Design, synthesis, and structure-activity relationship studies of fluorescent inhibitors of cycloxygenase-2 as targeted optical imaging agents.

Cycloxygenase-2 (COX-2) is an attractive target for molecular imaging because it is an inducible enzyme that is expressed in response to inflammatory and proliferative stimuli. Recently, we reported that conjugation of indomethacin with carboxy-X-rhodamine dyes results in the formation of effective, targeted, optical imaging agents able to detect COX-2 in inflammatory tissues and premalignant and malignant tumors (Uddin et al. Cancer Res. 2010, 70, 3618-3627). The present paper summarizes the details of the structure-activity relationship (SAR) studies performed for lead optimization of these dyes. A wide range of fluorescent conjugates were designed and synthesized, and each of them was tested for the ability to selectively inhibit COX-2 as the purified protein and in human cancer cells. The SAR study revealed that indomethacin conjugates are the best COX-2-targeted agents compared to the other carboxylic acid-containing nonsteroidal anti-inflammatory drugs (NSAIDs) or COX-2-selective inhibitors (COXIBs). An n-butyldiamide linker is optimal for tethering bulky fluorescent functionalities onto the NSAID or COXIB cores. The activity of conjugates also depends on the size, shape, and electronic properties of the organic fluorophores. These reagents are taken up by COX-2-expressing cells in culture, and the uptake is blocked by pretreatment with a COX inhibitor. In in vivo settings, these reagents become highly enriched in COX-2-expressing tumors compared to surrounding normal tissue, and they accumulate selectively in COX-2-expressing tumors as compared with COX-2-negative tumors grown in mice. Thus, COX-2-targeted fluorescent inhibitors are useful for preclinical and clinical detection of lesions containing elevated levels of COX-2.

Introduction

Early detection is one of the critical challenges for contemporary clinical cancer care. Detection of small premalignant lesions and early stage primary tumors is crucial for effective cancer therapy. For this reason, interest in highly sensitive imaging techniques for clinical oncology has increased tremendously in recent decades. Many groups have studied the in vivo detection of tumors using contrast agents appropriate for various imaging modalities.[1−3] Methods have been developed for specific targeting in order to maximize the localization of ligands in the tumor and to minimize the uptake in the surrounding normal tissues to achieve high signal-to-noise ratios.[4−6] These approaches include targeted imaging of tumors using monoclonal antibodies that specifically bind to receptors on the tumor cells,[7,8] protease-activated near-infrared fluorescent probes for detection of tumors,[9] folate receptor-targeted in vivo imaging of tumors using a near-infrared (NIR) dye-folate conjugate,[10] and integrin-targeted imaging of lung and liver metastases using a 64Cu-1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA) dimeric arginylglycylaspartic acid peptide radiotracer.[11]

Cyclooxygenase (COX) isozymes catalyze the biotransformation of arachidonic acid into a wide variety of prostaglandins that are important biological mediators of inflammatory diseases.[12] COX-1 is constitutively expressed in most normal tissues, where it modulates housekeeping functions, such as hemostasis, vascular tone, and cytoprotection of the gastric mucosa.[13] COX-2 is absent or expressed at very low levels in most epithelial cells but is found at high levels in inflammatory lesions and many premalignant and malignant tumors. COX-2 mRNA is detected in the early stages of tumorigenesis, such as adenomatous polyps in the colon or Barrett’s esophagus.[16−19] COX-2 expression promotes tumor growth, angiogenesis, and metastasis of cancer cells.[14,15] Recent work has shown that selective COX-2 inhibitors are useful in the prophylaxis of various human cancers.[20,21]

The high level of COX-2 in premalignant and malignant tumors compared to surrounding normal tissue suggests that it is an ideal target for molecular imaging by radiolabeled or fluorescent COX-2 inhibitors. To test this hypothesis, we and other laboratories have studied imaging of COX-2 expression in inflammation and cancer using 123I- or 18F-labeled COX-2 selective inhibitors.[22−26] In addition, we recently described a conjugate-based conversion of the nonsteroidal anti-inflammatory drug (NSAID), indomethacin, into fluorescent carboxy-X-rhodamine (ROX) conjugates that are capable of inhibiting COX-2 selectively and potently in both purified protein and intact cells. These conjugates, which do not significantly inhibit COX-1, enabled the optical imaging of COX-2 in inflammatory lesions and COX-2-expressing tumors in vivo.[27] The present paper describes the details of the structure–activity relationship (SAR) studies performed for optimization of these lead compounds from a wide range of fluorescent conjugates of NSAIDs or COX-2-selective inhibitors (COXIBs). The conjugates were evaluated as COX-2-targeted imaging agents, and active molecules were validated in vitro and in vivo as COX-2-targeted agents in cells and tumors. The results provide critical background and further support for our previous report,[27] which reveals the first fluorescently labeled optical imaging agent validated for COX-2-targeted in vivo imaging of inflammation and cancer.

