2.1. Free aspartic acid (Table 2, Experiment 1)

The free aspartic acid content was determined with HPLC after 12 months of bottle ageing according to the method proposed by Park et al. (2000) and modified by Bosso et al. (2015), based on a pre-column derivatisation reaction between ortho-phthalaldehyde (OPA) and 2-aminoethanol as derivatising agents, and with 2-methylaspartate as internal standard.

2.2. Dissolved oxygen (Table 2, Experiment 4)

The concentration of dissolved oxygen in the white wines was measured with a luminescence-based technology (NomaSense™ O₂ Trace, PreSens GmbH, Regensburg, Germany).

2.3. Filterability

The filterability test (Table 2, Experiment 5) consisted in measuring the time needed to filter increasing volumes of wine: 100 mL (T₁₀₀); 200 mL (T₂₀₀); 300 mL (T₃₀₀); 400 mL (T₄₀₀); and 500 mL (T₅₀₀). Cellulose acetate filters (Sartorius Minisart NML, 0.45 µm, diameter 28 mm) and filtering apparatus from Sartorius (Sartorius Stedim srl, Italy) were used. The filterability index (FI) and the modified filterability index (MFI) were calculated: FI = T₄₀₀ - 2T₂₀₀ and MFI = (T₅₀₀ - T₁₀₀) - 2(T₃₀₀ - T₁₀₀). Filterability is generally considered good when FI < 10, acceptable when 10 20.

2.4. Cold test

The tartaric acid content in the wines was determined with HPLC (Cane, 1990) before and after storage at -4 °C for 6 days. After cooling and before the analysis, the wines were filtered (0.45 µm) at a low temperature to avoid the solubilisation of the KHT salts. The concentration of tartaric acid (H₂T) was measured before and after the period of cold storage, and the difference between the two values due to the precipitation of KHT was calculated: ΔH₂T represents an index of the stability/instability degree of wines (Motta et al., 2009). The wines are considered stable (no loss of tartaric acid by precipitation) when ΔH₂T is ≤ 0.1 (Bosso et al., 2016).

2.5. Conductivity test (mini-contact test)

The tartaric stability of the wines was assessed by measuring the drop of electric conductivity (Δχ expressed as µS/cm) of 100 mL wine at 0 °C, 4 min after the addition of finely micronised KHT (10 g/L) as a precipitating agent: the higher the difference in conductivity, the higher the tartaric instability (Bosso et al., 2016). The test was performed with Check Stab a-2001 Millennium (Delta Acque, Firenze, Italy). The measurements were taken with a probe made up of two platinum electrodes, which were periodically recalibrated with a 0.01 N KCl solution to ensure a high degree of reliability. The temperature was controlled to a resolution of 0.1 °C. The tartaric stability thresholds, at the described operating conditions, were equal to 150 and 100 µS/cm for white and red wines, respectively.

2.6. Physicochemical analyses

Ethanol concentration, total extract, pH, titratable acidity, free and total SO₂ were determined according to EU methods (EEC Regulation 2676/90), and acetaldehyde was measured with a colorimetric method (Di Stefano and Ciolfi, 1982). For white wines, the absorbance at 420 nm (A420) was measured with a quartz cuvette (10 mm o.p.) after filtration with a 0.45 µm polypropylene filter. CIELAB indices (cylindrical coordinates: L∗ lightness, C∗ chroma, h∗ hue) were determined according to Piracci (1994). Wine turbidity was measured with Turbiquant 3000 IR and expressed as nephelometric turbidity units (NTU). The content of iron (Fe) and copper (Cu) was determined by atomic absorption spectroscopy with a Perkin Elmer 5100 PC AA spectrophotometer (Perkin Elmer, Norwalk, CT, USA) according to EU methods (EEC Regulation 2676/90). The content of total polyphenols was determined by spectrophotometry with the Folin-Ciocalteu method (Di Stefano et al., 1989).

2.7. Statistical elaboration

Data was processed with ANOVA. The comparison between the mean values of at least three theses was performed with the Tukey’s post hoc test. SPSS for Windows version 15.0 (SPSS Inc., Chicago Il USA, 2004) was used.

2.1. Effect on tartaric stability

Table 4 shows the results of the cold and conductivity tests performed for the W1 white wine.

After storage for 6 days at -4 °C (cold test), the control was significantly unstable compared to the trials with the two PAs added at two different doses (stable). The same result was obtained with the conductivity test (the stability thresholds are reported in the Materials and methods section). No significant differences were observed between the two PAs and the two doses used with the cold test; on the contrary, there were statistically significant differences between the trials with 100 mg/L and the trials with 200 mg/L of PAs with the conductivity test.

After 6 months of bottle aging (Table 4), the results were unchanged: the PAs trials were stable and statistically different from the control (unstable), with both the cold test and the conductivity test. Moreover, with the cold test no differences were observed between the different PAs nor between the different doses, whereas with the conductivity test the wines with the higher dose and NaPA-15 had significantly lower Δχ values compared to the wines with the lower dose and KPA.

