A Novel Interaction between Procaspase 8 and SPARC Enhances Apoptosis and Potentiates Chemotherapy Sensitivity in Colorectal Cancers*

Chemotherapy resistance accounts for the high mortality rates in patients with advanced cancers. We previously used a genomics approach to determine novel genes associated with this phenomenon and identified secreted protein acidic and rich in cysteine (SPARC) as a chemosensitizer capable of reversing therapy resistance in colorectal cancer cells by enhancing apoptosis in vitro and tumor regression in vivo. Here, we examined the mechanisms by which SPARC enhances apoptosis in the presence of chemotherapy. We show that SPARC potentiates apoptosis by augmenting the signaling cascade in a caspase-8-dependent manner, because apoptosis can be abolished by caspase 8 small interfering RNA in the presence of SPARC. This occurs independently of death receptor activation and leads to downstream involvement of Bid and subsequent apoptosis. Interestingly, this results from an interaction between SPARC and the N terminus of the procaspase-8 DED-containing domain. These exciting findings provide an initial map of the apoptosis signaling events mediated by SPARC and how this can ultimately result in the reversal of chemotherapy resistance and enhanced tumor regression. This signaling cascade can be exploited therapeutically and may have potential clinical implications for patients with advanced and therapy-refractory cancers.

High mortality rates continue to be observed in advanced colorectal cancers (CRC) 3 because of therapy-refractory disease. In the last decade, the use of newer agents, such as irino-tecan (CPT-11) (1, 2) and oxaliplatin (3,4) have led to significant improvements in response rates, from ϳ20% to ϳ50%. However, the median survival has remained poor at Ͻ17 months, partly because of chemotherapy resistance. Therefore, novel therapies are needed that can resensitize therapy-unresponsive tumors to chemotherapy to achieve better overall survival in patients with advanced cancers. SPARC (secreted protein acidic and rich in cysteine) is a secreted matricellular glycoprotein (5) involved in development, remodeling, and tissue repair (6). It inhibits cell proliferation, adhesion, and cell cycle progression (7). Differential expression of SPARC has been observed in various human cancers, and it is unclear why it has variable effects on tumor growth in different tissues (8). However, higher SPARC expression has been observed in human CRCs that are sensitive to chemotherapy, in comparison with therapy-refractory tumors (9). SPARC appears to function as a tumor suppressor in ovarian (10), pancreatic cancers (11), and acute myeloid leukemia cells (12). Moreover, in tumor xenograft models, greater growth of pancreatic cancer cells are observed in SP Ϫ/Ϫ mice in comparison with wild type SP ϩ/ϩ (11). One proposed mechanism responsible for retarding the growth of tumors is its ability to enhance apoptosis. This has been demonstrated in studies where exogenous exposure to SPARC resulted in enhanced apoptosis in ovarian cancer cells (10), whereas its absence endogenously diminished this event (11). Our own laboratory investigated the chemosensitizing properties of SPARC and found it to induce apoptosis in the presence of significantly lower concentrations of chemotherapy in CRCs overexpressing SPARC (9). The signaling events involved in SPARC-mediated apoptosis are the focus of the current study.
The effectiveness of chemotherapy can be assessed by its ability to enhance tumor cell death. This can be accomplished by activating signaling pathways involved in cell death (13). We know that SPARC is capable of promoting tumor regression by activating apoptosis, which may explain its ability to resensitize therapy-refractory cancer cells to chemotherapy. A variety of stimuli can trigger apoptosis and two major signaling pathways, "extrinsic" and "intrinsic," converge biochemically leading to its execution (14 -16). The extrinsic pathway is triggered by the activation of death receptors, such as Fas; the tumor necrosis factor-related apoptosis-inducing ligand death receptors, death receptors 4 or 5; or tumor necrosis factor receptor, following binding with their natural ligands (17). This recruits adaptor proteins, such as Fas-associated death domain (FADD), which recruits pro-caspase 8 to form death-inducing signaling complexes (DISCs) (18,19). Caspase 8 is activated at DISCs, leading to downstream pro-apoptotic events (20). Activation of the initiator caspases 8 and 10 of the extrinsic pathway is associated with the release of its prodomain from catalytically active fragments, which is also observed in other caspases involved in the apoptotic cascade (21,22). The intrinsic pathway is centered around the mitochondria, which is key in regulating the balance between pro-and anti-apoptotic factors, such as anti-apoptotic members Bcl-2, Bcl-XL, and pro-apoptotic members Bax, Bak, and Bok (23). It can be triggered by a number of stimuli, including agents that cause DNA damage (24) or growth factor deprivation (25). This leads to the permeabilization of the mitochondrial membrane and the release of cytochrome c into the cytosol (26), which then interacts with APAF-1 to recruit caspase 9, resulting in cleavage of executioner caspases and apoptosis (27). The convergence of the extrinsic and intrinsic pathways occur when caspase 8 activates Bid, a Bcl-2 family member that can trigger downstream targets to initiate the intrinsic apoptotic pathway (28).
