Exploring Nitric Oxide (NO)-Releasing Celecoxib Derivatives as Modulators of Radioresponse in Pheochromocytoma Cells

COX-2 can be considered as a clinically relevant molecular target for adjuvant, in particular radiosensitizing treatments. In this regard, using selective COX-2 inhibitors, e.g., in combination with radiotherapy or endoradiotherapy, represents an interesting treatment option. Based on our own findings that nitric oxide (NO)-releasing and celecoxib-derived COX-2 inhibitors (COXIBs) showed promising radiosensitizing effects in vitro, we herein present the development of a series of eight novel NO-COXIBs differing in the peripheral substitution pattern and their chemical and in vitro characterization. COX-1 and COX-2 inhibition potency was found to be comparable to the lead NO-COXIBs, and NO-releasing properties were demonstrated to be mainly influenced by the substituent in 4-position of the pyrazole (Cl vs. H). Introduction of the N-propionamide at the sulfamoyl residue as a potential prodrug strategy lowered lipophilicity markedly and abolished COX inhibition while NO-releasing properties were not markedly influenced. NO-COXIBs were tested in vitro for a combination with single-dose external X-ray irradiation as well as [177Lu]LuCl3 treatment in HIF2α-positive mouse pheochromocytoma (MPC-HIF2a) tumor spheroids. When applied directly before X-ray irradiation or 177Lu treatment, NO-COXIBs showed radioprotective effects, as did celecoxib, which was used as a control. Radiosensitizing effects were observed when applied shortly after X-ray irradiation. Overall, the NO-COXIBs were found to be more radioprotective compared with celecoxib, which does not warrant further preclinical studies with the NO-COXIBs for the treatment of pheochromocytoma. However, evaluation as radioprotective agents for healthy tissues could be considered for the NO-COXIBs developed here, especially when used directly before irradiation.


Introduction
Adrenal pheochromocytomas and extra adrenal paragangliomas (PPGLs) are rare catecholamine-producing tumors of chromaffin cell origin that form solid organ metastases mainly in lymph nodes, bones, lungs and the liver [1,2]. Recommended treatment options for metastatic PPGLs include, e.g., surgery, different combinations of chemotherapy, endoradiotherapy using [ 131 I]metaiodobenzylguanidine and external radiation therapy, but they are often considered as palliative. Somatostatin type 2 receptor (SSTR2)-targeting endoradiotherapy using particularly [ 177 Lu]Lu-DOTA-(Tyr 3 )octreotate ([ 177 Lu]Lu-DOTA-TATE) offers potential for treating metastatic PPGLs [3,4]. Because of response rates usually in the range of 30% and 60% [5], this approach is likely to benefit from combination with adjuvant, for example, radiosensitization therapy [6]. In this regard, we previously demonstrated The free radical and endogenous signaling molecule nitric oxide (NO • ) is produced by three NO synthases: the endothelial (eNOS), the neuronal (nNOS), and the inducible NO synthase (iNOS) [28]. Under physiological conditions, NO • is involved in the stimulation of mucus secretion from gastric cells and the regulation of vascular tone by acting as a vascular smooth muscle relaxant [29][30][31][32]. Some pathologies have been correlated with high levels of NO • produced by iNOS such as septic shock or inflammatory events, whereas others have been found to be connected to reduced levels of NO • such as endothelial dysfunction. For treatment of the latter, clinically approved drugs exist such as glyceryl nitrate or isosorbide dinitrate. With regard to the treatment using COX inhibitors, the concomitant release of NO • is hypothesized to counteract both gastrointestinal and cardiovascular side effects [31,[33][34][35]. Starting from naproxcinod [36,37], a variety of novel compounds now known as COX-inhibiting nitric oxide donors (CINODs) or nitric oxide-releasing COXIBs (NO-COXIBs) have been developed [31]. As a common principle for this class, known NSAIDs or COXIBs were functionalized directly or indirectly via a linker structure to an nitric oxide releasing moiety, e.g., nitro ester [38][39][40][41], diazeniumdiolate "NONO" [23], furoxane [40], or sulfohydroxame [41,42] functionalities ( Figure 1). NO • shares many actions of cytoprotective prostaglandins, as it stimulates mucus secretion from gastric cells. In this context, NO • has been proven to be protective against NSAID-induced gastropathy [35,43].  [39,44,45]. NO • has been shown to be involved in the cellular response to ionizing radiation leading to the activation of a number of stress proteins such as JNK or MAPK. Upregulation of iNOS is observed after irradiation in a variety of tumor cells and tissues, and it is regarded as one source of endogenous NO • in this context. While low levels of NO • are discussed to be cytoprotective, for higher levels of NO • or reactive nitrogen species, cell damaging effects are discussed. [46] NO • acts as reaction partner for the formation of other  [39,44,45].
Molecules 2022, 27, 6587 3 of 25 NO • has been shown to be involved in the cellular response to ionizing radiation leading to the activation of a number of stress proteins such as JNK or MAPK. Upregulation of iNOS is observed after irradiation in a variety of tumor cells and tissues, and it is regarded as one source of endogenous NO • in this context. While low levels of NO • are discussed to be cytoprotective, for higher levels of NO • or reactive nitrogen species, cell damaging effects are discussed [46]. NO • acts as reaction partner for the formation of other highly reactive oxygen/nitrogen species, such as peroxynitrite (ONOO − ) [47], thereby, e.g., contributing to the fixation of radiation-induced DNA lesions even in highly radioresistant hypoxic tumor areas [48]. From our point of view, the latter property combined with COX inhibition would build an interesting platform for the development of adjuvants for radiation therapy. In this context, we recently developed two novel NO • -releasing celecoxib derivatives (5a and 5e, Scheme 1) [49] and evaluated their radiosensitizing properties in comparison to a COX-2 selective celecoxib analog (3a, Scheme 1) in vitro [50]. In "COX-2-positive" (A2058) and "COX-2-negative" (Mel-Juso) human melanoma cell lines showing upregulated COX-2 expression under experimental hypoxic conditions and after irradiation, NO-COXIBs 5a and 5e showed more potent radiosensitizing efficacy compared to the COXIB 3a under normoxic and hypoxic conditions at low applied inhibitor concentrations of 10 µM. For example, by administration of 5a, the required radiation dose for 10% survival could be reduced from 6.6 (DMSO control) to 5.2 Gy for A2058 cells and from 4.2 (DMSO control) to 3.2 Gy for Mel-Juso cells [49].
To extend the knowledge about the structure-activity relationship on this substance class, we synthesized eight novel derivatives based on the NO-COXIBs 5a bearing either a 4-tolyl-or 4-methoxyphenyl ring at position 5 of the pyrazole core as well as a methylsulfonyl, an aminosulfonyl or an N-propionylsulfonamide group ( Figure 2). COX inhibition and NO • release was measured, and radiosensitizing effects were investigated in vitro. COX-1 and COX-2 inhibition potency was found to be comparable to the lead NO-COXIBs, and NO-releasing properties were found to be mainly influenced by the substituent in the 4-position of the pyrazole. NO-COXIBs showed similar but throughout more radioprotective effects compared to celecoxib.

