Cluster of differentiation 44 (CD44) is a single-pass transmembrane glycoprotein that is a widely distributed cell-surface adhesion molecule. CD44 undergoes ectodomain cleavage by membrane-associated metalloproteinases in breast cancer cells. Cleavage plays a critical role in cancer cell migration by mediating the interaction between CD44 and the extracellular matrix. To explore inhibitors of CD44 ectodomain cleavage, we developed two bioluminescent sensors for the detection of CD44 ectodomain cleavage. The sensors were designed as two-transmembrane proteins with split-luciferase fragments, one of which was cyclized by protein trans-splicing of a DnaE intein. These two sensors emit light by the cyclization or the spontaneous complementation of the luciferase fragments. The luminescence intensities decreased upon cleavage of the ectodomain in breast cancer cells. The sensors revealed that castanospermine, an α-glucosidase inhibitor, suppressed the ectodomain cleavage of endogenous CD44 in breast cancer cells. Castanospermine also inhibited breast cancer cell invasion. Thus, the sensors are beneficial tools for evaluating the effects of different inhibitors.
Breast cancer is the most frequent cancer diagnosed in women and has a global incidence of 24.5% (Sung et al., 2021). Triple-negative breast cancer (TNBC) is considered to be the most aggressive form and is negative for the estrogen receptor, progesterone receptor and human epithelial growth factor receptor 2 (Gadag et al., 2020). The recurrence rate of TNBC tends to be higher after diagnosis, and the subtypes of TNBC are linked to the majority of breast cancer-related deaths (Sulaiman et al., 2019). Among the subtypes, the mesenchymal and mesenchymal stem-like subtypes show similar profiles of gene expression, particularly those related to cell motility, cellular differentiation and growth pathways (Lehmann et al., 2011). These pathways are critical for the stem cell-like properties of the CD44+/CD24low population in nontumorigenic mammary epithelial cells (Mani et al., 2008). Because tumorigenic breast cancer cells are enriched in the CD44+/CD24−/low population (Al-Hajj et al., 2003), CD44 appears to be an effective target for the treatment of these TNBC subtypes (Chen et al., 2018).
Cluster of differentiation 44 (CD44), a single-pass transmembrane glycoprotein that functions as a major cellular adhesion molecule for hyaluronic acid, is widely expressed throughout the body and is involved in lymphocyte homing and activation, cell–matrix interactions, cell migration, tumor-cell invasion, angiogenesis and metastasis (Nagano and Saya, 2004; Senbanjo and Chellaiah, 2017). The CD44 gene contains 20 exons in mice and humans, wherein ten variant exons are alternatively spliced in various combinations, which generates numerous CD44 splice variant isoforms (CD44v) (Naor et al., 1997). The CD44 standard isoform (CD44s) is the most abundant form and splices out all variant exons (Azevedo et al., 2018). The ectodomain of CD44 undergoes proteolytic cleavage by membrane-type 1 matrix metalloproteinase (MT1-MMP; also known as MMP14) (Nakamura et al., 2004). Cleavage is critical for migratory stimulation (Kajita et al., 2001). In addition, the CD44 ectodomain is cleaved by a disintegrin and metalloproteinase domain-containing protein 10 (ADAM10), which is activated via extracellular Ca2+ influx (Nagano et al., 2004). ADAM10-mediated CD44 cleavage is also induced by epidermal growth factor through activation of the small GTPase Rac1 and mitogen-activated protein kinase (Murai et al., 2006). Furthermore, ADAM17 activated by protein kinase C (PKC; also known as PRKC) and Rac1 cleaves the ectodomain of CD44 (Nagano et al., 2004). In patients with cancer, the concentrations of the cleaved CD44 fragments in serum are correlated with the tumor burden and metastasis (Guo et al., 1994). In breast carcinoma, CD44 ectodomain cleavage has been detected in both CD44s and CD44v, and all the samples from the patients expressing CD44v exhibit CD44 cleavage products (Okamoto et al., 2002). Additionally, CD44 cleavage in tumors expressing CD44v is enhanced in lung, colon and ovarian carcinomas (Okamoto et al., 2002). Moreover, CD44 ectodomain cleavage triggers intramembranous cleavage caused by presenilin 1-dependent γ-secretase (Lammich et al., 2002; Murakami et al., 2003), and this liberated intracellular domain (ICD) translocates into the nucleus and thereby potentiates gene transcription of the CD44 gene (Medrano-González et al., 2021; Okamoto et al., 2001). Overexpression of CD44 ICD enhances transcriptional activation of the stemness factors Sox2 and Oct4 and increases the size and number of mammospheres, which promotes tumorigenesis and metastasis (Cho et al., 2015). Thus, the inhibition of CD44 ectodomain cleavage could suppress breast cancer progression. According to a previous report, tunicamycin (an N-linked glycosylation inhibitor) suppresses the cleavage of the CD44s ectodomain in the glioblastoma cell line U251MG (Okamoto et al., 1999). Therefore, we focus on glycosylation inhibitors to inhibit CD44 ECD cleavage.
Bioluminescent sensors are widely applied for the identification of specific inhibitors from the high-throughput screening of chemical libraries because the sensors generally exhibit a low background, a high signal-to-noise ratio and signals with wide dynamic ranges (Ozawa et al., 2013). A technique named split-luciferase complementation is based on the reconstitution of two unstructured luciferase fragments to recover the enzymatic activity, which is beneficial for the analyses of protein–protein interactions and protein localization and for high-throughput chemical screening (Wehr and Rossner, 2016). Firefly luciferase is extensively used in biological experiments because the substrate D-luciferin has higher chemical stability than Cypridina, krill and dinoflagellates (Shimomura, 2012). In addition, the emission wavelength of D-luciferin with firefly luciferase (Photinus pyralis) is relatively long (λmax=562 nm). These advantages allow for the development of a cyclic luciferase to monitor caspase-3 activation, and this developed luciferase has been widely used for in vitro and in vivo measurements (Kanno et al., 2007; Niu et al., 2013; Ozaki et al., 2012). Another engineered luciferase derived from a luminous deep-sea shrimp, NanoLuc (Nluc), requires only oxygen and the substrate furimazine for the luciferin–luciferase reaction (Hall et al., 2012). Unlike luciferases that use D-luciferin as a substrate, such as fireflies and click beetles, the Nluc luciferin–luciferase reaction does not require adenosine 5′-triphosphate or Mg2+. Hence, Nluc split proteins (Dixon et al., 2016) are ideal for the detection of extracellular events, such as CD44 ectodomain cleavage. No sensor has been generated to detect the CD44 ectodomain cleavage.
Here, we developed split-luciferase sensors to detect CD44 ectodomain cleavage in two breast cancer cell types; MDA-MB-231 cells were selected as a TNBC cell line, and MCF-7 cells, which are estrogen receptor positive (Fernandez et al., 2013), were used for comparison with the TNBC cells. The two sensors were composed of two-transmembrane proteins containing CD44s and split-Nluc fragments: one is cyclized by protein trans-splicing of a naturally split cyanobacterium DnaE intein (Evans et al., 2000; Scott et al., 1999; Wu et al., 1998), and the other maintains a linear (U-shaped) form. The cyclization of the sensor is assumed to enhance the complementation of Nluc fragments. Nluc fragments in the linear sensor may be spontaneously complemented by post-translational protein folding of the sensor. Our developed bioluminescent sensors revealed that an α-glucosidase inhibitor, castanospermine (Saul et al., 1985, 1984), is a candidate inhibitor of CD44 ectodomain cleavage. Therefore, we investigated the effects of castanospermine on the cleavage of the endogenous CD44 ectodomain in breast cancer cells.
