Proteomic Profiling Identified Multiple Short-lived Members of the Central Proteome as the Direct Targets of the Addicted Oncogenes in Cancer Cells

“Oncogene addiction” is an unexplained phenomenon in the area of cancer targeted therapy. In this study, we have tested a hypothesis that rapid apoptotic response of cancer cells following acute inhibition of the addicted oncogenes is because of loss of multiple short-lived proteins whose activity normally maintain cell survival by blocking caspase activation directly or indirectly. It was shown that rapid apoptotic response or acute apoptosis could be induced in both A431 and MiaPaCa-2 cells, and quick down-regulation of 17 proteins, which were all members of the central proteome of human cells, was found to be associated with the onset of acute apoptosis. Knockdown of PSMD11 could partially promote the occurrence of acute apoptosis in both MiaPaCa-2 and PANC-1 pancreatic cancer cells. These findings indicate that maintaining the stability of central proteome may be a primary mechanism for addicted oncogenes to maintain the survival of cancer cells through various signaling pathways, and quick loss of some of the short-lived members of the central proteome may be the direct reason for the rapid apoptotic response or acute apoptosis following acute inhibition of the addicted oncogenes in cancer cells. These findings we have presented can help us better understand the phenomenon of oncogene-addiction and may have important implications for the targeted therapy of cancer.

"Oncogene addiction" is an unexplained phenomenon in the area of cancer targeted therapy. In this study, we have tested a hypothesis that rapid apoptotic response of cancer cells following acute inhibition of the addicted oncogenes is because of loss of multiple short-lived proteins whose activity normally maintain cell survival by blocking caspase activation directly or indirectly. It was shown that rapid apoptotic response or acute apoptosis could be induced in both A431 and MiaPaCa-2 cells, and quick down-regulation of 17 proteins, which were all members of the central proteome of human cells, was found to be associated with the onset of acute apoptosis. Knockdown of PSMD11 could partially promote the occurrence of acute apoptosis in both MiaPaCa-2 and PANC-1 pancreatic cancer cells. These findings indicate that maintaining the stability of central proteome may be a primary mechanism for addicted oncogenes to maintain the survival of cancer cells through various signaling pathways, and quick loss of some of the short-lived members of the central proteome may be the direct reason for the rapid apoptotic response or acute apoptosis following acute inhibition of the addicted oncogenes in cancer cells. Although malignant carcinomas frequently contain multiple genetic and epigenetic abnormalities (1)(2)(3)(4), their sustained proliferation and/or survival are often dependent on a single activated oncogenic protein or pathway. Acute disruption of the oncogenic activity of the addicted oncoprotein or pathway can cause tumor cells to undergo rapid apoptosis, or sometimes growth arrest and differentiation (5,6). This phenomenon was first coined as "oncogene addiction" by Bernard Weinstein (5), and now it has been observed in multiple genetically engineered mouse models of human cancers, mechanistic studies in human cancer cell lines, and clinical experience involving specific molecular targeted agents (7), highlighting its potentially important implications of this phenomenon in the treatment of cancer.
To explain oncogene addiction, it has been suggested that the rapid apoptotic response observed in tumor cells on acute disruption of an oncogene product results from differential decay rates of various short-lived prosurvival (such as phospho-ERK, -Akt, and -STAT3/5), and longer-lived proapoptotic signals (such as phospho-p38 MAPK) emanating from the oncoprotein (such as EGFR or BCR-ABL) following its inactivation. Although this theory has circumstantial evidence from experimental findings in several systems, the exact molecular mechanism of how these proapoptotic and prosurvival signals were integrated to lead to rapid apoptosis following acute inhibition of the addicted oncogenes is still poorly understood.
In recent years, several research groups have documented that inhibition of protein synthesis with cycloheximide alone could also induce rapid apoptosis within 2-4 h in a variety of cancer cell lines (8 -12), or could markedly accelerate vinblastine induced apoptosis in several leukemia cell lines with cells dying in 4 h from all phases of the cell cycle, and it has been coined as "acute apoptosis" by Alan Eastman (13) to distinguish it from the delayed apoptosis, which is associated with cell cycle arrest. These research findings suggest that the rapid apoptotic response following acute inhibition of the addicted oncogenes in cancer cells may be caused by loss of multiple short-lived proteins whose activity normally maintains cell survival by blocking caspases activation directly or indirectly. Thus identifying these short-lived proteins can help us better understand the phenomenon of oncogene addiction.
