Caffeoylquinic acids: chemistry, biosynthesis, occurrence, analytical challenges, and bioactivity.

INTRODUCTION
Caffeoylquinic acids (CQAs) – esters of caffeic acid with quinic acid – are specialized bioactive metabolites derived from the phenylpropanoid biosynthesis pathway. In plants, CQAs play a defensive role against biotic or abiotic stress (Uleberg et al., 2012). In humans, CQAs have a wide range of potential benefits with therapeutic applications. These compounds have been reported to exhibit antioxidant (Hamed et al., 2020; Nzekoue et al., 2020; Trendafilova et al., 2020), antibacterial (Hamed et al., 2020; Nzekoue et al., 2020), cancer-related (Bourgou et al., 2017; Bulgakov et al., 2018; Giorgio et al., 2015; Liu, Li, Zu, et al., 2020; Murad et al., 2015; Taira et al., 2014; Zheleva-Dimitrova et al., 2017), antiviral, anti-Alzheimer (Matthews et al., 2020; Tsunoda et al., 2018), and neuroprotective activity (Gray et al., 2014; Metwally et al., 2020; Sasaki et al., 2019). Our group has recently shown that CQAs can reverse cognitive deficits in a mouse model of Alzheimer’s disease (Matthews et al., 2020). Furthermore, in preclinical studies, CQAs are able to inhibit α-glucosidase and were proposed to mitigate the incidence of so-called lifestyle-related diseases, such as diabetes (Imai et al., 2020). Every day, we ingest CQAs from plant-derived food, including fruits and vegetables. An important source of these compounds is coffee beverages. There is a strong association between coffee consumption and lower incidence of several degenerative diseases, as well as a positive association between longevity and the presence of CQAs (Farah and de Paula Lima, 2019). However, despite the substantial amount of evidence regarding the benefits of CQAs, randomized, double-blind, placebo-controlled studies are still lacking.
Despite the diversity of reported qualitative and quantitative methodologies for the study of CQAs (Liu, Li, Zu, et al., 2020; Wianowska and Gil, 2019), comprehensive characterization methods for CQAs are still in an early stage of development. There are several challenges associated with the reliable characterization of individual CQAs, including degradation and transformation through sample processing, a lack of commercially available authentic standards, matrix interferences, and the presence of multiple isomers with similar physicochemical properties that make the unambiguous characterization of isomers challenging.
This review focuses on the chemistry, biosynthesis, occurrence, analytical challenges, and bioactivity of CQAs. Here we describe the most recently updated biosynthetic pathways for the production of CQAs in plants, including a new pathway involving a peroxidase-type p-coumaric acid 3-hydroxylase found in Brachypodium distachyon (Barros et al., 2019), the recently discovered GDSL lipase-like enzyme, and the BAHD acyltransferase recently reported in sweet potato (Ipomoea batatas) (Miguel et al., 2020). We focus on caffeic acid substituents in this class of compounds. Other esterifications of hydroxyl and carboxylic groups of the quinic acid moiety, such as methyl, ethyl, butyl, and ferulic, cumaric, and malonic acid groups, are beyond the scope of this focused review. Some current advances for other substituents are reported elsewhere (Liu, Li, Zu, et al., 2020; Wianowska and Gil, 2019).

CHEMISTRY
The chemical structure of CQAs consists of quinic acid ((1S,3R,4S,5R)-1,3,4,5-tetrahydroxycyclohexane-1carboxylic acid) acylated with one to four caffeic acid moieties. Quinic acid possesses four stereocenters (carbons 1,3,4,5). Each one of these positions can be substituted by caffeic acid, giving rise to 15 possible combinations (Table 1), encompassing four monoCQAs, six diCQAs, four triCQAs, and one tetraCQA.
There is prevalent confusion regarding the nomenclature of mono-, di-, and triCQAs. According to IUPAC rules, for multi-substituted rings, the numbering is in the direction that gives the lowest numbers for substituents. To illustrate this, let us consider the case for 3-, and 5-monoCQAs. 3-O-CQA ((1S,3R,4S,5R)-1-(((E)-3-(3,4-dihydroxyphenyl)acryloyl)oxy)3,4,5-trihydroxycyclohexane-1-carboxylic acid) corresponds to caffeoyl substitution at the IUPAC position 3 (bold in IUPAC name) of the quinic acid. The same is true for the IUPAC name for 5-O-CQA ((1R,3R,4S,5R)-3-(((E)-3-(3,4dihydroxyphenyl)acryloyl)oxy)-1,4,5-trihydroxycycl ohexane-1-carboxylic acid). Nevertheless, they differ in stereochemistry at position 1 (underlined in the IUPAC name). This apparent IUPAC ‘ambiguity’ can be explained as follows. If we transfer caffeic acid from position 3 to position 5, the new position becomes position 3 (the direction that gives the lowest numbers according to IUPAC guidelines). This same principle applies from one to three substituents, and this may contribute to some of the ambiguities concerning some of the diCQAs, such as 1,3- and 1,5-diCQAs, whosenaming is inconsistent across the literature. For instance, 1,3- and 1,5-diCQAs have the same structural position of the caffeic acid moieties; however, they differ in stereochemistry. To avoid this ambiguity, for this review, we will keep consistent the numbering of the quinic acid moiety, according to Table 1. This numbering is currently in use by the U.S. National Library of Medicine of the NIH (Kim et al., 2020d). This nomenclature is strongly encouraged to decrease the confusion when sharing future research.
