Tetraspanin-induced death of myeloma cell lines is autophagic and involves increased UPR signalling

Background: Multiple myeloma (MM) therapy is hindered by the interaction of the heterogeneous malignant plasma cells with their microenvironment and evolving drug resistance. We have previously shown that the membranal tetraspanins, CD81 and CD82, are under-expressed in MM cells and that their reintroduction causes massive non-apoptotic death. In this study, we aimed to characterise the tetraspanin-induced MM death. Methods: Multiple myeloma cell lines were transiently transfected with eGFP–CD81N1/CD82N1 fusion proteins and assessed for death mode by flow cytometry (propidium iodide, ZVAD-fmk, 3MA), activation of unfolded protein response (UPR), and autophagy (immunoblot, RT–PCR). Results: Cell death induced by CD81N1 and CD82N1 in MM cell lines was autophagic and involved endoplasmic reticulum (ER)-stress manifested by activation of UPR pathways, PERK (protein kinase-like ER kinase) and IRE1 (inositol-requiring 1). We also established the relative X-box binding protein 1 baseline expression levels in a panel of MM cell lines and their general dependence on autophagy for survival. Timeline of UPR cascades and cell fate supported our results. Interpretation: This is the first publication implicating tetraspanins in UPR signalling pathways, autophagy, and autophagic death. Integration of our findings with published data highlights the unifying dependence of MM cells on ER–Golgi homoeostasis, and underscores the potential of tetraspanin complexes and ER-stress as leverage for MM therapy.

Multiple myeloma (MM) is an incurable malignancy of end-stage B lymphocytes (plasma cells). Effective disease treatment is hampered by the complex interactions of the cells with the bone marrow microenvironment and difficulties in defining genes that are integral to the malignant phenotype (Anderson, 2007). It is accepted that effective MM treatment will need to address the cells in context of their supportive microenvironment, as well as target compound signalling cascades to overcome cell heterogeneity and evolving resistance (Anderson, 2007). A unifying and unique feature of plasma and MM cells is their extensive protein synthesis and concomitant cellular adaptations (Barnhart and Simon, 2007).
Protein synthesis, maturation, assembly, and delivery are executed in the endoplasmic reticulum (ER) and Golgi apparatus. Multiple myeloma cells are characterised by expanded ER (Cenci and Sitia, 2007). Misfolded proteins in the ER are directed to the proteasome for degradation (ER-associated degradation -ERAD) (Ding et al, 2007). Insufficient ERAD causes proteins to accumulate in the ER, induces ER-stress, and activates signalling cascades. This series of events is termed the unfolded protein response (UPR). The UPR consists of three major pathways initiated by GRP78/BiP activation of ER-stress sensors (activating transcription factor (ATF6), high inositol-requiring 1 (IRE1), and double-stranded RNA-activated protein kinase-like ER kinase (PERK)), and ending in transcriptional modifications that promote the ER capacity for effective protein folding (volume and chaperones) and diminish the protein synthesis rate (Ron and Walter, 2007). In fact, recent studies have showed that elevation of the IRE1-induced transcription factor, X-box-binding protein 1 (XBP1), is essential for terminal B-cell maturation and that an animal model overexpressing XBP1 developed MM (Reimold et al, 2001;Carrasco et al, 2007). It was also shown in two MM cell lines that partial UPR is constitutively activated (Patterson et al, 2008).
Unfolded protein response activation, while having the capacity to rescue cells by providing a window of opportunity to overcome the stress source, can also result in cell death when the stress is not resolved (prolonged and/or too severe) (Ron and Walter, 2007). The ensuing cell death can be apoptotic, necrotic, autophagic, or a combination of these. Crosstalk between autophagy and apoptosis exists as well, resulting in mutual dependence and some redundancy (Gozuacik and Kimchi, 2007).
Autophagy is a highly regulated process usually activated in response to adverse environment during which cytoplasmic materials are enclosed in double membrane-bound vesicles (autophagosomes) that are then targeted by the lysosome for degradation. It is an essential process that allows cell conservation under stress conditions such as nutrient deprivation, and is implicated in both cell death and survival (Ding et al, 2007).
Previously, we showed that MM cells and cell lines underexpress tetraspanin members, CD81 and CD82, compared with normal plasma and peripheral blood B cells (Tohami et al, 2004(Tohami et al, , 2007Drucker et al, 2006). Tetraspanin proteins facilitate the spatial organisation and localisation of multi-protein complexes in distinct membranal microdomains that are important to exteriorinterior cell signalling. As molecular coordinators, the tetraspanins are involved in many fundamental biological pathways and are correlated with the malignant process and prognosis (Lazo, 2007). The importance of CD81 and CD82 to MM was shown in a study that reintroduced these tetraspanins in fusion vectors with eGFP (CD81N1, CD82N1) into MM cell lines. The tetraspanin overexpression resulted in significant cell death, described as primarily necrotic and caspase-independent (Tohami et al, 2007).
In the current study, we aimed to characterise further, the mode of CD81N1/CD82N1-induced MM death and to delineate the signals and constituents that are influenced by tetraspanins. Our major findings showed that the tetraspanins caused ER-stress, UPR activation, and autophagic cell death. This is the first publication implicating tetraspanins in autophagic signalling pathways and autophagic death. Furthermore, our findings underscore the potential of tetraspanin complexes, ER-stress, and their combination as possible means for effective MM therapy.