In this study we showed that rapid apoptotic response or acute apoptosis could be induced in both A431 cells and pancreatic cancer MiaPaCa-2 cells when treated with corresponding signaling inhibitors, and proteomic profiling identified that the quick down-regulation of 17 short-lived proteins, which were all members of central proteome of human cells, was associated with the onset of acute apoptosis in both A431 and MiaPaCa-2 cells. Knockdown of PSMD11 could partially promote the occurrence of acute apoptosis in both MiaPaCa-2 and PANC-1 pancreatic cancer cells. Based on these and additional findings described below, we conclude that maintaining the stability of central proteome may be a primary mechanism for addicted oncogenes to maintain the survival of cancer cells through various signaling pathways, and quick loss of some of the short-lived members of the central proteome may be the direct reason for the rapid apoptotic response or acute apoptosis following acute inhibition of the addicted oncogenes in cancer cells.
Inhibitors-LY294002 and Cycloheximide (CHX) 1 were purchased from Sigma-Aldrich Institute of Biotechnology [42 Sichou Road, Haimen, Jiangsu, China] and was resuspended in DMSO at a stock concentration of 50 mM and 10 mg/ml, respectively. PD98059, the caspase inhibitor zVAD-fmk, SP600125 were purchased from Beyotime Institute of Biotechnology [42 Sichou Road,Haimen,Jiangsu,China] and was resuspended in DMSO at a stock concentration of 40 mM, 42 mM and 30 mM respectively. MG132 and Wortmannin were purchased from Selleck and were resuspended in DMSO at a stock concentration of 10 mM and 2 mM respectively. All inhibitors were stored in small aliquots at Ϫ20°C.
Induction of Acute Apoptosis by Oncogene Inactivation-Cells were plated in Dulbecco's modified Eagle medium containing 10% fetal bovine serum and used for experimentation when they had reached a confluence of 80%. Acute apoptosis was induced in each cell line of our cell line panel. At various times after induction of acute apoptosis, the cells were harvested and the activation state of the relevant signal transducers was assessed by Western blotting.
Immunoblot Analysis-Cells were washed with ice-cold PBS after treatment and lysed on ice using RIPA. After centrifugation at 13,000 ϫ g for 15 min, the protein concentration of the supernatant was analyzed using a BCA assay kit. For the analysis of protein contents, 30 g of total protein was electrophoresed on a 10% SDS-PAGE gel, transferred to nitrocellulose membranes (Millipore). Membranes were blocked with 5% nonfat milk in TBS and 0.05% Tween20, and were probed with the appropriate primary antibody for 1 h at RT. Subsequently, membranes were washed in TBS with 0.05% Tween 20, and then incubated with secondary antibody conjugated to horseradish peroxidase. Proteins were visualized by enhanced chemiluminescence (Millipore). GAPDH was used as loading controls in Western blots. Images were digitalised with imageScanner (Amersham Biosciences) at 300 dpi resolution.
Cell Survival Measurement-Cells were incubated with the inhibitors for the indicated times. Subsequently, all cells (adherent and floating) were collected at 1500 rpm for 5 min., then washed with PBS. The pooled cell pellets were resuspended and mixed with 2% trypan blue dye and incubated for 5 min. The numbers of trypan blue negative cells and total cells were counted using a hemocytometer. Viability was expressed as a percentage of the total number of cells counted. To assess chromatin condensation, cells were incubated with 2 g/ml Hoechst 33342 for 20 min at 37°C and visualized with a fluorescent microscope. At least 200 cells were scored from each sample and data were expressed as the percentage of cells with condensed chromatin.
DNA Laddering-Cells were harvested by trypsinization, washed in PBS then suspended in TNE (10 mM Tris-Cl pH 8.2, 400 mM NaCl, 2 mM EDTA (ethylenediaminetetraacetic acid)) at a concentration of 5 ϫ 10 6 per ml. Cells were lysed by addition of 0.5% SDS and proteins digested by adding 0.25 mg/ml proteinase K and incubating at 37°C overnight. The lysates were extracted with water saturated phenol then chloroform and precipitated with 1.5 volumes of 100% ethanol and spun for 5 min in a microfuge. After a 70% ethanol wash the resulting nucleic acids were electrophoresed through a 1.5% agarose gel containing ethidium bromide. The gels were treated with 20 g/ml RNase A for 3 h at 37°C before visualization by UV transillumination.
