Materials and Methods

↓17

2.1  Purification of ribosome nascent chain complexes (RNCs)

↓18

For the generation of purified RNCs a wheat germ in vitro translation system (Ambion) was used programmed with truncated mRNA coding for the 90 N-terminal amino acids of the type-II membrane protein dipeptidylpeptidase B (DPAP-B) from S. cerevisiae. The mRNA carrying the code for an HA-tag was translated in the system and purified on a metal affinity column resulting in highly enriched RNCs.

2.2 Generation of DNA fragments by polymerase chain reaction

To make mRNA, a DNA fragment with N-terminal His- and HA-tags was generated by PCR from yeast genomic DNA using the forward primer DPHisHA and reverse primer DP90.

Oligonucleotide

Sequence (5’→ 3’)

Comment

DPHisHA

taatacgact cactataggg accaaacaaa acaaataaaa caaaaacaca atgtctcatc atcatcatca tca tacccat agatgttcca gattacgctga aggtggcgaa gaagaagttg

His tag, HA tag

DP90

ttgcagctcg tgatatttgg gatg

↓19

The PCR easy kit was used to amplify DNA. The concentration of oligonucleotide primers was 1 μM with ca 10 nM template concentration. Prior to the reaction start Taq-polymerase (50 units/ml) was added. Polymerase chain reaction was made in 30 cycles with 45 s of denaturation at 95 oC, followed by 45 s of primer annealing at 60oC, and 30 s of polymerase reaction at 72oC. Reaction was finished with 1 min at 72oC.

PCR products were checked on agarose gel.

2.2.1  Agarose gel electrophoresis

DNA and RNA are negatively charged molecules, and are moved by an electric field through a matrix of agarose. The migration of molecules depends on their size and on the size of pores of the agarose matrix which depends on agarose concentration.

↓20

The gels were made with 1-2% agarose (Seakem LE Agarose (Biozym, Hess. Oldendorf)) in TAE buffer and run for 20-40 min at 50 V. DNA/RNA molecules were stained with SybrGreen I/II (Molecular Probes) and visualized with 300 nm UV light.

2.2.2 Generation of RNA by DNA transcription

Subsequently, capped mRNA was synthesized using the Message Machine kit (Ambion). 1 μg of DNA was used in 20 μl reaction transcribing into 15-20 μg of mRNA.

2.2.3 Translation and RNC purification

To purify translating ribosomes, the mRNA was translated in a wheat germ in vitro translation system (Ambion). 6x 200 µl reactions were incubated for 45 min at 27ºC and terminated with 2 µl of 10 mg/ml cycloheximide. Reactions were spun through four 600 µl high salt sucrose cushion (50 mM Tris.Cl pH 7.0, 500 mM KOAc, 25 mM Mg(OAc)2, 2 mM DTT, 1 M sucrose, 10 µg/ml cycloheximide) at 355000xg for 45 min (TLA100.2 at 100k). The supernatant was quickly removed to prevent resuspension of the pellet. Each pellet was resuspended in the 200 µl ice-cold 250 buffer (50 mM Tris.Cl pH 7.0, 250 mM KOAc, 25 mM Mg(OAc)2, 0.1% (w/v) Nikkol, 5 mM ß-ME, 10 µg/µl cycloheximide, 250 mM sucrose) for 30 min on ice and transferred on 1.5 ml Talon metal affinity resin (Clontech) into the column. The resin was equilibrated with 5 ml 250 buffer before the addition of the ribosomes. The column with resin and resuspended ribosomes was agitated for 5 min to increase the interaction and binding of His-tagged nascent chains. The resin was washed with 10 ml 250 buffer, and 2 ml 500 buffer (250 buffer with 500 mM KOAc) to remove unspecific bound ribosomes. RNCs were eluted with 2.5 ml 100 mM imidazol pH 7.1 in 250 buffer and spun through the 400 µl high salt sucrose cushion for 45 min at 355000xg (TLA100.3 or TLA100.4 at 100000 rpm). The resulting pellet was slowly resuspended for 30 min in ca 50 µl G buffer (20 mM Tris.Cl pH 7.0, 50 mM KOAc, 10 mM Mg(OAc)2, 1 mM DTT, 125 mM sucrose, 100 µg/ml cycloheximide, 0.05% (w/v) Nikkol, 0.5% (w/v) EDTA-free complete protease inhibitor pill [Boehringer] and 0.2 U/µl RNasin [Ambion]), shock-frozen and stored at -80°C. From 1.2 ml translation reaction 0.7 OD260 of RNCs (~15 pmol) were isolated.

