The Death Effector Domains of Caspase-8 Induce Terminal Differentiation

The differentiation and senescence programs of metazoans play key roles in regulating normal development and preventing aberrant cell proliferation, such as cancer. These programs are intimately associated with both the mitotic and apoptotic pathways. Caspase-8 is an apical apoptotic initiator that has recently been appreciated to coordinate non-apoptotic roles in the cell. Most of these functions are attributed to the catalytic domain, however, the amino-terminal death effector domains (DED)s, which belong to the death domain superfamily of proteins, can also play key roles during development. Here we describe a novel role for caspase-8 DEDs in regulating cell differentiation and senescence. Caspase-8 DEDs accumulate during terminal differentiation and senescence of epithelial, endothelial and myeloid cells; genetic deletion or shRNA suppression of caspase-8 disrupts cell differentiation, while re-expression of DEDs rescues this phenotype. Among caspase-8 deficient neuroblastoma cells, DED expression attenuated tumor growth in vivo and proliferation in vitro via disruption of mitosis and cytokinesis, resulting in upregulation of p53 and induction of differentiation markers. These events occur independent of caspase-8 catalytic activity, but require a critical lysine (K156) in a microtubule-binding motif in the second DED domain. The results demonstrate a new function for the DEDs of caspase-8, and describe an unexpected mechanism that contributes to cell differentiation and senescence.


Introduction
Caspase-8 is an initiator protease recruited to the death inducing signaling complex during apoptosis initiated by death receptors. Homotypic interactions, mediated by the aminoterminal death effector domains (DEDs) of caspase-8, are required for recruitment and subsequent maturation and dimerization of caspase-8 initiating the extrinsic apoptosis cascade. In addition to this role in death receptor-mediated apoptosis, cumulative evidence suggests that caspase-8 performs other non-apoptotic functions in development [1], including proliferation [2,3], cell migration [4][5][6][7] and differentiation [3,8]. We and others previously reported that caspase-8 has the capacity to localize to a number of different cellular locations, including the cytosolic compartment [9], actin-rich ruffles [10,11], endosomes [12], including those at the front of migrating cells [13], focal adhesions [4] and stable microtubule structures, such as centrosomes [14]. Interestingly, the different domains of caspase-8 appear to favor localization to different cellular compartments. It is possible that these different preferred locations may ultimately influence caspase-8 function(s).
Differentiation, senescence, and apoptosis are critical programs for the development and maintenance of cellular homeostasis.
Disruption of any of these essential processes is an important component in the pathogenesis of many diseases, including cancer. A common characteristic of human cancer is disrupted cellular differentiation [15]. In some cancer cells, the induction of differentiation with therapeutic agents terminates uncontrolled proliferation [16,17]. Even among therapies which do not aim to specifically induce cell differentiation or senescence, it nonetheless appears to be a common mechanism limiting tumor growth [18]. Therefore, cellular senescence is not only a physiological program that limits the proliferative capacity of an indiviudal cell, but plays an important physiological role as a natural barrier to suppress pathogenic transformation or proliferation [19].
In the current study we observed in epithelial, endothelial, myeloid and tumor cells, that lack of caspase-8 exhibited a disrupted pattern of cellular differentiation/senescence. Specifically, we find that caspase-8 expressing myeloid, endothelial or epithelial cells display an accumulation of caspase-8 DEDs during the onset of differentiation and senescence. Silencing of caspase-8 correlates with disruption of differentiation, and increase in replicative capacity, while expression of the DEDs is sufficient to promote differentiation and proliferative arrest. In particular, these events occur independent of caspase catalytic activity, but require the ability of DEDs to bind to microtubules. The microtubule interaction capacity of caspase-8 DEDs, in turn, promotes aberrant mitosis, multinucleation, and induces apoptosis, senescence or terminal differentiation. Our results support the notion that caspase-8 is a multifunctional tumor suppressor protein, demonstrating that the DEDs of caspase-8 act to regulate not only cell death, but also differentiation and senescence.