Principle of the detection of CD44s ectodomain cleavage using bioluminescent sensors
To evaluate the cleavage of the CD44 ectodomain, we designed bioluminescent sensors using split-luciferase complementation techniques (Fig. 1A). The two split-luciferase fragments of Nluc (named NlucN and NlucC) were connected to CD44s in the vector, including the signal peptide sequence at the amino terminus. The vector harbors sequences of two CD44s transmembrane domains. To efficiently complement the luciferase fragments, the bioluminescent sensor was cyclized via a protein trans-splicing reaction with naturally split DnaE intein, DnaEn and DnaEc (Fig. 1B). The proximity between NlucN and NlucC upon cyclization allows for spontaneous complementation and thus for efficient recovery of enzymatic activity (named Cecto-C, a circular sensor for CD44 ectodomain cleavage detection) (Fig. 1A,B; Table S1). Cecto-C includes seven predicted N-linked glycosylation sites: six are located in CD44s, and one is found in the DnaEn intein (Table S2).
To confirm the cyclization of Cecto-C, a noncyclized sensor (named Cecto-L, a linear sensor of Cecto), which possessed an amino acid (aa) mutation in the DnaEn intein (C1A: first Cys is replaced with Ala), was constructed. In principle, it is impossible to induce protein trans-splicing using the mutated DnaEn intein. In addition, the N-linked glycosylation site, asparagine, in the DnaEn intein was replaced with isoleucine (N26I) to prevent the glycosylation of Cecto-L (Fig. 1A). Cecto-L may recover luciferase enzymatic activity through proximity between luciferase fragments caused by post-translational protein folding of the sensor (Fig. 1B). To further confirm the cyclization of Cecto-C by trans-splicing of the DnaE intein, two HA epitope tags were added adjacent to the DnaEn and DnaEc inteins in Cecto-C (named Cecto-HA) (Fig. 1A). The HA epitope tags are removed by protein trans-splicing of the DnaE intein; that is, Cecto-HA retaining the HA epitope tags indicates that cyclization has not occurred after the expression of the construct in the cells.
To ascertain the cleavage of the CD44 ectodomain in Cecto-C and Cecto-L by MT1-MMP, we constructed a vector harboring the MT1-MMP sequence with a 3× FLAG epitope tag (Fig. 1C). Additionally, to evaluate whether the difference in MT1-MMP activation affects the ectodomain cleavage of the two sensors, we constructed a vector harboring the mutated MT1-MMP sequence (Fig. 1C). PGD↓L50 and RRKR111↓Y112 cleavage is required for the activation of MT1-MMP in cancer cells (Golubkov et al., 2007, 2010). Thus, these sites in MT1-MMP were mutated to L50D and AAAA111. In addition, the E240A mutation was added as a catalytically inert mutant (Rozanov et al., 2001).
Cecto-C, Cecto-L and Cecto-HA are localized in the cell membrane of breast cancer cells
We investigated the cellular localization of the three sensors. Indirect immunofluorescence was performed using MCF-7 and MDA-MB-231 breast cancer cell lines with and without cell permeabilization (Fig. 2; Figs S1 and S2). Both the V5 and Myc epitope tags in Cecto-C, Cecto-L and Cecto-HA were recognized outside the cell membrane in the nonpermeabilized breast cancer cells (Fig. 2A,B; Fig. S2A). The results indicate that the three sensors are transmembrane proteins with tags outside the cells. Although the sensors were mainly localized on the cell membranes, they were also detected inside the permeabilized cells (Fig. 2C,D; Fig. S2B). A portion of the overexpressed sensors accumulated in the endoplasmic reticulum (ER) because the expression of the three sensors was strongly driven by a cytomegalovirus promoter. Thus, the constructed sensors were localized on the cell membrane due to the presence of the two transmembrane domains.
Cecto-C and Cecto-HA are cyclized or oligomerized by protein trans-splicing of a naturally split DnaE intein
We subsequently confirmed the endogenous CD44 protein abundance in untransfected MCF-7 and MDA-MB-231 cells by western blotting (Fig. S3A). We detected both CD44s and CD44v in both cell lines, although CD44s was more abundant in MDA-MB-231 cells than in MCF-7 cells. In these cell lines, we evaluated the cyclization of Cecto-C and Cecto-HA using Cecto-L as a noncyclized control (Fig. 3). The cells were transfected with each of the three sensors, and the V5 or HA epitope tag was recognized with their respective antibodies. The protein bands of Cecto-L detected by the anti-V5 antibody indicated noncyclized sensors. A band shift to a high molecular mass (the expected molecular mass without the signal peptide sequence is ∼87 kDa) was observed, and this shift possibly originated from certain glycosylations in CD44s. Moreover, several protein bands of Cecto-HA were detected using an anti-HA antibody. These bands do not contain the cyclized form of Cecto-HA because two HA tags in the sensor were removed by protein trans-splicing of the DnaE intein. The two bands showing lower molecular masses are noncyclized monomers, and the molecular mass was the same as that of Cecto-L. The other bands of Cecto-HA imply the presence of sensors with unexpected trans-splicing reactions, which show the tandemly oligomerized forms of the sensors because the tandemly oligomerized forms retain one HA tag per sensor. Thus, the protein bands of Cecto-C and Cecto-HA, with the exception of the noncyclized monomers and oligomers, indicate cyclized sensors. The molecular masses of Cecto-C and Cecto-HA without the signal peptides were ∼87 kDa and 89 kDa, respectively, whereas those of the cyclized sensors were ∼69 kDa. Glycosylation additionally increases the molecular mass by ∼50–60 kDa if endogenous CD44s has an estimated molecular mass of 85–95 kDa (Chen et al., 2018; Kajita et al., 2001; Okamoto et al., 2002); the molecular mass of CD44s, with the exception of the signal peptide, is ∼37 kDa in the absence of glycosylation. Therefore, the cyclized CD44 sensor containing glycosylations will be ∼120–130 kDa. In the case of Cecto-C and Cecto-HA, the molecular masses might be affected by the glycosylation to each N-linked glycosylation site in the DnaEn intein. In addition, the bands shift to a higher molecular mass due to cyclization (Kanno et al., 2007). In summary, the two bands that were not detected by the anti-HA antibody represent the cyclized sensors.
Cecto-C and Cecto-L are cleaved by overexpressed MT1-MMP, which leads to a decrease in the luminescence intensities
To examine the variation in the luminescence intensities of Cecto-C and Cecto-L due to their cleavage, breast cancer cells were then co-transfected with each of the sensors along with MT1-MMP, MT1-MMP (Mut) or mock (empty vector), and the luminescence intensities were measured (Fig. 4A,B). Compared with the respective mock controls, the luminescence intensity of the cells expressing MT1-MMP was significantly decreased in both cancer cell lines. Of the two sensors, the luminescence intensity of Cecto-C with MT1-MMP showed a larger decrease than that of Cecto-C with mock in MCF-7 cells, although a difference in the luminescence intensity between Cecto-C and Cecto-L was not observed in MDA-MB-231 cells. No difference between the basal luminescence intensity of Cecto-C and Cecto-L with mock was detected 24 h after transfection in either cancer cell line (Fig. 4C). On the other hand, 24 h and 48 h after the transfection of Cecto-C in both cell lines, the luminescence intensity of the cells expressing MT1-MMP (Mut) was significantly higher than that of cells expressing MT1-MMP (Fig. 4A,B). The luminescence intensity of Cecto-L with MT1-MMP (Mut) was significantly higher than that of Cecto-L with MT1-MMP in both cell lines at 24 h and 48 h, with the exception of MDA-MB-231 cells at 24 h. To exclude the possibility that the decrease in luminescence intensity was related to cell damage, the live cells were counted using Trypan Blue staining in the same experimental condition as Fig. 4A and B (Mock and MT1-MMP). No significant difference in the number of live cells was observed (Fig. 4D). These results indicate that the luminescence intensity of the two sensors was reduced by their cleavage.
