Abstract

   Radiotherapy, as a vital means of esophageal cancer treatment, has
   benefited countless cancer patients, but owing to the occurrence of
   radio-resistance, its therapeutic efficiency has been dramatically
   mitigated. Discovering key biomarkers governing radio-tolerance in
   esophageal cancer and revealing their inherent molecular mechanisms
   will be of great significance for clinical cancer treatment. Here, we
   have found roundabout guidance receptor 1 (ROBO1) was significantly
   upregulated in esophageal cancerous tissues and showed enhanced
   expression with the development of cancer staging. Cellular experiments
   demonstrated ROBO1 directly interacted with eukaryotic translation
   initiation factor 3A (eIF3A) and accelerated its degradation in
   esophageal cancer cells after irradiation treatment. Mass spectrum
   analysis further revealed that in response to irradiation, ROBO1, eIF3A
   and G3BP2 (Ras GTPase-activating protein-binding protein 2) formed a
   hetero-complex and triggered lysosomes-mediated protein degradation.
   Knocking down of G3BP2 abrogated the influence of ROBO1 on eIF3A
   instability. Besides, ROBO1-mediated eIF3A degradation interrupted P53
   translation process which in turn provoked downstream mTOR signaling
   and increased DNA repair associated genes expressions, resulting in
   radio-resistance enhancement in cancer cells. In conclusion, our
   findings revealed a novel role of eIF3A in modulating P53/mTOR
   signaling activity and provided a drug candidate (ROBO1) for overcoming
   radio-resistance in esophageal cancer.

   Subject terms: Cancer of unknown primary, Radiotherapy

Introduction

   As one of the most aggressive malignant tumors in the world [[30]1],
   esophageal cancer is always diagnosed at the advanced stage,
   dramatically reducing the patients’ chances of receiving effective
   treatment. Among all kinds of esophagus cancer types, squamous cell
   carcinoma is the most common subtype in China [[31]2]. The guidelines
   of cancer diagnosis and treatment have clearly stated surgery and
   radiotherapy are the most important means for esophageal cancer
   treatment [[32]3]. However, some patients showed increased resistance
   to radiotherapy, which greatly dampened its treatment efficacy [[33]4],
   and will inevitably lead to the recurrence and poor prognosis of the
   cancer [[34]5]. What causes the occurrence of radiation resistance? DNA
   double-strand breaks triggered by ionizing radiation are the leading
   cause of cell death. To avoid radiation-induced cell death, cancer
   cells can recruit DNA repair and cell cycle checkpoints machinery to
   enhance radio-resistance [[35]6]. Recently, the crucial homologous
   recombination protein, MRE11 is demonstrated to be lactylated at K673
   by the CBP acetyltransferase in response to radiation-induced DNA
   damage, and its lactylation promotes its binding to DNA, facilitates
   DNA end resection [[36]7]. What’s more, AEG upregulation confers
   radioresistance to esophageal squamous cell carcinoma by slowing
   PARP1-mediated homologous recombination and DNA repair associated
   proteins degradation [[37]8]. Up to date, some progress have been made,
   but elucidating the molecular mechanisms underlying ESCC
   radiosensitization and identifying potential targets to overcome
   radioresistance are still challenges for cancer researchers.

   Ring-binding receptor 1 (ROBO1) was first identified as a conserved
   transmembrane receptor protein that played a key role in
   neurodevelopment and signal transduction [[38]9]. Later studies have
   confirmed that ROBO1 is wildly expressed in a variety of non-neural
   tissues, such as esophagus, breast, lung, digestive systems, etc
   [[39]10]. SLIT2 was reported as the major ligand for ROBO1 and their
   interaction stimulated downstream signal cascades in cancer cells to
   exert contrasting function in different cancer types [[40]11]. In
   breast cancer cells, PRRG4 via NEDD4-mediated ubiquitination to
   downregulate ROBO1, resulting in the activation of Src and FAK and
   enhanced breast cancer metastasis [[41]12]. But in small cell lung
   cancer, SLIT2 and ROBO1 showed opposing expression pattern in cancerous
   and normal tissues, molecular analysis showed that ROBO1 knockout or
   SLIT2 overexpression suppressed the transforming growth factor beta
   1/β-catenin signaling pathway in tumor cells [[42]13]. Even in the same
   cancer type, the outcomes induced by the coadoption of SLIT2/ROBO1 are
   different. YY1 inhibited pancreatic cancer growth through upregulating
   ROBO1 transcription [[43]14], but the existence of ROBO1 also
   contributed to cancer cell liver metastatic dissemination [[44]15].
   Endothelial-derived SLIT2 protein and its receptor ROBO1 promoted the
   migration of lung cancer cells towards endothelial cells and invasion,
   but deletion of tumoral SLIT2 enhanced metastatic progression [[45]16].
   In summary, all of this data indicated the receptor independent
   characteristics of ROBO1.

   However, in this study, we have found that in response to irradiation
   stress, ROBO1 directly interact with eIF3A proteins and accelerated its
   degradation through orchestrating G3BP2-mediated lysosomes degradation
   pathway in ESCC cells. Reduced eIF3A was unable to maintain P53
   translation in ESCC, relieving mTOR signaling activity and amplifying
   DNA repair associated genes expressions. All of these results revealed
   the onco-promoting role of ROBO1 in strengthening ESCC radioresistance
   and paved a new way to improve the anti-tumor efficacy of radiotherapy
   in ESCC through targeting eIF3A-mediated P53 signaling.

