radical stabilization energies for enzyme engineering

doi.org/10.26434/chemrxiv.7379945.v2

Radical Stabilization Energies for Enzyme Engineering – Tackling theSubstrate Scope of the Radical Enzyme QueEChristian Suess, Floriane Martins, Anna Croft, Christof Jaeger

Submitted date: 21/05/2019 • Posted date: 21/05/2019Licence: CC BY-NC-ND 4.0Citation information: Suess, Christian; Martins, Floriane; Croft, Anna; Jaeger, Christof (2018): RadicalStabilization Energies for Enzyme Engineering – Tackling the Substrate Scope of the Radical Enzyme QueE.ChemRxiv. Preprint.

Experimental assessment of the reaction mechanisms and profiles of radical enzymes can be severelychallenging due to the reactive nature of the intermediates, and sensitivity of cofactors such as iron sulfurclusters. Here we present an enzyme-directed computational methodology for the assessment ofthermodynamic reaction profiles and screening for radical stabilization energies (RSEs) for the assessment ofcatalytic turnovers in radical enzymes. We have applied this new screening method to the radical SAMenzyme CPH4 synthase (QueE), following a detailed molecular dynamics (MD) analysis that clarifies the roleof both specific enzyme residues and bound Mg2+, Ca2+ or Na+. The MD simulations provided the basis for astatistical approach to sample different conformational outcomes. RSE calculation at the M06-2X/6-31+G*level of theory provided the most computationally cost-effective assessment of enzyme-based energies,facilitated by an initial triage using semi-empirical methods. The impact of intermolecular interactions on RSEwas clearly established and application to the assessment of potential alternative substrates (focusing onradical clock type rearrangements) proposes a selection of carbon-substituted analogues that would react toafford cyclopropylcarbinyl radical intermediates, as candidates for catalytic turnover by QueE.

File list (2)

download fileview on ChemRxivvRSE_QueE_revised-05042019_chemRxiv.pdf (1.60 MiB)

download fileview on ChemRxivSI-vRSE-resubmit-050419.pdf (5.57 MiB)

http://doi.org/10.26434/chemrxiv.7379945.v2

https://chemrxiv.org/authors/Anna_Croft/4832163

https://chemrxiv.org/authors/Christof_Jaeger/5674895

https://chemrxiv.org/ndownloader/files/15194426

https://chemrxiv.org/articles/Radical_Stabilization_Energies_for_Enzyme_Engineering_Tackling_the_Substrate_Scope_of_the_Radical_Enzyme_QueE/7379945/2?file=15194426



Radical Stabilization Energies for Enzyme Engineering – Tackling the Substrate Scope of the Radical Enzyme QueE

Christian J. Suess†, Floriane L. Martins†, Anna K. Croft†, Christof M. Jäger†*

† The University of Nottingham, Department of Chemical and Environmental Engineering, University Park, Nottingham,

NG7 2RD, United Kingdom.

ABSTRACT: Experimental assessment of catalytic reaction mechanisms and profiles of radical enzymes can be severely challenging

due to the reactive nature of the intermediates, and sensitivity of cofactors such as iron sulfur clusters. Here we present an enzyme-

directed computational methodology for the assessment of thermodynamic reaction profiles and screening for radical stabilization

energies (RSEs) for the assessment of catalytic turnovers in radical enzymes. We have applied this new screening method to the

radical SAM enzyme CPH4 synthase (QueE), following a detailed molecular dynamics (MD) analysis that clarifies the role of both

specific enzyme residues and bound Mg2+, Ca2+ or Na+. The MD simulations provided the basis for a statistical approach to sample

different conformational outcomes. RSE calculation at the M06-2X/6-31+G* level of theory provided the most computationally cost-

effective assessment of enzyme-based energies, facilitated by an initial triage using semi-empirical methods. The impact of

intermolecular interactions on RSE was clearly established and application to the assessment of potential alternative substrates

(focusing on radical clock type rearrangements) proposes a selection of carbon-substituted analogues that would react to afford

cyclopropylcarbinyl radical intermediates, as candidates for catalytic turnover by QueE.

INTRODUCTION

Radical intermediates are extremely versatile for chemical

functionalization and transformation reactions. Due to their

high reactivity, radicals can facilitate these reactions with

otherwise non-activated, unreactive substrates. However, this

advantage comes at the cost that these highly reactive

intermediates are particularly hard to control. This can lead to a

multitude of possible unwanted side reactions and is one of the

reasons why radical chemistry is predominantly found in

industry for processes where either such side reactions are

desirable or side reactions can be controlled or eliminated, for

example in the downstream processing of crude oil (cracking)

or in polymerization chemistry.

In nature, radical reactions play an important role in enzyme

catalysis. The radical SAM enzyme superfamily is one group of

enzymes that is capable of exploiting the potential of radical

reactions in a very controlled way. These enzymes are able to

initiate radical formation and direct their reaction by both

preventing side reactions and facilitating the desired reaction

simultaneously. This result of millions of years of evolution

harnesses key similarities in catalytic mechanism, and yet

results in a broad chemical reaction space (see reviews by

Broderick et al.1, Dowling et al.2, and Jaeger and Croft3 for a

critical summary). Radical SAM enzymes catalyze reactions

that include C-C bond formations,4-6 decarboxylation

reactions,7 functional group migrations (1,2-shifts),8-9 sulfur

insertions,10-13 methylations,14-15 and more complex radical

rearrangement mechanisms,16-19 with these radical-mediated

transformations involved in a multitude of biochemical

synthesis routes that include compounds with antibiotic and

antiviral activity. As such, it would be highly beneficial to gain

access to and adapt these biotransformations for their use in

industrial biotechnological applications, facilitating sustainable

routes towards fine chemicals, pharmaceuticals, or bulk

chemicals that would be highly challenging to synthesize by

alternative methods.3

The key commonality for the catalysis of radical SAM

enzymes is that they use S-adenosylmethione (SAM) either as

cofactor or co-substrate. SAM is bound to a central Fe4S4 iron

sulfur cluster responsible for initiating the redox reaction. The

cluster is embedded via binding to cysteine residues of a

conserved CX3CXφC motif (with φ representing a conserved

aromatic) and transfers an electron upon reduction to the SAM

molecule which subsequently cleaves to afford the 5’-adenosyl

radical (Ado•) and methionine (Met), which remains bound to

the cluster. The Ado• radical represents the first reactive

intermediate that then abstracts a hydrogen from a bound

substrate to initiate catalysis.

Generally the control of radical intermediates in radical SAM

enzymes is based on perfect positioning of the substrate towards

the cluster bound SAM and the stabilization of the radicals. The

argument is that more stable radicals are less likely to undergo

unwanted side reactions, for example with the enzyme itself.

However we recently showed with the example of the radical

rearrangement in 7-carboxy-7-deazaguanine (CDG) synthase

(QueE)20 that this assumption is not always true. In the case of

QueE, the unmodified substrate radical is so stable that the

energy barrier for the subsequent radical rearrangement is too

high for efficient catalysis, unless the radical is structurally or

electronically perturbed.21

The reaction catalyzed by QueE represents a central step in

queosine synthesis and facilitates the formation of the 7-

deazapurine scaffold through a radical-mediated ring

contraction.22 Scheme 1 shows the radical rearrangement

during the catalysis of QueE propagating through an

azacycloproylcarbinyl intermediate followed by NH2

elimination. This mechanism has been first described as one of

two potential pathways by Drennan and coworkers,20 and was

confirmed computationally by Zhu and Liu23 and us.21 The

rearrangement of CPH4 proceeds after initial hydrogen

abstraction from the C6 position through a cyclic

azacyclopropylcarbinyl intermediate (3), before hydrogen re-

abstraction from AdoH and deamination to form the final

product (6). Intermediate 3 and its analogues also represent a

class of structures known as radical clocks,24 which undergo

quick unimolecular rearrangements and have been described

extensively by experiment and theory.25-32

We recently demonstrated that it is necessary to hold the

substrate radical 2 in an energetically unfavorable configuration

to overcome the rearrangement barrier for the ring conversion.

This conformation is achieved by binding the substrate already

in this conformation (which represents for the substrate a local

energy minimum only slightly higher in energy than the

unbound conformer) and hold this conformation after hydrogen

abstraction through electrostatic fixation by a Mg2+ ion in the

active site of the enzyme. Without this constraint, the radical

would fall into a very stable, planar conformation and the

energy barrier for the rearrangement would be too high for

efficient reaction turnover. In other words, the substrate radical

needs to be destabilized in order to facilitate the reaction.

This observation was also demonstrated by evaluating the

stabilization of the radicals represented by radical stabilization

energies (RSEs). By calculating the energy of a formal

hydrogen transfer reaction (as described in the Experimental

Section), RSEs inform on the reactivity of radicals in

comparison to a given reference species. Thus, the higher the

radical stabilization energy, the less energized and reactive the

corresponding radicals are, also diminishing their role in

unwanted side reactions. The concept of calculating RSEs has

been used extensively for evaluating accurate absolute bond

dissociation energies (BDEs) for a broad range of radicals

including many systems related to peptides and enzyme

catalysis.33-40

More recently Hioe and Zipse also demonstrated the

usefulness of calculating radical stabilization energies for

radical SAM enzymes.40 They have calculated the

thermodynamic reaction profiles for selected radical SAM

enzymes on the basis of high-level RSE calculations in gas

phase and demonstrated that enzymes using SAM as cofactor

generally combine an initial exothermic hydrogen abstraction

step with a subsequent endothermic step, while enzymes using

SAM as co-substrate perform significantly exothermic H-

abstraction reactions. What was not incorporated in this study

was the effect on the radical stabilities and on the reaction

profiles of binding to the enzyme and additional cofactors. As

shown above, QueE is one such example where the

intermolecular environment has a significant influence on these

observables.

Scheme 1. Proposed ring contraction mechanism in the radical

SAM enzyme QueE.

In this study, a combination of molecular dynamics (MD)

simulations and RSE calculations is presented in order to

investigate the effect of intermolecular binding on RSE

estimation. It is demonstrated that a quick and affordable

strategy can highlight if the enzyme environment facilitates

significant changes to the reaction profile in comparison to the

reaction assessed in gas phase or solution. Further, this

information can subsequently be used to screen alternative

substrate candidates for their potential for conversion, by

comparing their reaction profiles to the native reaction.

Potential enzyme mutation candidates can also be screened that

may stabilize radical intermediates differently to the wild type

enzyme. Thus, the combination of MD sampling and RSE

calculations has the potential to be used as an initial tool for in

silico enzyme design of radical SAM enzymes, bringing these

enzymes one step closer to efficient biotechnological

applications.

RESULTS AND DISCUSSION

MD SIMULATIONS

As previously indicated, QueE is found to facilitate catalysis

only effectively in presence of Mg2+.20, 22 To investigate the

effects on substrate binding and conformation in the presence

of different ions in the enzyme, a series of molecular dynamics

simulations have been carried out. Starting from the crystal

structures of QueE20 in complex with SAM, the substrate CPH4

(1) and Na+ (pdb code 4NJH) or Mg2+ (pdb code 4NJI) in the

active site, the corresponding ions were either left unaltered,

switched between the two structures, or substituted by Ca2+. All

simulations were run in 5 replicas and for at least 100 ns and

maximal ~1 µs resulting in a total simulation time of

approximately 8 μs.

Figure 1. Analysis of 200 ns simulation time for CPH4 bound to QueE in the presence of Mg2+ (top), Ca2+ (middle), and Na+ (bottom).

The RMSD analysis (left) show high overall structural stability during simulation. The simulation time presented starts after 9 ns of

restrained equilibration and the equilibration and the rmsd is calculated against the crystal structure configuration. The binding

analysis of CPH4 (right) depicts significant differences and changes for the binding position towards hydrogen abstraction (C6-C5’

distance) and binding conformation (represented by the O-C-C6-H5 dihedral angle in CPH4).

During the timeframe of all simulations, binding of SAM to

the iron-sulfur cluster and the overall behavior of the dimeric

enzyme showed high stability and low fluctuations. In a small

number of simulations, conformational changes for SAM could

be observed after several hundred nanoseconds, which was

followed by the movement of a loop involved in SAM binding

(see Figure S20 in the Supporting Information). Substrate

unbinding, on the other hand, was observed in all cases after

varying simulation time.

All simulations could reproducibly demonstrate significant

differences for different ions in the active site. Based on the

results of our previous study, which highlighted the need to

bring the substrate into a bent conformation for optimal

catalysis, the initial simulation analysis focused on analyzing

this conformation and the positioning of the substrate for the

initial hydrogen abstraction step.

Only simulations with Mg2+ in the active site kept the

substrate in the bent conformation for significant simulation

times, with the average retention time in the preferred

conformation of 146 ns. The range covered binding of only

10 ns in one case and over 226 ns in other. Such long retention

times were rarely seen for simulations containing Ca2+ and even

less frequently for those containing Na+, demonstrated by more

flexible binding and much shorter retention times of the correct

binding mode. The observed average retention times for these

ions were 18 ns for Ca2+ (0-64 ns) and 2 ns (0-15 ns) for Na+

based on the analysis of 5 replicate simulation for each of the

systems containing the substrate and one of the ions Na+, Mg2+,

Ca2+, accounting for the sampling of 10 unbinding events for

each system (see Table S25 in the Supporting Information for

details).

It should be noted that these retention times based on

simulations starting from the bound conformers cannot be used

to accurately calculate the binding affinities of the substrate to

the enzymatic binding site. This would require a large set of

very long free simulations sampling multiple binding and

unbinding events and was not within the scope of this paper. In

our approach we were interested in sampling possible Michaelis

complexes that represent encounter complexes necessary to

facilitate the subsequent hydrogen abstraction and radical

rearrangement steps and then assess the effects of different

encounter complexes on the thermodynamic reaction profiles of

the subsequent catalytic reactions. Therefore, the observed

differences in binding retention times should not be used as a

truly quantitative measure but as a trend. The observed binding

features and retention times, however, still resemble observed

experimental binding affinities for the formation of the

Michaelis complex very well which have been reported in the

form of kinetic spectroscopy methods and characterize a

relatively slow enzyme with an overall kM of 20±7 µM and

turnover of 5.4 ± 1.2 min-1.41

Figure 1 depicts the essential binding analysis of substrate

binding for three selected simulations over 200 ns each (more

detailed analysis for all simulations can be found in Section 1.4

of the Supporting Information). All three simulations started

from the crystal structure 4NJI with the substrate bound in the

reactive conformation and the analysis started after careful and

extensive restrained equilibration for 9 ns. The RMSD of all

simulations remains low (Figure 1 left) and the SAM molecule

remains bound to the iron sulfur cluster interacting with the

unique iron through one carboxylate oxygen and the nitrogen of

the methionine moiety. The carbon-carbon distance, essential

for the hydrogen transfer between the C5’ carbon of SAM and

the C6 carbon of CPH4, already depicts significant variation and

differences. Without the need to calculate the transition state

structure and energy for the hydrogen abstraction and on basis

of structural sampling with SAM and not the cleaved Ado•

radical, this C-C distance can already provide information on

the relative likeliness for this initial hydrogen abstraction.

While for the Mg2+ simulation the substrate stays in adequate

position for almost the entire 200 ns in both active sites of the

dimeric enzyme, both substrates leave the correct position in the

Ca2+ simulation within a few nanoseconds and even earlier

during the simulation with Na+. In most cases, leaving the initial

position is directly coupled to changing into the more planar

conformer of CPH4 (also depicted in Figure 1). Only in some

cases is the substrate seen leaving the pocket together with the

tightly bound Mg2+ before the substrate changes its

conformation. In contrast, in simulations with Ca2+ and Na+ ions

left the binding pocket without the substrate.

A more detailed look at substrate binding with different ions

reveals further key differences (Figure 2). In all simulations, the

substrate is fixed in the active site by hydrogen bonds between

its carboxylate oxygens and the side chains of Arg27 and Thr90.

Additional frequently found hydrogen bonds, common for all

simulations, involve the side chains of residues Gln13 and

Glu15 and the backbone of Gly14. The main variations between

the simulations arise from the ion coordination within the active

site.

Mg2+ coordinates throughout the whole simulation to one

carboxylate and carbonyl oxygen, fixing the substrate in the

bent conformation (Figure 2a). The ion is further coordinated

by 4 water molecules but with no direct interaction observed

with residues in the active site. Coordination to Asp50 and

Thr51 is mediated through first shell water molecules. This is

in slight disagreement with the crystal structure 4NJI, which

suggests a direct coordination between Mg2+ and Thr51, but

with unusual long interatomic distances of 2.7 and 2.9 Å (in the

two binding sites respectively.20 The two monomers in the

crystal structure dimer also depict differences in the electron

density in their active sites, which might be attributed to

differing positions of water molecules or varying ion

occupation over the average of the whole crystal (please refer

to sction 1.5 of the Supporting Information for a more detailed

comparison of simulation and crystallographic data).

Interestingly, an additional simulation with the proposed

intermediate 5 and Mg2+ in the active site (Figure 2d) resulted

in a coordination chemistry resembling the crystal structure,

with direct coordination between the ion and Thr51, 3 water

molecules and the substrate. Further, the hydrogen bonding

interaction between the carboxylate of the C-terminal Pro209

residue and the substrate, described in the crystal structure, is

indicated as being much stronger in this intermediate-

containing simulation. This contrasts with simulations

containing bound substrate, where this interaction is loose and

the C-terminus shows flexibility during the simulations.

Distinct from Mg2+, Ca2+ coordinates the carbonyl oxygen of

CPH4 and Asp50 directly, while the contact to the carboxylate

of the substrate is water-mediated (Figure 2b). The frequency

of hydrogen bonding to Glu13 and Gln15 is also significantly

reduced. This suggests additionally that, during the period

where the substrate adopts the bent conformation, it is fixed less

tightly in this conformation. This is certainly also the case in the

very short time of correct binding observed with Na+. The ion

either leaves the pocket quite rapidly or is coordinated to Asp50

and the carboxylate of the substrate. Thus, the substrate flips

rapidly into the planar conformation and also migrates from the

optimal position for hydrogen abstraction.

In summary, clear differences for the binding of the substrate

and the different ions in the active site are seen by MD

simulation analysis. Mg2+ seems to place the substrate in much

better position for abstraction and in the correct conformation,

relative to what is seen for the other ions. However, the question

as to what this effectively means for the radical rearrangement

remains open.

RSE ASSESMENT

The calculation of radical stabilization energies (RSEs)

relating to different substrates and intermediates can provide a

clearer thermodynamic picture of radical reactions in radical

SAM enzymes. These calculations alone, though, lack

information about the influence of enzyme binding on the

thermodynamic reaction profile. From the MD simulations and

our previous work on a DFT model system21 we already know

that intermolecular interactions seem to have an important

influence on the conformational space available for the reaction

and thus on the stability of specific intermediates.

This study uses MD simulations as a sampling method for the

system incorporating the bound substrate to then subsequently

calculate the radical stabilization energies from multiple

snapshots of these simulations. Firstly, this is applied without

any further optimization of the structures from the simulations

in order to get a rapid insight into changes in radical

stabilization energies upon enzyme binding. Thus, these

calculated values do not represent adiabatic radical stabilization

energies, but vertical radical stabilization energies (vRSEs)

neglecting geometrical relaxation after hydrogen abstraction.

Figure 2. Substrate and ion binding in the active site of QueE.

Snapshots taken from simulations with a) Mg2+, b) Ca2+, c) Na+.

d) depicts the binding of intermediate 5 within the pocket in the

presence of Mg2+.

Due to the lack of structural relaxation of the substrate and,

more importantly, the radical, absolute vRSE values cannot be

compared to accurate RSE values from high level DFT gas

phase optimizations. However, when compared to the similar

sampling in the gas phase it can provide direct information

regarding how conformational restrictions in the active site

might influence the stability of the intermediate generated

during formation. In other words, it can signpost significant

influences brought about by enzyme-induced binding and

interactions.

In a second step, the protein environment is added to the

calculations to look for further effects influencing radical

stability. Additional residues can either explicitly be taken into

account in the calculations or be represented by partial point

charges. In that way, the impact of electrostatic effects in the

active site in further influencing radical stability can be

observed.

