Journal of Chromatography B, 980 (2015) 1–7

Contents lists available at ScienceDirect

Journal of Chromatography B journal homepage: www.elsevier.com/locate/chromb

Separation of isomeric short-chain acyl-CoAs in plant matrices using ultra-performance liquid chromatography coupled with tandem mass spectrometry Randy W. Purves a,b,∗ , Stephen J. Ambrose b , Shawn M. Clark b,c , Jake M. Stout d , Jonathan E. Page b,c,e a

Plant Sciences Department, University of Saskatchewan, 51 Campus Drive, Saskatoon, SK, Canada S7N 5A8 National Research Council of Canada, 110 Gymnasium Place, Saskatoon, SK, Canada S7N 0W9 c Department of Biology, University of Saskatchewan, 112 Science Place, Saskatoon, SK, Canada S7N 5E2 d Department of Biological Sciences, University of Manitoba, 50 Sifton Road, Winnipeg, MB, Canada R3T 2N2 e Department of Botany, The University of British Columbia, #3529-6270 University Boulevard, Vancouver, BC, Canada V6T 1Z4 b

a r t i c l e

i n f o

Article history: Received 28 August 2014 Received in revised form 26 November 2014 Accepted 8 December 2014 Available online 17 December 2014 Keywords: Acyl-CoAs UPLC Tandem mass spectrometry Hops Plants Metabolism

a b s t r a c t Acyl coenzyme A (acyl-CoA) thioesters are important intermediates in cellular metabolism and being able to distinguish among them is critical to fully understanding metabolic pathways in plants. Although significant advances have been made in the identification and quantification of acyl-CoAs using liquid chromatography tandem mass spectrometry (LC–MS/MS), separation of isomeric species such as isobutyryl- and n-butyrl-CoA has remained elusive. Here we report an ultra-performance liquid chromatography (UPLC)–MS/MS method for quantifying short-chain acyl-CoAs including isomeric species n-butyryl-CoA and isobutyryl-CoA as well as n-valeryl-CoA and isovaleryl-CoA. The method was applied to the analysis of extracts of hop (Humulus lupulus) and provided strong evidence for the existence of an additional structural isomer of valeryl-CoA, 2-methylbutyryl-CoA, as well as an unexpected isomer of hexanoyl-CoA. The results showed differences in the acyl-CoA composition among varieties of Humulus lupulus, both in glandular trichomes and cone tissues. When compared with the analysis of hemp (Cannabis sativa) extracts, the contribution of isobutyryl-CoAs in hop was greater as would be expected based on the downstream polyketide products. Surprisingly, branched chain valeryl-CoAs (isovaleryl-CoA and 2-methylbutyryl-CoA) were the dominant form of valeryl-CoAs in both hop and hemp. The capability to separate these isomeric forms will help to understand biochemical pathways leading to specialized metabolites in plants. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Acyl coenzyme A (acyl-CoA) thioesters are important intermediates in cellular metabolism that are formed through the conjugation of CoA with carboxylic acids. It has been estimated that in eukaryotic systems, approximately 4% of all enzymatic activity utilizes an acyl-CoA as a substrate [1]. Our research on the biosynthesis of polyketide-derived natural products in plants has led to a need to identify and quantify acyl-CoA precursors that are used by type III polyketide synthases [2,3]. Understanding

∗ Corresponding author at: University of Saskatchewan, Plant Sciences Department, 4D36 Agriculture Building, 51 Campus Drive, Saskatoon, SK, Canada S7N 5A8. Tel.: +1 306 966 2933. E-mail addresses: [email protected], [email protected] (R.W. Purves). http://dx.doi.org/10.1016/j.jchromb.2014.12.007 1570-0232/© 2014 Elsevier B.V. All rights reserved.

these key intermediates and the pathways by which they are formed, may enable the metabolic engineering of natural product biosynthesis in both plants and microorganisms [4]. Because of the complexity of the plant matrix, any quantitative method will need to be sensitive, selective, and robust. Early quantification methods for acyl-CoAs involved the use of LC with UV detection [5,6]. LC–UV methods suffered from low selectivity and sensitivity, and to overcome these limitations, a wide variety of other methods have been employed (reviewed in Haynes [7]). Among these techniques, methods based on GC–MS [8] and LC with fluorometric detection [9] provide improved sensitivity, however these methods require time consuming derivatization steps. Conversely, it was shown that acyl-CoAs could be analyzed intact by MS when ionized using fast atom bombardment [10], matrix-assisted laser desorption ionization (MALDI) [11], or electrospray ionization (ESI) [11]. These techniques, when combined

