Bio Imaging with X

Bio Imaging with X-ray FELs!
!
!
!
!
!
!
Henry Chapman
Center for Free-Electron Laser Science
DESY and University of Hamburg!
!
!
Workshop on Advanced X-Ray FEL
Development, 21st May 2014
Last week the 100,000th structure was deposited in
the protein databank
ribosome
Cumulative number of structures in the PDB
FELs
Structures
virus
myosin
12
nucleosome
antibody
transfer
RNA
actin
hemoglobin
myoglobin
C. Zardecki - PDB!
C. Abad-Zapatero - Acta Cryst D68 (2012)
Year
X-ray
NMR
Electron
High radiation dose causes changes in molecular structure
Crystal of Bovine enterovirus 2 (BEV2) after
subsequent exposures of 0.5 s, 6 ⨉ 108 ph/μm2
300 kGy dose
Room temperature
!
Cryogenic cooling gives 30 MGy tolerance
Axford et al. Acta Cryst. D68 592 (2012)
Diamond Light Source (courtesy Robin Owen &
Elspeth Garmen)
1 Gy = 1 J/kg!
!
1 MGy ! ≈ 1 eV / Da absorbed!
!
≈ 0.16 eV / atom!
(about one ionization per
!
10 amino-acid residues)!
!
!
≈ 2 ⨉ 109 ph/μm2
(assuming we need 25 eV to ionize)
X-ray free-electron lasers may enable atomicresolution imaging of biological macromolecules
2 fs
5 fs
10 fs
20 fs
50 fs
R. Neutze, R. Wouts, D. van der Spoel, E. Weckert, J. Hajdu, Nature 406 (2000)
X-ray FELs bring experimental innovations to protein
structure determination
1. Ultrashort pulses: outrun radiation damage
2. High intensities: measurable diffraction from small samples
3. High repetition: Hundreds of independent measurements per second
that can be averaged to increase the signal in proportion to the noise
XFEL pulse
Noisy
diffraction
pattern
Hard X-ray experiments show high-resolution diffraction
Photosystem)I)
9.3$keV$
Single$shot$pa8ern$$
~1$mJ$(5$×$1011$photons)$
40$fs$
2$×$1017$W/cm2$
25$GW$XJray$pulse
3.0$Å$resolu,on
crystals$prepared$
by$Petra$Fromme
Nanocrystallography is carried out in a flowing water
microjet
Boutet et al Science (2012)
Chapman et al Nature (2011)
Gas-focused
liquid jet
pump laser
X-ray FEL
Focusing
optic
Detector
0.2 - 2 µm crystals
4 µm jet
3 µm X-ray beam
Recent hard X-ray experiments show high-resolution diffraction
Each pattern is indexed
Tom White (CFEL)
Rick Kirian (ASU)
c*
b*
a*
We have merged indexed patterns into a 3D
diffraction pattern
CrystFEL software available:
http://www.desy.de/~twhite/crystfel/index.html
Tom White (CFEL)
CrystFEL processes diffraction patterns to give
merged intensities
http://www.desy.de/~twhite/crystfel
Free and open-source software (GNU GPLv3)
Documentation including a full tutorial online
T. White et al. J. Appl. Cryst. 45 (2012)
Chapman et al.: Initial
proof of concept
Boutet et al.: High
resolution proof of
concept with lysozyme
In-vivo IMPDH (Nass,
Redecke et al)
Photosystem II at 9.3 keV
NEWSFOCUS
THE DISCOVERY OF THE Higgs Boson
Demirci et al.: Feasibility
demonstration with 30S ribosomal
subunit microcrystals
Johansson et al.: B.
viridis reaction center
act by exchanging other particles that convey three forces: the electromagnetic force;
the weak nuclear force, which spawns
neutrinos; and the strong nuclear, which
binds quarks.
But there’s a catch. At first blush, the
standard model appears to be a theory of
massless particles. That’s because
simply assigning masses to the
particles makes the theory go
haywire mathematically. So
mass must somehow emerge
from interactions of the otherwise massless particles
themselves.
