Here, three different low-resolution (21–35 Å; 1 Å=10−10 m) three-dimensional reconstructions of CFTR particles (2000–4000 particles) have been produced.

The first three-dimensional density map was generated from nucleotide-free, detergent-solubilized CFTR particles (Figures 1A and 1B) with no symmetry imposed. The resolution of this structure is relatively limited (∼35 Å) as determined by FSC (Fourier shell correlation) [16] between two subsets of the data, and the resulting structure shows little detail.

Figure 1 Three-dimensional density maps generated by SPA of CFTR under three different conditions

(A) Purple netting shows the low-resolution (35 Å) three-dimensional reconstruction of nucleotide-free CFTR. (B) is related to (A) by a 90° rotation around the vertical axis. (C) Red netting shows the 25 Å resolution three-dimensional reconstruction of Ni-NTA nanogold-bound CFTR and blue netting shows the Ni-NTA nanogold. (D) is related to (C) by a 90° rotation around the vertical axis. (E) Red netting shows a 21 Å resolution three-dimensional map of Ni-NTA nanogold-bound CFTR in the presence of p[NH]ppA nucleotide and blue netting shows the Ni-NTA nanogold. (F) is related to (C) by a 90° rotation around the vertical axis. Scale bar, 5 nm.

The second three-dimensional density map was generated from Ni-NTA (Ni2+-nitrilotriacetate) nanogold-bound CFTR particles (Figures 1C and 1D) and it shows that the Ni-NTA nanogold binds at a unique site in the CFTR particle, shown by the blue netting. Since it is likely that the Ni-NTA nanogold binds to the decahistidine tag located at the C-terminal end of the protein [12], this makes a discrimination of the TMD and NBD regions possible since the C-terminus is at the end of NBD2. The resolution of this structure is at approx. 25 Å, determined as above by FSC.

The third three-dimensional density map was generated from Ni-NTA nanogold-bound CFTR particles in the presence of p[NH]ppA nucleotide (Figures 1E and 1F) and also shows that the Ni-NTA nanogold binds to a unique site in the CFTR particle shown by the blue netting. The resolution of this structure is approx. 21 Å (Figure 2A). Compared with the second map (Figures 1C and 1D), some differences are observed in the Ni-NTA nanogold binding position, which is less buried/more at the surface in this structure. These data suggest that the p[NH]ppA nucleotide may make subtle changes in the CFTR structure, in agreement with single-channel recordings [12].

Figure 2 Resolution and model fitting

(A) An example of plots of FSC for refinement iterations in the generation of the three-dimensional reconstruction for Ni-NTA nanogold-bound CFTR in the presence of p[NH]ppA nucleotide. The thick black line shows the FSC calculated between two subsets of the data and the intercepting blue lines represent the resolution (Å) estimated at FSC=0.5. (B) The three-dimensional reconstruction of Ni-NTA nanogold-bound CFTR in the presence of p[NH]ppA nucleotide red (netting), and the insertion of two copies of a homology model of CFTR (white wireframe trace). The Ni-NTA nanogold location (blue netting) implies that the C-terminus associates closely with NBD2. Scale bar, 50 nm.

These findings show similarities between the second and third maps (independently) generated with Ni-NTA nanogold, which supports the reliability of the work and that all maps come from the same molecule (CFTR). Furthermore, the use of Ni-NTA nanogold appears to improve the resolution of the SPA method, perhaps by providing a high-contrast reference point in the analysis, thus improving the alignment. Insertion of a homology model based on Vibrio cholerae MsbA for CFTR (J. Campbell and M.S.P. Sansom, personal communication) into the maps (Figure 2B) showed that two putative CFTR molecules could be accommodated within either of the higher-resolution maps. This also shows that the SPA methodology, when combined with Ni-NTA nanogold labelling, can generate low-resolution three-dimensional structures that are consistent with the expected dimensions of CFTR.

