3D correlation averaging of melibiose permease crystals Philip J. B. Koeck, Pasi Purhonen, Ronny Alv

1
3D correlation averaging ofmelibiose permease
crystalsPhilip J. B. Koeck, Pasi Purhonen,Ronny
Alvang, Björn Grundberg, Hans Hebert
Abstract Few 2-dimensional protein crystals can
be used to determine high-resolution structures,
wheras most electron crystallography projects
remain at a resolution around 10 Å. This might be
partly due to lack of flatness of many
2-dimensional crystals. We have investigated this
problem and suggest single particle processing of
locally averaged unit cells to improve the
quality and resolution of 3-dimensional maps.
Applying this method to melibiose permease we
have calculated a 3-dimensional map that is
clearer and easier to interpret than the map
derived using purely electron-crystallographic
methods.
Methods, results, validation, discussion and
everything else
(b)
(a)
(c)
(e)
(d)
Fig.1 Conventional processing of crystal images
(using e.g. The MRC programs) produces a
cross-correlation map showing the positions of
all unit cells.At these positions we extracted
images of individual unit cells from locally
averaged (lattice-filtered) images such as shown
in figure 2.
Fig. 3 31177 unit-cell cutouts from 50 such
lattice-filtered crystal images recorded at
tilt-angles ranging from 0 to 45 degrees were
iteratively subjected to conventional
single-particle processing using Eman and
shrink-wrapping using Khoros. This lead to a
3-dimensional map with a fourier shell
correlation indicating a resolution of about 10.5
Ångström (thick curve in (e), the thin curves
monitor convergence). The views presented in (a)
and (b) will in the following be called
side-view (looking along the short
crystallographic axis) and top-view
respectively. The rectangular (p2221) unit cell
is 189 by 49Å. From (a) and (b) it is clear that
the single particle model (yellow) has a much
lower noise level than the crystallographic
reconstruction calculated in MRC (white). (c) and
(d) show the x-ray structure of a Na-antiporter
manually fitted to the single-particle map in
side-view andtop-view respectively.
Fig.4 A comparison of map-projections (left
column) and image class-averages (right column)
shows good consistency of the 3d-map with the
image data-set. Difference-images between
projections and class-averages are shown in the
central column.
Fig.5 The angular distribution of views within
the assymetric triangle of the C2 point-group
(latitude 0 to 88 degrees, longitude 0 to 180
degrees , angluar step 4.4 degrees) is very even,
although images were actually only recorded at
tilt-angles of 0, 20, 25, 30 and 45 degrees.
Fig.6 (a)Aligning the unit cells of the crystal
shown in figure 2 with the single-particle 3D-map
shows a distribution of tilt angles around the
expected 0 degrees. The plot shows the assymetric
triangle covering latitude 0 to 88 and longitude
0 to 180 with an angluar step of 10
degrees. (b)The 5 major tilt classes are
distributed over the crystal in a systematic way.
Euler angles (and number of unit-cells) for the
classes black 0, 0 (100 cells) yellow 5, 60
(53 cells) green 5, 120 (75 cells) blue 10,
30 (50 cells) red 20,30 (76 cells)
Fig.2 The lattice-filtered (locally averaged)
crystal image from which the correlation map
shown in figure 1 was calculated. For
lattice-filtering Gaussian masks with a standard
deviation of 10 Fourier-pixels were used. This
image was recorded from an untilted crystal.
(b)
(a)
Karolinska Institutet Philip Koeck Philip.Koe
ck_at_csb.ki.se Department of Bioscience at
Novum 141 57 Huddinge, Sweden