That’s where the Higgs
comes in. Physicists assume
that empty space is filled with
a “Higgs field,” which is a bit
like an electric field. Particles interact with the Higgs field to acquire energy
Redecke, Nass et al.: High
Online
resolution structure of TbCatB
Higgs, researchers at the European particle
physics laboratory, CERN, near Geneva, built
the $5.5 billion, 27-kilometer-long LHC. To
spot the Higgs, they built gargantuan particle
detectors—ATLAS, which is 25 meters tall
and 45 meters long, and CMS, which weighs
12,500 tonnes. The ATLAS and CMS teams
boast 3000 members each. More
than 100 nations have a hand
in the LHC.
Perhaps most impressive
is the fact that theorists predicted the existence of the
new particle and laid out its
properties, right down to
the rates at which it should
decay into various combinations of other particles. (To
test whether the particle really
is the Higgs, researchers are measuring those rates now.) Physicists have made
Barends et al.:
Anomalous signal from
sulphur (at SACLA)
cember 21, 2012
NO RECENT SCIENTIFIC ADVANCE HAS
generated more hoopla than this one. On
4 July, researchers working with the world’s
biggest atom smasher—the Large Hadron
Collider (LHC) in Switzerland—announced
that they had spotted a particle that appears
to be the long-sought Higgs boson, the
last missing piece in physicists’ standard
model of fundamental particles and forces.
The seminar at which the results were presented turned into a media circus, and the news
captured the imagination of people around the
world. “[H]appy ‘god particle’ day,” tweeted
will.i.am, the singer for pop group The Black
Eyed Peas, to his 4 million Twitter followers.
Yet, for all the
hype, the discovery of
the Higgs boson eassciencemag.org
ily merits recognition
For an expanded
as the breakthrough
version of this sec- of the year. Hypoth-
Microcrystals are often measured with the
attenuated LCLS beam
Human$serotonin$5JHT2B$receptor$bound$to$the$agonist$ergotamine$
Crystals$for$
Synchrotron
Maximum$dose:$25$MGy,$30$fs$pulse$
(detector$limited)
X,ray)FEL)(RT)
Synchrotron
100$μm
Side)chain
Crystals$for$XJray$
FEL
H,bonds
Salt)bridge
Liu,...,$Cherezov,$Science$(2013)
Averaging improves the signal to noise ratio
SNR in 2.06 Å shell
1.4
1.2
SNR
1.0
0.8
0.6
0.4
0.2
0.0
0
25,000
50,000
Number of patterns
Cornelius Gati
75,000
100,000
Larger samples are inertially confined
0$fs
90$fs
30$fs
60$fs
Simulation by C. Caleman,
using Cretin (H. Scott, LLNL)
100 fs
6 GGy
(60 MGy/fs)
Atomic cross sections of neutral Carbon atoms
Photon energy (eV)
Cross section (µm2/atom)
10-8
10-10
10
-12
10
-14
102
103
Photoabsorption
104
105
106
At 6 keV:
6 keV
2Å
σA = 2 ×10-14 µm2
σS = 6 × 10-16 µm2
30 photoionisation
events for every
scattered photon
1/(Enough photons to
photoionise every atom)
Dose =
Coherent
Compton
A
I0 : photons ×
10-16
10-18
103
NA
I0
mA
(energy/photon)/
area
102
101
100
Photon wavelength (A)
10-1
10-2
NA
6keV ·
= 50 GGy
mA
Henke (1993) and Hubbell (1975)
The chances for hitting heavier elements is higher
C
N
100$GGy
O
S
Fe
Mn
29$GGy
2$GGy
3$GGy
350$MGy
5$fs
1$fs
0.5$fs
0.6$fs
55$GGy
10$fs
7$fs
Satura,on$dose$to$a$protein$(one$coreJshell$
photoionisa,on$for$every$atom$of$this$species)
Auger$decay$,me
The principal result after photoionisation is the
generation of a cascade of electrons
Incident X-rays
(6 keV)
Ejected KShell
electron
5.7 keV
500 Å/fs
Emitted
Auger
electron
(280 eV)
The principal result after photoionisation is the
generation of a cascade of electrons
Ejected KShell
electron
5.7 keV
500 Å/fs
350
300
Nelectrons
Incident X-rays
(6 keV)
Cascade thermalises in about 10 fs
Produces 200 ionisations (~10 core)
Over a radius of 500 Å
Range
400
8.5 keV
5.0 keV
1.5 keV
0.5 keV
1000 Å
250
200
500 Å
150
100
150 Å
50
0
1
Emitted
Auger
electron
(280 eV)
100 Å/fs
10
Time (fs)
100
MD simulations by Carl Caleman
and Nic Timneanu
see ACS Nano 5, 139 (2011)
A crystal only gives Bragg diffraction when it is
a crystal! Selecting Bragg peaks filters the data
)0)fs
25)fs
50)fs
!
