{"paper_id":"7624647b-16c3-498b-bcee-223cb34fc73a","body_text":"Carcinoma of ovaries\nand pancreas, neuroblastoma, and medulloblastoma\nremain life-threatening. 1  Analyses of the\ncancer patient’s microarray databases demonstrate that Vitamin-D\nreceptor (VDR) mRNA is overexpressed in the carcinomas of pancreas,\novaries, bladder, glioma, liver, and lungs and in neuroblastoma and\npredicts poor prognosis. VDR is also enriched in hyperplastic polyps\nand endometriosis and in early stages of tumorigenesis. 2 − 5  Causes of VDR overexpression in malignancies, polyps, and other\ndisease states are unclear and require further investigations. VDR,\na class-III nuclear receptor (NR), mediates physiologic actions of\ncalcitriol ( 1 ) 2  ( Figure  1 ), the hormonally\nactive form of Vitamin-D. Calcitriol has been tested in human trials\nfor the treatment of various malignancies, 6 − 12  but it induces hypercalcemia, which is an undesirable side effect\nin patients. 13 1  is currently\nused in the management of plaque psoriasis, hyperparathyroidism, and\nnephropathy. 14 , 15\nChemical structures of calcitriol ( 1 ) and the representative\nliterature-described VDR antagonists TEI-9647 ( 2 ), TEI-9648\n( 3 ), and MT19c ( 4 ). Chemical structure of\nthe new VDR antagonist MeTC7 ( 5 ).\nFurther, in addition to VDR, RXRα, the heterodimerization\npartner, which is necessary for DNA binding and recruitment of coregulators,\nis altered in malignancies and predicts poor prognosis. 16 , 17  Similarly, Importin-4, essential for nuclear internalization of\nVDR, is aberrantly altered in malignancies and exhibits poor prognosis. 18  Furthermore, VDR was shown to induce MYCN overexpression. 19  MYCN is overexpressed in over 70% of the malignancies. 20  In addition, our ongoing studies and published\nliterature show that Vitamin-D/VDR upregulates PD-L1 in cancer cells. 21  Therefore, we postulate that a VDR antagonist\nis needed to block tumorigenesis orchestrated by aberrant VDR and\nthe associated signaling nodes VDR/RXRα/Importin-4, VDR/MYCN,\nand VDR/PD-L1.\nProgress in inhibiting VDR has remained hampered\nby the unavailability\nof pharmacologically pure VDR antagonists. 22 , 23  Currently known VDR antagonists can exhibit residual agonistic effects, 22 , 23  and their therapeutic effects have yet to be evaluated in animal\nmodels for malignancies. In addition, the synthesis of the literature-described\nVDR antagonists including TEI-9647 ( 2 ) and TEI-9648 ( 3 ) ( Figure  1 ) requires multiple synthesis steps and is challenging. To overcome\nthese problems, we have attempted to develop easily synthesized VDR\nantagonists. 24  In this context, we had\npreviously reported MT19c ( 4 ) ( Figure  1 ) as a new class of VDR antagonists that\nactually showed very strong antitumor activity in animal models. 25  However, the low VDR antagonist activity of  4  remained an issue.\nIn this report, we describe our\nefforts to identify a novel VDR\nantagonist MeTC7 ( 5 ).  5  can be synthesized\nfrom 7-dehydrocholesterol (7DHC) ( 6 ) in two steps. We\ninvestigate the VDR selectivity of  5  and perform  in silico  studies to understand how  5  affects\nthe VDR-ligand-binding domain (VDR-LBD) versus calcitriol.  In vitro  and  in vivo  experiments using  5  were performed to understand the effect of VDR inhibition\non RXRα and Importin-4 and MYCN expression, the critical VDR\ndownstream signaling nodes, and to examine its effects on the growth\nof ovarian cancer, neuroblastoma, pancreatic cancer, and medulloblastoma\ncells. Based on the described role of VDR in MYCN’s expression\nin neuroblastoma, 19  we examine the effects\nof  5  treatment on the growth of spontaneous neuroblastoma\nusing a homozygous tyrosine hydroxylase (TH)-MYCN transgenic model 26  and investigate its effects on the population\nof hematopoietic cells as a measure of off-target effects. Our studies\nshow that  5  is a selective VDR antagonist endowed with\npromising antitumor effects against xenograft and transgenic spontaneous\ntumor models.\n\nOur approach of developing a novel\nVDR antagonist pharmacophore,\nof which MT19c ( 4 ) 24 , 25  is a previously described\nderivative, involves heterocyclizing the biological activity-endowed\nsecosteroidal scaffolds (Vitamin-D2/D3 and ergocalciferol 24 , 25 ) via Diels–Alder reactions with dienophiles (MTAD, PTAD)\nto (1) disable interactions with 1-α hydroxylase, which pivots\nthe classical Vitamin-D signaling including calcium regulations; and\n(2) convert the purely carbonaceous scaffold to heteroatom-rich druglike\npharmacophores carrying balanced charge/lipophilicity ratios. This\nstrategy generates unique heterocycle-fused conformationally constrained\nnovel pharmacophores in a short-path and atom-economy manner that\ncan be variously derivatized further to generate bioactive compounds\nthat are antagonists to VDR, are void of residual agonistic effects,\nand are highly nuclear receptor-selective. Our strategy differs from\nthe one followed for the development of TEI-9647 ( 2 )\nand TEI-9648 ( 3 ), which carry the Michael acceptor lactone\nring in their side chains keeping the A-ring unaltered. Probably,\nresidual VDR agonistic effects in  2  and  3  arise because the A-ring is accessible for interactions with 1-α\nhydroxylase; for similar reasons,  2  and  3  may also exhibit hypercalcemia if administered in animals. To improve\nupon the weak VDR antagonist activity of  4 , we derivatized\n7DHC ( 6 ) to build our new pharmacophore and MeTC7 ( 5 ) as the key derivative.  5  was designed to carry\nout additional rigidity in the backbone structure compared to Vitamin-D2/D3\nand ergocalciferol.\nMeTC7 ( 5 ) was synthesized by\nthe method shown in  Scheme  1 . 7DHC ( 6 ) was reacted with  N -methyl-1,2-4-triazolinedione (MTAD) in dichloromethane (DCM) at\n0 °C to afford  7  at 52% yield. Then,  7  was reacted with bromoacetic acid in the presence of  N , N ′-dicyclohexylcarbodiimide (DCC) and 4-dimethylaminopyridine\n(DMAP) in anhydrous dichloromethane, and the reaction mixture was\npurified by a preparative thin-layer chromatography plate to give  5  at 67% yield. Characterization data for  5  and  7  are shown in  Figure S1 .\nMeTC7\n( 5 ) showed potent VDR inhibition (IC50 = 2.9 ±\n0.1 μM) ( Figure  2 a, left) in a fluorescence polarization (FP) assay performed using\nVDR-LBD, SRC2-3 Alexa Fluor 647 and 10 nM of 1. 27 , 28  7-Dehydrocholesterol (7DHC,  6 ) and 7DHC-adduct ( 7 ) did not inhibit VDR (IC 50  > 100 μM).\nFluorescence\npolarization studies showed that  5  is void of any VDR\nagonistic activity ( Figure  2 a, right). In a cell-based transactivation assay,  5  inhibits VDR transactivation in the concentration range of (IC 50  = 20.8 + 8.3 μM) ( Figure  2 b).  