PROANTHOCYANIDIN COMPOSITION AND EVOLUTION DURING GRAPE RIPENING AS AFFECTED BY VARIETY : NEBBIOLO AND BARBERA CV

Methods and results : PAs were extracted separately from seeds and skins, before véraison until maturity. Intact extracts were fractionated according to PA degree of polymerization. Nebbiolo grapes presented a significantly higher PA content in both seeds and skins. PA depolymerization reactions as well as a weaker decrease in polymeric PAs occurred in Barbera skins. Oligomeric PAs prevailed over polymeric ones in Barbera seeds. At harvest, Nebbiolo skins presented a significantly higher percentage of oligomeric PAs, while monomeric and oligomeric flavanol percentages were significantly higher in Barbera seeds.


INTRODUCTION
Grape tannins represent a significant organoleptic component of wines and are an important contributor to wine quality.
The term tannin, in wine, describes proanthocyanidins (PAs) which are oligomers and polymers of flavan-3-ol units (Downey et al., 2003).PAs are present mainly in skins (Souquet et al., 1996) and seeds (Prieur et al., 1994) of grape berries.Their accumulation occurs early in berry development and is completed when ripening starts (Bogs et al., 2005).Flavan-3-ol monomers and PAs share the same biosynthetic pathway with anthocyanidins, which are red grape pigments, and with flavonols, which contribute to both bitter taste and color stability of red wine (Baranac et al., 1997).
In grape berries, the phenolic and PA content depends on climatic and geographical factors, cultural practices, ripening stage, plant vigor and, consequently, on vine cultivar, clone and rootstock (Pérez Magariño and González-Sanjosé, 2002).
As anthocyanin, flavonol and hydrocinnamate profiles, PA composition could be used for the characterization and thus the classification of grape cultivars (Ferrandino et al., 2012).Grape variety determines the polyphenolic profile in different cultivars.In fact, it is known that the patterns of some classes of flavonoids, such as anthocyanins, are under strict genetic control and that their distribution varies considerably among different grape cultivars (Mattivi et al., 2006).Their profiles for each variety are relatively stable, while absolute concentrations can vary between different vintages.
Moreover, the diversity in wine grape PA composition leads to the employment of different oenological techniques (Kennedy and Jones, 2001 ;Mattivi et al., 2003).As a consequence, the PA variation in grape skins and seeds, the rate of their decrease during fruit ripening and their extractability have to be considered when grapes at different maturity stages are used in winemaking (González-Manzano et al., 2006).The mean degree of polymerization of PAs affects their relative bitterness and astringency.Monomers and small oligomers are more bitter than astringent, whereas the opposite is true for larger molecules (Bordiga et al., 2011 ;Kallithraka et al., 1997).Therefore, extracted during red winemaking, PAs are mostly responsible for the astringency, bitterness and structure of wines (Gawel, 1998;Haslam, 1974;Mc Rae and Kennedy, 2011), while they also play an important role in red wine aging due to their high reactivity (polymerization, condensation with anthocyanins and oxidation reactions) (Ricardo-da-Silva et al., 1991).These reactions with anthocyanins may be responsible for the loss of astringency during wine aging (Cheynier et al., 1997;Remy et al., 2000).
Vitis vinifera L. cv.Nebbiolo and Barbera are two cultivars widely planted in northern Italy and particularly in.Barolo DOCG (Denominazione di origine controllata e garantita) and Barbaresco DOCG wines are produced with Nebbiolo grapes and represent two of the most famous wine regions in the world, both situated in.These wines are characterized by a light color and a high level of acids and tannins.In a sense, Nebbiolo can be compared to Pinot Noir, which also produces light-colored, aromatic wines with high acidity.However, Nebbiolo wines tend to be more tannic in their youth, the best examples often requiring decades to reach a balance between the aromatic complexity and the vibrant acids and tannins.They are long-lived wines that improve with age.On the other hand, the Barbera grape variety is widely planted in Piedmont and in other Italian regions as well as overseas.Barbera wines tend to be much more colored, fruitier, riper and less tannic than Nebbiolo wines.The full characterization of skin and seed PAs of those two important Piedmont cultivars is then fundamental to better understand their composition and evolution during fruit ripening and to chose the best time for harvest and modulate their extraction into wine.
Previous studies have focused more on anthocyanin profile characterization and other physico-chemical differences between these two cultivars (Di Stefano et al., 2002).Recently, anthocyanin, flavonol and hydrocinnamate profiles have been studied at maturity and their importance from a taxonomical point of view for the classification of a great number of Italian V. vinifera cultivars, including Nebbiolo and Barbera, was investigated (Ferrandino and Guidoni, 2010;Ferrandino et al., 2012;Mattivi et al., 2006).
While little information is available about Nebbiolo and Barbera PA content and evolution during berry development (Di Stefano et al., 2002), no study involving PA fractionation is mentioned.
The aim of this study is the evaluation of PA potential and composition as affected by grapevine cultivar at both maturity and different stages of berry development.Potential differences between the two cultivars are investigated too.

