Targeting Troponin C with Small Molecules Containing Diphenyl Moieties: Calcium

Sensitivity Effects on Striated Muscle and Structure Activity Relationship

Eric R. Hantz

, Svetlana B. Tikunova

, Natalya Belevych

, Jonathan P. Davis

, Peter J. Reiser

3, §

Steffen Lindert

1, §,

Department of Chemistry and Biochemistry, The Ohio State University, Columbus, OH, 43210

Davis Heart and Lung Research Institute and Department of Physiology and Cell Biology, The

Ohio State University, Columbus, OH, 43210

Division of Biosciences, College of Dentistry, The Ohio State University, Columbus, OH 43210

§ denotes co-senior authors

* Correspondence to:

Department of Chemistry and Biochemistry, Ohio State University

2114 Newman & Wolfrom Laboratory, 100 W. 18

Avenue, Columbus, OH 43210

614-292-8284 (office), 614-292-1685 (fax)

lindert.1@osu.edu

Abstract

Despite large investments from academia and industry, heart failure, which results from a

disruption of the contractile apparatus, remains a leading cause of death. Cardiac muscle

contraction is a calcium-dependent mechanism, which is regulated by the troponin protein complex

(cTn) and specifically by the N-terminal domain of its calcium binding subunit (cNTnC). There is

an increasing need for the development of small molecules that increase calcium sensitivity

without altering systolic calcium concentration, thereby strengthening cardiac function. Here, we

examined the effect of our previously identified calcium sensitizing small molecule, ChemBridge

compound 7930079, in the context of several homologous muscle systems. The effect of this

molecule on force generation in isolated cardiac trabeculae and slow skeletal muscle fibers was

measured. Furthermore, we explored the use of Gaussian accelerated molecular dynamics in

sampling highly predictive receptor conformations based on NMR derived starting structures.

Additionally, we took a rational computational approach for lead optimization based on lipophilic

diphenyl moieties. This led to the identification of three novel low affinity binders, which had

similar binding affinities to known positive inotrope trifluoperazine. The most potent identified

calcium sensitizer was compound 16 with an apparent affinity of

117 ± 17$𝜇𝑀

Introduction

The occurrence of heart failure has been shown to increase over time with aging of the population,

and its impact was estimated at $216 billion in direct cost from 2016 - 2017

. Heart failure (HF)

has been generally characterized as the heart’s inability to pump and/or fill with enough blood to

meet the demands of circulation

. Heart failure can be divided into three subtypes: HF with

reduced ejection fraction (HFrEF), heart failure with preserved ejection fraction (HFpEF), and HF

with mid-range ejection fraction (HFmrEF). These subtypes have been differentiated based on the

percentage of ejection fraction (EF) from the left ventricle, where HFrEF has an EF below 40%,

HFpEF has an EF greater than 50%, and HFmrEF has an EF ranging from 40-50%

3, 4

. HF has

been classically treated with diuretics and blood pressure medication to reduce blood volume.

However, the focus for many experts is the design of small molecules that restore cardiac

.CC-BY-NC-ND 4.0 International licenseavailable under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprintthis version posted February 6, 2023. ; https://doi.org/10.1101/2023.02.06.527323doi: bioRxiv preprint

contractility and functionality in efforts to treat HFrEF. A popular class of small molecules with

this intended purpose are positive inotropes

5, 6, 7, 8, 9

. Some of these compounds have been shown

to utilize the beta-adrenergic pathways to create an increased systole concentration of Ca

ions.

However, increasing the systolic [Ca

] can potentially lead to arrhythmias in patients

10, 11

. The

current direction of the field has shifted towards the identification and development of positive

inotropes that increase Ca

sensitivity of the proteins involved in muscle contraction without

affecting the concentration of systolic Ca

2+ 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22

-dependent heart muscle contraction is regulated by cardiac troponin (cTn). The cTn complex

consists of three subunits: troponin T (cTnT) which interacts with tropomyosin and anchors the

protein complex to the thin filament, troponin I (cTnI) known as the inhibitory subunit, and

troponin C (cTnC) which facilitates Ca

binding. cTnI plays a key structural role in anchoring

cTnC to the rest of the troponin complex, in addition to having two amphiphilic peptides that play

important functional roles. Under unsaturated Ca

conditions, the cTnI inhibitory peptide

(residues 137-148) interacts with actin, thereby inhibiting actomyosin binding

. Additionally, the

cTnI switch peptide (cTnI

residues 149-164) interacts with the N-terminal domain of cTnC

stabilizing the open conformation of the domain when saturated with Ca

23, 24, 25

. cTnC is a

dumbbell-shaped protein with two globular domains connected by a flexible linker, each

containing two EF-hand motifs

24, 26

. The C-terminal domain (cCTnC), termed the structural

domain, contains two high-affinity binding sites (sites III and IV) which are constantly saturated

by Ca

or Mg

under physiological conditions

. The N-terminal domain (cNTnC), known as

the regulatory domain, contains only one active binding site (site II)

28, 29

. Upon Ca

binding to

site II, a large conformational change of the troponin complex is initiated by cNTnC. One feature

of this structural rearrangement is the opening of a hydrophobic patch (residues 20, 23, 24, 26, 27,

36, 41, 44, 48, 57, 60, 77, 80, and 81) in cNTnC, which becomes stabilized upon binding of the

cTnI

. The hydrophobic patch serves as an attractive target for increasing Ca

sensitivity via

small molecules. Additionally, the hydrophobic patch and entire troponin complex have been the

focus of numerous in silico studies to capture the proteins dynamics of hydrophobic patch opening

30, 31, 32, 33, 34, 35, 36, 37

, and cTnI

binding

38, 39, 40, 41, 42

We have knowledge of several compounds that bind to the hydrophobic patch and modulate the

sensitivity of Ca

binding: levosimendan

, one of the most well-known Ca

sensitizers, and its

analog pimobendan

, 3-methyldiphenylamine (3-mDPA)

, bepridil

, W7

47, 48

, dfbp-o (2′,4′-

difluoro(1,1′-biphenyl)-4-yloxy acetic acid)

, trifluoperazine (TFP)

, National Cancer Institute

(NCI) database compounds NSC147866

, NSC600285

, NSC611817

, and ChemBridge

compounds 6872062, 7930079, and 9008625

. The two-dimensional structures of these

compounds are shown in Figure 1. There are currently no FDA approved Ca

sensitizers, despite

several of these compounds proceeding to clinical trial. Levosimendan has been utilized in Europe

for decades

, but has yet to receive FDA approval. Compounds trifluoperazine and bepridil failed

clinical trials due to off-target effects

. This underscores the need for continual improvement in

positive inotrope efficacy and safety.

In previous work we identified three Ca

sensitizing compounds with respective

𝐾

values < 100

𝜇𝑀

, with the lead compound being ChemBridge compound 7930079 (2-[(4,5-diphenyl-1H-

imidazol-2-yl)thio]-N-1,3-thiazol-2-ylacetamide). This compound is one of the most potent and

high affinity cNTnC binders known to date. In this work, we examined the sensitizing effects of

2-[(4,5-diphenyl-1H-imidazol-2-yl)thio]-N-1,3-thiazol-2-ylacetamide in rat cardiac trabeculae

and in slow skeletal muscle fibers. We found this compound induced a small but significant shift

.CC-BY-NC-ND 4.0 International licenseavailable under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprintthis version posted February 6, 2023. ; https://doi.org/10.1101/2023.02.06.527323doi: bioRxiv preprint

in cardiac trabeculae and a much larger (four-fold) shift in slow skeletal fibers. Furthermore, we

evaluated the use of Gaussian accelerated Molecular Dynamics (GaMD) in generating predictive

receptor conformations for virtual screening based on initial NMR model starting structures.

Additionally, we further explored the structure activity relationship of potential Ca

sensitizers

based on the lipophilic character of the known high affinity Ca

sensitivity modulators. We found

three compounds to slow the rate of Ca

dissociation from cNTnC within a cTnC-cTnI chimera,

with apparent disassociation constants ranging from

117 − 482$𝜇𝑀

, similar to that of many

known Ca

modulators.

Methods

Receptor Selection and Preparation

For our computational docking studies, we seek to identify cNTnC – cTnI

conformations that are

highly predictive. We prioritized the inclusion of the cTnI

to accurately model the cTnC-cTnI

chimera utilized in our biochemical studies. Thus, we performed an exhaustive search for human

cNTnC – cTnI

structures with known Ca

sensitivity modulating small molecules bound in the

hydrophobic patch (holo) and ligand unbound (apo) structures in the RCSB protein data bank. We

obtained apo structures from two PDB entries: 1MXL

where cNTnC is in complex with the

cardiac isoform of the TnI switch peptide and 2MKP

where cNTnC is in complex with the fast

skeletal TnI switch peptide. We obtained holo cNTnC conformations of several known calcium

sensitivity modulators: bepridil (PDB: 1LXF

), W7 (PDBs: 2KFX

, 2KRD

, and 6MV3

dfbp-o (PDB: 2L1R

), and 3-mDPA (PDBs: 5WCL

and 5W88

). All selected PDB entries

were obtained from NMR experiments and contained multiple conformers; each structure was

extracted resulting in 166 receptor conformations. We examined all 166 receptor conformers in

our docking studies.

