Structures of the OmpF porin crystallized in the presence of foscholine-12

Abstract

The endogenous Escherichia coli porin OmpF was crystallized as an accidental by-product of our efforts to express, purify, and crystallize the E. coli integral membrane protein KdpD in the presence of foscholine-12 (FC12). FC12 is widely used in membrane protein studies, but no crystal structure of a protein that was both purified and crystallized with this detergent has been reported in the Protein Data Bank. Crystallization screening for KdpD yielded two different crystals of contaminating protein OmpF. Here, we report two OmpF structures, the first membrane protein crystal structures for which extraction, purification, and crystallization were done exclusively with FC12. The first structure was refined in space group P21 with cell parameters a = 136.7 Å, b = 210.5 Å, c = 137 Å, and β = 100.5°, and the resolution of 3.8 Å. The second structure was solved at the resolution of 4.4 Å and was refined in the P321 space group, with unit cell parameters a = 215.5 Å, b = 215.5 Å, c = 137.5 Å, and γ = 120°. Both crystal forms show novel crystal packing, in which the building block is a tetrahedral arrangement of four trimers. Additionally, we discuss the use of FC12 for membrane protein crystallization and structure determination, as well as the problem of the OmpF contamination for membrane proteins overexpressed in E. coli.

Introduction

The outer membrane of gram-negative bacteria protects from harsh environmental conditions and toxic molecules such as antibiotics, proteases, and heavy metals. It is comprised of complex lipopolysccharides and a relatively small number of proteins which all form β-barrel structures. The majority of these proteins form hydrophilic pores that allow small nutrients and signaling molecules to permeate and reach the cell, and are thus called porins. In Escherichia coli, the major porins are the outer membrane proteins (OMP) OmpF, OmpC, and OmpE (also known as PhoE). The three proteins are highly homologous, with 62% sequence identity between OmpF and OmpE, and 58% between OmpF and OmpC (mature forms). OmpF and OmpE were among the first membrane protein structures determined,¹ following the determination of the photosynthetic reaction centers of Rhodospeudomonas viridis² and Rhodobacter sphaeroides,³ and the porin from Rhodobacter capsulatus.⁴ All general porin monomers form 16-stranded β-barrels that associate in tightly packed trimers.^1,5 Their unusual stability, explained by the robustness of their structure, has been critical for their crystallization and subsequent structure determination. OmpF protein is one of the best studied membrane proteins up to now (Table I). The mature protein is comprised of 340 amino acids, and forms aqueous, voltage-gated channels that span the outer membrane and allow the diffusion of small polar molecules. Eight short β-hairpins on the periplasmic side, and eight long irregular loops (L1–L8) on the extracellular side connect the antiparallel β-strands.¹ L3 is particularly long, consisting of 33 residues and folds inside the pore, restricting its accessibility. The pore dimensions at this narrowing are about 7 × 11 Å, allowing the passage of molecules <600 Da, with slight preference for those with a positive charge.

Table I.

Structurally Characterized OmpF or OmpF Mutants

PDB code	Resolution (Å)	Space group	Chains/AU	Crystallization detergent	Mutation/Complex	References
1HXU	3	P 3 2 1	1	octyl-POE	V18K, G131K	Ref. 6
1HXT	2.4	P 3 2 1	1	octyl-POE	R42A, R82A, D113N, E117Q, R132A	Ref. 6
1HXX	2.2	P 3 2 1	1	octyl-POE	Y106F	Ref. 6
1BT9	3	P 3 2 1	1	octyl-POE	D74A	Ref. 7
1MPF	3	P 3 2 1	1	octyl-POE	G119D	Ref. 8
2OMF	2.4	P 3 2 1	1	octyl-POE		Ref. 1
1GFM	3.5	P 3 2 1	1	octyl-POE	D113G	Ref. 9
1GFN	3.1	P 3 2 1	1	octyl-POE	DEL(109-114)	Ref. 9
1GFO	3.3	P 3 2 1	1	octyl-POE	R132P	Ref. 9
1GFP	2.7	P 3 2 1	1	octyl-POE	R42C	Ref. 9
1GFQ	2.8	P 3 2 1	1	octyl-POE	R82C	Ref. 9
2ZFG	1.6	P 3 2 1	1	octyl-POE		Ref. 10
2ZLD	3	P 63	2	octyl-POE	Complex with N-terminal segment of colicin E3	Ref. 10
1OPF	3.2	P 42	6	β-OG, (C10DAO)		Refs.11,12
3FYX	3.4	P 3 2 1	1	octyl-POE	K16C, chemically modified	Ref. 13
3K1B	4.4	P 3 2 1	4	foscholine-12		This work
3K19	3.8	P 21	12	foscholine-12		This work

