Antiproliferative Acylated Glycerols from New Zealand PropolisStephen Bloor,*,† Owen Catchpole,†,‡ Kevin Mitchell,† Rosemary Webby,† and Paul Davis§

†Callaghan Innovation, 69 Gracefield Road, PO Box 31310, Lower Hutt 5040, New Zealand‡Manuka Health NZ Ltd, PO Box 87429, Meadowbank, Auckland 1742, New Zealand§Trinity Bioactives, Lower Hutt 5040, New Zealand

*S Supporting Information

ABSTRACT: Previous work has shown that a number ofphenolic components of NZ propolis possess antiproliferativeactivity against certain human gastrointestinal cancer cell lines.Here we report on a series of acylglycerols isolated from thenonpolar fraction of propolis resin, which represent furtherbioactive constituents unrelated to the more usual phenoliccompounds generally found in propolis. NZ propolis issourced from poplar trees, and the acylglycerols have beenshown to be present in the leaves and buds of some commonpoplars. The compounds are a series of monoglycerides containing 3,8-dihydroxy fatty acids, many of which are further acylatedwith acetic acid residues. The dihydroxy fatty acids are C18 to C24, with the most abundant being C20 and C22. Theseacylglycerols were found to have strong antiproliferative activity against three human gastrointestinal cell lines, particularlygastric cancer cell line NCI-N87, where one example shows an IC50 of less than 50 μM.

Propolis is a heterogeneous material consisting of resincollected by honey bees from the leaf, buds, and bark of

certain tree species, which is then admixed with beeswaxproduced from the hypopharyngeal glands of the worker bees.Propolis is used by bees to defend the hive against invaders andto reduce air flow into the hive to retain heat. For human use,crude propolis is typically extracted with an aqueous ethanolicsolvent to separate the more polar resin from the nonpolarwax. The resultant tincture is used to make a variety ofproducts that take advantage of antioxidant and anti-inflammatory activity for oral consumption (tablets, capsules,lozenges, throat sprays, tinctures) or antibacterial activity forexternal application (toothpastes, soaps, skin care products,hair care products).There are around eight or nine distinct varieties of propolis

according to the botanical source of the resin.1−3 The mainbotanical sources of resins and characteristic chemicalcomponents are poplars (Europe, North America, southernSouth America, China) rich in aglycone flavonoids, birch(Russia and Eastern Europe) rich in flavones and flavonols,Baccharis spp. (Brazilian “green” propolis) rich in prenylatedderivatives of coumaric acid, Cupressacaea (Mediterraneanpropolis) rich in labdane-type diterpenoids, and Clusia spp.(central South America and southern Brazil, “red propolis”)rich in propolones.2

New Zealand propolis can be categorized as “European”, i.e.,obtained by honey bees mainly from the exudates of poplars. Acomparison of the antioxidant activity and total phenol andflavonoid and individual phenol and flavonoid composition forEtOH-extracted propolis samples from 14 countries showedthat New Zealand-sourced propolis is similar in composition topropolis from Bulgaria, Uzbekistan, and Hungary and to

propolis from three South American countries: Chile, Uruguay,and Argentina.1

In previous work we showed that the antiproliferativeactivity of New Zealand “poplar”-type propolis against a groupof human gastrointestinal cancer cell lines could be ascribed tomore than a dozen phenolic compounds.4 We also showed thatantiproliferative activity was retained by encapsulation ofpropolis in cyclodextrins.5 Apart from the well-known phenolicbioactives such as chrysin and CAPE (caffeic acid phenethylester) this work identified a number of flavonoids and caffeic-type esters with strong activity.4 The major contribution to thetotal activity of the propolis tincture and resin was attributed tothe main active compounds (chrysin, pinocembrin, galangin, 3-O-acetylpinobanksin), as these compounds are the mostdominant phenolic compounds in New Zealand propolis.Caffeic acid and its esters also collectively contributed to theactivity of propolis. 1,1-Dimethylallyl caffeate, benzyl caffeate,and 3-methyl-3-butenyl caffeate all had strong activity and arepresent in NZ propolis in higher concentration than the knownbioactive compound CAPE. Phenolic glycerides have also beenreported in poplar-type propolis.6,7 The esters comprise 2-acetyl-1,3-diglycerides of ferulic and/or cinnamic acid6 and a 2-acetyl-1-feruloyl-3-(3,16)-dihydroxypalmitoylglycerol.7 Thesecompounds had moderate antiproliferative and anti-inflamma-tory activity in in vitro testing.In the first stage of the previous work we also noted that one

of the most antiproliferative fractions was still underinvestigation. This fraction (S#8) had no identifiable phenolics

Received: July 10, 2018

Article

pubs.acs.org/jnpCite This: J. Nat. Prod. XXXX, XXX, XXX−XXX

© XXXX American Chemical Society andAmerican Society of Pharmacognosy A DOI: 10.1021/acs.jnatprod.8b00562

J. Nat. Prod. XXXX, XXX, XXX−XXX

Dow

nloa

ded

via

O C

atch

pole

on

Sept

embe

r 11

, 201

9 at

21:

46:5

4 (U

TC

).Se

e ht

tps:

//pub

s.ac

s.or

g/sh

arin

ggui

delin

es f

or o

ptio

ns o

n ho

w to

legi

timat

ely

shar

e pu

blis

hed

artic

les.

and required further work to identify the active compounds.This paper reports the results of further work on the activefraction leading to the identification of some novelacylglycerols identified for the first time not only in propolisbut in the source resins from poplar trees. The chemistry of thenew compounds is reported here as well as the antiproliferativeactivity of these acylglycerols against a range of humangastrointestinal cell lines.

