Aplospojaveedins A (1) was obtained as colorless crystals, with UV absorptions at λ_max 208, 255 and 317 nm. Its molecular formula was established as C₂₆H₃₅O₅N₃S on the basis of a prominent pseudomolecular ion peak at m/z 502.2366 [M + H]⁺ in the HRESIMS spectrum, indicating 11 degrees of unsaturation. The NMR spectra of 1 were recorded in DMSO-d₆ (Table 1) and in CD₃OD (Supplementary Table S1). The former solvent revealed two nitrogen-bearing protons at δ_H 8.66 (s, NH-20) and 8.27 (t, NH-27), three olefinic protons at δ_H 5.54 (br d, H-1), 5.32 (ddd, H-2), and 5.03 (q, H-13), as well as four methyl signals at δ_H 1.52 (Me-22), 1.43 (d, Me-14), 1.36 (s, Me-15) and 0.87 (d, Me-11). The ¹³C NMR spectrum of 1 displayed four carbonyl carbons at δ_C 196.6 (C-16), 171.9 (C-26), 171.1 (C-29), 167.0 (C-21) and six olefinic carbons at δ_C 174.9 (C-18), 135.3 (C-12), 132.3 (C-1), 128.1 (C-2), 123.0 (C-17), and 122.5 (C-13), accounting for seven degrees of unsaturation. Thus, compound 1 was suggested to bear a tetracyclic ring skeleton. In COSY spectrum (Figure 2), the correlations of H-1/H-2/H-3/H-4/H-5/H₂-6/H₂-7/H-8/H₂-9/H-10, and between H-5/H-10, H-8/Me-11, indicated the existence of an octahydronaphthalene ring with a double bond at C-1/C-2 and a methyl group at C-8. The COSY correlation between H-13 and Me-14, together with the HMBC correlations from Me-15 to C-12, C-13 and C-3, confirmed the attachment of a but-2-en-2-yl group at C-3. Furthermore, the HMBC correlations from NH-20 to C-16, C-17, C-18, C-19, C-21 and from H-4 to C-16 indicated the presence of an α, β-unsaturated lactam and its linkage with the above mentioned octahydronaphthalene ring via carbonyl group C-16. According to the molecular formula, two additional nitrogen atoms and a sulfur atom must reside in the remaining substructure. The presence of a glycine-cysteine moiety was supported by the COSY correlations between NH-27/H₂-28 (δ_H 3.80), H-24/H₂-25, together with the HMBC correlations from H₂-28 to C-29, and from H-24, H₂-25, NH-27 and H₂-28 to C-26. Moreover, the HMBC correlations from H-24 to C-19, from H₂-25 to C-18, and from Me-22 to C-18 and C-19, along with the HMBC correlations from Me-22 to N-20 and N-23 in ¹H-¹⁵N-HMBC confirmed the linkage between C-19/N-23 and C-18/S-25 to form the fourth ring in 1. Thus, the planar structure of 1 was elucidated as shown.

Compound 1 contains two isolated blocks of chirality: an octahydronaphthalene and a tetrahydropyrrolo [3,2-b][1,4] thiazin-6(2H)-one moiety with five and two chirality centers, respectively. The relative configurations in the isolated blocks were established by ROESY but the two blocks could not be correlated. An E configuration was assigned for the double bond between C-12 and C-13 due to the ROESY correlation between H-3 and H-13. The ROESY relationships between H-3/H-4, H-4/H-6b (δ_H 0.68), H-6b/H-10, H-10/H-4, H-10/H-8 suggested that these protons are present on the same face of the octahydronaphthalene moiety whereas the ROESY correlations between H-5 and H-6a (δ_H 1.75) established that these latter protons are on the opposite face of the ring. In the heterocyclic moiety, ROESY correlations between Me-22/H-25a (δ_H 3.08), H-24/H-25b (δ_H 3.05) as well as the coupling constants between H-24/H-25ab (²J_H-24/H-25a = 11.5 Hz, ²J_H-24/H-25b = 5.6 Hz) suggested that H-24 adopted axial orientation and Me-22 and H-24 are on opposite faces of the ring (Supplementary Figure S60).