Synthesis of NO • -Releasing COXIBs
Nitroesters 5a-d were synthesized following the synthetic route previously described by us with minor modifications to increase the yield (Scheme 1). [49] Starting from the respective 4-methyl-or 4-methoxy-substituted acetophenone derivative, the 2,4-dioxobutyrate ester derivatives 1a-b were synthesized by condensation with dimethyl oxalate after deprotonation with sodium methoxide. The pyrazole based heterocycles 2a-d were

Synthesis of NO • -Releasing COXIBs
Nitroesters 5a-d were synthesized following the synthetic route previously described by us with minor modifications to increase the yield (Scheme 1) [49]. Starting from the respective 4-methyl-or 4-methoxy-substituted acetophenone derivative, the 2,4-dioxobutyrate ester derivatives 1a-b were synthesized by condensation with dimethyl oxalate after deprotonation with sodium methoxide. The pyrazole based heterocycles 2a-d were generated by reaction of 1a-b with the 4-sulfonyl-substituted phenyl hydrazines as commonly applied for the synthesis of celecoxib derivatives. Subsequent reduction of the methyl ester functionality using lithium aluminum hydride in tetrahydrofuran at room temperature furnished the alcohols 3a-d. Conversion to the 3-chloromethyl-4-chloropyrazoles 4a-d was achieved by reaction with neat thionyl chloride. For this reaction, we prolonged the reaction time from 1 h to 8 h compared to our previously published method [49] because for 4a-d after 1 h the mono-chlorinated side product was still observed in considerable amounts. Of note, it was found that the reaction of 3a and 3b with thionyl chloride in dichloroethane at 80 • C favored the conversion to the dichloro-substituted products, while stirring at room temperature in dichloroethane exclusively resulted in the formation of 3-chloromethylpyrazoles 4e-f. Conversion to the nitroester derivatives 5a-d was achieved by application of silver nitrate in acetonitrile (for discussion of NMR spectra, see Appendix A).

Synthesis of NO • -Releasing COXIBs
Nitroesters 5a-d were synthesized following the synthetic route previously described by us with minor modifications to increase the yield (Scheme 1). [49] Starting from the respective 4-methyl-or 4-methoxy-substituted acetophenone derivative, the 2,4-dioxobutyrate ester derivatives 1a-b were synthesized by condensation with dimethyl oxalate after deprotonation with sodium methoxide. With the aim to synthesize more soluble derivatives of the envisaged NO-COXIBs, the sulfonamide derivatives 4a, 4b, 4e, and 4f were converted to the respective N-propionamides followed by conversion of the chloromethyl group into the nitroesters 6a-d (Scheme 2). For the N-propionamide synthesis, sulfuric acid on silica gel was utilized as "dry immobilized" acid as previously described by Massah et al. [51]. This furnished the propionamides as crude products, which were directly reacted with silver nitrate to give the final products 6a-d.
With the aim to synthesize more soluble derivatives of the envisaged NO-COXIBs, the sulfonamide derivatives 4a, 4b, 4e, and 4f were converted to the respective N-propionamides followed by conversion of the chloromethyl group into the nitroesters 6a-d (Scheme 2). For the N-propionamide synthesis, sulfuric acid on silica gel was utilized as "dry immobilized" acid as previously described by Massah et al. [51] This furnished the propionamides as crude products, which were directly reacted with silver nitrate to give the final products 6a-d. Scheme

