SUSD4 expression inhibits tumor growth in a syngeneic mouse model
We previously demonstrated that the expression of SUSD4 by tumor cells was correlated with a better prognosis for breast cancer patients [2]. To further evidence a tumor suppressive effect of SUSD4 in vivo, the murine breast cancer cell line 4 T1-Luc2 stably expressing mouse SUSD4-FLAG or mock control cells were transplanted into BALB/c mice. The previously identified complement inhibitory function of SUSD4 could theoretically be beneficial for the tumor [1]. As such, the rationale behind the choice of a syngeneic model was to study the effect of SUSD4 expression in model with non-aberrant immune system. Two doses of cells (2 × 106 or 5 × 106) were injected in the mammary fat pad and tumor growth was then monitored. While no tumors could be detected in the group transplanted with 2 × 106 SUSD4-expressing cells, seven out of 10 mice transplanted with mock control cells developed tumors and a significant difference in tumor volume between the two groups was observed 56 days post transplantation (Fig. 1A). In contrast to the four out of 10 mice transplanted with 5 × 106 mock control cells that developed tumors, only one mouse out of 10 transplanted with 5 × 106 SUSD4-expressing 4 T1-Luc2 cells developed a tumor and a significant difference in tumor volume between the two groups could be observed after 7 weeks (Fig. 1B). The syngeneic model thus provides further support for a tumor suppressive effect of SUSD4. Moreover, upon in vitro characterization, 4 T1-Luc2 cells stably expressing mouse SUSD4-FLAG showed a lower migratory capacity (Fig. 1C) and invasiveness through Matrigel (Fig. 1D) compared to control mock cells. These results are in accordance with previous observations in human breast cancer cell lines [2].
Expression of SUSD4 affects cellular EGFR levels
The triple negative breast cancer cell line BT-20 was stably transfected with a vector encoding human SUSD4 or a mock control vector. Based on the amino acid sequence, the estimated molecular weight of SUSD4 is 54 kDa, yet consistent with previous results [2], SUSD4 could be detected by western blot at approximately 70 kDa. SUSD4 is predicted by the NetNglyc server to have four potential N-glycosylation sites in its extracellular domain. After treatment of protein lysates of SUSD4-expressing BT-20 cells and mock control cells with PNGase F or Protein Deglycosylation Mix II, the former cleaving only N-linked oligosaccharides whereas the latter is also cleaving some O-linked oligosaccharides, the molecular weight of SUSD4 decreased (Fig. 1E), indicating that the protein carries N-linked oligosaccharides.
As SUSD4 is a poorly described protein, a broad screen of proteins differentially expressed in the presence of SUSD4 was employed to identify its underlying tumor-suppressive mechanism in breast cancer. The Human XL Oncology Array enabled for a protein profile comparison between the SUSD4-expressing BT-20 cells and mock control cells. This revealed a marked elevation of FoxO1 (Forkhead Box O1), Amphiregulin, Cathepsin D and Kallikrein 6 (Fig. 1F&S1A) levels in cells expressing SUSD4. Of note, the largest difference between the two groups could be seen in the upregulation of EGFR in SUSD4-expressing cells. Interestingly, Amphiregulin is a ligand for the EGFR [19] while FoxO1 is a downstream inhibitor of EGFR signaling [20]. To gain more insight into the connection between SUSD4 and EGFR we assessed their expression in individual cell types present in breast cancer using data from the Broad Institute Single Cell Portal [21]. Both proteins were highly expressed by breast cancer tumor cells such as luminal progenitor cells, cycling cancer cells, Her2 enriched cancer cells and basal-like breast cancer cells (Fig. 1G). Cancer associated fibroblasts in turn expressed EGFR but not SUSD4.