The W2 white wine was unstable (ΔH₂T = 0.15 g/L and Δχ = 209 µS/cm). After the addition of PAs, the wines were stable for both the cold test (ΔH₂T = 0.0 g/L for both doses) and the conductivity test (Δχ = 45.80 and 37.10 µS/cm for the trials with 100 and 200 mg/L of KPA, respectively). The results of the cold test did not change after 12 months, while the Δχ values (conductivity test), especially for the PAs trials, increased (Δχ = 156.6 and 111.7 µS/cm for the trials with 100 and 200 mg/L of KPA, respectively).

The poor reproducibility of the conductivity test during aging, tartaric stability being equal (when measured with the cold test), suggests that the conductivity test is not suitable for measuring over time the tartaric stability of PAs added wines.

2.2 Effect on colour stability and on wine oxidation process

Two experiments were carried out to assess the influence of PAs on the colour of white wines and the oxidative evolution, associated with the possible chelating effect for iron and copper. Therefore, all trials were oxygenated before bottling, at the dose of 4 mg/L, corresponding to an amount of oxygen 4-5 times greater than the dose that is normally dissolved in the wines during correct bottling operations.

At the beginning of the first experiment (W1 wine, Table 4), all trials were similar for the content of free and total SO₂ and colour intensity (A420). After 6 months of bottle aging, only small significant differences in acetaldehyde content were present. SO₂ and acetaldehyde concentrations are related to the wine oxidation process: the free and total SO₂ content tends to decrease by oxidation, while the acetaldehyde content increases (Fenton’s reaction) (Elias and Waterhouse, 2010). The absence of important differences between the trials for these parameters suggests that PAs have no role in the wine oxidation process. During the first 15 days following the oxygen supplies, the concentration of dissolved molecular oxygen was also monitored (Supplementary material: Figure A). All wines showed a high oxygen consumption rate, similar for all trials. Danilewicz (2011) suggested that the consumption rate of molecular oxygen, soon after the oxygen supplies, is influenced by the presence of metals and in particular iron. In fact, the molecular oxygen reacts directly and rapidly with iron (Fe⁺⁺), reducing itself to peroxyl radical capable of oxidizing phenols, whereas SO₂ and phenols participate later in the oxidation process by reacting with the peroxyl radical that is formed by iron. Moreover, it seems that copper acts as a catalyser of the reaction between O₂ and Fe⁺⁺ (Danilewicz, 2011). When increasing the concentration of iron and copper in wines, an increase was observed in the consumption rate of dissolved oxygen and in the loss of free SO₂ (Danilewicz, 2007).

According to our results, a chelating effect of PAs for iron and copper should be excluded, because otherwise a decrease of the initial oxygen consumption rate should have been observed in the PAs trials, particularly with the higher dose. The same can be deduced by considering another step of the oxidation process, the Fenton’s reaction, during which the hydrogen peroxide that is formed by reduction of the peroxyl radical in the presence of phenols or SO₂ is further reduced to hydroxyl radical in the presence of Fe⁺⁺ (Elias and Waterhouse, 2010). A decrease in the rate of this reaction limits the production of acetaldehyde via oxidation of ethanol by the hydroxyl radical. Conversely, in our case the acetaldehyde content did not decrease, but rather it slightly increased in the PAs trials compared to the control.

For this first experiment, iron and copper were added to a wine (W1) in order to accelerate the oxidation reaction and to emphasise the possible differences due to the chelating effect of PAs for metals. The second experiment (W2), without addition of exogenous iron and copper, was performed to verify whether an excess of these metals could have had a masking effect on the chelating action of PAs.

A white wine was used with an original content of 1.5 and 0.075 mg/L of iron and copper, respectively (W2). KPA was added to the wine at two different doses (100 and 200 mg/L). The results (Table 5) confirmed what was observed in the first experiment: no important differences were noticed for free and total SO₂, acetaldehyde and colour parameters after 1.5, 3, 6 and 12 months of bottle aging. Moreover, the addition of KPA did not slow down the oxygen consumption rate that, on the contrary, resulted increased compared to the control (Supplementary material: Figure B).

The presence of KPA did not influence white wine colour and its oxidation process during bottle aging. Unlike for PAs, some authors (Guise et al., 2014) observed a stabilising effect of CMC on wine colour, variable with the type and dose of CMC, which was probably due to the removal of polyphenols. On the contrary, this effect was not observed during our experiments with PAs. No differences between the PAs trials and the control were observed soon after the addition of PAs nor after 6 months (first experiment) and 1 year (second experiment) of bottle aging. As an example, soon after the addition of PAs all wines of the first experiment had an average total polyphenols content of 110-111 mg/L, and after 6 months the average concentration dropped to 103 mg/L in the control and 98 and 101 mg/L in the trials with 100 and 200 mg/L of PAs, respectively.

4.1. White wine (W1)

Table 7 reports the average results of the tartaric stability tests (cold test and conductivity test) for the white wines whose alcoholic degree and pH were artificially modified. All trials had the same content of tartaric acid and potassium.