2-D SDS-PAGE and MALDI-TOF/TOF Protein Identification-2-D SDS-PAGE and MALDI-TOF/TOF protein identification were performed as published previously (14). Briefly, cells were scraped and harvested, then grinded with a pestle thoroughly with 100 l extraction solution containing 7 M urea, 2 M thiourea, 4% CHAPS, 60 mM dithiotreitol, 1% IPG buffer (containing carrier ampholytes, pH 3-10, Amersham Biosciences), 1 mM PMSF, 2% protease inhibitor mixture (Sigma, St Louis, MO) and 1.5% Triton X-100. The mixture was centrifuged at 14,000 rpm at 4°C for 15 min to remove cell debris. The supernatant was collected as extracted protein and the concentration was determined by the 2-D Quant Kit (Amersham Biosciences) with BSA as standard, and repeated three times to guarantee accuracy. Aliquots of protein samples were stored at Ϫ80°C until further use.
with non-linear pH ranges of 3-10 were used for protein separation according to the manufacturer's instructions. Five hundred milligrams protein were mixed with a rehydration buffer containing 7 M urea, 2 M thiourea, 2% CHAPS, 60 mM DTT, 1% IPG buffer (pH 3-10, Amersham Biosciences), 1.5% Triton X-100 and a trace of bromphenol blue (Amerco, OH, USA) to a total volume of 340 l, then the mixtures were loaded onto the IPG strip. After rehydration for 12 h at 30 V, IEF was carried out using the following conditions: (1) 400 V, 1.5 h; (2) 1000 V, 0.5 h; (3) 3000 V, 1 h; (4) 5000 V, 1 h; (5) 8000 V, 12 h. Before the second dimension separation, strips were equilibrated in two successive buffers containing 6 M urea, 2% SDS, 50 mM Tris-HCl (pH 8.8), and 30% glycerol (Amresco). The first buffer also contained 1% w/v dithiotreitol, the second contained 2.5% w/v iodoacetamide. Each equilibration was for 15 min with continuous agitation with 10 ml buffer per IPG gel. Strips were then rinsed in electrophoresis buffer (25 mM Tris base, 192 mM glycine, and 0.1% w/v SDS) and applied to 12.5% acrylamide gels and sealed with melted agarose (0.5% w/v agarose in electrophoresis buffer containing a trace of bromphenol blue). Electrophoresis was carried out using an Ettan Dalt II apparatus (Amersham Biosciences), with initial separation at a constant 10 mA current per gel for 30 min followed by 30 mA current per gel until the dye front had migrated to the bottom at 25°C. The total run time was about 5 h. After electrophoresis, proteins on preparative gels were visualized by staining with Coomassie brilliant blue G350 staining solution (containing 0.1% w/v Coomassie brilliant blue G350 dye, 10% v/v acetic acid) heated to 95°C in advance for 15 min, then gels were destained briefly in 10% acetic acid two times each for 15 min and then destained overnight. Coomassie brilliant blue G350 stained gel images were digitalized with imageScanner (Amersham Biosciences) at 300 dpi resolution; image analysis was conducted with ImageMaster 2D Platinum software 5.0 (Amersham Biosciences). For normalization, the volume for each spot in a gel was divided by the total volume of all spots, and spot intensities were expressed as a percentage of the total sum of spot volumes. Image analysis was performed comparing the quantity of matched spots in the experimental gels, only those spots that had statistical significance in differential expression were selected for further MS identification.
Protein spots of interest were excised manually and digested with sequencing grade modified porcine trypsin (Promega Madison, WI). Briefly, gel pieces were washed three times in 100 mM bicarbonate buffer before destaining in potassium ferricyanate, and then were dehydrated in acetonitrile and dried in a speed-vac before being rehydrated on ice for 40 min in trypsinization buffer (15 ng/ml trypsin, 25 mM NH 4 HCO 3 , pH8.0). Proteins were digested overnight at 37°C. Peptides were extracted with 20 mM NH 4 HCO 3 , acetonitrile and 0.1% trifluoroacetic acid (v/v) sequentially, and then lyophilized and redissolved in 0.1% trifluoroacetic acid. After desalting with Millipore ZIP plate (Millipore Corporation, Milford, MS), samples were dissolved in 5 mg/ml a-cyano-4-hydroxycinamic acid matrix in 50% acetonitrile, 0.1% trifluoroacetic acid, and subjected to MALDI-TOF/TOF mass spectroscopy on a ABI 4700 Proteomics Analyzer MALDI-TOF/TOF mass spectrometer (Applied Biosystems) operating in a result-dependent acquisition mode. Peaklist-generating and peak-picking for MS and MS/MS spectra were conducted with GPS explorer TM software V.3.0. Peptide mass maps were acquired in reflectron mode (1 KeV accelerating voltage), with 1000 laser shots per spectrum. Six external standards (mass standard kit for the 4700 proteomics analyzer calibration mixture, part No. 4333604, Applied Biosystems) were used to calibrate each spectrum to a mass accuracy within 50 ppm. The MS/MS data were acquired with stop conditions so that 3000 -7000 laser shots were accumulated for each spectrum. The MS together with MS/MS spectra were searched against the international protein index human database v3.10 (57478 sequences; 25254519 residues) using GPS explorer TM software V.3.0 and MASCOT 2.0 software (Matrix Science). Searches were performed without restriction of protein molecular mass (M r ) or pI, and with mandatory carbamidomethylation of cysteines and variable oxidation of methionine residues. One trypsin misscleavage was allowed. Peptide mass tolerance and fragment mass tolerance were set to 150 ppm and Ϯ0.4 Da respectively. Probability scores in MASCOT search above 53 are defined as significant. The MASCOT search results were available as supplementary data.
