The selection of the optimal moment of grape ripeness may be considered as the most crucial decision in winemaking. Optimal ripeness will vary depending on the style of wine that is being produced (sparkling, still, rosé, etc.) (González-Caballero et al., 2012). In viticulture, ripeness constitutes the optimal characteristic of wine grapes to be harvested. Once the grapes are harvested, the physical and chemical components of the grape that will influence wine quality are essentially set.

As the grapes go through veraison, sugars in the grapes will continue to rise as acid levels fall. The balance between sugar (and so the potential alcohol level) and acids is considered one of the most critical aspects of producing quality wine. Thus, both the relative density and “total acidity” of the must, as well as the pH of the grapes, are evaluated to determine ripeness.

Towards the end of the 20th century, winemakers began focusing on the concept of achieving “phenolic” ripeness in the grapes. This is described as a more complete ripeness of tannins and other polyphenolic compounds like anthocyanins (coloured compounds). The polyphenolic composition of grape skin, seed and pulp changes during ripening and consequently influences red wine astringency.

There is still a gap in the real influence of ripening on astringency perception because grape tannin composition is affected not only by extrinsic factors like agronomical conditions or grape development stage but also by different intrinsic factors like genotype. Furthermore, the mean degree of polymerization (mDP) decreases after veraison, but it was recently described that the differences observed in tannin mDP could be overestimated by the analytical method (Meléndez et al., 2013). In general, tannin accumulation in grape seed and skin occurs early in berry development; it is completed when ripeness starts and, after that, tannin content remains unchanged or decreases after veraison (Quijada-Morín et al., 2016; Bogs et al., 2007).

Besides total tannin composition, it is important to elucidate if polyphenol extractability also changes during ripening. The extractability of polyphenols is defined as the percentage of total polyphenols extracted during winemaking. This determines the polyphenolic content of wines and is a key element in astringency perception. The extractability of skin and seed tannins during berry development is also different between varieties: tannins become easier to extract during the ripening of Tempranillo berries (Meléndez et al., 2013) but harder during the ripening of Monastrell berries (Fournand et al., 2006). In Sangiovese grapes, total tannin content decreases after veraison but extractable tannins remain unchanged (Allegro et al., 2016). Tannin structure also affects their extractability. Allegro et al. (2016) showed that while extractable (epi)catechin decreases after veraison, extractable epigallocatechin and epicatechin gallate increases along ripeness.

Anthocyanin behaviour is opposite to the one observed for tannins. Anthocyanin accumulation in skins starts at veraison and is maximal around harvest (Fournand et al., 2006; Mateus et al., 2002).. Some studies describe a decrease in total anthocyanins just before harvest and/or during over-ripening. In Tempranillo grapes, anthocyanin extractability peaks increase in overripe berries even though the total anthocyanin content has already declined (Fournand et al., 2006). In contrast, no changes in anthocyanin extractability were observed from veraison to harvest in Shiraz berries (Del Llaudy et al., 2008). So, if the polyphenolic composition evolves at ripening, the astringency will also depend on the ripening moment.

As abovementioned, the balance between sugar and acids is considered critical, but the evaluation of the sugar content and acid profile alone does not fully express the real oenological potential of grapes. Indeed, a prediction of the possible astringency intensity could only be estimated by polyphenolic composition. The knowledge of the polyphenolic characteristics of the grapes allows the maceration and winemaking process to be planned in order to fully exploit the potentiality that the grape reaches in the vineyard. Thus, the term “phenolic ripeness” was introduced to describe the concentration of polyphenolic compounds in grapes and the ease with which they are released (Cagnasso et al., 2008).

However, recent researches suggested that astringency perception could be also related to other wine compounds, namely carbohydrates (Wratelot et al., 2017; Boulet et al., 2016). Choosing the optimal ripening date to control wine astringency affects not only phenolic ripeness but also other compounds, namely carbohydrates. However, keeping in mind that there is no balance between phenolic and sugar (carbohydrates) contents during ripeness, the harvest date is difficult to choose.