To further confirm the results obtained from the protein array, the expression of amphiregulin and EGFR in the SUSD4-expressing BT-20 cells and mock control cells was assessed by quantitative real-time PCR (Fig. S1B). No significant differences could be seen, indicating that the observed differences at protein level were not due to alterations in mRNA expression. The difference in EGFR at protein level indicated by the array was thereafter verified by western blot while simultaneously assessing receptor phosphorylation at Tyr1086 and Tyr1045 (Fig. 1H&I). Both BT-20 cells and MDA-MB-468 cells stably expressing human SUSD4 displayed higher levels of EGFR compared to the corresponding mock control cells. Moreover, SUSD4-expressing cells had higher levels of phosphorylated EGFR at both Tyr1086 and Tyr1045, but this was likely a mere consequence of higher EGFR levels rather than increased activation as no difference in the ratio of phosphorylated EGFR to total EGFR could be observed (Fig. 1J&K).
To then assess if the excess EGFR in SUSD4-expressing cells is plasma membrane-associated or present intracellularly, cell surface proteins of MDA-MB-468 cells expressing or lacking SUSD4 were biotinylated and separated from the intracellular protein fraction. The great majority of SUSD4 was revealed to be present intracellularly and, additionally, the SUSD4-expressing MDA-MB-468 cells had higher level of EGFR in both compartments (Fig. S1C). To address a potential effect on EGFR trafficking, receptor ubiquitination was investigated as it is involved in EGFR internalization. EGFR immunoprecipitation and western blot analysis of receptor ubiquitination in serum-starved EGF treated BT-20 cells expressing SUSD4 or mock control cells proved the EGFR to be ubiquitinated faster and to a greater extent in SUSD4-expressing cells (Fig. 1L). Next, a possible effect of SUSD4 expression on EGFR degradation was investigated by treating serum-starved cells with EGF prior to western blot analysis of EGFR levels. However, the BT-20 SUSD4 and mock control cells exhibited a similar decrease in EGFR levels, thus indicating that the expression of SUSD4 does not affect EGFR degradation (Fig. S1D&E).
SUSD4 expression promotes autophagy
The EGFR plays a well-characterized role in the regulation of autophagy [10, 14] and therefore we hypothesized that SUSD4 may also influence this degradative pathway. To study the effect of SUSD4 expression on autophagy we treated cells with the lysosomotropic agent chloroquine, which inhibits autophagosome fusion with lysosomes by raising the lysosomal pH. Consequently, this leads to an accumulation of the well-established marker for autophagy LC3B-II. Following chloroquine treatment for up to 5 hours, LC3B-II levels assessed by western blot were compared between SUSD4-expressing BT-20 (Fig. 2A&C) or MDA-MB-468 cells (Fig. 2B&D) and corresponding mock control cells. Strikingly, both BT-20 cells and MDA-MB-468 cells expressing SUSD4 had significantly higher LC3B-II levels than the mock control cells even absent chloroquine exposure. Onwards, the LC3B-II levels increased with the chloroquine treatment time but remained significantly higher in the SUSD4-expressing cells, indicating a clear effect of SUSD4 on autophagy. The same phenomenon was also observed for the triple negative mouse breast cancer cell line 4 T1-Luc2 stably expressing mouse SUSD4-FLAG, which had been treated with chloroquine for 3 hours (Fig. S1F). Furthermore, LC3B-II levels were also evaluated in the luminal A cell lines T47D and MCF-7 stably expressing SUSD4. The T47D cells expressing SUSD4 had significantly higher LC3B-II levels after 2 hours of chloroquine treatment compared to the mock control cells (Fig. S1G). Expression of SUSD4 in MCF-7 did not seem to have effect on LC3B-II levels (Fig. S1H). Notably, there are large differences in EGFR expression levels between the used triple-negative and luminal A cell lines according to GOBO database [22].
This prompted subsequent assessment of SUSD4’s effect on autophagy using p62 as a marker. As a substrate for autophagy, p62 levels decline in the cell upon induction of autophagy while accumulating when the autophagic flux is inhibited. SUSD4-expressing BT-20 or MDA-MB-468 cells and mock control cells were treated with chloroquine prior to western blot-analysis of cellular p62 levels (Fig. 2E-H). In concurrence with the observations made for LC3B-II, a significant difference indicating higher levels of autophagy in SUSD4-expressing cells was seen. The mock control cells exhibited significantly higher p62 levels initially in both cell lines, a difference that persisted over the course of the chloroquine treatment for the MDA-MB-468 cells.