Cell lines
Multiple myeloma cell lines RPMI 8226 and U266, purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA), and ARP1, ARK, and CAG (provided by Professor Epstein (Little Rock, AR, USA)) were cultured in RPMI 1640 supplemented with 20% heat-inactivated fetal bovine serum (FBS) and antibiotics (Biological Industries, Kibbutz Beit Haemek, Israel). PC3 and Jurkat cell lines (human prostate cancer and T-cell Leukaemia, respectively) were provided by Collgard Company (Petach Tikva, Israel) and cultured in RPMI 1640 supplemented with 10% FBS. Jurkat cells were also supplemented with sodium pyruvate, HEPES buffer, and nonessential amino acids (Biological Industries).

Flow cytometry
Cell survival Transfected and untransfected cells were harvested 18, 24, or 48-h post-treatment/transfection and stained with 1 mg ml À1 propidium iodide (PI) for 10 min. PI À /PI þ cells were enumerated by FACS. Propidium iodide negative or eGFP þ /PI À were considered the surviving cell fraction (untransfected and transfected, respectively), whereas the PI þ and eGFP þ /PI þ were referred to as the fraction of dead cells (untransfected and transfected, respectively).

Cell sorting
Transiently transfected cells were harvested 18 or 24 h posttransfection and passed several times through a syringe for clump dispersion. Next, eGFP þ and eGFP À cells (5 Â 10 6 cells per ml in PBS supplemented with 10% FBS) were isolated and collected using a BD FACSAria cell sorter (BD Biosciences, Sparks, MD, USA). Matching cells treated with respective transfection reagent only, were considered eGFP À and used for calibration of eGFP þ cell threshold.

Reverse transcription polymerase chain reaction
Total RNA was extracted from 18, 24, or 48-h transfected cells with Purescript (Gentra Systems, Minneapolis, MN, USA). Total RNA (1 mg) was reverse transcribed (Reverse-iT 1st strand synthesis kit, ABgene, Epsom, UK) and amplified for XBP1/XBP1s and housekeeping b-actin (3 and 1.5 ml cDNA, respectively). The PCR was optimised at 941C for 2 min, followed by 38 cycles of 15 s at 941C, 60 s at 601C, and 30 s at 721C, using 4 pmol XBP1/XBP1s primers (F, 5 0 -CCTTGTAGTTGAGAACCAGG-3 0 and R, 5 0 -GGGGCTT GGTATATATGTGG-3 0 ) and 2 pmol b-actin primers (F, 5 0 -GAG ACCTTCAACACCCCAGC-3 0 and R, 5 0 -GCTCATTGCCAATGGT GATG-3 0 ). Products were electrophoresed on 2.2% agarose gels stained with ethidium bromide and visualised with Gel Doc 2000 and Multi-analyst software (Bio-Rad, Hercules, CA, USA). Amplification products of XBP1 observed in gels included the spliced, unspliced, and hybrid (duplex of full-length and spliced) forms (Lin et al, 2007). Spliced XBP1 was calculated from the sum of quantified XBP1s band and half the quantity of the hybrid form; total XBP1 was deduced from the combined sum of all bands. X-box binding protein 1/XBP1s expression was normalised to respective b-actin. Average expression of all myeloma and nonmyeloma cell lines was calculated (arbitrary units) and statistically compared. CD81N1/CD82N1 samples were compared with mock (N1)-transfected controls and ratios were expressed as fold change.