To further investigate the cleavage of Cecto-C by MT1-MMP, we performed western blotting under the same experimental conditions as the luminescence measurement. First, we detected endogenous MT1-MMP proteins in MCF-7 and MDA-MB-231 cells (Fig. S3B). Endogenous MT1-MMP proteins are theoretically detected as proenzymes (∼65.8 kDa), intermediates (∼60.7 kDa, at the cleavage site between the 49th and 50th aa) and mature enzymes (∼53.8 kDa, at the cleavage site between the 111th and 112th aa). The three expected protein bands of endogenous MT1-MMP were detected in MDA-MB-231 cells, whereas no bands were observed in MCF-7 cells. In contrast, the degradation product (∼40 kDa) was detected in both cell lines, and this finding was almost the same as previously reported results (∼45 kDa) (Golubkov et al., 2010). Subsequently, Cecto-C or Cecto-L, along with MT1-MMP or mock, was expressed in breast cancer cells. Cecto-L was used for evaluation of the noncyclized sensors (Fig. 4E). In both cancer cell lines, one noncyclized monomer protein band disappeared under MT1-MMP-overexpression conditions, whereas the other protein bands of Cecto-C were attenuated. Of these two monomers, the larger-sized monomer caused by N-linked glycosylations may represent a protein structure to which the endogenous metalloproteinases are more accessible, or, alternatively, the glycosylation itself may be required for the recognition from endogenous metalloproteinases. MT1-MMP with a 3× FLAG epitope tag was detected in both cell lines. To confirm whether cyclized Cecto-C is cleaved by MT1-MMP, we evaluated the cleaved form by western blotting. MDA-MB-231 cells were transiently co-transfected with Cecto-C with mock or MT1-MMP for 24 h and then treated with bafilomycin A1 for 24 h to inhibit the protein degradation. Bafilomycin A1 is a specific potent inhibitor of vacuolar-type H+-ATPases, which blocks lysosomal acidification and protein degradation in lysosomes (Yoshimori et al., 1991). The protein bands of the harvested cells were shown (Fig. 4F). When the cyclized Cecto-C is cleaved, the upward band shift does not occur. Therefore, the molecular mass of the western blotting band is ∼120-130 kDa. We detected the cleaved cyclized monomers in the cells overexpressing MT1-MMP under bafilomycin A1 treatment. The increase in the cleaved Cecto-C protein coincided with the decrease in the luminescence intensity of Cecto-C shown in Fig. 4B. The result indicates that the cyclization of Cecto-C elevates the bioluminescence intensity. These results suggest that Cecto-C and Cecto-L were cleaved by the overexpression of MT1-MMP, and this cleavage reduced the luminescence intensity of the sensors. Although we can use the two sensors to evaluate CD44 ectodomain cleavage, Cecto-C was selected for the subsequent experiments with MCF-7 and MDA-MB-231 cells due to its slightly higher sensitivity in MCF-7 cells.
BB-94 suppresses the cleavage of Cecto-C caused by endogenous membrane-associated metalloproteinases and overexpressed MT1-MMP
To ascertain whether the cleavage of Cecto-C is suppressed by the MMP inhibitor BB-94, breast cancer cells were transfected with Cecto-C along with MT1-MMP or mock. After 24 h of incubation, the cells were treated with BB-94 or vehicle for 24 h. The luminescence intensity of the cells transfected with Cecto-C and mock was significantly increased by BB-94 treatment compared with that of the cells treated with the vehicle, and this finding was obtained for both breast cancer cell lines (Fig. 5A). These results indicate that the suppression of sensor cleavage by BB-94 is attributed to endogenous metalloproteinases. Additionally, the luminescence intensity of the BB-94-treated cells overexpressing MT1-MMP was significantly higher than that of the respective vehicle controls. These results suggest that BB-94 suppresses the sensor cleavage caused by the overexpression of MT1-MMP. We then investigated whether the protein abundance of Cecto-C and Cecto-L was affected by BB-94 treatment under the same experimental conditions as those used for the luminescence measurement (Fig. 5B; Fig. S4A). Under MT1-MMP-overexpressing conditions, one noncyclized monomer band of Cecto-C disappeared in both cell lines, whereas the monomer band of the sensor was detected in the cells treated with BB-94. Similarly, the monomer band intensity of the mock-transfected cells treated with BB-94 was higher than that obtained with the vehicle control. These results were consistent with the variations in the luminescence intensity. The other bands may have been mainly cleaved within 24 h before BB-94 treatment because there was no apparent difference in the amount of the protein. In addition, the mature enzyme of MT1-MMP with a 3× FLAG epitope tag accumulated in the BB-94-treated MCF-7 cells, whereas the degradation product of MT1-MMP accumulated in the vehicle-treated cells. Moreover, the degradation product of MT1-MMP disappeared from MDA-MB-231 cells treated with BB-94. These results indicate that BB-94 inhibits the breakdown of mature MT1-MMP. BB-94 also increased the upper monomer band of Cecto-L, regardless of whether either cell line was under MT1-MMP-overexpressing conditions (Fig. S4A). We also examined whether BB-94 influences endogenous MT1-MMP in both breast cancer cell lines (Fig. S4B,C). In MCF-7 cells, mature endogenous MT1-MMP was not detected, regardless of whether the cells were treated with BB-94 or not. In contrast, endogenous MT1-MMP in MDA-MB-231 cells showed mature enzymes in the presence or absence of BB-94 treatment, which suggests that endogenous MT1-MMP cleaves CD44 in MDA-MB-231 cells. The breakdown of the mature enzyme of ADAM10 in MDA-MB-231 cells treated with BB-94 might be suppressed because the mature enzyme amount showed a 1.9-fold increase; however, a significant difference was not detected. The protein levels of CD44, ADAM17 and mature MT1-MMP were not changed under BB-94 treatment in either breast cancer cell line.