The wines with pH 3.00 were more stable to KHT precipitations than those with pH 3.70. The pH influences the dissociation degree of tartaric acid: for the same concentration of tartaric acid a decrease of wine pH causes a decrease in the concentration of HT^- ions. Considering the pK values (pK = -logK, where K is the dissociation constant of an acid) of tartaric acid, that is pK₁ = 3.04 and pK₂ = 4.37 as reported by Usseglio-Tomasset (1995), it is possible to calculate the percentages of the three forms of tartaric acid (H₂T, HT^- and T^2-) present in a wine at a given pH. In particular, the percentage of HT^- ions that form insoluble salts with K⁺ ions varies from 46.75 % to 69.80 % (50 % variation) of the tartaric acid present in wine when its pH varies from 3.00 to 3.70. After the addition of KPA all wines became stable, but the effect of KPA seemed weaker at low pH values. However, as the wines were originally less unstable at a low pH than at a high pH, a correct evaluation was not possible.

To date, no studies have yet been conducted to define the mechanism of action of KPA in wines. Some authors (Lankhorst et al., 2017) have recently studied the mechanisms of action of CMC, MTA and MP in a model solution and in a deionised white wine, spiked with KHT, with the use of a method based on dynamic light scattering. The method can detect the presence of wine colloids with sizes ranging from 10-³ to 10-¹ μm, which are already present in wine, and it can distinguish them from the KHT crystals. The results showed that these additives cannot prevent the nucleation of KHT crystals, but rather intervene by slowing down their growth.

Crachereau et al. (2001) studied the effect of CMC on the growth of KHT crystals in a supersaturated solution of the salt: CMC caused a slowing down of the KHT crystals growth and a modification of their shape. According to the authors, this effect was due to the interactions between the negatively charged CMC and the faces of the KHT crystals, positively charged by the accumulation of potassium ions, in competition with the bitartrate ions. It could be hypothesised that also the stabilising action of KPA depends on its negative charge; indeed, all additives used for the tartaric stabilisation of wines, including mannoproteins, are negatively charged. Wine pH could therefore influence the stabilising efficacy of KPA by modifying its negative charge, which decreases with decreasing pH.

The increase of the alcoholic degree did not influence the results of the conductivity test (Δχ), but conversely a negative and significant effect on tartaric stability was observed with the cold test (Table 7). Ethanol influences KHT solubility: when the ethanol content increases, the solubility of KHT in wines decreases. Considering as constant all the other chemical parameters, and applying the formulas proposed by Usseglio-Tomasset et al. (1992), it can be estimated that an increase of 3 % ethanol from 12 to 15 % v/v causes a 25 % reduction of the KHT solubility product in a wine with a saturation temperature ranging from 10 to 15 °C. During the cold test, the appearance and precipitation of KHT crystals in unstable wines are due to the formation of crystallisation nuclei (primary crystallisation) and to their subsequent growth. The formation of the crystallization nuclei depends on the supersaturation state of the wines, that in turn is influenced by the alcohol content. On the contrary, the alcohol content does not influence the conductivity test, during which exogenous KHT crystals are added as crystallisation nuclei, and the measured conductivity drop is only caused by the growth of the added KHT crystals. This is probably the reason for the differences in results observed between the two tests.

After the addition of KPA (independently of the dose) all wines became stable, regardless of their initial level of tartaric instability. Statistically significant differences between wines with different alcoholic degree were observed with the conductivity test, but they were modest and of no practical interest. In all cases, a dose of 100 mg/L was sufficient to stabilise the wines. No differences were observed with the cold test between the trials with different doses of KPA, whereas some significant differences, but without any practical interest, were noticed with the conductivity test.

4.2. Red wine (R3)

The experiment was repeated with a Montepulciano red wine (R3) with an initial low degree of tartaric instability. After the adjustment of wine pH, the trials acidified at pH 3.00 became stable, whereas those at pH 3.70 became more unstable than the original wine. As with the white wine, the increase in alcoholic degree caused an increase of tartaric instability, which in this case was significant with both the cold and the conductivity tests. After the addition of KPA, all wines became stable and no dose effect was observed (Table 7).

Finally, wine turbidity was measured before and after the addition of KPA. Statistically significant interactions between pH and alcohol content were observed (Supplementary material: Table A). The initial wine turbidity was on average 5 NTU; after the addition of KPA the turbidity remained virtually unchanged in the trials at pH 3.00, whereas it increased in the trials at pH 3.70, particularly in those with the higher alcoholic degree. The appearance of haze after the addition of KPA was a rare event throughout the whole research, observed only in this case (R3 red wine) during this work. The appearance of haze observed with an increase of pH could be the consequence of electrostatic interactions between molecules with opposite charges. The destabilising effect of the high alcoholic degree was additive to the pH effect: the trials with the highest alcoholic degree were also the most unstable. However, the pH effect prevailed on the alcohol effect: at low pH no differences in turbidity were observed between the trials with different alcoholic degree.