[email protected] Studying protein-protein interactions Ed Evans, T-cell biology group

[email protected] www.t-cellbiology.org/teaching

Studying protein-protein interactionsEd Evans, T-cell biology group


Studying Protein-Protein Interactions

A. INDIRECT (looking for functional association)1. Correlated mRNA Expression2. Computational Approaches3. Phylogenetic Profiling4. Synthetic Lethality

B. QUALITATIVE1. The Two-Hybrid Method2. Mass Spectrometry of Affinity-Purified Complexes3. FRET & BRET

C. QUANTITATIVE1. SPR (BIAcore)2. AUC3. Calorimetry


Indirect detection of interactions(looking for implied functional

association NOT direct interaction)


A. 1. Correlated mRNA expression


A. 2. Computational approaches

e.g. “Rosetta Stone”


A. 2. Computational approaches


A. 3. Phylogenetic Profiling


A. 4. Synthetic Lethality


Qualitative detection of protein-protein interactions


B. 1. The Two-Hybrid Method


B. 2. Mass Spectrometry of Affinity Purified Complexes


•Immunoaffinity•TAP tagging•2D gel•Formaldehyde crosslinking•etc…..

Gel

MS compatibleSilver stain,SYPRO stainCoomassie stain

>100 fmol protein

Specific Proteasee.g. trypsin

LC MSMSPROTEIN IDENTIFICATION

Q-ToF Micro Mass Spectrometer – LC MSMS

ProteinDigest

Nano HPLC system

NanosprayIon source

Quadrupole Time-of-flight mass spectrometer

Data acquisition

Peptides

CID

Peptidefragments

Peptide sequence

Basic Workflow


“Mass-fingerprint” Indentification


Non covalent protein complex

Thiol cleavable cross-linker

Covalently cross-linked complex

Digest with Protease

Thiol reagent

MALDI MS

MALDI MS

Differential peptide mapping

Non reduced

Reduced

Cross-linking the interaction


Summary of current effort in yeast


...and the bad news


=> BE WARNED!

These techniques (along with e.g. Co-immuniprecipitation) give lots of false positives


Förster (Fluorescence) Resonance Energy Transfer (FRET)

In this strategy, excitation of GFP will result in emission from a nearby protein such as blue fluorescent protein (BFP) if it is physically close enough. The best FRET pairs are actually the cyan and yellow mutants of GFP, referred to as CFP and YFP.

B. 3. a. FRET


Power of FRET

1. Probe macromolecular interactionsInteraction assumed upon fluorescence decay

2. Study kinetics of association / dissociation between macromolecules

3. Estimation of distances (?)4. In vitro OR on live cells5. Single molecule studies


FRET


Live cell FRET imagingDoes CD4 specifically associate with the TCR/CD3 complex on triggering?

Non-specific peptide Specific peptide

* marks contacts between cells. High FRET signal between CD4 and CD3 when correct antigen is present but not with non-specific antigen.


DeepBlueC hf1 hf2

Luciferase >10nm

GFP2

B. 3. b. BRET: Bioluminescence Resonance Energy Transfer


• BRET analysis can be achieved at physiological levels of protein expression

• No problems with photobleaching or photoconversion as seen in FRET techinques (no laser stimulation)

• Both methods involve the same physical processes and so can be analysed in a similar manner

• BRET cannot be used in microscopy-based techniques such as FRAP or FLIP, or FACS-based analysis

BRET vs FRET


• The gene of interest is fused to both luciferase (donor) and GFP (acceptor) in two separate vectors

• A positive control is used to determine maximal BRET

Construction of Fusion Proteins


B7-1luc:B7-1YFP

CTLA-4luc:CTLA-4YFP

B7-1luc

B7-1luc:CTLA-4YFP

YFP

luc

B7-1YFPB7-1luc

substratehu2 (530 nm)

hu1 (470 nm)

e.g. B7-1 BRET


Energy transfer can occur solely by random interactions

e.g. BRET on B7 family


Strong dimers

Weak dimer

Monomers

Comparison to T cell surface molecules with known oligomerisation status!


0 1 2 3 4 50.0

0.1

0.2

0.3

0.4

0.5

BR

ET

Rat

io

GFP / Rluc

hCD80 - CTLA-4 hCD80 + CTLA-4 hCD86 - CTLA-4 hCD86 + CTLA-4

Specific ligand engagement can be observed when receptor is presented in solution or cell-surface bound

Ligand binding causes specific increase in dimerisation


Measure Quantitative Properties

SPR(BIAcore)

AUC ITC(microcalorimetry)

Surface Plasmon

Resonance

AnalyticalUltracentrifugatio

n

IsothermalCalorimetry


Measuring key properties of protein-protein interactions

Property AUC BIAcore Calorimetry

Affinity + ++ +

Enthalpy no + ++

Entropy no + ++

Heat capacity no + ++

Kinetics no ++ no

Stochiometry + + ++

Size & Shape + no no


C. 1. SPR / BIAcore(Surface Plasmon Resonance)


Advantages of SPR on the BIAcore

1. No labelling is necessary2. Real-time analysis allows equilibrium

binding levels to be measured even with extremely rapid off-rate.