CD44 ectodomain cleavage is induced by TFP in breast cancer cells
We evaluated whether Cecto-C was cleaved by ADAM10 or ADAM17. Breast cancer cells transfected with Cecto-C were treated with trifluoperazine (TFP; a calmodulin inhibitor) for ADAM10 activation or with 12-O-tetradecanoylphorbol 13-acetate (TPA; a PKC activator) for ADAM17 activation. The luminescence intensity in the cells treated with TFP for 30 min was significantly lower than that in the respective control cells (Fig. 6A). In contrast, no significant difference in luminescence intensity was found between the TPA-treated cells and nontreated cells in either cell line (Fig. 6B). To rule out the possibility that the luminescence intensity was affected by cell death, the live cells were counted under the same experimental condition as the luminescence measurement. It was found that 100 μM TFP-treated cells were damaged by trypsinization. Therefore, 50 μM TFP- or vehicle-treated cells were counted (Fig. 6C). No significant difference between these live-cell counts was observed. In addition, there was no significant difference between the number of live cells treated with 100 ng/ml TPA and vehicle (Fig. 6D). We also examined the protein abundance under the same experimental conditions as the luminescence measurement. TFP treatment reduced the intensity of one noncyclized monomer band of Cecto-C in both cell lines (Fig. 6E), whereas no difference was found in the protein abundance of the TPA-treated cells (Fig. 6F). We could not completely exclude the possibility that the decrease in luminescence intensity of 100 μM TFP-treated cells is related to cell damage. However, this decrease in luminescence may not entirely depend on the cell damage because the western blotting results of 100 μM TFP-treated cells show a decrease in the amount of the upper noncyclized monomer band (Fig. 6E). Endogenous mature ADAM10 and ADAM17 were detected in both cell lines treated with TFP and TPA, respectively. These results of Cecto-C protein bands were consistent with the variations in the luminescence intensity under the same experimental conditions.
We then examined the endogenous protein abundance in untransfected breast cancer cells treated with TFP or TPA (Fig. S5A,B). Endogenous CD44v and CD44s were decreased by TFP treatment in MCF-7 cells, whereas endogenous CD44s was increased in MDA-MB-231 cells. In contrast, a change in the endogenous CD44v levels in MDA-MB-231 cells was not observed regardless of the presence or absence of TFP treatment. However, the membrane-bound C-terminal fragment was detected in MCF-7 and MDA-MB-231 cells under TFP treatment. The fragment size was ∼25 kDa, which was the same as the previously reported size (Okamoto et al., 1999). In contrast, TPA did not affect the cleavage of endogenous CD44. Mature ADAM10 and ADAM17 were detected in the cells treated with TFP and TPA, respectively. However, TFP and TPA did not affect the amounts of these proteins. Although mature MT1-MMP was detected in MDA-MB-231 cells with both TFP and TPA treatments, the protein amount did not change significantly regardless of the drug treatment. These results obtained from western blotting of the membrane-bound C-terminal fragment were consistent with the luminescence changes of Cecto-C under TFP and TPA treatment. Thus, Cecto-C has potential application for the evaluation of CD44 cleavage.
The replacement of asparagine at N-linked glycosylation sites with isoleucine affects the localization of the mutated Cecto-C
To further evaluate the properties of Cecto-C, we investigated whether a lack of N-linked glycosylation affects the localization of Cecto-C. We constructed three Cecto-C molecules containing one, five and six aa mutations (Fig. S6A) because the absence of N-linked glycosylation may cause protein accumulation in the ER. Asparagine at the N-linked glycosylation sites was replaced with isoleucine. CD44s contains six predicted N-linked glycosylation sites, as shown in Table S2. MDA-MB-231 cells were transfected with each of the constructed sensors for 24 h and cultured for 24 h. The fixed and permeabilized cells were observed by indirect immunofluorescence. Cecto-C with five and six aa replacements of asparagine [named Cecto-C-(NI-5) and Cecto-C-(NI-6), respectively] accumulated in the ER, whereas Cecto-C with one aa replacement of asparagine [named Cecto-C-(NI-1)] was localized in the cell membrane (Fig. S6B). These results suggest that the N-linked glycosylation site (255th aa in CD44s) in Cecto-C may not be required for cell membrane localization, whereas the positions (25th, 57th, 100th, 110th and/or 120th aa in CD44s) of N-linked glycosylation in Cecto-C are necessary for cell membrane localization.
Subsequently, to examine whether Cecto-C undergoes N-linked glycosylation, we performed western blot analysis of Cecto-C and Cecto-C-(NI-6). MDA-MB-231 cells were transfected with Cecto-C or Cecto-C-(NI-6), and western blotting was performed. We found that the protein bands of Cecto-C shifted to higher molecular masses than those of Cecto-C-(NI-6) (Fig. S6C). Moreover, these two noncyclized monomer bands of Cecto-C became one band in the analysis of Cecto-C-(NI-6), which suggested that the two bands of monomers obtained with Cecto-C were derived from the difference in N-linked glycosylation. As mentioned above, the molecular mass of Cecto-C without the signal peptide was ∼87 kDa, which was consistent with the band size of the monomer obtained with Cecto-C-(NI-6). These results indicate that Cecto-C was modified by N-linked oligosaccharides and thereby localized to the cell membrane.
Castanospermine increases the luminescence intensity of Cecto-C in MDA-MB-231 cells
Subsequently, to evaluate whether the glycoprotein-processing inhibitors affect CD44 ectodomain cleavage, the cells were transfected with Cecto-C for 24 h and treated with each glycoprotein-processing inhibitor for 24 h. Upon treatment with tunicamycin (an N-linked glycoprotein inhibitor), the luminescence intensity was slightly reduced in MDA-MB-231 cells, whereas no significant difference in the luminescence intensity was detected in MCF-7 cells (Fig. S7A). In contrast, treatment with 40 μM castanospermine (an α-glucosidase inhibitor) increased the luminescence intensity by ∼1.5-fold compared with that obtained with the vehicle control in MDA-MB-231 cells (Fig. S7B). In the case of MCF-7 cells, no difference was found between castanospermine treatment and the vehicle control, probably because Cecto-C in the cells was slightly cleaved by endogenous metalloproteinases under standard culture conditions, as shown in Fig. 5A (Mock). In addition, no change in the luminescence intensities was observed after treatment with either 1-deoxynojirimycin (an inhibitor of α-glucosidase) or swainsonine (an inhibitor of α-mannosidase in the Golgi complex) (Fig. S7C,D). To exclude the possibility that the luminescence intensity of the castanospermine-treated cells was affected by cell number, the live cells were counted under the same experimental conditions as the luminescence measurement (Fig. S7E). No significant difference in the cell number was observed. These results suggest that castanospermine suppresses CD44s cleavage caused by endogenous metalloproteinases in MDA-MB-231 cells.
Castanospermine allows the cell membrane localization of endogenous CD44 but suppresses the cleavage of CD44v induced by TFP treatment
We evaluated whether castanospermine suppresses the ectodomain cleavage of endogenous CD44 in MDA-MB-231 cells because castanospermine increases the luminescence intensity of Cecto-C, as shown in Fig. S7B. First, to investigate whether endogenous CD44 treated with castanospermine localizes to the cell membrane, indirect immunofluorescence was performed using castanospermine-treated MDA-MB-231 cells. In cells treated with castanospermine for 24 h, endogenous CD44 was specifically observed in the cell membrane (Fig. 7A). We then determined whether castanospermine suppresses the cleavage of endogenous CD44. MDA-MB-231 cells were treated with castanospermine for 24 h, and the endogenous CD44 protein abundance was detected (Fig. 7B). Castanospermine treatment did not affect the increase in CD44 protein abundance; hence, we confirmed whether castanospermine suppresses TFP-induced endogenous CD44 cleavage. Castanospermine-treated MDA-MB-231 cells were treated with TFP for 30 min. Notably, the cleavage of endogenous CD44v was suppressed by castanospermine treatment. However, the membrane-bound C-terminal fragment of CD44 was detected (Fig. 7C). The detection of the cleaved product indicates endogenous CD44s cleavage because CD44v cleavage was suppressed by castanospermine treatment. Therefore, both CD44 isoforms and their N-linked glycosylation are likely to affect the ectodomain cleavage of CD44. In contrast, no significant difference in the amounts of CD44s and ADAM10 protein was observed in the cells treated with castanospermine and TFP.