Experimental Procedures

Chemistry

Standard methods were utilized for the synthesis of fluorescent derivatives of NSAIDs and COXIBs. A wide variety of carboxylic acid-containing core compounds (e.g., indomethacin, an iodoindomethacin, an indolyl carboxamide analog of indomethacin (reverse indomethacin), flurbiprofen, ketoprofen, or a carboxylic acid derivative of celecoxib) were tethered through a series of alkyl, aryl piperazinyl, or polyethylene glycol linkers to a diverse range of bulky organic fluorophore moieties. The fluorophores included dansyl, dabsyl, coumarin, fluorescein, rhodamine, alexa-fluor, nile blue, cy5, cy7, near IR, and IR dyes, as well as some lanthanide chelators (Figure 1). Synthetic procedures, in addition to analytical and spectroscopic characterization of all compounds, are described in Supporting Information.

Figure 1

Fluorescent COX-2 probes. Carboxylic acid containing core NSAIDs, COXIBs, or their derivatives were used as building blocks for conjugate chemistry. These compounds were tethered through a series of alkyl, piperazine, and other linkers to a diverse range of fluorophore moieties to synthesize fluorescent conjugates as COX-2-selective fluorescent imaging agents.

Fluorometry

Steady state fluorescence excitation and emission spectra were determined for each compound with a Spex 1681 Fluorolog spectrofluorometer, equipped with a 450 W xenon arc lamp. The excitation and emission monochromator slit widths were 1–2 mm. The solvent used was pH 7 buffer.

Inhibition Assay Using Purified COX-1 and COX-2

Inhibition of purified ovine COX-1 or mouse COX-2 by test compounds was assayed by a previously described method,[27] which quantifies the conversion of [1-14C]arachidonic acid to [1-14C]prostaglandin products. Reaction mixtures of 200 μL consisted of hematin-reconstituted protein in 100 mM Tris-HCl, pH 8.0, 500 μM phenol, and [1-14C]arachidonic acid (50 μM, ∼55–57 mCi/mmol, Perkin-Elmer). For the time-dependent inhibition assay, hematin-reconstituted COX-1 (44 nM) or COX-2 (66 nM) was preincubated at 25 °C for 17 min and 37 °C for 3 min with varying inhibitor concentrations in dimethylsulfoxide followed by the addition of [1-14C]arachidonic acid (50 μM) for 30 s at 37 °C. Reactions were terminated by solvent extraction in diethyl ether/methanol/1 M citrate buffer, pH 4.0 (30:4:1). The phases were separated by centrifugation at 2000g for 2 min, and the organic phase was spotted on a thin-layer chromatography plate (EMD Kieselgel 60, VWR). The plate was developed in ethyl acetate/methylene chloride/glacial AcOH (75:25:1) at 4 °C. Radiolabeled products were quantified with a radioactivity scanner (Bioscan, Inc., Washington, DC. The percentage of total products observed at different inhibitor concentrations was divided by the percentage of products observed for protein samples preincubated for the same time with dimethyl sulfoxide.

Cell Culture and In Vitro Intact Cell Metabolism Assay

Inhibition of COX-2 in intact cells by test compounds was assayed by a previously described method.[27] Briefly, RAW264.7, ATCC TIB-71 murine macrophage-like cells (passage number 8–15, mycoplasma negative by a polymerase chain reaction detection method) were grown in Dulbecco’s Modified Eagle Medium (DMEM) + 10% heat-inactivated fetal bovine serum to 40% confluence (6-well plates, Sarstedt). The cells were activated for 7 h in 2 mL serum-free DMEM with 200 ng/mL bacterial lipopolysaccharide (Calbiochem) and 10 U/mL interferon gamma (Calbiochem) to induce COX-2 expression. Human 1483 head and neck squamous cell carcinoma (HNSCC) cells (passage 8–18, mycoplasma negative by a polymerase chain reaction detection method) were cultured in DMEM/F12 + 10% fetal bovine serum + antibiotic/antimycotic in 6-well plates to 60% confluence. Serum-free medium (2 mL) was added, and the cells were treated with inhibitor dissolved in DMSO (0–5 μM, final concentration) for 30 min at 37 °C followed by the addition of [1-14C]-arachidonic acid [10 μM, ∼55 mCi/mmol] for 20 min at 37 °C. Reactions were terminated and analyzed by thin layer chromatography as described above.