Our earlier findings that SPARC enhances chemosensitivity by increasing apoptosis led us to investigate the signaling events involved in SPARC-mediated apoptosis in this study. Here, we examined the involvement of SPARC in relation to the extrinsic pathway of apoptosis and showed that there was enhanced activation of this pathway when CRC cells were exposed to chemotherapy in the presence of high levels of SPARC. Even more interesting is our finding of an interaction between SPARC and pro-caspase 8 that augments apoptosis in cancer cells, which begins to explain the ability of SPARC to reverse chemotherapy resistance.
RT-PCR-The cells were seeded at 150,000 cells/well in 6-well plates. After 24 h, the cells were incubated with 1000 M 5-FU for 4 h, and RNA was isolated with TRIzol (Invitrogen) (9). 1 g of total RNA was used to generate cDNA (Superscript III, Invitrogen). To assay for mRNA stability, the cells were incubated with actinomycin D (10 g/ml) (Sigma) for 0 -20 h, and RNA was isolated for RT-PCR. Specific primers were used as previously described for SPARC and glyceraldehyde-3-phosphate dehydrogenase (9), and others are listed in supplemental Table S1. PCR products were separated on a 1.5% agarose gel electrophoresis, ImageJ was used for quantitation of expression levels (9), and the values were normalized to ␤-actin.
Chromatin Immunoprecipitation (ChIP) Assay-Monolayers of MIP/ZEO and MIP/SP cells were fixed by adding formaldehyde to a final concentration of 1% for 10 min at 37°C. The cells were rinsed and collected in 1ϫ phosphate-buffered saline. The cell pellets were resuspended in 200 l of SDS lysis buffer and lysed on ice for 10 min, followed by sonication, and centrifugation at 13,000 rpm for 5 min. The supernatant was diluted in ChIP buffer. The cell lysates were precleared, and input fractions were put aside. Immunoprecipitations were carried out using IRF-1 and IRF-2 antibodies (Santa Cruz) with overnight incubation at 4°C. 100 l of protein A-Sepharose beads (Sigma) were added and incubated for 1 h at 4°C and washed with low salt wash buffer and TE buffer. The beads were extracted with 50 l of elution buffer, and 15 l of 1 M NaHC0 3 was added to the input samples. Eluates and input fractions were incubated overnight at 65°C. DNA was purified with PCR purification kit (Qiagen) and used for PCR as described above. All of the buffer constituents are described in supplemental Table S2.
RNA Interference-Initially, to assess the efficiency of caspase 8 gene expression knock-down by siRNA, MIP/SP and HCT116 cells were seeded (6-well plate). 24 h later, the cells were transiently transfected with 20 -60 nM scramble oligonucleotide sequence (control) or caspase 8 siRNA (Stealth RNAi, Invitrogen), and the cells were collected at various time intervals following transfection. 40 nM of siRNA yielded the most efficient knock-down (14-fold decrease in caspase 8 expression at 48 -96 h; supplemental Fig. S1). For all subsequent experiments, 40 nM of siRNA or scramble control was used, and transient transfections were carried out for at least 72 h. Following caspase 8 siRNA transfection, the cells were assessed for cell viability and apoptosis using either Caspase 3/7 assay or TUNEL assay.
TUNEL Assay-The cells were seeded (24-well plates) to achieve ϳ60% confluence 24 h later for transient transfection with caspase 8 siRNA. 36 h later, the cells were incubated with 1000 M 5-FU for 36 h, harvested (suspension and attached cells), and fixed onto glass slides with Shandon cytospin at 2000 rpm for 10 min and stained as per the manufacturer's instructions (Promega). The number of TUNEL-positive cells was counted and averaged from four different fields (n ϭ 4 independent experiments, with slides read independently by two individuals in a blinded fashion).