6  and  7  did\nnot inhibit VDR transactivation (IC 50  > 100 μM)\n( Figure  2 b). Since\nlack of\nNR selectivity is a major challenge in developing NR modulators, 29  next, we investigate whether  5  binds\nRXRα, the heterodimeric binding partner of VDR. Fluorescence\npolarization assay 30  showed that  5  treatment does not bind RXRα, while Bexarotene (IC 50  = 3.74 μM) and 9- cis -retinoic acid\n(IC 50  = 3.45 μM) showed potent binding ( Figure  2 c). Similarly,  5  did not show agonistic or antagonistic effects against PPARγ\n( Figure  2 d), another\nNR, until the tested doses of 100 μM.\nPharmacologic characterizations\nof MeTC7 ( 5 ). (a)\nFluorescence polarization assay showing VDR inhibition by  5  (left) without induction of agonistic effects (right). (b)  5  inhibited VDR transactivation in a HEK293 cell-based assay.\n(c)  5  did not bind to RXRα. (d)  5  did\nnot exhibit agonistic or antagonistic effects against PPAR-γ.\nTo elucidate the structural\nmechanisms\nforming antagonistic attributes of  5 ,  in silico  studies are performed to determine the interactions of  5  with VDR-ligand-binding domain (VDR-LBD) residues. The crystal structure\nof VDR-LBD (1DB1) cocrystallized with 2-α-(3-hydroxy-1-propyl)\ncalcitriol 31  is used for docking studies.  1  is visualized to consist of a ring-A, a conjugated linker,\nring C, and ring D along with a flexible chain ( Figure  3 a, left). The simultaneous interactions mediated\nby C1, C3, and C25-hydroxyl groups are crucial for the super agonistic\nbehavior of  1 . 31  Synchronized\ninteraction is possible only if the correct spacing exists between\nthe hydroxyl groups, which is achieved by proper folding within the\nmolecular structure of  1  to favor the correct orientation\nof the hydroxyl group. Structurally,  5  ( Figure  3 a, right) is much larger in\nsize and is a highly conformationally rigid system compared to calcitriol.\nThree-dimensional (3D) binding of  5  (green) in LBD (brown)\nof VDR (white surface) is shown by the surface diagram and overlaid\nwith  1  (yellow) in  Figure  3 b. Binding interactions of  5  with VDR-LBD\nare shown in  Figure  3 b. An overlaid superposition for  1  and  5  is shown in  Figure  3 c. To handle this large and rigid molecule, the induced-fit (IF)\ndocking strategy 32 − 35  was implemented to reveal the possible binding modes of  5  with VDR-LBD.  1  shows the involvement of the −OH\ngroup in H-bonding with VDR residues (Ser278 and His305) ( Figure  3 d). Binding of  5  shows the recruitment of C–H...π, C–H...O,\nand H-bonding with VDR-LBD active site residues Trp286, Tyr147, Asp144,\nand Ser237 ( Figure  3 e) due to the major changes in the binding motifs (M1–M4),\nwhich potentially result in an antagonistic effect. 32  The shortening and removal of the C25-OH group near the\nM1 motif in  5  cause the loss of interaction with His305.\nHis305 along with Arg274 residues play a crucial role in determining\nthe agonistic behavior of  1  ( Figure  3 d). The loss of the conjugated linker system\nin  5  followed by the shortening distance (2.6 Å)\nin comparison to  1  (3.7 Å) between rings A and C\nappears to induce the antagonistic behavior in  5 . Thus,\nthe triazolidine-dione moiety and the hydrophobic  N -methyl group occupy a similar spatial position within VDR as was\nplaced by the hydrophilic C1-OH of calcitriol. Thus,  5  loses H-bonding with Arg274; however, carbonyl group’s interaction\nwith Ser237 supports the strong binding of  5  ( Figure  3 e,f). Further, the\nconjugated diene linker of  1  enters tightly into the\nhydrophobic cavity of the VDR and agonizes the system 31  ( Figure  3 d). In contrast, the triazoline-3,5-dione moiety of  5  increases the volume of this VDR-LBD cavity and locks the conformational\nfreedom of VDR due to the deeper binding ( Figure  3 e,f). Further, interactions of Tyr143 and\nArg274 with\n1,3-OH of the A-ring of  1  (A-ring subsite) were lost\nwhen  5  bonded with VDR-LBD ( Figure  3 e,f).  5  interacts with a strong\nH-bonding with a backbone of Asp144 and Ser237 in the ring-A subsite\n( Figure  3 e).  5  utilizes only hydrophobic residues in the 25-OH-subsite,\nwhereas  1  uses a hydrophilic interaction (Ser306 and\nHis305) as well.  5  interacts with Leu230, Ala303, and\nVal380 through favorable van der Waal’s contacts shown by a\nmesh surface diagram (green for LBD residues and yellow for  5 ). The videos ( 1  and  2 ) exhibit\nthe interactions of  1  and  5  with VDR-LBD\nresidues.\nPutative 3D-binding modes of  5  in VDR-LBD with noncovalent\ninteractions. (a) For  in silico  binding to VDR-LBD,  1  and  5  were marked into four structurally relevant\nzones (M1–M4). Binding motifs (M) in  1  and  5  are highlighted by red dotted circles. The bond connection\nbetween atoms 9 and 10 of  5  (red) is shown far to get\nbetter clarity of the M2 motif. However, atoms 9 and 10 are close\nto the normal C–C bond in  5  in which A and B rings\nfuse together to form the conformationally rigid ring system. (b)\nSurface view with the VDR-active site showing the binding mode overlay\nof  1  (yellow) (movie 1) and  5  (green) in\nLBD (brown) (movie 2). (c) Superposition overlay of  1  (yellow, cocrystal structure) and  5  (green) into the\nVDR-ligand-binding domain (LBD). (d) 3D-binding mode of  1  (crystallographic structure) showing major noncovalent interactions\nwith VDR-LBD. (e) 3D-binding mode of  5  showing major\nnoncovalent interactions such as C–H...π, C–H...O,\nand H-bonding. (f)  5  interacts with Leu230, Ala303, and\nVal380 through favorable van der Waal contacts shown by the mesh surface\ndiagram (green for LBD residues and yellow for  5 ).\nKaplan–Meier survival\nanalyses at the system-selected expression cutoffs (microarray data\nand tools available at R2-Genomics Analysis and Visualization Platform  https://hgserver1.amc.nl/cgi-bin/r2/main.cgi ) showed that VDR mRNA enrichments correlated with increased mortalities\nin lung ( p  = 0.0043) and pancreatic cancer patients\n( p  = 0.004), neuroblastoma patients ( p  = 1.7 × 10 –5 ), breast cancer ( p  = 0.011), glioma ( p  = 0.0048), cervical cancer\n( p  = 0.055), liver cancer ( p  = 0.048),\novarian cancer ( p  = 0.09), and bladder cancer ( p  = 0.05) patients ( Figure S2A–I ). The effect of disease stages on the association of VDR mRNA enrichments\nwith decreased mortalities in cancer patients was analyzed. Among\nstage IIa/IIb pancreatic cancer patients, VDR mRNA enrichment was\nstrongly associated with increased mortalities (data not shown). The\nidentity of microarray databases analyzed is described in the Materials\nand Methods section.\nScreening a panel\nof SKOV-3, OVCAR-3, OVCAR-8, CaOV-3, IGROV-1, and 2008 ovarian cancer\ncell-lines by immunoblotting identifies 2008 and SKOV-3 cells as the\nhigh VDR expressor cell-lines suitable for VDR/ 5  signaling\nstudies ( Figure  4 a,\nupper). The expression of α-tubulin as the loading control in\nthese cells is shown ( Figure  4 a, lower).  