■ RESULTS AND DISCUSSION

Initial Separation from Propolis Resin. In earlier workwe described the in vitro bioactivity-guided fractionation of“Bio30” propolis using both anti-inflammatory (TNF-α, COX-1, COX-2) and anticolon cancer (DLD-1 colon cancer cellviability) assays and determined the phenolic compoundsresponsible for the activity.4 This work identified a range ofphenolic compounds responsible for much of the observedactivity. One fraction in particular (fraction S#8) was notpursued at the time, as the activity did not relate to anyphenolic-type compounds. Identification of the active com-pounds in this fraction is now reported in this work.The active fraction of interest was one of the most nonpolar

fractions eluted from a reverse-phase (RP) silica chromato-graphic separation. This fraction showed good levels ofantiproliferative activity against a human colon cancer cellline, DLD-1. Further fractionation into four subfractionsshowed activity to mainly reside in just one of thesesubfractions. This subfraction showed no UV-absorbingpeaks in HPLC; however evaporative light-scattering detection(ELSD) indicated the presence of several peaks in the area ofinterest. Chromatograms showing the relative retention time ofthese peaks in a typical propolis tincture sample are shown inFigure 1. In order to observe these peaks in the ELSD, theamount of tincture injected was much higher than for a typical

analysis, as can be seen in some of the peaks “topping out” inboth UV and ELSD channels. The workup of more propolisresin through a series of chromatographic steps including RPand normal-phase silica gel chromatography allowed theseparation of sufficient material for structural analysis.

Compound Identification. The nonpolar sharp and well-resolved peaks observed in the ELSD could be linked to a setof peaks seen in the LCMS runs of the crude propolis andenriched fractions. No UV absorption is associated with thesepeaks. Mass spectrometry shows prominent peaks in positivemode where [M + H]+ and [M + Na]+ ions were observed.The main components of this set of compounds havemolecular weights of 460, 474, 488, and 502 Da. This suggestsa homologous series where the components differ by amethylene group. All the mass spectra have commonfragmentation ions at m/z 159 and 117. In the negativemode the compounds showed formate adduct ions at [M +45]+. High mass accuracy MS suggested a molecular formula ofC25H48O7 for the 460 Da compound and extra CH2 units forthe others in the series.The mixture of compounds was quite complex and proved

difficult to separate from the background “polymeric” material.Peracetylation of the crude fractions enriched in thesecompounds proved to be useful. Not only did this allowseparation of the acetylated compounds from the othermaterial but the spectroscopic data were more definitive.The mass spectra of the acetylated mixture showed thecompounds had gained three acetyl groups, as indicated by theMW increase by 3 × 42 amu. The compounds of interest werealso readily separated by silica gel chromatography to yieldsufficient material for NMR spectroscopic analysis. Althoughthe material was still a mixture of the four major compounds,the NMR spectrum was surprisingly well resolved. The spectrashowed there were in fact four acetate groups, indicating one of

Figure 1. RP-HPLC chromatogram of New Zealand propolis tincture with detection at 268 and 320 nm and ELSD. Boxed area in the ELSDchromatogram contains the set of acylglycerol peaks.

Journal of Natural Products Article

DOI: 10.1021/acs.jnatprod.8b00562J. Nat. Prod. XXXX, XXX, XXX−XXX

B

the acetate moieties was present in the natural compound. Atrisubstituted glycerol was present, and a fatty acid attached atone of the hydroxy groups of the glycerol moiety was alsohydroxylated at C-3. The rest of the molecule was composed ofthe remaining fatty acid chain, which also had a secondhydroxy group. The likely chain length of the fatty acid,therefore, was C20, C21, or C22 for the main compounds.Detailed analysis of the NMR data (Table 1) identified the

most likely placement of the second hydroxy group on the fattyacid to be at C-8. C-2 of the fatty acid is an isolated methylenegroup and shows long-range HMBC correlations to C-3 (70ppm) and C-4 (34 ppm). The H-3 methine proton showslong-range correlations to C-4 (34 ppm) and C-5 (25 ppm).The other methine proton (H-8) shows long-range correla-tions to C-6 and C-10 (both at 25 ppm) and C-7 and C-9(both 34 ppm). Also, the C-4, C-7, and C-9 protons showHMBC correlations to the multiple carbons at 25 ppm (C-5,-6, and -10). No correlation is seen between H-4 and any ofthe carbons with a chemical shift around 29 ppm. The C-4protons are a clear multiplet, while H-7 and H-9 areoverlapped in an isolated multiplet. No direct H−H couplingis seen between H-4 and H-7 or H-9. A long-range correlationis seen between H-7/9 and a carbon with a chemical shift at 29ppm, which must be C-10.These data fit a pattern of chemical shift order that is

consistent with a 3,8-diacetoxy fatty acid. For C-8 the chemicalshifts of both the α carbons are at 34 ppm, while the β carbonsresonate at 25 ppm. Similarly, C-4 and C-5 resonate at 34 and25 ppm, respectively. The chemical shift pattern would besomewhat different if the second acetoxy group was placed atC-9 (the C-5 signal would have a higher chemical shift and aprobable long-range correlation with H2-4). The 3,8-

disubstitution pattern is also supported by MS data analysis(see below).The positive ion mass spectra (LCMS, ESIMS/MS) all show

fragment ions at m/z 159 and m/z 117. These are most easilyaccounted for by a C7H11O4 fragment (three indices ofhydrogen deficiency) comprising the C-1−C-5 part of the fattyacid with an acetate group at C-3. The loss of the acetate thengives the other common fragment at 117 amu. The proposedstructure for the peracetylated glyceride is shown in Figure 2with the putative mass spectrometric fragmentation pathway.

