[YFP] (AU)...Fluorescence Microscopy: Fluorescence images of Sf21 cells were recorded using a Nikon TE200S microscope equipped with metal halide lamp and CCD camera. Automated filter

Förster’s theory shows that the relation between FRET efficiency (E) and thedistance between two chromophores (RDA) is described by:

E = 1 / [1 + (RDA/R0)6] (1)

The Förster distance (R0) is experimentally determined and is calculated from:

R0 = 9780[J(λ)2η-4φD]

1/6

Where J(λ) is the spectral overlap integral, 2 is the orientation factor (assumed tobe 2/3 for random orientation), η is the index of refraction of the medium, and φD isthe donor quantum yield. For the CFP-YFP pair R0(

2/3) = 49.2Å.

FRET efficiencies can be calculated from molecular simulations by determining thedonor-acceptor distance RDA, and the orientation factor

2. The angles θA, θD, and φcalculated for each donor-acceptor pair are used to calculate the orientation factoras:

2 = (sin θD sin θA cos + 2 cos θD cos θA)2 (2)

An adjusted R0 based on molecular simulation is calculated from each structure as:

R0() = R0(2/3) × (

3/2)1/3 (3)

The 2 adjusted R0 is then used in Eq 1 to calculate the FRET efficiency (E). This datacan be used to compare with experimental results to determine whether a structuralmodel based on a crystal structure or MD simulation is in agreement with observeddata.

Table 2SERCA/SLN/PLB

Experimental

CFP-SERCA/

YFP-SLN

Calculated

CFP-SERCA/

YFP-SLN

Experimental

CFP-SERCA/

YFP-PLB

Calculated

CFP-SERCA/

YFP-PLB (T)

Calculated

CFP-SERCA/

YFP-PLB (R)

FRET Efficiency 0.20 0.093 0.38 0.035 0.032

RDA Distance (Å) 62 81.7 53 101.1 103.4

Orientation factor (2) 0.667 0.651 0.667 0.672 0.659

(Predicted values are shown in Italics)

Center-to-center distances (above) and other parameters were measured and FRET was calculated (Table 2).

Simulated FRET values are significantly lower than experimentally observed, suggesting that there exists someassociation between the XFP’s or between the XFP’s and SERCA. The simulation protocol assumes that there are nospecific interactions between proteins and that the XFP’s can move freely. A more crowded environment may causemore protein-protein association in the case of SERCA-SNL/PLB than for the CerFP-VenFP constructs.

1. Singh, D.R., Dalton, M.P., Cho, E.E., Pribadi, M.P., Zak, T.J., Seflova, J., Makarewich, C.A.,

Olson, E.N., and Robia, S.L. (2019) J Mol Biol 431, 4429-4443.

2. Koushik, S.V., Chen, H., Thaler, C., Puhl III, H.L., and Vogel, S.S. (2006) Biophys J 91, L99-

L101.

3. Autry, J.M., Rubin, J.E., Pietrini, S.D., Winters, D.L., Robia, S.L., and Thomas, D.D. (2011) J

Biol Chem 286, 31697-31706.

4. Humphrey, W., Dalke, A., and Schulten, K. (1996) J Mol Graph 14, 33-38, 27-38.

5. Winters, D.L., Autry, J.M., Svensson, B. and Thomas, D.D. (2008) Biochemistry 47, 4246–4256.

6. Svensson, B., Autry, J.M., and Thomas, D.D. (2016) Methods Mol Biol-P-type ATPases

protocols 1377, 503-502.

7. Chaing, J., Li, I., Pham, E., and Truong, K. (2006) Proceedings of 28th Annual International

IEEE Engineering Conference in Medicine and Biology.

8. Ansbacher, T., Srivastava, H.K. Stein, T., Baer, R., Merkx, M., and Shurki, A. (2012) Phys Chem

Chem Phys 14, 4109-4117.

Simulation and FRET Analyses of SERCA, Phospholamban, and Sarcolipin Complexes

Bengt Svensson, Joseph M. Autry, Tory M. Schaaf, Răzvan L. Cornea, and David D. ThomasDepartment of Biochemistry, Molecular Biology & Biophysics, University of Minnesota, Minneapolis MN, USA.

We have used molecular modeling and experimental FRETconstraints of fluorescent protein fusion constructs to study molecularinteractions and small-molecule effects on SERCA and itstransmembrane regulatory peptide phospholamban (PLB) andsarcolipin (SLN).

The sarcoplasmic reticulum calcium transport ATPase (SERCA) isreversibly inhibited in heart and muscle by PLB and SLN, withphosphorylation-induced relief of inhibition.

