Structural and Functional Analysis of Bacterial Sulfonosphingolipids and Rosette‐Inducing Factor 2 (RIF‐2) by Mass Spectrometry‐Guided Isolation and Total Synthesis

Abstract

We have analyzed the abundance of bacterial sulfonosphingolipids, including rosette‐inducing factors (RIFs), in seven bacterial prey strains by using high‐resolution tandem mass spectrometry (HRMS²) and molecular networking (MN) within the Global Natural Product Social Molecular Networking (GNPS) web platform. Six sulfonosphingolipids resembling RIFs were isolated and their structures were elucidated based on comparative MS and NMR studies. Here, we also report the first total synthesis of two RIF‐2 diastereomers and one congener in 15 and eight synthetic steps, respectively. For the total synthesis of RIF‐2 congeners, we employed a decarboxylative cross‐coupling reaction to synthesize the necessary branched α‐hydroxy fatty acids, and the Garner‐aldehyde approach to generate the capnine base carrying three stereogenic centers. Bioactivity studies in the choanoflagellate Salpingoeca rosetta revealed that the rosette inducing activity of RIFs is inhibited dose dependently by the co‐occurring sulfonosphingolipid sulfobacins D and F and that activity of RIFs is specific for isolates obtained from Algoriphagus.

Similarities across species: MS‐guided isolation and the first total synthesis of bacterial sulfonosphingolipids allowed bioactivity studies on their rosette‐inducing and inhibiting capabilities in Salpingoeca rosetta. The most abundant sulfonosphingolipids inhibited rosette inducing factor‐related activity in a concentration‐dependent manner thus suggesting that sulfobacins are likely competing for the same cellular target in S. rosetta.

Introduction

Choanoflagellates, the closest known living single‐celled relatives of animals, are water‐dwelling predators of bacteria that have emerged as new model systems for studying eukaryotic development.[ ¹ , ² ] While predominately unicellular, several choanoflagellate species such as Salpingoeca rosetta have multicellular‐like life stages, including rosettes, that are formed upon sensing of specific signaling biomolecules released by their prey bacteria (Figure 1).[ ³ , ⁴ , ⁵ ] The bacterial signals responsible for the morphological switch in S. rosetta were identified as specific sulfonosphingolipids (rosette‐inducing factors, RIFs) produced by the co‐occurring prey bacterium Algoriphagus machipongonensis.[ ⁶ , ⁷ ] Isolated RIF‐1 and RIF‐2 induced rosette formation in a small subfraction of all cells, and a mixture of inducing RIFs, including yet unidentified congeners, promoted rosette development in 20–30 % of cells. ^[7] A. machipongonensis also produced two lysophosphatidylethanolamines (LPEs), metabolic intermediates in bacterial phospholipid turnover and known signaling molecules, ^[8] which synergized with RIFs to enhance rosette development in about 80 % of all cells.[ ⁷ , ⁹ ] These chemically/ecologically driven studies exemplified that a specific set of bacterial membrane lipids, which are structurally highly related to eukaryotic signaling molecules,[ ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ ] can act as cross‐kingdom signals and induce developmental changes in the eukaryotic recipient. ^[15]

Figure 1.

Cross‐kingdom interactions between S. rosetta and A. machipongonensis. Bacterial sulfonosphingolipids trigger the morphological switch from single cells to rosette‐like colonial cell type.

The tight structure‐activity relationship of RIFs raised intriguing questions about the abundance and morphogenic activities of RIF‐like metabolites derived from other known prey bacteria.

Here, we questioned to what extent RIFs, related sulfonosphingolipids, and LPEs are produced by Bacteroidetes species (Algoriphagus spp., Cyclobacterium marinum, pacifica) previously evaluated within the interaction with S. rosetta (Tables S4 and S5 in the Supporting Information). ^[4] Thus, we set out to pursue a high‐resolution tandem mass spectrometry (HRMS²)‐ and Global Natural Product Social Molecular Networking (GNPS)‐based ^[16] metabolomic analysis of bacterial prey species and aimed to complement the analysis by a total synthesis of RIF‐2 to verify its structural assignment and if possible, deduce RIF‐2’s absolute stereochemistry. ^[7]

