1. Binding Quantification with Thermophoresis
Seidel S, Dijkman PM, Lea WA, van den Bogaart G, Jerabek-Willemsen M, Lazic A, Joseph JS, Srinivasan P, Baaske P, Simeonov A, Katritch I, Melo FA,Ladbury JE, Schreiber G, Watts A, Braun D, Duhr S
Susanne Seidel and Dieter Braun from Systems Biophysics, and Philipp Baaske, Moran Jerabek –Willemsen and Stefan Duhr from NanoTemper Technologies.
Speaker: Christian Niederauer 1

2. Outline Introduction Theoretical Background Experimental Approach
Signal Analysis 2

3. What characterizes MST?
microscale thermophoresis
quantifies biomolecular interactions based on thermophoresis
various molecular properties are influencing MST
low sample consumption
flexible assay design (with/without fluorescent labels)
measurements in cell lysate or complex buffers possible
Binding events involving proteins and other biomolecules play a
central role in all fields of life science, from molecular physiology
and pathology to diagnostics and pharmacology because they allow for fundamental insights into molecular biology.
difference to other binding quantification which measure only the size (f.e. electrophoretics)
f.e. isothermal titration calorimetrie: high sample consumption 3

4. Theoretical background
Thermophoretic flow: Diffusion flow: Steady State: Integration: Concentration changes due to thermophoresis  readout trough measurement of fluorescence
Carl ludwig: directed movement of molecules in a temperature gradient
Identified the solvation entropy and the hydration shell of molecules as the driving force.
How excactly is not relevant for MST binding quantification.
Its I a non-equilibrium effect, still there is a steady state (but temp. is not equilibrated!!) 4

5. MST Setup TC: temperature-controlled tray OBJ: objective
FO: fluorescence observation
IR: IR-laser
HM: IR-reflecting hot mirror
 all-optical approach
Sample solution is kept inside a capillary placed on a temperature-controlled tray (TC)
Fluorescence of solution in capillary is observed (FO) through Objective (OBJ)
-> the capillary is locally heated with an IR-laser (IR) which is coupled into path of fluorescence excitation and emission with IR reflecting “hot”mirror
The heating laser is focused through the same objective used for fluorescence detection. This allows a precise local microscopic heating of the sample within the capillary and simultaneously and observation of local changes of fluorescence intensity due to the motion of labeled molecules in the glass capillaries. 5

6. If you haven‘t been to the sideroom, here is a picture of the NanoTemper Nanolith.

7. Optics Setup LED-filter combinations for fluorophore usage :
blue (excitation 460nm-480nm, emission nm)
green (excitation 515nm-525nm, emission nm)
red (excitation nm, emission nm)
LED-filter combination for label-free approach:
excitation 280nm, emission 360nm (both UV)
IR-laser: 1480nm creates temperature gradient
volume heated: 2nl by 1K-6K
Label free: to excite and detect the intrinsic UV-fluorescence of proteins
focused on exactly the same spot where fluorescence intensity is measured. [water absorbs IR really good, as we know from the greenhouse effect]
IR laser can easily be focused -> strong and localized beam, and overall temperature therefore remains low.
1/e extension: 25µm

8. Fluorescent Labeling Fluorescent labeling provides high sensitivity:
GFP
Fluorescent labeling provides high sensitivity:
sub-nM concentrations detectable
fluorescent dye coupled to crosslinker
crosslinker binds covalently to functional groups
non-natural amino acids already carrying a dye
fusion to a recombinant fluorescent proteins (GFP)
thermophorectic movement is detected through fluorescence
of one of the binding partners

9. Intrinsic Fluorescence
labels may influence binding interactions
label-free MST using intrinsic fluorescence
as low concentration as 100nM possible with > 2 TRP
quantifiable
Two or more tryptophane sufficient..
Label free of course not in complex liquids like cell lysate possible due to high protein concentration -> background signal too high
Tryptophane

10. Dilution series Capillaries
non-fluorescent partner titatred against fixed concentration of fluorescent partner
minimal concentration: unbound state dominant
maximal conc.: saturation of fully bound states ( )
Capillaries
variation of inner diameter less than 1µm
no diffraction & constant absorption of laser power
constant heat conduction
hydrophilic/hydrophobic coating possible
Initial fluorescence should be constant throughout the serial dilution, unless:
fluorophore is close to the binding sites or there are problems with aggregation

11. Signal Analysis
Fluorescence in the focal area of the IR beam is recorded during turning the laser on and off again (on time: constant power) after around 30s
Typical curve, the fluorescence decays when turning the IR laser on because labeled molecule concentration decays due to thermophoresis
But theres also the possibility of rasing concentration creating a increasing curve..later
The following stages are recorded for each sample: fluorescence signal before turning the IR laser on, fast temperature-dependent changes in fluorescence intensity, thermophoresis and back diffusion after switching the laser off. 11