Foscholine detergents are lipid-like zwitterionic molecules, widely used in the investigation of membrane proteins. They have the same head group as phospholipids, but a single hydrophobic tail, which leads to the formation of micelles rather than bilayers. This is the main reason why these detergents are extremely efficient in E. coli membrane solubilization,^14,15 and foscholine-12 (FC12) with 12 methylene hydrophobic groups is one of the most effective membrane solubilization agents. FC12 has also been extremely useful and popular in NMR studies of membrane proteins aiming to obtain high-resolution spectra in mixed micelles,¹⁶ and has been shown to play a crucial role in refolding misfolded membrane proteins.¹⁷ When it comes to membrane protein crystallization, octyl glucoside (OG) remains the most extensively used detergent, with lauryldimethylamine oxide (LDAO) following closely.¹⁶ Although crystallization of a membrane protein in the presence of FC12 has been reported,¹⁶ there is no crystal structure of a protein that would be both purified and crystallized with FC12. Therefore, it is believed that FC12 is of limited use for membrane protein crystallography. However, a recent analysis of 43 eukaryotic membrane proteins expressed in Saccharomyces cerevisiae, showed that 70% of the well-expressed membrane proteins were stable, targeted to the correct organelle, and monodisperse in either FC12 or n-dodecyl-β-d-maltoside.¹⁸ Refuting the negative opinion about the use of FC12 in membrane protein crystallography we present here two different crystal forms of the structure of the OmpF porin, for which the protein was exclusively treated with FC12 throughout the extraction, purification, and crystallization processes. We also report two new modes of crystal packing of OmpF in the presence of FC12.

Results and Discussion

OmpF identification and structure solution

In the process of crystallization of the E. coli histidine kinase receptor KdpD, we obtained small crystals (about 50 μm) of different morphology [Fig. 1(A,B)] that diffracted to about 10–15 Å. Crystals have been further optimized to diffract X-rays to 3.5–5 Å. Our efforts to solve the structure by molecular replacement using the crystal structure of the N-terminal region (19–243) of KdpD from Pseudomonas syringae pv. tomato str. DC3000 [Protein Data Bank (PDB) code: 2R8R] were futile. Although the protein preparation looked very clean on sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS–PAGE) [Fig. 1(C,D)], we suspected contamination from the endogenous E. coli porin OmpF since this protein is known to express at high levels during overexpression of membrane proteins in E. coli,¹⁹ including histidine kinase receptors [Fig. 2(A) and Ref. (20)]. N-terminal amino acid sequencing had previously confirmed that the highly expressed protein is mature OmpF (N terminus at amino acid 29). The N-terminal sequencing reading was long enough (12 residues: AEIYNKDGNKVD) to ensure that the protein is OmpF and not some other closely related Omp protein. To identify the crystallized protein, we carefully washed the crystals and ran a silver stain gel that showed a protein of about ∼38 kDa [Fig. 2(B)], compatible with the expected OmpF monomer size. Molecular replacement using the protein sequence for the E. coli BL21(DE3) strain OmpF (Uniprot C6EI53) [which is identical to the sequence for the E. coli K12 strain OmpF (Uniprot P02931)] did indeed give a clear solution, with MR scores 23.13 for the P21 crystal form, and 18.91 for the P321 crystal form.