To determine the absolute configuration, compound 1 was crystallized in methanol, and a single-crystal X-ray diffraction analysis was performed. Compound 1 crystallizes in the non-centrosymmetric orthorhombic space group (necessary for a chiral molecule) P2₁2₁2₁.The absolute configuration of compound 1 is (3R, 4S, 5R, 8R, 10S, 19R, 24R) (Figure 3). The asymmetric unit contains a molecule of 1 together with a water molecule of crystallization. In the solid state the O-H and N-H groups are engaged in inter-molecular hydrogen bonding (Supplementary Figure S1; Supplementary Table S5).

To elucidate the absolute configuration of 1 independently and test the combination of TDDFT-ECD (Superchi et al., 2018; Mándi and Kurtán, 2019) and DFT-NMR methods to assign isolated blocks of chirality (Li et al., 2020; Kovács et al., 2023), MMFF conformational searches were applied on stereoisomers (3R,4S,5R,8R,10S,19R,24R)-1, (3R,4S,5R,8R,10S,19R,24S)-1, (3R,4S,5R,8R,10S,19S,24R-1) and (3R,4S,5R,8R,10S,19S,24S)-1. The computed 199, 126, 99 and 154 initial conformers were re-optimized at the B3LYP/6–31 + G(d,p) and the ωB97X/TZVP PCM/MeOH levels. ¹³C NMR chemical shift data computed for the low-energy (≥ 1% Boltzmann populated) 7, 9, 10 and 7 B3LYP conformers were corrected for DMSO and MeOH and then compared with the experimental ¹³C NMR chemical shifts (Supplementary Tables S6, S7). The DMSO results showed better agreement for the (3R,4S,5R,8R,10S,19S,24S) and (3R,4S,5R,8R,10S,19R,24R) stereoisomers than for the other two, which confirmed cis-(19R*,24R*) relative configuration of the C-19 and C-24 substituents but the two isolated blocks could not be correlated. It is noteworthy that the presence of the sulfur heavy atom in the vicinity of chirality centers and the coordinating DMSO or MeOH solvent molecules can have a relatively large impact on the chemical shifts (Sikandar et al., 2023).

TDDFT-ECD calculations were performed at various levels of theory on the low-energy (≥ 1% Boltzmann population) 11, 9, 5 and 11 ωB97X conformers of the four stereoisomers, which indicated that the ECD spectrum was governed by the C-19 chirality center above 240 nm. The (3R,4S,5R,8R,10S,19R,24R) and (3R,4S,5R,8R,10S,19R,24S) stereoisomers reproduced equally well the experimental ECD spectrum of 1, while the (3R,4S,5R,8R,10S,19S,24R) and (3R,4S,5R,8R,10S,19S,24S) stereoisomers gave a mismatch in this region (Figure 4; Supplementary Figures S2–S4). Below 240 nm, all four computed stereoisomers reproduced the intense negative Cotton effect (CE), which suggested that this region was determined by the (4S) chirality center of the octahydronaphthalene moiety. Thus, ECD transitions above 240 nm reported the AC of C-19 chirality center of the heterocyclic scaffold, while the C-4 chirality center of the octahydronaphthalene moiety was reflected in the intense CE below 240 nm. Thus, combined with the known relative configuration of the isolated blocks, TDDFT-ECD calculations allowed determining the absolute configuration of 1 as (3R,4S,5R,8R, 10S,19R,24R), which corroborates the result of X-ray analysis.

Compound 2 shared the same molecular formula as 1 as deduced from HRESIMS. The planar structure of 2 was confirmed to be identical to that of 1 after detailed analysis of the 2D NMR spectra of 2. The obvious differences of NMR data between 1 and 2 were the chemical shifts for H-24 and H-25a which differed by +0.55 and + 0.26 ppm in 2 when compared to 1. Further differences were observed with regard to the coupling constants between H-24 and H-25ab, suggesting a different configuration of 2 compared to 1 in the nitrogen containing moiety. In the ROESY spectrum of 2, the correlations from Me-22 (δ_H 1.58) to H-24 (δ_H 3.99) and H-25α (δ_H 3.34) indicated that they were present on the same face of the ring, whereas Me-22 and H-24 were on the opposite side of the ring in 1 (Supplementary Figure S61).

