Untargeted Identification of Alkyne Containing Natural Products using Ruthenium Catalyzed Azide Alkyne Cycloaddition Reactions Coupled to LC-MS/MS

Abstract

Alkyne containing natural products have been identified from plants, insects, algae, fungi, and bacteria. This class of natural products have been characterized as having a variety of biological activities. Polyynes are a sub-class of acetylenic natural products that contain conjugated alkynes and are underrepresented in natural product databases due to the fact that they decompose during purification. Here we report a workflow that utilizes alkyne azide cycloaddition (AAC) reactions followed by LC-MS/MS analysis to identify acetylenic natural products. In this report, we demonstrate that alkyne azide cycloaddition reactions with p-bromobenzyl azide result in p-bromobenzyl substituted triazole products that fragment to a common brominated tropylium ion. We were able to identify a synthetic alkyne spiked into the extract of Anabaena sp. PCC 7120 at a concentration of 10 μg/ml after optimization of MS/MS conditions. We then successfully identified the known natural product fischerellin A in the extract of Fischerella muscicola PCC 9339. Lastly, we identified the recently identified natural products protegenins A and C from Pseudomonas protegens Pf-5 through a combination of genome mining and RuAAC reactions. This is the first report of RuAAC reactions to detect acetylenic natural products. We also compare CuAAC and RuAAC reactions and find that CuAAC reactions produce less by-products compared to RuAAC but is limited to terminal alkyne containing compounds. In contrast, RuAAC is capable of identification of both terminal and internal acetylenic natural products, but by-products need to be eliminated from analysis by creation of an exclusion list. We believe that both CuAAC and RuAAC reactions coupled to LC-MS/MS represent a method for the untargeted identification of acetylenic natural products, but each method has strengths and weaknesses.

Graphical Abstract

Introduction

Acetylenic natural products have been identified in extracts of plants, insects, algae, fungi, and bacteria. One of the first acetylenic natural products isolated was dehydromatricaria ester from Artemisia sp., in 1826.¹ The majority of previously described acetylenic natural products have been identified in plants, fungi, algae, insects, and marine invertebrates including lentinamycin (1), an acetylenic natural product displaying antibiotic properties produced by Lentinula edodes (shiitake mushroom), falcarinol (2) an alkyne natural product displaying anticancer activity from Daucus carota (wild carrot), (Z)-13-hexadecen-11-ynoic acid (3), a sex pheromone from Thaumetopoea pityocampa (pine processionary moth), and the cytotoxic petrosynoic acids isolated from a marine sponge (Petrosia sp.).^{2, 3}

Comparatively, acetylenic natural products characterized from bacteria are seemingly less prevalent.⁴ Many of the currently described bacterial acetylenic natural products, such as the cyanobacterial natural products jamaicamide B (5),⁵ anaephene B (6),⁶ and georgamide (7),⁷ feature a single terminal alkyne moiety. Much less common are bacterial conjugated polyacetylenic natural products (polyynes), as to date only a handful have been isolated and structurally characterized.⁸ The low frequency of discovered polyynes can be attributed to the fact that some polyynes have been described to undergo decomposition during purification and characterization of these compounds is difficult.⁸ This lead us to the hypothesis that they may be underrepresented in databases of characterized natural products, which is supported by the prevalence of proteins putatively involved in the biosynthesis of alkynes in bacterial genomes.⁹ Recently, copper catalyzed azide-alkyne cycloaddition reactions (CuAAC), also referred to as copper click chemistry, have been employed to stabilize polyynes containing a terminal alkyne moiety by generating a 1,4-disubstituded triazole from the coupling of the terminal alkyne moiety with an azide. The cycloadducts are typically more stable than the unreacted polyynes, which allows isolation and structural characterization of the resulting 1,2,3-triazole.⁴ This approach was recently utilized for isolation and structure determination of caryoynencin A (8),⁴ collimonin A (9),⁸ and related compounds. CuAAC has also been utilized to aid in the initial discovery of acetylenic natural products from crude extracts by coupling CuAAC with fluorescence or LC-MS based analyses.¹⁰

While CuAAC has undoubtedly aided in the study and identification of several acetylenic natural products, it neglects a subset of acetylenic natural products, those containing internal alkyne moieties as these compounds do not undergo a copper catalyzed cycloaddition. Examples of natural products containing internal alkynes include fischerellin A (10),¹¹ microcarbonin A (11),¹² and ergoyne A (12).¹³

Ruthenium mediated azide-alkyne cycloaddition reactions (RuAAC) were first described in 2005¹⁴ and can catalyze the cycloaddition of azides with both terminal and internal alkynes to generate the 1,5-disubstituted and 1,4,5-trisubstituted triazoles respectively (Fig. 1) in contrast to the 1,4-disubstituted-1,2,3-triazole product formed during CuAAC reactions. While RuAAC has a broader scope of substrate reactivity, it has not been previously utilized in natural products discovery, most likely due to byproducts produced under reaction conditions.¹⁵

Figure 1.

Catalytic mechanisms of CuAAC (left) and RuAAC (right) reactions.

In this paper, we report a method utilizing RuAAC for the untargeted discovery of acetylenic natural products from unfractionated bacterial extracts, which relies solely on the unique fragmentation pattern of a p-bromobenzyl triazole. As a proof of concept, both fischerellin A and B were located in the unfractionated extract of the cyanobacterium Fischerella muscicola PCC 9339 using this method. Coupling the RuAAC with the genome mining of acetylenases allowed us to develop a workflow to identify acetylenic natural products that we describe in the manuscript. Through genome mining approaches followed by the described RuAAC method, two acetylenic natural products produced by the plant protective bacterium Pseudomonas protegens Pf-5 were identified, which have recently been reported by the Kai group as protegenins A and C.¹⁶ This discovery demonstrates the advantage of utilizing RuAAC as a powerful tool to screen unfractionated extracts for acetylenic natural products discovery.