Results and Discussion

Sulfonolipid isolation

First, we pursued a comparative HRMS²‐based GNPS analysis of five rosette‐inducing Algoriphagus species (A. machipongonensis PR1, A. aquimarinus LMG 21971, A. ratkowskyi LMG 21435, A. marincola SW‐2, A. antarcticus LMG 21980, Table S1), which revealed overall highly similar sulfonolipid and LPE profiles across all species (Figures 2 and S1–S10).[ ⁶ , ⁷ ] Whereas signals for RIF‐1 (m/z 606.441 [M−H]⁻) and RIF‐2 (m/z 604.425 [M−H]⁻) were found only in very low abundance, non‐inducing sulfonosphingolipids like sulfobacin B, D and F (m/z 574.451, 590.445, and 588.430 [M−H]⁻) dominated the sulfonolipid profiles. ^[6] An LPE‐related GNPS cluster was detected in all five strains with synergizing LPE‐450 (m/z 450.263 [M−H]⁻) and LPE‐464 (m/z 464.278 [M−H]⁻) being produced by most of the tested bacterial strains.

Figure 2.

HRMS²‐based GNPS cluster (negative ion mode) of cell membrane extracts with nodes for RIF‐1 (m/z 606.441 [M−H]⁻), RIF‐2 (m/z 604.425 [M−H]⁻), sulfobacin F (m/z 588.430 [M−H]⁻), sulfobacin D (m/z 590.445 [M−H]⁻), sulfobacin B (m/z 574.451 [M−H]⁻) in A) A. machipongonensis PR1 (red), A. antarcticus LMG 21980 (black), A. marincola SW‐2 (blue), A. ratkowskyi LMG 21435 (purple) and A. aquimarinus LMG 21971 (green) and B) A. machipongonensis PR1 (red), C. marinum LMG 13164 (black) and E. pacifica KMM 6172 (blue). Stacked C) ¹H and D) ¹³C NMR spectra (CD₃OD, 600 MHz) of EP‐606 (blue), CM‐606 (green), and RIF‐1.

We then compared the sulfonosphingolipid profile of rosette‐inducing C. marinum LMG 13164 with A. machipongonensis PR1 and the non‐inducing prey species E. pacifica KMM 6172. As shown in Figure 2B, rosette‐inducing bacteria A. machipongonensis and C. marinum share a highly similar sulfonolipid‐related molecular ion pattern with the non‐inducing strain E. pacifica including m/z signals assigned to RIF‐1 and RIF‐2. An LPE‐related GNPS cluster with nodes corresponding to LPE‐450 (m/z 450.263 [M−H]⁻) and LPE‐464 (m/z 464.278 [M−H]⁻) was observed in all three strains.

To clarify if the detected molecular ions assigned to sulfobacin F (m/z 588.430 [M−H]⁻), RIF‐1 (m/z 606.441 [M−H]⁻) and RIF‐2 (m/z 604.425 [M−H]⁻) in C. marinum and E. pacifica relate to the same stereoisomers as those characterized from A. machipongonensis,[ ⁶ , ⁷ ] we pursued an MS‐guided purification of lipid extracts (Figure S11, Table S2).

Sulfonosphingolipids, which were isolated from E. pacifica and C. marinum and showed the same molecular mass as RIF‐2 (m/z 604.425 [M−H]⁻) were assigned as CM‐604 and EP‐604, respectively. Similarly, sulfonosphingolipids having the same molecular mass as RIF‐1 (m/z 606.441 [M−H]⁻) were assigned as CM‐606 and EP‐606, while sulfobacin F congeners with an m/z 588.430 [M−H]⁻ were named CM‐588 and EP‐588, respectively.

We then compared the 1D NMR data obtained from RIF‐2, EP‐604 and CM‐604 (Tables S3 and S4, CD₃OD or [D₆]DMSO) and observed nearly identical ¹H and ¹³C chemical shift patterns as well as 2D correlations for all three compounds (Scheme 5, below), which suggested the same structural, and possibly, the same stereochemical assignment for all three isolated compounds.

Scheme 5.