COX Inhibition
COX inhibition (Table 1) was evaluated in vitro using purified ovine COX-1 and recombinant human COX-2 enzyme in a commercially available "COX Fluorescent Inhibitor Screening Assay KIT" (item no. 700100, Cayman Chemical, Ann Arbor, MI, USA). Celecoxib served as a reference and inhibition data concerning 4a, 4e, 5a and 5e were previously reported by us and is herein discussed for comparison.

COX Inhibition
COX inhibition (Table 1) was evaluated in vitro using purified ovine COX-1 and recombinant human COX-2 enzyme in a commercially available "COX Fluorescent Inhibitor Screening Assay KIT" (item no. 700100, Cayman Chemical, Ann Arbor, MI, USA). Celecoxib served as a reference and inhibition data concerning 4a, 4e, 5a and 5e were previously reported by us and is herein discussed for comparison. The chloromethyl-substituted celecoxib-derivatives 4a-f, except of 4b, were found to be selective COX-2 inhibitors with IC 50 (COX-2) ranging between 0.22 and 1.27 µM, while no considerable COX-1 inhibitory potency was found (IC 50 > 100 µM). Only 4b showed a COX-1 inhibitory potency (IC 50 = 0.29 µM) in this subset of compounds and hence a rather non-selective COX inhibition profile. The conversion of the chloromethyl group to the nitroester group of 5a-f was well tolerated and did not markedly influence the inhibitory profile. Compounds 5a-f showed half-maximal inhibition of COX-2 in the range of 0.28-2.12 µM and, in a direct comparison to their respective chloro-analogs, only 2-3 times lower COX-2 inhibitory potency. Only 5b showed considerable COX-1 inhibition potency while the others did not inhibit COX-1. Both 4b and 5b contain an methoxy-and sulfonamide-group at their respective phenyl rings as well as the chloro-substituent at the pyrazole core so that both compounds are electron rich derivatives of this class. In this regard, antioxidative properties that principally disturb the fluorescence-based assay by giving apparent inhibition might be considered in this regard but must be further evaluated.
The conversion of sulfonamide-residues to their respective N-propionamide analogs 6a-d abolished COX inhibitory potency against COX-1 and COX-2. This finding is in line with the general importance of amino-and methyl sulfonyl group for COX-2 selective binding [52][53][54] as well as other literature reports for the synthesis of N-propionamide based prodrugs [55][56][57] of cyclooxygenase-inhibitors.
In summary, COX inhibitory potency was found to be consistent with the previously determined inhibition pattern [49], but unfortunately no markedly improved COX-2 inhibition was achieved by the herein performed chemical modifications.

LogD pH7.4, HPLC
The lipophilicity of all synthesized compounds was determined as logD 7.4, HPLC value by an HPLC method originally described by Donovan and Pescatore [58]. Compounds 4a-f and 5a-f were found to be characterized by logD 7.4, HPLC values in the range of 3.30 to 4.04 while N-propionamides 6a-d exerted markedly lower lipophilicity in the range between 1.55 and 2.27. In a direct comparison, ∆logD 7.4, HPLC between the sulfonamides and their N-propionamides were found to be between 1.5 and 1.8, which is comparable to our previous study focusing on (dihydro)pyrrolo[3,2,1-hi]indole-based COX inhibitors [55].

