Chlorogenic acids and the acyl-quinic acids: discovery, biosynthesis, bioavailability and bioactivity †

This review is focussed upon the acyl-quinic acids, the most studied group within the ca. 400 chlorogenic acids so far reported. The acyl-quinic acids, the ﬁ rst of which was characterised in 1846, are a diverse group of plant-derived compounds produced principally through esteri ﬁ cation of an hydroxycinnamic acid and 1 L -( (cid:1) )-quinic acid. Topics addressed in this review include the confusing nomenclature, quanti ﬁ cation and characterisation by NMR and MS, biosynthesis and role in planta , and the occurrence of acyl-quinic acids in co ﬀ ee, their transformation during roasting and delivery to the beverage. Co ﬀ ee is the major human dietary source world-wide of acyl-quinic acids and consideration is given to their absorption and metabolism in the upper gastrointestinal tract, and the colon where the microbiota play a key role in the formation of catabolites. Evidence on the potential of the in vivo metabolites and catabolites of acyl-quinic acids to promote the consumer's health is evaluated.

A wider denition of CGAs includes compounds formed with various quinic acid epimers, quinic acid methyl ethers, alkyl quinates, deoxyquinic acid, 2-hydroxyquinic acid and shikimic acid and its epimers, plus the analogous compounds that are esteried with a hydroxybenzoic acid (6), hydroxyphenylacetic acid (7) or 3-(4 0 -hydroxyphenyl)propionic acid (8).Aliphatic acid substituents may also be present, occasionally in the absence of an aromatic residue.A fuller account of these less common CGAs is available. 1he origin of 'chlorogenic acid' is probably related to the use of the term 'chlorogen acid' by Payen 2,3 in 1846 (1846a, 1846b) who reported the isolation of a crystalline potassium caffeine chlorogenate that formed up to 5% of green coffee (Coffea arabica) beans.He proposed an empirical formula of C 14 H 8 O 7 which is now known to be C 16 H 18 O 9 , and described its conversion to a green pigment on alkaline oxidation.In retrospect, one can deduce that there were a few earlier studies with the same or a very similar fraction prepared from green coffee beans.The earliest of these appears to be a paper by Robiquet and Boutron in 1837, 4 in which they reported the isolation of an acidic substance that turned green when treated with ferric chloride.Rochleder 5 isolated an acidic substance from green coffee beans in 1844 that associated with caffeine and which precipitated as a lead salt.In 1846 Rochleder further reported that this fraction Mike Clifford is the Emeritus Professor of Food Safety at the University of Surrey and has investigated the analysis and characterisation, absorption and metabolism of phenols, polyphenols and tannins.He has published several books, including monographs on tea and coffee, over 250 research papers of which over 70 relate to chlorogenic acids.These latter include a new unambiguous nomenclature for the chlorogenic acids, plus several technical reports, citing in excess of 1500 references thereon, that are freely available on Researchgate.In 2009 he was awarded the Mars Prize for his research on "Polyphenol Analysis" at the Fourth International Conference on Polyphenols and Health, in Harrogate, UK.
was yellow in ammoniacal solution and became green on exposure to oxygen, and suggested C 16 H 9 O 8 for the free acid. 6It was not until 1907 that pure white crystals were obtained which had a melting point of 206-207 C, and from which quinic acid (1)  and caffeic acid (3) were released by alkaline hydrolysis.An empirical formula of C 32 H 38 O 19 was proposed in 1908 and to reconcile this with the evidence from hydrolysis Gorter 7 proposed that quinic acid combined with caffeic acid to form hemichlorogenic acid, two molecules of which condensed losing one molecule of water to produce chlorogenic acid.This would correspond to an empirical formula of C 16 H 20 O 10 .Freudenberg 8 reported in 1920 that chlorogenic acid was a substrate for the enzyme tannase and that hydrolysis released equimolar amounts of quinic acid and caffeic acid.In 1932 Fischer and Dangschat 9 proposed that this substance was 3-O-caffeoylquinic acid (3-CQA).
Acyl-quinic acids are widespread dietary components being found, for instance, in coffee, cherries (Prunus avium), blueberries (Vaccinium spp.), aubergine (Solanum melongena), apples (Malus pumila) oregano (Origanum vulgare), spearmint (Mentha spicata), chicory (Cichorium intybus) and sunower (Helianthus annus) seeds 28,[35][36][37][38][39] with high levels in globe artichoke. 40,41The herbal tea maté, made from infusion of dry leaves of Ilex paraguariensis, contains substantial amounts of CQAs and diCQAs (Table 1). 42Coffee beverage, rather than fruits and vegetables is probably the main dietary source of CQAs for many people with intakes in excess of 1 g per day being readily attained. 43he hydroxycinnamic acid moiety of acyl-quinic acids is predominantly in the trans form.Some cis isomers are known, with cis-5-O-p-coumarylquinic acid (15) being detected in ower buds of herbal aster (Aster ageratoides Turcz). 47and a wide range of cis-isomers have now been detected in other species. 34It has been suggested that the cis derivatives originate from plant tissues where the trans isomer has been exposed to relatively strong UV-irradiation which induces geometric isomerisation. 48here is good experimental evidence to support this view. 49owever, direct synthesis cannot be ruled out, and there is some evidence from research with cell culture to suggest that UV-irradiation is not essential. 50Nomenclature Acyl-quinic acids exhibit congurational isomerism, conformational isomerism and regio-isomerism, plus geometric isomerism for those containing a cinnamic acid residue.Consequently, it is not easy to describe unambiguously the structures of acyl-quinic acids that may appear almost identical when drawn in 2D or projected in 3D.Various systems for cyclitols were evaluated by IUPAC who recommended that the most common natural form of quinic acid be described as 1L-1(OH),3,4/5-tetrahydroxycyclohexanecarboxylic acid, with the trivial names (À)-quinic acid or L-quinic acid. 51In the IUPAC system, Fischer and Dangschatt's 3-CQA 9 became 5-CQA (9).
Unfortunately, papers are still published in which non-IUPAC numbering is used, the numbering system used is not stated, or worse, the numbering is IUPAC but the structure shown is not, or vice versa.Kremr et al. 52 commented that some published structures make no attempt to depict the spatial arrangement of the substituents, and Clifford recorded that some authors discuss previously published data unaware that different numbering systems have been used, for example 3-CQA (non-IUPAC) and 3-CQA (IUPAC) are treated as the same compound. 34,53Note that Wikipedia and many other online sources, plus many catalogues listing acyl-quinic acid preparations, use non-IUPAC nomenclature.Many examples of misleading or incorrect descriptions have been compiled. 1,34,53or the avoidance of doubt, the IUPAC and non-IUPAC structures are presented in Fig. 1.From this point onwards the IUPAC numbering system will be used in this review.Trivial names further complicate the literature but a glossary is available, 1,54 and this is presented as Table S1 in the ESI.† To this must be added chrysanthemorimic acids, a series of recently discovered diCQAs, in which one caffeic acid residue has undergone a [5 + 2] cycloaddition of D-glucose. 55he IUPAC system, however, has limitations when applied to acyl-quinic acids and Abrankó and Clifford 54 have proposed a combination of IUPAC cyclitol numbering 51 the Cahn-Ingold-Prelog (CIP) sequence rules 56 plus the use of a to dene a hydroxyl trans to the quinic acid carboxyl (and b to dene a hydroxyl cis to the quinic acid carboxyl) to describe the orientation of a substituent on a carbon atom which is not a centre of chirality, as favoured by Eliel and Ramirez. 57This approach can accommodate all eight quinic acid stereo-isomers, with, for example, IUPAC (À)quinic acid described as 1L-(À)-quinic acid 3R,5R-(1a,3a,4a,5b). 54 ESI Table S2 † provides a comprehensive set of structures for the various quinic acids, which in addition to using different styles of presentation (Fischer-Tollens, 2D, alternative chair conformations), also considers the perspective from which a structure is viewed.

Characterisation and quantification of acyl-quinic acids
This section of the review draws on three open-access documents which contain extensive tabulations and discussion of the topics presented here. 1,53,56

Extraction
Extraction of acyl-quinic acids from plant material generally employs aqueous alcohol, usually 70% MeOH.Artefacts may  arise from acyl migration, alkylation, hydrolysis, water addition across the cinnamic acid double bond and/or trans-cis isomerisation. 59Acyl migration is favoured by higher water content in the solvent and sample, and when extracting undried plant material 100% MeOH is advisable for the rst stage, especially if 1,5-diCQA might be present, as it rapidly converts to 1,3-diCQA.
In contrast, controlled acyl migration and partial hydrolysis can usefully generate regio-isomers that are not otherwise conveniently available (e.g.1,4-diCQA, 1-CQA), 26,60,61 as can UVirradiation for preparing cis-isomers. 48Any novel acyl-quinic acid should be examined by controlled acyl migration in order to access associated regio-isomers.