2

R.W. Purves et al. / J. Chromatogr. B 980 (2015) 1–7

with tandem mass spectrometry, showed that acyl-CoAs undergo characteristic fragmentation in both positive and negative modes thereby offering very high selectivity. Liquid chromatography was first combined with electrospray tandem mass spectrometry (LC–MS/MS) to analyze long chain acyl-CoAs using negative ionization [12] and subsequently with positive ionization [13]. Magnes et al. [13] also showed that the characteristic fragmentation pattern for acyl-CoAs could be used to investigate acyl-CoA profiles in complex mixtures by using a neutral loss scan to probe for the loss of phosphoadenosine diphosphate (507 Da). Because of the high specificity and sensitivity of LC–MS/MS, it has become the technique of choice for quantification of acyl-CoAs and methods have been developed to analyze a variety of acyl-CoAs in different biological matrices including Escherichia coli [14], animal cells and tissues [15–20], and plants [2,21–23]. Despite these significant advances in the analysis of acyl-CoAs, none of the aforementioned LC–MS/MS methods are able to distinguish among isomeric acyl-CoAs as the isomers are not resolved chromatographically (i.e., they co-elute) or by MS/MS techniques (since they generate the same fragment ions). The capability not only to detect acyl-CoAs, but also to distinguish among isomeric acyl-CoAs in plant tissues is critical to fully understanding metabolic pathways. For example, isobutyryl-CoA is an important polyketide precursor in hop (Humulus lupulus) [3], whereas n-butyryl-CoA serves in this role in cannabis (Cannabis sativa), where it gives rise to propyl side-chain cannabinoids such as 9 -tetrahydrocannabivarinic acid [24]. Recently, we implemented an ultra-performance liquid chromatography (UPLC)–MS/MS method for quantifying short-chain acyl-CoAs [2] since UPLC has been shown to improve separations compared with conventional HPLC [25]. Our method was used to quantify the hexanoyl-CoA precursor in the cannabinoid biosynthetic pathway [2]. Here, we have expanded this method to analyze seven other short-chain acyl-CoAs including the first published separation of two isomers of butyryl-CoA and three isomers of valeryl-CoA. Finally, quantification of acyl-CoAs is carried out in extracts of hop, which contain mainly branched-chain acyl-CoAs (e.g., isovalerylCoA), and cannabis, which mainly contains n-chain acyl-CoAs (e.g., hexanoyl-CoA). 2. Experimental 2.1. Materials All acyl-CoA standards and 13 C labelled standards were purchased from Sigma Aldrich (St. Louis, MO, USA). KH2 PO4 was purchased from EM Science (Gibbstown, NJ, USA). Trifluoroacetic acid (TFA) was purchased from Sigma (Oakville, ON, Canada). Isopropanol (IPA), acetic acid, methanol, acetonitrile, and triethylamine (TEA) were purchased from Fisher Scientific (Waltham, MA, USA). 2.2. Plant growth conditions Samples of hop lupulin glands and cones were collected as described previously [3]. Samples of flowers, leaves, and roots were obtained from the hemp cultivar “Finola” as described previously [2]. 2.3. Extraction of acyl-CoAs from plant tissues As is described in more detail in Section 3.2, the method that we used previously to extract acyl-CoAs from plant tissues [2] was changed because of improved recoveries of acyl-CoAs that were obtained using a weak anion exchange solid phase extraction (WAX

Table 1 Optimized UPLC gradient for separating isomeric short-chain acyl-CoAs. Time (min)

Flow rate (mL/min)

