CD147 reinforces [Ca2+]i oscillations and promotes oncogenic progression in hepatocellular carcinoma.

Oscillations in intracellular Ca2+ concentrations ([Ca2+]i) mediate various cellular function. Although it is known that [Ca2+]i oscillations are susceptible to dysregulation in tumors, the tumor-specific regulators of [Ca2+]i oscillations are poorly characterized. We discovered that CD147 promotes hepatocellular carcinoma (HCC) metastasis and proliferation by enhancing the amplitude and frequency of [Ca2+]i oscillations in HCC cells. CD147 activates two distinct signaling pathways to regulate [Ca2+]i oscillations. By activating FAK-Src-IP3R1 signaling pathway, CD147 promotes Ca2+ release from endoplasmic reticulum (ER) and enhances the amplitude of [Ca2+]i oscillations. Furthermore, CD147 accelerates ER Ca2+ refilling and enhances the frequency of [Ca2+]i oscillations through activating CaMKP-PAK1-PP2A-PLB-SERCA signaling pathway. Besides, CD147-promoted ER Ca2+ release and refilling are tightly regulated by changing [Ca2+]i. CD147 may activate IP3R1 channel under low [Ca2+]i conditions and CD147 may activate SERCA pump under high [Ca2+]i conditions. CD147 deletion suppresses HCC tumorigenesis and increases the survival rate of liver-specific CD147 knockout mice by regulating [Ca2+]i oscillations in vivo. Together, these results reveal that CD147 functions as a critical regulator of ER-dependent [Ca2+]i oscillations to promote oncogenic progression in HCC.


INTRODUCTION
Various cancer types differ significantly in terms of morphology, cell of origin, physiology, and pharmacology, but one thing common to most cancer cells is the requirement for intracellular Ca 2+ signaling to maintain malignant phenotypes: sustaining proliferation, resisting cell death, enabling replicative immortality, inducing angiogenesis, and activating invasion and metastasis [1][2][3][4]. Ca 2+ -signaling pathways are remodeled or deregulated in cancer that result in changes in their physiology and distinguish them from non-malignant cells. Remodeling or deregulation of Ca 2+ -signaling pathways can provide means by which cancer cells can overcome systemic anticancer defense mechanisms [5][6][7]. Numerous studies have now established that some cancers are associated with major changes in the expression of specific Ca 2+ channels and pumps and that inhibition of some of these proteins inhibits the oncogenicity of cancer cells [8][9][10]. Although the dysregulation of Ca 2+ signaling has been observed in tumor progression, the tumor-specific regulators of [Ca 2+ ] i oscillations are poorly characterized.
Depending upon the stimulus, Ca 2+ signals are transmitted in the form of repetitive time-dependent changes in its concentration, known as [Ca 2+ ] i oscillations [11]. [Ca 2+ ] i oscillations are proposed to convey information in their amplitude and frequency, leading to the activation of specific downstream targets in response to sustained stimulation by extracellular signaling molecules [12,13]. The simplest mechanism proposed to describe [Ca 2+ ] i oscillations involves an increase in cytosolic Ca 2+ (upstroke) due to the release of Ca 2+ from intracellular stores following activation of extracellular Ca 2+ influx; and a subsequent decrease in cytosolic Ca 2+ (downstroke) occurs due to the refilling of the stores with Ca 2+ following inactivation of extracellular Ca 2+ influx [14,15].
As a cancer biomarker, CD147 is highly expressed in hepatocellular carcinoma (HCC) but weakly expressed in normal liver tissues [16]. CD147 is also significantly expressed in various cancers and appears to have prognostic significance, rendering it a possible cancer-associated biomarker for pathological diagnosis, prognostic evaluation,targeted therapy and radioimmunoimaging of a broad range of cancer types. [17] Prior studies have demonstrated that CD147 plays important roles in the progression of HCC, including migration, proliferation, tumor recurrence and poor prognosis [16,17]. Our prior study indicated that CD147 activates the FAK-PI3K-calcium (Ca 2+ ) signaling pathway by interacting with a3b1 integrin and disrupts the NO/ cGMP-mediated negative regulation of store-operated Ca 2+ entry, thus increasing the intracellular level of Ca 2+ level in HCC cells [18,19]. In this study, we demonstrated that CD147, functions as a critical regulator of ER-dependent [Ca 2+ ] i oscillations to correlate with the progression in HCC.