In a final step, the substrate and its radical form are optimized

within a field of point charges representing the enzyme. This

computationally more expensive QM/MM-type step confirms if

the radical is able to substantially change its conformation once

formed, and if this would have an effect of its stability and the

energetic profile of the radical reaction.

An initial simulation where the substrate is bound correctly

with Mg2+ was investigated. From this simulation, five thousand

snapshots along a simulation time of 100 ns were extracted. For

comparison, the same number of snapshots have been generated

from a 500 ns simulation of the substrate in the gas phase.

Subsequently, RSEs have been calculated for the C6 hydrogen

abstracted radical (2) of the substrate for all snapshots on basis

of the closed shell molecular mechanics geometries with a

number of methods and basis sets.

Three exchange functionals (and standard Hartree-Fock

calculations) and six basis sets have been considered and the

results are shown in Table 1. Double and triple-ζ basis with

polarisation and diffuse functions have been cross-analysed

with functionals of varying degrees of sophistication. As

expected, the Hartree-Fock (HF) calculations performed purely

as standard implementations of HF are known to overestimate

energies in protein systems and fail to describe dispersion

correctly, making it an inappropriate method for

thermochemical calculations. The Minnesota suite of

functionals have been well received as suitable candidates for

studying kinetics, thermochemistry and non-covalent

interactions,42 hence why two (M06-2X and M11) have been

included in the preliminary calculations. Here the M06-2X

functional was designed for these type of calculations, and the

M11 has been presented as an improvement.43 M06-2X was

used for further calculations as it was a more dependable

solution than M11, which had difficulty converging for this

particular set of systems. All further calculations have also been

corrected for dispersion by use of Grimme’s D3 dispersion

correction with correction values taken from literature.44

As the focus of this study depends more on relative shifts

rather than absolute values, the choice of functional is not as

significant as the choice of basis set. As long as a sensible

choice is made, relative energy shifts are similar (see also

Table S26 in the Supporting Information). In summary,

calculations at the M06-2X/6-31+G* level of theory presented

best-balanced cost-accuracy relation of all DFT methods tested

and all following DFT results are presented using this approach.

Table 1. Averaged vRSE values for 5000 snapshots of a MD

simulation with Mg2+ and bound substrate (MD simulation ID1,

chainA) at different levels of theory. Energies are presented in

kJ mol-1.

HF B3LYP M06-

2X

M11

6-31G* -237.3 -170.4 -146.5 -147.5

6-31+G* -141.8 -37.7 -33.7 -40.6

6-31++G* -238.2 -37.5 -33.5 -40.5

6-311+G* -147.5 -34.2 -30.2 -30.3

6-311++G* -147.2 -33.7 -29.8 -31.2

G3Large -139.9 -25.9 -22.3 -20.9

RSE values have also been tested with semi-empirical (SE)

methods to confirm their applicability for a quick RSE

screening from MD data since the use of SE is significantly

faster, with a speed up of around 1200-fold. A positive linear

correlation between the calculated RSE values from both DFT

(M06-2X/6-31+G*) and SE (PM6-D3) (r2 = 0.86Å) could be

observed for the given example (see Figure S25 of the

Supporting Information). This correlation allows the

subsequent construction of a Gaussian distribution of the SE

RSE’s and then use of QM to calculate the RSEs of higher

accuracy on a subset of selected frames near the Gaussian peak

position (see Table S28 of the Supporting Information). This

delivers RSE values of high accuracy at reduced calculation

time.

The RSE values have then been calculated for the Mg2+-

bound substrate with and without optimization in the

electrostatic field of the protein and have been compared to the

initially calculated vertical RSE values. As can be seen from the

data presented in Table 2, the vertical radical stabilization of the

substrate radical bound correctly in the active site of QueE

together with Mg2+ drops significantly by 34.0 kJ mol-1 in

comparison to gas phase sampling only considering the

substrate/radical itself in the conformation retrieved from the

MD simulations directly. Thus, it can clearly be seen that the

main contribution to the change in stability originates from the

conformational change. Further, the standard deviation of the

energies appears to be higher (SD ±20.2) for the bound

conformations. This is also expected, as these radical

conformations are not close to a stable conformational

minimum and thus slight geometric changes represent larger

energetic changes on the potential energy surface.

Table 2. Radical stabilization energies from 100ns MD

simulations of CPH6 in vacuum or bound to QueE with Mg2+ in

the active site (MD simulation ID2, chainA) at the M06-2X/6-

31+G*(D3) level of theory. Energies presented in kJ mol-1.

RSE SD Shift

Vacuum -54.8 ±13.1 0.0

Vacuum, optimized -104.4 ±2.3 0.0

Protein -20.8 ±20.2 34.0

Protein + point charges -19.4 ±15.6 35.5

Protein + pchg, opt -33.7 ±23.3 70.7

When the same procedure is repeated optimizing both the

substrate and the radical in the electrostatic field of the protein,

represented by atomic partial charges, the RSE values drop to

more negative values. The shift in RSE upon enzyme binding

even increases to -70.7 kJ mol-1 and while the standard

deviation for the unconstrained gas phase system drops to a very

low value of ±2.3, it remains high (±23.3) for the bound system.

This once more indicates a radical in an uncomfortable

conformation far from the preferred optimum and thus with

large energy changes upon small structural changes.

When including the electrostatic field without further

optimization (and comparison to the gas phase system) it can be

seen that the electrostatic field seems to only have a small effect

on radical stabilization for this example. Comparing the results

after QM/MM optimization between DFT and SE calculations

moreover shows that optimizations at the SE level are not

reliable for calculating relative shifts of RSE values for

constrained molecules (see Table S27 in the Supporting

Information). Thus, the semi-empirical sampling can only be

suggested for an initial sampling of MD trajectories as

described before.

When comparing simulations of the substrate bound in the

active site together with different cations, the consequences of

the structural and dynamic differences seen in the MD

simulations on the radical stability of the substrate in the

enzyme are demonstrated very clearly. The average

stabilization in cases where the substrate is bound in the correct

conformation is shown to be significantly lower compared to

unreactive binding of the substrate. As can be seen from Table 3

in a simulation where unbinding can be observed during the

simulation with Ca2+ in the binding pocket, the vRSE value

drops by over 25 kJ mol-1 from -33.9 (bound) to -59.1 kJ mol-1

(unbound).

Table 3. Vertical RSE values for MD simulations of substrate

bound to active site with different cations at the M06-2X/6-

31+G*(D3) level of theory. Energies presented in kJ mol-1.

System1 Sampling

time [ns]

Average

RSE

SD

4NJI Mg2+ 200 -20.8 20.2

4NJI Ca2+ (bound) 55 -33.9 12.3

4NJI Ca2+ (unbound) 45 -59.1 11.1

4NJH Ca2+ 100 -51.6 12.5

4NJH Na+ 100 -51.2 11.5

1 Snapshots for RSE calculations taken from simulation ID 2 for Mg2+,

1 for Ca2+, and 1 for Na+.

The effect of unbinding on the radical stability of the

substrate (and thus on the thermodynamic reaction profile) can

also be monitored in form of QM post-processing of the

underlying MD simulation. In this way, changes of this central

feature can be monitored in quasi-real (simulation) time without

other analysis needed. Figure 3 demonstrates this for the Ca2+

simulation. The stability of the substrate radical shifts

significantly after ~56 ns. Additional structural analyses

confirm that this shift is correlated with movement of the

substrate slightly out of the pocket, which changes the

complexation to the cation and results in conformational change

of the substrate. While the substrate is still anchored in the

active site by the strong salt bridge between Arg27 and the

substrate’s carboxylate, it is not binding in a reactive fashion

anymore. As demonstrated in the MD discussion above, this

behavior is much more likely when Mg2+ is substituted by other

cations in the active site.

POTENTIAL SUBSTRATE SCOPE OF QueE

In the view of the possibility to apply the method to calculate

vRSE values for radical enzymatic reactions in the context of

protein engineering, a set of alternative substrates have been

tested for their potential to be converted through a similar

reaction mechanism to the natural substrate QueE. A set of

structures were selected that are also able to react via an

analogous radical rearrangement through either an

azacyclopropylcarbinyl radical or their carbon- and sulfur-

substituted analogues. The structures were chosen to represent

different substitutions next to the radical center that either

stabilize or destabilize the radical by electron pushing or pulling

effects, or might influence substrate binding through the

presence or absence of functional groups necessary for

hydrogen bonding. The selected structures are listed in

Figure 4.

Figure 3. Representative snapshots of MD simulation of

substrate CPH4 (orange licorice) bound in QueE with Ca2+

(orange sphere) and SAM (licorice representation) in (a) active

and (b) inactive (unbound) conformation. Distribution of M06-

2X/6-31+G* vRSE values at over a time window of 1 ns after

50 and 100 ns simulation (simulation ID 1), including Gaussian

fit to data (c, d) and plot of crucial H-abstraction C-C distance

(e) and vRSE values (f) over 100 ns simulation.

The gas phase radical stabilization energies of the alternative

substrates were evaluated in an analogous fashion to our

previous paper21 at the M06-2x/6-311++G(3df,3p)//B3LYP/6-

31+G(d) level of theory and have been compared to the

corresponding radical rearrangement barriers at the same level

of theory (and to even more accurate G3B3 data, where

affordable). The results presented in Table 4 and Figure 5

clearly show two remarkable trends. Firstly, the radical

rearrangement barriers for azacyclopropylcarbinyl appear to be

significantly higher compared to their cyclopropylcarbinyl

counterparts. Secondly, the trend that higher radical

stabilization correlates with higher rearrangement barriers is

confirmed for the selected examples. The trend is more

prominent for rearrangements through the

azacycloproylcarbinyl and outliers can mainly be attributed to

heterocyclic structures that either contribute with additional

spin delocalization within the ring or represent structural

constraints due to the ring structure, which additionally hinder

the rearrangement.

Subsequently, the alternative substrates examined were

docked into the crystal structure of QueE using the docking

program GLIDE from the Schroedinger suite of programs. For

the resulting enzyme substrate complexes, the radical

stabilization energies were then calculated as single point

energies of the docked substrate conformations and included

optimizing the substrate and substrate radical structures in the

electrostatic field of the enzyme.

Figure 4. Alternative substrate radicals considered for docking and

RSE studies.

The results were evaluated taking into account the following

selection criteria for potential alternative substrates of QueE: 1)

the correct positioning of the substrate carbon involved in the

necessary first hydrogen abstraction step between the Ado•

radical and the potential substrate to initiate the radical

rearrangement; 2) the corresponding docking score as an

indicator for substrate affinity to the pocket (This score is also

compared to the lowest docking score of the corresponding

substrate to see whether alternative binding conformations are

more likely); 3) the radical stabilization energy of the substrate

in order to evaluate if a high rearrangement barrier is to be

expected; 4) the RSE after optimization in the electrostatic field.

This last assessment adds information about the likelihood of a

substrate to be bound in a preferred conformation for catalysis,

but where the radical would undergo quick relaxation to form a

stable inactive radical intermediate.

All these four criteria do not only give valuable information

about whether a potential substrate might be a good candidate

for catalysis, but also indicate potential ways to improve the

candidacy by signposting additional mutations within the

enzyme’s active site. The full docking results are presented in

the Table S29-30 in the Supporting Information and the most

interesting results are briefly discussed below.

From the docking calculations, the three conformers of each

molecule with the best docking score have been taken for

further RSE analysis. Additional docking studies without

including the Mg2+ ion in the active site were also performed to

investigate if some substrates might be suitable for

transformation without the support of the ion.

Table 4. Calculated radical stabilization energies and

rearrangement barriers (Ea) for QueE. Energies presented in

kJ mol-1.

RSE Ea

M06-2Xc G3B3 M06-2Xc G3B3

N1 -60.8 -57.6 99.5

N2 -73.9 -73.8 105.6 102.1

N3 -87.2 -82.2 132.0 123.9

N4 -98.5 117.8

N5 -137.1 174.1

N6 -107.0 -95.5 138.1 130.3

N7 -104.6 143.4

N8 -94.0 143.3

C1 -26.4 -23.6 43.2

C2 -90.3 -88.9 82.4 83.8

C3 -50.2 -49.1 39.5 38.8

C4 -54.0 50.2

C5 -78.0 93.1

C6 -47.2 -38.9 65.5 68.3

C7 -41.4 95.6

C8 -47.8 101.8

S1 -47.3 65.2

S2 -52.1 -56.0 69.9 69.7

S3 -53.5 -58.7 71.1 70.5

S7 -64.7 136.3

S8 -71.4 138.7

Figure 5. Calculated radical stabilization energies and

rearrangement barriers (Ea) for QueE. The trend for the

correlation between the activation barriers and RSE values

values is highlighted in yellow. The green area represents

rearrangement barriers approximately suitable for catalysis.

The red circles highlight the values for the natural substrate

CPH4 in gas phase and in the model system from our previous

study.21

Figure 6. a) natural substrate radical 2 (orange ball and stick

representation) after docking and geometry optimization in the

active site of QueE superimposed with the crystal structure of

CPH4 (1, green licorice). b) alternative ligand C3 after docking

and geometry optimization in QueE without Mg2+.

Applying this protocol to the natural substrate CPH4 (N8)

(see Table 5) delivered reasonable docking poses with and

without Mg2+ in the active site. Subsequent RSE calculations

including QM/MM geometry optimisations could confirm an

increase in radical stabilization but the structures did not relax

to a very stable planar conformation, making the rearrangement

in principle possible.

Across the complete dataset, substrates docked in preferential

position and conformation often drop into a more preferable and

less reactive radical conformation with high radical

stabilization upon optimization. This indicates that these

substrates are unlikely to undergo a catalytic rearrangement and

that additional mutations of active site residues might be

necessary to further maintain reactive conformations. Further,

the observation that the single point calculations including the

point charge (PC) field, but neglecting further optimizations,

differ much more significantly from single point calculations

without the PC field (compared to the results from vRSE single

point calculations based on MD simulations) indicates that

equilibration after docking is necessary for an adequate

interpretation and reliable results, either through MD or

geometry optimizing the structures.

Unsurprisingly, substrates without carbonyl or carboxylic

functional groups cannot be attached strongly to residues in the

active site (in particular Arg27) and thus result in very low

docking scores and variable docking orientations (Table S30 in

the Supporting Information). Thus, substrates C1, C2, C4, and

C5, for example, show relative low RSE values, but do not bind

strongly enough or in the correct position for catalysis and are

very unlikely to act as alternative substrates.

For structures able to facilitate supporting anchoring by

Arg27, on the other hand, catalytic turnover might be possible

(see ligands C3, C6, C7, and C8 in Table 5) with improved

positioning and even without Mg2+ support in some cases.

Figure 6b depicts ligand C3 docked and subsequently

optimized in its radical form in the pocket of QueE without

Mg2+ demonstrating adequate positioning and radical

stabilization for enzymatic turnover. In general, the

cyclopropylcarbinyl precursors, that do not need to overcome a

similar high transition barrier to the aza-cyclopropylcarbinyl

analogues, might function better without Mg2+ due to better

positioning. Also, ligand N3 (which includes the anchoring

carbonyl but lacks the ring structure of the natural substrate)

might undergo turnover without Mg2+ as its radical does not

need to be fixed in a tightly bended conformation.

Table 5. Selected RSE values with and without QM/MM geometry optimization in the point charge (PC) environment, based on

docked structures of alternative substrates from Figure 4. XPscore denotes the extra precision Glide docking score, and the C6-C5’

distance indicates the crucial distance for the initial hydrogen abstraction reaction in the docked structure.

On the other hand, structures that are placed well for initial

hydrogen abstraction but where significant stabilization occurs

upon optimization effectively might act as inhibitors. Although

they show the principle potential for rearrangement, they are

RSE [kJ mol-1]

Ligand

(docking

rank)

With /

without Mg2+

XPscore

(kJ/mol)

C6-C5’

distance (Å)

SP (vRSE) SP with PC Opt with PC RMSD

C3 (1) Mg2+ -26.6 4.57 -9.4 -6.7 -51.9 1.04

C3 (3) Mg2+ -27.6 4.47 -11.3 -5.7 -56.1 0.89

C3 (1) - -14.3 4.00 -4.9 4.6 -20.7 0.41

C3 (1)

MD 150 ns

Mg2+ -26.6 4.59

±0.27

-18.4

±9.9

-9.9

±10.5

-37.5

± 15.3

C6 (1) Mg2+ -28.3 4.60 -4.7 7.6 -86.4 1.14

C6 (2) - -21.4 3.92 -18.3 -14.8 -46.9 0.21

C7 (1) - -17.5 3.65 -8.1 -2.5 -19.0 0.33

C8 (2) Mg2+ -33.7 5.21 -5.9 -8.1 -65.5 0.68

C8 (1) - -15.5 3.75 -8.1 -2.9 -17.3 0.29

N3 (1) Mg2+ -28.6 5.42 -20.2 -17.9 -26.2 1.19

N3 (3) Mg2+ -28.8 3.91 -14.4 -12.1 -112.2 0.85

N3 (2) - -12.9 4.27 -21.4 -11.2 -48.8 0.53

N8 (1) Mg2+ -28.1 3.80 -43.9 -46.9 -77.1 0.17

N8 (2) Mg2+ -32.2 5.56 -14.4 -12.6 -33.4 0.17

N8 (1) - -30.6 3.76 -38.9 -38.5 -84.9 0.19

N8 (2) - -21.8 5.01 -42.1 -43.5 -59.9 0.22

trapped in a low energy minimum after hydrogen abstraction

and thus are inhibited from further turnover. Examples showing

this behavior for all conformers analyzed are ligands C4 and N4

(see Table S29-30 in the Supporting Information for details).

To further prove the potential suitability of alternative

substrate C3 we have conducted another set of MD simulations

based on the docked structures with and without Mg2+ and

followed up by RSE assessments. The simulations, performed

in triplicate, showed that the initial assumption of good binding

without Mg2+ resulted in relatively rapid unbinding of the

alternative substrate in all three simulations. With Mg2+ in the

active site, the simulations showed stable binding for at least

70 ns in one case and for the full simulation time of 150 ns in

the other two cases. Subsequent RSE analysis of 300 snapshots

(shown in Table 5) of one of the simulations confirmed the

observations from the docking simulations demonstrating

suitable binding and RSE stabilization for turnover. Therefore,

the MD simulations added a crucial binding equilibration

assessment, not available by simple docking assessments and

should be added to any workflow applied to selected hit

structures.

In summary, this first combined docking and RSE assessment

gives new insights into how the radical enzyme QueE controls

the central catalytic radical rearrangement. The protocol

developed here offers a new and suitable approach to assess the

thermodynamic reaction profile of natural and alternative

substrates in radical SAM enzymes. Excitingly, it has good

potential to serve as a pre-screening tool for alternative

substrates in radical enzymes, as a first step towards enzyme

engineering in this underexploited domain.

CONCLUSION

Radical stabilization plays an important role in the enzymatic

catalysis of radical SAM enzymes. Often the way intermediate

radicals are stabilized or destabilized within the active site

determines the rate determining steps of the catalysis of these

enzymes. In cases where the reaction mechanism is known,

evaluating the thermodynamic reaction profiles for these

enzymes offers an affordable gateway into computational

predictions of substrate scope and initial steps into

reengineering substrate scope, turnover, or promiscuity of these

enzymes.

In this study on QueE we show that the combination of MD

simulations and the evaluation of radical stabilities delivers

deeper understanding in how the enzyme controls enzymatic

turnover and what the crucial role of the central Mg2+ ion is,

namely that tight control of the reactive radical conformation

and stability is only supported by Mg2+ ions but not by other

cations. This is represented by tighter binding of the substrate

in the reactive Michaelis complex conformations as

demonstrated by longer retention times in this conformation and

the significant lowering of the radical stabilization of the

substrate radical in this conformation that correlates with a

lowering of the transition barrier for the radical rearrangement.

Applying a combination of substrate docking and radical

stabilization assessment on a set of alternative substrates

delivered detailed information about potential alternative

substrates and inhibitors. We could confirm that only ligands

that show significant lower stabilization in their preferred

conformation, or substrates that can be stabilized in a less

preferred conformation in the enzyme, are likely to act as

alternative substrates. Cyclopropylcarbinyl precursors were

shown to be more likely to undergo turnover than heteroatom

substituted analogues, based on their smaller radical

stabilization and lower rearrangement barriers. Also, anchoring

to Arg27 by a functional group of the ligands (preferably by

carboxylates) is necessary to bring the ligands in optimal

position (for initial hydrogen abstraction) and stabilize reactive

conformations for potential turnover.