Castanospermine suppresses gelatin degradation and slightly affects cell migration in MDA-MB-231 C1 clones
To evaluate whether castanospermine suppresses extracellular matrix (ECM) degradation, we performed gelatin degradation assays using an MDA-MB-231 C1 clone with higher invasiveness. The MDA-MB-231 C1 clone was cultured on fluorescein-gelatin under castanospermine or vehicle treatment for 24 h, and the fixed cells were detected by immunofluorescence using anti-CD44 antibody along with TRITC-phalloidin (for F-actin staining) and 4′,6-diamidino-2-phenylindole (DAPI) (Fig. 8A). The gelatin degradation area by the cells treated with 20 μM or 40 μM castanospermine was significantly reduced compared with that by the cells treated with the vehicle (Fig. 8B). CD44 was localized on the cell membranes regardless of whether the cells were treated with castanospermine or not, which suggested that the completed N-linked glycosylation of CD44 might be necessary for ECM degradation.
We then examined whether castanospermine affects the cell migration of the C1 clone. Cells on a gelatin-coated 24-well plate were treated with castanospermine or vehicle for 24 h and then scratched. The media of the cells were replaced with new media containing castanospermine or vehicle, and the cells were cultured for 24 h (Fig. 8C,D). Castanospermine treatment at 40 μM slightly suppressed cell migration compared with that observed under all other conditions (Fig. 8C,D). Therefore, castanospermine may contribute to the suppression of ECM degradation by breast cancer cells with high invasiveness rather than their migration.
Castanospermine suppresses the ectodomain cleavage of Cecto-L in the MDA-MB-231 C1 clone with high invasiveness
We evaluated whether Cecto-C and Cecto-L function in the MDA-MB-231 C1 clone. The cells were transiently co-transfected with Cecto-C or Cecto-L along with MT1-MMP or mock, and the luminescence intensity of the cells was measured 24 h after transfection (Fig. 8E). The luminescence intensity of Cecto-C or Cecto-L co-expressing MT1-MMP was reduced compared with that obtained with the respective mocks. Of the two sensors, the cleavage of Cecto-L showed a relatively larger difference than that obtained with the cleavage of Cecto-C. Therefore, we used Cecto-L for the subsequent experiment with the MDA-MB-231 C1 clone. To exclude the possibility that the decrease in luminescence intensity was affected by cell death, the live cells were counted under the same experimental conditions as the luminescence measurement (Fig. S8A). No significant difference in the cell number was observed. To investigate whether the luminescence intensity of Cecto-L is altered by castanospermine treatment, the luminescence intensity of the cells treated with castanospermine or vehicle for 24 h was measured (Fig. 8F). Treatment with 20 μM or 40 μM castanospermine significantly increased the luminescence intensity compared with that obtained with the vehicle control, which suggests that the ectodomain cleavage of Cecto-L is suppressed by castanospermine treatment. To rule out the possibility that the increase in the luminescence intensity was dependent on the cell number, we counted the untransfected MDA-MB-231 C1 clones treated with castanospermine for 24 h (Fig. S8B). No significant difference in the cell number was observed. Therefore, the increase in the luminescence intensity of the cells treated with castanospermine was independent of the cell number.
We subsequently evaluated the endogenous protein amounts under the same experimental conditions as those used for the luminescence measurement with castanospermine treatment. Castanospermine did not affect the protein amounts of endogenous CD44, ADAM10, ADAM17 and MT1-MMP (Fig. S8C,D). In addition, castanospermine might not inhibit endogenous CD44 ectodomain cleavage because the C-terminal fragment of CD44 was detected regardless of the presence or absence of castanospermine treatment. However, in the C1 clone, the ectodomain cleavage of endogenous CD44 by endogenous metalloproteinases is modest under standard culture conditions. Therefore, even if castanospermine suppresses the ectodomain cleavage of endogenous CD44 in the C1 clone, the difference in the protein amount is assumed to be slight. Therefore, castanospermine might slightly affect endogenous CD44 ectodomain cleavage in the MDA-MB-231 C1 clone.
We developed bioluminescent sensors to detect CD44s ectodomain cleavage using the luciferase fragment complementation technique. Western blotting revealed the presence of several protein bands with Cecto-C, and, with these bands, we were able to distinguish between the cyclized and linear forms of Cecto-C. We speculated that the oligomerization of Cecto-C depended on the distance between DnaEn and DnaEc in each adjacent sensor. Although Cecto-L did not possess the cyclized form, luminescence was detected upon its expression in the cells. The results indicate that the two Nluc fragments, NlucN and NlucC, were in close proximity and sufficiently complementary to recover the luminescence reaction. In addition, the cleavage of Cecto-L released the cleaved ectodomain fragment into the medium, which possibly changed the configuration of NlucC. Such phenomena may have led to luminescence changes in Cecto-L. Similarly, the noncyclized monomers and oligomers of Cecto-C are inferred to exhibit changes in luminescence upon cleavage of the sensors.
Cecto-C was cleaved by MT1-MMP overexpression or TFP treatment. In addition, the luminescence intensity of Cecto-C with mock was increased upon BB-94 treatment in MDA-MB-231 and MCF-7 cells; however, a slight increase in luminescence was detected in MCF-7 cells. These results imply that CD44 was cleaved by endogenous metalloproteinases in MDA-MB-231 cells but not in MCF-7 cells under standard culture conditions. In addition, the results obtained with the MCF-7 cells indicate that Cecto-C allowed for the detection of modest changes in ectodomain cleavage. Cecto-L is also available for the detection of CD44 ectodomain cleavage caused by overexpressed MT1-MMP in breast cancer cells containing the MDA-MB-231 C1 clone with high invasiveness. Thus, Cecto-C and Cecto-L allow sensitive detection of ectodomain cleavage in CD44 caused by endogenous metalloproteinases, which indicates that techniques using the sensors might have potential applications for screening drugs outside the cell membrane.
We showed that the luminescence intensity of Cecto-C and Cecto-L was increased by castanospermine treatment in MDA-MB-231 cells and the C1 clone, respectively. The effect of castanospermine was not detected in MCF-7 cells because Cecto-C in the cells was slightly cleaved by endogenous metalloproteinases under standard culture conditions; unlike MDA-MB-231 cells, endogenous mature MT1-MMP was not detected in MCF-7 cells. Hence, BB-94 had only a modest effect on the cleavage of Cecto-C by endogenous metalloproteinases in MCF-7 cells, as shown in Fig. 5A (Mock). Therefore, the ectodomain cleavage of Cecto-C in MCF-7 cells was not assumed to be affected by castanospermine treatment. We also found that castanospermine suppressed endogenous CD44v ectodomain cleavage under TFP treatment in MDA-MB-231 cells. These results indicate that the completed N-linked glycosylation of CD44v may be essential for the recognition or cleavage of CD44v by endogenous metalloproteinases, probably ADAM10. Castanospermine has been reported to prevent angiogenesis and tumor growth (Pili et al., 1995). The inhibition of CD44 reduces vascular tube formation and cell proliferation in endothelial cells (Savani et al., 2001; Trochon et al., 1996). CD44-null mice injected with melanoma and ovarian tumor lines show a decrease in both tumor growth and associated tumor angiogenesis compared with those observed in wild-type mice (Cao et al., 2006). In human gastric cancer cells, a high level of CD44v8-10 facilitates the resistance of reactive oxygen species by stabilizing xCT (a light-chain subunit of a cysteine–glutamate exchange transporter) and thereby promotes tumor growth; however, CD44v8-10 mutations in the consensus motifs for an N-linked glycosylation site, CD44v (S301A), fail to interact with xCT (Ishimoto et al., 2011). Taken together, these facts and our results imply that the suppression of angiogenesis and tumor growth by castanospermine depends on the inhibition of CD44 ectodomain cleavage or, alternatively, on suppression of the interaction between CD44 and other proteins.