5  inhibited the expression of RXRα\nin 2008 ovarian cancer cell-lines ( Figure  4 b). RXRα expression correlates with\npoor prognosis in ovarian cancer patients ( Figure S3 ). Importin-4 mediates nuclear translocation of VDR. 36  VDR showed colocalization with Importin-4 in\novarian cancer tissues ( Figure S4a ) and\nindicated poor prognosis in neuroblastoma ( Figure S4b ).  5  (250 nM, 12 h) treatment suppressed Importin-4\nexpression in 2008 cell-lines ( Figure  4 c).\n(a) VDR expression in a panel of 2008, IGROV-1, CaOV-3,\nOVCAR-3,\nOVCAR-8, and SKOV-3 ovarian cancer cell-lines. Expression of α-tubulin\nas a loading control is shown. Normalized western blot densitometric\ndata are shown numerically. (b) Treatment with  5  (250\nnM, 18 h) reduces the expression of RXR-α in 2008 cells. Expression\nof GAPDH as a loading control is shown. Normalized western blot densitometric\ndata is shown numerically. (c) Treatment with  5  (250\nnM, 18 h) reduces the expression of Importin-4 in 2008 cells. Expression\nof α-tubulin as a loading control is shown. Normalized western\nblot densitometric data are shown numerically. (d)  5  reduces\nthe viability of SKOV-3, IGROV-1, CAOV-3, OVCAR-3, OVCAR-8, and 2008\novarian cancer cell-lines during 24 h of treatment. Data + standard\nerror of the mean (SEM) are expressed as the mean of the triplicate\ndeterminations as the % of absorbance by dimethylsulfoxide (DMSO)-treated\ncells set equal to 100%. (e) Treatment with  5  (250 nM,\n18 h) increases cleaved PARP1 expression in 2008 cells. Expression\nof α-tubulin as a loading control is shown. Normalized western\nblot densitometric data are shown numerically. (f-left)  5  ( n  = 10, 10 mg/kg. M-F, IP) treatment reduces the\ngrowth of ES2 ovarian carcinoma-derived xenografts in NSG mice compared\nto the vehicle ( n  = 10). At the baseline, the  5  group has a statistically larger volume to begin with compared\nto the vehicle ( p  = 0.02). At the second and third\ntime points, the  5  group has a statistically smaller\ntumor volume compared to the controls ( p  = 0.0303\nand 0.0119, respectively). At the final time point, there is no statistical\ndifference between the treatment groups with respect to the tumor\nvolume. (f-right): There is no statistical difference in animal weights\nacross time between treatment groups ( p  = 0.5437).\nA panel of SKOV-3, OVCAR-3,\nOVCAR-8, IGROV-1, CAOV-3, and 2008 ovarian cancer cell-lines showed\na dose-dependent response to  5  treatment ( Figure  4 d). OVCAR-8, 2008, SKOV-3,\nCAOV-3, and IGROV-1 cells were sensitive to  5  treatment,\nwhile OVCAR-3 was relatively resistant against  5  treatment.\nCleaved PARP1 expression in 2008 cells upon treatment with  5  was observed ( Figure  4 e).\nMeTC7 ( 5 ) treatment slowed the growth of the clear-cell\novarian carcinoma cell-line ES2-derived xenografts growing in NSG\nmice despite starting with significantly higher basal tumor sizes\nin the treatment group ( p  < 0.05 for the first\ntwo treatments,  Figure  4 f, left). The difference between tumor sizes in the control and treatment\ngroups at the final treatment is not statistically significant. The\ndifference between the animal weights in the treatment and control\ngroups is not statistically significant ( Figure  4 f, right). While ovarian cancer cell-lines\nshow sensitivity to  5  treatment, HepG2 (hepatocellular\ncarcinoma) cell-lines and HEK293T (immortalized human embryonic kidney,\nHEK, cell-line) exhibit resistance to  5  treatment until\n100 μM concentrations ( Figure  5 a,b). In addition to ES2,  5  treatment\nreduces the growth of SKOV-3 serous ovarian cancer cell-derived xenograft\ntumors ( p  = 0.02,  Figure  5 c). The rate of weight gain was significantly\nhigher in the  5  group than in the control group over\nthe time period ( p  = 0.005) ( Figure  5 d). Similar results were obtained using percentage\nchange in weights relative to the baseline. The Kaplan–Meier\nanalysis showed statistically greater survival prospects for the treatment\ngroups ( Figure  5 e).\nImmunohistochemistry (IHC) shows reduced VDR expression in a randomly\nselected  5 -treated SKOV-3 xenograft tumor  in\nvivo  ( Figure S5 ).  5  treatment also reduces the rate of growth of medulloblastoma D283\ncell-derived xenografts in NSG mice, compared to the control ( p  = 0.032,  Figure  5 f). Harvested tumors post euthanasia showed a tendency to\nform smaller tumors in the treatment group than in the control ( p  = 0.093,  Figure  5 f, inset). Further,  5  treatment reduces the %\nchange in the growth of PANC-1 ( p  = 0.036 on day\n12) and BXPC-3 ( p  = 0.0063 on day 18) pancreatic\ncancer cell-derived xenografts in NSG mice ( Figure  5 g,h).\nSelective antiproliferative functions\nand antitumor activities\nof MeTC7 ( 5 ) in xenograft animal models. (a)  5  does not inhibit the proliferation of HepG2 cells in the dose ranges\ntested. (b)  5  does not inhibit the proliferation of HEK293T\ncells in the dose ranges tested. (c)  5  (10 mg/kg, M-F,\nIP) treatment reduces the growth of SKOV-3 cell-derived xenografts\nin nude mice. (d) Animal weights of the mice undergoing treatment\nwith vehicle or  5  increase during the period of observations.\n(e) The control group witnessed a death on day 25, indicated by the\ndrop in the line. The dashed line for  5  indicates 100%\nsurvival. (f-left)  5  (10 mg/kg, M-F, IP) treatment reduces\nthe growth of D283 medulloblastoma cell-derived xenografts in NSG\nmice. % change in average tumor volume in the treatment group was\nlower compared to the vehicle on day 12. (f-right) Mice were euthanized,\nand tumors were extracted and weighed. The tumor weights in the treatment\ngroup show a tendency to be smaller (unpaired  t -test,  p  = 0.093). (g, h)  5  (10 mg/kg, M-F, IP) treatment\nreduces % change in the growth of PANC-1 and BXPC-3 cell-derived xenografts\ngrowing in NSG mice (PANC-1, day 12,  p  = 0.0361)\nand (BXPC-3, day 18,  p  = 0.0063). Statistical differences\nbetween the groups were analyzed using GraphPrism.\nVDR mRNA overexpression is prognostic\nin neuroblastoma ( Figure S2c ).  5  treatment suppresses the viability of neuroblastoma cell-lines Lan-5,\nSK-N-AS, SHEP-1, BE(2)C, Kelly, and SH-SY5Y dose-dependently ( Figure  6 a). Immunoblotting\nshows that VDR and MYCN are expressed in neuroblastoma cell-lines\n( Figure  6 b). Lan-5,\nKelly, and BE(2)-C express both VDR and MYCN, whereas SK-N-AS and\nSHEP-1 are VDR-positive but MYCN-negative.  5  treatment\nreduces VDR and MYCN expressions in BE(2)C cell-lines ( Figure  6 c).  