Golgi Phosphoprotein 3 Mediates Radiation-Induced Bystander Effect via ERK/EGR1/TNF-α Signal Axis

The radiation-induced bystander effect (RIBE), an important non-targeted effect of radiation, has been proposed to be associated with irradiation-caused secondary cancers and reproductive damage beyond the irradiation-treated area after radiotherapy. However, the mechanisms for RIBE signal(s) regulation and transduction are not well understood. In the present work, we found that a Golgi protein, GOLPH3, was involved in RIBE transduction. Knocking down GOLPH3 in irradiated cells blocked the generation of the RIBE, whereas re-expression of GOLPH3 in knockdown cells rescued the RIBE. Furthermore, TNF-α was identified as an important intercellular signal molecule in the GOLPH3-mediated RIBE. A novel signal axis, GOLPH3/ERK/EGR1, was discovered to modulate the transcription of TNF-α and determine the level of released TNF-α. Our findings provide new insights into the molecular mechanism of the RIBE and a potential target for RIBE modulation.


Introduction
Malignant tumors are the main cause of death in the world, and radiotherapy is widely used in tumor treatment. In radiotherapy, DNA damages are induced by ionizing radiation (IR) in the targeted cells, including cancer cells and the surrounding normal cells, which eliminate the tumor and/or cause normal tissue injury. Except for the targeted effects, some biological events, including genomic instability, epigenetic changes, DNA methylation, tumor formation, etc. [1], in the non-irradiated cells/tissues were observed beyond the irradiated area. These effects in non-irradiated cells or tissues, termed the radiation-induced bystander effect (RIBE), implies that the radiation-induced biological effect on cells or tissues is not limited to the targeted cells. It is becoming clear that the RIBE could be triggered by low-or high-dose radiation and occurs not only in adjacent tissues but also in distant tissues [2]. Furthermore, mutations and genomic instability in the bystander cells tend to increase the risk of carcinogenesis [3]. Epidemiological studies indicate that approximately 1 in 70 patients surviving more than 10 years after prostate cancer radiotherapy will develop secondary cancer. In addition to the tissues adjacent to the radiation field, secondary tumor formation is also found in the tissues distant from the radiation field. Conceptually, RIBE contributes to radiation-induced secondary carcinogenesis [4]. Mancuso et al. [3] further reported that DNA double-strand breaks (DSBs) and apoptotic cell death in the non-irradiated cerebellum were caused by irradiation, insights to understand the underlying mechanism of RIBE and give some hints to protect normal tissues in radiotherapy.

Irradiation
The cells were irradiated with a XHA600D X-ray irradiator (SHINVA, Zibo, Shandong, China) at a dose rate of 0.189 Gy/min. Before irradiation, the culture medium was replaced with the fresh medium. A dose of 10 Gy was used in all subsequent experiments.

Model of RIBE In Vitro
In the Transwell co-culture system, cells were seeded into the Transwell insert (2 × 10 5 cells/insert) and the corresponding 6-well plate (5 × 10 4 cells/well), respectively. After overnight incubation, the culture medium was replaced with fresh medium. Subsequently, the cells in the Transwell insert were irradiated with X-rays, immediately followed by co-culture with the cells (bystander) in the plate (Figure 1a).
In the medium transfer experiment, the conditioned medium was collected from the irradiated cells at the indicated time points, followed by centrifugation at 3000 g for 5 min. The supernatants were collected and then transferred to treat the bystander cells ( Figure 1b).

Knockout of GOLPH3 by Using CRISPR/Cas9
The sgRNA (5-CACCGCCGCTACCGTGAGTTCGTG-3) targeted to GOLPH3 was cloned into lentiCRISPRv2 vector (Addgene, MA, USA, Cat#52961) and validated sequencing. The constructed sgRNA lentiviral vector and the packaging plasmids (psPAX2 and pMD2.G) were co-transfected into HEK293T cells to produce lentiviruses. The lentiviruses were used to infect A549 cells, followed by puromycin selection (2 µg/mL) for 7 days. After single-cell culture, single-cell clones were obtained and subjected to sequencing for the target sequence. Finally, the GOLPH3 knockout cell line (A549/GOLPH3-KO) was generated and then utilized for further experiments.