All receptor conformations were imported into Schrödinger’s Maestro

and prepared using the

Protein Preparation Wizard

; where for all applicable conformers the C-terminus was capped by

the addition of an N-methyl amide and the N-terminus was capped by the addition of an acetyl

group. The protonation states of all titratable residues were assigned using EPIK

60, 61

with a pH

constraint of

7.4 ± 1.0

GaMD Simulations and Clustered Receptor Generation

In previous work we have demonstrated the utility of GaMD to generate highly predictive receptor

conformations based on X-ray crystallographic structures

. Here, we evaluated the impact of this

computational technique for increased sampling of troponin’s structural ensemble when the initial

frame was derived from an NMR experiment. For each of the beforementioned nine NMR PDB

deposited structures (1MXL, 2MKP, 1LXF, 2KFX, 2KRD, 6MV3, 2L1R, 5WCL, and 5W88), we

extracted model one (the representative receptor conformer) which served as the initial starting

frame for a 300 ns GaMD simulation performed with Amber20

63, 64

. For ligand-bound conformers,

the small molecule was parameterized using the second generation of the generalized amber

forcefield (GAFF2)

and Amber’s antechamber software. All protein complexes were

parameterized with the amber forcefield ff14SB

, solvated with TIP3P

water molecules in a 12

Å

octahedron, and neutralized with sodium ions.

All systems were minimized with restraints (20 kcal/mol) on the protein and ligand (if applicable)

using 2500 steps of steepest decent minimization followed by 2500 steps of conjugate gradient

decent. A second unconstrained minimization was sequentially performed using 2500 steps of

steepest decent minimization followed by 2500 steps of conjugate gradient decent. The system

.CC-BY-NC-ND 4.0 International licenseavailable under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprintthis version posted February 6, 2023. ; https://doi.org/10.1101/2023.02.06.527323doi: bioRxiv preprint

was then heated to 310 K over a span of 1 ns using the Langevin thermostat

. For each system, a

short conventional MD simulation was performed for equilibration under constant temperature

(310 K) and pressure (1 bar) using the Langevin thermostat and Berendsen barostat

, prior to the

GaMD preparation simulations. During the GaMD preparation simulation statistics were collected

to calculate the appropriate boosts to apply to the dihedral and total potential energies. These

statistics were obtained from a second 10 ns conventional MD run, the initial boosts then applied,

and subsequently updated during a 50 ns GaMD biasing run. The final GaMD restart parameters

(VmaxP, VminP, VavgP, sigmaVP, VmaxD, VminD, VavgD, and sigmaVD) were then read in

for a 300 ns GaMD production run. The upper limit for the dihedral and total boost potentials was

set to 6 kcal/mol. All simulations were performed with a 12

Å

cutoff for electrostatic and van der

Waals interactions, and a 2 fs timestep with the SHAKE algorithm

. Coordinates were saved

every 2 ps, resulting in 150,000 frames.

The final structures used for docking studies were obtained by clustering each of the 300 ns GaMD

simulations individually. Prior to clustering the individual trajectories, all water molecules, ions,

and ligands were removed. Clustering was performed over every fourth frame of the original

150,000 frames, resulting in 37,500 frames available for clustering. The density-based clustering

algorithm (DBScan)

implemented in Amber’s CPPTRAJ was used to cluster the processed

trajectories to obtain approximately 11 representative conformers per simulation. Trajectories were

clustered using the backbone atoms of the hydrophobic patch residues (residues 20, 23, 24, 26, 27,

36, 41, 44, 48, 57, 60, 77, 80, and 81). In total, 101 clustered conformers were generated and used

in further active/decoy docking studies (see below). All receptor conformers were loaded into

Maestro and prepared using the Protein Preparation Wizard tool, as detailed above.

Receptor Grid Generation

Receptor grids for ligand-bound receptor conformers were generated by selecting the ligand within

the Maestro workspace and using the center of mass of the ligand as the center of the receptor grid.

For apo receptor conformers, the center of the search space was determined by submitting a PDB

file of the conformer to the FTMap webserver

72, 73

, where fragments were globally docked to

identify potential small molecule binding sites. The resulting output file (PDB format) was

imported into PyMOL

, where the align function was utilized to overlay the FTMap PDB file

with a holo receptor conformer. Upon visual inspection, the fragment cluster that overlapped the

position of the known calcium modulator was identified, and the center of mass of the cluster was

calculated. The center of mass served as the three-dimensional coordinates for the center of the

receptor grid. The search area was centered on the ligand’s center of mass or the supplied three-

dimensional coordinates, respectively, and allowed the centroids of any docked compound to fully

explore a

10 × 10 × 10$Å

inner search space, while the periphery of the ligand was able to extend

out into a

20 × 20 × 20$Å

search space. The OPLS3e forcefield

was used to generate the search

grid and all hydroxyl groups were selected to be freely rotatable within the search area.

Ligand Preparation

The three-dimensional coordinates of all small molecules (known binders) were extracted from

the representative conformation of their respective PDB entries. Small molecules from the

Schrödinger decoy sets, the ChemBridge EXPRESS-Pick Collection, the ChemBridge Core

Library, and NCI database used in our docking protocol were obtained from their respective

databases in the form of SDF files containing their three-dimensional structural information. For

the known binders levosimendan, pimobendan, and trifluoperazine, we generated their respective

structures using Schrödinger’s 2D Sketcher tool. Schrödinger’s LipPrep

tool was used to prepare

.CC-BY-NC-ND 4.0 International licenseavailable under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprintthis version posted February 6, 2023. ; https://doi.org/10.1101/2023.02.06.527323doi: bioRxiv preprint

each ligand for docking. All tautomers, protomers, and stereoisomers were generated for each

ligand. Protonation states were assigned using EPIK with a pH value of

7.4 ± 1.0

, identical to that

of the protein preparation step.

Active/Decoy Screening and Receptor Performance Analysis

A total of 267 receptor conformations (166 NMR receptor conformers and 101 GaMD clustered

conformers) were evaluated by active/decoy screening. The 13 known sensitivity modulators were

used as known actives in the active/decoy docking process, and the presumed decoy small

molecules (1,000) were obtained from the Schrödinger decoy set

with an averaged molecular

weight of 360 g/mol. The decoy set was confirmed to match the molecular properties of the known

binders based on the following chemical properties: molecular weight, LogP, formal charge,

number of hydrogen bond donors, and number of hydrogen bond acceptors. These properties were

calculated using the RDKit (version 2020.09.1)

Chem Descriptor ExactMolWt, Crippen model,

GetFormalCharge, Lipinski NOCount, and Lipinski NHOHCount modules, respectively (see

Figure S1). All compounds were subject to ligand preparation as detailed above. The active and

decoys compounds post-LigPrep were docked into all receptor conformations using Schrödinger’s

Glide SP

77, 79, 80

. Default parameters from the Schrödinger 2018-3 release were utilized for all

docking simulations.

The docked poses of all small molecules were ranked by their respective docking score.

Subsequently, the top docked pose of each compound was kept, and all other

protomers/tautomers/stereoisomers of the compound removed. The performance of every receptor

conformer was evaluated with three metrics: receiver operating characteristic (ROC) curve,

enrichment factor, and a weighted enrichment factor. The ROC curve was generated by plotting

the true positive rate (TPR) against the false positive rate (FPR) and the area under the ROC curve

(AUC) was calculated using Python’s scikit-learn library (ver.0.22.1)

. The ROC AUC metric

evaluates the predictiveness of the receptor conformation for use in future blind screenings.

Enrichment factor (EF) was calculated to evaluate the number of known binders for a predefined

early recognition period (in our case: top 40 compounds). The early recognition period was

determined based on the number of compounds we typically screen in one iteration of our in vitro

studies. EF was calculated via the following equation:

𝐸𝐹 =

𝑁

#$%&'()*&+*%,-*./

𝑁

0($,1)

+ 𝑁

#$%&'()

𝑁

#$%&'()

𝐸𝐹 =

𝑁

#$%&'()*&+*%,-*./

1000 + 13

= 𝑁

#$%&'()*&+*%,-*./

× 1.948

For cNTnC, nine of the thirteen known actives were classified as poor binders (

𝐾

> 100𝜇𝑀

and four actives (ChemBridge compounds 6872062, 7930079, 9008625, and 3-mDPA) were

classified as high – moderate affinity binders (

𝐾

< 100𝜇𝑀

). We aimed to minimize the influence

of the poor binders with regards to the EF score and subsequent conformer identification, as we

believed this would steer the computational screenings towards the identification of additional

poor binders. Therefore, in order to prioritize the identification of the high affinity known binders

in the early recognition period, we formulated a weighted enrichment factor (

𝐸𝐹

2(&34%(0

) score:

!"#$%&"'

%#$%()**#+#&,(

)-&#."/(#+(&01(23

40!()**#+#&,(

)-&#."/(#+(&01(23

(

'"-0,/

%#$%()**#+#&,()-&#."/(

40!()**#+#&,()-&#."/

)-&#."/

.CC-BY-NC-ND 4.0 International licenseavailable under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprintthis version posted February 6, 2023. ; https://doi.org/10.1101/2023.02.06.527323doi: bioRxiv preprint