Materials and methods

Cell culture

   Human esophageal carcinoma cell lines KYSE450 and TE1 (HyCyte, Suzhou,
   China) were incubated in RPMI Medium 1640 basic (Gibco, USA) containing
   10% fetal bovine serum (OriCell, China) and 1% Penicillin-Streptomycin
   Solution (Biosharp, China). HEK293T cells kindly provided by Shanghai
   Cell Bank, Chinese Academy of Sciences, were cultured in Dulbecco’s
   modified Eagle’s medium (Gibco, USA) supplemented with 10% fetal bovine
   serum. All cells were incubated at 37 °C in a humidified atmosphere
   with 5% CO2.

Cell transfection

   The plasmids of pMD2G, psPAX, shRNA (Vigene Biosciences, Shandong,
   China) were co-transfected into HEK293T with Lipofectamine 3000
   transfection kit (Invitrogen, USA) according to the manufacturer’s
   instructions. At 48 h after transfection, the cell culture supernatant
   was collected, concentrated using a 100 kDa filter (Millipore, USA) and
   filtered through a 0.22 µm membrane (BIOFIL, Guangzhou, China). Next,
   viral solution was added into the esophageal cancer cells culture
   medium and continued to cultivate cells for 24 h. Subsequently, the
   cells successfully expressing shRNAs were screened with Puromycin
   treatment (Solarbio, Beijing, China) at 5 μg/ml for one week. The
   protocol for siRNA transfection (GenePharma, Shanghai, China) is same
   to the plasmids using Lipofectamine 3000 transfection kit (Invitrogen
   Carlsbad, CA, USA). All of the sequences used for shRNAs and siRNAs are
   provided in the Supplements.

Real-time quantitative PCR assays

   The total RNA was extracted with an RNA-easy isolation reagent (Vazyme,
   Nanjing, China) as described in the protocol, and the complementary
   first-strand DNAs were synthesized with HiScript III RT SuperMix kit
   (Vazyme, Nanjing, China) following the manufacturer’s instructions.
   Polymerase Chain Reaction (PCR) for each gene was completed using ChamQ
   Universal SYBR qPCR Master Mix (Vazyme, Nanjing, China) containing
   specific primers. Target genes relative expression levels were
   calculated using ΔΔCt method and normalized by GAPDH
   (Glyceraldehyde-3-phosphate dehydrogenase).

Western blot assays

   After sonication and centrifugation, protein samples dissolved in RIPA
   Lysis buffer (Bioss, Beijing, China) were quantified with BCA Protein
   Quantification Kit (Beyotime, Shanghai, China). Protein lysates were
   separated in 4–20% density gradient gel (ACE Biotechnology, Nanjing,
   China) and electro-transferred to PVDF membranes (Millipore, USA).
   After blocking in TBST solution containing 5% skimmed milk for 2 h, the
   membrane were subjected to primary antibody incubation at 4 °C
   overnight. Next, the secondary antibodies (Seracare, USA) was added and
   incubated at room temperature for 2 h. Finally, the bands for target
   proteins were detected using ECL chromogenic solution (Yeasen Biotech,
   Shanghai, China).

Cell counting Kit-8 assays

   About 2000 cells were seeded into a well of 96-well plate (100 μL/well)
   and the density of the cells were measured at the indicated time.
   Briefly, 10 μL of Enhanced Cell Counting Kit-8 regent (Bioss, Beijing,
   China) were added into medium and incubated for 1.5 h in the darkness.
   Light Absorbance for each well was measured at 450 nm.

Colony formation assays

   About 2000 cells were seeded into a well of a six-well plate. At 24 h
   after cell seeding, the cells were exposed to irradiation and continued
   cell culture for 7 ~ 14 days. Cell colonies were fixed for 20 min with
   4% paraformaldehyde and stained with crystal violet (Bioss, Beijing,
   China) for 30 min. Finally, the cell colonies were photographed and
   counted under a microscope.

Cell apoptosis

   About 1 × 10^5 cells were seeded into each well of a 6-well plate and
   irradiated at 24 h after seeding. Later, cells were collected and
   re-suspended in Binding buffer premixed with Annexin V-FITC (APExBIO,
   USA) and Propidium Iodide (APExBIO, USA). Next, the cells were
   incubated at room temperature in the darkness for 20 min, and then
   subjected to flow cytometry analysis, immediately.

Immunoprecipitation assays

   For immunoprecipitation assays, about 1 × 10^7 cells were suspended in
   Cell lysis buffer (Beyotime, Shanghai, China) containing 1× PMSF
   (Bioss, Beijing, China) and lysed on a rotator at 4 °C for 4 h. After
   lysis, the cell lysate was collected by centrifuging at 15,000 × g for
   10 min and divided into two centrifuge tubes equally. One tube was
   incubated with primary antibody overnight and another centrifuge tube
   was incubated with IgG antibody (Beyotime, Shanghai, China) as a
   control. Then, the target protein was pulled down with Protein A + G
   Agarose beads (Beyotime, Shanghai, China) from the cell lysates by
   incubating at 4 °C for 3 h. Beads were washed with Cell lysis buffer
   (Beyotime, Shanghai, China) two times, and bound proteins were eluted
   in SDS-loading buffer (Beyotime, Shanghai, China) and subjected to
   Western blot analysis.