ELISA Assay
Concentrations of human TGF-β (Elabscience, Wuhan, China, Cat#E-EL-0162c), TNF-α (Elabscience, Cat#E-EL-H0109c) and IL-6 (Elabscience, Cat#E-EL-H6156) secreted from the irradiated cells were measured with ELISA kits, in accordance with the instructions of the manufacturer. Briefly, cell culture media were collected and centrifuged at 1000 g (20 min, 4 • C) to remove cell debris. The supernatants were collected for ELISA assays.

Micronucleus (MN) Test
The frequency of micronucleus formation was assessed with the cytokinesis-block micronucleus technique [28]. Briefly, 5 × 10 4 bystander cells were seeded into each well of 6-well plate. After co-culture with irradiated cell, cytochalasin B (Sigma, Saint Louis, Missouri, USA, Cat#C6762) was added into the culture medium with a final concentration of 1 µg/mL, and then the cells were fixed with 4% paraformaldehyde (Sigma, Cat#P6148-100G) 48 h later (two doubling times) and stained with 0.1% acridine orange (Sigma, Cat#A6014) solution. Finally, the cells were viewed under a DMI4000B fluorescence microscope (Leica, Wetzlar, Germany). The micronuclei in at least 1000 bi-nucleated (BN) cells were counted, and the frequencies of MN per 1000 BN cells were calculated.

Statistical Analysis
All experiments were repeated independently at least 3 times. The results were expressed as mean ± standard deviation. The difference was calculated with Student's t-test or one-way ANOVA using SPSS v21.0 software (SPSS, USA). p < 0.05 indicates a statistically significant difference.

GOLPH3 Plays a Key Role in the Production of RIBE Signals Released from Irradiated Cells
To explore the relationship between GOLPH3 and RIBE in vitro, we stably knocked down GOLPH3 expression in A549 and H1299 cells with two independent short hairpin RNAs (GOLPH3-RNAi#1 and GOLPH3-RNAi#2), validated by Western blot (Figure 1c). The RNAi-NC and GOLPH3-RNAi cells were irradiated and cocultured with the bystander cells. The bystander A549 or H1299 cells were used to determine the RIBE in cancer cells, and the bystander NHLF cells were used to determine the RIBE in normal lung cells. The frequency of MN and DSBs (γ-H2AX foci positive cells) was then used to assess the RIBE. The results in Figure 1d show that the increase in MN yields was observed in the bystander cells (A549 or NHLF) after co-culture with the irradiated GOLPH3 wild-type cells (A549 and A549/RNAi-NC), whereas knockdown of GOLPH3 in irradiated cells (A549/GOLPH3-RNAi) significantly attenuated the formation of MN in the bystander cells. A similar phenomenon was also observed in the bystander cells (H1299 or NHLF) co-cultured with GOLPH3 wild-type or knockdown H1299 cells ( Figure 1e). In addition, the fraction of γ-H2AX positive cells also increased in the bystander cells (A549, H1299 or NHLF), which were co-cultured with irradiated GOLPH3 wild-type cells, whereas the fraction of γ-H2AX-positive cells was dramatically reduced in the bystander cells cocultured with irradiated GOLPH3 knockdown cells (Figure 1f,g). These results indicate that knocking down GOLPH3 in the irradiated cells significantly blocks the occurrence of the RIBE. This was also supported by the results that COX-2 protein, an important mediator of the RIBE, was upregulated in the bystander cells by the RIBE signals released from irradiated GOLPH3 wild-type cells, but the induction of COX-2 was reversed after knockdown of GOLPH3 in the irradiated cells (Figure 1h,i). To further confirm the key role of GOLPH3 in RIBE, the rescue experiment was performed by re-expressing exogenous GOLPH3Res-3×Flag in GOLPH3 knockdown cells (Figure 2a,b). Our results showed that the MN yields in the bystander cells were restored after re-expression of GOLPH3 in the irradiated GOLPH3 knockdown cells (Figure 2c,d). Moreover, re-expression of GOLPH3 in the irradiated A549/GOLPH3-RNAi#2 or H1299/GOLPH3-RNAi#2 cells significantly elevated the expression of COX-2 in the bystander cells (Figure 2e,f). These results indicate that the re-expression of GOLPH3 in the irradiated GOLPH3 knockdown cells effectively rescues the RIBE, confirming that GOLPH3 is crucial to the production of the RIBE signals in the irradiated cells. To further confirm the key role of GOLPH3 in RIBE, the rescue experiment was performed by re-expressing exogenous GOLPH3Res-3×Flag in GOLPH3 knockdown cells (Figure 2a,b). Our results showed that the MN yields in the bystander cells were restored after re-expression of GOLPH3 in the irradiated GOLPH3 knockdown cells (Figure 2c,d). Moreover, re-expression of GOLPH3 in the irradiated A549/GOLPH3-RNAi#2 or H1299/GOLPH3-RNAi#2 cells significantly elevated the expression of COX-2 in the bystander cells (Figure 2e,f). These results indicate that the re-expression of GOLPH3 in the irradiated GOLPH3 knockdown cells effectively rescues the RIBE, confirming that GOLPH3 is crucial to the production of the RIBE signals in the irradiated cells.