Sampling and extraction of phenolic compounds from solid parts of the grape
Nebbiolo and Barbera grape berries grown during the 2011 harvest season were collected in Cisterna d'Asti, Piedmont (Italy), every fifteenth day during berry development, from July 15 th (before véraison) until September 28 th (commercial maturity).The second sampling date corresponded to the beginning of véraison.All samples consisted of randomly collected berries gathered with their pedicles, from different sides of the bunches, both shaded and sunlit, within the same vineyard block for each cultivar.
Three replicates of 25 berries were chosen each time from a pool of 300 -400 berries for each cultivar and then weighed.
The seeds and skins were separated from the berry mesocarp, and the intact tissues (i.e., whole seed or skin) were extracted separately.
Preliminary tests have shown that 50 mL of acetone/water (60 :40) were enough to extract the totality of phenolic compounds from grape samples of both cultivars (data not reported).
Briefly, the berry samples were processed as follows: isolated skins and seeds were homogenized (Ariete Magic Choppy 1824) separately with 30 mL of acetone/water (60:40).The extraction was carried out at room temperature for 30 min, then the suspension was centrifuged ( x 15 min) and the pellet washed twice with the same solvent.The final volume of extracts was 50 mL for each sample.The oxidation of phenolic compounds was limited by saturation with nitrogen during the preparation of both skin and seed extracts.In particular, the solvents were saturated with nitrogen and the same gas was used to saturate the headspace during both the homogenization and extraction processes.
Afterwards, acetone was removed under reduced pressure, then the residue was dissolved in a tartaric buffer pH 3.2 ( L -1 tartaric acid, 22 mL NaOH 1 N, 12 % v/v ethanol and 100 mg L -1 Na 2 S 2 O 5 ) and brought up to 30 mL of total volume.

Spectrophotometric determinations
Among the proposed methods to study flavanols, spectrophotometric measurement after acidic hydrolysis or reactivity towards vanillin represent cheap and easy-to-use measurement approaches.Moreover, the chemistry behind those analytical methods provides different information and the two results cannot be equivalent.Acidic catalysis carried out during the analysis of total PAs evaluates oligomeric and polymeric structures.The evaluation of flavanols by the vanillin assay method (V) takes into account both monomeric flavan-3-ols as well as oligomeric and polymeric structures accessible to the electrophilic addition of vanillin (6 and 8 free positions).Even if « the higher the amount in oligomeric and polymeric flavanols, the greater the response in acidic hydrolysis and reactivity to vanillin », the indexes provide different information.The former reaction hydrolyzes flavanols into their monomeric components and measures them.The vanillin assay takes into account the property of flavanols to react with aldehyde.This property is strictly related to the spatial conformation and sterical hindrance of PAs.Thus, on the one hand, the two spectrophotometric methods provide complementary but different information, on the other hand, the V/PAs ratio of intact extracts and of each fraction also provides some information about PA structure during berry development (Di Stefano et al., 2002).

Total proanthocyanidin (PA) determination
Total PAs were determined by spectrophotometry according to Di Stefano (Di Stefano et al., 1989;Di Stefano and Cravero, 1991) using a Beckman D.U. 640 spectrophotometer.Briefly, 0.2 mL of each intact extract was added to 12.3 mL ethanol and 12.5 mL 37 % HCl containing 300 mg L -1 FeSO4*7H 2 O.The absorbance at 532 nm was measured before and after hydrolysis (50 min in a thermostatic bath at ), and the concentration of PAs was determined according to a calibration curve obtained with cyanidin-3-glucoside.The results are expressed as mg per kg of fresh weight.

