Fifty Years of Diazeniumdiolate Research: A Tribute to Dr. Larry K. Keefer

Abstract

The pioneering studies of Dr. Larry Keefer and colleagues with diazeniumdiolates or NONOates as a platform have unraveled the chemical biology of many nitric oxides and have led to the design of a variety of promising therapeutic agents in oncology, gastroenterology, antimicrobials, wound healing, and the like. This dedication to Dr. Larry Keefer briefly highlights some of his studies using the diazeniumdiolate platform in the cancer arena.

I. INTRODUCTION

The odyssey of diazeniumdiolate chemistry started in 1961 when Russell Drago and Bruce Karstetter described the reaction of nitric oxide (NO) with selected nucleophiles.¹ One of these was diethylamine, whose reaction with NO produced a white powder of formula 1 [Eq. (1)] that, on standing exposed to air overnight, was converted to the potent carcinogen N-nitrosodiethylamine, formula 2 [Eq. (1)].²

$$\begin{gathered} {\left(\mathrm{C}{\mathrm{H}}_3\mathrm{C}{\mathrm{H}}_2\right)}_2\mathrm{N}\mathrm{H}\stackrel{\mathrm{N}\mathrm{O}}{\to }{\left(\mathrm{C}{\mathrm{H}}_3\mathrm{C}{\mathrm{H}}_2\right)}_2\mathrm{N}-\underset{1}{{\mathrm{N}}_2{\mathrm{O}}^{-}}{\left(\mathrm{C}{\mathrm{H}}_3\mathrm{C}{\mathrm{H}}_2\right)}_2\mathrm{N}{\stackrel{+}{\mathrm{H}}}_2 \\ \stackrel{{\mathrm{O}}_2}{\to }\underset{2}{{\left(\mathrm{C}{\mathrm{H}}_3\mathrm{C}{\mathrm{H}}_2\right)}_2\mathrm{N}-\mathrm{N}\mathrm{O}} \end{gathered}$$ (1)

Mechanistic studies showed that in solution, compound 1 goes on to regenerate the NO and diethylamine, with the NO then being autoxidized to the well-known nitrosating agent N₂O₃, which recombines with the free amine to form formula 2.

Keefer and colleagues reasoned that if formula 1 and its analogues indeed hydrolyze to produce molecular NO in solution, then these ions might be pharmacologically active. As a platform for testing this hypothesis, Keefer and colleagues started by preparing a selection of salts of structure X-N(O) = NO⁻ M+ and systematically cataloguing their physicochemical properties. Structural studies confirmed the substantial double bond character of the N–N linkage of many X-N(O) = NO⁻ ions, with the oxygens being cis to one another; this and other features of their structures led to their formal designation as “diazeniumdiolates,”³ although earlier nomenclature referring to them as “NOC compounds” or “NONOates” continues to be used.

As a platform, diazeniumdiolates are very versatile, targeting NO to specific sites in the body for important therapeutic applications (Fig. 1). They provide reliable fluxes generated in solution with reproducible half-lives ranging from 2 seconds to 20 hours at physiological pH and temperature. Certain diazeniumdiolates were found to generate nitroxyl (HNO, the one-electron reduction product of NO and another known bioeffector; see below) as well as NO on hydrolysis. This dedication to Dr. Larry Keefer highlights some of the studies using the diazeniumdiolate platform in the cancer arena.

FIG. 1:

Diazeniumdiolate chemistry as a platform for designing clinical applications based on nitric oxide (NO) and nitroxyl (HNO). Shown are four clinical areas in which diazeniumdiolate-based proof-of-concept studies suggest considerable practical significance on the part of this class of compounds.

II. DIAZENIUMDIOLATE-BASED NO-RELEASING COMPOUNDS

Diazeniumdiolate prodrugs release NO upon hydrolysis or metabolic activation to form the parent anion, which further decomposes to release up to two moles of NO and the parent amine (Fig. 2A).^4,5 The therapeutic applications of these prodrugs are diverse and largely depend on the O-2 protecting group (“R” in Fig 2A) and its mechanism of activation. For example, vinyl protected prodrug V-PYRRO/NO (Fig. 2B) is activated by cytochrome P450 to release NO, and shows hepatoprotective properties against a variety of toxins.⁶ Glutathione (GSH)-activated arylated prodrug JS-K (Fig. 2C) has anticancer activity.⁷ The primary amine diazeniumdiolate prodrug AcOM-IPA/NO (Fig. 2D)^8,9 was reported to release nitroxyl (HNO), another potent bioeffector molecule with possible applications in treating heart failure, alcohol abuse, and cancer.^10–12 Secondary amine diazeniumdiolate ions are protonated at N-3 (see Fig. 2A for numbering) to release NO,⁴ whereas primary amine diazeniumdiolates release nitroxyl^8,9 (HNO) on protonation at N-2. Below, the anticancer properties of these diazeniumdiolate prodrugs are presented.