One of the sp2-iminosugar-type castanospermine analogs, 1-C-octyl-2-oxa-3-oxocastanospermine (CO-OCS), suppresses breast cancer cell migration through stromal-interaction molecule 1, β1-integrin and focal adhesion kinase signaling pathways (Gueder et al., 2017). Similarly, we showed that castanospermine inhibited cell migration in the MDA-MB-231 C1 clone, although the inhibitory effect was modest. Additionally, castanospermine significantly suppressed gelatin degradation by the MDA-MB-231 C1 clone. This result was consistent with the inhibitory effect of CD44 ectodomain cleavage in the luminescence measurements. Therefore, the inhibition of modest endogenous CD44 cleavage may affect the gelatin degradation process. In addition, CD44s interacts with MT1-MMP at sites of invadopodia, actin-based protrusions of the cell membrane, and CD44s-mediated matrix degradation requires MT1-MMP activity (Zhao et al., 2016). MT1-MMP is the key protease for the degradation activity in invadopodia in breast cancer cells (Artym et al., 2006), and, hence, the completed N-linked glycosylation of CD44s might be required for ECM degradation via MT1-MMP.
In ovarian cancer, cell surface transmembrane CD44v8-10 is associated with an epithelial phenotype and longer survival; however, the detection of the CD44v8-10 ectodomain in ascites samples of patients is correlated with a worse prognosis (Sosulski et al., 2016), and the poor prognosis induced by the cleavage of CD44v8-10 likely depends on increased cell motility. Therefore, castanospermine administration has the potential to improve the prognosis of patients with ovarian cancer expressing CD44v due to the inhibition of CD44v cleavage. In addition, castanospermine inhibits the metastasis of cancer cells in mice by blocking tumor colonization (Humphries et al., 1986). In a clinical study, the recurrences of breast tumors with CD44+/CD24−/low characteristics tend to show a positive correlation with bone metastasis (Abraham et al., 2005). CD44 enhances cluster formation of circulating tumor cells (CTCs) in breast cancer cells; CTC cluster formation is mediated by intercellular CD44–CD44 homophilic interactions, which are highly metastatic (Liu et al., 2019). Therefore, incomplete N-linked glycosylation of CD44 by castanospermine might prevent intercellular CD44–CD44 homophilic interactions and thereby inhibits the metastasis of cancer cells. Thus, the castanospermine treatment of breast cancer cells is anticipated to contribute to the suppression of cancer cell metastasis.
In contrast, tunicamycin-treated MDA-MB-231 cells showed a slight decrease in the luminescence intensity of Cecto-C. Tunicamycin causes the accumulation of unfolded proteins in the ER and induces ER stress in CD44+/CD24− breast cancer cells (Nami et al., 2016). Both Cecto-C-(NI-5) and Cecto-C-(NI-6) accumulated in the ER due to the lack of N-linked glycosylation. These results indicate that Cecto-C accumulated in the ER because of tunicamycin treatment. Therefore, the decrease in the luminescence intensity of Cecto-C with tunicamycin in MDA-MB-231 cells might be due to the degradation of Cecto-C in the ER. Altogether, castanospermine has the potential to contribute to breast cancer therapy, and Cecto-C and Cecto-L might be useful tools for the screening of drugs to inhibit CD44 ectodomain cleavage.
MATERIALS AND METHODS
A DNA fragment containing restriction enzyme sites, V5 and Myc epitope tags, the signal peptide of CD44s (1–20 aa), the DnaEc intein-coding region derived from cyanobacterium Synechocystis sp. strain PCC6803 (Evans et al., 2000; Scott et al., 1999; Wu et al., 1998), and a transmembrane domain of CD44s was commercially synthesized (Eurofins Genomics K.K., Tokyo, Japan). The synthesized DNA fragment was inserted into the modified pcDNA4/V5-His B vector (Invitrogen-Thermo Fisher Scientific, Carlsbad, CA) containing the puromycin resistance gene (Noda et al., 2017). The multiple cloning site and the two epitope tags in the vector were deleted prior to fragment insertion. A DNA fragment encoding human CD44s (NCBI reference sequence: NP_001001391.1) was amplified by PCR using Human XG Lung Carcinoma (LX-1) QUICK-Clone cDNA (7201-1, Clontech-Takara, Mountain View, CA) as the template. Because the amplified DNA fragment contained a variant region, the region was deleted by inverse PCR. In addition, the ectodomain in CD44s (21–268 aa) and the transmembrane and intracellular domains (269–361 aa) were inserted into the vector. The C-terminal fragment of Nluc (NlucC, 160–170 aa) (Dixon et al., 2016), (GGGGS)3 flexible linkers, and an alpha-helical linker containing di-glycine as a hinge [(EAAAR)2EAAGG(EAAAR)2EAA (Merutka et al., 1991)] were generated by oligonucleotide annealing. The N-terminal fragment of Nluc (NlucN, 1–159 aa) was commercially synthesized (Integrated DNA Technologies, Coralville, IA). The coding region of the DnaEn intein was amplified by PCR. All the fragments were inserted into the vector. The coding sequences of DnaE extein-N (KFAEY) and extein-C (CFNK) were inserted into the vector (Cecto-C) by inverse PCR of the vector. The N-linked glycosylation sites in Cecto-C were predicted using the NetNGlyc 1.0 server (https://services.healthtech.dtu.dk/service.php?NetNGlyc-1.0). The first amino acid cysteine of the DnaEn intein in Cecto-C was replaced with alanine using inverse PCR. In the mutated Cecto-C, the 26th amino acid of the DnaEn intein, asparagine, was replaced with isoleucine using inverse PCR to prevent N-linked glycosylation (Cecto-L). HA epitope tags were generated by oligonucleotide annealing. The two HA epitope tags were added adjacent to the DnaEn and DnaEc inteins in Cecto-C using the In-Fusion HD Cloning Kit (Clontech-Takara) (Cecto-HA). The N-linked glycosylation sites (25th, 57th, 100th, 110th and 120th aa in CD44s) of Cecto-C were replaced with isoleucine (Bartolazzi et al., 1996); these vectors were amplified by inverse PCR, and ligation between the amplified DNA and a synthesized DNA fragment (Integrated DNA Technologies) was performed. The 255th site was replaced with isoleucine by inverse PCR [Cecto-C-(NI-1), Cecto-C-(NI-5), and Cecto-C-(NI-6)]. The coding sequence of human MT1-MMP (NCBI reference sequence: NP_004986) was commercially synthesized as two DNA fragments (Integrated DNA Technologies). The two fragments were inserted into the modified pcDNA3.1/Myc-His B vector (Noda and Ozawa, 2018) using an In-Fusion HD Cloning Kit (Clontech-Takara). The vector included a 3× FLAG epitope tag and a hygromycin resistance gene instead of a Myc epitope tag, polyhistidine tag and neomycin resistance gene. The amino acids (L50, RRKR111 and E240) in MT1-MMP were mutated by inverse PCR of the vector [MT1-MMP (Mut)].