Immunofluorescence assays

   About 1 × 10^4 cells were seeded into a small glass-bottom round dish
   (Corning, USA) one day before experiment. First of all, the cells were
   fixed in 4% paraformaldehyde for 20 min, followed by permeabilization
   for 10 min with 0.3% Triton X-100 (Biosharp, China). After washing with
   PBS, the dishes were blocked with PBS containing 0.5% Bovine Serum
   Albumin (Solarbio, Beijing, China) and incubated with primary
   antibodies at 4 °C overnight. Subsequently, dishes were incubated with
   fluorescent group-conjugated secondary antibodies for 2 h in the
   darkness, and finally stained with DAPI (Beyotime, Shanghai, China) for
   10 min. Images were visualized and recorded using a confocal laser
   scanning microscope (Zeiss, Germany).

Protein degradation assays

   In order to characterize the protein degradation rate of EIF3A,
   cycloheximide (CHX, MCE, USA) was added into the culture medium at
   100 μg/ml final concentration, which resulted in the blockade of
   protein synthesis. Next, cell samples were collected at specific time
   points and the expressions of target proteins was detected using
   western blot.

Lysosomes extraction and purification

   Lysosomes were purified from cancer cells using a lysosomes isolation
   kit (Solarbio, China, Cat: EX2670) according to Lin’s method [[46]17]
   with minor modifications. Briefly, about 2 × 10^7 cells were washed
   with PBS and collected by gentle scraping. Then, the cell sediment was
   re-suspended with 4 ml of lysosomal extraction reagent A (Solarbio,
   Beijing, China) and homogenized with a Dounce’s homogenizer for about
   40 strokes. Following homogenization, the cell extracts were
   transferred to new 1.5 ml tubes and centrifuged at 1000 × g for 10 min
   at 4 °C twice to ensure cell debris and nucleus clear. Then, the rough
   mitochondria and lysosomes fraction (MLF) was sedimented by
   centrifugation at 13,000 × g for 10 min at 4 °C and re-suspended in
   600 μl lysosomal extraction reagent A containing protease inhibitor
   cocktail. The MLF was divided into two quits and one was incubated with
   IgG, another was incubated with 10 μl LAMP2 antibodies at 4 °C for two
   hours. Following rotation, equal amounts of PBS-washed BeyoMag™ Protein
   A + G Magnet beads (Beyotime, Shanghai, China) was added into each
   group and continued the incubation at 4 °C for another 2 h. Finally,
   the lysosomes are collected and washed by physical magnetic enrichment
   and subjected to Western blotting.

Animals

   Xenografted mice models were established through subcutaneously
   injecting cancer cells into BALB/c nude mice. Briefly, cancer cell
   suspensions were mixed with the matrix gel (Corning, USA) at the ratio
   of 4:1, and then subcutaneously injected into the right dorsal side of
   4-week-old female nude mice at a dose of 5 × 10^6/mice. When tumor
   volume reaches 200 mm^3, mice were exposed to radiation treatment at a
   dose of 2 Gy every five days. If tumors were larger than 1 cm in
   length, mice were sacrificed to collect tumors for subsequent
   experiments.

Ethics approval and consent to participate’ statement

   All clinical tissue specimens were collected from Shandong Cancer
   Hospital. This study was approved by the Ethics Committee of Shandong
   Cancer Hospital and Institute (approval number: SDTHEC2023009012). The
   authors state that informed consent was obtained from all participants.
   All animal experiments were conducted in accordance with guidelines of
   a protocol approved by the Animal Care and Use Committee of Shandong
   Cancer Hospital.

Statistics

   The Graph Pad Prism 8 software was used for statistical analysis. The
   differences between two independent groups were assessed with the
   Student’s t test analysis. For the comparison of samples above two
   groups, which was analyzed by One-way ANOVA. At p < 0.05, the results
   were calculated as being statistically meaningful.

Results

High expression of ROBO1 is associated with poor prognosis in ESCC

   In order to explore the function of ROBO1 in ESCC, the expression level
   of ROBO1 in patient-derived samples was analyzed by referring to some
   online databases, including GEPIA ([47]http://gepia.cancer-pku.cn/),
   TNMplot ([48]https://tnmplot.com) and Kaplan–Meier Plotter
   ([49]http://kmplot.com). The results showed that ROBO1 was
   significantly up-regulated in ESCC cancerous tissues, especially in
   metastasis, compared with the normal tissues (Fig. [50]1A, C). With the
   progression of ESCC, ROBO1 expression also showed an obvious
   upregulation (Fig. [51]1B). Besides, ESCC patients with higher ROBO1
   expression always exhibited shorter survival time when compared with
   ROBO1-lower group (Fig. [52]1D). In accordance with the online database
   analysis, the majority of clinical ESCC cancerous tissues we collected
   also presented higher ROBO1 expression relative to the paired normal
   tissues (Fig. [53]1E, F). All of this data indicated ROBO1 may play an
   onco-promoting role in ESCC progression. To further investigate the
   function of ROBO1 in ESCC, the level of ROBO1 in several cancer cell
   lines was detected. KYSE450 and TE1 showed the highest ROBO1 expression
   among all selected cell lines (Fig. [54]1G), which indicated these two
   cell lines are more suitable for the next research work.