NO • -Release
The nitric oxide releasing properties of the nitroester-derivatives 5a-d, 5f, and 6a-d were evaluated using a fluorometric Griess assay in the presence and absence of cysteine to determine thiol-dependent and spontaneous release of nitric oxide ( Figure 3).  . Spontaneous and thiol-dependent NO • release from NO-COXIBs; measured using fluorometric Griess assay (n = 2); relative release indicates the percentage of soluble nitrogen species (nitrate + nitrite) detected in the incubation buffer after 24 hours compared to the initial molar amount of compounds; NO • release of NONOate (positive control) exceeded the maximum detection limit; data presented as mean ± range.
In both blank (DMSO) and negative control (4f), release of NO • was not detectable, while the positive control NONOate showed nearly quantitative NO • release in the spontaneous experimental setup. In the thiol-dependent experimental setup, NO • release of NONOate exceeded the assay's detection limit. For all nitroester derivatives, the thiol dependent NO • -release was found to be higher than the spontaneous NO • release, which is . Spontaneous and thiol-dependent NO • release from NO-COXIBs; measured using fluorometric Griess assay (n = 2); relative release indicates the percentage of soluble nitrogen species (nitrate + nitrite) detected in the incubation buffer after 24 h compared to the initial molar amount of compounds; NO • release of NONOate (positive control) exceeded the maximum detection limit; data presented as mean ± range. In both blank (DMSO) and negative control (4f), release of NO • was not detectable, while the positive control NONOate showed nearly quantitative NO • release in the spontaneous experimental setup. In the thiol-dependent experimental setup, NO • release of NONOate exceeded the assay's detection limit. For all nitroester derivatives, the thiol dependent NO •release was found to be higher than the spontaneous NO • release, which is consistent with a favored thiol-dependent reaction. The chloro-pyrazoles 5b and 5c as well as its derived N-propionamide 6b furnished no detectable spontaneous NO • release after 24 h while thioldependent release was found to be at 20%, 17% and 56%, respectively. In comparison, the structurally more heterogenous group of 5a, 5d, 5f, and 6a showed spontaneous NO • release in the range of 10-22%, which was approximately half as high compared to the respective thiol-dependent release ranging from 29-53%. Finally, the N-propionamide based pyrazoles 6c and 6d sharing a pyrazole with unsubstituted 4-position showed highest spontaneous NO • release of 26% and 40%, respectively, which reached approximately 2/3 compared to the thiol-dependent release of 40% and 65%, respectively.
The time-dependent degradation behavior of compounds 4f, 5a-f, and 6a-d as well as celecoxib was followed by UPLC analysis over a time course of 8 days ( Figure 4). For that, compounds were incubated at 37 • C in the presence or absence of cysteine in the assay buffer used for the Griess assay. Samples were withdrawn at distinct time points (1, 5, 23, 47, 70, and 191 h), diluted into a mixture of acetonitrile/water, and analyzed.
Furthermore, HPLC-HRMS was performed after 24 h to obtain more detailed information about the identity of the degradation products. For all investigated compounds, the spontaneous NO • release conditions resulted in the conversion to a more polar product corresponding to the respective alcohol formally resulting from the hydrolysis of the nitroester bond. In comparison, the thiol-dependent release condition resulted in the formation of this and another even more polar product that could be identified to be the S-cysteinyl,S'-pyrazolylmethyl thioether.
In the presence of cysteine, only 6d and the chloromethyl-derivative 4f were quantitatively consumed after 70 h while for the other nitroester derivatives still between 6-25% were found to be intact. In the absence of cysteine, only 6d was consumed after that time while the amount of intact substance varied to a higher extend, i.e., between 16-85%. For the most stable compound 5c, still 57% of the compound was found to be intact after 8 days (192 h). Following the time-dependent degradation by UPLC revealed, especially for the substitution pattern in the 4-position of the pyrazole, a pronounced susceptibility to undergo either spontaneous or thiol dependent transformation. While chloro-substituted pyrazoles preferentially reacted to the cysteinyl derivatives and were converted to a lesser extent spontaneously to the hydroxy derivatives, the ratio clearly turned when the 4position of the pyrazole was unsubstituted. In these cases, spontaneous conversion to the hydroxy derivatives was dominant, and the formation of the cysteinyl-substituted product was diminished. Of note, because the spontaneous conversion can be regarded as the inherent instability of the compounds to undergo hydrolysis of the nitroester group, a stabilizing effect can be deduced from the chloro-substitution at the pyrazole.

Effects on Sensitivity of Tumor Spheroids to External Beam (X-Ray) Irradiation and Lu-177
During acute response to irradiation, inhibition of COX-2 has been reported to suppress DNA repair mechanisms, to inhibit metastatic behavior and to induce apoptosis [13]. Herein, the radiosensitizing properties of NO-COXIBs were tested in vitro on genetically modified HIF2α-expressing mouse pheochromocytoma (MPC-HIF2a) tumor spheroids. In these pseudohypoxic MPC-HIF2α cells, basal expression of the Ptgs2 gene encoding COX-2 was further stimulated under hypoxic conditions compared to the "empty vector" control cell line MPC-EV (Supplementary Material Figure S67). In MPC-HIF2α spheroids, stabilization of HIF2α in regions of intrinsic hypoxia has been reported to promote a radiation resistant phenotype via yet unknown downstream mechanisms [59]. In various cancer cells, direct upregulation of COX-2 through HIF2α has been reported [60,61]. hydroxy derivatives was dominant, and the formation of the cysteinyl-substituted product was diminished. Of note, because the spontaneous conversion can be regarded as the inherent instability of the compounds to undergo hydrolysis of the nitroester group, a stabilizing effect can be deduced from the chloro-substitution at the pyrazole.