TNF-α Acts as an Intercellular Molecule in GOLPH3-Mediated RIBE
To explore the nature of signal molecule(s) released by the irradiated cells, we transferred the conditioned medium (CM) at the indicated time points after IR to treat the bystander cells. As shown in Figure 3a, the CM harvested at 1.5 h after IR elevated MN frequency to a peak level in the bystander cells. It is well-known that RIBE could be mediated by the soluble factors (such as TNF-α, TGF-β, IL-6, etc.) released by irradiated cells [8][9][10][11][12]. Hence, a GOLPH3 knockout cell line (A549/KO-GOLPH3) was established (Supplementary Figure 1A,B), and then the TNF-α, TGF-β and IL-6 in the 1.5 h CM harvested from the irradiated A549 or A549/KO-GOLPH3 were measured with ELISA. Interestingly, a

TNF-α Acts as an Intercellular Molecule in GOLPH3-Mediated RIBE
To explore the nature of signal molecule(s) released by the irradiated cells, we transferred the conditioned medium (CM) at the indicated time points after IR to treat the bystander cells. As shown in Figure 3a, the CM harvested at 1.5 h after IR elevated MN frequency to a peak level in the bystander cells. It is well-known that RIBE could be mediated by the soluble factors (such as TNF-α, TGF-β, IL-6, etc.) released by irradiated cells [8][9][10][11][12]. Hence, a GOLPH3 knockout cell line (A549/KO-GOLPH3) was established (Supplementary Figure S1A,B), and then the TNF-α, TGF-β and IL-6 in the 1.5 h CM harvested from the irradiated A549 or A549/KO-GOLPH3 were measured with ELISA. Interestingly, a significant increase in TNF-α and TGF-β rather than IL-6 was observed in the 1.5 h CM harvested from the irradiated A549 cells, and knocking out GOLPH3 decreased the amount of TNF-α significantly (Figure 3b). These results suggest that GOLPH3 regulates the release of TNF-α from irradiated cells. To further verify the role of TNF-α in the transduction of the GOLPH3-mediated RIBE, a neutralizing antibody against TNF-α was added into the 1.5 h CM harvested from GOLPH3 wild-type cells (A549 or H1299) before treating the bystander cells. The addition of the neutralizing antibody against TNF-α significantly suppressed the formation of MN (Figure 3c) in the bystander cells, demonstrating that blocking TNF-α inhibited the transduction of the RIBE. This was also supported by the results that the addition of the neutralizing antibody against TNF-α also significantly attenuated the COX-2 induction in the bystander cells treated by the 1.5 h CM harvested from irradiated the A549 or H1299 cells (Figure 3d,e). These results indicate that the TNF-α released from irradiated cells may act as an intercellular signal molecule to mediate the RIBE transduction. significant increase in TNF-α and TGF-β rather than IL-6 was observed in the 1.5 h CM harvested from the irradiated A549 cells, and knocking out GOLPH3 decreased the amount of TNF-α significantly (Figure 3b). These results suggest that GOLPH3 regulates the release of TNF-α from irradiated cells. To further verify the role of TNF-α in the transduction of the GOLPH3-mediated RIBE, a neutralizing antibody against TNF-α was added into the 1.5 h CM harvested from GOLPH3 wild-type cells (A549 or H1299) before treating the bystander cells. The addition of the neutralizing antibody against TNF-α significantly suppressed the formation of MN (Figure 3c) in the bystander cells, demonstrating that blocking TNF-α inhibited the transduction of the RIBE. This was also supported by the results that the addition of the neutralizing antibody against TNF-α also significantly attenuated the COX-2 induction in the bystander cells treated by the 1.5 h CM harvested from irradiated the A549 or H1299 cells (Figure 3d,e). These results indicate that the TNF-α released from irradiated cells may act as an intercellular signal molecule to mediate the RIBE transduction.

GOLPH3 Regulates the Transcription of TNF-α via ERK/EGR1 Pathway
To determine how GOLPH3 regulates TNF-α, we firstly detected the IR-induced transcriptional activation of TNF-α after knocking down GOLPH3. Intriguingly, our results showed that IR enhanced the transcription of TNF-α, but loss of GOLPH3 significantly abolished it after IR (Figure 4a). Given that Egr-1, an "immediate-early response" protein after IR treatment, is a critical transcription factor of TNF-α, the expression of EGR1, at both the mRNA and protein levels, was detected after IR in RNAi-NC and RNAi-GOLPH3 cells. As expected, knocking down GOLPH3 disabled upregulation of the protein and mRNA levels of EGR1 after IR (Figure 4b, c), suggesting the key role of GOLPH3 in EGR1 activation. To explore the possible mechanism of GOLPH3 modulating EGR1, we screened the activation of several pathways previously reported to be associated with GOLPH3. As shown in Figure 4d, ERK1/2 (but not GSK3β, STAT3 or IκB) was activated after IR, and GOLPH3 knockdown significantly blocked the activation of ERK1/2. Furthermore, a selective ERK1/2 inhibitor, PD98059, was used to inhibit the activation of