Fluorescence Microscopy of 1483 HNSCC Cells

Fluorescence imaging of human 1483 HNSCC cells by compound 58 was performed by a previously described method.[27] Briefly, human 1483 HNSCC cells were grown to 60% confluence. The cells were incubated in 2.0 mL Hank’s balanced salt solution (HBSS)/Tyrode’s with 200 nM compound 58 for 30 min at 37 °C. The cells were then washed briefly three times and incubated in HBSS/Tyrode’s for 30 min at 37 °C. Following the required washout period, the cells were imaged in 2.0 mL fresh HBSS/Tyrode’s on a Zeiss Axiovert 25 Microscope with the propidium iodide filter (0.5–1.0 s exposure, gain of 2). All treatments were performed in duplicate dishes in at least three separate experiments. To block the COX-2 active site, the cells were preincubated with 5 or 10 μM indomethacin for 20 min prior to the addition of the test compound.

Establishment of Xenograft Tumors in Nude Mice

Human 1483 HNSCC cells and HCT116 colorectal carcinoma cells were used to grow tumor xenografts in nude mice using a previously described method.[27] Female nude mice, NU-Fox1nu, were purchased at 6–7 weeks of age from Charles River Laboratories. Human 1483 HNSCC cells and HCT116 colorectal carcinoma cells were trypsinized and resuspended in cold PBS containing 30% Matrigel such that 1 × 106 cells in 100 μL were injected subcutaneously on the left flank. The HCT116 and 1483 xenografts required only 2–3 weeks of growth.

In Vivo Imaging of Nude Mice with Xenografts

Fluorescence imaging of tumors by test compounds was performed by a previously described method.[27] Female nude mice bearing medium-sized 1483 or HCT116 xenograft tumors on the left flank were dosed by intraperitoneal injection with 2 mg/kg compound 58. The animals were lightly anesthetized with 2% isoflurane for fluorescence imaging in the Xenogen IVIS 200 with the DSRed filter at 1.5 cm depth and 1 s exposure (f2).

Results

Synthesis of Fluorescent COX-2 Inhibitors

The synthesis of NSAID- or COXIB-diamide imaging agents targeted to COX-2 first required the conjugation of the carboxylate functional group of the NSAID or COXIB nucleus to a diamine linker. Diamide linkages were chosen rather than mixed amide-ester linkages to minimize the potential for hydrolysis in intact cells or in vivo. Selective amidation of only one of the two available amino groups present in the diamine tether necessitated protection of one of the groups. This was accomplished by the use of the mono tert-butoxycarbonyl (BOC)-protected alkyldiamine. Reaction of indomethacin with a series of mono BOC-alkyldiamines in the presence of ethyl-1-[3-(dimethylamino)propyl]-3-ethylcarbodiamide followed by treatment with HCl (gas) gave the corresponding indomethacin-alkylamine hydrochloride salts in high yield. Similarly, indomethacin-piperazine hydrochloride and indomethacin-phenylenediamine hydrochloride were synthesized by substituting mono BOC-alkyldiamine with mono BOC-piperazine or mono BOC-phenylenediamine, respectively. For synthesizing indomethacin-polyethylene glycolyl (PEG) carboxylic acid, t-butyl-PEG4-amine was used in place of mono BOC-alkyldiamine, followed by deprotection with trifluoroacetic acid at room temperature. As described above, other NSAIDs, COXIBs, or appropriate analogs were linked to a chosen tether (alkyldiamine, PEG, piperazine, or phenylene diamine) to form the corresponding conjugates having a terminal primary or secondary amine or a carboxylic acid group. The isothiocyanate, sulfonylchloride, or succinimidyl ester of the desired fluorophore was conjugated with the amino group of the tether-linked-NSAID or -COXIB using triethylamine as a base. Alternatively, formation of a carboxylamide from the reaction of a carboxylic acid with an amino-group required either ethyl-1-[3-(dimethylamino)propyl]-3-ethylcarbodiamide or N,N,N,N-tetramethyl-O-(N-succinimidyl)uronium tetrafluoroborate coupling reactions. Using this general strategy, 5-ROX-acid was activated using N,N,N,N-tetramethyl-O-(N-succinimidyl)uronium tetrafluoroborate in the presence of triethylamine at room temperature and coupled with the free amino group of the tether-linked-NSAIDs or COXIBs to afford the target fluorescent conjugates. All other fluorescent dyes were conjugated with the respective NSAID or COXIB scaffolds using a similar coupling strategy in good yields (60–70%). The structure of all compounds was established by NMR and mass spectrometry. HPLC analyses in two different solvent systems of all representative fluorescent compounds indicated a minimum purity of 96.0%. Most of the compounds were in excess of 98.5% purity. The synthetic procedure and compound characterization are described in detail in Supporting Information.