12. Signal Analysis I. Initial fluorescence, constant for all samples 12
Approximately constant, minor errors (pipetting the dye) dont matter because of ratio
If F0 not constant for the samples , probably binding events affect the fluorophore (close to it f.e.)
-> we can infer the K_D from measuring F_0 for a titration series 12

13. Signal Analysis II. T-Jump due to temperature dependent fluorescence
II: This is no thermophoresis effect, ist just the temperature dependency. (heating is almost instantan because of low volume….few hundred ms)
T-jump can be also influenced by binding event, same as in I. 13

14. Signal Analysis III. Thermophoresis creates concentration gradient
Diffusion limited process .. few ten seconds. Lower concentration of labeled molecules  fluorescence decays
 reaches plateau when counterbalanced by mass diffusion 14

15. Signal Analysis
IV. Inverse T-Jump
Same as in region II 15

16. Signal Analysis V. Backdiffusion compensates concentration gradient
 initial fluorescence is nearly recovered
Turning IR-Laser off: mass diffusion compensates concentration gradient and initial fluoerescence is approximately reached (minus bleaching of fluorophores) 16

17. Signal Analysis Ratio: 17
F_1 is a few seconds after T-Jump. You do not have to reach the plateau, just has to be the same timespan for the whole titration process
F_0 can be initial fluorescence in area I or shortly after (1s) T-Jump.
Typically both ways achieve same results but if T-jump is influenced by binding, better choose F_0 as region I
Then the change of the T-jump during titration is included as F_0 is without T-jump and F1 is with. Otherwise both had T-Jump in, and therefore cancelling out when taking ratio
Including T-Jump can provide a better signal to noise ratio 17

18. Signal Analysis . Ratio: fraction bound 18
Relative fluorescence F_Norm is used to quantify binding
We can split the fluorescence up into unbound and bound fluorescences. We define B as the reaction partner (which may be labeled and) whose concentration is constant. Then AB divided by B is the fraction of the bound molecules.
We can write this in this way because the fluorescences add up linearly.
We can multiply the equation out and get that the normalized fluorescence is directly proportional to the bound fraction plus a constant.
For most applications the law of mass action is sufficient (but one can add things like Hill’s eq. for cooperativity etc) and we get a equilibrium dissociation constant K_D as shown. As we don’t know the free concentrations we have to use total concentration [A] meaning free partner plus concentration of complex. 18

19. Signal Analysis . Ratio: 1) 19
We can multiply the equation out and get that the normalized fluorescence is directly proportional to the bound fraction plus a constant.
For most applications the law of mass action is sufficient (but one can add things like Hill’s eq. for cooperativity etc) and we get a equilibrium dissociation constant K_D as shown. As we don’t know the free concentrations we have to use total concentration [A] meaning free partner plus concentration of complex. 1) 19

20. Signal Analysis with Law of Mass action: 𝐴+𝐵⇋𝐴𝐵 2) 20
For most applications the law of mass action is sufficient (but one can add things like Hill’s eq. for cooperativity etc) and we get a equilibrium dissociation constant K_D as shown. As we don’t know the free concentrations we have to use total concentration [A] meaning free partner plus concentration of complex. 2) 20

21. Solve 2) for fraction bound:
Signal Analysis
Solve 2) for fraction bound:
Now we solve for fraction bound.
K_D being the single free parameter. 21

22. Solve 2) for fraction bound:
Signal Analysis
Solve 2) for fraction bound:
F_Norm from the MST measurement linearly reports FB and can thus directly be fitted to this equation.
This way, we receive a value for K_D.
linear 1)
fit 2) 22

23. Plot is plotted on linear y-axis in ‰
x-axis is log10 of concentration of titrated partner
sigmoid-shape with bound & unbound plateaus
We get F_norm_unbound by extrapolating the F_Norm plot for the unbound plateau.

24. Plot is plotted on linear y-axis in ‰
x-axis is log10 of concentration of titrated partner
sigmoid-shape with bound & unbound plateaus
is revealed and can be subtracted, getting
determine 𝐾 𝐷 by fitting
F_Norm is linear correspondating with unbound fraction. As we saw [slide19 | 1)] , by substracting F_norm_unbound we receive Delta F Norm which is directly proportial to fraction unbound (and not only linear as in slide 22).
Delta F Norm then is directly proportional to unbound fraction. Hence, we fit [*] the equation (with sqrt etc) to the mean values of Delta F Norm from independent measurments. K_D value is received with error estimation (assuming Gaussian error distribution)

25. Plot
By fitting the change in thermophoretic depletion upon titration of wt-BLIP to a constant amount of wt-TEM1 labeled with the fluorescent dye NT657 to the quadratic solution of the mass action law, a binding constant of K_D =3.8 \pm 0.8 nM was determined.