Synthesis of 7-¹⁵N-Oroidin and Evaluation of Utility for Biosynthetic Studies of Pyrrole-Imidazole Alkaloids by Microscale¹H-¹⁵N HSQC and FTMS

Abstract

Numerous marine-derived pyrrole-imidazole alkaloids (PIAs), ostensibly derived from the simple precursor oroidin, 1a, have been reported and have garnered intense synthetic interest due to their complex structures and in some cases biological activity; however very little is known regarding their biosynthesis. We describe a concise synthesis of 7-¹⁵N-oroidin (1d) from urocanic acid and a direct method for measurement of ¹⁵N incorporation by pulse labeling and analysis by 1D ¹H-¹⁵N HSQC NMR and FTMS. Using a mock pulse labeling experiment, we estimate the limit of detection (LOD) for incorporation of newly biosynthesized PIA by 1D ¹H-¹⁵N HSQC to be 0.96 μg equivalent of ¹⁵N oroidin (2.4 nmole) in a background of 1500 μg unlabeled oroidin (about 1 part per 1600). 7-¹⁵N-Oroidin will find utility in biosynthetic feeding experiments with live sponges to provide direct information to clarify the pathways leading to more complex pyrrole-imidazole alkaloids.

The polycyclic pyrrole-imidazole alkaloid (PIA) family has received much attention recently from the synthetic organic communityⁱ because of their challenging molecular structure, dense functionality and potential biological activity. All are believed to derive from a simple common sponge metabolite; most likely oroidin (1a) or its analogs hymenidin (1b) and clathrodin (1c), the structures of which differ only in bromine content.ⁱⁱ Representative of the members of this family are the dimeric PIAs sceptrin (2),ⁱⁱⁱ ageliferin (3),^iv massadine (4),^v palau'amine (5)^vi and axinellamines A and B (6a-b),^vii and higher-order tetrameric PIAs.^viii In addition, compact monomeric polycyclic PIAs, such as agelastatins A,^ix B,^x C and D^xi (7a-d) and dibromophakellin^xii appear to be derived from oxidative intramolecular cyclizations of oroidin (1a) or hymenidin (1b). PIAs are of contemporary interest due to their potent biological activity and intriguing biosyntheses. For example, (–)-agelastatin A (7a), an antitumor PIA^x is 1.5-16 times more active than cisplatin against a range of tumor cell lines, and potently inhibits osteopontin, an adhesion glycoprotein transcriptionally regulated by the Wnt β-catenin pathway and implicated in tumor progression, dissemination, and metastases.^xiii

The widely accepted hypothesis for biosynthesis of dimeric and tetrameric PIAs, involves dimerization of oroidin (1a) followed by consecutive oxidative transformations leading to more densely functionalized, polycyclic alkaloids.^viii In 1982, Büchi reported a ‘biomimetic’ synthesis of the simple PIA, racemic dibromophakellin, by oxidative cyclization of 9,10-dihydrooroidin.^xiv The structure of sceptrin (2) is highly suggestive of a formal [2+2] cycloaddition of two molecules of hymenidin (1b) to 2; however, despite many attempts, Faulkner, Clardy and co-workers were unable to achieve photodimerization of 1b to 2,ⁱⁱⁱ suggesting that an alternative non-concerted mechanism may be responsible for the biosynthesis. Recent speculation on the order of biosynthesis of 2 and 3 revolves around ionic mechanisms of ring-expansion, supported by the demonstrated conversion of 2 to 3 under thermal conditions, ^if,xv,xvi or a ring-contraction pathway from an ageliferin-type dimer generated by a Diels Alder type cycloaddition of two molecules of hymenidin (1b).ⁱⁱ For example, one speculative hypothesis invokes a cycloaddition in the biosynthesis of palau'amine^vib and axinellamine, followed by a stepwise electrophilic chlorination and ring contraction. Remarkably, none of the above hypotheses have been tested experimentally in live sponges.

We describe here a synthesis of ¹⁵N-labeled oroidin (1d) and optimization of conditions for detection of ¹⁵N-incorporation by microscale FTMS and ¹H-¹⁵N HSQC using a ‘mock’ pulse-labeling experiment with 1d and natural oroidin (1a). We estimate the limits of detection of these methods and determine their suitability for use in field incorporation for biosynthetic studies of complex PIAs with living sponges. This information can be used to answer biosynthetic questions of the biosynthesis of PIAs by sponges in the genera Agelas, Stylissa and Ptilocaulis.