Flavan-3-ol (V) determination by the vanillin assay (V)
The analysis was carried out according to Di Stefano et al. (1989) and Di Stefano and Cravero (1991).In particular, 0.5 mL of each intact extract previously diluted with methanol was added to 3 mL vanillin (4 % in methanol) and then acidified with 1.5 mL 37 % HCl.The electrophilic reaction of vanillin with flavanols and PAs was followed by measuring the absorbance at 500 nm after 15 min reaction at room temperature.A calibration curve was obtained by known amount of (+) catechin.The results are expressed as mg per kg of fresh weight.

Fractionation of phenolic compounds
The fractionation of Nebbiolo and Barbera grape PAs according to their degree of polymerization was performed as reported by Sun and coworkerset al. (1995;1996) with minor modification.Briefly, 1 mL of each intact extract (skins or seeds) was fractionated on a 500 mg C 18 Sep-Pak (Waters) (previously conditioned with 2 mL methanol and 4 mL deionised water) into three fractions by different organic solvents.Monomeric flavan-3-ols were eluted with 25 mL of diethyl ether, oligomeric PAs with 25 mL of ethyl acetate, and polymeric PAs with methanol, collected directly in a 10 mL flask.
The first two fractions were taken to dryness, then dissolved with methanol and transferred in 10 mL flasks.Sample fractionation and analyses were carried out in triplicate.
Since monomeric flavan-3-ols are not transformed in anthocyanins and thus not measured by the acidic hydrolysis, their content in the diethyl fraction was evaluated using the vanillin assay ; oligomeric and polymeric flavanols in the ethyl acetate and methanol fractions were characterized by PAs and V indexes in order to assess both their total content in flavanols as well as their reactivity towards vanillin.Combined values of V and PAs were used to gain information about flavanol structure.The results are expressed as mg per kg of fresh weight.

Statistics
All data are expressed as the arithmetic mean ± standard deviation (SD) of three replicates.At harvest time, the percentages of oligomeric and polymeric PAs of skin and seed flavanols were calculated, and the differences between the cv.were tested for significance by applying the t student test.

Skin intact extracts
Barbera grape skins were significantly less rich in PAs and flavan-3-ols measured by the vanillin assay than Nebbiolo grape skins (Figure 1a).
PA and V measurements showed that PA content in skin was high prior to véraison, then decreased during ripening, as previously reported for other cultivars (Downey et al., 2003) as well as for Nebbiolo and Barbera (Di Stefano et al., 2002).This decrease may be due to the deviation of the intermediate metabolites (cyanidin and delphinidin) towards the synthesis of anthocyanins as they share the same biosynthetic pathway (Baranac et al., 1997) or to little-known phenomena involving PA transformation and oxidation.
From a varietal point of view, the decrease was stronger and probably started before the first sampling in Nebbiolo grapes in comparison to Barbera (Figure 1a).
The decrease of V value in both varieties at véraison suggests that during this period PAs also modify their structure.A more rapid decrease was noticed for Nebbiolo skins (Figure 1a).Moreover, PA structure, as revealed by the V/PAs ratio, appeared quite different for the two grape cultivars : PAs in berry skin identified Barbera and Nebbiolo with respect to both their content and structure (Table 1).
Moreover, great differences between V and PAs indexes can be observed in both Barbera and Nebbiolo intact skin extracts, particularly at the end of maturation.This suggests that large flavanols which compose oligomeric and polymeric PAs are not reactive towards vanillin due to sterical hindrance or the involvement of nucleophilic sites in linkages between flavanols: polymeric and complex structures are abundant in grape skins.