CD147 promotes HCC cells invasion, migration and proliferation by regulating [Ca 2+ ] i oscillations
We investigate the expression of CD147 in a range of hepatoma cancer-derived and nonmalignant liver cells. CD147 was highly upregulated in hepatoma cancer-derived cell lines (Supplementary Figure 1A). To assess the effect of CD147 on [Ca 2+ ] i oscillations, we used lentiviral delivery of shRNA constructs to knock down expression of endogenous CD147 proteins in HepG2 and SMMC-7721 cells (Supplementary Figure 1B-1E), and examined changes in [Ca 2+ ] i over time in individual cells, and examined changes in [Ca 2+ ] i over time in individual cells. [Ca 2+ ]i oscillations were shown in both cells; interestingly, the amplitudes of the Ca 2+ oscillations induced by EGF stimulation were lower in CD147 knockdown cells relative to control cells. In addition, CD147 knockdown cells displayed a lower frequency of [Ca 2+ ] i oscillations than control cells. Approximately 20% of responding control cells displayed ten [Ca 2+ ] i spikes or more, whereas CD147 knockdown cells responded with fewer than seven spikes over the same time period ( Figure  1A & 1B). Together, these data suggested that CD147 modifies the nature of Ca 2+ signals: CD147 knockdown triggers lower-amplitude and lower-frequency [Ca 2+ ] i oscillations in HCC cells, leading to an overall reduction in Ca 2+ signal.