A) Final steps in the synthesis of RIF‐2 congeners: a) 8, HBTU, DIPEA, DMF, 0 °C to RT, 1 h, 71 %; b) PPh₃, DIAD, AcSH, THF/CH₂Cl₂, 0 °C to RT, 2 h, 73 %; c) NaSMe, MeOH, 0 °C, 5 min, then m‐CPBA, CH₂Cl₂, 0 °C, 5 min, then SMe₂, 0 °C, 5 min, 43 %; d) Red‐Al, THF, 0 °C, 6 h, 5.9 %; B) Oxidation of sulfobacin F yielding 27 a/27 b: e) SeO₂, 1,4‐dioxane, H₂O, 130 °C, microwave, 4 h, 14 %, dr (6R)‐27 a/(6S)‐27 b 1.7 : 1. Stacked C) ¹H and D) ¹³C spectra ([D₆]DMSO, 600 MHz) of sulfonosphingolipids.

Similarly, NMR‐based comparison of isolated and synthetic RIF‐1, ^[6] CM‐606 and EP‐606 (Figure 2C, Tables S8 and S9) unraveled almost identical chemical shifts and coupling constants, which again suggested the same relative stereochemistry of all three isolated compounds. A similar conclusion was inferred when NMR spectra of sulfobacin F like compounds (sulfobacin F, CM‐588 and EP‐588) were compared (Tables S10 and S11). The comparative NMR analysis was complemented by experimental CD measurements, which showed identical absorption patterns for each sulfonolipid‐type (Figures S12–S14).

Thus, we hypothesized at this stage that sulfonosphingolipids from C. marinum and E. pacifica might be of the same stereochemical configuration as those derived from A. machipongonensis.

Bioactivities

In a next step, we investigated the influence of the most abundant bacterial sulfonosphingolipids (sulfobacins D and F) on the rosette‐inducing capacity of RIFs obtained from A. machipongonensis using S. rosetta cell lines that are kept in co‐culture with a non‐inducing bacterial prey E. pacifica.

As expected from previous studies, the addition of purified sulfonosphingolipids RIF‐1 and RIF‐2, as well as outer‐membrane vesicles (OMVs) isolated from A. machipongonensis (positive control) reliably induced the formation of stable rosettes in a dose‐dependent fashion without any sign of growth inhibition, while supplementation of either sulfobacin D or sulfobacin F caused no phenotypic change in S. rosetta (Figure 3A). ^[6]

Figure 3.

A) Rosette development of S. rosetta in the presence of increasing concentrations of RIF‐1, RIF‐2 and sulfobacin F. B) Rosette development of S. rosetta in the presence of RIF‐1 (10 μM) or RIF‐2 (3 μM) and increasing concentration of sulfobacin D and F, respectively. C) Cell density of S. rosetta in the presence of increasing concentrations of EP‐604, EP‐606 and CM‐604 and CM‐606. Outer‐membrane vesicles (OMVs) from A. machipongonensis PR1 served as positive and DMSO as negative control (data not shown). D) Structures of sulfonosphingolipids isolated from A. machipongonensis PR1, C. marinum LMG 13164 (CM) and E. pacifica KMM 6172 (EP).

However, addition of sulfobacin‐like lipids to a constant concentration of inducing RIF‐1 (10 μM) or RIF‐2 (3 μM) resulted in a dose‐dependent reduction in rosette formation (Figure 3B). A similar effect was also previously observed for the sulfonolipid‐based inhibitor IOR‐1A from A. machipongonensis.[ ⁵ , ⁷ ] Thus, structurally related sulfobacins D and F appear to act as competitive inhibitors of RIFs, and thus might also serve as a counter measure to avoid predation of A. machipongonensis by rosettes of S. rosetta.[ ⁶ , ⁷ , ¹⁷ ]

Next, we evaluated the inducing effects of RIF‐like sulfonosphingolipids isolated from the non‐inducer strain E. pacifica (EP‐604 and EP‐606) and the rosette‐inducing strain C. marinum (CM‐604 nor CM‐606). To our surprise, none of the isolated RIF‐like sulfonosphingolipids caused rosette formation at any given concentration, but instead, three RIF‐like sulfonosphingolipids EP‐604, CM‐604 and CM‐606 caused the reduction of cell density, despite their structural resemblance to RIF‐1 and RIF‐2 (Figure 3C).