Figure 1.

Purification and crystallization of KdpD protein. (A) KdpD-M and (B) KdpD-NM crystals. Both were proved to be OmpF. (C) KdpD-M and (D) KdpD-NM fractions after size exclusion chromatography, used for crystallization trials.

Figure 2.

(A) OmpF expression in BL21(DE3) cells. NuPAGE (4–12%) Bis-Tris SDS gel. Lane 1: See blue plus 2 protein standard (Invitrogen). Lanes 2 and 4: Cells before induction of recombinant protein expression. Lanes 3 and 5: Cells after induction of recombinant membrane protein expression, and after induction of the same membrane protein in fusion with Mistic, respectively. Lane 6: Control BL21(DE3) cells grown in parallel with the transformed and induced cells. Sample taken at the same time point as samples in lanes 2 and 4. Lane 7: Control BL21(DE3) cells grown in parallel with the transformed and induced cells. Sample taken at the same time point as samples in lanes 3 and 5. Lane 8: Control BL21(DE3) whole cell lysate. Lane 9: Control BL21(DE3) supernatant after centrifugation for 1 h at 100,000g. In black boxes are the expressed target proteins. In the brown box the highly expressed OmpF porin is highlighted. (B) SDS–PAGE of OmpF, silver stained. The molecular weigh of the OmpF monomer is about 37 kDa.

Crystal packing of OmpF prepared in FC12

OmpF crystallized in two different space groups, P21 and P321, with crystal packings not observed previously. The most common packing observed for 14 of 15 previously observed OmpF structures is the linear stacking of OmpF trimers, with neighboring columns running in opposite directions. This arrangement is observed in all 13 P321 space groups reported so far, and in the P63 space group in the complex of OmpF with the N-terminal segment of colicin E3 (Table I, Fig. 3). In contrast, the building block of the crystal packing observed in this study is a tetrahedral arrangement of OmpF trimers for both space groups (Fig. 4). The tetrahedral arrangement was observed in space group P42 previously reported.^11,12 In that case crystals were grown from solutions containing 0.9% β-OG and 0.09% C8E6-11. The asymmetric unit contains two trimers, and two pairs of trimers from adjacent asymmetric units form a tetrahedron. In our structures the same tetrahedron is the building block for both packings: in the P21 crystal the tetrahedron is formed by the four tetrahedrally arranged trimers of the asymmetric unit, while in the P321 crystal the asymmetric unit consists of a trimer and a monomer, and three adjacent asymmetric units assemble into one tetrahedron identical to the one observed in the P21 space group. Thus, the same tetrahedron, arranged so that the rough end of each monomer is oriented towards the center of the tetrahedron, is the building unit of all three crystal packings mentioned above (P42, P21, and P321). Remarkably, each tetrahedron forms four contacts in the crystal lattices of these three space groups. There are two possible types of contacts observed so far: corner-to-corner contact, made between the smooth ends of adjacent tetrahedra [Fig. 5(A)] and detergent mediated corner-to-center contact with the smooth end of one tetrahedron oriented towards the center of the neighboring one [Fig. 5(B)]. In the previously reported P42 crystal form,¹¹ each tetrahedron forms identical crystallographically related lattice contacts at its four corners [Fig. 6(A)]. In the structures we report here, each tetrahedron forms two corner-to-corner and two corner-to-center lattice contacts in the P21 crystal form [Fig. 6(B)], or three corner-to-corner and one corner-to-center contacts in the P321 crystal form [Fig. 6(C)], resulting in dramatically different packing arrangements. This difference in crystal contacts must be due to the effects of crystallization conditions, such as ionic strength and pH, on detergent-induced surface properties.

Figure 3.

Previously reported crystal packing in the common P321 crystal form. Linear stacking of OmpF trimers, with neighboring columns running in opposite directions, and 90° rotation. PDB code: 2OMF (1).

Figure 4.