In vivo ,  5  (10 mg/kg, M-F, IP) treatment reduces the growth\nrate of the SH-SY5Y xenograft as measured on day 5 ( p  = 0.018) and day 10 ( p  = 0.012) in NSG mice ( Figure  6 d, upper). Similarly,\n% change in BE(2)C tumor volumes of  5 -treated mice is\nsignificantly lower than in vehicle-treated NSG mice (day 5,  p  = 0.0052; day 8,  p  = 0.0034) ( Figure  6 d, lower and  Figure S6 ).\n(a) MeTC7 ( 5 ) treatment for 48 h decreases the viability\nof LAN-5, SK-N-AS, SHEP-1, BE(2)-C, Kelly, and SH-SY5Y neuroblastoma\ncell-lines dose-dependently. Data represent ±SEM. (b) Immunoblot\nanalysis shows relative VDR and MYCN levels in LAN-5, SK-N-AS, SHEP-1,\nBE(2)-C, Kelly, and SH-SY5Y cell-lines. Expression of β-actin\nas the protein loading control is shown. (c)  5  (1 μM,\n24 h) treatment reduces VDR, MYCN, and cyclin-D expressions in the\nBE(2)-C cell-line. Expression of β-actin or α-tubulin\nas a protein loading control is shown. Normalized western blot densitometric\ndata are shown numerically. (d-upper)  5  (10 mg/kg, M-F,\nIP) treatment reduced the growth rate of SH-SY5Y tumors growing in\nNSG mice.  T -test analysis showed that % changes in\ntumor volumes are lower in the treatment group (day 5:  p  = 0.0176; day 10:  p  = 0.0102) than in the control\n(d-lower).  5  (10 mg/kg, M-F, IP) treatment reduced the\ngrowth rate of BE(2)-C xenograft tumors growing in NSG mice.  T -test analysis showed that % changes in tumor volumes are\nlower in the treatment group ( p  = 0.0057 on day 5;  p  = 0.0034 on day 8) than in the control.\nMYCN overexpression predicts poor overall and event-free survival\nin patients with ovarian cancer and neuroblastoma 37  ( Figure S7 ) and other malignancies. 38 − 45  We investigate whether targeting VDR by  5  can inhibit\nMYCN expression and, in turn, control the MYCN-orchestrated neuroblastoma\ngrowth. Prior to testing antineuroblastoma activities of  5  using the well-established TH-MYCN+/+ transgenic mice, which spontaneously\ndevelop neuroblastoma and recapitulate human neuroblastoma disease\nclosely, 26 , 46  we establish via immunohistochemistry that\nceliac ganglia harvested from homozygous TH-MYCN+/+ mice exhibit positive\nexpressions of VDR, MYCN, and TH antigens ( Figure  7 a–c). Similarly, an MTS cell viability\nassay run on the tumor cells derived from three independent homozygous\nTH-MYCN+/+ mice shows reduced viability of the spheroid cell’s\nviability upon  5  treatment ( Figure  7 d upper). Images of the tumor spheroids treated\nwith vehicle or  5  are shown ( Figure  7 d lower).\nImmunohistochemistry in the tissues from\nhomozygous TH-MYCN mice\nshowed strong expressions of VDR, MYCN, and tyrosine hydroxylase (TH).\n(a) Compared to wild-type, celiac ganglia isolated from the TH-MYCN+/+\nmice (postnatal days 1, 7, and 14) showed increased VDR expression.\n(b) Positive and negative controls used for IHC in postnatal mice.\nSerial section stained with normal immunoglobulin G (IgG) was used\nas a negative control for IHC staining. The expression of VDR in the\npancreas of TH-MYCN mice was used as a positive control to verify\nthe VDR antibody retrieval method and staining in the same sagittal\nsection as the celiac ganglia. (c) Advanced TH-MYCN tumor isolated\nfrom a 5.5-week old mouse showed VDR, MYCN, and TH expressions in\nthe tumors. (d-top)  5  treatment reduces the viability\nof TH-MYCN tumor cells isolated from the three independent mice (tags:\n1, 6, and 8) during three days of treatment. Data represents ±SEM.\n(d-bottom)  5  treatment dose-dependently decreases the\nviability of murine neuroblastoma tumor spheres isolated from TH-MYCN\nmouse (tag: 1). Representative images of tumor spheres from the same\ncell viability experiment. Images were captured using an Olympus BX41\nlight microscope with an Olympus DP70 camera and CellSens digital\nsoftware.\nNext, we investigate the effect\nof  5  against the growth\nof tumors in TH-MYCN+/+ transgenic mice. The response of the drug\nin alive mice was monitored using ultrasound. Images were reconstructed\nto capture the 3D tumor volume using inbuilt software.  5  reduces the tumor growth compared to the vehicle group ( Figure  8 a). Analysis of the\nestimated tumor volumes ( p  = 0.033) ( Figure  8 b) and harvested tumor weights\n( p  = 0.053) ( Figure  8 c) exhibits the reduced tumor burden upon treatment\nwith  5 .\n(a) Antitumor activity of MeTC7 ( 5 ) in the\nTH-MYCN\nmodel of spontaneous neuroblastoma. Compared to the vehicle (upper,  n  = 7),  5  (lower, 10 mg/kg, daily IP,  n  = 8) treatment reduced the growth of TH-MYCN-driven neuroblastoma\nin transgenic mice. Tumor burden was measured by an ultrasound imaging\ninstrument. (b)  5  treatment dose-dependently reduced\nthe TH-MYCN tumor volume at the end of the treatment. (c) Mice were\neuthanized, and extracted tumors were weighed. Tumor sizes and weights\nin the control versus treated groups were compared using a  T -test.\nRuling out whether  5  treatment exerts off-target effects\nagainst hematopoietic cells, the common side effects of chemotherapies,\nthe population of various CD45+ cell subtypes isolated from TH-MYCN\nmice was analyzed by flow cytometry using characterized markers. The\nanalysis shows that  5  treatment does not affect populations\nof CD45+, CD4+, CD8+, macrophages, patrolling monocytes/DC+, and CD11b+\ncells compared to the control ( Figure S8 ).\nBased\non the data described in the report above, the cartoon ( Figure  9 ) summarizes the putative mechanism\nof action of MeTC7 against VDR.\nCartoon outlining the putative mechanism\nof action of MeTC7 against\nVDR.\n\nCompared to agonists, the development of VDR antagonists has lagged\nbehind. 23 , 24  Following our previously described approach\nof the Diels–Alder modification of secosteroidal scaffolds\nthat generated earlier classes of VDR antagonists [MT19c ( 4 ) and PT19c], 24 , 25 , 47 5  was synthesized.  5  showed superior VDR\ninhibition than  4  and PT19c without incurring any agonistic\nactivity and exhibited noteworthy NR selectivity against PPAR-γ\nand RXRα, the two closely related members of the VDR-NR family\n( Figure  2 c,d).  In silico  studies show that heterocyclization of  6  by 1-methyl-1,2,4-triazolinedione instills enormous structural rigidity\nin  5 , which disrupts VDR-LBD. It has been shown that\nupon binding to the LBD of VDR, the antagonist complex converts into\na transcriptionally inactive form. 35  Loss\nof TYR143, HIS305, and ARG274 interactions, which are critical for\nhydrogen-binding interactions of  1 , may account for the\nantagonistic attributes of  5 , similar to the effects\nof mutations at HIS305 along with ARG274 residues, which was shown\nto generate antagonistic effects 35  ( Figure  3 e–g). Structurally,  5  differs from the literature 23 , 24 -described\nVDR antagonists TEI-9647\n( 2 ) and TEI-9648 ( 3 ).  