CHX Chase Assay-Equal numbers of A431 or MiaPaCa-2 cells were seeded into 12-well plates and allowed to attach for 18 h. The cells were treated with CHX (20 g/ml) and harvested at indicated time points. The cell lysates were prepared in SDS sample buffer directly, and an equal amount of lysates was analyzed by immunoblotting. To examine the effect of PI3K inhibition, 50 M LY294002 were added at the same time with CHX.
RNA Interference-RNA interference transfections were conducted using N-TER TM Nanoparticle siRNA Transfection System (Sigma-Aldrich) and indicated predesigned siRNAs (final concentration, 20 nM) from Sigma-Aldrich. For RNA interference experiments, the transfection mixtures were prepared in 100l Nanoparticle Formation Solution (NFS) by adding 2 l of N-TER peptide and either 3.25 l 5 ⌴ nonspecific siRNA (Sigma-Aldrich negative control siRNA) or indicated three PSMD11 and three RPS3a siRNAs (Sigma-Aldrich siRNA I. D. numbers SASI_Hs01_00123189, SASI_Hs01_00123190, SASI_Hs01_00123193, SASI_Hs02_00305064, and RPS3a1 and RPS3a2), mixed, and incubated at room temperature for 20 min. The transfection mixtures were then diluted with serum-containing medium to a final siRNA concentration of 20 nM, and transferred to 24-well plates with 600 l/well. The cells were then placed at 37°C, 5% CO 2 for the indicated times before harvest for immunoblot analysis, or assessing chromatin condensation. Each siRNA consisted of 19-nucleotide double-stranded RNA with two 3Ј-dT overhangs on each strand. siRNA sense sequences and Sigma-Aldrich ID numbers were as follows: PSMD11 ID SASI_Hs01_00123189, GCAUUUGAGGGUUAU-GACUdTdT, SASI_Hs01_00123190, CAGAAGAUGUCCAGGCUUUd-TdT, SASI_Hs01_00123193, CUGACAUAGAGUUGGAUCUdTdT, RPS3a ID SASI_Hs02_00305063, GUCUCAAGGGUCGUGUGUUdTdT, RPS3a 1, GUGCUAAAG UUGAACGAGC, RPS3a 2, GGAUCUUAC-CCGUGACAAA, Negative Control, UUCUCCGAACGUGUCACGUTT.

RESULTS
Inhibiting PI3K Activity Could Induce Acute Apoptosis in A431 Cells-PI3K signaling pathway is a critical regulator of many cellular processes that promote the transformation of a normal cell to a cancer cell. Suppression of PI3K activity with various PI3K inhibitors have been shown to inhibit proliferation and induce apoptosis in a variety of cancer cells through G0/G1 cell cycle arrest (15)(16)(17)(18)(19)(20). Here, however, we showed that inhibiting PI3K activity with 10 -60 M LY294002 or 100 nM-1 M Wortmannin could induce more than 70% of A431 cells to die within 6 h (Figs. 1A-1E) in regular medium containing 10% fetal bovine serum, and apoptosis-specific fragmented cytosolic DNA and cleaved PARP can be detected as early as 2 h after LY294002 addition (Figs. 1F, 1G), demonstrating that inhibiting PI3K could induce A431 cells to die through apoptosis. The rapid and extensive apoptosis of A431 cells after PI3K inhibition indicated that the survival of A431 cells was absolutely dependent on the PI3K signaling pathway, and it could represent an excellent example of oncogene addiction for the oncogene of PI3K. In addition, the FIG. 1. Inhibiting PI3K signaling with LY294002 could induce acute apoptosis in A431 cells. A, B, A431 cells were incubated with LY294002 or Wortmannin for the indicated time at the indicated concentration ranging from 5-60 ⌴ and 50 nM-1⌴ respectively, and survival cells were scored by trypan blue dye exclusion test. C, D, A431 cells were incubated with LY294002 or Wortmannin for 1 h or 2 h at the indicated concentrations ranging from 5-60 ⌴ and 50 nM-1 ⌴ respectively, Cells were harvested and their proteins were analyzed by Western blotting using antibodies directed against PARP(treated for 2 h), phospho-Akt(Ser473), phospho-Akt(Thr308), and pan-Akt(treated for 1 h). The relative migration of the relevant proteins are indicated on the left-hand side of each autoradiogram and to the right of each autoradiogram are indicated the relative migration of molecular weight standards. E, Time course of acute apoptosis of A431 cells photographed at the indicated times after addition of 50 M LY294002. F. A431 cells were left untreated or treated with 50 M LY294002 for the indicated times. Cytosolic nucleic acids were isolated and DNA fragmentation assessed by gel electrophoresis. G, A431 cells were left untreated or treated with 50 M LY294002 for the indicated times and the cells were harvested and their proteins were analyzed by Western blotting using antibodies directed against PARP and GAPDH. extensive apoptosis of A431 cells also indicated