1 H-NMR
4][65] The denitive studies of Pauli et al., 63,64 and the data tabulated by Clifford, 53 show that if best practice is followed, mono-and di-acyl-quinic acids can be successfully identiedd 4 -MeOH is the preferred solvent, to which a small amount of D 2 O or d 6 -DMSO (not exceeding 10%) can be added, followed by analysis at not less than 500 MHz.Identication of the acyl residue(s) is straightforward, assignment of the quinic acid conguration and the position(s) of acylation less so, because signals for H2 and H6 methylenes and H3 and H5 methines oen overlap, especially in tri-or tetra-acyl-quinic acids, and/or when an aliphatic substituent is present.Temperature, analyte concentration, and especially solvent inuence the proton chemical shis, the order in which the shis occur, and the conformation of the quinic acid moiety, and hence the coupling constants for the relevant protons. 63,64,66][69][70][71][72][73] These novel compound(s) are rarely observed in company with a full set of the commoner 1L-(À)-quinic acid derivatives.This does not per se refute the claim, but might suggest that one of the commoner acyl-quinic acids has been wrongly identied.Reports by Wang et al. 74,75 of an acyl-iso-quinic acid (¼ 4,5-diCepi-QA IUPAC) are convincing because the quinic acid moiety released by saponication did not co-chromatograph with 1L-(À)-quinic acid, and subsequent 500 or 600 MHz NMR in d 4 -MeOH is distinctive. 75,76The biosynthesis of 1L-(À)-epi-quinic acid requires only that D-threose-4-phosphate replace Derythrose-4-phosphate in the pathway to 1L-(À)-quinic acid. 58here are LC-MS data for four incompletely characterised CQA, 32,77 which plausibly are derivatives of a quinic acid isomer.[81]

Chromatography
A reverse-phase LC column packing and shallow linear solvent gradient providing structure-diagnostic relative capacity factors or relative retention times (RRTs) is the best strategy for characterising acyl-quinic acids.Once characterised, a revised gradient may speed up the separation for routine use.The behaviour of acyl-quinic acids is determined by (i) orientation of the free quinic acid hydroxyls, and (ii) number and hydrophobicity of the acyl moieties.For a given acyl moiety, 1-acyl and 3acyl regio-isomers (two free equatorial hydroxyls) are wellresolved from the 4-acyl and 5-acyl regio-isomers (two free axial hydroxyls).1-Acyl regio-isomers almost always elute rst, but 4-acyl and 5-acyl regio-isomers vary with column packing.Similarly 1,3-diacyl regio-isomers (two free equatorial hydroxyls) elute rst and 4,5-diacyl regio-isomers elute last (two free axial hydroxyls).The other four diacyl regio-isomers elute close together just before the 4,5-diacyl isomer, but their sequence varies with column packing. 31,82For diacyl-quinic acids with different substituents, the isomer with the more hydrophilic substituent more equatorially positioned, elutes rst. 58elative to free quinic acid, acylation delays elution.Cinnamic and benzoic acids with more ring hydroxyls elute faster, and for a given number of ring substituents, methylation of the hydroxyl slows elution, i.e. for a given regio-isomer the sequence is CQA, p-coumaroylquinic acid (pCoQA), feruloylquinic acid (FQA), and (3 0 ,4 0 -dimethoxycinnamoyl)quinic acid (DQA), etc. cis-3-Cinnamoyl and cis-4-cinnamoyl elute before their transcounterparts but cis-5-cinnamoyl elute later.The methyl esters of the mono-acyl-quinic acid regio-isomers, but not the diacylquinic acid regio-isomers, elute in the reverse order compared with the free acids. 83Glycosidation of the acyl moiety markedly speeds the elution.Although the presence of an aliphatic dicarboxylic acid, e.g.succinic acid, may speed elution the behaviour is less predictable because of internal hydrogen bonding in some regio-isomers. 84ith that exception, these RRTs are sufficiently consistent to be used as a structure-diagnostic tool, easily locating cis-5-CQA which is oen reported as trans-1-CQA, [85][86][87][88][89] and early-eluting CQA-glycosides which are oen reported as diCQA. 90,91RRT values should always be considered when assigning the structure of a novel acyl-quinic acid.

LC-MS
Mass spectroscopy was long considered blind to isomers but the development of LC-ion trap-MS changed perceptions, 92 especially with regard to acyl-quinic acids, 58 for which negative ion MS is preferable.The methods and hierarchical keys developed by Clifford and Kuhnert using mild fragmentation conditions produce distinctive patterns at MS 2 , MS 3 and MS 4 , allowing regioisomers to be assigned condently.Critically, these methods have been applied successfully to previously unknown acyl-quinic acids.The geometric isomers of cinnamoyl-quinic acids fragment identically, 48 but the cis-isomers can be distinguished by UVirradiation, 62 sodium-adduct MS, 93 and ion mobility-MS. 94Using the hierarchical keys it is possible to identify at regio-isomer level in excess of 20 acyl-quinic acids in a single run subject to adequate analyte concentration. 31,77,78,83, It isfalse economy to characterise CGAs and avonoids in a single run, because avonoids require harsher MS conditions which mask the structure-diagnostic fragmentations of the acyl-quinic acids, and invalidate the hierarchical keys.
The hydrogen-bonding networks responsible for these distinctive fragmentations have been deduced. 58For some regio-isomers the charge may be on a phenolic hydroxyl rather than the quinic acid carboxyl. 116Accurate mass may be helpful, but is not essential to distinguish diCQA from CQA-glycosides because the glycosides yield characteristic MS 2 ions (m/z 341 and 323) 84 the latter as the base peak indicating a 3 0 -glycoside. 103,107riCQA, diCQA-glycosides and CQA-biosides, as well as acylisocitric acids and acyl-quinic acids, can be distinguished similarly, but only fragmentation can distinguish between isobaric methyl-CQ, FQA and isoFQA, and acyl-quinides and acyl-shikimic acids. 58Methyl-CQ and methyl-diCQ fragment very differently from CQA and diCQA, 83 which coupled with the 'reversed' order of elution for the methyl CQ, necessitates care in regio-isomer assignment.][119][120] Full assignment of tri-acyl-and tetra-acyl-quinic acids requires MS 4 and perhaps MS 5 spectra, particularly where there are two or more different acyl moieties (e.g.dicaffeoylferuloylquinic acids or caffeoyl-sinapoyl-feruloylquinic acids).MS 8 is required to characterise depsidic GQA bearing up to eight galloyl residues. 98Targeted rather than automated fragmentation may be essential with more complex structures such as 4-methoxyoxalyl-1,5-dicaffeoylquinic acid and 4methoxyoxalyl-3,5-dicaffeoylquinic acid, 112 or when a 'dehydrated' base peak (e.g.m/z 349 rather than m/z 367) occurs, more frequent with increasing methylation of the acyl moieties. 58,102hese structure-diagnostic protocols were developed using a ThermoFinnigan LCQ deca+ or a Bruker Daltronics HCT Ultra ion trap mass spectrometer, but should be easily transferred to similar instruments provided that the ionisation potential and fragmentation energy are appropriate.Even on the instruments originally used changes to the instrument parameters, such as increasing the ion spray voltage from 3.5 kV to 4.5 kV, generates many additional fragments and invalidates the hierarchical keys as published. 121,122ome investigators have reported that with QTOF-MS all CQA regio-isomers fragment identically. 123,124However, Madala and co-workers have demonstrated that with careful control of the collision energy it is possible to obtain the same MS 2 and MS 3 fragmentation data as an ion trap. 125,126When adapting the ion trap-MS hierarchical key methods to a non-ion trap-MS, a surrogate standard, such as a green coffee extract, 127 should be analysed and operating parameters adjusted until identical fragmentations are achieved. 58Whatever equipment is used there is scan-to-scan variation in fragment intensity, being most prominent for weak signals, i.e. low analyte concentrations, and higher order spectra, and for a reliable assignment at least 20 scans should be taken.If necessary use a more concentrated extract or larger injection.
Only limited success has been achieved with triple quadrupole instruments because of the much increased fragmentation energy employed.Matsui et al. successfully ngerprinted several CQA, FQA and diCQA regio-isomers using two different collision energies in positive ion mode (15 and 30 eV) and three different collision energies in negative ion mode (20, 40 and 60 eV), 128 but this is too cumbersome for routine use.Similar limitations are apparent in the methods reported by Lin and Harnly, 129,130 and Willems et al. 131 Ion-trap-MS has its limitations, most apparent where there is an aliphatic dicarboxylic acid substituent (e.g.succinic acid) because the fragmentations can be driven also by the distal carboxyl of that substituent in addition to the quinic acid carboxyl. 84,95,111,112Reproducible ngerprints are obtained, but even with targeted fragmentations, full regio-isomeric assignment is not always possible. 58Scopoletin ( 16) and the CQAs are indistinguishable by accurate mass, and there are currently insufficient fragmentation data for scopoletin to judge whether it could easily be distinguished. 58

Calibrants
Quantication requires one or more pure calibrants.The molar absorbance of any set of acyl-quinic acid regio-isomers differs comparatively little, e.g.CQA AE 4% of mean (18 500), pCoQA AE 2.5% of mean (20 400), FQA AE 2.5% of mean (19 000) and diCQA AE 6% of mean (33 300).A good quality 5-CQA (for which the molar absorbance should be quoted) can be used for all of the foregoing with arithmetic corrections for the subgroups if required, although this is only signicant for the diCQA and other acyl-quinic acids with two or more aromatic substituents.
In contrast to molar absorbance values obtained aer meticulous purication with corrections for ash content, water content, etc., published calibration curve response factors for acyl-quinic acids can vary by some 300%, indicating that these commercial preparations are far from pure, containing non-UVabsorbing salts, solvents, etc. 58 Such preparations are unsuitable for quantitative analysis.Some commercial standards are incorrectly described at regio-isomer level, 1 and use of surrogate standards such as a green coffee extract, an alcoholic cider, or artichoke extract, characterised by LC-MS fragmentation and relative retention times should be considered. 130

