*2.3. Sample Port Integration into Tumor Oral Mucosa (TOM) Models*

Tumor oral mucosa (TOM) models were prepared as described elsewhere [14] but adopted to a 6-well-plate design to handle the integration of the sample port (Figure 1a). In brief, 0.3 × 10<sup>6</sup> oral fibroblasts were embedded in collagen G and 1 × 10<sup>6</sup> SCC-25 cells were seeded on top of these lamina propria equivalents one week after. The model growth medium was changed three times a week and replaced by model differentiation medium one week after seeding the tumor cells [14]. The sampling port was created by placing a 24-well insert (400 nm pore size) into the TOM model before the collagen started to solidify. The tumor cells proliferated and migrated into the collagen matrix around the sampling port for seven days, before docetaxel was applied. The 24-well insert was fixed by a custom-made metal support and filled with 600 µL serum-free growth medium. The top of the 6-well plate was sealed with aluminum foil (VWR, Darmstadt, Germany) instead of using the standard plastic lid. TOM models were incubated at 37 ◦C inside the autosampler of the UHPLC-MS/MS device (Agilent Technologies GmbH, Waldbronn, Germany) for the 96 h observation period in the final week of TOM model culture.

**Figure 1.** Experimental design and morphology of tumor oral mucosa (TOM) models. (**a**) Schematic cross-section of (1) sampling port with the needle of the autosampler, (2) TOM model with tumor cells (brown) and fibroblasts (magenta) within lamina propria, (3) Reservoir with differentiation medium, supplemented with docetaxel. The arrows indicate drug diffusion equilibria. Hematoxylin and eosin (H&E) staining of TOM models following two applications of (**b**) the vehicle control and (**c**) 7000 ng/mL docetaxel. Images were representative of four batches; scale bar = 250 µm.

#### *2.4. Docetaxel Treatment of TOM Models*

Docetaxel was dissolved in DMSO to a 70 mg/mL stock solution and diluted with construct differentiation medium to 7; 70; 700; 7000 ng/mL. DMSO, 0.01% in model differentiation medium, served as vehicle control since this was the maximum DMSO concentration among all samples (0.00001%; 0.0001%; 0.001%; and 0.01% DMSO for 7; 70; 700; 7000 ng/mL docetaxel). Docetaxel solutions were applied two times per construct with an application interval of 48 h.

#### *2.5. Morphological Analysis*

TOM models were snap frozen at the end of the 96 h observation period and cut into 7 µm thick slices using a cryotome (Leica CM 1510 S; Leica, Wetzlar, Germany). Cryosections were analyzed by hematoxylin and eosin (H&E) staining and pictures were taken with a microscope (BZ-8000; Keyence, Neu-Isenburg, Germany).

#### *2.6. UHPLC-MS*/*MS Analyses*

Method A: For automated real-time quantitation of docetaxel, an Agilent 1290 UHPLC coupled to an Agilent 6495 triple quadrupole tandem mass spectrometer equipped with a Jet Stream electrospray ionization (ESI) source was used (Agilent Technologies GmbH, Waldbronn, Germany). Separation of docetaxel was achieved on an Agilent Poroshell Phenyl Hexyl column (50 mm × 2.1 mm, 1.9 µm particle size) equipped with a corresponding guard column (5 mm × 2.1 mm, 1.9 µm particle size) using water (solvent A) and acetonitrile (solvent B) each containing 0.1% formic acid (*v*/*v*) as mobile phase. At a flow rate of 0.350 mL/min, the following gradient was applied: 5% B for 0.5 min, to 100% B at 4 min, 1 min hold, 5% B at 5.1 min, stop time 6.50 min. The column compartment was kept at 40 ◦C. The injection volume was 5 µL and the autosampler temperature was set to 37 ◦C. A needle wash (acetonitrile/methanol/isopropanol/water, 25% each, *v*/*v*/*v*/*v*) was applied for 20 s while an additional needle seat backflush using an Agilent Flex Cube was used to minimize carry over (15 s at 2 mL/min with needle wash solvent, pure isopropanol, and a mixture of water/acetonitrile (95/5, *v*/*v*) containing 0.1% formic acid). The total run time was 9.75 min.

The mass spectrometer was operated in multiple reaction monitoring (MRM) acquisition mode. Positive electrospray ionization mode (ESI+) yielded the sodium adduct of docetaxel [M + Na]<sup>+</sup> and was detected at *m*/*z* 830.3. Source and MRM parameters were optimized using Mass Hunter Source Optimizer software (version 1.1, Agilent Technologies Inc., Santa Clara, CA, USA). Final source parameters were as follows: drying gas temperature: 230 ◦C, drying gas flow: 20 L/min (nitrogen), nebulizer pressure: 40 psi (nitrogen), sheath gas temperature: 390 ◦C, sheath gas flow: 12 L/min (nitrogen), capillary voltage: +4,500 V, nozzle voltage: +300 V, high pressure radio frequency (HPRF): 210 V, low pressure radio frequency (LPRF): 160 V. MRM details are listed in Table 1. MassHunter (Quant) software (version B08, Agilent Technologies Inc., Santa Clara, CA, USA) was used for data acquisition and processing.



Method B: For identification of degradation products, an Agilent 1290 II HPLC connected to an Agilent 6550 iFunnel QTOF with Agilent Jet Stream source was used (Agilent Technologies Inc., Santa Clara, CA, USA). Separation of docetaxel and its metabolites was achieved on an Agilent Poroshell Phenyl Hexyl column (50 mm × 2.1 mm, 1.9 µm particle size) equipped with a corresponding guard column (5 mm × 2.1 mm, 1.9 µm particle size) using water (solvent A) and acetonitrile (solvent B) each containing 0.1% formic acid (*v*/*v*) as mobile phase. At a flow rate of 0.350 mL/min, a longer gradient was applied: 5% B for 0.5 min, to 37% B at 5 min, 50% B at 10 min, to 98% B at 15 min, 2 min hold, back to 5% B at 17.1 min, stop time 19 min. The column compartment was kept at 40 ◦C. The injection volume was 5 µL. A needle wash (acetonitrile, methanol, isopropanol, water) was applied for 20 s. The mass spectrometric parameters were as follows: drying gas temperature: 230 ◦C, drying gas flow 14 L/min (nitrogen), nebulizer pressure 40 psi (nitrogen), sheath gas temperature: 375 ◦C, sheath gas flow: 12 L/min (nitrogen), capillary voltage +4,500 V, nozzle voltage +300 V, high pressure radio frequency 200 V, low pressure radio frequency 100 V, fragmentor 365 V. Data acquisition was performed in auto MS/MS mode using a mass range of *m*/*z* 100–1000 at a scan rate of 1 spectrum/s for MS1 and *m*/*z* 50–1000 for MS2 experiments at 3 spectra/s. The collision energy was adjusted depending on the target *m*/*z* value (offset 4 eV, slope 3 eV/*m*/*z* 100).