2.2.4 Protein precipitation and SDS PAGE

↓21

Proteins were precipitated with 6% TCA and 0.0125% Na-deoxycholate and separated using SDS PAGE (Sodiumdodecylsulfate polyacrylamid gel electrophoresis) (Leammli (1970)) for approximately 1h at 150 V. 12% PA gels were used. The size of proteins was determined by comparison with broad range protein marker (P7702S, New England Biolabs) .

Protein staining was done with Coomassie Brilliant Blue R250, or Sypro Orange (1:5000) (Molecular Probes).

2.2.5 Western Blot analysis

To check the enrichment of translating ribosomes Western blot analysis was performed. Proteins were transferred onto a nitrocellulose membrane with a semi-dry blotting procedure in transfer buffer(20% MeOH, 48 mM Tris, 39 mM Gly), 0.037% SDS) for 45 min at 1 mA/cm2 (50 mA). The nitrocellulose membrane was incubated first with fat free milk (5% w/v) for 30 min to prevent unspecific antibody interaction. As primary antibody, for the detection of the HA-tag, monoclonal anti-HA.11 16B12 from mouse (Babco) was used in dilution 1:500 in 5% w/v milk. As secondary antibody rabbit anti-mouse IgG-POD (DIANOVA) was used at a dilution of 1:5000 in 5% w/v milk. For the chemiluminescence reaction, the nitrocellulose membrane was incubated for 1 min with ECL (100mM Tris pH 8.5, 1.25 mM aminophtalhydrazide (Luminol, Fluka), 0.2 mM Coumarinacid, 0.01% H2O2). Signals were detected with Kodak Biomax MR film.

2.3 Reconstitution of SRP-RNC complex

2.3.1  Reconstitution and sucrose gradient

↓22

RNC-SRP complexes were reconstituted by incubating 1.5 pmol mammalian SRP (isolated according to [86] and further purified by sucrose density gradient centrifugation[87]) and 0.5 pmol RNCs. Prior to the incubation the KOAc concentration of RNC buffer (G buffer) and SRP buffer was increased to 350 mM by mixing with K500 buffer (25 mM HEPES (pH 7.5), 500 mM KOAc, 5 mM DTT, 5 mM Mg(OAc)2, 100 mM sucrose, 0.02% Nikkol, 100 µg/ml cycloheximide, and 1% of EDTA-free complete protease inhibitor pill). After mixing, buffer conditions were adjusted to 25 mM HEPES (pH 7.5), 150 mM KOAc, 5 mM DTT, 5 mM Mg(OAc)2, 100 mM sucrose, 0.02% Nikkol, 100 µg/ml cycloheximide, and 1% of EDTA-free complete protease inhibitor pill (with K0 buffer which is equal to the K500 except that it lacks KOAc). After 15 min of incubation at RT, the reaction was brought back to 500 mM KOAc (with K1 buffer which is equal to the K500 except that KOAc concentration is 1 M), and spun through 10%-40% high salt sucrose cushion for 80 min in SW60 (Beckmann) at 55k (310000xg) (buffer conditions as for incubation except 500 mM KOAc) and analyzed by SDS-PAGE. Alternatively, instead of applying the complex onto the 10%-40% sucrose gradient, it was spun through 400 l of 1 M sucrose cushion in a TLA100.2 rotor for 45 min at 100k (355000xg). SR-SRP-RNC complexes were reconstituted by incubating 3 pmol mammalian SRP with 5 pmol SR (from Irmgard Sinning, Biochemie-Zentrum Heidelberg) and 0.5 pmol RNCs. Buffer conditions were identical to SRP-RNC reconstitution with addition of SR and 200 nmol GMP-PNP after SRP-RNC complex formation. After 15 minutes of additional incubation with SR, the complex was analyzed in the same way as the SRP-RNC complex.

