Absolute structure determination of Berkecoumarin by X-ray and electron diffraction

The absolute structure of Berkecoumarin, a natural product from a Berkeley Pit Lake Penicillium sp. is reported from both X-ray and electron diffraction data.


Introduction
The Stierle lab has dedicated nearly 30 years to investigating extremophilic fungi derived from an acid mine waste lake in Butte, Montana.Situated within the largest United States Environmental Protection Agency Superfund site, the Berkeley Pit Lake system encompasses an abandoned openpit copper mine, measuring 1500 feet in depth and one mile across.As infiltrating groundwater interacts with the pit, rich veins of pyrite and other minerals dissolve, resulting in acid generation.The Pit holds nearly 35 billion gallons of water, with a daily inflow of >2.5 million gallons, characterized by an acidic nature (pH 2.7) and contamination with elevated metal sulfates (e.g.1000 ppm iron, 150 ppm copper, and 600 ppm zinc) (Gammons & Duaime, 2006) (Fig. 1).
While research on the chemistry and potential remediation strategies of the Berkeley Pit Lake spans almost 40 years, the microbial ecology was neglected until the Stierles began their investigation of the secondary metabolites of the resident fungal extremophiles.Although the Berkeley Pit was assumed to be too toxic to support life due to the low pH and high metal content, the Stierles, in collaboration with Grant Mitman, isolated over 40 fungi, protists, algae, protozoans, and bacteria from its water and sediments (Mitman, 1999).Despite the toxic conditions for conventional aquatic biota, the Pit Lake system provides an ideal environment for extremophiles, potentially fostering new species to produce unique secondary metabolites.The challenge of natural products drug discovery lies in devising methods to target the bioactive compounds within these organisms.
The bioactivity of Berkecoumarin has been explored.One study demonstrated the ability of Berkecoumarin to traverse cell membranes and inhibit caspase-3, suggesting a potential neuroprotective effect post-stroke (Stierle et al., 2017).Despite previous studies, the absolute configuration of Berkecoumarin remained elusive.In this article, we present the absolute structure of Berkecoumarin, employing both X-ray diffraction methods and dynamical refinement of microcrystal electron-diffraction data.

Metabolite generation and isolation
The collection, extraction, and isolation of Berkecoumarin has been described previously (Stierle et al., 2004).

X-ray data collection and processing
Crystal data, data collection, and structure refinement details are summarized in Table 1.All non-H atoms were refined with anisotropic displacement parameters.It was possible to identify H-atom positions from the difference Fourier maps.H atoms bound to O atoms were placed and refined.Those bound to C atoms were placed in geometrically calculated positions and refined using a riding model.Isotropic displacement parameters of the placed H atoms were fixed at 1.2 times the U eq value of the atoms to which they are linked (1.5 times for methyl groups).

MicroED data collection and processing
Very fine needles of Berkecoumarin, obtained by slow evaporation of a deuterated chloroform solution, were ground, then deposited on a pre-clipped continuous carbon film on Cu 200 mesh (Ted Pella 01840).The grid was then plunged into liquid nitrogen, and transferred under cryogenic conditions to the microscope.Continuous rotation electrondiffraction data were recorded using a Thermo Fisher Scientific Glacios Cryo Transmission Electron Microscope (operating at 200 keV) equipped with a CETA-D detector.Automated tilt series data collection was carried out using Leginon software (Cheng et al., 2021).A total of nine diffraction data sets were collected under parallel illumination conditions and under cryogenic temperature (' 105 K).After visual inspection, four data sets were removed due to poor quality, leaving a total of five data sets for data reduction and further analysis.A 20 mm condenser aperture was used during data collection, resulting in a ' 0.6 mm diameter beam on the specimen.

Figure 1
Berkeley Pit Lake.
splitting.These traits are unfavorable for dynamical refinement, which is, in its current implementation, based on the assumption of a perfect crystal.In the case of imperfect crystals, the results of the dynamical refinement tend to be less accurate.However, the absolute structure determination is sufficiently robust to provide reliable results even in these unfavorable cases.Therefore, the best three data sets were selected for the dynamical refinement.Their processing statistics are summarized in Table 2.

Molecular structure and packing (X-ray)
Small needles suitable for X-ray diffraction were obtained by slow evaporation of a deuterated chloroform solution of Berkecoumarin.Berkecoumarin crystallized in the orthorhombic space group P2 1 2 1 2 1 and Fig. 2 highlights the asymmetric unit.
The molecule contains two alcohol groups, each participating in hydrogen-bonding interactions [Fig.3(a) and Table 3].The phenolic alcohol group interacts with the tertiary alcohol group of an adjacent molecule, with a hydrogen-bond distance and angle for the O3-H3� � �O5 i interaction of 2.723 (3) A ˚and 161 (4) � , respectively.This hydrogen bond forms helical chains that propagate along the crystallographic a axis.This chain described in graph-set notation is C(10) [Fig. 3(b)].The helix is right-handed and seems like a main building block in the crystal assembly.In fact, this helix is further supported by a hydrogen bond between the tertiary alcohol group and the coumarin carbonyl group of a molecule directly above it in the helical column assembly [Fig.3(c)].The hydrogen-bond distance and angle of this interaction (O5-H5� � �O1 ii ) are 2.915 (3) A ˚and 170 (4) � , respectively.
Together we suspect this is what ultimately leads to the needle morphology of the crystals, as evaluations of packing diagrams highlight minimal strong intermolecular interactions between adjacent helical columns (Fig. 4).The interaction that is most striking is a C-H� � �O hydrogen bond from the methyl ether group to the phenol O atom; the C13-H13B� � � O3(À x + 1, y À 1 2 , À z + 3 2 ) hydrogen-bond parameters are 2.60 A ˚and 162.1 � .The distance between the H and O atoms is less than the sum of the van de Waals radii, with an angle greater than 130 � .This interaction is categorized as strong according to the parameters put forth by Johnson and coworkers (Fargher et al., 2022).Besides this interaction, there are minimal additional inter-column interactions.

