A Stereocontrolled Total Synthesis of Lipoxin B4 and its Biological Activity as a Pro-Resolving Lipid Mediator of Neuroinflammation

Abstract

Two stereocontrolled, efficient, and modular syntheses of eicosanoid lipoxin B4 (LXB₄) are reported. One features a stereoselective reduction followed by an asymmetric epoxidation sequence to set the vicinal diol stereocentres. The dienyne was installed via a one-pot Wittig olefination and base-mediated epoxide ring opening cascade. The other approach installed the diol through an asymmetric dihydroxylation reaction followed by a Horner-Wadsworth-Emmons olefination to afford the common dienyne intermediate. Finally, a Sonogashira coupling and an alkyne hydrosilylation/proto-desilylation protocol furnished LXB₄ in 25% overall yield in just 10 steps. For the first time, LXB₄ has been fully characterized spectroscopically with its structure confirmed as previously reported. We have demonstrated that the synthesized LXB₄ showed similar biological activity to commercial sources in a cellular neuroprotection model. This synthetic route can be employed to synthesize large quantities of LXB₄, enable synthesis of new analogs, and chemical probes for receptor and pathway characterization.

Graphical Abstract

Two stereocontrolled, efficient, and modular syntheses of eicosanoid lipoxin B4 (LXB₄) are reported along with full spectral characterization of the natural product for the first time. Synthetic LXB₄ exhibited significant neuroprotection in a cellular model of retinal ganglion cell survival assay. Our synthetic routes enable syntheses of new analogs and chemical probes for receptor and pathway identification.

Introduction

Lipoxin B4 (LXB₄) is a non-classic eicosanoid and member of the specialized pro-resolving mediator (SPM) family of polyunsaturated fatty acids, derived from arachidonic acid (AA) through a series of oxidation steps by 5-lipoxygenase (5-LOX) and 12/15-LOX.^[1,2] Extensive chromatographic evidence along with comparisons of biological activities to related eicosanoids led to the structural and configurational assignment of LXB₄^[3,4] and related lipoxin A4 (LXA₄) (Figure 1).^[5] Both lipoxins have been associated with the promotion of anti-inflammatory and pro-resolution processes since their initial discovery in 1984.^[6] Of the two structurally similar lipoxins, LXA₄ has been well-studied and shown to primarily bind to the G-protein coupled formyl peptide receptor 2 (ALX/FPR2) and GPR32, which initiates downstream anti-inflammatory effects.^[7,8] However, endogenous receptors for LXB₄ have so far not been identified. Additionally, spectroscopic analysis to confirm the structure of LXB₄ remains ambiguous, even though it has been the subject of previous total syntheses and is reported to exhibit potent bioactivities.^[9–11]

Figure 1.

Chemical structures of lipoxins A4 (1) and B4 (2).

The first total synthesis of LXB₄ was reported by Leblanc and co-workers in 1985 via a chiral pool strategy from 2-deoxy-D-ribose.^[12a] LXB₄ and its all-trans isomer were reported in order to assign the stereochemistry relative to an authentic natural sample. Shortly thereafter in 1986, Morris and co-workers utilized a very similar strategy to synthesize LXB₄ and its methyl ester for the same purpose.^[12b] In that same year, Nicolaou and co-workers independently reported a stereocontrolled total synthesis of LXB₄ and its isomers using Sharpless asymmetric epoxidation and asymmetric reduction strategies to install the stereocentres in the natural product.^[12c] Since then, the Depezay and Spur groups have also reported formal and total syntheses of LXB₄.^[12d–f] However, previously reported syntheses to date are generally low yielding, together with long synthetic routes and lack full spectral characterization for LXB₄.

Recently, both LXA₄ and LXB₄ have been implicated in the regulation of neuroinflammation and neurodegeneration.^[13] Specifically, therapeutic LXB₄ treatment was significantly more potent and efficacious than LXA₄ in promoting direct neuroprotection in a variety of neuronal cell types, and from acute and chronic injury models of the neurodegenerative disease glaucoma.^[14] Based on these data, our group has initiated a drug discovery program to study LXB₄ signaling. To enable the identification and deconvolution of LXB₄’s putative biological targets, we have developed a novel and efficient total synthesis of the natural product LXB₄. This modular and efficient synthetic approach has been used to synthesize appreciable quantities of LXB₄, allowing us to study its role in neuronal signaling mechanisms and evaluate its therapeutic potential as a treatment for neurodegeneration. Finally, for the first time, we have comprehensively characterized the structure of LXB₄ using a combination of 1D and 2D ¹H and ¹³C NMR, high resolution mass spectrometry and optical rotation, confirming that the structure of LXB₄ is indeed (5S,6E,8Z,10E,12E,14R,15S)-5,14,15-trihydroxyicosa-6,8,10,12-tetraenoic acid (2), as shown in Figure 1.