Presently it could be anticipated that polysaccharides influence astringency through a direct mechanism in which cell wall polysaccharides bind to tannins reducing their extractability or an indirect way through a competition process at a sensory level. In fact, the previously referred differences in anthocyanin extractability among varieties could be correlated with the polysaccharide content of the skin. With regard to different grape varieties, there is no consistent trend in polysaccharide composition associated with ripeness. Polysaccharide composition varies among grape varieties but the individual influence of each polysaccharide is not clear. Some authors have stated that there is no link between polysaccharide composition and tannin binding capacity of cell walls (Hanlin et al., 2010), whereas other studies revealed a different effect in tannin–cell wall binding depending on the polysaccharide tested (Le Bourvellec et al., 2005). However, in general, several polysaccharide families have been described to be able to interact with tannins (Le Bourvellec et al., 2005). Extractability increases when there are low concentrations of pectin, cellulose, arabinoxylan, arabinogalactan and xyloglucan. Polysaccharides would therefore limit the concentration of available proanthocyanidins and, thus, astringency would be reduced.

At a sensory level, it is known that polysaccharides could inhibit the interaction of red wine tannins with proteins, as it has been previously proposed for the loss of astringency in ripening fruits (Ozawa et al., 1987) Several in vitro studies have shown that this inhibition could occur by two different mechanisms: (i) polysaccharides could interfere with tannins' ability to bind proteins through encapsulation or (ii) some polysaccharides could inhibit the precipitation process by the formation of ternary complexes (aqueous solubility is enhanced) (Carvalho et al., 2006).

Several structural and chemical changes occur during ripening, leading to the softening of the cellular structure of many fruits. These changes affect especially hemicelluloses, pectins and cellulose microfibrils of the cell wall (Fasoli et al., 2016). Pectin modification is mainly responsible for the progressive loss of firmness during ripening. Fruit softening allows the release of small pectin soluble fragments of the cellular structure during ripening. Furthermore, the xyloglucan content of hemicellulose decreases significantly during berry growth and ripening, whereas cellulose levels are only marginally affected (Fasoli et al., 2016).

Sensory analysis has been also used to study the influence of polysaccharides on astringency perception in model wine solutions, showing that all polysaccharide families reduce the perception of astringency to some degree. These studies revealed that acidic polysaccharides have a greater impact on the reduction of astringency perception. Neutral polysaccharides also tend to decrease the intensity of astringency attributes. Nevertheless, in that study, the differences between model wine and the fraction containing a mixture of mannoproteins and type II arabinogalactan proteins isolated from wine were not statistically significant (Quijada-Morín et al., 2014).

Overall, the effect of ripening time on astringency perception is quite complex. In general, several oenological techniques have been used to enhance the composition deficiencies derived from a trade-off harvest data. Harvesting grapes at different times has been also used to improve wine sensory properties.

Grape seeds and skin tannins are transferred to must/wine during the maceration step of winemaking. The maceration process could be induced by using hydrolytic enzymes, mainly against pectins, cellulose and hemicellulose. Hydrolytic enzymes are supposed to break down the interaction between proanthocyanidins and cell walls. The effective extraction of proanthocyanidins from grapes will depend on having enough knowledge of the nature of these interactions and on the ability to disrupt or manage these associations. Hence, the higher tannin extraction efficiency, the higher tannin content in wines, which, in turn, will participate in the astringency process. As regards the nature of these associations, it has been found that electrostatic or ionic interactions do not seem to play any important role in the association between proanthocyanidins and cell-wall material, rather the adsorption mechanism seems to involve the establishment of weak interactions, more precisely hydrogen bonds and hydrophobic interactions (Castro-López et al., 2016). Enzymes exert their effect by deconstructing the polysaccharide network of the cell wall and by allowing the extraction of the phenolic compounds located inside the vacuoles (Castro-López et al., 2016; Bautista-Ortín et al., 2016 a). The use of maceration enzyme results in an increase in the proanthocyanidin content of must and wine, not only by favouring the extraction of these compounds from skin cell vacuoles but also by promoting a lower adsorption on cell walls (Castro-López et al., 2016; Bautista-Ortín et al., 2016 a). It was therefore concluded that the use of enzymes during maceration could promote higher astringent wines. Therefore, it is very important to acquire a deeper knowledge of the nature of cell walls–tannins interactions and the role played by the main enzymatic activities found in the commercial enzymes. This would allow a better understanding of the ability of these enzymes to disrupt or manage these associations, as well as the effect of the polysaccharides released after the action of these enzymes on the quantities of proanthocyanidins in solution.

