KDEL-like motif is essential for ER protein-induced cell proliferation and migration
Increasing evidence has shown that secreted ER proteins are important for cell survival, growth, and progression in various cancer cell lines [3, 4, 6, 32, 33]. However, it is currently unknown that whether the common receptor for ER chaperones, KDELR, is involved in mediating these functions. There are three isoforms of KDELR (KDELR1, 2, and 3), all of which appear to expressed on the cell surface in mammalian cells [15]. Since KDELR1 is the most studied and best representative of the family, we focus on KDELR1 and refer to it as KDELR in the rest of this study.
As the last four amino acids of ER proteins are essential for KDELR binding, we purified recombinant GRP78, both wildtype and KDEL-deleted mutant (GRP78ΔKDEL), from E. coli and used them in cell counting (CCK-8) and wound healing experiments to investigate whether KDEL motif influences cell proliferation and migration (Sup. Figure 1A).
For the cell proliferation (CCK-8) assay, HT1080 cells were seeded on a 96-well plate and incubated with DMSO, EGF, GRP78, or GRP78ΔKDEL. The proliferation rate was evaluated by the optical density (OD 450 nm) measurements on 0, 1, and 2 days after seeding. As shown in Fig. 1A, EGF and GRP78-incubated cells showed a significantly higher growth rate than DMSO-treated control cells, whereas GRP78ΔKDEL-treated cells revealed similar proliferation rate as control cells. These results indicated that the recombinant ER protein, GRP78, induce cell proliferation, which is probably dependent on their binding to surface expressed KDELR.
Next, we tested cell migration rate during the wound-healing process. HT1080 cells were seeded onto a 96-well plate and grown until full confluency before cells were scratched and incubated with DMSO, EGF, GRP78, and GRP78ΔKDEL. Images of wound closure were captured at 6, 12, 18, and 24 hours by phase contrast microscopy and the migration distance was measured using ImageJ software. Strikingly, GRP78-treated cells were observed to migrate at a higher rate than cells incubated with DMSO on both day 1 and day2, but GRP78ΔKDEL had no effect (Fig. 1B&C), suggesting that the last four amino acids are essential for GRP78-stimulated cell migration.
Another KDELR ligand protein, mesencephalic astrocyte-derived neurotrophic factor (MANF), has been described to promote cell differentiation, migration, and regeneration as well [17, 34, 35]. We investigated whether the KDEL-like motif on MANF was required for its function using HeLa cells. The results of CCK-8 assay showed that recombinant MANF, but not MANFΔRTDL, induced cell proliferation (Sup. Figure 1B&C). Cell migration rate was evaluated by the transwell experiments. HeLa cells growing on the upper layer were induced with DMSO, EGF, MANF or MANFΔRTDL for 18 hours. Then the cells that migrated through permeable membrane were stained with crystal violet. The results showed that purified His-tagged MANF induced significant cell migration while MANFΔRTDL did not (Sup. Figure 1D&E). Taken together, our results suggested that cell surface KDELR is the receptor for secreted ER proteins-induced cell proliferation and migration.
KDELR on the plasma membrane does not interact with Gα proteins
KDELR appears to serve as a G protein-coupled receptor (GPCR) at the Golgi by binding α subunits of heterotrimeric G proteins and activating their downstream signaling pathways [36–39]. Therefore, we first examined whether cell surface localized KDELR behaves like a GPCR and binds to Gα proteins using bimolecular fluoresecence complementation (BiFC) assay [40, 41]. As illustrated in Sup. Figure 2A, the N-terminal half of a fluorescent protein, Venus, and a flag tag were fused to the C-terminus of Halo-KDELR that contains a signal peptide as a membrane insertion signal (Halo-KDELR-flag-VN). All Gα proteins were individually subcloned to upstream of a myc linker and the C-terminal half of Venus (Gα-myc-VC). It is well known that heterotrimeric G protein interacts with GPCR in the absence of a signal, when Gα is bound to guanosine diphosphate (GDP). Upon agonist activation, the receptor serves as a guanine nucleotide exchange factor (GEF) for the Gα subunit, leading to dissociation of active Gα-GTP and GPCR. Thus, we also introduced a single amino acid mutation in Gα proteins (Gαs S54C, Gαq S53C, Gαo S47C) to mimic their GDP-bound state (Gα-GDP-myc-VC) for improved binding with KDELR. HeLa cells were co-transfected with Halo-KDELR-flag-VN and Gα-myc-VC, or Gα-GDP-myc-VC for 18 hours, followed by confocal microscope analysis. Venus signal was not detected on cell membrane in our experiments, suggesting that KDELR may not function as a GPCR on plasma membrane (Sup. Figure 2B&C).