Seed intact extracts
As for skin extracts, Nebbiolo seed intact extracts were characterized by higher PA and V content in comparison to Barbera extracts but the trend during grape ripening was similar for both cultivars (Figure 1b).In Barbera, PAs reached a maximum value at the first sampling (3119.55mg kg -1 ), while flavan-3-ols by the vanillin assay (V), as for the skins, reached a maximum value at the second sampling.Similar behavior can be noticed for Nebbiolo seed extracts.
During this trial, V values increased until the second sampling then decreased, while PAs continuously decreased for both varieties: this suggests that (i) the accumulation of flavans determined by the vanillin assay continues for a longer period than that of polymeric structures or (ii) PAs continuously change towards more complex, probably less extractable structures.
Degradation processes are the main reason for PA decrease (Figure 1b).In particular, since it is highly unlikely that the decrease of seed flavanols is due to metabolic reactions, their decrease during ripening could be due to oxidation phenomena and subsequent formation of covalent bonds between the product quinones and polysaccharidic or proteic structures of seeds, making them more difficult to extract during ripening with organic solvents (Kennedy et al., 2000b ;Pourcel et al., 2005).Decreased PA extractability after véraison has been previously reported (Kennedy and Waterhouse, 2000a ;Mateus et al., 2001).
The V/PAs ratio in seed extracts was different considering the two grape cultivars (Table 1).Despite this, it is noteworthy that V and PAs indexes were similar in seeds at least at maturity, in both Nebbiolo and Barbera intact extracts, which was not the case for skins.This result highlights that almost the whole flavanols which compose PAs are reactive towards vanillin : this suggests a modest degree of polymerization or more linear structures in seeds.

Fractionation
The fractions obtained were characterized for their total PA content as well as their V content (as determined by the vanillin assay).In order to evaluate the recovery of the fractionation method employed, the amount of PAs and V of intact extract was compared to the sum of PAs and V determined in diethyl ether, ethyl acetate and methanol fractions (Table 2).
A good correspondence was observed between the PAs and V indexes of skin and seed intact extracts and the sum of the same indexes measured in the fractions (Table 2) : the fractionation method is not responsible for degradation processes or flavanol losses.The method was therefore employed for a first evaluation of the flavanol composition of grape skins and seeds and its evolution during ripening.

Barbera skin extract fractionation
In the monomeric fraction of Barbera skins, flavan-3ol monomers (V) increased during ripening, showing the highest value (57.17 mg kg -1 ) at the last sampling (Figure 2a).Similar results were noticed by Fujita et al. (2007) for Cabernet-Sauvignon but not for Sagrantino grape skins (Di Stefano et al., 2008), where the V value in the diethyl ether fraction remained unchanged during grape ripening.véraison, which then decreased slightly or remained constant during the last period of grape ripening (Figure 2a) : the accumulation of oligomeric and polymeric flavanols occurred during pre-véraison, then their content decreased.Similar results were reported for Sagrantino grapes (Di Stefano et al., 2008).
PAs changed their structure during grape ripening.In particular, more complex structures which are less reactive to the vanillin addition were present.The main result was a lower V/PAs ratio in the skin methanolic fraction (Table 3).

Barbera seed extract fractionation
A clear difference in flavan-3-ol monomers (V) and oligomeric and polymeric PAs was noticed in seed extract fractionation in comparison to skin extracts.
In particular, a higher amount of V and polymeric PAs was measured in seeds (Figure 2b).An increase  of flavan-3-ol monomers in seeds prior to véraison followed by a decrease towards the harvest stage was highlighted, similar to what was observed for other cultivars (Di Stefano et al., 2002 and2008;Fujita et al., 2007;Kennedy et al., 2000b).This trend suggests that monomeric flavan-3-ols are also involved in degradation processes, mainly oxidation reactions.Their involvement in polymerization reactions for the synthesis of PAs seems unlikely, according to the characterization of oligomeric and polymeric fractions.
Actually, the polymeric PAs and V underwent a progressive decrease from the first to the fourth sampling, then it remained almost constant until ripening (Figure 2b).Similar results were reported in literature (Di Stefano et al., 2008).
The total amount of flavano-3-ol monomers and PAs reached its maximum close to the véraison stage, as for Cabernet-Sauvignon seeds (Fujita et al., 2007).
In Barbera seeds, the amount of oligomeric PAs was higher than that of polymeric PAs during the entire ripening period (Figure 2b), except the first sampling, and significantly higher at harvest (Figure 4).Similar results were obtained for Carménère seeds (Obreque-Slier et al., 2010) during ripening but not at harvest when no significant differences between oligomers and polymers were noticed.
This result is in contrast with some previous works where polymeric PAs are the greatest fraction in red wines (Sun et al., 1996), as well as in pulp, seeds and skins of some varieties (Monagas et al., 2003).A possible explanation for this result could be the spectrophotometric method employed to measure flavanols in the oligomeric and polymeric fractions.Usually, in the cited works, the vanillin assay is used to determine flavanols in fractions, while in this work the acidic hydrolysis has been employed.As reported above, a clear difference in the obtained results can be noticed, especially in extracts where a huge amount of oligomeric and polymeric flavanols is present: the hindrance or the involvement of nucleophilic sites in polymers could be responsible for low response in vanillin assay, which does not occur in acidic hydrolysis.