These results show that there is significant potential in the

presented methodology to be used as a screening approach for

enzyme engineering of radical SAM enzymes, other radical

enzymes/processes, and more broadly for enzymes that proceed

via other highly reactive intermediates (e.g. those proceeding

through reactive cationic intermediates). Screening

thermodynamic reaction profiles is easier and quicker to

perform than more costly transition state searches for multistep

reactions, like the reaction presented here, and the use of

intermediate energies in such reactive systems is supported by

Hammond’s postulate, provided similar systems are compared.

Further, the use of this type of screening is particularly relevant

in these cases, as the reactivity of the intermediate often means

that its generation is also the rate determining step in the

enzyme process.

The next step for this methodology is to screen for mutations

in the active site, either supporting turnover for alternative

substrates or altering turnover for the natural substrate, in this

case CPH4. Such alterations may improve reactivity, or provide

leverage for greater selectivity for a mixture of similar

substrates, alongside providing a greater insight into the

contributions of both individual and groups of residues on the

mechanism. Having a rapid screen at hand to look at mutants

will serve as the entry point for the computational design of

radical SAM enzymes and other processes proceeding through

reactive species, facilitating the development of reactions with

non-natural substrates for the generation of novel, bioactive

compounds.

In this spirit, we propose a general computational screening

methodology for alternative substrates for radical SAM

enzymes and mutants of radical SAM enzymes as outlined in

Figure 7. The methodology starts with identifying features of

the natural enzyme-substrate complex. Subsequent docking

screens and MD equilibrations of hit structures are followed by

further alternative mutant screens to find potential alternative

substrates and enzyme mutants capable to react with these

substrates. A more detailed technical workflow can also be

found in Section 4 of the Supporting Information and is freely

available in form of a Jupyter Notebook application.

As mentioned before, the central RSE evaluation within the

workflow serves as a rapid evaluation of the thermodynamic

energy difference between a highly reactive intermediate and

the substrate that correlates to the kinetic reaction barrier when

the system follows Hammond’s postulate. Therefore, it is

necessary to know if a relationship between the thermodynamic

reaction data and the bottleneck of the catalytic reaction exists.

Figure 7. Proposed computational workflow for screening for

alternative rSAM enzyme substrates and alternative mutants based

on the rapid assessment of radical stabilization energies.

We also propose that this screening workflow is transferable

to a wide variety of enzyme engineering applications if the

above-mentioned requirement is satisfied and the chemical and

structural change between the substrate and the high energy

intermediate is small (as given for a wide range of hydrogen or

proton transfer and other reactions). The latter is necessary to

ensure that MD sampling of the enzyme-substrate complex is

sufficient to be able to calculate the thermodynamic reaction

properties as described. Otherwise, it would be necessary to

perform the MD sampling with the substrate and the

intermediate, which would double the computational effort.

When transferring the workflow to other examples only the

central RSE assessment needs to be replaced by the

corresponding thermodynamic reaction profile of the target

reaction. This should make this approach easily adaptable to

other protein and catalyst engineering applications.

EXPERIMENTAL SECTION

Additional information and more detailed methods are provided in

the Supporting Information. Simulation input files, important

output files, analysis and scripts used for analysis are also provided

on on the figshare repository (DOI:

https://doi.org/10.6084/m9.figshare.c.4290332.v1). Computational

workflows developed for the calculation of vRSE values from MD

data on GitHub (https://github.com/ChrisSuess/RSE-Calc).

Molecular Dynamics Simulations

All molecular-dynamics simulations were performed using the

GPU implementation45-47 of the Amber1648 molecular dynamics

package. The force field parameters for SAM are based on

electrostatic reparametrized force field parameters from Saez and

Vöhringer-Martinez49 as described and tested previously.50 The

parameters for the 4Fe4S cluster are based on a recent

parametrisation of biological relevant iron-sulfur clusters by

Carvalho and Swart.51 These parameters showed good structural

identity of the clusters as shown in the MD analysis in Section 1.4

of the Supporting Information. The interaction of SAM and the

cluster was treated by electrostatic interactions only. No restraints

have been applied to the cluster, the substrate, or SAM apart from

the simulation equilibration.

All simulations have been performed in explicit solvent, using

the SPC/E52 water model. The simulations were conducted at a

temperature of 300 K using periodic boundary conditions.

Thorough restrained equilibration as described in Section 1.2 of the

Supporting Information was followed by multiple simulations in

between 100 and over 1 µs.

DFT and RSE calculations

All accurate RSE calculations for alternative substrates were

performed analogue to our previous publication on QueE21 and as

outlined for reactions in radical SAM enzymes by Hioe and Zipse.40

Their calculations represent a formal hydrogen abstraction between

closed shell precursors (e.g. the substrate CPH4) and a reference

radical, such as CH3• for carbon centered radicals and NH2• for

nitrogen centered radicals, as given in Equations 1 and 2.

Based on these processes, the resulting reaction enthalpies can

be calculated as defined in Equation 3:

These energy values are also referred to as radical stabilization

energies (RSEs), which technically represent the relative stability

of the radicals against a given reference, and can be referenced

against accurate bond dissociation energies (BDEs) through

accurate experimental values of the reference systems. The

calculated RSE values were corrected with unscaled zero-point

energies on the level of their geometry optimization. RSE energies

were then calculated applying thermal corrections to enthalpies at

298.15 K at the level of their geometry optimization. All stationary

points have been characterized by frequency calculations.

For the calculation of the vertical radical stabilization energy

(vRSE) values snapshots from the trajectories of the dynamic

simulations were analyzed with single point energy calculations, at

both a semi-empirical (SE) and density functional level of theory

(DFT). All semi-empirical calculations reported were carried out

using MOPAC53 a semi-empirical quantum chemistry package

based on Dewar and Thiel’s NDDO approximation54 with the PM6-

D3 method which uses Grimme’s D3 dispersion corrections for

correlation.44 All DFT calculations use the quantum chemistry

package Q-Chem55 at an M062X/6-31+G(d) level of theory.

Grimme’s D3 dispersion corrections are applied where: s6 = 1.0,

sr,6 = 1.619 and s8 = 0.0.

Docking calculations

The prepared set of substrates was docked into the receptor QueE

using a combined standard precision (SP) extra precision (XP)

protocol with Glide56-57 as described in detail in Section 3.1 of the

Supporting Information. Following an exhaustive sampling search

to predict orientation, conformation and binding position of a

structure inside the rigid receptor pocket by Glide SP, the best

conformers of each substrate were selected and docked with Glide

XP to retain more accurate results. The OPLS3 force field58 was

used for the docking calculations and no constraints were applied.

ASSOCIATED CONTENT

Detailed description of the simulation, DFT, and docking setups,

complete results of the docking and vRSE calculations, further

graphical analysis, Cartesian coordinates, and force field

parameters used are given in the Supporting Information. This

includes references to primary literature essential for this study.59-80

This material is available free of charge via the Internet at

http://pubs.acs.org.

AUTHOR INFORMATION

Corresponding Author

* [email protected]

Author Contributions

The manuscript was written through contributions of all authors.

All authors have given approval to the final version of the

manuscript.

Funding Sources

EU FP7 Marie Curie Actions - People, Co-funding of Regional,

National and International Programmes (COFUND) under Grant

Agreement no PCOFUND-GA-2012-600181. European

Cooperation in Science and Technology (COST) network

CM1201.

ACKNOWLEDGMENT

AKC and CMJ would like to acknowledge support from the

European Cooperation in Science and Technology (COST)

network CM1201. CMJ acknowledges funding through the

Nottingham Advanced Research Fellowship and EU FP7 Marie

Curie Actions - People, Co-funding of Regional, National and

International Programmes (COFUND) under Grant Agreement no

PCOFUND-GA-2012-600181. We also gratefully acknowledge

support and access to the University of Nottingham High

Performance Computing Facility.

ABBREVIATIONS

BDEs, bond dissociation energies; DFT, density functional theory;

MD, molecular dynamics; pdb, protein database; QM/MM,

quantum mechanics/molecular mechanics; QueE, 7-carboxy-7-

deazaguanine synthase ; RMSD, root mean square deviation; RSEs,

radical stabilisation energies; SAM, S-adenosylmethione.

REFERENCES

(1) Broderick, J. B.; Duffus, B. R.; Duschene, K. S.; Shepard, E. M.,

Radical S-adenosylmethionine enzymes. Chem. Rev. 2014, 114, 4229-

4317.

(2) Dowling, D. P.; Vey, J. L.; Croft, A. K.; Drennan, C. L.,

Structural diversity in the AdoMet radical enzyme superfamily.

Biochim. Biophys. Acta Proteins Proteom. 2012, 1824 (11), 1178-

1195.

(3) Jäger, C. M.; Croft, A. K., Anaerobic Radical Enzymes for

Biotechnology. ChemBioEng Reviews 2018, 5 (3), 143-162.

(4) Magnusson, O. T.; Toyama, H.; Saeki, M.; Schwarzenbacher, R.;

Klinman, J. P., The Structure of a Biosynthetic Intermediate of

Pyrroloquinoline Quinone (PQQ) and Elucidation of the Final Step of

PQQ Biosynthesis. J. Am. Chem. Soc. 2004, 126 (17), 5342-5343.

(5) Barr, I.; Latham, J. A.; Iavarone, A. T.; Chantarojsiri, T.; Hwang,

J. D.; Klinman, J. P., Demonstration That the Radical S-

Adenosylmethionine (SAM) Enzyme PqqE Catalyzes de Novo

Carbon-Carbon Cross-linking within a Peptide Substrate PqqA in the

Presence of the Peptide Chaperone PqqD. J. Biol. Chem. 2016, 291

(17), 8877-8884.

(6) Cooper, L. E.; Fedoseyenko, D.; Abdelwahed, S. H.; Kim, S.-H.;

Dairi, T.; Begley, T. P., In Vitro Reconstitution of the Radical S-

Adenosylmethionine Enzyme MqnC Involved in the Biosynthesis of

Futalosine-Derived Menaquinone. Biochemistry 2013, 52 (27), 4592-

4594.

(7) Bruender, N. A.; Bandarian, V., The radical S-adenosyl-l-

methionine enzyme MftC catalyzes an oxidative decarboxylation of the

C-terminus of the MftA peptide. Biochemistry 2016, 55 (20), 2813-

2816.

(8) Lepore, B. W.; Ruzicka, F. J.; Frey, P. A.; Ringe, D., The x-ray

crystal structure of lysine-2,3-aminomutase from Clostridium

subterminale. Proc. Natl. Acad. Sci. U.S.A. 2005, 102 (39), 13819-

13824.

(9) Ruzicka, F. J.; Frey, P. A., Glutamate 2,3-aminomutase: a new

member of the radical SAM superfamily of enzymes. Biochim.

Biophys. Acta 2007, 1774 (2), 286-296.

(10) Lanz, N. D.; Lee, K.-H.; Horstmann, A. K.; Pandelia, M.-E.;

Cicchillo, R. M.; Krebs, C.; Booker, S. J., Characterization of Lipoyl

Synthase from Mycobacterium tuberculosis. Biochemistry 2016, 55

(9), 1372-1383.

(11) Cicchillo, R. M.; Iwig, D. F.; Jones, A. D.; Nesbitt, N. M.;

Baleanu-Gogonea, C.; Souder, M. G.; Tu, L.; Booker, S. J., Lipoyl

Synthase Requires Two Equivalents of S-Adenosyl-l-methionine To

Synthesize One Equivalent of Lipoic Acid. Biochemistry 2004, 43 (21),

6378-6386.

(12) Berkovitch, F.; Nicolet, Y.; Wan, J. T.; Jarrett, J. T.; Drennan,

C. L., Crystal Structure of Biotin Synthase, an S-Adenosylmethionine-

Dependent Radical Enzyme. Science 2004, 303 (5654), 76-79.

(13) Fugate, C. J.; Jarrett, J. T., Biotin synthase: Insights into radical-

mediated carbon–sulfur bond formation. Biochim. Biophys. Acta

Proteins Proteom. 2012, 1824 (11), 1213-1222.

(14) Boal, A. K.; Grove, T. L.; McLaughlin, M. I.; Yennawar, N. H.;

Booker, S. J.; Rosenzweig, A. C., Structural Basis for Methyl Transfer

by a Radical SAM Enzyme. Science 2011, 332 (6033), 1089-1092.

(15) Layer, G.; Moser, J.; Heinz, D. W.; Jahn, D.; Schubert, W.-D.,

Crystal structure of coproporphyrinogen III oxidase reveals cofactor

geometry of Radical SAM enzymes. EMBO J. 2003, 22 (23), 6214-

6224.

(16) Hänzelmann, P.; Schindelin, H., Crystal structure of the S-

adenosylmethionine-dependent enzyme MoaA and its implications for

molybdenum cofactor deficiency in humans. Proc. Natl. Acad. Sci.

U.S.A. 2004, 101 (35), 12870-12875.

(17) Chatterjee, A.; Li, Y.; Zhang, Y.; Grove, T. L.; Lee, M.; Krebs,

C.; Booker, S. J.; Begley, T. P.; Ealick, S. E., Reconstitution of ThiC

in thiamine pyrimidine biosynthesis expands the radical SAM

superfamily. Nat. Chem. Biol. 2008, 4 (12), 758-765.

(18) Sicoli, G.; Mouesca, J.-M.; Zeppieri, L.; Amara, P.; Martin, L.;

Barra, A. L.; Fontecilla-Camps, J. C.; Gambarelli, S.; Nicolet, Y., Fine-

tuning of a radical-based reaction by radical S-adenosyl-L-methionine

tryptophan lyase. Science 2016, 351 (6279), 1320-1323.

(19) Yu, Y.; Duan, L.; Zhang, Q.; Liao, R.; Ding, Y.; Pan, H.;

Wendt-Pienkowski, E.; Tang, G.; Shen, B.; Liu, W., Nosiheptide

Biosynthesis Featuring a Unique Indole Side Ring Formation on the

Characteristic Thiopeptide Framework. ACS Chem. Biol. 2009, 4 (10),

855-864.

(20) Dowling, D. P.; Bruender, N. A.; Young, A. P.; McCarty, R.

M.; Bandarian, V.; Drennan, C. L., Radical SAM enzyme QueE defines

a new minimal core fold and metal-dependent mechanism. Nat. Chem.

Biol. 2014, 10 (2), 106-112.

(21) Jäger, C. M.; Croft, A. K., Radical Reaction Control in the

AdoMet Radical Enzyme CDG Synthase (QueE): Consolidate,

Destabilize, Accelerate. Chem. Eur. J. 2017, 23 (4), 953-962.

(22) Bandarian, V.; Drennan, C. L., Radical-mediated ring

contraction in the biosynthesis of 7-deazapurines. Curr. Opin. Struct.

Biol. 2015, 35, 116-124.

(23) Zhu, W.; Liu, Y., Ring Contraction Catalyzed by the Metal-

Dependent Radical SAM Enzyme: 7-Carboxy-7-deazaguanine

Synthase from B. multivorans. Theoretical Insights into the Reaction

Mechanism and the Influence of Metal Ions. ACS Catal. 2015, 5 (7),

3953-3965.

(24) Griller, D.; Ingold, K. U., Free-radical clocks. Acc. Chem. Res.

1980, 13 (9), 317-323.

(25) Newcomb, M.; Esala, R.; Chandrasena, P.; Lansakara-P, D. S.

P.; Kim, H.-Y.; Lippard, S. J.; Beauvais, L. G.; Murray, L. J.; Izzo, V.;

Hollenberg, P. F.; Coon, M. J., Desaturase Reactions Complicate the

Use of Norcarane as a Mechanistic Probe. Unraveling the Mixture of

Twenty-Plus Products Formed in Enzyme-Catalyzed Oxidations of

Norcarane. J. Org. Chem. 2007, 72 (4), 1121-1127.

(26) Valentine, A. M.; Le Tadic-Biadatti, M. H.; Toy, P. H.;

Newcomb, M.; Lippard, S. J., Oxidation of ultrafast radical clock

substrate probes by the soluble methane monooxygenase from

Methylococcus capsulatus (Bath). J. Biol. Chem. 1999, 274, 10771-

10776.

(27) Newcomb, M.; Le Tadic-Biadatti, M. H.; Chestney, D. L.;

Roberts, E. S.; Hollenberg, P. F., A nonsynchronous concerted

mechanism for cytochrome P-450 catalyzed hydroxylation. J. Am.

Chem. Soc. 1995, 117, 12085-12091.

(28) Kemball, M. L.; Walton, J. C.; Ingold, K. U., The

conformations of cycloalkylmethyl radicals and barriers to internal

rotation. J. Chem. Soc., Perkin Trans. 1982, 2, 1017-1023.

(29) Smith, D. M.; Nicolaides, A.; Golding, B. T.; Radom, L., Ring

opening of the cyclopropylcarbinyl radical and its N- and O- substituted

analogues: A theoretical examination of very fast unimolecular

reactions. J. Am. Chem. Soc. 1998, 120 (39), 10223-10233.

(30) Datta, A.; Hrovat, D. A.; Borden, W. T., Calculations Predict

Rapid Tunneling by Carbon from the Vibrational Ground State in the

Ring Opening of Cyclopropylcarbinyl Radical at Cryogenic

Temperatures. J. Am. Chem. Soc. 2008, 130 (21), 6684-6685.

(31) Jäger, C. M.; Hennemann, M.; Clark, T., The Effect of a

Complexed Lithium Cation on a Norcarane-Based Radical Clock.

Chem. Eur. J. 2009, 15 (10), 2425-2433.

(32) Jäger, C. M.; Hennemann, M.; Mieszała, A.; Clark, T., An ab

initio and Density Functional Theory Study of Radical-Clock

Reactions. J. Org. Chem. 2008, 73 (4), 1536-1545.

(33) Hioe, J.; Savasci, G.; Brand, H.; Zipse, H., The Stability of Cα

Peptide Radicals: Why Glycyl Radical Enzymes? Chem. Eur. J. 2011,

17 (13), 3781-3789.

(34) Hioe, J.; Zipse, H., Radicals in enzymatic catalysis - A

thermodynamic perspective. Faraday Discuss. 2010, 145, 301-313.

(35) Hioe, J.; Zipse, H., Radical stability and its role in synthesis and

catalysis. Org. Biomol. Chem. 2010, 8 (16), 3609-3617.

(36) Wood, G. P. F.; Gordon, M. S.; Radom, L.; Smith, D. M.,

Nature of Glycine and Its α-Carbon Radical in Aqueous Solution: A

Theoretical Investigation. J. Chem. Theory Comput. 2008, 4 (10),

1788-1794.

(37) Croft, A. K.; Easton, C. J.; Kociuba, K.; Radom, L., Strategic

use of amino acid N-substituents to limit α-carbon- centered radical

formation and consequent loss of stereochemical integrity.

Tetrahedron Asymmetry 2003, 14 (19), 2919-2926.

(38) Croft, A. K.; Easton, C. J.; Radom, L., Design of radical-

resistant amino acid residues: A combined theoretical and experimental

investigation. J. Am. Chem. Soc. 2003, 125 (14), 4119-4124.

(39) Wood, G. P. F.; Moran, D.; Jacob, R.; Radom, L., Bond

Dissociation Energies and Radical Stabilization Energies Associated

with Model Peptide-Backbone Radicals. J. Phys. Chem. A 2005, 109,

6318-6325.

(40) Hioe, J.; Zipse, H., Hydrogen Transfer in SAM-Mediated

Enzymatic Radical Reactions. Chem. Eur. J. 2012, 18 (51), 16463-

16472.

(41) McCarty R.M.; Krebs C.; and Bandarian V., Spectroscopic,

steady-state kinetic, and mechanistic characterization of the radical

SAM enzyme QueE, which catalyzes a complex cyclization reaction in

the biosynthesis of 7-Deazapurines. Biochemistry 2013, 52 (1), 188–

198.

(42) Mardirossian, N.; Head-Gordon, M., How Accurate Are the

Minnesota Density Functionals for Noncovalent Interactions,

Isomerization Energies, Thermochemistry, and Barrier Heights

Involving Molecules Composed of Main-Group Elements? J. Chem.