Further information as to the structural features of the fattyacid moiety were obtained by analysis of the mixture of fattyacids obtained after base hydrolysis of the glyceride mixture.The free fatty acids were liberated by treatment with LiOH andmethylated by diazomethane treatment to yield a fatty acidmethyl ester (FAME) mixture.8 GCMS analysis of thetetramethylsilyl (TMS) derivatives of the FAME mixtureshowed a set of 13 acid derivatives (Figure 3). The main peaksrepresent a homologous series. Peak 4 has a characteristic ionat m/z 383 (corresponding to loss of methyl and TMSOH),and similar ions are seen for peaks 5 (m/z 397), 7 (m/z 411),

Table 1. 1H NMR and 13C NMR Data (δ) for 1, 2, and 8 (500 MHz) (δ in ppm, J in Hz)a

1b 2c 8c

pos. δC, type δH (J in Hz) δC δH (J in Hz) δC δH (J in Hz) HMBC

1′ 171.6, C 172.7 169.92′ 42.4, CH2 2.39, dd, (15, 8); 2.49, dd

(15,4.4)41.6 2.45, dd, (15, 8); 2.55 dd,

(15, 4)39.1 2.57, AB dq (9.6, 7.3) 1′, 3′, 4′

3′ 67.8, CH 4.00, br s 68.2 4.02, br s 70.2 5.18, tt, (7.2, 5.4) 2′, 4′, 5′4′ 37.1, CH2 36.6 1.5−1.6, m 33.9 1.6, m 3′, 5′5′ 25.6, CH2 25.5 1.25−1.5, m 25.0 1.25−1.4, m6′ 25.6, CH2 25.7 1.25−1.5, m 25.0 1.25−1.4, m7′ 37.7, CH2 37.3 1.4−1.5, m 33.9 1.51, m8′ 70.6, CH 3.53, br s 71.9 3.6, br s 74.1 4.85, q (7.1) 10′ (or 6′), 7′,

9′9′ 37.6, CH2 37.6 1.4−1.5, m 34.1 1.51, m10′ 25.5, CH2 25.5 1.25−1.5, m 25.3 1.25−1.4, m11′-17′ 29.3, CH2 29.3 1.25, br s 29−30 1.25, br s18′ 31.7, CH2 31.9 1.25, br s 31.9 1.25, br s19′ 22.4, CH2 22.6 1.25, br s 22.7 1.25, br s20′ 13.42,

CH2

0.87, t (7.0) 14.1 0.88, t (7.0) 14.1 0.87, t, (6.5)

1 63.4, CH2 4.1−4.3, m 65.1 4.15−4.3, m 62.5 4.29, 4.32. both dd(12.3, 4.4)

2, 3

2 69.9, CH 3.82, m 68 4.1, m 69.4 5.23 1, 33 63.2, CH2 3.59, br s 65.1 4.15−4.3, m 62.2 4.15, 4.16 both dd (10.6,

4.4)2, 3

acetatesCH3

20.8 2.11, s 20.6, 20.8, 21.0, 21.2 2.09, 2.07, 2.03, 2.03 s

acetate CO 171.1 170.0, 170.3, 170.4,170.9

aHMBC correlations are from proton(s) stated to the indicated carbon. bAcetone-d6 solvent.cCDCl3 solvent.

Figure 2. Structure of C20 glyceride peracetate (8) and MSdegradation to give m/z 159 and m/z 117 ions characteristic ofglyceride series.

Journal of Natural Products Article

DOI: 10.1021/acs.jnatprod.8b00562J. Nat. Prod. XXXX, XXX, XXX−XXX

C

10 (m/z 425), 11 (m/z 439), and 12 (m/z 453), so peaks 4−9correspond to the C19−C24 dihydroxy fatty acids (methyl esterTMS ethers). Another feature was fragments at m/z 301 andm/z 247 seen for all the peaks. These fragments, especially thefragment with m/z 301, are characteristic of 3,8-dihydroxyacids.9 The other possible dihydroxy acids (3,6; 3,7; or 3,9)would be expected to have different fragmentation patternsbased on published data.9 For example, the large peak 5 (C20dihydroxy) has a mass spectrum [m/z (rel intensity), 397(43),301(100), 271(76), 247(37), 73(71)] that is similar to thatdescribed for methyl 3,8-bis[trimethylsilyloxy]eicosanoate[literature values: 487(1), 397(5), 333(2), 301(26), 271(24),247(3), 243(1), 211(4), 175(7), 143(3), 129(13), 73(100)].Thus, the dihydroxy acids in the glyceride fraction are theC19−C24 3,8-dihydroxy fatty acids.Note that the relative peak size indicates that the C20 (peak

5) and C22 (peak 10) chain lengths are probably the dominantcompounds. Smaller minor peaks with highly similar MSfragmentation to the larger peaks were also seen (somedifference in relative abundance of low m/z fragments). Peak 9is similar to 10 and peak 12 to 13. It is probable these minorpeaks are the iso-versions of the C22 and C24 acids (i.e., a

branched methyl at the terminal end). This fits with evidencefrom NMR data on the peracetylated mixture, which suggestedminor amounts of methyl branching.The LCMS analysis of the crude propolis or partially

purified glyceride fractions is discussed next. As shown above,the acylglycerols have characteristic MS patterns and showsome common fragment ions, which facilitate analysis usingLCMS data. The major compounds are the monoacetates withMWs of 460, 474, 488, and 502 Da, corresponding to C20−C23fatty acid glycerides with one acetate group.A typical propolis acylglycerol profile is shown in Figure 4.

This is a crude propolis resin sample obtained by evaporationto dryness of Manuka Health Bio-30 tincture. The chromato-grams are those resulting from multiple reaction monitoring(MRM) transitions specific for each type of compound, andonly the relevant portion of the total chromatogram is shown.Generally, the MRMs chosen were those for the molecular ionthat yielded an ion at m/z 117 or 159. Peak 1 is the non-acetylated C20 glyceride. Peaks 2, 3, 5, and 6 are themonoacetates of the C20−C23 acylglycerols. Peak 4 is probablythe iso-analogue of the C22 glyceride monoacetate, as it has thesame MW as the compound in peak 5. Peak 7 represents the

Figure 3. GCMS chromatogram TIC of TMS derivatives of methylated fatty acids from the glyceride fraction.

Figure 4. RP-LCMS-MRM chromatogram of acylglycerols from propolis (only segment shown).