Effects on Sensitivity of Tumor Spheroids to External Beam (X-Ray) Irradiation and Lu-177
During acute response to irradiation, inhibition of COX-2 has been reported to suppress DNA repair mechanisms, to inhibit metastatic behavior and to induce apoptosis. [13] Herein, the radiosensitizing properties of NO-COXIBs were tested in vitro on genetically modified HIF2α-expressing mouse pheochromocytoma (MPC-HIF2a) tumor spheroids. In these pseudohypoxic MPC-HIF2α cells, basal expression of the Ptgs2 gene encoding COX-2 was further stimulated under hypoxic conditions compared to the "empty vector" control cell line MPC-EV (supplementary material Figure S67). In MPC-HIF2α spheroids, . Spontaneous and thiol-dependent conversion of NO-COXIBs followed by HPLC-UV (λ = 254 nm); Y-axis shows relative fraction of intact compound and the respective conversion products, i.e., the hydroxy derivative (CH 2 OH) and the cysteine adduct (CH 2 Cys) detected in the incubation buffer at the indicated time points in the presence or absence of cysteine.
As a first pilot study, growth responses of MPC-HIF2α spheroids were investigated after two different radiation treatments such as single-dose external beam irradiation (Xray) and incubation with [ 177 Lu]LuCl 3 , respectively. NO-COXIBs were tested at 10 µM for additional effects on radiation treatments against a vehicle-treated control (w/o) and celecoxib (Ce). All compounds were applied at a non-toxic concentration that showed no growth attenuation on non-irradiated spheroids (Supplementary Material Table S2).
In combination with X-ray treatment at a single dose of 15 Gy, NO-COXIBs showed both radioprotective or radiosensitizing effects on tumor spheroids, depending on the time of administration before or after irradiation ( Figure 5A and Supplementary Material Figure S68). The log 2 -fold change in spheroid diameter is given as a standardized parameter for spheroid growth, which is positive for growing, near zero for stagnating and negative for shrinking spheroids. Furthermore, regrowth probabilities are given as a parameter to compare how many of the overall studied spheroids in one group recovered within 14 days. Radioprotective effects occurred when compounds were added 2 h before irradiation. In this setup, all compounds accelerated the increase in spheroid size during the regrowth phase compared to controls ( Figure 5A, upper panel). Regrowth probabilities of 67-100% under NO-COXIB treatment remained however similar to controls (83%). Radiosensitizing effects occurred when compounds were added 2 h post irradiation. In this setup, all compounds decelerated the increase in spheroid size during the regrowth phase compared to controls. Celecoxib and 6b most effectively reduced the regrowth probability (0%) followed by 5a,c,d,f (17%) and 6a,c,d (33-40%) compared to controls (83%). Compounds showed no additional effects on spheroid regrowth when added 24 h after irradiation. Overall, the tested NO-COXIBs were found to be more radioprotective than celecoxib in MPC-HIF2α spheroids in response to external beam irradiation, and no significant correlation to COX inhibition potency could be delineated. performed to unravel the potential application of COXIBs [72] for radioprotection, e.g., to mitigate radiation-induced fibrosis in mice [73] or to reduce the severity of oral mucositis in human [74], as well as for radiosensitization, e.g., during the treatment in FaDu squamous cell carcinoma in mice [75]. Altogether, the NO-COXIBs introduced in this study showed the same tendency as celecoxib, exerting both radioprotective and radiosensitizing effects on MPC-HIF2α spheroids when applied shortly before or after irradiation. However, the effects of NO-COXIBs did not exceed those of celecoxib in terms of radiosensitization. This contrasts the bifunc- Radionuclide therapy is a fast-evolving treatment modality gaining more importance in treating cancer [62][63][64]. For treatment of pheochromocytomas and paragangliomas, in particular, somatostatin type 2 receptor-targeted radionuclide therapy with [ 177 Lu]Lu-(Tyr 3 )octreotate is currently investigated in clinical controlled trials (e.g., NCT04029428, NCT03923257, NCT03206060). To assess the additional effects of NO-COXIBs on the outcome of Lu-177 therapy, a simplified in vitro setup was used here, where MPC-HIF2α tumor spheroids were treated with 10 µM of the compounds and simultaneously 0.25 MBq of [ 177 Lu]LuCl 3 for six days. In this setup, most of the NO-COXIBs showed a trend toward radioprotective effects ( Figure 5B, Supplementary Material Figure S67). In combination with Lu-177 treatment, compounds 5a,c,d,f, and 6a,c,d prevented spheroids from shrinking in size, whereas 6c and celecoxib had almost no additional effect on growth response compared to controls. Consistent with this, compounds 6b and 6d most effectively increased the regrowth probability of spheroids (100%), followed by 5a,c,d,f (60-80%), whereas 6c and celecoxib showed no considerable effects (40%) compared to controls (50%). Overall, the tested NO-COXIBs, except of 6c, were found to be more radioprotective than celecoxib in MPC-HIF2a spheroids in response to [ 177 Lu]LuCl 3 treatment, and no significant correlation to COX inhibition potency could be delineated.