Theory Comput. 2016, 12 (9), 4303-4325.

(43) Peverati, R.; Truhlar, D. G., M11-L: A Local Density

Functional That Provides Improved Accuracy for Electronic Structure

Calculations in Chemistry and Physics. J. Phys. Chem. Lett. 2012, 3

(1), 117-124.

(44) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H.; S., G.; J., A.; T.,

S.; C., M.-L., A consistent and accurate ab initio parametrization of

density functional dispersion correction (DFT-D) for the 94 elements

H-Pu. J. Chem. Phys. 2010, 132 (15), 154104.

(45) Salomon-Ferrer, R.; Götz, A. W.; Poole, D.; Le Grand, S.;

Walker, R. C., Routine Microsecond Molecular Dynamics Simulations

with AMBER on GPUs. 2. Explicit Solvent Particle Mesh Ewald. J.

Chem. Theory Comput. 2013, 9 (9), 3878-3888.

(46) Le Grand, S.; Götz, A. W.; Walker, R. C., SPFP: Speed without

compromise—A mixed precision model for GPU accelerated

molecular dynamics simulations. Comput. Phys. Commun. 2013, 184

(2), 374-380.

(47) Götz, A. W.; Williamson, M. J.; Xu, D.; Poole, D.; Le Grand,

S.; Walker, R. C., Routine Microsecond Molecular Dynamics

Simulations with AMBER on GPUs. 1. Generalized Born. J. Chem.

Theory Comput. 2012, 8 (5), 1542-1555.

(48) Case, D. A.; Botello-Smith, R. M. B., W.; Cerutti, D. S.;

Cheatham, III, T.E.; Darden, T.A.; Duke, R.E.; Giese, T.J.; Gohlke, H.;

Goetz, A.W.; Homeyer, N., et al. University of California, San

Francisco, 2016.

(49) Saez, D. A.; Vöhringer-Martinez, E., A consistent S-

Adenosylmethionine force field improved by dynamic Hirshfeld-I

atomic charges for biomolecular simulation. J. Comput. Aided Mol.

Des. 2015, 29 (10), 951-961.

(50) Bame, J.; Hoeck, C.; Carrington, M. J.; Butts, C. P.; Jäger, C.

M.; Croft, A. K., Improved NOE fitting for flexible molecules based

on molecular mechanics data – a case study with S-

adenosylmethionine. Phys. Chem. Chem. Phys 2018, 20 (11), 7523-

7531.

(51) Carvalho, A. T. P.; Swart, M., Electronic Structure

Investigation and Parametrization of Biologically Relevant Iron–Sulfur

Clusters. J. Chem. Inf. Model. 2014, 54 (2), 613-620.

(52) Berendsen, H. J. C.; Grigera, J. R.; Straatsma, T. P., The

missing term in effective pair potentials. J. Phys. Chem. 1987, 91 (24),

6269-6271.

(53) Stewart, J. J. P., MOPAC2009. Stewart Computational

Chemistry: Colorado Springs, 2008.

(54) Stewart, J. J. P., Optimization of parameters for semiempirical

methods VI: more modifications to the NDDO approximations and re-

optimization of parameters. J. Mol. Model. 2013, 19 (1), 1-32.

(55) Shao, Y.; Gan, Z.; Epifanovsky, E.; Gilbert, A. T. B.; Wormit,

M.; Kussmann, J.; Lange, A. W.; Behn, A.; Deng, J.; Feng, X.; et al.

Advances in molecular quantum chemistry contained in the Q-Chem 4

program package. Mol. Phys. 2015, 113 (2), 184-215.

(56) 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.

(57) 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.

(58) Harder, E.; Damm, W.; Maple, J.; Wu, C.; Reboul, M.; Xiang,

J. Y.; Wang, L.; Lupyan, D.; Dahlgren, M. K.; Knight, J. L.; et al.

OPLS3: A Force Field Providing Broad Coverage of Drug-like Small

Molecules and Proteins. J. Chem. Theory Comput. 2016, 12 (1), 281-

296.

(59) Gordon, J. C.; Myers, J. B.; Folta, T.; Shoja, V.; Heath, L. S.;

Onufriev, A., H++: a server for estimating pKas and adding missing

hydrogens to macromolecules. Nucleic Acids Res. 2005, 33 (Web

Server issue), W368-W371.

(60) Bayly, C. I.; Cieplak, P.; Cornell, W. D.; Kollman, P. A., A

well-behaved electrostatic potential based method using charge

restraints for deriving atomic charges – The RESP model. J. Phys.

Chem. 1993, 97 (40), 10269-10280.

(61) Goldman, P. J.; Grove, T. L.; Booker, S. J.; Drennan, C. L., X-

ray analysis of butirosin biosynthetic enzyme BtrN redefines structural

motifs for AdoMet radical chemistry. Proc. Natl. Acad. Sci. 2013, 110

(40), 15949-15954.

(62) Umitsu, M.; Nishimasu, H.; Noma, A.; Suzuki, T.; Ishitani, R.;

Nureki, O., Structural basis of AdoMet-dependent

https://pubs.acs.org/author/Krebs%2C+Carsten

https://pubs.acs.org/author/Bandarian%2C+Vahe

aminocarboxypropyl transfer reaction catalyzed by tRNA-wybutosine

synthesizing enzyme, TYW2. Proc. Natl. Acad. Sci. 2009, 106 (37),

15616-15621.

(63) Becke, A. D., Density‐functional thermochemistry. III. The role

of exact exchange. J. Chem. Phys. 1993, 98, 5648.

(64) Lee, C.; Yang, W.; Parr, R. G., Development of the Colle-

Salvetti correlation-energy formula into a functional of the electron

density. Phys. Rev. B: Condens. Matter Mater. Phys. 1988, 37, 785.

(65) Stephens, P. J.; Devlin, F. J.; Chabalowski, C. F.; Frisch, M. J., Ab Initio Calculation of Vibrational Absorption and Circular

Dichroism Spectra Using Density Functional Force Fields. J. Phys.

Chem. 1994, 98, 11623-11627.

(66) Ditchfield, R.; Hehre, W. J.; Pople, J. A., Self‐Consistent

Molecular‐Orbital Methods. IX. An Extended Gaussian‐Type Basis for

Molecular‐Orbital Studies of Organic Molecules. J. Chem. Phys. 1971,

54 (2), 724-728.

(67) Dunning, T. H., Gaussian basis sets for use in correlated

molecular calculations. I. The atoms boron through neon and hydrogen.

J. Chem. Phys. 1989, 90 (2), 1007-1023.

(68) Clark, T.; Chandrasekhar, J.; Spitznagel, G. W.; Schleyer, P. V.

R., Efficient diffuse function-augmented basis sets for anion

calculations. III. The 3-21+G basis set for first-row elements, Li–F. J.

Comput. Chem. 1983, 4 (3), 294-301.

(69) Tomasi, J.; Mennucci, B.; Cammi, R., Quantum mechanical

continuum solvation models. Chem. Rev. 2005, 105 (8), 2999-3093.

(70) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.;

Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci,

B.; Petersson, G. A.; et al. Gaussian 09. 2009.

(71) Li, P.; Roberts, B. P.; Chakravorty, D. K.; Merz, K. M., Rational

Design of Particle Mesh Ewald Compatible Lennard-Jones Parameters

for +2 Metal Cations in Explicit Solvent. J. Chem. Theory Comput.

2013, 9 (6), 2733-2748.

(72) Horinek, D.; Herz, A.; Vrbka, L.; Sedlmeier, F.; Mamatkulov,

S. I.; Netz, R. R., Specific ion adsorption at the air/water interface: The

role of hydrophobic solvation. Chem. Phys. Lett. 2009, 479 (4), 173-

183.

(73) Wang, J.; Wang, W.; Kollman, P. A.; Case, D. A., Automatic

atom type and bond type perception in molecular mechanical

calculations. J. Mol. Graph. Model. 2006, 25 (2), 247-260.

(74) Darden, T.; York, D.; Pedersen, L., Particle mesh Ewald: An

N⋅log(N) method for Ewald sums in large systems. J. Chem. Phys.

1993, 98 (12), 10089-10092.

(75) Boese, A. D.; Martin, J. M. L., Development of density

functionals for thermochemical kinetics. J. Chem. Phys. 2004, 121 (8),

3405-3416.

(76) Zhao, Y.; Truhlar, D. G., The M06 suite of density functionals

for main group thermochemistry, thermochemical kinetics,

noncovalent interactions, excited states, and transition elements: two

new functionals and systematic testing of four M06-class functionals

and 12 other functionals. Theor. Chem. Acc. 2008, 120, 215-241.

(77) Baboul, A. G.; Curtiss, L. A.; Redfern, P. C.; Raghavachari, K.,

Gaussian-3 theory using density functional geometries and zero-point

energies. The Journal of Chemical Physics 1999, 110 (16), 7650-7657.

(78) Meastro, S., LLC: New York, 2016.

(79) Madhavi Sastry, G.; 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.

(80) Olsson, M. H. M.; Søndergaard, C. R.; Rostkowski, M.; Jensen,

J. H., PROPKA3: Consistent Treatment of Internal and Surface

Residues in Empirical pKa Predictions. J. Chem. Theory Comput.

2011, 7 (2), 525-537.

Table of contents graphic:

download fileview on ChemRxivvRSE_QueE_revised-05042019_chemRxiv.pdf (1.60 MiB)



1

Radical Stabilization Energies for Enzyme

Engineering – Tackling the Substrate Scope of the

Radical Enzyme QueE

Christian J. Suess, Floriane L. Martins, Anna K. Croft, Christof M. Jäger*

1. MD simulations ...................................................................................................................... 6

1.1. MD simulation setup and parameters ............................................................................. 6

1.2. MD equilibration and production setup .......................................................................... 8

1.3. MD simulation parameter files ..................................................................................... 10

1.4. MD analysis results: ...................................................................................................... 21

1.5. Comparison of MD geometries to experimentally obtained crystal structure 4NJI ..... 53

2. Quantum chemical calculations ........................................................................................... 57

2.1 High level RSE and rearrangement barrier calculations for alternative substrates ....... 57

2.2 vRSE calculations .......................................................................................................... 57

2.3. RSE calculations of docked alternative structures ........................................................ 62

3. Alternative substrate docking protocol ................................................................................ 65

3.1. Docking protocol .......................................................................................................... 65

3.2. Docking results ............................................................................................................. 66

4. Combined MD equilibration and RSE assessment workflow ............................................. 69

2

Figure S1. Atomic coordinates, atomic charges and Amber force field atom types for SAM

molecules. HF parameterisation, net charge 0. ................................................................ 10

Figure S2. Atomic coordinates, atomic charges and GAFF force field atom types for the

natural substrate CPH4. .................................................................................................... 11


substrate analogue CP6. ................................................................................................... 11

Figure S4. Atomic coordinates, atomic charges and Amber force field atom types for the

substrate intermediate. ..................................................................................................... 12

Figure S5. Force field modification file for SAM. ................................................................. 13

Figure S6. Force field modification file for the iron-sulfur cluster Fe4S4. ............................. 14

Figure S7. Force field modification file for the natural substrate CPH6. ............................... 14

Figure S8. Force field modification file for the substrate intermediate CP6. ......................... 15

Figure S9. Force field modification file for the substrate intermediate. ................................. 15

Figure S10. Force field library file for adapted cysteine residues connected to the 4Fe4S

cluster. Charges and connectivity have been adapted. ..................................................... 17

Figure S11. Force field library file for the 4Fe4S cluster based on the parameterisation by

Carvalho and Swart23 with adapted charges for the unique Fe atom. .............................. 20

Figure S12. Graphically summarized MD analysis for the first 100 ns of a simulation of

QueE (pdb 4NJI) with the substrate CPH6 and Mg2+ (simulation ID 1). The simulation

time presented starts after 9 ns of restrained equilibration and the equilibration and the

rmsd is calculated against the crystal structure configuration. ........................................ 22










QueE (pdb 4NJI) with the substrate CPH6 and Ca2+ (simulation ID 1). The simulation








QueE (pdb 4NJI) with the substrate CPH6 and Na+ (simulation ID 1). The simulation




QueE (pdb 4NJI) with the substrate CPH6 and Na+ (simulation ID 2). The simulation




QueE (pdb 4NJG) with the substrate analogue CP6. The simulation time presented starts

after 9 ns of restrained equilibration and the equilibration and the rmsd is calculated

against the crystal structure configuration. ...................................................................... 29

3





















Figure S25. Comparison of crystal structure, MD, and DFT model system geometries for the

active site of QueE containing the substrate CPH4, Mg2+, and water molecules.

Experimentally observed electron density in the active site of a) chain A and b) chain B

of QueE (pdb code 4NJI); c) superimposed structures of crystal structure and

equilibrated MD geometries; d) distance analysis of Mg2+ coordination from crystal

structure 4NJI and DFT models as presented in reference 26. ........................................ 55

Figure S26. Correlation of vRSE values at the PM6-D3 and the M06-2X/6-31+G(d) level of

theory based on 5000 snapshots taken from 100 ns MD of bound substrate to QueE with

Mg2+. Energies given in kJ mol-1. .................................................................................... 60

Figure S27. RSE value distribution at the M06-2X/6-31+G(d) level of theory derived on

5000 snapshots taken from 100 ns MD of bound substrate to QueE with Mg2+. ............ 61

Figure S28. Overview of the MD simulation and RSE evaluation workflow available on

GitHub. ............................................................................................................................ 70

Figure S29. Screenshot of the introductory overview and first steps within the Jupiter

Notebook for MD and RSE calculations. ........................................................................ 71

Figure S30.Screenshot of the graphical MD minimization analysis within the Jupiter


Figure S31. Screenshot of the ligand selection and RSE calculation routine within the Jupiter


Table S1. Hydrogen bonding summary for first 55 ns of a simulation of QueE (pdb 4NJI)

with the substrate CPH6 (MOL) and Ca2+ (simulation ID 2, monomer B). Substrate fully

bound. Contacts between Ca2+ and the substrate and protein. ......................................... 35

Table S2. Hydrogen bonding summary for first 55 ns a simulation of QueE (pdb 4NJI) with

the substrate CPH6 (MOL) and Ca2+ (simulation ID 2, monomer B). Substrate fully

bound. Contacts between substrate and the protein. ........................................................ 35



bound. Key hydrogen bonds of central residues (>10% fraction for amino acids). ........ 36

4



bound. Hydrogen bonding of key residues with solvent (sum of >10% fraction). .......... 36


with the substrate CPH6 (MOL) and Mg2+ (simulation ID 4, monomer A). Substrate

fully bound. Contacts between Mg2+ and the substrate, protein. ..................................... 37


with the substrate CPH6 (MOL) and Mg2+ (simulation ID 4, monomer A). Hydrogen

bonds between Substrate and protein. .............................................................................. 38


with the substrate CPH6 (MOL) and Mg2+ (simulation ID 4, monomer A). Key

hydrogen bonds of central residues (>10% fraction for amino acids). ............................ 38


with the substrate CPH6 (MOL) and Mg2+ (simulation ID 4, monomer A). Hydrogen

bonding of key residues with solvent (sum of >10% fraction). ....................................... 39


with the substrate CPH6 (MOL) and Mg2+ (simulation ID 4, monomer B). Contacts

between ion and substrate, amino acids. .......................................................................... 40


with the substrate CPH6 (MOL) and Mg2+ (simulation ID 4, monomer B). Hydrogen

bonds between Substrate and protein. .............................................................................. 40


with the substrate CPH6 (MOL) and Mg2+ (simulation ID 4, monomer B). Key hydrogen

bonds of central residues (>10% fraction for amino acids). ............................................ 41


with the substrate CPH6 (MOL) and Mg2+ (simulation ID 4, monomer B). Hydrogen

bonding of key residues with solvent (sum of >10% fraction). ....................................... 41

Table S13. Hydrogen bonding summary for first 100 ns of a simulation of QueE (pdb 4NJG)

with the substrate analogue CP6 (2KA) and Mg2+ (simulation ID 1, monomer A).

Contacts between ion and substrate, amino acids. ........................................................... 42



Hydrogen bonds between Substrate and protein (> 1% occupancy). .............................. 43


with the substrate analogue CP6 (2KA) and Mg2+ (simulation ID 1, monomer A). Key




Hydrogen bonding of key residues with solvent (sum of >10% fraction). ...................... 44


with the substrate analogue CPH6 (MOL) and Na+ (simulation ID 2, monomer A).

Contacts between ion and amino acids. ........................................................................... 45



Contacts between ion and amino acids. Hydrogen bonds between Substrate and protein

(> 1% occupancy). ........................................................................................................... 45



Contacts between ion and amino acids. Key hydrogen bonds of central residues (>10%

fraction for amino acids). ................................................................................................. 46

5



Contacts between ion and amino acids. Hydrogen bonding of key residues with solvent

(sum of >10% fraction). ................................................................................................... 47


with the substrate analogue CPH6 (MOL) and Na+ (simulation ID 2, monomer B).

Contacts between ion and amino acids. Contacts between Ion and amino acids. ........... 48



Hydrogen bonds between Substrate and protein (> 1% occupancy). .............................. 48


with the substrate analogue CPH6 (MOL) and Na+ (simulation ID 2, monomer B). Key




Hydrogen bonding of key residues with solvent (sum of >10% fraction). ...................... 50

Table S25. Summary of residence time analysis of the substrate in correct positioning and

conformation bound insight the binding pocket of QueE. ............................................... 51

Table S26. Comparison of calculated average RSE values and relative vRSE shifts for

simulations of CPH6 in vacuum and bound to QueE with Mg2+ for DFT and

semiempirical calculations. .............................................................................................. 58


simulations of the product radical (5) in vacuum and bound to QueE with Mg2+ for DFT

and semiempirical calculations. ....................................................................................... 58

Table S28. Comparison of the accuracy of average RSE calculation when selecting fewer

‘selective’ snapshots near the peak of a Gaussian fit to a larger data set at lower

computational level (PM3). Pchg, depicts calculations including the point charge

environment of the enzyme. ............................................................................................. 59

Table S29. RSE values on docked alternative substrates. ....................................................... 62

Table S30. Three best conformers of each new substrate, ranked by Emodel score, inside

QueE and QueE without Mg2+. Their binding score (XPscore) and the distance between

C6 from substrate and C5' from SAM are also presented. ............................................... 67

6

1. MD simulations

1.1. MD simulation setup and parameters

All simulation coordinates, input and important analysis files (in particular the MD interatomic

distance analysis) are also available online on the figshare repository (DOI:

https://doi.org/10.6084/m9.figshare.c.4290332.v1). Additional analysis scripts (perl) and

cpptraj input files are also presented there.

All molecular-dynamics simulations were performed using the GPU implementation1-3 of the

Amber164 molecular dynamics package. The Amber force field FF12SB has been applied for

the enzyme and the SPC/E5 water model has been used for explicit solvation. The protonation

state of titratable amino acids has been determined using the software of the H++ server.6 The

force field parameters for SAM are based on electrostatic reparametrized parameters from Saez

and Vöhringer-Martinez7 as described and tested previously.8 Electrostatic point charges were

reparametrised following the restrained electrostatic potential (RESP) fitting procedure by

Kollman et al.9 and are based on multiconfigurational fitting of three different conformers. The

structures for RESP fitting were taken from the crystal structures of butirosin biosynthetic

enzyme,10 BtrN (pdb entry 4M7T), tRNA-wybutosine synthesising enzyme, TYW2 (pdb entry

3A25),11 and 7-carboxy-7-deazaguanine synthase, QueE (pdb entry 4NJI),12 and have been

picked in order to represent different bent and stretched SAM conformations. After geometry

optimisation at the B3LYP13-15/6-31+G(d)16-17 level of theory including diffuse functions18 and

applying the polarisable continuum model (PCM)19 as implicit solvation model with

Gaussian0920 two sets of charges were derived. The first set was prepared following the

standard RESP procedure at the HF/6-31G(d) level, and a second set was generated based on

PCM-B3LYP/cc-PVTZ17 calculations in implicit solvent with a dielectric constant of 4.335,

that is suitable for representing the electrostatic environment in a protein more closely.