CD147 contributes to ER Ca 2+ release through FAK-Src pathway-mediated IP3R1 channel activation in human HCC cells
Normally, [Ca 2+ ] i oscillations consist of two phases, an upstroke phase, contributed by ER Ca 2+ release following the extracellular Ca 2+ influx, and a downstroke phase, due to ER Ca 2+ refilling. We next examined the roles of CD147 in ER Ca 2+ release. EGF stimulation induced an increase in Ca 2+ release in control cells, but this increase was minor in CD147 knockdown cells ( Figure 2A). Upon addition of extracellular Ca 2+ , control cells displayed increased [Ca 2+ ] i . This increase in [Ca 2+ ] i was blunted in CD147 knockdown cells ( Figure 2A). Together, these data suggest that CD147 may increase Ca 2+ release from ER stores and correspondingly increase the extracellular Ca 2+ influx. To rule out the possibility of a difference in ER Ca 2+ content, we measured the [Ca 2+ ] i of control cells and CD147 knockdown cells after treatment with ionomycin, which could transport calcium from the ER into the cytosol. In Ca 2+ -free medium, the peak levels of Ca 2+ released by ionomycin were identical in control cells and CD147 knockdown cells ( Figure 2B). The above results suggest that CD147 increases Ca 2+ release from ER stores but does not alter the ER Ca 2+ content.
IP3R is predominant ER Ca 2+ releasing channel. We found that CD147 knockdown reduced Ca 2+ release from ER stores. However, CD147 knockdown did not reduce Ca 2+ release from ER stores following IP3R inhibitor Xestospongin C (XeC) treatment ( Figure 2C). The result suggests that IP3R is involved in CD147-promoted Ca 2+ release. Moreover, we found that the major IP3R isoform expressed in HCC is IP3R1 (Supplementary Figure 2A) and CD147 did not influence IP3R1 expression levels ( Figure 2D). It is known that IP3R1 can be phosphorylated at tyrosine (Tyr), serine (Ser) and threonine (Thr). We then investigated the effects of CD147 on phosphorylated Tyr, Ser and Thr of IP3R1. The results showed that CD147 knockdown markedly decreased phosphorylated Tyr of IP3R1, however, CD147 knockdown did not change phosphorylated Ser and Thr of IP3R1 (Figure2E). To directly address this issue, we constructed a Tyr353mutated IP3R1 (IP3R1-Y353F). Immunoprecipitation (IP) showed that, compared with CD147 co-transfection with IP3R1 (WT), CD147 co-transfection with Tyr353mutated IP3R1 (Y353F) suppressed IP3R1 Tyr353 phosphorylation levels ( Figure 2F), indicating that CD147 may induce IP3R1 phosphorylation at Tyr353. To further detect whether the CD147-promoted cell oncogenicity was mediated by [Ca 2+ ] i oscillations in HCC cells, we tested the proliferation, invasion and migration in HepG2 cells which were transfected with IP3R1(WT) or IP3R1-Y353F vector. We found that CD147 promoted proliferation, invasion and migration in HepG2 cells transfected with IP3R1(WT) vector. However, CD147 did not apparently promote proliferation, invasion and migration in HepG2 cells transfected with IP3R1-Y353F vector (Supplementary Figure 2B-2D). The results strengthen the conclusion that CD147-related tumor behavior is [Ca 2+ ] i oscillation -related.
The tyrosine phosphorylation of IP3R1 is dependent on Src activity. We demonstrated that CD147 knockdown markedly decreased the Src phosphorylation in HepG2 cells ( Figure 2G). Then we found that the Src inhibitor markedly decreased Tyr phosphorylation of IP3R1, and CD147 knockdown did not further reduce Tyr phosphorylation of IP3R1 after Src inhibitor treatment ( Figure 2H). It is known that FAK activates Src through The expression levels of IP3R1 were examined. E. Cell lysates were immunoprecipitated with IP3R1 antibody and detected with a phospho-Tyr-specific antibody or a phospho-Ser-specific antibody or a phospho-Thr-specific antibody. F. Cell immunoprecipitates (IP) were analyzed with a general anti-phospho-Tyr antibody or IP3R1 antibody in cells expressing WT IP3R1 or IP3R1-Y353F mutant alone or in combination with CD147. G. The expression and phosphorylation levels of Src were examined. H. Analysis of phosphorylated Tyr in lysates from immunoprecipitates of IP3R1 in cells that were or were not pretreated with the Src inhibitor. I. The expression and phosphorylation levels of FAK were examined. J. Western blot analysis of phosphorylated Src in cells that were or were not pretreated with an FAK inhibitor. K. Analysis of phosphorylated Tyr in lysates from immunoprecipitates of IP3R1 in cells that were or were not pretreated with the FAK inhibitor. Bars represent each sample performed in triplicate, and the error bars represent the standard deviations. *p < 0.05 by Student's t-test. occupancy of the Src SH2 domain. We further confirmed that CD147 knockdown markedly decreased the FAK phosphorylation levels ( Figure 2I). Moreover, FAK inhibitor significantly decreased the Src phosphorylation and Tyr phosphorylation of IP3R1, and CD147 knockdown did not further reduce Src phosphorylation and Tyr phosphorylation of IP3R1 following FAK inhibitor treatment ( Figure 2G & 2K). These results suggested that CD147 promotes the Tyr phosphorylation of IP3R1 by activating FAK-Src pathway, thus inducing Ca 2+ release from ER stores. CD147 contributes to ER Ca 2+ refilling through CaMKP-PAK1-PP2A-PLB pathway mediated SERCA pump activation in human HCC cells To assess whether CD147 regulates the downstroke of [Ca 2+ ] i oscillations, the ER stores of HCC cells were depleted of Ca 2+ using BHQ (2,5-di-t-butyl-1,4ben-zohydroquinone), which could lead to ER store depletion. After completely depleting Ca 2+ from the ER, we removed BHQ and added 2mM Ca 2+ to initiate ER Ca 2+ refilling. Intriguingly, we found the downstroke of [Ca 2+ ] i oscillations in CD147 knockdown cells was slowed ( Figure 3A). These results suggest that CD147 may accelerate the downstroke of [Ca 2+ ] i oscillations. It is known that the downstroke of [Ca 2+ ] i oscillations was influenced by the refilling of the ER stores with Ca 2+ [20]. We measured the ER intraluminal Ca 2+ concentration ([Ca 2+ ] ER ) with the low-affinity fluorescent Ca 2+ indicator mag-fura-2 AM. We found that the refill rate was severely depressed in CD147 knockdown cells ( Figure 3B). These results indicated that CD147 may accelerate ER Ca 2+ refilling.
To address whether CD147-mediated Ca 2+ refilling occurs through the SERCA pump, thapsigargin (Tg), an irreversible SERCA inhibitor, was used. The addition of Tg completely abolished CD147-mediated Ca 2+ refilling, indicating that CD147-mediated Ca 2+ refilling requires SERCA pump activity ( Figure 3B). However, we found CD147 did not alter the expression of SERCA in HepG2 cells ( Figure 3C). It has been well documented that phospholamban (PLB) negatively regulates SERCA activity in the ER by direct association. The phosphorylation of PLB at Ser16 causes its disassociation from SERCA, and the subsequent relief of SERCA inhibition [9]. After initiating ER Ca 2+ refilling, we found that PLB Ser16 phosphorylation levels were markedly decreased in CD147 knockdown cells ( Figure 3D). Corresponding co-IP tests further demonstrated that CD147 knockdown significantly increased the binding of PLB to SERCA ( Figure 3E). These results demonstrated that CD147 causes PLB disassociation from SERCA and subsequent relief of SERCA inhibition by promoting the phosphorylation of PLB at Ser16. Activated PP2A and PP1 could associate with PLB and dephosphorylation of PLB at Ser16. The phosphorylation of protein phosphatase1 (PP1) and phosphatase type-2A (PP2A) leads to inhibition of PP1 and PP2A enzyme activity, respectively [11]. The result that CD147 knockdown markedly increased the phosphorylated PP2A but did not influence the phosphorylated PP1 suggested that CD147 may suppress PP2A activity ( Figure 3F). Moreover, CD147 knockdown did not further reduce phosphorylated PLB levels following PP2A inhibitor treatment ( Figure 3G). Correspondingly, co-IP tests further demonstrated that CD147 knockdown did not increase the binding of PLB and SERCA following PP2A inhibitor treatment ( Figure  3H). These results indicate that CD147 may promote PLB phosphorylation and enhance PLB dissociation from SERCA by suppressing PP2A activity.
The phosphorylation of p21-activated kinase 1 (PAK1) at Thr-423 induces a conformational change in PP2A that promotes PP2A dephosphorylation and increases its activity [11]. CD147 knockdown markedly increased the phosphorylation of Thr423 in PAK1 ( Figure  3I). PAK1 siRNA led to a marked reduction of the PAK1 levels in HepG2 cells (Supplementary Figure 3A & 3B), and CD147 knockdown did not further increase PP2A phosphorylation following PAK1 depletion ( Figure 3J). The result suggested that CD147 suppressed PP2A activity through inhibiting PAK1 phosphorylation.
Ca 2+ -dependent protein kinase phosphatase (CaMKP) associates with and dephosphorylates PAK1, leading to the inactivation of this kinase [21]. The CaMKP inhibitor treatment increased PAK1 phosphorylation in HepG2 cells ( Figure 3K). Treatment of cells with CaMKP inhibitor blocked the CD147-induced dephosphorylation of PAK1 and CD147-induced phosphorylation of PP2A and PLB ( Figure 3L). CD147 knockdown did not increase the binding of PLB to SERCA following CaMKP inhibitor treatment ( Figure 3M). These data indicate that CD147 may inhibit PAK1 phosphorylation by activating CaMKP. Together, CD147 activates CaMKP to inhibit PAK1 phosphorylation and subsequently suppresses PP2A activity. CD147-inactivated PP2A increases phosphorylation of PLB and enhances disassociation of PLB from SERCA, thus promoting SERCA pump activity and rapid ER Ca 2+ refilling.