Biosynthesis of acyl-quinic acids
The initial steps in the biosynthesis of CQAs are via the phenylpropanoid pathway and the enzymes catalysing the conversions that produce 5-CQA (9) are well established but there is less clarity about the later stages of the pathway leading from 5-CQA to other acyl-quinic acids.
The conversion of phenylalanine to p-coumaroyl-CoA, with cinnamic acid and p-coumaric acid acting as intermediates, is catalysed sequentially by phenylalanine ammonia lyase (PAL), cinnamate 4 0 -hydroxylase (C4H) and 4-cinnamoyl-CoA ligase (4CL) (Fig. 2).PAL is encoded by a multi-gene family and typically exists as multiple isoforms in plants, ranging from a few members, such as three in Arabica coffee (Coffea canephora), 132 four in Arabidopsis thaliana, and nine in rice (Oryza sativa) 133 to more than a dozen copies in tomato (Lycoperscion esculentum). 134Individual PAL isoforms might be associated with specic metabolic activity during plant growth and development, and in plant-environment interactions. 135In western balsam poplar (Populus trichocarpa), there are ve genes (PtrPAL1-5) that are differentially expressed.PtrPAL2, 4 and 5 are mainly expressed in xylem and root tips, predominantly responsible for the production of lignin while PtrPAL1 and 3 are responsible principally for the production of condensed tannins. 136However, in plants such as tomato, most of the isoforms of PAL present are functionally redundant and underutilized.There are 20 PAL-encoding genes in tomato but only one is expressed strongly throughout the plant with others being silenced. 134Since PAL is the entry point enzyme, its control and regulation is crucial to mediate carbon ux from primary metabolism into biosynthesis of acyl-quinic acids.The regulation of PAL have been shown to occur at multiple levels, and has been reviewed in detail by Zhang and Liu. 135n Arabidopsis thalania the C4H catalysing the 4 0 -hydroxylation of cinnamic acid to p-coumaric acid (Fig. 2) is encoded by a single gene, mutation of which causes a dwarf phenotype, male sterility and swellings at branch junctions, and also results in the accumulation of cinnamoylmalate (17), which is not found in wild-type plants. 137However, when the gene encoding C4H in sweet sagewort (Artemisia annua) was silenced through RNAi technology, it caused accumulation of cinnamic acid accompanied with signicant reductions in p-coumaric acid (1), total phenolics and anthocyanins, highlighting its role as a major ux controlling enzyme in the phenylpropanoid pathway (Kumar et al. 2016). 138The next step in the pathway is the conversion of p-coumaric acid to p-coumaroyl-CoA catalysed by a small gene family of 4CL ligases (Fig. 2).p-Coumaroyl-CoA is a key branch point in phenylpropanoid biosynthesis acting as the immediate precursor of avonoids and stilbenes, as well as being channeled into the production of methoxy guaiacyl-and syringyl-monolignols. 139 Four isoforms of 4CL have been iden-tied in Morus notabilis (mulberry) 140 and Arabidopsis. 141The subcellular localization of specic isoforms of 4CL plays a central role in redirecting the biosynthetic pathway either towards the p-coumaroyl-CoA or towards other branch points in the phenylpropanoid pathway. 141e conversion of p-coumaroyl-CoA to 5-CQA involves the enzymes hydroxycinnamoyl CoA:quinate hydroxycinnamoyl transferase (HQT) and a cytochrome P450 oxidase p-coumaroyl-3 0 -hydroxylase (C3H) (Fig. 2).The HQT-catalysed metabolism of p-coumaroyl-CoA produces 5-O-p-coumaroylquinic acid (5-pCoQA) which then undergoes C3H-mediated 3 0 -hydroxylation to yield 5-CQA.Alternatively, C3H-mediated hydroxylation of pcoumaroyl-CoA produces caffeoyl-CoA which is then converted to 5-CQA via the action of HQT (Fig. 2).Two HQT-encoding genes have been isolated from globe artichoke and when expressed in Escherichia coli each produced a recombinant protein, HQT1 and HQT2, with acyltransferase activity.Kinetic data and in silico homology modeling and docking analyses suggested that the two enzymes may be involved in different steps in the 5-CQA biosynthesis pathway with HQT1 catalysing the esterication of caffeoyl-CoA with quinic acid to produce 5-CQA, and HQT2 being involved in the conversion of pcoumaroyl-CoA to 5-pCoQA (Fig. 2). 142This esterication is reversible.However, as a result of the genome-wide identication of BAHD acyltransferases in globe artichoke, a third isoform of HQT enzyme was detected. 143When HQT1 was subjected to virus-induced gene silencing, a marked reduction of both CQAs and diCQAs occurred.In contrast, transient overexpression of all three isoforms in leaves of Nicotiana benthamiana had the opposite effect, 63 highlighting the role of all three isoforms in the production of acyl-quinic acids.
5-CQA ( 9) is the dominant regio-isomer in most plants.However, most species also contain 4-CQA (10) and 3-CQA ( 11), while 1-CQA ( 12) is present in a limited number of plants, and 3-CQA is the dominant isomer in some species, most notably in plums (Prunus sp.) belonging to the Rosaceae family, 27 and some Brassicaceae. 144From an enzymic perspective information is available only on the biosynthesis of 5-CQA ( 9) and the perceived wisdom is that other CQAs are derived from 5-CQA although there are no data on the seemingly specic isomerases involved in such conversions.Little is also known about the biosynthesis of di-and triCQAs or acyl-quinic acids containing substituents other than caffeic acid (3), although the in vitro synthesis of diCQAs from 5-CQA and CoA, mediated by a recombinant HCT enzyme cloned from coffee, has been reported. 145In tomato, the enzyme HQT converted 5-CQA to diCQAs in vitro.It was proposed that the HQT enzyme has a dual role in vivo catalysing different reactions in two subcellular compartments: in the cytoplasm acting as a quinate transferase while in the vacuole favouring chlorogenate:chlorogenate transferase (acyl-quinate:acyl-quinate transferase) activity producing diCQA. 146esearch with globe artichoke, switchgrass (Panicum virgatum) and chicory has established a further route to acyl-quinic acids that contains a shikimic acid shunt 142,147 It involves an hydroxycinnamoyl-CoA:shikimate hydroxycinnamoyl transferase (HCT1; EC 2.3.1.133),or possibly two such enzymes in switchgrass 148 that catalyse the conversion of p-coumaroyl-CoA to 5-O-p-coumaroylshikimic acid, which is further converted to 5-O-caffeoylshikimic acid (dactylifric acid) by p-coumaroylshikimate-3 0 -hydroxylase (C3H) (Fig. 3).The caffeoylshikimate so generated could be converted to caffeoyl-CoA by an HCT acting in the reverse direction.In vivo functional analysis of the gene encoding HCT from chicory when transiently overexpressed in tobacco conrmed the involvement and the role of HCT in this step. 147Alternatively a caffeoylshikimate esterase could release caffeic acid which is converted to caffeoyl-CoA by a ligase.Conversion of caffeoyl-CoA to 5-CQA in globe artichoke and other dicots probably involves an HQT. 143,147The gene encoding this enzyme is not present in switchgrass and phylogenetic analysis suggests that in monocots the conversion may be catalysed by enzymes more closely related to HCT than HQT (Fig. 3). 148s far as the production of FQAs, such as 5-FQA is concerned, cDNA encoding S-adenosyl-L-methionine:caffeoyl-CoA-3-Omethyltransferase (CCoAOMT) an enzyme which catalyses the conversion of caffeoyl-CoA to feruloyl-CoA (Fig. 2) has been cloned from oats (Avena sativa) 149 and other plants, including Coffea species.150 Although the main role of the CCoAOMT family was initially ascribed to lignin biosynthesis, 151 other studies have indicated its multifunctional role in catalysing steps in other biosynthetic pathways including those producing the coumarin, scopoletin ( 16), 152 anthocyanins 153 and other avonoids.154 In the scopoletin biosynthetic pathway, a recombinant CCoAOMT1 protein possessed methylating activity in vitro targeting caffeoyl-CoA and converting it to feruloyl-CoA.152 A third pathway to 5-CQA from cinnamic acid has been proposed on the basis of data obtained with tubers of sweet potato (Ipomoea batatas Lam.) which converted [2-14 C]cinnamic acid to a radiolabeled intermediate that was metabolized to 5-CQA and further converted to 3,5-diCQA 155 both of which are major native acyl-quinic acids in sweet potato.156 The radiolabeled intermediate was identied as 1-O-cinnamoyl-glucose 157 and it was proposed that it was converted to 5-CQA via 1-O-pcoumaroyl-glucose and 1-O-caffeoyl-glucose as illustrated in Fig. 4. 157,158 Enzymes which catalyse the rst three steps in this postulated pathway, a UDP-glucose:cinnamate glucosyltransferase, 159 a p-coumaroyl-glucose hydroxylase 160 and a hydroxycinnamoyl glucose:quinate hydroxycinnamoyltransferase have been detected in sweet potato roots.158,161 In addition, a partially puried enzyme has been isolated from sweet potato which catalyses the single step conversion of 5-CQA to 3,5-diCQA.162 The production of cinnamoyl-glucose has also been reported in strawberry (Fragaria Â ananassa) where a cDNA encoding a UDP-glucose:cinnamate preferentially catalysed cinnamic acid in vitro.The cDNA transcript was found to accumulate during strawberry fruit ripening and this positively correlated with the in planta concentration of cinnamoyl, p-coumaroyl-, and caffeoyl-glucose.163 Similar results were obtained when the gene encoding this enzyme was overexpressed in transgenic Populus where it led to accumulation of hydroxycinnamate glucose esters which further increase under N-limiting conditions.164 It has long been known that exposure of trans-cinnamoylquinic acids to UV light readily produces the associated cis isomers, and such isomers have been found in tissues, such as the leaves of tobacco and coffee, that are exposed to comparatively strong UV-irradiation, suggesting that such exposure might explain their presence.48 3,5-DiCQA has been shown to undergo isomerisation upon UV exposure where the naturally occurring 3-trans, 5-trans-diCQA isomer (18) gives rise to the 3-cis, 5-trans-diCQA ( 19), 3-trans, 5-cis-diCQA (20), and 3-cis, 5-cis-diCQA ( 21) isomers, allowing them to have a wider synergistic biological activity when present together.165 Recent studies of 120 samples of Stevia rebaudiana leaves have established a strong positive correlation between sunshine hours prior to harvesting and the content of cis isomers in the harvested leaves, 166 effectively conrming this hypothesis.It has also been shown that treatment of cultured tobacco cells with several inducers of plant stress leads to an approximate doubling of cis 5-CQA (22) in the absence of UV irradiation.50 There are many other details yet to be elucidated and several intriguing questions remain to be answered regarding the factors that control the biosynthesis of acyl-quinic acids and the signicance of the observed variations.