In principle, these results are in line with other reports demonstrating that application of selective COX-2 inhibitors can have either radioprotective or radiosensitizing effects on irradiated cells depending on a myriad of factors such as cell type, basal and inducible COX-2 levels, modality of irradiation and, the timing of complementary COXIB treatment [13,65,66]. Celecoxib as the main reference compound used in this study was for example shown to enhance radiosensitivity of normoxic and hypoxic glioblastoma cells [67], nasopharyngeal carcinoma cells [68], lung cancer cells (NSCLC) [69], human colorectal carcinoma cells [70] as well as to be radioprotective for human lymphocytes [71] or melanoma (MelJuso) cells [50]. Accordingly, preclinical and clinical studies were performed to unravel the potential application of COXIBs [72] for radioprotection, e.g., to mitigate radiation-induced fibrosis in mice [73] or to reduce the severity of oral mucositis in human [74], as well as for radiosensitization, e.g., during the treatment in FaDu squamous cell carcinoma in mice [75].
Altogether, the NO-COXIBs introduced in this study showed the same tendency as celecoxib, exerting both radioprotective and radiosensitizing effects on MPC-HIF2α spheroids when applied shortly before or after irradiation. However, the effects of NO-COXIBs did not exceed those of celecoxib in terms of radiosensitization. This contrasts the bifunctional effect of 5a (referred to as compound 5 in our previous report [50]) on clonogenic cell survival under normoxia and hypoxia in melanoma cells showing high (A2058) and low (MelJuso) COX-2 expression. In this monolayer model, in both cell lines, radiosensitizing effects were observed when 5a was applied at a 10 µM concentration shortly before irradiation and radiosensitivity was enhanced by the additional NO • -release compared to celecoxib. Of note, while irradiation upregulated COX-2 expression and no influence of celecoxib or NO-COXIBs was observed on COX-2 protein synthesis in melanoma cells, also in this model, radiosensitization was found to be rather COX-2 independent [50]. Recently, we also demonstrated for COX-2-knockout A2058 melanoma cells (A2058-COX2 KO) that COX-2 independent pathways are most likely involved in the response to X-ray during concomitant treatment with celecoxib and rofecoxib at 1 or 10 µM treatment dose. As an example, radiosensitization was only observed after treatment with 10 µM of celecoxib while all other settings resulted in radioprotection of A2058 and A2058-COX2 KO cells [12]. Principally, a variety of COX-2 independent effects have been described for radiosensitization by celecoxib ranging from cell cycle arrest and prominent autophagy under hypoxic conditions by ER stress loading [67], G2-M phase arrest and apoptosis induction [68], blocking of vasculogenic mimicry through inhibiting off-targets aminopeptidase N (APN) and integrin alpha-V (ITAV) [69] to upregulation of BCCIP (BRCA2 and CDKN1A interacting protein) [70]. Herein, we found celecoxib to exhibit radioprotective effects on MPC-HIF2α spheroids when present during irradiation treatment and to exert radiosensitizing effects only when incubated 2 h after irradiation. Celecoxib did not show any scavenging effects on OH • , O 2 •− , and OClas shown previously by us [76] so that radioprotection exerted when given before irradiation probably does not rely on antioxidant effects. NO-COXIBs showed in comparison to celecoxib more radioprotective effects when applied 2 h before or after irradiation treatment, but no considerable effect when applied after 24 h. In addition, marked radioprotective effects in comparison to celecoxib were observed in combination with [ 177 Lu]LuCl 3 treatment. Generally, the inflammatory state mediated by COX-2 can be active for several days [65,72], whereas DNA damage repair and induction of apoptotic pathways is completed within a few hours after radiation-induced damage [77]. Furthermore inhibition of COX-2 with celecoxib was reported to inhibit the recovery from irradiation-induced injury e.g., by modulating homologous recombination [13,70]. Hence, COX-2 dependent effects are probable factors to modulate the acute radioresponse but as found herein are not the main drivers within MPC-HIF2α cells. In contrast, the time dependency of the treatment was found to prominently influencing the outcome for external beam radiation. Exposure of cells or tissue to ionizing radiation causes damage of DNA, proteins, and lipid membranes directly and indirectly by reactive oxygen as well as reactive nitrogen species, resulting from the radiolysis of water [78]. While radiation-induced cell and protein damage such as DNA double-strand breaks occur in both external beam irradiation and [ 177 Lu]LuCl 3 treatment, less frequent occurrence is expected for the low dose rate of Lu-177 exposure because a lower dose (estimated 2 Gy) is delivered over six days compared to irradiation where 15 Gy was delivered within minutes. Hence, impairment of cellular repair mechanisms by the COX-2-inhibiting properties of the compounds may be less relevant compared to the combination with the high dose rate external beam irradiation. Furthermore, the relatively slow kinetics of NO • release from the NO-COXIBs match with the relatively slow-dose delivery from Lu-177 exposure, indicating that the radioprotective effects of the compounds result from NO • release rather than from COX-2 inhibition.