1.5
$"#
Diffraction intensity relative to
12344&5+36&7+(3)&'2389:0
“40 fs”
Ratio of Bragg decay
!
40
100
150
200
250
300
$"!
1
1.0
;!&<,
$!!&<,
$#!&<,
=!!&<,
=#!&<,
>!!&<,
40 fs
100 fs
!"#
0.5
0.5
150 fs
Photosystem)I,)2)keV,)6)GGy
0.0
!"!
0 !"!
0.0
0
!"#
0.5
0.5
Barty et al. Nature Photon 6, 35–40 (2012)q
%&'()(*+,-./0
q (nm-1)
(cycles/nm)
300 fs
$"!
1.0
1
$"#
1.5
1.5
The explosion accelerates during the pulse
RMS displacement (nm)
I(q) =
2
I0 re
|F0 (q)|
T
2
e
q2
2
(t)
dt
1020 W/cm2
0
60 GGy/fs
1.000
6 GGy
6000 MGy/fs
600 MGy/fs
0.100
60 MGy/fs
6 GGy
0.010
6 GGy
0.001
1
Carl Caleman & Nic Timneanu
30 MGy
10
time (fs)
1016 W/cm2
6 MGy/fs
Diffraction at d=3Å
resolution turns off at
RMSD = d/2π = 0.5Å
100
Bragg diffraction will also stop when there’s no more
electrons bound to atoms (transparent sample)
Average ionization per C atom
10
1020 W/cm2 = 1 TW/µm2
60 GGy/fs
6 GGy/fs
1
0.1
600 MGy/fs
60 MGy/fs
6 MGy/fs
0.1
1
10
Time (fs)
1016 W/cm2
= 0.1 GW/µm2
100
Higher dose rates (i.e. higher X-ray intensities)
should give larger Bragg signals
Intensity at 6 keV(GW/µm2)
20
200
2000
20,000
200,000
1014
1013
1012
1011
1010
Maximum effective dose
Hollow S Hollow C
atoms
atoms
300 GGy
30 GGy
100 fs
100 fs
3 GGy
10 fs
300 MGy
1 fs
pristine atoms
30 MGy
1 fs
Turn off time (for 3Å resolution)
Effective fluence at 6 keV (ph/µm2)
0.2
0.1 fs
0.6
6
60
600
6,000
Dose rate (MGy/fs)
60,000
600,000
Son, Young, Santra PRA 2011
We propose to use the fluence dependence of
heavier element scattering factors for phasing
Dose (MGy)
1.2
20
200
2,000
Average ionisation
of 1
20,000
S saturation
200,000
O saturation
H
1.0
C
N
MAD ã
0.8
O
0.6
S
“Low”
intensity
0.4
@6 keV
0.2
0.0
1010
“High”
intensity
1011
1012
2
Fluence (photons/µm )
1013
1014
S.K. Son, R. Santra CFEL
Scattering signals increase with increasing
wavelength
Relative signal
1000.0
6
3
2
1.5
1
wavelength (Å)
100.0
Selenium edge
10.0
1.0
0.1
2
Sulfur edge
4
6
8
10
12
Photon energy (keV)
14
Work at the longest wavelength that supports the achievable or
desirable resolution
Atomic-resolution diffraction from single
2
particles should be possible with 1014 ph/μm
Laser polarisation (E)
28 nm
25
3 Å resolution
1014 ph/μm2
60 GGy
6000 MGy/fs ⨉ 10 fs
!
RMS displacement: 0.5Å
half electrons ionized
The resolution limit is determined by the
ability to average diffraction patterns
Anthrax lethal factor
protein
Simulated diffraction patterns with 1012 photons in (0.1 μm)2
Average of 100 patterns
Average of 10000 patterns
(50Å radius)
Atomic resolution can be achieved for relatively small particles and
fluences achievable at LCLS (1012 photons/(0.1 μm)2)
The highest
resolution is
limited by the
ability to
group patterns
of similar
orientation
Averaging improves signal to background
(a)
Single frame,
1000 photons
(b)
(c)
9
Total
photons
in
dataset:
1.5x10
(a) Photographs
the target object, a plastic toy figure about 50 mm tall. (b) TypEMC ofalgorithm:
ame of data containing 96 photons in a 396x266 pixel detector. This translates to
10 4 photons/pixel.