Nebbiolo skin extract fractionation
The PAs determined in the methanol fraction (polymeric PAs) were the most abundant, as previously reported for Barbera skin extract.Unlike the observation in Barbera cv., flavan-3-ol monomers (V) decreased during ripening in Nebbiolo skin (Figure 3a).The evolution of monomeric fraction in grape skins seems to be a varietal character: different trend was observed for Sagrantino cv.(Di Stefano et al., 2008), where the content of monomeric flavan-3ols was constant.
Both oligomeric and polymeric fractions were characterized by a decrease in PAs during grape ripening.The highest PA content was measured at the first sampling date.It can be assumed that already in the first sampling oxidation reactions were active.An increase in V/PAs ratio was measured during ripening in the oligomeric fraction, thus suggesting a modification in the molecular structure (Table 4).
The evaluation of the PAs in the polymeric fraction during ripening showed a decrease in both PAs and V values.The decrease of such values was stronger in Nebbiolo skin extracts in comparison to Barbera ones, thus suggesting that this behavior depends on the grape variety.
Nebbiolo skins presented a higher oligomeric PA content, as a percentage of total skin flavanols, in comparison to Barbera skins during ripening, and a significantly higher content at harvest (Figure 5).
-  As for the oligomeric fraction, a decrease in the V/PAs ratio was observed during ripening for the polymeric structures : an increase in the molecular complexity seems to occur.From a varietal characterization point of view, the polymeric PAs showed minor molecular complexity in Nebbiolo skin extracts (V/PAs: 0.44) in comparison to Barbera ones (V/PAs: 0.33) at harvest (Tables 3, 4).

Nebbiolo seed extract fractionation
Similarly to Barbera seed extract fractionation, clear differences were noticed in the amount of monomeric, oligomeric and polymeric fractions in comparison to skin extract.Also for Nebbiolo seed extracts, the amount in monomeric and oligomeric fractions was high.Similar results are reported in two different studies for Cabernet-Sauvignon seeds (Fujita et al., 2007;Obreque-Slier et al., 2010), while opposite data were shown for Graciano, Tempranillo and Cabernet-Sauvignon cv.(Monagas et al., 2003), where polymeric PAs prevailed both in seeds and skins.In the mentioned studies the same method for the extraction and determination (vanillin assay) of phenolic compounds was used.Thus, environmental parameters could also influence the relative content of monomeric, oligomeric and polymeric flavanols, especially in seeds (Downey et al., 2003).
As observed previously for Barbera and already reported for other varieties (Kennedy et al., 2000b), an increase in flavan-3-ol monomers (V value in the diethyl ether fraction) was observed prior to véraison, then the content in such flavanols progressively decreased during ripening.
In the oligomeric fraction, both oligomeric PAs and V reached a maximum level at the second sampling, thus close to véraison, then their value decreased during ripening (Figure 4b).At the same time, V/PAs ratio remained quite constant (Table 4), thus suggesting that the molecular complexity for the oligomeric fraction did not vary.
The polymeric fraction was characterized by a decrease in both PA and V values during ripening :  the accumulation of these compounds stopped early probably before the first sampling.As reported for the Barbera cv., transformation reactions of flavanols could be the main reason for the decline of such compounds in grape seeds.Moreover, these reactions contribute to a decrease in flavanol extractability (Kennedy et al., 2000b).
The V/PAs ratio at harvest was 0.62 in the polymeric fraction and 0.75 in the oligomeric fraction (Table 4).Polymeric PAs are characterized by a higher molecular complexity than the oligomeric ones.
In this study, the data were obtained under the same environmental and analytical conditions.Thus, from a varietal point of view, Barbera seed oligomeric PAs were more abundant than the polymeric PAs, as mentioned before, both during ripening and at the harvest date, while no significant differences were noticed for Nebbiolo seeds (Figures 2b,3b,4).
Besides important PA quantitative differences between the two cv.at harvest (Figure 4), significant qualitative differences were noticed when the percentages (%) of monomers, oligomers and polymers were calculated in skins and seeds.Nebbiolo skins were characterized by a significantly more important % of oligomeric PAs (P < 0.05) in comparison to Barbera skins, while Barbera seeds contained a significantly higher % (P < 0.01) of flavan-3-ol monomers and oligomeric PAs than Nebbiolo skins at harvest (Figure 5).