Quinic acid is considerably more polar than caffeic acid. The topological polar surface area for quinic acid is 118 Ų versus 77.8 Ų for caffeic acid (Kim et al., 2020d). Consequently, the polarity of CQAs decreases with the degree of esterification with caffeic acid: monoCQAs > diCQAs > triCQAs > tetraCQA. CQAs are soluble in water; however, solubility decreases rapidly with the number of caffeic acid substituents due to the increased lipophilicity.
Each caffeic acid moiety possesses five conjugated double bonds. These conjugated bonds of CQAs have an important role in stabilizing radicals by resonance, and therefore potentially govern their bioactivity. The total number of conjugated double bonds in CQAs can offer an explanation of why CQAs with a higher degree of esterification have a greater capacity to decrease oxidative stress by scavenging free radicals (Liang et al., 2019). Caffeic acid has strong UV-VIS absorption bands due to its conjugated double bonds, making it useful for matrix-assisted laser desorption/ionization mass spectrometry (MALDI-MS) applications. However, CQAs are photosensitive and chemically unstable in general. Xue et al. (2016) observed exposure of CQAs to light for 7 days resulted in significant cis-trans isomerization and degradation, probably due to the higher stability of some of the isomers or degradation products. Yet, monoCQAs are far more stable than diCQAs, likely due the fact that the ester bond linkage in the quinic acid moiety is less stable in an axial bond configuration than in an equatorial bond configuration (Xue et al., 2016). TriCQAs and tetraCQA are far less abundant in nature and less studied than mono- or diCQAs. However, it is expected that CQAs with a higher degree of esterification degrade to diCQAs, then to monoCQAs, and finally to quinic and caffeic acid under harsh extraction conditions.
Another remarkable property of CQAs is their capacity to undergo structural modification by spontaneous acyl migration – migration of caffeoyl residues – among the hydroxyl groups of the quinic acid. Acyl migration plays a key role in the diversity of CQAs naturally occurring in plants since intramolecular acyl migration can take place in the vacuole (Moglia et al., 2014) or in aqueous solution influenced by the pH and temperature (Deshpande et al., 2014; Xue et al., 2016).
In addition to light exposure, heating of CQAs causes not only their isomerization by acyl migration but also their transformation. For example, heating 5-CQA in the presence of water will produce a mixture of 5-, 3-, and 4-CQA and the hydrolysis products quinic acid and caffeic acid (Dawidowicz and Typek, 2017). Therefore, plant extracts obtained by Soxhlet extraction, decoction, infusion, and ultrasound-assisted and microwave-assisted extractions may exhibit significant differences in the isomeric ratios of CQAs from the original plant. Nevertheless, the resulting extract is likely to be more stable for storage and characterization, producing more reproducible results. An important recommendation to reduce these shortcomings is the standardization of extraction procedures followed by a comprehensive characterization of the end product. It should be stressed that characterization should include quantification of potential bioactives if known, as well as a chemical fingerprint of the extract determined by untargeted analysis (Alcazar Magana et al., 2020; Kim, Kim, et al., 2020; Nie et al., 2019; Sun et al., 2018; Villalon-Lopez et al., 2018; Wan et al., 2017; Wang, Gao, et al., 2019; Zengin et al., 2017).

BIOSYNTHESIS
The synthesis of CQA components, quinic acid and caffeic acid, begins in the shikimic acid pathway within plastids (Figure 1) (Maeda and Dudareva, 2012). This pathway first synthesizes 3-dehydroquinic acid from phosphoenolpyruvic acid and erythrose-4-phosphate produced by glycolysis and the pentose phosphate pathway, respectively. 3-Dehydroquinic acid is subjected to reversible reduction by quinic acid dehydrogenase (QDH) to produce quinic acid as a secondary metabolite of the shikimic acid pathway (Carrington et al., 2018; Maeda and Dudareva, 2012). QDH has a similar structure as shikimic acid dehydrogenase (SDH), which catalyzes shikimic acid production in the major route of the shikimic acid pathway. Based on the phylogenetic analysis of QDH and SDH in plant species, Carrington et al. (2018) found that SDH was universally conserved, whereas QDH was confined to seed plants, suggesting that QDH evolved from the ancestral enzyme of SDH (Carrington et al., 2018).
Since CQAs have been suggested to play a role in the storage of lignin monomer intermediates (Schoch et al., 2001), plants may have acquired a QDH to improve the lignin biosynthetic system during evolution. Phenylalanine and tyrosine produced as the final products of the shikimic acid pathway are transported from the plastid to the cytosol by the plastidial cationic amino acid transporter (Widhalm et al., 2015). Recently, Qian et al. (2019) found a sub-pathway in which chorismic acid is transported to the cytosol and then utilized to produce phenylalanine in the cytoplasm. As shown in Figure 1, among the compounds synthesized in the plastidial shikimic acid pathway, quinic acid, shikimic acid, chorismic acid, tyrosine, and phenylalanine are transported to the cytoplasm (Maeda and Dudareva, 2012). A quinic acid transport protein, the H+-symporter QutD, has been found in fungi (Aspergillus nidulans and Neurospora crassa) (Enquist-Newman et al., 2014; Geever et al., 1989), but has not yet been reported in plants.