Role of acyl-quinic acids in planta
With one or two notable exceptions, the in planta role(s) of acylquinic acids are not known with any certainty.It seems almost certain that their function in tissues containing ca. 100 g kg À1 is different to their function in tissues containing only mg kg À1 concentrations.Radiotracer studies have shown that 5-CQA ( 9) is incorporated into lignin in Xanthium 167 which suggests that 5-CQA may accumulate as a store of cinnamic acid for lignin biosynthesis.It has also been proposed that FQAs are important indirect plant defence agents acting as lignin precursors, the lignin per se making tissues more resistant to attack or invasion. 168g. 4 Proposed pathway for the biosynthesis of 5-O-caffeoylquinic acid in sweet potato.Enzymes: UDP-glucose:cinnamate glucosyltransferase (CGT), cinnamate-glucose 4 0 -hydroxylase (CG4H), p-coumaroyl-glucose 3 0 -hydroxylase (CG3H) and hydroxycinnamoyl glucose:quinate hydroxycinnamoyltransferase (HCGQT).
A high acyl-quinic acid content has been linked to the allelopathic potential and resistance to root disease of some sweet potato clones, 169 via an inhibition of fungal toxin production. 170An elevated 5-CQA content of transgenic tomato plants has been associated with enhanced tolerance to paraquat-induced oxidative stress and resistance to infection by the pathogen Pseudomonas syringae. 171Acyl-quinic acids acting as defence metabolites may have possible links to phytohormone-activated defense signaling networks as implied in Nicotiana tabacum cells treated with salicylic acid (23) and methyl-jasmonate (24). 172The defense signaling networks produce a range of acyl-quinic acids to combat pathogens and some are capable of inhibiting cell wall degrading enzymes, such as polygalacturonase and cutinase, produced by the pathogens. 173 tomato plants CQA has been associated with a protective ability against oxidative damage when exposed to salinity or a combination of salinity and heat. 174[185] 6 Acyl-quinic acids in coffee Much of the literature on acyl-quinic acids focuses on coffee and green beans which contain substantial amounts.In addition to the major CQAs, FQAs, diCQAs and p-CoQAs (Fig. 5) many quantitatively minor acyl-quinic acids have also been identied. 33,102,114,186,187Currently, green coffee, with 72 acylquinic acids, is second only to Lonicera japonica (Japanese honeysuckle) which contains a record 111 in the form of 35 trans-cinnamoyl-quinic acids, 17 cis isomers and 49 acyl-quinic acid glucosides. 188he principal component in green coffee beans is always 5-CQA (9) and is accompanied by numerous diacyl-quinic acids and some triacyl-quinic acids.All the cinnamic acid derivatives so far characterised in green coffee have the cinnamate moiety in the trans conguration.To date, no 1-acyl-quinic acids have been found in the green coffee bean.The triacyl-quinic acids have so far been found only in Robusta coffee beans. 33Robustas usually have a greater content of any isomer but there are exceptions, the best known being a tendency for Arabicas to contain more 5-p-CoQA (25)  than Robustas.From a chemotaxonomic standpoint it is interesting to note that acyl-quinic acids containing 3 0 ,4 0 -dimethoxycinnamic acid (26), 3 0 ,4 0 ,5 0 -trimethoxycinnamic acid ( 27), 3 0 -methoxy-4 0 ,5 0 -dihydroxycinnamic acid (5-hydroxyferulic acid) (28), and 3 0 ,5 0 -dihydroxy-4 0 -methoxycinnamic acid (29) are scarce in nature and may be unique to coffee.Cinnamoyl-amino acid conjugates also have been characterised, 26,187,189,190 and together with the acyl-quinic acids have been used as criteria in chemo-taxonomic studies 191,192 for distinguishing Robustas from Arabicas, 193 and to some extent the prole can be used to dene the geographic origin of green coffee beans, especially Robustas from Angola, 25,197 Ethiopia, 194 Uganda, Vietnam, Cameroun and Indonesia, and for distinguishing South American from African Arabicas. 193g. 5 The main acyl-quinic acids in coffee beans.
analyses, in comparison with the 72 derivatives detected in the green coffee bean. 30,31,33,102,114,115,186,187,190Acyl migration and hydrolysis during coffee roasting changes the relative proportions of each subgroup, being most obvious within the CQAs, FQAs and diCQAs, with 3-CQA being very susceptible to hydrolysis.For example, it has been reported that aer 5 min of roasting the content of 5-CQA ( 9) decreased substantially while the levels of 4-CQA (10) and 3-CQA ( 11) increased to twice their original values.The same behavior was observed for FQAs.It is also noted that partial hydrolysis of diCQAs to CQAs occurs in addition to isomerization. 22,195The triacyl-quinic acid derivatives of the green coffee bean have not been found in the roasted bean, and presumably are lost via partial hydrolysis.The clearest evidence for acyl migration during coffee roasting is the appearance of 1-acyl derivatives, absent from the green beans, 196 either as the free acids or as the lactones 1-O-caffeoylquinic-1,5lactone (1-CQL) (30) and 1-O-feruloylquinic-1,5-lactone (1-FQL) (31) 197 with similar transformations also occurring during brewing and instant coffee manufacture. 198,199antitatively, the main acyl-quinic acids in the roast bean and, instant coffee powder and beverage are CQAs, FQAs and diCQAs (Fig. 5). 23Despite the losses of acyl-quinic acids that occur during roasting, coffee beverage is still a rich, probably the richest, dietary source. 200However, because of the use of a diversity of roasting methods and infusion/manufacturing protocols, a 'cup of coffee' is an extremely variable commodity which is best illustrated by modern data for the acyl-quinic acids contents.Espresso coffees bought at 20 commercial outlets near the University of Glasgow in the UK delivered between 24 mg and 423 mg per serving (Table 4). 43A similar but more extensive study of coffees from 104 commercial outlets in Scotland, Italy and Spain observed that one cup delivered between 6 mg and 188 mg of acyl-quinic acids. 201Standardised brewing of eight commercial roast and ground coffees (20 g/900 mL) delivered between 27 and 95 mg of acyl-quinic acids per cup, whereas standardised preparation of ten commercial instant coffees available in the UK (1.8 g/200 mL) delivered between 37 and 121 mg. 202An individual preferring weak instant coffee would likely consume rather less than these gures suggest.It is well known that coffee brews consumed on the American west coast are appreciably weaker than on the American east coast.This variation is such that, conceivably, a person consuming 10 cups per day might actually receive less acyl-quinic acids than another consuming only one cup per day.While this is undoubtedly an extreme and rare event, a four to ve-fold variation can surely be expected.Under such circumstances data from epidemiological studies that link 'cups consumed' with an effect on health, whether advantageous or otherwise, can only be treated as a hypothesis until further investigations are performed under conditions that are as controlled as possible.

Bioavailability and metabolism of acyl-quinic acids in humans
Because of the dominance of coffee as the source of dietary acylquinic acids it has been used in most human studies on their absorption and metabolism, supported by in vitro investigations using cultured human cells.A study using cultured gastric epithelial cells established that CQAs, FQAs and CQLs cross the epithelium.The 3-acyl and 5-acyl regio-isomers probably cross by passive diffusion via the paracellular route, whereas facilitated transport might operate for 4-CQA and 4-FQA. 203DiCQAs crossed the membrane even more rapidly, a behaviour that could be attributable to their greater hydrophobicity, although in the case of 3,5-diCQA (18) there was evidence for a carrier-mediated efflux. 203These observations may help to explain the results of feeding studies, discussed below, where minor acyl-quinic acids in the coffee beverage had a higher peak plasma concentration (C max ) than the dominant regio-isomers.For example, Stalmach et al. 44 observed that 3-FQA (32) and 4-FQA (33) had C max values of 16 AE 2 nmol L À1 and 14 AE 2 nmol L À1 , respectively, while the beverage dominant 5-FQA (34) peaked at 6 AE 2 nmol L À1 .
In vitro studies have also shown that acyl-quinic acids differ in their susceptibility to intestinal chlorogenate esterase (acylquinate esterase), with 5-CQA (9) being hydrolysed more readily than 3-CQA (11), and 4-CQA (10) being particularly resistant to hydrolysis. 204Consistent with this enzymic hydrolysis occurring in the stomach or upper gastrointestinal (GI) tract, signicant amounts of 1L-(À)-quinic acid, in excess of the amount consumed in the free form, were found in the ileostomy effluent of volunteers who consumed cloudy apple juice, apple smoothie, or coffee. 205uch in vivo resistance to enzymic and chemical hydrolysis postabsorption would also predispose to greater plasma concentrations of 4-acyl-quinic acids.It follows that commodities with acylquinic acid proles substantially different from that of coffee beverage are likely to produce plasma proles of acyl-quinic acid metabolites that differ quite markedly from those produced when coffee is consumed, as recently reported for maté which contains proportionately more 3-CQA (11) and diCQAs. 206he presence of low nmol L À1 concentrations of CQAs in plasma and their low level excretion in urine aer oral intake of coffee, 44,[207][208][209] artichoke, 210 and 5-CQA (9) 211 suggest that the bioavailability of acyl-quinic acids per se is limited.In contrast, Monteiro et al. 212 reported the presence of unmetabolised CQAs in the circulatory system with a C max of 7.7 mmol L À1 , 2.3 h (T max ) aer acute ingestion of a coffee containing 3395 mmol of acyl-quinic acids.Despite the high C max of the CQAs, CQAs were not detected in urine collected 0-24 h aer coffee intake.In a subsequent study by the same group, in which volunteers consumed a green coffee extract containing a much lower 451 mmol of acyl-quinic acids, 5-CQA (9) and 4-CQA (10) were detected in urine from some, but not all, subjects. 213The CQAs were also detected in plasma although there were unusually marked variations in the plasma pharmacokinetic proles of the individual subjects with, in two instances, an exceedingly high C max of >20 mmol L À1 being recorded.The coffee used in this study contained 43.2 mmol of diCQAs which aer consumption were reported to have attained a C max of 6.6 mmol L À1 .Assuming an average plasma volume of 3 L per person, this corresponds to 45.8% of intake.Despite there being some evidence from in vitro studies that diCQAs are more rapidly absorbed than monoacyl-quinic acids. 203It is difficult to reconcile these studies with the results obtained by other investigators.Aer volunteers consumed maté delivering 210 mmol of diCQAs no diacyl-quinic acids were detect in plasma. 206talmach et al. 44,214 carried out coffee feeding studies with healthy humans and ileostomists who have had their colon removed surgically for medical reasons, typically Crohn's disease or ulcerated colitis.In these investigations, samples were analysed using HPLC-MS 2 without recourse to the use of the more typical glucuronidase/sulfatase treatments prior to analysis.It was observed that during passage through the body extensive metabolism of acyl-quinic acids occurs, with some compounds being absorbed in the stomach and/or small intestine and others in the colon.Studies with ileostomists have made a major contribution to our understanding of colonic (poly)phenol metabolism in humans with an intact colon.Viewed simply, a transformation associated with the colonic gut microora will not be seen in ileostomists.In reality, it is important to recognise some colonisation of the upper GIT can occur in ileostomists and, as a consequence, some limited microbial transformations may persist.An ileostomy involves major surgery and is likely to have some patho-physiological effects on other organs and tissues, and although the consequences have barely been studied, it would not be surprising if there were some alteration to (poly)phenol metabolism in, for example, the upper GIT, liver and kidney of ileostomists.