Results and discussion

Evaluating alkyne reactivity and MS/MS fragmentation.

As some acetylenic natural products contain conjugated alkyne moieties (either in conjugation with an alkene or alkyne) and the reactivity of conjugated alkynes has not been extensively evaluated with RuAAC, we synthesized a small library of conjugated alkynes for testing (Fig. 2A–B and Supplementary Figure 1). The synthetic compounds with internal diyne moieties were then subject to a RuAAC reaction containing p-bromobenzyl azide and Cp*RuCl(PPh₃)₂ in 1,4-dioxane using a procedure based on that described by Boren et al.¹⁷ The catalyst Cp*RuCl(PPh₃)₂ was chosen as it has been shown to be one of the more stable ruthenium catalysts in aerobic environments.¹⁷ 1,4-Dioxane was used as the solvent as previous research has shown success with employing 1,4-dioxane as the solvent with the Cp*RuCl(PPh₃)₂ catalyst.¹⁷ We chose p-bromobenzyl azide as the azide for three reasons. First, the bromine atom is easily traceable via mass spectrometry due to the characteristic isotope pattern. Second, we anticipated that the p-bromobenzyl moiety would result in a common fragment produced from the resulting cycloadducts (vide infra). Third, we envisioned that the p-bromobenzyl group would aid in crystallization of the cycloadducts in later steps. With all of our synthetic alkynes we observed the formation of a triazole and the resulting cycloadducts were purified, structurally characterized, and subjected to MS/MS analysis. The synthetic triazoles all showed a consistent fragmentation pattern to yield the brominated tropylium ion ([M]⁺ of 168.96/170.96) as well as the non-brominated tropylium ion with an [M]⁺ of 90.04 (Fig. 2A–C and Supplementary Figure 1). Since the 168.96/170.96 signals were present in all MS/MS spectra with the highest abundance, we reasoned that this fragment ion would serve as an excellent reporter ion for the detection of derivatized alkynyl natural products in extracts.

Figure 2. Cycloadducts formed during RuAAC have a common fragmentation pattern.

A. MS/MS fragmentation of synthetic 13 showing the common fragments of 168.96, 170.96, and 90.04; B. MS/MS of synthetic 14 showing the common fragments of 168.96, 170.96, and 90.04; C. Proposed fragmentation and structures of the observed common fragments; D. Extracted ion chromatogram (390.0–390.2 m/z) of the RuAAC reaction containing diphenylacetylene, p-bromobenzyl azide and Cp*RuCl(PPh₃)₂ in redissolved Anabaena 7120 extract in 1,4-dioxane showing the presence of the cycloadduct 13; E. Extracted ion chromatogram (390.0–390.2 m/z) of the RuAAC reaction containing diphenylacetylene and Cp*RuCl(PPh₃)₂ in redissolved Anabaena 7120 extract in 1,4-dioxane showing the absence of the cycloadduct 13; F. HRESIMS of compound 13 seen in panel D.

Detecting standard compounds in an extract.

We then tested to see if the Cp*RuCl(PPh₃)₂ catalyst was compatible for RuAAC reactions in 60% aq. methanol extracts of cyanobacterial cells. We chose to use Anabaena sp. PCC 7120 as our organism to generate our test extract for the following reasons. Anabaena sp. PCC 7120 is a fairly fast-growing cyanobacterium, and it is not known to produce acetylenic natural products. It also contains chlorophyll a, carotenoids, and other typical cellular compounds that could poison the catalyst or complicate downstream LC-MS/MS analysis. We extracted a lyophilized cell mass of Anabaena sp. PCC 7120 with 60% aq. methanol. The particulate was removed via filtration and the extract was concentrated to dryness. The resulting dried material was resolubilized in 1,4-dioxane and diphenyl acetylene was spiked into the extract at a concentration of 10 μg/ml. To limit substrate usage, by-product formation, and mass spectrometry contamination, one-tenth of the amount of p-bromobenzyl azide from the previous RuAAC reactions was used as it was still thought to be present in the reaction mixture in excess allowing for derivatization of the diphenyl acetylene. The Cp*RuCl(PPh₃)₂ catalyst was added at 4 mol% of the p-bromobenzyl azide as it was easier to measure at the reduced p-bromobenzyl azide concentration. With these conditions, compound 13 was clearly observed in the RuAAC reaction mixture using an extracted ion chromatogram (Fig. 2D) but was not observed in any of the control reactions lacking catalyst, azide, or alkyne (Fig. 2E and Supplementary Figure 2). Examination of the extracted ion chromatograms (179.0700–179.1000) of the full reaction and the reaction lacking p-bromobenzyl azide showed that the majority of diphenylacetylene was consumed in the full reaction (Supplementary Figure 3). This suggested that RuAAC could be adapted for the discovery of acetylenic natural products in extracts without fractionation or purification.

Detection of known acetylenic natural products in extracts.