Immunoblot Analysis-48 h after seeding, the cells were incubated with 1000 M 5-FU and collected at 0 -12 h for protein as previously described (9,29). 40 g of total protein/sample was loaded, separated on a 12% SDS-PAGE, and then trans-ferred to polyvinylidene difluoride membranes (Bio-Rad). Immunodetection was performed using antibodies against caspase 8, cleaved caspase 8, FADD, p-FADD, caspase 10, BID, caspase 9, and caspase 3 (all 1:1000, Cell Signaling Technologies) followed by incubation with the appropriate secondary antibody. All of the immunoblots were also probed with anti- Primer positions (forward-p and reverse-p) and amplification products for ChIP assays are also indicated. E, ChIP assays to detect binding of transcription factors IRF-1 and IRF-2 to caspase 8 promoter. F, assessment of caspase 8 mRNA stability following incubation with actinomycin D for 0 -20 h (PCR, 40 cycles). G and H, changes in the expression of genes involved in apoptosis were assessed following treatment with 5-FU after various timed intervals at the protein level. Total proteins were isolated from cells harvested at 0 -12 h following treatment with 1000 M 5-FU and probed for various proteins shown by immunoblots (in G, activated caspases are represented by their cleaved fragments, with the exception of caspase 10 because antibody does not recognize the cleaved fragment) and also caspase 3/7 activity (I) (all results representative of three or four experiments). *, statistical difference, where p Ͻ 0.05, Student's t test.
Subcellular Fractionation and Immunoprecipitation-MIP/SP and HCT116 cells were grown until ϳ80% confluence, incubated with 1000 M 5-FU, and isolated at 4 h. The cells were separated into nuclear, cytosolic, and membrane fractions using ProteoExtract Subcellular Proteome Extraction Kit (EMD Biosciences Inc.). In addition, to verify the site of interaction of caspase 8 with SPARC, MIP/SP, and MIP/ZEO cells were also incubated with antibodies against caspase 8 targeting its C terminus (Cell Signaling) or N terminus (Abcam) in vitro for 24 h at 1.5-3.0 g prior to collecting and fractionating the cell lysates, for immunoprecipitation, and as well, caspase 3/7 assay (as described previously). 250 g of the individual cellular fractions were incubated with antibodies against SPARC (10 g/ml; Hematologic Technologies), caspase 8 (1:100; Cell Signaling Technology (C terminus) or Abcam (N terminus)), or a nonspecific anti-mouse IgG antibody as control (Cell Signaling Technologies), in phosphate-buffered saline overnight (4°C) with gentle agitation. Protein-antibody mixture was then incubated with 30 l of protein A-protein G (Sigma) (1:1) beads for 4 h (4°C). The proteins were also incubated with EZView Red His-Select HC Nickel affinity gel (Sigma) for immunoprecipitation of His-tagged SPARC protein. For all complexes, the beads were washed five times with phosphate-buffered saline, eluted with 40 l of 2ϫ SDS loading buffer, and used for immunoblotting against SPARC and caspase 8.
Animal Studies-Tumor xenografts harvested from National Institutes of Health nude mice (6 weeks old; Taconic Laboratories) were used for histology, RT-PCR, or immunoblot. 2 ϫ 10 6 MIP101 cells were injected into the left flank. Once tumors reached 100 mm 3 , the mice were treated with chemotherapy using 3-week cycle regimen (ϫ2 cycles) as previously described (9). Experimental groups (2 animals/group) for this study included treatment with: SPARC, SPARC ϩ 5-FU, 5-FU only, and saline. In addition, tumor xenografts of MIP/ZEO and MIP/SP cells from mice treated with either 5-FU (three consecutive days) or saline were collected after the first cycle of treatment and homogenized (Kinematica, POLYTRON-Aggregate). The lysates were then prepared for immunoblot or RT-PCR. All of the mice received care according to standard animal care protocol and guidelines. For histology, tissue sections were processed for immunohistochemistry based on previously established protocols (9). Caspase 8 antibody (1:50) was used and incubated overnight at 4°C and counterstained with DAPI. A Zeiss Axioplan 2 fluorescence microscope was used for image capture.
Statistics-Statistical difference between experimental groups were calculated and analyzed using Student's t test. Statistical significance was defined as p Ͻ 0.05, using Smith's statistical package.