To gain further confirmation, SUSD4-expressing BT-20 and MDA-MB-468 cells together with the corresponding mock control cells were transfected with a vector encoding GFP-tagged LC3B, allowing for the visualization of LC3B coated vesicles, i.e. autophagosomes, as puncta. Compared to the mock control cells, a significantly higher number of LC3B puncta per cell was observed for the SUSD4-expressing cells with or without a three-hour chloroquine treatment preceding the fluorescent imaging of the cells (Fig. 2I&J).
The significant differences in both the number of LC3B puncta and LC3B-II levels seen between the SUSD4-expressing cells and the mock control cells in the absence of chloroquine, indicate that SUSD4 may play a role in initiating autophagy, but could also be explained by aberrations in the autophagic flux. To address this, the Premo Autophagy Tandem Sensor RFP-GFP-LC3B was used to evaluate the effect of SUSD4 expression on the fusion between autophagosomes and lysosomes. The dual fluorescent labelling of LC3B with the acid-sensitive GFP and acid-insensitive RFP allowed for the membranous compartment where LC3B is localized to be identified. The presence of both GFP and RFP causes autophagosomes to stain yellow while the degradation of the acid-sensitive GFP causes autophagolysosomes to stain red (Fig. 2K). No difference in the colocalization of red- and green-stained vesicles could be seen between the SUSD4-expressing cells and mock control cells (Fig. 2L&M), indicating that the fusion process between autophagosomes and lysosomes is not affected by SUSD4 expression and that SUSD4 likely plays a role in initiating autophagy.
SUSD4 interacts with growth factor receptors
We next assessed a possible interaction between SUSD4 and EGFR since both are transmembrane proteins. For this purpose, proximity ligation assay and confocal microscopy were employed and pointed to an interaction between SUSD4 and EGFR in both BT-20 and MDA-MB-468 cells (Fig. 3A). Albeit indicative of an interaction, the proximity ligation assay merely shows that the target proteins are in close vicinity of one another. Sandwich ELISA, whereby an antibody against EGFR was used for capturing and an antibody against SUSD4 was used for detection, was therefore used to further evidence an interaction between SUSD4 and EGFR in both cell lines (Fig. 3B&C). Additionally, SUSD4 was assessed for its ability to interact with another growth factor receptor, PDGFRα. By using the same approach, SUSD4 was found to interact with PDGFRα using both proximity ligation assay (Fig. 3D) and sandwich ELISA (Fig. 3E). We next used the CRISPR/Cas9 system to delete EGFR from BT-20 cells (Fig. S1I). These cells were used to test if SUSD4 interacts independently with PDGFRα or if a heterotrimeric complex with the EGFR is present. Although the presence of a heterotrimeric complex was not disproven, SUSD4 was found to interact directly with PDGFRα in cells lacking EGFR using sandwich ELISA (Fig. 3F). Moreover, the interaction between SUSD4 and EGFR was evident upon co-immunoprecipitation (Fig. 3G).
Effect of SUSD4 on autophagy depends on its interaction with the EGFR
The aforementioned CRISPR/Cas9-mediated EGFR knockout BT-20 cells stably expressing SUSD4 and mock control cells were used to assess whether the effect of SUSD4 on autophagy depends on its interaction with the EGFR. Cells were treated with chloroquine for up to 5 hours prior to assessing LC3B-II levels by western blot. Consistent with the previous results, the cells expressing both SUSD4 and EGFR had significantly higher levels of LC3B-II than the EGFR-expressing mock control cells at all time-points (Fig. 4A&B). This effect was abolished in the EGFR knockout cells, which displayed similar LC3B-II levels as the EGFR-expressing mock control cells, regardless of SUSD4-expression, thus showing that SUSD4’s autophagy-promoting effect is mediated through its interaction with the EGFR.