Results and Discussion

Several syntheses of oroidin have been described in the literature and the synthetic approaches utilized can be classified into two categories.^xvii The first involves synthesis of the required 2-aminoimidazole ring by condensation of a guanidine derivative^xviii or cyanamide^xix with an α-haloketone. An alternative strategy involves amination of an imidazole ring via azido or diazonium transfer reagents.^xx We elected the latter strategy for brevity and the utility of the commercially available, starting material urocanic acid, as demonstrated by Lovely.^io,xxi This strategy enabled a late-stage introduction of the ¹⁵N-label which is ideal for cost considerations (Scheme 1). Disconnection of the amide bond leads to the known 4,5-dibromo-2-trichloroacetylpyrrole 8,^xxib,c readily synthesized from commercially available 2-trichloroacetylpyrrole, and the ¹⁵N-labeled allylic amine 9 which could be obtained by a lithiation/azidation sequence at the 2-position of the N-protected urocanic acid ester 10. The ¹⁵N label would be introduced towards the end of the overall sequence by Gabriel synthesis employing the potassium salt of commercially available ¹⁵N-phthalimide and an activated derivative of the alcohol, that in turn, could be obtained by reduction of the carbomethoxy group of ester 10.

Scheme 1.

Retrosynthetic analysis of 7-¹⁵N-oroidin (1d) from urocanic acid.

Fisher esterification of urocanic acid with methanol (Scheme 2) and subsequent protection gave the known N-trityl imidazole derivative 10. ^xxii Reduction of ester 10 (DIBAL-H) afforded the corresponding alcohol 11 in excellent yield. Initial attempts to execute a direct azidation of alcohol 11 at the 2-imidazole position with the unprotected alcohol using known procedures (n-BuLi/ TsN₃),^xxc-e gave none of the desired azido product. Instead, the product obtained was tentatively assigned the structure of a triazene-imidazole derivative based on ¹H NMR and mass spectrometric analysis.^xxiii When the TBS protected imidazole 12 was subjected to the previously employed conditions for azido-transfer, azide 13 was obtained in excellent yield. Deprotection of silyl ether 13 with TBAF afforded alcohol 14, which was then subjected to mesylation (MsCl, Et₃N). Without purification, the mesylate was treated with the potassium salt of ¹⁵N-phthalimide (>98% atom ¹⁵N) to provide the ¹⁵N-labeled pthalimide 15. Cleavage of phthalimide 15 with NH₂NH₂ furnished the ¹⁵N-allylic amine 9 which was acylated with 4,5-dibromo-2-trichloroacetylpyrrole (8) to provide ¹⁵N-pyrrolecarboxamide 16 in good overall yield. Removal of the trityl group from 16 with anhydrous methanolic HCl, generated in situ from acetyl chloride and MeOH, afforded the corresponding 2-azido imidazole^xxc that was hydrogenated in the presence of Lindlar's catalyst to provide 7-¹⁵N-oroidin (1d, >98% atom ¹⁵N).

Scheme 2.

Synthesis of ¹⁵N-labeled oroidin.^a

^aReagents and conditions: (a) AcCl, MeOH, reflux, quant.; (b) TrCl, Et₃N, DMF; (c) DIBAL-H, THF; (d) TBSCl, imidazole, 74% for 3 steps; (e) n-BuLi, TsN₃, THF, 94%; (f) TBAF, THF, 82%; (g) MsCl, Et₃N, THF; (h) potassium phthalimide-¹⁵N, DMF, 42% for 2 steps; (i) NH₂NH₂, EtOH, 50 °C; (j) 4,5-dibromo-2-trichloroacetylpyrrole, Na₂CO₃, DMF, 76% for 2 steps; (k) AcCl, MeOH/EtOAc, 99%; (l) H₂/Pd-Lindlar, MeOH/THF, 84%.

The ¹H, ¹³C and ¹⁵N NMR spectra and mass spectra were fully consistent with the structure and location of the ¹⁵N label as depicted. The FTMS of natural oroidin (1a, Figure 2) shows the expected 1:2:1 pattern for the pseudomolecular ion [M+H]⁺ associated with the ⁷⁹Br and ⁸¹Br isotopomers (m/z 387.93990, Δmmu = –0.41 mmu). The FTMS spectrum of 7-¹⁵N-oroidin (1d) (Figure 2b) shows the same ion pattern, only the distribution is increased by 1 amu (m/z 388.93719, Δmmu = –0.16). Less than 2% of the ¹⁴N isotopomer was detected which is consistent with the reported purity of the ¹⁵N-pthalimide (>98% atom ¹⁵N) employed in the synthesis.