In the cytosol, various compounds, including CQAs as well as lignin monomers, flavonoids, and stilbenes, are synthesized by the phenylpropanoid pathway (Figure 1) (Vogt, 2010). CQA synthesis has also been found both in the cytosol and in chloroplasts in some plants, e.g., Coffea canephora and Lonicera japonica (Li et al., 2019; Mondolot et al., 2006). This pathway has been studied mainly as a biosynthetic pathway for lignin monomers, and a vast amount of knowledge has been accumulated (Boerjan et al., 2003; Vanholme et al., 2019). Phenylalanine is converted to p-coumaric acid by phenylalanine ammonia-lyase (PAL) and cinnamic acid 4-hydroxylase (C4H). In grasses, tyrosine deamination by tyrosine ammonia-lyase is an alternative route to yield p-coumaric acid (Barros et al., 2016). Then p-coumaroyl-CoA is generated by the ligation of coenzyme A (CoA) with p-coumaric acid by 4-coumaric acid:CoA ligase (4CL).
p-Coumaroyl-CoA is a major branch for the production of various phenolic compounds, including flavonoids and stilbenes, in addition to lignin monomers and CQAs. P-Coumaroyl-CoA is transesterified with shikimic acid or quinic acid to produce corresponding esters by hydroxycinnamoyl-CoA:shikimic acid/quinic acid hydroxycinnamoyltransferase (HCT) or hydroxycinnamoyl-CoA: quinic acid hydroxycinnamoyltransferase (HQT) (Hoffmann et al., 2003). The routes to produce p-coumaroyl shikimic acid and p-coumaroyl quinic acid from p-coumaroyl-CoA are called the ‘shikimic acid shunt’ and the ‘quinic acid shunt’, respectively (Ha et al., 2016) (Figure 1). Of these routes, p-coumaroyl-CoA is basically converted to p-coumaroyl shikimic acid, since lignin monomer synthesis is operated via the shikimic acid shunt (Volpi e Silva et al., 2019). Both HCT and HQT catalyze the same reaction, but prefer shikimic acid and quinic acid as substrates, respectively (Hoffmann et al., 2003; Niggeweg et al., 2004). Also, both enzymes catalyze the reverse reaction, allowing the corresponding CoA ester to be produced from the hydroxycinnamoyl esters of shikimic acid and quinic acid (Hohlfeld et al., 1995; Niggeweg et al., 2004). p-Coumaroyl shikimic acid and p-coumaroyl quinic acid are oxidized at the 3′ position of the aromatic ring by cytochrome P450-type p-coumaric acid ester-specific 3′-hydroxlase (C3′H, CYP98) to produce the corresponding caffeoyl ester. In the quinic acid shunt, a typical CQA, 3-CQA (chlorogenic acid), is generated by oxidation of p-coumaroyl quinic acid. In contrast, caffeoyl shikimic acid proceeds to the synthesis of guaiacyl/syringyl-type lignin monomers or 3-CQA via multi-branched pathways involving various enzymes such as caffeoyl shikimic acid esterase (CSE), 4CL, HCT/HQT, caffeic acid O-methyltransferase (COMT), and caffeoyl-CoA O-methyltransferase (CCoAOMT) (Figure 1).
The expression of phenylpropanoid biosynthetic pathway genes differs between plant species. It is also known that the expression of the lignin monomer synthetic pathway varies among plant species (Vanholme et al., 2019). This means that the expression of the synthetic pathway of CQAs also differs depending on the plant species. As shown in Figure 2, there are four typical 3-CQA synthetic routes starting from p-coumaric acid. Routes 1 (blue) and 2 (yellow) are common until the generation of caffeoyl shikimic acid through the shikimic acid shunt. In route 1, caffeoyl shikimic acid is hydrolyzed by CSE to produce caffeic acid and shikimic acid (Vanholme et al., 2013). The resultant caffeic acid is then converted to caffeoyl-CoA by 4CL, and subsequently transesterified with quinic acid by HQT/ HCT to generate 3-CQA. Gene suppression analysis of HQT in tomato (Solanum lycopersicum) and artichoke (Cynara cardunculus var. scolymus) showed that HQT plays a major role in 3-CQA production from caffeoyl-CoA (Payyavula et al., 2015; Rhodes and Wooltorton, 1976). By contrast, in route 2, caffeoyl-CoA is generated directly from caffeoyl shikimic acid by the reverse reaction of HCT/HQT. CSE plays a large role in the conversion of caffeoyl shikimic acid in angiosperms such as Arabidopsis, Medicago, and Populus and gymnosperms such as Larix kaempferi (Ha et al., 2016; Saleme et al., 2017; Vanholme et al., 2013; Wang, Chao, et al., 2019). Also, the CSE gene is conserved in most plants (Ha et al., 2016). From these facts, it is considered that most plants use route 1 (Figure 2) to synthesize 3-CQA. On the other hand, some plant species, including B. distachyon and Zea mays, do not have the CSE gene (Ha et al., 2016). In 2019, a novel peroxidase-type p-coumaric acid 3-hydroxylase (C3H) was found in B. distachyon (Barros et al., 2019). In route 3 (green), 3-CQA is generated via caffeic acid, which is directly produced from p-coumaric acid by C3H (Figure 2). Plants with no CSE gene may produce 3-CQA via route 2 or route 3. Route 4 (pink) is the simplest route; 3-CQA is synthesized from p-coumaroyl-CoA by reactions of HQT/HCT and C3′H in the quinic acid shunt. This route would be a by-pass branch since the essential lignin biosynthetic pathway basically uses the shikimic acid shunt. In addition to these routes, a route through cinnamoyl glucose has been reported in I. batatas (Kojima and Uritani, 1972a, 1972b). In this pathway, cinnamic acid is first converted to cinnamoyl-glucose by UDP-glucose:cinnamic acid glucosyltransferase, and then caffeoyl-glucose is produced by hydroxylation at positions 3 and 4 of the aromatic ring. Hydroxycinnamoyl glucose: quinic acid hydroxycinnamoyl transferase transfers quinic acid to caffeoyl-glucose to synthesize 3-CQA (Villegas and Kojima, 1986).