With our test case successful, we then turned our attention to the detection of known natural products. Extraction of the cyanobacterium Fischerella muscicola PCC 9339, which is a known producer of the internal diyne-ene containing secondary metabolites fischerellin A¹¹ and B,¹⁸ and hapalindole-like compounds,¹⁹ was performed with 60% aq. methanol as this has been reported to extract both fischerellins and the hapalindole-like compounds. After concentration to dryness, the extract was solubilized in 1,4-dioxane and the RuAAC reaction was performed as described above. However, the reaction was performed in scintillation vials without stirring in a heat block. This was done to determine if the RuAAC reaction could potentially be used to derivatize multiple extracts simultaneously with minimal space. Following the RuAAC derivatization of the extract, LC-MS/MS data was collected using the AutoMS/MS setting (data dependent MS/MS) for both the RuAAC reaction and the underivatized extract. We were able to observe the formation of the expected triazole product (15, Fig. 3) in the RuAAC reaction, but not in the control reactions (Supplementary Figure 4), and we could not detect fischerellin A in the extract after the full RuAAC reaction (Supplementary Figure 5). However, in the initial LC-MS/MS run, 15 was not selected by the computer for fragmentation. We hypothesized that the intensity of 15 was never one of the top five (5) ions in the MS (one of our program parameters) and was therefore not chosen for fragmentation. We reasoned that the generation of an exclusion list would aid in the detection of lower abundance cycloadduct products.

Figure 3. Identification of fischerellin in the extract of F. muscicola PCC 9339.

A. RuAAC reaction with fischerellin A and the expected product; B. (i) Extracted ion chromatogram for precursor ions that generate fragment ions between 168.8–170.9 m/z (the common fragment previously observed) in an RuAAC reaction containing Fischerella muscicola PCC 9339 extract, Cp*RuCl(PPh₃)₂, and p-bromobenzyl azide; (ii) Extracted ion chromatogram for the fischerellin A cycloadduct in an RuAAC reaction containing Fischerella muscicola PCC 9339 extract, and Cp*RuCl(PPh₃)₂. C. HRESIMS spectrum showing the [M+H]⁺ and the [M+Na]⁺ for the fischerellin A cycloadduct in the reaction mixture; D. MS/MS fragmentation of the protonated ion 620.25 m/z (isolation width 4 m/z) showing the fragmentation pattern of 15.

We examined the chromatogram produced by identifying the precursor ions that generated the brominated tropylium fragment. We noted the presence of multiple brominated compounds throughout the chromatogram (Supplementary Figure 5a). Based on the exact masses we suspected that these compounds were produced by off-pathway reactions of p-bromobenzyl azide with the ruthenium catalyst. To optimize our workflow and triazole product formed we tested various solvents with diphenylacetylene as the substrate (Supplementary Figure 6). We note that 1,4-dioxane, dimethylformamide and acetonitrile resulted in similar levels of 15 formed. We chose 1,4-dioxane as our preferred solvent because (1) it forms an azeotrope with water, allowing for water removal if necessary; (2) the boiling point (101 °C) allows evaporation at a slow enough rate that concentrating the extract to dryness can be avoided, as concentration to dryness has resulted in the degradation of polyynes; and (3) 1,4-dioxane is more easily removed compared to dimethylformamide, which allows concentration of the alkyne-azide cycloaddition reactions (when the polyyne is stabilized as the triazole product), allowing the detection of low concentration compounds. To reduce the background of unwanted byproducts of the RuAAC reaction, we examined the effect of diluting the catalyst (Supplementary Figure 7), and reducing the azide concentration (Supplementary Figure 8). However, dilution of p-bromobenzyl azide or catalyst did not produce a lower ratio of unwanted byproduct compared to triazole formation. Efforts were then put into identifying the major byproducts of the reaction and generating an exclusion list so they would be excluded from fragmentation during LC-MS/MS analysis. To do this, the major byproducts were identified and manually added to an exclusion list. The derivatized extract was re-analyzed by LC-MS/MS using the manually curated exclusion list of the major byproducts. However, the mass corresponding to the fischerellin A cycloadduct was still not fragmented during the analysis. This posed a major problem as there were too many byproducts yielding the signature ion fragment during LC-MS/MS fragmentation to manually identify.

We then utilized iterative MS/MS to analyze our samples. Iterative MS/MS is an integrative feature of the Agilent MS operating software but was originally used for proteomics workflows to increase coverage.²⁰ Iterative MS/MS automatically adds ions selected for fragmentation to an exclusion list for subsequent analyses. In this way an operator can delve deeper into the data and fragment molecular ions of lower abundance in subsequent runs with minimal operator involvement thereby increasing automation, while simultaneously performing fragmentation over the entire elution period. Using iterative MS/MS, 15 was successfully fragmented in the second iteration of the derivatized extract producing an MS/MS spectrum for 15. However, the peak for 15 was not easily identifiable in the EIC of precursor ions that generated the brominated tropylium fragment (Supplementary Fig. 9). While obtaining an MS/MS spectrum for the fischerellin A cycloadduct was considered a success and indicated that iterative MS/MS was a potential solution, we needed to further optimize our analytical method.