7

The parameters for the divalent cations were taken from Li et al.,21 those for monovalent ions

to neutralise the total charge of the system from Horinek et. al.22 and the parameters for the

4Fe4S cluster are based on a recent parametrisation of biological relevant iron-sulfur clusters

by Carvalho and Swart.23 These parameters showed good structural identity of the clusters as

shown in the MD analysis. For embedding the cluster into the enzyme the sulfur charges of

connected cysteine residues have been adapted to the charges from the model compounds used

for the parametrisation and the charge of the unique iron atom have been adjusted to remain

the total charge of +1 for the reduced cluster.

The interaction of SAM and the cluster was treated by electrostatic interactions only. No

restraints have been applied to the cluster, the substrate, or SAM apart from the simulation

equilibration. Only in some rare cases the unique (non-bonded) iron of the cluster inverted

when no SAM molecule was bound. This might be due to the reason that the parametrisation

included very low angle force constants and was based on fully bound clusters without unique

Fe atoms.

The parameters of the substrate CPH4 and the substrate intermediate (2K8) have been

parametrised on the basis of the general amber force field (GAFF)24 using the same RESP

fitting procedure as described above.

Parameter files are given below and on the figshare repository for download.

8

1.2. MD equilibration and production setup

All simulations have been performed in explicit solvent, using the SPC/E5 water model.

Electrostatic long range interactions were treated with the Particle Mesh Ewald (PME)25 and

the simulations were conducted at a temperature of 300 K using periodic boundary conditions

method and a 12 Å cut-off for nonbonding interactions. The temperature in all simulations was

controlled by coupling the system with the Langevin thermostat with collision frequency set to

2 ps‐1. An integration time step of 2 fs was used and the SHAKE algorithm was employed to

constrain bonds involving hydrogen atoms during the MD simulation

Firstly, all structures have been minimized in four minimization steps of combined steepest

descent and conjugate gradient minimisation for 1000 steps each following a partial release of

positional restraints with a 50 kcal mol-1 for constant applied. Restraints have been applied for

minimizations 1-4 as follows: 1) Protein, 4Fe4S clusters, central ion, SAM, ligands, crystal

water; 2) protein, 4Fe4S clusters, central ion, SAM, ligands; 3) protein, 4Fe4S clusters, central

ion, SAM; 4) 4Fe4S clusters, central ion, SAM. This was followed by a heat up phase of the

system were the protein, SAM, the 4Fe4S clusters, and the central ions have been restrained

with a force constant of 20 kcal mol-1 for 100 ps of simulation. Finally, the system was further

equilibrated at constant pressure (NPT) molecular dynamics at one atmosphere using a

Langevin directed dynamics for pressure control with restraints of 20 kcal mol-1 on 4Fe4S

clusters, central ion, and SAM for another 900 ps, and with weaker restraints of 10 kcal mol-1

on the same residues for another 8 ns. The final production runs for the simulations were

performed for at least 100 ns and over 1500 ns in maximum. The simulations have been

repeated several times starting from the same starting coordinates, but applying different initial

velocities to the atoms.

9

All MD simulations are based on the following crystal structures:

4NJI: QueE from Burkholderia multivorans in complex with AdoMet (SAM) and 6-carboxy-

5,6,7,8-tetrahydropterin (CPH6) and Mg2+.

4NJH: QueE from Burkholderia multivorans in complex with AdoMet (SAM) and 6-carboxy-

5,6,7,8-tetrahydropterin (CPH6) and Na+.

4NJJ: QueE from Burkholderia multivorans in complex with AdoMet, 6-carboxy-5,6,7,8-

tetrahydropterin (CPH4), and Mn2+.

4NJG: QueE from Burkholderia multivorans in complex with AdoMet (SAM) and 6-

carboxypterin (CP6).

4NJK: QueE from Burkholderia multivorans in complex with AdoMet (SAM), 7-carboxy-7-

deazaguanine (CDG), and Mg2+.

10

1.3. MD simulation parameter files

1 N1 -4.6880 -2.9390 0.1850 NT 1 SAM -0.985345 2 C1 -4.0880 -2.0900 -0.8540 CT 1 SAM 0.266729 3 C2 -2.5360 -2.2310 -0.8400 C 1 SAM 0.664757 4 O1 -2.0680 -3.3300 -0.4380 O2 1 SAM -0.727188 5 O2 -1.8730 -1.2420 -1.2780 O2 1 SAM -0.727188 6 C3 -4.5880 -0.6400 -0.7380 CT 1 SAM -0.122480 7 C4 -4.4240 0.0270 0.6280 CT 1 SAM 0.272300 8 S1 -2.6810 0.3820 1.1410 SP 1 SAM 0.399335 9 C5 -2.9940 1.3770 2.6360 CT 1 SAM -0.579883 10 C6 -2.1650 1.6420 -0.0930 CT 1 SAM -0.457032 11 C7 -0.8860 2.3740 0.3190 CT 1 SAM 0.313050 12 O3 0.1000 1.4230 0.7810 OS 1 SAM -0.455480 13 C8 -0.2540 3.1250 -0.8660 CT 1 SAM 0.354359 14 O4 0.3170 4.3330 -0.3840 OH 1 SAM -0.688357 15 C9 0.8530 2.1520 -1.3260 CT 1 SAM 0.029885 16 O5 1.9540 2.8170 -1.9230 OH 1 SAM -0.654644 17 C10 1.2920 1.5150 0.0030 CT 1 SAM 0.169327 18 N2 1.8940 0.2050 -0.1050 N* 1 SAM -0.128158 19 C11 1.3120 -0.9650 -0.5710 CK 1 SAM 0.231447 20 N3 2.1270 -1.9960 -0.5290 NB 1 SAM -0.524991 21 C12 3.3080 -1.4890 -0.0090 CB 1 SAM -0.140252 22 C13 4.5580 -2.0770 0.2820 CA 1 SAM 0.795683 23 N4 4.8100 -3.3950 0.0940 N2 1 SAM -0.899841 24 N5 5.5300 -1.2960 0.8010 NC 1 SAM -0.808920 25 C14 5.2660 0.0030 1.0210 CQ 1 SAM 0.622312 26 N6 4.1320 0.6730 0.7850 NC 1 SAM -0.797992 27 C15 3.1840 -0.1250 0.2680 CB 1 SAM 0.544320 28 H1 0.2740 -1.0060 -0.8940 H5 1 SAM 0.129732 29 H2 6.0840 0.5820 1.4430 H5 1 SAM 0.062185 30 H3 -4.1130 -3.7770 0.2640 H 1 SAM 0.362620 31 H4 -5.6230 -3.2360 -0.0920 H 1 SAM 0.362620 32 H5 -4.3780 -2.4250 -1.8670 H1 1 SAM 0.023164 33 H6 -5.6720 -0.6320 -0.9160 HC 1 SAM 0.023596 34 H7 -4.1420 -0.0330 -1.5310 HC 1 SAM 0.023596 35 H8 -4.7890 -0.6330 1.4200 H1 1 SAM -0.013132 36 H9 -4.9560 0.9810 0.6720 H1 1 SAM -0.013132 37 H10 -3.4880 0.7190 3.3540 H1 1 SAM 0.217957 38 H11 -3.6230 2.2400 2.4080 H1 1 SAM 0.217957 39 H12 -2.0290 1.6880 3.0380 H1 1 SAM 0.217957 40 H13 -2.9860 2.3510 -0.2310 H1 1 SAM 0.197714 41 H14 -2.0210 1.0440 -0.9960 H1 1 SAM 0.197714 42 H15 -1.0850 3.0820 1.1300 H1 1 SAM 0.062387 43 H16 -0.9800 3.3330 -1.6620 H1 1 SAM 0.000792 44 H17 1.0680 4.5450 -0.9710 HO 1 SAM 0.437487 45 H18 0.4410 1.3900 -1.9950 H1 1 SAM 0.071022 46 H19 1.8440 2.8270 -2.8880 HO 1 SAM 0.461202 47 H20 2.0330 2.1530 0.4940 H2 1 SAM 0.182663 48 H21 4.1690 -3.9520 -0.4540 H 1 SAM 0.404074 49 H22 5.7660 -3.7190 0.1630 H 1 SAM 0.404074

Figure S1. Atomic coordinates, atomic charges and Amber force field atom types for SAM

molecules. HF parameterisation, net charge 0.

11

1 O6B 3.1050 -0.8780 1.2550 o 1 MOL -0.793049 2 C6A 2.5280 0.1670 0.9000 c 1 MOL 0.766329 3 O6A 2.2420 1.1830 1.5540 o 1 MOL -0.793049 4 C6 2.1570 0.2470 -0.6400 c3 1 MOL 0.338581 5 N5 1.0120 1.1120 -0.9330 nh 1 MOL -0.773960 6 C4A -0.2130 0.5400 -0.5510 cc 1 MOL -0.055212 7 C7 1.8950 -1.1380 -1.2430 c3 1 MOL -0.014734 8 N8 0.6300 -1.6920 -0.7720 nh 1 MOL -0.584506 9 C8A -0.3790 -0.8320 -0.4260 cd 1 MOL 0.531070 10 N1 -1.5320 -1.4270 0.0340 nd 1 MOL -0.815539 11 C2 -2.5280 -0.6490 0.2990 cc 1 MOL 0.964451 12 N2 -3.7450 -1.1800 0.7390 nh 1 MOL -0.983559 13 N3 -2.4840 0.6930 0.1250 n 1 MOL -0.793389 14 C4 -1.3320 1.3960 -0.2800 c 1 MOL 0.683281 15 O4 -1.3790 2.6170 -0.3580 o 1 MOL -0.632230 16 H1 3.0330 0.6790 -1.1450 h1 1 MOL -0.054647 17 H2 1.1240 1.9950 -0.4400 hn 1 MOL 0.381414 18 H3 2.7030 -1.8020 -0.9310 h1 1 MOL 0.045011 19 H4 1.8690 -1.0700 -2.3410 h1 1 MOL 0.045011 20 H5 0.6780 -2.5230 -0.2010 hn 1 MOL 0.351384 21 H6 -4.1540 -0.6760 1.5170 hn 1 MOL 0.387409 22 H7 -3.6310 -2.1620 0.9560 hn 1 MOL 0.387409 23 H8 -3.3010 1.2690 0.2760 hn 1 MOL 0.412524


natural substrate CPH4.

1 N1 2.0010 -1.3570 0.0690 nc 1 MOL -0.697272 2 C1 2.8990 -0.4370 0.0180 cd 1 MOL 0.706587 3 N2 4.2450 -0.7750 -0.0130 nh 1 MOL -0.899680 4 N3 2.6260 0.9010 -0.0560 n 1 MOL -0.567909 5 C2 1.3330 1.4530 -0.0790 c 1 MOL 0.557332 6 O1 1.2060 2.6500 -0.1600 o 1 MOL -0.556187 7 N4 -1.0000 0.7950 0.0010 nb 1 MOL -0.449495 8 C3 -1.9180 -0.1510 0.0180 ca 1 MOL 0.202558 9 O2 -3.7730 1.2810 0.4370 o 1 MOL -0.762201 10 O3 -4.1280 -0.7820 -0.4120 o 1 MOL -0.762201 11 C4 -3.4470 0.1710 0.0160 c 1 MOL 0.741327 12 C5 -1.5120 -1.5210 0.0290 ca 1 MOL 0.156238 13 N5 -0.2550 -1.9150 0.0580 nb 1 MOL -0.556455 14 C6 0.6790 -0.9450 0.0320 ca 1 MOL 0.573244 15 C7 0.2860 0.4100 -0.0130 ca 1 MOL 0.096029 16 H1 4.3800 -1.7550 0.1960 hn 1 MOL 0.383628 17 H2 4.8660 -0.1690 0.5050 hn 1 MOL 0.383628 18 H3 3.3720 1.5740 -0.1760 hn 1 MOL 0.355798 19 H4 -2.2910 -2.2790 0.0170 h4 1 MOL 0.095031


substrate analogue CP6.

12

1 N1 -5.786 82.907 24.558 nc 1 MOL -0.641516 2 C1 -5.692 81.610 24.645 cd 1 MOL 0.708381 3 N2 -4.599 81.048 25.291 nh 1 MOL -0.937916 4 N3 -6.580 80.756 24.093 n 1 MOL -0.407072 5 C2 -7.778 81.146 23.413 c 1 MOL 0.388410 6 O1 -8.489 80.256 22.969 o 1 MOL -0.594902 7 N4 -9.450 83.132 21.504 n3 1 MOL -1.116382 8 C3 -9.051 83.422 22.886 c3 1 MOL 0.389468 9 O2 -9.974 83.486 25.100 o 1 MOL -0.824674 10 O3 -11.409 83.391 23.366 o 1 MOL -0.824674 11 C4 -10.287 83.403 23.900 c 1 MOL 0.821007 12 C5 -8.408 84.858 23.014 c3 1 MOL 0.112275 13 N5 -7.095 84.663 23.650 nh 1 MOL -0.667127 14 C6 -6.893 83.338 23.882 cc 1 MOL 0.446795 15 C7 -7.861 82.721 23.918 cd 1 MOL -0.121219 16 H1 -4.794 80.211 25.824 hn 1 MOL 0.385467 17 H2 -4.093 81.744 25.822 hn 1 MOL 0.385467 18 H3 -6.430 79.756 24.125 hn 1 MOL 0.311627 19 H4 -9.031 85.512 23.632 h1 1 MOL 0.028127 20 H5 -8.284 85.287 22.012 h1 1 MOL 0.028127 21 H6 -6.747 85.323 24.328 hn 1 MOL 0.360091 22 H7 -10.463 83.274 21.503 hn 1 MOL 0.385120 23 H8 -9.330 82.132 21.356 hn 1 MOL 0.385120

Figure S4. Atomic coordinates, atomic charges and Amber force field atom types for the

substrate intermediate.

frcmod file for Adomet(SAM) with parameters of Markham et al. Biochemistry 2002, Saez et al 2015, Yildirim et al 2010 SP 32.7798 2.900 CS 12.010 0.878 BOND CT-SP 345.28 1.78 same as CT-SP,Markham 2002 SP-CS 345.28 1.78 same as CT-SP,Markham 2002 CS-H1 340.00 1.090 same as CT-H1 ANGLE CT-CT-SP 96.058 96.058 Markham 2002 CT-SP-CS 289.219 100.546 Markham 2002 CT-SP-CT 289.219 100.546 Markham 2002 SP-CT-H1 33.045 104.439 Markham 2002 SP-CS-H1 33.045 104.439 Markham 2002 H1-CS-H1 35.000 109.500 Markham 2002 DIHE CT-CT-SP-CT 2 0.400 315.000 -3.000 Saez 2015 CT-CT-SP-CT 2 0.200 50.000 -2.000 Saez 2015 CT-CT-SP-CT 2 1.000 350.000 1.000 Saez 2015 CT-CT-SP-CS 2 0.300 100.000 -3.000 Saez 2015

13

CT-CT-SP-CS 2 0.100 325.000 -2.000 Saez 2015 CT-CT-SP-CS 2 0.900 300.000 1.000 Saez 2015 CT-SP-CS-H1 1 0.000 0.000 0.000 ATTN, need revision CT-SP-CT-H1 1 0.000 0.000 0.000 ATTN, need revision CS-SP-CT-H1 1 0.000 0.000 0.000 ATTN, need revision OS-CT-N*-CK 2 0.152425 0.000 -4.000 Yildirim 2010 OS-CT-N*-CK 2 -1.699430 0.000 -3.000 Yildirim 2010 OS-CT-N*-CK 2 0.504875 0.000 -2.000 Yildirim 2010 OS-CT-N*-CK 2 1.355570 0.000 -1.000 Yildirim 2010 CT-CT-N*-CK 2 0.228818 0.000 -4.000 Yildirim 2010 CT-CT-N*-CK 2 1.267980 0.000 -3.000 Yildirim 2010 CT-CT-N*-CK 2 -0.278197 0.000 -2.000 Yildirim 2010 CT-CT-N*-CK 2 1.603540 0.000 1.000 Yildirim 2010

IMPROPER CT-O2-C -O2 10.5 180.0 2.0 Using general improper torsional angle X-O2- C-O2, penalty score= 3.0) H5-N*-CK-NB 1.1 180.0 2.0 Using the default value CA-CB-CB-NB 1.1 180.0 2.0 Using the default value CA-H -N2-H 1.0 180.0 2.0 Using general improper torsional angle X- X-N2- H, penalty score= 6.0) H5-NC-CQ-NC 1.1 180.0 2.0 Using the default value CB-N*-CB-NC 1.1 180.0 2.0 Using the default value

NONBON SP 2.3890 0.0053 CT 1.9080 0.1094

Figure S5. Force field modification file for SAM.

# force field for Fe4S4 cluster MASS FC 55.845 FU 55.845 SC 32.065 BOND FC-SC 55.2 2.31 FU-SC 55.2 2.31 FC-SH 60.9 2.31 FU-FH 60.9 2.31 ANGLE SC-FC-SC 8.200 105.60 SC-FU-SC 8.200 105.60 SC-FC-SH 11.000 113.20 FC-SC-FC 8.800 71.50 FU-SC-FU 8.800 71.50 FU-SC-FC 8.800 71.50

14

FC-SH-CT 15.600 105.60 DIHE X -SC-FC-X 1 0.000 180.000 3.000 X -SC-FU-X 1 0.000 180.000 3.000 X -SC-SH-X 1 0.000 180.000 3.000 X -SH-FC-X 1 0.000 180.000 3.000 NONBON FC 1.20 0.050 # iron from heme FU 1.20 0.050 SC 2.00 0.250

Figure S6. Force field modification file for the iron-sulfur cluster Fe4S4.

remark goes here MASS BOND ANGLE nh-cc-c 67.790 115.620 same as c -ce-nh DIHE IMPROPER c3-o -c -o 10.5 180.0 2.0 corrected for

carboxylates (cmj) c3-cc-nh-hn 1.1 180.0 2.0 Using default value c -cd-cc-nh 1.1 180.0 2.0 Using default value c3-cd-nh-hn 1.1 180.0 2.0 Using default value cc-nd-cd-nh 1.1 180.0 2.0 Using default value n -nd-cc-nh 1.1 180.0 2.0 Using default value cc-hn-nh-hn 1.1 180.0 2.0 Using default value c -cc-n -hn 1.1 180.0 2.0 General improper

torsional angle (2 general atom types) cc-n -c -o 10.5 180.0 2.0 General improper

torsional angle (2 general atom types)

NONBON

Figure S7. Force field modification file for the natural substrate CPH6.

remark goes here MASS BOND

15

ANGLE DIHE IMPROPER n -nc-cd-nh 1.1 180.0 2.0 Using default value cd-hn-nh-hn 1.1 180.0 2.0 Using default value c -cd-n -hn 1.1 180.0 2.0 General improper

torsional angle (2 general atom types) ca-n -c -o 10.5 180.0 2.0 General improper

torsional angle (2 general atom types) c -ca-ca-nb 1.1 180.0 2.0 Using default value ca-o -c -o 10.5 180.0 2.0 Carboxylate adjusted

(cmj) ca-h4-ca-nb 1.1 180.0 2.0 Using default value ca-nb-ca-nc 1.1 180.0 2.0 Using default value

NONBON

Figure S8. Force field modification file for the substrate intermediate CP6.

remark goes here

MASS BOND ANGLE DIHE IMPROPER n -nc-cd-nh 1.1 180.0 2.0 Using default value cd-hn-nh-hn 1.1 180.0 2.0 Using default value c -cd-n -hn 1.1 180.0 2.0 General improper

torsional angle (2 general atom types) cd-n -c -o 10.5 180.0 2.0 General improper

torsional angle (2 general atom types) c3-o -c -o 10.5 180.0 2.0 corrected for

carboxylate improper torsion (cmj) c3-cc-nh-hn 1.1 180.0 2.0 Using default value cd-nc-cc-nh 1.1 180.0 2.0 Using default value c -c3-cd-cc 1.1 180.0 2.0 Using default value

NONBON

Figure S9. Force field modification file for the substrate intermediate.