3. Small volumes allow efficient use of protein. Important when very high concentrations are required.

4. No wash steps => weak interactions OK5. All types of binding data obtained –

including kinetics as its real-time.


Principle of Surface Plasmon Resonance

Angle of ‘dip’ affected by:1) Wavelength of light2) Temperature3) Refractive index n2

Dip in light intensity


Surface Plasmon Resonance in the BIAcore


2 Main options:• Direct:

Covalently bind your molecule to the chip• Indirect:

First immobilise something that binds your molecule with high affinity e.g. streptavidin / antibodies

Direct: Indirect:

Immobilisation


Sensorgram for ligand binding


• Each chip has four ‘flow-cells’• Immobilise different molecules in each flow-cell• Must have a ‘control’ flowcell• ‘Specific binding’ is the response in flow-cell of

interest minus response in the control flowcell

“Specific” Binding

Response in control / empty flowcell due to viscosity of protein solution injected – therefore ‘control’ response DOES increase with concentration (this is NOT binding!!)

Specific response in red flowcell

Measured response

Is it specific?


Binding curve can be fitted with a Langmuir binding isotherm (assuming a 1:1 binding with a single affinity)

d

Max

KA

ARBound

][

][

Scatchard plot: rearrangement of binding isotherm to give a linear plot. Not so good for calculating Kd, as gives undue weight to least reliable points (low concentration)

Plot Bound/Free against BoundGradient = 1/Kd

dd

Max

K

Bound

K

R

A

Bound

][

Equilibrium Binding Analysis


HarderCase:2B4

binding CD48

Kinetics


• Protein Problems: Aggregates (common)Concentration errorsArtefacts of construct (eg Fc

linked)• Importance of controls: Bulk refractive index issues

Control analyteDifferent levels of immobilisationUse both orientations (if pos.)

• Mass Transport: Rate of binding limited by rate of injection: kon will be

underestimated• Rebinding: Analyte rebinds before leaving

chipkoff will be underestimated

Last two can be spotted if measured kon and koff vary with immobilisation level (hence importance of controls)

Potential pitfalls


1. Temperature dependence of binding

van’t Hoff analysis: STHKRTG a )ln(

R

S

TR

HKa

1

)ln(

Gradient

Intercept

Less common applications


1. Temperature dependence of binding

Non-linearvan’t Hoff analysis:

0,0,,, ln)(

00 T

TCTTTCSTHG vHpvHpTvHTvH



2. Combination with mutagenesis

Q30R Q40K R87A

Binding of CD2 by CD48 mutants at 25°C (WT Kd = 40M)


Reduce / abolish bindingDo not affect bindingNot tested


3. Estimation of valency



4. Screening

Newer BIAcore machines are capable of high throughput injection. With target immobilised, many potential partners / drugs can be tested for binding.

5. Identification of unknown ligands

Mixtures e.g. cell lysates, tcs, food samples etc. can be injected over a target and bound molecules can then be eluted into tandem mass spectroscopy for identification.



CD48 binding to immobilised CD2(van der Merwe et al.)

What a lot of people would have used(straight out of the freezer)

Correct result

One last warning: take care


2. AUC(Analytical Ultracentrifugation)


Theory: The Svedberg equation

1. Consider a particle m in a centrifuge tube filled with a liquid.

2. The particle (m) is acted on by three forces:

a) FC: the centrifugal force b) FB: the buoyant force

(Archimedes principle) c) Ff: the frictional force

between the particle and the liquid

3. Will reach constant velocity where forces balance:


• Define s, the sedimentation coefficient:

s =

• s is a constant for a given particle/solvent, has units of seconds, but use Svedberg (S) units (10–13 s).

• Cytochrome c has s=1S, ribosome s=70S, composed of 50S and 30S subunits (s does not vary linearly with Mr)

• Values for most biomolecules between 1 and 10000 S



S =

f RT

ND

D = diffusion coefficient, N = Avogadro’s number

sm0(1 )

RT NDor

RTs NDm0(1 )

Mr RTs

D(1 )

• Therefore can directly determine Mr in solution by measuring physical properties of the particle (s and v) under known experimental conditions (D, T and ),

• c.f. PAGE, chromatography – comparative & non-native

(Because Mr = Nm0)



AUC – analytical ultracentrifugation

•Spin down protein at various concentrations and follow its distribution in the cell by OD.