Fig. 1. High expression of ROBO1 is associated with poor prognosis in ESCC.

   [55]Fig. 1
   [56]Open in a new tab

   A The mRNA expression level of ROBO1 in esophageal cancerous tissues
   (n = 182) and normal tissues (n = 13) was analyzed using GEPIA
   database. *p < 0.05. B The expression of ROBO1 in patients with ESCC at
   different stages was investigated. C The relationship between ROBO1
   expression and ESCC metastasis were evaluated by online TNMplot
   database. D Kaplan–Meier curves showed the correlation between ROBO1
   expression and overall ESCC patient survival by Kaplan–Meier Plotter. E
   ROBO1 protein expression in esophageal cancerous tissues and
   pericarcinomatous tissues was detected by western blot. F
   Representative images of ROBO1 and Ki67 staining in esophageal
   cancerous and pericarcinomatous tissues were presented. Scale bars,
   50 μm. G Different ESCC cell lines were subjected to ROBO1 detection by
   western blot (upper) and qRT-PCR (bottom).

Silencing ROBO1 expression improved ESCC cells radio-sensitivity

   To elucidate the relationship between ROBO1 and ESCC malignancy, the
   function of ROBO1 in mediating cancer cell radio-resistance was
   explored. Western blot results showed that in response to irradiation,
   ROBO1 level was dose-dependently increased, indicating a potential role
   of ROBO1 in regulating ESCC radio-sensitivity (Fig. [57]2A).
   Subsequently, ESCC cell lines with stably decreased ROBO1 were
   established and the shRNA knockdown efficiency was validated at mRNA
   level and protein level, respectively (Fig. [58]2B, C). As expected,
   when ROBO1 was knocked down, the proliferation rate and colony
   formation capacity of ESCC cells was attenuated under radiation
   condition (Fig. [59]2D, E). As the recruiter of DNA damage repair
   machinery, γ-H2A.X’s amount and nucleic distribution pattern were
   usually used to characterize the status of genome. In our results,
   compared with control group, ROBO1 depletion induced γ-H2A.X higher
   expression and more focal particles formation in nucleus under
   irradiation treatment at same time point (Fig. [60]2F, G), which
   suggested ROBO1 knockdown aggravated irradiation-induced DNA damage.
   Meanwhile, flow cytometry analysis uncovered stronger cell apoptosis
   occurred in ROBO1-silenced cells following irradiation exposure
   relative to the control cells (Fig. [61]2H). In accordance with the
   results of flow cytometry, BCL2, an apoptosis inhibitor, showed
   significantly down-regulation after ROBO1 silence (Fig. [62]2I). All of
   these findings indicated that ROBO1 played a vital role in driving ESCC
   malignancy through enhancing cancer cell radio-resistance.

Fig. 2. Silencing ROBO1 expression improved ESCC cells radio-sensitivity.

   [63]Fig. 2
   [64]Open in a new tab

   A The influence of radiation on ROBO1 expression detected by Western
   blot. B, C The knockdown efficiency of ROBO1 in ESCC cells was
   determined by a western blot and RT-qPCR assays. D, E ESCC cell
   proliferation status was evaluated by clone formation assays and CCK-8
   assays after radiation exposure. F Representative immunofluorescence
   images of γ-H2A.X distribution in cells with or without ROBO1 depletion
   under radiation condition. G The influence of ROBO1 knockdown on
   γ-H2A.X expression upon radiation treatment. H Impact of ROBO1
   silencing on cell apoptosis detected by flow cytometry analysis. I The
   expression of apoptosis-associated protein was detected in
   ROBO1-knockdown cells by Western blot.

ROBO1 accelerated eIF3A degradation in ESCC cells

   To gain insight into the molecular mechanism by which ROBO1 mediated
   ECSS cell radioresistance, immunoprecipitation coupled with mass
   spectrometry analysis were employed to profile proteins interacted with
   ROBO1 (Fig. [65]3A). Among all identified proteins, the family of
   eukaryotic translation initiation proteins and stress granules
   associated proteins were found to be enriched in anti-ROBO1 sediments
   (Fig. [66]3B). To further validate this endogenous interaction, we
   incubated ESCC cell lysates with anti-ROBO1 antibodies and found eIF3A
   and G3BP2 were co-precipitated with ROBO1 in both cell lines (Fig.
   [67]3C). Besides, immunofluorescent staining demonstrated the
   colocalization between ROBO1 and eIF3A in ESCC cancer cells (Fig.
   [68]3D). All of this results demonstrated the interaction between ROBO1
   and eIF3A was genuinely existed in ESCC cells. Intriguingly, we found
   ROBO1 knockdown significantly diminished the expression of eIF3A at
   protein level, but with no effect on its mRNA expression (Fig. [69]3E,
   F). Therefore, translation inhibitor cycloheximide (CHX) was used to
   block cellular de novo peptides synthesis and found silencing ROBO1
   significantly decreased the degradation rate of eIF3A in ESCC cancer
   cells (Fig. [70]3G). In conclusion, ROBO1 directly interacted with
   eIF3A to accelerate its degradation in ESCC cells. To reveal the
   connection between of eIF3A expression and esophagus cancer
   progression, the expression level of eIF3A was investigated by
   immunohistochemical method, and we have found the level of eIF3A in
   esophagus cancerous tissues was decreased relative to the paired
   para-carcinoma tissues (Fig. [71]3H). Besides, TCGA database analysis
   showed eIF3A expression in ESCC decreased with the development of
   cancer stages, lower eIF3A expression always associated with dismal
   survival status for ESCC patients (Fig. [72]3I, J), which further
   indicated eIF3A functioned a tumor suppressing role in ESCC confronted
   with ROBO1-mediated malignancy.

