characterization of TtOmp85 from Thermus thermophilus HB27
4.3.3 Crystallization of purified TtOmp85 and crystallographic analysis
No crystals were identified for preparations of pure detergent-solubilzed TtOmp85 in sitting-drop crystallization set-ups using commercially available screens. However, using buffers of the Iwata-screen containing 20% PEG 400 and a pH of 7.5 and 8.5, respectively, in hanging-drop crystallization set-ups showers of very small crystals were observed for TtOmp85 preparations in C8E4. Crystallization conditions were further refined and crystal growth could be somewhat improved, but the achievable final crystal size remained limited to plates with dimensions of 25 µm × 25 µm × 5 µm (fig. 4.4). Refined crystallization conditions contained 100 mM 4-(2-Hydroxyethyl)piperazine-1-ethanesulfonic acid (HEPES) (pH 7.5), 24 % PEG 400 and 200 mM LiSO4. Crystals appeared within a few days. The crystals could be reproduced with protein solutions from different preparations, but further refinement did not lead to growth of bigger crystals up to now.
A partial dataset with a resolution of 5.0 Å was recorded with the best diffracting crystal.
Reasonable diffraction of the plate-shaped crystals was observed only for a limited range of crystal rotation (ϕtotal) due to the very limited thickness of the crystals (fig. 4.5): diffraction was best for orientations of the plate-shaped crystals parallel to the X-ray beam as the volume
Figure 4.5: Diffraction images of crystallized TtOmp85. Upper image: Cryofrozen crystal in nylon loop, mounted on beamline ID23-1 of the ESRF. Bars represent dimensions of 20µm and 50µm, respectively.
Lower images: Diffraction images of various∆ϕorientations of the crystal during data collection.
participating in scattering becomes maximal in this orientation. Accordingly, diffraction was weakest after a rotation of the crystals of 90° from that orientation. Crystals were accumulating radiation damage during proceeding data collection and in consequence observable diffraction weakened further. However, a dataset of 50°ϕtotal with 0.5° ∆ϕper frame could be recorded, sufficient for determination of unit-cell-dimensions and spacegroup of the crystals.
Crystals were identified to belong to spacegroup C222 with unit cell dimensions of a = 96.3 Å, b = 416.1 Å, c = 94.3 Å;α=β=γ= 90°. The completeness of the collected data was far too limited for determintion of even a low-resolution structure due to the anisotropic scattering be-haviour, accumulating radiation damage and limited number of collected frames of the dataset:
For spacegroup C222 a dataset of minimum 90° ϕtotal unaffected by anisotropy or radiation damage would be required for complete data.
Calculation of the corresponding Matthews-coefficient (Matthews, 1968) verified the unit-cell dimensions to be large enough for either one or two monomers of TtOmp85: 78% sol-vent content for a monomer and 55% for a crystallographic dimer of TtOmp85, respectively.
Samples used for successful crystallization (fig. 4.2 B, lane 3) were verified to contain highly purified TtOmp85 by MALDI-MS.
4.3.4 In-vitro characterization of purified TtOmp85: TEM, FFEM and single channel conductance measurements
TtOmp85 solubilized in 20 mM Tris (pH 8.5), 150 mM NaCl, 0.05% Cymal-6 was reconstituted into liposomes formed by dimyristoylphosphatidylcholine (DMPC) orE. colilipids, negatively stained either with 2% uranyl acetate or 2% phosphotungstic acid, and examined in a TEM (for details see Nesperet al., 2008). In all cases the incorporated protein was visible as rings on the liposome surface (fig. 4.6).
Both to confirm this result and to demonstrate reconstitution when starting from another detergent system, FFEM was employed using TtOmp85 in 20 mM Tris (pH 8.5), 150 mM NaCl, 0.35% C8E4 and E. colilipids (for details see Nesper et al., 2008). A very high lipid-to-protein ratio of 40:1 was chosen for this experiment to ensure that individual protein entities would be visible on the liposome surface. A typical proteoliposome is shown in fig. 4.7.
Figure 4.6: Negative-stain TEM images of TtOmp85 reconstituted into lipid vesicles. (A) TtOmp85 re-constituted into DMPC liposomes and negatively stained with 2% uranyl acetate. The rere-constituted protein is visible as rings on the liposome surface (arrows). The insets show averages of the two different types of TtOmp85 rings observed; the diameter of the upper averages approximately 5.1 nm, and that of the lower averages approximately 5.9 nm. Bar: 5 nm. (B) TtOmp85 reconstituted into DMPC liposomes and nega-tively stained with 2% phosphotungstic acid. TtOmp85 protrudes roughly 4 to 5 nm from the membrane, giving the edge of the liposomes a rather lacy appearance (arrows). The inset shows liposomes prepared in the absence of protein. (C) TtOmp85 reconstituted intoE. coliliposomes and negatively stained with 2%
uranyl acetate. End views (black arrow) and side views (white arrow) of the protein can be distinguished.
