MMP-9 Reinforces Radiation-Induced Delayed Invasion and Metastasis of Surviving Neuroblastoma Cells Through Second-Signaling Positive Feedback With NFkB Via Both ERK and IKK Activation

Neuroblastoma (NB) progression is branded with hematogenous metastasis and frequent relapses. Despite intensive multimodal clinical therapy, outcomes for patients with progressive disease remain poor, with negligible long-term survival. Therefore, understanding the acquired molecular rearrangements in surviving cells with therapy pressure and developing improved therapeutic strategies is a critical need to improve the outcomes for high-risk NB patients. We investigated the rearrangement of MMP9 in NB with therapy pressure, and unveiled the signaling that facilitates NB evolution. Radiation-therapy (RT) signicantly increased MMP9 expression/activity, and the induced enzyme activity was persistently maintained across NB cell lines. Further, RT-triggered NFkB transcriptional activity and this RT-induced NFkB were required/adequate for MMP9 maintenance. RT-triggered NFkB-dependent MMP9 actuated a second-signaling feedback to NFkB, facilitating a NFkB-MMP9-NFkB positive feedback cycle (PFC). Critically, MMP9-NFkB feedback is mediated by MMP9-dependent activation of IKKβ and ERK phosphotransferase activity. Beyond its tumor invasion/metastasis function, PFC-dependent MMP9 lessens RT-induced apoptosis and favors survival pathway through the activation of NFkB signaling. In addition, PFC-dependent MMP9 regulates 19 critical molecular determinants that play a pivotal role in tumor evolution. Interestingly, seven of 19 genes possess NFkB-binding sites, demonstrating that MMP9 regulates these molecules by activating NFkB. Collectively, these data suggest that RT-triggered NFkB-dependent MMP9 actuates feedback to NFkB though IKKβ- and ERK1/2-dependent IkBα phosphorylation. This RT-triggered PFC prompts MMP9-dependent survival advantage, tumor growth, and dissemination. Targeting therapy-pressure-driven PFC and/or selective inhibition of MMP9 maintenance could serve as promising therapeutic strategies for treatment of progressive NB that dees current clinical therapy. The cell microarray (CMA) approach allows us to measure changes in protein translation across the treated samples, without inter-sample assay discrepancies. For this, human IMR-32 cells (i) exposed to mock-IR, (ii) RT (2Gy), (iii) PMA, (iv) after ectopic expression of p50/p65, (v) treated with GM6001 and exposed to RT, (vi) treated with aprotinin and exposed to RT, and (vii) transfected with ΔIkB and exposed to RT were collected 24 h post-RT. CMA construction, sectioning, and IHC were performed in our Tissue Pathology Core of the Stephenson Cancer Center following standard protocols, as described in our earlier work (Somasundaram et al. 2019). Appropriate histology controls (H&E) and negative controls with no primary antibody were examined in parallel. Expression and localization of MMP9, NFκB- p105, RelB, pP65, p65, IκBα, pIκBα, ERK, and pERK are investigated. The slides were digitally scanned into virtual slides using a Zeiss Axio Scan Z1 slide scanner at 40× magnication. The whole slide images were then group-analyzed for protein-specic positivity using Aperio image analysis and quantication software (Aperio Technologies, Inc., Buffalo Grove, IL, USA) with the appropriate algorithms for protein specic cellular localization. Group-wise comparisons were performed using GraphPad Prism. In conclusion, in this study, we report that a clinical dose of RT signicantly increases MMP9 activity and the induced activity is persistently maintained across human NB cell lines investigated. Further, we showed that RT triggers NFκB phosphorylation, nuclear localization, and transcriptional activity, and this RT-induced NFκB is required and adequate for the maintenance of MMP9. RT-triggered NFκB-dependent MMP9 actuates a second-signaling feedback to NFκB signaling, thereby facilitating a NFκB-MMP9-NFκB PFC. Our results demonstrated that MMP9-NFκB second signaling feedback is mediated by MMP9-dependent activation of IKKβ and ERK activity. Beyond its regular tumor invasion and metastasis function, PFC-dependent MMP9 activation lessens RT-induced apoptosis through the activation of NFκB signaling. More importantly, this study identied that NFκB-MMP9-NFκB-dependent MMP9 regulates 19 critical molecular determinants that play a pivotal role in tumor evolution. Interestingly, seven of these molecules have binding sites for NFκB, indicating that MMP9 regulates these molecules by activating NFκB. Collectively, these data suggest that acquired maintenance of MMP9 after therapeutic pressure could contribute heavily for tumor evolution, and selective inhibition of activated MMP9 maintenance could serve as a promising therapeutic strategy for progressive NB that dees current clinical therapy.


