MATERIALS AND METHODS

Vinification of wine. Vitis vinifera cv Pinotage grapes were sourced from vineyards in the Boland region of South Africa, during the 2012 and 2013 vintages. Fermentations were done in triplicate in 2012 and quadruplicate in 2013 in polypropylene buckets with 5 kg de-stemmed and crushed grapes per bucket (weighed, de-stemmed and crushed individually; one bucket is one biological repeat). After the grapes were crushed, sodium metabisulfite was added (30 ppm SO₂) and the total acid was adjusted to 6.5 g/L. The yeast strain, Saccharomyces cerevisiae VIN13 (Anchor Yeast, Johannesburg, South Africa) were rehydrated and inoculated according to the manufacturer’s recommendations and yeast nutrition was supplemented with Fermaid K^TM (Lallemand, Bellville, South Africa). Where applicable, Rapidase^ Ex Color, CB and -Expression (DSM Food Specialties, Heerlen, The Netherlands) were added to the crushed grapes at a dosage of 0.03 g/L. The manufacturer of these enzymes declares (Product data sheets, www.dsm-foodspecialities.com) that all the mixes have primarily pectinolytic activity but that Rapidase^CB has been formulated with enhanced polygalacturonase activity, Rapidase®

Ex Color has additional hemicellulases for optimum grape skin extraction during maceration and Rapidase® Expression formulations contain arabinases and rhamnogalacturonases.

Fermentations were conducted at 25°C and the grape skin caps were punched down twice

daily. At the end of alcoholic fermentation the wines were pressed and transferred to 4.5 L glass bottles fitted with fermentation caps. The pomace was stored at 4 C until analysis (up to one week). Freeze dried Oenococcus oeni, Lalvin VP41 (Lallemand, South Africa), supplemented with OPTI-MALO-PLUS^ (Lallemand), was inoculated in the wine to induce malolactic fermentation (MLF) and the fermentations were incubated at 20C. At the end of MLF the wine was racked off the lees, sulfited (80 ppm SO₂), stabilised at 4°C and then filtered and bottled.

General oenological parameters. Three 50 mL samples of the juice were randomly selected from the buckets after the grapes were crushed, and analyzed for sugar and acid concentration using Fourier transform infrared (FT-IR) spectroscopy with a WineScan™ FT120 Basic (FOSS Analytical,Hillerød, Denmark). The same instrument was used to analyse the pH, volatile acidity, total acid, glucose + fructose and ethanol of two 50 mL wine samples from each biological repeat at the end of MLF. The FT-IR scans were done in duplicate per sample (two technical repeats). Wine samples for color analysis were collected three months after the wines have been bottled. The pH of all samples was adjusted to 3.5 and the color characteristics were determined according to Iland et al.²⁶ using a Specord 50 UV/VIS spectrophotometer (Analytik Jena, Jena, Germany). The analysis was done in triplicate.

Pinotage skin cell wall preparation. For the fresh grape skin samples the grapes were selected from different harvest crates, from different bunches and from the top, middle and bottom of a bunch to mimic the variability that is found in a commercial harvest batch. The berries were frozen and stored at -20C before the skins were obtained by manually peeling the grapes with a scalpel and flash-freezing the skins in liquid nitrogen. Twenty-one samples of 10 g each (wet weight) were collected in 2012 and twelve in 2013, kept at -80 C before freeze-drying. The fermented grape skin samples (with and without enzyme treatment) were collected at the end of alcoholic fermentation after pressing (at least two samples per biological repeat) and stored at 4C before excess pulp was removed from the skins with a scalpel and the skins were freeze-dried. Samples of 10 g (wet weight) were milled and homogenized with a Retsch MM400 mixer mill (Retsch, Haan, Germany) at a frequency of 30Hz for 30 seconds and the resulting powder was used for the isolation of the cell walls using the method described by Ortega-Regules et al. ²⁷ with the following changes. The first acetone step after the buffered phenol extraction was omitted. Furthermore, after the third acetone wash the alcohol insoluble residue (AIR) was air-dried and then mixed with a volume of MilliQ water, roughly equal to the pellet size of the AIR, frozen at -80C and then freeze-dried.

Mid-infrared spectroscopy. AIR isolated from fresh and fermented Pinotage skins (with and without the addition of maceration enzymes) was analyzed by infrared spectroscopy. The samples were placed in direct contact with the diamond window and the spectra between 4000 and 600 cm^-1 were acquired using a NEXUS 670 (Thermo Scientific, MA, USA) fitted with a Golden Gate Diamond ATR (attenuated total reflectance). A Geon-KBr beam splitter and DTGS/Csl detector were used and each sample was scanned 128 times. The FT-MIR spectra data were filtered with a MSC (multiplicative scatter correction) filter to compensate for light scattering and differences in the effective path length. The spectra for the fingerprint region (1400 – 800 cm^-1) where every polysaccharide has its own unique pattern, as well as wavelengths from 1400 up to 1800 cm^-1 that conveys information about functional groups ^28,29 were compared between treatments. The spectra shown in the results section represent the average absorbance from at least five AIR samples per treatment. Technical repeats were done once every tenth scan to control for technical reproducibility.

Monosaccharide analysis of cell wall samples. The monosaccharide composition of the AIR isolated from the grape skins were analyzed according to the method described in ³⁰ and used on grape cell walls in ²⁴ including a few modifications. Approximately 5 mg of AIR was hydrolyzed to monosaccharides using 2M TFA and incubated for 2 hours at 110C³⁰. This was followed by derivatization to methoxy sugars at 80 C for 16 h. After silylation with HMDS + TMCS + Pyridine 3:1:9 (Sylon HTP) kit (Sigma-Aldrich, MO, USA) the derivatives were separated and analyzed in a gas chromatograph, Agilent 6890 N (Agilent Technologies, CA, USA) coupled to a Agilent 5975 MS mass spectrometer detector, using a polar (95%

dimethylpolysiloxane) ZB-Semivolatiles Guardian (30 m, 0.25 mm ID, 0.25 µm film thickness) GC column. The oven temperature was maintained at 70 C for 2 min, ramped at 1 C/min to 76

C, then at 8 C/min to 300 C and then held for 5 min. The monosaccharide composition was expressed as the mole percentage contribution of each monosaccharide in relation to the nine monosaccharides present. Error bars in the histograms or the standard deviation in the tables represents the mean of at least three biological samples.

Comprehensive microarray polymer profiling (CoMPP) analysis of cell wall fractions. The pectin- and hemicellulose-rich fractions of the cell wall samples were sequentially extracted from 10 mg AIR samples using the solvents CDTA (diamino-cyclo-hexane-tetra-acetic acid) and NaOH respectively according to the procedure described elsewhere²⁴. Arrays printed with the different fractions were probed individually with 26 different monoclonal antibodies (mAbs) and carbohydrate binding modules (CBMs) as previously described²⁵. A mean spot signal was

calculated where after it was normalized to the highest signal (set as 100) in the dataset and a cut-off value of five was used.

Multivariate and univariate statistics. Univariate statistical analysis were done (ANOVA, with P = 0.05) under the guidance of the Centre for Statistical Consultation at Stellenbosch University (Prof Martin Kidd) using Statistica 10 (StatSoft Southern Africa - Analytics, Sandton, South Africa). Multivariate analysis by means of principle component analysis (PCA)³¹ and orthogonal projections to latent structures (OPLS) were performed using SIMCA 13 software package (Umetrics AB, Umeå, Sweden).