EGFR and TfR are identified as potential interacting proteins of surface KDELR
Our previous study showed that cell surface KDELR is able to be internalized through clathrin-mediated endocytic pathway [15]. However, it lacks the canonical dileucine-based sorting signal for clathrin adaptor protein 2 (AP-2), which is responsible for cargo recognition [19, 42]. Therefore, it is likely that KDELR has a co-receptor on the plasma membrane for its endocytosis and signaling. Although EGFR-STAT3 signaling is the most frequently reported downstream pathway of surface-bound ER proteins, several other pathways may be involved as well. For example, MANF has been described to activate platelet-derived growth factor (PDGF)-like signaling in the retina of flies and mice [35].
To identify the potential co-receptor of surface KDELR, we revisited our previous data of KDELR interactome identified in the mass spectrometry of BioID experiment, based on protein proximity to C-terminally biotin ligase-fused KDELR (KDELR-BirA*) [20]. In this database, we found a plasma membrane receptor, EGFR, and two clathrin-related proteins, clathrin interactor 1 (CLINT1) and phosphatidylinositol binding clathrin assembly protein (PICALM), suggesting that EGFR may bind intracellular KDELR (Fig. 2A).
Since most of KDELR-BirA* resides at the Golgi apparatus, our BioID assay is expected to identify KDELR-interacting proteins close to the Golgi. To selectively study the interactome of the surface-expressed KDELR, we transfected HeLa cells with previously used N-terminally Halo-tagged KDELR (Halo-KDELR, [15]), selectively labeled cell surface KDELR with non-membrane permeable biotin-conjugated Halo ligands in living cells, and purified KDELR-associated proteins with streptavidin agarose, prior to mass spectrometry analysis. This experiment revealed proteins involved in membrane trafficking, such as sorting nextin 5 and 6 (SNX5 and SNX6), as well as two surface receptors, EGFR and transferrin receptor (TfR) (Fig. 2B).
Next, we tested whether the receptors identified in mass spectrometry experiments appeared in CCVs as the recycling of KDELR between cell surface and the Golgi was shown to be mediated by clathrin[15]. To include a negative control in our experiments, we took advantage of a previous study showing that a KDELR mutant, D91A/T92A, causes complete ER retention [43]. Since this mutant is not able to reach the Golgi, it will not be found in purified CCVs. We transfected HeLa cells with Halo-KDELR or Halo-KDELR D91A/T92A mutant and purified CCVs from HeLa cells according to a published protocol [44]. As expected, western blots analysis using specific antibodies indicated that clathrin and the µ1 subunit of clathrin adaptor protein 1 (AP1M1), but not the β subunit of COPI (β-COP), were enriched in CCVs obtained from both wildtype and mutant KDELR transfected cells. EGFR and TfR were also accumulated in CCVs from both kinds of cells, consistent with the fact that they are constantly internalized and recycled even in the absence of their ligands [45, 46]. On the other hand, only wildtype KDELR was found in CCV fractions, suggesting that internalized KDELR may be transported in CCVs (Sup. Figure 3A).
To further confirm that EGFR and TfR are interacting receptors for KDELR, HeLa cells were transfected with mCherry or KDELR-mCherry for 18 hours, followed by immunoprecipitation (IP) using anti-red fluorescent protein (anti-RFP) beads. KDELR-bound proteins were analyzed by western blots. EGFR, TfR, and ArfGAP1, which is a known KDELR binding protein, were pulled down by KDELR, while a cell surface protein, E-cadherin, was not found in the co-immunoprecipitated fraction (Fig. 2C).