Caspase-8 Regulates Terminal Differentiation
Caspase-8 has been shown to play non-apoptotic roles in many tissues [1,3,20,21]. Caspase-8 is particularly enriched in the skin [20], with a notable accumulation in the differentiating layers of the epidermis ( Figure 1A, B and Figure S1A and B). Interestingly, the skin of caspase-8 knockout mice shows clear signs of hyperproliferation, particularly through the basal and spinous layers ( Figure 1A). Examining the skin of K14 Cre /C8 flox/flox animals, we found that loss of caspase-8 expression is associated with a deficient commitment of the basal cells to undergo terminal differentiation. In addition to epidermal thickening, the keratinocyte marker K5 is not restricted to a single cell layer, but rather is maintained throughout several cell layers ( Figure 1A). Moreover, in caspase-8-deficient animals, the microtubules of the keratinocytes remained largely cytosolic rather than peripheral ( Figure 1B), suggesting that the uncoupling of microtubules from the centrosome, associated with keratinocyte exit from proliferation, was disrupted [22]. Together, these results indicated that caspase-8 plays an important role in attenuating skin proliferation and/or modulating skin differentiation.
To examine this further, we next used a keratinocyte model of induced differentiation [23][24][25]. Differentiation of HaCat cells did not induce significant DEVDase (caspase) activity, nor did it promote accumulation of caspase-8 activation products such as the cleaved fragment p43 ( Figure 1C and data not shown). The level of pro-caspase-8 expression did not change during this process, however, we did detect a modest accumulation of the DED domains among differentiating and non-proliferative cells ( Figure 1C, right). We also detected the DEDs of caspase-8 at the centrosome, and this persisted during reorganization of the microtubules upon differentiation ( Figure 1C, left).
To evaluate whether the DEDs of caspase-8 influence cell differentiation, we next examined myeloid differentiation, in which caspase-8 has been implicated in vivo [3,8]. Phorbol-ester-induced differentiation of U937 cells resulted in DED accumulation, as determined via immunoblot analysis with a monoclonal antibody to the DEDs of caspase-8 ( Figure 2A). No caspase-8 DEDs were observed in differentiated U937 cells in which caspase-8 expression was silenced (U937-shC8) ( Figure 2B, inset). Differentiation of the U937-shC8 cells was consistently compromised relative to control cells expressing caspase-8 ( Figure 2B and C), both, by expression of the macrophage marker CD11b when scored by flow cytometry ( Figure 2B) or by morphological criteria ( Figure 2C). Importantly, re-expression of the DEDs was sufficient to rescue the differentiation process ( Figure 2C). Together, these results support the notion that the DEDs of caspase-8 regulate terminal differentiation.

Caspase-8 DED Suppress Tumor Growth
Our observations suggested that the DEDs of caspase-8 were responsible for influencing terminal differentiation events. Neoplasms such as neuroblastoma can arise in part due to the failure of precursor cells (neuroblasts) to differentiate correctly [19]. Aggressive disease is associated with the loss of caspase-8 [26], while spontaneously resolving tumors (neuroblastoma type IV-S) will frequently express caspase-8 [27]. We confirmed that ectopic expression of DEDs among cells that already express caspase-8 induces apoptosis among caspase-8 positive cells [14,28]. Many of these cells express readily detectable levels of DEDs, and appear to be intolerant to additional expression. Therefore, we next used caspase-8-deficient neuroblastoma cells to evaluate the role of caspase-8 DEDs in tumor progression. Interestingly, the growth of DED-GFP reconstituted neuroblastoma tumors in the chick chorioallantoic membrane was significantly inhibited relative to control tumors expressing GFP alone ( Figure 3A). Accordingly, we found that the DED-GFP expressing cells did proliferate significantly less than control cells expressing GFP ( Figure 3B). Thus, DED expression influenced neuroblastoma proliferation in vitro and in vivo. The effect did not appear to be related to ''toxicity'' of DEDs, as the reconstituted neuroblastoma expressed similar levels of DEDs compared to other cell lines examined (COS-7, NB16; reference 14).