For the configurational assignment of 2, the computed ¹³C DFT-NMR and TDDFT-ECD data of four stereoisomers of 2 were compared with the experimental data. The ¹³C DFT-NMR results compared to the experimental data of 2 showed a preference for the cis-(19R*,24R*) relative configuration over the trans-(19S*,24R*) one, which contradicted the experimental ROE results (Supplementary Tables S8, S9). Above 260 nm, a broad positive ECD band was measured at about 310 nm, which was reproduced well by the computed TDDFT-ECD of the stereoisomers (3R,4S,5R,8R,10S,19S,24R)-2 and (3R,4S,5R,8R,10S,19S,24S)-2, while the stereoisomers (3R,4S,5R,8R,10S,19R,24R)-2 and (3R,4S,5R,8R,10S,19R,24S)-2 gave a mismatch, allowing elucidation of the (19S) absolute configuration. All four stereoisomers reproduced the negative ECD transitions below 260 nm, which confirmed that the C-4 chirality center and hence the octahydronaphthalene moiety had the same absolute configuration as that of 1 (Figure 5; Supplementary Figures S5–S7). The negative highest wavelength n-π* transition above 350 nm could be well estimated by the (3R,4S,5R,8R,10S,19S,24R)-2, while it could not be reproduced by the stereoisomer (3R,4S,5R,8R,10S,19S,24S)-2 in accordance with the experimental ROE results. The combination of TDDFT-ECD calculations and experimental ROE data determined the absolute configuration of 2 as (3R,4S,5R,8R,10S,19S,24R).

The molecular formula of 3 was the same as for compounds 1 and 2 based on the HRESIMS data. The ¹H and ¹³C NMR data of the octahydronaphthalene ring in 1, 2 and 3 were likewise very similar, suggesting that they shared the same substructure. However, significant differences were found for chemical shifts for C-17 and C-19 (−24.9 and − 10.3 ppm, respectively) in 3 when compared to 2. In addition, H-25ab showed HMBC correlations to C-19 in 3 instead of to C-18 as in compounds 1 and 2, indicating exchange of the positions of S and NH in 3. Thus, the planar structure of 3 was elucidated as shown. The ROESY relationship between Me-22 (δ_H 1.78) and H-24 (δ_H 4.53) suggested they were on the same side of the ring (Supplementary Figure S62).

The above DFT-NMR and TDDFT-ECD protocols were performed on four stereoisomers of 3 by changing the configuration of the C-19 and C-24 chirality centers; (3R,4S,5R,8R,10S,19R,24R)-3, (3R,4S,5R,8R,10S,19R,24S)-3, (3R,4S,5R,8R,10S,19S,24R) -3 and (3R,4S,5R,8R,10S,19S,24S)-3. The comparison of the computed and experimental ¹³C NMR chemical shifts showed a clear preference for the trans-(19S*,24R*) relative configuration of the heterocyclic moiety in line with the experimental ROE results (Supplementary Tables S10, S11). TDDFT-ECD calculations of the stereoisomers (3R,4S,5R,8R,10S,19S,24R)-3 and (3R,4S,5R,8R,10S,19S,24S)-3 gave acceptable agreement with the experimental ECD spectrum of 3, while stereoisomers (3R,4S,5R,8R,10S,19R,24R)-3 and (3R,4S,5R,8R,10S,19R,24S)-3 gave a mismatch (Figure 6; Supplementary Figures S8–S10). The combination of experimental ROE data, ¹³C DFT-NMR and TDDFT-ECD calculations allowed determining the absolute configuration of 3 as (3R,4S,5R,8R,10S,19S,24R).