Cell culture and transfection
The human breast cancer cell lines MDA-MB-231 and MCF-7 were obtained from the European Collection of Authenticated Cell Cultures (Public Health England, Salisbury, UK) and as a kind gift from the Umezawa laboratory (The University of Tokyo). Examination of these cells was negative for mycoplasma contamination. These cells were maintained in RPMI-1640 medium supplemented with 10% fetal bovine serum (FBS) under 5% CO2 at 37°C. The MDA-MB-231 C1 clone was a kind gift from the Ohta laboratory of Kitasato University in Japan. MDA-MB-231 cells from American Type Culture Collection (Manassas, VA) were cultured at 10–20 cells per well in 48-well plates, and the clones were evaluated for their ability to degrade gelatin. The clone showing high invasiveness was selected (Named C1 clone). The cells were not recently authenticated or tested for contamination. The C1 clone was maintained in Dulbecco's modified Eagle medium (DMEM) (high glucose) supplemented with 10% FBS under 5% CO2 at 37°C. These MDA-MB-231 and MCF-7 cells were transiently transfected with the constructed vectors for 24 h using Lipofectamine LTX (Invitrogen-Thermo Fisher Scientific) and TransIT-LT1 reagent (Mirus-Takara, Madison, WI), respectively, according to the manufacturers’ instructions.
MDA-MB-231 and MCF-7 cells were plated on poly L-lysine-coated coverslips in a six-well plate and cultured in RPMI-1640 medium supplemented with 10% FBS for 24 h under 5% CO2 at 37°C. The cells were transiently transfected with each Cecto for 24 h and cultured for 24 h. The cells were fixed with prewarmed 4% paraformaldehyde for 20 min at 37°C. Wheat germ agglutinin, Alexa Fluor™ 680 Conjugate (Invitrogen-Thermo Fisher Scientific) (20 μg/ml) was used for cell membrane counterstaining. The nonspecific binding sites were blocked with 2% bovine serum albumin (BSA) for 1 h at 25°C. The coverslips were incubated with anti-V5 tag antibody [EPR12989] (ab182008, Abcam, Cambridge, UK) (1:200) and anti-Myc tag monoclonal antibody (R950-25, Invitrogen-Thermo Fisher Scientific) (1:500) for 1 h at 25°C. The immune complexes were fluorescently labeled with Alexa Fluor™ 488 chicken anti-rabbit IgG (1:5000) and Alexa Fluor™ 568 goat anti-mouse IgG1 (1:2000) (Life Technologies-Thermo Fisher Scientific, Eugene, OR) for 45 min at 25°C. Subsequently, to confirm the localization of the sensors in breast cancer cells, the fixed cells on the coverslips were permeabilized with 0.5% Triton X-100 on ice for 5 min. Immunostaining was performed as described above. Concanavalin A, Alexa Fluor™ 594 Conjugate (Invitrogen-Thermo Fisher Scientific) (50 μg/ml) was used for ER counterstaining. The localization of endogenous CD44 protein in MDA-MB-231 cells was detected using anti-CD44 antibody [EPR18668] (ab189524, Abcam) for recognition of the N-terminus (1:250) and Alexa Fluor™ 488 chicken anti-rabbit IgG (1:5000). Images of the samples were acquired using a laser-scanning confocal microscope (IX81-FV1000-D, Olympus, Tokyo, Japan) with a 60× oil-immersion objective (1.35 NA).
MDA-MB-231 and MCF-7 cells were lysed with radioimmunoprecipitation assay buffer containing a complete EDTA-free protease inhibitor cocktail (Roche, Mannheim, Germany) and centrifuged at 20,400 g and 4°C for 10 min. The cell supernatants were separated by SDS-PAGE and transferred to a nitrocellulose membrane. The membrane was blocked with 1–2% skim milk for 1 h at 25°C. Western blot analysis was performed using anti-V5 antibody (R960-25, Novex-Thermo Fisher Scientific, Carlsbad, CA) (1:10,000), anti-HA tag antibody [EPR4095] (ab182009, Abcam) (1:1000), anti-CD44 antibody [ERP18668] (ab189524, Abcam) for recognition of the N-terminus (1:1000), anti-CD44 antibody (ab157107, Abcam) for recognition of the C-terminus (1:4000), anti-ADAM10 antibody [EPR5622] (ab124695, Abcam) (1:1000), anti-ADAM17 antibody (ab28233, Abcam) (1:1000), anti-MMP14 antibody [EP1264Y] (ab51074, Abcam) (1:5000), anti-DDDDK-tag polyclonal antibody (PM020, Medical & Biological Laboratories Co., Ltd, Aichi, Japan) (1:1000) and anti-β-actin antibody (A1978, Sigma-Aldrich, St Louis, MO) (1:10,000). Beta-actin was used as an internal standard. The protein bands were detected using a Lumino image analyzer (LAS 4000 Mini, GE Healthcare, Chalfont St Giles, UK). The intensity of the protein bands was analyzed using ImageJ Fiji software.
To ascertain whether the ectodomain cleavage of Cecto-C and Cecto-L is inhibited by a metalloproteinase inhibitor, breast cancer cells were co-transfected with Cecto-C or Cecto-L along with MT1-MMP or mock for 24 h and treated with Batimastat (BB-94) (Millipore, Burlington, MA) in serum-free medium for 24 h. To examine whether Cecto-C was cleaved by ADAM10 or ADAM17, breast cancer cells were transiently transfected with Cecto-C for 24 h and cultured for the next 24 h. The cells were treated with TFP (Santa Cruz Biotechnology, Dallas, TX) or TPA (Merck-Millipore, Burlington, MA) for 30 min in serum-free medium. Untransfected cells were treated with BB-94, TFP or TPA under the same experimental conditions as the transfected cells. To evaluate whether Cecto-C is cyclized, MDA-MB-231 cells were transiently co-transfected with Cecto-C along with mock or MT1-MMP for 24 h. The cells were treated with RPMI-1640 medium supplemented with 10% FBS containing bafilomycin A1 (Cayman Chemical Co., Ann Arbor, MI) or vehicle for 24 h to inhibit protein degradation.
To evaluate the cytotoxicity of overexpressed MT1-MMP, breast cancer cells were plated in 48-well plates and cultured in RPMI-1640 medium supplemented with 10% FBS for 24 h under 5% CO2 at 37°C. The cells were transiently co-transfected with Cecto-C or Cecto-L along with mock or MT1-MMP for 24 h. The medium of the cells was replaced with fresh RPMI-1640 medium supplemented with 10% FBS and cultured for 24 h. The cells were trypsinized and stained with 0.2% Trypan Blue. The cells were counted using hemocytometers. To exclude the possibility that the variation in the luminescence intensity was affected by cell death, the cells were transiently transfected with Cecto-C for 24 h, and the cells were treated with TFP, TPA or castanospermine under the same experimental conditions as the luminescence measurement. The cells were trypsinized, and the live cells were counted after Trypan Blue staining. In the MDA-MB-231 C1 clones, the untransfected cells were plated in 48-well plates and cultured in DMEM supplemented with 10% FBS for 24 h under 5% CO2 at 37°C. The cells were treated with castanospermine for 24 h, and then the trypsinized cells were counted after Trypan Blue staining.