EGFR is a novel KDELR’s interacting protein
Since EGFR has been reported to bind TfR, it is difficult to determine which receptor, EGFR or TfR, is the co-receptor on cell surface for KDELR using conventional methods [47]. Therefore, we performed proximity-based BiFC assays of KDELR with EGFR, TfR and the hepatocyte growth factor receptor (HGFR, a negative control). HeLa cells were co-transfected with Halo-KDELR-Flag-VN and EGFR-, HGFR- or TfR-myc-VC overnight, prior to incubation with DMSO or TAEKDEL peptide for 30 minutes. Confocal images showed that Venus signal was detected on cell surface and at the Golgi only in cells co-expressing Halo-KDELR-Flag-VN with EGFR-myc-VC, but not TfR-myc-VC, or HGFR-myc-VC. No background signals were observed when individual construct was transfected in cells (Sup. Figure 3B&C). Moreover, the Venus signal generated by EGFR-KDELR association was greatly enhanced after TAEKDEL peptide addition, indicating the interaction between EGFR and KDELR was improved by KDEL-KDELR binding (Fig. 2D&E).
To confirm our finding in BiFC experiments, we used split-ubiquitin-based membrane yeast two-hybrid (MYTH) system. KDELR, the “bait” protein, was fused with the C-terminal fragment of ubiquitin (Cub) and a transcription factor. Acyl-CoA binding domain-containing protein 3 (ACBD3), EGFR, glucose transporter 4 (GLUT4), and integrin subunit α5 (ITGA5) were fused to a mutant of N-terminal fragment of ubiquitin (NubG), which carries an isoleucine to glycine point mutation to avoid the automatic association of Cub and Nub (Fig. 3A) [48]. ACBD3 was used as a positive control as it has been shown to bind KDELR directly [20]. Cell membrane receptors, GLUT4 and ITGA5, were included as negative controls.
The results from our MYTH assay indicated that co-expression of KDELR with ACBD3 or EGFR allowed yeast colonies to grow on synthetic dropout (SD) growth media that are supplemented with X-Gal, but do not contain Tryptophan, Leucine, Adenine, and Histine (Fig. 3B, QDO + X-Gal panel). Nubl, which binds Cub spontaneously, was used as a positive control in MYTH system. On the other hand, yeasts expressing KDELR with GLUT4 or ITGA5 grew on SD-Tryptophan-Leucine media (DDO) only (Fig. 3B). Co-expression of KDELR and pPR3-N prey vector, or prey proteins and pBT3-SUC bait vector was also contained as negative controls. Taken together, our data suggested that KDELR interacts with EGFR specifically.
The C tail of KDELR interacts with EGFR
As a small protein equipped with seven transmembrane domains, KDELR has approximately 79% of its residues buried in the membrane. We first tested whether the interaction between EGFR and KDELR is mediated by their transmembrane regions. To this end, we swapped the transmembrane spans of EGFR and HGFR to make a chimeric protein (EGFR-(HGFR-TM)-mCherry) that is composed of the ectodomain and cytoplasmic region of EGFR, but the transmembrane motif of HGFR (Fig. 3C). HeLa cells co-expressing KDELR-Flag and EGFR-(HGFR-TM)-mCherry chimera or EGFR-mCherry, were analyzed in co-immunoprecipitation (co-IP) experiment using anti-Flag beads. The chimera protein was co-immunoprecipitated by KDELR as well as wildtype EGFR, suggesting that EGFR does not use its transmembrane motif to bind KDELR (Fig. 3D).
Since the C tail of KDELR is known to be important for its interaction with other proteins, we next asked whether it is required for EGFR binding. We transfected HeLa cells with mCherry, KDELR-mCherry, or KDELR-ΔCT-mCherry (C tail deleted) and immunoprecipitated mCherry tagged proteins with anti-RFP agarose. A significant amount of EGFR was found in the precipitate only in cells expressing wildtype KDELR. When C tail of KDELR was deleted, EGFR was no longer detected to be pulled down by KDELR-ΔCT-mCherry. ArfGAP1, which has been described to bind the C-terminus of KDELR, was also included as a positive control (Fig. 3E).
Then, we performed GST pulldown assays to further confirm the result of our co-IP experiment. GST and GST tagged KDELR C tail were purified from E. coli, immobilized onto glutathione beads, and incubated with cell lysates prepared from a epidermoid carcinoma cell line, A431, which has a high endogenous level of EGFR. EGFR as well as ArfGAP1 were pulled down by the C tail of KDELR efficiently, confirming that EGFR binds the C-terminus of KDELR (Fig. 3F) [49].