Results and Discussion

Lipoxins differ in their aliphatic sidechains but share a characteristic and synthetically challenging conjugated tetraene backbone with an E,Z,E,E-configuration (Figure 1), alongside three key alcohol stereocentres. Formal and total syntheses of LXB₄ have been previously reported using stereospecific,^[12c,g] enantioselective^[12d,h] and chiral pool synthetic strategies.^[12b,e,f,i] However, we required a robust and modular synthesis of LXB_4, to both provide unambiguous spectroscopic evidence for its structure, and to enable the future exploration of synthetic analogs for structure-activity relationship (SAR) studies to probe the biological function of LXB₄ in neuroprotection and neurodegeneration.

Based on our retrosynthetic analysis, a Z-selective semi-hydrogenation of alkyne 3 was envisaged to afford LXB₄ (2) as depicted in Scheme 1. Intermediate 3 would be constructed via a Sonogashira coupling of two key fragments, dienyne 4 and vinyl iodide 5. Two routes relying on carbonyl olefinations of functionalized aldehydes were envisioned to dienyne 4, whereas vinyl iodide 5 can be synthesized from commercially available starting materials following a literature route.^[15] The first route to dienyne 4 depended on a Wittig olefination between phosphonium salt 6 and aldehyde 7, itself the product of a Wittig one-carbon homologation/isomerization cascade on epoxy aldehyde 11. We initially elected to define the stereochemistry of the vicinal diol moiety via asymmetric carbonyl reduction and Sharpless asymmetric epoxidation process,^[16] and constructed 11 from hexanoyl chloride 12 using an approach similar to that taken by Kobayashi and co-workers in their syntheses of hydroxyeicosatetraenoic acids (HETEs)^[17] and resolvins.^[18,19] An alternative route to dienyne 4 depended on the Horner-Wadsworth-Emmons (HWE) reaction of phosphonate 9 with aldehyde 8; similar olefinations have been used in the syntheses of various other triene and tetraene natural products and derivatives.^[20a–g] In turn, it was anticipated that aldehyde 8 could be prepared in a few steps from octenal, with the alcohol stereocentres installed in the first step via an organocatalytic asymmetric dihydroxylation.^[21]

Scheme 1.

Retrosynthetic analysis of LXB₄ (2).

The synthesis of advanced intermediate epoxy aldehyde 11 was initially achieved from hexanoyl chloride. Addition of PMB-protected propargyl alcohol to hexanoyl chloride 12 afforded alkynone 13 in 76% yield. The use of readily available acyl chlorides highlights the modularity of this approach, as any aliphatic acyl chloride can be incorporated for SAR studies. Asymmetric reduction of 13 using (S)-alpine borane gave the propargyl alcohol 14 in 91% yield and 92% e.e., which was followed by a selective reduction to the E-olefin in 75% yield using Red-Al. The desired epoxy alcohol 16 was then obtained in 77% yield using Sharpless asymmetric epoxidation conditions. The resulting alcohol was protected as the TIPS ether 17, which was subjected to sequential PMB removal with DDQ followed by a DMP oxidation to the desired epoxy aldehyde 11 in excellent yields.

With aldehyde 11 in hand, a Wittig olefination with (methoxymethyl)triphenylphosphonium chloride 10 furnished the desired hydroxy enal 7 in 50% isolated yield in one step, instead of yielding the expected homologated epoxy aldehyde 19 (Scheme 2A). We initially proposed to access 7 from a base-mediated epoxide ring-opening of 19; however, we observed that the enol ether product of the olefination reaction 18 was hydrolyzed during aqueous workup or silica gel chromatography to the corresponding enol 20, which presumably undergoes an acid or base catalyzed rearrangement, simultaneously forming the enal system and opening the epoxide to fruitfully give 7 as a single diastereomer in moderate yield. Synthesis of the TMS-protected dienyne fragment 22 was completed by a Wittig olefination on hydroxy enal 7 with commercially available ylide 6, followed by a TIPS protection to furnish the globally protected advanced intermediate as a single diastereomer with a 3:1 E/Z ratio (Scheme 2A) (49% over two steps).