To optimize our workflow, we utilized iterative MS/MS of a reaction containing Cp*RuCl(PPh₃)₂ and p-bromobenzyl azide to generate an exclusion list. An alternate approach would be to utilize standard metabolomics software to identify ions present in the full reaction mixture and absent from the control reactions. The ions present in only the full reaction would then be added to an inclusion list for targeted fragmentation.²¹ Using three successive iterative MS/MS runs of the reaction containing Cp*RuCl(PPh₃)₂ and p-bromobenzyl azide (no F. muscicola extract) generated a large list of byproducts from the reaction that were then programmed into an exclusion list for subsequent runs (Supplementary figure 10 and Supplementary file 2). The RuAAC-derivatized extract of F. muscicola PCC 9339 was analyzed once again using this method in conjunction with celite filtering. Two peaks corresponding to fischerellin A cycloadducts are prominent peaks in the chromatogram (Fig. 4, Supplementary Figure 11–12). We also observed two peaks corresponding to the predicted [M+H]⁺ of the fischerellin B cycloadduct, which had not been observed previously. The multiple peaks for both fischerellin A and B are suspected to be E and Z isomers (Fig. 4 and Supplementary figure 11). We suspect the earlier eluting peaks are the (Z) isomers of fischerellin A and B, which would be consistent with preliminary NMR analysis we have obtained as they have a coupling constant of 10.7 Hz between the alkene proton signals (data not shown). Full structural elucidation of these peaks is currently underway in our laboratory and will be described in a future manuscript. We then investigated which alkyne had been derivatized by isolating (Z)-15 and obtaining ¹H-¹³C HMBC data at 3 and 8 Hz. The observed correlations support the formation of the triazole at the alkyne distal to the pyrroline (dihydropyrrole) ring (Supplementary figure 28). The appearance of the fischerellin B cycloadducts and the E/Z isomers of both fischerellin A and B in our chromatograms indicates that the multiple iterations allowed for more compounds to be identified. The successful identification of both fischerellin A and B and the suspected (Z) isomers demonstrated that this method was ready to screen bacteria with putative alkyne producing biosynthetic gene clusters.

Figure 4. Use of an exclusion list allows the facile detection of RuAAC cycloadducts.

A. Extracted ion chromatogram of precursor ions that produce ions between 168.8–170.9 m/z (the common fragment previously observed) in an RuAAC reaction containing F. muscicola PCC 9339 extract, Cp*RuCl(PPh₃)₂, and p-bromobenzyl azide with celite filtering allows for cycloadducts of fischerellin A (15) and B (16) to be clearly identified. The appearance of multiple peaks is suspected to be E and Z isomers of fischerellin A and B. The chromatogram shown is raw data demonstrating the ease of identification of cycloadduct peaks from the unprocessed chromatogram.; B. F. muscicola PCC 9339 extract showing no cycloadducts; C. HRESIMS spectrum of compound 16 (right peak).

Identification of acetylenic natural products produced by Pseudomonas protegens Pf-5 using RuAAC.

We next identified putative producers of alkynyl natural products through genome mining. As described in previous studies, some acetylenic natural products, such as polyacetylenes, lack a polyketide synthase, non-ribosomal peptide synthetase, and other enzymes characteristic of natural products biosynthesis.⁴ Coupled with the fact that only a handful of polyyne natural product biosynthetic gene clusters (BGCs) have been identified to date makes it difficult to identify putative polyyne producing organisms using antiSMASH.²² Previous studies have been conducted on the biosynthesis of bacterial acetylenic natural products.^{4, 23, 24} These studies have determined that a desaturase is most likely responsible for the formation of the alkyne bond. Along with a desaturase, an acyl carrier protein, and an AMP-dependent ligase are thought to play important roles in the biosynthesis of the alkyne moiety.²³ The sequence of ColB, a desaturase involved in alkyne biosynthesis of the collimonin group of polyynes from the bacterium Collimonas fungivorans Ter6, was used as a BLAST query in the EFI-EST web tool to generate a network of similar enzymes based on sequence homology (Supplementary Figure 13A). In this network, putative desaturase protein sequences from many of the characterized bacterial polyacetylenic producers clustered together (Cluster 2, Supplementary Figure 13A). Within this cluster were also sequences of putative desaturases from various other bacteria in which polyacetylenes have not been characterized, including Pseudomonas protegens Pf-5. The putative polyyne BGC in P. protegens Pf-5 (PFL0261-0268) was shown to be very similar to that of other polyacetylenic gene clusters such as the caryoynencins and collimonins (Supplementary Figure 13B).²⁵ As we have previously researched the secondary metabolites produced by P. protegens Pf-5, we chose this organism as our initial candidate for screening with the described RuAAC-LC-MS/MS method.

Polyynes are known to be unstable, and it was thought that subjecting them to elevated temperatures could cause degradation thereby preventing their detection. Therefore, we tested to see if the RuAAC derivatization could proceed at room temperature with a higher catalyst loading. It was determined that using 0.0044 mmol of Cp*RuCl(PPh₃)₂ was sufficient for catalyzing triazole formation with 0.04 mmol p-bromobenzyl azide and 0.04 mmol diphenyl acetylene at room temperature without stirring (determined by TLC). These conditions were then adopted for screening the extract of Pseudomonas protegens Pf-5. Utilizing the described RuAAC method, with the exclusion list obtained from the Cp*RuCl(PPh₃)₂ and p-bromobenzyl azide iterative run, yielded two prominent peaks with [M+H]⁺ values of 480.1277 and 478.1122 producing the signature brominated tropylium ion as seen in the MS/MS chromatogram (Fig. 5). These masses of the underivatized compounds were calculated by subtracting the mass of the p-bromobenzyl azide tag from the observed masses in the derivatized extract resulting in the calculated [M+H]⁺ values of alkynes as 269.1532 and 267.1377, which corresponded to the molecular formulae of C₁₈H₂₀O₂ and C₁₈H₁₈O₂ respectively. These masses were shown to be present in the underivatized extract (Fig. 5 and Supplementary Figure 14). Closer analysis of the mass spectrum of the earlier eluting compound showed the presence of a co-eluting compound that based on accurate mass of the [M+2+H]⁺ peak is presumed to be triazole product of a compound with the molecular formula C₁₈H₂₂O₂ (obs. 484.1407, calc. 484.1418, 2.2 ppm error). We could not calculate the mass of the presumed [M+H]⁺ peak as the overlap between the triazole products of these two compounds were not resolved on our mass spectrometer resulting in a broad, inaccurate peak at m/z 482.1335.