The tetrahedron arrangement: (A) Four trimers (shown here in different colors) form a tetrahedron. (B) The tetrahedron in the P21 structure reported here (PDB code: 3K19). (C) The tetrahedron in the P321 structure reported here (PDB code: 3K1B). In (B) and (C) the contents of the asymmetric unit are represented in different colors. In (C), trimers forming one arm of the tetrahedron and one monomer of the arm of a neighboring trimer form the asymmetric unit.

Figure 5.

The two types of contacts between tetrahedra in the crystal packings (P21 and P321) reported here. (A) Corner-to-corner contact involving the edges of two adjacent tetrahedra. (B) Corner-to-center contact with the smooth end of one tetrahedron oriented toward the center of the neighboring one, where the rough ends of all four trimers are clustered.

Figure 6.

Tetrahedron contacts. (A) The P42 crystal form (PDB code: 11PF) (11) and 90° rotation. (B) The P21 crystal form (PDB code: 3K19) and 90° rotation. (C) The P321 crystal form (PDB code: 3K1B) and 90° rotation. The asymmetric unit contents are shown in different colors. In (A), the asymmetric unit content is a pair of trimers, in (B) it is a tetrahedron (two pairs of trimers), and in (C) it is one trimer forming one arm of the tetrahedral unit plus one monomer of the neighboring arm of the tetrahedron. Neighboring tetrahedra are shown in white.

In general, the impact of the type of detergent on crystal packing can be very powerful. The widely used N-octyl-polyoxyethylene (octyl-POE) (Table I and Refs.^1,6,13), results in linear stacking of OmpF trimers, whereas β-OG and FC12 give rise to tetrahedral arrangements of trimers, which shows that FC12 can work successfully in membrane protein crystallization, just like the detergent β-OG. To our knowledge, the reported structures are the first membrane protein structures for which all steps of extraction, purification and crystallization were performed in the presence of FC12, a very potent detergent in membrane protein studies, especially when an E. coli expression system is used. According to the Membrane PDB,¹⁶ two membrane proteins have been crystallized in the presence of FC12, but one of them was extracted and purified in a different detergent^21,22 and the second was crystallized in a mixture of two detergents.²³ In particular, the zinc transporter YiiP (PDB code: 2QFI) was crystallized in the presence of FC12 in the reservoir solution,²¹ but it was extracted and purified in the presence of DDM,²² so DDM was present in the crystallization sample as well. In the second case, the rat monoamine oxidase A was extracted and purified in FC12, but 0.8% (w/v) dimethyldecylphosphine oxide was also present during crystallization, in addition to 0.05% (w/v) FC12.²³ There is one reported case of a membrane protein being extracted, purified, and crystallized in another foscholine detergent, FC14, the E. coli mechanoselective channel McsC.^24,25

OmpF overexpression during recombinant membrane protein expression in E. coli

Recenly, the first crystal structure of a human membrane protein expressed in E. coli^26,27 has renewed interest in E. coli-based expression of mammalian membrane proteins.^28,29 OmpF contamination is a major issue that needs to be seriously considered. We have observed and verified by N-terminal sequencing that OmpF overexpresses abundantly during recombinant E. coli expression of different families of membrane proteins, including human membrane proteins. OmpF expression does not seem to be linked to the induction of particular recombinant protein overexpression: we have grown BL21(DE3) cells, as a control, in parallel with cultures expressing recombinant membrane proteins, and the amount of expressed OmpF is very comparable between the control cells and the cells expressing the recombinant proteins [Fig. 2(a)]. Moreover, we see similar OmpF expression patterns for both Misticated and non-Misticated proteins,²⁰ so fusion with Mistic does not seem to affect OmpF expression. Small contamination by OmpF in a membrane protein preparation has been reported before in the case of the vitamin B12 transporter BtuB.³⁰ OmpF overexpression is not surprising as porins are one of the most abundant protein families in bacteria. Depending on the bacterial species and the growth conditions, gram-negative bacteria typically have about 10⁵ copies of general porins per cell.³¹ The presence of OmpF is, however, a cause for concern when purifying membrane proteins for structural studies. As we report here, OmpF crystallization can occur opportunistically even from trace amounts in a solution of target protein concentrations of 10–15 mg/mL, due to its unusual stability. Due to the minimal amount of OmpF in the solution, crystallization was slow and it took 4–6 weeks for the crystals to grow, which prompted us to check the identity of the crystallized protein by silver stain SDS–PAGE [Fig. 2(B)]. The choice of cell lysis method doesn't impact on OmpF contamination significantly, as evidenced by comparable levels of contamination either using a microfluidizer or a sonication protocol. The solubilization stage, however, is critical for minimizing the contamination risk. Special care is needed in cases where the only detergent that can efficiently extract the protein of interest is also a good extractor for porins or other endogenous E. coli membrane proteins, as was the case for KdpD.