!!index array str "CYF"

16

!entry.CYF.unit.atoms table str name str type int typex int resx int flags int seq int elmnt dbl chg "N" "N" 0 1 131073 1 7 -0.415700 "H" "H" 0 1 131073 2 1 0.271900 "CA" "CT" 0 1 131073 3 6 0.021300 "HA" "H1" 0 1 131073 4 1 0.112400 "CB" "CT" 0 1 131073 5 6 -0.123100 "HB2" "H1" 0 1 131073 6 1 0.111200 "HB3" "H1" 0 1 131073 7 1 0.111200 "SG" "SH" 0 1 131073 8 16 -0.677700 "C" "C" 0 1 131073 10 6 0.597300 "O" "O" 0 1 131073 11 8 -0.567900 !entry.CYF.unit.atomspertinfo table str pname str ptype int ptypex int pelmnt

dbl pchg "N" "N" 0 -1 0.0 "H" "H" 0 -1 0.0 "CA" "CT" 0 -1 0.0 "HA" "H1" 0 -1 0.0 "CB" "CT" 0 -1 0.0 "HB2" "H1" 0 -1 0.0 "HB3" "H1" 0 -1 0.0 "SG" "SH" 0 -1 0.0 "C" "C" 0 -1 0.0 "O" "O" 0 -1 0.0 !entry.CYF.unit.boundbox array dbl -1.000000 0.0 0.0 0.0 0.0 !entry.CYF.unit.childsequence single int 2 !entry.CYF.unit.connect array int 1 9 !entry.CYF.unit.connectivity table int atom1x int atom2x int flags 1 2 1 1 3 1 3 4 1 3 5 1 3 9 1 5 6 1 5 7 1 5 8 1 9 10 1 !entry.CYF.unit.hierarchy table str abovetype int abovex str belowtype int

belowx "U" 0 "R" 1 "R" 1 "A" 1 "R" 1 "A" 2 "R" 1 "A" 3 "R" 1 "A" 4 "R" 1 "A" 5 "R" 1 "A" 6 "R" 1 "A" 7 "R" 1 "A" 8 "R" 1 "A" 9 "R" 1 "A" 10 !entry.CYF.unit.name single str

17

"CYS" !entry.CYF.unit.positions table dbl x dbl y dbl z 3.325770 1.547909 -1.607204E-06 3.909407 0.723611 -2.739882E-06 3.970048 2.845795 -1.311163E-07 3.671663 3.400129 -0.889820 3.576965 3.653838 1.232143 2.496995 3.801075 1.241379 3.877484 3.115795 2.131197 4.309573 5.303523 1.366036 5.485541 2.705207 -4.398755E-06 6.008824 1.593175 -8.449768E-06 !entry.CYF.unit.residueconnect table int c1x int c2x int c3x int c4x int c5x

int c6x 1 9 0 0 0 0 !entry.CYF.unit.residues table str name int seq int childseq int startatomx

str restype int imagingx "CYS" 1 12 1 "p" 0 !entry.CYF.unit.residuesPdbSequenceNumber array int 0 !entry.CYF.unit.solventcap array dbl -1.000000 0.0 0.0 0.0 0.0 !entry.CYF.unit.velocities table dbl x dbl y dbl z 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

Figure S10. Force field library file for adapted cysteine residues connected to the 4Fe4S

cluster. Charges and connectivity have been adapted.

!!index array str "SF4" "a" !entry.SF4.unit.atoms table str name str type int typex int resx int flags

int seq int elmnt dbl chg "FE1" "FC" 0 1 131073 1 26 0.258150 "FE2" "FC" 0 1 131073 2 26 0.258150 "FE3" "FU" 0 1 131073 3 26 0.889350 "FE4" "FC" 0 1 131073 4 26 0.258150 "S1" "SC" 0 1 131073 5 16 -0.496625 "S2" "SC" 0 1 131073 6 16 -0.496625 "S3" "SC" 0 1 131073 7 16 -0.496625

18

"S4" "SC" 0 1 131073 8 16 -0.496625 !entry.SF4.unit.atomspertinfo table str pname str ptype int ptypex int pelmnt

dbl pchg "FE1" "FC" 0 -1 0.0 "FE2" "FC" 0 -1 0.0 "FE3" "FU" 0 -1 0.0 "FE4" "FC" 0 -1 0.0 "S1" "SC" 0 -1 0.0 "S2" "SC" 0 -1 0.0 "S3" "SC" 0 -1 0.0 "S4" "SC" 0 -1 0.0 !entry.SF4.unit.boundbox array dbl -1.000000 0.0 0.0 0.0 0.0 !entry.SF4.unit.childsequence single int 302 !entry.SF4.unit.connect array int 0 0 !entry.SF4.unit.connectivity table int atom1x int atom2x int flags 1 6 1 1 7 1 1 8 1 2 5 1 2 7 1 2 8 1 3 5 1 3 6 1 3 8 1 4 5 1 4 6 1 4 7 1 !entry.SF4.unit.hierarchy table str abovetype int abovex str belowtype int

belowx "U" 0 "R" 1 "R" 1 "A" 1 "R" 1 "A" 2 "R" 1 "A" 3 "R" 1 "A" 4 "R" 1 "A" 5 "R" 1 "A" 6 "R" 1 "A" 7 "R" 1 "A" 8 !entry.SF4.unit.name single str "default_name" !entry.SF4.unit.positions table dbl x dbl y dbl z -20.371000 83.065000 19.300000 -19.102000 82.008000 21.471000 -18.519000 84.592000 20.710000 -20.969000 83.988000 21.798000 -18.917000 83.819000 22.825000 -20.593000 85.237000 19.932000 -21.344000 81.902000 20.997000 -18.116000 82.665000 19.530000 !entry.SF4.unit.residueconnect table int c1x int c2x int c3x int c4x int c5x

int c6x

19

0 0 0 0 0 0 !entry.SF4.unit.residues table str name int seq int childseq int startatomx

str restype int imagingx "SF4" 301 9 1 "?" 0 !entry.SF4.unit.residuesPdbSequenceNumber array int 1 !entry.SF4.unit.solventcap array dbl -1.000000 0.0 0.0 0.0 0.0 !entry.SF4.unit.velocities table dbl x dbl y dbl z 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 !entry.a.unit.atoms table str name str type int typex int resx int flags int

seq int elmnt dbl chg "FE1" "FC" 0 1 131075 1 26 0.258150 "FE2" "FC" 0 1 131075 2 26 0.258150 "FE3" "FU" 0 1 131075 3 26 0.258150 "FE4" "FC" 0 1 131075 4 26 0.258150 "S1" "SC" 0 1 131075 5 16 -0.496625 "S2" "SC" 0 1 131075 6 16 -0.496625 "S3" "SC" 0 1 131075 7 16 -0.496625 "S4" "SC" 0 1 131075 8 16 -0.496625 !entry.a.unit.atomspertinfo table str pname str ptype int ptypex int pelmnt

dbl pchg "FE1" "FC" 0 -1 0.0 "FE2" "FC" 0 -1 0.0 "FE3" "FU" 0 -1 0.0 "FE4" "FC" 0 -1 0.0 "S1" "SC" 0 -1 0.0 "S2" "SC" 0 -1 0.0 "S3" "SC" 0 -1 0.0 "S4" "SC" 0 -1 0.0 !entry.a.unit.boundbox array dbl -1.000000 0.0 0.0 0.0 0.0 !entry.a.unit.childsequence single int 302 !entry.a.unit.connect array int 0 0 !entry.a.unit.hierarchy table str abovetype int abovex str belowtype int

belowx "U" 0 "R" 1 "R" 1 "A" 8 "R" 1 "A" 7 "R" 1 "A" 6 "R" 1 "A" 5

20

"R" 1 "A" 4 "R" 1 "A" 3 "R" 1 "A" 2 "R" 1 "A" 1 !entry.a.unit.name single str "default_name" !entry.a.unit.positions table dbl x dbl y dbl z -20.371000 83.065000 19.300000 -19.102000 82.008000 21.471000 -18.519000 84.592000 20.710000 -20.969000 83.988000 21.798000 -18.917000 83.819000 22.825000 -20.593000 85.237000 19.932000 -21.344000 81.902000 20.997000 -18.116000 82.665000 19.530000 !entry.a.unit.residueconnect table int c1x int c2x int c3x int c4x int c5x

int c6x 0 0 0 0 0 0 !entry.a.unit.residues table str name int seq int childseq int startatomx str

restype int imagingx "SF4" 301 9 1 "?" 0 !entry.a.unit.residuesPdbSequenceNumber array int 1 !entry.a.unit.solventcap array dbl -1.000000 0.0 0.0 0.0 0.0 !entry.a.unit.velocities table dbl x dbl y dbl z 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

0.0 0.0 0.0

Figure S11. Force field library file for the 4Fe4S cluster based on the parameterisation by

Carvalho and Swart23 with adapted charges for the unique Fe atom.

21

1.4. MD analysis results:

The standard analysis for a set of MD simulations of the enzyme QueE with bound substrates

or substrate analogue are shown graphically below. Important output files are also presented

in the figshare repository for this publication. The figures below show a) the overall

flexibility of the enzyme demonstrated by overall RMSD vs time and RMSF fluctuations of

individual residues for the dimeric protein and for the individual monomers; b) the crucial

distance for H-abstraction between the substrate carbon C6 and C5’ of SAM and the dihedral

angle of CPH6 depicting the conformational change of the substrate; c) the rmsd of SAM and

the iron-sulfur cluster; d) the crucial distances for the coordination between SAM and the

cluster; and e) crucial dihedral and pucker angles depicting conformational changes of SAM

(the orientation of the base, the conformation of the sulfur, and the puckering of the sugar).

Further, analysis for all crucial hydrogen bonding interactions in the binding pocket are

presented. All analysis are presented for repeated simulations (numbered) with different

divalent cations the substrate or the intermediate based in the different available crystal

structures:

4NJI: including CPH6 and Mg2+

4NJG: including the substrate analogue (CP6)

22


QueE (pdb 4NJI) with the substrate CPH6 and Mg2+ (simulation ID 1). The simulation time

presented starts after 9 ns of restrained equilibration and the equilibration and the rmsd is

calculated against the crystal structure configuration.

23





24





25


QueE (pdb 4NJI) with the substrate CPH6 and Ca2+ (simulation ID 1). The simulation time



26





27

Figure S17. Graphically summarized MD analysis for the first 100 ns of a simulation of QueE

(pdb 4NJI) with the substrate CPH6 and Na+ (simulation ID 1). The simulation time presented

starts after 9 ns of restrained equilibration and the equilibration and the rmsd is calculated

against the crystal structure configuration.

28


QueE (pdb 4NJI) with the substrate CPH6 and Na+ (simulation ID 2). The simulation time



29


QueE (pdb 4NJG) with the substrate analogue CP6. The simulation time presented starts after

9 ns of restrained equilibration and the equilibration and the rmsd is calculated against the

crystal structure configuration.

30





31





32





33





34





35

Table S1. Hydrogen bonding summary for first 55 ns of a simulation of QueE (pdb 4NJI) with

the substrate CPH6 (MOL) and Ca2+ (simulation ID 2, monomer B). Substrate fully bound.

Contacts between Ca2+ and the substrate and protein.

Residue Atom name Fraction

MOL 213 O4 0.9996

Asp 49 OD2 0.7261

Asp 49 OD1 0.3859



Contacts between substrate and the protein.

Atom name Fraction

MOL O6B Arg 26 NH1 0.5065

MOL O6A Arg 26 NH1 0.4718

MOL O6A Thr 89 OG1 0.4608

MOL N1 Gly 13 N 0.4289

MOL O6B Thr 89 OG1 0.4153



Leu 11 O MOL H7 0.1403

Leu 11 O MOL H6 0.0646

Gln 12 OE1 MOL H5 0.0331

Glu 14 OE2 MOL H2 0.0109

36



Key hydrogen bonds of central residues (>10% fraction for amino acids).

Gly 90 O Arg 26 NH1 0.9181

Arg 26 O Lys 5 N 0.7301

Glu 6 O Arg 26 N 0.5839

Glu 6 OE1 Arg 26 NE 0.3609

Glu 6 OE2 Arg 26 NE 0.1939

Thr 89 O Phe 27 N 0.8947

Cys 25 O Thr 89 N 0.5365

Thr 89 OG Gly 90 N 0.1064

SAM N7 Asp 49 N 0.3994

Asp 49 OD1 Gln 201 HE21 0.2910

Asp 49 OD1 Lys 204 NZ 0.1807

Asp 49 OD1 Lys 204 NZ 0.1793

Asp 49 OD1 Lys 204 NZ 0.1768



Hydrogen bonding of key residues with solvent (sum of >10% fraction).

MOL Acceptor O6A, O6B, N1, N3,

O4, N2, N8, N5

3.0205

MOL Donor N5, N8, N2, N3 1.0611

Arg 26 Acceptor O, NH2, N, NH1, NE 1.4551

37

Arg 26 Donor NH1, NH2, NE, N 0.8723

Thr 89 Acceptor O, N, OG1 0.8525

Thr 89 Donor OG1, N 0.2788

Asp 49 Acceptor O, N, OD1(~0.5),

OD2(~0.5)

1.8863

Asp 49 Donor N 0.1333

Thr 50 Acceptor O, N, OG1(~1.0) 1.8725


Pro 209 Acceptor O, OXT, N 1.1703


with the substrate CPH6 (MOL) and Mg2+ (simulation ID 4, monomer A). Substrate fully

bound. Contacts between Mg2+ and the substrate, protein.

Residue Atom Fraction

MOL 213 O6A 1.0000

MOL 213 O4 1.0000

38


with the substrate CPH6 (MOL) and Mg2+ (simulation ID 4, monomer A). Hydrogen bonds

between Substrate and protein.

Atom Fraction



MOL N1 Gly 13 N 0.4026

Gln 12 OE1 MOL N8 0.3876

Glu 14 OE2 MOL N5 0.3557


Leu 11 O MOL N2 0.1656

Leu 11 O MOL N2 0.1381


with the substrate CPH6 (MOL) and Mg2+ (simulation ID 4, monomer A). Key hydrogen

bonds of central residues (>10% fraction for amino acids).

Gly 90 O Arg 26 NH1 0.8780

Arg 26 O Lys 5 N 0.7331

Glu 6 O Arg 26 N 0.4932

Glu 6 OE2 Arg 26 NE 0.2081

Thr 89 O Phe 27 N 0.8973

Cys 25 O Thr 89 N 0.4875

Thr 89 OG Gly 90 N 0.1274

39

Asp 49 OD1 Gln 201 NE2 0.5050

Asp 49 OD2 Gln 201 NE2 0.2794

Asp 49 OD1 Lys 204 NZ 0.1416

SAM N7 Asp 49 N 0.1300

Asp 49 OD1 Lys 204 NZ 0.1286

Asp 49 OD1 Lys 204 NZ 0.1282


with the substrate CPH6 (MOL) and Mg2+ (simulation ID 4, monomer A). Hydrogen bonding

of key residues with solvent (sum of >10% fraction).

MOL Acceptor N1, N2, N3, N5, N8,

O6A, O6B, O4

2.5359

MOL Donor N2, N3, N5, N8 0.8324

Arg 26 Acceptor N, O, NH1, NH2, NE 1.6118

Arg 26 Donor NH1, NH2, NE, N 0.9419




OD2(~1.4)

3.4804


Thr 50 Acceptor O, N, OG1 (~1.0) 1.7114


40



with the substrate CPH6 (MOL) and Mg2+ (simulation ID 4, monomer B). Contacts between

ion and substrate, amino acids.


MOL 213 O6B 1.0000

MOL 213 O4 1.0000


with the substrate CPH6 (MOL) and Mg2+ (simulation ID 4, monomer B). Hydrogen bonds

between Substrate and protein.

Atom Fraction



MOL N1 Gly 13 N 0.3184

Gln 12 OE1 MOL N8 0.2962


Glu 14 OE2 MOL N5 0.2096

Leu 11 O MOL N2 0.1515

Leu 11 O MOL N2 0.1047

41

Table S11. Hydrogen bonding summary for first 100 ns of a simulation of QueE (pdb 4NJI) with the

substrate CPH6 (MOL) and Mg2+ (simulation ID 4, monomer B). Key hydrogen bonds of central

residues (>10% fraction for amino acids).

Gly 90 O Arg 26 NH1 0.8976

Arg 26 O Lys 5 N 0.7349

Glu 6 O Arg 26 N 0.5812

Glu 6 OE1 Arg 26 NE 0.2871

Glu 6 OE2 Arg 26 NE 0.2791

Thr 89 O Phe 27 N 0.8941

Cys 25 O Thr 89 N 0.5570

Thr 89 OG1 Gly 90 N 0.1373

Asp 49 OD2 Gln 201 NE2 0.7936

Asp 49 OD2 Lys 204 NZ 0.2391

Asp 49 OD2 Lys 204 NZ 0.2225

Asp 49 OD2 Lys 204 NZ 0.2046

SAM N7 Asp 49 N 0.1489


with the substrate CPH6 (MOL) and Mg2+ (simulation ID 4, monomer B). Hydrogen bonding

of key residues with solvent (sum of >10% fraction).

MOL Acceptor O6A, O6B, O4, N1,

N2, N3, N5, N8

2.5878

42

MOL Donor N2, N3, N5, N8 0.85954


Arg 26 Donor NH1, NH2, NE, N, 0.9394




OD2(~1.3)

3.4236


Thr 50 Acceptor O, N, OG1(~1.0) 1.8207




with the substrate analogue CP6 (2KA) and Mg2+ (simulation ID 1, monomer A). Contacts

between ion and substrate, amino acids.


2KA 214 O3 1.0

2KA 214 O1 1.0

Thr 50 50 OG1 1.0

43


with the substrate analogue CP6 (2KA) and Mg2+ (simulation ID 1, monomer A). Hydrogen

bonds between Substrate and protein (> 1% occupancy).

Atom Fraction

2KA O2 Arg 26 NH1 0.9641

2KA O2 Thr 89 OG1 0.7054

Glu 14 OE2 2KA N4 0.4779

MOL N1 Gly 13 N 0.3191

SAM O4 2KA N4 0.3061

2KA O2 Arg 26 NH2 0.3004

Pro 209 OXT 2KA N3 0.2733

Leu 11 O 2KA N2 0.1645

Pro 209 O 2KA N2 0.1635

Pro 209 O 2KA N3 0.1538

Leu 11 O 2KA N2 0.1509

Pro 209 OXT 2KA N2 0.1474

Pro 209 O 2KA N2 0.0849

Pro 209 OXT 2KA N2 0.0108


with the substrate analogue CP6 (2KA) and Mg2+ (simulation ID 1, monomer A). Key

hydrogen bonds of central residues (>10% fraction for amino acids).

Gly 90 O Arg 26 NH1 0.8406

Arg 26 O Lys 5 N 0.7711

Glu 6 OE1 Arg 26 NE 0.6806

Glu 6 O Arg 26 N 0.5813

Glu 6 OE1 Arg 26 NH2 0.1475

44

Thr 89 O Phe 27 N 0.8873

Cys 25 O Thr 89 N 0.5782

Thr 89 OG1 Gly 90 N 0.1803

Asp 49 OD1 Thr 50 OG1 0.9885

Asp 49 OD2 Gln 201 NE2 0.7062

Asp 49 OD2 Lys 204 NZ 0.3231

Asp 49 OD2 Lys 204 NZ 0.2337

Asp 49 OD2 Lys 204 NZ 0.2306

Thr 50 OD1 Asp 49 OD1 0.9885

Pro 209 O Thr 10 OG1 0.2808

Pro 209 OXT Thr 10 OG1 0.1303


with the substrate analogue CP6 (2KA) and Mg2+ (simulation ID 1, monomer A). Hydrogen

bonding of key residues with solvent (sum of >10% fraction).

MOL Acceptor O1, O2, O3, N1, N2,

N3, N4, N5

3.6815

MOL Donor N2, N3, N4, N5 1.014

Arg 26 Acceptor O, NH2, N, NH1, NE

(~ 5*0.34)

1.7039


45




OD2(~1.2)

2.5912


Thr 50 Acceptor O, N, OG1(each

~0.34)

1.0252




with the substrate analogue CPH6 (MOL) and Na+ (simulation ID 2, monomer A). Contacts

between ion and amino acids.


MOL 214 O4 0.0359



46

between ion and amino acids. Hydrogen bonds between Substrate and protein (> 1%

occupancy).

Atom Fraction

Gln 12 OE1 MOL N8 0.6347





Leu 11 O MOL N2 0.4373


MOL N1 Gly 13 N 0.3842


Leu 11 O MOL N2 0.2336

Pro 209 O MOL N3 0.0810

Pro 209 OXT MOL N2 0.0471



between ion and amino acids. Key hydrogen bonds of central residues (>10% fraction for

amino acids).