%A

%B

Curve

Initial 2.0 20.0 22.0 24.0 25.0 27.0 29.0 32.0 32.1 35.0

0.20 0.20 0.20 0.20 0.35 0.50 0.50 0.40 0.20 0.20 0.20

98 98 88 70 2 2 2 98 98 98 98

2 2 12 30 98 98 98 2 2 2 2

6 6 6 11 6 6 6 6 6 6

SPE) clean-up step. Our new method for extraction of acyl-CoAs from plant tissues is as follows: Plant samples were weighed, immediately immersed in liquid nitrogen, and ground with the addition of internal standards (13 C3 -malonyl-CoA, butenoyl-CoA and benzoyl-CoA). The samples were extracted two times with 10 mL of 5% (v/v) TFA followed by evaporation of the solvent and resuspension in 900 ␮L of acetonitrile:isopropanol:100 mM KH2 PO4 :aqueous acetic acid (9:3:4:4, v/v/v/v) by vortexing and sonication. A clean-up step involving a Waters (Milford, MA) Oasis WAX SPE was then used and the protocol was as follows: (1) Activate cartridge with 1.0 mL of methanol. (2) Equilibrate cartridge with 1.0 mL acetonitrile:isopropanol: water:acetic acid (9:3:4:4, v/v/v/v). (3) Load reconstituted sample (or standard). (4) Wash with 2.0 mL acetonitrile:isopropanol:water:acetic acid (9:3:4:4, v/v/v/v). (5) Elute with 3 mL of 250 mM NH4 OH in 80% methanol (pH = 11.4). (6) Samples were dried and stored at −20 ◦ C. (7) The residue was reconstituted in 100 ␮L of water:acetonitrile (95:5, v/v) containing 5 mM triethylamine (TEA) and 3 mM acetic acid as well as the recovery standard 13 C4 -octanoyl-CoA. (8) The reconstituted sample was vortexed, sonicated, centrifuged, and transferred to an injection vial. 2.4. UPLC–MS/MS analysis The acyl-CoAs were investigated using a Waters Acquity UPLC binary system coupled to a Thermo Fisher TSQ Quantum (San Jose, CA). Chromatography was carried out using a Waters BEH-C18 (1.7 ␮m, 2.1 × 100 mm) column with a Vanguard precolumn (BEH Shield RP18, 1.7 ␮m). The column and autosampler were operated at 35 ◦ C and 4 ◦ C, respectively. The amount of extract injected was 7.5 ␮L (using needle overfill). The mobile phases were (A) 5 mM TEA and 3 mM acetic acid in water, and (B) 5 mM TEA and 0.3 mM acetic acid in 95:5 (v/v) acetonitrile:water. The optimized gradient that was used is shown in Table 1. Initial studies using the TSQ Quantum mass spectrometer were carried out in both positive and negative ionization modes by using flow injection analysis of individual standards (dissolved in 1:1 A:B) injected into a solvent stream of 1:1 A:B. Full scans (m/z 700–1200) were used to confirm the identity of the protonated molecular ion. The use of an ion-pairing reagent, TEA, resulted in the presence of adduct peaks in positive ionization mode. Thus, the capillary, tube lens, and skimmer voltages were adjusted to give the maximal sensitivity for the [M + H]+ ions, while minimizing the appearance of [M + H + TEAn ]+ peaks (where n was typically 1 or 2) in the mass spectrum. Product ion scans (MS/MS) were carried out to determine the most sensitive fragment ions for each acyl-CoA that subsequently would be used in selected reaction monitoring (SRM). Using

R.W. Purves et al. / J. Chromatogr. B 980 (2015) 1–7

3

Table 2 SRM settings used for quantification. Segment

Compound

Parent m/z

Fragment m/z

CE (eV)

1 1 1 2 2 2 3 3 3 4

Acetyl-CoA Malonyl-CoA 13 C3 -malonyl-CoA Propionyl-CoA 2-butenoyl-CoA Butyryl-CoA Valeryl-CoA Hexanoyl-CoA Benzoyl-CoA 13 C4 -octanoyl-CoA

810 854 857 824 836 838 852 866 872 898

303 347 350 317 329 331 345 359 365 391

24 27 27 27 25 27 30 27 25 25

SRM, optimal collision energies (CE) were determined using flow injection analysis of individual standards. Table 2 shows the most intense transition for each acyl-CoA that was used for SRM quantification along with the corresponding optimized collision energy (CE). As will be described in more detail in the following section, quantitative data was acquired only using positive mode. Since UPLC–SRM is a targeted method and the retention times of the acylCoAs are known, to maximize mass spectrometer acquisition time on eluting peaks (and thereby maximize sensitivity for quantification) four different time segments were employed. Segment 1 was from 0 to 11 min, segment 2 from 11 to 16.5 min, segment 3 from 16.5 to 23 min, and segment 4 from 23 to 35 min (segment number for each acyl-CoA is indicated on Table 2). For each acyl-CoA, 100 ms was used as the dwell time and the peak widths for both Q1 and Q3 were set to 0.7 FWHM. The collision gas used in this study was argon and the Q2 CID gas was set to 1.5 mTorr. The MS parameters used in acquiring SRM data were as follows: the electrospray needle was set to 4000 V; the tube lens was 210 V; the skimmer was 15 V; and the capillary offset was 100 V. The auxiliary gas and sheath gas were set to 50 and 30, respectively; and the vaporizer and capillary temperatures were 300 ◦ C and 225 ◦ C, respectively. Quantification of the acyl-CoAs is described in Section 3.3. 3. Results and discussion 3.1. Method development The method described herein builds upon our previous work that was used to quantify hexanoyl-CoA in cannabis [2,26] and focuses on short-chain acyl-CoAs (up to six carbons) with emphasis on the separation of isomeric acyl-CoAs; in particular n-butyryl-CoA and isobutyryl-CoA as well as n-valeryl-CoA and isovaleryl-CoA. As described in Section 2.4, tandem mass spectrometry conditions for the short-chain acyl-CoAs were individually optimized using standard mixtures prepared in equal volumes of solvent A and B. Signal intensities using full scan (MS) mode were comparable for the pseudomolecular ions of the acyl-CoAs using either positive or negative polarity. However, product ion scans (and hence SRM analyses) were found to be approximately five to ten times more sensitive in positive mode compared with negative mode. This difference in signal intensity is illustrated in Fig. 1 for hexanoylCoA. In both modes, the value of CE used to acquire the spectrum was set to the optimum value of the most intense fragment; i.e., CE = 25 V in Fig. 1A and CE = 30 V in Fig. 1B. In Fig. 1A (positive ionization mode), the most intense fragment ion (m/z 359) has a signal intensity that is approximately seven times greater than the signal intensity of the most intense fragment ion (m/z 408) acquired in negative mode. In addition, the spectrum is less complicated as a much larger number of fragments were observed using negative ionization mode. Since all acyl-CoAs studied herein behaved