Figure 5.

Results of RuAAC reaction with Pseudomonas protegens Pf-5 extract. A. Chromatogram of the ions that produced the brominated tropylium [M+H]⁺ fragment shows two prominent peaks, determined to be cycloadducts of protegenins A and C in the extract of P. protegens Pf-5 wild type subjected to RuAAC reaction (trace I), Chromatogram of the ions that produced the brominated tropylium [M+H]⁺ fragment in the underivatized extract of P. protegens Pf-5 wild type (trace II), Chromatogram of the ions that produced the brominated tropylium [M+H]⁺ fragment in the extract of P. protegens Pf-5 LK164 subjected to RuAAC reaction (trace III). B. MS spectrum of the protegenin C cycloadduct peak. This spectrum shows a brominated peak with a mass of 480.1277. Calculated mass of the underivatized natural product [M+H]⁺ (480.1277 – 210.9745 = 269.1532). C. MS spectrum of the protegenin A cycloadduct peak. This spectrum shows a brominated peak with a mass of 478.1122. Calculated mass of the underivatized natural product [M+H]⁺ (478.1122 – 210.9745 =267.1377).

A third peak with an [M+H]⁺ value of 524.1002 was also present in the chromatogram (Fig. 5) corresponding to an underivatized [M+H]⁺ value of 313.1257. In the extract, an [M+H]⁺ value of 313.1254 was present and this mass was diminished in the RuAAC derivatized extract indicating the presence of another alkyne containing natural product from Pseudomonas protegens Pf-5 (Supplemental Figures 15–17). However, the production of this compound was inconsistent, so efforts to characterize the structure were not pursued. To demonstrate that these compounds were the product of the PFL0261-0268 biosynthetic gene cluster, a deletion mutant was constructed. The compounds with [M+H]⁺ values of 269.1532 and 267.1377 were not identified in the extract of the deletion mutant, thereby demonstrating that these compounds were biosynthesized by the BGC (Supplementary Figure 14). However, the compound with the [M+H]⁺ value of 313.1257 was observed in the deletion mutant suggesting that this compound is biosynthesized by another, currently unknown biosynthetic gene cluster in P. protegens Pf-5 (Supplementary Figures 14, 16, and 17). While this manuscript was being finalized, a publication was released reporting the isolation and structure characterization of protegenins A-D (17-20) from P. protegens Cab57 (aka MAFF 212077).¹⁶ These compounds were isolated and their structure determined via bioassay guided fractionation. The compounds identified in this study are most likely protegenin A (m/z 267.1), protegenin C (m/z 269.1) and protegenin D (co-eluting with protegenin C). We did not identify a peak consistent with protegenin B in either the RuAAC reaction or crude extract. The lack of protegenin B can be explained by the fact that it is most likely not a biosynthetic product as the compound was isolated as a racemic mixture.¹⁶ We were unable to identify protegenin D in the crude extract, and it is possible that the lack of signal corresponding was due to the extremely low titers of production of this compound coupled to a low ionization efficiency. A second publication was released characterizing protegenin A and the BGC (pgnA-K which includes PFL_0261-0268).²⁶ However, in this second study, failure to detect protegenins B, C and D is not unexpected based on their low production titer. This demonstrates the effectiveness of utilizing RuAAC for the discovery of acetylenic natural products in an untargeted manner.

CuAAC derivatization of P. protegens extract.

The results from the RuAAC derivatization of the P. protegens Pf-5 extract show the presence of two naturally occurring acetylenic natural products, subsequently named protegenins A and C and a third currently unidentified compound. To acquire additional structural information, we subjected the P. protegens Pf-5 extract to CuAAC. Processing the LC-MS/MS data by searching for fragmented ions that produce the brominated tropylium fragment ion identified the cycloadduct of 17 ([M+H]⁺/ ([M+2+H]⁺ = 478.11/480.11) as a prominent peak in the computer-generated chromatogram but not 19 (([M+H]⁺/ ([M+2+H]⁺ = 480.12/482.13). We were unable to detect the cycloadduct of 19 by an extracted ion chromatogram but were able to clearly observe the [M+H]⁺ = 269.13 (19) in the reaction mixture. This shows that while both compounds were present in the original extract and react during the RuAAC reaction, only 17 reacted during the CuAAC reaction. This confirms that 19 contains a terminal alkene group compared to the terminal alkyne group found in 17, which is supported by the increase in mass of +2 Da and the co-occurrence of terminal alkyne/terminal alkene natural products in other organisms (e.g. viequeamide B and C²⁷).

Comparison of RuAAC and CuAAC.