2.3.2 Grid preparation

For cryo-EM 1.8 pmol of SRP (1.5 μl of 1.25 μM SRP) were adjusted to ca 400 mM KOAc with K500 buffer (3 μl) and 0.5pmol of RNCs (7 μl of 6OD/ml) to ca 330 mM KOAc with K1 buffer (3 μl). Both components were mixed (14 μl) and salt concentration was reduced to 180 mM by adding the same amount of K0 buffer resulting in a total volume of 28 μl under the described conditions.

2.4 Electron microscopy

Samples were applied to carbon coated holey grids as described [88]. Micrographs of the SRP-RNC complex were recorded under low-dose conditions on a Tecnai F30 field emission gun electron microscope in Albany (USA) at 300 kV and on a Tecnai F20 at 160 kV in a defocus range between 1.0 µm and 4.5 µm. The micrographs were scanned on a Heidelberg drum scanner resulting in a pixel size of 1.63 Å on the object scale. SR-SRP-RNC micrographs were recorded on a Tecnai F30 field emission gun electron microscope in Berlin at 300 kV and scanned at a pixel size of 1.21 Å on the object scale.

2.5 Image processing

↓23

Power spectra and defocus determination

The data were analyzed with the SPIDER software package [89]. Firstly, important parameters were saved in the document params.rib (using the script p_makeparams.srp). The structure of the document is described below:

key

name

values or units

values in SRP-RNC project

1

zip flag

0 : do not unzip

1 : needs to be unzipped

0

2

file format

0 : SPIDER

1 : HiScan tif

2 : Perkin Elmer

3 : ZI scanner

1

3

width

(of micrograph, in pixels)

4

height

(of micrograph, in pixels)

5

pixel size

(in Angstroms)

1.63

6

electron energy

(in keV)

300

7

spherical aberration

(mm)

2.0

8

source size

(1/A)

9

defocus spread

(A)

10

astigmatism

(A)

11

azimuth

(degrees)

12

amplitude contrast ratio

(0..1)

13

Gaussian envelope halfwidth

(1/A)

14

(reserved)

(-)

15

(reserved)

(-)

16

(reserved)

(-)

17

window size

(pixels)

276

18

actual size

(pixels)

200

19

interpolation/decimation factor

(1…)

2

↓24

The entries 1-6, 17 and 18 are entered interactively while others have standard values or can be changed by editing the params.rib document. If the interpolation/decimation factor is an integer number, decimation will be used. The decimation will sum values of two neighbouring pixels resulting in an increased the signal to noise ratio, which is the preferred way to reduce the size of images. The document micnum.rib containing the list of micrographs used for processing was created (using the SPIDER command doc create).

For all scanned images (micrographs) the matching contrast transfer function (CTF) and defocus value were determined with the program ctffind3 [90] (using scripts p_ctffind3.srp, p_convert1.srp, ctffind.sh, p_readmrc.py). The script p_ctffind3.srp prepares an image for ctf determination and it converts it into the mrc file format which can be used by the software. Ctffind.sh is executed by the script p_ctffind3.srp and it determines defocus values of micrographs while python script p_readmrc.py converts ctffind output file into the spider document file format. The defocus values for each micrograph were saved in defocus.rib document.

Ctffind3 creates the power spectrum images of micrographs with estimated model on the left and the real data on the right. Power spectra were visually inspected in Web (part of SPIDER software package) and only micrographs with acceptable power spectra (without or with very low drift and astigmatism), and images containing information in the frequency range below 15 Å were selected and used for further processing. Unwanted micrographs were removed from micnum.rib document and the document key was renumbered (using the SPIDER command doc ren). Altogether 150 micrographs were selected, 100 from the F30 and 50 from the F20 microscope, and used for further processing.

2.5.1  Particle picking

↓25

Since only particles over a thin layer of carbon film contain proper information, a mask for the hole on the grid was created. For easier handling the images were decimated 20-fold (p_decimate.srp) and 3 coordinates of the circle of the hole were determined visually in Web (using the command pixel) and saved into a document file. These coordinates were used to create a matching circular mask (p_3coordcircle.srp) for every individual hole.