!"#$%&"'

%#$%()**#+#&,(

)-&#."/(#+(&01(23

40!()**#+#&,(

)-&#."/(#+(&01(23

=???@=A

%#$%()**#+#&,(

)-&#."/(#+(&01(23

40!()**#+#&,(

)-&#."/(#+(&01(23

:=B>CD

Computational Investigation of Compounds with Diphenyl Groups

Based on the two-dimensional structure of the four known high affinity Ca

sensitizers (3-mDPA

and ChemBridge compounds 6872062, 7930079, and 9008625), we hypothesized that the diphenyl

rings are crucial for ligand binding. In order to preserve the lipophilic interactions, we created

three chemical motifs (diphenyl-imidazole, diphenyl-triazine, and diphenylmethane; see Figure

S2) for two-dimensional structural similarity searches of large chemical libraries. In preparation

for our database searches, we filtered both the ChemBridge EXPRESS-Pick Library (501,916

compounds) and Core Library (812,681 compounds) with a cheminformatics approach. Similarly

to our previous work

, compounds found to have a molecular weight greater than 400 g/mol or a

calculated LogP value less than 2.0 or greater than 4.0 were removed. In a second filtering step,

compounds found to violate both of the following rules were removed as violators: number of

hydrogen bond donors > 5 and number of hydrogen bond acceptors > 10. Additionally, the libraries

were screened for Pan-assay interference (PAINS) compounds and violators were sequentially

removed. Upon completion of library filtering, the EXPRESS-Pick library was reduced to 206,726

compounds and the Core library was reduced to 412,866 compounds. For each of the remaining

compounds (619,592 compounds), we calculated the Tanimoto coefficient between that compound

and the three diphenyl motifs. We identified all compounds that had a Tanimoto coefficient

≥

0.6

to any of the diphenyl functional groups, resulting in 89 compounds. The compounds were further

filtered by removing any compounds we had tested in our previous study

to avoid overlap,

compounds that did not contain a diphenyl motif via visual confirmation, and any compounds that

added functional groups exclusively to the diphenyl rings. The last filtering step was implemented

to prioritize identification of compounds with extensions from the imidazole, triazine, or methane

in efforts to retain similar protein-ligand interactions as compared to the thiazole acetamide tail of

2-[(4,5-diphenyl-1H-imidazol-2-yl)thio]-N-1,3-thiazol-2-ylacetamide (see Figure S3). These

criteria further reduced the number of compounds to 36 compounds. All remaining 36 compounds

were docked into the top three performing receptor conformations (1LXF M6, 1LXF M23, and

1LXF M28) based on the results from our active/decoy docking simulations. The compounds were

all subjected to ligand preparation as detailed above, and docked using Glide SP with default

settings, and the docking score was averaged across the three conformers to select a top set of

potentially promising compounds for experimental testing. Compounds that exhibited an averaged

docked score across the three conformers of

>$−5.0$𝑘𝑐𝑎𝑙/𝑚𝑜𝑙

were removed from experimental

testing consideration. This resulted in 23 compounds to be ordered for in vitro testing based on

Tanimoto coefficients.

In addition to identifying potential sensitizing compounds based on two-dimensional similarity,

we performed an exhaustive blind docking of the ChemBridge Core Library. We utilized the

filtered version of the Core Library as described above. The 412,866 compounds were prepared as

described in the Ligand Preparation methods section, and subsequentially docked into the top three

receptor conformations (1LXF M6, 1LXF M23, and 1LXF M28) using Glide SP with default

settings. Upon completion of docking, the docking Z-score of each compound in each receptor

was calculated and averaged over all conformers. We have previously shown that compound

selection with an unbiased ranking such as Z-score, can lead to high success rates in blind virtual

screenings

62, 82

. Based on ranking by averaged Z-score, the two-dimensional structures of the top

.CC-BY-NC-ND 4.0 International licenseavailable under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprintthis version posted February 6, 2023. ; https://doi.org/10.1101/2023.02.06.527323doi: bioRxiv preprint

40 compounds were visually inspected and diphenyl motifs were identified in 17 of the

compounds. We selected these 17 diphenyl containing compounds for experimental testing. In

total, 40 compounds (23 based on Tanimoto coefficients and 17 from the Core library blind

screening) were ordered from ChemBridge and tested in vitro.

Preparation of Proteins for Biochemical Studies

The cTnC-cTnI chimera (herein designated as chimera) was generated as previously described

17,

. The purified chimera was labeled with the environmentally sensitive fluorescent probe IAANS

on Cys53 of the cNTnC (within the chimera) with C35S, T53C and C84S substitutions, as

previously described

Stopped-Flow Fluorescence Measurements

All kinetic measurements were carried out in stopped-flow buffer (10 mM MOPS, 150 mM KCl,

pH 7.0) at 15

C using an Applied Photophysics Ltd. (Leatherhead, UK) model SX.18MV stopped-

flow apparatus with a dead time of ~1.4 ms. IAANS fluorescence was excited at 330 nm and

monitored using a 420-470 nm band-pass interference filter (Oriel, Stratford, CT). Ca

chelator

EGTA (10 mM) in stopped-flow buffer was used to remove Ca

(200 µM) from the chimera (0.5

µM) also in stopped-flow buffer in the absence or presence of compounds. Varying concentration

of each compound were individually added to both stopped-flow reactants. Data traces were fit

using a program by P.J. King (Applied Photophysics, Ltd.), that utilizes the non-linear Lavenberg-

Marquardt algorithm. Each k

off

represents an average of at least three separate experiments ±

standard error, each averaging at least three shots fit with a single exponential equation.

Calcium-sensitivity of Force Generation

The care and use of the animals (four male Sprague Dawley rats, 7-9 months old) used for this

project were approved by the Institutional Animal Care and Use Committee of Ohio State

University. The preparation of cardiac trabeculae (n=10) and slow skeletal muscle (soleus) fibers

(n=10) was as described in Tikunova et al

. The force measurements were also as described in

Tikunova et al

. Briefly, active force generation was measured in a series of activating solutions

with different pCa (-log of [Ca

]) values, first without, then with the Compound 7930079 (20

µM). The force versus pCa data were fit as described in Mahmud et al. 2022

. The pCa that

corresponded to 50% of maximal force generation (pCa

) was determined from the curve fits and

the difference in the value between with and without the compound was calculated for each cardiac

trabecula and muscle fiber. The paired Student’s t-test was used to evaluate the statistical

significance of the difference in pCa

with and without the compound. The fiber type of the soleus

fibers that were used for the force/pCa measurements was determined with SDS-PAGE, as

described in Bergrin et al., 2006

Results and Discussion

Effect of Compound 7930079 on Force Generation in Isolated Cardiac Trabeculae and Slow

Skeletal Muscle Fibers.

All of the soleus fibers that were selected for force measurements were verified to be slow-type,

based upon the results from SDS-PAGE (Figure S4). A summary of the force/pCa data obtained

in cardiac trabeculae and in skeletal slow fibers, with and then without 20

𝜇

M 7930079, is shown

in Figure 2. This compound significantly (P<.01) shifted the curve to higher sensitivity in cardiac

trabeculae (0.05 ± 0.04 pCa unit) and in slow fibers (0.20 ± 0.03 pCa unit). We also tested whether

7930079 impacted maximal force generation in cardiac trabeculae and in slow fibers. Every third

activation in the series of force/pCa measurements was in pCa 4.0 activating solution which yields

.CC-BY-NC-ND 4.0 International licenseavailable under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprintthis version posted February 6, 2023. ; https://doi.org/10.1101/2023.02.06.527323doi: bioRxiv preprint

maximal force generation in both types of preparation. The peak force generated in pCa 4.0

solution was averaged (n=6 activations), without then with, the compound. Compound 7930079

had no effect (P=0.39) on average maximal force generation in slow fibers. There was a significant

(P=0.001) decrease of 8% in the average force in the presence of 7930079 in cardiac trabeculae.

However, in a control set of measurements in six other cardiac trabeculae, in which the complete

force/pCa relationship was measured twice with the compound, the average maximal force in pCa

4.0 activating solution was 6%, lower, on average, in the second, compared to the first, series of

measurements in each trabecula. We, therefore, conclude that 7930093 caused a very small (~2%)

reduction in maximal force generation in cardiac trabeculae. While these are promising results,

this underscores the need for further identification of Ca

sensitizers and of particular interest in

this study, the exploration of structure-activity relationships around the diphenyl motifs.

Top Receptor Conformations Identified Based on Active/Decoy Docking.

Prior to virtual screens of small molecule libraries, we sought to further characterize the

predictiveness of PDB-deposited cNTnC-cTnI

receptor conformations to correctly identify

known cNTnC binders. We identified 166 NMR-derived conformers for characterization,

expanding our search from previous works

16,17

. Additionally, we sought to determine the

effectiveness of Gaussian accelerated Molecular Dynamics (GaMD) generated conformers based

on NMR starting structures. Recently, we have shown that clustering over GaMD simulations may

result in the identification of additional highly predictive receptor conformations

. However, the

starting structures for this study were almost exclusively from x-ray crystallography structures

deposited in the PDB. Here, for the cNTnC-cTnI

system, we aimed to determine if this enhanced

sampling technique could capture additional protein dynamics that are currently absent from

known NMR PDB structures. Therefore, based on the representative NMR model of each PDB

entry (as described in Methods), we performed a 300 ns GaMD simulation and clustered over each

individual trajectory, resulting in 101 additional receptor conformations.