that it was not associated with G0/G1 or G2/M cell cycle arrest, instead it could occur from all phases of the cell cycle. For similar phenomenon of this kind of rapid apoptosis have been observed in multiple cancer models following inactivation of the addicted oncoproteins such as Src, BCR-ABL and EGFR (21), or following treatment with other agents (13,22), and has been coined as "acute apoptosis" by Alan Eastman (13) to distinguish it from the delayed apoptosis, which is associated with cell cycle arrest, we hypothesized that "acute apoptosis" might be a general and widespread mechanism, which was responsible for the marked death of cancer cells following inactivation of the addicted oncogenes.
To understand the mechanism of acute apoptosis occurred in A431 cells on PI3K inhibition, we first investigated the effects of PI3K inhibition on signaling downstream of PI3K, then characterized whether the onset of acute apoptosis in A431 cells was dependent on the down-regulation of one or set of specific regulatory protein(s) by proteasome such as Mcl-1, activation of JNK(c-Jun-NH 2 -terminal kinase) or activation of caspases. As shown in Fig. 2A, treatment of A431 cells with 50 M LY294002 induced rapid and time-dependent apoptosis in Ͼ95% of the cells within 6 h, and apoptosisspecific cleavage of the caspase substrate PARP could be detected as early as 1 h after LY294002 addition. In the absence of LY294002, A431 cells exhibited easily detectable phosphorylation of Akt and GSK3␤, inhibition of PI3K with LY294002 resulted in dramatic decline in GSK3␤ phosphorylation but not in Akt phosphorylation. Conversely, phosphorylation of Akt appeared to have increased modestly, especially when A431 cells were treated with 5, 10, and 20 M LY294002 (Fig. 1A). Therefore the effect of PI3K to maintain the survival of A431 cells seemed not to be dependent on the activity of AKT, which was consistent with the findings of Faber AC (23) and Vasudevan KM (24). Moreover, it was found that the general caspase inhibitor zVAD-fmk could completely suppress acute apoptosis through 6 h after addition of 50 M LY294002, the proteasome inhibitor MG132 could only partially reverse acute apoptosis, however, the JNK inhibitor SP600125 has no effect on the acute apoptosis. These data indicated that the rapid apoptotic response of A431 cells following acute inhibition of PI3K may be caused by loss of some short-lived proteins whose activity normally maintains cell survival by blocking caspases activation directly or indirectly.

Time Course Analysis of Differently Expressed Proteins During the Early Phase of Acute Apoptosis in A431 Cells by 2-DE and MALDI-TOF/TOF Protein Identification-To identify the
short-lived proteins that may play important role in inhibiting acute apoptosis, time course two-dimensional PAGE was run in the early phase (0 -2 h) of acute apoptosis in A431 cells on IPG strips of pH 3-10 and analyzed by ImageMaster 2D Platinum software 5.0 (Amersham Biosciences). The reproducibility of samples was examined at each of the time points after treatment respectively. In total, 8 gels, containing at least two replicates for each sample, were compared with each other; After statistical analysis of the normalized quantities of matched spots, 17 proteins, which were all members of the central proteome of human cells (25), were identified to be down-regulated significantly with intensity change greater than twofold (Fig. 3 and Table I) from the about 1100 protein spots we have aligned. Among of them, six proteins including eIF3g, eIF-2-alpha, EF-1-alpha-1, EF-1-gamma, EF-1-delta, and AspRS have been shown to be involved in protein synthesis (26 -30), two proteins including RPS3a and GRP-78 have been shown to be able to inhibit apoptosis through various mechanism (31, 32), and two proteins including PSMD11 and TXNL1 are components of the proteasome that have been shown to play a key role in regulating proteasome activity (33,34), the other five proteins also play important role in pathways such as tumor invasion and migration(RhoC) (35) protein modification (LS) (36), cell proliferation (PCNA) (37), macromolecular complex assembly (TCP-1-alpha) (38), and pre-mRNAs processing (hnRNP H and hnRNP H3) (39,40), indicating that these vital processes of the cell, such as proteostasis, primary metabolism, cell cycle and death, are all under the control of PI3K in A431 cells through maintaining the expression level of these short-lived central proteome members.