Fig. 1. (A) MS/MS spectrum of hexanoyl-CoA in positive ionization mode with CE = 25 V, (B) MS/MS spectrum of hexanoyl-CoA in negative ionization mode with CE = 30 V, and (C) sites of bond cleavage resulting in the two most intense fragments of hexanoyl-CoA in positive ionization mode.

in a similar manner, analyses were carried out exclusively in positive ion mode. Fig. 1C indicates the location of the bond cleavages in hexanoyl-CoA that are necessary to form the two most intense fragment ion peaks in Fig. 1A. The formation of the m/z 428 fragment ion and the fragment ion having a neutral loss of 507 Da were observed by Norwood and co-workers for acetyl-CoA using FAB [10] and then used for LC–MS/MS for quantification by Magnes et al. [13] on a triple quadrupole mass spectrometer. Since these two specific transitions give rise to fragment ions formed from completely different locations within an acyl-CoA, the requirement of detecting both fragment ions gives rise to an extremely selective method and therefore very high confidence. In our study, both transitions were employed for identification, however, to maximize sensitivity, only the transition for the neutral loss of 507 Da was used for quantification. To maximize the chromatographic separation efficiency, UPLC methodology was employed in this study. A small sample of UPLC columns were tested and the Waters BEH-C18 1.7 ␮m, 2.1 × 100 mm column was selected for this study. The chromatographic method was developed by injecting mixtures of standard solutions of the short-chain acyl-CoAs (and monitoring using the optimum SRM conditions determined above) while testing different solvent combinations. Acetonitrile was chosen as the organic solvent as sharper peaks were observed with acetonitrile compared with methanol. Different modifiers were also tested based on previous published methods that employed acidic conditions [22], basic conditions [17], and the use of ion pairing reagents [16]. The sharpest chromatographic peaks were observed using a basic pH; thus the method developed herein was based on using the ion pairing reagent triethylamine (TEA) since significant column bleed using NH4 OH was observed. Thus, by using a UPLC system with a BEHC18 column, acetonitrile as the organic solvent, basic conditions, and TEA, implementing the optimized gradient shown in Table 2 resulted in the separation of all eight acyl-CoA standards, including n-butyryl and isobutyryl-CoA and n-valeryl and isovaleryl-CoA. The n- and iso-species were injected separately and as mixtures to confirm and optimize the separation. Fig. 2 shows a representative

4

R.W. Purves et al. / J. Chromatogr. B 980 (2015) 1–7

Fig. 2. UPLC–SRM chromatograms for the short chain acyl-CoA standards using the optimized gradient given in Table 1. The inset better shows the near baseline separation (valley is ∼3% of full maximum) of isobutyryl-CoA and n-butyryl-CoA. The peak labelled “int” results from the isotopomer of the 2-butenoyl-CoA standard.

chromatogram of an injection of a 1 ␮M solution of the eight shortchain acyl-CoAs (and the four standards) used in this study. The figure clearly shows complete baseline separation of n-valeryl-CoA and isovaleryl-CoA and near baseline separation of n-butyryl-CoA and isobutyryl-CoA. The inset in Fig. 2 shows that interference is minimal as the height of the valley between the peaks is ∼3.5% the maximum intensity of isovaleryl-CoA. Note that the peak labelled as “int” in the trace for butyryl-CoA is a consequence of an isotopomer peak of 2-butenoyl CoA.

into our extraction protocol [28]. With the use of the WAX SPE cartridges, total recovery experiments for the acyl-CoAs improved to 78%-90% [28]. The details for the updated extraction protocol using the Waters Oasis WAX SPE are described in the experimental section and based on this procedure, estimated recoveries for the eight acyl-CoAs are given in Table 3. Note that during the optimization of the WAX SPE protocol, the reconstitution solvent was found to be a critical variable that could affect retention times. 3.3. Plant tissue analysis