Whether the effect of this interaction on autophagy depends on the receptor’s kinase activity was thereafter investigated. The EGFR KO cells expressing or lacking SUSD4, were transiently transfected with a vector encoding canonical EGFR, a kinase-dead mutant EGFR, or an empty control vector. Following a subsequent chloroquine treatment, LC3B-II levels were assessed by western blot while simultaneously verifying the absence or presence of EGFR and the kinase-dead mutant EGFR in the cells (Fig. 4C&D). No differences in LC3B-II levels could be seen absent chloroquine exposure and the presence of EGFR or the kinase-dead mutant EGFR did not affect LC3B-II levels in cells lacking SUSD4 expression exposed to chloroquine. In contrast, higher LC3B-II levels following a four-hour chloroquine treatment could be seen in the SUSD4-expressing cells transfected with either the canonical EGFR or the kinase-dead mutant, relative to the control cells. Hence, the effect on autophagy that stems from the interaction between SUSD4 and EGFR does not depend on the receptor’s kinase activity.
To test whether SUSD4 promotes initiation of autophagy by interacting with the intra- or extracellular domain of EGFR, the CRISPR/Cas9 mediated EGFR KO cells expressing or lacking SUSD4 were transiently transfected with constructs enabling the expression of Myc-tagged intra- or extracellular domain of EGFR. The expression of Myc-tagged domains of EGFR was confirmed by western blot and LC3B-II levels with or without a preceding chloroquine treatment was assessed (Fig. 4E&F). Irrespective of the expression of SUSD4, no differences in LC3B-II levels could be observed between cells expressing the intra- or extracellular EGFR domain and cells transfected with an empty vector, both in the presence and absence of chloroquine. Thus, the complete EGFR, rather than solely the intra- or extracellular portion of the receptor, is required for SUSD4-mediated promotion of autophagy. Moreover, two phosphorylation sites in the intracellular domain of SUSD4 have been suggested to be important for protein function [23]. Site-directed mutagenesis was used to address the importance of these phosphorylation sites for SUSD4’s ability to influence autophagy. Untreated and chloroquine-treated BT-20 cells stably expressing SUSD4, SUSD4-LSPF (Y379F) or SUSD4-PPAF (Y414F) were assessed for LC3B-II levels. Western blot analysis revealed a marked decrease in apparent molecular weight for the SUSD4-LSPF mutant while no difference in migration could be seen for the SUSD4-PPAF mutant, relative to the native protein (Fig. 4G). Onwards, the assessment of LC3B-II levels demonstrated a cardinal role for the LSPY phosphorylation site for SUSD4’s ability to promote autophagy (Fig. 4G&H). While no difference in LC3B-II levels, irrespective of chloroquine treatment, could be seen between the SUSD4-expressing cells and the SUSD4-PPAF mutant, significantly lower LC3B-II levels, resembling the mock control cells, were seen for the SUSD4-LSPF mutant, even for the untreated cells. Deletion mutants were constructed to determine, which domains are important for the regulation of autophagic flux. One CCP domain was deleted in each mutated protein, yielding cells expressing SUSD4 ΔCCP1, ΔCCP2, ΔCCP3 or ΔCCP4 (Table S1). Deletion of CCP1 and CCP2 led to impaired induction of autophagy compared to BT-20 wt SUSD4 expressing cells, irrespective of chloroquine treatment (Fig. 4I&K).
SUSD4 expression leads to activation of signaling complexes promoting autophagy
SUSD4 promotes autophagy through its interaction with the EGFR, but these features are yet to be mechanistically linked. To gain insight into the molecular pathway involved, the Human Phospho-Kinase Antibody Array was utilized as it enabled a comparison in phosphorylation status between the SUSD4-expressing BT-20 cells and mock control cells. The largest differences in phosphorylation status were observed for AMPKα1 (Tyr183), AMPKα2 (Tyr172) as well as JNK1/2/3 (Thr183/Tyr185, Thr221/Tyr223) (Fig. 5A&S1J). Consequently, a role for AMPKα1 was further investigated as it is known to promote autophagy. BT-20 cells and MDA-MB-468 cells stably expressing SUSD4 and corresponding mock control cells were treated with two AMPKα1 activators, AICAR (5-aminoimidazole-4-carboxamide riboside) and A-769662. AMPKα1 activation was subsequently compared by western blot through detection of phospho-AMPKα1 (Tyr172). While no difference in AMPKα1 activation could be seen between the SUSD4-expressing cells and mock control cells for either cell line absent treatment, both AICAR and A-769662 led to significantly higher levels of phospho-activated AMPKα1 in SUSD4-expressing cells compared to mock (Fig. 5B-E).