Overexpression of SPARC in MIP101 Cells Augment the Expression of Genes Involved in the Extrinsic Pathway of
Apoptosis-Overexpression of SPARC in MIP 101 CRC cells (MIP/SP) leads to increased sensitivity to 5-FU and CPT-11 chemotherapy by diminishing cell survival and enhancing apoptosis (9). To understand the mechanisms involved in this SPARC-mediated effect, we began by examining the relative contribution of genes involved in apoptosis at the transcriptional level. We noted that caspase 8 and 10 and FADD appeared significantly higher in two different clones of MIP/SP cells (Fig. 1A). The greatest difference in gene expression was observed with caspase 8 and 10, which were ϳ3.4and 5.9-fold higher in MIP/SP than MIP/ZEO, respectively, thereby suggesting that differential expression of SPARC positively influenced genes involved in the extrinsic pathway of apoptosis.
In an earlier study, we observed a greater number of cells undergoing apoptosis when either sensitive MIP101 or 5-FUresistant MIP101 cells (MIP/5-FU) were exposed to SPARC in combination with 5-FU in vitro and in vivo (9). Based on these findings, we proceeded to assess the effect of 5-FU exposure on the extrinsic pathway in cells with variable levels of SPARC (highest in MIP/SP, moderate in MIP/ZEO, and lowest in MIP/ 5-FU) (9). Higher levels of caspase 8 and 10 gene expression were observed in MIP/SP cells, and this increased following exposure to 5-FU 1000 M (Fig. 1B). In cells with low SPARC expression, caspase 8 and 10 were not observed either basally or following exposure to 5-FU in either MIP/ZEO or MIP/5-FU cells. FADD gene expression increased in all cells following treatment with 5-FU. It was interesting to note that by reducing caspase 8 gene expression with siRNA, there was also a demonstrable reduction in FADD gene expression, but no change was observed with caspase 10 (Fig. 1C).
The elevation in caspase 8 mRNA in cells overexpressing SPARC was interesting, yet the mechanisms involved were unclear. The higher basal levels of caspase 8 mRNA indicated a transcriptional up-regulation in SPARC-overexpressing cells as a potential mechanism. Up-regulation of caspase 8 transcription has been shown to occur following binding of transcription factors IRF-1 and IRF-2 to caspase 8 promoter (Fig. 1D) (30 -33), and indeed, we show by ChIP assays that in MIP/SP cells, there was significantly more binding of IRF-1 and IRF-2 to the (Ϫ162/ϩ76) DNA region at the 5Ј-end of caspase 8 gene (   NOVEMBER 23, 2007 • VOLUME 282 • NUMBER 47 JOURNAL OF BIOLOGICAL CHEMISTRY 34461 1E), suggesting a SPARC-mediated effect in the up-regulation of caspase 8 transcripts. We also examined whether some of this effect could be due to a decrease in mRNA degradation by harvesting cells at various timed intervals following incubation with actinomycin D, a RNA polymerase II inhibitor. We noted that in lower SPARC-expressing MIP/ZEO cells, a reduction in basal caspase 8 mRNA levels was detected at 4 h, whereas in two different clones of MIP/SP and the higher SPARC-expressing HCT 116 cells, a reduction in caspase 8 mRNA level only began to occur after 8 -12 h of exposure. This also suggested the involvement of SPARC in improving the stability of caspase 8 mRNA (Fig. 1F).

SPARC Enhances Apoptosis via Caspase 8
Cells Overexpressing SPARC Have Enhanced Activation of the Extrinsic Pathway of Apoptosis-This heightened basal expression of caspase 8 and 10 at the transcriptional level in cells overexpressing SPARC translated to the protein level, where MIP/SP cells again showed an abundance of pro-caspase 8 and 10, in comparison with MIP/ZEO and MIP/5-FU cells prior to exposure to 5-FU (Fig. 1G). Conversion of pro-caspase 8 to its cleaved products occurred following exposure to 1000 M of 5-FU. There were also higher basal levels of Bid in MIP/SP cells, which peaked at 4 h after incubating with 5-FU, followed by a gradual decline over the next 8 h. Interestingly, activation of caspases 9 and 3 was prominently observed in MIP/SP cells following 5-FU incubation and to a lesser degree in MIP/ZEO, and even less in MIP/5-FU cells. FADD was basally expressed in all cell lines; however, significantly greater phosphorylated FADD was seen in MIP/SP as early as 2 h after incubation with 5-FU in comparison with either MIP/ZEO (Fig. 1H) or MIP/ 5-FU cells (Fig. 1G). This observation that MIP/SP cells were more likely to undergo apoptosis following incubation with 5-FU than MIP/ZEO cells was further supported by significantly higher levels of caspase 3/7 activity in MIP/SP cells at 12 h after incubation with 5-FU than in MIP/ZEO cells (16397.0 Ϯ 2787.6 versus 9954.0 Ϯ 1104.8, p ϭ 0.0003) (Fig. 1I).
