For determination of engraftment of human CD3+ CD8+ T cells in NRG mice and their anti-pp65 reactivity, peripheral blood samples were treated with erythrocyte lysis buffer (0.83% ammonium chloride/20mMHepes, pH 7.2) for 1 min, washed with PBS and stained with fluoro-conjugated tetramers and antibodies; PE-conjugated pp65-reactive tetramers HLA-A*0201 (NLVPMVATV) and HLA-B*0702 (TPRVTGGGAM)

(Beckman Coulter), APC-conjugated anti-human CD3 and FITC-conjugated anti-human CD8 were incubated with cells for 15 min at room temperature followed by erythrocyte lysis buffer incubation (Becton Dickinson). The BKM120 cell line FACS acquisition was performed in a FACS Calibur flow cytometer (Becton Dickinson) and the analysis was performed PFT�� mouse using CellQuest software. For functional T cell assay, spleen cells were harvested

and stained with APC-conjugated anti-human CD3 for 30 min in the dark. After washing off unbound antibodies, human CD3+ T cells were sorted from splenocytes with a FACSAria IIu apparatus (Becton Dickinson) and further analyzed with ELISPOT assay. 10,000 CD3+ T cells were seeded on IFN-γ antibody-coated 96 wells plate, restimulated overnight with a pool of pp65 peptides or CEF peptides and the plates were further developed as described above. Viability of iDCs in vivo was determined at different time points with in vivo bio-luminescence imaging analyses. NRG mice were subcutaneously injected at hind flank with 5 × 105 SmyleDCs or SmartDCs, marked with firefly luciferase after co-transduction however with LV-fLUC. Mice were anesthetized

and intraperitoneally injected with aqueous solution of D-Luciferin (150 mg/kg) 5 min before imaging. The imaging was performed on day 7, 14, 30 and 90 days after iDC injection using a CCD camera (IVIS, Caliper Life Sciences, Mainz, Germany). Quantified bioluminescence consisted of averaged photon radiance on the surface of the animal and was expressed as photons/sec/cm2/sr (sr = steradian). Parametric (t test) statistical analysis was used for determining statistical significance. All tests were two-sided, and p < 0.05 was considered significant. Data was analyzed with GraphPad Prism 5 software (San Diego, CA, USA). We constructed bicistronic self-inactivating lentiviral vector backbones co-expressing human GM-CSF/IFN-α (LV-G2α) or GM-CSF/IL-4 (LV-G24) containing 2A elements interspacing the transgenes (Fig. S1a). Through a ribosomal skipping mechanism, a peptidic bond is missing between the 2A glycine and 2B proline sites, resulting in synthesis of two individual proteins [24] and [22]. Using routine production methods [25], both vectors could be consistently packaged as integration-competent lentiviral vectors (IC-LVs) in 293T cells at high titers (Fig. S1b). Packaging of ID-LVs in 293T cells was performed with a construct expressing the HIV gag/pol mutated at the integrase gene (D64V).