Anthocyanin content is also affected by maceration. It has been demonstrated that knowing the quantities of anthocyanins in grape skins is not enough for estimating wine anthocyanin concentrations. This lack of correlation has been commonly attributed to the partial retention of these anthocyanins in the skin cells due to the barrier effect of the cell walls (Bautista-Ortín et al., 2016b). Anthocyanins were also described to affect tannin extractability (Bautista-Ortín et al., 2016b), so they could have an indirect influence on wine astringency. Recent studies suggested that the incorporation of anthocyanins in polymeric procyanidins seems to increase the amount of tannin retained in wine, thus explaining the larger quantities of tannins found in red wines and their higher astringency extractability (Bautista-Ortín et al., 2016b). Two mechanisms were proposed: (i) the higher solubility of polymers formed from anthocyanins and condensed tannins or (ii) the competition between anthocyanins and tannins by cell-wall binding sites.

During winemaking, tannin extractability could considerably vary even under identical fermentation conditions. Furthermore, alcohol and surfactants in food or beverages may increase the solubility of the hydrophobic phenol–protein complexes, decreasing astringent sensations that may depend upon precipitation of salivary proteins (Lee and Lawless, 1991). The ethanol formed during fermentation not only modifies the solubility of must/wine compounds (mainly tannins), but is also directly related to astringency. As the ethanol level increases in model solution, a decrease in perceived astringency and in some astringent sub-qualities was observed (Vidal et al., 2004). In case of red wine, the astringency sensation was reduced by ethanol (Meillon et al., 2009). In fact, tannin–salivary protein interaction, which is the main mechanism in astringency development, depends on solvent composition. In the wine matrix, overall astringency usually decreases with increasing ethanol concentrations (from 0 to 15% ethanol). Moreover, ethanol can disrupt tannin–cell wall binding of apple pulp at concentrations between 20 and 40%, pointing out the influence of ethanol on noncovalent tannin–carbohydrate binding (McRae et al., 2015.

Tannin extractability in wine fermentation was also related to the presence of grape-derived proteins (Springer et al., 2016). Removing proteins through prefermentation bentonite fining can promote two parallel events: (i) reduction of grape-derived protein and (ii) increase of wine tannin concentration. Sensory changes, mainly astringency changes, were observed as a consequence of the increase of phenolic concentration before fermentation.

Wine fermentation is usually promoted by inoculation with specific strains. Saccharomyces cerevisi a e was the most common yeast used in controlled fermentation. Over the last few years the use of multi-starter inocula has become an attractive biotechnological practice in the search for wine with high flavour complexity or distinctive characters (Lencioni et al., 2016. The use of Saccharomyces and non-Saccharomyces yeasts influences the ethanol and polysaccharide content in wine. Thus, mixed fermentations with S. cerevisiae and a selected strain of Z. florentina (formerly Zygosaccharomyces florentinus) caused an increase in the production of polysaccharides (Domizio et al., 2011). Non-Saccharomyces yeasts have shown potential for producing wines with lower ethanol content. These yeast species, whose main sources are vineyard and grape surface, are present in the early stages of fermentation but, in general, are not capable of completing alcoholic fermentation. Sequential inoculations with Saccharomyces and non-Saccharomyces at different fermentation steps were proposed to obtain low ethanol concentrations in wines (Contreras et al., 2014).

Given the influence of both ethanol and carbohydrates on astringency, it could be ascertained that the yeast used could also affect wine astringency. Hence, several studies proposed the astringency loose perception in wines obtained after mixed fermentations (Lencioni et al., 2016). Despite the knowledge of the differences in carbohydrate content, ethanol levels and astringency sensation between using single or multi-starter inocula, there is still a gap in our knowledge of how the compositional changes could affect the overall astringency.

Grape skin is also a source of polysaccharides in wine. These compounds are localized in a highly soluble form within the grape intracellular spaces and are extracted into wine during maceration and fermentation. However, polysaccharides could also derive from yeast cell wall like mannoprotein polysaccharides (Pérez-Través et al., 2016). Mannoprotein, a main component of S. cerevisiae cell wall (25-50% dry weight), comprising roughly 70-80% polysaccharide and 20-30% protein, may represent as much as 35% of wine polysaccharides (López-Solís et al., 2017). It was already described that mannoproteins decreased the intensity of some astringent attributes and contributed to the fullness of model wine solutions (Vidal et al., 2004). In this case, different mechanisms were proposed in the decrease of astringency modulated by mannoproteins: interaction with wine phenolics, reducing the levels of free reactive proanthocyanidins, or increasing salivary viscosity (Guadalupe et al., 2010).

