Combined use of SANS and SAXS - Eventsevents.embo.org/13-SAXS/.../Arleth_EMBO_SAS_Grenoble_May13.pdf · Combined use of SANS and SAXS Lise Arleth StructuralBiophysics Niels Bohr Institute

Combined use of SANS and SAXS

Lise ArlethStructural BiophysicsNiels Bohr InstituteUniversity of Copenhagen

Just finished SAXS/SANS experimenthere in Grenoble. Pic from lunch yesterday (May 8, 2013) aftersuccessful termination of SANS experiment

….And still a bit hung over….

Agenda

• Motivation: Why and which type of samples• The experiment: How to combine SAXS and SANS in practice

• The data analysis: How to get the most out of your data

• Detailed example:

Nanodiscs: Lipid-protein complexes• Conclusions and outlook

X-ray contrast Neutron contrast 1

Neutron contrast 3

Neutrons give the possibility of not seing everything at the same time…. X-rays help assembling the whole picture

Why combining neutrons and X-rays

Neutron contrast 2

D2O-based solvent

D2O-based solvent

Deuterated membrane

H2O/D2O-based solvent

Small-angle scattering:Different structures – Different scattering patterns

Spheres:R = 60 Å

Rods: R = 60 ÅL = 1200 Å

I(Q

) 1/

cm

Q 1/Å

I(Q

) 1/

cm

Q 1/Å

Worms: R = 18 ÅL = 5000 Å,

Kuhn Length = 300 Å

I(q)=φφφφ V (∆ρ)∆ρ)∆ρ)∆ρ)2222 P(q) S(q)

P(q): Particle form factorS(q): Structure factorφ: Concentration (vol/vol)V: Volume of single particles (Å3)∆ρ: Excess scattering length density

SANS versus SAXS

SAXSX-rays are scattered by electrons (with a scattering length of 2.82 fm/electron):→Scattering length densitydetermined by electron density

Water: 0.33 electrons/Å3 Proteins: ~0.43 electrons/Å3X-rays ”see” electron density fluctuations

SANSNeutrons are scattered by nuclei. Most bio-relevant isotopes have a scattering length around 5 fm.

One very important exception:

Neutrons ”see” hydrogen density fluctuations.

I(q)=φφφφ V (∆ρ)∆ρ)∆ρ)∆ρ)2222 P(q) S(q)

P(q): Particle form factor S(q): Structure factorφ: Sample conc. (vol/vol) V: volume of single particles∆ρ: Excess scattering length density

Hydrogen

Isotope

Scattering Length

b (fm) 1H* -3.7409

2D 6.674

3T 4.792

Different contrast situations for SAXS and SANS

Consequence: Internal structure gives different SAXS and SANS data on the same sample

DSPE-PEG5000:EYPC 50:50

Rod shaped polymer modified

micelles

Lc=120 Å

Nagg=280

RG, PEG=30 Å

SAXS

SANS

L. Arleth, B. Ashok, H. Onyuksel, P. Thiyagarajan, J. Jacob, and R. Hjelm.

Langmuir, 2005, 21(8), 3279-3290.

SANS: IPNS, ArgonneSAXS: APS, Argonne (Il, USA)

In which systems is it relevant to combine SAXS and SANS?

Monomeric non-glycosylated Proteins in solution – No point

Glycosylated proteins - Yes(Homo/hetero)-Oligomeric proteins in solution – Yes, in combination with

specific deuterationProtein-DNA/RNA complexes? -YesSurfactant micelles? -YesBlock co-Polymer systems? - Yes Lipid-protein complexes? -Yes

But also to check if deuteration of a sample or its buffer has changed sample structure (avoid different samples at different contrast)

Lots of good reasons☺☺☺☺

Should we use SAXS or SANS to investigateproteins in solution?Small-angle scattering from protein in buffer:

Proteins are extremely monodisperse, have a mass density of ~1.35-1.38 g/cm3 and are homogenous on length-scales above~10-20 Å (See e.g. Mylonas and Svergun, J. Appl. Cryst. (2007), 40, 245–249).

NB: In 100% D2O

SAXS: SANS:

Should we use SAXS or SANS to investigateproteins in solution?Small-angle scattering from protein in buffer:

SAXS SANS∆ρprotein 3.0 × 1010 1/cm2 3.4 × 1010 1/cm2

Sample volume (exposed volume)

~40 µL (1-20 µL) ~200 µL (150 µL)

Beam intensity 1e7 - 1e12 photons/s < 1e8 neutrons/s

Measurement time 10 ms-2 hours 2 minutes – 2 hours

Radiation damage Yes, sometimes No

Typical (good) q-range 0.007-0.5 1/Å(π/q): 6 Å-450 Å

0.001 – [0.2;0.5] 1/Å(π/q): [6Å;16Å]-3000 Å

180 Å

Method of choicefor standard ”proteins in solution” exps.