Statistical analysis
Paired Student's t-tests were used to analyse differences between cohorts. A P-value of less than or equal to 0.05 was considered significant. An antagonistic effect was verified by drugs' inter- antagonist; q41.15 -synergist; 1.154q40.85 -additive) (Su et al, 2004) assuming that tetraspanin's transfection is the first treatment (A) and 3MA treatment is the second (B). All experiments were repeated separately three to seven times.

RESULTS
Previously we have reported that reintroduction of the tetraspanins, CD81N1 and CD82N1, into CAG and RPMI 8226 MM cell lines resulted in significant non-apoptotic cell death compared with mock-transfected cells (Tohami et al, 2007). Acknowledging the necessity of MM cells to incur changes that support the intensive protein synthesis and ER -Golgi equilibrium typical to them, we hypothesised that the tetraspanins' effect may involve an autophagic form of cell death due to ER-stress and UPR activation. The relevance of our hypothesis was preliminarily tested by addressing the activation of UPR and autophagy in MM cell lines' panel (untransfected).

Autophagy facilitates basal MM cell lines survival
We assessed the importance of autophagy to MM cell lines homoeostasis. We found that inhibiting autophagosome formation using 3MA resulted in elevated levels of cell death (up to 10%, Po0.05) in all MM cell lines ( Figure 1A), but not in the prostate cancer and T-cell leukaemia cell lines (PC3 and Jurkat, respectively). We also determined XBP1 transcript level in MM cell lines and established a relative baseline for further reference ( Figure 1B and C). All myeloma cell lines express detectable levels of total XBP1 (mean ¼ 0.6) that are significantly higher than non-myeloma cell lines (PC3, Jurkat) (mean ¼ 0.41, Po0.05), which is in agreement with published data (Carrasco et al, 2007;Patterson et al, 2008).
The connection between ER-stress and activation of UPR and the induction of autophagy is well established, yet autophagy may function as a means of cell preservation as well as a mechanism of cell death (Ogata et al, 2006;Ding et al, 2007). Moreover, it was shown that increased autophagy levels in yeast facilitate the removal of excess ER after UPR activation, thus promoting ER-homeostasis (Kincaid and Cooper, 2007). Taken together, our results indicate that MM cell lines display a general dependency on autophagy for survival, which may be maintained through a basal condition of ER-stress and activation of UPR signalling cascades.