Journal of Natural Products Article

DOI: 10.1021/acs.jnatprod.8b00562J. Nat. Prod. XXXX, XXX, XXX−XXX

D

diacetate of the C20 glyceride. The general pattern ofcompounds is somewhat similar to that seen above for theGCMS analysis of the methylated free fatty acids.At this stage the location of the fatty acid attachment to the

glycerol moiety was not resolved, the site of acetylation wasuncertain, and none of the individual glyceride compounds hadbeen purified. Using a combination of preparative RPHPLCand size exclusion chromatography several of the individualacylglycerols were obtained in purity suitable for NMRspectroscopic analysis. The NMR data for two of thesecompounds, the major non-acetylated compound 1 and themajor isolated monoacetate 2, are presented in Table 1. TheNMR spectra of the individual purified monoacetates 2, 3, and5 were difficult to distinguish from each other, as the minordifferences were buried in the aliphatic part of the NMRspectrum. The position of the acetate group in themonoacetate 2 could be assigned by comparison of theNMR data (Table 1) for the non-acetylated C20 compound, 1,the monoacetylated C20 compound, 2, and the peracetate, 8. Inparticular, only the C-3′ protons of the glycerol unit in themonoacetate 2 show chemical shifts consistent with acetylation(3.59 in 1 vs 4.15−4.30 in 2). An HMBC correlation betweenthe C-3′ protons and the acetate carbonyl was also observed.This shows the monoacetate compounds have the glycerolacylated at the 3 and the acetyl group at the 1 position; that is,compound 2 is 1-O-acetyl-(3′,8′-dihydroxyeicosanoyl) glycer-ol. The LCMS fragmentation pattern initially suggestedacetylation at the C-3 of the fatty acid, as this explains thefragment ion at m/z 159. It is possible that this fragment ionarises from trans-acetylation before loss of the glycerol moiety.A sufficient quantity of the diacetate could not be isolated fordefining the location of the second acetate group in compound7.Based upon analysis of propolis samples using LCMS and a

standard of the major purified acylglycerol, 2, the acylglycerolsare present in a typical propolis tincture at approximately 0.2%(w/w) of the weight of dewaxed propolis solids.A search of the literature using the molecular formulas and

structures determined for the acylglycerols in propolis showedonly one other example of this specific type of compound, aseries of acylglycerols isolated from the leaf exudate ofPaulownia sp.9 Thirty acylglycerols were isolated fromPaulownia, and many were acetylated. The acetates werevariously located on either the fatty acid or the glycerol moiety,resulting in the large number of analogues. However, onhydrolysis, only C18 and C20 mono- and dihydroxy fatty acidswere seen. In the New Zealand propolis it is likely a similarrange of degrees of acetylation is present, but the fatty acids areC20, C21, and C22 hydroxy fatty acids. This is somewhatunusual, as generally fatty acids are synthesized from twocarbon units, and hence the resultant series would normally beC18, C20, C22, etc. Also, the hydroxy fatty acids in compoundsfrom Paulownia are positioned at the C-2 position of theglycerol moiety.Stereochemistry. Further work was required to determine

the configuration of the oxygenated stereogenic centers of the3,8-dihydroxy fatty acids. As the length of the fatty acid chainhas no effect upon the chemical shift of the protons at C-2, -3,and -8, the crude glyceride mixture was used for this work. Thecrude mixture was hydrolyzed using LiOH as describedpreviously. The free fatty acid was converted to the methylester using diazomethane. Samples of the methyl ester werethen derivatized using R- or S-α-methoxyphenylacetic acid

(MPA). In each case a diester was formed with the chiralreagent attached to the hydroxy groups at C-3 and C-8. Theproducts were examined by 1H NMR. Only the signals for H-2,H-3, and H-8 are sufficiently well separated for detailedanalysis.The changes in the shifts of these derivatives [Δδ(S−R),

Table 2] can then be compared with relevant examples from

the literature. There are three examples where directcomparisons are valid; these are the floral oil compoundsbyrsonic acid10 and oncidinol8 and the acylglycerols fromPaulownia.9

In the case of byrsonic acid, hydrolysis produced a 3,7-dihydroxy fatty acid. Preparation of the diesters using chiralMosher’s acid [methoxy(trifluoromethyl)phenylacetic acid]produced changes (Table 2) that permitted assignment ofthe acid as (3R,7R)-3,7-dihydroxydocosanoic acid. Similarly,for oncidinol, the acid was assigned as a (3R,6R)-dihydroxyacid based on the results shown in Table 2 using Mosher’sacid. In both examples, the results are similar to those seen forthe propolis acylglyceride sample in Table 2. The Δδ(S−R)values are negative for H-2 and H-3 and positive for the remotehydroxy position.Paulownia has a range of dihydroxy fatty acids acylated to

glycerol including 3,8-dihydroxy fatty acids as seen in thepropolis acylglyceride fraction, enabling a direct comparison ofthe same hydroxylation pattern. Chiral derivatives wereprepared from these compounds.9 However, the chiral reagentused was a naphthalene derivative that is not commerciallyavailable. However, the same general shielding/deshieldingeffects should be observed. In the work described for thePaulownia acids the individual 3- and 8-monoesters wereprepared, and the chemical shift differences are shown in Table3.The authors concluded a 3R, 8R absolute configuration

based on these results. Along with preparing these esters theyalso prepared Mosher’s acid derivatives using the 3-acetylversions of the acylglycerols; that is, only the C-8 position wasderivatized with the chiral reagent. The Δδ(S−R) values seen

Table 2. 1H NMR Chemical Shift Differences between (R)-and (S)-MPA Esters for 3,8-Dihydroxy Fatty Acid MethylEsters from Propolis Compared with a Similar Experimentwith Byrsonic Acid10 and Oncidinol (Using Mosher’s Acid)8

hydrogen

δdihydroxyFAME

δ di ResterFAME

δ di SesterFAME

Δδ(S−R)

Δ δ(S−R)

byrsonic

Δ δ(S−R)

oncidinol

Ha-C(2) 2.5 2.6 2.43 −0.17 −0.11 −0.11Hb-C(2) 2.41 2.5 2.35 −0.15 −0.13 −0.12H-3 4.01 5.22 5.15 −0.07 −0.08 −0.06H-8 3.57 4.76 4.86 +0.10 +0.07

(H-7)+0.08(H-6)

