injection 3 times weekly, and 0.4 mg afatinib or vehicle control was given daily by oral gavage. P values were generated using a Mann–Whitney test for nonparametric data. Five thousand cells were plated in the inner well of a Matrigel invasion chamber (BD Biosciences) in serum-free media. Wells were placed into media containing 10% FBS, and drugs were added to both chambers where indicated. After 24 hours, cells invading through the Matrigel-coated membrane were stained and counted. P values were generated using a homoscedastic 2-tailed Student’s t test. Immunoblots were carried out on cell lysates collected 48 hours after plating in drug-free media. Lysates were resolved on SDS-PAGE gels and transferred to nitrocellulose membranes prior to antibody staining with the following antibodies: EGFR (BD Transduction Laboratories); HER2 and 611-CTF (clone F11, sc-7301; Santa Cruz); pHER2 and p611-CTF at Y1248 [(2247s); Cell Signaling]; p-serine (BD Transduction Laboratories); and cortactin (Upstate XL765 

Biotechnology). Densitometry was carried out using ImageJ software, and P values were generated by a Student’s t test. Tumors were initially snap frozen then fixed in 4% paraformaldehyde overnight, followed by 30% sucrose overnight before embedding and cryosectioning. Tissue sections were stained using the TumorTACS Apoptosis Detection Kit (Trevigen) on the basis of terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling (TUNEL) staining and according to the manufacturer’s protocol. Lentiviral particles were provided by Dr. R.W. Sobol and the University of Pittsburgh Cancer Institute (UPCI) Lentiviral Facility. Virus stocks were generated by cotransfection of the short EGFR(HER) inhibition hairpin RNA (shRNA) expression plasmid (pLK0.1; Mission shRNA library from Sigma) into 293-FT cells together with the packaging plasmids pMD2.g (VSVG), pRSV-REV, and pMDLg/pRRE. Forty-eight hours posttransfection viral particles were collected in the

culture supernatant, filtered (0.45 mmol/L), and stored at80Cor used immediately to transduce the target cells. To study mechanisms of cetuximab resistance, we created a preclinical model on the basis of the previously published in vivo generated model of trastuzumab resistance (13). Subcutaneous tumor xenografts were established using 5 cetuximab-sensitive epithelial cancer cell lines (T24, CAL33, A431, OSC-19, and SCC1) as well as a previously described cetuximab-resistant epithelial cancer cell line, SCC1c8 (15). Xenograft-bearing athymic nude mice were treated with increasing EGFR(HER) inhibitor in clinical trials concentrations of cetuximab over the course of 3 months. Animals were initially treated with moderate doses of cetuximab that are equivalent to 4 times that of a human dose (0.8 mg 2 times/wk). This was increased to doses equivalent to 6 times the standard human dose of cetuximab (0.8 mg 3 times/wk) over the course of 3 months. A majority of the epithelial carcinoma– derived xenografts regressed with cetuximab treatment, including the head and neck cancer cell line SCC1 and its in vitro derived cetuximab-resistant clone SCC1c8 (Fig. 1A). Although most xenografts treated with cetuximab were cetuximab-sensitive, 4 cetuximab-resistant tumors (T24PR1–4) emerged out of the 12 original

xenografts from T24 bladder carcinoma cells (Fig. 1A). Cetuximabresistant tumors T24PR1–4 were surgically removed from sacrificed animals and digested into single-cell suspensions that were used to generate cell lines of the same name in vitro and additional xenografts in vivo. Xenografts from the cetuximab-resistant cells persisted despite EGFR(HER) inhibitor drug treatment with doses of cetuximab equivalent to 12 times the human dose of cetuximab (0.8 mg 3 times/wk) immediately upon tumor formation (Fig. 1B). The persistent growth of tumors derived from in vivo generated cetuximab-resistant cells as compared with in vitro generated cetuximab-resistant cells in high doses of cetuximab shows the validity of in vivo generation for models of drug resistance, especially for therapeutic agents such as monoclonal antibodies that are known to have antitumor effects that cannot be reproduced under cell culture conditions. To distinguish acquired resistance to cetuximab from intrinsic resistance, we compared cetuximab sensitivity between the cetuximab-sensitive parental cells and the cetuximab-resistant clones. To test this in vivo, athymic nude mice were inoculated with sensitive cells on one flank and resistant cells on another flank. Following tumor formation, animals were randomized on the basis of tumor volumes and treated with high concentrations of cetuximab (2.0 mg 3 times/wk).