Scheme 2.

A) Synthesis of common dienyne intermediate 22 from hexanoyl chloride 12 via an asymmetric reduction with (S)-alpine borane followed by a Sharpless asymmetric epoxidation sequence and a proposed one-pot Wittig olefination/base-mediated epoxide ring opening cascade. B) Synthesis of common dienyne intermediate 22 from octenal 13 via an asymmetric dihydroxylation with Jørgensen’s catalyst and a Horner-Wadsworth-Emmons reaction. PMB-: p-methoxybenzyl-; Bu-: butyl-; THF: tetrahydrofuran; rt: room temperature; o/n: overnight; PhMe: toluene; L-(+)-DIPT: (+)-diisopropyl L-tartrate; TIPS-: triisopropylsilyl-; DMAP: 4-dimethylaminopyridine; DMF: dimethylformamide; DDQ: 2,3-dichloro-5,6-dicyano-1,4-benzoquinone; aq.: aqueous; DMP: Dess-Martin periodinane; TMS-: trimethylsilyl-; Me-: methyl-; PTSA: p-toluenesulfonic acid.

While the synthesis of fragment 22 was achieved from hexanoyl chloride in 10 steps, the modest E/Z ratio and overall yield (6%) prompted the exploration of an alternative route that would reduce the number of steps, while remaining amenable to the incorporation of varying aliphatic chains for future SAR studies. Thus, a route that hinges on an organocatalytic asymmetric dihydroxylation of readily available enals was designed (Scheme 2B).^[21] Starting from octenal 13, the asymmetric dihydroxylation proceeded via an epoxyaldehyde intermediate 24 that was formed upon treatment of 13 with the second generation Jørgensen catalyst (23) and hydrogen peroxide; chloral hydrate acted as a phase transfer catalyst for this first step.^[22–24] While epoxide 24 can be isolable, it can also be directly converted to trans diol 25 by the addition of excess sodium methoxide with concurrent protection of the aldehyde as a dimethyl acetal.^[21] Although the conversion of 13 to 25 is reported as a one-pot procedure, we decided to add a peroxide quenching step to eliminate the possibility of any base-catalyzed epoxidation of trace/unconverted octenal, which would negatively impact the e.e. of 25. Using this modified procedure, 25 was isolated as a single diastereomer in 80% yield and >99% e.e. (see Supporting Information) following simple aqueous workup. Subsequent protection of the diol with excess TIPSOTf gave the globally protected intermediate 26 in 80% yield. Chemoselective transacetalization of the dimethyl acetal to yield aldehyde 8 was achieved using catalytic PTSA in acetone and very mild heating. The completion of the dienye intermediate from aldehyde 8 proceeded via a HWE reaction with phosphonate 9 followed by an isomerization of the resulting protected dienyne using catalytic iodine.^[25] Indeed, similar olefination-isomerization strategies have been successfully used in the synthesis of other conjugated polyenes and lipoxin analogues.^{[12c,20b,e,f,20h–l]} In our hands, the HWE reaction of aldehyde 8 yielded the desired dienyne in 11:10 E/Z ratio, but isomerization using trace iodine in benzene gave 22 in a 13:1 E/Z ratio and 81% isolated yield over two steps. This alternative route from octenal 13 was achieved in 5 steps with a much improved 50% overall yield.

Having established two viable routes to the desired dienyne intermediate, the carbon backbone of LXB₄ was achieved smoothly with the quantitative removal of the TMS group followed by Sonogashira coupling of enyne 4 with vinyl iodide 5,^{[15,20m–q,26]} yielding the coupled product 28 in 89% yield (Scheme 5). The TIPS groups were globally removed with TBAF delivering the triol 28 in good yield, followed by hydrolysis of the methyl ester to give the penultimate product 3. The corresponding lactone 29 was also isolated in low quantities but it can easily be converted back to the methyl ester 28 or directly hydrolyzed to 3.^[12c]

Scheme 5.

Synthesis of tritiated LXB₄ (2T) via hydrogenation with Lindlar’s catalyst and tritium gas.