Total synthesis

To solve the contradicting structure‐activity results (Figure 3D), we pursued the total synthesis of two RIF‐2 diastereomers having either S or R configuration at C‐6. The remaining stereocenters at C‐2, C‐3 and C‐2’ were anticipated to be determined as (2R,2’R,3R) due to the conserved sphingolipid biosynthesis and were also previously verified in sulfonosphingolipid syntheses of RIF‐1 and sulfobacins.[ ¹⁸ , ¹⁹ ] Inspired by our previous synthetic studies and earlier reports on structurally related 6‐hydroxy sphingosine or 6‐hydroxy ceramides[ ²⁰ , ²¹ , ²² , ²³ , ²⁴ ] we envisaged building up the capnine base by the alkynylation of Garner's aldehyde using prochiral propargylic ketones and/or enantiopure propargylic alcohols (Scheme 1). Although the outlined strategy would require a stereoselective reduction of the triple bond at the final stage of the synthesis, it allows for the use of acidic and oxidative reaction conditions to introduce the sulfono‐headgroup.

Scheme 1.

Retrosynthetic considerations for RIF‐2.

In a first step, we analyzed options for a step‐effective synthesis of the α‐hydroxy‐fatty acid as many enantioselective approaches required the use of auxiliaries, protecting group sequences and final oxidation steps to the acid.[ ⁶ , ²⁵ ] To explore new synthetic avenues, we decided to make use of a recently described Ni‐catalyzed decarboxylative cross‐coupling strategy,[ ²⁶ , ²⁷ ] and treated acetonide‐protected malic acid 1 (single step from D‐malic acid) with di(10‐methylundecyl)zinc as coupling partner, which afforded fatty acid 3 in not yet optimized 19 % yield after deprotection (Scheme 2).[ ²⁸ , ²⁹ ] Functionalized acid 4 was obtained in a similar fashion in 18 % yield in only two synthetic steps. Despite the low yields, the remarkably short reaction sequence motivated us to explore a reaction sequence that would allow for the formation of a silyl‐protected fatty acid. Hence, TIPS‐protected methyl ester 5 (2 steps from malic acid, 48 % yield) was reacted with dipentadecylzinc under decarboxylative cross‐coupling conditions which afforded the desired product in 54 % yield. Subsequent saponification yielded the linear fatty acid 6 in 12 % yield over two steps. To avoid by‐product formation and allow for modifications upon demand, di(oct‐7‐en‐1‐yl)zinc was used as a coupling partner for the cross‐coupling reaction affording methyl ester 7 in 32 % yield. The desired elongated and branched fatty acid 8 was obtained by cross metathesis, reduction of the double bond and saponification in 95 % yield over three steps. Overall, despite moderate yields for decarboxylative cross‐coupling reactions, the reaction sequences allow for the direct synthesis of the free acid from the chiral pool without use of an auxiliary and/or oxidation to the acid.

Scheme 2.

Synthesis of α‐hydroxylated fatty acids. a) CITU, NMM, DMF, RT, 10 min, then NiCl₂⋅glyme, 4,4′‐di‐tBu‐bpy, DMF, di(10‐methylundecyl)zinc, THF, RT, 16 h, 32 %; b) 1 M HCl, THF/H₂O, 110 °C, 5 min; 59 %; c) DIC, TCNHPI, 4‐DMAP, CH₂Cl₂, RT, 10 min, then NiCl₂⋅glyme, 4,4’di‐tBu‐bpy, DMF, dipentadecylzinc, THF, RT, 16 h, 54 %; d) LiOH, THF/MeOH/H₂O, 0 °C to RT, 16 h, 23 %; e) CITU, NMM, 10 min, then NiCl₂⋅diglyme, 4,4′‐di‐tBu‐bpy, DMF, di(oct‐7‐en‐1‐yl)zinc, THF, RT, 16 h, 32 %; f) Hoveyda‐Grubbs catalyst 2nd‐generation, 4‐methylpent‐1‐ene, CH₂Cl₂, 40 °C, 3 h, 98 %; g) H₂, Pd/C, EtOAc, RT, 1 h, 97 %; h) LiOH, THF/MeOH/H₂O, 0 °C to RT, 16 h, quant.