To confirm their change of expression, we selected PSMD11 and RPS3a for further confirmation by time-course immunoblotting in the early phase(0 -2 h) of acute apoptosis in A431 cells. In addition, for Mcl-1 have been reported to be able to block apoptosis in several cancer cell lines (13,22), its change of expression was also determined during the early phase of acute apoptosis in A431 cells. As shown in Fig. 2B, RPS3a, PSMD11, and Mcl-1 could all be detected easily in A431 cells. When the activity of PI3K was inhibited with 50 M LY294002, the level of Mcl-1 and PSMD11 all decreased quickly to basal level just before the onset of apoptosis. However, the level of RPS3a showed first increase then decrease before the onset of apoptosis. Inhibition of caspases with zVAD-fmk could not rescue the expression of RPS3a, PSMD11, and Mcl-1, indicating that the down-regulation of RPS3a, PSMD11, and Mcl-1 is a cause but not a consequence of apoptosis. Consistent with its partial suppression of acute apoptosis in A431 cells, inhibition of the proteasome activity with MG132 could only block the loss of Mcl-1 but not PSMD11 and RPS3a, implicating that the rapid down-regulation of these short-lived proteins we have identified may play important role in acute apoptosis.
Furthermore, these data also indicated that maintaining the expression level of these short-lived members of central proteome may be a primary mechanism for addicted-oncogenes to maintain the survival of cancer cells.
To demonstrate that these proteins are indeed short-lived proteins, we determined the half-life of RPS3a and PSMD11 with CHX chase experiments in A431 cells. A431 cells were treated with a protein synthesis inhibitor, CHX, alone or in combination with the PI3K inhibitor LY294002, and the endogenous RPS3a and PSMD11 were monitored for 2.5 h. As shown in Fig. 4A, RPS3a and PSMD11 degradation all occurred rapidly with a half-life ϳ2 h. When CHX was combined with LY294002, the half-life of RPS3a and PSMD11 was shortened to 1 h and Ͻ0.5 h respectively, confirming that RPS3a and PSMD11 are indeed short-lived proteins and PI3K could inhibit the degradation of RPS3a and PSMD11 directly through an unknown mechanism.
Acute Apoptosis in a Panel of Pancreatic Cancer Cell Lines-To further verify the findings above, we next investigated whether acute apoptosis could also be induced a panel of pancreatic cancer cell lines, including MiaPaCa-2, BxPC-3, Panc-1, CAPAN-2, and CFPAC-1 cells, which all harbor activating mutations in K-ras except for BxPC-3 cells (41,42). As mutationally activated K-Ras was generally regarded as the addicted oncogene in pancreatic cancer cells and simultaneous inhibition of its two downstream signaling pathways including PI3K and mitogen-activated protein kinase/ERK1/2 (43)(44)(45) has been shown to induce cell cycle arrest and apoptosis in pancreatic cancer cells (46), we selected PI3K inhibitor LY294002 and MEK1/2 inhibitor PD98059 to treat these pancreatic cell lines separately or in combination to see whether acute apoptosis could also be induced. Moreover, as these protein we have identified are all short-lived proteins, an alternative approach to suppress their expression is to inhibit protein synthesis. Accordingly, we also investigated whether inhibiting protein synthesis with CHX alone or in combination with LY294002 and/or PD98059 could induce acute apoptosis in these pancreatic cancer cell lines.