Introduction
Matrix metalloproteinase (MMPs) are zinc-containing endopeptidases that degrade extracellular matrix (ECM) machinery (Klein, Bischoff 2011). Constitutively, MMPs reside as latent zymogens that become active through multiple mechanisms (e.g., cysteine, allosteric, furin, MMPs, plasmin) (Ra, Parks 2007) and prompt the degradation of matrix barriers. Programmed remodeling of ECM is a key step in cancer evolution, and the MMP-dependent degradation of matrix barrier is critical for tumor cell migration, invasion, and metastasis (Stetler-Stevenson 1990). High expression levels and enzyme activity of MMPs are evident in in malignant tumors, compared with normal, benign, or even in premalignant tissues (Sternlicht, Werb 2001). Functionally, MMPs are involved in tumor cell proliferation (by release of growth factors [GF], cleaving GF-binding proteins and GF-receptors); invasion (by generating an α1-antitrypsin cleavage product); epithelial to mesenchymal transition (EMT); angiogenesis (release of angiogenesis factors); cell cycle checkpoint control; genomic instability; and clonal selection (anchorage-independent and apoptosis-resistant) (Lukashev, Werb 1998;Suzuki et al. 1997;Fowlkes et al. 1994; Levi et al. 1996; Kataoka et al. 1999;Tlsty 1998; Thomasset et al. 1998; Sternlicht et al. 2000). The function of MMPs in altering the metastatic state of cancer cells and the association of MMPs expression with poor prognosis has been recognized in different cancers, including neuroblastoma (NB) ( Ara et al. 1998). In particular, MMP-9, a 92kDa type IV collagenase that plays a vital role in cancer cell invasion and metastasis, is well characterized (Kalavska et al. 2021). However, the function of MMP9 in NB evolution, particularly its reinforcement under therapeutic pressure and the signaling involved, is thus far unrealized.
Neuroblastoma, the most aggressive extra-cranial solid tumor occurring in childhood, remains a major cause of cancer death in infancy (Matthay et al. 2016). Clinically, neuroblastoma progression is branded with hematogenous metastasis and frequent relapses, with a rapidly decreasing timeline (Santana et al. 2008). Despite the current intensive and multimodal therapeutic strategies (chemotherapy, radiotherapy, surgery, immunotherapy, stem cell transplant, 13-cisretinoic acid), outcomes for progressive heterogeneous disease remain poor, with negligible long-term survival (London et al. 2011). Therefore, unveiling the acquired molecular rearrangements in response to current clinical therapy and developing new therapeutic strategies are critical to improve the clinical outcomes for NB patients. Radiation therapy (RT) is one of the mainstream treatment modalities for NB treatment and provides many bene ts, such as: shrinking tumors prior to surgery to enable ease of surgery; treating larger tumors that affect breathing; treating tumors that do not respond quickly to chemotherapy; destroying leftover NB cells after stem cell transplant for high-risk disease; and relieving pain caused by advanced disease. Ultimately, the goal of RT is to kill as many cancer cells as possible while limiting harm to nearby healthy tissue.
Conversely, RT-induced radioresistance in cancer cells is a fundamental barrier limiting the effectiveness of RT. We provided rst evidence on the existence of radioresistance in NB cells after clinically relevant fractionated RT (Madhusoodhanan et al. 2009; Veeraraghavan et al. 2011). Further, we unveiled the intercellular molecular cross talk and the signaling mechanisms involved in NB cell radioresistance Veeraraghavan et al. 2011;Aravindan et al. 2014). It is pertinent to understand how this RT-induced signaling translates into cellular function that contributes to NB evolution.
MMP-9 protein contains ve domains (hemopexin-like, catalytic, signal peptide, hinge region, and propeptide region), of which the catalytic domain with active site and zinc-binding region plays a critical role in enzyme activity and the bronectin site is key for substrate binding and degradation (Huang 2018). The hemopexin-like domain is critical for speci city during recognition and interacts with gelatin and collagen (Roeb et al. 2002). Further, this domain is key in forming a complex with tissue inhibitor of metalloproteinases (TIMPs) and preventing the activation of MMP-9 ( About two decades ago, production of MMP9 by NB cells; its role in induced angiogenesis, tumor progression, and metastasis; and association of MMP9 with advanced disease stage and poor clinical outcomes were documented (Sugiura et al. 1998;Ribatti et al. 1998;Ara et al. 1998 Aravindan et al. 2017). However, a greater understanding of the mechanism(s) involved in acquired maintenance and/or activated MMP9 in therapy-surviving cells is warranted for development of an improved and targeted therapeutic strategy for progressive NB that de es the current intensive multi-modal clinical therapy.
The human MMP-9 gene contains 13 exons and 12 introns, and is located in chromosome 20q13.12. Transcriptionally, the MMP-9 gene is regulated by multiple cis acting factors binding to their matched promoter elements. Transcription factors (TFs) for MMP9 include AP-1 (two recognition sites that bind fos/jun family members), ETS (multiple PEA3 elements), SP1, and NFκB (Benbow, Brinckerhoff 1997;. However the stand-alone AP1 recognition is not su cient and requires substantial NFκB and SP-1 recognition for activation (Benbow, Brinckerhoff 1997;. For instance, TNFα-induced MMP9 promoter activation mandates AP-1, PEA3, NFκB, and Sp-1 recognition (Lauricella- Lefebvre et al. 1993). However, the role of speci c (e.g., NFκB) TFs in MMP9 regulation remains uncharted. As early as the recognition of MMP9's role in NB progression and clinical outcomes (Sugiura et al. 1998;Ribatti et al. 1998;Ara et al. 1998), it has been shown that NFκB-dependent transcriptional activation of MMP9 is key in enhancing basement membrane invasivity in NB (Farina et al. 1999 Further, we reported that NB cells that survive RT have high MMP9 activation, and selectively targeting RTinduced nuclear translocation of NFκB regulates RT-induced MMP9 in these cells (Aravindan et al. 2013b). However, it is not clear how MMP9 activation is maintained in the therapy-resistant NB cells.