As mentioned, mannoprotein polysaccharides are extracted from yeast cell walls either during alcoholic fermentation or later by enzymatic action during wine contact with yeast lees (Pérez-Través et al., 2016). Inactive dry yeasts were therefore used as an oenological treatment in the astringency sensation control. Inactive dry yeast products are mostly derived from cultured S. cerevisiae strains that are subsequently subjected to thermal or enzymatic inactivation. Interaction of an inactive dry yeast product (CP10) with either salivary protein or a proanthocyanidin-rich extract (binary mixtures) or with both of them (ternary mixtures) was verified (López-Solís et al., 2017). The use of commercial dry yeast preparations improves some sensory characteristics of white and red wines, probably due to the increase of neutral polysaccharides (Del Barrio-Galán et al., 2012). In red wines, dry yeast preparations mainly reduced green tannins, increasing the softness on the palate. Therefore, they could be useful in order to improve the roundness and softness in the mouth, especially in young wines that are more astringent (Del Barrio-Galán et al., 2012). It has also been reported that supplementation with commercial inactive yeasts in grape juice during winemaking and oak aging can significantly decrease the proanthocyanidin content of red wines coinciding with a decrease in high molecular weight mannoproteins (Del Barrio-Galán et al., 2016). This suggests that the co-aggregates mannoprotein–tannin precipitate during treatment (González-Royo et al., 2017). According to Gawel et al. (2016), the effect of polysaccharides on mouthfeel was independent of the concentration of phenolic substances. Polysaccharides have a relatively small effect on mouthfeel compared with the effect of wine pH and ethanol. Other authors have confirmed the interaction between mannoproteins and phenolics (Mekoue Nguela et al., 2016). Given this controversy, further studies are required to better establish the influence of these commercial products on sensory attributes, mainly on astringency, according their chemical composition and purity.

Wine stabilization and clearness is obtained after winemaking through the precipitation of unstable compounds and the sedimentation of the clouding particles. Wine clarification is naturally produced but this process is slow and may not be enough for proper clarity and stability of the wine. This process is therefore often improved by using different fining agents that will interact with the wine components, reaching a better clearness in less time and improved stability of the wines.

Fining agents could be protein based and, therefore, could induce some organoleptic changes. Astringency and bitterness of wine can decline due to its interaction with tannins. Colour changes could be also promoted. The fining process directly occurs from the precipitation of proanthocyanidins by these fining agents and it is influenced by the chemical characteristics of the protein used. The interactions between proanthocyanidins and fining agents therefore depend of the molecular weight and aminoacid composition of the proteins used (Maury et al., 2001). Different protein types are used for wine fining. Gelatin, β-lactoglobulin, ovalbumin and casein are the proteins from animal origin most commonly used as fining agents. Different types of gelatin remove different amounts of tannins (9-16%) depending on the wine and gelatin composition (Smith et al., 2015). General observations set that high polymerized tannins as well as high galloylated tannins were preferentially removed (Maury et al., 2001).

Several authors observed that the addition of polysaccharides influenced the clarification process (Maury et al., 2001; Vidal et al., 2004). It was again reflected that other components of the wine matrix, mainly polysaccharides, are also important factors in mouthfeel perception. The effect appears to be dependent on the type of polysaccharide (Cheynier et al., 2006). Polysaccharides influence the fining process since they influence the aggregation between tannins and proteins. The aggregations that can potentially occur may involve proanthocyanidin–proanthocyanidin, proanthocyanidin–protein, proanthocyanidin–polysaccharide and proanthocyanidin–protein–polysaccharide interactions (Scollary et al., 2012).