The 3-CQA biosynthesis pathways have been metabolically engineered in agricultural crops such as tomato and sweet potato to increase the production of 3-CQA (Niggeweg et al., 2004; Yu et al., 2021). These engineered plants exhibit enhanced 3-CQA production, which is considered beneficial to human health (Niggeweg et al., 2004). Overexpression of I. batatas PAL (IbPAL1) in sweet potato has been reported to increase the CGA levels in leaves (Yu et al., 2021). Although the expansion of the secondary xylem was also stimulated in stems, the formation of storage roots was inhibited. Considering that the biosynthesis pathway of CQAs is closely related to that of flavonoids and lignin monomers, strict spatiotemporal control of the metabolic pathway is important to increase 3-CQA production without growth penalty. In bacteria, research efforts were conducted to produce CQA using engineered Escherichia coli mutants that express 3-CQA biosynthesis genes from plants (Kim et al., 2013; Li et al., 2021).
HQT expression levels have been associated with 3-CQA production levels in several plants, such as tomato, artichoke, and potato (Solanum tuberosum) (Navarre et al., 2013; Niggeweg et al., 2004; Sonnante et al., 2010). Li et al. found that the spatiotemporal expression pattern of HQT is directly correlated with the distribution of 3-CQA biosynthesis sites in L. japonica flowers (Li et al., 2019). 3-CQA is synthesized mainly in the cytoplasm and chloroplasts and then transferred to the vacuole for long-term storage. The transport of CQA from the cytoplasm to the vacuole is expected to be mediated by vesicles in L. japonica (Li et al., 2019). In C. canephora, 3-CQA seems to play a role in the storage of caffeine in the vacuole (Nic-Can et al., 2015).
DiCQAs and triCQAs are often found in plants, but little is known about the biosynthesis of these compounds. HCT of Coffea spp. and HQT of S. lycopersicum have been shown to not only catalyze normal 3-CQA synthesis but also 3,5-diCQA production from 3-CQA (Lallemand et al., 2012; Moglia et al., 2014) (Figure 1). In S. lycopersicum, HQT produces 3,5-diCQA by two kinds of reactions: (i) a second acyltransferase reaction using 3-CQA and caffeoyl-CoA as an acyl donor or (ii) an acyltransferase reaction using two 3-CQAs molecules as both the acyl donor and acyl acceptor. The diCQA isomers 4,5-diCQA and 3,4-diCQA were produced by spontaneous acyl migration of 3,5-diCQA in the vacuole (Moglia et al., 2014). Based on the fact that both HQT and 3-CQA are localized in the vacuole as well as the cytosol, the optimum pH of HQT to produce 3,5-diCQA is low, and the pH in the vacuole is also relatively low. 3,5DiCQA has been suggested to be synthesized by HQT in the vacuole of S. lycopersicum (Moglia et al., 2014). Very recently, Miguel et al. (2020) discovered that a GDSL lipase-like enzyme from I. batatas, isochlorogenic acid (3,5-diCQA) synthase (IbICS), catalyzes the production of 3,5-diCQA (Miguel et al., 2020). Unlike S. lycopersicum HQT, IbICS synthesizes 3,5-diCQA by only an acyltransferase reaction using two 3-CQA molecules (Figure 1). The recombinant IbICS produced in Nicotiana benthamiana showed optimum reaction conditions of pH 6.3 and 39.9°C, and the Km value was determined to be 3.5 mM for 3-CQA (S. lycopersicum HQT, 16 mM). By expressing IbICS in Pichia pastoris, this yeast produced 1.2 g L−1 of 3,5-diCQA from 2.6 g L−1 of 3-CQA (61.3% mol efficiency). Since IbICS showed significantly higher activity to produce 3,5-diCQA than HCT of I. batatas, IbICS has been hypothesized to play a major role in the production of 3,5-diCQA. GDSL lipase-like family enzymes related to IbICS have been found in Rauvolfia serpentina and Alopecurus myosuroides (Miguel et al., 2020).

OCCURRENCE
CQAs are among the most abundant polyphenols. Relatively high concentrations of monoCQAs, ranging from 46 to 1662 mg L−1, have been reported in coffee brews (Moeenfard et al., 2014). They also contain several forms of other CQAs. Among the CQAs, chlorogenic acid (3-CQA) is the most abundant in plants such as Coffea arabica and Camellia sinensis. MonoCQAs can account for about 10% of the mass in green coffee beans (Tajik et al., 2017). The monoCQA concentration in different green coffee extracts ranges between 131 and 221 g kg−1 (Jeszka-Skowron et al., 2016). Plant species from the Coffea genus, species from the Asteraceae family, C. sinensis, Ilex paraguariensis, and Centella asiatica are particularly rich in CQAs in both quantity and diversity. Our group focuses on the study of C. asiatica – a botanical recognized for enhancing cognition (Gray et al., 2018). Our findings support its potential as a phytotherapeutic for aging (Wright et al., 2020) due to its high content of CQAs and triterpenoids (Alcazar Magana et al., 2020). In C. asiatica water extract, we found a wide range in percent composition across different plant accessions. The total relative content of monoCQAs ranged from 0.15% to 0.47%, and the total relative content of diCQAs ranged from 0.12% to 1.4%, with 4,5-diCQA being the most abundant, accounting for 0.46% of the dry mass of the water extract. Furthermore, we found similar concentrations for 3,4-, 1,5-, and 3,5-diCQAs (Alcazar Magana et al., 2020).