Fig. 3. ROBO1 accelerated eIF3A degradation in ESCC cells.

   [73]Fig. 3
   [74]Open in a new tab

   A ROBO1 was specifically immunoprecipitated with antibodies in ESCC
   cells. B LC-MS analysis. C Co-immunoprecipitation assays were performed
   to demonstrate the interaction of ROBO1 with eIF3A and G3BP2. D
   Immunofluorescent staining displayed the co-localization of ROBO1 and
   eIF3A in ESCC cells. E, F The influence of ROBO1 knockdown on eIF3A
   expression was detected at protein level and mRNA level. **p < 0.01. G
   The degradation rate of eIF3A was monitored in ESCC cells with or
   without ROBO1 depletion. H Immunohistochemical staining of eIF3A in
   esophagus cancerous tissues and para tissues. Scales, 50 μm. I The
   expression levels of eIF3A in ESCC patients at different stages. J
   Overall survival rate of ESCC patients with different eIF3A expressions
   were analyzed using a online database
   ([75]https://smuonco.shinyapps.io/PanCanSurvPlot/).

ROBO1 elicited lysosomal eIF3A degradation through interacting with G3BP2

   The finding that knocking down ROBO1 expression induced eIF3A
   upregulation at protein level but not mRNA level led us to hypothesize
   whether ROBO1 was involved in regulating eIF3A protein stability in
   ESCC cells. Therefore, the canonical protein degradation pathways
   including proteasome and lysosomes, were blocked using small molecule
   inhibitors, respectively. Intriguingly, chloroquine (inhibitor of
   lysosome), but not MG132(inhibitor of proteasome), successfully
   compromised the effects of ROBO1 silencing-induced eIF3A upregulation
   in ESCC cells, indicating that lysosomes may participate in modulating
   eIF3A protein stability (Fig. [76]4A). Then, we re-checked the mass
   spectrometry (MS) data of anti-ROBO1 and G3BP2 (Ras GTPase-activating
   protein-binding proteins 2) was screened out owning to its reported
   function in anchoring tuberous sclerosis complex (TSC) protein complex
   to lysosomes [[77]18]. Immunoprecipitation and confocal assays
   demonstrated the direct interaction between G3BP2 and ROBO1 in ESSC
   cells which is in accordance with the results of MS (Fig. [78]4B, C).
   To explore whether G3BP2 was engaged in regulating eIF3A degradation
   like TSC, G3BP2 expression was suppressed using targeted siRNAs in ESCC
   cells. It was found that G3BP2 depletion significantly induced eIF3A
   upregulation at protein level, but not mRNA level (Fig. [79]4D, E). Due
   to eIF3A protein expression can be influenced by translation or
   degradation processes, different inhibitors targeting these processes
   were employed to detect their influence on eIF3A expression. Similar to
   the results of ROBO1, chloroquine was able to block G3BP2
   silence-mediated eIF3A upregulation in cancer cells exposed to
   irradiation (Fig. [80]4F). To further interrogate whether lysosomes
   were implicated in orchestrating eIF3A degradation, the lysosomes were
   isolated from ESCC cells and subjected to Western blot. We can see from
   the results (Fig. [81]4G), ROBO1, G3BP2 and eIF3A were simultaneously
   presented in lysosomes fraction without endoplasmic reticulum,
   mitochondria and nucleus contamination. In line with the results of
   lysosomes fraction, immunofluorescent confocal assays demonstrated that
   all of G3BP2, ROBO1 and eIF3A were co-localized with lysosomes marker
   (LAMP2) in irradiation-treated ESCC cells (Fig. [82]4H, I).
   Collectively, all of mentioned results indicated that ROBO1 modulated
   eIF3A stability in ESCC cells through interacting with G3BP2 to trigger
   lysosomal degradation pathway.

Fig. 4. ROBO1 elicited lysosomal eIF3A degradation through interacting with
G3BP2.