In the Presence of SPARC, Caspase 8 Decreases Cell Viability in Response to 5-FU-The relative contribution of the extrinsic pathway in SPARC-mediated apoptosis was examined by reducing caspase 8 expression using siRNA. Following transient transfection with caspase 8 siRNA, we noted that caspase 8 gene expression knock-down in MIP/ZEO cells did not affect their response to 5-FU 1200 M, because cell viability decreased from 87.8 Ϯ 5.8 to 52.1 Ϯ 1.7% (p ϭ 0.0002) in comparison with control cells transfected with scramble siRNA (Fig. 2A). Interestingly, caspase 8 gene silencing in MIP/SP abolished the effect of 5-FU by preventing a decrease in cell viability after exposure to 5-FU (97.2 Ϯ 1.5% viable untreated cells versus 97.6 Ϯ 1.4% after 5-FU treatment, p ϭ 0.8783) ( Fig. 2A). This dramatic effect of increasing cell viability after reducing caspase 8 by siRNA persisted following longer exposure to even higher concentrations of 5-FU in MIP/SP and the intrinsically SPARCexpressing HCT116 cells in comparison with cells not trans-fected with caspase 8 siRNA (Fig. 2B). However, even in the presence of caspase 8 siRNA, a significant decrease in cell viability could now be observed in both MIP/SP and HCT116 cells following 5-FU, suggesting that a caspase 8-independent mechanism could now be contributing to this reduction in cell viability.
These initial results suggested a caspase-8-mediated effect in promoting apoptosis in cells overexpressing SPARC and that in the absence of caspase 8, the possibility that the intrinsic pathway could be contributing to this event was raised. Using chemical inhibitors that display some specificity against caspase 8-like (z-IETD-fmk) and caspase 9-like (z-LEHD-fmk⅐ trifluoroacetate salt) activities, we again observed that inhibition of caspase 8-like activity affected MIP/SP cells more dramatically than control MIP/ZEO cells. In response to an exposure to 1000 M of 5-FU for 24 h only, cell viability decreased in both control MIP/ZEO cells and MIP/SP cells (Fig. 2C). However, in MIP/SP cells, preincubation with 10 M of a caspase 8-like inhibitor abolished this decrease in cell viability observed following exposure to 5-FU (Fig. 2C) (Fig. 2D). These results further support a caspase 8-dependent activation of apoptosis in SPARC overexpressing cells and suggest that the involvement of caspase 9 occurs downstream of this event.
To further assess whether the effect of caspase 8 gene silencing was dependent on SPARC expression, we examined the effect of a different agent, CPT-11 on several CRC cell lines expressing variable levels of SPARC, intrinsically high SPARC-expressing HCT116 cells (9) and high SPARCexpressing MIP/SP cells, and compared them to low SPARC expressing MIP/ZEO and even lower MIP/CPT cells. Again, we noted that MIP/SP cells had a reduction in cell viability after treatment with CPT-11 100 M (100.0 Ϯ 0.0001% versus 63.4 Ϯ 4.9%, p ϭ 0.00003), which was again abolished after transfection with caspase 8 siRNA despite the presence of CPT-11 (98.4 Ϯ 3.1% versus 97.2 Ϯ 4.0%, p ϭ 0.8092) (Fig. 2D). The most interesting finding was that a similar effect of caspase 8 gene silencing was observed with high SPARC-expressing HCT116 cells, with decreased sensitivity to CPT-11 in comparison with untreated cells (Fig. 2D), for example, 104.8 Ϯ 1.8%  (Fig. 3A). However, in MIP/SP cells, an increase in caspase 3/7 activity was only observed in control MIP/SP cells (untreated 327.7 Ϯ 30.9 RLU versus 5-FU-treated 5501.0 Ϯ 800.0 RLU, p ϭ 0.0001), and this effect was abrogated when treated cells were initially transfected with caspase 8 siRNA (untreated 371.0 Ϯ 58.9 RLU versus treated 291.0 Ϯ 34.7 RLU, p ϭ 0.31). No caspase 3/7 activity was observed in MIP/5-FU cells following incubation with 5-FU, following transfection with either caspase 8 siRNA or