As shown in Fig. 5, among the five pancreatic cancer cell lines, acute apoptosis was only observed in MiaPaCa-2 cells when treated with LY294002ϩPD98059ϩCHX and LY294002ϩ CHX combinations with 70 and 52% of the cells dying within 6 h respectively, and the apoptosis-specific cleaved PARP can be detected as early as 4h. The other treatments including LY294002ϩPD98059, LY294002, PD98059, PD98059ϩCHX, and CHX only showed marginal apoptosis after treatment for 6 h although weak apoptosis-specific cleaved PARP could also be detected at 4h after treatment except for treatment with PD98059. Correspondingly, quick degradation of PSMD11 and Mcl-1, and similar first increase and then decrease of RPS3a could be observed following treatment with LY294002ϩPD98059ϩCHX, but not following treatment with LY294002ϩPD98059, which, on the contrary, increased the expression of PSMD11. Similarly, treatment with LY294002   . 4. RPS3a and PSMD11 are short-lived proteins and PI3K could inhibit the degradation of RPS3a and PSMD11 in both MiaPaCa-2 and PANC-1 cells. A, Time course of RPS3a and PSMD11 degradation in A431 cells. A431 cells were treated with 20 g/ml CHX alone or in combination with 50 M LY294002, and the endogenous RPS3a and PSMD11 were monitored for 2.5 h and ␤-actin was used as the loading control. The half-life of RPS3a and PSMD11 was calculated to be 2 h. When CHX was combined with LY294002, the half-life of RPS3a and PSMD11 was shortened to 1 h and Ͻ0.5 h respectively B, Time course of RPS3a and PSMD11 degradation in MiaPaCa-2 cells. MiaPaCa-2 cells were treated with 20 g/ml CHX alone or in combination with 50 M LY294002, and the endogenous RPS3a and PSMD11 were monitored for 4 h and ␤-actin was used as the loading control. The half-life of RPS3a was calculated to be 2 h, and PSMD11 showed a half-life ϳ1 h with de novo synthesis. When CHX was combined with LY294002, accelerated degradation of both RPS3a and PSMD11 could be seen despite the obvious de novo synthesis of both proteins.
suggesting that protein synthesis cannot be completely inhibited by cycloheximide (22), the mechanism and role for the induction of PSMD11 by inhibition of protein synthesis deserves further investigation.
To investigate whether the turn-over of RPS3a and PSMD11 are also controlled by PI3K in MiaPaCa-2 cells, we measured the half-life of RPS3a and PSMD11 with CHX chase experiments. The same as that in A431 cells, MiaPaCa-2 cells were treated with 20 g/ml CHX, alone or in combination with the 50 M LY294002, and the endogenous RPS3a and PSMD11 were monitored for 4 h. As shown in Fig. 4B, when MiaPaCa-2 cells were treated with 20 g/ml CHX, RPS3a was found to be degraded quickly with a half-life of ϳ2 h, the same as that in A431 cells. However, PSMD11 showed first decrease then increase with a half-life ϳ1 h. When CHX was combined with LY294002, the degradation rate of RPS3a and PSMD11 was all accelerated despite the obvious de novo synthesis of both proteins. Confirming that PI3K could inhibit the degradation of both RPS3a and PSMD11 directly in MiaPaCa-2 cells either, but the detailed mechanism needed to be further studied.
These findings above indicated that the onset of acute apoptosis of MiaPaCa-2 cells was also associated with the down-regulation of these short-lived members of the central proteome, but it seemed that only simultaneous inhibition of PI3K and/or MEK1/2 and protein synthesis could induce these proteins to undergo quick down-regulation, indicating that the expression level of these short-lived members of the central proteome are controlled by different signaling pathways in MiaPaCa-2 cells by both inhibition of degradation and maintaining protein synthesis.
As in A431 cells, inhibition of PI3K could decrease the phosphorylation of GSK3␤ but not of AKT in MiaPaCa-2 cells, suggesting again that AKT is dispensable for the activity of PI3K to suppress acute apoptosis. As expected, inhibition of MEK1 with PD98059 could suppress phosphorylation of ERK1/2 markedly. For treatments with PD98059 and PD98059ϩCHX could also decrease the expression of Mcl-1, it was thought that Mcl-1 might not be a determinant factor for the onset of acute apoptosis in MiaPaCa-2 cells.
In further support of the findings above, the other four pancreatic cancer cell lines in which acute apoptosis could not be induced by all the treatments, including BxPC  CAPAN-2, and CFPAC-1 cells. A, C, E, G, BxPC-3, PANC-1, CAPAN-2  when it combined with PI3K, MEK1/2, and protein synthesis inhibition, acute apoptosis could be partially induced in both MiaPaCa-2 and PANC-1 cells by siRNA 189 and 190 against PSMD11 as assessed by PARP cleavage and chromatin condensation (Figs. 7A, 7B). These experiments confirmed that some of the short-lived members of central proteome had important role in acute apoptosis, and they may be novel therapeutic targets for pancreatic cancer or other cancers.
In conclusion, these findings above support the hypothesis that the rapid apoptotic response of cancer cells following acute inhibition of the addicted oncogenes is caused by loss of multiple short-lived proteins whose activity normally maintains cell survival by blocking caspases activation directly or indirectly.