Given that OmpF is not rich in histidine residues (one in 362 residues, one in 340 for the mature protein), and that the molecular weight of the OmpF trimer is about 110 kDa versus 234 kDa of the KdpD-M dimer and 204 kDa of the KdpD-NM dimer, it is striking that OmpF is still present (although in amounts not detected by SDS–PAGE) in our samples after the immobilized metal ion affinity chromatography and the size exclusion purification steps. To avoid OmpF contamination in the future, we have incorporated in our routine purification protocol an ion exchange purification step. This helps ensuring good separation in cases where the pI of the protein of interest is at least 2 pH units different from the OmpF pI (theoretical pI: 4.6).

The expression of OMP proteins is tightly regulated so that the total porin amount remains constant. The OmpF/OmpC ratio is dependent on factors such as osmolarity and pH of the growth medium, antibiotic concentration, cell density, and bacterial strain.³² In media of high osmolarity, OmpC is expressed at the expense of OmpF. The smaller pore of OmpC slows down the diffusion rate of small molecules and this might be beneficial in the bacterium's natural environment, the intestine, where nutrient concentration, but also the concentration of toxic molecules such as bile salts, is relatively high. In conditions where nutrient concentration is low, OmpF becomes the major porin.³³ In Gram-negative bacteria the development of resistance to β-lactam antibiotics depends, in addition to the production of lactamase enzymes, on down-regulation of porin expression, or porin functional modification through mutation.^34–36 Nonetheless, it has been reported that under antibiotic stress, OmpF was preferentially lost in E. coli K-12, but OmpC was overproduced, and the strain showed resistance to only some β-lactams.³⁷ In the light of this finding, increasing the antibiotic concentration during recombinant protein expression when ampicillin/carbenicillin resistance is used, might help with reducing the amount of expressed OmpF, but the total porin amount is likely to remain the same.

In this study we used the E. coli strain BL21(DE3), one of the most commonly used strains for recombinant protein expression. Analysis of the outer membrane fraction of different E. coli strains has previously shown that OmpF expression is particularly elevated in BL21(DE3), compared to strains HB101, JM101, MC4100, XL1-Blue, and W3110.³⁸ It has been reported that the difference in OmpF expression between the E. coli BL21(DE3) and K12 strain might be explained by two mutations at positions –18 and –363 in the ompF promoter and its upstream region. Position –363 is within the OmpR binding site,³⁹ so the mutation seems to affect OmpR binding, and thus the repression of ompF expression. We have prepared a BL21(DE3) ompF knock-out strain for protein expression for a future comparison study. Nonetheless, caution must be exercised because it has been reported that in E. coli BL21(DE3) ompF knock-outs, OmpF expression is abolished but the expression of OmpC is increased.³⁸

Materials and Methods

All chemicals were purchased from Sigma, unless otherwise noted. NuPAGE gels and the SeeBlue +2 protein standard were from Invitrogen. Detergents were purchased from Anatrace.