Table 3. 1H NMR Chemical Shift Differences between (R)-and (S)-Naphthylmethoxyacetic Acid Esters for SelectedPaulownia Compounds (3,8-Dihydroxy Fatty Acids)9

hydrogen 3-mono ester Δδ(S−R) 8-mono ester Δδ(S−R)Ha-C(2) +0.11 +0.18Hb-C(2) +0.1 +0.17H-3 +0.1 +0.25H-8 −0.23

Journal of Natural Products Article

DOI: 10.1021/acs.jnatprod.8b00562J. Nat. Prod. XXXX, XXX, XXX−XXX

E

for H-2 and H-3 with the Mosher’s reagent were negative,showing that the naphthyl reagent produces the oppositeΔδ(S−R) values to the Mosher’s esters for the samestereogenic center. As the reagent we used for the chiralderivatization of the propolis glyceride dihydroxy fatty acid issimilar to Mosher’s acid (a trifluoromethyl group at the α-position vs a hydrogen), we can confidently assign the propolisglyceride dihydroxy fatty acid as (3R,8R)-3,8-dihydroxyeicosa-noic acid (or heneicosanoic/dodecanoic acid for the othermajor fatty acids) as shown in Figure 5.

The assignment of the (3R, 8R) absolute configuration isalso consistent with known similar compounds, suggestingthese compounds are probably derived from a similarbiosynthetic pathway.Ideally individual monoesters should also be prepared for

the compounds under study. This would allow for any effectsdue to the overlapping effects of the two aromatic estermoieties. However, in this case it was difficult to just formmonoesters, as the reaction tended to go to completion quitequickly and a 1:1 mixture of the acid and ester also producedmostly diester. As there are suitable examples of other similardiester compounds, attempts to prepare monoesters werediscontinued.Plant Source of Acylglycerols. The source of the

acylglycerol compounds in NZ propolis is from the exudatesof some species and hybrids of poplar trees. Analysis of leaf andtwig exudates using LCMS shows the presence of the sameacylglycerols in the tree resin extracts at about the same relativequantity as seen in propolis resin. Table 4 presents the resultsof analysis of leaf and twig extracts of a range of commonpoplar trees found in New Zealand. The analysis showed that

the same dihydroxy fatty acid acylglycerols seen in propoliswere present in many of the more commonly planted NewZealand poplar cultivars analyzed. The acylglycerols werefound on leaves, twigs, and buds of many poplars and appearedto be generally associated with the resin on these plant parts. Itis anticipated that the individual acylglycerols could be isolatedfrom poplar exudates in a similar manner to that used forpropolis. Poplar cultivars such as “Tasman”, “Selwyn”, and“Fraser” are typical examples of the widely planted Populus ×euramericana. The dihydroxy fatty acid acylglycerols weregenerally found to be absent in some of the other cultivars suchas the P. maximowiczii × nigra clones and the Chinese poplarspecies P. yunnanensis. Acylglycerols were absent in hybrids notcontaining a cross with P. deltoides. These types of glycerideswere also absent in samples of willow leaf and buds, while inPaulownia plant parts sampled in New Zealand low levels ofdifferent types of acylglycerols were present, showing thatthese plant species are not the source of the acylglycerols inNew Zealand propolis. Figure 6 shows an LCMS-MRMchromatogram for acylglycerols from P. euramericana cultivar“Fraser” leaf extract, which qualitatively shows a similar patternto the corresponding propolis glyceride fraction in Figure 4.Although the Paulownia compounds are structurally the

closest to the propolis acylglycerols, there are a number ofreports of other acylglycerols from other plant exudates.11−13

In each case the fatty acid was reported as attached at the 2-position. Although acyl migration may readily occur duringextraction and workup, it would appear unlikely that thepropolis acylglycerols have undergone acyl migration to themore stable C-1 analogues since the source of poplar exudatesshows the same pattern of compounds. In poplar, the resins arereportedly produced by stipules, which start producing resin atthe bud stage,14 whereas the acylglycerols mentioned abovearise from glandular trichome exudates.11−13

Antiproliferative Activity. The relative antiproliferativebioactivities of the two major monoacetates, 2 and 3, as well asthe acylglycerol-rich concentrate fraction were examinedagainst three human gastrointestinal cancer cell lines toconfirm that the isolated compounds were the source of theantiproliferative activity observed in earlier work.4 Samples of 1and the deacetylated versions of 2 and 3 were also tested(these were isolated from preparative HPLC and derived fromhydrolysis during workup). The initial bioassay-guidedfractionation work was performed using DLD-1 colonadenocarcinoma cells. These compounds were also tested foractivity versus two other cell lines, KYSE-30 (esophagealsquamous cancer) and NCI-N87 (gastric carcinoma), andcompared with the positive controls glycerol monopalmi-tate15,16 and the anticancer drug 5-fluorouracil. The results areshown in Table 5 and confirm the high level of activity in theseassays for the acylglycerols. The level of activity is similar to orgreater than that which we reported for other propoliscompounds, such as pinocembrin and the various caffeates.4

However, unlike the phenolic compounds, which are less activeagainst the gastric carcinoma NCI-N87 cell line, theacylglycerols are highly inhibitory of proliferation and cytotoxicat 110 μM.

■ EXPERIMENTAL SECTIONGeneral Experimental Procedures. Propolis resin was obtained

from Manuka Health NZ Ltd. as bulk samples and as Bio-30 tincture(commercially available). R- and S-Methoxy-phenyl acetic acid, N-(3-(dimethylamino)propyl)-N′-ethylcarbodiimide, and N,N-dimethyla-

Figure 5. Compounds 1 (R = H, n = 10), 2 (R = Ac, n = 10), 3 (R =Ac, n = 11), and 4 (R = Ac, n = 12).