High [Ca 2+ ] i inhibits CD147-induced IP3R1 channel activation by antagonizing CD147induced FAK phosphorylation
The above results demonstrated that CD147 contributes to both ER Ca 2+ release and ER Ca 2+ refilling in human HCC cells. To investigate how CD147 is able to achieve this biphasic effect, we tested the role of CD147 in low [Ca 2+ ] i conditions and high [Ca 2+ ] i conditions, as time-dependent changes in [Ca 2+ ] i concentration is the key feature of [Ca 2+ ] i oscillations. We generated low levels of [Ca 2+ ] i with BAPTA-AM, which is a membrane-permeable Ca 2+ chelator, and generated high levels of [Ca 2+ ] i with Tg, which acts to irreversibly deplete ER Ca 2+ stores and then activating a sustained Ca 2+ influx.
We found that CD147 knockdown markedly decreased the tyrosine phosphorylation of IP3R1 in low [Ca 2+ ] i conditions, but had no effect in high [Ca 2+ ] i conditions ( Figure 4A). Furthermore, CD147 knockdown markedly decreased phosphorylated Src and phosphorylated FAK levels in the absence but not in the presence of high [Ca 2+ ] i ( Figure 4B & 4C). To further confirm these results, we examined FAK phosphorylation after a switch from low [Ca 2+ ] i to high [Ca 2+ ] i . The level of phosphorylated FAK decreased after the culture medium was switched from low to high [Ca 2+ ] i ( Figure 4D, lane 2,3) but returned to its initial high levels after Ca 2+ was chelated by BAPTA-AM ( Figure 4D, lane 4). These results demonstrate that the dephosphorylation of FAK by elevated Ca 2+ antagonizes the effect of CD147 inducing FAK phosphorylation.
It has been well documented that Ca 2+ can activate ERK1/2. Activated ERK1/2 may stimulate FAK dephosphorylation [12]. We found that phosphorylated ERK1/2 was less in low [Ca 2+ ] i conditions and increased in high [Ca 2+ ] i conditions ( Figure 4E). The membrane-permeant ERK1/2 inhibitor, U0126 inhibited the dephosphorylation of FAK in high [Ca 2+ ] i conditions ( Figure 4F). In addition, U0126 enabled CD147 to phosphorylate FAK in high [Ca 2+ ] i conditions ( Figure 4G). Previous studies showed that ERK1/2 induces PTP-PEST phosphorylation. Isomerized PTP-PEST interacts with and dephosphorylates FAK at Y397. To assess whether PTP-PEST regulates the levels of FAK dephosphorylation, HepG2 cells were depleted of PTP-PEST with specific siRNA. Introduction of PTP-PEST specific siRNA in HepG2 cells resulted in marked reduction of PTP-PEST mRNA and protein (Supplementary Figure 4A & 4B). We found that the depletion of PTP-PEST was shown to increase FAK phosphorylation in HepG2 cells ( Figure  4H). The phosphorylation of FAK induced by CD147 was inhibited in high [Ca 2+ ] i conditions, but PTP-PESTspecific siRNA inhibited the dephosphorylation of FAK in the presence of high [Ca 2+ ] i conditions and enabled CD147 to induce FAK phosphorylation ( Figure 4I). These results demonstrate that in high [Ca 2+ ] i conditions, Ca 2+ -activated ERK1/2 promotes FAK dephosphorylation by PTP-PEST, thus antagonizing the phosphorylation of FAK by CD147 and blocking the activation of the IP3R1 channel by CD147. Collectively, these data show that CD147 may induce IP3R1 channel activation and increase Ca 2+ release only under low [Ca 2+ ] i conditions.