DED Expression Causes Mitotic Defects
Interestingly, DED association with microtubule-containing structures can influence cell responses to paclitaxel (14), and thus, DED association with microtubules might influence cell division and/or proliferative arrest. The DEDs of caspase-8, but not c-FLIP, interact with stable microtubule structures ( Figure S2), including centrosomes, spindle poles and midbodies in neuroblastoma cells [14] as well as keratinocytes and other epithelial cells ( Figure S3A). Microtubule dynamics play a key role in regulating normal cell division, therefore, we tested the effect of DED expression on mitotic progression. Notably, we found an increased incidence of aberrant mitotic figures in DED-expressing neuroblastoma cells ( Figure 3C). This was associated with an enrichment of DED-expressing cells in pre-metaphase relative to GFPcontrols, suggesting mitotic delay ( Figure 3D). In post-mitotic cells, DED-GFP expression was associated with multinucleation ( Figure 3E, F, and Supplementary Figure S3B). Intriguingly, other cells which exhibit high basal levels of multinucleation, such as COS-7 [29] also expressed abundant caspase-8 DEDs ( Figure  S2C). Thus, the DEDs of caspase-8 influence normal cellular mitotic function, impacting cell proliferation.
In contrast, the decreased proliferation was not explainable simply by mitotic defects leading to apoptosis, since faster proliferation and larger tumors in vivo were derived from cells expressing a pro-apoptotic (wt) form of caspase-8. This caspase-8 GFP construct exhibited higher levels of apoptosis than the DED-GFP cells, but nonetheless grew larger in vivo and faster in vitro ( Figure S4C, D and E). These studies also demonstrated that DEDs expressed as part of an inactive caspase-8 holoprotein (Casp8*-GFP) showed no deleterious effects on the cell ( Figure  S3C and E), possibly due to intra-domain interactions in the native procaspase. Moreover, DED expression in other caspase-8deficient tumors (small cell lung carcinoma) [30], similarly impacted proliferation and resulted in multinucleation ( Figure  S4F), thereby validating the concept that DEDs contribute to mitotic crisis and cell cycle arrest.

DEDs Activate the p53 Pathway and Induce Apoptosis or Cell Cycle Arrest
We considered that selection pressure might result in the loss of caspase-8 DEDs, and increased proliferation and tumor formation might result with continued passage of the cells. To our surprise, we observed that proliferation of neuroblastoma cells did not increase, and that DED-expressing cultures arrested ,30 passages after initial selection. Beyond early passages (p5-p12), multinucleation was increasingly common among the DED-expressing Immunoblot analysis of caspase-8 expression among U937 undifferentiated monocytes (control) and differentiated macrophages. Lysates were probed with antibody recognizing the DEDs of caspase-8, with analysis of tubulin shown as a loading control. B. Monocytes were induced to differentiate by treatment with phorbol esters after infection with lentivirus encoding shRNA to caspase-8 or a control shRNA. Flow cytometry of control and caspase-8 knockdown U937 cells using CD11b expression as a reporter of differentiation. DED expression during differentiation was assessed by immunoblot analysis of these lysates with an antibody to caspase-8 DED domain (inset). C. Evaluation of U937 differentiation was performed by direct microscopic assessment using standard morphological criteria (round undifferentiated and spread differentiated). Data were analyzed with Fisher's Exact Test (*, significant difference, p#0.05). doi:10.1371/journal.pone.0007879.g002 cells ( Figure 4A upper panels). At later passages, cells either died or exhibited striking changes in phenotype, including elongate morphology and loss of proliferation ( Figure 4A lower panels). Testing these cells for classical markers of cell cycle arrest [18], we observed an accumulation of both p21 and p53 in DED-GFP expressing cells ( Figure 4B and Figure S4A, left panel). In agreement with our previous results in myeloid cells ( Figure 2) and keratinocytes (Figure 1), this induced proliferative arrest that was accompanied by expression of neuronal differentiation markers such as MAP2 ( Figure 4B) and bIII tubulin ( Figure S5A, right panel). Therefore, the DEDs of caspase-8 appeared to influence both, multinucleation and the acquisition of cellular differentiation in the neuroblastoma cell model. To further validate a role of the DEDs in proliferative arrest, we used primary endothelial cell cultures, a known model of replicative senescence ( Figure S5B and C). Endothelial cells also accumulated caspase-8 DEDs, and exhibited multinucleation at later passages (p6-p7) ( Figure S5B). Knockdown of caspase-8 did not prevent senescence among the endothelial cells, but did reduce multinucleation and typically permitted additional 2-3 passages prior to arrest ( Figure S5C). Together, the results support a model in which DED interaction with stable microtubule structures arrests proliferation and induces differentiation.