Crucial in a range of ”difficult” cases

General characteristics of ”Bio”-Samples:

Small sample volumes and low sample concentrations:< 1 mg of dry protein< 200 µµµµL of sample (@ 1 to 5 mg/ml)

Samples degrade fast and do not withstand freezing =>Demands for freshly purified and prepared samples (= a lot of planning and long time in a well equipped sample prep. lab ahead of measurements)

Combining SANS and SAXS

Simultaneous experiments

Disadvantage: Co-scheduling of SANS and SAXS is logistically demanding (advantage if SAXS and SANS instruments are physically near)

Advantage: The exact same samples are measured in the two different contrast situations

=> Optimal basis for simultaneous analysis

Serial experiments

Advantage: Easy to plan due to decoupling of SANS and SAXS scheduling –Samples are prepared freshly for each exp.

Disadvantage:Difficult/impossible to check whether the two samples are really totally similar

=> Forms a weak basis for simultaneous analysis (in difficult cases)

Simultaneous SANS and SAXSCo-localized facilities:Grenoble: ILL and ESRFPaul Scherrer Institute (Switzerland): SINQ and SLSLund (Sweden): ESS and Max-IV (from 2019)Paris: Soleil and LLB

ESRF+ILL: My present favorite combination

NB: Combined FRMII (Munich) and ESRF experiment successfully carried out in September 2012.

SANS before SAXS? –Or vice versa?

SAXS before SANS-SANS sample volume is larger than SAXS sample volume. -Radiation damage with SAXS, => SAXS samples can not be recycled for SAXSTotal sample volume=

SANS sample volume (280µµµµL)+SAXS sample volume (30µµµµL)

BioSaxs: ~200 samples/24h SANS: ~20 samples/24h SAXS allows for prescreening the quality of the SANS samples

=>SAXS before SANS is my preferred order (at present) despite slightly larger sample volume requirements

SANS before SAXS-SANS sample volume is larger than SAXS sample volume. -No radiation damage with SANS, SANS samples can be recycled for SAXSTotal sample volume=SANS sample volume (280µµµµL)

But:

Contamination issues: SANS samples may (theoretically) be neutron activated and become radioactive => Samples are checked for contamination before exit from ILL, and left to “cool down” and then rechecked for radioactivity (by ESRF) upon entrance to the ESRF

What: Phospholipid ”disc” stabilized by two amphipatic protein belts (His-tagged 220 AA 8-alpha helical structure)

-Nanodiscs are interesting in their own right and from a membrane physics point of view-But really interesting as a ”sample-holder” for structural and functional studies of membrane proteins

AND: Lots of internal structure, which makes the particlesideally suited for combined SAXS and SANS

See Pie Huda’s poster

Data analysis: Example/Model system:Nanodiscs – Empty and membrane protein loaded

Fundamental: The geometrical description of the particles is the same in SAXS and SANS – only the contrast is different.

IFT analysis: Still works – but our intuition from homogenous particles breaks down.

Further analysis: Nanodiscs: Use Geometrically based modelling for Nanodiscs. -But same principles apply to any type of modelling (cf lectures earlier this course)

Simultaneous analysis of SAXS and SANS data

Geometrically based modelling -The primitive approach:

Same geometrical shapes of different parts of particle, different contrasts as determined from model-fit (MONSA-type approach)

Simultaneous analysis of SAXS and SANS data

Works sometimes and would work more often if we did not have the uniqueness problem in Small-angle scattering:

-Infinitely many structures will give rise to a given I(q)

-And there are still many solutions to the problem of fitting the three contrasts with a given fixed geometrySolution: Start by removing the unphysical solutions by incorporation of molecular constraints

Constraining the analysis to the physical solutions: Incorporation of molecular constraints

Using molecular constraints <=> Exploit thatnanodiscs are build by well-known building blocks:-Hydrophobic core consists of alkyl-chains-Hydrophilic caps consists of PC headgroups and hydration water-Belts consists of MSP

-The molecular volumes and scattering lengths of all constituents are known-Sample concentrations incl. Belt:Phospholipid ratios are knownNB: Both with experimental uncertainties of a few %

Calculating the X-ray contrast of POPC from firstprinciples:

POPC

Palmityl-oleyl-Phosphatidyl choline

X-ray contrast

νννν, Å3 b, cm ρ=ν/ρ=ν/ρ=ν/ρ=ν/b, , , ,

cm/cm3∆ρ=ρ∆ρ=ρ∆ρ=ρ∆ρ=ρs−−−−ρρρρw, , , ,

cm/cm3

Water 30 2.82e-12 9.4e10 0

Head-group 319 4.62e-11 14.5e10 5.1e10

Hydrated head-group (Nh=4)

319+4×30=439

5.75e-11 13.1e10 3.7e10

Hydrophobic tail

928 6.71e-11 7.2e10 -2.2e10

Contrasts and scattering volumes can not vary freely!Scattering volume depends on molecular volumes. Contrast depends on molecular scattering lengths and molecular volumes. –The two are interrelated

Calculation of scattering data on an absolute scale:

P(0)=1. and at low concentrations S(q)�1.

So at q=0 and low c:

- We know the sample concentration – or should be able to determine it.− ∆ρtot and V comes out of our (molecular constrained) model=> I(0) and I(q) can be calculated on absolute scale

A note on software: Implementation of molecular constraints in computer programs for analysing SAXS/SANS data

Most existing soft-ware is based on purely geometrical form-factors (e.g. SASFIT (Kohlbrecker,PSI), Bodies (ATSAS package)).

Nice and quick - but too general and usually not sufficient.

=>Normally necessary to code your own models + fitting routines or adapt existing software to your problem.

Program structure for programs that incorporate molecular constraints:

• Load SAXS and SANS data (incl. experimental error bars)• Load other data: Sample concentration, molecular volume and

scattering lengths of constituents• Establish model to be fitted to SAXS/SANS data and link to information

from other data: Define fit parameters and deduced parameters for a given geometry. Implement molecular constraints, calculate volumes and excess scattering lengths of the different parts of the object

• Calculate analytical form factor and make fit in terms of free fit parameters

Lise Aleth, Biophysics, KU-Life

What: Phospholipid ”disc” stabilized by two amphipatic protein belts (His-tagged 220 AA 8-alpha helical structure)

System developed by Sligar group at U. Illinois, and derivedfrom APO-A1 high density lipoproteins (HDL)

Detailed Example: The nanodisc system

From the Sligar lab homepage:http://sligarlab.life.uiuc.edu/nanodisc.html


NB: Almost all the studies are functional studies. These are apparently a lot easier than structural studies

Why studying membraneProteins

Structural Biophysics

• ~30% of the proteins in the cell

• ~50% of all drug targets today

• Extremely challenging to study

• 80 000 protein structures in PDB

~366 Unique membrane proteins

The Nobel Prize in Chemistry 2012

Robert Lefkowitz and Brian Kobilka

GPCR(2012-11-19)

Aquaporin-tetramer in DDM micelles.

Berthaud, Manzi, Perez, Mangenot, JACS, 134(24), 10080, 2012

Challenges associated with SAXS/SANS studies of membrane proteins

Niels Bohr Institute - Structural Biophysics

Dias 24

Potassium Channel KcsA in DDM micelles.

Calcutta, Jessen, Behrens, Oliveira, Pedersen et al, Biophysica BiochemActa, 1818, 2290, 2012

-Sample handling is more difficult than for water soluble proteins-SAXS/SANS data analysis is non-trivial because the entire MP +

carrier construct has to be accounted for explicitly

NanodiscBicelle Peptide discDetergent

Dias 25

Different means of reconstituting membrane proteins

-Complicated reconstitution-More native-like lipid environment-Monodisperse => Optimal for SAXS/SANS analysis

-Simple reconstitution-Non-native lipid environment-Polydisperse =>Less optimal for SAXS/SANS analysis

The monodispersity of Nanodiscs and their native-like lipid environment make them ideal as molecular sample holders of membrane proteins for structural characterization (In theory)

Research goal: Develop phospholipid nanodiscs into a platform for Small-angle scattering based low-resolution structural studies of membrane proteins

Main targets:

+ Leucine transporter: 12 TM’s

GPCR’s: 7 TM’s(Use Bacteriorhodopsinin the developmentphase)

Central challenges:Incorporation of Membrane proteins into Nanodiscs: Obtaining sufficiently”pure and well-defined” samples.

Data analysis: Development of experimental and computational methodsto extract relevant structural information about the membrane proteins.