Figure 2.

MALDI FTMS (7T) expansions of the pseudomolecular ion [M+H]⁺ of oroidin for (a) natural 1a (C₁₁H₁₂Br₂¹⁴N₅NO) and (b) 7-¹⁵N-oroidin (1d), C₁₁H₁₂Br₂Br₂¹⁴N₄¹⁵NO, >98% atom ¹⁵N. Mass accuracy = 1ppm.

Two general methods are available for measuring incorporation of the stable isotope ¹⁵N into natural products by pulse labeling – mass spectrometry and NMR. In both techniques, the level of isotopic enrichment from de novo biosynthesis of derived PIAs must exceed the background ¹⁵N present of extant PIAs within the sponge and, for MS in particular, ¹⁵N background contributions from other nitrogen atoms in the molecule. As a stable isotope label, ¹⁵N is preferable to ¹³C because the natural abundance of the heavier to lighter isotope for N (~0.3%) is only about 1/4 of C (~1.1%). ¹⁵N incorporation will, therefore, be easier to detect. Ideally for ¹⁵N incorporation experiments, mass spectrometry should differentiate the contributions to the M+1 peak from ¹⁵N and ¹³C,^xxiv a task best-suited to Fourier transform MS (FTMS) with exceptionally high resolution (7T field, R = 240,000) that should be capable of resolving the nominal M+1 peak of the pseudomolecular ion into individual isotopomer contributions from species containing one ¹³C or one ¹⁵N. Baseline separation of the isotopomers was readily achieved by MALDI FTMS (Figure 3), allowing quantitation of the ¹⁵N isotopomer peak. For oroidin (1a), the critical difference in the mass of ¹²C-¹⁵N containing ions versus the heavier ¹³C-¹⁴N isotopomer amounts to only 0.00632 amu, a baseline separation not readily achievable by high-resolution mass spectrometers with conventional analyzers (e.g. double sector, TOF).

Figure 3.

MALDI FTMS (7T) expansions of m/z and peak height, h, for the ‘M+1’ peak of the pseudomolecular ions of oroidin (C₁₁H₁₁Br₂N₅O). Mass accuracy = 1 ppm (R = 240,000). (a) natural 1a. (b) 1a + 1d (0.04 mol equiv). (c) 1a + 1d (0.10 mol equiv). Peak assignments of [M+H]⁺ in (c): m/z 388.94372, ¹²C₁₀¹³CH₁₂Br₂N₅O⁺; m/z 388.93739; ¹²C₁₁H₁₂Br₂N₄¹⁵NO⁺. The unrelated peak at m/z 388.9305 (indicated with an ‘*’) is tentatively assigned to some loss of H from the dominant ⁷⁹Br/⁸¹Br isotopomer, C₁₁H₁₀⁷⁹Br⁸¹BrN₅O⁺ ([M]⁺, Δmmu = –0.5).

Quantitation of the ¹⁵N isotopomer can be made from integration of ion current intensities of the [M+1] components, ¹⁵N-¹²C and ¹⁴N-¹³C (Figure 3). Assuming the ¹³C abundance in natural and synthetic ¹⁵N-labeled oroidin are identical, the ¹³C peak intensity may be used as the denominator in the ratio h (Equation 1), where h_N and h_C are the respective measured peak heights of the ¹⁵N-¹²C and ¹⁴N-¹³C component ions.

$$h=h_{\mathrm{N}}∕h_{\mathrm{C}}$$ (Eqn 1)

Consequently, h can be used to reflect small changes in the ¹⁵N content.^xxv From the ion current peak heights measured in the FTMS spectrum of natural oroidin (1a), h was found to be 0.1487±0.0175. A prepared ‘mock’ labeled sample (‘Mix1’), consisting of a 1:0.005 mixture of oroidin (1a): 7-¹⁵N-oroidin (1d), showed an enhanced h of 0.1881±0.0135 corresponding to a ~4% ¹⁵N enrichment. A second sample (‘Mix2’, 1:0.100, 1a:1d) gave h = 1.024±0.0369, corresponding to ~88% ¹⁵N enrichment. We estimate that the limit of detection (LOD, ±3 σ) for ¹⁵N incorporation corresponds to an increase in intensity, Δh ~5.2%. Given that natural abundance of ¹⁵N is approximately 0.3% atom, this would amount to observation of as little as 10.4 parts per thousand of ¹⁵N incorporation per N atom of oroidin (1a) by de novo synthesis. Detection of this small amount will, of course, be made more difficult by the presence of unlabeled oroidin (1a) within the sponge.