Each micrograph contains several hundreds or even thousands of single ribosomes which have to be isolated. Particles were automatically picked from micrographs (p_pickCCM.srp, p_pickparticles.srp, p_convert1.srp) by a local fast correlation method where local cross-correlations are calculated with Fourier methods according to Alan Roseman [91]. This procedure needs a reference 3D volume similar to particles that should be searched, and generates one or more projections as template images for the search. Only one projection image was used as a template for these datasets. The procedure sorts particles dependent on the cross correlation with best fitting particles showing up first. This method reduced the time for visual inspection of the particles since low quality particles usually end up clustered together either at the top of the list (contamination with high contrast) or at the end (high noise).

Automatically picked particles were visually inspected in Web and good particles were selected. Prior to the visual inspection particles were low pass filtered depending on defocus value (p_filt.srp). Bad particles were removed from the dataset and good particles were renumbered (p_copygood.srp). A total of 35488 particles were selected as good and used later for the reconstruction.

↓26

Selected micrographs were sorted according to the defocus value and a defocus group document defgrp.rib was created (p_makedefgrpfile.srp). Micrographs with similar defocus values were assigned to same defocus group (third column in defgrp.rib) with an average defocus not more than 250 Å distant from defocus values of the single micrographs. Micrographs from two microscopes were kept separately. Altogether 51 defocus groups were created, 33 for F30 dataset and 18 for F20 dataset.

2.5.2 Alignment

In the first alignment step particles were aligned (p_alidef.srp) to projections of the existing reference of the Sec61-80S ribosome complex from yeast. For each micrograph the reference volume was distorted with corresponding CTF function which depends on the defocus value of the micrograph. Initial alignment was done at an angular accuracy of 15 degrees which generates 83 projections. Shifts in x and y directions were as large as possible to ensure proper positioning of particles. To speed up the alignment particles were decimated by a factor of 2. The output document of the alignment includes for each particle the best fitting projection, and the shift and rotation parameters necessary to apply in order to match the projection.

2.5.3 3D-reconstruction

Rotation parameters and shifts were then used to create the new set of particles used for 3D-reconstruction (p_spinnem2.srp, p_rotate.srp, p_angles.srp). Two percent of particles with lowest cross-correlation coefficients were removed (p_howmanyvo2.srp) from the dataset. Particles were backprojected using parameters from the alignment (bp32f.srp). For each defocus group three volumes were created; one was backprojected with all particles and two additional ones were backprojected with two independent half’s of all particles.

↓27

All odd and all even volumes were CTF corrected and added to create two volumes each containing half of the particles. These two volumes were then compared and the Fourier shell correlation, which is used for resolution determination, was calculated. The cut-off in the Fourier shell correlation curve used for resolution determination was 0.5. Volumes created with all particles in each defocus group were ctf corrected and summed up resulting in the final volume. This volume was filtered to the resolution and used as an initial volume in the refinement procedure.

2.5.4 Refinement

In the refinement particles are iteratively aligned to new references created by those particles (ref_sortref.srp). Before the refinement, stack files containing aligned particles have to be created for each micrograph. Stack files have to be interpolated or decimated to the desired pixel size if necessary. Decimation factor of 2 was used giving a pixel size of 3.26 Å on the object scale. Prior to the refinement, transformation files have to be created (p_maketrans.srp). Transformation files contain shifts and rotation for each particle which have to be applied to particles to fit the reference projection. To avoid subsequent interpolation, after each refinement round original particles are rotated and shifted using transformation files.

In first round of refinement particles were aligned to the volume created in the first reconstruction with angular accuracy of 2 degrees without angular restriction. This procedure offers all possible references to each particle, however, in cost of the speed. In the next rounds particles were compared only with projections inside defined angular restriction and shifts were allowed to position them even more accurate. Angular restriction and angular accuracy were slowly reduced in following rounds allowing better alignment of particles.