3.2. Extraction protocol Plant tissue samples of hop and cannabis were obtained through in house cultivation (cannabis) and field collection (hop). In addition to providing a comparison with our previous work [2], short-chain acyl-CoAs from cannabis are believed to contain mostly n-based acyl-CoAs, whereas acyl-CoAs from hop are believed to be mostly branched chains based on the predominance of bitter acids derived from these precursors [3]. The procedure for the extraction of the short-chain acyl-CoAs from plant tissues we described previously employed a strong cation exchange solid phase extraction step [2]. Experiments investigating the total recoveries for the short-chain acyl CoAs (using standards) showed that the recoveries ranged from 53% (malonyl-CoA) to 68% (n-butyryl-CoA) using this method [26]. Minkler et al. reported a novel isolation procedure for acyl-CoAs [27] that employed a weak anion exchange solid phase extraction (WAX SPE) step. We subsequently investigated different WAX SPE cartridges and implemented the Waters Oasis WAX

3.3.1. Chromatography Using the optimized procedure described above, plant tissues derived from hop and cannabis were analyzed to determine the levels of the various acyl-CoAs and a typical chromatogram for the analysis of the trichomes (lupulin glands) from hop is shown in Table 3 Estimated % recovery for the eight acyl-CoAs. Acyl-CoA

% recovery*

Malonyl Acetyl Propionyl n-Butyryl iso-Butyryl n-Valeryl iso-Valeryl Hexanoyl

82 78 82 86 83 80 84 87

*

%RSD is 5–8%.

R.W. Purves et al. / J. Chromatogr. B 980 (2015) 1–7

5

Table 4 Precision and accuracy for isovaleryl-CoA.

Fig. 3. UPLC–SRM chromatograms monitoring for the short chain acyl-CoAs present in hop trichomes. The peaks labelled as “new” are believed to be 2-methylbutyrylCoA (valeryl-CoA trace) and a hexanoyl-CoA isomer (hexanoyl-CoA trace). The insets better show the separation of structural isomers in the channels of butyryl-CoA, valeryl-CoA, and hexanoyl-CoA.

Fig. 3. Note, the retention times for the acyl-CoAs among these plant samples were observed to change, and therefore the retention times in Fig. 3 have been adjusted as is described later in this section. Three insets are shown in Fig. 3 to better illustrate the separation of the various species within the three traces. Compared with the chromatographic traces for the standard solutions shown in Fig. 2, there are some noticeable differences. A very striking difference between the chromatograms is the existence of additional peaks in the traces for both valeryl-CoA and hexanoyl-CoA; these peaks are labelled as “new”. The trace for the valeryl-CoA channel yielded three distinct peaks, the adjusted retention times for two of which corresponded to n-valeryl and isovaleryl-CoA. The third peak also gave characteristic fragmentation peaks at m/z 428 and at m/z 345 (corresponding to a loss of 507). As mentioned previously, since these two fragment ions are formed from two completely different regions of valeryl-CoA, this extremely high selectivity provides very convincing evidence that the third peak is a structural isomer of valeryl-CoA. The literature provides further strong evidence that the identity of this species is 2-methylbutyryl-CoA [3] due to the presence of both adhumulone and adlupulone in hop lupulin glands that are downstream metabolites in the bitter acid biosynthesis pathway. Unfortunately, we are not able to 100% confirm the identity of this species since there is no commercial standard that is presently available for 2-methylbutyryl-CoA. The presence of the additional peak labelled as “new” in the trace for hexanoyl-CoA was unexpected and demonstrates the power of being able to distinguish between the various acyl-CoA structural isomers. The inset of the hexanoyl-CoA trace shows that the height of the valley between hexanoyl-CoA and the peak for the isomer is ∼5% of the peak height for hexanoyl-CoA. Since the peak for the isomer gave characteristic fragmentation peaks at m/z 428 and at m/z 359 (corresponding to a loss of 507) and again the fragmentation pattern of this peak is virtually identical to that of n-hexanoyl-CoA (not shown) this strongly suggests that this peak represents a structural isomer of hexanoyl-CoA (this peak will be referred to as an hexanoyl-CoA isomer as it is not meant to imply any specific isomeric structure). Note that in setting up the method, the slope of the gradient was dramatically changed after the elution of n-valeryl-CoA, and therefore separation of hexanoyl-CoA from the hexanoyl-CoA isomer could be further improved by decreasing the slope of the gradient in this

Standard #

[Specified] (nM)

[Actual] (nM)

Precision (%)

Accuracy (%)