•Equilibrium Analysis:Spin slowly - centrifugal force and back-diffusion reach equilibrium. Distribution depends on average mass. If this increases with concentration then association is occurring and affinity can be estimated.

•Velocity Analysis: Spin fast & follow speed of boundary descent. Depends on mass and shape– can fit multiple distributions to estimate number of species and their properties. Dependence on concentration again gives affinity.


AUC – analytical ultracentrifugation

• Generally less precise than others.

• Key advantages are:

1. Works well for homomeric association, which is hard to follow with other techniques

2. Estimates size & shape – useful. In its own right and also for quality assessment


Equilibrium sedimentation

1. Moderate centrifuge speed2. After sufficient time, an

equilibrium is reached between sedimentation & diffusion, resulting in a montonic solute distribution across the cell

Cell bottomMeniscus

• Non-linear curve fitting can rigorously determine:– the solution molecular

weight– association state– equilibrium constant for

complex formation


Data modeling

1. A plot of ln(c) vs r2 should be a straight line with a slope proportional to molecular weight

Single ideal homogeneous species Mp(1- ) = d ln(c) 2RT d r2 2


Testing for monomorphic protein

little or no curvature

10 ºC, 200 mM NaCl 40 ºC, 100 mM NaCl

26K

19K

31K

40K

obvious curvature = variation in mass i.e. unstable protein leading to aggregation


Protein concentration (mg/ml)

6

5

4

3

20 1.0 2.0

Mw

,ap

p(D

a/1

04)

sB7-1

B7-1 : an equilibrium dimer


sB7-2 sLICOS

Concentration (mg/ml) Concentration (mg/ml)

Mw(k

Da)

Mw(k

Da)

0 1 2 3 4 0 1 2 3 4

80

60

40

20

0

80

60

40

20

0

B7-2 and LICOS are monomeric


Velocity sedimentation

• High centrifuge speed• Forms a sharp boundary between

solute depleted region (at top) and a region of uniform solute concn (at bottom)

• The concentration gradient (dc/dr) defines the boundary position

Non-linear curve fitting can rigorously determine:• number of mass species • molecular weight • shape information for a molecule of known mass


g(s*) distribution

Velocity sedimentation - data analysis


The example of SLAM (CD150)1. Claimed to self-associate with nM Kd raising serious

problems for models of cell surface protein interactions2. Equilibrium data can’t be fitted – high concentrations!3. Velocity data confirmed shape of complex and

approximate strength of association


3. ITC(Isothermal Titration Calorimetry)


Isothermal Titration Microcalorimetry:Using the heat of complex formation to

report on a binding interaction.The Basic Experiment:1. Fill the upper syringe with ligand at

high concentrations.2. Fill the larger lower reservoir with

protein at a lower concentration.3. Titrate small aliquots of ligand into

protein.4. After each addition, the instrument

returns the reservoir temperature to the temperature of the control cell and measures the heat required to cause this change.

5. Typically, subtract appropriate blank titrations (ligand into buffer & buffer into protein) to control for heats of dilution.


Microcalorimetry

1. Two proteins are mixed and the heat release upon binding is measured

2. Provides a direct measure of the H (whereas van’t Hoff analysis is indirect)

3. Allows more accurate measurement of C

4. Can also determine G and => T S

5. Its disadvantage compared with the BIAcore is that very large amounts of protein are required and no kinetic data are provided


ITC Data AnalysisGet a plot of heat (J or Cal) / s following each injection, integrate peaks for total heat released and plot against concentration of protein injected – binding isotherm.

c = concn / Kd


Data Analysis – e.g. of B7-1 & CTLA-4

0 1 2 3 4

-12

-8

-4

0

kcal/m

ole

of

inje

ctant

molarratio

H = -11.6 G = -8.9 TS = -2.7 kcal/mol-1

1. Curve fitting gives values for H (enthalpy) and G (Gibbs free energy, related to affinity) – from these one can also calculateS (entropy).


Calculating heat capacity

1. H and S are not constant with temperature, hence direct measurement by ITC is better than deriving them from binding data across several temperatures (e.g. by SPR)

2. Relationship of H to temperature can be used to calculate heat capacity change on binding (Cp)


Studying Protein-Protein Interactions

A. INDIRECT1. Correlated mRNA Expression2. Computational Approaches3. Phylogenetic Profiling4. Synthetic Lethality

B. QUALITATIVE1. The Two-Hybrid Method2. Mass Spectrometry of Affinity-Purified Complexes3. FRET & BRET

C. QUANTITATIVE1. SPR (BIAcore)2. AUC3. Calorimetry

Bulk screeninge.g. For databaseNEED TESTINGAFTERWARDS

When looking for/at a (or a few) specificinteractions

Documents

[email protected] Studying protein-protein interactions Ed Evans, T-cell biology group