The safety and feasibility of adoptive immunotherapy, using CD19-specific T cells that have been genetically modified to express a chimeric antigen receptor (CAR) and numerically expanded ex vivo, need to be addressed. Second-generation trials are being developed incorporating improvements into the design of the CAR as well as the manufacturing processes. Here we describe a platform for propagating CD19-specific T cells through an artificial antigen presenting cell (aAPC) which co-expresses CD19 and T-cell co-stimulatory molecules to provide a fully-competent T-cell activation signal leading to T-cell proliferation. K562 cells were selected as the platform for the aAPCs since (i) they have previously been used in compliance with current good manufacturing practice (cGMP), (ii) they express the desired endogenous T-cell adhesion molecules, and (iii) they fail to express classical HLA class I/II molecules and thus are not targets for a T-cell mediated allogeneic immune response. Therefore, K562 cells were genetically modified to co-express CD19 and both of the T-cell co-stimulatory molecules 4-1BBL (CD137L) and MICA. We then tested the ability of these K562 aAPCs to expand T cells expressing a new CD19-specific CAR designated CD19RCD28. This CAR utilizes a CD19-specific scFv to bind to CD19 independent of MHC and confers an activation signal to genetically modified T cells through both CD28 and CD3-ζ. The CD19RCD28+ T cells could be rapidly expanded (50-fold in 14 days) when cultured in the presence of recombinant human IL-2 and irradiated K562 aAPCs (1:50 ratio, T cell to aAPC). The use of freshly thawed aAPCs improved the practicality of using this antigen-driven expansion method in compliance with cGMP. The numerical expansion of the genetically modified T cells was associated with an increased CAR cell-surface expression, from 17 ± 11% (mean ± SD) before co-culture compared with 44 ± 8% (mean ± SD) after co-culture with the aAPCs, which is consistent with T-cell activation through the CAR. A 3H-thymidine incorporation assay was used to demonstrate that CD19 on the K562 aAPC was necessary, but not sufficient, to proliferate CD19RCD28+ T cells. Furthermore, this proliferation assay demonstrated that co-expression of both 4-1BBL (CD137L) and MICA along with CD19 resulted in the most efficient proliferation of the genetically modified T cells. The propagation of CAR+ T cells on antigen+ aAPCs may thus (i) avoid the need for allogeneic peripheral blood mononuclear feeder cells, which are expensive and time-consuming to prepare in compliance with cGMP, (ii) select in vitro for genetically modified T cells with proven CAR-dependent replicative capacity, and (iii) provide conditions for the outgrowth of subpopulations of T cells within a bulk culture that have increased transgene expression. The feasibility of this new T-cell propagation method using aAPC will be tested in the upcoming clinical trials.