FIG. 2:

(A) Activation of diazeniumdiolate prodrugs to release NO or HNO. (B) Structure of V-PYRRO/NO. (C) Structure of JS-K. (D) Structure of AcOM-IPA/NO.

III. NONO-NSAIDs

NONO-NSAIDs are based on linking a N-diazen-1-ium-1,2 diolate functional group to a classical NSAID (Fig. 3) yielding compounds that are not likely to lead to “nitrate tolerance”^13–15 and also have the potential to generate two equivalents of NO.¹⁶ The production of NO from nitrate esters, discussed earlier, requires a three-electron reduction.¹⁷ However, NONO-NSAIDs do not require redox activation before the release of NO.¹⁶ Another attractive attribute of these classes of NO-releasing compounds is their rich derivatization chemistry that facilitates the targeting of NO to specific target organ and/or tissue sites.¹⁸

FIG. 3:

Chemical structures of NONO-aspirin, IPA/NO-aspirin, DEA/NO-aspirin, PABA/NO, and diazeniundiolate/OA hybrid. Hybrid ester prodrugs possessing a 1-(pyrrolidin-1yl)diazen-1-ium-1,2-diolate, A(1) or 1-(N,N-dimethylamino)diazen-1-ium-1,2-diolate, A(2), moiety attached via a one-carbon methylene space to the carboxylic acid group of aspirin. (B) NO-releasing moiety is O²-acetoxymethy 1-[N-(2-hydroxyethyl)-N-methylamino]diazen-1-ium-1,2-diolate. (C) NO-releasing moiety is O²-acetoxymethy 1-[(2-hydroxymethyl) pyrrolidin-1-yl]diazen-1-ium-1,2-diolate. (D) Two NONO-NSAIDs prepared by derivatizing both a primary and secondary amine diazeniumdiolate with aspirin to produce O²-(acetylsalicyloyloxymethyl)-1-(N-isopropylamino)-diazen-1-ium-1,2-diolate (IPA/NO-aspirin), and O²-(acetylsalicyloyloxymethyl)-1-(N,N-diethylamino)-diazen-1-ium-1,2-diolate (DEA/NO-aspirin). (E) O²-[2,4-dinitro-5-(N-methyl-N-4-carboxyphenylamino) phenyl] 1-N,N-dimethylamino)diazen-1-ium-1,2-diolate (PABA/NO) designed to be activated for anticancer effects by glutathione-S-transferase (GST)-induced release of NO. (F) A hybrid comprising of O²-(2,4-dinitrophenyl)diazeniumdiolate and oleanolic acid (OA).

The first agent reported in this compound class had a NONOate (O²-unsubstituted N-diazen-1-ium-1,2-diolate) attached via a one-carbon methylene spacer to the carboxylic acid group of a traditional NSAIDs (aspirin, ibuprofen, indomethacin) (Fig. 3A).¹⁹ In vitro, these agents did not inhibit the enzymatic catalytic activity of COX-1 or COX-2; however, they were equipotent to their traditional NSAID counterparts when evaluated in the carrageenan-induced rat paw anti-inflammatory model. Also, unlike their traditional parent NSAID, these agents had no significant gastric toxicity when given orally.¹⁹ A series of NONO-NSAIDs (aspirin, ibuprofen, indomethacin) was subsequently made that possessed O²-acetoxymethyl-1-[N-(2-hydroxyethyl)-N-methylamino]diazen-1-ium-1,2-diolate moiety as the NO donor (2-HEMA/NO) (Fig. 3B).¹⁶ Here, the NO-donating moiety was attached via a two-carbon ethyl spacer to the carboxylic acid of the traditional NSAID, and because a secondary dialkyamine was used in their synthesis, the number of possible new NONO-NSAIDs was enormous. Like their predecessors, these agents were nonulcerogenic and in vitro did not inhibit either COX-1 or COX-2 activity, but showed even better anti-inflammatory properties, suggesting that they were acting as prodrugs, requiring metabolic activation by an esterase to release the parent NSAID. A potential limitation was that hydrolysis would also release one equiv of the corresponding nitrosoamines that are biologically toxic. To overcome this concern, a second-generation of O²-acetoxymethyl-protected (PROLI/NO) releasing NONO-NSAIDs was developed where a diazeniumdiolate ion obtained from an amine such as L-proline was used, the N-nitroso derivative of which is nontoxic (Fig. 3C).²⁰ These agents were also nonulcerogenic and had better anti-inflammatory properties and effective analgesic activity. They also produced up to 1.9 mol of NO/mol of compound.²⁰