   [83]Fig. 4
   [84]Open in a new tab

   A ROBO1 regulated eIF3A expression relying on lysosomes. MG132, a
   protease inhibitor. Chloroquine, an autophagy lysosome inhibitor. B
   G3BP2 interacted with ROBO1 and eIF3A in ESCC cells. C
   Immunofluorescence staining showed the colocalization of ROBO1 (green)
   and G3BP2 (red) in ESCC cells. Scale bars, 10 μm. D, E G3BP2 knockdown
   increased eIF3A expression at protein level, but not at mRNA level.
   **p < 0.01. F G3BP2 promoted eIF3A degradation relying on lysosomes
   activity. G Lysosomes were purified from cells and subjected to Western
   blot. ERp72, TOM20 and Histon3 were used as the markers of endoplasmic
   reticulum, mitochondria and nucleus, respectively. H–J
   Immunofluorescence assays demonstrated the co-localization of G3BP2,
   ROBO1, eIF3A with lysosomal marker protein LAMP2 (red) in ESCC cells.
   Nucleus were stained with DAPI (blue). Scale bars, 10 μm.

G3BP2 knockdown impaired ESCC tolerance to irradiation

   G3BP2 was reported to participate in organizing stress granules, a kind
   of molecular condensates that form in response to various stresses.
   Whether G3BP2 functioned in orchestrating ESCC radio-resistance?
   Immunofluorescence experiments were performed and found that G3BP2
   formed more particles at 2 hours after radiation exposure and declined
   with time, suggesting the potential of G3BP2 as a irradiation responsor
   (Fig. [85]5A). Then, cell lines with stable decreased G3BP2 was
   established using lentivirus-mediated shRNA transfection (Fig. [86]5B).
   Upon irradiation treatment at different doses, ESCC cell with decreased
   G3BP2 presented obviously attenuated proliferation ability and
   clonogenicity compared with control cells (Fig. [87]5C, D). To explore
   whether G3BP2 was implicated in regulating irradiation-induced DNA
   damage repair process, the expression of γ-H2A.X was detected using
   Western blot and immunofluorescent staining. Irradiation exposure
   induced higher γ-H2A.X expression and longer existing time in
   G3BP2-depleted cells when compared with the control cells (Fig. [88]5E,
   F), suggesting that silencing G3BP2 impaired cancer cell DNA repair
   capability. What’s more, flow cytometry analysis identified more
   obvious cell apoptosis in G3BP2-depleted ESCC cells triggered by
   radiation, which is in accordance with the Western blot results of
   BCL2, an inhibitor of cell apoptosis (Fig. [89]5G, H). Taken together,
   high expression of G3BP2 rendered ESCC cells strong resistance to
   radiotherapy.

Fig. 5. G3BP2 knockdown impaired ESCC tolerance to irradiation.

   [90]Fig. 5
   [91]Open in a new tab

   A Irradiation triggered more G3BP2 (green) particles formation. Nucleus
   were stained with DAPI (blue). B G3BP2-silenced cells were constructed
   using shRNAs. **p < 0.01. C The survival status of ESCC cells were
   monitored after experiencing different doses of radiation. D The colony
   formation ability of ESCC cells was detected. E, F The influence of the
   G3BP2 inhibition on γ-H2A.X expression was detected by Western blot and
   immunofluoresent staining at 2 h after irradiation. The impact of G3BP2
   silencing on cell apoptosis was detected by flow cytometry (G) and
   Western blot (H).

ROBO1 stimulated mTOR signaling pathway through blocking eIF3A-mediated P53
translation

   To further elucidate the mechanisms by which ROBO1 and eIF3A regulated
   ESCC radioresistence, cancer cells with or without decreased ROBO1
   expression were subjected to RNA sequencing. Kyoto Encyclopedia of
   Genes and Genomes (KEGG) analysis showed that genes with altered
   expressions were specifically enriched in P53 signaling pathway (Fig.
   [92]6A). Furthermore, silencing ROBO1 expression in ESCC cells
   obviously induced P53 upregulation at protein level, but no significant
   difference was observed at mRNA level (Fig. [93]6B, C), indicating that
   ROBO1 post-transcriptionally regulated P53 expression. Therefore, the
   proteasome inhibitor (MG132) and translation blockade (CHX) were
   employed to interrupt the process that potentially engaged in
   ROBO1-mediated P53 alteration. It was shown that CHX, but not MG132,
   effectively compromised the upregulation of P53 induced by ROBO1
   silence in ESCC cells (Fig. [94]6D). We hypothesized whether ROBO1
   regulated P53 expression through direct molecular binding? but
   anti-ROBO1 immunoprecipitation assays failed to trace the direct
   interaction between ROBO1 and eIF3A. Surprisingly, suppressing eIF3A
   expression using siRNAs completely abrogated ROBO1 depletion-triggered
   P53 upregulation in ESCC cells (Fig. [95]6E), suggesting the
   indispensable role of eIF3A in modulating P53 elevation. After
   irradiation treatment, cancer cells with decreased eIF3A expression
   were exposed to the several inhibitors treatment, and we have found the
   addition of the translation inhibitor (CHX) dramatically mitigated P53
   elevation initiated by eIF3A alteration (Fig. [96]6F), indicating that
   the exist of eIF3A contributed to P53 translation. Large evidence has
   demonstrated that P53 participated in modulating mTOR (mammalian target
   of rapamycin) signaling activity and suppressing its downstream genes
   expressions, including DNA nucleotide excision repair (NER) genes
   [[97]19, [98]20]. Here, we have found P53 silencing significantly
   elevated mTOR phosphorylation and NER genes expressions in ESCC cells,
   which was consistent with others’ reports (Fig. [99]6G). To make it
   clear whether ROBO1 participated in regulating radiosensitivity through
   eIF3A/P53/mTOR pathway, the expressions of ROBO1 and eIF3A were
   suppressed in ESCC cells. We have found eIF3A silencing obviously
   stimulated mTOR signaling activity and enhanced NER gene expression
   (Fig. [100]6H), contrary to the results observed in G3BP2 or ROBO1
   abrogated cells (Fig. [101]6I, J). All of these data demonstrated that
   ROBO1 tuned ESCC radiosensitivity through eIF3A/ P53/mTOR axis.