CQAs are also ubiquitous among fruits and vegetables. They are present in pomes (Rosaceae) and drupes (Prunus), berries (Rubus), tropical fruits, Apiaceae and Brassicaceae vegetables, root vegetables, and seeds (Rothwell et al., 2013). Furthermore, CQAs have been reported in roots, rhizomes, flowers, leaves, fruits, stems, and seeds of several plants (Table 1).
For plants, the KNApSAcK database, a comprehensive species–metabolite relationship database (Afendi et al., 2011), contains reports for hundreds of plant species containing CQAs. Among the most prevalent, we can find 3-CQA, 4-CQA, and 5-CQA, which have been reported in 169, 54, and 15 plant species, respectively (updated in March 2021).
The Phenol-Explorer database, a comprehensive database on polyphenol content in foods (Neveu et al., 2010), contains 1348 determinations for mono- and diCQAs in common foods (updated in March 2021). 3-CQA was reported for 899 foods, at concentrations ranging from 0.02 mg/100 g fresh weight (FW) in olive oil to 454.48 mg/ 100 g FW in sunflower seed; 4-CQA was reported in 128 measurements, at concentrations ranging from 0.1 mg/ 100 g FW (in blackberry [Rubus plicatus]) to 31.25 mg/ 100 g FW (in plums [Prunus domestica]); and 5-CQA was reported in 282 measurements, at concentrations varying from 0.07 mg/100 g FW (in carrot [Daucus carota subsp. sativus]) to 118.59 mg/100 g FW (in plums). Similarly, diCQAs were listed for 56 measurements, at concentrations ranging from 0.44 mg of 4,5-diCQA/100 g FW (in quince [Cydonia oblona]) to 18.56 mg/100 ml of coffee beverage.
TriCQAs occur mainly in the Asteraceae family. For example, the triCQAs, 1,3,5-triCQA, 1,3,4-triCQA, 3,4,5-triCQA, and 1,4,5-triCQAs have been found in Duhaldea nervosa – a plant used for traumatic injury in Chinese traditional medicine (Liu et al., 2018). The same four triCQAs were found in Anthemis tinctoria – a plant with broad use for pharmaceutics, extensively used in Europe for its antispasmodic, anti-inflammatory, and antibacterial activities (Orlando et al., 2019). 1,3,4-TriCQA was also reported in Chrysanthemum flower, another traditional Chinese medicine commonly used for dissipating cold (Zhang et al., 2018). Recently, 1,3,4-triCQA and 3,4,5-triCQA were reported in propolis, a beehive product used medicinally against infections and as an anti-inflammatory agent (do Nascimento Araujo et al., 2020). Amongst all CQAs included in this review, 1,3,4,5-tetraCQA is the least reported of all. However, it was recently reported in aged tea leaves of C. sinensis (Liu et al., 2020c).
It is important to mention that in the same plant species, the concentration of CQAs can be significantly affected by the plant’s ontogenetic stage and genetics, biotic or abiotic stress, geography, storage conditions, the fermentation process, and the part of the plant used, among other factors. Table 1 shows the 15 CQAs naturally present in plants, resulting from the different degrees of esterification of quinic acid with caffeic acid, along with citations of recent studies (2015–2021) involving the CQAs.

DETECTION AND ANALYTICAL CHALLENGES
Several qualitative and quantitative analytical methods for CQAs are available in the literature with detection limits in the nanomolar range. A comprehensive review was recently provided by Wianowska and Gil (2019). Current methods for evaluating the occurrence of phenolic compounds, including CQAs, commonly comprise extraction and enrichment steps followed by different analytical platforms such as liquid chromatography with diode array detection (LC-DAD) (Baeza et al., 2016; Tanriseven et al., 2020), MALDI time-of-flight (TOF) MS (Liu et al., 2012), LCMS (Magana et al., 2015; Zheleva-Dimitrova et al., 2017), gas chromatography-MS (Keser et al., 2020), and nuclear magnetic resonance (NMR) (Cho et al., 2016). Among them, hyphenated techniques are the most robust. Ultrahigh-performance LC (UPLC) coupled to a triple quadrupole mass spectrometer is the golden standard for targeted analysis (Olennikov et al., 2019). For untargeted chemical fingerprinting analysis, high-resolution accurate mass spectrometry instruments equipped with TOF analyzer (Garcia et al., 2018) or Orbitrap analyzers (Zhan et al., 2020) are advantageous.