5  carries\na highly constrained heteroatom-rich-tetracyclic ring system derived\nfrom the conjugated diene system of  6 , the precursor\nof Vitamin-D, whereas  2  and  3  retain the\nclassic Vitamin-D scaffold but have the C25 carbon converted into\na five-membered lactone ring.  5  sports an alkylating\nbromoacetoxy functionality, whereas a Michael acceptor ring likely\nforms the basis of  2  and  3  functions.\nThe rationale to identify a VDR antagonist such as MeTC7 ( 5 ) and investigate its antitumor effects stemmed from (1)\nVDR mRNA overexpression in ovarian, breast, lung, pancreatic, neuroblastoma,\nand bladder malignancies and association with poor prognosis; (2)\nthe role of Vitamin-D/VDR in increased immune checkpoint inhibitor\nligand PD-L1 expression in head-neck cancer, leukemia, 23  and ovarian cancer cells (unpublished data);\nand (3) VDR’s role in expression of MYCN, 19  an oncogene dysregulated in ∼70% of human cancers. 37 − 45  Further, RXRα and Importin-4, the critical downstream signaling\nnodes of VDR, are also shown to be altered in malignant cells and\npredict poor prognoses in malignancies 16 − 18  ( Figures S3 and S4 ).\nThe anticancer effects of VDR antagonists\nare not well understood. 22 , 23 , 48 , 49  Our study shows that ovarian\ncancer, neuroblastoma, medulloblastoma,\nand pancreatic cancers respond to  5  treatment  in vitro  and  in vivo  ( Figures  4 d,f,  5 c–h, and  6 a,d).  In vivo , animals treated with  5  did not experience detrimental\neffects on their weights ( Figures  4 f and  5 d) or general demeanors.  5  actions are VDR-dependent, as stably VDR knockdown SKOV-3\ncells show diminished responses compared to stably VDR overexpressor\nSKOV-3 cell-lines, which respond better than their null vector or\nwild-type cell counterparts (data not shown). Importantly,  5  reduced VDR expression in the SKOV-3 cell-line-derived xenografts,\ndemonstrating target engagement  in vivo  ( Figure S5 ). Not only xenograft tumors but also\nsyngeneic TH-MYCN murine transgenic spontaneous neuroblastoma that\nmore closely recapitulates human neuroblastoma disease showed reduction\nin tumor growth ( Figure  8 ). In terms of signaling,  5  reduced the expression of\nRXRα and Importin-4, the two pivotal nodes of the VDR signaling\npathway ( Figure  4 b,c).\nThe probable mechanism of action of MeTC7 is outlined in  Figure  9 . Since  5  does not directly inhibit RXRα ( Figure  2 c), it is likely that RXRα released\npost VDR inhibition is degraded. Decreased expression of RXRα\nis therapeutically important because RXRα is prognostic in renal\ncancer ( p  < 0.00056), melanoma ( p  = 0.034), ovarian cancer ( p  = 0.0017), endometrial\ncancer ( p  = 0.072), and thyroid cancer ( p  = 0.015) ( www.proteinatlas.org ). Similarly,  5  inhibits Importin-4 in 2008 ovarian\ncancer cells. Importin-4 executes nuclear internalization of VDR.\nVDR and Importin-4 colocalize in ovarian cancer tissues ( Figure S4a ), and we postulated that targeting\nthe VDR/Importin-4 axis may be crucial to controlling VDR/Importin-4-orchestrated\nmalignancies because, similar to VDR, Importin-4 mRNA expression independently\npredicts poor prognosis in neuroblastoma ( Figure S4b ) and hence the need to inhibit it. Further,  5  inhibits MYCN expression in neuroblastoma cells. MYCN expression\nis altered in over 70% of human cancers 38 − 45  and predicts poor prognoses in ovarian cancer ( p  = 0.0033) and neuroblastoma ( Figure S7 ). MYCN-driven cancers are aggressive and chemoresistant and await\na targeted therapy. 20 − 22  VDR is a key regulator of MYCN expression, 19  and therefore, targeting the VDR/MYCN axis can\nbe exploited to control such malignancies. For example,  5  treatment blocked the growth of TH-MYCN transgenic neuroblastoma  in vivo  and its spheroids  in vitro . Next,\nwe examine the effects of  5  on immune cells in TH-MYCN\nmice as a measure of off-target effects because immune cells express\nVDR and function calcitriol-dependently. The flow cytometric analysis\nof the tumors and immune cells isolated from mice carrying TH-MYCN\ntumors that were treated with vehicle/ 5  showed unaltered\npopulations of CD45 +  CD4 + , CD8 + ,\nmacrophages, patrolling monocyte, DCs, and CD11b +  cells\nin mice ( Figure S8 ), suggesting that VDR\ninhibition by  5  spares immune cells, which is notable\nbecause chemotherapies often cause indiscriminate cytotoxicity against\nnormal hematopoietic cells, imposing life-threatening side effects.\nFinally, notwithstanding the challenges associated with the currently\nknown VDR agonists, the functions of VDR and its agonists remain an\nongoing inquiry for cancer treatment. Notably, Sherman et al. 50  showed that as an adjuvant Vitamin-D reprograms\ntumor stroma transcriptionally and enables chemotherapeutic responses\nin pancreatic ductal adenocarcinoma (PDA). Similarly, while the role\nof Vitamin-D/VDR in upregulation of PD-L1 on ovarian cancer (unpublished\ndata) and leukemia and head and neck cancer cells 23  is concerning in the context of malignancies, there lies\na very promising opportunity to use  1  to convert a cold-tumor\ntype like PDA and ovarian cancer into a PD-L1-enriched hot-tumor type\nthat can be better targeted by immune checkpoint antibodies and/or  5 .\nCompound  5  differs significantly from\nthe previously\nreported VDR antagonists in that it can be synthesized in only two\nsteps from readily available raw materials. This has enabled us to\ndemonstrate the efficacy of VDR antagonists  in vivo . It is anticipated that the data presented in this study will contribute\nto the creation of easily synthesized VDR antagonists structurally\nsimilar/dissimilar to  5 .\n\nReagents and solvents were purchased from\ncommercial sources without further purification. The final compounds\nwere purified by preparative thin-layer chromatography (Analtech #Z513059).\nAll compounds were >95% pure by HPLC analysis. The progress of\nreactions\nwas monitored by thin-layer chromatography (TLC). NMR spectra were\nobtained from a 400 or 600 MHz Bruker spectrometer. Electrospray ionization\nmass spectrometry (ESI-MS) was performed on an Agilent 1100 LC-MS\nspectrometer. Melting points were determined with a Yanagimoto hot-stage\nmelting point apparatus and are uncorrected. Purity of  5  was analyzed by a Dionex UltiMate 3000 LC system.\nTo a solution\nof 7DHC ( 6 ) (Sigma Aldrich, 300 mg, 0.7 mM) in dichloromethane\nwas added  N -methyl-1,2-4-triazolinedione (87 mg,\n0.7 mM) at 0 ° C. During the following 4–5 h,\nthe pink color of the reaction medium was discharged. The separated\nproduct (adduct) was filtered, washed with hexane (10 mL × 5),\nand dried in vacuum in a desiccator overnight. Weight: 205 mg (52%).