The disadvantage of analysis of ¹⁵N-content in oroidin and other PIAs by MS is the presence of at least five N atoms, each contributing to the intensity of ¹⁵N-¹²C component of the M+1 peak. Consequently, it may be expected the actual LOD by FTMS will be higher.

Direct detection of ¹⁵N incorporation by ¹⁵N NMR spectroscopy is impractical due to low natural abundance and severely compromised sensitivity of this I= ½ nucleus which possesses a low-magnitude, negative magnetogyric ratio (γ = – 4.31726570 MHz T^–1).^xxvi Conveniently, ¹H-¹⁵N HSQC circumvents most of these problems by indirect detection of bonded ¹H-¹⁵N couplets through magnetization transfer from ¹⁵N and observation of the more sensitive ¹H nucleus. Because four of the five N atoms in oroidin (1a) are bonded to H, 2D ¹H-¹⁵N HSQC can provide bond-specific information about fractional enrichment at an individual N atom in oroidin and, by extension, more complex PIAs by measurement of the ¹H-¹⁵N couplet intensity. The intensity of cross peaks in the 2D ¹H-¹⁵N HSQC spectrum of oroidin will vary for each N-H couplet depending on efficiency of magnetization transfer that largely depends upon the magnitude of the ¹J_NH scalar coupling constant. Nevertheless, the intensity of each crosspeak will scale linearly as a function of ¹⁵N abundance.

When put to practice, it is easier and more meaningful to measure ¹⁵N enrichment from the volume integration of cross peaks of 2D HSQC spectra or – best of all – from the 1D version of the HSQC experiment where evolution of the F1 dimension (¹⁵N chemical shift) is eliminated and acquisition time is devoted to improving S/N in the detected dimension F2 (¹H chemical shift). We define here ΔI as the difference of cross peak intensity, I, of the ¹⁵N-labeled N-H couplet of interest (in this case, the NH(C=O) signal of 1d) in a sample of oroidin (1a), spiked with a known amount of 7-¹⁵N-oroidin (1d), compared to the intensity, I₀, of oroidin recorded under the same conditions (Equation 2). The values of I are normalized to the intensity of the pyrrole N-H couplet of oroidin, I_py, in each experiment.

$$\Delta I=(I-I_0)∕I_0$$ (Eqn. 2)

We elected to use the pyrrole N-H couplet of oroidin (1a) (δ_H = 13.14, s; δ_N169.1) as the control because of its ready assignment from ¹H-¹⁵N HMBC, lack of tautomerism, and likely origination in a different biosynthetic pathway unaffected by biosynthetic incorporation of ¹⁵N into the amino-imidazole fragment (see Discussion).

The exquisite mass-sensitivity of the triple-tuned {¹³C,¹⁵N}¹H 1.7 mm 600 MHz microcryoprobe was used to full advantage in this work for characterization of the ¹⁵N content of oroidin samples. This low-volume, high-mass sensitivity cryoprobe allowed measurements of samples of only ~1.5 mg in 30 μL (acquisition times, ~4 h). Figure 4 shows the 2D ¹H-¹⁵N HSQC experiment for oroidin (1a) with respective N assignments (also, see the Experimental for accurate δ's). A single dominant cross peak was observed in the 2D ¹H-¹⁵N HSQC of 7-¹⁵N-oroidin (1d) (not shown), confirming the location of the ¹⁵N label at the amide group N-H (δ_H = 8.75, t, ³J_HH = 6.0 Hz, H7; δ_N = 107.8, Ν7), as expected.

Figure 4.

2D ¹H-¹⁵N HSQC (600 MHz, 1:1 DMSO-d₆/benzene-d₆) of natural oroidin (1a, depicted tautomer is arbitrary). N locants are labeled following the numbering scheme of Assman and Köck (Ref.^xxxii). ns = 16, T2, T1 = 2K × 256; F2, F1 = 2K × 1K; d1 =1.5s

Figure 5a-c shows 1D ¹H-¹⁵N HSQC experiments of oroidin (1a) spiked with different amounts of 7-¹⁵N-oroidin (1d) and the integrals of each N-H couplet, normalized to the N1-pyrrole couplet are reported in Table 1 (note that the integrals are not exactly proportional to the stoichiometric ratios of attached H's). As predicted, the value of ΔI increases linearly with the amount of added 7-¹⁵N-oroidin (1d) and the slope of this ‘standards addition’ linear regression (Figure 6) gives the proportionality constant c = 0.34 for ΔI per μg of added 1d. It should be noted that this value will be highly dependent upon the pulse sequence parameters and conditions used for NMR acquisition.