Cloning and expression

The kdpD gene from E. coli K12 MG1655 was amplified and Gateway (Invitrogen) adapted by performing two successive rounds of polymerase chain reaction (PCR). In the first round, gene-specific primers and a single cell E. coli colony as template were used. After PCR purification (Qiagen), the product of the first PCR served as the template for the second PCR, and a pair of generic primers was used to introduce the attB sites required for Gateway cloning and an N-terminal thrombin cleavage site. The final PCR product was gel-purified (Qiagen), and used to set up a BP reaction with the pENTR207 vector, following the manufacturer's instructions (Invitrogen). Two different expression constructs were prepared, with or without the membrane-targeting protein Mistic⁴⁰ fused to the N-terminus of the protein. Both vectors are pET derived, Gateway adapted, and add an N-terminal histidine tag to the expressed proteins. The recombinant plasmids were used to transform E. coli BL21(DE3) cells (Invitrogen). Cells from an overnight 5 mL culture were inoculated in 6 L of Terrific Broth medium and grown at 37°C to an OD 600 nm⁻¹. The temperature was then lowered to 18°C and protein production was induced by the addition of 0.5 mM β-d-thiogalactopyranoside, followed by overnight (about 16 h) incubation. Cells were lysed either by three sonication cycles 1 min each, using 1 s pulses, on ice, or by three microfluidizer cycles (M-110L Microfluidizer Processor from Microfluidics) at 18,000 psi.

The membrane isolation protocol has been previously reported.²⁰ In brief, it consists of an initial high-speed centrifugation to remove soluble proteases as soon as possible after cell disruption, followed by a soft spin to remove inclusion bodies, and after overnight solubilization in a high salt buffer to remove electrostatically bound proteins, we pellet the membrane fraction by a final high-speed centrifugation.

Protein purification

KdpD was extracted from the membrane with 20 mM FC12 for 2 h at 4°C in a buffer containing 20 mM Tris pH 8, 200 mM NaCl, 1 mM MgCl₂, 5 mM β-mercaptoethanol (β-ME), 10% v/v glycerol, and one Complete Mini EDTA-free protease-inhibitor cocktail tablet (Roche) per 20 mL. The solubilized membrane protein fraction was separated by centrifugation for 60 min at 100,000g. The supernatant was filtered through a 0.22 μm membrane and was incubated with 5 mL Ni-NTA (Qiagen) beads for 40 min at 4°C. The beads were then packed into a column, and the flow through was collected and reloaded to the column. To remove unbound proteins, the column was washed with 50 mL of 20 mM Tris pH 8.0, 200 mM NaCl, 20 mM imidazole, 5% (v/v) glycerol, 2 mM β-ME, 5 mM FC12. The protein was eluted in three column volumes with 20 mM Tris pH 8.0, 200 mM NaCl, 300 mM imidazole, 5% (v/v) glycerol, 2 mM β-ME, and 5 mM FC12. After SDS–PAGE, the KdpD containing fractions were pooled and concentrated using a 100 kDa molecular weight cut-off concentrator (Vivaspin) and subsequently purified by gel filtration (Superdex 200, 16/60; GE Healthcare) in a buffer containing 20 mM Tris pH 8.0, 200 mM NaCl, 5%(v/v) glycerol, 1 mM dithiothreitol, and 2 mM FC12. The peak fractions were analyzed by SDS–PAGE [Fig. 1(C,D)], pooled and concentrated. The purity of the protein was estimated by SDS–PAGE and its oligomeric state was probed by static light scattering and refractive index.

