Single-particle reconstruction in the absence of symmetry

1
Single-particle reconstruction in the absence of
symmetry
Absence of symmetry means much larger data
collection Absence of crystal packing
means larger degree of variability,
i.e., heterogeneity of particle
set Low-resolution density maps must be
interpreted in terms of atomic structures
2
High-resolution Single-Particle Cryo-EMLow
(nonexisting) symmetry

• Bridget Carragher Case studies in automation
• Jose-Maria Carazo Dealing with conformational
• variability by advanced classification and
  alignment
• methods
• Joachim Frank (Quasi-) atomic models of the
• ribosome in different functional states
  flexible fitting

3
(Quasi-) Atomic Models of the Ribosome in
Different Functional States, by Cryo-EM and
Flexible Fitting

• Joachim Frank
• Howard Hughes Medical Institute, Health Research,
  Inc., Wadsworth Center, Albany, New York
• Department of Biomedical Sciences, State
  University of New York at Albany
• Supported by HHMI, NIH R01 GM55440, NIH R37
  GM29169, and the National Center for Research
  Resources (NCRR/NIH)

4
Elongation of the Polypeptide Chain by One Amino
Acid
peptidyl transfer gtgt
translocation gtgt
aa-tRNA gtgt accommodation
aa-tRNA gtgt accommodation
5
The Role of the Elongation Factors EF-G and EF-Tu
tRNA selection
translocation
6
7.8 ?
110,000 projections
Spahn et al. (2003) in prep.
11.5 ?
73,000 projections
Gabashvili et al. (2000) Cell
7
Wheat germ
52,000 projections

9.5 Å
Halic et al., NSMB 2005
8
The 70S Ribosome, Seen by Two Modalities of
Imaging
X-ray structure (T. thermophilus)
Cryo-EM map (E. coli) (Yusupov et al., Science
2001) (Gabashvili et al.,
2000)
9
The elongation cycle as seen by cryo-EM using
antibiotic and GTP nonhydrolyzable analogs (no
X-ray studies thus far)
peptidyl transfer
translocation
decoding
fusidic acid thiostrepton GDPNP
kirromycin GDPNP
10
Functional dynamics of the ribosome

• In each binding event observed so far, both
  ribosome and the ligand (e.g., the elongation
  factor) undergo conformational changes.
• Induced fit phenomenon

11
Example the Ratchet Motion

• EF-G

EF-G GDPNP
1) When EF-G binds to the ribosome, the small
subunit rotates counter-clockwise
relative to the large subunit. (Frank and
Agrawal, Nature 2000) 2) EF-G is no longer in
X-ray GDP conformation (Agrawal et al., PNAS 1998)
12
(No Transcript)
13
What is the Purpose of the Ratchet Motion?
Mechanism of mRNA Translocation, in Two Phases
PHASE I mRNA moves along with 30S, relative to
50S (lock is closed)
PHASE II 30S moves back, relative to mRNA and
50S (lock is open)
14
Flexible Fitting

• We wish to explain the conformational changes
  observed in the low-res density map in terms of
  changes in the atomic structure. We achieve
  this by molding the X-ray structure of the
  static ribosome into the density map, by a
  process of flexible fitting. The following two
  computational methods can be used
• Normal Mode Analysis-guided flexible fitting
  (NMFF) molecule is modeled as an elastic network
  (balls connected with springs) only small
  amplitudes allowed.
• Real-space refinement provides multi-fragment
  docking, preserving structural integrity as much
  as possible.
• The resulting quasi-atomic models have enormous
  heuristic value,
• allowing dynamic changes of the system to be
  followed, and
• testable hypotheses to be formulated.

15
Normal Mode Analysis Applied to X-ray Structures
The preferred modes of motion are implicit in
the gross molecular architecture. For example,
the ratchet motion triggered by the binding of
EF-G is predicted by normal mode
analysis Relative Rotation of Small
Subunit L1 stalk pivoting Normal-mode Analysis
guided fitting deforms structure along its normal
modes such that optimal agreement is reached with
the density map.
Animation
Tama et al., PNAS
Tama et al., PNAS 100 (2003) 9319 Jernigan, J.
Struct. Biol. (2004) Wriggers NMA of low-res.
density maps.
16
Fitting via Real-Space Refinement (Chapman,
1995)
Rgeom stereochemical term
17
Real-space Refinement
geometry restraint
density restraint
R?
Rgeom
energy minimization, TNT, CNS
18
Dynamic events we have analyzed by real-space
refinement

• Translocation EF-G-induced ratchet motion,
  motion of a factor-binding component of the
  ribosome called GTPase-associated center (GAC),
  and motion of L1 stalk
• Decoding/tRNA selection motion of GAC and
  kinking/distortion of the tRNA
• Signal peptide (SecM)-induced translational
  arrest for the ribosome to allow lateral
  insertion of membrane-intrinsic protein in
  co-translational protein translocation
• Each analysis consists of a comparison of two
  maps via RSR. PDB-formatted coordinates can then
  be displayed using any molecular graphics
  package.
• Very effective and informative display modes
• 1) animation rotate Ribbons representations
  while alternating between of the two versions of
  the structure.
• 2) color the Ribbons representation of one
  structure according to the magnitude of the RMSD
  between the two structures .
• 3) color the secondary structure diagram of one
  structure according to the magnitude of the RMSD
  between the two structures