Materials and Methods
All reagents and solvents were commercially available and used without further purification. Column chromatography was performed using silica gel (mesh size 40-63µm). Dry column vacuum chromatography (DCVC) followed the procedure described by Pedersen and Rosenbohm [79]. Melting points were determined with a melting point apparatus (Galen TM III, Cambridge Instruments, London, UK; Testotherm testo 700, Titisee-Neustadt, Germany; heater: Leica) and are uncorrected. Thin-layer chromatography (TLC) was carried out on Merck silica gel F-254 aluminum plates (Merck KGaA, Darmstadt, Germany), while visualization was performed using UV light (254 nm/366 nm; Benda, UV lamp NU-4). Nuclear magnetic resonance spectra (NMR) were recorded on a 400 MHz spectrometer (Unity INOVA 400 MHz, Varian (now Agilent Technologies, Santa Clara, CA, USA)). The chemical shifts are reported relative to the signal of the residual solvent for 1 H and 13 C spectra as internal standard. Copies of NMR spectra are given in Supplementary Material Figures S1-S55.
Analytical HPLC was carried out on the following systems: (System 1) column Luna

Chemistry
Synthesis of starting material 1a was performed as described in the literature [80], and analytical data were in accordance with published literature findings. The synthesis and characterization of compounds 2a, 3a, 4a, 4e, 5a, and 5e was recently described by us [49]. Selected results obtained within this study are described in the respective section below.
Preparation of silica sulfate catalyst: Sulfuric acid (96%, 3 mL) was added to silica gel (9.964 g) in diethyl ether (50 mL) in a round bottom flask. The suspension was mixed at room temperature at a rotary evaporator with a rotation speed of 150 rpm. All volatile components were removed under reduced pressure to yield the catalyst quantitively as a fine white powder.
Methyl 2,4-dioxo-4-(4-methoxyphenyl)butanoate (1b): Sodium methylate was freshly prepared by dropwise addition of methanol (5 mL) to sodium (0.936 g, 40.7 mmol, 1.22 equiv, stored over benzene) in a cooled flask, evaporation to dryness followed by addition of toluene (25 mL). Under nitrogen atmosphere, to the resulting suspension was added at 5-10 • C dimethyloxalate (4.4 g, 37.3 mmol, 1.11 equiv) followed by addition of 4methoxyacetophenone (5.0 g, 33.3 mmol, 1.0 equiv) in toluene (5 mL) over a period of 30 min. The solution was allowed to stand overnight. The crude product was isolated by filtration and washing with toluene (2 × 20 mL). Water (20 mL) was added to the solid, and the suspension was acidified with 6 M HCl (15 mL). After extraction with EtOAc, the organic phase was separated, washed with brine (2x), and dried over sodium sulfate. Evaporation under reduced pressure gave the purified product 1b as a yellow solid (6.

General Synthetic Procedure C
Under Schlenk conditions, to (5-(4-aryl)-1-(4-(sulfonyl)phenyl)-1H-pyrazol-3-yl)methanol (0.84 mmol, 1.00 equiv) was added thionyl chloride (2.20 mL) and the mixture was heated at 80 • C for 8 h. After that, thionyl chloride was distilled off under reduced pressure into a cooling trap where it was afterward deactivated with water. During synthesis and deactivation process, the exhaust gas of the apparatus was passed via an oil filled bubble counter to an alkaline aqueous solution to neutralize acidic gases. If indicated, the crude product was dissolved in EtOAc, adsorbed on silica gel and purified by column chromatography as specified below.      The mixture was heated to 80 • C and stirred for 24 h under reflux.* Afterward, the mixture was filtered to remove precipitated silver chloride and the filter residue was washed with EtOAc (15 mL). Both eluates were combined, and the organic phase was washed with water (3 × 3 mL), dried over sodium sulfate and the crude product adsorbed on silica gel. Purification was performed by column chromatography (petroleum ether/EtOAc 72.5/27.5).

Determination of COX Inhibition
COX inhibition activity against ovine COX-1 and human COX-2 was determined using the fluorescence-based COX assay "COX Fluorescent Inhibitor Screening Assay Kit" (catalog number 700100; Cayman Chemical, Ann Arbor, MI, USA) according to the manufacturer's instructions. Compounds 6a-d were screened at a concentration of 100 µM, all others were assayed in a concentration range of 0.1 nM to 100 µM in duplicate.

Fluorometric NO • Release Assay (Griess Assay)
Nitric oxide release of compounds within 24 h at 37 • C was measured using the fluorometric nitric oxide microplate assay kit ab65327 (Abcam, Cambridge, UK). For measurement of spontaneous release, 1.8 nmol of compounds were incubated in 75 µL Dulbecco's PBS containing 0.05% v/v residual DMSO and 2.5 mM DTPA (pH = 7.1) for 24 h at 37 • C. Thioldependent release was measured accordingly in presence of 2.5 mM cysteine. Two series of nitrate standards were prepared (0.2−1 nmol) in 75 µL incubation buffer, in presence or absence of cysteine, respectively. According to the manufacturer's instructions, all nitrate content in samples and standards was enzymatically converted into nitrite through incubation with nitrate reductase for 2 h at room temperature, and the resulting nitrite content (total nitric oxide) was determined through reaction with 2,3-diaminonaphthalene followed by measurement of fluorescence intensity at λ ex/em = 360/450 nm using a microplate reader. Nitrite content in samples was calculated from linear regression of standards and expressed as relative nitric oxide release (%) normalized to the initial molar amount of compound. NONOate served as positive control.