P450: Membrane anchored protein’s

MD Simulation (Shih, Sligar, Schulten et al, Biophys. J 2005)=>Circular discs. But model did not agree well with SAXS data

SANS and SAXS study of cholesterol containingPOPC-APO-A1 (Wu et al, JBC, 2009)=> Double super helix model

Structure and control of the empty nanodisc system

At project start in 2009: Not full consensus in the literatureabout what nanodisc’s/HDL’s looked like:

Small-Angle Neutron Scattering study ofDMPC-APO-A1 (Nakano et al, JACS, 2009)Geometrical based modelling=> Circular discs. But very low res. data

Initial experimental work: SAXS and SANS measurements to determine the structure of the empty nanodiscs:

DLPC NanodiscsBelt: MSP1D1 (His-tagged membrane scaffold protein belt)Lipid: DLPC (Di-Lauryl-Phosphatidylcholine)(+ DMPC and POPC)

SANS in 100% D2O:Bulk-contrast of entire ND

SANS in 42% D2O:Protein belt matched out. Bulk-contrast of lipid core

SAXS contrast:Complex contrast situation: Different contrasts of protein belts, of hydrophobic alkyl chains of lipid and of hydrophilic headgroups.

Data from D11-ILL/ID14.3-ESRF, Nov 2009

Lise Arleth - Structural Biophysics Mathematical model for the ND’s to interpret SAXS data

1. Derive analytical expressions to describe the scattering from nanodiscs (trivial but very long equations …)

2. Incorporate molecular constraints: Exploit that Nanodiscs are build by well-known building blocks of well-known chemistry and scattering length:-Hydrophobic core consists of alkyl-chains-Hydrophilic caps consist of PC headgroups and hydration water-Belts consist of MSPAll information about concentrations and chemical composition along with estimates for the partial specific molecular volumes is build into the mathematical model to secure self-consistency.

Lise Arleth - Structural Biophysics

Excellent simultaneous model fits when using a fully molecular constrained analytical model for elliptical, His-tagged nanodiscs!

Conclusion: The Nanodiscshave a flat elliptical disc shape. The His-tags are protruding and clearly visible!

Simultaneous fits to SANS/SAXS data => detailed structure of Nanodiscs:

Red: elliptical cross sectionBlue: circular cross section

Data from D11-ILL/ID14.3 ESRF, Nov 2009


Conclusions from SAXS/SANS study of empty nanodiscs

-Combined SAXS and SANS gives the detailed structure (and temperature dependence) of empty nanodiscs-The nanodiscs are disc-shaped and structurally very homogeneous with an elliptical cross-section (ε~1.4)-POPC is laterally stretched and DLPC is laterally compressed when located in nanodiscs (data not shown today)Stretching and compression of lipids can be understood as minimization of the “hydrophobic mismatch” between protein belt and lipid bilayer

Figure from Mouritsen, OG. 1984

POPC DLPC

Publications:-Skar-Gislinge, Simonsen, Mortensen, Feidenhans'l, Sligar, Lindberg Møller, Bjørnholm and Arleth, J. Am. Chem. Soc, 2010.-Skar-Gislinge and Arleth, Phys. Chem. Chem. Phys, 2011.

Immobilized mono-meric avidin resin

Next step: Reconstitution of a membrane protein into nanodiscs

size exclusion column

1. Mix detergent solubilized membrane protein, phospholipids, reconstitution detergent and MSP

2. Add biobeads to remove reconstitution detergent (or remove detergent by dialysis) 3. Remove Biobeads

and start purifying

Purify good Nanodiscs

Affin

ity column

Purify loadedNanodiscs

See more description e.g. in Denisovet al, JBC, VOL.282(10), pp.7066–7076, 2007


Note: Reconstitution of Membrane proteins into ND’s is difficult and still only poorly understood….

SAXS and SANS data from bR incorporated into POPC:POPG 2:1 nanodiscs


Dias 33

Dmax of ~120Å

SANS/SAXS data from D11-ILL/ID14.3-ESRF

Hybrid modelling approach

-Geometrical approach to describe the nanodiscs-Bead modelling approach (Sverguntype) to describe the incorporated membrane proteins.

NB: Full bead modelling approach also tested, but without success so far.


Dias 34

Details of the modelling approach


Dias 35

• Exploit that the crystal structure of bR is known. Only the position and orientation of the membrane protein is fitted in relation to the surrounding nanodisc.

• The excess scattering length of the different subdomains of the membrane protein are adapted as to which medium (water, lipid) the subdomain has excluded.

• The structure of the surrounding nanodisc is explicitly fitted and allowed to adapt to a lense shape.