EGFR mediates the endocytosis of KDELR
Although we showed that EGFR is an interacting protein for KDELR so far, we have not demonstrated whether EGFR is the co-receptor on the cell surface that is responsible for mediating KDELR endocytosis. To selectively label surface-localized KDELR, we co-transfected HeLa cells with Halo-KDELR and mCherry-clathrin for 18 hours, stained the cell membrane KDELR with non-membrane permeable fluorescent Halo ligands in living cells at 4oC, and then incubated cells with TAEKDEL peptide or DMSO for 0, 15, and 30 minutes at 37oC, prior to fixation and staining with anti-EGFR antibody. Confocal results showed that few surface-expressed Halo-KDELR co-localized with EGFR and clathrin at steady state. Upon addition of TAEKDEL ligand, co-localization of KDELR with EGFR and clathrin was improved dramatically over time, confirming that internalization of KDELR-EGFR complex undergoes clathrin-mediated endocytic pathway as reported before (Fig. 4A-B) [15]. A two-way ANOVA analysis indicated that co-localization of EGFR and KDELR was statistically significant only after TAEKDEL treatment, compared to DMSO controls (Fig. 4C).
In order to further confirm that EGFR mediates endocytosis of KDELR via CCV, we investigated internalization of KDELR in EGFR-depleted HeLa cells. As in wildtype HeLa cells, overexpressed Halo-KDELR was observed to localize to the plasma membrane in EGFR knockout cells. However, almost all of surface-labeled KDELR was retained on the plasma membrane after incubation with TAEKDEL peptide for 30 minutes, whereas more than 60% of KDELR were co-localized with EGFR and clathrin in endosomes and at the Golgi area in cells expressing EGFR-mCherry (Fig. 4D&E). Interestingly, a small fraction of KDELR was observed in endosomes close to the plasma membrane in EGFR-depleted cells, although the majority of KDELR remained on the cell surface (Fig. 4D).
As the C-tail of KDELR is seemingly responsible for EGFR binding, we also tested whether C tail-deleted KDELR is endocytosed upon KDEL ligand binding using cell surface biotinylation assay, as previously described [15]. HeLa cells stably knockdown of KDELR were transfected with Halo-KDELR or Halo-KDELR-ΔCT and incubated with KDEL ligand at 4oC for 30 minute, prior to incubation at 37oC for 0 or 30 minutes. All plasma membrane proteins were labeled by non-membrane permeable Sulfo-NHS-LC-Biotin, followed by streptavidin pulldown and western blot analysis. Incubation with TAEKDEL peptide induced significant reduction of surface-expressed EGFR and KDELR (Fig. 4F, lane 7&8). In contrast, in cells expressing Halo-KDELR-ΔCT, KDEL ligand did not change the surface fraction of either EGFR or KDELR-ΔCT, suggesting that KDEL ligand-induced internalization of surface-expressed KDELR requires its C-terminal tail (Fig. 4F, lane 9 &10).
KDEL ligand binding improves the interaction between EGFR and KDELR
As we observed that co-localization of EGFR and KDELR was considerably enhanced by KDEL peptide incubation in confocal images, we next checked whether the binding of EGFR and KDELR increases after KDEL ligand addition. HT1080 cells stably overexpressing KDELR-mCherry were incubated with DMSO or TAEKDEL peptide for 30 minutes, prior to co-IP with anti-RFP agarose beads. As expected, addition of TAEKDEL peptide significantly increased EGFR precipitated by KDELR compared to DMSO treated samples (Fig. 4G), confirming that KDEL ligands may trigger the interaction between EGFR and KDELR.
We also observed that TAEKDEL ligand incubation induced a perinuclear localization pattern for clathrin, EGFR, and KDELR, whereas DMSO incubation did not (Fig. 4A&B). To further confirm whether this perinulcear localization is in the Golgi, we stained cells with a specific antibody against a trans-Golgi network (TGN) marker, TGN46, after KDEL ligand or DMSO treatment. Indeed, a subset of EGFR was localized to the TGN after TAEKDEL incubation, but not by DMSO addition (Sup. Figure 4A&B).
To determine whether endocytosed EGFR induced by KDEL ligand undergoes lysosomal degradation, we incubated HeLa Halo-KDELR cells with 1.5 nM EGF (low concentration), 200 nM EGF (high concentration), TAEKDEL, or TAEAAAA peptides, respectively. As reported previously, a high dose of EGF treatment mostly led to lysosomal degradation of ~ 65% endocytosed EGFR, while a low dose of EGF resulted in much less degradation. Strikingly, KDEL ligand addition did not result in EGFR degradation at all (Sup. Figure 4C).