GOLPH3 Regulates the Transcription of TNF-α via ERK/EGR1 Pathway
To determine how GOLPH3 regulates TNF-α, we firstly detected the IR-induced transcriptional activation of TNF-α after knocking down GOLPH3. Intriguingly, our results showed that IR enhanced the transcription of TNF-α, but loss of GOLPH3 significantly abolished it after IR (Figure 4a). Given that Egr-1, an "immediate-early response" protein after IR treatment, is a critical transcription factor of TNF-α, the expression of EGR1, at both the mRNA and protein levels, was detected after IR in RNAi-NC and RNAi-GOLPH3 cells. As expected, knocking down GOLPH3 disabled upregulation of the protein and mRNA levels of EGR1 after IR (Figure 4b,c), suggesting the key role of GOLPH3 in EGR1 activation. To explore the possible mechanism of GOLPH3 modulating EGR1, we screened the activation of several pathways previously reported to be associated with GOLPH3. As shown in Figure 4d, ERK1/2 (but not GSK3β, STAT3 or IκB) was activated after IR, and GOLPH3 knockdown significantly blocked the activation of ERK1/2. Furthermore, a selective ERK1/2 inhibitor, PD98059, was used to inhibit the activation of ERK1/2. The increase in EGR1 after IR was attenuated in the presence of ERK1/2 inhibitor (Figure 4e), confirming the regulation of EGR1 by ERK. In addition, ERK1/2 inhibitor also significantly suppressed RIBE occurrence, displaying the decreased induction of COX-2 protein (Figure 4f) and MN in the bystander cells (Figure 4g). Collectively, a novel signaling pathway (GOLPH3/ERK/EGR1/TNF-α) in irradiated cells was activated in response to irradiation to trigger the occurrence of the RIBE ( Figure 5). ERK1/2. The increase in EGR1 after IR was attenuated in the presence of ERK1/2 inhibitor (Figure 4e), confirming the regulation of EGR1 by ERK. In addition, ERK1/2 inhibitor also significantly suppressed RIBE occurrence, displaying the decreased induction of COX-2 protein (Figure 4f) and MN in the bystander cells (Figure 4g). Collectively, a novel signaling pathway (GOLPH3/ERK/EGR1/TNF-α) in irradiated cells was activated in response to irradiation to trigger the occurrence of the RIBE ( Figure 5).  ERK1/2. The increase in EGR1 after IR was attenuated in the presence of ERK1/2 inhibitor (Figure 4e), confirming the regulation of EGR1 by ERK. In addition, ERK1/2 inhibitor also significantly suppressed RIBE occurrence, displaying the decreased induction of COX-2 protein (Figure 4f) and MN in the bystander cells (Figure 4g). Collectively, a novel signaling pathway (GOLPH3/ERK/EGR1/TNF-α) in irradiated cells was activated in response to irradiation to trigger the occurrence of the RIBE ( Figure 5).