Despite the numerous analytical methods available to study CQAs (Liu, Li, Zu, et al., 2020), there are still several challenges associated with the consistent characterization of individual compounds. One of the main challenges is that several naturally occurring compounds including CQAs readily degrade and transform into other compounds. Quinic acid ((1S,3R,4S,5R)-1,3,4,5-tetrahydroxycyclohexane-1-carboxylic acid) possesses four chiral carbons (shown in bold). Consequently, eight stereoisomers are possible as a product of extraction degradation and thermal processing (Deshpande et al., 2016). Conversion due to biotransformation by microbiota needs yet to be explored. The eight possible stereoisomers, together with the different degree of esterification presented in Table 1, can potentially yield more than 100 CQAs in processed biological samples. Another analytical challenge results from interference between CQA analytes and components of the biological matrix interferences. When analyzing CQAs, plant extracts and food comprise relatively simple matrices. However, upon ingestion, some of these compounds can reach the plasma or other biofluids and tissues. Biofluids and tissues increase the complexity of the analysis, where matrix interferences must be addressed, and analysis methods require extensive validation (Tsikas, 2018).
Typical HPLC or UPLC analytical columns, such as C18, C8, amide, Phenyl, Cyano, and Pentafluoro-Phenyl, lack the selectivity required to resolve stereoisomers that share similar chemical properties. Consequently, several reports in the literature may overestimate individual concentrations. A potential solution for this drawback is to take advantage of different mass spectral fragmentation patterns for isomers, yielding different relative intensities across fragment ions and different fragment ions in some cases (Gray et al., 2018). Figure 3 shows typical MS/MS fragment ion spectra obtained for CQAs using a quadruple TOF mass spectrometer. Figure 3a shows typical product ions for a diCQA (3,4-diCQA) using a collision energy of 35 V. The fragment ions with m/z 135.04, 155.03, 161.02, 173.04, 179.03, and 191.05 (Figure 3b) will be present in the negative ion mode for all compounds listed in Table 1. However, those fragment ions will have different abundances for each unique CQA. In addition to these fragment ions, for triCQAs, the fragment with m/z 515.11 will be present after the loss of one caffeic acid moiety. Similarly, tetraCQA will yield all previous fragment ions along with a fragment at m/z 677.14, due to the loss of one of the four caffeic acid moieties. Taking advantage of the fragmentation patterns and the difference in relative abundance, MS/MS data can be obtained by multiple reaction monitoring (MRM) in triple quadruple analyzers or by parallel reaction monitoring in TOF or Orbitrap instruments to obtain qualitative and quantitative information (Perez de Souza et al., 2017; Tada et al., 2020).
In our laboratory, we have developed an analytical method for chemical profiling of botanical samples that uses a phenyl-bonded phase. This stationary phase takes advantage of the presence of phenolic scaffolds in many phytochemicals (Alcazar Magana et al., 2020). In such case, chromatographic separations are governed by π–π interactions, and CQAs isomers, such as 1,3-diCQA, 3,4-diCQA, 1,5-diCQA, 3,5-diCQA, and 4,5-diCQA, can be resolved, and clean MS/MS spectra can be acquired (Figure 3c).
As mentioned before, CQAs are prone to photolytic, pH, and thermal degradation. In our experience, stock solutions of 1 mg ml−1 in methanol 70% containing 0.1% v/v formic acid are stable for 6 months if kept in the dark at −20°C (Alcazar Magana et al., 2020). Due to the confusion in the nomenclature of CQAs, we strongly recommend to verify the structure of authentic analytical standards provided by suppliers; use of CAS numbers and confirmation of standards by NMR is advised.

BIOACTIVITY
The relationship between bioactivity and bioavailability of CQAs is complex and depends largely on gut microbial conversion into metabolites with their own pharmacological profiles. For a well-studied CQA, chlorogenic acid, only 0.8% of the orally administered dose in rats was recovered in the 24-h urine (Gonthier et al., 2003). The urinary recovery of the hydrolysis product, caffeic acid, and its methylated metabolites, ferulic and isoferulic acid, did not exceed 0.5% of the administered dose. As with many other dietary polyphenols (Stevens and Maier, 2016), caffeic acid hydrolytically released from chlorogenic acid undergoes extensive microbial metabolism to form m-coumaric acid (de-hydroxylation), 3-phenylpropionic acid (de-hydroxylation and hydrogenation), benzoic acid (either derived from caffeic acid by β-oxidation or quinic acid), and hippuric acid (the glycine amide conjugate of benzoic acid). The urinary recovery of these microbial metabolites exceeded 50% of the administered dose (Gonthier et al., 2003). The relatively high concentration of these phenolic acids in the urine may reflect their intestinal absorption by monocarboxylate transporters (Ziegler et al., 2016). Therefore, any bioactivity from orally administered chlorogenic acid, and by extension other CQAs, is likely to be caused by their gut microbial metabolites (Clifford et al., 2020).
CQAs have been found to be particularly widespread in medicinal plants, and this may not be a coincidence as they possess potentially disease-modifying properties. CQAs are electron-rich compounds containing phenolic hydroxyl groups in a conjugated system with an even number of carbons. Therefore, CQAs can be oxidized in vivo to quinoids (active Michael acceptors). Consequently, CQAs can play roles as direct antioxidants (scavenging free radicals) or act as Michael acceptors and target the Keap1-Nrf2 pathway. Nrf2 is a central regulator of cellular resistance to oxidative stress and a therapeutic target in aging-related diseases (Brandes and Gray, 2020; Liang et al., 2019). Liang et al. (2019) have demonstrated that diCQAs have a great capacity of scavenging radicals in in vitro assays. They also showed in Caco-2 cells that Nrf2 activation capacity was CQA isomer-specific (Liang et al., 2019). Isomer-specific antioxidant capacity is a critical factor to consider during plant extract preparations. The lack of standardization practices can lead to different isomeric ratios in plant extracts, with the risk of obtaining variable results in activity tests when using different batches or plant accessions.