Taken together, our results suggested that EGFR seems to function as a co-receptor for KDELR on the plasma membrane, mediates its endocytosis via clathrin-coated vesicles upon KDEL-KDELR binding, leading to their Golgi-localization and cell surface-recycling of KDELR/EGFR, instead of degradation.
Dimerization of EGFR is moderately increased upon KDEL-KDELR binding
The canonical process of EGFR internalization starts with the binding of EGF with the extracellular domain of EGFR, followed by homo- or hetero-dimerization of EGFR, followed by phosphorylation of its C-terminus. Activated EGFR exits the cell surface through clathrin-mediated endosomal pathway for subsequent recycling back to the cell surface or degradation by the lysosomal compartment [26].
First, we tested whether KDEL ligand-activated EGFR undergoes dimerization as well. To this end, the interaction between two differently tagged EGFR monomers was examined by co-IP experiments. Cells were transiently transfected with EGFR-mCherry and EGFR-myc, followed by EGF or KDEL ligand incubation and IP of EGFR-mCherry using anti-RFP beads. As expected, EGFR-mCherry and EGFR-myc interacted with each other in the absence of ligand and their interaction was greatly enhanced by EGF incubation (Fig. 5A, lane 6&7). TAEKDEL peptide addition moderately increased co-IP of EGFR-mCherry and EGFR-myc, but to a much less extent than EGF addition (Fig. 5A, lane 6–8), suggesting that dimerization of EGFR is moderately improved by KDEL binding to KDELR.
The dimerization of EGFR was also evaluated by split Venus assay. HeLa cells overexpressing EGFR-myc-VC and EGFR-Flag-VN were incubated with TAEKDEL peptide or EGF for 0 and 30 minutes, prior to confocal analysis for Venus signals. As shown in Fig. 5B&D, EGFR dimerization was observed even before TAEKDEL or EGF addition. KDEL ligand was not able to enhance the fluorescent signals in the Venus channel. In some endosome-like compartments, co-localization of EGFR-myc-VC and EGFR-Flag-VN in confocal images did not generate green signals, suggesting that EGFR monomers are in close proximity but are not close enough to bring the two halves of Venus together (white arrowheads in the bottom panels of Fig. 5B). In contrast, Venus signals were largely enhanced upon EGF stimulation especially in endosomes (white arrows in Fig. 5C), suggesting EGFR dimers appear in endosomal pathways after EGF binding, as previously described.
KDELR oligomerization is greatly increased by KDEL ligand binding
Intracellular KDELR is known to self-oligomerize through its transmembrane domains [50]. On the cell surface, KDELR has been described to form clusters especially after cargo binding, suggesting that surface-expressed KDELR oligomerizes as well [12, 13]. To study oligomerization state of KDELR using split Venus assay, HeLa cells were co-transfected with Halo-KDELR-V5-VN and Halo-KDELR-myc-VC, followed by confocal microscope analysis for Venus signals. None of these two constructs had background venus signals when transfected individually in cells (Sup. Figure 3C&D). At steady state, weak Venus fluorescence was observed, suggesting that KDELR monomers oligomerize even when there is no ligand. The Venus signal increased almost 2 fold after 30 minutes of incubation with TAEKDEL peptide, whereas it remained unchanged with DMSO treatment (Fig. 5E-G). These results showed that KDEL binding to KDELR may facilitate the formation of KDELR dimers.
The kinase activity of EGFR is required for the endocytosis of KDELR
Phosphorylation of tyrosine residues in trans in the carboxyl-tail of EGFR dimer is critical for the receptor’s internalization and downstream signaling. To investigate the role of EGFR phosphorylation in the endocytosis of KDELR, we transfected HeLa cells with Halo-KDELR overnight and added TAEKDEL peptide in the medium for 30 minutes in the presence of DMSO, EGFR kinase inhibitor (PD153035), and HGFR kinase inhibitor (Crizotinib). Strikingly, addition of PD153035 restrained surface-expressed EGFR and KDELR from endocytosis even after KDEL peptide treatment, whereas both receptors in DMSO and Crizontinib treated cells showed significantly increased localization in endosomes and at the Golgi upon TAEKDEL incubation (Fig. 6A&B). These results suggested that EGFR kinase inhibitor completely blocked the endocytosis of EGFR and KDELR specifically, as HGFR kinase inhibitor did not affect the internalization of EGFR and KDELR effectively.
