Urine was collected by means of a purposefully-designed, two-tier litter tray system. A top tray of 1 mm stainless-steel wire mesh was placed inside a solid plastic litter tray that
allowed for the collection of uncontaminated unne. A complete unne sample was collected from each cat during the 24 h prior to the administration of the isoflavone tablet. For the duration of the trial (96 h) all urine voided was collected and pooled into 24 h batches, mixed by stirring and then weighed before being stored at -4°C, prior to analysis. Isoflavones are known to be stable in urine stored at room temperature for 1 4 days (Atkinson et al. 2005) such that it was not necessary to col lect urine at interval s shorter than this.
3.3.2.5. Faecal collection
Faeces were collected from the top wire tray of each eat's litter tray (described in Section 3 . 3 . 2 .4.) for 1 20 h. Faeces were collected in 24 h pooled samples, and then accurately weighed and frozen at -4°C, prior to analysis.
3.3.2.6. Plasma collection
3 . 3 . 2 .6 .a. Catheter placement
Catheter placement and blood collection technique was as described in Section 3 . 1 .2.3 .a. The only exception was that anaesthesia was induced with tiletamine and zolazepam (Zolatil 1 00; 1 2 mg/kg BW, sub-cutaneous) instead of propofol, since these cats did not tolerate the intra-venous injection required for propofol.
3 . 3 . 2 .6.b. Blood collection technique
3.3.2.7. Urine analysis
Urine was analysed for total and unbound genistein and daidzein by H igh Performance L iquid Chromatography ( H PLC) using methodology adapted from Cave et al. ( 2007a). All samples were analysed in duplicate. Conjugated isoflavones were quantified after enzymatic hydrolysis to yield total unbound equivalents . This hydrolysis step required 0.8 ml urine to be combined with 0.8 ml acetate buffer (pH 5 ), containing 1 990 U sulphatase/m! and 205 U glucuronidase/m! (Sigma-Aidrich, St. Louis, MO, USA). Samples were then incubated in a shaking water bath at 37°C for 1 h before being evaporated to dryness in an automatic speed-vac concentrator with a vapour net ( Savant, New Y ork, NY, USA). The enzyme incubation phase was eliminated when analysing unbound isoflavone content, and the conjugated proportion estimated by the subtraction of the unbound content from the total equivalent content.
Three methanol extractions were performed on each sample by the addition of 300 ).d methanol (BDH, Poole, UK). Suspensions were then mixed by sonication and vortex, before centrifugation at 1 3 ,400 rpm for 5 min. The supematant was taken to dryness before being re-suspended in 200 111 aqueous methanol ( 4.94mol/l) and the mixing and centrifugation procedure repeated. The supematant was then withdrawn, and 25 11 1 inj ected onto the HPLC system.
Analysis was conducted on an Alliance Waters 486 (Millipore Corp., Milford, MA., USA) HPLC, using a Luna 5 11 C l 8 reverse-phase column (4 x 250 mm; Phenomenex, Torrance CA, USA), with an in-line 4 x 3 mm C 1 8 guard column (Phenomenex, Torrance
CA, U SA). Samples were injected in 95 % buffer A ( 1 .75 mol/L HPLC-grade acetic
acid (BDH, Poole, U K)) and 5 % buffer B ( 1 9.2 mol/L HPLC-grade acetonitrile (BDH, Poole, UK)) and eluted using a linear gradient of 5 % B to 70 % B over 40 min with a 1 5 min equilibration at 70 % B before returning to starting conditions. The flow rate was set at 0.5 mllmin and isoflavones were detected by absorbance at 260 nm using a Waters W486 UV Detector (Millipore Corp., MA., USA). Quantification was performed using
Millennium software package (Waters Millipore Corp., Milford, MA, USA) based on peak area.
3.3.2.8. Faecal analysis
Faeces were accurately weighed, before being freeze-dried and re-weighed. Samples were homogeni sed using a food processor, further mixed by shaking and then pooled as follows; faeces voided in the period between time 0 and time 48 h were analysed as one sample, as were faeces voided between 48 and 96 h, with a final (third) pooled sample including all faeces voided between 96 and 1 20 h post-isoflavone ingestion.
Approximately 2 g of each pooled sample was weighed (to 2 dp) into a 50 ml centrifuge tube. Ten ml of 80% methanol (Burdick & Jackson, Michigan, USA) in water was added before being vortexed for 1 min. The sample was then centrifuged and an aliquot removed and filtered through a 0.45 � syringe filter ( B D Pore, Belgium). One ml was taken up for analysis and 1 0 �l of acetic acid added to prevent peak splitting over time (observed in preliminary experiments). An aliquot of 20 �l was injected onto the H P LC column (Shimadzu Corporation, Kyoto, Japan). Pre-treatment of samples with hydrochloric acid ( 1 0 ml of 6 M HCl i n methanol, 1 : 9) was performed on selected samples for extraction of total genistein, daidzein and equol.
A reverse phase C 1 8 HPLC column ( 1 00 x 4.6 mm, 5 � ) (Applied Biosystems, CA, U SA) fi tted with a guard column ( 1 5 x 3 .2 mm, 7 �) (Applied Biosystems, CA, USA) was used. The H PLC conditions involved a gradient elution with mobile phase A consisting of 40% methanol/water and mobile phase B consisting of 70% methanol/water. Over the first 1 0 min the gradient was ramped to 1 00% A, followed by a linear gradient to 1 00% over the subsequent 1 0 min. This was held for 9 m in before being ramped back to 1 00% A, over the fol lowing 1 min. The total run time was 40 min with a flow rate of 1 .5 mllmin. Ultra violet (diode array) detection (Shimadzu Corporation, Kyoto, Japan) was utilised with
wavelength set at 260 nm for daidzein and geni stein, and 280 nm for equol. Genistein and daidzein standards were supplied by Acros Organics (Geel, Belgium) and reported as 99% and 98% pure, respectively. Equol standard (99% purity) was supplied by Fluka (Sigma-Aldrich, Switzerland). Retention times for daidzein, genistein and equol were
1 3 .2 min, 1 7. 6 min and 1 4. 1 min, respectively.
3.3.2.9. Plasma analysis
Analysed as per methodology described in Section 3 . 1 .2.5. (LC-ESI-MS/MS).