Autophagic death induced by CD81N1 and CD82N1 in MM cell lines
Next, we examined the mode of cell death induced by CD81N1/CD82N1 in RPMI 8226 and CAG cells. Transfected cells supplemented with 3MA displayed a significant B25% (Po0.05) rescue from cell death of CD81N1-and CD82N1-transfected RPMI 8226 cells (48 h) and 12% rescue in CD81N1-transfected CAG cells (24 h) (Po0.05, qo0.85) compared with transfected cells not supplemented with 3MA ( Figure 2A). In addition, a significant elevation in the proportion of LC3II (vs LC3I) compared with the mock control was observed in CD81N1/CD82N1-transfected RPMI 8226 (24 h) and CD81N1-transfected CAG cells (18 h) (35-45%, Po0.05, Figure 2B). Our results also display significant increases in absolute LC3II levels (not relative to LC3I) compared with mocktransfected cells in CD81N1-transfected RPMI 8226 (38%) and CAG cells (60%), and in CD82N1-transfected RPMI 8226 cells (34%) (Po0.05), an analysis method suggested to be more reliable for determining autophagy (Mizushima and Yoshimori, 2007). It should be noted that CD82N1-transfected CAG cells did not display an increase in LC3II (absolute and relative), nor did they respond to 3MA compared with the mock control, all in accordance with a non-autophagic cell death mode (Figure 2A and B). Next, we co-administered the caspase inhibitor, ZVAD, to tetraspanintransfected MM cell lines treated with 3MA (data not shown). The combined application of 3MA and ZVAD did not differ from CD81N1/CD82N1-transfected cells treated with 3MA alone, indicating that there was no shift in the death mode of the MM cell lines when the autophagy and/or apoptotic cascades were blocked.
For additional validation of autophagic modulation, we examined levels of the established inhibitor of autophagy, mTOR, which normally converts metabolic and mitogenic signals into protein synthesis, and is a recognised target in MM therapeutics (Hu et al, 2003). Assessment of phosphorylated and total mTOR levels in CD81N1/CD82N1-transfected MM cell lines displayed decreased levels of active mTOR compared with mock-transfected control (24 h post-transfection, 35 -40%, Po0.05), which is in sync with autophagic activation ( Figure 2C). These findings are also in accordance with previous results that described a decrease in the number of tetraspanin-transfected MM cells expressing pmTOR (Lishner et al, 2008). Next, we assessed cellular levels of Beclin 1, an established component of the autophagic machinery (Cao and Klionsky, 2007), in the tetraspanin vs mock-transfected MM cell lines, but failed to determine significant changes (data not shown). Interestingly, several recent publications present evidence of a 'Beclin-independent autophagic pathway' distinguished with an ERK and/or JNK-induced autophagy (Chu et al, 2007;Dagda et al, 2008). Therefore, we examined the importance of ERK and JNK signalling to transfected MM cells.
In summary (Table 1), these results show several facets of UPR activation and, therefore, substantiate increased ER-stress in the tetraspanin-transfected MM cell lines. The consequential cell death was mostly autophagic except for the CD82N1-transfected CAG cells, which display ER-stress, yet their death mode is necrotic (by elimination: not autophagic ( Figure 2) and not apoptotic (Tohami et al, 2007)).

Timeline of tetraspanin-induced UPR and cell death in MM cell lines
The feasibility of our findings is further confirmed by the timeline of UPR signalling and ensuing death ( Figures 5 and 6), which is compatible with the model presented recently by Lin et al (2007). Lin and co-authors presented that the initial combined activation of IRE1, PERK, and ATF6 produces cytoprotective outputs (reduced translation, enhanced ER protein folding capacity), which provide a 'window of opportunity' for cells to readjust their ER to cope with stress. If these steps fail to re-establish homoeostasis, IRE1 signalling followed by ATF6 signalling is attenuated, creating an imbalance in which an unchecked PERK pro-apoptotic output channels the cell towards its termination (Lin et al, 2007). Figure 5 shows a temporary increase in XBP1s (Po0.05) and continued elevation of XBP1t in CD81N1/CD82N1transfected MM cell lines. In Figure 6, we present a sequential alignment of UPR markers with cell fate. Inositol-requiring 1 and PERK activation was determined 24/18 h after tetraspanins' transfection in RPMI 8226 and CAG cells, respectively. At 48/ 24 h post-transfection, PERK activation was still maintained in RPMI 8226 and CAG cells, respectively, whereas IRE1 activity had already decreased. At this time point, we also determined JNK activation and showed significant cell death. In addition, pJNK levels were not elevated at earlier time points (24 h in RPMI 8226 and 18 h in CAG cells) and, therefore, are in accordance with its activation in IRE1 progression and onset of cell death ( Figure 6 and data not shown). Arrows in XBP RT -PCR picture depict the unspliced (u), spliced (s) and hybrid (%) forms. Expression levels were normalised to b-actin, fold changes were calculated in total and spliced XBP1 levels relative to mock and expressed as mean±s.e. of at least four separate experiments. Statistically significant differences (*Po0.05, **Po0.01) are depicted. CD81N1/CD82N1 causes ER-stress and autophagic death in MM V Zismanov et al