Plant cell wall materials from grape and apple are also being used as fining agents. Moreover, in oenology, there is another possible technological interest arising from a deeper knowledge of proanthocyanidins–cell wall interactions, that is, using the pomace obtained after maceration as a fining agent to reduce the level of proanthocyanins in wines when they show a high degree of astringency (Guerrero et al., 2013; Bautista-Ortín et al., 2015). Pomace wastes from the winemaking process are widely available and a source of cell wall material. Besides, pomace cell wall material showed a very high degree of affinity for proanthocyanidins. How enzymes may be used to enhance or limit, if necessary, cell wall adsorption properties would be of great interest. Preliminary results suggest that the use of maceration enzymes could affect the retention capacity of the pomace cell walls and the extent to which high or low molecular mass proanthocyanidins are retained, which might have implications for the sensory properties of wine (Bautista-Ortín et al., 2016).

Table

The commercial value of aged wines is mainly correlated with their colour stability and mouthfeel perception. These sensory attributes are closely related to phenolic composition (Chira et al., 2011). An optimum balance between tannins and anthocyanins is essential to colour stability during red wine aging. Among sensory properties, wine aging has long been used not only to allow the development of focused flavours but also to get a mellow astringency (Chira et al., 2012). The main reactions occurring during aging and involving colour stabilization are well known, whereas the impact of reactions involving proanthocyanidins and anthocyanins with regard to the decrease of wine astringency is still not well established. Greater astringency intensity is directly associated with higher concentration of tannins with a higher mDP (Soares et al., 2007). During aging, astringency perception becomes mellower, so the reasons for the change in wine astringency could involve a decrease in tannin concentration accompanied by a decrease in tannin size because of subunit cleavage (Chira et al., 2012; Bindon and Smith, 2014).

The aging process could be induced by several ways that could influence wine composition and properties in a different way. Wine could evolve in bottle or in barrels of different age and different wood origin. The cost of maintaining barrels and long-term storage of wine has prompted much research into new technologies to accelerate the aging process. These cheaper alternatives to barrels were the use of oak staves and aging on wine lees. Traditionally, red wines undergo barrel aging in French or American oak prior to bottling. Different favourable flavour compounds as well as tannin changes are induced during wood contact. Oak tannins are extracted from wood to wine, and barrel porous oxygenation also induces tannin changes. Ellagitannins are the most common tannins extracted from oak. This extraction could be oenologically driven by using staves. Extraction of phenolic compounds from oak staves is low, with a concentration up to 250 mg/L compared to around 2000-4000 mg/L of tannin in red wines regardless of oak treatment (Smith et al., 2015). Instead of these compositional changes, a decrease in astringency during aging is supposed to be promoted to a higher extent by oxygenation.

Given the high reactivity of phenolics towards oxygen, many of the complex reactions involving phenolics are affected by oxygen exposure of wine. Thus, a moderate uptake of oxygen during aging can modulate specific reactions that further influence sensory properties, mainly astringency. Different methods of oxygen exposure were tried. Micro-oxygenation and nano-oxygenation were recommended instead of barrel oxygenation. Highly reactive quinones are formed after phenolic oxygen exposure. Tannin polymerization was further developed by direct and indirect condensation reactions between tannins and flavan-3-ols or anthocyanins (Sánchez-Ilárduya et al., 2012). These reactions were favoured with wine aging as a result of low wine pH. The chemical bonds between flavan-3-ol subunits are continually broken and reformed, which also changes the tannin structure in such a way that it reduces water solubility and induces their precipitation in the bottle. The level of precipitation can possibly be mediated by interactions with polysaccharides (Guadalupe and Ayestarán, 2007).

Thus, the combined effects of the structural changes that occur in tannins during wine aging contribute to the softening of astringency during red wine aging.

This thorough literature review focused on the major red wine oenological steps and how each step influences wine compounds, specially polyphenols, carbohydrates and ethanol (Figure 2). In conclusion, the modulation of each step and the combined effect of different oenological practices modulate wine astringency contributing to a final full bodied, flavourful and balanced wine.

Acknowledgments : The authors would like to thank the Fundação para a Ciência e Tecnologia for financial support through two fellowships (SFRH/BPD/88866/2012 and SFRH/BPD/85293/2012) and project PTDC/AGR-TEC/6547/2014. The authors would also like to thank the LAQV (UID/QUI/50006/2013 - POCI/01/0145/FEDER/007265) for financial support from FCT/MEC through national funds and co-financed by FEDER under Partnership Agreement PT2020.