In total 267 cNTnC-cTnI

conformers were used in our active/decoy screening, where all

compounds were docked utilizing the Glide SP docking methodology. The 13 known cNTnC Ca

sensitivity modulators served as the known actives and were docked along a set of 1,000 presumed

decoy small molecules. The decoy compounds were obtained from Schrödinger’s decoy set with

an average molecular weight of 360 g/mol. In order to avoid any unrealistic enrichment in the

active/decoy screenings, the decoys and actives were property-matched across five chemical

properties: molecular weight, LogP, formal charge, number of hydrogen bond acceptors and

number of hydrogen bond donors (see Figure S1). For all conformers, we calculated the respective

true positive rates (TPRs) and false positive rates (FPRs) and determined the AUC of the generated

ROC curves. The ROC AUC metric was used to evaluate the predictiveness of the conformer in

identifying known actives over the decoy compounds. Furthermore, we used enrichment factor

(EF) as a confidence metric for identifying the actives in a predefined early recognition period (top

40 compounds). However, particularly in the case of cTnC, many of the known actives can be

classified as weak binders (

𝐾

> 100$𝜇𝑀

) to the hydrophobic patch of cNTnC in the presence of

cTnI

, such as: levosimendan (

𝐾

≈ 700$𝜇𝑀

)

, bepridil (

𝐾

= 380 ± 60$𝜇𝑀)

, W7

(𝐾

150 − 300$𝜇𝑀)

, dfbp-o

(𝐾

= 270 ± 30$𝜇𝑀)

, TFP (

𝐾

= 110 ± 20$𝜇𝑀)

, and NCI

compound NSC147866 (

𝐾

= 379 ± 50$𝜇𝑀

)

. The experimental binding affinities of the other

two NCI compounds (NSC600285 and NSC611817) were not reported but were predicted to be

weak binders. In order to dampen the influence of the weak binders for the purpose of receptor

selection, we developed a weighted enrichment factor (EF

weighted

) score to prioritize the early

identification of known cNTnC Ca

sensitivity modulators with moderate to high affinity. Those

.CC-BY-NC-ND 4.0 International licenseavailable under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprintthis version posted February 6, 2023. ; https://doi.org/10.1101/2023.02.06.527323doi: bioRxiv preprint

included 3-mDPA (

𝐾

= 30$𝜇𝑀)

and ChemBridge compounds 6872062, 7930079, and

9008625 with respective apparent experimental binding affinities of

84 ± 30$𝜇𝑀

1.45 ±

0.09$𝜇𝑀

, and

34 ± 12$𝜇𝑀

. These four compounds contain diphenyl motifs, which served as the

inspiration for our two-dimensional similarity searches of the ChemBridge databases (see below).

The data for all three metrics (ROC AUC, EF, and EF

weighted

) for all 166 NMR conformers is

summarized in Table S1, and the data for all 101 clustered GaMD conformers is summarized in

Table S2.

Figure 3 shows the evaluation of all 267 receptor conformations, as well as the ROC curve of the

top performing receptor conformation (1LXF NMR model 6) with the respective values for ROC

AUC, EF, and EF

weighted

explicitly shown. The top three receptor conformations were determined

to be all NMR models of PDB entry 1LXF (1LXF model 6, 1LXF model 23, and 1LXF model

28), and were used in later virtual screens. Interestingly, the top performing conformers all belong

to the same PDB (1LXF), which contains cNTnC in complex with cTnI

and bepridil. We believe

that our focus on the diphenyl motifs (through use of the weighted EF) led to implicit bias of

selecting 1LXF receptor conformations. 1LXF is the only deposited structure of the cNTnC –

cTnI

complexes with a small molecule containing a diphenyl motif (benzyl methylaniline).

Therefore, we speculated, the various receptor conformations belonging to this PDB entry could

capture more favorable protein-ligand interactions between the hydrophobic patch residues and

that of compounds with a diphenyl structure. Additionally, only one GaMD clustered conformer

(5WCL clustered conformer 6) exhibited a relatively high ROC AUC of 0.750; however, the EF

and EF

weighted

values were low (5.89 and 5.98 respectively). As shown before, GaMD is a powerful

technique for structure-based drug discovery, with the most benefit seen when x-ray

crystallographic structures served as the initial frame of the simulation. Unfortunately, in our case,

there was no improvement for capturing ideal receptor conformations for ligand binding compared

to preexisting NMR based structures. The top three receptor conformers (1LXF model 6, 1LXF

model 23, and 1LXF model 28) were used for all virtual screenings going forward.

Exploration of Diphenyl Motifs Leads to Identification of Three cNTnC Calcium Sensitizers.

Based on our previous screenings, we theorized lipophilic phenyl rings to be an important

functional group for ligand binding into the cNTnC hydrophobic patch. Therefore, we focused our

efforts on identification of small molecules containing diphenyl structures. We created three motifs

of interest: diphenyl-imidazole (based on 2-[(4,5-diphenyl-1H-imidazol-2-yl)thio]-N-1,3-thiazol-

2-ylacetamide), diphenyl-triazine, and diphenyl-methane (similar to 3-mDPA). The two-

dimensional structures of these motifs are shown in Figure S2. We focused on two small molecule

libraries from ChemBridge, the EXPRESS-Pick Library and Core Library, and filtered each library

with a cheminformatics approach as described in the Methods section. For each compound of the

filtered libraries (619,592 compounds) we calculated the Tanimoto coefficient to each of the three

diphenyl motifs. Compounds found to have a high structural similarity (Tanimoto coefficient

≥

0.6) were further filtered to remove redundancies from previous testing by our group and based on

averaged Glide SP docking score across the three most predictive receptor conformations. This

method led to the identification of 23 compounds for further experimental testing. Additionally,

we performed a blind virtual screening of the filtered Core Library (412,866 compounds) and rank

ordered compounds based on averaged Z-Score across the three most predictive receptor

conformations. Upon visual inspection of the top 40 compounds, 17 were identified to have a

diphenyl motif and were selected for further testing. In total, we ordered 40 compounds from

ChemBridge for in vitro testing, the two-dimensional structures of all ordered compounds are

available in Figure S5.

.CC-BY-NC-ND 4.0 International licenseavailable under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprintthis version posted February 6, 2023. ; https://doi.org/10.1101/2023.02.06.527323doi: bioRxiv preprint

Initial stopped-flow kinetics revealed that 39 of the initial 40 compounds were soluble enough in

aqueous buffer that experimental measurements could be obtained. Of the 39 compounds that were

able to be experimentally tested, three showed at least a 15% decrease of the Ca

dissociation rate

for the chimera. The average Ca

dissociation rate observed for the chimera in the absence of

compounds was 68.6 ± 0.4 s

−1

. Compounds 2 (ChemBridge ID 7874460), 16 (ChemBridge ID

56598339), and 17 (ChemBridge ID 14233019) lead to a moderate slowing of the Ca

dissociation

rate to

51.8 ± 0.4$𝑠

49.0 ± 0.3$𝑠

, and

57.6 ± 0.9$𝑠

50$𝜇𝑀

, respectively. To further

characterize these hits, for each compound we performed stopped flow experiments using

increasing concentrations of the respective small molecule in order to get a dose response. Figure

4 illustrates the effects of each compound slowing the rate of Ca

disassociation from the chimera,

where panel 4A depicts the dose response and panel 4B depicts a representative stopped-flow trace

for each of the hit compounds at a

100$𝜇𝑀

concentration of the compound. Of the three

compounds, compound 17 was determined to have the lowest apparent experimental affinity of

482 ± 300$𝜇𝑀

, a similar range to that of bepridil and NSC147866. Compound 2 was found to

have an improved apparent experimental affinity (

154 ± 23$𝜇𝑀

) compared to compound 17.

Compound 16 performed the best of the tested compounds with an apparent experimental affinity

117 ± 17$𝜇𝑀

. Compounds 2 and 16 had comparable affinities to known modulators TFP and

W7, and showed significant improvement over levosimendan, bepridil, dfbp-o, and NSC147866.

While we did not identify additional high affinity compounds, the three compounds that were

identified meaningfully expand our knowledge of moderate affinity cNTnC binders.