In the case of P. protegens both RuAAC and CuAAC reactions were able to derivatize 17 but only RuAAC was able to form the desired cycloadduct with 19. This suggests that RuAAC could be utilized to identify a greater number of natural products than CuAAC, for example fischerellin A and B described earlier in the manuscript. The largest downside to utilizing RuAAC is the generation of azide derived byproducts. This necessitates the creation of an exclusion list from the reaction containing Ru catalyst and azide prior to analyzing the RuAAC reaction mixture. In contrast, the identification of 17 in the CuAAC reaction was accomplished without the need for an exclusion list, thereby shortening the total analysis time. While we were able to circumvent this issue, this suggests optimization of the RuAAC reaction or Ru catalyst would be a worthwhile investment to allow greater convenience for end users. However, the discovery of protegenin C by RuAAC and not CuAAC demonstrates that RuAAC is a powerful tool for identifying acetylenic natural products in an extract and that CuAAC can be used in conjunction for preliminary structural assignments on a small-scale culture before isolation. As this method was shown to be successful in locating uncharacterized alkyne natural products from a crude extract, it has potential to be used to screen various other organisms including fungi and plants in the future to discover natural products with both terminal and internal alkyne moieties as well as further modifying the method for a high throughput approach to screen organisms with both sequenced and un-sequenced genomes from various culture collections.

Experimental Section

General experimental procedures

All UV-vis spectroscopy was performed using a BioSpectrometer Kinetic (Eppendorf). IR spectra were obtained on a Nicolet IR100 FT-IR spectrometer (ThermoFisher) using NaCl plates. NMR spectra were obtained on a Bruker DPX-500 MHz instrument with a 5mm TXI triple resonance (HCN) probe or BBOF probe, or a Bruker Avance III-700 MHz instrument equipped with a 2-channel 5 mm carbon observe cryoprobe using TopSpin version 3.5pl7. Data was processed using TopSpin version 4.0 (Bruker). LCMS analysis was performed utilizing a 3200 QTrap mass spectrometer (AbSciex) downstream of an Agilent 1260 Infinity HPLC system consisting of a degasser, quaternary pump, autosampler (set to 12 ᵒC), and a diode array detector operated using the Analyst software package. Data was analyzed offline using Peakview version 2.2 software. High-resolution mass spectrometry was performed using an Agilent 6545 quadrupole time-of-flight (Q-ToF) mass spectrometer downstream of an Agilent 1260 Infinity HPLC system consisting of a degasser, quaternary pump, autosampler (set to 12 ᵒC), and heated column compartment (set to 30 °C). The instrument was operated using MassHunter software and data was processed offline using MassHunter Qualitative software (version 10).

LCMS grade water, methanol Lysogeny Broth (LB), and Silica gel 60 (0.063–0.200 mm) were purchased from MilliporeSigma, while HPLC grade water, acetonitrile, and LCMS grade formic acid were purchased from Fisher Scientific. Nutrient broth was purchased from Difco. All other chemicals were purchased from Sigma-Aldrich (synthesis) or VWR and used without further purification unless otherwise specified.

Source of Microorganisms

Anabaena sp. PCC 7120 was a kind gift from Prof. Sean Callahan (University of Hawaii at Manoa). Pseudomonas protegens Pf-5 was a gift to the Loper lab by Dr. Charles Howell (USDA).²⁸

Synthesis of Internal Diynes, General procedure²⁹

To a flask, Cu powder (0.6 mg, 0.01 mmol) and Tetramethylethylenediamine (6 μL, 0.04 mmol) were added to a flame dried flask followed by 0.1 mL dioxane and 0.3 mL CHCl₃. Terminal alkyne (0.4 mmol) was then added to the reaction mixture. The reaction mixture was then stirred at 50 ᵒC overnight open to atmosphere. The reaction was monitored by TLC. When the reaction was completed, 4 mL of CHCl₃ was added and the reaction mixture was washed with 3 mL of saturated NH₄Cl solution. The organic layer was then dried over Na₂SO₄. The coupled alkynes were then purified by silica flash chromatography eluting with hexanes:ethyl acetate (20:1).

Synthesis of p-bromobenzyl azide³⁰

4-bromobenzyl bromide (12 mmol) and sodium azide (48 mmol) were dissolved in 20 mL of acetone:H₂O (3:1) and stirred at room temperature for six hours. The reaction was then diluted with 40 mL H₂O and extracted with 40 mL of dichloromethane four times. The organic layer was then dried over MgSO₄ and then concentrated to yield p-bromobenzyl azide as a colorless oily substance.

RuAAC reaction with Synthetic Alkynes:

To a flask, p-bromobenzyl azide (0.4 mmol) and alkyne (0.4 mmol) were added. To another flask, Cp*RuCl(PPh₃)₂ (0.008 mmol) was added. Under N₂ gas, 1.5 mL of 1,4-dioxane was added to the vial with the alkyne and azide and 1.5 mL of 1,4-dioxane was added to the vial containing the Cp*RuCl(PPh₃)₂ catalyst. The alkyne and azide containing solution was then transferred to the flask containing the Cp*RuCl(PPh₃)₂ catalyst. The flask was then placed in a hot oil bath at 70 °C overnight and reaction progress was monitored by TLC. A small amount of silica gel was added, and the solvent was removed in vacuo. The silica gel was added to the top of a prepacked silica gel column and the compound was purified by silica flash chromatography eluting with hexanes:ethyl acetate.

Growth and extraction of Anabaena sp. PCC 7120

Anabaena sp. PCC 7120 was cultivated in a Hoffman incubation chamber (Hoffman Manufacturing, Inc, Corvallis, OR) at 28°C under 24 hours constant light illumination, with the light being supplied by two GE Lighting 49893 F40/PL/AQ Plant and Aquarium Tube bulbs with a light intensity of 25 microeinsteins per m² per s and the atmosphere contained 1% CO₂. The bacteria were cultivated in 50 mL BG11(Nitrate+) media at 200 rpm for 14 days. The cells were collected via filtration on a Whatman Glass fiber filter and frozen. The cells were then lyophilized overnight followed by extraction in 20 mL 60% aq. methanol overnight with vigorous stirring by a stir bar. The extract was concentrated down to dryness in vacuo and resuspended in minimal amount of 1,4-dioxane and solid particles were pelleted by centrifugation. The cleared liquid was concentrated to dryness in a tared vial. The residue was the was then resuspended in 1,4-dioxane at a concentration of 1 mg/mL.