DISCUSSION
Oncogene addiction has now been well documented in multiple experimental cancer models and appears to play an important role in the clinical response to various cancer targeted agents that have been recently developed. However, the exact molecular mechanism underlying the phenomenon of oncogene addiction is still not clear. In this study we have tested a specific hypothesis that the rapid apoptotic response following acute inhibition of the addicted oncogenes in cancer cells are because of loss of multiple short-lived proteins whose activity normally maintains cell survival by blocking caspase activation directly or indirectly. We showed that rapid apoptosis or acute apoptosis could be induced in both A431 cells and pancreatic cancer MiaPaCa-2 cells when treated with appropriate signaling inhibitors, and proteomic analysis showed that the quick down-regulation of 17 short-lived proteins, which are all members of the central proteome of human cells, was associated with the onset of acute apoptosis in both A431 and MiaPaCa-2 cells. Knockdown of PSMD11 but not RPS3a via RNA interference could promote the occurrence of acute apoptosis in both MiaPaCa-2 and PANC-1 pancreatic cancer cells. These observations suggest that the quick loss of some of the short-lived members of the central proteome may be the direct reason for the rapid or acute apoptosis following acute inhibition of the addicted oncogenes in cancer cells.
The human central proteome, which was first coined by Schirle et al. (25), refers to a collection of rather well conserved proteins (ϳ1124 proteins) commonly expressed by human cells that are mainly involved in vital processes of the cell, such as proteostasis, primary metabolism, cell cycle, and death. Because of the fact that all of the short-lived proteins we identified during the early phase of acute apoptosis in A431 cells are members of the central proteome of human cells (47), and they all have important role in multiple vital processes of the cell, we suppose that the protein levels of these short-lived members of the central proteome may all be controlled by the addicted oncogenes in cancer cells through suppressing protein degradation and maintaining protein syn-thesis or other mechanism(a state of "bizarre" (5)). When the addicted oncogenes in cancer cells were targeted with "rationally targeted" agents, these crucial members of central proteome will be lost quickly through various mechanism(and this may be a kind of short-lived pro-survival signals (48)), then the constitutively expressed pro-apoptotic signals will be released, then acute apoptosis will occur. However in normal cells, the protein levels of these short-lived members of the central proteome may be under the control of different kinases or signaling pathways, targeting the addicted oncogenes in cancer cells will not have or only have marginal effect on their expression in normal cells, therefore acute apoptosis will not occur. This may account for the less toxicity of these rationally targeted agents compared with conventional chemotherapeutic agents.
Acute apoptosis was first coined by Alan Eastman (13) to distinguish it from the delayed apoptosis induced by vinblastine alone. The characteristic of acute apoptosis is that it can be triggered within hours (Ϸ6h) and it can occur from all phases of the cell cycle, yet the ordinary apoptosis usually undergo cell cycle arrest before the onset of apoptosis and need more time than acute apoptosis. The quick and massive death of A431 and MiaPaCa-2 cells following treatment with corresponding signaling inhibitors in this study indicated that they all die through acute apoptosis. For similar phenomenon of this kind of rapid apoptosis have been observed in multiple cancer models following inactivation of the addicted oncoproteins such as Src, BCR-ABL, and EGFR (21), or following inhibition of protein synthesis (22), we suppose that "acute apoptosis" may be a general and widespread mechanism, which is responsible for the marked death of cancer cells following inactivation of the addicted oncogenes, and it should be regarded as a standard or goal when designing or selecting appropriate therapeutic strategies for individual cancer patients, for if acute apoptosis cannot be induced, it may indicate that the "Achilles Heel" of the cancer has not been identified.
The studies we have performed also have clinical implications regarding the use of cancer therapies that target oncogenic kinases. For the expression level of these short-lived members of central proteome is usually controlled by a single addicted oncogene in cancer cells through both suppressing their degradation and maintaining protein synthesis through different signaling pathways, only targeting certain downstream pathways, such as PI3K/Akt signaling pathway, may not be sufficient to induce these short-lived proteins to lose rapidly if the addicted oncogene in certain cancer patient is not successfully targeted, therefore co-administration of drugs that target multiple critical signaling pathways including protein synthesis may improve the therapeutic effect of these new molecularly targeted agents. In addition, for acute apoptosis can be triggered within hours, it can help us choose appropriate molecular targeted agents for individual cancer patients, for if a certain molecular targeted drug cannot in-duce acute apoptosis in a certain patient, it may indicate that other drugs should be considered. In summary, we believe that we have identified a significant mechanism for the phenomenon of oncogene addiction, and it may have important implications for the targeted therapy of cancer.