(c) Data frame with 1025 photons obtained by combining 10
•
utive frames from the previous data set. The sizes of the pixels recording photons
een enlarged
to improve visibility. Pixels with two photons are shown in red.
•
Start with random 3D model
Distribute frames in angle,
weighted by how they fit to
neration
model
omographic
reconstructions using the EMC algorithm were tested using a plastic
ly matching
the size of the detector, as the object of study. A 50 watt molybdenum
• Repeat
tube was used to generate x-rays (Tru-Focus TCM-5000M). The applied high volt-
kV and the current was set to 0.05 mA. A 400 micron zirconium filter was used to
the K line of molybdenum (17.5 keV) from the x-ray tube emission spectrum by
y attenuating lower energies and higher energies (beyond the K-edge of zirconium,
he distance between the x-ray source and the sample was 1.3 meters. The detector
directly after the sample. Detector pixels measured 150 µm ⇥ 150 µm. The perure time was chosen to be 4 ms with an average signal of 100 photons per frame
ns/pixel/frame).
c figure (Fig. 1(a)) was mounted on a post and attached to a rotation stage (Newport
K. Ayyer et al Opt. Express 22 (2014)
The assembly and orientation problem can be
solved by accumulation of self correlation functions
Number of patterns
required is proportional
to SNR2
This method can utilize all 30,000 pulses
per second, even if the detector can only
store 3,000. (Incoherent addition)
Dmitri Starodub
Summary
“Diffraction before destruction” holds to <1.9 Å resolution.
Hundreds of thousands of 2D diffraction measurements can be
combined into a dataset of average 3D structure factors
Global (isotropic) damage is dose rate dependent. The faster you
go the more you can outrun
Atomic motion:
10 – 100 fs
1 GW/μm2
Electron cascades:
1 – 10 fs
1 TW/μm2
Auger decay:
0.1 – 1 fs
1 PW/μm2
Bragg diffraction is self gating (choosing the Bragg peaks filters
out disorder), setting the exposure time for the measurement
Anomalous phasing works over a wide range of doses
High intensities may provide a way for de novo phasing
LCLS experiments and analyses are carried out as a
large collaboration
!
CFEL-DESY
A. Barty, M. Liang, T. White, D. Deponte, S. Stern, A. Martin, C. Caleman,
K. Beyerlein, R. Bean, R. Kirian, K. Nass, F. Stellato, C. Yoon, F. Wang,
H. Fleckenstein, L. Gumprecht, L. Galli, S. Bajt, M. Barthelmess, O. Yefanov,
D. Oberthür, C. Gati
ASU
J. Spence, P. Fromme, U. Weierstall, B. Doak, X. Wang, I. Grotjohann, R. Fromme,
N. Zatsepin, D. Wang, D. James, S. Basu
MPG Med. Res.
I. Schlichting, R. Shoeman, L. Lomb, S. Kassemeyer, S. Bari, T. Barends,
J. Steinbrener, S. Botha
SLAC-PULSE
M. Bogan, D. Starodub, R. Sierra, C. Hampton, D. Loh
SLAC-LCLS
S. Boutet, G. Williams, M. Seibert, C. Bostedt, M. Messerschmidt, and many
others
Uppsala
J. Hajdu, N. Timneanu, J. Andreasson, M. Seibert, F. Maia, M. Svenda, T. Ekeberg,
J. Andreasson, A. Rocker, O. Jonsson, D. Westphal
Euro XFEL
A. Aquila
LLNL
M. Frank, M. Hunter, S. Hau-Riege
LBNL
S. Marchesini, J. Holton
Gotheburg
R. Neutze, L. Johansson, D. Arnlund
U. Hamburg!
L. Redecke, C. Betzel
U. Auckland
P. Metcalf
U. Melbourne
A. Martin
CFEL DESY Theory
R. Santra, S.-K. Son

Download Report