CONCLUSIONS
Flavanols are accumulated in Nebbiolo and Barbera skins and seeds prior to véraison.Then, as reported for many varieties, their total content decreases due to the deviation of the intermediate metabolites towards the synthesis of anthocyanins or to little-known transformation and oxidation phenomena.
The amount of total flavanols in grape seeds and skins depends on the variety.In particular, Nebbiolo grapes accumulate higher flavanol content in both seeds and skins than Barbera.Moreover, differences between acidic hydrolysis and vanillin assay values suggest that flavanol composition and complexity in skin and seed extracts represent a varietal characteristic : further evaluation of the flavanol composition is needed.
The fractionation method used in this study has allowed an evaluation of the flavanol composition of grape skins and seeds and their evolution during ripening, revealing further differences between the cultivars.During grape ripening, flavan-3-ol monomers increase in Barbera skins, indicating a modest depolymerization reaction, while the opposite happens in Nebbiolo grapes.An increase of monomers was also noticed in Cabernet-Sauvignon skins, while a different trend was observed for Sagrantino cv.: the content of monomeric flavan-3ols was constant during ripening.
Nebbiolo skins present higher content in oligomeric PAs as a percentage in comparison to Barbera skins both during ripening and at harvest.Moreover, the decrease in polymeric PAs, weaker in Barbera skins in comparison to Nebbiolo, depends on the grape variety.
A rapid and consistent decrease of flavan-3-ol monomers for both Barbera and Nebbiolo seeds during grape ripening suggests that they are probably more involved in oxidation reactions.
Polymeric PAs prevailed in skins, while monomeric and oligomeric structures prevailed in seeds of both cv.The spectrophotometric method used for the evaluation of PA content is also the reason for this result.However, the abundance of monomeric, oligomeric and polymeric flavanols in seeds, as others studies on Cabernet-Sauvignon demostrate, has not been clearly illustrated.In addition to analytic method, environmental parameters seem to influence the relative content of each fraction in seeds.
In our study, the data collected under the same environmental and analytical conditions highlight that Barbera seed oligomeric PAs are more abundant than the polymeric ones during ripening and at harvest, while no significant differences are noticed for Nebbiolo seeds.According to literature, similar results to Barbera were obtained for Carménère seeds during ripening.Finally, Barbera seeds had also a significantly higher percentage of flavan-3-ol monomers and oligomeric PAs than Nebbiolo ones at harvest.
Differences evidenced in this study regarding the total amount of flavanols and their distribution in monomeric, oligomeric and polymeric fractions during grape ripening and at harvest could be the basis to better understand the astringency and bitter sensation in Barbera and Nebbiolo grapes and wines.Moreover, a good knowledge of skin and seed PA composition is important during the winemaking process because the extraction of PAs is not only dependent on the maceration time but also influenced by the characteristics of the grapes.

Figure 1 .
Figure 1.PA and V evolution in Barbera and Nebbiolo intact skin (a) and seed (b) extracts during grape ripening.Data reported as mg kg -1 of fresh weight, mean values ± SD (n = 3).

Figure 3 .
Figure 3. Flavan-3-ol monomer and oligomeric and polymeric PA content and evolution in Nebbiolo skins (a) and seeds (b) during ripening.Data reported as mg kg -1 of fresh weight, mean values ± SD (n = 3).

Figure 4 .
Figure 4. Content of flavan-3-ol monomers and oligomeric and polymeric PAs in Barbera and Nebbiolo skins and seeds at harvest.Data reported as mg kg -1 of fresh weight, mean values ± SD (n = 3).

Table 1 . V/PAs ratio in intact skin and seed extracts of Barbera and Nebbiolo grapes during ripening.
Both oligomeric (ethyl acetate fraction) and polymeric structures (methanol fraction) were characterized by the highest PA value before -63 -J.Int.Sci.Vigne Vin, 2015, 49, 59-69 ©Vigne et Vin Publications Internationales (Bordeaux, France)