Arg 26 O Lys 5 N 0.7522

Gly 90 O Arg 26 NH1 0.7342

Glu 6 OE1 Arg 26 NE 0.7101

Glu 6 O Arg 26 N 0.6748

Glu 6 OE1 Arg 26 NH2 0.1467

47

Thr 89 O Phe 27 N 0.8998

Cys 25 O Thr 89 N 0.5898

Thr 89 OG1 Gly 90 N 0.1727

Asp 49 All <10%



between ion and amino acids. Hydrogen bonding of key residues with solvent (sum of >10%

fraction).


N2, N3, N5, N8

1.9938

MOL Donor N2, N3, N5, N8 0.7049






OD2(~0.5)

1.6446



48




with the substrate analogue CPH6 (MOL) and Na+ (simulation ID 2, monomer B). Contacts

between ion and amino acids. Contacts between Ion and amino acids.

Residue Nr Atom Fraction

Asp 49 49 OD1 0.9997

Asp 49 49 OD2 0.9919

SAM 212 O4’ 0.1668


with the substrate analogue CPH6 (MOL) and Na+ (simulation ID 2, monomer B). Hydrogen

bonds between Substrate and protein (> 1% occupancy).

Atom Fraction

Gln 12 OE1 MOL N8 0.5461


MOL N1 Gly 13 N 0.5050



Leu 11 O MOL N2 0.4485


49



Leu 11 O MOL N2 0.1600

Pro 209 O MOL N3 0.1163




with the substrate analogue CPH6 (MOL) and Na+ (simulation ID 2, monomer B). Key

hydrogen bonds of central residues (>10% fraction for amino acids).

Arg 26 O Lys 5 N 0.7894

Gly 90 O Arg 26 NH1 0.7499

Glu 6 OE1 Arg 26 NE 0.7383

Glu 6 O Arg 26 N 0.5938

Thr 89 O Phe 27 N 0.9089

Cys 25 O Thr 89 N 0.5098

Thr 89 OG1 Gly 90 N 0.1750

Asp 49 OD1 Thr 50 OG1 0.4655

Asp 49 OD2 Lys 204 NZ 0.3341

Asp 49 OD2 Lys 204 NZ 0.2836

Asp 49 OD2 Gln 201 NE2 0.2734

Asp 49 OD2 Lys 204 NZ 0.2260

SAM N7 Asp 49

50

Asp 49 OD1 Thr 50 OG1 0.4655

Pro 209 OXT Thr 10 OG1 0.1163


with the substrate analogue CPH6 (MOL) and Na+ (simulation ID 2, monomer B). Hydrogen

bonding of key residues with solvent (sum of >10% fraction).


N2, N3, N5, N8

3.8209

MOL Donor N2, N3, N5, N8 1.0960

Arg 26 Acceptor O, NH2, N, NH1, NE


Thr 89 Acceptor O, N, OG1

Thr 89 Donor OG1, N

Asp 49 Acceptor O 1.5515

Asp 49 Donor N

Thr 50 Acceptor O, OG1 2.201

Thr 50 Donor OG1, N ~0

Pro 209 Acceptor O, OXT 5.2039

51

Table S25. Summary of residence time analysis of the substrate in correct positioning and

conformation bound insight the binding pocket of QueE.

Time (I)

Time

(II)

Simulation

Sim

ID Monomer [ns] [ns]

4NJI Mg 1 A 10 10 a1 i1

4NJI Mg 1 B 84 84 b2 i1 ++100ns

4NJI Mg 2 A 223 223 a2 i1

4NJI Mg 2 B 166 166 b2 i1 ++209ns

4NJI Mg 3 A 226 226 a1 i1

4NJI Mg 3 B 188 188 a1 i1

4NJI Mg 4 A 15 15 a2

4NJI Mg 4 B 146 146 a2 i1

4NJI Mg 5 A 219 219 a1 i1

4NJI Mg 5 B 180 196 a1 i1

mean: 145.7 147.3

STD 81.97 82.87

4NJI Ca 1 A 0 0 a2 i2

4NJI Ca 1 B 57 57 a1 i2

4NJI Ca 2 A 25 25 a2 i2 *

4NJI Ca 2 B 24 31 a2 i1 *,**24-31ns

4NJI Ca 3 A 5 5 a2 i2 *

4NJI Ca 3 B 0 0 a1 i2

4NJI Ca 4 A 4 4 i2

4NJI Ca 4 B 0 0 a1 i1

4NJI Ca 5 A 0 0 a1 + 500ns

4NJI Ca 5 B 69 69 a2 i1

mean: 18.4 19.1

STD 25.53 25.80

4NJI Na 1 A 0 0 a2 i2

4NJI Na 1 B 0 0 a2 i2

4NJI Na 2 A 0 26 a1 i2

4NJI Na 2 B 15 15 a1 i2

4NJI Na 3 A 0 67 a1 i2 ~3ns *1-67ns

4NJI Na 3 B 0 0 a2 i1 at 5ns

4NJI Na 4 A 0 0 a2 i2 *

4NJI Na 4 B 0 0 a2 i2 *,** 8-10ns

4NJI Na 6 A 0 0 b2 i2 *0-10ns

4NJI Na 6 B 8 8 b2 i2

mean: 2.3 11.6

STD 5.12 21.37

52

Time (I) Stable binding in correct position and conformation

Time (II) including fluctuating binding in correct position and conformation,

but with unstable conformation

a) simultaneously leaving catalytic position and switching conformation

b) first moving out than switching conformation

1) leaving in one step

2) leaving stepwise

i1: ion leaving with substrate

i2: ion leaving before substrate at

* flipping conformation frequently, including rebinding events

** getting back into conformation

+ remaining close inside the active site until ++ remaining close to ideal conformation and position until

53

1.5. Comparison of MD geometries to experimentally obtained crystal structure 4NJI

The original crystal structure of QueE with bound substrate and Mg2+ ion (pdb code 4NJI)

postulates coordination distances of the magnesium ion of 2.87 Å in average. These distances

are suspiciously long in comparison to average coordination data from the Cambridge

Structural Database (CSD) of 2.08 Å and also comparison to other magnesium binding

proteins from the pdb (as analysed preciously,26 and can originate from different causes. In

addition, our previously published results from DFT model calculations demonstrate a much

smaller average coordination distances of 2.10 Å in accordance to the above-mentioned data.

Therefore, the presented coordination represents a very strained system and can originate

from uncertainties in the crystal structure. This is also supported by a closer examination of

the electron density map that confirms differences between the two binding sites (see Figure

S25 below). Some of these reasons (as also confirmed by personal discussions with two of

the lead authors of the crystal structure paper) might originate from:

Firstly, the crystal structures originate from experiments in buffer solutions that offered an

excess of a given ion. This does not eliminate the possibility, that sometimes enzymes bind

other ions in the active site. The preference for magnesium, however, is demonstrated by

kinetic experiments. These, in turn cannot be facilitated completely isolated from magnesium

ion traces as the substrate itself is synthesized via a biosynthetic route and cannot be isolated

from magnesium ions necessary in the previous synthesis step. This means, that during

crystallisation a mixture of ions could be present in the active site which is also represented in

the resulting diffraction data, as confirmed by the authors of the experimental paper.

Secondly, it is possible that some enzymes in the structure also represent product or

intermediate states of the catalytic reaction. As our simulations with the intermediate

demonstrate the exact same water coordination as shown in the crystal structure for the

54

substrate it demonstrates how the water coordination can change during catalytic turnover,

which might be represented in the crystal structure as well.

Lastly, a crystal structure always represents the optimal structure under the given

crystallisation conditions. Therefore, slight differences in the presented water coordination

might also reflect the instabilities of the presented crystal structure model in solution (not

necessarily weaknesses in the description by the computational models).

This in summary makes it likely, that the overall crystal structure represents a mixture of

different states, however, clearly dominated by the postulated enzyme substrate complex

including the presented ion. Slight uncertainties for the water coordination, however, cannot

be ruled out and it cannot be concluded with 100% certainty if the present mismatch between

simulation and crystallographic structure can be addressed to weaknesses in only one of the

descriptions.

The overall excellent agreement between computational and experimental binding geometries

and with experimental kinetic studies give high confidence that the presented computational

model is of sufficient accuracy.

55

Figure S25. Comparison of crystal structure, MD, and DFT model system geometries for the

active site of QueE containing the substrate CPH4, Mg2+, and water molecules.

Experimentally observed electron density in the active site of a) chain A and b) chain B of

QueE (pdb code 4NJI); c) superimposed structures of crystal structure and equilibrated MD

56

geometries; d) distance analysis of Mg2+ coordination from crystal structure 4NJI and DFT

models as presented in reference 26.

57

2. Quantum chemical calculations

2.1 High level RSE and rearrangement barrier calculations for alternative substrates

All accurate RSE calculations for alternative substrates were performed analogue to our

previous publication on QueE27 and as previously outlined for reactions in radical SAM

enzymes by Hioe and Zipse.28 All geometry optimizations and frequency calculations of the

open-shell systems and their restricted counterparts were performed at the UB3LYP13-15/6-

31+G(d) and the UBMK29/6-31+G(2df,p) levels, including diffuse functions.18 Stationary

points were confirmed by calculating their normal vibrations. Higher level single point

calculations were performed on all B3LYP geometries at the M06-2X30/ 6-311++G(3df,3p)

level and with the more accurate G3B331 methodology wherever affordable. All relative

energies were corrected with unscaled zero-point energies on the level of their geometry

optimization. RSE energies were then calculated applying thermal corrections to enthalpies at

298.15 K at the level of their geometry optimization. All optimized geometries can be found

on the figshare repository.

2.2 vRSE calculations

For the calculation of the vRSE values snapshots from the trajectories of the dynamic

simulations were analyzed with single point energy calculations, at both a semi-empirical (SE)

and density functional level of theory (DFT). All semi-empirical calculations reported were

carried out using MOPAC32 a semi-empirical quantum chemistry package based on Dewar and

Thiel’s NDDO approximation33 with the PM6-D3 method which uses Grimme’s D3 dispersion

corrections for correlation.34 All DFT calculations use the quantum chemistry package Q-

Chem35 at an M062X/6-31+G(d) level of theory. Grimme’s D3 dispersion corrections are

applied where: s6 = 1.0, sr,6 = 1.619 and s8 = 0.0. All energies have been corrected for

58

dispersion by use of Grimmes D3 dispersion correction with correction values taken from

literature.


simulations of CPH6 in vacuum and bound to QueE with Mg2+ for DFT and semiempirical

calculations.

DFT (M06-2X/6-31+G*(D3)) PM3(D3)

RSE SD Shift RSE SD Shift

Vacuum -54.83 13.09 0.00 -64.62 17.50 0.00

Vacuum, OPT -104.41 2.33 0.00 -147.83 4.66 0.00

Protein -20.82 20.15 34.00 -28.22 13.31 36.40

Protein, Pchg -19.35 15.60 35.48 -22.70 19.77 41.93

Protein, Pchg, OPT -33.73 23.27 70.68 -246.66 1698.61 -98.83


simulations of the product radical (5) in vacuum and bound to QueE with Mg2+ for DFT and

semiempirical calculations.

DFT (6-31+G*, M06-2X) PM3

RSE SD Shift RSE SD Shift

Vacuum -29.36 7.80 0.00 -58.41 9.69 0.00

Vacuum, OPT -51.81 1.00 0.00 -61.28 4.43 0.00

Protein -48.05 11.28 -18.69 -64.43 10.16 -6.02

Protein, Pchg -46.12 9.30 -16.77 -68.68 18.75 -10.27

Protein, Pchg, OPT -55.46 2.07 -3.65 -154.67 1278.25 -93.39

59

Table S28. Comparison of the accuracy of average RSE calculation when selecting fewer

‘selective’ snapshots near the peak of a Gaussian fit to a larger data set at lower

computational level (PM3). Pchg, depicts calculations including the point charge

environment of the enzyme.

DFT (6-31+G*, M06-2X)

PM3(D3) SP SP, Pchg OPT, Pchg Snapshots

3717 -15.48 -18.87 -17.84 -15.63 1

4696 -15.84 -24.09 -24.76 -64.76 1

'Selective’ Average -15.48 -20.04 -17.74 -36.98 160

Average -15.84 -20.82 -19.35 -33.73 6212

60

Figure S26. Correlation of vRSE values at the PM6-D3 and the M06-2X/6-31+G(d) level of

theory based on 5000 snapshots taken from 100 ns MD of bound substrate to QueE with Mg2+.

Energies given in kJ mol-1.

61

Figure S27. RSE value distribution at the M06-2X/6-31+G(d) level of theory derived on 5000

snapshots taken from 100 ns MD of bound substrate to QueE with Mg2+.

62

2.3. RSE calculations of docked alternative structures

Table S29. RSE values on docked alternative substrates.

With Mg2+ Without Mg2+

System SP PCHG SP PCHG OPT SP PCHG SP PCHG OPT

C1_1 6.40 -0.31 -36.84 1.93 4.09 -

C1_2 -4.30 6.08 -64.82 7.27 6.91 -31.16

C1_3 -4.59 11.83 -16.13 -10.99 -6.60 -

C2_1 -5.85 -8.54 -58.26 -0.61 3.49 -43.17

C2_2 -1.05 10.81 -40.69 2.07 7.20 -25.75

C2_3 -7.20 -8.13 -36.91 -9.17 -9.60 -16.17

C3_1 -9.45 -6.68 -51.93 -4.87 4.65 -20.66

C3_2 -10.74 -7.30 - -8.96 -7.32 -46.63

C3_3 -11.35 -5.68 -56.12 2.79 6.20 -49.09

C4_1 -8.37 -13.31 -106.85 -14.16 -7.28 -40.18

C4_2 -3.37 -15.11 -137.77 -14.11 -7.78 -47.27

C4_3 -13.62 -20.32 -107.58 -19.26 -12.74 -54.12

C5_1 - - - -23.83 -16.03 -57.79

C5_2 - - - -31.64 -22.04 -86.71

C5_3 - - - - - -

C6_1 -7.73 -10.58 - -13.59 -7.66 -36.24

C6_2 -20.05 -19.35 - -18.34 -14.78 -46.94

C6_3 -4.66 7.65 -86.38 -18.76 -10.20 -22.71

C7_1 -9.47 -9.98 -232.40 -8.05 -2.49 -18.96

63

C7_2 -10.67 -18.42 -234.67 -3.50 -10.78

C7_3 -8.83 -10.94 -58.24 -20.49 -20.80 -50.17

C8_1 -4.72 -5.27 -65.47 -8.07 -2.93 -17.33

C8_2 -5.93 -8.09 -65.54 -4.34 1.62 -87.44

C8_3 -9.29 -17.40 - -12.87 -18.21 -105.35

N2_1 -11.95 -13.61 -109.61 -36.63 -36.94 -63.09

N2_2 -15.84 -17.47 -88.58 -50.63 -46.22 -63.77

N2_3 -46.78 -56.96 - -26.68 -21.44 -69.78

N3_1 -20.22 -17.86 -26.18 -19.19 -13.31 -47.76

N3_2 -14.02 -15.16 -86.12 -21.44 -11.20 -48.79

N3_3 -14.38 -12.12 -112.20 -47.30 -45.99 -72.08

N4_1 -15.38 -17.45 -111.85 -48.96 -47.71 -109.69

N4_2 -54.11 -48.58 -138.95 -53.77 -52.53 -113.30

N4_3 -55.74 -62.87 -120.39 -39.55 -34.30 -106.54

N5_1 -41.99 -33.45 -155.99 -60.66 -55.10 -88.21

N5_2 -67.85 -54.43 -163.04 -77.69 -71.99 -164.54

N5_3 - - - - - -

N6_1 -27.78 -29.88 -161.31 -38.07 -35.41 -82.73

N6_2 -56.48 -46.49 -134.43 -64.71 -83.28 -113.31

N6_3 - - - -38.73 -29.98 -91.43

N7_1 -70.97 17.59 -121.25 - - -

N7_2 - - - - - -

N7_3 - - - - - -

N8_1 -149.49 -15.49 -44.12 -146.28 -98.04 -128.15

64

N8_2 - - - -143.75 -164.74 -197.59

N8_3 - - - -145.24 -152.75 -210.87

S2_1 -15.09 -24.57 -42.49 -15.36 -6.60 -59.43

S2_2 -12.94 -19.32 -28.34 -14.93 -15.85 -53.70

S2_3 -13.25 -7.78 - -11.90 -11.83 -55.21

S3_1 26.50 14.69 - 36.59 22.12 4.64

S3_2 41.00 42.07 - 38.66 45.78 -

S3_3 35.26 35.25 6.84 36.15 27.83 22.03

S6_1 -15.82 0.52 -1.63 -22.82 -20.47 -74.01

S6_2 -24.31 -24.10 -78.23 -21.65 -17.95 -53.67

S6_3 - - - -24.15 -15.35 -30.78

S7_1 -7.37 -8.60 -50.86 -11.02 -3.65 -22.70

S7_2 -10.91 -14.49 - -29.84 -26.73 -44.61

S7_3 -11.90 -20.70 - -23.17 -34.12 -55.11

S8_1 -11.71 -13.13 -62.93 -11.30 -4.55 -24.21

S8_2 -14.14 -23.26 - -27.99 -25.00 -44.85

S8_3 - - - -17.36 -20.68 -

65

3. Alternative substrate docking protocol

3.1. Docking protocol

Ligand and protein preparation:

The alternative set of substrate was prepared for docking using Ligprep from Schrödinger suite

(Version 2016-3),36 with OPLS3 force field.37 The charge and stereochemistry of each structure

was kept as designed. The prepared substrate set was docked into the active site of the QueE

protein crystal structure, obtained from the Protein Data Bank (QueE PDB 4NJI) and prepared

by using the Maestro Protein Preparation Wizard in the Schrödinger suite.38 A first stage of

pre-process analysed the structural integrity, where missing hydrogen atoms and residue side

chains were added to the structure by Prime refinement program. During the refinement step,

water molecules with less than three hydrogen bonds to other atoms were removed (which

resulted in no more water in the binding site). Selection of the position of hydroxyl and thiol

hydrogen, the protonation/tautomer states and the “flip” assignment of Asp, Glu, Arg, Lys and

His were adjusted at pH = 7.0 using PROPKA.39 Finally, the whole structure was minimized

using the OPLS3 force field with a root mean square deviation (RMSD) of 0.3 Å for the

displacement of non-hydrogen atoms as convergence parameter. Following a similar process,

the QueE protein was also prepared without the natural magnesium ion, removed from the

crystal structure during the early protein preparation step.

Docking:

The binding area was defined by a grid, using the receptor grid generation application. The

enclosure box was define from the optimized protein crystal structure, at the centroid of the

active site (10 Å radius around co-crystallized ligand). No constraints were added. The standard

settings of a van der Waals scaling factor of 1.0 for nonpolar atoms was conserved. Nonpolar

atoms were define with absolute value of partial atomic charges ≤ 0.25.

66

The prepared set of substrates was docked into the receptor grid using an exhaustive sampling

search technique to predict orientation, conformation and binding position of a structure inside

the receptor pocket. The receptor was kept rigid and the ligands docked with flexibility. First,

a pre-screening by Glide SP (Standard Precision)40 method was applied, defining the ligands

that bind strongly to the ones that have little affinity. The default parameters were kept to define

the ligands docked and OPLS3 was used as the force field. No constraints were used. The best

conformer of each substrate was then docked with Glide XP (extra Precision) procedure,41 with

similar settings. This gave a more extended sampling methodology and an optimized scoring

function.

After completion of the two docking screens, the three best conformers were saved per

substrate, based on the Emodel score. This score function combines the non-bonded interaction

energy and the excess internal energy of the generated ligand conformation to discriminate

between the various conformers of a ligand. The quality and efficiency of this protocol was

tested by re-docking of the co-crystallised substrate into the protein and comparing the docking

poses through calculations of RMSD in Maestro panel (accurate when RMSD <2 Å).