Fig. 6. ROBO1 stimulated mTOR signaling pathway through blocking
eIF3A-mediated P53 translation.

   [102]Fig. 6
   [103]Open in a new tab

   A Pathway enrichment analysis for the influence of ROBO1 knockdown in
   ESCC. B, C The influence of ROBO1 silence on P53 expression was
   detected at protein level and mRNA level. **p < 0.01. D P53 protein
   degradation and translation was blocked in ESCC cells with or without
   ROBO1 depletion. MG132, a proteasome inhibtor. CHX, a translation
   blockade. E ROBO1 regulated the translation of P53 with the help of
   eIF3A. F The influence of EIF3A on regulating P53 translation and
   stability was explored by Western blot. G Silencing P53 initiated mTOR
   signaling cascades and induced downstream DNA nucleotide excision
   repair (NER) genes expressions in response to irradiation treatment.
   H–J The influence of G3BP2, ROBO1 or eIF3A abrogation on mTOR signaling
   activity was explored in ESCC cells upon irradiation exposure.

Disrupting ROBO1/G3BP2/eIF3A axis effectively overcame the resistance of ESCC
to radiation in vivo

   To further elucidate the function of ROBO1 in regulating ESCC
   tumorigenesis in vivo, xenografted tumor models with or without ROBO1
   knockdown were established subcutaneously. Compared with the control
   group, ROBO1 depletion significantly reduced tumor burdens and the
   tumor-suppressing effect was further amplified especially when mice
   were exposed to irradiation (Fig. [104]7B, C), suggesting that ROBO1
   functioned in strengthening radio-resistance in ESCC. In addition,
   immunohistochemical (IHE) staining results showed that silencing ROBO1
   significantly increased eIF3A and P53 expression and inhibited mTOR
   phosphorylation in ROBO1-deficient xenografted solid tumor tissues in
   relative to the control group, which is consistent with the observed
   cellular results (Fig. [105]7D). Whether G3BP2-mediated eIF3A lysosomal
   degradation was implicated in regulating ESCC cell growth and
   radiotherapy tolerance in vivo, tumor models with decreased G3BP2 were
   established in nude mice. Similarly, G3BP2 deficiency obviously
   mitigated tumor growth in mice and efficiently impaired the ability of
   cancer cell to tolerance radiotherapy in vivo (Fig. [106]7E, F). IHE
   staining revealed a negative correlationship between G3BP2 and eIF3A
   expressions in cancerous tissues (Fig. [107]7G). Besides, increased P53
   expression and downregulated mTOR phosphorylation were presented in
   G3BP2-decreased tissues, further demonstrating the critical role of
   G3BP2 in modulating ESCC response to irradiation. Collectively, all of
   these results demonstrated that disrupting ROBO1 or G3BP2 mediated
   eIF3A protein stability will be beneficial for improving the efficiency
   of radiotherapy and limiting the progression ESCC.

Fig. 7. Disrupting ROBO1/G3BP2/eIF3A axis effectively overcame ESCC
radio-resistance in vivo.

   [108]Fig. 7
   [109]Open in a new tab

   A Schematic illustration of tumor inoculation and irradiation process
   for building mice models. B Representative picture of xenografted solid
   tumors dissected from mice exposed to irradiation (IR) or not. C Tumor
   weights analysis. **p < 0.01. D Representative immunohistochemical
   staining pictures of Ki67, ROBO1, eIF3A, P53 and phosphorylated mTOR in
   xenografted solid tumor tissues with or without ROBO1 deficiency. E
   Representative images of xenografted solid tumors with or without eIF3A
   depletion. F Statistical analysis of tumor weights. **p < 0.01. G
   Representative immunohistochemical staining pictures of solid tumor
   tissues with or without G3BP2 decrease.

