Generation of PRAME-specific CTLs using PepMix. We used a peptide library of 125 pentadecapeptides spanning the entire PRAME protein to load APCs for the generation of PRAME-specific CTLs. (A) Specificity of the generated CTL lines using PRAME-PepMix as assessed by INFγ ELISpot assay against control PepMix, PRAME PepMix, and a pool of the 4 previously identified HLA-A*02–restricted peptides (P4).^8,18  Medium was used to evaluate background production of IFNγ by nonstimulated CTL. Each symbol represents 1 of the 23 individual PRAME-CTLs; horizontal lines represent the mean group value. (B) IFNγ production of CTLs generated from 4 donors using the PRAME-PepMix at a concentration of 1, 0.1, or 0.01 μg (corresponding to 6, 0.6, and 0.06 nmol each of the 125 15-mer composing PRAME-PepMix, respectively) against medium, irrelevant PepMix, or PRAME-PepMix. (C) The fold expansion of PRAME-PepMix CTLs primed with autologous APCs and stimulated with aAPCs for 3 weeks. Each symbol represents one of the 15 individual PRAME-CTLs; horizontal lines represent the mean group value. (D) PRAME-CTLs expanded from healthy donors were evaluated for their cytotoxic activity using a standard 4-hour ⁵¹Cr-release assay against autologous PHA blasts loaded with irrelevant (gray bar) or PRAME-PepMix (black bars). Data represent the means ± SD of PRAME-CTLs from 9 healthy donors. Shown is the killing at a 20:1 E:T ratio by PRAME-CTLs, which was significantly higher against PRAME PepMix–loaded targets compared with control PepMix targets. In addition, killing of PRAME PepMix–loaded autologous PHA blasts by CTLs was significantly inhibited by preincubation of the targets with an anti–HLA class I antibody, but not by an isotype control, indicating HLA-restricted killing.