In the present study, we investigated the mechanism(s) of MMP9 activation with therapeutic (RT) pressure in NB cells and how the activated MMP9 is persistently maintained in surviving cells. The outcomes of this study showed that, at least in post-treatment surviving NB cells, RT-triggered NFκB mediated MMP-9 reinforced ERK-and/or IKK-dependent MMP9-NFκB feedback, thereby maintaining sustained activation of MMP9.

Cell culture
The human NB cells SH-SY5Y, MC-IXC, SK-PN-DW and IMR-32 were obtained from ATCC (Manassas, VA). Culture and maintenance of the NB SH-SY5Y, IMR-32, and SK-PN-DW cells were performed as described in our earlier studies Aravindan et al. 2013a). For passaging and for all experiments, the cells were detached by using 0.25% trypsin/1% EDTA, resuspended in complete medium, counted (Countess, Invitrogen), and incubated in a 95% air/5% CO 2 humidi ed incubator. For all experiments, the cells were serum-starved by incubating in medium containing 2% serum for at least 12 h, unless otherwise speci ed.

Radiation and inhibition studies
For RT experiments, cells were exposed to 2Gy using a Gammacell 40 Exactor (Nordion International, Inc., Ontario, Canada) at a dose rate of 0.81Gy/min and were then incubated at 37°C. Mock-irradiated cells were treated identically except that cells were not subjected to RT. Irradiated cells were incubated for an with GM6001 or aprotinin for 3 h prior to RT exposure. In addition, as a reference standard for NFκB dependent activation of MMP9, we used Phorbol 12-myristate 13-acetate (PMA, 10 nM, Cayman Chemical), a polyfunctional diterpene phorbol ester. PMA is an activator of protein kinase C and upregulates MMP-9 in a PKCα-NF-κB dependent manner (Shin et al. 2007).
Plasmid preparation, DNA transfection, and luciferase reporter assay NFκB p65 and p50 subunits were transiently transfected into the NB cells following the lipofection method utilizing EffecteneTM reagent (Qiagen) as in our earlier studies (Mohan N et al. 2002). NFκB inhibition was achieved using transient transfection of s32A/s36A double mutant IκBα (ΔIκBα) as discussed earlier (Aravindan et al. 2013b). The mutated form of IκBα with a serine-to-alanine mutation at residues 32 and 36 does not undergo signal-induced phosphorylation, and thus remains bound to NFκB, subsequently preventing nuclear translocation and DNA binding. MMP9 overexpression with full-length human untagged MMP9 expression vector and MMP9 inhibition with human MMP9 shRNA plasmid kit (4 unique 29mer constructs) in retroviral RFP vector (Origene, Rockville, MD) were achieved by transient transfection utilizing TurboFectin transfection reagent. In addition, the plasmid construct pNFκB-Luc was ampli ed, puri ed, and transfected as in our earlier studies (Aravindan et al. 2013b). Cell lysates were assayed for luciferase activity as per the manufacturer's protocol (Biovision Research Products, Mountain View, CA).
Gelatin zymography SH-SY5Y, MC-IXC, and IMR-32 cells seeded in 100-mm plates were irradiated and the conditioned medium (CM) was collected after 24, 48, and 72 h. CM collected from mock-IR plates at 72 h was used as controls. CM was concentrated using nanosep 30K concentrators (Pall Biotech, Westborough, MA), and an equivalent volume was subjected to 10% SDS-PAGE containing gelatin (2 mg/ml). The gels were washed in 2.5% Triton X-100 (3X) and incubated in the buffer (50 mM Tris-HCl, pH 7.6; 10 mM CaCl 2 ; 50 mM NaCl and 0.05% Brij35) for 16 h at 37 °C. Gels were then stained with Coomassie brilliant blue R-250 (0.25% in 40% methanol and 10% acetic acid). The MMP activities were visualized as digested bands in a Canon RE350 video visualizer.

Kinetics of MMP9 activity using a uorogenic substrate
The kinetics of MMP9 activity were measured using a speci c uorogenic substrate (DNP-Pro-Leu-Gly-Leu-Trp-Ala-D-Arg-NH2, Calbiochem), a peptide with the uorescence on one end and a quencher on the other. After the cleavage into two separate fragments by MMP, the uorescence is reclaimed and can be quanti ed in real-time. Equal volumes of CM from the SH-SY5Y, IMR-32, and MC-IXC cells exposed to mock-IR or RT (2Gy) were collected after 2 h (selected for MMP9 activity) and concentrated using nanosep 30K concentrators (Pall Biotech). Concentrated CM (20 µL) in duplicate were mixed with uorogenic substrate (20 µM, dissolved in DMSO) in 96-well plates for a total volume of 50 µL using assay buffer (0.5 M Tris-HCl, pH 7.7; 5 mM CaCl 2 ; 0.2 M NaCl). Kinetics of the MMP9 activity were immediately and continuously (every 20 minutes for 20 h) quanti ed by measuring the uorescence intensity (excitation 280 nm; emission 360 nM) using a Synergy II micro plate reader (Biotek). Group-wise comparisons of MMP9 activity were performed using GraphPad Prism.