19
Steps to Follow in Real-Space Refinement

• 1) Decide on a division into stable fragments.
  Here are the choices for the ratchet motion
• 16S RNA 43 pieces
• 23S RNA 62 pieces
• 5S RNA 4 pieces
• Proteins most retained as single rigid units.
• exceptions S2, S7, S13 L2, L3, L5, L9, L11,
  L18, L24,
• which were cut into major domains.
• Total number of rigid pieces 162. Is this
  overfitting? No
• Number of degrees of freedom (100?/10?)3 4/3p
  4000
• 2) Use manual or automated rigid-body docking for
  pre-alignment
• 3) Use RSRef program
• Gao et al., Cell 113 (2003) 789-801

20
E. coli rRNA
16S
23S
5S
21
(No Transcript)
22
Real Space Refinement Using RSRefRatchet motion

• Initiation-like EF-G bound
• Map resolution 11.5 Å 12.3 Å
• Initial CCC 0.53 0.37
• Final CCC 0.71 0.67
• Initial R-factor 0.29 0.32
• Final R-factor 0.23 0.24
• Initial vdW close gt10,000 gt10,000
• Final vdW close 1,900 1,200
• Gao et al., Cell 113 (2003) 789-801

23
Gao et al., Cell 113 (2003) 789-801
24
(No Transcript)
25
Gao et al., Cell 113 (2003) 789-801
RSREF applied to EF-G-triggered ratchet-like
rotation Color mapping shows where changes occur
maximallly.
26
Dynamics of tRNA Selection and Accommodation
Cryo-EM Snapshots in Three States

unbound
A/T A/A
post-translocation Phe-tRNAPheEF-TuGDPkir
accommodated ready for next tRNA
tRNA selection tRNA
approved
Valle et al., NSMB 10 (2003) 899
27
Real-Space Refinement Using RSRef binding of
ternary complex (A/T state)

• Unbound A/T state
• Map resolution 11.5 Å 12.5 Å
• Initial CCC 0.53 0.48
• Final CCC 0.71 0.67
• Initial R-factor 0.29 0.36
• Final R-factor 0.23 0.26
• Initial vdW close gt10,000 gt10,000
• Final vdW close 1,900 4,000
• Sengupta et al., in preparation

28
H43/H44 (GAC)
Sengupta et al., in preparation
RSREF applied to A/T state (ternary complex
bound) and unbound state
GAC moves strongly
29
GAC (L11helices 42, 43, 44 of 23S rRNA)
movements in response to (1) GTP hydrolysis
(open ? half-closed) and (2) binding of ternary
complex (half-closed ? closed)
closed
half-closed
open
KT-42
Frank et al., FEBS Lett. 2004
30
translocon
lateral insertion into lipid
Co-translational insertion of transmembrane
protein, signaled by SecM signal sequence that
is in transit in the tunnel, requires
translational arrest
K. Mitra HHMI Wadsworth Center
31
SecM-induced conformational changes in the
ribosome analyzed by RSRef
(translational arrest)
K. Mitra et al., Mol. Cell 2006
Presence of SecM is probably sensed by L4 and L22
fingers, producing conformational signal.
32
Conformational changes, and their putative roles
in translational arrest
Mitra et al., Mol. Cell 2006
33
Control of dynamic study with RSRef use
identical samples re-do all steps of sample
prep, EM, image processing and RSRef
Mitra et al., Mol. Cell 2006
Result RMSD between pairs of coordinates of same
residue is below 2 Å everywhere, (see
Rossmanns rule of thumb ratio 1 to 5)
34
Overview over docking and fitting procedures
(Fabiola and Chapman, 2005)

• Global rigid body search for initial
  configuration
• SITUS (Wriggers and Chacon, 2001) use of code
  vectors
• COAN (Volkman et al., 2003) 6-D exhaustive
  search consider solution set
• DOCKEM (Roseman, 2000) 6-D exhaustive search
  local normalization of cross-correlation
• Final refinement
• URO (Navaza et al., 2002)refinement in reciprocal
  space
• NMFF-EM (Tama et al., 2004) normal-mode analysis
  guided fitting
• RSREF (Chapman et al., 1995 2005) real-space
  refinement
• In between
• EMFIT (Rossmann, 2000) variety of target
  functions refinement in reciprocal space
• SITUS (Wriggers and Chacon, 2001)
• CHARMM coarse-grained search combined with
  Monte-Carlo optimization (Wu et al., 2003)

35
Conclusions

• Real-space refinement can be used to construct
  quasi-atomic models depicting snapshots of a
  dynamic process.
• Such models have great heuristic value as they
  allow local conformational changes underlying
  global motions to be followed.
• The fit of the ribosome with a number of pieces
  in the order of 150 represents a conservative
  use of RSRef
• Insights have been obtained for translocation,
  tRNA selection, and SecM-induced translational
  arrest.

36
Contributors
Members of the group Haixiao Gao Bob
Grassucci Kakoli Mitra Mikel Valle now at CNB,
Madrid Jayati Sengupta Christian Spahn now at
Charite, Humboldt University, Berlin Collaborators
within Patrick Van Roey -- Wadsworth Rajendra
Agrawal -- Wadsworth Outside collaborators Andrei
Sali and Narayanan Eswar, UCSF Måns Ehrenberg and
Andrej Zavialov, Uppsala University Michael
Chapman, Felcy Fabiola, and Andrej Korostelev,
Florida State Steve Harvey and Scott Stagg,
Georgia Tech Charles Brook and Florence Tama,
Scripps