Figure 5.

1D ¹H-¹⁵N HSQC experiments (¹H δ, 1.7 mm microcryoprobe, 600 MHz) of 1a with N-H assignments. Labels indicate N atom assignments made from a separate ¹H-¹⁵N HMBC experiment (not shown). (a) 1500 μg of 1a, no added [¹⁵N]-1d. (b) 1500 μg 1a + 10 μg 1d. (c) 1500 μg 1a + 5.0 μg 1d. (d) 1500 μg 1a + 2.5 μg. Relaxation delay, d1 = 1.50 s, optimized for ¹J_N-H = 90 Hz; ¹H π/2 pulse = 12 μS, ¹⁵N π/2 pulse = 34 μS; NA = 6K; dummy scans = 16; NI = 1; T2 = F2= 8K points (no zero fill). See Supporting Information for calculations of limit of detection (LOD), mean and SD.

Table 1.

Integrals from ¹H-¹⁵N 1D HSQC of natural oroidin (1a, 1.5 mg/ 30 μL) and 1a 'spiked ' with 1d. Integrals of H-N couplets corresponding to N-12, N-14 and N-16 vary due to tautomerism.

Aliquot μg 1d	N-12	N-1	N-14	N-7	N-16
0.0	1.26	1.00	1.05	2.17	2.14
2.5	1.03	1.00	0.48	2.57	2.23
5.0	1.65	1.00	1.18	3.83	2.55
10.0	0.63	1.00	0.00	5.74	3.02

Figure 6.

Linear regression of ¹H-¹⁵N HSQC cross-peak integral ratio, (I–I₀)/I₀ for the NH(C=O) signal in ‘mock’ pulse labeling experiments with natural abundance 1a (1500 μg, I₀) spiked with measured aliquots of 1d (μg, ‘x-scale'). Observed values, I, for the NH-C=O ¹H-¹⁵N couplet are normalized to the intensity of I_py the pyrrole N-H cross peak. For the NH(C=O) peak in the natural abundance sample (no added 1d), I = I₀ and I₀/I_py = 1.00. Error bars are ± SD.

Constant c will be useful for estimation of absolute %N incorporations in PIAs labeled during in-field incorporation experiments of secondary metabolism within living sponges. Although it is advisable to measure the relative HSQC intensities of pyrrole N-H couplet and NH(C=O) couplets of the specific PIA in question, it may be reasonably expected that the magnetic properties of these couplets in more complex PIAs are comparable to those of oroidin (1a). We estimate the limit of detection (LOD) for ¹⁵N enrichment by HSQC in 1a to be approximately 1.5 parts per thousand (based on ±3 SD). This is more sensitive than the LOD of ¹⁵N-enrichment by FTMS, the reason being HSQC detection of couplet ¹⁵N-¹H at each N atom and eliminates background ‘dilution’ by other ¹⁵N's, however the precision is inferior to FTMS replicate measurements.

Biosynthetic Considerations

Two phases of PIA biosynthesis can be considered: the assembly of oroidin (1a), and downstream conversion of the latter to more complex PIAs. Despite much speculation, on the biosynthesis of PIAs, only one experimental study in living marine sponges has appeared to date. The precursor for the imidazole side chain is probably an amino acid and both histidine and ornithine have been proposed.^xxvii Kerr and co-workers found that cultured cells of Teichaxinella morchella (Styllisa caribica^xxviii) produced ¹⁴C-labeled stevensine (17),^xxix the simplest monomeric PIA related to oroidin (1a), upon exposure to U-¹⁴C-labeled histidine or C5-¹⁴C-ornithine, but not U-¹⁴C-arginine (Figure 7).^xxvii Paradoxically, U-¹⁴C-histidine was incorporated into 17, but this would require an additional C₁ unit from an unidentified source to complete the imidazole ring and a non-intuitive substitution of NH₂ at C-2 of the imidazole ring. No simple mechanism for this scenario presents itself. Alternatively, citrulline from the N-carbamoylation of ornithine, may be the precursor to the side chain of oroidin; however ornithine is also a precursor for the biosynthesis of the pyrrole group; this hypothesis would not be resolved by labeled ornithine incorporation experiments. Ornithine, if it is the precursor to oroidin (1a), must undergo decarboxylation, amide coupling with a suitable pyrrole carboxylate group, N-carbamoylation to citrulline, oxidative N-insertion into the side chain and aromatization of the heterocyclic product to give the nascent 2-aminoimidazole ring (Figure 7).

