*2.2. Analysis*

#### 2.2.1. Milk Clotting

Milk-clotting was evaluated visually when clotting began, in accordance with Uchikoba et al. [14]. During the visual evaluation, the time was measured between the addition of the coagulant solution and the first appearance of solid material against the background (CT, clotting time).

Milk-clotting properties were evaluated by a Formagraph ® (Foss Electric, Hillerød, Denmark). Following the Formagraph ® instructions, milk samples (10 mL) were heated to 35 ◦C, and 200 μL of a solution of rennet (NATUREN ® PLUS 215-215 IMCU/mL; chymosin 63%; pepsin 37%) (CHR HANSEN, Hoersholm, Denmark) with a strength of 40 IMCU/mL was added; 10 replicates per run and for each sample milk were performed. The same conditions were applied to evaluate the kiwifruit extract; however, 200 μL of a 40mg/mL solution of freeze-dried powder in distilled water was used. Measurements were stopped thirty minutes after the addition of the enzyme.

The principle of lacto-dynamography is based on the control of the oscillation that is driven by an electromagnetic field created by a swinging pendulum. During milk clotting, a pendulum is immersed into the milk container. The greater the extent of the coagulation, the smaller the pendulum swing. This analysis provided measurements of the clotting time (r) in min, curd firming time (k20) in min, and curd firmness (A30) in mm [15].

#### 2.2.2. Physical and Chemical Analysis

Cheese samples were analysed in terms of moisture, protein, fat, and ash following o fficial AOAC methods (AOAC, 2000). For the colour measurements, samples were placed on a standard white tile. Colour readings were taken at four randomly selected locations on the cranial surface of each piece to obtain a representative mean value. The cheese colour was measured in the CIE L\*a\*b\* space (CIE, 19876) with an area diameter of 8 mm, including the specular component, and 0% UV, D65 standard illuminant, observer angle 10◦, and a zero and white calibration using a Minolta CM 2006d spectrophotometer (Konica Minolta Holdings, Inc., Osaka, Japan). Lightness (L\*), greenness (a\*) and yellowness (b\*) were recorded [16,17]. The colour parameters were used to calculate the total colour di fferences between cheeses obtained with calf rennet and cheeses obtained with kiwifruit, using the following formula: ΔE\* = ((L\*)<sup>2</sup> + (a\*)<sup>2</sup> + (b\*)2) 1/2. Values were expressed as the mean ± standard deviation. Following Sanz [18], the colour di fferences for the human eye are not obvious if ΔE\* < 1; not appreciable if 1 < E\* < 3; and obvious if E\* > 3.

Calcium, iron, sodium, magnesium and potassium were determined by flame atomic-absorption spectroscopy on an iCE 3000 series AA spectrophotometer (Thermo-Scientific, Waltham, MA, USA) equipped with a deuterium lamp as a background-correction system. An acetylene-air flame was used, while the gas flow rates, and the burner height were adjusted in order to obtain the maximum absorbance signal for each element. The organic matter of the samples (0.5 g) was put in a mu ffle-furnace at 450 ◦C for 24 h to obtain ash. When cool, the residue was dissolved in 1 mL nitric acid and the volume was diluted to 10 mL with water. The wavelength of the spectrometer was set at 422.7 nm for Ca, 589.6 nm for Na, 248.3 nm for Fe, 766.5 nm for K and 285.2 nm for Mg. The slit width was 0.7 nm for all elements, except Fe with 0.2 nm. The volumes and corresponding concentrations of the samples were selected within the linear range of the instrument used (at least five concentrations).

#### 2.2.3. Lipid Composition of Kiwifruit and Cheeses

Total lipids (TL) of kiwifruit and cheeses were extracted with a chloroform/methanol solution (2:1, v/v), following Rodriguez-Estrada et al. [19].

Unsaponifiable matter was obtained following Sanders et al. [20]. Briefly, 300 mg of TL were cold-saponified by adding 4.5 mL of ethanolic KOH (4.8% w/v) solution and incubated at room temperature for 12 h. The unsaponifiable matter was isolated by two washes with 4.5 mL of water and 9 mL of hexane. The non-polar phase (upper phase) was transferred into a fresh tube and

dried by nitrogen gas. Finally, the samples were dissolved once again with 1 mL of methanol. Before saponification, 100 μL of a solution of dihydrocholesterol in chloroform (2 mg/mL) as internal standard for sterols were added to TL.

Sterols were then silylated adding a hexamethyldisilazane/chlorotrimethysilane/pyridine 2/1/5 v/v/v mixture, dried under a nitrogen stream and dissolved in 300 μL of n-hexane. The sterols were identified and quantified using a GC–FID (GC 2000 plus, Shimadzu, Columbia, MD, USA) equipped with a VF 1-ms apolar capillary column (25 m × 0.25 mm i.d., 0.25 μm film thickness; Varian, Palo Alto, CA, USA). A total of 2 μL of the sample in hexane were injected into the column with the carrier gas (hydrogen) flux at 1 mL/min and the split ratio was 1:10. The run was carried out in constant pressure mode. The oven temperature was held at 250 ◦C for 1 min, and increased to 260 ◦C over 20 min at the rate of 0.5 ◦C/min, and then increased to 325 ◦C over 13 min at the rate of 5 ◦C/min, and kept at 325 ◦C for 15 min. The injector and the detector temperatures were set at 325 ◦C. Chromatograms were recorded with LabSolution (Shimadzu, Columbia, MD, USA). Sterols were calculated by comparing the area of the samples and internal standards and expressed as mg/100 g of cheese.