Evaluation of Fluorescent COX-2 Inhibitors Using Purified Enzymes

The inhibitory activity of test compounds against ovine COX-1 or mouse COX-2 was evaluated by a thin layer chromatography assay,[27] as briefly described in Experimental Procedures.

Indomethacin Conjugates of Blue and Green Fluorophores

IC50 values for the inhibition of purified COX enzymes by test compounds are listed in Table 1. Compound 1, a conjugate of indomethacin and coumarin tethered through an ethylenediamide linker, displayed selective COX-2 inhibition (Table 1). Chain length extension of the linker group of 1 to higher alkyl homologues or replacement of coumarin with 7-diethylaminocoumarin revealed significant increases in potency and selectivity against COX-2 (4–5). For example, the n-butydiamide-conjugate 5 was more potent and selective as a COX-2 inhibitor than 1. Conjugation of organic functionalities (i.e., N,N-dimethylamino cinnamic acid, sulfathiazole, sulfadimethoxine, mycophenolic acid, or biotin) afforded a series of compounds (6–12), of which the trans-cinnamyl (6) or sulfathiazolyl (7) conjugates displayed potent and selective COX-2 inhibition. Compared to compounds 6 or 7, the dansyl (13–15) or dabsyl (16–17) derivatives were more potent and selective for COX-2 inhibition (13, λex = 355 nm, λem = 493 nm). The n-butydiamide-tethered nitrobenzoxadiazole (NBD)-conjugate (18) also displayed selective COX-2 inhibition. However, shortening of the alkyl chain or incorporation of perphenazine with compounds 19 and 20 showed complete loss of COX inhibitory activity (20, λex = 492 nm, λem = 505 nm). Although a similar loss of COX activity was observed with fluorescein-conjugates (e.g., 21, 22, 24–26, 29–31), compounds 23, 27, and 28 displayed COX-2 inhibitory activity to some extent. These results suggested that transformation of indomethacin into fluorescent conjugates can lead to selective COX-2 inhibition; however, the length and nature of the tether and the structure of the fluorophore have a significant impact on the COX inhibitory potential of the conjugates.

Table 1

In Vitro Purified COX-1 and COX-2 Enzyme Inhibition Assay Data of Compounds 1—31

IC50 values were determined as described in Experimental Procedures. Assays were run in duplicate.