Figure 7.

Possible biosynthesis of oroidin (1a) and conversion to stevensine (17)

These uncertainties in the origin of oroidin (1a) are, in theory, addressable by suitable ¹⁵Nlabeled amino acid incorporation experiments. Unfortunately, low levels of radioactivity and specific incorporation levels were observed with amino acids in the Kerr experiment ([U-¹⁴C]-His, 1460 dpm (0.026%) and [U-¹⁴C]-ornithine, 1300 dpm (0.024 %), respectively^xxvii), presumably, a consequence of the low numbers of cultured cells used (5 x 10⁷) and higher metabolic mobilization and dilution of ¹⁴C-labeled amino acids into other committed pathways of primary metabolism. Consequently, detection of ¹⁵N incorporation into PIAs using ¹⁵N-labeled amino acids may be precluded by the limits of detection. For the above reasons, late-stage precursors (e.g. 7-¹⁵N-oroidin (1d), sceptrin (2) and ageliferin (3) are attractive for mapping the origin of downstream polycyclic PIAs since these molecules should be committed intermediates. Measurement of their rates of incorporation into PIAs in living sponges of the genera Agelas, Stylissa and Ptilocaulis may also provide critical data on the rates of marine alkaloid secondary metabolism for which there is no current information.

The use of precursors labeled with the spin I=1/2 stable isotope ¹⁵N is particularly attractive as it provides bond-specific information regarding the flow of intermediates in biosynthesis of complex PIAs. Judicious design and placement of the label in ¹⁵N-labeled oroidin is important. Because the primary NH₂ group of the aminoimidazole group is potentially metabolically labile through deamination reactions, we chose to label the amide bond in ¹⁵N-oroidin. Quantitation of ¹⁵N incorporation into proteins has been made from integration of cross peak volumes of 2D HSQC experiments;^xxx however, in the present work with natural products we show that it is far simpler to integrate the corresponding 1D HSQC spectra.

Finally, although the ratio of sceptrin to ageliferin found in one sample of Agelas conifera is ~10:1,^xxxi the ratios of mono-brominated and dibrominated analogs of sceptrin and ageliferin vary slightly from approximately 1:1 ~ 3:4; however, these data do not clearly indicate an ordering for a bromination-dimerization sequence in the biosynthesis of dimeric PIAs. Propagation of 7-¹⁵N-oroidin (1d) along the biosynthetic pathway to complex PIAs will carry the ¹⁵N label through symmetrical and non-symmetrical intermediates and detection of residual ¹⁵N in strategically chosen N-H bonds in the molecule should reveal timing, ordering and molecularity of the C-C bond-forming events and the extent of coupling of de novo enzymatic reactions. HPLC analysis of PIA-containing extracts from A. conifera (Agelasidae), collected in the Florida Keys by Köck and coworkers,^xxxi show the presence of sceptrin (2) and ageliferin (3), however, some specimens of Agelas sceptrum collected by us contained sceptrin (2) but were devoid of the expected precursors 1a-c. If 1a-c are precursors to other PIAs, as is highly probable, this implies that the rates of conversion of 1a-c to higher-order PIAs are highly dependent upon the sponge species, their biosynthetic capacities, and possibly the habitats they occupy. 7-¹⁵N- Oroidin (1d) should find great value in systematic studies of the time-dependent biosynthesis of PIA natural products by pulse labeling in live sponges to answer some of these questions.