RuAAC with Anabaena PCC 7120 extract:

To a 10 mL scintillation vial, p-bromobenzyl azide (0.04 mmol) and 10 μl of a 1 mg/mL solution of diphenyl acetylene in 1,4-dioxane were added. To another 10 mL vial, Cp*RuCl(PPh₃)₂ (0.0016 mmol) was added followed by 1,4-dioxane (1 mL). Under N₂ gas, 0.75 mL of 1,4-dioxane was added to the vial with the diphenyl acetylene and p-bromobenzyl azide solution along with 0.25 mL of a 1 mg/mL solution of Anabaena sp. PCC 7120 extract. The alkyne, azide, and Anabaena extract containing solution was then transferred to the vial containing the Cp*RuCl(PPh₃)₂. The vial was then wrapped in parafilm and placed in a heated block at 70 °C overnight. The reaction was monitored LCMS.

Growth and extraction of Fischerella muscicola PCC 9339

Fischerella muscicola PCC 9339 was cultivated at 28°C under 24 hours constant light illumination, with the light being supplied by a cool white fluorescent bulb with a light intensity of 25 microeinsteins per m² per s. The culture was stirred by hand once per day and the bacterium was cultivated for 5 weeks. The cells were collected via filtration on a Whatman Glass fiber filter and frozen. The cells were then lyophilized overnight followed by extraction in 20 mL 60% aq. methanol overnight with vigorous stirring by a stir bar. The extract was concentrated down to dryness in vacuo and resuspended in minimal amount of 1,4-dioxane and concentrated to dryness in a tared vial.

RuAAC with Fischerella muscicola PCC 9339 extract:

To a 10 mL scintillation vial, p-bromobenzyl azide (0.04 mmol) and 0.25 mL of a 1 mg/mL solution of Fischerella muscicola PCC 9339 extract in 1,4-dioxane were added. Under N₂ gas, 0.75 mL of 1,4- dioxane was added to the vial with p-bromobenzyl azide and Fischerella muscicola PCC 9339 extract. To another 10 mL vial, Cp*RuCl(PPh₃)₂ (0.0016 mmol) was added followed by 1,4-dioxane (1 mL). The azide and Fischerella muscicola PCC 9339 extract solution was then transferred to the vial containing the Cp*RuCl(PPh₃)₂. The vial was then wrapped in parafilm and placed in a heated block at 70 °C overnight. The reaction was monitored LCMS.

Creation of Pseudomonas protegens Pf-5LK mutant strain

In-frame deletion of PFL_0261-0268 was created using an overlap extension PCR method as previously described,³¹ using the primers listed below.

Briefly, all PCRs were performed using KOD Hot Start DNA polymerase (Novagen, Madison, WI, USA). Up and Dn primers sets for each gene were used to generate the first round PCR products encompassing the 5′ and 3′ ends of the alkyne biosynthetic gene cluster (BGC). In the second round of PCR, 50 ng of each Up and Dn PCR product were combined for three rounds of amplification without added primers, creating the fused Up and Dn regions. After 3 cycles, the primers 0261 UpF-Xba and 0268 DnR-Xba were added to amplify the fused product for an additional 25 cycles, creating the deleted BGC product. The final PCR product was gel-purified and digested with XbaI and cloned into the corresponding site of pEX18Tc.³² All plasmids were first transformed into One Shot TOP10 Chemically Competent E. coli cells (Invitrogen, Carlsbad, CA, USA), then subsequently into the mobilizing strain Escherichia coli S17-1.³³ Biparental matings were performed with Pf-5 and S17-1 transformant. Transconjugants were selected on KMB with streptomycin (100 μg ml⁻¹, innate resistance of Pf-5) and tetracycline (200 μg ml⁻¹). Colonies were grown for 3 h without selection in Luria-Bertani (LB) broth and then plated on LB medium supplemented with 5% sucrose to favor resolved merodiploids growth. Colonies were patched onto KMB with tetracycline (200 μg ml⁻¹) to confirm resolution of merodiploids. Tetracycline-sensitive clones were screened for presence of the mutant allele by performing PCR across the recombination site using 0261 5’F and 0268 3’R. The resulting PCR product was also sequenced using the same primers to confirm correct incorporation of the mutant allele.

Primers	Sequence 5’−3’
0261 UpF-Xba	CAGCACTCTAGATGATCAGTTCGCTGGACAGG
0261–0268 UpR	GCAATCACCTTGACAGGCACGGCGTAGTTCATTCA
0261–0268 DnF	CTACGCCGTGCCTGTCAAGGTGATTGCCGATGC
0268 DnR-Xba	CTCCTCTCTAGACAAGTCGCTGTTCGTCAACG
0261 5’F	AGCATGCGTTTTGTGTCGAG
0268 3’R	GACCCACGACTTCACCAAGT

Culturing of Pseudomonas protegens Pf-5

Pseudomonas protegens Pf-5 cultures were inoculated onto King’s media B agar plates and incubated at 27 ᵒC. After sufficient growth was present, multiple colonies were picked and inoculated into 20 mL of nutrient broth + 1% glycerol. This was incubated at 27 ᵒC at 200 rpm for 16–18 hours. After 16–18 hours, the OD₆₀₀ was measured and the cells were centrifuged (3,500 × g, 30 min) and resuspended in nutrient yeast broth + 1% glycerol. The cells were then centrifuged again and resuspended in nutrient yeast broth + 1% glycerol with 0.35 mM zinc sulfate to obtain an OD₆₀₀ = 5.0 – 10.0. Enough culture was transferred to achieve starting OD₆₀₀ of 0.05 in 500 mL nutrient yeast broth + 1% glycerol with 0.35 mM zinc sulfate. The cultures were incubated at 20 ᵒC at 200 rpm (or 175 rpm if excessive foaming occurred) for 48 hours.