Low [Ca 2+ ] i inhibits CD147-induced SERCA pump activation by impairing CD147-induced CaMKP activation
To determine whether CD147 mediated ER Ca 2+ refilling was affected by changing [Ca 2+ ] i , the phosphorylation of PLB and PAK1 were examined in different [Ca 2+ ] i conditions. We found that CD147 knockdown markedly decreased the phosphorylation of PLB in the presence of high [Ca 2+ ] i but had no effect in the presence of low [Ca 2+ ] i ( Figure 5A). Furthermore, CD147 knockdown markedly increased phosphorylated PAK1 levels in the presence but not in the absence of high [Ca 2+ ] i ( Figure 5A). These results suggested that CD147 induces PAK1 dephosphorylation only under high [Ca 2+ ] i conditions. An analysis of CD147 complexes isolated from HepG2 cells confirmed that CD147 was co-immunoprecipitated with endogenous CaMKP under high [Ca 2+ ] i conditions but not low [Ca 2+ ] i conditions ( Figure 5B). Immunofluorescent staining further showed the constitutive membrane colocalization of CD147 and CaMKP under high [Ca 2+ ] i conditions ( Figure 5C). We then observed that CD147 knockdown did not alter the CaMKP expression levels ( Figure 5D). CaMKP activity tests showed that CaMKP incubated with CD147 was not significantly activated in either the presence or the absence of Ca 2+ (Figure 5E lane 2, 3). CaMKII, a Ca 2+active multifunctional protein kinase, which was found to phosphorylate and activate CaMKP [22,23], was added to the reaction mixture. CD147 increased CaMKP activation in the presence of CaMKII and Ca 2+ ( Figure 5E  lane 4, 5). The dephosphorylation of PAK1 induced by CD147 in high [Ca 2+ ] i conditions was largely blocked by pretreatment with CaMKII inhibitor KN-93 ( Figure 5F). In addition, the CD147-CaMKP interaction, which was found to be constitutive under high [Ca 2+ ] i conditions, was disrupted in cells treated with KN-93 ( Figure 5G). These results suggested that CD147 interacts with and activates CaMKP in the presence of high [Ca 2+ ] i and that this     CD147, also named basigin in mice. Mice with conditional null allele for basigin were generated in our lab [24]. To analyze whether CD147-promoted tumor development were mediated by [Ca 2+ ] i oscillations in vivo, we compared the phenotypes of livers with deletion of basigin. Using the Cre-loxP approach, basiginfloxed mice were crossed with Alb-Cre transgenic mice to derive Alb-Cre; Bsg fl/fl mice as described [24]. We designed PCR primers to amplify the deleted allele (using primers P1 : 5'-CTGGAACTCCTAGCAATC-3' and P2 : 5'-AGGTGGGTTTTCTGTAAGGT-3', 614-bp product in length) and undeleted allele (using primers P3 : 5'-CTCTGGGACTCAATGTGTGT-3' and P2: 5'-AGGTGGGTTTTCTGTAAGGT-3', 405-bp product in length). The PCR results showed no deleted band in liver obtained from control Bsg fl/fl mice, but a strong deleted band in genomic DNA from liver of the Alb-Cre; Bsg fl/fl mice ( Figure 7A). Real-time quantitative PCR and Western blotting showed that basigin was normal in livers of control Bsg fl/fl mice, but was undetectable in liver of the Alb-Cre; Bsg fl/fl mice ( Figure 7B & 7C). These results demonstrated that the Bsg fl allele could be specifically deleted in livers in Alb-Cre; Bsg fl/fl mice.

