and oncogenes, where it is critical for survival and for tumorigenesis (Guo et al. in the autophagy-deficient cell lines, which instead accumulated the autophagy substrate p62 (Figure 1B). Autophagy defects also caused accumulation of ER chaperones GRp170 and GRp78 and protein disulphide isomerase (PDI) in both iBMK cell lines (Figure 1C; top panel), consistent with previous findings (Mathew et al., 2009). We then examined the autophagy flux with the lysosomal inhibitor Bafilomycin A1, which resulted in the accumulation of LC3-II in the WT RGS3 cells, but not in the autophagy-deficient cells (Figure 1D). In tumor-derived cell lines (TDCLs) from the genetically engineered mouse model (GEMM) for non-small-cell lung cancer (NSCLC) (Guo et al., 2013a), starvation robustly induced autophagy in the WT cells, while autophagy defects resulted in p62 accumulation and elevated expression of ER stress markers (Figure 1B, 1C, and 1E). Thus the iBMK cells used for the SILAC are representative of autophagy functionality independent of tissue type and subfamily. SILAC-based mass spectrometry coupled with strong cation exchange (SCX) and off-gel fractionations (OG) and protein identification buy 187164-19-8 by MaxQuant (MQ) and Proteome Discoverer (PD) (Supplemental Experimental Procedures) enabled identification of 7184 proteins (~25% of the total estimated mouse proteome) present during at least one of the conditions tested (0, 3 and 5 hours of starvation) (Figure 2A; red) comparable to the most comprehensive description of the mouse kidney proteome to date (dotted circle) (Huttlin et al., 2010) (Figure 2A). Of these, 5300 proteins were identified by both MQ and PD algorithms, with 845 unique to PD and 1039 unique to MQ (Figure 2B). Similarly, 5441 proteins were identified by both fractionation techniques, with 931 unique to SCX and 812 unique to OG separations (Figure 2B), consistent with other observations of partial complementarity between the two fractionation techniques and algorithms (Barbhuiya et al., 2011; Chang et al., 2013). A substantial overlap between the proteins was identified in each starvation buy 187164-19-8 condition, indicating the relatively minor qualitative alterations to the proteomes (Figure 2B). Figure 2 Effect of autophagy deficiency on the cellular proteome We normalized the data in two ways; first we took into account difference in viability during starvation to which the autophagy-deficient cells are more sensitive, by normalizing the proteomes on a per-cell basis. The first observation was that a strikingly large percentage of the observed proteome was impacted by the functional status of autophagy, as evident by differential relative protein abundances between WT and autophagy-deficient cell lines consistent across the duration of starvation. We observed that autophagy was involved in the predicted degradation of nearly half of the overall proteome on a per cell basis within 3 hours of starvation that increased to 70% at 5 hours (Figure 2C). This suggests that autophagy is a significant mechanism for turnover and remodeling of the cellular proteome. Second, to examine the specificity by which autophagy impacts the global proteome, proteins in each SILAC channel were normalized to total protein for each cell type and examined for relative protein abundance ratios (ratios of levels in autophagy-deficient compared to WT; deletion in deficiency in HRasG12V-transformed cells significantly altered relative proteins levels of the majority of cellular proteins compared to those in WT cells, revealing the magnitude of the impact of autophagy on protein homeostasis. The major downstream consequence of disruption in protein homeostasis was two-fold. First, when autophagy was functional, there was selective elimination of proteins detrimental for cell survival to stress, while those that support cell survival were preserved. Second, defects in autophagy caused accumulation of putative autophagy substrates, many of which were members of signaling pathways detrimental to cell survival. Moreover, buy 187164-19-8 their accumulation was sufficient to mediate cell death, providing an explanation, at least in part, for the requirement of autophagy for tumor maintenance. Importantly, autophagy-deficient cells accumulate levels of PARP family members while depleting their catabolic counterpart, PARG (Figure 3C). This alteration could occur in response to alterations in NAD+ regulation and changes in cellular energy homeostasis and likely leads to increased ADP-ribosylation of PARP targets. Alteration in PARP levels are implicated in NAD+ depletion, mitochondrial dysfunction, inflammatory-response gene expression, senescence and susceptibility to cell death, all of which are phenotypes we and others have observed in autophagy-defective cells. Most striking was the observation that autophagy defects result in accumulation of proteins such as RIG-I.