Studies involving volunteers with and without a functioning colon
Stalmach et al. 44 carried out a study in which healthy humans with an intact colon consumed a 200 mL serving of instant coffee, containing 412 mmol (146 mg) of acyl-quinic acids, with CQAs comprising 65% of the total (Table 5), aer which plasma and urine samples were collected over a 24 h period.A total of 12 hydroxycinnamate derivatives, of which four were unchanged acyl-quinic acids and eight were metabolites, two of which were unconjugated, were identied and quantied in plasma.Their pharmacokinetic parameters are summarized in Table 6 and pharmacokinetic proles are illustrated in Fig. 6.
A similar array of metabolites has been detected in plasma aer the consumption of coffee by Scherbl et al. 209 Postingestion C max values in the Stalmach study ranged from 2.2 nM for 5-CQA (9) to 385 nM for dihydroferulic acid (35) 44 with the duration for T max extending from 0.6 h (ferulic acid-4 0sulfate [36] and a 3-CQL-sulfate) to 5.2 h (dihydroferulic acid [35]).The compounds detected in highest concentrations in plasma were free and sulfated conjugates of dihydroferulic acid (35) and dihydrocaffeic acid (37) with C max values ranging from 41 to 385 nmol L À1 .The T max for these compounds was in a narrow range from 4.7 to 5.2 h, implying absorption in the large intestine.Much shorter T max values of 0.6 to 1.0 h, indicative of stomach and/or small intestine absorption, were obtained with 5-CQA (9) and three FQAs (32-34) which had not been subject to phase II metabolism, plus ferulic acid-4 0 -sulfate (36), caffeic acid-3 0 -sulfate (38), and two CQL-sulfates, and all of which had relatively low C max values (Fig. 6 and Table 6).
As noted by Stalmach et al. 44 most of the acyl-quinic acidderived compounds were rapidly removed from the circulatory  Table 6 Pharmacokinetic parameters of acyl-quinic acid derivatives and metabolites circulating in plasma of healthy volunteers, 0-24 h following the ingestion of 412 mmol of acyl-quinic acids and derivatives contained in a 200 mL serving of instant coffee (after Stalmach et al. 44 ) a  4).c Apparent T 1/2 estimated aer oral intake rather than intravenous dosing.