Extraction of Pseudomonas protegens Pf-5

After 48 hours of culturing, Pseudomonas protegens Pf-5 cultures were pelleted and media was poured into a separatory funnel. The media (500 mL) was extracted with 250 mL of ethyl acetate twice. Centrifugation was employed to remove emulsion from organic layer and the organic layers were combined. Before in vacuo removal of the ethyl acetate, 8 mL of 1,4-dioxane was added to the combined ethyl acetate layer. The solvent was removed in vacuo. When the volume reached around 1 mL, an additional 4 mL of 1,4-dioxane was added to ensure water was removed the extract. This was then concentrated down to a final volume of 1 mL.

RuAAC with Pseudomonas protegens Pf-5 extract:

To a 10 mL screw cap vial, p-bromobenzyl azide (0.04 mmol) and 0.25 mL of a 1 mg/mL solution of Pseudomonas protegens Pf-5 extract in dioxane were added. To another 10 mL vial Cp*RuCl(PPh₃)₂ (0.0044 mmol) was added followed by 1 mL of 1,4-dioxane. Under N₂ gas, 0.75 mL of 1,4-dioxane was added to the vial with p-bromobenzyl azide and Pseudomonas protegens Pf-5 extract. The azide and Pseudomonas protegens Pf-5 extract solution was then transferred to the vial containing the Cp*RuCl(PPh₃)₂. The vial was then wrapped in parafilm and left to sit at room temperature overnight. The reaction was monitored by iterative LCMS.

LCMS analysis

Both low-resolution and high-resolution mass spectrometry were obtained using an Agilent 1260 HPLC upstream of an Agilent 6545 Q-ToF downstream of an Agilent 1260 Infinity HPLC system. Separation was achieved using Luna C18(2) column (150 × 2.00 mm, 3 μm) at a flow rate of 0.2 ml/min and the following gradient. Line A was water with 0.1% (v/v) formic acid and line B was acetonitrile with 0.1% (v/v) formic acid. The column was pre-equilibrated with 90% A/10% B. Upon injection the mobile phase composition was maintained for 1 minute followed by changing the mobile phase to 0% A/100% B over 25 minutes using a linear gradient.

Both low-resolution and high-resolution mass spectrometry were obtained using an Agilent 1260 HPLC upstream of an Agilent 6545 Q-ToF downstream of an Agilent 1260 Infinity HPLC system. Separation was achieved using Kinetex C18 column (50 × 2.1 mm, 2.6 μm) at a flow rate of 0.4 ml/min and the following gradient. Line A was water with 0.1% (v/v) formic acid and line B was acetonitrile with 0.1% (v/v) formic acid. The column was pre-equilibrated with 90% A/10% B. Upon injection the mobile phase composition was maintained for 0.75 minutes followed by changing the mobile phase to 0% A/90% B over 7.5 minutes using a linear gradient. The gradient was then increased from 0% A/ 90% B to 0% A/ 100% B over 3.65 minutes using a linear gradient.

The Agilent Q-ToF mass spectrometer was equipped with an Agilent JetSpray source operated with the following parameters: Auto MS/MS mode, Positive polarity; Gas Temp, 325 ᵒC; Drying gas, 7 L/min; Nebulizer, 20 psi; Sheath gas temp, 270 ᵒC; Sheath gas flow, 12 L/min; VCap, 4000 V; Nozzle voltage (Expt), 600 V; Fragmentor, 175 V; Skimmer, 65 V; Oct 1 RF Vpp, 750 V; Mass range, 100–3000 m/z; Acquisition rate, 10 spectra/s; Time, 100 ms/spectrum. The AutoMS/MS settings were as follows: Mass range, 50–3000 m/z; 10 spectra/s; 100 ms/spectrum; Transients/spectrum, 429; Isolation Width, Medium (~4 m/z); The Precursor selection parameters were as follows: 5 Max Precursor Per Cycle; Abs. Threshold 1000 counts, Rel. Threshold (%) 0.01%, Excluded after 4 spectra; Released after 0.01 min. The iterative MS/MS settings were as follows: Mass error tolerance (+/− ppm) 20; RT exclusion tolerance, 0.2 (-min) 0.2 (+min).

Authors

Daniel Back - Department of Pharmaceutical Sciences, College of Pharmacy, Oregon State University, Corvallis, Oregon, 97331, United States, https://orcid.org/0000-0002-6879-923X

Brenda T. Shaffer – Agricultural Research Service, US Department of Agriculture, 3420 N.W. Orchard Avenue, Corvallis, OR 97330, United States

Joyce E. Loper - Agricultural Research Service, US Department of Agriculture, 3420 N.W. Orchard Avenue, Corvallis, OR 97330, United States; College of Agricultural Sciences, Oregon State University, Corvallis, OR 97331, United States, ORCID https://orcid.org/0000-0003-3501-5969