To screen for soluble factors which induce further differentiation, islet cells were transduced with an insulin GSI-IX Gamma-secretase inhibitor promoter-DsRed2 reporter lentivirus . Loss of insulin promoter activity during cell expansion, coupled with DsRed2 half-life of 4.5 days, resulted in marker disappearance . Following expansion, cells were transferred to SFM containing various agents, and differentiation was evaluated in live cells by scoring fluorescence reappearance. Based on preliminary screening of individual agents and their combinations, a two-step differentiation protocol termed Redifferentiation Cocktail was optimized . The factors included activin A, exendin-4, nicotinamide, and high NVP-BEZ235 Glucose concentrations, which have been shown to promote beta-cell differentiation . N2, B27, and ITS, were included to prevent cell death in the absence of serum. Glucose concentration was reduced in Step 2 to increase cell sensitivity to glucose-stimulated insulin release . This treatment resulted in cell cluster formation similar to that seen with SFM alone . However, the number of DsRed2 + cells was 6-fold higher in RC, compared with SFM . Since islet cell cultures represent a mix of several cell types, the observed differentiation could result from redifferentiation of BCD cells, or de-novo differentiation of other cells. To determine the origin of the newly-generated insulin-expressing cells we used our inducible lineage tracing approach . BCD cells were labeled during the first days of culture with the RIP-Cre/ER and pTrip�CloxP-NEO-STOP-loxP-eGFP lentivirus vectors in the presence of tamoxifen as previously described . As seen in Figure S1, Cre was specifically expressed in C-pep + cells. Labeled islet cells at passages p4�C6 were treated with SFM or RC, and stained for human C-peptide and eGFP. Since the average beta-cell labeling efficiency is 57.568.9% , the expected incidence of C-pep/eGFP-double-positive cells in case of redifferentiation should be close to this value, while de-novo differentiation should result in 0% co-labeling in the absence of TM. The actual incidence of double-positive cells found following SFM treatment was 60616%, suggesting that redifferentiation was the predominant mechanism . Redifferentiation rate was relatively low, with 4.763.0% of GFP + cells expressing C-peptide .