Conclusion

A concise, gram-scale synthesis of ¹⁵N-labeled oroidin from commercially available and inexpensive urocanic acid has been achieved. ¹⁵N-Oroidin, or its debromoanalogs, may be utilized in biosynthetic feeding experiments to elucidate likely intermediates in the biosynthesis of complex members of the pyrrole-imidazole marine alkaloid family, including palau'amine and axinellamine. FTMS detection of ¹⁵N offers unparalleled sensitivity (picomole) and ¹H-¹⁵N HSQC provides bond-specific information. The utility of ¹H-¹⁵N HSQC experiments for detection of site-specific ¹⁵N incorporation in a mock pulse-labeling experiment has been demonstrated. Several advantages of ¹⁵N pulse labeling of biosynthetic pathways are apparent, especially for in-field experiments carried out with living marine invertebrates. The ¹⁵N NMR chemical shift was exploited through indirect ¹H-¹⁵N HSQC to afford enhancement of otherwise weak ¹⁵N signals by polarization transfer through N-H couplets. ¹⁵N incorporation was further amplified by the enhanced mass-sensitivity of the triple-tuned ¹H{¹³C,¹⁵N}-600 MHz 1.7 mm microcryoprobe. Biosynthetic studies with alkaloids bearing N-H groups that are not complicated by exchange or tautomerism afford the best cases for application of the technique. In principle, even alkaloids lacking N-H couplets (tertiary N compounds) are amenable to study by detection of long-range ¹H-¹⁵N couplets from HMBC experiments. The use of ¹⁵N stable isotope avoids more complex logistics associated with radioactive tracers (e.g. ¹⁴C, ³H) that often preclude laboratory operations aboard seagoing research vessels. The results of field incorporation experiments with 7-¹⁵N-oroidin (1d) will be reported in due course.

Experimental Section

General Experimental Procedures

IR spectra were recorded on thin films using a Jasco 4100 FTIR by attenuated total reflectance (ATR) on samples mounted on a 3 mm ZnSe plate. ¹H and ¹³C NMR spectra were recorded on a Bruker DRX 600 MHz equipped with 1.7-mm {¹³C,¹⁵N}¹H 600 MHz CPTCI microcryoprobe, JEOL 500 MHz spectrometer equipped a 5 mm {¹³C }¹H probe or Varian 400 and 500 MHz spectrometers equipped with 5 mm {¹³C}¹H room temperature probe and 5 mm {¹H}¹³C cryoprobe, respectively. NMR spectra are referenced to residual solvent signals (CHCl₃, ¹H, δ 7.26 ppm; CDCl₃, ¹³C, δ 77.16 ppm) or the internal offset for ¹⁵N assigned by the instrument manufacturer (Bruker). HRMS measurements of synthetic compounds were made at the Laboratory for Biological Mass Spectrometry (Texas A&M University), Scripps Research Institute (TOF-MS) or University of California, San Diego (EI-MS) mass spectrometry facilities. MALDI FTMS were recorded on a 7T Bruker Apex QFTMS. LCMS was carried out on a ThermoFisher Accela UHPLC coupled to an MSQ single quadrupole mass spectrometer operating in positive ion mode, unless otherwise stated. Semi-preparative HPLC was carried out on a Varian SD200 system equipped with a dual-pump and UV-1 UV detector under specified conditions. All solvents for purification were of HPLC grade. See Supporting Information for other procedures.

Synthesis of 7-¹⁵N-Oroidin (1d)

To a solution of amide 16 (380 mg, 0.58 mmol) in 20 mL of EtOAc was added MeOH (1.17 mL, 28.9 mmol) followed by dropwise addition of AcCl (2.06 mL, 28.9 mmol) (to generate HCl) at 0 °C. The solution was stirred for 1.5 h at 25 °C and during that time a precipitate formed. The reaction mixture was filtered to separate the solid product of azidoimidazole hydrochloride salt (240 mg, 99%), which was of sufficient purity for the next reaction: IR (thin film) 3115, 233, 2150, 1611, 1555, 1439 cm⁻¹; ¹H NMR (500 MHz, CD₃OD) δ4.11 (d, J = 5 Hz, 2H), 6.35 (dt, J = 16, 3.5 Hz, 1H), 6.43 (d, J = 16 Hz, 1H), 6.89 (d, J = 2 Hz, 1H), 7.31 (s, 1H); ¹³C NMR (125 MHz, CD₃OD) δ42.3 (d, J_C-N = 114 Hz), 100.8, 107.3, 115.2, 116.3, 117.5 (d, J_C-N = 22.5 Hz), 129.4 (d, J_C-N = 106 Hz), 131.8, 132.3, 142.6, 162.4 (d, J_C-N = 174 Hz); HRESIMS m/z 416.9171 [M+H]⁺ (calcd for C₁₁H₁₂Br₂N ¹⁵₆NO, 416.9163).