Of the 40 tested compounds only five (12.5%) contained a diphenyl-imidazole structure and of

these, one compound was identified to have moderate binding affinity: compound 2. Meanwhile

18 compounds (45%) contained diphenyl-triazine structures, which consistently failed to slow the

rate of Ca

dissociation despite several compounds with a near-identical structure to previously

identified modulators. We believe the substitution of a six-member ring disrupts the electrostatic

and van der Waals protein-ligand interactions, thereby preventing the small molecule to bind

deeply in the hydrophobic patch. Compounds containing diphenylmethane moieties accounted for

42.5% (17 compounds) of the screened compounds, and two of those small molecules were

identified as medium to low affinity binders: compounds 16 and 17. Based on these results, we

believe compounds containing a five-member ring (i.e., imidazole or triazole) or diphenylmethane

are optimal for maximal lipophilic protein-ligand contact. On the other hand, our results seem to

suggest that six-member rings (i.e., triazines) alter the sterics of the ligand and prevent deeper

binding in the hydrophobic pocket. Representative docked poses of the two identified Ca

sensitizers with moderate affinity in this study are shown in Figure 5. In this figure, high affinity

sensitizer 2-[(4,5-diphenyl-1H-imidazol-2-yl)thio]-N-1,3-thiazol-2-ylacetamide was docked

into the cNTnC hydrophobic patch with the results of our virtual screenings overlayed, where

panel 5A shows compound 16 (ChemBridge ID 56598339) and panel 5B shows compound 2

(ChemBridge ID 7874460). Compounds 2 and 16 were shown to have the prioritized diphenyl ring

structure dock similarly to 2-[(4,5-diphenyl-1H-imidazol-2-yl)thio]-N-1,3-thiazol-2-ylacetamide.

To confirm this, we determined the symmetry-corrected heavy atom RMSD of the diphenyl rings

(utilizing spyrmsd

) between 2-[(4,5-diphenyl-1H-imidazol-2-yl)thio]-N-1,3-thiazol-2-

ylacetamide and compounds 2 and 16 to be

0.64$Å

and

0.91$Å

, respectively. While the diphenyl

structural contacts of focus were maintained, it is likely that compounds 2 and 16 are missing other

important protein-ligand interactions beyond the lipophilic structure. Therefore, future work

.CC-BY-NC-ND 4.0 International licenseavailable under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprintthis version posted February 6, 2023. ; https://doi.org/10.1101/2023.02.06.527323doi: bioRxiv preprint

should consider only five-member heterocycles or diphenylmethanes with a focus on optimizing

ligand tail interactions with the cTnI

Conclusion

In this study we tested one of our previously identified cTnC Ca

sensitizers (2-[(4,5-diphenyl-

1H-imidazol-2-yl)thio]-N-1,3-thiazol-2-ylacetamide) for Ca

sensitization properties in the

context of two different muscle types, cardiac trabeculae and slow skeletal muscle fibers which

share the expression of cTnC and explored further virtual screening around a diphenyl motif. The

results demonstrate that compound 7930079 is a potent sensitizer of force generation in cardiac

trabeculae and in slow skeletal muscle fibers, with a much greater (four-fold) effect in the latter.

This demonstrates the potential of this compound as a possible therapeutic for the treatment of

heart failure and/or muscle weakness that is commonly associated with several myopathies,

including cancer cachexia, AIDS and sarcopenia associated with aging.

We examined the predictive utility of GaMD clustered models based on NMR initial starting

structures and, at least for cTnC, found no improvement compared to receptor conformers derived

directly from NMR experiments. Future work should focus on whether this trend can be

generalized across different receptor systems. Additionally, whether starting GaMD simulations

from the most predictive models of 1LXF might increase the AUC and EF could be explored.

Through further drug discovery efforts centered around diphenyl motifs, we identified additional

low affinity binders that increase Ca

sensitivity in cNTnC site II. While our approach did not

lead to the desired success of identifying additional high affinity Ca

modulators, the observed

binding affinities for two of the three identified compounds in this work were comparable to TFP,

and better than levosimendan, W7, dfbp-o, bepridil, and NCI compound NSC147866. Therefore,

we believe that the diphenyl-imidazole functional group is an important scaffold in cNTnC

binding, but in itself is insufficient for high affinity binding. Future work to improve upon 2-[(4,5-

diphenyl-1H-imidazol-2-yl)thio]-N-1,3-thiazol-2-ylacetamide will require collaboration with

medicinal chemists. Lead optimization efforts will be focused on optimizing the protein-ligand

interactions of the “tail” region, as this region was not sampled sufficiently in our present off the

shelf catalog-based optimization. The identification of three more cNTnC binders increases our

knowledge of the structure activity relationship in cTnC binding, which will be useful for follow

up work.

Acknowledgements

The authors would like to thank the members of the Lindert group for useful discussions. We

would like to thank the Ohio Supercomputer Center for valuable computational resources

. This

work was supported by the NIH (R01 HL137015 to S.L.).

.CC-BY-NC-ND 4.0 International licenseavailable under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprintthis version posted February 6, 2023. ; https://doi.org/10.1101/2023.02.06.527323doi: bioRxiv preprint

Figure 1. 2D structure of known cNTnC Ca

sensitivity modulators.

Figure 2. The Force/pCa relationships were measured in both rat cardiac trabeculae and slow

skeletal rat muscle fibers (10 each) before and after the addition of compound 7930079, our highest

affinity compound to date. Compound 7930079 significantly (P<0.05) sensitized both cardiac, and

more potently, slow skeletal muscle to Ca

, as evidenced by the increase in Ca

sensitivity

(positive ΔpCa50) compared to the absence of the compound (internal control).

.CC-BY-NC-ND 4.0 International licenseavailable under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprintthis version posted February 6, 2023. ; https://doi.org/10.1101/2023.02.06.527323doi: bioRxiv preprint

Figure 3. cNTnC receptor conformation characterization using active/decoy screening. Right:

ROC AUC (x-axis) of receptor conformations plotted against enrichment factor (y-axis) and

weighted enrichment factor (z-axis). Receptor conformations obtained from NMR models are

shown with a circle marker, and clustered GaMD conformers are shown with a triangle marker.

Left: ROC curve for the best performing conformer (1LXF NMR model 6) where the TPR is

plotted against the FPR. The inset region shows the TPR and FPR of the top 40 docked compounds,

the calculated EF, EF

weighted

, and highlights true positives.

Figure 4. Effect of the hit compounds on the Ca

dissociation rate from the chimera. Panel (A)

shows the plot of the apparent rates of Ca

dissociation from the chimera in the presence of

increased concentration of compounds 2, 16 or 17. Each data point represents an average of at least

three measurements ± standard error. Data were fit with an asymmetric sigmoid curve. Panel (B)

shows representative stopped-flow traces as Ca

is removed from the chimera in the absence or

presence of 100 µM of compounds 2, 16 or 17. The traces have been normalized and staggered for

clarity.

.CC-BY-NC-ND 4.0 International licenseavailable under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprintthis version posted February 6, 2023. ; https://doi.org/10.1101/2023.02.06.527323doi: bioRxiv preprint

Figure 5. Docked poses of Ca

modulators in cNTnC hydrophobic pocket. PDB 1LXF model 6

is shown in gray surface representation with the protein backbone explicitly shown in cartoon

representation, calcium sensitizer 2-[(4,5-diphenyl-1H-imidazol-2-yl)thio]-N-1,3-thiazol-2-

ylacetamide shown in cyan overlayed with compound 16 shown in yellow (panel A) and

compound 2 shown in magenta (panel B).

.CC-BY-NC-ND 4.0 International licenseavailable under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprintthis version posted February 6, 2023. ; https://doi.org/10.1101/2023.02.06.527323doi: bioRxiv preprint

References

(1) Virani, S. S.; Alonso, A.; Aparicio, H. J.; Benjamin, E. J.; Bittencourt, M. S.; Callaway, C.

W.; Carson, A. P.; Chamberlain, A. M.; Cheng, S.; Delling, F. N.; et al. Heart Disease and Stroke

Statistics-2021 Update: A Report From the American Heart Association. Circulation 2021, 143

(8), e254-e743. DOI: 10.1161/CIR.0000000000000950.

(2) Coronel, R.; de Groot, J. R.; van Lieshout, J. J. Defining heart failure. Cardiovasc Res 2001,

50 (3), 419-422. DOI: 10.1016/s0008-6363(01)00284-x.

(3) Savarese, G.; Lund, L. H. Global Public Health Burden of Heart Failure. Card Fail Rev 2017,

3 (1), 7-11. DOI: 10.15420/cfr.2016:25:2.

(4) Nussbaumerová, B.; Rosolová, H. Diagnosis of heart failure: the new classification of heart

failure. Vnitr Lek 2018, 64 (9), 847-851.

(5) Bourge, R. C.; Fleg, J. L.; Fonarow, G. C.; Cleland, J. G.; McMurray, J. J.; van Veldhuisen,

D. J.; Gheorghiade, M.; Patel, K.; Aban, I. B.; Allman, R. M.; et al. Digoxin reduces 30-day all-

cause hospital admission in older patients with chronic systolic heart failure. Am J Med 2013,

126 (8), 701-708. DOI: 10.1016/j.amjmed.2013.02.001.

(6) Tacon, C. L.; McCaffrey, J.; Delaney, A. Dobutamine for patients with severe heart failure: a

systematic review and meta-analysis of randomised controlled trials. Intensive Care Med 2012,

38 (3), 359-367. DOI: 10.1007/s00134-011-2435-6.

(7) Tuttle, R. R.; Mills, J. Dobutamine: development of a new catecholamine to selectively

increase cardiac contractility. Circ Res 1975, 36 (1), 185-196. DOI: 10.1161/01.res.36.1.185.