Table 4. Qualitative 3,8-Dihydroxy Fatty Acid AcylglycerolContent in New Zealand Poplar Cultivarsa

plant part

cultivar poplar description twig leaf

Kulu P. ciliata Wall. ex Royle np npOxford P. maximowiczii Henry × berolinensis Dipp. np npPecam P. maximowiczii × nigra np npSelwyn P. × euramericana Guinier (= × canadensis

Moench)+++ +++

Tasman P. × euramericana Guinier (= × canadensisMoench)

++ ++

LuisaAvanzo

P. × euramericana Guinier (= × canadensisMoench)

+++ +++

Fraser P. × euramericana Guinier (= × canadensisMoench)

+++ ++

Yunnan P. yunnanensis Dode np npGeyles P. maximowiczii × nigra np npItalica P. nigra L. np np

aLevels estimated from LCMS data on plant extracts: np =acylglycerols not present, ++ levels of acylglycerols somewhat lowerthan seen in propolis, +++ acylglycerol levels similar to those inpropolis.

Journal of Natural Products Article

DOI: 10.1021/acs.jnatprod.8b00562J. Nat. Prod. XXXX, XXX, XXX−XXX

F

minopyridine were purchased from Sigma-Aldrich (St. Louis, MO,USA). NMR spectra were recorded using a Bruker Avance III 500MHz NMR spectrometer (Bruker, Karlsruhe, Germany). LCMSanalysis was performed using a Shimadzu LCMS system with aNexera UPLC and an 8040 triple quad MS (Shimadzu, Kyoto, Japan).The system was operated using an ESI source in both negative andpositive modes. The other conditions used were nebulizer gas flow of3 L/min, desolvation line temperature of 250 °C, heating blocktemperature of 400 °C, and drying gas flow of 15 L/min. A WatersQTOF MS was also employed for direct infusion ESIMS analysis ofsome samples (Waters, Milford, MA, USA). Preparative HPLC wasperformed using a Gilson 321 pump and PDA detector (Gilson,Middleton, WI, USA). HPLC analysis was performed using a WatersH-class UPLC system equipped with PDA and ELSD detectors(Waters, Milford, MA, USA). The solvent program used for propolisanalysis (e.g., as shown in Figure 1) is similar to that described belowfor LCMS analysis of acylglycerols. The IR instrument used was aBruker Tensor with ATR insert (Bruker, Karlsruhe, Germany).GCMS analysis was performed with a Shimadzu QP2010 Ultrainstrument equipped with a Restek RTX5 ms 30 m column using atemperature program of 170−320 °C, with a split ratio of 20(Shimadzu, Kyoto, Japan).Bioactivity-Guided Isolation of Propolis Fraction. The 90%

aqueous EtOH elution fraction (propolis fraction S#8) produced inthe work reported earlier4 was further fractionated using preparativeHPLC as follows. The propolis fraction S#8 was dissolved in neatEtOH and chromatographed by preparative HPLC on a Phenomenex

Synergi C12 column (4 μm, RP Max 80 Å, 250 × 30 mm). Injectionvolumes between 0.5 and 1.5 mL and a flow rate of 20 mL/min wereemployed. Solvents were 80% aqueous MeOH (containing 0.1%trifluoroacetic acid) and EtOAc/MeOH (4:1 v/v). The initial eluentcomposition consisted of 80% aqueous MeOH. The solventcomposition was held for 5 min at the initial conditions before theEtOAc/MeOH solvent concentration was increased linearly to 100%over 35 min. The chromatography was carried out at roomtemperature (18−20 °C). Fractions were collected manually withonline detection carried out at 210, 268, and 327 nm and also byevaporative light scattering detection. The main components of thisfraction were quite nonpolar, eluting late in the chromatogram, andshowed minimal UV absorption at the wavelengths typically used foranalysis on other propolis fractions containing phenolics, i.e., 268 and327 nm. However, ELSD revealed a complex mixture of components.Owing to the destructive nature of ELSD, preparative HPLC fractionswere collected manually with online detection carried out at 210 nm.Based on the similarity of analytical profiles, the collected fractionswere pooled into four fractions. These pooled fractions weresubjected to the bioassay using DLD-1 cells as described earlier.4

The most active fraction was S#8 90% F3, which completely inhibitedproliferation and was cytotoxic at the concentration tested (200 μg/mL). Analysis of this active fraction by HPLC-ELSD and LCMSshowed that a series of sharp peaks with characteristic mass spectrawere concentrated in this fraction.

Purification of Acylglycerols. A large-scale extract was preparedfrom 1.1 kg of dewaxed propolis resin using EtOAc (3 L) as the

Figure 6. RP-LCMS chromatogram (MRM) showing acylglycerols in P. euramericana “Fraser” leaf extract.

Table 5. Antiproliferative Activity of Propolis Acylglycerols against Human Cell Lines DLD-1, KYSE-30, and NCI-N87 (%Inhibition of Proliferation versus Cells-Only Control)a

cell line (% inhibition of proliferation, 110, 55, or 22 μM)

DLD-1 NCI-N87 KYSE-30

compound 110 55 22 110 55 22 110 55 22

acylglycerol concentrate 82 83 44acylglycerol 1 80 100 52acylglycerol 1b 76 100 29acylglycerol 1c nd 100 82acylglycerol 2 94 13 nd 100 86 39 52 50 9acylglycerol 3 91 100 29MGP (glycerol mono- palmitate) 72 100 75

aAcylglycerols 1b and 1c are equivalent to compounds 2 and 3 without the acetyl group. bPositive control: 5-fluorouracil at 450 nM gave 21, 8, and17% inhibition against DLD-1, NCI-N87, and KYSE-30, respectively. cnd = not determined.