Discussion
In this study, we found that GOLPH3 mediated RIBE induction by regulating the transcription of TNF-α. Loss of GOLPH3 in the irradiated cells significantly inhibited the RIBE in vivo and in vitro, while re-expression of GOLPH3 in the irradiated GOLPH3 knockdown cells restored the RIBE. Mechanistically, a GOLPH3/ERK/EGR1 signaling axis modulates the transcription of TNF-α to regulate the RIBE. These findings enable a better understanding of the molecular mechanism underlying the RIBE. As we know, gene mutations and chromosomal instability caused by the RIBE may result in cancerous transformation and reproductive impairment [5,29,30]; thus, the occurrence of RIBE increases the health risks to normal tissues in the survivors, especially the young, after radiotherapy. The nausea and fatigue after radiotherapy are also thought to be associated with the RIBE [31]. However, the RIBE, unlike the direct effect of radiation, is unable to be prevented by physical protection. Blocking the generation or transmission of the RIBE intercellular signal molecules or inhibiting the activation of the RIBE signal pathways in the bystander cells are considered as effective strategies to prevent the RIBE. From this perspective, vitamin C, vitamin E, carbon monoxide (CO), etc., have been identified to protect the bystander cells or tissues against the attack of RIBE [32][33][34]. In fact, targeting the pathways of the RIBE in irradiated cells is also effective to prevent the RIBE, though sometimes this may affect the killing effect on the tumor due to the decrease in DNA damage or ROS production. In our study, we confirmed that GOLPH3 could be a potential molecular target for modulating the RIBE. Together with our recent identification of GOLPH3 as a radiosensitization target in lung cancer [27], the inhibition of GOLPH3 in radiotherapy appears to kill these "two birds" with one stone, so targeting GOLPH3 may be a promising strategy to achieve therapeutic benefit in radiotherapy.
Several studies have demonstrated that the cytokine profile of tumor cells changes after IR, and multiple soluble factors, known to induce RIBE, are induced and released into the extracellular space [35]. In general, irradiated cells exhibit a diversity of cytokine profiles depending on cell types, radiation doses, radiation types, etc., so the soluble factor involved in RIBE may be different. As we know, TGF-β released from irradiated HeLa or T98G cells acts as a signal molecule to induce the RIBE [36,37], whereas the IL-6 released from the irradiated PC3 cells triggers the RIBE [38]. In our model using the irradiated A549 or H1299, TNF-α (but not TGF-β or IL-6) acts as an intercellular signal molecule to mediate the transduction of the RIBE, while the release of TGF-β and TNF-α after irradiation was still increased. The finding also implies that the soluble factor(s) involved in the RIBE is determined by not only the irradiated cells but also the bystander cells, displaying cell-type specificity. Previously, the involvement of TNF-α in mediating the RIBE has been confirmed in vitro, and the TNF-α-induced activation of COX-2 in the bystander cells is reported to play a key role in RIBE transduction [39][40][41]. This conclusion is also supported by our findings that secreted TNF-α from the irradiated cells elevated the level of the COX-2 protein in the bystander cells.