Crystallographic analysis of nucleotide-free and Ni-NTA nanogold-bound CFTR

Two-dimensional crystals of nucleotide-free and Ni-NTA nanogold-bound CFTR were grown directly on a carbon-coated electron microscopy grid using a hanging droplet method [12,1719]. The crystals were negatively stained with 4% (w/v) uranyl acetate for this analysis, yielding low-resolution data to approx. 20 Å. Crystals that were one molecular layer thick and well ordered were chosen for structural analysis for each condition (Figure 3A).

Figure 3 Electron crystallography of CFTR

(A) An area of a negatively stained two-dimensional crystal of nucleotide-free CFTR. (B) Projection map of nucleotide-free CFTR and its unit cell parameters (below). (C) Projection map of crystal form 1 crystals of CFTR grown in the presence of Ni-NTA nanogold and its unit cell parameters (below). (D) Projection map of crystal form 2 crystals of CFTR grown in the presence of Ni-NTA nanogold and its unit cell parameters (below).

Nucleotide-free CFTR

The preliminary analysis of nucleotide-free CFTR two-dimensional crystals produced one crystal form with a projection structure map that was roughly hexagonal in shape (Figure 3B), and this map showed a high degree of similarity to a map of crystal form 1 published recently for CFTR in the presence of p[NH]ppA nucleotide [12].

Ni-NTA nanogold-bound CFTR

The analysis of the two-dimensional crystals from Ni-NTA nanogold-bound CFTR resulted in two crystal forms with projection structure maps shown in Figures 3(C) and 3(D). The projection maps that resulted here were nearly identical with the two projection maps derived from two crystal forms of CFTR grown in the presence of p[NH]ppA nucleotide [12]. Comparisons between all the CFTR projection maps show a degree of similarity between the projection map of nucleotide-free CFTR, the projection map of crystal form 1 of Ni-NTA nanogold-bound CFTR and the projection map of crystal form 1 from CFTR in the presence of p[NH]ppA nucleotide [12]. All these maps show a molecular outline of roughly hexagonal shape. The projection map of crystal form 2 from Ni-NTA nanogold-bound CFTR shows some degree of similarity to the projection map of crystal form 2 from CFTR in the presence of p[NH]ppA nucleotide [12]. These show a molecular outline of roughly triangular shape and the projection map of crystal form 2 from Ni-NTA nanogold-bound CFTR also show a distinct low-density area in the middle.

Conclusions

Previous SPA studies showed dimeric [9,15] or monomeric [10,14] associations of intact transporters. Although studies based on three- and two-dimensional crystals mostly reveal monomeric structures, there are two reports of dimeric association within the crystal packing [11,20]. In the present study, we present SPA data for CFTR that is consistent with a dimeric association in terms of the size of the particle; on the other hand, the labelling with a single Ni-NTA nanogold sphere (rather than two) may point to non-equivalence or asymmetry. Interestingly, the Ni-NTA nanogold label appears to improve the efficiency of the single-particle approach. Two-dimensional crystals of CFTR are formed by monomeric CFTR particles, and display two distinct crystal forms under certain experimental conditions. This behaviour may be a reflection of different, stable conformational states in the channel.

Footnotes

  • Mechanistic and Functional Studies of Proteins: A Focus Topic at BioScience2005, held at SECC Glasgow, U.K., 17–21 July 2005. Edited by S. Crosthwaite (Manchester, U.K.), M. Ginger (Oxford, U.K.), K. Gull (Oxford, U.K.), A. Lee (Southampton, U.K.), H. McWatters (Oxford, U.K.), J. Mottram (Glasgow, U.K.), P. Rich (University College London, U.K.), C. Robinson (Warwick, U.K.) and H. van Veen (Cambridge, U.K.).

Abbreviations: ABC, ATP-binding cassette; p[NH]ppA, adenosine 5′-[β,γ-imido]triphosphate; CFTR, cystic fibrosis transmembrane conductance regulator; FSC, Fourier shell correlation; NBD, nucleotide-binding domain; Ni-NTA, Ni2+-nitrilotriacetate; SPA, single-particle analysis; TMD, transmembrane domain

References

View Abstract