DISCUSSION
The primordial function of autophagy has long been recognised, but its significance in multiple cellular roles and various pathological conditions, including cancer, is only now being revealed (Gozuacik and Kimchi, 2007;Lerena et al, 2008;Mizushima et al, 2008). In this study, we showed that MM cell lines rely on autophagy for their survival and proliferation under normal culture circumstances. This was not evident in cell lines of other origins. The propensity of MM cells for increased autophagic activity is consistent with the published data that established their heightened ER-stress levels and the functional relationship between the two states (Cenci and Sitia, 2007). Despite the survival-promoting role of the basal autophagy, which we determined, it is well recognised that elevated and/or prolonged autophagy can culminate in cell death (Gozuacik and Kimchi, 2007;Galluzzi et al, 2008), a prospect of great significance in the death of resistant myeloma cells.
In this study, we establish that autophagic death is an achievable target in MM cell lines. We showed that tetraspanins induced ERstress, manifested by activation of UPR pathways, increased autophagy, and eventually, cell death. Moreover, chronological alignment of UPR pathways' expression, shutdown, and ensuing cell death is in agreement with the mechanism of cell fate regulation allocated with the UPR (Lin et al, 2007). The achievability of autophagic death in MM is strengthened by a recent study that reported that inhibition of p27 expression in MM caused death by autophagy (Chen et al, 2008). Yet, in this study, death was preceded by cell cycle arrest, whereas our experimental setting displayed no such effect (Tohami et al, 2007). Previously, we showed that the CD81N1/CD82N1 overexpression attenuated AKT activity and activated FoxO transcription factors (Lishner et al, 2008). In this project we again showed the involvement of JNK activity in the fate of transfected cells (Lishner et al, 2008). Taken together, these findings are compatible with a recent publication that depicted a role for FoxO1 in the death of ERstressed macrophages (Senokuchi et al, 2008). Interestingly, Senokuchi and colleagues showed that FoxOs failed to induce cell death in the absence of ER-stress. The involvement of JNK in ERstress response, as well as in the regulation of FoxO proteins has been described extensively (Ogata et al, 2006;Huang and Tindall, 2007;Lim et al, 2009).
The connection of tetraspanin circuitry with ER-stress and autophagy is novel and provides a direct link between the cancer microenvironment and fundamental cellular functions. In particular, positioning the tetraspanins up-stream of ER -Golgi homoeostasis underscores the significance of the integration of metabolic, environmental, and mitogenic cues to MM survival.
The biological validity of our experimental system is substantiated by multiple controls. In both the current and in previous Abbreviations: ERK ¼ extracellular signal-regulated kinase; IRE ¼ inositol-requiring 1; JNK ¼ c-Jun N-terminal kinases; MA ¼ methyladenine; MAPK ¼ mitogen-activated protein kinase; MM ¼ multiple myeloma; PERK ¼ protein kinase-like ER kinase; PI ¼ propidium iodide; UPR ¼ unfolded protein response.O , An effect was found. Â , no effect was evident. (Detailed in the results).   Figure 5 Fate of UPR signalling. Temporary increase in XBP1s (indicative of IRE1 pathway) and continued elevation of XBP1t (indicative of PERK pathway) in RPMI 8226 81N1 (A) 82N1 (B), and CAG 81N1 (C) and 82N1 (D) cells are presented. X-box binding protein spliced and total were detected by semi-quantitative RT -PCR. Expression levels of XBP1s and XBP1t were normalised to b-actin, fold changes were calculated relative to mock and expressed as mean±s.e. of at least three separate experiments. Statistically significant differences (*Po0.05) are depicted. (Tohami et al, 2007;Lishner et al, 2008) studies, we showed that cell death, as well as the activation of UPR pathways, can be determined only in MM cell lines transfected with CD81N1 and CD82N1 eGFP fusion vectors