We previously used fluorescent fusion-protein constructs toquantify the interactions between SLN, PLB, and SERCA1a. Resultsshow that SERCA has 3-fold higher affinity for SLN over PLB, eventhough the average maximum FRET between SERCA and PLB is twiceas high as for SERCA-SLN.

In order to interpret FRET results (dynamic populationdistributions) in a three-dimensional, structural context, weperformed molecular modeling and conformational samplingsimulations. Starting points for molecular simulations were x-raycrystal structures: SERCA-SLN (3W5A), and SERCA-PLB (4KYT).

Simulations generated an ensemble of conformations from whichthe inter-probe distance (RDA) and the orientation factor (κ

2) werecalculated. These two simulation-based parameters were then used tocalculate FRET from CFP to YFP tags, allowing quantitative comparisonbetween structure-based simulations and experimental results.

Fluorescence spectroscopy studies were carried out at the University of MinnesotaBiophysical Technology Center.

Computational resources were provided by the Minnesota SupercomputingInstitute.

This work was supported by NIH grants to DDT (GM027906, HL129814, AG026160).

Conflict of Interest Disclosure: David D. Thomas and Razvan L. Cornea hold equity inand serve as executive officers for Photonic Pharma LLC, a company that ownsintellectual property related to technology used in part of this research. Theserelationships have been reviewed and managed by the University of Minnesota inaccordance with its conflict-of-interest polices.

More information on this and related work can be viewed athttp://biochem.umn.edu

FRET Calculation From Simulations

Abstract

References Acknowledgments etc.

Conformational Sampling of CFP-SERCA and YFP-SLN/PLB

With the aim to correlate structural models with FRET resultsmodels for the SERCA-SLN/PLB fluorescent fusion proteinswere constructed. For CFP-SERCA/YFP-SLN, the crystalstructure (3W5A) was used. For SERCA-PLB, only the TM-helixof PLB was crystallized (4KYT). The cytoplasmic helix of PLBwas modeled as either embedded in the lipid head groups,PLB(T), or extended into the cytoplasm, PLB(R).

Conformational sampling of the XFPs was run on all threemodels.

To study the interaction between SERCA and it’s regulatory peptides SLN and PLB fluorescent fusion proteinswere designed. YFP and CFP were fused to the N-termini of SLN, PLB, and SERCA using recombinant DNAtechnology and baculovirus expression.

MethodsEngineering of Fluorescent Fusion Proteins: cDNA encoding SLN and SERCA were cloned from rabbit skeletal muscle andPLB from dog heart. cDNA encoding CFP and YFP were purchased from Clontech. CFP and YFP cDNA were fused to SLN,PLB, and SERCA cDNA using DNA ligation at engineered restriction sites, producing fluorescent fusion proteins with the XFPat the N-terminus [3].

Baculovirus/Insect Cell Expression: Fluorescent fusion proteins were expressed in Spodoptera frugiperda (Sf21) insect cellsby infection with recombinant baculoviruses encoding fluorescent fusion proteins. Confocal microscopy and SDS-PAGEdemonstrated that YFP-SLN, YFP-SLN, and CFP-SERCA are correctly targeted and efficiently expressed in endoplasmicreticulum of Sf21 cells.

Fluorescence Microscopy: Fluorescence images of Sf21 cells were recorded using a Nikon TE200S microscope equipped withmetal halide lamp and CCD camera. Automated filter wheels were used to acquire CFP, YFP, and FRET-selective images.Fluorescence resonance energy transfer (FRET) was measured on a cell-by-cell basis using 3-cube FRET. Protein bindingcurves were plotted as a function of FRET vs acceptor concentration, using hyperbolic fit to determine the maximum FRETand dissociation constant (Kd) of each protein interaction, where Kd is defined as the concentration at which half-maximalFRET is observed. CFP filters: Ex = 426-446 nm, dichroic mirror = 455 nm longpass, Em = 460-500 nm. YFP filters: Ex = 490-510 nm, dichroic = 515 nm LP, Em = 520-550 nm. FRET filters: Ex = 426-446 nm, dichroic mirror = 515 nm LP, Em = 520-550nm.

Molecular Visualization: Structure visualization and creation of figures were accomplished using VMD 1.9.3 [4].

SERCA-SLN/PLB Fusion Protein Modeling: Initial models for CFP-SERCA, YFP-SLN, and YFP-PLB were created with DSVisualizer 2019 (BIOVIA, San Diego, CA) using the CFP structure (1RM9), YFP structure (3DQO) and either SERCA-SLN(3W5A), or SERCA-PLB (4KYT) structures.

Conformational sampling were performed by in a similar fashion as done previously for CFP-SERCA [5, 6] using the FPMODmodeling tools which randomly generates models by rigid body rotation [7]. In the simulation the CFP and YFP wereconsidered to move independently. The FPMOD software discards conformations that show clashes within the linker regionor between the XFP and the rest of the protein. The software was modified so that in addition to that, conformations withclashes with the membrane were discarded. The membrane is represented by a 45Å thick geometric region.