scramble control (Fig. 3A). This reduction in apoptosis was further confirmed by TUNEL assay. In the absence of any caspase 8 inhibition, the highest percentage of cells to undergo apoptosis was observed with MIP/SP cells exposed to 5-FU, with 20.7 Ϯ 6.9% apoptotic cells, in comparison with the untreated group (5.1 Ϯ 2.0%, p ϭ 0.0003) (Fig. 3B). However, following caspase 8 gene knock-down, there was a decrease in the sensitivity of MIP/SP cells to 5-FU (Fig. 3B), with only 0.5 Ϯ 0.5% apoptotic MIP/SP cells detected. A larger percentage of MIP/ZEO cells continued to undergo apoptosis regardless of the presence or absence of caspase 8 gene expression following 5-FU treatment (no transfection with caspase 8 siRNA: 12.2 Ϯ 0.7% apoptotic cells versus transfection with caspase 8 siRNA: 12.5 Ϯ 2.4%, p ϭ 0.02). Therefore, in complete contrast to MIP/SP cells, knock-down of caspase 8 did not inhibit apoptosis in MIP/ZEO cells (Fig. 3, B and C). Resistant MIP/5-FU cells did not respond to 5-FU treatment significantly, and this was not influenced by changing the expression of caspase 8 gene expression (Fig. 3B and supplemental Fig.  S2).
SPARC Interacts with Pro-caspase 8-Based on the above results, there appeared to be recruitment of the extrinsic pathway in SPARC-mediated apoptosis, and in particular, this appeared to be associated with a prominent role for caspase 8. It is unclear how this occurs, but we wondered whether this effect FIGURE 4. Interaction between pro-caspase 8 and SPARC is detected at the cell membrane. Co-localization of SPARC and pro-caspase 8 was determined by co-IP studies following fractionation of MIP/SP (A) or HCT116 cells (B) into nuclear (N), cytoplasmic (C), and membrane (M) fractions with our without exposure to 1000 M of 5-FU for 4 h. 250 g of protein from each cellular fraction were immunoprecipitated (IP) with antibodies to SPARC, caspase 8 (C and N termini), His 6 (to detect His-tagged recombinant SPARC), or IgG (nonspecific control), and immunoblotted for SPARC or caspase 8. To determine the effect of chemotherapy on the interaction between SPARC and pro-caspase 8, cellular fractions obtained from cells treated with 5-FU were co-IP with antibodies against SPARC and the Western blot probed for pro-caspase 8. With exposure to 5-FU, no interaction between SPARC and the C terminus of caspase 8 is observed, and only the interaction between the N terminus of caspase 8 and SPARC is seen following cleavage of procascapse 8. C and D, to validate the region of caspase 8 that interacts with SPARC, MIP/SP cells were incubated with antibodies targeting the N and C terminus of caspase 8. C, the membrane fraction was used for co-IP for SPARC and caspase 8. Blocking the N terminus inhibits the SPARC-caspase 8 interaction, and diminishes apoptosis (D). These results provide a model for the site of interaction between SPARC and procaspase 8 (E).
was mediated through an interaction between SPARC and members of the death-receptor/caspase 8/DISC complex. We examined this possibility by assessing binding interactions with SPARC by co-immunoprecipitation (co-IP) studies using antibodies to SPARC and pro-caspase 8. Different subcellular fractions were examined and, interestingly, pro-caspase 8 co-IP with SPARC in a reciprocal fashion from the MIP/SP (Fig. 4A) and intrinsically SPARC-overexpressing HCT116 (Fig. 4B) cell membrane fractions. Moreover, this interaction between procaspase 8 and SPARC disappeared when MIP/SP cells were exposed to 1000 M 5-FU (Fig. 4A), when a caspase 8 antibody that recognizes the C-terminal sequence of the p18 fragment of the protein was used for the co-IP. However, using a caspase 8 antibody recognizing the N-terminal region of this protein, the interaction between SPARC and procaspase 8 could again be detected despite 5-FU exposure (Fig. 4B).