↓28

The density of SRP was visible at lower contour levels compared to the density of the ribosome showing lower occupancy of the ligand. To increase the occupancy computationally, the particles were iteratively sorted into two subsets, one containing the ligand and one without. For the initiation of the sorting a volume without SRP was manually created by masking away the density of SRP using a binary mask. Both volumes were offered for alignment to the particles resulting in two different cross correlation coefficients for each particle. The cross correlation coefficients were compared and, dependent on the best match, the particles were sorted into two subsets and backprojected separately. This procedure was repeated iteratively until particles stabilized in each subset. At the end, two subsets of particles were created, one with SRP containing 25397 particles and one lacking SRP containing 10097 particles. Since the sorting was not perfect due to the high level of noise, the SRP containing volume still contained ribosomes without SRP. Nevertheless, the SRP occupancy was significantly enriched.

After the final alignment particles were backprojected with the procedure bprp.srp which is using a slower real space backprojection algorithm resulting in a better signal to noise ratio, and in that way better resolution.

The final CTF-corrected reconstruction was at a resolution of 12.0 Å (7.7 Å) based on the Fourier shell correlation with a cut off value of 0.5 (3σ). This map was used for further interpretation and the model building.

2.6 Building the SRP model

↓29

Firstly, the final volume was adjusted in size, position and orientation to fit the yeast Sec61 volume which allowed usage of the existing models for the yeast ribosome. Orientation search was done in a first step manually to find an approximate orientation and then fine-tuned using SPIDER command OR 3Q. After the volume rotation, size and position were adjusted with the script vol_resize.srp which calculates the cross correlation between the volumes and searches for the highest peak.

For the modelling, the programme package O was used[92]. Since yeast and wheat germ ribosomes showed an extremely high degree of similarity, the molecular model of the yeast ribosome was used as a model for the ribosome (1K5X, 1K5Y, 1K5Z).

Several crystal structures of SRP components were used to make a model of mammalian SRP. First, a large fragment of mammalian S-domain containing 7SL RNA helix 6,7, 8, part of helix 5, SRP19 and the SRP54 M-domain [46] (1MFQ) was docked. The M-domain from this crystal structure was replaced by a different model [93] using the RNA binding moiety for alignment. This model, derived from site-directed mutagenesis, was a modification of the M-domain from the S-domain crystal structure and was fitting better into the density. The structure of a prokaryotic SRP54 NG-domain [94] (1JPJ) was docked into density present near the M-domain. A short α-helical peptide fragment was docked as a signal sequence in the empty density belonging to M-domain at a place predicted to bind a signal sequence. The X-ray structure of the mammalian Alu 5’RNP [40] (1E8O) was docked in intersubunit space and, for the missing part of 7SL RNA, three fragments from a model provided by the SRP-database [95] were used.

↓30

Densities for 60S, 40S, tRNA and SRP were isolated using binary masks. Amplitude correction for the final volume was done by Fourier filtering using B-factors. A higher B-factor was applied to the ribosomal density (150) then to the SRP density (100). For surface representation a lower contour level of the SRP density was applied. This reflects that the SRP density is underrepresented due to incomplete removal of SRP-free ribosomal particles from the final particle subset.

2.7 High resolution structure of SRP-RNC complex

To increase the resolution of the structure more images of SRP-RNC complex were collected on a Tecnai F30 microscope resulting in additional 25000 particles. Altogether 50000 particles were used for the high resolution project. The data from Tecnai F20 microscope were not used due to lower quality in higher frequencies. As the pixel size severely limits the resolution when 0.5 cut-off in Fourier shell correlation curve reaches spatial frequency of approximately 0.25 (describing features defined by 4 pixels), the high resolution project required smaller pixel size of the data. The pixel size was changed as the resolution was increasing, from 3.26 Å/pixel (decimation factor 2) used at the beginning, to 2.44 Å/pixel (interpolation factor 1.5) and finally to 2.04 Å/pixel (interpolation factor 1.24).

Because of the envelope function of the electron microscope higher frequencies are underrepresented and their contribution to cross-correlation coefficients used in alignment procedures is severely impaired. To reach higher resolution, it was necessary to increase the weight of higher frequencies by increasing the amplitude. Amplitude correction was done by Fourier filtering using B-factors and amplitude corrected volumes were subsequently used as references in the refinement procedures. B-factor values were varying between 60 and 140. To be used in the Fourier filter these values have to be divided by the square of the pixel size. Amplitude correction severely improved the resolution of the density.