\nNMR assignments:  1 H NMR (600 MHz, DMSO) δ 6.33 (d,  J  = 8.3 Hz, 1H), 6.23 (dd,  J  = 8.3, 0.8\nHz, 1H), 4.63 (s, 1H), 4.04 (dq,  J  = 10.3, 5.3 Hz,\n1H), 2.94 (ddd,  J  = 14.1, 5.0, 1.4 Hz, 1H), 2.78\n(s, 3H), 2.46–2.38 (m, 1H), 2.08 (dd,  J  =\n12.4, 6.5 Hz, 1H), 1.98–1.93 (m, 1H), 1.88–1.79 (m,\n2H), 1.67 (dd,  J  = 10.4, 5.6 Hz, 1H), 1.65–1.60\n(m, 1H), 1.57–1.50 (m, 2H), 1.53–1.41 (m, 2H), 1.41–1.23\n(m, 2H), 1.23–1.06 (m, 6H), 1.05–0.97 (m, 1H), 0.91\n(d,  J  = 6.5 Hz, 3H), 0.85 (d,  J  =\n6.6 Hz, 3H), 0.84 (d,  J  = 6.6 Hz, 3H), 0.83 (s, 3H),\n0.73 (s, 3H).  13 C NMR (150 MHz, DMSO) δ 150.02, 147.64,\n135.60, 128.01, 65.29, 64.84, 64.03, 54.67, 52.37, 49.14, 43.39, 40.28,\n38.88, 37.72, 35.37, 34.67, 34.18, 33.61, 29.80, 27.36, 27.12, 24.64,\n23.15, 22.63, 22.42, 22.37, 21.71, 18.71, 17.07, 12.59.\nTo a stirred\nsolution of bromoacetic acid (36 mg, 0.13 equivalent)\nin anhydrous dichloromethane (DCM) maintained in an ice bath was added\nDCC (87 mg, 0.16 equivalent) and purged with nitrogen. The reaction\nmixture was stirred for 10 min. To the suspension formed was added\nadduct ( 7 , 100 mg, 0.2 mM) and stirred. A catalytic amount\nof DMAP was also added and stirred overnight, during which the temperature\nof the reaction mixture was allowed to rise to room temperature. Dichloromethane\nwas removed using the Buchi rotavapor, and the crude product obtained\nwas purified using a preparative thin-layer chromatography plate.\nThe band containing the product was collected, and the compound was\nstripped off the silica gel by washing with MeOH/DCM (9:1). The solvent\nwas removed using a rotary evaporator, and the compound ( 5 , MeTC7) was collected after drying under a vacuum in a desiccator\nas an off-white powder (83 mg, 67%) and stored at −20 °C.\nPurity of  5  was analyzed by a Dionex UltiMate 3000 LC\nsystem using Develosil 250 × 4.6 mm 100Diol-5, 5 μm LC\ncolumn. A binary solvent system with solvent A (0.1% formic acid in\nwater) and solvent B (0.1% formic acid in acetonitrile) was used with\na linear gradient of 0% B to 50% B from 0 to 5 min; 50% B to 70% B\nfrom 5 to 7 min; isocratic elution of 70% B from 7 to 8 min; linear\ngradient of 70% B to 80% B from 8 to 10 min; 80% B to 85% B from 10\nto 15 min; 85% B to 100% B from 15 to 30 min; 100% B to 50% B from\n30 to 36 min at a flow rate of 1 mL/min. NMR assignments.  1 H NMR (CDCl 3 , 600 MHz) δ 6.35 (d,  J  = 8.3 Hz, 1H), 6.14 (d,  J  = 8.3 Hz, 1H), 5.55 (tt,  J  = 11.0, 5.4 Hz, 1H), 3.82 (s, 2H), 3.2 (ddd,  J  = 13.8, 5.1, 1.4 Hz, 1H), 2.95 (s, 3H), 2.55–2.48 (m, 1H),\n2.28–2.22 (m, 1H), 2.21–2.16 (m, 1H), 2.12–1.99\n(m, 3H), 1.78–1.64 (m, 4H), 1.57–1.47 (m, 2H), 1.47–1.20\n(m, 8H), 1.18–1.08 (m, 3H), 1.07–1.02 (m, 1H), 0.95\n(s, 3H), 0.92 (d,  J  = 6.6 Hz, 3H), 0.87 (d,  J  = 6.6 Hz, 3H), 0.86 (d,  J  = 6.6 Hz, 3H),\n0.77 (s, 3H).  13 C NMR (150 MHz, CDCl 3 ) δ\n166.20, 150.62, 148.11, 134.72, 129.20, 72.59, 64.87, 64.70, 55.04,\n52.79, 49.26, 43.95, 40.92, 39.46, 38.16, 35.86, 35.27, 33.55, 30.65,\n28.03, 27.51, 26.19, 25.64, 25.05, 23.69, 23.19, 22.81, 22.57, 22.38,\n18.94, 17.37, 12.93. HRMS: Calculated for [M + H] + : 618.2828,\nfound: 618.2882. Purity of  5  was assessed by both elemental\nanalysis and HPLC. Elemental analysis: calculated for C 32 H 48 BrN 3 O 4 : C, 62.13; H, 7.82; N,\n6.79. Found: C, 62.288; H, 7.718; N, 6.712. HPLC: Retention time (RT):\n19.62. Purity: 98.2%. Mp 177.1–178.0.  1 H, 13 C, 1 H– 1 H COrrelated SpectroscopY (COSY),\nnuclear Overhauser effect spectroscopy (NOESY), multiplicity edited\nheteronuclear single quantum coherence (HSQC), heteronuclear multiple\nbond correlation (HMBC), and selective HMBC spectrograms of the adduct\nand  5  (each in CDCl 3 , at 600 MHz) are provided\nin the Supporting Information Section ( Figure S1 ).\nThe assay 27 , 28  was conducted in 384-well black\npolystyrene microplates (Corning,\n#3573) using 20 μL of buffer per well (25 mM PIPES, 50 mM NaCl\n(Fisher), and 0.01% NP-40, at pH 6.75), 0.1 μM VDR-LBD, 17.5\nnM Alexa Fluor 647-labeled SRC2-3, and 10 nM, 1,25(OH) 2 D 3  or 5 μM PPARγ-LBD, Texas Red-labeled DRIP2\n(7 nM) and rosiglitazone (1 μM). Then, 10 mM stock solutions\nof synthesized compounds made in DMSO were serially diluted (1:2)\nand added with a Tecan Freedom EVO liquid handling system using a\n50H hydrophobic-coated pin tool that carried 100 nL (V&P Scientific).\nAfter 2 h of incubation, fluorescence polarization was detected at\nemission/excitation wavelengths of 635/685 nm (Alexa Fluor 647) and\n596/615 nm (Texas Red). Three independent experiments were carried\nout in quadruplicate, and data were analyzed using nonlinear regression\nwith a variable slope (GraphPrism).\nHuman embryonic\nkidney (HEK) 293T cells were cultured in 75 cm2 flasks (CellStar)\ncoated in matrigel (BD Bioscience, #354234). Cells were grown in Dulbecco’s\nmodified Eagle’s medium (DMEM)/high glucose (Hyclone, #SH3024301)\nmedia to which nonessential amino acids (Hyclone, #SH30238.01), 10\nmM HEPES (Hyclone, #SH302237.01), 5 × 10 6  units of\npenicillin and streptomycin (Hyclone, #SV30010), and 10% of heat-inactivated\nfetal bovine serum (Gibco, #10082147) were added. For the assay, cells\nat 70–80% confluency were transfected by lipid-based methods,\nwhere 2 mL of untreated DMEM/high glucose media (without additives)\ncontaining 0.7 μg of VDR-CMV plasmid, 16 μg of a CYP24A1-luciferase\nreporter gene, lipofectamine LTX (75 μL, Life Technologies,\n#15338020), and PLUSTM reagent (25 μL) were added to the flask.\nAfter 16 h of incubation at 37 °C with 5% CO 2 , the\ncells were harvested with 0.05% Trypsin (Hyclone, #SH3023601) and\nadded to sterile white, optical-bottom 384-well plates (NUNC, #142762),\nplates that were pretreated with a 0.25% matrigel solution. To each\nwell, 20 μL of cells was added to yield a final concentration\nof 15,000 cells per well. After 4 h, plated cells were treated with\ncompounds in DMSO solution using a Tecan Freedom EVO liquid handling\nsystem with a 50H hydrophobic-coated pin tool. In the competitive\ninhibition assay, 1,25(OH) 2 D 3  (10 nM) was also\nadded to the assay wells containing  5 . After 16 h of\nincubation at 37 °C with 5% CO 2 , 20 μL of Bright-Glo\nLuciferase assay kit (Promega, Madison, WI) was added to each well\nand the luminescence was read. At least two independent experiments\nwere performed in quadruplicate, and data were analyzed using nonlinear\nregression with variable slope (GraphPrism).\nFluorescence polarization\nbinding assays were done with an INFINITE\npro200. 30  The measurements were performed\nin 1% DMSO buffer (pH 7.9, 10 mM HEPES, 150 mM NaCl, 2 mM MgCl 2 ). To a 384-well plate (Greiner 784076), RXRα-LBD (10\nμL, 0.5 μM final concentration), CBTF-BODIPY (5 μL,\n0.3 μM final concentration), and  5  (5 μL,\n32, 16, 8, 4, 2, 1, 0.5, 0.25, 0.125, 0.0625, 0.03125 μM final\nconcentration) were added, and the plate was incubated at 25 °C\nfor 1 h. The excitation and emission wavelengths were read at 485\nnm and 535 nm, respectively. The IC 50  value of each test\ncompound was calculated using Prism 8.