• Molecular constraints are fully incorporated.• To gain computational speed, the amplitudes

from both the geometrical nanodisc and the MP are expanded in terms of spherical harmonics

• Software code implemented in a combination of C and Python. Equations checked for correctness by point based modelling (see Pedersen, Oliveira et al, Biophys J, 2012)

bR in Nanodiscs – Model fit resultsStructural Biophysics

bR slightly de-centered in the disc and slightly tilted. Tilt is in accordance with expectations from surface analysis and crystallography

Model gives information on both membrane protein and lipid environment

Insight into the perturbation of the lipid environment

-The loaded disc is significantly more expanded and has a larger circumference-Lipids are laterally more stretched in the loaded disc than in the empty discs


SAS2012, Sydney 2012Dias 37

Parameter Empty pc/pg Br Loaded

N lipids 126 130

Area per lipid (Å2) 63 78

Axis ratio 1.66 1.44

Comparison of lipid environment in bR-loaded and empty POPC/POPG discs:

Søren Kynde, Nicholas Skar-Gislinge, Martin Cramer Pedersen

MD simulation of P450 in lipid bilayer J Inorg Chem, (2012) 108, 150

Project in collaboration with Steve Sligar, Ilya Denisov and Xin Ye, U Illinois

SAXS data from BM29/ESRF


Dias 38

Cytochrome P450 3a4 in Nanodiscs – Experimental data

P450 in Nanodiscs – Fit results – “Dammin-type”Bead modelling


Initial condition:

Nicholas Skar-Gislinge, Søren Kynde, Martin Cramer Pedersen

After convergence:

Research goal: Develop phospholipid nanodiscs into a platform for Small-angle scattering based low-resolution structural studies of membrane proteins

Main targets:

+ Leucine transporter: 12 TM’s

Bacteriorhodopsin7TM

Central challenges:Incorporation of Membrane proteins into Nanodiscs: Obtaining sufficiently”pure and well-defined” samples.

Data analysis: Development of experimental and computational methodsto extract relevant structural information about the membrane proteins.

✔ ✔

✔

÷P450: Membrane anchored protein’s

✔

(✔)

Selma Maric and Søren Roi Midtgaard


Improving the conditions for the structural analysis:

⇒ Use specific deuteration of bio-building blocks and obtain ”stealthnanodiscs”, i.e. Nanodiscs that are invisible to neutrons so that onlythe membrane protein is seen.Collaboration with the D-lab at ILL, Grenoble

⇒ MSP1D1 expressed in E-coli under 85% deuterated conditions⇒ Deuterated PC based unsatured lipids obtained from geneticallymodified E-coli. Systematic ”feeding” of the strain with mixtures of deuterated/non-deuterated nutrients allows for controlling the deuteration level in heads and tails separately.


Stealth Nanodiscs - SANS Contrast Variation

SANS forward scattering as a function of the D2O in the buffer.

SANS data at different levels of D2O in solution.

Data from KWS2, FRMII, Munich Sep 2012

60%

70%

75%

80%

85%90%95%

100%

Selma Maric et al

Next step: Get membrane protein incorporated and investigate feasibility…

Conclusions & Outlook

Combined analysis by SAXS and SANS allows for obtaining unprecedented structural information about particles with internal structure.

The combination is very far from being fully exploited by the scientific community!

=>Careful experimental planning is required=>Careful data analysis is also required (fun+challenging☺☺☺☺)



Department of Micro and Nanotechnology

The SAS&Nanodisc group at U. Cph:

• Nicholas Skar-Gislinge, PhD student• Søren Kynde, PhD student• Martin Cramer Pedersen, PhD student• Selma Maric, PhD student• Søren Roi-Midtgaard, PhD student • Pie Huda, PhD student• Søren Skou Nielsen, Post Doc• Grethe V Jensen, Post Doc• Lise Arleth, Professor

• Kell Mortensen, Professor

Acknowledgements

Collaborators:-Steve Sligar and Ilya Densiov, U. Illinois-Several UNIK Synthetic Biology project partners at U. Copenhagen-Javier Perez and Gabriel David Synchrotron Soleil, Paris-Ralph Schweins, Peter Lindner, D11, ILL Grenoble-Adam Round, Petra Pernot, ID14-3, ESRF Grenoble-Martine Moulin, Trevor Forsyth, Michael Haertlein, D-lab, ILL Grenoble-Henrich Frielinghaus, KWS-2, FRMII, MunichFunding:

UNIK synthetic Biology project, Lundbeck Foundation, Danish Research Council,Danscatt, ESS

Thank you!

Documents

Combined use of SANS and SAXS - Eventsevents.embo.org/13-SAXS/.../Arleth_EMBO_SAS_Grenoble_May13.pdf · Combined use of SANS and SAXS Lise Arleth StructuralBiophysics Niels Bohr Institute