Crystallization, data collection and structure determination

The protein was concentrated to 10–15 mg/mL and initial crystallization trials were performed using our in-house Honeybee crystallization robot (Genomic Solutions). Crystals were obtained at 20°C after 4–6 weeks in 0.4 M NaH₂PO₄/1.6 M K₂HPO₄, 0.1 M Imidazole pH 8, 0.2 M NaCl (Emerald Biosciences Wizard I condition 20) for the non-Misticated KdpD (KdpD-NM) sample, and in 0.1 M acetate pH 4.5, 0.8 M NaH₂PO₄/1.2 M K₂HPO₄ (Emerald Biosciences Wizard II condition 35), later optimized to 0.1 M Acetate pH 3.9, 0.48 M NaH₂PO₄/0.72 M KH₂PO₄, for the Misticated KdpD (KdpD-M) sample. Diffraction data were collected at the Advanced Light Source (Berkeley, CA) on beamline 8.2.1 using an ADSC Q315r 3 × 3 CCD detector at −100°C. Data were processed and scaled with the HKL2000 package.⁴¹ The crystal harvested from the KdpD-NM crystallization drop diffracted X-rays to 3.8 Å resolution and belongs to space group P21 with cell parameters a = 136.7 Å, b = 210.5 Å, c = 137 Å, and β = 100.5°. The crystal harvested from the KdpD-M crystallization drop diffracted to 4.4 Å resolution and belongs to the P321 space group, with unit cell parameters a = 215.5 Å, b = 215.5 Å, c = 137.5 Å, and γ = 120°. Our various attempts to solve the structure by molecular replacement using the structure of the N-terminal region (residues 19–243) of P. syringae pv. tomato str. DC3000 KdpD (PDB code 2R8R) as a search model, with the program MOLREP⁴² and the molecular replacement automatic suit BALBES,⁴³ failed. Surprisingly, however, molecular replacement using the amino acid sequence of OmpF with BALBES did give a clear solution, using the OmpF structure 1HXX⁶ as the search model. The MR scores were 23.13 for the P21 crystal form, and 18.91 for the P321 crystal form. The structures were refined with REFMAC5⁴⁴ and model building was performed using COOT.⁴⁵ The structures were refined to Rcryst and Rfree factors of 26.4% and 32.9% for the P321 structure, and 28% and 28.8% for theP21 structure. Data processing and refinement statistics are summarized in Table II. Atomic coordinates and structure factors have been deposited in the RCSB PDB, under the accession codes 3K1B and 3K19. Figures were prepared using the programs PyMol⁴⁶ and O.⁴⁷

Table II.

Refinement and Model Statistics for OmpF

	3K1B	3K19
Space group	P 3 2 1	P 21
Cell dimensions	a = 215.5 Å, b = 215.5 Å, c = 137.5 Å, γ = 120°	a = 136.7 Å, b = 210.5 Å, c = 137 Å, β = 100.5°
Resolution limits (Å)	49.23-4.39 (4.5-4.39)	49.58-3.79 (3.89-3.79)
Completeness (%)	99 (98.14)	98.9 (87.5)
Data cutoff [F/s(F)]	0.0	0.0
Number of reflections	23,572	74,966
Working set	22,361	71,179
Test set	1211	3787
Rcryst (%)	26.4 (28)	28 (34.4)
Rfree (%)	32.9 (33.5)	28.8 (32.4)
Number of atoms	10,500	31,492
R.m.s. deviations
Bonds (Å)	0.027	0.004
Angles (°)	2.86	0.650
Average B factor for protein (Å2)	126.423	78.133
Ramachandran plot
Most favoured (%)	73.3	89.74
Additionally allowed (%)	15.9	6.71

Values in parenthesis indicate the highest resolution shell.

Concluding Remarks

Two structures of the OmpF porin with different packing properties were solved. These are the first structures of membrane proteins exclusively treated with FC12, showing that FC12 proves a useful detergent for crystallization screening.

OmpF contamination can be a potentially systematic problem in E. coli-based membrane protein expression. Approaches to minimize the problem might include careful monitoring of growth conditions, selection of a low porin-expressing strain, and sucrose gradient centrifugation to separate inner and outer membranes. After expression, an optimized two-step extraction protocol might be applied, when the target protein can be extracted by a small number of specific detergents.

Glossary

Abbreviations:

β-ME

β-mercaptoethanol

FC12

foscholine-12 (aka DPC, n-dodecylphosphocholine)

M

Mistic fused protein, Misticated

NM

non-Mistic fused, non-Misticated

OG

octyl glucoside

OMP

outer membrane proteins

PCR

polymerase chain reaction

SDS–PAGE

sodium dodecyl sulfate polyacrylamide gel electrophoresis.