Upon ligand-induced activation, the EGFR signals through the PI3K-Akt-mTOR axis, leading to Akt-mediated suppression of AMPKα1. As SUSD4-expressing cells demonstrated increased activation of AMPKα1, the modulatory implications of the interaction with SUSD4 on EGFR downstream signaling were evaluated by checking the phosphorylation of Akt at Tyr308 and Ser473, and hence its activation. BT-20 cells expressing SUSD4 exhibited significantly lower levels of phosphorylated Akt at both Tyr308 and Ser473. In contrast, however, no difference in phospho-Akt could be seen between cells expressing SUSD4 and corresponding mock control cells for neither MDA-MB-468 cells nor HS-578 T cells at either phosphorylation site (Fig. S2A). Conversely, MDA-MB-468 cells expressing SUSD4 displayed lower levels of phospho-mTOR (Ser2448) than corresponding mock control cells while no difference in mTOR activation could be seen between BT-20 cells or HS-578 T cells expressing SUSD4 and corresponding mock control cells (Fig. S2B).
While activated AMPKα1 can activate the Unc-51 like kinase-1 (ULK1) through phosphorylation of Ser555 and Ser317 to promote autophagy, mTORC1 can suppress autophagy by phosphorylating ULK1 at Ser757. Comparing the levels of phospho-ULK1 between SUSD4-expressing cells and mock control cells at all three phosphorylation sites revealed significantly lower levels of ULK1 phosphorylated at Ser757 in both BT-20 cells and MDA-MB-468 cells expressing SUSD4, but no difference in ULK1 phosphorylated at Ser555 or Ser317 in either cell line (Fig. 5F&G). Explicitly, SUSD4-expressing cells displayed lower levels of mTORC1-mediated suppression of ULK1 while no difference could be seen in AMPKα1-mediated phospho-activation.
Next, the effect of SUSD4 expression on the activation of complexes involved in autophagosome formation was investigated. Compared to mock control cells, SUSD4-expressing BT-20 and MDA-MB-468 cells exhibited higher levels of activated Beclin-1 as indicated by its phosphorylation at Ser30 (Fig. 5H&I). Similarly, increased phosphorylation and activation of Atg14 could also be observed for BT-20 cells and MDA-MB-468 cells expressing SUSD4 relative to the corresponding mock control cells (Fig. 5J&K). Moreover, LKB1 is an upstream regulator of AMPKα1 that upon activation through phosphorylation at Ser428, can in turn phospho-activate AMPKα1. In BT-20 cells, expression of SUSD4 resulted in increased activation of LKB1 (Fig. 5L), an effect abolished in CRISPR/Cas9-mediated EGFR knockout cells (Fig. 5M). Additionally, increased activation of LKB1 because of SUSD4 expression was further evidenced in MDA-MB-468 cells (Fig. 5N).
Intracellular localization of SUSD4 and EGFR
In addition to the plasma membrane, both SUSD4 and EGFR were found intracellularly (Fig. S1C). To further investigate the intracellular compartments harboring SUSD4 and EGFR, their colocalization with various endosomal markers was investigated in BT-20 and MDA-MB-468 cells by confocal microscopy (Fig. 6A, B & S3,4). The localization of both SUSD4 and EGFR positively correlated with the localization of caveolin and clathrin-coated vesicles (Fig. 6C&D). Furthermore, while EGFR colocalized with the early endosomal markers EEA1 and Rab5a in MDA-MB-468 cells, a significant colocalization could only be seen with Rab5a in BT-20 cells. Additionally, SUSD4 was found to colocalize with EEA1 in both BT-20 cells and MDA-MB-468 cell, but not with Rab5a in either cell line. Both EGFR and SUSD4 were found to colocalize with the marker for late endosomes, Rab7 and Rab11, a marker for recycling endosomes. The elevated EGFR levels in SUSD4-expressing cells could not be explained by a difference in mRNA expression (Fig. S1B) or receptor degradation (Fig. S1D&E). Therefore, it is possible that SUSD4 plays a role in the recycling of EGFR to the plasma membrane.