Immunoblotting
Total protein extraction and immunoblotting were performed as described in our earlier study ). The protein-transferred membranes were incubated with mouse monoclonal anti-MMP-9, IKKβ, p65 (Santa Cruz Biotechnology, Inc., Santa Cruz, CA), rabbit polyclonal anti-ERK and phosphor ERK antibodies, and developed with the appropriate anti-mouse/anti-rabbit antibodies (BioRad Laboratories, Hercules, CA). Blots were stripped and reblotted with mouse monoclonal anti-α-tubulin antibody (Santa Cruz) to determine equal loading of the samples. Band intensity analysis was quanti ed using Quantity One 1D image analysis software (BioRad). Group-wise comparisons were performed using GraphPad Prism.

Cell-microarray construction and immunohistochemistry
The cell microarray (CMA) approach allows us to measure changes in protein translation across the treated samples, without inter-sample assay discrepancies. For this, human IMR-32 cells (i) exposed to mock-IR, (ii) RT (2Gy), (iii) PMA, (iv) after ectopic expression of p50/p65, (v) treated with GM6001 and exposed to RT, (vi) treated with aprotinin and exposed to RT, and (vii) transfected with ΔIkB and exposed to RT were collected 24 h post-RT. CMA construction, sectioning, and IHC were performed in our Tissue Pathology Core of the Stephenson Cancer Center following standard protocols, as described in our earlier work (Somasundaram et al. 2019). Appropriate histology controls (H&E) and negative controls with no primary antibody were examined in parallel. Expression and localization of MMP9, NFκB-p105, RelB, pP65, p65, IκBα, pIκBα, ERK, and pERK are investigated. The slides were digitally scanned into virtual slides using a Zeiss Axio Scan Z1 slide scanner at 40× magni cation. The whole slide images were then group-analyzed for protein-speci c positivity using Aperio image analysis and quanti cation software (Aperio Technologies, Inc., Buffalo Grove, IL, USA) with the appropriate algorithms for protein speci c cellular localization. Group-wise comparisons were performed using GraphPad Prism.

Cell death (TUNEL) analysis
RT-triggered NFκB-dependent MMP9 activation and maintenance-associated induced NB cell death, if any, was quanti ed at the single-cell level based on labeling DNA strand breaks with terminal deoxynucleotidyl transferase (TUNEL Assay). All TUNEL assay procedures were performed on the customized CMA (constructed as discussed above) in the SCC-Tissue Pathology Core using a commercially available In Situ cell death detection kit (MilliporeSigma, St. Louis, MO, USA). Appropriate positive (recombinant DNAse I treatment before TUNEL labeling) and negative (without Tdt enzyme mix) controls were included. The slides were micro-digitally scanned using an Aperio Scanscope (Aperio Technologies, Inc., Buffalo Grove, IL, USA) slide scanner. TUNEL positivity was observed using NIH ImageJ, plotted with GraphPad Prism, and compared between groups using ANOVA with Tukey's post-hoc correction.
QPCR pro ling of tumor invasion/metastasis signaling and bioinformatics analysis Total RNA extraction and real-time QPCR pro ling were performed as described in our earlier studies The ΔΔct values were calculated by normalizing the gene expression levels to positive controls (β-actin, GAPDH, Hprt1), compared between groups, and the relative expression level of each gene was expressed as a fold change. Differential gene expression analysis with stringent criteria (log2 fold change) coupled with false discovery rate calculation were used identify the genes altered with RT, with and without inhibition of RT-activated MMP9. To investigate the functional relevance of modulated genes in tumor progression, we utilized ingenuity pathway analysis (IPA). Core analysis on the genes that showed signi cant differential changes with/without MMP9 after RT were selectively annotated (de ned or predicted data availability) and subjected to downstream analysis by the IPA. Core analysis was performed with criteria including a direct relationship with causal path scoring for networks and upstream regulator analysis and experimentally observed con dence level. Signi cant associations of the genes with diseases, molecular and cellular functions, and canonical pathways and networks were examined.