To further validate that the interaction between SPARC and procaspase 8 occurs within the N terminus region of procaspase 8, MIP/SP cells were incubated in vitro with antibodies to caspase 8 targeting its N or C terminus. Cell lysates subjected to reciprocal co-IP with SPARC or caspase 8 showed that cells incubated with antibodies targeting the N terminus of caspase 8 no longer interacted with SPARC, whereas incubation with antibodies targeting the C terminus had no such effect (Fig. 4C). By interfering with this interaction, MIP/SP cells became less responsive to 5-FU as indicated by a significant reduction in apoptosis (Fig. 4D).

SPARC in Combination with 5-FU Increases Caspase 8 Expression in Tumor
Xenografts-We previously demonstrated that the combination of SPARC and 5-FU treatment conferred the greatest tumor regression in mouse xenografts of MIP101 cells, and this correlated with higher numbers of apoptotic cells (9). We next examined whether the interaction between SPARC and caspase 8 could also be detected in vivo. Tumor xenografts bearing MIP/SP cells showed intrinsically higher levels of caspase 8 gene expression than MIP/ZEO cells (Fig.  5A). Moreover, in keeping with our in vitro observations, caspase 8 protein activity was highest in MIP/SP tumors harvested from animals treated with 5-FU (Fig. 5B). We also examined tumors from MIP 101 mouse xenografts that had been previously treated with a combination of SPARC and 5-FU, or as single agents, for caspase 8 expression and again only observed higher levels of caspase 8 in tumors harvested from mice that were administered SPARC and even more significantly following combination treatment with SPARC and 5-FU (Fig. 5C). This up-regulation of caspase 8 expression in animals exposed to both SPARC and 5-FU appears to be restricted to the tumor xenografts, because livers harvested from the same mice did not have elevated caspase 8 expression (Fig. 5C).

DISCUSSION
SPARC, best known for its role as a matricellular protein, is expressed in many different cell types (34). Its role in tumorigenesis has been examined in a variety of cancers (9,(35)(36)(37)(38) including hematological malignancies (12), with diverging actions that may be attributed to tissue-related differences. Previously, we demonstrated that SPARC gene and protein expression changes the sensitivity of therapy-refractory MIP101 CRC cells to chemotherapy in vitro and in vivo by enhancing apoptosis and tumor regression (9). Induction of apoptosis in the presence of SPARC has also been observed in ovarian cancer cell lines (10), lung and pancreatic tumors implanted in SPARC ϩ/ϩ mice (11), and kidney tumor xenografts (39). Our previous observations, together with others, led us to assess the apoptotic signaling events mediated by SPARC, and in particular, the potential involvement of the extrinsic pathway.
Several mechanisms, including inhibition of apoptosis, can contribute to tumor progression in therapy-unresponsive tumors. Most chemotherapies induce apoptosis via the intrinsic pathway (40). However, there is growing evidence of the involvement of the extrinsic pathway of apoptosis in chemotherapy-induced cell death. In particular, there have been reports of caspase 8 activation in both death receptor-dependent (41) and -independent fashions (42). In non-small cell lung cancer and leukemia, caspase 8 can be activated by chemotherapies such as cisplatin, etoposide, gemcitabine, and topotecan, independent of FADD and death receptors (43). Absence or reduced expression of caspase 8 has been correlated with resistance to apoptosis in a variety of tumors, including neuroblasto- mas, and small cell lung cancers (44). In laryngeal squamous cell carcinoma, cisplatin-resistant HEp-2 cells had reduced caspase 8 activation (45), whereas doxorubin-resistant MCF-7 breast cancer cell line had defects in caspases 8, 9, and 10 (46). In our study, we show that in SPARC-overexpressing MIP 101 cells (MIP/SP) that were more sensitive to chemotherapy, higher levels of expression of the genes involved in the extrinsic pathway of apoptosis (caspase 8, 10, FADD) were noted, which increased even more significantly following exposure to 5-FU. This elevated expression at the mRNA level translated to the protein level, where similar increases were seen in MIP/SP cells. Interestingly, higher caspase 8 levels were also observed in tumor xenografts that experienced the most dramatic regression in mice treated exogenously with SPARC and 5-FU, further supporting our in vitro observations. Activation of caspase 8 appears to involve and converge downstream with the mitochondrial pathway through activation of Bid, because higher levels of the full-length protein were observed in MIP/SP cells, which gradually declined following incubation with 5-FU, as one would expect with activation and truncation of Bid. The downstream effects can be seen through a significantly greater activation of caspase 9 and 3 in MIP/SP cells than in MIP/ZEO, whereas no significant activation is seen in MIP/5-FU. Therefore, our findings indicate that in the presence of SPARC, apoptosis occurs as a caspase 8-dependent event that later converges with the intrinsic pathway to involve caspase 9 (Fig. 6A). However, in the absence of SPARC, apoptosis occurs by activating the intrinsic pathway independently of caspase 8 (Fig.  6B). This recruitment of caspase 8 in the presence of SPARC, in addition to the involvement of the intrinsic pathway, further augments and amplifies the death signal induced by chemo-therapy and explains in part the chemosensitizing effect of SPARC.