Journal of Natural Products Article

DOI: 10.1021/acs.jnatprod.8b00562J. Nat. Prod. XXXX, XXX, XXX−XXX

G

extraction solvent. The propolis resin was coarsely chopped andsoaked in EtOAc at room temperature with mechanical stirring for 2h. The resulting solution was filtered but not concentrated. Around20% of this extract solution was chromatographed using silica gel,after first absorbing the extract onto a portion of silica gel (50 g) andevaporating the EtOAc using a rotary evaporator. The preabsorbedsolid was applied to the top of a large silica gel column (100 × 45mm) and eluted with 1:1 hexanes/Et2O (v/v), and 8 × 150 mLfractions were collected. Then the column was eluted with Et2O tocollect a further four 150 mL fractions before beginning elution withEtOAc. The four EtOAc fractions #13−#16 (F13−F16) were muchdarker in color than the earlier fractions. A final fraction F17 wascollected by elution with acetone. Silica TLC (visualization usingphosphomolybdic acid in EtOH followed by heat) and LC-MS wereused to track progression of the purification process. The acylglycerol-rich fractions from silica gel chromatography (fractions 13−17, wt 1.6g) were combined and subjected to RP column chromatography(Machery-Nagel Chromabond C18 70 mL 10 g SPE type column)using mixtures of water and EtOH without acid to yield a finalacylglycerol-rich fraction. This process was repeated with theremaining EtOAc extract as required.A portion of the combined fraction was used for preparative HPLC

followed by Sephadex LH-20 size exclusion chromatography to obtaintwo monoacetates, 2 and 3 (90 and 74 mg, C20 and C21 dihydroxyfatty acid monoacetates, respectively), and three non-acetylatedcompounds, 1, 7, and 8 (45, 16, and 23 mg; C20, C21, and C22dihydroxy fatty acid monoglycerides, respectively). The separationprogress was tracked using HPLC with ELSD, and the purity of theindividual compounds isolated was assessed to be >95% by HPLC,LC-MS, and NMR. Preparative HPLC was performed using aPhenomenex Synergi 4 μm-RP Max 80 Å 250 × 30 mm C12 column(gradient elution using water with 0.1% formic acid and MeCN), andfractions were analyzed offline using TLC and LC-MS. Data arereported for 1 and 2 only. Other compounds show almost identicalNMR data.Compound 1 (C20 non-acetylated): cream solid; [α]20D −2 (c 0.2,

acetone); FTIR (ATR) νmax 3350, 2915, 2848, 1728, 1455 cm−1;NMR data see Table 1; HRMS m/z 441.3201 (calcd for C23H46O6Na,441.3193).Compound 2 (C20 monoacetate): cream solid; [α]20D −0.1 (c 0.3,

CHCl3); FTIR (ATR) νmax 3345, 2914, 2849, 1722, 1380, 1264 cm−1;

NMR data see Table 1; HRMS (QTOF) m/z 483.3296 (calcd forC25H48O7Na, 483.3298).Hydrolysis of Acylglycerol Fraction. Hydrolysis of a sample of

the acylglycerol-rich fraction was performed as described in Reis etal.10 Approximately 100 mg of the acylglycerol-rich fraction was mixedwith 4 mL of THF/MeOH/H2O (3:1:1) and 55 mg of LiOH. Thiswas stirred at 0 °C for 30 min and then at room temperature for 48 h.The reaction mixture was acidified with HOAc (to pH ca. 2.8) andextracted with CHCl3. The dried extract was esterified withdiazomethane and subjected to silica gel chromatography usinghexanes/EtOAc mixtures. A fraction (40 mg) comprising one mainspot on silica TLC (Rf ca. 0.3 with 1:1 hexanes/EtOAc, visualizationwas by phosphomolybdic acid in EtOH dip followed by heat) wascollected.Preparation of Peracetate. Peracetylation of the acylglycerol-

rich fraction was carried out by dissolving the partially purifiedacylglycerol fraction (approximately 150 mg) in Ac2O (20 mL) andadding a small amount of N,N-dimethylaminopyridine. After stirringat room temperature overnight the reaction mixture was worked up byadding MeOH and toluene (approximately 20 mL of each) andevaporated to dryness using a rotary evaporator. The residue waschromatographed on silica gel using hexanes/EtOAc mixtures toobtain a purified peracetate mixture (8, approximately 80 mg).Peracetate (8): oil; [α]20D 0 (c 0.2, CHCl3); FTIR(ATR) νmax

2924, 2853, 1737, 1370, 1217 cm−1; 1H NMR data see Table 1;QTOF MS (positive ion m/z) 665.6, 651.5, 637.5, 623.5, 609.5, MS/MS on 665.6 (m/z) 605.6 (loss of acetic acid), 545.5 (loss of furtheracetic acid), 507.5 (loss of glycerol residue), 447.5 (loss of further

acetic acid); HRMS (QTOF) m/z 609.3618 (calcd for C31H54O10Na,609.3615).

LCMS Analysis of Acylglycerols. Chromatography wasperformed using a Supelco Ascentis Express C18 column (150 × 3mm, 2.7 μm, 90 Å) with a H2O (A):MeCN (B) gradient at a flow rateof 0.22 mL/min. The water contained 0.1% formic acid. The solventgradient time program was as follows: initial 20% B, hold for 3 min,then linear increase to 30% at 8 min, 40% at 37 min, 60% at 46 min,80% at 52 min, 100% at 8 min, and hold for 4 min before returning tostarting conditions. A shorter gradient was employed for more rapidanalysis. The Shimadzu 8040 MS detector was operated using aelectrospray ion source, and acylglycerols were analyzed by MRM inpositive mode. Transitions were m/z 461 to m/z 117 for the mainmonoacetate, with other transitions to the same daughter ion beingm/z 475, 489, and 503 for other monoacetates, 503, 517, and 531 fordiacetates, and 410 and 419 for non-acetylated glycerides. The otherconditions used were nebulizer gas flow of 3 L/min, desolvation linetemperature of 250 °C, heating block temperature of 400 °C, anddrying gas flow of 15 L/min. MRM dwell times and collision energieswere all set at 20 ms and −35 V.

Chiral Ester Derivatization. The method used was thatdescribed by Freire et al.17 A 10 mg amount of the free fatty acidmethyl ester mixture was dissolved in approximately 0.8 mL ofCH2Cl2 (dry) in a 5 mL reaction vial. The solution was stirred atroom temperature (RT), and 3 equiv (22.3 mg) of R-methoxyphenyl-acetic acid, 2 equiv (10 μL) of N-(3-(dimethylamino)propyl)-N′-ethylcarbodiimide, and a catalytic amount of N,N-dimethylaminopyr-idine were added. The reaction was worked up by washing thereaction solution with water, 1 M HCl, NaHCO3, and water. Theorganic layer was dried under vacuum and purified using a small silicagel column to obtain the diester product (approximately 8 mg). TheS-diester was obtained in a similar fashion. Samples for 1H NMRanalysis were dissolved in CDCl3 and run at 500 MHz.