1 2 3 4 5 6 7

6.42 24.1 80.5 321 642 1285 2570

6.47 23.9 80.7 318 643 1295 2562

18.4 8.3 12.5 6.6 5.3 4.7 2.3

100.8 99.1 100.3 99.2 100.2 100.8 99.7

portion of the chromatographic run. This was not done in this study since the separation was sufficient to enable independent quantification of these two species and also since improving separation would increase run time. As was mentioned above, especially in the early part of the chromatographic run, the retention times were observed to shift in the plant matrix, as for example the retention time of malonyl-CoA was observed to shift by as much as 0.5 min. We hypothesize that this shift is due to differences in the very complex matrices associated with the samples, which contain high amounts of polyketide and terpenoid metabolites. Fortunately, the presence of the internal standards made it straightforward to compensate for these shifts. The retention time for each internal standard in Fig. 3 was adjusted to the retention time for each standard observed in Fig. 2. For each of the eight analytes, the retention time was then adjusted by the same amount that was used for its corresponding internal standard. The result is that the retention times for each of the eight analyte peaks in Fig. 3 are virtually identical (e.g., all butyryl- and valerylCoAs were within 0.05 min). In addition, by comparing the insets of Figs. 2 and 3, it is readily observed that the separation between the isomers is maintained in the plant matrix and therefore the individual acyl-CoAs were readily quantified. 3.3.2. Quantification and method validation Quantification of acyl-CoA analytes was achieved by running seven point calibration curves. Concentrations for each of the acylCoAs were varied over approximately two and a half orders of magnitude and the concentration ranges were established using estimates of acyl-CoA amounts based on preliminary analyses. For example, malonyl-CoA was present at a higher concentration among the acyl-CoAs and a calibration curve ranging from 25 nM to 10 ␮M was used, whereas propionyl-CoA was present at relatively lower concentrations and thus the calibration range was 5 nM to 2 ␮M. A mixture of internal standards were added to samples and standards at concentrations of 4 ␮M for 13 C3 -malonyl-CoA and 2 ␮M for both 2-butenoyl-CoA and benzoyl-CoA. The standards were used both for quantification and to adjust for retention time shift caused by the matrix as was described earlier. One micro molar 13 C octanoyl-CoA was used as an external standard to monitor 4 recovery, and was present in the reconstitution solvent for standards and samples providing a common reference. Quantification of 2-methylbutyryl-CoA was carried out under the assumption that it has a similar response factor to isovaleryl-CoA since a standard was not available. Similarly, the hexanoyl-CoA isomer was quantified using the assumption that it has a similar response factor as hexanoyl-CoA. For quantification of the acyl-CoAs, a linear calibration curve was employed and the points were weighted 1/X. Fig. 4 shows calibration curves for two of the acyl-CoAs, isovaleryl-CoA and nvaleryl-CoA, and these curves illustrate the linear response over the calibration range used in this study. Validation characteristics [29], namely linearity, lower limit of quantification (LLOQ), range, precision, and accuracy, for all eight acyl-CoAs were determined. Table 4 shows the precision and accuracy data for isovaleryl-CoA, whereas Table 5 summarizes the linearity, LLOQ, and range for the

6

R.W. Purves et al. / J. Chromatogr. B 980 (2015) 1–7

below the LLOQ are listed as BLOQ (below limit of quantification). The calibration curves shown in Fig. 4 for isovaleryl-CoA and n-valeryl-CoA were also chosen to illustrate the similarity in their responses. This result is expected for these ions since the charge will reside within the CoA moiety; changes related to branching within the acyl-chain would not be expected to significantly impact the ionization efficiency of the compound. For the quantification of 2-methylbutyryl-CoA, the calibration curve for isovaleryl-CoA was used since a commercial standard of 2-methylbutyryl-CoA was not available and since isovaleryl-CoA is the most closely related branched chain acyl-CoA. Because of the expected similarity in ionization efficiencies, we do not anticipate introducing any significant error by using this calibration curve instead. Using a similar argument, the hexanoyl-CoA isomer was quantified using the calibration curve for hexanoyl-CoA.

Fig. 4. Calibration curves for isovaleryl CoA and n-valeryl CoA with best fit lines based on a linear regression and 1/X weighting. As expected, because the ionization efficiencies for these compounds are similar, the curves are also similar and therefore the isovaleryl-CoA calibration curve was used also to quantify 2-methylbutyryl-CoA.