DED Microtubule Binding Is Required for Cell Cycle Arrest and Differentiation
Caspase-8 DEDs contain a microtubule-association motif (KLD) present in the second DED (DEDb) [14]. The motif is required for caspase-8 localization to centrosomes, microtubules, spindle poles and midbodies ( Figure S2A) [14]. To test whether this microtubule binding function was required for the differentiation and anti-proliferative function of the DEDs, we next evaluated the proliferative capacity of NB7 cells expressing a mutant caspase-8 DED construct (K156R) that selectively lacks microtubule association, but still associates with other DED-containing proteins [14]. In contrast to neuroblastoma expressing wildtype DEDs, those expressing the K156R mutant were unaffected by DEDK156R expression. These cells did not differentiate, and proliferated similar to untransfected controls with no increased incidence of mitotic defects ( Figure 5A and B). Accordingly, in tumors grown in the chick CAM, the DEDK156R cells grew similar to those expressing GFP ( Figure 5C), while the expression of DEDs arrested growth, with a corresponding decrease in cell proliferation ( Figure 5D). These results demonstrate that microtubule association is critical for DED-mediated differentiation and arrest of proliferation. Together, these data reveal a new role for caspase-8 DEDs in the regulation of cell differentiation and senescence.

Discussion
In this study, we characterize a new function for the death effector domains of caspase-8 in cell cycle regulation. First, we find that caspase-8 DEDs accumulate in cells undergoing terminal differentiation. Silencing of caspase-8 disrupts or delays differentiation, while reintroduction of DEDs restores differentiation potential. We further report that DED expression is sufficient to impair tumor growth and cell proliferation, promoting mitotic defects that foster cell death or cell cycle arrest and terminal differentiation. Finally, we demonstrate that these events require a critical lysine (K156) in a microtubule binding motif in the second DED (DED-b). These observations together suggest that caspase-8 DEDs function as a tumor suppressor, acting as an antiproliferation and differentiation-inducing element. It is tempting to speculate that this type of mechanism may act in some spontaneously regressing tumors, such as stage IV-S neuroblastoma, however, such a mechanism would be co-dependent upon other classic tumor suppression pathways, including p53 and p21. Interestingly, the cell lines which can maintain higher levels of DEDs, such as COS-7, also have documented defects in p53 signaling.
Reconstitution of DED expression in the caspase-8 deficient NB7 cells similarly leads to an increase in multi-nucleation. This effect resembles the effect of microtubule-directed agents which disrupt mitotic spindle formation and chromosome segregation. However, these microtubule disrupting agents do not efficiently induce differentiation, suggesting that the DEDs may play additional roles following microtubule binding. Collaborative effect is observed between caspase-8 DEDs and microtubule stabilizing agents, but not microtubule-disrupting drugs [14]. It is likely that this occurs because microtubule-stabilizing agents act to increase DED association with microtubules [14].
Death domain proteins are a large and evolutionarily ancient family. Defined by Tube and Pelle, each domain consists of six peptide helices, with the specific homology placing family members into subfamilies that include the DEDs, the classic death domain (DD) proteins, the pyrin domains proteins, and the caspase recruitment domain (CARD) families [31]. Interactions among these proteins are frequently homotypic or within a subfamily, but interactions between different families, and with unrelated proteins, are unknown. Microtubules have several surface helices with which microtubule-binding proteins interact, as well as cell surface clefts bound by KLD/KID-containing proteins such as kinesins, Tau, and MAP2C [32,33]. While other DED proteins lack the KLD motif in the turn between helix 4 and 5 of the DED structure, it remains possible that some might still associate with microtubules. For example, at least one death domain protein with a CARD fold, CARD6, interacts with microtubules [34].
It is also possible that DED-mediated effects can influence differentiation or proliferation arrest among cells in which the apoptotic pathway is compromised. For example, it is common for tumor cells to express high levels of anti-apoptotic proteins that interrupt the catalytic cascade [35], or which otherwise alter the threshold of caspase-8 activation required for apoptosis [36,37]. Alternatively, low level activation of caspase-8 may contribute to aberrant cell growth, since activation of the extrinsic death pathway at levels which do not induce apoptosis can foster NF-kB signaling [38]. Since caspase-8 maturation occurs concomitant with the accumulation of DEDs, DED accumulation may act to oppose NF-kB signaling, functioning as a tumor suppressor via their microtubule-associated function. Given the caveats of the studies performed here, it would appear that the DEDs have a better stability than the catalytic domain, which is a target of RING protein-mediated clearance [37].
In summary, we describe a novel function for caspase-8 as an orchestrator of not only apoptosis but also differentiation and senescence. The function is unexpectedly associated with the amino-terminal death effector domains, and does not require caspase-8 catalytic activity. The ermerging roles of this multifunctional protein, and its surprising interactions at the nexus of the cellular migration, proliferation, differentiation and apoptosis pathways, continue to offer key insights into the signaling crosstalk that regulates cell fate. After treatment, transmission light pictures were taken with an inverted Microscope (Eclipse, Nikon Inc, USA) using MetaMorph software (Molecular Devices, Sunnyvale, CA). Adherent cells were harvested by first incubating the cells with Versene, a non enzymatic dissociation buffer (Gibco, Carlsbad, CA) for 5 min, followed by collecting the cells using a cell scraper. Cellular differentiation was quantified by flow cytometry. Briefly, cells were resuspended in PBS + 5% FBS, human IgG (1:100) was added and cells were incubated on ice during 45 min to block unspecific binding. Cells were stained using an anti human CD11b-APC antibody (M1/70, Miltenyi Biotech Auburn, CA).