\nThe available\ncrystal structure of Vitamin-D receptor (VDR) with  1  (PDB\ncode: 1DB1) 31  provided the platform for structural modeling\nand studies. Most of the modeling and simulation were carried out\nusing the modeling suite from Schrödinger 2021. 32  The crystallographic waters were removed to\navoid conformational discrepancies associated with water sampling\nduring the simulations. Overall, the implicit-water model was used\nduring minimization, conformational search, molecular and induced-fit\ndocking, loop refinement, and energetic calculation. A restrained\nminimization job with an RMSD constraint of 0.3 Å was carried\nout on the preprocessed structure using the OPLS3e force field to\nfurther refine the structure. The starting structures of  1  and  5  were obtained by performing a 5000-step conformational\nsearch with 0.05 kJ/mol convergence criteria using the Polak–Ribiere\nConjugate Gradient (PRCG) method using the MMFF force field. The chemical\nstructure of  5  is larger in size and conformationally\nrigid molecule than  1 . To handle this large and rigid\nmolecule, the induced-fit (IF) docking strategy was implemented. Therefore,\nthe IF docking protocol 33 , 34  was used to conduct\ndocking of  5  to the  1  site in VDR followed\nby side-chain refinement through Prime 32  to allow receptor flexibility according to the binding mode. Overall\ndocking strategies were checked by reference molecule  1 , and several docked poses were generated, and the best receptor- 5  docked complex was selected for the final minimization in\nusing the OPLS3e force field to relax and optimize to reveal the possible\nbinding mode of  5 .\nSKOV-3\n(ATCC, HTB77), OVCAR-3 (ATCC, HTB-161),\nOVCAR-8 (inherited from Laurent Brad’s previous laboratory),\nand CAOV-3 (ATCC, HTB75) ovarian cancer cells were grown in complete\nDMEM media (Gibco, 11965). IGROV-1 (Sigma, SCC203) and 2008 (kindly\nprovided by Dr. François X. Claret, University of Texas M.D.\nAnderson Cancer Center) ovarian cancer cells were grown in complete\nRPMI medium (Gibco, 22400). ES2 (ATCC, CRL-1978) was grown in McCoy’s\n5A complete medium (ATCC, 30-2007). BE(2)C (ATCC, CRL-2268), SH-EP1\n(ATCC, CRL-2269), SH-SY5Y (ATCC, CRL-2266), KELLY (Sigma, 92110411),\nSK-N-AS (Sigma, 94092302), and LAN-5 (COG,  http://www.cogcell.org ) neuroblastoma\ncell-lines were maintained in RPMI1640 media (Gibco, 11875) supplemented\nwith 10% heat-inactivated FBS. TH-MYCN+/+ cells were derived by mechanical\ndissociation of tumors obtained from TH-MYCN homozygous mice 51 − 53  and were maintained in RPMI1640 media (Gibco, 11875) supplemented\nwith 20% heat-inactivated FBS, 10–5 mM 2-mercaptoethanol, 1\nmM sodium pyruvate, and 1× nonessential amino acids (Gibco, 11140076).\nAll animal experiments were conducted at the\nUniversity of Rochester under the approval of the Institutional Animal\nCare and Use Committee (IACUC). NSG mice, 6–8 weeks old, bred\nin-house were used in ES2, BE(2)C, and SH-SY5Y xenograft studies.\nTH-MYCN hemizygous mice (129×1/SvJ-Tg(TH-MYCN)41Waw/Nci) 51 − 53  were initially obtained from the NCI Mouse Repository (strain code\n01XD2) and maintained in a 129×1/SvJ background through cross-breeding\nwith either wild-type 129×1/SvJ mice obtained from The Jackson\nLaboratory (stock number 000691) or other TH-MYCN hemizygous mice.\nTH-MYCN homozygous mice were identified through genotyping as previously\ndescribed. 51 − 53  All mice were maintained on a breeder diet (Labdiet\n5021), and tumor-bearing mice were further supplemented with Diet\nGel 67A (ClearH2O).\nSurvival\nanalyses of patients\ndiagnosed with pancreatic, lung, bladder, esophageal, and bladder\ncancers as well as neuroblastoma and other malignancies ( Figure S2 ) were generated by analyzing the mRNA\ndata available at the R2-Genomics Analysis and Visualization platform\n( http://hgserver1.amc.nl ) or the Human Protein Atlas ( https://www.proteinatlas.org/ ). Best system-recommended cutoffs\nwere opted. Databases analyzed for this study include the following:\nbladder cancer: Higlund-308-custom-ilmnht12v3; breast cancer: TCGA-1097-rsem-tcgars;\ncervical cancer: TCGA-305-rsem-tcgars; Glioma: TCGA-540-Mas5.0-u133a;\nliver cancer: TCGA-371-rsem-tcgars; lung cancer: Bild-114-Mas5.0-u133p2;\nneuroblastoma: Virsteeg-88-Mas5.0-u133p2; ovarian cancer: Mcdonald-45\n(fRMA-u133p2), Wong-77 (fRMA-u133p2), Mechta-Grigoriou-107-Mas5.0-u133p2;\nand pancreatic cancer: Badea-78(Mas5.0-u133p2), Wang-51 (Mas5.0-u133p2),\nTCGA-178-rsem-tcgars, and Yeh-132-custom-4hm44k.\nViability of ovarian cancer and neuroblastoma\ncell-lines exposed to  5  treatment was determined by the\nCellTiter 96 AQueous One Solution assay (Promega Corp, Madison, WI).\nCells were seeded into a 96-well plate at 5,000 cells/100 μl/well\ndensity in a complete cell culture medium, allowed to attach overnight\nat 37 °C with 5% CO 2 , in a humidified incubator, and\nwere treated with a complete medium containing the indicated concentration\nof  5  dissolved in DMSO ( Figures  4 d and  6 a). The final\nconcentration of DMSO did not exceed 0.2% (v/v). At planned hours,\nexisting media were replaced with fresh RPMI media containing the\nMTS reagent (1:10 dilution) and incubated for 2–4 h. Absorbance\nwas read at 490 nm using the iMark microplate reader (BioRad). Viability\nof HepG2 and HEK293T cells after  5  treatment was measured\nby the CellTiter-Glo (Promega) assay ( Figure  5 a,b). Cells were plated in quadruplicate\nin 384-well plates and treated with indicated concentrations of  5 , 7DHC ( 6 ), and 7DHC-adduct ( 7 ).\nCells were incubated for 18 h at 37 °C. CellTiter Glo (Promega)\nwas added. The number of live cells was quantified by luminescence\nusing a Tecan M1000 plate reader. DMSO (negative) was used as the\ncontrol. Data were analyzed using nonlinear regression with the variable\nslope (GraphPadPrism) assay.\nNeuroblastoma tissues\nwere fixed in 10% neutral buffered saline for several days and then\ndehydrated into paraffin using a Sakura VIP tissue processor and Sakura\nTissue Tek 5 embedding center. Sections of 5–10 μm in\nthickness were cut using a Leica RM2265 microtome. Immunohistochemical\nstains were performed using the GBI Polink-2 antirabbit HRP Plus Detection\nSystem (GBI International, D39) or the Mouse-on-Mouse HRP-Polymer\nBundle (BioCare Medical) and were counterstained with hematoxylin.\nPrior to primary antibody addition, sections were rehydrated, followed\nby 30 min antigen retrieval in sodium citrate buffer pH 6.0, and blocked\nof endogenous peroxidase with hydrogen peroxide. Primary antibodies\nused for immunohistochemistry were mouse rabbit anti-VDR (Abcam, ab3508),\nmouse anti-VDR (Santa Cruz Biotechnology, SC-13133), mouse anti-MYCN\n(Santa Cruz Biotechnology, SC-53993), rabbit anti-tyrosine hydroxylase\n(TH) (Millipore, AB152), normal rabbit IgG (Millipore, 12-370), and\nnormal mouse IgG (Millipore, 12-371). Slides were visualized using\nan Olympus BX41 light microscope and imaged with an Olympus DP70 camera.