RT triggered and maintained MMP9 activity in surviving NB cells
To de ne whether a clinically relevant dose of RT triggers and maintains MMP9 activity, NB cells (SH-SY5Y, IMR-32, and MC-IXC) exposed to RT (2Gy) were assessed for changes in MMP9 activity at 24 through 72 h post-RT (Fig. 1A). Zymography analysis revealed a robust increase in MMP9 activity in all NB cells investigated. The induced MMP9 activity was maintained consistently at 48 and 72 h. Interestingly, a similar increase in and maintenance of MMP2 activity was observed in those cells that survived RT. Further, (i) to shed light on the triggering time line of MMP9 activity after RT; (ii) whether activation is transient or persistent; and (iii) temporal uctuations (phases) of activity, if any, we assessed the kinetics of MMP9. Mock-irradiated controls showed a stable cell line-independent and timeindependent base-line MMP activity. Compared with mock-IR, RT profoundly increased MMP9 activity in these cells (Fig. 1B). Interestingly, the heightened MMP9 activity peaked around 4-6 hours. This increased activity was persistently maintained for at least for 20 h without any time-dependent uctuations/phases (Fig. 1B). These results corroborate well with the zymography results, where we observed a signi cant maintenance of MMP9 activity even after 72 h. PMA was used as control to assess the magni cation of RT-induced/maintained MMP9 activity. While RT-induced activity is near-50% of PMA-induced activity in SH-SY5Y and identical in IMR-32, the activity is robust in MC-IXC in relation to PMA treatment (Fig. 1B). To de ne the MMP9 speci city, we examined the activity kinetics in cells pretreated with GM6001 (synthetic MMP inhibitor) prior to RT. Treatment with GM6001 completely reverted the observed RT-triggered and maintained MMP9 activity ( Fig. 2A). GM6001-affected MMP9 activity is consistent and near-baseline (mock-IR) levels in all three cell lines investigated ( Fig. 2A). On the other hand, aprotinin treatment did not revert RT-induced/maintained MMP9 activity to near-baseline levels (data not shown).  (Fig. 2B-D) and CMA-IHC in IMR-32 ( Fig. 4A-B) showed a signi cant increase in the expression of NFκB p50 and p65 and phosphorylation of p65. In addition, we observed a signi cant increase in the expression of other NFκB family (RelB and p105) proteins (Fig. 4A-B). Conversely, treating cells with GM6001 signi cantly inhibited RT-induced p50, p65, RelB, and p105 in all three NB cell lines investigated (Fig. 2B-D; Fig. 4A-B). Interestingly, we observed a signi cant inhibition in IMR-32 and SK-PN-DW cells, but not in MC-IXC cells.
On the other hand, PMA (which activates MMP9) resulted in a signi cant increase in p50 and p65 expression and p65 phosphorylation in NB cells, indicating that MMP9 activation could lead to increased expression and phosphorylation of p65. Further, to validate whether MMP9-induced expression and phosphorylation of p65 translates the transcriptional activity of NFκB, we performed an NFκB luciferase assay (Fig. 4C). Compared with mock-IR controls, RT signi cantly (P < 0.005) increased NFκB transcriptional activity in NB cells. shRNA-mediated silencing of MMP9 resulted in complete (nearbaseline) inhibition of RT-induced NFκB transcriptional activity (Fig. 4C). Conversely, forced expression of MMP9 in NB cells showed a robust (P < 0.002) increase in NFκB activity when compared with the mock-IR. This MMP9-dependent increase in transcriptional activity is signi cantly (P < 0.008) repressed in the presence of GM6001 (Fig. 4C). Together, these results clearly indicate that RT activated MMP9 prompts NFκB expression and/or phosphorylation, which lead to increased NFκB transcriptional activity.
RT-triggered NFκB orchestrated the NFκB-MMP9 feedback cycle Recognizing that (i) RT-induces NFκB transcriptional activity, (ii) MMP9 have NFκB binding sites, and (iii) RT-induced MMP9 causally orchestrates a feedback loop to NFκB (Figs. 2 and 4), it is pertinent to understand the causal therapeutic pressure trigger. The hypothesis is that therapy (RT)-triggered NFκB orchestrates the MMP9-NFkB positive feedback cycle, which leads to the maintenance of high enzyme activity and tumor evolution. Changes in RT-induced MMP9 expression and activity were examined after selectively inhibiting the nuclear translocation of p65. Cells transfected with ΔIκBα (s32A/s36A double mutant IκBα) and exposed to RT showed a signi cant decline in the levels of RT-phosphorylated p65 ( Fig. 2B-D). Consistently, we observed a signi cant decrease in RT-induced MMP9 expression in NB cells.
Likewise, MMP9 activity analysis showed complete inhibition (near-identical to mock-IR) of MMP9 activity in all NB cells transfected with ΔIκBα (Fig. 5). Conversely, forced expression of p50/p65 in neuroblastoma cells signi cantly increased MMP9 expression (Fig. 2B-D) and MMP9 enzymatic activity (Fig. 5) when compared with the mock-IR exposed cells. The expression and activity was near-identical to the levels induced by RT. Together, these results indicate that RT-induced NFκB is required for the RTinduced NFkB-MMP9-NFkB feedback cycle.