CD147 increases the amplitude of [Ca 2+ ] i oscillations through the FAK pathway and increases the frequency of [Ca 2+ ] i oscillations through the CaMKP pathway
The control Bsg fl/fl and Alb-Cre; Bsg fl/fl mice were injected with DEN. Bsg fl/fl mice and Alb-Cre; Bsg fl/fl mice were both randomized into the two groups: the BAPTA-AM treatment group and the untreated group (Control). The results showed that deletion of basigin markedly reduced tumor weight and the numbers of nodules in vivo. But deletion of basigin did not reduce tumor weight and the numbers of nodules following BAPTA-AM treatment ( Figure 7D & 7E). We also investigated the survival rate in Bsg fl/fl mice and Alb-Cre; Bsg fl/fl mice with established orthotopic hepatomas after treatment with BAPTA-AM. The results showed that the survival rate were markedly increased in Alb-Cre; Bsg fl/fl mice compared with Bsg fl/fl mice. However, deletion of basigin did not increase the survival rate following BAPTA-AM treatment ( Figure  7F) (Figure 8). Deletion of CD147 inhibites HCC development, thus increasing the survival rate of Alb-Cre; Bsg fl/fl mice. Therefore, CD147 functions as a critical regulator of ER-dependent [Ca 2+ ] i oscillations to promote HCC oncogenicity.
Ca 2+ -mediated signaling pathways have also been shown to play important roles in carcinogenesis such as transformation of normal cells to cancerous cells, tumor formation and growth, angiogenesis, invasion and metastasis [1][2][3][4]. Our findings that CD147 regulates [Ca 2+ ] i oscillations may account for the various biological functions of CD147 in HCC tumorigenesis.
IP3R1 is an intracellular Ca 2+ -release channel predominantly located on the ER and responsible for a controlled release of Ca 2+ in the cytoplasm. Functional experiments confirm that tyrosine phosphorylation of IP3R1 leads to a 5-fold increase in affinity for IP3 and promotes IP3R1 open probability. We demonstrated that during ER Ca 2+ release stage, the higher level of IP3R1 phosphorylation in the presence of CD147, which is mediated by the FAK-Src pathway, would lead to a high level of cytoplasmic calcium following the main Ca 2+ release from the ER and subsequently a greater level of Ca 2+ influx, thus enhancing the amplitude of the Ca 2+ oscillations. Refill of Ca 2+ into the ER is predominantly mediated by SERCA pump, which is widely distributed within the ER of most cell models. PLB regulates the active transport of Ca 2+ into the ER via a reversible inhibitory association with the SERCA. Experiments confirm that phosphorylation of PLB at Ser-16 causes relief of SERCA inhibition and permits rapid refill of Ca 2+ into the ER. We demonstrated that during ER Ca 2+ refilling stage, CD147 inhibits PAK1-PP2A pathway activity, promoting PLB phosphorylation at Ser16 and dissociation from SERCA, thus enhancing SERCA activity and rapid Ca 2+ refilling into the ER, therefore increasing the frequency of [Ca 2+ ] i oscillations. It is likely that the high expression of CD147 in HCC cells promotes the recycling of the internal Ca 2+ store, both enhancing Ca 2+ release and promoting more efficient Ca 2+ refilling. Thus, differences in the ability of the ER Ca 2+ stores to recycle Ca 2+ can account for the apparent differences in the mechanism of CD147-promoted [ In the present study, we discovered that CD147 acts as a critical regulator of ER-dependent [Ca 2+ ] i oscillations to promote oncogenic activities and influence the progression of HCC. The inhibition of CD147 may therefore represent a novel approach to inhibit the dysregulation of Ca 2+ signaling and HCC metastasis and proliferation. In addition to its effects in HCC cells, we have also established that CD147 is highly expressed in other cancers, including breast, lung and bladder cancer gliomas and laryngeal squamous cell, ovarian, renal cell and skin carcinomas [17]. As CD147 is a widely expressed cancer biomarker, its molecular signaling cascade, as well as the mechanism and consequences of its activation in HCC, could also exist in other molecule in developing therapies. In this regard, CD147 may represent a novel anti-cancer target in tumor treatments.