Viability Cell Cycle Analysis
Cell viability was analyzed by flow cytometry following propidium iodide (PI) staining, essentially as described before [34]. Briefly, cells were cultured for 48 hours and harvested, resuspended in ice cold PBS containing 10 mg/ml of Propidium iodide (PI) and analyzed by flow cytometry. The extent of apoptosis was determined by plotting PI fluorescence versus the forward scatter parameter, using the Cell Quest program. For DNA content analysis (cell cycle distribution), cells were previously permeabilized in methanol at 220uC, for 15 minutes and resuspended in PBS containing RNase A and 10 mg/ml of PI. Samples containing roughly 2610 4 cells were analyzed using the Cell Quest program.

Immunoprecipitation and Immunoblotting
Cells were lysed in either NP40 lysis buffer (150 mM NaCl, 50 mM Tris Base pH 7.4, 1% NP40) or RIPA buffer (50 mM Tris pH 7.4, 100 mM NaCl, 0.1% SDS) supplemented with complete protease inhibitor mixture (Roche), 50 mM NaF and 1 mM Na 3 VO 4 and centrifuged at 13,000 g for 10 min at 4uC. Protein concentration was determined by BCA assay. For immunoprecipitation, 500 mg of protein was incubated with 2 mg of rabbit anti-GFP antibody (Abcam) overnight, at 4uC. Complexes were precipitated with 25 ml of protein A/G (Pierce). Beads were washed five times, eluted in boiling Laemmli buffer, resolved on 10% SDS-PAGE and immunoblotting was performed with mouse anti-alpha tubulin antibody (1:1000). For immunoblot analysis, 30 mg of protein was boiled in Laemmli buffer and resolved on 10% gel.

Cell Culture and Transfections
Human neuroblastoma cells (NB7), deficient in caspase-8 were cultured in RPMI supplemented with 10% fetal bovine serum, glutamine and non-essential aminoacids. Human keratinocytes (HaCat), human epithelial cells (A549) and monkey fibroblasts (COS-7), were cultured in DMEM, supplemented as previously described. Human endothelial cells (HUVECs) were cultured in M199 with endothelial cell supplement, 10% fetal bovine serum, glutamine and minimal essential aminoacids. NB7 cells deficient in caspase-8 were transfected using the Fugene reagent following manufacturer's protocol. Stable cell lines were selected with 500 mg/ml G418 (Gibco) and sorted by Flow cytometry for GFP positive cells. Caspase-8 expression was confirmed by immunoblotting with caspase-8 N-terminus (BD Pharmingen) and C-terminus antibodies (hybridomas, clones C5 and C15, were a gift from M. Peter, University of Chicago, USA). To create caspase-8 deficient and control cell lines, U937 and HUVECs cells were infected with lentivirus encoding shRNA to caspase-8 (Open Biosystems) or a control shRNA as described previously (Wrasidlo et al., 2008).