Transcriptional activation of the anti-inflammatory Keap1/Nrf2 pathway has been reported for orally administered chlorogenic acid in Wistar rats challenged with thioacetamide to induce acute liver inflammation and toxicity (Chen et al., 2018; Hussein et al., 2021). Considering the extensive metabolism of CQAs, it is notable that circulating metabolites containing 1,2-dihydroxyphenol moieties can also be oxidized to electrophilic o-quinones which form Michael-type adducts with cysteine thiols of Keap1 (Stevens et al., 2018). The covalent modification of Keap1 leads to inhibition of Nrf2 ubiquitination in the cytoplasm, enhancement of Nrf2 levels in the nucleus, and transcriptional activation of the antioxidant response element (ARE) genes. This chain of events leads to ARE-dependent expression of antioxidant molecules, such as glutathione.
CQAs have been shown to exert neuroprotective effects in several in vitro models (Gray et al., 2018). Mono- and diCQAs protect rodent primary neurons against glutamate excitotoxicity (Kim, Bolton, et al., 2012; Mikami and Yamazawa, 2015) potentially by normalizing calcium homeostasis. 1,5-O-dicaffeoyl-3-O-[4-malic acid methyl ester]-quinic acid protects SH-SY5Y neuroblastoma cells against N-methyl-D-aspartic acid (NMDA) toxicity, inhibiting NMDA-induced phosphorylation of ERK1/2, p38 mitogen-activated protein kinase (MAPK), and JNK1/2m and restoring the NMDA-impaired activation of CREB, AKT, and GSK3B (Tian et al., 2015). Both 1,5- and 3,5-diCQA protect both MC65 and SH-SY5Y cells from beta-amyloid (Aβ)-induced cell death. This was associated with increased expression of Nrf2 and its target genes for both compounds. 3,5-diCQA also increased mitochondrial respiration and expression of mitochondrial genes encoding enzymes in the electron transport chain (Gray et al., 2015).
In vivo, 3,5-diCQA improves learning and memory deficits in the SAMP8 mouse model of aging, which was accompanied by an increased expression of glucose metabolism enzymes (Han et al., 2010; Sasaki et al., 2013), while a mixture of CQAs improved cognition in the 5xFAD mouse model of Aβ accumulation (Matthews et al., 2020). Chlorogenic acid and 3-O-CQA improved memory in scopolamine- and Aβ1–42-treated mice, respectively. Both compounds decreased lipid peroxidation and chlorogenic acid treatment also inhibited acetylcholinesterase activity in the brains of treated animals (Choi et al., 2014; Kwon et al., 2010).
Metabolites of CQAs also exert cognitive effects. Caffeic acid reduces infarct volume, attenuates neurological deficits, and improves working, spatial, and long-term aversive memory deficits in a mouse model of stroke (Pinheiro Fernandes et al., 2014). In a rat model of global cerebral ischemia/reperfusion injury, caffeic acid reduces memory deficits and reduces hippocampal cell injury and death. Both anti-inflammatory and antioxidant effects were observed, including reductions in 5-lipoxygenase expression, pro-inflammatory markers, and lipid peroxidation and increased antioxidant enzyme activity (Liang et al., 2015). Caffeic acid treatment improves memory in the Tg2576 mouse model of Aβ accumulation as well as in Aβ25–35-injected mice and Aβ1–40-injected rats, while reducing markers of inflammation, oxidative damage, and p38 MAPK activity and increasing synaptophysin expression (Kim, Lee, et al., 2012; Sasaki et al., 2013; Subash et al., 2015).
Other metabolites of CQAs exert anti-inflammatory effects. In Caco-2 cells challenged with TNF-α – which represents a model of the inflamed gut – treatment with benzoic acid or 3-phenylpropionic acid results in reduced production of the pro-inflammatory cytokine interleukin-8 (IL-8), while the anti-inflammatory effect of these chlorogenic acid metabolites was synergistic when used in combination (Zheng et al., 2021). As gut inflammation can lead to systemic inflammation due to gut inflammation related loss of epithelial barrier function, a local effect of the metabolites of CQAs can exert a systemic anti-inflammatory effect without absorption. Conversely, the higher absorption of CQA metabolites, compared to the parent polyphenol, could be expected to account for anti-inflammatory effects in organ tissues, by the same inhibition of the pro-inflammatory NF-κB pathway that releases IL-8 and other cytokines from macrophages (Monagas et al., 2009). In addition, 3-phenylpropionic acid is a known inhibitor of cyclo-oxygenase 2 (COX-2), an enzyme that forms pro-inflammatory eicosanoids and that is widely expressed in peripheral tissues as well as in the brain. Inhibition of COX-2 in the brain by phenylpropionic acid could confer neuroprotection of the brain under chronic inflammatory stress, as is observed in Alzheimer’s disease. Other possible mechanisms of neuroprotection by phenylpropionic acid include suppression of the NF-κB and MAPK signaling pathways (Karlsson et al., 2005). While dietary CQAs can be observed in the micromolar range in the human body, glutathione and ascorbic acid are present at concentrations in the millimolar range in the cytosol. Consequently, CQAs may exert most of their bioactivity as indirect antioxidants and as pro-oxidants. As in vivo pro-oxidants and Nrf2 activators, CQAs may promote GSH synthesis, protecting cells from ROS (Stevens et al., 2018).