MMP9 regulated NFκB through IKK and ERK signaling
To understand the mechanism by which RT triggered NFκB-dependent MMP9 feedback NFκB activation, we explored its role in altering NFκB canonical signaling. Since IKKβ is speci c for NFκB canonical signaling (IKKα is for non-canonical signaling), we investigated its modi cation, if any, in NB cells, with and without blocking RT-induced MMP9. Compared with the mock-IR controls, we observed a signi cant increase in IKKβ localization in cells exposed to RT and in cells with forced expression of p50/p65. Muting RT-induced MMP9 with GM6001 resulted in completely reduced levels of IKKβ in both NB cell lines investigated (Fig. 2B-D). Consistent with the MMP9-dependent regulation of IKKβ, we found a substantial decrease in the IκBα phosphorylation when compared with RT (Fig. 6). Total IκBα levels were high in GM6001-treated RT-exposed cells as opposed to the cells with only RT exposure, indicating the loss of IKKβ-dependent phosphorylation. Next, assessing the effect of MMP9 on ERK activity and ERK-dependent NFκB activation, we examined the levels of ERK phosphorylation in NB cells. Compared with the mock-IR control, RT signi cantly increased the phosphorylation of ERK in MC-IXC, SK-PN-DW (Fig. 2B-D), and IMR-32 cells (Fig. 7). Forced expression of p50/p65 in these cells showed a signi cant increase in ERK phosphorylation and served as the positive control. However, muting RT-induced MMP9 with GM-6001 partly inhibited RT-induced ERK phosphorylation (Figs. 2B-D; 7A-B). Directly assessing the active ERK/phosphotransferase immune complex, we found signi cant ERK activity in cells that survived RT (Fig. 7C). More importantly, when we silenced the RT-induced MMP9, the RT-induced ERK phosphotransferase activity was completely reduced (P < 0.001). Conversely, cells treated with PMA showed a signi cant increase in ERK activity when compared with mock-IR control (Fig. 7C). Together, these results indicate that RT-induced MMP9 regulates NFκB through IKKβ in one part and through ERK on the other, at least in the NB setting.

Role of RT-triggered NFκB-dependent MMP9 in NB evolution
It has been reported that targeting MMP9 could prompt NFκB regulation-dependent apoptosis. To determine whether RT-triggered NFκB-MMP9-NFκB PFC-dependent activation of MMP9 contributes to tumor evolution, we examined the apoptotic alterations in NB cells exposed to RT with/without MMP9 blocking. Compared with the mock-IR controls, RT signi cantly induced apoptosis (Fig. 8A-B). However, when RT-induced MMP9 was inhibited, we observed a robust increase in NB cell death (vs. RT), indicating that MMP9-dependent NFκB activation regulates RT-induced cell survival.
Activated MMP9 heavily contributes to tumor invasion and metastasis through matrix degradation. However, its role in tumor progression beyond its enzymatic reaction is thus far unrealized. We examined the effect of RT-induced NFκB-MMP9-NFκB PFC-dependent MMP9 in facilitating tumor invasion and dissemination beyond ECM degradation. Examining 93 tumor invasion and metastasis signaling genes, we observed a heightened metastatic state of NB cells that survive RT, with 31 genes showing signi cant upregulation (Fig. 9A). Nineteen of 31 RT-upregulated genes were downregulated when cells were treated with GM6001 and exposed to RT ( Fig. 9B and C). RT induced MMP9 in NB cells and this RT-induced MMP9 was downregulated in the presence of GM6001 (Fig. 9C). Interestingly, seven (ID1, ILIB, KISS1, MCAM, MMP1, MMP9, VEGFA) of 19 genes regulated in the presence of GM6001 are known to contain NFκB response elements. Ingenuity pathway core analysis revealed that this small subset of tightly interregulated molecular targets showed in uential participation in many canonical signaling pathways and demonstrated de ned roles in tumor invasion and metastasis. IPA-data mining considering only relationships where con dence = experimentally observed; these molecules exhibited their role in at least 245 different canonical pathways exerting nearly 150 biological functions. Interestingly, in light of tumor progression and dissemination, we observed a signi cant association (-log(p-value)) of these molecules in key pathways including the tumor microenvironment, role of tissue factor in cancer, HIFα signaling, HOTAIR regulatory pathway, regulation of the EMT by growth factors pathway, regulation of EMT pathway, stat3 pathway, epithelial adherens junction signaling, molecular mechanisms of cancer, NFκB signaling, IGF signaling PPAR signaling, RAR activation, mTOR signaling, and cancer drug resistance (Additional le 1: Figure S1; Table S1). An all-encompassing overview of these molecules, including information on their symbol, name, subcellular location, protein functions, binding, regulating, regulated by, targeted by miRNA, role in cell, molecular function, biological process, cellular component, disease, and role in tumor progression and metastasis, are provided in Additional le 2: Table S2.

Discussion
Understanding the acquired molecular rearrangements in therapy-surviving cells is critical in developing improved therapeutic measurements for the progressive NB that de es current clinical therapy. We provided rst evidence that clinically relevant fractionated RT activates NFκB in surviving NB cells (Aravindan et al. 2008;Madhusoodhanan et al. 2009). Further, our studies unveiled NFκB-dependent intraand inter-cellular mechanisms that coordinate survival advantage, clonal expansion, and tumor progression (Aravindan et  signaling involved in tumor cells has been extensively documented over the years, and is beyond the scope of the current study. However, our results demonstrated that RT-induced NFκB is required for the activation and maintenance of MMP9 in this setting. As discussed earlier, MMP-9 is transcriptionally regulated by multiple TFs, including AP-1, ETS, SP1, and NFκB (Benbow, Brinckerhoff 1997;. It is also evident that AP1 or other TFs-driven MMP9 transcription require substantiation of NFκB recognition (Benbow, Brinckerhoff 1997; (Lauricella-Lefebvre et al. 1993). Conversely, our present data demonstrate that inhibiting RT-induced NFκB transcriptional activity by blocking IκB phosphorylation completely prevents MMP9 expression and consequent activity. A forced increase of p50/p65 in these cells mimicked the RT-induced effect on MMP9 expression. In parallel, the use of PMA, which is known to upregulate MMP9 through the PKCα-NFκB cascade, serves as the best positive control for the study (Shin et al. 2007). These outcomes indicate that RT-induced NFκB is both required and adequate to activate MMP9 in NB cells.