In addition to the suppressive effect on autophagy following ligand-induced activation of EGFR and downstream signaling via the PI3K-Akt-mTOR axis, inactive EGFR has been shown to promote autophagy [14]. This pathway entails internalization of inactive EGFR because of serum-starvation, sequestration in endosomes where it can interact with LAPTM4B, which ultimately leads to the dissociation of Beclin-1 from Rubicon and autophagy initiation. To investigate if SUSD4 plays a role in this pathway, proximity ligation assay was employed to assess the interaction between Beclin-1 and Rubicon in the BT-20 cells and MDA-MB-468 cells expressing SUSD4 as well as the corresponding mock control cells. Under normal conditions, no difference in the interaction between Beclin-1 and Rubicon could be observed between the SUSD4-expressing cells and mock control cells in either cell line (Fig. S5A). However, upon serum-deprivation, a decreased level of interaction between Beclin-1 and Rubicon could be seen in SUSD4-expressing BT-20 cells compared to the mock control cells (Fig. S5B). In contrast, no difference could be seen between the MDA-MB-468 cells expressing SUSD4 and the mock control cells (Fig. S5B).
SUSD4 expression in breast cancer cells correlates with improved patient prognosis
The expression of SUSD4 was previously correlated with a better prognosis for breast cancer patients [2]. Data now obtained from the Gene expression-based Outcome for Breast cancer Online (GOBO) database [22] and assessed by Kaplan-Meier survival analysis further corroborated a positive outcome related to SUSD4 expression (Fig. 7A). Following the division of patients into three groups based on their expression of SUSD4: high expression (log2 expression 0.738, 5.120), intermediate expression (log2 expression − 0.557, 0.738) and low expression (log2 expression − 4.731, − 0.557), a high expression of SUSD4 could be correlated with improved distant metastasis free survival (Log-rank p = 0.00852). Furthermore, a low expression of SUSD4 was identified as an independent prognostic factor (p = 0.038) using multivariant analysis (Fig. 7B). Looking at the expression of SUSD4 in various subtypes of breast cancer revealed an upregulation in the subtypes luminal A and luminal B, which have a better prognosis (Fig. 7C). Additionally, SUSD4 was found to be downregulated in HER2 enriched breast cancer and unchanged in Normal-like breast cancer.
According to the Kaplan Meier plotter online survival analysis tool [24] for breast cancer patients, a microarray analysis of mRNA expression in patients (n = 3951) examined with Kaplan-Meier survival analysis revealed a longer relapse free survival for patients with high expression of SUSD4 (Log-rank p = < 0.0001) (Fig. 7D). The same online tool used for a Kaplan-Meier survival analysis of mRNA sequencing data of breast cancer patients in the pan-cancer project (n = 936), further corroborated a longer relapse free survival correlated with high expression of SUSD4 (Log-rank p = 0.034) (Fig. 7D). In accordance, patients with a high expression of SUSD4 and upregulated expression of autophagy markers (LC3B or p62) had longer relapse free survival compared with those with low expression of SUSD4 and upregulated expression of autophagy markers (Fig. S5C&D).
Moreover, comparing the expression of SUSD4 in tumorous tissues and matched normal tissues using the Gene Expression Profiling Interactive Analysis (GEPIA) database [25], showed an upregulation of SUSD4 in various cancer types, including breast cancer, while it was downregulated in others, for example ovarian cancer (Fig. 7E). Onwards, using single cell RNA sequencing data obtained from the Broad Institute Single Cell Portal [21], enabled for the expression of SUSD4 in both epithelial cells of various breast cancer subtypes (Fig. 7F) and in different stromal cells (Fig. 7G) to be evaluated. Correlating the level of SUSD4 expression seen in blue to the left in each panel, with the color-annotated breast cancer subsets (Fig. 7F) or color-annotated stromal cell subsets (Fig. 7G), shows that SUSD4 is expressed by breast cancer epithelial cells of the various subtypes of breast cancer, but not by the stromal cells. Hence indicating that SUSD4 is expressed by the cancerous cells rather than the tumor-associated stromal cells.