Indomethacin Conjugates of Red Fluorophores

The fluorescent derivatives of indomethacin with alexa fluor, tetramethylrhodamine, or bis-iodobenzylcarboxyrhodamine (compounds 32–35) showed no COX inhibitory activity, even at the higher concentrations (Table 2). However, COX-2 inhibitory activity was achieved with the PEG-containing tetraethyl 5- or 6-sulforhodamine conjugates 36 and 37. Poor COX inhibition was observed with 38 having an n-pentylsulfonamido n-butyldiamide linker (36, λex = 571 nm, λem = 593 nm). Selective COX-2 inhibition by the 5- or 6-ROX derivatives was extremely sensitive to the length and electronic properties of the linker moieties. For instance, the 5-ROX conjugate 39, containing an ethylenediamide linker, was a weak inhibitor of COX-2. As the alkyl chain length increased, COX-2 inhibitory potency and selectivity increased substantially, with the n-butyldiamide derivative 41 (Fluorocoxib A, λex = 580 nm, λem = 605 nm) exhibiting the best combination of COX-2 inhibitory potency and selectivity and identified as one of the most potent derivatives in the series.[27] A further increase in alkyl chain length resulted in a dramatic reduction in COX-2 inhibitory potency (e.g., 42 and 43). Interestingly, the phenylene derivatives 44 and 45 showed similar potencies to 41. Replacement of the n-butyldiamide linker of compound 41 with a piperazine linker yielded a very weak inhibitor (46) of COX-2. Likewise, indomethacin-6-ROX conjugates (47–56), linked through alkyl, piperazine, or phenylene tethers, showed similar potencies to those of the corresponding 5-ROX conjugates. Notably, the regioisomer 49 (Fluorocoxib B, λex = 581 nm, λem = 603 nm) exhibited selective COX-2 inhibition.27 However, fluorescent conjugates 57–64, containing bulkier rhodamine dyes, showed significantly lower potency or complete loss of COX-2 inhibitory activity with the exception of compound 58 (λex = 572 nm, λem = 595 nm), which displayed activity comparable to the 5- and 6-ROX conjugates. In summary, a four-carbon n-alkyl linker with 5- or 6-ROX provided the best balance for COX-2 inhibitory activity and selectivity of the fluorescent indomethacin conjugates.

Table 2

In Vitro Purified COX-1 and COX-2 Enzyme Inhibition Assay Data of Compounds 32–64

IC50 values were determined as described in Experimental Procedures. Assays were run in duplicate.

Indomethacin Conjugates of NIR Fluorophores

Conjugation of a nile blue dye with the carboxylic acid moiety of indomethacin tethered through a PEG4 linker afforded compound 65 (λex = 618 nm, λem = 670 nm), a perchlorate salt, which was identified as a poor COX-2 inhibitor (Table 3). A complete loss of COX inhibitory activity was observed with the Cy5 conjugate 66. Interestingly, the Cy7 conjugate 67 inhibited COX-2 with poor potency, while inhibitory activity but lack of selectivity was observed with the NIR641 conjugate (68), a chloride salt. No significant COX-2 inhibition was discernible with the ethyl- or propyldiamide-linked indomethacin conjugates of NIR664, such as compounds 69 and 70, respectively. Nonselective COX inhibition was observed with the piperazine-linked conjugate 71, an inner salt (zwitterion). However, a complete loss of COX activity was documented with the higher alkyl homologues (72, 73). A similar loss was observed with the NIR667 conjugates 74–78 (chloride salts). Conjugation of indomethacin with NIR700, an inner salt, afforded the selective COX-2 inhibitor 79 (λex = 710 nm, λem = 731 nm), which contained an n-butyldiamide linker. When the length of the linker was increased, conjugates 80–81 showed a dramatic loss of COX-2 inhibitory activity. Interestingly, a poor potency was achieved when the n-butyl linker was replaced by a phenylene-alkyl hybrid linker in compound 82, and no COX-2 inhibition was observed with the NIR782 or IRDye800 derivatives, compounds 83 and 84, respectively. A similar lack of activity was observed with DOTA-, Eu-Quinoline(F)-DOTA-, Eu-Quinoline(CH3)-DOTA-, or Gd-DOTA-derivatives (compounds 85–106), even at very high concentrations (Table 4). Thus, conjugation of indomethacin with a zwitterionic fluorophore tethered through an n-butyldiamide linker is most suitable to achieve COX-2 inhibitory activity. Highly polar organic functional groups, such as DOTA, or its analogous lanthanide chelators, are not suitable for COX-2 inhibitory activity of the conjugates.

Table 3

In Vitro Purified COX-1 and COX-2 Enzyme Inhibition Assay Data of Compounds 65–84

IC50 values were determined as described in Experimental Procedures. Assays were run in duplicate.

Table 4

In Vitro Purified COX-1 and COX-2 Enzyme Inhibition Assay Data of Compounds 85–106

IC50 values were determined as described in Experimental Procedures. Assays were run in duplicate.