Absolute structure determination analysis from X-ray data
From the X-ray diffraction data, we have determined the Flack parameter to be 0.01 (11) (Parsons et al., 2013) (Table 1).Calculation of the Friedif(Cu) value (36) suggests that the u value obtained here is about the best we could obtain given the chemical make-up of Berkecoumarin and the use of Cu K� radiation (Flack & Shmueli, 2007;Flack, 2008).The standard uncertainty (u) (0.11) is on the edge of what is considered to be acceptable for an established enantiopure compound (Flack & Bernardinelli, 2000, 2008).While the u value obtained is 0.01 units beyond the recommendation, we feel confident that we have determined the proper enantiomer.One reason is that chiral natural products are often produced in an optically pure form and cases of generating enantiomeric or scalemic products are rare (Finefield et al., 2012).Furthermore, analysis of the absolute structure using likelihood methods (Hooft et al., 2008)  The asymmetric unit of Berkecoumarin with the atomic numbering scheme.Displacement ellipsoids are presented at the 50% probability level.Notes: (a) range of usable frames, not the entire recorded range.(b) The dynamical refinement proceeds against unmerged data and, therefore, the R and wR values are calculated on unmerged data.Therefore, the mR and mwR are also reported.These values are calculated on the merged data (Klar et al., 2023).

Table 3
Hydrogen-bond geometry (A ˚, � ).ment, with a Hooft parameter of 0.02 (0.9).Finally, the probability statistics indicate that the absolute configuration has been correctly assigned, with a P2(true) value of 1.00.

Absolute structure determination from electron-diffraction data
There is no anomalous dispersion for electron-diffraction data, so determination of the enantiomer is not possible with a kinematical refinement of the data.However, dynamical refinement has proven to be a powerful and reliable method for determining the absolute configuration of chiral molecules (Bra ´zda et al., 2019;Klar et al., 2023;Palatinus, Petr ˇı ´c ˇek et al., 2015;Palatinus, Corre ˆa et al., 2015).
Three data sets were imported in JANA2020.The model obtained from the X-ray refinement was used as a starting model, although the structure could also be solved by ab initio methods directly from the MicroED data.A wedge-shaped crystal model was used to model the thickness variation (Palatinus, Petr ˇı ´c ˇek et al., 2015).The refinement proceeded smoothly, and the refinement statistics are summarized in Table 2.The overall R1(obs) value calculated on all three data sets is 12.82%.This is a relatively large number for dynamical refinement (likely attributable to the high mosaicity of the samples), but it can still be considered acceptable.
The absolute structure was determined by a method described previously (Klar et al., 2023).Once the refinement of the S-enantiomorph was finalized, an inverted model was created, and, without changing any parameters, it was also refined with the dynamical refinement approach.those of the R-enantiomer (Table 4).The reliability of this qualitative assessment can be quantified by the z-score method (Klar et al., 2023), which provides the confidence level of the hypothesis that one of the enantiomorphs is the correct one.The results in Table 4 show that each of the three data sets alone provides statistically significant evidence for the S-enantiomorph (z-score larger than 3).The combined z-score calculated from all three data sets is 6.39, which corresponds to the probability of an incorrect absolute structure assignment of <10 À 6 .The absolute structure is thus unambiguously determined.

Conclusion
Here we have reported the absolute structure configuration of Berkecoumarin, a natural product isolated from extremophilic microbes living in a toxic mining pit lake in Butte, Montana.The chemical make-up of this light-atom molecule pushes the limits of a routine in-house X-ray diffraction absolute structure determination from anomalous scattering.A combination of Flack and Hooft parameters, and probability statistics, indicate the S-enantiomer.To further support this finding, MicroED data were collected, and dynamical refinement was conducted.Despite the high mosaicity and low completeness, the dynamical method was able to determine the absolute configuration as the S-enantiomer as well, further confirming the assignment.Overall, this work further demonstrates that dynamical refinement of MicroED structures is a powerful and robust method for the absolute structure elucidation of lightatom chiral molecules.

Figure 3
Figure 3 Hydrogen-bond images of Berkecoumarin.(a) The two different hydrogen bonds within the Berkecoumarin structure.Hydrogen bonds are displayed by both red and neon green dotted lines.(b) Highlighted in neon green is the C(10) helical chain formed by the hydrogen bond of the phenolic alcohol group of one molecule to the tertiary alcohol group of an adjacent species.(c) The hydrogen bond (red dotted line) of the tertiary alcohol group to the coumarin carbonyl group.Molecules of similar color schemes are 'above' each other.

Figure 4
Figure 4Packing diagram of Berkecoumarin as viewed down the crystallographic a axis.
et al. � Berkecoumarin by X-ray and electron diffraction 5 of 5

Table 1
Experimental details.

Table 2
MicroED processing and dynamical refinement experimental details.

Table 4
Absolute structure determination by the dynamical refinement.Values of z-score above 3 indicate, in a statistically significant manner, that the corresponding enantiomorph is the correct one.