#### 2.2.4. Phenol Extraction, Quantitation and Characterization

A liquid–liquid extraction was used to isolate the phenolic fraction from the cheeses and kiwifruit, following Suarez et al. [21] with some modifications. Briefly, 10 mL of methanol/water (80/20, *v*/*v*) were added to 5 g of sample and homogenized for 2 min with an ULTRATURRAX (IKA ®-Werke GmbH & Co. KG, Staufen, Germany). After this, two phases were separated by centrifugation at 637× *g* for 10 min and the supernatant (hydroalcoholic phase) was transferred to a balloon. This step was repeated with 5 mL of methanol and the extracts were combined in the balloon. The hydroalcoholic extracts were then rotary evaporated to a syrupy consistency at 31 ◦C and dissolved in 5 mL of acetonitrile. Subsequently, the extract was washed with 10 mL of n-hexane and the rejected n-hexane was treated with 5 mL of acetonitrile. The acetonitrile solution was finally rotary evaporated to dryness. It was then re-dissolved in 5 mL of methanol and maintained at −20 ◦C before the chromatographic analysis.

Total phenol concentration extracts were determined spectrophotometrically by the Folin–Ciocalteu assay [22] using gallic acid as a standard. An aliquot of 1 mL of each extract was mixed with 5 mL of H2O and 1 mL of Folin–Ciocalteu phenol reagen<sup>t</sup> 1N. The reaction had a duration of 7 min. A total of 10 mL of saturated Na2CO3 solution (7.5%) and 5 mL of H2O were then added and allowed to stand for 90 min before the absorbance of the reaction mixture was measured in triplicate at 750 nm. The total phenol content was expressed as mg of polyphenols per 100 g of cheese.

Individual polyphenol profiles by HPLC analysis were determined according to Kim et al. [23] with slight modifications. Briefly, 20 μL of each sample were analysed using a Prostar HPLC (Varian) with UV-DAD and a C18 reverse phase column (ChromSep HPLC Columns SS 250 mm × 4.6 mm including Holder with ChromSep guard column Omnispher 5 C18). The PDA acquisition wavelength was set in the range of 200–400 nm, with an analogue output channel at wavelength 280 nm width 10 nm. The gradient elution was performed by varying the proportion of solvent A (water–acetic acid, 97:3 v/v) to solvent B (methanol), with a flow rate of 1 mL min−1. The initial mobile phase composition was 100% solvent A for 1 min, followed by a linear increase in solvent B to 63% in 27 min. The mobile phase composition was then brought back to the initial conditions in 2 min for the next run. All the solutions prepared were filtered through 0.45 μm membranes.

#### 2.2.5. Volatile Organic Compounds Analysis

The volatile organic compounds were determined by solid phase microextraction–gas chromatography–mass spectrometry (SPME-GC/MS), according to Serra et al. [24]. Briefly, volatile organic compounds (VOCs) were extracted from 5 g of a finely-ground sample in a 20-mL glass vial closed with an aluminium cap equipped with a PTFE-septum. Samples were incubated for 15 min and then VOCs were collected using a divinylbenzene/carboxen/polydimethylsiloxane (DVB/Carboxen/PDMS) Stable Flex SPME fibre (50/30 μm; 2-cm long) (Supelco, Bellefonte, PA, USA). The SPME fibre was exposed to headspace for 30 min. The conditioning and exposure were carried out at 60 ◦C [25]. The fibre was inserted into the injector of a single quadrupole GC/MS (TRACE GC/MS, Thermo-Finnigan, Waltham, MA, USA) set at 250 ◦C, 3 min in splitless mode, keeping the fibre in the injector for 30 min to obtain complete fibre desorption.

The GC programme conditions were the same as those described by Serra et al. [24]. The GC was coupled with a Varian CP-WAX-52 capillary column (60 m × 0.32 mm; coating thickness 0.5 μm). The transfer-line and the ion source were both set at 250 ◦C. The filament emission current was 70 eV. A mass range from 35 to 270 *m*/*z* was scanned at a rate of 1.6 amu/s. The acquisition was carried out by electron impact, using the full scan (TIC) mode. Three replicates (*n* = 3) were run per sample. The VOCs were identified in three different ways: (i) comparison with the mass spectra of the Wiley library (version 2.0-11/2008); (ii) injection of authentic standards; and (iii) calculation of the linear retention index (LRI) and matching with reported indexes [26–28]. Data were expressed as the peak percentages of the total VOCs.

#### *2.3. Statistical analysis*

JMP software (SAS Institute Inc., Cary, NC, USA) was used for the statistical analysis. Data were analysed with the following mixed linear model:

$$\mathbf{y}\_{i\mathbf{j}} = \boldsymbol{\mu} + \mathbf{C}\_{i} + \mathbf{B}\_{\mathbf{j}} \left( \mathbf{C}\_{i} \right) + \boldsymbol{e}\_{i\mathbf{j}} \tag{1}$$

where yij = the dependent variables (physico-chemical component, lactodimography data, fatty acids, sterols, polyphenols) relative to the ith coagulant and to the jth batch; μ = the mean; Ci = the fixed effect of the ith coagulant (BM-C vs. BM-K or SM-C vs. SM-K); Rj (Ci) = the random effect of the jth batch (1 or 2) nested within Ci; and *e*ij = the random residual.