Table 5 Validated calibration curve summary. Acyl-CoA

LOQ (nM)

R2 of fit

Linear range (nM)

Malonyl Acetyl Propionyl n-Butyryl iso-Butyryl n-Valeryl iso-Valeryl Hexanoyl

25 25 5 5 8.4 4.6 6.4 4.0

0.998 0.990 0.992 0.998 0.997 0.996 0.999 0.996

25–10,000 25–10,000 5.0–2000 5.0–2000 8.4–3371 4.6–1834 6.4–2570 4.0–1456

eight acyl-CoAs. Table 4 shows that the precision and accuracy for isovaleryl-CoA were within method validation guidelines (15% for all points except the LLOQ, for which the guideline is 20%) for all points on the curve. The precision and accuracy for all eight acylCoAs were similar in that all points on the curves were within the limits of the guidelines (data not shown). As can be seen from Table 5, R2 values were greater than 0.99 and the range is ∼2.5 orders of magnitude for all of the acyl-CoAs. Any results obtained

3.3.3. Analysis of hop and cannabis samples Fig. 5 shows a plot of the concentration of ten short chain acylCoAs in two different tissues of hops: (A) trichome and (B) cone. In the cone samples, three of the acyl-CoAs were BLOQ (propionyl, n-valeryl, and the hexanoyl-CoA isomer) and are labelled as such in the figure. As was expected, the overall levels of the branched chain acyl-CoAs are higher in hops with relatively very little n-butyryl and n-valeryl. Within the hop samples, the levels of acyl-CoAs in the trichome are the greatest, which parallels the biosynthesis and accumulation of bitter acids derived from these precursors [3]. Being able to distinguish the isomeric acyl-CoAs helps to decipher the chemical pathways within plants. Flowers of Cannabis sativa were also analyzed and Fig. 6 compares UPLC–SRM traces of hop (cultivar ‘Apollo’) and cannabis for the isomeric acyl-CoAs (only butyryl and valeryl isomers are shown). As expected, Fig. 6a shows that the formation of isobutyryl-CoA is favoured in hops, whereas n-butyryl-CoA is greater in cannabis. Furthermore, nvaleryl-CoA is BLOQ in the cannabis samples, as shown in Fig. 6b. This fits with the observation that this acyl-CoA is not used as a substrate in cannabinoid biosynthesis. The presence of high quantities of isovaleryl-CoA and 2-methylbuturyl-CoA in the cannabis samples is surprising. They may simply be a product of basal branched chain amino acid catabolism, or may represent cryptic metabolites in cannabis trichome metabolism. Given that the detection and quantification of these compounds has not been possible prior to advent of the method described, it is unknown

Fig. 5. The top traces show the concentrations of ten short chain acyl-CoAs in the (A) trichomes and (B) cone of hop. The bottom traces in each of (A) and (B) show the same plots with 10X magnification. The three hops plants were ‘Taurus’ (Germany), ‘Taurus’ (Saskatoon), and ‘Apollo’ (Yakima).

R.W. Purves et al. / J. Chromatogr. B 980 (2015) 1–7

7

Although the dominance of n-butyryl-CoA over isobutyryl-CoA was as expected in hemp, the presence of high quantities of isovaleryl-CoA and 2-methylbuturyl-CoA in the hemp flowers was unexpected. Chromatographic separation of the isomeric acyl-CoAs made these interesting results possible, illustrating the utility and importance of this method in understanding biochemical pathways that can ultimately be used to create specialized metabolites in plants. Acknowledgements The authors thank Thermo Fisher for the loan of the TSQ Quantum mass spectrometer as part of a collaborative agreement. We thank Sandra Polvi (NRC) for growing hemp plants and Dr. Paul Matthews (Hopsteiner) for help in obtaining hop samples. References

Fig. 6. Comparison of the UPLC–SRM plots for the isomeric compounds of (A) butyryl-CoA (m/z 838–331) and (B) valeryl-CoA (m/z 852–345) in the flowers of cannabis and the trichomes of hop (‘Apollo’).

what the pool sizes of these metabolites are in other plants and their tissues. Further quantification of acyl-CoA pools using the method described should clarify this. 3.4. Conclusions A UPLC–MS/MS method capable of separating and quantifying isomers of short-chain acyl-CoAs has been developed. The method results in near baseline separation of n-butyryl and isobutyryl-CoA isomers and baseline separation of n-valeryl and isovaleryl-CoA. The method was applied to hop extracts and showed the presence of two isomers of butyryl-CoA, three isomers of valeryl-CoA, and two isomers of hexanoyl-CoA. Strong evidence supports the assignment of the third isomer of valeryl-CoA as 2-methylbutyryl-CoA. The levels of the branched chain acyl-CoAs were greatest in hops as was expected based on downstream metabolites (e.g., humulone). Hemp extracts were also examined since the cannabinoid pathway suggests the predominance of straight-chained acyl-CoA intermediates leading to end-products such as cannabidiolic acid.