Chlorogenic acids and metabolites
system with apparent elimination half-life (T 1/2 ) values of 0.3 to 1.9 h (Table 6).The only compounds with an extended T 1/2 were dihydroferulic acid-4 0 -sulfate (39) (4.7 h), dihydrocaffeic acid-3 0sulfate (40) (3.1 h) and ferulic acid-4 0 -sulfate (36) which had an unusual biphasic plasma prole with dual T max values at 0.6 h and 4.3 h.It is of note that the free acid, dihydroferulic acid (35), as opposed to the more typical glucuronide and sulfate metabolites, was the principal component to accumulate in plasma which also contained dihydrocaffeic acid (38) in a lower concentration.
Fig. 6 Plasma pharmacokinetic profiles of circulating acyl-quinic acids and metabolites, following the ingestion of 200 mL of coffee by healthy human subjects (based on Stalmach et al. 44 ).
The true T 1/2 values can be determined only by intravenous dosing of the metabolite.Estimates based on elimination aer oral intake overestimate the true T max because the metabolite is still entering the plasma when the elimination is being estimated.A further complication arising with gut ora metabolites for which absorption, and hence elimination, cannot commence until several hours aer substrate ingestion, is the lack of sufficient data points in the declining period between 8 and 24 hours.The lack of a reliable value for the true T max effectively precludes modelling of multiple doses at say three hour intervals, i.e. the manner in which many people drink coffee.Nevertheless, with such a pattern of repeat consumption there is clearly a potential for a sizable accumulation of gut ora metabolites in the plasma because a subsequent dose of substrate will have entered the colon before the previous intake has been eliminated.
In a further study, ileostomists drank an instant coffee with a very similar 385 mmol (136 mg) acyl-quinic acid content and prole to that ingested by the healthy subjects. 214Plasma metabolites were not investigated, but data on the amounts of acylquinic acids and their metabolites in ileal uid collected over a 0-24 h period aer the ingestion of coffee are presented in Table 7.The highest recovery of unmetabolised acyl-quinic acids compared to their intake was FQAs (77%), followed by CQAs (59%) and p-CoQAs and diCQAs (46%).The recoveries of acyl-quinic acid metabolites were much lower, ranging from 3.6 to 8.8%, except for CQL metabolites which corresponds to 56% of CQL intake.
Of the 385 mmol of acyl-quinic acids ingested by the ileostomists, 274 mmol (71%) was recovered in the 0-24 h ileal uid as the parent compounds and metabolites (Table 7).This indicates that $30% of intake is absorbed in the stomach and/or small intestine, and that in subjects with a functioning colon $70% of the ingested acyl-quinic acids pass from the small to the large intestine where they will be subjected to the action of the colonic microora.These observations are in line with the ndings of Olthof et al. 211 who fed 2.8 mmol of 5-CQA ( 9) to humans with an ileostomy and recovered $70% of the CQA intake in ileal uid.The data from both studies, albeit at substantially different doses, imply that around one third of ingested 5-CQA is absorbed and enters the bloodstream from the small intestine.In vitro studies support this conclusion as 5-CQA is not extensively degraded when incubated with gastric juice, duodenal uid and ileostomy effluent 211,215 although as noted below interesterication can occur. 205In keeping with these ndings, when 3,5-, 3,4-, and 4,5-diCQAs (18, 41, 42) from Ilex kudingcha were incubated with arti-cial saliva, gastric and pancreatic uids, they were not degraded.When incubated with a human fecal slurry under anaerobic conditions, the diCQAs were hydrolysed to CQAs and caffeic acid, which was then further catabolised to dihydrocaffeic acid. 216imal and in vitro cell culture studies indicate that postabsorption acyl-quinic acids are subjected to the action of epithelial esterases in the stomach and small intestine. 204,217,218In incubations with cultured gastric epithelial cells, there is some release of caffeic acid (3), ferulic acid (4) and 3 0 ,4 0 -dimethoxycinnamic acid (26) from CQAs, FQAs and dimethoxycinnamoylquinic acids (DQAs).Caffeic acid is metabolised primarily to isoferulic acid (43)  and a lesser amount of ferulic acid but is not transported unchanged to the basal side.In contrast, ferulic, isoferulic and dimethoxycinnamic acids are transported, and must be the primary source of these substrates in ileostomists. 219,220e absorption of ferulic acid and subsequent conjugation in the liver 221 is consistent with the early plasma T max (0.6 h) for ferulic acid-4 0 -sulfate (36) observed in feeding studies 44 while the later secondary T max (4.3 h) (Fig. 6 and Table 6) plausibly reects the absorption and sulfation of ferulic acid released by gut microora-mediated hydrolysis of unabsorbed FQA.Escherichia coli, Bidobacterium lactis and Lactobacillus gasseri have the requisite cinnamoyl esterase activity. 222The plasma prole was not recorded for the ileostomists and, thus, it is not known whether that too was biphasic, but because ileostomists could not be absorbing ferulic acid released in the colon there would have to be another compensatory source to maintain ferulic acid-4 0 -sulfate excretion.The plasma prole (Fig. 6) and the excretion of substantial amounts of dihydroferulic acid (35)  by volunteers with an intact colon continues until at least 8 h aer coffee consumption, 44 indicating that a signicant part of this is derived from 77% of FQA intake that reaches the colon (Table 7).The necessary hydrolysis of FQAs and hydrogenation of ferulic acid to dihydroferulic acid could involve either microbial and/or human enzymes post absorption. 223 0 ,4 0 -Dimethoxycinnamic acid (26) and 3 0 ,4 0 -dimethoxycinnamoylquinic acid (44) are minor components in coffee.However, when a coffee containing $400 nmol of these compounds was ingested by volunteers free 3 0 ,4 0 -dimethoxycinnamic acid was detected in plasma with a 500 nmol L À1 C max and a T max of $0.5 h.219 The C max is higher than that of the metabolites derived from CQAs that occur in coffee in much higher quantities (see Fig. 6).3-(3 0 ,4 0 -Dimethoxyphenyl)propionic acid (45) appeared later, predominantly 8-12 h aer coffee intake with a C max of 97 nmol L À1 .The high C max relative to dose probably reects the comparatively high hydrophobicity of dimethoxycinnamic acid derivatives facilitating passive absorption.225 In contrast to the behaviour of ferulic acid and dimethoxycinnamic acid, the failure of cultured gastric cells to transport caffeic acid (3) 219 seemingly conicts with the $1 h plasma T max for caffeic acid-3 0 -sulfate (39) (Table 6).44 The most plausible explanation is that absorbed CQAs may be hydrolysed and the caffeic acid conjugated in hepatocytes.
The quantities of acyl-quinic acids and their metabolites excreted in urine by healthy subjects and ileostomists over a 24 h period aer ingestion of coffee are summarised in Table 8.It is apparent that absence of a colon had minimal impact on urinary excretion of CQL-sulfates and FQAs, as well as caffeic, ferulic and isoferulic acid sulfates.The ileostomists excreted in urine 30.8 mmol of acyl-quinic acid metabolites equivalent to 8.0% of the amount ingested.In contrast, the volunteers with an colon excreted a total of 120.2 mmol which corresponds to 29.2% of intake.This is almost certainly an under estimate of acyl-quinic acid bioavailability because in this study phenolic catabolites, such as C 6 -C 2 hydroxy-and methoxy-phenylacetic acids (46, 47), C 6 -C 1 benzoic acids (48)  and hippuric acids (49, 50) were not quantied.As well as these phenolics being catabolites of caffeic acid, 215,[224][225][226] there is growing evidence that they are also a feature of the catabolism of many avonoids including avonols, 224,227,228 anthocyanins, [229][230][231] avanones [232][233][234] and avan-3-ols. 224,235,236he data in Table 7 show that aer coffee consumption a total of 46.2 mmol of ferulic acid-based compounds (ferulic acid [3], FQAs [32-34], ferulic acid-4 0 -sulfate [36], FQA-O-glucuronides FQAsulfates) were present in the 0-24 h ileal uid.In healthy subjects these compounds would pass to the large intestine and the quantity of ferulic acid metabolites excreted in the urine of volunteers with a functioning colon (feruloylglycine [51], and dihydroferulic acid [35] and its 4 0 -sulfate [39] and 4 0 -O-glucuronide [52]) totalled 51.2 mmol (Table 8).This is not greatly in excess of the 46.2 mmol of ferulic acid-based compounds entering the large intestine (Table 8).In contrast to in vitro cell culture studies, 203 arguably, this suggests that in vivo the ferulic acid metabolites may be derived principally from the ingested FQAs rather than via 3 0methylation of caffeic acid derivatives, formed from CQAs, diCQAs and CQLs.However, the presence of isoferulic acid [43] and dihydroisoferulic acid [53] metabolites in urine signies that caffeic acid undergoes 4 0 -methylation.Excretion of these metabolites by subjects with and without a colon (Table 9) points to 4 0 -methylation of caffeic acid [3] producing isoferulic acid [43] occurring in the upper GI tract.The human gastric epithelium is capable of such a methylation, 203 while formation of dihydroisoferulic acid (54) takes place in the distal GI tract.On the basis of coffee feeding studies with healthy volunteers and ileostomists, Stalmach et al. 44,214 proposed a series of metabolic pathways.These have been extended to incorporate data obtained when coffee was incubated with fecal slurries and the breakdown of acyl-quinic acids monitored. 226The proposed pathways are illustrated in Fig. 7 and 8.In the preparation of these pathways the following points were taken into consideration.Some, but not complete hydrolysis of CQAs and CQLs can occur in the small intestine, catalysed by mammalian esterases Table 8 Urinary excretion of acyl-quinic acids and their conjugated metabolites in 0-24 h urine of healthy subjects (n ¼ 11) and ileostomists (n ¼ 5) following the ingestion of 200 mL of coffee (after Stalmach et al. releasing caffeic acid. 237Further, hydrolysis in the colon is probably due to the action of bacterial esterases.Small amounts of CQAs and substantial amounts of FQAs and CQLs are absorbed in the small intestine with the CQLs appearing in the circulatory system as sulfates (Fig. 6).The released caffeic acid is subjected to sulfation forming caffeic acid-4 0 -sulfate and smaller quantities of caffeic acid-3 0 -sulfate.Caffeic acid is also methylated producing isoferulic acid, but also possibly ferulic acid, both of which in turn form glucuronide and sulfate derivatives.FQAs are absorbed in the small intestine and are also hydrolysed to some degree with ferulic acid-4 0 -sulfate [36]  appearing in the bloodstream with an initial C max of 0.6 AE 0.1 h.
Ferulic acid (4) undergoes glycination forming feruloylglycine (51) and although this involves mammalian enzymes the conversion is reduced by 90% in ileostomists indicating that it is primarily colonic in origin.Methyl, glucuronyl, sulfate and glycine conjugation steps involve mammalian enzymes.Dehydroxylation and demethoxylation are almost certainly mediated by the gut microora while demethylation and hydrogenation steps can be mediated by both microbial and mammalian enzymes.For convenience, the pathways in Fig. 7 and 8 show C 6 -C 3 catabolites being converted by two a-oxidations to C 6 -C 1 compounds by microora and/or mammalian enzymes.However, it is possible the C 6 -C 3 catabolites progress directly to Fig. 7 Proposed metabolism of caffeoylquinic acids and caffeoylquinic lactones following the ingestion of coffee by volunteers.5-CQA is the illustrated structure but the respective 3 0 -and 4 0 -isomers and diCQAs would be metabolized in a similar manner.Likewise 4-CQAL is the illustrated lactone but 3-CQAL would be metabolised in a similar manner.COMT, catechol-O-methyltransferase; ET, esterase; RA, reductase; GT, UDP-glucuronyltransferase; ST, sulfuryltransferase; Co-A, co-enzyme A. Arrows: boldmajor routes, dottedminor pathways; redmicrobial enzymes, bluemammalian enzymes.Steps blocked in subjects with an ileostomy and hence occurring principally, but not necessarily exclusively, in the colon are indicated.* Intermediates that did not accumulate in detectable amounts (based on data of Stalmach et al. 44,139 and Ludwig et al. 226 ).
Table 9 Impact of acyl-quinic acid dose on ileal and urinary excretion of acyl-quinic acids and conjugated metabolites as a percentage of intake, and on metabolite sulfate : glucuronide (S : GlcUA) ratio, after ingestion of coffees and a fruit drink containing apple juice by ileostomists a (after Stalmach et al. 214  C 6 -C 1 structures via a mammalian catalysed b-oxidation and that C 6 -C 2 catabolites arise independently.In reality, further complexity is introduced as there are multiple points which catabolites might be absorbed.For example, a percentage of some C 6 -C 3 catabolites could be absorbed and undergo boxidation and/or mammalian phase II conjugation while the balance is subjected to microbial hydrogenation and/or aoxidation prior to absorption and mammalian conjugation.Also for some catabolites mammalian conjugation either does not occur or is incomplete.As noted above, aer feeding coffee containing trace levels of 3 0 ,4 0 -dimethoxycinnamic acid (26) and dimethoxycinnamoylquinic acids to volunteers with a functioning colon, 3 0 ,4 0 -dimethoxycinnamic acid appears rapidly in plasma and this is followed by a delayed appearance of a smaller amounts of 3-(3 0 ,4 0 -dimethoxyphenyl)propionic acid (45). 219These observations are in keeping with esterases in the small intestine hydrolysing the dimethoxycinnamoylquinic acids releasing free dimethoxycinnamic acid which is readily absorbed in the upper GI tract, while the unabsorbed methoxy acid reaching the colon it is converted to 3-(3 0 ,4 0 -dimethoxyphenyl)propionic acid.A potential pathway involved in such conversions is illustrated in Fig. 9. Currently there is no evidence for further phase II metabolism of these products.
Sulfates are the main acyl-quinic acid metabolites and glucuronides are relatively minor components (Fig. 6

and Table
Fig. 8 Proposed metabolism of feruloylquinic acids following the ingestion of coffee by volunteers.5-FQA is the illustrated structure but the respective 3 0 -and 4 0 -isomers would be metabolized in a similar manner.EST, esterase; RA, reductase; GT, UDP-glucuronyltransferase; ST, sulfuryltransferase; Co-A, co-enzyme A. Arrows: boldmajor routes; redmicrobial enzymes, bluemammalian enzymes.Steps blocked in subjects with an ileostomy and hence occurring principally, but not exclusively, in the colon are indicated (based on data of Stalmach et al. 44,139 and Ludwig et al. 226 ).9).Transformations that occur in the colon facilitate the absorption, as microbial catabolites, of 70% or more of the acylquinic acids consumed, clearly the importance of the colon in their bioavailability.It is also apparent that ileostomists are less able to utilise acyl-quinic acids and other dietary phenolics and are, thus, less able to access any benets that might otherwise accrue.
Further clarication of the proposed pathways in Fig. 7-9 will require feeding studies, not with coffee, but with substrates such as 5-CQA (9) and 5-FQA (34) in which the hydroxycinnamate moiety is labelled with 13 C.This approach has been used successfully with [ 13 C 5 ]cyanidin-3-O-glucoside to obtain detailed proles of the metabolism and catabolism of the anthocyanin in the small and large intestine and in the process provided novel details of involvement of the microbiota in the lower bowel. 229,230Human feeding studies are the gold standard and while animal studies can be helpful, because of species differences in gut microbiota and endogenous metabolism, they can produce data that are difficult to apply to human studies as has recently been shown with the radiolabelled avan-3-ol [2-14 C](À)-epicatchin. 236Likewise, data obtained in vitro with isolated cells and tissue extracts should be treated with caution as substrates can come into contact with enzymes and conditions to which they would not be exposed in vivo when cellular compartmentation is maintained.