To a solution of the azidoimidazole hydrochloride salt (210 mg, 0.50 mmol) in 30 mL of MeOH/EtOAc (1 : 1) was added Lindlar's catalyst (100 mg, 5% Pd on CaCO₃, poisoned with Pb). The resulting mixture was stirred under atmospheric hydrogen at 25 °C for 6 h then filtered with MeOH through a pad of Celite. The filtrate was concentrated under reduced pressure to leave a residue, which was purified by flash column chromatography on silica gel (gradient elution: 10:1:0 → 1:1:0.01 CH₂Cl₂/MeOH/NH₄OH) to furnish 7-¹⁵N-oroidin (1d) (165 mg, 84%): IR (thin film): 3304, 2363, 2339, 1620, 1507, 1415 cm⁻¹; ¹H NMR (500 MHz, CD₃OD) δ 4.00 (d, J = 6 Hz, 2H), 5.87 (dt, ³J_NH = 16 Hz, ³J_HH= 6 Hz, 1H), 6.30 (d, J = 16 Hz, 1 H), 6.47 (s, 1H), 6.83 (d, J = 1 Hz, 1H); ¹³C NMR (125 MHz, CD₃OD) δ 43.1 (d, ¹J_C-N = 106 Hz), 100.7, 107.0, 115.1, 118.5, 122.2, 123.6 (d, ²J_C-N = 15 Hz), 129.7 (d, ¹J_C-N = 99 Hz), 132.2, 152.8, 162.4 (d, ¹J_C-N = 175 Hz); HRESIMS m/z 390.9289 [M+H]⁺ (calcd for C₁₁H₁₂Br₂N₄¹⁵NO, 390.9258).

See Supporting Information for additional procedures.

Isolation of Natural Oroidin (1a) from Stylissa caribica

A sample of Stylissa caribica (wet wt. 24.5 g) was collected in the Bahamas (June 2008) and extracted with 1:1 CH₂Cl₂/MeOH (rt, overnight, 2 × 500 mL). The solvent was removed under reduced pressure and the residue partitioned between hexanes (3 × 250 mL) and 9:1 MeOH/H₂O (250 mL). The aqueous MeOH layer was concentrated and further partitioned between n-BuOH (3 × 250 mL) and H₂O (250 mL). Removal of the solvent from the upper layer gave an n-BuOH extract (787 mg), a portion of which (191 mg) was adsorbed onto a solid phase extraction cartridge (C₁₈) and eluted with 4:1 CH₃CN/H₂O, followed by THF. The first-eluting fraction (116 mg) was separated by gradient HPLC (C₁₈, 10:90 to 60:40 CH₃CN/H₂O + 0.1% TFA) to give oroidin 1a (8.8 mg, 1.0% wt weight).

MALDI FTMS of Oroidin (1a) and 7-¹⁵N-Oroidin (1d)

Matrix solution (1 μL of saturated α-cyano-4-hydroxycinnamic acid (CHCA) solution in 50:50:0.2 MeOH:H₂O:TFA) was added to 1 μL of the sample solution (1 pmol/μL) and the sample (approximately 1 pmol) deposited onto a stainless steel target. For each sample, 200 shots were acquired for each mass spectrum (acquisition time, ~1 minute each), and 24 spectra were co-added. Data were processed using Bruker software and error analysis was carried out using standard deviation of replicate MS measurements (N≥5).

1D ¹H-¹⁵N HSQC Experiments

Samples for mixing experiments were prepared by addition of measured aliquots of a standard solution of 1d (1.25 mg/10 mL MeOH) to a sample of 1a (1500 μg). The samples were concentrated by removal of solvent (MeOH + 0.05% TFA), and redissolved in 1:1 DMSO-d₆/benzene-d₆ (35 μL) before transferring ~30 μL to a NMR tube (ϕ = 1.7 mm) using a gas-tight syringe. 1D ¹H-¹⁵N HSQC NMR spectra were acquired with 8K data points in T2 (no zero-fill) and optimized for ¹J_N-H = 90 Hz. Each experiment was preceded by 16 dummy scans and ~6K scans were collected. The corresponding ¹H-¹⁵N integrals were averaged and normalized against the N1 pyrrole couplet of 1 (I_P = 1.000). See Table 1 (numbering system follows that of Assman and Köck^xxxii). ¹H-¹⁵N assignments: δ_H, δ_N (1:1 DMSO-d₆/C₆D₆): 13.14, 169.1 (N1); 8.75, 107.8 (N7); 13.54, 134.7 (N12); 12.78, 137.7 (N14); 8.21, 62.2 (N16).

Figure 1.

Oroidin (1a) and other representatives of the pyrrole-imidazole alkaloid (PIA) family from marine sponges.