Chlorogenic acid (3-CQA) has been extensively studied as a nutraceutical for treating metabolic syndrome and associated disorders, such as diabetes, obesity, dyslipidemia, and hypertension, with some positive results. For instance, it has been suggested that 3-CQA reduces the risk of developing type 2 diabetes by activation of adenosine monophosphate kinase (AMPK), in which AMPK acts as a switch between ATP expenditure and ATP production (Santana-Galvez et al., 2017). However, most clinical studies used coffee as a source of 3-CQA, which limited the scope and interpretation of the results (Santana-Galvez et al., 2017; Santos and Lima, 2016).
In vitro studies are difficult to translate to in vivo bioactivity due to physicochemical limitations such as solubility, which may influence bioavailability. Additionally, liver and gut transformation produces a set of modifications in the original compounds, including sulfation, glucuronidation, methylation, and hydroxylation (Stevens et al., 2018). The in vitro transformation of some CQAs by gut microbiota has been investigated (Mills et al., 2015; Tomas-Barberan et al., 2014). Yet, the in vivo transformation of CQAs by gut microbiota is currently unclear (Clifford et al., 2017, 2020). Nonetheless, in our research group, we found some agreement between in vivo and in vitro studies. We studied the neuroprotective properties of CQAs in C. asiatica. CQAs were found to reverse cognitive decline in 5XFAD mice, a model of Alzheimer’s disease (Matthews et al., 2020); these in vivo findings are in agreement with in vitro studies where neuroprotective effects of CQAs against Aβ toxicity were observed (Gray et al., 2014). 3,5-DiCQA and 1,5-diCQA were found to be the most active in protecting MC65 cells from Aβ-induced cell death. Both compounds showed neuroprotective activity in MC65 and SH-SY5Y cells. In addition to mitigating Ab-induced cell death, diCQAs from C. asiatica were able to attenuate Ab-induced alterations in tau expression and phosphorylation in MC65 and SH-SY5Y cells. These data suggest that CQAs are associated with the neuroprotective activity observed for C. asiatica. Additionally, CQAs showed beneficial effects on spine density and dendritic arborization in primary neurons (Gray et al., 2017). In line with these findings, there are other reports supporting the role of CQAs in cognition and memory improvement (Mikami and Yamazawa, 2015; Sasaki et al., 2019; Tian et al., 2015).
Sasaki et al. (2019) reported that 3,4,5-triCQA stimulates adult neurogenesis in a senescence-accelerated aging mouse model through activation of the bone morphogenetic protein signaling pathway. Furthermore, 3,4,5-triCQA was found to improve the deficit of learning and memory by increasing neurogenesis of the hippocampal dentate gyrus (Sasaki et al., 2019). These discoveries open up opportunities to synthesize analogs of triCQAs for therapy of neurodegenerative disorders. It is important to emphasize that the ubiquitous presence of CQAs in plants complicates assessment of the bioactivity in preclinical studies. Plant-based laboratory animal diets can compromise a CQA-free baseline for the control groups. In such a case, a synthetic diet is advised.
Polyphenols, including CQAs, are also known for their antiviral activity. Catechol groups in CQAs may inhibit severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) papain-like cysteine protease (PLpro) by interacting with the PLpro thiol nucleophile (Ospanov et al., 2020; Park et al., 2012). Smith et al. (2016) tested small-molecule inhibitors of TbCet1 – a RNA triphosphatase – essential for the development of viruses. Among 2879 compounds, 1,3,4,5-tetraCQA was the most potent inhibitor (IC50, 13 nM). By contrast, 3,5-diCQA (with only two catechol groups), while still significantly potent (IC50, 70 nM), was about five times less effective than tetraCQA. Recently, Paraiso et al. (2020) suggested that viral proteases may be inhibited by polyphenols via hydrogen bonding. At the chemical level, CQAs have the possibility to inhibit viral proteases, opening the interest for in vitro and in vivo studies to evaluate whether CQAs can reduce viral infections, such as COVID-19, caused by SARS-CoV-2.

CONCLUSIONS AND PERSPECTIVES
Interest in the study of CQAs is growing dramatically due to their potential activity in ameliorating cognitive decline and lifestyle-related diseases, as well as their emerging use as antivirals with huge potential for drug development. Although CQAs are widely known as antioxidants, their physiological levels reached after ingestion are well below the concentrations of other more potent antioxidants, such as glutathione and ascorbic acid. Therefore, it is expected that their bioactivity is due to a pro-oxidant effect or indirect mechanisms such as Nrf2 regulation; nonetheless, information on these mechanisms is limited. Furthermore, in vivo transformation of CQAs by gut microbiota is an open and important field for future research.
The discovery of new enzymes and biosynthetic pathways will pave the road to increase CQA production and support applications of CQAs. Foreseeable applications of CQAs as nutraceuticals or phytotherapeutic agents require a comprehensive qualitative and quantitative characterization of these compounds with a unified, unambiguous nomenclature, including the absolute configuration (i.e., IUPAC names). Despite the lack of an official method for the analysis of CQAs, hyphened analytical techniques, such as LC-MS detection, are among the most reliable options. Finally, the growing body of preclinical literature supporting health-improving and disease-modifying properties of CQAs must be validated by randomized, double-blind, placebo-controlled clinical trials.