The synthesis of the capnine base commenced with the diastereoselective addition (Felkin‐Anh model) of the protected propargylic ketone 9 to Garner's aldehyde (Scheme 3). ^[30] Subsequent protection of the secondary alcohol and deprotection of the ketone afforded 11 over three steps in more than 55 % yield. We then applied Noyori's asymmetric hydrogen transfer protocol to reduce the ketone group. ^[31] While a matched effect was observed for the S,S enantiomer of the ligand yielding alcohol 12 b within 30 min at room temperature in 63 % yield, a mismatched effect was observed for the R,R enantiomer resulting in 12 a in only 37 % yield after 7 h. Here, it needs to be noted that we did not pursue a selective reduction of the triple bond at this stage as the resulting double bond derivative proved unstable towards later used deprotection‐oxidation sequences. Thus, as next steps, the secondary alcohol was TIPS‐protected using TIPSOTf and 2,6‐lutidine, and the Boc group removed using our previously reported silyl triflate‐induced cleavage to yield the free capnine bases 13 a and 13 b, respectively. ^[18] In addition, selective acetonide cleavage resulted in the formation of Boc‐protected capnine base 14.

Scheme 3.

Synthesis of capnine bases. a) nBuLi, 9, THF/DMPU, −78 °C to RT, 16 h, 65 %; b) TBSCl, DIPEA, DMF, RT, 16 h, 87 %; c) p‐toluene sulfonic acid (10 mol %), acetone, RT, 3 h, 98 %; d) for 12 a: RuCl[(R,R)‐TsDPEN](mesitylene) (5 mol %), FA, Et₃N, iPrOH, RT, 7 h, 37 %; for 12 b: RuCl[(S,S)‐TsDPEN](mesitylene) (5 mol %), FA, Et₃N, iPrOH, RT, 30 min, 63 %; e) for 13 a: 1) TIPSOTf, 2,6‐lutidine, CH₂Cl₂, 0 °C to RT, 16 h, 88 %; 2) TMSOTf, 2,6‐lutidine, CH₂Cl₂, 0 °C, 5 h, 69 %; for 13 b: 1) TIPSOTf, 2,6‐lutidine, CH₂Cl₂, 0 °C to RT, 16 h, 80 %; 2) TMSOTf, 2,6‐lutidine, CH₂Cl₂, 0 °C, 4 h, 50 %; g) PPTS, MeOH, 60 °C, 1 h, 22 % (66 % brsm); h) nBuLi, 10, THF/DMPU, −78 °C to RT, 16 h, 47 %, dr>19 : 1; i) TBSOTf, 2,6‐lutidine, CH₂Cl₂, 0 °C, 5 h, then TMSOTf, 0 °C to RT, 3 h, 27 %.

Our second strategy employed the diastereoselective addition of the (3R)‐propargylic alcohol 10 to Garner's aldehyde, which afforded 15 in moderate yields (47 %, dr >19 : 1). Boc‐deprotection to the TBS‐protected capnine base 16 was achieved in a similar fashion as shown for the branched compounds. At this stage of the synthesis, we considered ways to introduce the sulfonic acid at C‐1. While nucleophilic replacement reactions of the primary alcohol with, for example, sodium sulfite were not successful, ^[17] conversion of model substrate 17 (syn‐capnine base) ^[18] with thioacetate under Mitsunobu conditions resulted in the formation of thiol 19 (36 % yield; Scheme 4). From a mechanistic point of view, the formation of an intermediate thioester 18 and subsequent acyl migration to form the stable amide bond appeared most likely. Motivated by the finding that thiol 19 was readily oxidized with (NH₄)₆Mo₇O₂₄ to sulfonic acid[ ²⁰ , ³² ] we investigated shortly if longer acyl chains would undergo acyl‐migration (thioligation). The required TIPS‐protected thioacid 21 was obtained by treating fatty acid 8 with Lawesson's reagent in almost quantitative yield. In contrast, unprotected α‐hydroxy acid formed stable but unreactive adducts with Lawesson's reagent (for details, see the Supporting Information SI‐C12).