3.2. Docking results

The docking protocol was verified by RMSD calculation of re-docking of the co-crystallized

ligand, where a good docking pose is highlighted by an RMSD < 2 Å. The best conformer (best

Emodel score) obtained into QueE shows an RMSD of 0.915Å (RMSD 2nd= 3.978 Å; RMSD

3rd= 0.424 Å) and into QueE without Mg2+ RMSD was of 0.862Å (RMSD 2nd = 3.571 Å;

RMSD 3rd = 3.871 Å). The alternative substrates have then been docked into the crystal

structure of QueE and QueE without Mg2+.

67

Table S30. Three best conformers of each new substrate, ranked by Emodel score, inside

QueE and QueE without Mg2+. Their binding score (XPscore) and the distance between C6

from substrate and C5' from SAM are also presented.

QueE QueE wo Mg2+

Ligands XPscore (kJ/mol)

Emodel score

(kJ/mol)

C6-C5’ distance (Å)

Ligands Xpscore (kJ/mol)

Emodel score

(kJ/mol)

C6-C5 distance (Å)

Cocrystallized 3.9 Cocrystallized 3.8

N1

-4.1 -120.6 5.786

N1

-6.3 -88.8 7.3

-5.1 -111.8 6.1 -6.2 -80.4 7.1

-3.8 -105.1 6.1 -5.0 -80.4 5.8

N2

-29.2 -170.8 4.6

N2

-9.3 -76.2 5.3

-28.8 -162.0 4.3 -8.7 -69.1 4.8

-27.6 -148.2 4.6 -6.5 -58.2 5.9

N3

-28.6 -202.2 5.4

N3

-14.0 -118.1 4.6

-30.7 -190.5 4.6 -12.9 -108.9 4.3

-28.8 -187.6 3.9 -19.4 -106.8 5.0

N4

-32.9 -159.1 4.4

N4

-15.1 -106.3 4.0

-22.5 -144.0 6.7 -5.6 -93.4 5.9

-21.6 -141.1 6.3 -10.6 -92.9 6.1

N5

-10.1 -134.8 6.2

N5

-13.0 -144.9 4.9

-8.8 -112.2 5.7 -14.2 -137.7 6.1

N6

-25.7 -200.5 7.1

N6

-17.2 -135.2 3.8

-27.0 -182.1 4.1 -15.7 -131.5 5.4

-13.7 -126.0 6.9

N7

-36.7 -257.5 3.5

N7

-26.5 -235.7 3.8

N8 -30.6 -304.4 3.8

N8 -30.6 -270.9 3.8

-32.2 -292.2 5.6 -21.8 -238.6 5.0

C1

-0.9 -87.5 7.1

C1

-3.5 -80.0 7.0

-1.0 -84.6 7.5 -3.1 -79.5 6.0

-0.8 -82.5 7.4 -2.6 -77.9 6.8

C2

-27.1 -152.8 5.4

C2

-9.4 -70.8 5.1

-26.5 -136.5 7.5 -9.2 -67.4 4.8

-24.8 -134.4 6.3 -3.3 -55.3 5.4

C3

-26.6 -200.5 4.6

C3

-14.3 -119.3 4.0

-27.6 -182.5 4.5 -10.1 -113.9 5.2

-27.6 -182.5 4.5 -13.5 -112.2 5.3

C4 -27.0 -169.1 5.3 C4 -16.1 -211.4 5.2

68

-22.4 -167.1 6.3 -11.6 -91.3 5.9

-26.4 -166.2 4.8 -8.6 -88.3 6.0

C5

-9.6 -142.8 6.0

C5

-12.0 -135.2 5.2

-12.5 -116.8 5.6

C6

-26.2 -208.1 7.0

C6

-16.5 -144.9 4.3

-25.3 -202.2 7.2 -21.4 -128.1 3.9

-28.3 -196.8 4.6 -13.8 -116.4 7.0

C7

-27.4 -186.7 8.9

C7

-17.5 -174.2 3.7

-28.6 -173.3 3.9 -16.1 -158.7 6.3

-28.6 -171.2 5.1 -12.1 -155.3 6.4

C8

-34.7 -235.3 5.9

C8

-15.5 -179.6 3.8

-33.7 -227.3 5.2 -13.0 -161.2 4.3

-32.8 -212.3 4.0 -13.6 -157.4 6.5

S1

-9.80 -106.8 6.8

S1

-3.7 -111.8 8.1

-10.2 -106.3 6.0 -3.0 -85.8 7.6

-8.5 -100.1 4.1 -3.2 -85.0 7.7

S2

-27.0 -137.3 5.3

S2

-10.4 -91.7 5.0

-25.4 -134.8 4.9 -10.0 -79.1 5.0

-22.1 -129.0 7.0 -8.2 -73.3 5.0

S3

-27.1 -179.2 4.5

S3

-13.6 -111.8 4.5

-26.6 -169.1 3.8 -9.5 -111.0 6.1

-27.4 -166.2 5.0 -14.5 -109.7 4.6

S6

-31.1 -218.1 7.5

S6

-16.9 -145.7 3.7

-30.2 -213.1 7.1 -21.3 -144.4 3.8

-14.0 -121.0 7.3

S7

-30.7 -177.5 5.4

S7

-14.4 -156.6 3.6

-27.0 -164.5 8.1 -11.6 -155.3 6.8

-33.3 -159.1 3.9 -18.0 -152.4 6.4

S8

-32.2 -224.8 5.4

S8

-15.4 -158.3 3.7

-38.7 -203.1 4.0 -13.9 -149.1 7.0

-14.5 -147.4 3.7

69

4. Combined MD equilibration and RSE assessment workflow

A workflow for a combined MD equilibration and RSE assessment has been made available

on GitHub in form of a Jupiter notebook (https://github.com/ChrisSuess/RSE-Calc/). The

workflow reads different receptor files generated directly from pdb structures or previous MD

equilibrations and ligand files from a different set of pdb structures (e.g. from docking results).

Different types of MD equilibrations can be set up and performed with submission interfaces

to cloud computing. Subsequently the reaction for which the thermodynamic reaction energy

should be calculated (e.g. RSE calculations) can be defined, performed and analyzed by using

an additional script to be plugged into the workflow. A general overview of the workflow is

shown below in Figure S27 and screenshots from the Jupiter Notebook application in

Figures 28-30.

The workflow makes use of XbowFlow a workflow language designed to take advantage of

cloud technologies and can be run on the cloud using ‘Xbow’ (see

https://github.com/ChrisSuess/Project-Xbow/ for more details). Detailed explanation of this

specific workflow are as follows: Stage 1. A PDB can be loaded from a users own trajectory

own download straight from https://www.rcsb.org/. Stage 2a. The workflow ‘cleans’ the PDB

using ‘pdb4amber’. 2b. Parameterises the systems using tleap. 2c. Performs a minimisation

using Amber. 2d. Runs a production MD simulation using Xbow. Stage 3a. Extracts co-

ordinates of the ligand of interest using MDTraj.42 3b. Turns rest of the protein into point

charges. 3c. Creates QM inputs for QChem. 3d. Runs QChem using Xbow. Stage 4. Calculates

the radical stabilisation energy.

https://github.com/ChrisSuess/RSE-Calc/

https://github.com/ChrisSuess/Project-Xbow/

https://www.rcsb.org/

70

Figure S28. Overview of the MD simulation and RSE evaluation workflow available on

GitHub.

71

Figure S29. Screenshot of the introductory overview and first steps within the Jupiter

Notebook for MD and RSE calculations.

72

Figure S30.Screenshot of the graphical MD minimization analysis within the Jupiter Notebook

for MD and RSE calculations.

73

Figure S31. Screenshot of the ligand selection and RSE calculation routine within the Jupiter

Notebook for MD and RSE calculations.

74

References:

1. Salomon-Ferrer, R.; Götz, A. W.; Poole, D.; Le Grand, S.; Walker, R. C., Routine

Microsecond Molecular Dynamics Simulations with AMBER on GPUs. 2. Explicit Solvent

Particle Mesh Ewald. J. Chem. Theory Comput. 2013, 9 (9), 3878-3888.

2. Le Grand, S.; Götz, A. W.; Walker, R. C., SPFP: Speed without compromise—A mixed

precision model for GPU accelerated molecular dynamics simulations. Computer Physics

Communications 2013, 184 (2), 374-380.

3. Götz, A. W.; Williamson, M. J.; Xu, D.; Poole, D.; Le Grand, S.; Walker, R. C., Routine

Microsecond Molecular Dynamics Simulations with AMBER on GPUs. 1. Generalized Born.

J. Chem. Theory Comput. 2012, 8 (5), 1542-1555.

4. Case, D. A.; Botello-Smith, R. M. B., W.; Cerutti, D. S.; Cheatham, III, T.E.; Darden,

T.A.; Duke, R.E.; Giese, T.J.; Gohlke, H.; Goetz, A.W.; Homeyer, N.; et al. University of

California, San Francisco, 2016.

5. Berendsen, H. J. C.; Grigera, J. R.; Straatsma, T. P., The missing term in effective pair

potentials. J. Phys. Chem. 1987, 91 (24), 6269-6271.

6. Gordon, J. C.; Myers, J. B.; Folta, T.; Shoja, V.; Heath, L. S.; Onufriev, A., H++: a

server for estimating pKas and adding missing hydrogens to macromolecules. Nucleic acids

research 2005, 33 (Web Server issue), W368-W371.

7. Saez, D. A.; Vöhringer-Martinez, E., A consistent S-Adenosylmethionine force field

improved by dynamic Hirshfeld-I atomic charges for biomolecular simulation. J. Comput

Aided Mol Des 2015, 29 (10), 951-961.

8. Bame, J.; Hoeck, C.; Carrington, M. J.; Butts, C. P.; Jäger, C. M.; Croft, A. K.,

Improved NOE fitting for flexible molecules based on molecular mechanics data – a case study

with S-adenosylmethionine. Phys. Chem. Chem. Phys 2018, 20 (11), 7523-7531.

9. Bayly, C. I.; Cieplak, P.; Cornell, W. D.; Kollman, P. A., A well-behaved electrostatic

potential based method using charge restraint for deriving atomic charges: the RESP model. J.

Phys. Chem. 1993, 97 (40), 10269-10280.

10. Goldman, P. J.; Grove, T. L.; Booker, S. J.; Drennan, C. L., X-ray analysis of butirosin

biosynthetic enzyme BtrN redefines structural motifs for AdoMet radical chemistry. Proc.

Natl. Acad. Sci. USA 2013, 110 (40), 15949-15954.

11. Umitsu, M.; Nishimasu, H.; Noma, A.; Suzuki, T.; Ishitani, R.; Nureki, O., Structural

basis of AdoMet-dependent aminocarboxypropyl transfer reaction catalyzed by tRNA-

wybutosine synthesizing enzyme, TYW2. Proc. Natl. Acad. Sci. USA 2009, 106 (37), 15616-

15621.

12. Dowling, D. P.; Bruender, N. A.; Young, A. P.; McCarty, R. M.; Bandarian, V.;

Drennan, C. L., Radical SAM enzyme QueE defines a new minimal core fold and metal-

dependent mechanism. Nat. Chem. Biol. 2014, 10 (2), 106-112.

13. Becke, A. D., Density-functional thermochemistry.III. The role of exact exchange. J.

Chem. Phys. 1993, 98, 5648.

14. Lee, C.; Yang, W.; Parr, R. G., Development of the Colle-Salvetti correlation-energy

formula into a functional of the electron density. Phys. Rev. B: Condens. Matter Mater. Phys.

1988, 37, 785.

15. Stephens, P. J.; Devlin, F. J.; Chabalowski, C. F.; Frisch, M. J., Ab initio caluclation of

vibrational absorption and circular dichroism spectra using density functional force fields. J.

Phys. Chem. 1994, 98 (45), 11623-11627.

16. Ditchfield, R.; Hehre, W. J.; Pople, J. A., Self‐Consistent Molecular‐Orbital Methods.

IX. An Extended Gaussian‐Type Basis for Molecular‐Orbital Studies of Organic Molecules. J.

Chem. Phys. 1971, 54 (2), 724-728.

75

17. Dunning, T. H., Gaussian basis sets for use in correlated molecular calculations. I. The

atoms boron through neon and hydrogen. J. Chem. Phys. 1989, 90 (2), 1007-1023.

18. Clark, T.; Chandrasekhar, J.; Spitznagel, G. W.; Schleyer, P. V. R., Efficient diffuse

function-augmented basis sets for anion calculations. III. The 3-21+G basis set for first-row

elements, Li–F. Journal of Computational Chemistry 1983, 4 (3), 294-301.

19. Tomasi, J.; Mennucci, B.; Cammi, R., Quantum mechanical continuum solvation

models. Chemical Reviews 2005, 105 (8), 2999-3093.

20. Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman,

J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A.; Nakatsuji, H.; Caricato, M.;

Li, X.; Hratchian, H. P.; Izmaylov, A. F.; Bloino, J.; Zheng, G.; Sonnenberg, J. L.; Hada, M.;

Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao,

O.; Nakai, H.; Vreven, T.; Montgomery, J. A.; Peralta, J. E.; Ogliaro, F.; Bearpark, M.; Heyd,

J. J.; Brothers, E.; Kudin, K. N.; Staroverov, V. N.; Kobayashi, R.; Normand, J.; Raghavachari,

K.; Rendell, A.; Burant, J. C.; Iyengar, n.; Tomasi, J.; Cossi, M.; Rega, N.; Millam, N. J.; Klene,

M.; Knox, J. E.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann,

R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Martin, R. L.;

Morokuma, K.; Zakrzewski, V. G.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Dapprich, S.;

Daniels, A. D.; Farkas, Ö.; Foresman, J. B.; Ortiz, J. V.; Cioslowski, J.; Fox, D. J., Gaussian

09. 2009.

21. Li, P.; Roberts, B. P.; Chakravorty, D. K.; Merz, K. M., Rational Design of Particle

Mesh Ewald Compatible Lennard-Jones Parameters for +2 Metal Cations in Explicit Solvent.

J. Chem. Theory Comput. 2013, 9 (6), 2733-2748.

22. Horinek, D.; Herz, A.; Vrbka, L.; Sedlmeier, F.; Mamatkulov, S. I.; Netz, R. R.,

Specific ion adsorption at the air/water interface: The role of hydrophobic solvation. Chemical

Physics Letters 2009, 479 (4), 173-183.

23. Carvalho, A. T. P.; Swart, M., Electronic Structure Investigation and Parametrization

of Biologically Relevant Iron–Sulfur Clusters. J. Chem. Inf. Model. 2014, 54 (2), 613-620.

24. Wang, J.; Wang, W.; Kollman, P. A.; Case, D. A., Automatic atom type and bond type

perception in molecular mechanical calculations. Journal of Molecular Graphics and

Modelling 2006, 25 (2), 247-260.

26. Dokmanic, I.; Sikic, M.; Tomic. S., Metals in proteins: correlation between the metal-

ion type, coordination number and the amino-acid residues involved in the coordination. Acta

Cryst. 2008, D64, 257-263.

25. Darden, T.; York, D.; Pedersen, L., Particle mesh Ewald: An N⋅log(N) method for

Ewald sums in large systems. J. Chem. Phys. 1993, 98 (12), 10089-10092.

27. Jäger, C. M.; Croft, A. K., Radical Reaction Control in the AdoMet Radical Enzyme

CDG Synthase (QueE): Consolidate, Destabilize, Accelerate. Chem. Eur. J. 2017, 23 (4), 953-

962.

28. Hioe, J.; Zipse, H., Hydrogen Transfer in SAM-Mediated Enzymatic Radical

Reactions. Chem. Eur. J. 2012, 18 (51), 16463-16472.

29. Boese, A. D.; Martin, J. M. L., Development of density functionals for thermochemical

kinetics. J. Chem. Phys. 2004, 121 (8), 3405-3416.

30. Zhao, Y.; Truhlar, D. G., Theor. Chem. Acc. 2008, 120, 215.

31. Baboul, A. G.; Curtiss, L. A.; Redfern, P. C.; Raghavachari, K., Gaussian-3 theory

using density functional geometries and zero-point energies. J. Chem. Phys 1999, 110 (16),

7650-7657.

32. Stewart, J. J. P., MOPAC2009. Stewart Computational Chemistry: Colorado Springs,

2008.

http://scripts.iucr.org/cgi-bin/citedin?search_on=name&author_name=Dokmanic,%20I.

http://scripts.iucr.org/cgi-bin/citedin?search_on=name&author_name=Sikic,%20M.

http://scripts.iucr.org/cgi-bin/citedin?search_on=name&author_name=Tomic,%20S.

76

33. Stewart, J. J. P., Optimization of parameters for semiempirical methods VI: more

modifications to the NDDO approximations and re-optimization of parameters. J. Mol. Model.

2013, 19 (1), 1-32.

34. Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H.; A consistent and accurate ab initio

parametrization of density functional dispersion correction (DFT-D) for the 94 elements H-Pu.

J. Chem. Phys. 2010, 132 (15), 154104.

35. Shao, Y.; Gan, Z.; Epifanovsky, E.; Gilbert, A. T. B.; Wormit, M.; Kussmann, J.;

Lange, A. W.; Behn, A.; Deng, J.; Feng, X.; et al. Advances in molecular quantum chemistry

contained in the Q-Chem 4 program package. Mol. Phys. 2015, 113 (2), 184-215.

36. Maestro, Schrodinger, LLC, New York, 2016.

37. Harder, E.; Damm, W.; Maple, J.; Wu, C.; Reboul, M.; Xiang, J. Y.; Wang, L.; Lupyan,

D.; Dahlgren, M. K.; Knight, J. L.; Kaus, J. W.; Cerutti, D. S.; Krilov, G.; Jorgensen, W. L.;

Abel, R.; Friesner, R. A., OPLS3: A Force Field Providing Broad Coverage of Drug-like Small

Molecules and Proteins. J. Chem. Theory Comput. 2016, 12 (1), 281-296.

38. Madhavi Sastry, G.; 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.

39. Olsson, M. H. M.; Søndergaard, C. R.; Rostkowski, M.; Jensen, J. H., PROPKA3:

Consistent Treatment of Internal and Surface Residues in Empirical pKa Predictions. J. Chem.

Theory Comput. 2011, 7 (2), 525-537.

40. 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.; Shaw, D. E.; Francis, P.; Shenkin, P.

S., 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.

41. 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.

42. McGibbon, R.T.; Beauchamp, K.A.; Harrigan, M.P.; Klein, C.; Swails, J.M.;

Hernández, C.X.; Schwantes, C.R.; Wang, L.P.; Lane, T.J.; Pande, V.S., MDTraj: A Modern

Open Library for the Analysis of Molecular Dynamics Trajectories. Biophys J. 2015, 109 (8),

1528-1532.

https://www.ncbi.nlm.nih.gov/pubmed/?term=McGibbon%20RT%5BAuthor%5D&cauthor=true&cauthor_uid=26488642

https://www.ncbi.nlm.nih.gov/pubmed/?term=Beauchamp%20KA%5BAuthor%5D&cauthor=true&cauthor_uid=26488642

https://www.ncbi.nlm.nih.gov/pubmed/?term=Harrigan%20MP%5BAuthor%5D&cauthor=true&cauthor_uid=26488642

https://www.ncbi.nlm.nih.gov/pubmed/?term=Klein%20C%5BAuthor%5D&cauthor=true&cauthor_uid=26488642

https://www.ncbi.nlm.nih.gov/pubmed/?term=Swails%20JM%5BAuthor%5D&cauthor=true&cauthor_uid=26488642

https://www.ncbi.nlm.nih.gov/pubmed/?term=Hern%C3%A1ndez%20CX%5BAuthor%5D&cauthor=true&cauthor_uid=26488642

https://www.ncbi.nlm.nih.gov/pubmed/?term=Schwantes%20CR%5BAuthor%5D&cauthor=true&cauthor_uid=26488642

https://www.ncbi.nlm.nih.gov/pubmed/?term=Wang%20LP%5BAuthor%5D&cauthor=true&cauthor_uid=26488642

https://www.ncbi.nlm.nih.gov/pubmed/?term=Lane%20TJ%5BAuthor%5D&cauthor=true&cauthor_uid=26488642

https://www.ncbi.nlm.nih.gov/pubmed/?term=Pande%20VS%5BAuthor%5D&cauthor=true&cauthor_uid=26488642

download fileview on ChemRxivSI-vRSE-resubmit-050419.pdf (5.57 MiB)



radical stabilization energies for enzyme engineering

Documents