Fluorescent Conjugates of Miscellaneous NSAIDs and COXIBs

Lack of isoform selectivity or complete loss of COX-2 inhibitory activity was often observed with fluorescent conjugates of NSAIDs and COXIBs other than indomethacin (Table 5). For instance, the iodoindomethacin-5-ROX conjugates 107 and 109 showed no COX-2 selectivity. A complete loss of COX inhibitory activity was observed with the 5-ROX or NIR664 conjugates 108, 110, and 111. We synthesized O-des-methyl- or 2-des-methyl-indomethacin-5-ROX conjugates 112 and 113, which were selective COX-2 inhibitors. However, the 2-des-methyl-, N-des-4-chlorobenzoyl-, or N-4-chlorobenzyl-indomethacin and 5- or 6-ROX conjugates 114–117, containing an n-butyldiamide linker, showed no COX-2 inhibition. A loss of COX-2 inhibitory activity was also observed with fluorescent derivatives of reverse-indomethacin, 118–122. An exception was identified with the reverse-indomethacin-NIR700 conjugate (123), which exhibited selective COX-2 inhibition with a moderate potency. A complete loss of COX-2 inhibitory activity was observed with the majority of fluorescent derivatives of celecoxib, flurbiprofen, or ketoprofen (compounds 124–139). However, conjugation of the celecoxib analog with nile blue or sulforhodamine (e.g., 124 and 125) resulted in selective COX-2 inhibition with moderate to poor potencies. The COX-2 inhibitory activity of indomethacin- or celecoxib-like building blocks (140–152) is delineated in Table 6. Although the derivatives of indomethacin and celecoxib, such as iodoindomethacin and carboxypropionyl celecoxib, retain the COX-2 inhibitory activity of the parent compound, their fluorescent derivatives were not always COX-2 inhibitors. So, unlike the parent indomethacin, conversion of indomethacin- or celecoxib-like molecules into fluorescent amide derivatives does not readily generate a family of highly selective COX-2 inhibitors.

Table 5

In Vitro Purified COX-1 and COX-2 Enzyme Inhibition Assay Data of Compounds 107–123

IC50 values were determined as described in Experimental Procedures. Assays were run in duplicate.

Table 6

In Vitro Purified COX-1 and COX-2 Enzyme Inhibition Assay Data of Compounds 124–152

IC50 values were determined as described in Experimental Procedures. Assays were run in duplicate.

Evaluation of Fluorescent COX-2 Inhibitors in Cell Culture

The ability of fluorescent derivatives to inhibit COX-2 in intact cells was assayed in activated RAW264.7 macrophages or 1483 HNSCC cells using previously reported methods,[27] as briefly described in Experimental Procedures. In this assay, many of the structurally diverse fluorescent compounds inhibited COX-2 in LPS-activated RAW264.7 cells with IC50 values in the nanomolar range (Table 7). For instance, the IC50 values for inhibition of prostaglandin synthesis by 41 and 58, were 0.31 and 0.34, μM, respectively. In 1483 HNSCC cells, which constitutively express COX-2, compounds 41 and 58 retained their COX-2 inhibitory activity with IC50 values of 0.09 and 0.38 μM, respectively. Thus, conjugation of an NSAID, indomethacin, with zwitterionic fluorophores, tethered through an n-butydiamide linker afforded fluorescent conjugates that were capable of passing through the cell membrane to inhibit COX-2 effectively.

Table 7

In Vitro Cell Line Assay Data of Promising Compounds

IC50 values were determined as described in Experimental Procedures for RAW264.7 macrophage-like cells.

IC50 values were determined as described in Experimental Procedures for human head and neck squamous cell carcinoma (HNSCC), 1483 cells. The NT designation indicates that cell line assay results were not tested.

In Vitro Optical Imaging of Cells

The human head and neck cancer cell line, 1483, which expresses high levels of COX-2,[28] was treated with compound 58. After washout, the cells were imaged by fluorescence microscopy and exhibited strong red fluorescence due to the accumulation of the rhodamine derivative (Figure 2A). Preincubation of the cells with the COX inhibitor indomethacin blocked uptake and labeling of 1483 cells by compound 58 (Figure 2B). These results are consistent with those reported previously for compound 41.[27]

Figure 2

Labeling of COX-2-expressing cells by compound 58. The experimental protocols are described in Experimental Procedures. (A) 1483 HNSCC cells treated with 200 nM compound 58 for 30 min. (B) 1483 HNSCC cells pretreated with 5 μM indomethacin for 20 min prior to compound 58 treatment.