Scheme 4.

Synthesis of thioester for native ligation‐type reaction. a) PPh₃, DIAD, AcSH, CH₂Cl₂, 0 °C to RT, 16 h, 36 %; b) (NH₄)₆Mo₇O₂₄, H₂O₂, MeOH, RT, 20 min, 11 %; c) Lawesson's reagent, quant; d) 14, PPh₃, DIAD, THF/CH₂Cl₂, 0 °C to RT, 27 %; e) for tested conditions, see the Supporting Information.

To avoid spontaneous N‐acylation, coupling of thioacid 21 was pursued with Boc‐protected capnine base under Mitsunobu conditions yielding thioester 22 in moderate yields. To initiate the acyl‐transfer, thioester 22 was first subjected to Boc‐deprotection using TMSOTf; ^[33] however, neither deprotection nor migration of the acyl side chain was observed. Similar results were obtained with model substrates and reaction conditions known to selectively induce Boc‐deprotection in the presence of silyl‐protecting groups; in all cases either decomposition was observed or starting material recovered (Tables S1–S3).

In light of the low‐yielding Mitsunobu coupling reaction with long‐chain thioacids, we decided to reverse the reaction sequence by first acylating the capnine base and then pursuing the necessary substitution reaction at C‐1 (Scheme 5A). Boc‐deprotected capnine base 13 a was coupled with fatty acid 8 using peptide coupling conditions (HBTU in DMF) to yield 25 a, followed by the introduction of a thioacetate at C‐1 under Mitsunobu conditions in good overall yield.^[6]] To allow for mild oxidation conditions, acetate protected thiol was deprotected with NaSMe ^[34] and then oxidized with m‐CPBA in a one‐pot fashion at 0 °C, which caused the simultaneous TBS‐deprotection of the secondary alcohol yielding 26 a. ^[35] Although TBS‐cleavage was preferred under the given reaction conditions, additional TIPS‐deprotection (2’‐position) was observed occasionally with prolonged reaction times. With propargyl alcohols of type 26 in hands, the selective reduction of the triple bond was performed with Red‐Al® and the remaining TIPS‐groups were then removed with TBAF in THF. Here, it is important to note that the reduction step afforded in general higher yields if the hydroxy group at position 2′ and 6′ were TIPS protected. Overall, three sulfonosphingolipids were synthesized using the outlined synthetic route, including the 6R stereoisomer 27 a, the 6S stereoisomer 27 b (each required a total of 15 steps) and the unbranched 6R congener 28 (a total of eight steps). Comparison of NMR spectra of stereoisomer 27 a (and 28) showed a nearly perfect match to those of RIF‐2 indicating a 6R stereochemistry (Tables S8 and S9). ^[36]

Based on the hypothesis that RIF‐2 might originate from the selective enzymatic oxidation of the allylic position of the structural congener sulfobacin F, we also pursued a biomimetic approach and subjected sulfobacin F to Riley oxidation conditions (Scheme 5B), ^[22] which indeed yielded an inseparable mixture of sulfonic acids 27 a/27 b in 14 % yield.

To deduce the most likely stereochemical assignment of RIF‐2, we compared ¹H and ¹³C NMR spectra of synthesized and isolated RIF‐2 congeners (EP‐604 and CM‐604, Tables S3–S7). As depicted in Scheme 5C, ¹H NMR spectra of products obtained from the oxidation of sulfobacin F showed two clear sets of chemical shifts each belonging to one C‐6 diastereomer. While one signal set matched chemical shifts assigned to sulfonosphingolipid 27 a and RIF‐2, the second signal set exhibited chemical shift differences similar to 27 b (Δδ ${}_{{}^1\mathrm{H}}$ ∼0.02–0.05 ppm for 6‐OH, N−H, H‐3, H‐4, H‐5 and H‐6; Δδ ${}_{{}^{13}\mathrm{C}}$ ∼0.1–0.3 ppm, Tables S5 and S6). The same chemical shift patterns were obtained when NMR spectra of linear homolog 28 were compared to spectra of oxidized sulfobacin F, RIF‐2, EP‐604 and CM‐604 (Tables S7). Based on these findings, the most likely stereochemical assignment of RIF‐2 was deduced to be (2R,2’R,3R,6R) as depicted for sulfonolipid 27 a.