Vectors and Constructs
DNA for caspase-8 was kindly provided by Guy Salvesen, Burnham Institute, La Jolla, USA. Full length caspase-8-GFP (C8-GFP) and inactive mutant C360A (C8*-GFP) fusion proteins were cloned into C1pEGFP (Clontech Laboratories). Death effector domain from caspase-8 (DED-GFP), DED1-GFP and DED2-GFP fusion proteins were generated using 59 and 39 primers containing unique HindIII and BamHI restriction sites respectively, and cloned into N2pEGFP. K156R, S109A and D135A mutations were introduced in DED using the QuikChange Mutagenesis kit (Stratagene) with appropriate mutagenesis primers. DED myc-His fusion protein was made with the same primers but cloned into pcDNA3.1 myc-His (Invitrogen).

Tumor Growth
For avian tumor studies, 5610 6 neuroblastoma cells suspended in 40 ml of complete medium were seeded on 11-day-old chick chorioallantoic membrane [35]. Tumors were left to develop for 7 days and were then resected and weighed.

Immunocytochemistry, Immunohistochemistry and Confocal Microscopy
Cells were permitted to attach to coverslips, fixed with 4% PFA and permeabilized in PBS containing 0.1% triton for three minutes, blocked for 60 minutes, at room temperature with 2% BSA in PBS. Cell were stained with monoclonal antibody to amino terminus death effector domain of caspase-8 (Pharmingen and Calbiochem), p53 (Cell signaling), p21 (Santa Cruz) polyclonal antibody specific for alpha tubulin (Abcam), gamma tubulin (Abcam), pericentrin (Abcam) or lamin B (Santa Cruz). All primary antibodies were used at 1:100 dilution, for two hours at room temperature. After washing several times with PBS, cells were stained for two hours at room temperature, with secondary antibody fluorescently labelled in green or red, specific for mouse, rabbit or goat (Invitrogen) and diluted 1:300. In some cases, cells were co-incubated with the blue DNA dye TO-PRO-3 (1:1000) (Invitrogen).
Mouse skin isolated from P1 and P10 wildtype and knockout animals were frozen in OCT (Tissue-Tek). Caspase-8, a tubulin and epidermal differentiation markers, K5, K1 and loricrin were stained in 8 mm frozen sections after tissues were fixed for 10 min in cold acetone or cold acetone/methanol (for a tubulin staining). For nuclear staining, TO-PRO-3 (invitrogen) was added to the secondary antibody dilution. Immunofluorescence was detected using Alexa-fluor 488 and Alexa-fluor 568 secondary antibodies (Invitrogen). Samples were mounted in Vectashield hard set mounting media (Vector Laboratories) and imaged on a Nikon Eclipse C1 confocal microscope. Bright field and confocal microscopy images of Human Umbilical Vein Endothelial Cells (HUVECs) at early passages (up to P4) and late passages (P6-P8). Nuclear p53 staining (red channel) confirms accumulation of senescent cells at late passages. Lamin B staining (red channel) shows accumulation of binucleated cells at late passages (yellow arrows) (scale bar = 10 mm). Immunoblot analysis with a caspase-8 DEDs specific antibody on HUVECs shows accumulation of DED at late passages. Immunoblot analysis of tubulin is shown as loading control. Quantification of multinucleated HUVECs at early and late passages. C. Quantification of multinucleation in wild type and lentivirus encoding shRNA to caspase-8 infected endothelial cells, at early and late passages. Immunoblot analysis showing caspase-8 expression in these cells (inset). Data were analyzed with U-Mann-Whitney Test (significant differences, *, p#0.05; **, p#0.01). Found at: doi:10.1371/journal.pone.0007879.s005 (21.02 MB TIF)