Bar: 50 nm.
Figure 4.7: FFEM of reconstituted TtOmp85. (A) Proteoliposome formed from TtOmp85 andE. coli lipids. It is peppered with shadowed particles, revealing the distribution of the TtOmp85 protein on the lipidic surface. The individual particles have an average diameter of 9 nm, compatible with the insertion of monomeric TtOmp85 in the lipid bilayer. (B) Control liposomes formed by pureE. colilipids. Note the orange-skin appearance of their surface and the absence of larger shadowed particles. Bar: 100 nm.
Figure 4.8: Single-channel conductance recordings indicate that TtOmp85 forms ion channels in black lipid films. (A) Stepwise increase of the conductance (G) recorded as a function of time (t), showing that TtOmp85 exhibits two channel states. (B) Expanded time window of panel A, indicating frequent channel openings with a conductance∆G2of∼0.4 nS. These openings required a preceding conductance change
∆G1of∼0.65 nS. (C) Histogram of TtOmp85 channel conductance events, indicating that about half of all channel events show this conductance∆Gof∼0.4 nS. Conductance events with∆G1 of∼0.65 were far less frequent. A total of 640 channel events were recorded and evaluated.
To date, all OMPs of the Omp85 family have been found to form ion channels, with single-channel conductances ranging from 0.4 nS for YaeT fromE. colito 2.1 nS for nOmp85 from Nostoc(Bredemeieret al., 2007; Ertelet al., 2005; Méliet al., 2006; Paschenet al., 2003; Robert et al., 2006). Pore formation by Omp85Tt was tested in planar lipid bilayers (for details see Nes-peret al., 2008). Purified native TtOmp85 in 20 mM Tris (pH 8.5) containing 100 mM NaCl and 0.05% Cymal-6 was reconstituted into black lipid membranes of diphytanoylphosphatidyl-choline (diPhPC). Conductance measurements revealed the opening and closing of pores (fig.
4.8).
4.4 Discussion
The first protein of the Omp85 family from a thermophilic bacterium was analyzed. TtOmp85 was overexpressed and purified as a native, stable protein from the OM of T. thermophilus.
TtOmp85 showed a high content of β-sheet secondary structure, similar to most OMPs of Gram-negative bacteria, indicating that it forms a transmembrane β-barrel. The composition of the secondary structure (∼55% β-sheet, ∼16% α-helix, and ∼29% random coil) closely resembles that of FhaC (∼49%β-sheet and∼9.1%α-helix) (Clantinet al., 2007). The signifi-cant amount ofα-helix secondary structure detected in TtOmp85 agrees well with the predicted presence of five POTRA domains within the periplasmic N-terminus. The structures of several POTRA domains were recently solved (Clantinet al., 2007; Kimet al., 2007). Although the primary sequence similarity between different POTRA domains is very low, their structures are similar and comprise a three-strandedβ-sheet and twoα-helices. POTRA domains are known to mediate protein-protein interactions, and the presence of at least one is essential for the func-tion of proteins of the Omp85 family (Bos et al., 2007b; Clantin et al., 2007; Habib et al., 2007). They are reported to interact with unfolded substrates (Clantinet al., 2007; Habibet al., 2007) and to mediate homo-oligomerization (e.g., nOmp85 trimerizes via the POTRA domains (Ertelet al., 2005)) and hetero-oligomerization (e.g., YaeT interacts with the four lipoproteins NlpB, SmpA, YfiO, and YfgL via the POTRA domains (Kimet al., 2007)). Incorporated into liposomes, TtOmp85 was found to protrude∼4 to 5 nm from the membrane, giving a rough idea of the size of the N-terminus with its five predicted POTRA domains.