Discussion

   Over the past decades, numerous studies have revealed the
   tumor-promoting role of SLIT2/ROBO1 in several cancer types, including
   prostate cancer, hematopoietic neoplasm and lobular breast cancer, etc
   [[110]21, [111]22]. Traditionally, the interaction of SLIT2 and ROBO1
   was able to drive carcinogenesis through directly triggering a series
   of downstream signaling pathways in cancer cells. In tumor environment,
   melanoma cells could recruit endothelial cells to construct vessels in
   a ROBO1-dependent manner through releasing soluble SLIT2 proteins
   [[112]23], which further increased the functional complexity of ROBO1.
   However, the marker of gene silencing expression, hypermethylation, was
   uncovered within the promoter regions of ROBO1 and SLIT2 in non-small
   cell lung, ovarian, glioma, hepatocellular, and colorectal cancers
   [[113]21, [114]24], which may indicating the detrimental function of
   SLIT2/ROBO1 for cancer progression. In our study, we have found ROBO1
   was upregulated in ESCC compared with normal tissues and rendered
   cancer cells enhanced resistance to radiotherapy. Knocking down of
   ROBO1 obviously restricted ESCC cancer cell proliferation in vitro and
   in vivo especially when exposed to irradiation. In breast cancer cells,
   Slit induced redistribution of Robo1 from intracellular compartments to
   the plasma membrane dependent on USP33 [[115]25], and then ROBO1 was
   cleaved by metalloproteinases and c-secretase and migrated into the
   nucleus [[116]26], indicating that the cellular location of ROBO1 was
   not limited to cytoplasm membrane. Our results confirmed ROBO1
   interacted with eIF3A and G3BP2 and anchored to lysosomes to regulate
   eIF3A stability in ESCC cytosol, but the possibility of ROBO1 function
   as a membrane receptor during radiotherapy should not be fully ruled
   out and more research works need to be performed.

   Translation is a multi-step process that in charge of almost all
   protein-encoding genes expressions. During this process, the eukaryotic
   translation initiation factors are major players with at least 12
   members participation in a specific spatiotemporal manner [[117]27].
   Among all of them, eIF3 is the largest and most complex factor,
   comprising 13 subunits designated from eIF3a to eIF3m [[118]28]. As the
   largest isoform of eIF3 family, eIF3A plays a key role in translation,
   which has been widely reported to be implicated in tumorigenesis, cell
   differentiation, DNA synthesis and repair, fibrosis, etc [[119]29]. In
   human non–small cell lung cancer, eIF3a promotes glucosemetabolism and
   cell proliferation via promoting small GTPase Rheb synthesis to
   orchestrate AMPK activation [[120]30]. Depletion of eIF3a significantly
   reduced HIF1α protein level and cellular glycolysis ability in
   hepatocellular carcinoma cells through controlling internal ribosomal
   entry site dependent translation [[121]31]. However, more and more
   evidences pointed that eIF3a is not deemed essential for the function
   of eIF3, and not all of eIF3a is linked with ribosomes, indicating that
   it may have functions beyond protein translation. It has been reported
   that eIF3A was able to bind with SHC and Raf-1, two components of the
   ERK pathway, to competitively confront EGF-induced neuron cell
   differentiation [[122]32]. Besides, eIF3a was required for non-small
   cell lung cancer stem cell-like traits maintenance through directly
   regulating β-catenin and YY1 expressions at mRNA level, but not at the
   translational level [[123]33]. Although the function of eIF3A in
   various cancers was extensively studied, but its role in ESCC and
   specially in orchestrating cancer cell radio-resistance is still
   elusive.

   RNA sequencing revealed the p53 signaling pathway was influenced
   significantly by ROBO1 knocking down in ESCC cells. P53 is the single
   most commonly mutated gene in human tumors and has been linked to
   various biological processes regulations. In response to DNA damage,
   P53 usually activates damage-specific repair mechanisms, or in
   irreparable damage, activates cellular senescence, or programmed cell
   death to maintain genome integrity [[124]34]. In retinoblastoma models,
   the mTOR activity was markedly increased in p53-deficient tumors and
   rapamycin treatment suppressed tumor cell growth, identifying mTOR
   inhibition as a critical p53 tumor suppressive function [[125]35].
   Here, we have revealed that eIF3A is required for boasting P53
   translation in ESCC to trigger cell apoptosis in response to
   irradiation treatment. Mechanically, ROBO1 interact with eIF3A and
   G3BP2 to form a heterocomplex neighboring lysosomes to accelerate eIF3A
   degradation, resulting in P53 translation disruption. When P53-mediated
   mTOR signaling inhibition was relieved, downstream signaling initiated
   the upregulation of DNA nucleotide excision repair (NER) associated
   genes, including XPC and XPA, which in turn enhanced ESCC cells ability
   to cope with irradiation treatment (Fig. [126]8). Overall, the findings
   of this study implied that ROBO1 contributed to the enhancement of ESCC
   radioresistance through regulating P53/mTOR signaling and provided
   ROBO1 as a potential target for the treatment and prognosis of ESCC.

Fig. 8. Schematic diagram deciphering the mechanisms by which ROBO1, G3BP2
and eIF3A form hetero-complex to mediate radio-resistance in ESCC cells.

   Fig. 8
   [127]Open in a new tab

   Mechanistically, ROBO1 interacted with G3BP2 and eIF3A to form
   heterocomplex and accelerated eIF3A degradation through activating
   lysosome signaling pathway in ESCC cells. Under irradiation exposure,
   free eIF3A inactivated mTOR signaling and downreglated its downstream
   DNA nucleotide excision repair (NER) genes expression through directly
   promoting P53 translation, which finally attenuated ESCC
   radioresistance and retarded cancer severity.

Supplementary information

   [128]Original Data^ (14.3MB, pdf)
   [129]supplements^ (4.2MB, doc)

Acknowledgements