Compounds 1–3 show limited structural similarities to known fungal constituents including equisetin (Phillips et al., 1989), phomasetin (Singh et al., 1998), and Ascosalipyrrolidinone A (Osterhage et al., 2000). However, in compounds 1–3, the substituents on the octahydronaphthalene moiety are different from those present in known compounds. Furthermore, 1–3 show significant differences in the alkaloid part, implying that the former compounds feature a distinctive novel skeleton. A plausible biogenetic pathway for the formation of 1–3 involves iterative hybrid polyketide synthases–non ribosomal peptide synthetases (PKS–NRPS) with the following key steps (Supplementary Figure S59), which can be partially referenced from the known biogenetic pathway of equisetin (Fisch, 2013). Intermediate B was built from an octaketide and an alanine moiety. The beta carbon of the lactam ring in intermediate B was attacked either by S atom of the cysteine moiety to form 1 and 2 or by N atom of the cysteine moiety to give 3.

To identify potential targets for the three compounds 1–3, we used the target prediction software packages SwissTargetPrediction (Daina et al., 2019) and SuperPred (Nickel et al., 2014). Targets identically predicted by both approaches (Supplementary Tables S2–S4) were subjected to blind docking experiments individually (“inverse docking”). These probe ligand binding across the whole receptor to unravel whether potential binding sites of compounds 1–3 are similar to known binding sites.

The blind docking revealed a high propensity of the compounds to bind to two different sites: The orthosteric binding site of G-protein coupled receptors (GPCRs) and domains similar to the β-propeller domain found in integrins (Figures 7A,B) (Campbell and Humphries, 2011). A few binding poses in the orthosteric binding pocket in the GPCR adenosine receptor A_2B and the cannabinoid CB₂ (Figure 7A) receptor address transmembrane helices (TMs) 3 and 6 and in the chemokine receptor CCR1 TM 6. Such poses, which might be suitable for an agonist (A_2B, CB₂) or antagonist (all), were identified for all compounds. However, the number of poses identified in the orthosteric binding pockets was too small to count as valid binding poses (see Methods for criteria). In the GPCRs A_2B and formyl peptide receptor 1 (FPR1) (Figure 7C), binding modes at the extracellular loop 2 + 3 (EL2 + 3), a potential allosteric binding site (Peeters et al., 2012), were additionally identified. For the integrin α₅β₁ complex, binding modes of the compounds inside the β-propeller (Figure 7B) or the β-propeller-like domain of Kelch-like ECH-associated protein 1 were identified (Figure 7D). In the X-ray crystal structure of cyclin-dependent kinase 5 (CDK5), a binding mode of deprotonated compound 1 at the position of the co-crystallized R-roscovitine (Figure 7E) was identified. Despite this target being found for compound 3 in the ligand similarity-based screening only, the high similarity to compound 1 still renders CDK5 a viable higher-priority target for compound 1. In the X-ray crystal structure of mitotic rotamase Pin1 (Peptidyl-prolyl cis-trans isomerase NIMA-interacting 1), a binding mode of deprotonated compound 3 at the position of the co-crystallized proline (Figure 7F) was identified.

We predicted a diverse set of potential targets for compounds 1–3 based on a search for chemically and structurally similar ligands. Accordingly, proteins known to bind ligands similar to compounds 1–3 were suggested as potential targets for the respective compounds. The set of potential targets is structurally diverse, including GPCRs, integrins, kinases, and histone deacetylases. Despite the high structural similarity of, e.g., the epimeric compounds 1 and 2, varying targets were identified for all three compounds. To counteract potential biases of the screening methods and to identify targets that are more likely to be influenced by the compounds, a blind docking of all compounds to all potential targets identically identified by both screening methods was conducted (“inverse docking”). The blind docking suggested several putative targets for all three compounds, e.g., the GPCR cannabinoid receptor 2 or the integrin α₅β₁ complex; the predicted compound binding might functionally impact these targets.

Note, however, that computational target prediction (Katsila et al., 2016) and inverse docking (Vasseur et al., 2015) are considered challenging. Further investigations could thus consider compute-intensive molecular simulations of free ligand binding (Gohlke et al., 2013a; Kaiser et al., 2023) to confirm the predicted binding modes, long-time molecular simulations to probe functional effects (Gohlke et al., 2013b), or experimental validation of the putative agonistic or antagonistic activity of compounds 1–3 with a higher priority toward those targets.