(8) Packer, M.; Carver, J. R.; Rodeheffer, R. J.; Ivanhoe, R. J.; DiBianco, R.; Zeldis, S. M.;

Hendrix, G. H.; Bommer, W. J.; Elkayam, U.; Kukin, M. L. Effect of oral milrinone on mortality

in severe chronic heart failure. The PROMISE Study Research Group. N Engl J Med 1991, 325

(21), 1468-1475. DOI: 10.1056/NEJM199111213252103.

(9) Endoh, M. Amrinone, forerunner of novel cardiotonic agents, caused paradigm shift of heart

failure pharmacotherapy. Circ Res 2013, 113 (4), 358-361. DOI:

10.1161/CIRCRESAHA.113.301689.

(10) Amin, A.; Maleki, M. Positive inotropes in heart failure: a review article. Heart Asia 2012,

4 (1), 16-22. DOI: 10.1136/heartasia-2011-010068.

(11) Kass, D. A.; Solaro, R. J. Mechanisms and Use of Calcium-Sensitizing Agents in the Failing

Heart. Circulation 2006, 113 (2), 305-315. DOI: 10.1161/CIRCULATIONAHA.105.542407

(acccessed 2023/02/01).

(12) Perrone, S. V.; Kaplinsky, E. J. Calcium sensitizer agents: a new class of inotropic agents in

the treatment of decompensated heart failure. Int J Cardiol 2005, 103 (3), 248-255. DOI:

10.1016/j.ijcard.2004.12.012.

(13) Tikunova, S. B.; Cuesta, A.; Price, M.; Li, M. X.; Belevych, N.; Biesiadecki, B. J.; Reiser,

P. J.; Hwang, P. M.; Davis, J. P. 3-Chlorodiphenylamine activates cardiac troponin by a

mechanism distinct from bepridil or TFP. J Gen Physiol 2019, 151 (1), 9-17. DOI:

10.1085/jgp.201812131.

(14) Li, M. X.; Hwang, P. M. Structure and function of cardiac troponin C (TNNC1):

Implications for heart failure, cardiomyopathies, and troponin modulating drugs. Gene 2015, 571

(2), 153-166. DOI: 10.1016/j.gene.2015.07.074.

(15) Mahmud, Z.; Tikunova, S.; Belevych, N.; Wagg, C. S.; Zhabyeyev, P.; Liu, P. B.; Rasicci,

D. V.; Yengo, C. M.; Oudit, G. Y.; Lopaschuk, G. D.; et al. Small Molecule RPI-194 Stabilizes

Activated Troponin to Increase the Calcium Sensitivity of Striated Muscle Contraction. Front

Physiol 2022, 13, 892979. DOI: 10.3389/fphys.2022.892979.

.CC-BY-NC-ND 4.0 International licenseavailable under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprintthis version posted February 6, 2023. ; https://doi.org/10.1101/2023.02.06.527323doi: bioRxiv preprint

(16) Aprahamian, M. L.; Tikunova, S. B.; Price, M. V.; Cuesta, A. F.; Davis, J. P.; Lindert, S.

Successful Identification of Cardiac Troponin Calcium Sensitizers Using a Combination of

Virtual Screening and ROC Analysis of Known Troponin C Binders. J Chem Inf Model 2017, 57

(12), 3056-3069. DOI: 10.1021/acs.jcim.7b00536.

(17) Coldren, W. H.; Tikunova, S. B.; Davis, J. P.; Lindert, S. Discovery of Novel Small-

Molecule Calcium Sensitizers for Cardiac Troponin C: A Combined Virtual and Experimental

Screening Approach. J Chem Inf Model 2020, 60 (7), 3648-3661. DOI:

10.1021/acs.jcim.0c00452.

(18) Lindert, S.; Li, M. X.; Sykes, B. D.; McCammon, J. A. Computer-aided drug discovery

approach finds calcium sensitizer of cardiac troponin. Chem Biol Drug Des 2015, 85 (2), 99-106.

DOI: 10.1111/cbdd.12381.

(19) Li, M. X.; Gelozia, S.; Danmaliki, G. I.; Wen, Y.; Liu, P. B.; Lemieux, M. J.; West, F. G.;

Sykes, B. D.; Hwang, P. M. The calcium sensitizer drug MCI-154 binds the structural C-terminal

domain of cardiac troponin C. Biochem Biophys Rep 2018, 16, 145-151. DOI:

10.1016/j.bbrep.2018.10.012.

(20) Tikunova, S. B.; Davis, J. P. Designing calcium-sensitizing mutations in the regulatory

domain of cardiac troponin C. J Biol Chem 2004, 279 (34), 35341-35352. DOI:

10.1074/jbc.M405413200.

(21) Tikunova, S. B.; Liu, B.; Swindle, N.; Little, S. C.; Gomes, A. V.; Swartz, D. R.; Davis, J.

P. Effect of calcium-sensitizing mutations on calcium binding and exchange with troponin C in

increasingly complex biochemical systems. Biochemistry 2010, 49 (9), 1975-1984. DOI:

10.1021/bi901867s.

(22) Shettigar, V.; Zhang, B.; Little, S. C.; Salhi, H. E.; Hansen, B. J.; Li, N.; Zhang, J.; Roof, S.

R.; Ho, H.-T.; Brunello, L.; et al. Rationally engineered Troponin C modulates in vivo cardiac

function and performance in health and disease. Nature Communications 2016, 7 (1), 10794.

DOI: 10.1038/ncomms10794.

(23) Takeda, S.; Yamashita, A.; Maeda, K.; Maéda, Y. Structure of the core domain of human

cardiac troponin in the Ca(2+)-saturated form. Nature 2003, 424 (6944), 35-41. DOI:

10.1038/nature01780.

(24) Dvoretsky, A.; Abusamhadneh, E. M.; Howarth, J. W.; Rosevear, P. R. Solution structure of

calcium-saturated cardiac troponin C bound to cardiac troponin I. J Biol Chem 2002, 277 (41),

38565-38570. DOI: 10.1074/jbc.M205306200.

(25) Kobayashi, T.; Solaro, R. J. Calcium, thin filaments, and the integrative biology of cardiac

contractility. Annu Rev Physiol 2005, 67, 39-67. DOI:

10.1146/annurev.physiol.67.040403.114025.

(26) Herzberg, O.; Moult, J.; James, M. N. G. CONFORMATIONAL FLEXIBILITY OF

TROPONIN C11Supported by the Medical Research Council of Canada, an Alberta Heritage

Foundation of Medical Research Fellowship (to O.H.) and an Alberta Heritage Foundation for

Medical Research Visiting Scientist Fellowship (to J.M.). In Calcium-Binding Proteins in Health

and Disease, Norman, A. W., Vanaman, T. C., Means, A. R. Eds.; Academic Press, 1987; pp

312-322.

(27) Zot, H. G.; Potter, J. D. A structural role for the Ca2+-Mg2+ sites on troponin C in the

regulation of muscle contraction. Preparation and properties of troponin C depleted myofibrils. J

Biol Chem 1982, 257 (13), 7678-7683.

.CC-BY-NC-ND 4.0 International licenseavailable under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprintthis version posted February 6, 2023. ; https://doi.org/10.1101/2023.02.06.527323doi: bioRxiv preprint

(28) Sia, S. K.; Li, M. X.; Spyracopoulos, L.; Gagné, S. M.; Liu, W.; Putkey, J. A.; Sykes, B. D.

Structure of cardiac muscle troponin C unexpectedly reveals a closed regulatory domain. J Biol

Chem 1997, 272 (29), 18216-18221. DOI: 10.1074/jbc.272.29.18216.

(29) Spyracopoulos, L.; Li, M. X.; Sia, S. K.; Gagné, S. M.; Chandra, M.; Solaro, R. J.; Sykes, B.

D. Calcium-Induced Structural Transition in the Regulatory Domain of Human Cardiac Troponin

C. Biochemistry 1997, 36 (40), 12138-12146. DOI: 10.1021/bi971223d.

(30) Lindert, S.; Kekenes-Huskey, P. M.; Huber, G.; Pierce, L.; McCammon, J. A. Dynamics and

calcium association to the N-terminal regulatory domain of human cardiac troponin C: a

multiscale computational study. J Phys Chem B 2012, 116 (29), 8449-8459. DOI:

10.1021/jp212173f.

(31) Lindert, S.; Kekenes-Huskey, P. M.; McCammon, J. A. Long-timescale molecular dynamics

simulations elucidate the dynamics and kinetics of exposure of the hydrophobic patch in troponin

C. Biophys J 2012, 103 (8), 1784-1789. DOI: 10.1016/j.bpj.2012.08.058.

(32) Dewan, S.; McCabe, K. J.; Regnier, M.; McCulloch, A. D.; Lindert, S. Molecular Effects of

cTnC DCM Mutations on Calcium Sensitivity and Myofilament Activation-An Integrated

Multiscale Modeling Study. J Phys Chem B 2016, 120 (33), 8264-8275. DOI:

10.1021/acs.jpcb.6b01950.