With 3 in-hand, various conditions were explored to the synthesis of LXB₄ via reduction of the internal alkyne (Scheme 5). Unfortunately, various conditions using a Boland reduction of 2 failed to generate the desired E,Z,E,E-conjugated tetraene, contrary to previous reports on similar internal alkynes.^{[20h,k,27–32]} In our hands, the Boland reduction conditions led to starting material recovery. Next, we explored semi-hydrogenation to generate the delicate tetraene system. Unfortunately, no reduction was observed when using P–2 Ni with ethylene diamine.^[33,34] As well, the use of various poisoned Pd catalysts, including Lindlar’s catalyst, consistently produced a mixture of LXB₄, residual starting material, and over-reduced by-products that were difficult to separate. Attempts to tune the reactivity by modifying the Pd source, additives, and solvent sources were unsuccessful; thus, we opted to pursue alternative conditions for installing the tetraene system. We were inspired by the Karstedt alkyne hydrosilylation/proto-desilylation approach used by Hansen and co-workers in their synthesis of resolvin D1 congeners.^[20q,35] Treatment of globally protected 27 with Karstedt’s catalyst and ethoxy(dimethyl)silane led to a mixture of hydrosilylated regioisomers 30a & 30b. The regioisomers were subjected to a global silyl deprotection followed by hydrolysis to furnish LXB₄ in 56% isolated yield over three steps via semi-preparative HPLC purification. (Scheme 4).

Scheme 4.

Synthesis of LXB₄ (2) via hydrosilylation/proto-desilylation of 27. Karstedt’s catalyst: platinum(0)-1,3-divinyl-1,1,3,3-tetramethyldisiloxane.

Instability of LXB₄ presented a substantial obstacle in its synthesis and purification. LXB₄ is highly unstable to air and light, and rapid degradation was observed when LXB₄ was stored as a solid even at low temperature (−20 °C). Telescoping the final transformations from 27 to 2, preventing light exposure, and storing all crude intermediates in a methanol solution were required to obtain LXB₄ of sufficient purity. LXB₄ could be stored as a solution in methanol or ethanol at −20 °C for several weeks or at −80 °C for one to three months before substantial degradation was observed. Using this route, we were able to synthesize 100–200 mg (~0.25–0.5 mmol) of LXB₄. However, because of the instability of LXB₄, synthesis of 2 on larger scales is not prudent. Instead, we were able to generate the advanced intermediate 27 on an approximately 1 mmol scale. This intermediate can be stored at −20 °C for at least six months with minimal decomposition and can be used to rapidly generate pure LXB₄.

In conducting a review of the literature around the synthesis of LXB₄, we were surprised to discover that incomplete spectroscopic data have been reported for this natural product. ¹H NMR, mass spectrometry, IR, UV, and optical rotation data have been reported for the LXB₄ methyl ester (LXB₄-ME), but data provided for the free acid was non-existent. Given the advances in analytical instrumentation since the total synthesis of LXB₄ was first reported, we were able to obtain extensive spectroscopic data (¹H, ¹³C, COSY, HSQC, NOESY, and HMBC), high-resolution mass spectrometry, HPLC, and optical rotation for this natural product. We were gratified to observe that the spectra for the material we obtained were consistent with the proposed structure of LXB₄, and also matched the spectral data previously reported for similar tetraene systems, primarily LXB₄-ME,^[12c] LXA₄,^[3,12g] resolvin D1,^[18] and resolvin D2.^[29] Furthermore, we have been able to resolve several NMR peaks at 6.38, 6.28, 5.83, and 5.73 ppm as doublet of doublets which were previously assigned as multiplets in the spectra of LXB₄-ME (see Supporting Information). To further confirm the structure of LXB₄, our synthesized material (UHN) was compared to a commercial sample (Cayman Chemical)^[36] and found to be identical by HPLC (mobile phase: acetonitrile + 0.1% TFA and MilliQ water + 0.1% TFA, gradient from 10:90→95:5, retention time 9.97 min for 2, 9.95 min for commercial sample) and HRMS (ESI HRMS calculated for C₂₀H₃₂O₅Na [M+Na]⁺, predicted m/z 375.2147, found 375.2152, commercial sample 375.2165).^[37]