More importantly, the results presented here revealed that RT-triggered and -maintained MMP9 enzyme activity facilitates the onset of second signaling feedback to NFκB. We showed that blocking RT-triggered NFκB-dependent MMP9 resulted in the regulation of NFκB family proteins and reduced phosphorylation of p65. Rao and colleagues indicated that targeting MMP9 inhibits RT-induced NFκB and consequently affected apoptosis in a breast cancer setting . For the rst time, here we showed that MMP9 activity corroborates with the NFκB transcriptional activity in NB cells and beyond. Our outcomes from the NFκB luciferase assay showed a signi cant decrease in RT-induced NFκB transcriptional activity in both MMP9-muted cells and in cells treated with GM6001. Substantiating our observations, studies have shown that NFκB activation was prevented by selective MMP9 inhibition Dwir et al. 2020). More importantly, our results indicated that forced expression of MMP9 in NB cells perpetrates a robust NFκB transcriptional activity. To our knowledge, for the rst time, our results portray that activated MMP9 communicates with the NFκB signaling pathway and leads to increased NFκB transcriptional activity. Earlier, we showed that blocking RT-induced second-signaling feedback- affects ERK phosphorylation in medulloblastoma (Bhoopathi et al. 2008) and meningioma (Gogineni et al. 2009). Here we provide rst evidence that MMP9 activates ERK in NB cells that survive RT. Our results showed high levels of ERK phosphotransferase activity, which is completely lost with selective inhibition of MMP9. Conversely, RT-mimicking levels of ERK phosphotransferase activity with PMA, a NFκBdependent MMP9 inducer, bolster our claims.
Because ERK activation actuates NFκB signaling by regulating IκK and thereby increasing IκBα phosphorylation (Chen et al. 2016), we next investigated the effect of RT-activated MMP9 on IKK signaling and IκB phosphorylation. The IKK complex is the pivotal regulator of inducible NFκB signaling and includes a regulatory sub-unit NEMO and two kinases, IKKα and IKKβ. While NEMO and IKKβ contribute to the canonical NFκB signaling pathway, IKKα facilitates non-canonical mechanisms (Solt, May 2008). To that end, our results showed a signi cant induction of IKKβ in irradiated cells, while this induction was completely alleviated when RT-induced MMP9 activity was inhibited. Substantiating the MMP9-dependency of IKKβ alterations, our results con rmed a signi cant MMP9-IκB phosphorylation association in NB cells. Together, these outcomes clearly portray the unforeseen role of MMP9 in the second signaling feedback to NFκB activation after RT and suggest that MMP9 exploits ERK-and/or IKKβ-dependent IκBα phosphorylation for the cause-effect (Fig. 10). Next, looking into its upstream regulatory capabilities, our custom archived QPCR pro ling identi ed at least 19 (of 32 RT-activated) tumor evolution-related oncotargets that are speci cally regulated by MMP9. Since individually discussing the oncogenic role of each of these 19 MMP9-regulated molecules will be elaborate and is beyond the scope of this study, we examined their integrated role in tumor cell signaling and tumor evolution function. Bioinformatic analysis with IPA clearly portrayed the high signi cance of these molecules in cancer evolution and drug resistance, with direct implications in 245 canonical signaling and 150 biological functions. Critically, these molecules are directly involved in tumor microenvironment, drug resistance, EMT, survival, clonal selection, progression, and metastasis, directly affecting key pathways such as tissue factors, HIFα, HOTAIR, stat3, IGF, RAR, mTOR, and NFκB. To our knowledge, this is the rst report identifying a group of tumor progression targets that could be regulated by MMP9 in NB cells under therapeutic pressure, in this case RT. More importantly, seven (ID1, ILIB, KISS1, MCAM, MMP1, MMP9, VEGFA) of the regulated genes are known to possess NFκB binding sites. This a rms that MMP9-dependent second-signaling feedback to NFκB is critical for the transcription of these molecules. Together, these diverse aspects of MMP9 function work in concert to effect the signaling dysregulation in NB cells that survive therapy, and contribute to the resistance, survival, growth, and spread of NB.
The authors acknowledge the limitations of this study, including the requirement for a preclinical animal model coupled with the induction (or maintenance) phase combination chemotherapy appropriate in conjunction with RT for any meaningful clinical translation. However, this proof-of-concept in vitro study is warranted, and provides the rst evidence on existence of the RT-triggered NFκB-MMP9-NFκB feedback cycle; the PFC-maintained MMP9 activity; and the molecular signaling involved. The authors also acknowledge that the use of single-dose RT is a limitation and a clinically relevant fractionated dose regimen is required. As discussed above, unveiling the existence of the signal transduction will now allow us to translate this to preclinical spontaneous NB in vivo systems with mimicking clinical therapy regimens and to measure the in uence in real time in light of tumor progression and spreading.