(33) Bowman, J. D.; Lindert, S. Molecular Dynamics and Umbrella Sampling Simulations

Elucidate Differences in Troponin C Isoform and Mutant Hydrophobic Patch Exposure. J Phys

Chem B 2018, 122 (32), 7874-7883. DOI: 10.1021/acs.jpcb.8b05435.

(34) Bowman, J. D.; Coldren, W. H.; Lindert, S. Mechanism of Cardiac Troponin C Calcium

Sensitivity Modulation by Small Molecules Illuminated by Umbrella Sampling Simulations. J

Chem Inf Model 2019, 59 (6), 2964-2972. DOI: 10.1021/acs.jcim.9b00256.

(35) Wang, D.; Robertson, I. M.; Li, M. X.; McCully, M. E.; Crane, M. L.; Luo, Z.; Tu, A. Y.;

Daggett, V.; Sykes, B. D.; Regnier, M. Structural and functional consequences of the cardiac

troponin C L48Q Ca(2+)-sensitizing mutation. Biochemistry 2012, 51 (22), 4473-4487. DOI:

10.1021/bi3003007.

(36) Wang, D.; McCully, M. E.; Luo, Z.; McMichael, J.; Tu, A. Y.; Daggett, V.; Regnier, M.

Structural and functional consequences of cardiac troponin C L57Q and I61Q Ca(2+)-

desensitizing variants. Arch Biochem Biophys 2013, 535 (1), 68-75. DOI:

10.1016/j.abb.2013.02.006.

(37) Jayasundar, J. J.; Xing, J.; Robinson, J. M.; Cheung, H. C.; Dong, W. J. Molecular dynamics

simulations of the cardiac troponin complex performed with FRET distances as restraints. PLoS

One 2014, 9 (2), e87135. DOI: 10.1371/journal.pone.0087135.

(38) Lindert, S.; Cheng, Y.; Kekenes-Huskey, P.; Regnier, M.; McCammon, J. A. Effects of

HCM cTnI mutation R145G on troponin structure and modulation by PKA phosphorylation

elucidated by molecular dynamics simulations. Biophys J 2015, 108 (2), 395-407. DOI:

10.1016/j.bpj.2014.11.3461.

(39) Cheng, Y.; Rao, V.; Tu, A. Y.; Lindert, S.; Wang, D.; Oxenford, L.; McCulloch, A. D.;

McCammon, J. A.; Regnier, M. Troponin I Mutations R146G and R21C Alter Cardiac Troponin

Function, Contractile Properties, and Modulation by Protein Kinase A (PKA)-mediated

Phosphorylation. J Biol Chem 2015, 290 (46), 27749-27766. DOI: 10.1074/jbc.M115.683045.

(40) Cool, A. M.; Lindert, S. Umbrella Sampling Simulations Measure Switch Peptide Binding

and Hydrophobic Patch Opening Free Energies in Cardiac Troponin. J Chem Inf Model 2022.

DOI: 10.1021/acs.jcim.2c00508.

.CC-BY-NC-ND 4.0 International licenseavailable under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprintthis version posted February 6, 2023. ; https://doi.org/10.1101/2023.02.06.527323doi: bioRxiv preprint

(41) Cool, A. M.; Lindert, S. Computational Methods Elucidate Consequences of Mutations and

Post-translational Modifications on Troponin I Effective Concentration to Troponin C. J Phys

Chem B 2021, 125 (27), 7388-7396. DOI: 10.1021/acs.jpcb.1c03844.

(42) Genchev, G. Z.; Kobayashi, M.; Kobayashi, T.; Lu, H. Molecular dynamics provides new

insights into the mechanism of calcium signal transduction and interdomain interactions in

cardiac troponin. FEBS Open Bio 2021, 11 (7), 1841-1853. DOI: 10.1002/2211-5463.13009.

(43) Kivikko, M.; Lehtonen, L. Levosimendan: a new inodilatory drug for the treatment of

decompensated heart failure. Curr Pharm Des 2005, 11 (4), 435-455. DOI:

10.2174/1381612053382043.

(44) Takeda, N.; Hayashi, Y.; Arino, T.; Takeda, A.; Noma, K. Effect of pimobendan in patients

with chronic heart failure. Exp Clin Cardiol 2001, 6 (4), 195-199.

(45) Cai, F.; Li, M. X.; Pineda-Sanabria, S. E.; Gelozia, S.; Lindert, S.; West, F.; Sykes, B. D.;

Hwang, P. M. Structures reveal details of small molecule binding to cardiac troponin. J Mol Cell

Cardiol 2016, 101, 134-144. DOI: 10.1016/j.yjmcc.2016.10.016.

(46) MacLachlan, L. K.; Reid, D. G.; Mitchell, R. C.; Salter, C. J.; Smith, S. J. Binding of a

calcium sensitizer, bepridil, to cardiac troponin C. A fluorescence stopped-flow kinetic, circular

dichroism, and proton nuclear magnetic resonance study. J Biol Chem 1990, 265 (17), 9764-

9770.

(47) Cai, F.; Hwang, P. M.; Sykes, B. D. Structural Changes Induced by the Binding of the

Calcium Desensitizer W7 to Cardiac Troponin. Biochemistry 2018, 57 (46), 6461-6469. DOI:

10.1021/acs.biochem.8b00882.

(48) Adhikari, B. B.; Wang, K. Interplay of troponin- and Myosin-based pathways of calcium

activation in skeletal and cardiac muscle: the use of W7 as an inhibitor of thin filament

activation. Biophys J 2004, 86 (1 Pt 1), 359-370. DOI: 10.1016/S0006-3495(04)74112-0.

(49) Robertson, I. M.; Sun, Y. B.; Li, M. X.; Sykes, B. D. A structural and functional perspective

into the mechanism of Ca2+-sensitizers that target the cardiac troponin complex. J Mol Cell

Cardiol 2010, 49 (6), 1031-1041. DOI: 10.1016/j.yjmcc.2010.08.019.

(50) Ogawa, Y. Calcium binding to troponin C and troponin: effects of Mg2+, ionic strength and

pH. J Biochem 1985, 97 (4), 1011-1023. DOI: 10.1093/oxfordjournals.jbchem.a135143.

(51) Papp, Z.; Agostoni, P.; Alvarez, J.; Bettex, D.; Bouchez, S.; Brito, D.; Černý, V.; Comin-

Colet, J.; Crespo-Leiro, M. G.; Delgado, J. F.; et al. Levosimendan Efficacy and Safety: 20 Years

of SIMDAX in Clinical Use. J Cardiovasc Pharmacol 2020, 76 (1), 4-22. DOI:

10.1097/FJC.0000000000000859.

(52) Li, M. X.; Robertson, I. M.; Sykes, B. D. Interaction of cardiac troponin with cardiotonic

drugs: a structural perspective. Biochem Biophys Res Commun 2008, 369 (1), 88-99. DOI:

10.1016/j.bbrc.2007.12.108.

(53) Li, M. X.; Spyracopoulos, L.; Sykes, B. D. Binding of cardiac troponin-I147-163 induces a

structural opening in human cardiac troponin-C. Biochemistry 1999, 38 (26), 8289-8298. DOI:

10.1021/bi9901679.

(54) Robertson, I. M.; Pineda-Sanabria, S. E.; Holmes, P. C.; Sykes, B. D. Conformation of the

critical pH sensitive region of troponin depends upon a single residue in troponin I. Arch

Biochem Biophys 2014, 552-553, 40-49. DOI: 10.1016/j.abb.2013.12.003.

(55) Wang, X.; Li, M. X.; Sykes, B. D. Structure of the regulatory N-domain of human cardiac

troponin C in complex with human cardiac troponin I147-163 and bepridil. J Biol Chem 2002,

277 (34), 31124-31133. DOI: 10.1074/jbc.M203896200.

.CC-BY-NC-ND 4.0 International licenseavailable under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprintthis version posted February 6, 2023. ; https://doi.org/10.1101/2023.02.06.527323doi: bioRxiv preprint

(56) Hoffman, R. M.; Sykes, B. D. Structure of the inhibitor W7 bound to the regulatory domain

of cardiac troponin C. Biochemistry 2009, 48 (24), 5541-5552. DOI: 10.1021/bi9001826.

(57) Oleszczuk, M.; Robertson, I. M.; Li, M. X.; Sykes, B. D. Solution structure of the regulatory

domain of human cardiac troponin C in complex with the switch region of cardiac troponin I and

W7: the basis of W7 as an inhibitor of cardiac muscle contraction. J Mol Cell Cardiol 2010, 48

(5), 925-933. DOI: 10.1016/j.yjmcc.2010.01.016.

(58) Schrödinger Release 2021-1 : Maestro; Schrödinger, LLC: New York, NY, 2021.

(accessed.

(59) Sastry, G. M.; Adzhigirey, M.; Day, T.; Annabhimoju, R.; Sherman, W. Protein and ligand

preparation: parameters, protocols, and influence on virtual screening enrichments. J Comput

Aided Mol Des 2013, 27 (3), 221-234. DOI: 10.1007/s10822-013-9644-8.

(60) Schrödinger Release 2021-1 : Epik; Schrödinger, LLC: New York, NY, 2021 (accessed.

(61) Shelley, J. C.; Cholleti, A.; Frye, L. L.; Greenwood, J. R.; Timlin, M. R.; Uchimaya, M.