Conversion in NO • -Assay Buffer followed by UPLC and HPLC-HRMS
The time-dependent conversion of NO-COXIBs 5a-f and 6a-d as well as reference compounds 4f and celecoxib in the presence or absence of cysteine in assay buffer was investigated by following the degradation of intact compound and identification of selected degradation products by UPLC. For that, samples were incubated at 37 • C in the assay buffer in the presence or absence of cysteine and at the indicated time point; 5 µL assay buffer of each sample was withdrawn and diluted with 45 µL MeCN/water (50/50 v/v). The samples were thoroughly vortexed and then subjected to UPLC analysis. All samples of one time point were analyzed within a total analysis time of approximately 2.5 h. Conversion was monitored at 254 nm using the following UPLC system: column Aquity UPLC ® HSS C18 column (Waters Corporation, Milford, MA, USA; 100 × 2.1 mm, 1.  45/55). For the identification of degradation products, samples taken after 24 h were additionally subjected to HPLC-HRMS analysis using HPLC system 6. In brief, the NO-COXIBs were converted to the respective hydroxymethyl-substituted compounds and in the presence of cysteine additionally to the cysteine adduct, which could be identified by the respective ∆m/z shifts. A detailed comparison of calculated and found m/z-signals is given in Supplementary Material Table S1, and copies of HPLC-HRMS chromatograms and selected spectra are given in Supplementary Material Figures S56-S66.

Spheroid Regrowth Assay
Genetically modified mouse pheochromocytoma (MPC) cells expressing a codonoptimized Epas1 gene (encoding HIF2α) were cultured as three-dimensional tumor spheroids as published elsewhere [83]. External X-ray beam radiation treatment and [ 177 Lu]LuCl 3 radionuclide treatment was performed as reported previously [59] with some modifications. In brief, spheroids were irradiated with a 200 kV photon beam using a Maxishot X-ray system equipped with a Y.TU/320-D03 tube (YXLON, Hamburg, Germany), growth of spheroids (n = 6) with diameters between 300−400 µm was measured in response to one fixed radiation dose, 15 Gy for X-ray treatment, and 0.25 MBq/mL of [ 177 Lu]LuCl 3 for radionuclide treatment corresponding to a cumulative dose of approximately 2 Gy delivered for 6 days. The applied radiation doses were selected based on the half-maximal spheroid control doses (SCD 50 ) as previously determined for MPC HIF2α spheroids by us [59]. Spheroids were considered as regrown at 100 µm increase in diameter compared to treatment start or at a diameter growth rate of 15 µm per day. Results are reported as spheroid regrowth probability (%).

Statistical Analyses
Statistical analyses were performed using Prism 8 (GraphPad, La Jolla, CA, USA). Data were visualized as mean ± SEM.

Summary and Conclusions
The aim of the present work was to extend the knowledge about NO-COXIBs based on the previously described lead 5a. COX-inhibition potency of the newly synthesized COXIBs (5b-d,f) differing in the periphery was comparable to their lead 5a, while the introduction of an N-propionamide residue at the sulfamoyl group of 6a-d abolished inhibition potency as expected, as part of a prodrug strategy that lowered lipophilicity considerably. All synthesized compounds were found to release nitric oxide as determined by Griess assay, and differences in the time-dependent release were found to be mainly influenced by the substitution at the pyrazole. The effect of NO-COXIBs on genetically modified HIF2αpositive mouse pheochromocytoma (MPC-HIF2α) tumor spheroids was tested to evaluate the application of the compounds with respect to external beam irradiation or radionuclide therapy. In this artificial model, NO-COXIBs were found to have a radioprotective effect. This raises the question, if radioresponse to treatment would be similar or different in the in vivo situation of this cancer entity. Furthermore, future evaluation as radioprotectant if applied before radiation events on normal tissue should be considered based on the principal COX-2 selective inhibition and nitric oxide-releasing properties as well as the potential to use the respective more hydrophilic prodrugs introduced herein. In this regard, determining the optimal dosage at an optimal time window for treatment is a very important point to consider when planning the use of coxibs and only NO-COXIBs. This aspect has already been highlighted for other potential applications of coxibs and NO-COXIBs as adjuvants, for example in the modulation of inflammatory processes in bone healing [84].
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/molecules27196587/s1, Copies of NMR spectra, HPLC chromatograms and HRMS data after incubation in NO assay buffer, expression levels of Ptgs1 and Ptgs2 genes in genetically modified MPC cell lines, growth-rate of tumor spheroids in the presence of NO-COXIBS without radiation treatment, and detailed growth response of tumor spheroids in the presence of test compounds.