\nPhotographs were captured using CellSens digital software.\nBriefly, sixteen-bit images were\nacquired with a Nikon E800 microscope (Nikon Inc., Melville, NY) using\na 40× PlanApo objective. A Spot II digital camera (Diagnostic\nInstruments, Sterling Heights MI) was used to acquire the images.\nThe camera’s built-in green filter was used to increase the\nimage contrast. Camera settings were based on the brightest slide.\nImages were acquired with the same settings. Image processing and\nanalysis were performed using iVision (BioVision Technologies, version\n10.4.11, Exton, PA.) image analysis software. Positive staining was\ndefined through intensity thresholding, and integrated optical density\n(IOD) was calculated by examining the thresholded area multiplied\nby the mean. All measurements were performed in pixels. Confocal images\nwere acquired with a Nikon C1si confocal (Nikon Inc., Melville, NY.)\nusing diode lasers 402, 488, and 561. Serial optical sections were\nperformed with EZ-C1 computer software (Nikon Inc., Melville, NY).\nZ series sections were collected at 0.3 μm with a 40× PlanApo\nlens and a scan zoom of 2. The gain settings were based on the brightest\nslide and kept constant between specimens. Deconvolution and projections\nwere done in Elements (Nikon Inc. Melville, NY) computer software.\nES2, SH-SY5Y, and BE(2)C cells\nisolated from 70 to 80% confluent Petri dishes were spun down (1000\nrpm, 5 min). Media was removed, and cells (calculated 250,000/mice\nfor ES2 and 1 million/mice for SH-SY5Y and BE(2)C) were suspended\nin cold matrigel/serum-free RPMI media mix (1:1) and implanted subcutaneously\nin the right flank of the NSG mice. Prior to inoculation, NSG mice\nwere shaved at the inoculation site using a clean shaving machine,\nand skin was disinfected and cleaned using commercially available\nalcohol swabs. SKOV-3, D283, and PANC-1 cells were grown to semiconfluence\nin complete DMEM media. BXPC-3 cells were cultured in complete RPMI\nmedia to 70–80% confluence. Trypsinized cells were harvested,\ncentrifuged, and suspended in precooled matrigel/DMEM media (1:1)\nand subcutaneously implanted in nude mice (SKOV-3) or NSG mice (D283,\nPANC-1, BXPC-3) each at 1 million cells/animal rate. Once tumors became\npalpable, mice were treated with vehicle or  5 . Tumors\nin each case were allowed to grow until the volume [(length ×\nwidth 2 )/0.5] in one or more mice reached 2000 mm 3 , and then the entire group of animals was sacrificed. Tumor sizes\nand animal weights were recorded periodically except in BE(2)C and\nES2 animals, which necessitated alternate day monitoring due to rapid\ntumor growth. Tumors from the control and drug groups were harvested,\nweighed, snap-frozen, and stored in liquid nitrogen.\nTH-MYCN hemizygous\nmice (129×1/SvJ-Tg(TH-MYCN)41Waw/Nci) were\ninitially obtained from the NCI Mouse Repository (strain code 01XD2)\nand maintained in a 129×1/SvJ background though cross-breeding\nwith either wild-type 129×1/SvJ mice obtained from The Jackson\nLaboratory (stock number 000691) or other TH-MYCN hemizygous mice.\nTH-MYCN homozygous mice were identified through genotyping as previously\ndescribed. 52 , 53  All mice were maintained on a\nbreeder diet (Labdiet 5021), and tumor-bearing mice were further supplemented\nwith Diet Gel 67A (ClearH2O). Control mice ( n  = 7)\nand  5  (10 mg/kg,  n  = 8) mice were treated\nintraperitoneally with indicated doses. Mice in the control and  5  (10 mg/kg) group received six treatments in total, whereas\nthe mice in the  5  (100 mg/kg) group were given just three\ntreatments to see the effect of escalated drug dose on the safety\nof animals at 10× dose and to monitor for changes in the tumor\nburden. The tumor burden in each mouse was estimated using ultrasound\nimaging instrumentation as described below.\nTumors in vehicle/drug-treated\ngroups were visualized by abdominal ultrasound using a Vevo 3100 Imaging\nSystem and MX550D transducer (FUJIFILM VisualSonics, Inc). Animals\nwere anesthetized (1–3% isoflurane and oxygen mixture) and\nrestrained on a heated stage with monitors for respiration and heartbeat.\nVentral hair was removed with a depilatory cream prior to monitoring\nwith an ultrasound probe. The 3D volume measurements were carried\nout using Amira 6.1 software with an XImagePAC extension (FEI).\nTo analyze the statistical difference\nbetween vehicle and  5 -treated ES2 xenograft tumors ( Figure  4 f), a repeated-measures\nanalysis of variance was performed using maximum likelihood estimation\nwith group, day, and the interaction between group and day as fixed\neffects. The correlation of repeated measures on the same subject\nover time was handled using an unstructured covariance, which was\nallowed to vary by treatment condition. Model assumptions were verified\ngraphically. Analysis was conducted using SAS v9.4 Proc Mixed (Cary,\nNC). Assumptions made by the mixed model analysis were verified by\nexamining the distribution of residuals, or unexplained variation.\nIdeally, the residuals are approximately normally distributed with\na mean of zero and no obvious patterns. Finally, we used a nonparametric\ntest (Wilcoxon rank sum test) to compare the tumor volumes between\ngroups at each time point. Results corroborated those seen with the\nregression model, which makes more assumptions. The statistical differences\nbetween vehicle and  5 -treated SKOV-3 xenograft tumors,\naverage animal weights, and tumor sizes were compared between the\ncontrol and  5  groups at the baseline by Student’s  T -test. Weights and tumor sizes were compared by group over\nthe observation period using linear mixed effect regression. Random\nintercepts and slopes were included to model within-animal response\ntrajectories. Group differences in the rate of weight change or tumor\ngrowth were tested by an interaction term between the treatment group\nand day of treatment. Residuals were examined to assess model fit.\nAnimal survival was plotted using the Kaplan–Meier method.\nTwo-tailed p-values less than 0.05 were considered statistically significant\n( Figure  5 c–e).\nThe difference between the % change in tumor volumes of D283 medulloblastoma\nand pancreatic cancer (PANC-1, BXPC-3) and neuroblastoma BE(2)C and\nSH-SY5Y xenografts treated with control or  5  treatment\nwas analyzed by Student’s  T -test ( Figure  5 f–h). Tumor\nweights in BE(2)C and SH-SY5Y xenografts were compared by Student’s  T -test. TH-MYCN tumor sizes and weights in the control and\ntreatment groups were compared by Student’s  T -test ( Figure  8 b,c).","source_license":"CC-BY-4.0","license_restricted":false}