Epik: a software program for pK( a ) prediction and protonation state generation for drug-like

molecules. J Comput Aided Mol Des 2007, 21 (12), 681-691. DOI: 10.1007/s10822-007-9133-z.

(62) Hantz, E. R.; Lindert, S. Actives-Based Receptor Selection Strongly Increases the Success

Rate in Structure-Based Drug Design and Leads to Identification of 22 Potent Cancer Inhibitors.

J Chem Inf Model 2022. DOI: 10.1021/acs.jcim.2c00848.

(63) Amber20; University of California, San Francisco. , 2021. (accessed.

(64) Miao, Y.; Feher, V. A.; McCammon, J. A. Gaussian Accelerated Molecular Dynamics:

Unconstrained Enhanced Sampling and Free Energy Calculation. Journal of Chemical Theory

and Computation 2015, 11 (8), 3584-3595. DOI: 10.1021/acs.jctc.5b00436.

(65) Vassetti, D.; Pagliai, M.; Procacci, P. Assessment of GAFF2 and OPLS-AA General Force

Fields in Combination with the Water Models TIP3P, SPCE, and OPC3 for the Solvation Free

Energy of Druglike Organic Molecules. Journal of Chemical Theory and Computation 2019, 15

(3), 1983-1995. DOI: 10.1021/acs.jctc.8b01039.

(66) Maier, J. A.; Martinez, C.; Kasavajhala, K.; Wickstrom, L.; Hauser, K. E.; Simmerling, C.

ff14SB: Improving the Accuracy of Protein Side Chain and Backbone Parameters from ff99SB.

Journal of Chemical Theory and Computation 2015, 11 (8), 3696-3713. DOI:

10.1021/acs.jctc.5b00255.

(67) Jorgensen, W. L.; Madura, J. D. Quantum and statistical mechanical studies of liquids. 25.

Solvation and conformation of methanol in water. Journal of the American Chemical Society

1983, 105 (6), 1407-1413. DOI: 10.1021/ja00344a001.

(68) Loncharich, R. J.; Brooks, B. R.; Pastor, R. W. Langevin dynamics of peptides: the

frictional dependence of isomerization rates of N-acetylalanyl-N'-methylamide. Biopolymers

1992, 32 (5), 523-535. DOI: 10.1002/bip.360320508.

(69) Berendsen, H. J. C.; Postma, J. P. M.; van Gunsteren, W. F.; DiNola, A.; Haak, J. R.

Molecular dynamics with coupling to an external bath. The Journal of Chemical Physics 1984,

81 (8), 3684-3690. DOI: 10.1063/1.448118 (acccessed 2021/03/24).

(70) Ryckaert, J.-P.; Ciccotti, G.; Berendsen, H. J. C. Numerical integration of the cartesian

equations of motion of a system with constraints: molecular dynamics of n-alkanes. Journal of

Computational Physics 1977, 23 (3), 327-341. DOI: https://doi.org/10.1016/0021-

9991(77)90098-5.

(71) Ester, M.; Kriegel, H.-P.; Sander, J.; Xu, X. A density-based algorithm for discovering

clusters in large spatial databases with noise. Proc. Second Int. Conf. Knowledge Disc. Data

Mining (KDD-96) 1996, 226-231.

.CC-BY-NC-ND 4.0 International licenseavailable under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprintthis version posted February 6, 2023. ; https://doi.org/10.1101/2023.02.06.527323doi: bioRxiv preprint

(72) Brenke, R.; Kozakov, D.; Chuang, G. Y.; Beglov, D.; Hall, D.; Landon, M. R.; Mattos, C.;

Vajda, S. Fragment-based identification of druggable 'hot spots' of proteins using Fourier domain

correlation techniques. Bioinformatics 2009, 25 (5), 621-627. DOI:

10.1093/bioinformatics/btp036.

(73) Kozakov, D.; Grove, L. E.; Hall, D. R.; Bohnuud, T.; Mottarella, S. E.; Luo, L.; Xia, B.;

Beglov, D.; Vajda, S. The FTMap family of web servers for determining and characterizing

ligand-binding hot spots of proteins. Nat Protoc 2015, 10 (5), 733-755. DOI:

10.1038/nprot.2015.043.

(74) The PyMOL Molecular Graphics System; Schrödinger, LLC: (accessed.

(75) Roos, K.; Wu, C.; Damm, W.; Reboul, M.; Stevenson, J. M.; Lu, C.; Dahlgren, M. K.;

Mondal, S.; Chen, W.; Wang, L.; et al. OPLS3e: Extending Force Field Coverage for Drug-Like

Small Molecules. J Chem Theory Comput 2019, 15 (3), 1863-1874. DOI:

10.1021/acs.jctc.8b01026.

(76) Schrödinger Release 2021-1 : LigPrep; Schrödinger, LLC: New York, NY, 2021. (accessed.

(77) Halgren, T. A.; Murphy, R. B.; Friesner, R. A.; Beard, H. S.; Frye, L. L.; Pollard, W. T.;

Banks, J. L. Glide: a new approach for rapid, accurate docking and scoring. 2. Enrichment

factors in database screening. J Med Chem 2004, 47 (7), 1750-1759. DOI: 10.1021/jm030644s.

(78) RDKit: Open-source cheminformatics; https://www.rdkit.org (accessed.

(79) Friesner, R. A.; Murphy, R. B.; Repasky, M. P.; Frye, L. L.; Greenwood, J. R.; Halgren, T.

A.; Sanschagrin, P. C.; Mainz, D. T. Extra precision glide: docking and scoring incorporating a

model of hydrophobic enclosure for protein-ligand complexes. J Med Chem 2006, 49 (21), 6177-

6196. DOI: 10.1021/jm051256o.

(80) Friesner, R. A.; Banks, J. L.; Murphy, R. B.; Halgren, T. A.; Klicic, J. J.; Mainz, D. T.;

Repasky, M. P.; Knoll, E. H.; Shelley, M.; Perry, J. K.; et al. Glide: a new approach for rapid,

accurate docking and scoring. 1. Method and assessment of docking accuracy. J Med Chem

2004, 47 (7), 1739-1749. DOI: 10.1021/jm0306430.

(81) Pedregosa, F.; Varoquaux, G.; Gramfort, A.; Michel, V.; Thirion, B.; Grisel, O.; Blondel,

M.; Prettenhofer, P.; Weiss, R.; Dubourg, V.; et al. Scikit-learn: Machine Learning in Python.

The Journal of Machine Learning Research 2011, (12), 2825-2830.

(82) Kim, S. S.; Aprahamian, M. L.; Lindert, S. Improving inverse docking target identification

with Z-score selection. Chem Biol Drug Des 2019, 93 (6), 1105-1116. DOI:

10.1111/cbdd.13453.

(83) Siddiqui, J. K.; Tikunova, S. B.; Walton, S. D.; Liu, B.; Meyer, M.; de Tombe, P. P.;

Neilson, N.; Kekenes-Huskey, P. M.; Salhi, H. E.; Janssen, P. M.; et al. Myofilament Calcium

Sensitivity: Consequences of the Effective Concentration of Troponin I. Front Physiol 2016, 7,

632. DOI: 10.3389/fphys.2016.00632.

(84) Davis, J. P.; Norman, C.; Kobayashi, T.; Solaro, R. J.; Swartz, D. R.; Tikunova, S. B.

Effects of thin and thick filament proteins on calcium binding and exchange with cardiac

troponin C. Biophys J 2007, 92 (9), 3195-3206. DOI: 10.1529/biophysj.106.095406.

(85) Bergrin, M.; Bicer, S.; Lucas, C. A.; Reiser, P. J. Three-dimensional compartmentalization

of myosin heavy chain and myosin light chain isoforms in dog thyroarytenoid muscle. Am J

Physiol Cell Physiol 2006, 290 (5), C1446-1458. DOI: 10.1152/ajpcell.00323.2005.

(86) Robertson, I. M.; Baryshnikova, O. K.; Li, M. X.; Sykes, B. D. Defining the binding site of

levosimendan and its analogues in a regulatory cardiac troponin C-troponin I complex.

Biochemistry 2008, 47 (28), 7485-7495. DOI: 10.1021/bi800438k.

.CC-BY-NC-ND 4.0 International licenseavailable under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprintthis version posted February 6, 2023. ; https://doi.org/10.1101/2023.02.06.527323doi: bioRxiv preprint

(87) Hoffman, R. M.; Li, M. X.; Sykes, B. D. The binding of W7, an inhibitor of striated muscle

contraction, to cardiac troponin C. Biochemistry 2005, 44 (48), 15750-15759. DOI:

10.1021/bi051583y.

(88) Meli, R.; Biggin, P. C. spyrmsd: symmetry-corrected RMSD calculations in Python. J

Cheminform 2020, 12 (1), 49. DOI: 10.1186/s13321-020-00455-2.

(89) Ohio Supercomputer Center. Ohio Technology Consortium of the Ohio Board of Regents,

1987.

.CC-BY-NC-ND 4.0 International licenseavailable under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprintthis version posted February 6, 2023. ; https://doi.org/10.1101/2023.02.06.527323doi: bioRxiv preprint