In conclusion, in this study, we report that a clinical dose of RT signi cantly increases MMP9 activity and the induced activity is persistently maintained across human NB cell lines investigated. Further, we showed that RT triggers NFκB phosphorylation, nuclear localization, and transcriptional activity, and this RT-induced NFκB is required and adequate for the maintenance of MMP9. RT-triggered NFκB-dependent MMP9 actuates a second-signaling feedback to NFκB signaling, thereby facilitating a NFκB-MMP9-NFκB PFC. Our results demonstrated that MMP9-NFκB second signaling feedback is mediated by MMP9dependent activation of IKKβ and ERK activity. Beyond its regular tumor invasion and metastasis function, PFC-dependent MMP9 activation lessens RT-induced apoptosis through the activation of NFκB signaling. More importantly, this study identi ed that NFκB-MMP9-NFκB-dependent MMP9 regulates 19 critical molecular determinants that play a pivotal role in tumor evolution. Interestingly, seven of these molecules have binding sites for NFκB, indicating that MMP9 regulates these molecules by activating NFκB. Collectively, these data suggest that acquired maintenance of MMP9 after therapeutic pressure could contribute heavily for tumor evolution, and selective inhibition of activated MMP9 maintenance could serve as a promising therapeutic strategy for progressive NB that de es current clinical therapy.   (A) Line graphs obtained from uorogenic substrate speci c activity assay showing MMP9 activity kinetics in the conditioned medium of NB cells (SH-SY5Y, IMR-32 and MC-IXC) exposed to mock-IR, RT or, treated with GM-6001 (a potent cell permeable MMP9 inhibitor) and exposed to RT. Kinetics of the MMP9 activity was quanti ed continuously for every each 20minutes constantly for 20h. (B) Representative immunoblots showing the expression of IKKβ ERK1/2, pERK1/2, NFkB-p50, NFkB-p65, NFkB-p65 phosphorylated, and MMP9 in MC-IXC and SK-PN-DW NB cells exposed to mock-RT, RT (2Gy), PMA, after ectopic expression of p50/p65, treated with GM6001 and exposed to RT, treated with aprotinin and exposed to RT or, transfected with IkB and exposed to RT. Blots were stripped and reblotted with mouse  MMP9 expression in IMR-32 cells exposed to mock-IR, RT (2Gy), PMA, treated with GM6001 and exposed to RT or, treated with aprotinin and exposed to RT. No antibody IgG controls (blank) and H&E staining are included for each condition. Images are at 10x and the inserts are at 40x magni cation. (B) Histograms from the aperio spectrum image analysis quanti cation of the digitally scanned virtual images showing MMP9 expression in IMR-32 cells exposed to mock-IR or RT with/without MMP9 inhibition. Expression   Line graphs obtained from uorogenic substrate speci c activity assay showing MMP9 activity kinetics in the conditioned medium of NB cells (SH-SY5Y, IMR-32 and MC-IXC) exposed to mock-IR, RT, NFkB inhibition with s32A/s36A double mutant IkBα (ΔIkBα) and exposed to RT, or after NFkB (NFkB-p50/p65) overexpression. Kinetics of the MMP9 activity was quanti ed continuously for every each 20minutes constantly for 20h.  and phosphorylation in IMR-32 cells exposed to mock-IR or RT with/without MMP9 inhibition. Expression (mean and SEM) is graphed as fold change over mock-IR control. Groupwise comparisons were made with two-way ANOVA with Tukey's post-hoc comparison. A P value of 0.05 was considered signi cant. (C) MMP9-dependent ERK phosphotransferase activity. Representative immunoblot from MAP Kinase/ERK Kinase assay showing alterations in ERK phosphotransferase activity in SH-SY5Y cells exposed to mock-IR, RT, PMA, NF B inhibition with s32A/s36A double mutant I Bα (ΔI Bα) and exposed to RT, treated with GM6001 and exposed to RT, treated with aprotinin and exposed to RT, or after NF B (NF B-p50/p65) overexpression. Histograms obtained from 1D gel analysis of immunoblot showing altered levels of ERK phosphotransferase activity. Groupwise comparisons were made with ANOVA with Tukey's post-hoc comparison. A P value of ≤ 0.05 was considered signi cant.  Volcano plots from custom archived QPCR pro ling of 93 genes pertaining to tumor invasion and metastasis signaling showing genes modulated (up/down) genes in SH-SY5Y cells (A) exposed to RT (2Gy) and (B) treated with GM6001 and exposed to RT. The ΔΔct values were calculated by normalizing the gene expression levels to positive controls (β-actin, GAPDH, Hprt1), compared between groups, and the relative expression level of each gene was expressed as a fold change. Differential gene expression analysis with stringent criteria (log2 fold change) coupled with false discovery rate calculation were used identify the genes altered with RT with/without inhibition of RT-activated MMP9.  PFC signaling leads to the maintenance of activated MMP9, which leads to the NF B-dependent survival advantage and MMP9-dependent tumor growth invasion and metastasis.