We further demonstrate that this caspase 8-dependent event may occur because of an interaction between SPARC and pro-caspase 8, taking place at the p43/p41 fragment of caspase 8 containing the N-terminal DED domains and the catalytic site, because this interaction can no longer be observed upon cleavage of caspase 8 following exposure to chemotherapy when a caspase 8 antibody recognizing the C terminus was used. Also, in vitro exposure to caspase 8 antibody targeting the N terminus interferes with its ability to interact with SPARC and thereby abrogates its effect on apoptosis. These observations lend further support that the interaction between the N terminus of caspase 8 and SPARC is important in mediating apoptosis. We are currently investigating whether SPARC interacts with a larger complex that contains pro-caspase 8. This is particularly relevant, because studies have shown that Bcl-2 can inhibit caspase 8-mediated apoptosis by binding and inhibiting pro-caspase 8 (47). Interestingly, exposure to SPARC can induce autosomal dominant polycystic kidney disease cystlining epithelial cells to undergo apoptosis as a result of a reduction in the ratio of Bcl-2 and Bax (34). Therefore, one wonders whether an interaction between SPARC and pro-caspase 8 may prevent its sequestration by Bcl-2, thereby enhancing apoptosis, as observed in this study.
The presence of higher caspase 8 gene expression in cells overexpressing SPARC was also particularly interesting. The transcriptional regulation of caspase 8 involves binding of transcription factors IRF-1 and IRF-2 to the interferon-sensitive response element site within its promoter (32,33). Interestingly, our ChIP assay showed greater binding of IRF-1/2 to the promoter regions of the caspase 8 gene in cells overexpressing SPARC, thereby indicating that, in the presence of higher levels of SPARC, there is transcriptional up-regulation of caspase 8 in these cells. In addition, there also appeared to be diminished caspase 8 mRNA stability in cells with reduced SPARC expression, because higher SPARC-expressing cells only began to show a decrease in basal caspase 8 mRNA levels after 8 -12 h after exposure to actinomycin D. This time frame is comparable with those observed in MCF-7 cancer cells, with caspase 8 mRNA levels diminishing after 6 -12 h of actinomycin D exposure (32). These results suggest a SPARC-mediated effect in the regulation of caspase 8 mRNA, and further studies are required to elucidate these interesting findings.
Our current observations support a caspase 8-dependent event in SPARC-mediated apoptosis (Fig. 6A), and we show that this occurs as a result of its interaction with pro-caspase 8. . SPARC-induced apoptosis: A model. In this study, we demonstrate how SPARC influences apoptosis. A, in the presence of SPARC, apoptosis occurs as a caspase 8-dependent event through an interaction (*) between SPARC (yellow) and pro-caspase 8 (green) that results in cleavage of procaspase 8 (blue) following exposure to chemotherapy . This leads to the activation of the mitochondrial pathway of apoptosis via Bid. This SPARC-mediated event enhances apoptosis and augments chemosensitivity in CRC cells. B, at lower SPARC levels, apoptosis only occurs through the intrinsic pathway, independent of caspase 8.
This mechanism may also prove to be important in other malignancies where SPARC has also been shown to enhance tumor regression and to induce apoptosis (10 -12, 48). The results presented in this study are exciting because they provide an initial insight into a potential mechanism by which SPARC mediates its chemosensitizing effect in cancers refractory to therapy. The caspase 8-dependent signaling events mediated by SPARC to promote tumor regression not only supports its therapeutic potential in cancer but also allows us to identify potential downstream targets that can be similarly exploited to enhance tumor regression and overcome chemotherapy resistance in patients with advanced, therapy-refractory CRC.