The transcriptional factor EGR1 has been previously shown to activate the transcription of TNF-α [40]. Here, in agreement with the previous study, the positive correlation between EGR1 and TNF-α was also observed in the irradiated A549 and H1299 cells, supporting the role of EGR1 in the activation of TNF-α. Furthermore, we found an important role of GOLPH3 in the activation of EGR1 and identified ERK1/2 as a link between GOLPH3 and EGR1. In previous studies, GOLPH3 have been reported to be involved in the activation of multiple signaling pathways, such as NF-κB, GSK3β, STAT3 and ERK1/2 [42,43]. However, we only observed the activation of ERK1/2 rather than NF-κB, GSK3β or STAT3 at 1.5 h after irradiation. These results were in line with those of previous studies, that ERK1/2 [44] was activated at 1 h after IR, whereas activation of NF-κB (phosphorylation of IκB-α) [45] and STAT3 [46] was observed at least 3 h after IR, suggesting an important role for ERK in the early response of cells after IR. Inhibition of ERK1/2 activity with its specific inhibitor in the irradiated cells suppressed the activation of ERG1 and blocked the induction of RIBE, further confirming the key role of ERK1/2 in RIBE induction. These findings are also supported by several previous studies, that EGR1 is the target gene of ERK1/2 [47,48]. Importantly, our results showed that GOLPH3 knockdown blocked the activation of ERK1/2 and EGR1 after irradiation and then inhibited the RIBE, indicating that the GOLPH3/ERK/EGR1/TNF-α signal axis regulates the RIBE. In addition, growth factor receptors, such as epidermal growth factor receptor (EGFR) and platelet-derived growth factor receptor (PDGFR), have been identified to be involved in the substantial activation of the ERK induced by oxidative stress [49,50]. In our previous study, we confirmed that GOLPH3 maintains the stability of the EGFR protein. Therefore, we speculate that GOLPH3 may act as a sensor of the oxidative stress induced by IR, and EGFR may be the mediator between GOLPH3 and the activation of ERK1/2. Certainly, further investigations are needed to verify this speculation.

Conclusions
Our study reveals that GOLPH3 plays a key role in RIBE induction. Furthermore, a novel signal pathway, GOLPH3/ERK/EGR1, was identified to modulate the transcriptional upregulation of TNF-α after IR, which then caused the release of TNF-α from the irradiated cells to induce the RIBE. Our findings provide new insights into understanding the molecular mechanism of the RIBE and suggest that GOLPH3 could be a potential molecular target for the prevention of the RIBE. A novel combination of GOLPH3 inhibitor and radiotherapy offers an exciting therapeutic strategy to overcome the radioresistance of cancer and reduce the risk of secondary cancer caused by the RIBE.

Data Availability Statement:
The data presented in this study are available in article and supplementary material.

Conflicts of Interest:
The authors declare no conflict of interest.