[1] C.F. Lee, A.F. Chen, in: J. Everse, B. Anderson, K. You (Eds.), The Pyridine Nucleotide Coenzymes, Academic Press, New York, NY, 1982, p. 189. [2] J.M. Stout, Z. Boubakir, S.J. Ambrose, R.W. Purves, J.E. Page, Plant J. 71 (2012) 353. [3] S.M. Clark, V. Vaitheeswaran, S.J. Ambrose, R.W. Purves, J.E. Page, BMC Plant Biol. 13 (2013) 12. [4] P.J. Facchini, J. Bohlmann, P.S. Covello, V. De Luca, R. Mahadevan, J.E. Page, D. Ro, C.W. Sensen, R. Storms, V.J.J. Martin, Trends Biotechnol. 30 (2012) 127. [5] M.T. King, P.D. Reiss, Anal. Biochem. 146 (1985) 173. [6] G. Woldegiorgis, T. Spennetta, B.E. Corkey, J.R. Williamson, E. Shrago, Anal. Biochem. 150 (1985) 8. [7] C.A. Haynes, Biochim. Biophys. Acta 1811 (2011) 663. [8] C.S. Tamvakopoulos, V.E. Anderson, Anal. Biochem. 200 (1992) 381. [9] T.R. Larson, I.A. Graham, Plant J. 25 (2001) 115. [10] D.L. Norwood, C.A. Bus, D.S. Millington, J. Chromatogr. B: Biomed. Sci. Appl 527 (1990) 289. [11] J.A. Hankin, R.C. Murphy, Int. J. Mass Spectrom. Ion Process 165–166 (1997) 467. [12] T. Mauriala, K.H. Herzig, M. Heinonen, J. Idziak, S. Auriola, J. Chromatogr. B: Analyt. Technol. Biomed. Life Sci. 808 (2004) 263. [13] C. Magnes, F.M. Sinner, W. Regittnig, T.R. Pieber, Anal. Chem. 77 (2005) 2889. [14] J.W. Armando, B.A. Boghigian, B.A. Pfeifer, Lett. Appl. Microbiol. 54 (2012) 140. [15] P.E. Minkler, J. Kerner, T. Kasumov, W. Parland, C.L. Hoppel, Anal. Biochem. 352 (2006) 24. [16] L. Gao, W. Chiou, H. Tang, X. Cheng, H.S. Camp, D.J. Burns, J. Chromatogr. B: Analyt. Technol. Biomed. Life Sci. 853 (2007) 303. [17] C. Magnes, M. Suppan, T.R. Pieber, T. Moustafa, M. Trauner, G. Haemmerle, F.M. Sinner, Anal. Chem. 80 (2008) 5736. [18] C.A. Haynes, J.C. Allegood, K. Sims, E.W. Wang, M.C. Sullards, A.H. Merrill, J. Lipid Res. 49 (2008) 1113. [19] F. ter Veld, M. Mueller, S. Kramer, U. Haussmann, D. Herebian, E. Mayatepek, M.D. Laryea, S. Primassin, U. Spiekerkoetter, PLoS One 4 (2009) e6449. [20] R.R. Gilibili, M. Kandaswamy, K. Sharma, S. Giri, S. Rajagopal, R. Mullangi, Biomed. Chromatogr. 25 (2011) 1352. [21] O. Hayashi, K. Satoh, Biosci. Biotechnol. Biochem. 70 (2006) 2676. [22] M.A.D.N. Perera, S.-Y. Choi, E.S. Wurtele, B.J. Nikolau, J. Chromatogr. B: Analyt. Technol. Biomed. Life Sci. 877 (2009) 482. [23] A.V. Qualley, B.R. Cooper, N. Dudareva, Anal. Biochem. 420 (2012) 182. [24] Y. Shoyama, H. Hirano, I. Nishioka, Phytochemistry 23 (1984) 1909. [25] C.C. Leandro, P. Hancock, R.J. Fussell, B.J. Keely, J. Chromatogr. A 1103 (2006) 94. [26] R.W. Purves, J.M. Stout, S.J. Ambrose, J.E. Page, Proc. 59th ASMS Conf. Mass Spectrom., Allied Top., Denver, CO, 2011. [27] P.E. Minkler, J. Kerner, S.T. Ingalls, C.L. Hoppel, Anal. Biochem. 376 (2008) 275. [28] S.J. Ambrose, Q. Li, Z. Jitao, R.W. Purves, Proc. 60th ASMS Conf. Mass Spectrom., Allied Top., Denver, CO, 2012. [29] US FDA Center for Drug Evaluation and Research. Guidance for Industry, Analytical Procedures and Methods Validation for Drugs and Biologics, February 2014, available from: http://www.fda.gov/downloads/drugs/ guidancecomplianceregulatoryinformation/guidances/ucm386366.pdf