For each of the SERCA structural states 1,000 conformations were generated for each XFP. All pairwise interactions wereconsidered after excluding conformations with structure clashes between CFP and YFP. Typically about 950,000 pair-interactions were used.

Calculation of FRET parameters was done by software developed in house. The transition dipole vector for the XFPchromophores was chosen in analogy with published data [8].

Cerulian-Venus modeling: The initial model was built using DS Visualizer using the CerFP structure (2WSO) and the VenFPstructure (1MYW). Conformational sampling was done with FPMOD with no additional filtering generating 25,000conformations. Calculation of FRET parameters were done as above.

SR

lumen

cytoplasm

Fluorescent Fusion Proteins

FRET Measures Binding Affinity and DistanceFRET between the fluorescent fusion proteins CFP and YFP was detected using acceptor photobleaching usingthe 3-cube method in Sf21 insect cells. FRET binding curves determine binding affinity (Kd) and intermoleculardistance.Binding affinity: SLN shows ~3-fold higher association with SERCA than PLB does with SERCA.Maximum FRET: PLB shows ~2-fold higher maximum FRET with SERCA than SLN does with SERCA, indicatingthat the CFP-YFP distance is ~10Å closer in the SERCA-PLB regulatory complex.

0 10 20 30 400.0

0.1

0.2

0.3

0.4

FR

ET

(E

)

[YFP] (AU)

Model Hill

Equationy=Vmax*x^n/(k^n+x^n)

Reduced Chi-Sqr

3.88209E-4 2.01039E-4

Adj. R-Squar 0.73935 0.9458

Value Standard Err

B

Vmax 0.19772 0.01491

k 4.75157 1.26612

n 1 0

F

Vmax 0.38074 0.02036

k 15.08964 1.83901

n 1 0

Table 1. Cer17Ven FRET Efficiency Ave. Donor-AcceptorDistance

Orientation factor (2)

Experiment 0.38 ± 0.03 58 0.667

Calculation 0.28 61.5 0.634

Conformational Sampling of a CerFP-VenFP FRET Standard

To validate the simulation and FRET calculation results, amodel was constructed for a simpler system. TheCeruleanFP-VenusFP joined by a 17 residue linker has beenproposed to be used as a FRET standard 2].

/ / /

Conformational sampling has generated 25,000 structures.VenFP was kept fixed, and CerFP was allowed to move as arigid body at the end of the 17 residue linker. Here 15representative conformations are shown.

The center-to-center distances (above) andother parameters were measured and FRETcalculated (Table 1).

Calculated FRET from the simulationsis slightly lower than experimentallyobserved FRET, suggesting that theremay be some association betweenthe XFP’s or that more extendedlinker conformations are favored inthe simulations.

Average distance = 61.5Å

RDA Distance (Å)C

ou

nt RDA Distance (Å)

Co

un

t

RDA Distance (Å) RDA Distance (Å)

Ave. = 103.4ÅAve. = 101.1ÅAve. = 81.7Å

(Predicted values are shown in Italics)

1000 conformations were generated. Here, 25representative conformations are shown for CFP-SERCA/YFP-SLN (side view).

25 representative conformations are shown for CFP-SERCA/YFP-SLN, CFP-SERCA/YFP-PLB(T), and CFP-SERCA/YFP-PLB(R)(top view). Despite the quite different location in space where YFP is attached to SLN or PLB, the distributions don’t looksignificantly different. This may be due to the flexible C-ter of the XFP plus linker, 12-14 residues, which can span a distanceof up to ~30Å.

Three starting models, CFP-SERCA/YFP-SLN, CFP-SERCA/YFP-PLB(T), and CFP-SERCA/YFP-PLB(R) used for conformationalsampling and FRET calculations.

SERCA-PLB

SERCA-SLN

RDA

Related to these results is a recent paper from the Olson and Robia labs [1]. FRET was measured betweenCerulean labelled SERCA2a and YFP labelled SLN and PLB, showing FRETMax of 39±7 and 27±4, i.e. higher FRETfor SLN than for PLB. Kd values also show the reverse results with 2.9±0.8 (SLN) and 1.5±0.4 (PLB). Are thesedifferences isoform dependent? Further studies may be needed.

o Simulation of fluorescent fusion proteins using FPMOD and our analysis tools provide usefulinformation for a more rigorous interpretation of FRET data.

o Good agreement between simulated and experimental results is demonstrated for CerFP-VenFP.

o Current simulations are being improved by incorporating probabilities for XFP-XFP, XFP-proteininteractions and using new models for SERCA-SLN/PLB complexes.