Revisiting bioactivities

Synthesized sulfonosphingolipids (27 a, 27 b, 28 and the 27 a/b mixture obtained from oxidation of sulfobacin F) were then tested for rosette‐inducing activity. Despite repeated purification by high‐pressure liquid chromatography (HPLC), neither 27 a, 27 b, 28 nor mixtures thereof induced rosette formation. Although the results were disappointing at first, the observations are in strong agreement with our activity studies of isolated RIF‐like sulfonosphingolipids EP‐604, CM‐604, and EP‐606, CM‐606 as none of them induced rosette formation despite nearly identical NMR datasets.

As only RIF‐1 and RIF‐2 isolated from Algoriphagus spp. induced rosette formation, we currently hypothesize that other yet undetermined and highly potent or potentially synergizing signals might be produced by the rosette‐inducing bacterium Algoriphagus spp., ^[7] and we are currently elucidating the peculiar chemical nature of these stimulating factors. Furthermore, it can be speculated that the absolute structures of RIF‐2 might still differ from synthetic 27 a (2R,2’R,3R,6R) despite highly similar NMR pattern. Here, we acknowledge that deduction of stereochemistry in sulfonosphingolipids purely based on comparative NMR analysis requires critical evaluation as micelle formation and pH dependencies both influence ¹H and ¹³C chemical shift values. Using the outlined optimized synthetic strategy current (bio)chemical studies are now directed towards the synthesis of other stereoisomers to test this hypothesis.

Conclusion

In summary, our metabolomic and organic‐synthetic driven studies resulted in three major findings. First, HRMS²‐based analysis of seven Bacteroidetes species from three different genera revealed highly similar sulfonosphingolipid profiles across species; this allowed the first isolation and structural characterization of six sulfonosphingolipid congeners from the rosette‐inducing strain C. marinum (CM‐604/606/588) and the non‐inducing strain E. pacifica (EP‐604/606/588).

Secondly, we found that the most abundant sulfonosphingolipids in Algoriphagus, sulfobacin D and F, inhibited RIF‐related activity in a concentration‐dependent manner; a similar phenomenon to that observed for the structurally related sulfonolipid IOR‐1A.[ ⁷ , ¹⁷ ] This finding suggests that sulfobacins are likely competing for the same cellular target in S. rosetta; a finding that we will exploit for future target identification.

Thirdly, starting from a new synthetic approach of branched α‐hydroxylated fatty acids, the first total syntheses of three RIF‐2 isomers were accomplished in eight and 15 steps. In addition, a biomimetic synthesis of RIF‐2 diastereomers through the allylic oxidation of sulfobacin F was pursued.

Detailed comparative NMR analyses of RIF‐1‐ (RIF‐1, CM‐606, EP‐606) and RIF‐2‐type sulfonosphingolipids (RIF‐2, CM‐604, EP‐604, 27 a/27, 28) were performed and showed nearly identical chemical shift patterns; thus the same planar structure for each compound group was proposed. Based on the acquired dataset, the absolute structure of RIF‐2 was tentatively assigned as 2R,2′R,3R,6R, but further biological and isolation studies are needed to solidify the assignment. The synthetic strategy and metabolic analysis established herein allow us to explore the basis of the observed structure‐activity relations in this ancient predator‐prey interaction in future more deeply.

Experimental Section

The datasets supporting this article have been uploaded as part of the Supporting Information and contain details of chemical procedures, 1D and 2D NMR of described compounds, as well as HRMS data and bioassay data.

Conflict of interest

The authors declare no conflict of interest.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

D. Leichnitz, C.-C. Peng, L. Raguž, F. U. N. Rutaganira, T. Jautzus, L. Regestein, N. King, C. Beemelmanns, Chem. Eur. J. 2022, 28, e202103883.

Data Availability Statement

The data that support the findings of this study are available in the supplementary material of this article.