Three different techniques, DLS, BN-PAGE, and STEM, indicated that native TtOmp85 is present mainly as a monomer in detergent solutions. On reconstitution into lipid bilayers, TtOmp85 also appeared as single particles scattered across liposomes by FFEM (E. colilipids) or as monomeric, ring-like particles with a central cavity by negative-stain TEM (DMPC and E. coli lipids). In the presence of detergent, higher-molecular-mass forms of TtOmp85 were observed by BN-PAGE analysis and STEM, suggesting that TtOmp85 has the capability to form homo-oligomers. Compared to those obtained by the other techniques, the STEM data suggested a relatively high content of oligomeric particles (around 31%). However, STEM is a single-molecule- rather than a bulk-technique, and not every particle imaged could be measured
due to the close proximity of neighbours, which would have influenced the statistics. Some purified Omp85 family proteins have been reported to exist as multimers. In native form the TpsB-type Omp85 protein HMW1B was found to be a dimer (Li et al., 2007). The cyanobac-terial Omp85 protein nOmp85 was reported to exist as homotrimers when reconstituted into artificial membranes (Bredemeieret al., 2007). Purified, refolded YaeT was suggested to form tetramers (Robert et al., 2006). In contrast, FhaC was crystallized as a monomer, suggesting that it is also predominantly monomeric in solution (Clantinet al., 2007). Omp85 family pro-teins are thus able to oligomerize, but there are clearly differences in the degree. At present it is not understood whether the protein-trafficking activity of Omp85 family proteins in vivo depends on this property. Further, oligomerization might also be dependent on the association of other proteins to form one active complex.
TtOmp85 reconstituted into liposomes appeared as a ring-like structure with a stain-filled cavity, implying a pore with a maximum diameter of approximately 15 Å. The pore is most likely occluded since channel activity of TtOmp85 incorporated into planar lipid bilayers does not suggest such a large pore. The most frequent change in conductance at 1 M KCl found for TtOmp85 (∆G2, ∼0.4 nS) is similar to that published for Omp85 proteins involved in the insertion ofβ-barrel proteins, e.g., 0.4 nS (Bredemeier et al., 2007) or 0.5 nS (Robert et al., 2006) for the YaeT ofE. coliand 0.56 nS for the Sam50 ofDrosophila melanogaster mitochon-dria (Bredemeier et al., 2007). Larger channel conductances have been observed for Omp85 proteins proposed to be involved in the translocation of proteins, e.g., 2.1 nS for the nOmp85 of the cynaobacteriumNostoc, 2 nS for the Toc75 ofPisum sativumchloroplasts (Bredemeier et al., 2007), and 1.2 nS for FhaC ofBordetella pertussis(Méliet al., 2006).
Crystals grown from highly pure preparations of TtOmp85 from outer membranes ofT. ther-mophilus HB27 showed diffraction patterns typical for protein crystals. The spacegroup and unit cell dimensions were determined using a dataset collected with the best diffracting crys-tal. Unit cell dimensions provide enough space for either one or two molecules of TtOmp85 which is in accordance with the oligomeric state of TtOmp85 in solution as identified by BN-PAGE, DLS and STEM. The unregular threedimensional shape of the plate-like crystals and the very limited overall-size of the crystals leads to anisotropic diffraction during rotation of the crystals in the beam. Additionally, accumulation of radiation damage consecutively
de-creased diffraction of the crystals during data collection. In consequence, collected datasets were of only partial completeness, unsufficient for solution of a low-resolution structure. How-ever, the presence of reproducible, diffracting crystals shows the potential of preparations of C8E4-solubilized TtOmp85 to grow crystals suitable for successful structure solution after fur-ther refinement of the present or screening for additional crystallization conditions with future preparations of TtOmp85.
As separation of TtOmp85 from other co-solubilized proteins by ion-exchange chromato-graphy was rather efficient but still limiting the yield of purified TtOmp85, attempts were done in designing constructs for His-tagged TtOmp85 (data not shown). N- or C-terminal His tags did not lead to successful overexpression of His-TtOmp85. Consequently, further His tag inserts were designed which are located in regions of predicted loop regions of theβ-barrel domain of TtOmp85. Positions for insertion of hexa-His tags were estimated by secondary structure pre-diction done with PRED-TMBB (http://biophysics.biol.uoa.gr/PRED-TMBB/) according to the amino acid sequence of TtOmp85. Predicted large coil regions, assumed to represent extracellular loops, were chosen as first targets for His tag insertion. These positions could be refined upon publication of the structure of the Omp85-family protein FhaC (Clantin et al., 2007): combined with sequence alignments of the respectiveβ-barrel domains of FhaC and TtOmp85, prediction of intra- and extracellular loop regions could be optimized. Success-ful but limited overexpression was achieved for a number of constructs but only two showed binding to Ni-affinity media. His tags for these two constructs are located in the last (sequence-wise) putative intra- and the last extracellular loop. Purified His-TtOmp85 showed tendencies towards oligomerization and eluted mainly as tetramers (data not shown). Thus, further en-gineering is necessary to obtain His-tagged TtOmp85 overexpressed in significantly increased amounts and, preferably, in monomeric form.