Determination of Plant Sources of Acylglycerols. Samples ofpoplar buds, twigs, and leaf material were collected from the curatedPlant and Food Research national poplar collection at Aokautere, nearPalmerston North, New Zealand, from September 2014 to February2015. Most samples were from coppiced trees, as these were easilyaccessed. Leaf and twig samples were weighed and extracted bydipping in absolute EtOH at RT for approximately 2 min. Thesamples were extracted as is without maceration or slicing. Theextracts were dried and weighed, and samples of each prepared foranalysis by making up solutions in EtOH at 5 mg/mL. Samples of leafand flower material were collected from a Paulownia tomentosa tree inLower Hutt, NZ (Tirohanga Rd., GPS −41.201586, 174.90840). Theplant material was extracted by dipping the plant material into ethylacetate to extract the waxy outer layers only.

Antiproliferative (Cell Viability) Assays: DLD-1, KYSE-30,and NCI-N87 Gastrointestinal Cancer Assays. Human gastro-intestinal cancer cells DLD-1 (colorectal adenocarcinoma cells[ATCC CCl-221, obtained from ATCC]); NCI-N87 (gastriccarcinoma cells [ATCC CRL-5822, obtained from ATCC]), KYSE-30 (esophageal squamous cell carcinoma [ECACC, obtained fromSigma-Aldrich]) were revived from cryostorage and cultured in thepresence of the test and reference samples. The culture conditions forthe cells were those described by the supplier of the cells, and theassay methodology was based on previously reported procedures.4

Briefly, working solutions of the test compounds and positive controlswere prepared by dissolving the test fractions in 15% EtOH/Hank’sbalanced saline solution to a concentration of 2 mg/mL solids. Thefinal concentration of each sample in the test well was 110, 55, or 22μM. The 5-fluorouracil positive control test well concentrations were450, 150, and 50 nM. After incubation an MTT cell viability assay wasperformed on the cultures to determine the effect of the samples onthe inhibition of cell proliferation. Results were expressed as thepercentage proliferation of cells cultured in the presence of the samplein comparison to the cells-only control. Six replicates were used forboth test and control samples.

Journal of Natural Products Article

DOI: 10.1021/acs.jnatprod.8b00562J. Nat. Prod. XXXX, XXX, XXX−XXX

H

■ ASSOCIATED CONTENT*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/acs.jnat-prod.8b00562.

NMR data for compounds 1, 2, and peracetate, 8 (PDF)

■ AUTHOR INFORMATIONCorresponding Author*E-mail: stephen.bloor@callaghaninnovation.govt.nz.ORCIDStephen Bloor: 0000-0002-1936-5857NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSMr. T. Jones (Plant and Food Research, NZ) is thanked foraccess to poplar collection, colleagues Dr. M. Vyssotski and Dr.K. Lagutin for advice on lipid analysis and absoluteconfiguration, and Manuka Health New Zealand Ltd forproviding the samples of “Bio30” propolis and ongoing supportin propolis research. This work was also supported by fundingfrom the New Zealand Government through CallaghanInnovation.

■ REFERENCES(1) Kumazawa, S.; Hamasaka, T.; Nakayama, T. Food Chem. 2004,84, 329−339.(2) Toreti, V. C.; Sato, H. H.; Pastore, G. M.; Park, Y. K. Evidence-based Comp. Alt. Med. 2013, 2013697390.(3) De Groot, A. C. Dermatitis 2013, 24, 263−282.(4) Catchpole, O.; Mitchell, K.; Bloor, S.; Davis, P.; Suddes, A.Fitoterapia 2015, 106 (3257), 167−174.(5) Catchpole, O.; Mitchell, K.; Bloor, S.; Davis, P.; Suddes, A. J.Funct. Foods 2018, 41, 1−8.(6) Banskota, A. H.; Nagaoka, T.; Sumioka, L. Y.; Tezuka, Y.; Awale,S.; Midorikawa, K.; Matsushige, K.; Kadota, S. J. Ethnopharmacol.2002, 80, 67−73.(7) Shi, H.; Yang, H.; Zhang, X.; Sheng, Y.; Huang, H.; Yu, L. J.Agric. Food Chem. 2012, 60, 10041−10047.(8) Reis, M. G.; De Faria, A. D.; Do Amaral, M. D. C. E.; Marsaioli,A. J. Tetrahedron Lett. 2003, 44 (46), 8519−8523.(9) Asai, T.; Hara, N.; Kobayashi, S.; Kohshima, S.; Fujimoto, Y.Helv. Chim. Acta 2009, 92 (8), 1473−1494.(10) Reis, M. G.; De Faria, A. D.; Dos Santos, I. A.; Amaral, M. D. C.E.; Marsaioli, A. J. J. Chem. Ecol. 2007, 33 (7), 1421−1429.(11) Ohkawa, A.; Sakai, T.; Ohyama, K.; Fujimoto, Y. Chem.Biodiversity 2012, 9 (8), 1611−1617.(12) Asai, T.; Hara, N.; Fujimoto, Y. Phytochemistry 2010, 71 (8−9),877−894.(13) Okawa, A.; Ohyama, K.; Fujimoto, Y. Nat. Prod. Res. 2013, 27(15), 1372−1377.(14) Curtis, J. D.; Lersten, N. R. Am. J. Bot. 1974, 61 (8), 835−845.(15) Kato, A.; Ando, K.; Suzuki, S.; Tamura, G.; Arima, K. J.Antibiot. 1969, 22 (2), 83−84.(16) Philippoussis, F.; Arguin, C.; Mateo, V.; Steff, A.-M.; Hug, P.Blood 2003, 101, 292−294.(17) Freire, F.; Seco, J. M.; Quinoa, E.; Riguera, R. Chem. - Eur. J.2005, 11 (19), 5509−22.

Journal of Natural Products Article

DOI: 10.1021/acs.jnatprod.8b00562J. Nat. Prod. XXXX, XXX, XXX−XXX

I