Cytoplasmic [Ca 2+ ] i and ER [Ca 2+ ] i measurements
Cells were plated onto coverslips coated with L-polylysine. For Cytoplasmic [Ca 2+ ] i measurements, the coverslips were mounted on an imaging chamber and washed with Ca 2+ -free Ringer's solution; the cells were then incubated with Fluo-8/AM (0.1 mg/ml) for 40 min at 37°C. Ca 2+ signaling was induced by the addition of the indicated reagents. The ER [Ca 2+ ] i was measured as described previously [26]. Briefly, cells were loaded by incubation in culture medium containing AM derivatives of Mag-Fura2 for 60 minutes at 37°C and then removed the cytoplasmic fraction of the dye. The cells were subsequently treated with BHQ or Tg to deplete the ER Ca 2+ stores, and 2mM Ca 2+ was added to initiate ER Ca 2+ refilling. [Ca 2+ ] i were recorded from Fluo-8/AM or Mag-Fura 2/AM loaded cells, which were excited at 340 nm and 380 nm and imaged with 510 nm filters, and the F340/F380 ratios of the fluorescence intensities were calculated after subtraction of background fluorescence and correction for bleaching. The fluorescence signal was monitored and recorded by an FV300 laser scanning confocal microscope (Olympus, Japan, Tokyo). All experiments were repeated 3 to 6 times. Statistical significance was determined with Student's t-test analysis; p < 0.05 was considered significant. All data are shown as the average ± SEM. www.impactjournals.com/oncotarget

In vitro invasion assays
The assay was performed using chambers with polycarbonate filters (8 μm pore size; Millipore). The upper side of a polycarbonate filter was either coated or not coated with Matrigel to form a continuous thin layer. HCC cells (1×10 5 ) were resuspended in 300 μL of 0.1% serum medium and added to the upper chamber. The lower chamber was filled with 10% FBS medium (200 μL). After 24 h incubation, the cells on the upper chamber of the filter were removed with a cotton swab, and the cells on the underside were stained and counted.