In Vivo Optical Imaging of Tumor Xenografts

We evaluated the ability of compound 58 to target COX-2 in human tumor xenografts. Nude mice bearing 1483 or HCT116 xenograft tumors on the left flank were dosed by intraperitoneal injection with compound 58 (2 mg/kg). At 60 min postinjection, no fluorescence was observed in the tumor. Signal was detected in the COX-2 expressing 1483 tumors starting at 3 h (Figure 3A). In contrast to the selective uptake of 58 into 1483 xenografts, minimal uptake was observed at 3 h in HCT116 xenografts, a human colon tumor that does not express COX-2.[29] An ex vivo imaging was performed to confirm the in vivo imaging results and to verify the 1483 tumor signal was due to the tumor uptake of 58, and not from the nearby normal tissues or skin. A bright fluorescence signal was detected in the ex vivo 1483 tumor as compared with the HCT116 tumor (Figure 3C,D). The signal enrichment of 58 in the 1483 tumors compared to HCT116 or contralateral leg muscle was established to be >10:1 (Figure 3E). Further, we performed a COX-2 blocking experiment using indomethacin, in which nude mice with 1483 xenografts were pretreated with either DMSO or indomethacin in DMSO (2 mg/kg, intraperitoneal) prior to compound 58 dosing (2 mg/kg, intraperitoneal). At 3 h postinjection, the DMSO-pretreated mice showed strong fluorescence in their tumors, as compared to weak signals in the tumors of the indomethacin-pretreated mice (SI Figure S1). So, COX-2 expression in the 1483 tumor seems to be required for this selective uptake, although other factors besides the level of COX-2 protein may contribute to the relative enrichment over the control tissues. This result supports the hypothesis that the difference in labeling of 1483 and HCT116 xenografts is due to their differential in COX-2 expression, and confirms those reported previously for compound 41.[27]

Figure 3

In vivo labeling of COX-2-expressing xenografts by compound 58. (A) Nude mice with 1483 xenograft (COX-2 positive) or HCT116 xenograft (B, COX-2 negative) on the left flank were dosed intraperitoneally with compound 58 (2 mg/kg) and imaged at 3 h postinjection. (C) Ex vivo fluorescence image of 1483 xenograft tumor at 4.5 h postinjection of 58. (D) Ex vivo fluorescence image of HCT116 xenograft tumor at 4.5 h postinjection of 58, showing a significant tumor uptake difference between COX-2 positive versus COX-2 negative tumor models. (E) Software measurement of light emission from the 1483 versus HCT116 xenograft tumor at 4.5 h postinjection of 58, as compared with leg muscle (n = 4). RFU, relative fluorescence units.

Discussion

The present report describes our SAR studies of fluorescent derivatives of NSAIDs or COXIBs as COX-2-selective inhibitors. The compounds were prepared using a conjugate-based strategy.[27] Some, but not all, fluorescent derivatives of NSAIDs or COXIBs, such as indomethacin or celecoxib, exhibit the ability to inhibit COX-2 selectively. This is consistent with our observation that nonselective carboxylic acid-containing NSAIDs can be transformed into COXIBs by converting them into amide or ester derivatives.[30] Relatively large fluorophore moieties containing an alkyl, aryl, aralkyl, polyethylene glycolyl, or heterocyclic esters or amides of indomethacin or celecoxib exhibit COX-2 inhibitory activity. It is noteworthy that, among the fluorescent conjugates of NSAIDs or COXIBs that were evaluated in this study, only indomethacin derivatives bind tightly to COX-2 without binding to COX-1. Therefore, tethering fluorescent tags at the carboxylic acid site of indomethacin generates optical probes that are potent and selective COX-2 inhibitors.

The COX active site is located at the top of a deep channel that runs from the membrane-binding domain into the catalytic domain.[31] The initial portion of the channel has a large volume, termed the lobby, which narrows at the top into a constriction comprising residues Arg-120, Tyr-355, and Glu-524.[32] These residues must open and close in order for substrates and inhibitors to pass into or out of the COX active site, which is located above it. The use of the long alkyl, aryl, phenylene, polyethylene glycol, or heterocyclic chain allows the indomethacin functionality to fully insert into the binding pocket of COX-2, while the bulky fluorescent secondary amide functional group projects through the constriction at the base of the active site and into the wide lobby in the membrane-binding domain. This model, which is consistent with the crystal structure of a complex of COX-2 with a carboxyl chain-extended analogue of zomepirac,[33] helps to explain how such large moieties as 5- and 6-ROX can be conjugated to indomethacin with retention of inhibitory activity.