KDEL ligand binding to KDELR activates STAT3 signaling
EGFR and its interactors contribute to a complex network of signaling cascades, initiated by a variety of ligands. We tested activation of major EGFR signaling pathways with specific antibodies against phosphorylated proteins in western blotting experiment. HT1080 cells stably overexpressing KDELR-mCherry were incubated with DMSO, EGF or TAEKDEL peptides, prior to cell disruption and western blot analysis. As expected, EGF incubation greatly increased phosphorylation of EGFR (Tyr 1068, Tyr 1086, Tyr 845) and extracellular signal-regulated kinase (ERK), but not JAK, Src or STATs. In contrast, KDEL peptide incubation increased phosphorylation of STAT3 (Tyr705, but not Ser727) by 44% compared to the control cells, whereas JAK, Src, and ERK kinases were not activated, indicating that STAT3 activity may be specifically increased after TAEKDEL binding to KDELR on the cell surface (Fig. 6C). Note that we also observed a 55% increase of EGFR phosphorylation on Tyr1068 in cells incubated with KDEL ligand compared to control cells, which is similar to the phosphorylation level detected upon low concentration of EGF stimulation and may be important for EGFR signaling [51].
As a transcription factor, activated STAT3 has been described to initiate cell proliferation and migration by translocating to the nucleus and turning on transcription of target genes [52, 53]. To assess the function of activated STAT3 by KDEL ligand, we selected four STAT3-regulated genes known to be involved in cell proliferation/migration and investigated their transcription by reverse transcription- polymerase chain reaction (RT-PCR) assay. Total mRNA samples isolated from HT1080 cells were reverse transcribed and quantified by real-time PCR. As shown in Fig. 6D, mRNA levels for Cyclin-D1 and Bcl-2 gene, which induce cell proliferation and inhibit apoptosis, were upregulated in GRP78-incubated cells. Two genes that promote cell migration were differently affected by GRP78 treatment. Transcription of Vimentin gene was greatly increased by GRP78 incubation, while ICAM-1 gene transcription was not significantly changed.
Inhibitor of STAT3 suppresses cell proliferation and migration activated by KDEL ligand
To further dissect the function of STAT3 signaling activated by KDEL peptide, we first investigated whether STAT3 activation was necessary for the endocytosis of KDELR. Cells expressing Halo-KDELR were incubated with TAEKDEL peptide in the presence of DMSO or cryptotanshinone, a STAT3 inhibitor, for 30 minutes, followed by staining with anti-GM130, a Golgi marker, and anti-clathrin antibodies. Addition of KDEL ligand increased co-localization of KDELR with GM130 and clathrin, which was not changed by cryptotanshinone incubation, indicating that STAT3 is unlikely to be involved in the endocytosis of KDELR (Sup. Figure 5A&B).
Next, we asked whether STAT3 signaling is responsible for ER proteins-induced cell proliferation using CCK-8 assay. The growth rate of HT1080 cells seeded on a 96-well plate was evaluated by OD450 measurements on 0, 1, and 2 days, in growth medium supplemented with DMSO, ERp57, ERp57ΔQEDL with or without cryptotanshinone, a chemical that strongly inhibits phosphorylation of Tyr705 on STAT3. Similar to what we observed with GRP78 and MANF, ERp57 induced cells to grow at a rate significantly higher than DMSO- and ERp57ΔQEDL-treated cells on both day 1 and day 2. However, ERp57/cryptotanshinone incubated cells revealed similar proliferation rate as control cells (Fig. 7A). These results suggested that ERp57-induced cell proliferation is controlled by STAT3 signaling.
Impact of cryptotanshinone was also evaluated using the wound healing assay. Confluent HT1080 cells were scratched and grown in medium supplemented with DMSO, EGF, purified ERp57 or ERp57ΔQEDL in the absence or presence of cryptotanshinone. Cells incubated with ERp57 only were observed to heal faster than control cells at all time points tested after a scratch. In contrast, cells treated with ERp57 and cryptotanshinone showed about the same healing speed as DMSO-treated cells, suggesting that STAT3 inhibitor completely abrogated the effect of ERp57 on cell migration (Fig. 7B&C).