Impact of dose on acyl-quinic acid bioavailability
The impact of dose on acyl-quinic acid absorption in the small intestine has been assessed in a study by Erk et al. on coffee consumption by ileostomists where acyl-quinic acid-derived compounds were analysed in plasma, urine and ileal uid aer the ingestion of coffees containing 1053, 2219 and 4525 mmole of CQAs, diCQAs and FQAs. 238The prole of acyl-quinic acids and their metabolites identied in ileal uid and urine by HPLC-MS was very similar to that detected by Stalmach et al. 214 aer the ingestion of a coffee containing 385 mmol of acyl-quinic acids.Ileal excretion reported by Erk et al. 238 was $70% of intake irrespective of dose (Table 9), which was also the level reported by Stalmach et al. 214 In keeping with these observations the $3 : 1 ratio of urinary excretion of metabolites absorbed in the distal and proximal GI tract was not inuenced to any extent by dose following coffee acyl-quinic acid intakes of 412-795 mmol (Table 10). 239Dose did, however, have a noticeable impact on the amount of conjugated metabolites appearing in ileal uid where they were equivalent to 22.3% of the acyl-quinic acids at the lowest dose and 6.7-8.9% at the three higher intakes. 238Sulfated metabolites were  a Data in nM AE standard error.Analysis of acyl-quinic acids at the three higher doses was aer enzyme hydrolysis 162 and without enzyme hydrolysis for the lowest dose. 139n.d., not detected; n.a., not analysed.
15.5 times more prevalent than glucuronides at the lowest dose but as intake increased this changed and with the ingestion of 4525 mmol of acyl-quinic acids the ratio of metabolites to glucuronides in ileal uid fell to 8.2 (Table 10).In keeping with these observations, in a further study in which ileostomists consumed a (poly)phenol-rich drink containing 46 mmol of 5-CQA derived from apples, only sulfate metabolites of caffeic acid, dihydrocaffeic acid and ferulic acid were detected in urine 0-24 h aer intake. 240hese dose-related changes presumably reect enzyme saturation, limited transport capacities into and out of the enterocyte in the small intestine and/or differences in GI tract transit times.The increasing proportion of glucuronidation with increasing acyl-quinic acid dose was even more marked in urine (Table 9), arguably indicative of phase II UDPglucuronyltransferase activity in the small intestine epithelium and/or in the liver and kidneys.
Data on the effect of dose on plasma C max of acyl-quinic acids and metabolites absorbed in the small intestine aer ingestion of coffee by healthy subjects with a colon (lowest dose) and ileostomists are presented in Table 11.This reveals a clear trend towards higher C max values with increasing acyl-quinic acids intakes.

Matrix effects and acyl-quinic acid bioavailability
There is one report on the consumption by humans of black coffee and black coffee made with milk rather than water in which urinary excretion of acyl-quinic acids and metabolites was 68% of intake with the black coffee and 40% aer ingestion of coffee with milk.The two gures were not statistically different but it was hypothesized that milk may have impaired the absorption of coffee acyl-quinic acids. 241Although an in vitro investigation suggests that the addition of milk fat to coffee may increase acyl-quinic acids bioavailability, 242 a feeding study by Renouf et al. 243 revealed no difference in the pharmacokinetic proles of plasma acyl-quinic acid metabolites aer drinking black coffee with or without 10% whole milk.Thus, although 5-CQA has been reported to bind to certain proteins in vitro, such as albumin and casein, 244,245 milk would appear not to have a signicant impact on the overall absorption of coffee acylquinic acids.However, adding a mixture of sugar and nondairy creamer to the black coffee resulted in lower C max values for caffeic acid (3) and isoferulic acid (43) accompanied by longer T max times for ferulic acid (4) and isoferulic acid. 243ugar 246 and lipids 247 are known to delay gastric emptying and this may have delayed absorption of the coffee acyl-quinic acids resulting in an extended T max for two of the three metabolites.
Changes in the CQA prole in the upper GI tract have been monitored by feeding ileostomists coffee, cloudy apple juice and an apple smoothie and analysing the CQAs excreted in ileal uid over an 8 h period. 205The data obtained are summarised in Table 12.With coffee, in keeping with other investigations, there was an overall recovery of 76% of intake and the ratios of the individual CQA isomers were not affected by inter-esterication reactions during passage through the proximal GI tract. 214,238There was a similar 77% recovery of apple smoothie CQAs but whereas the drink contained only 5-CQA ( 9) and 4-CQA (10) while the ileal uid also contained 3-CQA (11)  and 1-CQA (12).The same occurred following consumption of cloudy apple juice but in this instance the overall recovery of CQAs was only 26%.
The matrix in which the CQAs were ingested, therefore, appears to have had an impact on their fate as they pass through the proximal GI tract with interesterication of CQAs occurring with the apple products but not coffee.Such reactions have been reported previously 248,249 and seemingly are caused by the pH increasing to above pH 6 during passage through the small intestine.Erk et al. 238 suggest that the CQA prole of the coffee was closer to the interesterication equilibrium reported by Trugo and Macrae 250 and so did not alter substantially during gastric transport (Table 12).The susceptibility of individual acyl-quinic acids to inter-esterication and hydrolysis was discussed in Section 3.1.
There was also a $3-fold lower recovery of CQAs in ileal uid aer cloudy apple juice consumption compared with the apple smoothie (Table 12). 205The CQA content of the two beverages was similar and the drinks differed only in that cloudy apple juice was pressed and unltered whereas the smoothie comprised 60% cloudy juice and 40% apple purée which contains a much higher proportion of cell wall constituents.How this resulted in the enhanced ileal recovery of CQAs from the apple smoothie is unclear.Although neither plasma nor urine were analysed in this study to conrm the point, it appears unlikely that it is a consequence of enhanced absorption in the small intestine of the cloudy apple juice CQAs.It is also unclear as to what constituents in the smoothie result in the CQAs being less prone to degradation and/or irreversible binding in the apple matrix.Scherbl et al. reported a study in which volunteers consumed (i) black coffee, (ii) a carbohydrate-rich black coffee and two bread rolls and honey ($626 kcal) and (iii) a fatrich black coffee and one bread roll and peanut butter ($626 kcal). 209Plasma and urine collected 0-24 h postingestion were analysed and a detailed spectrum of acylquinic acid metabolites was obtained.The fat-and carbohydrate-rich supplements resulted in a statistically signicant, but relatively small, reduction in the levels of acylquinic acid metabolites.This contrast with more marked effects obtained in an anthocyanin study in which a strawberry drink was ingested either 2 h before or 2 h aer a breakfast meal which comprised a croissant with butter and apple jelly, cereal, whole milk and sausages (838 kcal). 251Coingestion of the drink with the meal resulted in the C max of the main anthocyanin, a pelargonidin-O-glucuronide, falling signicantly to 12.8 AE 2.1 nmol L À1 compared to 38.0 AE 6.6 and 35.5 AE 2.1 nmol L À1 when the drink was ingested, respectively, before and aer the meal.Co-ingestion with the meal also extended the T max from $1.8 h to 2.9 h.

Inter-individual variations in acyl-quinic acids excretion aer coffee intake
Data obtained on urinary excretion of acyl-quinic acids and their metabolites following coffee consumption showed substantial inter-individual variation.With a 412 mmole intake by 11 volunteers the mean excretion was equivalent to 26 AE 4% of intake while individual values ranged from 12 to 58% urinary recoveries.Similar gures were obtained with intakes of 635 and 793 mmole (Table 13).It is noteworthy that while there was some variation, high excretors such as volunteer 4, maintained the condition at all three doses as did low excretors such as volunteers 1-3.Much of this person-to-person variation probably reects variation in the composition and biochemical competence of the gut microbiota. 239The inter-individual variation in the bioavailability of dietary (poly)phenols in general, is an area of increasing interest because the knowledge of the variation in the capacity to absorb and metabolize these compounds is thought to be a key factor in obtaining a better knowledge of the benecial effects of plant bioactive compounds against diseases, and particularly, an understanding of their role in healthy ageing and cardiometabolic risk reduction. 252