Inhibition of N-linked glycosylation
Tunicamycin (Abcam), castanospermine (Merck-Millipore and FUJIFILM Wako Pure Chemical Corp., Osaka, Japan), 1-deoxynojirimycin (hydrochloride) (Cayman Chemical Co., Ann Arbor, MI) and swainsonine (FUJIFILM Wako Pure Chemical Corp.) were used for N-linked glycosylation inhibition. Breast cancer cells were transfected with Cecto-C for 24 h and treated with each inhibitor in serum-free medium for 24 h. Untransfected MDA-MB-231 cells were treated with castanospermine for 24 h for indirect immunofluorescence and western blotting. After castanospermine treatment, TFP treatment was performed for 30 min in serum-free medium with castanospermine for western blotting. The untransfected MDA-MB-231 C1 clone cells were treated with DMEM supplemented with 10% FBS and castanospermine at 10–40 μM or vehicle for 24 h. The harvested cells were used for western blotting.
MDA-MB-231 and MCF-7 cells were plated on 96-well solid white microplates and cultured in RPMI-1640 medium supplemented with 10% FBS for 24 h. The cells were transiently co-transfected with Cecto-C or Cecto-L along with MT1-MMP, MT1-MMP (Mut) or mock vector for 24 h. The medium was replaced with Phenol Red-free RPMI-1640 medium supplemented with 10% FBS, 25 mM HEPES and Nano-Glo® Luciferase Assay Substrate (Nano-Glo® Luciferase Assay System, Promega, Madison, WI) (1:1000). The luminescence intensity of the cells was measured using a microplate reader (Mithras LB 940, Berthold Technologies, Bad Wildbad, Germany). After 24 h, the medium was replaced with fresh medium containing the luciferase substrate. The luminescence intensity of the cells was measured repeatedly. The medium of the cells treated with BB-94, TFP or TPA was replaced with Phenol Red-free RPMI-1640 medium supplemented with the luciferase substrate and 25 mM HEPES. Subsequently, the luminescence intensity was measured. In experiment using N-linked glycosylation inhibitors, breast cancer cells were transfected with Cecto-C for 24 h and treated with each inhibitor in serum-free medium for 24 h. The medium was replaced with Phenol Red-free RPMI-1640 medium supplemented with the luciferase substrate and 25 mM HEPES, and the luminescence intensity was measured using the microplate reader.
The cells of the MDA-MB-231 C1 clone were transiently co-transfected with Cecto-C or Cecto-L with MT1-MMP or mock vector for 24 h. The medium was replaced with Phenol Red-free DMEM supplemented with 10% FBS and Nano-Glo® Luciferase Assay Substrate (1:1000), and the luminescence intensity of the cells was measured using the microplate reader. For castanospermine treatment, the C1 clone was transiently transfected with Cecto-L for 24 h, and the medium of the cells was replaced with DMEM supplemented with 10% FBS and castanospermine at 10, 20 or 40 μM or vehicle. After 24 h, the medium of the cells was replaced with Phenol Red-free DMEM supplemented with 10% FBS and Nano-Glo® Luciferase Assay Substrate (1:1000), and the luminescence intensity of the cells was measured using the microplate reader.
Gelatin matrix degradation assays
The glass bottoms (diameter, 12 mm) in 35-mm dishes were coated with fluorescein-gelatin using a QCM™ Gelatin Invadopodia Assay (Green) (Millipore). MDA-MB-231 C1 clones (0.6×104) were plated on the gelatin-coated glass bottoms (diameter, 12 mm) in 35-mm dishes. The dishes were placed on a clean bench for 20 min, and the cells were incubated for 30 min under 5% CO2 at 37°C. DMEM (high glucose) containing 10% FBS and vehicle or castanospermine was added to the dishes to a volume of 2 ml. The final concentrations of castanospermine were 10, 20 and 40 μM. After 24 h, the cells were fixed with prewarmed 4% paraformaldehyde for 30 min at 37°C. The fixed cells were then permeabilized with 0.5% Triton X-100 on ice for 5 min. The nonspecific binding sites were blocked with 2% BSA for 1 h at 25°C. The cells were incubated with an anti-CD44 antibody (ab189524, Abcam) for recognition of the N-terminus (1:200) for 1 h at 25°C. The immune complexes were fluorescently labeled with Alexa Fluor™ 647 goat anti-rabbit IgG (Life Technologies-Thermo Fisher Scientific) (1:2000) along with TRITC-phalloidin and DAPI [QCM™ Gelatin Invadopodia Assay (Green)] for 45 min at 25°C. Images of the cells were acquired under phosphate-buffered saline using a laser-scanning confocal microscope (IX81-FV1000-D, Olympus) with a 60× oil-immersion objective (1.35 NA). The gelatin degradation areas were calculated using ImageJ Fiji software.
Cell migration assays
MDA-MB-231 C1 clones (1.5×105 per well) were plated on a 0.1% gelatin-coated (190-15805, Wako, Osaka, Japan) 24-well plate and cultured in DMEM supplemented with 10% FBS for 24 h under 5% CO2 at 37°C. The clone was treated with DMEM containing 10% FBS and vehicle or 10, 20 or 40 μM castanospermine. After 24 h, the clone was scratched using a CELL Scratcher® scratch stick and scratch guide (AGC Techno Glass, Shizuoka, Japan) and rinsed with 300 μl Phenol Red-free DMEM once. The medium was replaced with Phenol Red-free DMEM supplemented with 10% FBS and vehicle or 10, 20 or 40 μM castanospermine for 24 h. The images of the cells were acquired using a digital camera (OLYMPUS STYLUS XZ-2, Olympus) and an inverted microscope with a 4× phase-contrast objective (0.13 NA) (CKX31, Olympus). The areas of the cell-free region were calculated using ImageJ Fiji software.
The paired group data were analyzed using Welch's t-test (Statcel4, an Excel add-in software, OMS Publishing Co.). Data from multiple groups and two series of data were analyzed by one-way and two-way analysis of variance, respectively, with a post hoc multiple comparison test (Bonferroni–Dunn) using Stat View software (version 5.0, SAS Institute). Details of the sample sizes (n) are shown in the figure legends. The extent of significant differences is also shown in the figure legends.
We are grateful to Dr Rintaro Shimada, Dr Masaki Takeuchi and Mr Tomoki Nishiguchi of The University of Tokyo for vector design and data interpretation. We also thank Prof. Yasutaka Ohta and Dr Koji Saito of Kitasato University for providing the MDA-MB-231 C1 clone.
Conceptualization: N.N., T.O.; Methodology: N.N.; Validation: N.N.; Formal analysis: N.N.; Investigation: N.N.; Writing - original draft: N.N., T.O.; Supervision: T.O.; Project administration: N.N.; Funding acquisition: N.N., T.O.
This work was supported by the Japan Science and Technology Agency [Core Research for Evolutional Science and Technology (JPMJCR1752 to T.O.)] and the Japan Society for the Promotion of Science [KAKENHI Grant-in-Aid for Scientific Research (A; 19H00900 to T.O.) and a Grant-in-Aid for Challenging Research (Exploratory; 18K19573 to N.N.)].
Peer review history
The peer review history is available online at https://journals.biologists.com/jcs/article-lookup/doi/10.1242/jcs.259314.
The authors declare no competing or financial interests.