The systematic evaluation of a large number of conjugates generated from a wide variety of carboxylic acid-containing core molecules, tethers, and fluorophores provides some key insights into the structural requirements for COX-2-selective inhibition. For example, the results showed that as the number of methylene groups in the alkyl linker increased from 2 to 4, the inhibitory activity of conjugates was increased in an ordered fashion. A further increase (n = 5 to n = 14) in methylene groups or incorporation of additional amide groups in the tether reversed the trend of inhibitory activity, suggesting that an n-butyl linker was optimal for selective COX-2 inhibition by these conjugates. In addition, the activity varied depending on the number of amide-linkages in the chain. However, hybrid linkers (i.e., alkyl and PEG or alkyl and piperazine) were tolerated to some extent. The activity of conjugates also depended on the size and electronic properties of the fluorophores. Fluorophores having molecular weight of <650 without a charged atom afforded effective COX-2 inhibitory potencies. For example, coumarin or dansyl fluorophores are uncharged and have a molecular weight of <650. Indomethacin-coumarinylamide (5) was a highly potent and selective COX-2 inhibitor, and as the coumarin fluorophore was replaced by an uncharged dansyl fluorophore, the extent of COX-2 selectivity and potency of the conjugate (15) remained unchanged. In contrast, fluorophores with a favorable molecular weight but containing a charged atom that ion-pairs with an external oppositely charged ion afforded indomethacin conjugates (e.g., 74–78) that did not inhibit COX isozymes. Furthermore, conjugating the polar IRDye800 trisodium salt or polycarboxylic lanthanide chelators with indomethacin yielded inactive compounds (85–106). Interestingly, zwitterionic fluorophores, such as 5- or 6-ROX, conjugated with indomethacin through an n-butyl linker afforded selective and potent COX-2 inhibitors (e.g., 41 and 49). ROX fluorophores proved to be the optimal functional groups for effective COX-2 binding of their conjugates. Although indomethacin conjugates of dansyl, dabsyl, coumarin, or related fluorophores exhibited promising COX-2 inhibition in purified protein and in intact cells, they did not possess fluorescence properties suitable for in vivo imaging.

Among the compounds that emerged from our development pathway, only the 5-ROX- and 6-ROX-based conjugates (linker = n-Butyl) exhibited the necessary combination of adequate photophysical properties and effective COX-2 inhibitory potencies. As we previously reported,[27] the 5-ROX and 6-ROX conjugates (compounds 41 and 49) show great promise as in vivo imaging agents for inflammation and cancer. Here, we have expanded those findings, through the demonstration that compound 58 also displays a very high degree of selectivity of uptake by intact cells in tissue culture and tumors in live animals. The selective delivery of compound 58 requires the expression of COX-2 at the target site and is not seen when COX-2 is not expressed. Thus, these fluorocoxibs are capable of targeting COX-2 in intact cells and tumors that express COX-2 in both in vitro and in vivo settings. Thus, conversion of NSAIDs or COXIBs into their fluorescent derivatives provides a facile strategy for generating effective imaging agents for detection of COX-2 in premalignant and malignant tumors. Also, this facile strategy provides proof-of-principle for the development of COX-2-specific NSAID or COXIB–toxin conjugates (Chemocoxibs) enabling selective delivery of cytotoxic agents into neoplastic cells for the treatment of premalignant and malignant tumors.

Conclusions

We describe the SAR of a series of optical imaging agents comprising bulky fluorescent functional groups tethered onto NSAIDs or COXIBs to create dual function fluorescent COX-2 inhibitors. A wide variety of fluorescent conjugates were synthesized and evaluated as COX-2-selective inhibitors against both purified protein and COX-2-expressing human cancer cells. The SAR study revealed that indomethacin conjugates afford the best COX-2 inhibitors compared to conjugates of other carboxylic acid-containing NSAIDs or COXIBs. A four-carbon n-alkyl linker is optimal for conjugating bulky zwitterionic fluorescent functionalities, such as ROX, to provide the best balance between selectivity and potency of COX-2 binding. Fluorophores that are metal or halide salts, such as IRDye800 or NIR667, or organic polycarboxylic acid-containing functional groups that are highly polar, such as lanthanide chelators, are not suitable for developing COX-2-targeted imaging agents.