and not with the empty eGFP construct (mock) or with the CD81C1 and CD82C1 oriented fusion plasmids (Tohami et al, 2007). Taken together, our results can be interpreted as presenting specific signals initiated by the cloned tetraspanins (in their N1 conformation). Concurrent with this line of thought, studies underway in our laboratory are examining the possible involvement of phosphatidylinositol 4 kinase type II (PI4KIIa/b) by direct or indirect (through other tetraspanin members) association with CD81N1/CD82N1. Phosphatidylinositol 4 kinase type II-a is a particularly attractive target because it is one of the few signalling molecules known to bind to tetraspanins and it is critically important for ER-Golgi homoeostasis. Moreover, its product, the phosphatidyl inositol-4-phosphate, functions as a docking/binding domain for multiple proteins (Weixel et al, 2005;Matteis and D'Angelo, 2007). The differential association of CD81 (binding) and CD82 (non-binding) (Yauch and Hemler, 2000) to PI4KII and the presence of CD81 in RPMI 8226 cells may result in diverse tetraspanin microdomains (qualitatively and quantitatively).
Our results can also be viewed from a totally different perspective. It is possible that ER-stress and autophagy death are instigated by the sheer burden of protein synthesis and trafficking, regardless of any specific signals that may originate from the tetraspanin fusion protein. It should also be taken into account that CD81/CD82 eGFP fusion proteins are regulated by the powerful CMV promoter, and that MM cells are already in a sensitised state of elevated protein synthesis and increased IRE/XBP1 expression. The differences between the pEGFP-N1-oriented plasmids and the empty or pEGFP-C1-oriented cloned tetraspanins may be attributed to decreased efficiency of folding and/or trafficking. If so, the transfected proteins would accumulate in the ER and eventually cause stress. Proof-of-concept will indicate that the mere burdening of ER -Golgi function may present a therapeutic target in protein-secreting cells such as myeloma. The effect of CD81N1 and CD82N1 on protein synthesis in the transfected MM cell lines is currently under investigation in our laboratory.
Targeting protein synthesis and secretion in MM has been addressed previously (Carew et al, 2006;Nakamura et al, 2006;Davenport et al, 2007;Patterson et al, 2008). Several drugs abrogating ER -Golgi stability have been used in vitro to induce MM cell death. In addition, a recent publication reported that the proteosome inhibitor, Bortezomib, used clinically causes protein accumulation in the ER by blocking ERAD as one  Inositol-requiring 1 and PERK were determined by semi-quantitative RT -PCR of XBP1s and total XBP1, respectively; JNK activation was deduced from rescue of tetraspanin-transfected cells (relative mock) by SP600125; and cell death was analysed by FACS based on PI exclusion of eGFP þ cells. The A panel is a schematic presentation and the B panel depicts mean fold change±s.e. of IRE1, PERK, and cell death (depicted in the legend under the graphs) relative to mock at different time points (x-axis). Significant elevation of IRE1 and PERK were at 24 h/18 h posttransfection in RPMI 8226 and CAG MM cells, respectively (Po0.05). As time progressed (48 h/24 h post-transfection in RPMI 8226 and CAG cells, respectively), it can be determined that, although IRE1 splicing of XBP1 was diminished, the PERK-regulated expression of total XBP1 was maintained. At least three separate experiments were conducted at each time point.
CD81N1/CD82N1 causes ER-stress and autophagic death in MM V Zismanov et al of its action modes (McCloskey et al, 2008). Our findings suggest the possibility that ER-stress may be achieved by increased protein synthesis and interestingly, this may be the path of least resistance, because it is in sync with built-in MM characteristics, hence the 'Achilles' heel' of these cells. Future studies will be needed to determine effective ways to induce catastrophically elevated and clinically achievable protein synthesis for therapeutic purposes.