Wound healing assay
HCC cells (2×10 6 ) were plated in six-well plates and cultured to approximately 90% confluence. The cells were scraped with a pipette tip, washed several times in serumfree medium, and then examined under a phase contrast microscope (Olympus, Japan, Tokyo). The cells were refed with 10% FBS medium for 24 h, and images were obtained.

Cell growth tests
HCC cells were cultured in 96-well plate (1×10 4 per well) for 24 h. Each condition was performed in triplicate. After incubation, the cells were harvested and counted at every 24 h. The doubling time (T) was calculated according to the formula T = t×log 2/ (log N t -log N 0 ). N t indicates the number of cells after culture for t hours, N 0 indicates the number of cells at the beginning of the culture, and t indicates the culture time.

Western blot analysis
Cells were lysed in 1% OG buffer (20 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1% OG, 1 mM EDTA, 10 µg/ml leupeptin, 2 µg/ml aprotinin and 1 mM PMSF). A BCA Protein Assay Kit (Pierce Biotechnology, Rockford, Illinois) was then used to determine the total protein concentration, and equal amounts of protein were separated by 10% SDS-PAGE and transferred to a polyvinylidene fluoride (PVDF) microporous membrane (Millipore, Billerica, MA). The membrane was blocked with 5% non-fat milk and then incubated for 2 h at room temperature with the designated antibody. The Western-Light chemiluminescent detection system (Applied Biosystems, Foster, CA) was used for immunodetection.

Co-immunoprecipitation
Cells were washed and resuspended in NPBS and then lysed in 1% OG buffer. A BCA Protein Assay Kit (Pierce Biotechnology, Rockford, IL, USA) was then used to determine the total protein density. Lysates (1 mL) were immunoprecipitated by incubation with 2 µg of the designated antibody and 25 µL of protein A-agarose overnight at 4 °C, followed by Western blotting with the designated antibody for 2 h at room temperature. The Western-Light chemiluminescent detection system (Applied Biosystems, Foster City, CA, USA) was used for immunodetection.

Protein phosphatase assay using MUP as a substrate
The protein phosphatase assays were carried out at 30°C for 20 min in a reaction mixture containing 50 mM Tris-HCl (pH 8.0), 20 mM dithiothreitol, 10 mM MnCl 2 , 0.01% Tween 20, 25 μM Methylumbelliferyl Phosphate (MUP), and 1 μg of CaMKP. The reaction was started by the addition of CaMKP and terminated by the addition of EDTA. MUP would be converted to the soluble fluorescent reaction product methylumbelliferone when they are hydrolyzed by CaMKP. The fluorescence of methylumbelliferone present in the reaction mixture was measured by a spectrofluorometer (Molecular devices, CA, USA) with excitation at 360 nm and emission at 440 nm.

Animal model
The conditional null allele for basigin in mice were generated in our lab [24]. Alb-Cre transgenic mice were purchased from Model Animal Research Center of Nanjing University. Basigin-floxed mice were crossed with Alb-Cre transgenic mice to generate Alb-Cre;Bsg fl/ fl mice (backcrossed to C57BL/6 background for more than eight generations). Male mice were injected intraperitoneally with 25 mg/kg body weight of DEN (Sigma-Aldrich, Munich, Germany) at 14 days of age and then 90 mg/kg body weight of DEN at 6 weeks. Mice were killed at 5 months after birth. All of the animal work was approved by the Animal Care and Use Committee of the Fourth Military Medical University and handled in strict accordance with good animal practice as defined by the relevant national animal welfare bodies.

Statistical analysis
Statistical analysis was performed using SPSS 13.0 statistical software. The Student's t-test or one-way ANOVA were used to evaluate the statistical significance in the groups. The data are presented as mean ± SD and p < 0.05 was considered as significant.