Biomarkers of acyl-quinic acids intake
The data presented in Table 8, and that of Gomez-Juaristi et al. 206 on urinary excretion of hydroxycinnamate metabolites aer the respective ingestion of coffee and maté by healthy volunteers clearly show that dihydrocaffeic acid-3 0 -sulfate (40) and feruloylglycine (51) would serve as very sensitive biomarkers for the consumption of relatively small amounts of acyl-quinic acids.A more detailed ngerprint for coffee could be obtained by the additional analysis of 3 0 ,4 0 -dimethoxycinnamic acid (26), dihydroferulic acid (35), dihydroferulic acid-4 0 -sulfate (39), ferulic acid-4 0 -sulfate (36) and dihydroferulic acid-4 0 -O-glucuronide (52).However, whether coffee and/or maté consumption can be accurately assessed in population studies from measurements of urinary biomarkers is doubtful.These colonic catabolites are not unique to acyl-quinic acids as some can be derived from other (poly)phenols including orange juice avanones and berry anthocyanins.Intake of CQAs from coffee or maté is, however, likely to yield much higher levels of the catabolites than consumption from alternative sources.4][255] However, relatively little information exists about their mechanisms of action.For many years, mechanistic studies linked the protective effects of dietary (poly)phenols to their antioxidant activity.Acyl-quinic acids, particularly the dominating CQAs, are frequently referred to as powerful antioxidants, and this might be true in vitro.Aer coffee consumption by humans, unmetabolised CQAs achieve transient and very low nmol L À1 peak plasma concentrations (Table 6), thus, they cannot realistically compete with the more substantial concentrations of other dietary antioxidants, such as the far more powerful ascorbate and atocopherol. 256As discussed in Section 7.1, aer ingestion acylquinic acids are extensively metabolised and the concept that health benets might arise from direct-antioxidant activity has been challenged repeatedly, [256][257][258] and it has now been discarded and alternative mechanisms are being investigated.
(Poly)phenols have been found to exert modulatory effects in cells through selective action on multiple cell-signalling pathways involved in the pathogenesis of degenerative diseases, indicating that the health effects go beyond simple antioxidant activity. 259,260However, most in vitro studies investigating the biological activity of acyl-quinic acids have used the unmetabolised parent compounds present in foods and beverages and, in many cases, at supra-physiological concentrations that are never attained in the circulatory system, and organs or tissues, except perhaps in the GI tract.This section will, therefore, cover studies using catabolites of acyl-quinic acids and their phase II metabolites that have been tested at physiologically relevant sub-and low mmol L À1 concentrations.As noted in Section 8, the relevant catabolites are not unique to acyl-quinic acids so studies that originally focussed on other dietary polyphenols are included.
Several such studies have investigated benecial effects, on markers of cardiovascular diseases and the possible molecular mechanisms involved.,Van Rymenant et al. 261 compared the effects of ferulic acid (4) and ferulic acid-4 0 -sulfate (36) on vasorelaxation ex vivo using mice saphenous artery, femoral artery and aorta.While ferulic acid was inactive, its ferulic acid-4 0 -sulfate, caused a concentration-dependent relaxation in all three tissues that were signicant at sub-mmol L À1 concentrations.Further analyses demonstrated that soluble guanylate cyclase (sGC) and voltage-gated K + -channels, are involved in the ferulic acid-4 0 -sulfate-dependent vasorelaxation.Whether these effects also occur with human test systems remains to be determined.It is of interest, however, that following human consumption of coffee containing 412 mmol of acyl-quinic acids, ferulic acid-4 0 -sulfate, but not ferulic acid, appears in the circulatory system with a dual C max of 76 nmol L À1 aer 0.6 h and 46 nmol L À1 aer 4.3 h (Fig. 6 and Table 6). 44min et al. 262 studied the anti-inammatory effects of ferulic acid, and protocatechuic acid (3,4-dihydroxybenzoic acid) (54)  and its glucuronide and sulfate conjugates in human umbilical vein endothelial cells (HUVECs) stimulated with either oxidized LDL or a cluster of differentiation 40 ligand.Protocatechuic acid and its phase II metabolites were effective in modulating the production of key inammatory mediators IL-6 and vascular cell adhesion molecule-1 (VCAM-1) at dietary relevant concentrations as low as 100 nmol L À1 , with maximum reduction observed for the sulfate conjugates.In keeping with the data of Van Rymenant et al. 261 these results indicate that phase II conjugation may unexpectedly increase bioactivity.However, in a similar study by Warner et al. 263 that analysed the effects of phenolic catabolites on secretion of VCAM-1 in HUVEC-stimulated with tumor necrosis factor alpha (TNF-a), a statistically signicant reduction of 17.2% was observed with 1 mmol L À1 protocatechuic acid, but not its phase II metabolites.Protocatechuic acid at 0.5 mmol L À1 has also been shown to signicantly reduce adhesion of monocytes to HUVECs by 28.7%.Protocatechuic acid-3-sulfate (55) and protocatechuic acid-4-sulfate (56) also signicantly reduced VCAM-1 secretion but at the much higher concentrations of 10 and 100 mmol L À1 , respectively. 264In the same study, 1 mmol L À1 ferulic acid (4) signicantly reduced monocyte adhesion by 21.3%.Adhesion of monocytes to endothelial cells and their subsequent transendothelial migration into the vascular wall initiate the atheroma formation which leads to cardiovascular diseases.
Baeza et al. 266 studied the inhibitory effects of caffeic acid (3), ferulic acid (4), dihydrocaffeic acid (37), and dihydroferulic acid (35) on platelet activation in human blood samples by measuring ADP-induced P-selectin expression.Excessive platelet activation has been associated with development of chronic inammation and has, therefore, been proposed as a risk factor for cardiovascular diseases. 267A signicant decrease in P-selectin expression was observed with dihydrocaffeic acid at 1 mmol L À1 while ferulic acid and dihydroferulic acid were without effect at concentrations up to 10 and 20 mmol L À1 , respectively.
González-Sarrías et al. 268 evaluated the potential neuroprotective effects of 19 (poly)phenol-derived metabolites, including dihydrocaffeic acid (37) and 3 0 ,4 0 -dihydroxyphenylacetic acid (46).At 1 mmol L À1 no protective effects were observed but at 5 mmol L À1 both CQA catabolites signicantly attenuated the H 2 O 2 -induced cytotoxicity in SH-SYS neuroblastoma cells.The strongest neuroprotective effects occurred at 10 mmol L À1 and at this concentration they were also able to signicantly reduce ROS levels in the SH-SY5Y cells and decrease oxidative stress-induced apoptosis by preventing caspase 3-activation via mitochondrial apoptotic pathway.
In a similar study Verzelloni et al. 269 investigated the ability of an array of (poly)phenol catabolites generated in vivo from different food sources, including coffee, to counteract neurotoxicity linked to oxidative stress in human SK-N-MC neuroblastoma cells.The compounds were tested at concentrations ranging from 0.1 to 20 mmol L À1 .Dihydrocaffeic acid (37)  exhibited signicant protective effects at 0.5 mmol L À1 , while at the highest tested concentration dihydroferulic acid (35) and feruloylglycine (51) also protected neurons against dimethoxy-1,4-naphthoquinone-induced toxicity, with increased survival aer oxidative stress ranging from 6.5 to 17.0%.The possible synergistic effects were investigated by evaluating the neuroprotective effects of different combinations of catabolites.When assayed in this manner at concentrations of 0.5 mmol L À1 , the three acyl-quinic acids catabolites induced a statistically signicant 16% increase in cell viability.
It is unfortunate that there are still comparatively few studies of relevant in vitro metabolites and catabolites at concentrations close to those known to occur in humans consuming real-world diets.Nevertheless, the results summarised here suggest that at least some gut ora catabolites and their phase II metabolites may indeed have anti-inammatory and neuroprotective effects at such concentrations.Although the consumption of acylquinic acids is not an absolute pre-requisite for the production of these catabolites and metabolites, the fact that acylquinic acids-rich beverages such as coffee and maté are typically consumed repeatedly at short intervals throughout the day, day aer day, makes the acyl-quinic acids particularly important.This frequent and long term consumption results in several bolus doses being available simultaneously to the gut microbiota and greatly increasing the period of time during which catabolite absorption can occur.Accordingly there is a potential for a somewhat higher plasma C max than would be achieved by a single bolus dose, but more importantly, a signicant concentration is achieved for several hours rather than a few minutes. 270For these reasons plasma AUC values and 24 hour urinary excretion values are better indicators of physiological relevance than C max values per se.Such data for freeliving healthy and clinically compromised volunteers, in addition to volunteers given dened supplements, would be a valuable adjunct to epidemiological and in vitro studies.

Conflicts of interest
There are no conicts to declare.
acyl-quinic acids 5Role of acyl-quinic acids in planta 6Acyl-quinic acids in coffee 7Bioavailability and metabolism of acyl-quinic acids in humans 7.1 Studies involving volunteers with and without a functioning colon 7.2 Impact of dose on acyl-quinic acid bioavailability 7.3 Matrix effects and acyl-quinic acid bioavailability 7.4 Inter-individual variations in acyl-quinic acids excretion aer

Fig. 9
Fig. 9 Potential metabolism of 5-O-(3 0 ,4 0 -dimethoxycinnamoyl) quinic acid following the ingestion of coffee by volunteers.The 5-Oester is the illustrated structure but the respective 3-O-and 4-Oisomers would be metabolized in a similar manner.EST, esterase; RA, reductase.Blue arrows mammalian enzymes, red arrowmicrobial enzymes (based on data of Farrell et al. 144 ).

Table 1
Acyl-quinic acid content of coffees, apples juices, globe artichoke and mat éa a Data expressed as mean values (n ¼ 3).

Table 5
44antities of acyl-quinic acids in a 200 mL serving of instant coffee fed to volunteers (after Stalmach et al.44

Table 7
214ntities of acyl-quinic acids and metabolites recovered in ileal fluid collected 0-24 h after the consumption of 200 mL of instant coffee containing 385 mmol of acyl-quinic acids by humans with an ileostomy (after Stalmach et al.214) a a Data presented as mean values AE standard error (n ¼ 5).n.dnot detected.This journal is © The Royal Society of Chemistry 2017 Nat.Prod.Rep., 2017, 34, 1391-1421 | 1407 Review Natural Product Reports Open Access Article.Published on 21 November 2017.Downloaded on 8/10/2024 6:25:26 AM.This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.
44,214) a a Data represent mean values in mmol AE standard error.n.d.not detected.Different superscripts within rows indicate a statistical difference between the two sets of volunteers (two-sample t-test, P-value < 0.05).b Figures in bold italicised parentheses indicate excretion as a percentage of acyl-quinic acids intake.
205 )rk et al.205 ) a Data presented as mean values AE standard error.

Table 10
Urinary excretion of acyl-quinic acid metabolites absorbed in the upper and lower gastrointestinal tract by volunteers with a functioning colon after the consumption of 200 mL of coffee beverage containing 412, 635 and 795 mmol of acyl-quinic acids.(afterStalmachet al.239) a a Data expressed as mean values in mmol AE SE (n ¼ 11).Italicised gures in parentheses represent the percentage absorption taking place in the upper and lower gastrointestinal tract.

Table 11
238act of acyl-quinic acid dose on plasma C max of acyl-quinic acids and metabolites absorbed in the small intestine after ingestion of coffees by healthy volunteers with a colon (lowest dose) (Stalmach et al.44) and ileostomists (three higher doses) (after Erk et al.238) a

Table 12
205mary of caffeoylquinic acid intake by ileostomists and 0-8 h CQA ileal excretion after the ingestion of coffee, apple smoothie and cloudy apple juice.(afterErket al.205) a a Data presented in mmoles as mean values.Bold gures in parentheses represent total caffeoylquinic acid recovery in ileal uid as a percentage of intake.n.d., not detected.

Table 13
239mary of the quantities of total acyl-quinic acids and their metabolites, excreted in urine 0-24 h after the consumption of 200 mL of coffee beverage containing 412, 635 and 795 mmol of acyl-quinic acids by 11 volunteers with an intact functioning colon (based on unpublished data of Stalmach et al.239) a a Data expressed as a percentage of intake.