↓31

A modified sorting procedure has been used to generate two subsets of particles. As sorting criteria the presence of the ligand and also the contribution of the particle in the high frequency region were used. Two subsets of particles were created, the first containing SRP and particles with highest correlation in high frequencies, and the second containing ribosomes without the ligand and particles with dominant low frequencies or weakly aligned high frequencies. These particles did not contribute to the signal in high frequencies, however, they increased the noise. Therefore, their removal resulted in increased resolution. Altogether, approximately 20000 particles, which were used for the final reconstruction, sorted to the positive volume leaving 30000 particles in the negative volume.

To obtain high frequency information (significantly below 10 Å) the contrast transfer function correction has to be done as precise as possible. In first steps, the contrast transfer function was determined from micrographs based mainly on the signal from the carbon film which results in a shift of the defocus. To correct that, the defocus of each micrograph was determined again from volumes backprojected from particles from each micrograph. The volumes offer a better signal to noise ratio of the object of interest itself and in that way, a more precise defocus determination.

The final CTF-corrected reconstruction is at an overall resolution of 9.5 Å (6.9 Å) based on the Fourier shell correlation with a cut-off value of 0.5 (3σ). The resolution of the ribosome is at 8.8 Å with SRP density being at lower resolution due to lower occupancy and possibly lower rigidity.

2.7.1  L30 localization and the model

↓32

The high-resolution structure of the ribosome has α-helical secondary structure clearly resolved allowing the localization of the eukaryotic ribosomal protein L30e. The fold of L30e could be visually identified in the cryo-EM map and the crystal structure from Thermococcus celer could be docked. To confirm the localization the signature search procedure was used (sigsearch.srp) [96]. In the first step search was done at 15 degrees allowing all possible orientations of L30e to roughly be localized in the map. In the second step the search was done at 2 degrees with restricted L30e orientation to fine tune the fit. When L30e was localized, the crystal structure was replaced with the wheat germ homology model. As the template for homology modelling the crystal structure of yeast L30e in complex with maltose-binding protein was used (1NMU, chain D) [97]. The homology model was manually docked using the program package O with further manual adjustment of poorly fitting regions. Firstly, a flexible region between residues 70 and 86 was adjusted to fit into the density. The main chain was manually placed into the corresponding density with side chains positioned in their most common orientation from the O rotamer database. Both main and side chains were refined in O to follow stereochemical constraints. The N- and C-terminal helices were slightly shifted towards the flexible region. The model was completed by positioning missing residues of the N- and the C-terminus in the corresponding density. Due to the limited resolution of the map, these N- and C-terminal residues, the loops connecting helix 4 of L30e and all side chains could not be positioned precisely in an unambiguous manner.

2.8 Structure of SR-SRP-RNC complex

The SR-SRP-RNC dataset was recorded at a Tecnai F30 field emission gun electron microscope at 300 kV in a defocus range between 0.9 µm and 3.2 µm, and scanned on a Heidelberg drum scanner resulting in a pixel size of 1.22 Å on the object scale. Altogether 116 micrographs were selected and used for processing resulting in a total of 73000 particles. The processing was done in a very similar way as the processing of the high resolution structure of the SRP-RNC complex. Sorting according to the presence of ligands resulted in two datasets, one containing SRP and SR (55000 particles) and one without them (18000).

The final CTF-corrected reconstruction is at an overall resolution of 8.8 Å (6.3 Å) based on the Fourier shell correlation with a cut-off value of 0.5 (3σ).


© Die inhaltliche Zusammenstellung und Aufmachung dieser Publikation sowie die elektronische Verarbeitung sind urheberrechtlich geschützt. Jede Verwertung, die nicht ausdrücklich vom Urheberrechtsgesetz zugelassen ist, bedarf der vorherigen Zustimmung. Das gilt insbesondere für die Vervielfältigung, die Bearbeitung und Einspeicherung und Verarbeitung in elektronische Systeme.
DiML DTD Version 4.0Zertifizierter Dokumentenserver
der Humboldt-Universität zu Berlin
HTML generated:
13.12.2006