In order to confirm the bioactivity of this material, we tested its neuroprotective effects in a neuronal injury model. Recent clinical failures of a variety of neuroprotection trials have intensified interest in new treatment strategies. Evidence is accumulating that lipoxins and other lipid mediators can have an important impact on neuroinflammation and neuronal survival.^[13,38] We previously reported that in addition to their known pro-resolution functions, lipoxins demonstrate an exciting novel protective bioactivity directly on neurons.^[14] In particular, LXB₄ was surprisingly 20-fold more potent than LXA₄ in an established neuronal injury model,^[14,39] and also active in a variety of primary neurons, including cortical and hippocampal cells, and retinal ganglion cells. These broad neuroprotective effects were subsequently correlated with demonstrated LXB₄ efficacy in vivo using both acute and chronic models of retinal ganglion cell death associated with the common neurodegenerative disease glaucoma, a leading cause of vision loss and blindness worldwide.^[40–42] Yet the lack of efficient syntheses of LXB₄ coupled with the expensive cost from commercial sources (USD$806/100μg, Cayman Chemical)^[36] have made this new mechanism of action challenging to modify and study. In this regard with our new synthetic route towards LXB₄, we profiled the neuroprotective activity of newly synthesized LXB₄ (UHN) and compared it against the commercial source.^[36] For this assay, HT22 neuronal cells were pre-treated with LXB₄ from both sources (commercial vs. in-house), followed by a glutamate challenge to induce excitotoxic cell death.^[14] Cell viability was then measured using an XTT assay. Both commercial and in-house samples of LXB₄ were significantly neuroprotective in this assay, demonstrating a similar three-fold recovery in neuronal survival at 1 μM (Figure 2A). In a parallel dose-response experiment, the EC₅₀ of the newly synthesized LXB₄ was determined to be 292.8 nM (Figure 2B).

Figure 2.

In vitro validation studies between synthesized LXB₄ vs. commercial LXB₄ from Cayman Chemical. A) HT22 cells were treated with 1μM of LXB₄ from either a commercial source (Cayman) or internally synthesized (UHN). Both showed significant protection in a glutamate injury model (n=8, *p<0.05, bars are S.E.). B) A dose response curve for the internally synthesized LXB₄ (UHN) in the same assay indicates an EC₅₀ of 292.8 nM (n=8, bars are S.E.).

In the interests of further characterization of LXB₄ bioactivity, we are pursuing screening technologies that required the ligand (LXB₄) to be radiolabelled with tritium. We envisaged incorporating tritium in the last synthetic step from advanced intermediate 3 via hydrogenation with tritium gas (Scheme 5). Although the transformation resulted in a mixture of unreacted LXB₄, tritiated-LXB₄, and over-reduced LXB₄, the desired tritiated product 2T was isolated in small quantities via semi-preparative HPLC purification with 96% radiochemical purity along with trace amounts of over-reduced material (see Supporting Information). Nonetheless, the radiolabelled sample contained sufficient specific activity of 37 Ci/mmol, exceeding the 20 Ci/mmol required to screen >6,200 human plasma membrane monomers, heterodimers, and secreted proteins within the panel.^[43] Results of this study, along with target identification and deconvolution using photoaffinity-labelled LXB₄ and aromatic mimics, will be described in subsequent upcoming publications by our group.

Conclusion

In summary, we have developed a stereocontrolled and modular total synthesis of LXB₄ to facilitate target deconvolution and advancement of SAR studies. The chromatographic data of LXB₄ are consistent with previous reports^[1,2,12c,e] and commercial sources.^[36] More importantly, we have now reported unambiguous spectroscopic data to confirm the configurational structure of LXB₄ to be (5S,6E,8Z,10E,12E,14R,15S)-5,14,15-trihydroxyicosa-6,8,10,12-tetraenoic acid.^[37] Two routes to the advanced enyne intermediate 4 were developed, both of which will allow for the preparation of future derivatives with varying aliphatic chains. Coupling of 4 with known vinyl iodide 5 proved an excellent strategy for building the carbon backbone of LXB₄. Moreover, we have explored several strategies for the conversion of the internal alkyne precursors (3 or 27) to the sensitive tetraene system of LXB₄. Ultimately, the hydrosilylation/proto-desilylation protocol with Karstedt’s catalyst yielded the desired tetraene system of the natural product. Final hydrolysis furnished 100–200 mg of the natural product LXB₄ in 10 steps with an overall yield of 25%, constituting the shortest, most efficient route of this natural product to date. With this new route in hand, LXB₄ analog synthesis, as well as the synthesis of LXB₄-derived chemical biology probes are currently underway to further investigate the biological function of LXB₄ in neuroprotection and neurodegeneration. These analogs will also be used for target receptor and pathway identification purposes, which will be the subject of forthcoming reports.

Scheme 3.

Synthesis of triol intermediate 3 via Sonogashira coupling and attempted conversions to LXB₄ via semi-hydrogenation. TBAF: tetra-n-butylammonium fluoride.