NONO-NSAIDs are an attractive class of compounds; however, there is no data on their chemopreventive potential. Based on the NO-NSAIDs, one might expect these compounds to have potent chemoprevention properties. The antiulcerogenic, anti-inflammatory, analgesic, and antipyretic effects of an NONO-NSAID were compared directly to that of m-NO-ASA, together with effects on relevant biological markers such as gastric PGE₂ and lipid peroxidation levels, superoxide dismutase activity, and TNF-α levels. In all aspects, the two classes of compounds were similar, suggesting that there may be a threshold for NO above which no further beneficial effects may be apparent in the higher levels of NO.²¹

In a study evaluating the potential use of NONO-NSAIDs as antimetastasis agents, the effects of NONO-aspirin and NONO-naproxen were compared with the effects of their respective parent NSAIDs on avidities of human melanoma M624 cells.²² Both NONO-NSAIDs, but not their corresponding parent NSAIDs, reduced M624 adhesion on vascular cellular adhesion molecule-1 (VCAM-1) and fibronectin under fluid flow conditions and static conditions, respectively. The NONOate moiety on the NSAIDs was shown to be critical for their function in reducing the avidity of melanoma cells. These findings suggest that NONO-NSAIDs are potentially a new class of antimetastasis drugs for cancer treatment.

IV. HNO-NSAIDs

Decomposition of diazeniumdiolates (also known as NONOates) can lead to formation of nitroxyl (HNO) and/or NO.⁴ HNO has emerged as an important pharmacological agent with beneficial effects in overcoming heart failure,¹¹ preconditioning against myocardial infarction,²³ and treating alcohol abuse.¹⁰ Using Angeli’s salt (Na₂N₂O₃) to generate HNO, the first anticancer activity of HNO was reported in 2008,²⁴ when it was shown that it suppressed the proliferation of both MCF-7 estrogen receptor (ER)-positive and MDA-MB-231 ER-negative human breast cancer cell lines under both hypoxic and normoxic conditions. In mice bearing ER-negative breast cancer xenografts, HNO treatment caused significant reductions in tumor growth accompanied by increases in apoptosis and suppression of angiogenesis.²⁴ Recently, two new NONO-NSAIDs were prepared by derivatizing both a primary and secondary amine diazeniumdiolate with aspirin to produce O²-(acetylsalicyloyloxymethyl)-1-(N-isopropylamino)-diazen-1-ium-1,2-diolate (IPA/NO-aspirin) (Fig. 3D) and O²-(acetylsalicyloyloxymethyl)-1-(N,N-diethylamino)-diazen-1-ium-1,2-diolate (DEA/NO-aspirin).²⁵ Both of these agents had enhanced GI safety profiles, showed strong anti-inflammatory properties, and exhibited significantly enhanced cytotoxcity compared to either aspirin or the parent diazeniumdiolate toward non-small-cell lung carcinoma cells (A549), but they were not appreciably toxic toward endothelial cells (HUVECs), which suggests cancer-specific sensitivity. The HNO-NSAID prodrug also inhibited cylcooxgenase-2 and glyceraldehyde 3-phosphate dehydrogenase activity (GAPDH). Both IPA/NO-aspirin and DEA/NO-aspirin were also shown to reduce the growth of breast cancer cell lines MDA-MB-231, MDA-MB-468 (both ER-negative), and MCF-7 (ER-positive) more effectively than the parent compounds while not being appreciably cytotoxic in a related nontumorigenic MCF-10A cell line.¹² Notably the cytotoxicity of these agents was about the same in ER-negative and ER-positive cell lines, suggesting an effect independent of the estrogen status of the cells. In xenografts of GFP-transfected MDA-MB-231 cells, which allow noninvasive monitoring of tumor size, DEA/NO-aspirin caused a substantial reduction in fluorescence intensity; however, this was not statistically significant. On the other hand, treatment with IPA/NO-aspirin led to a significant decrease in both fluorescence intensity and tumor mass, suggesting that the HNO-donating derivative not only effectively inhibited tumor progression but was also tumoricidal.²⁵ The high cytotoxicity of IPA/NO-aspirin was in part due to increased oxidant levels leading to DNA damage and to inhibition of GAPDH, which led to caspase-3-mediated induction of apoptosis.

V. JS-K AND PABA/NO

O²-(2,4-dinitrophenyl)1-[(4-thoxycarbonyl)piperazin-1-yl]diazen-1-ium-1,2-diolate (JS-K; Fig 2C) and O²-[2,4-dinitro-5-(N-methyl-N-4-carboxyphenylamino) phenyl] 1-(N,N-dimethylamino)diazen-1-ium-1,2-diolate (PABA/NO; Fig. 3E) are members of the diazeniumdiolate class of nitric oxide prodrugs that were designed to be activated for anticancer effects by glutathione-S-transferase (GST)-induced release of NO.²⁶ The rationale for this was based on the observation that GST (specifically GST-π), a key phase II detoxification enzyme, is frequently overexpressed in cancer tissue.^27,28 JS-K^29,30 and PABA/NO^31,32 have shown promise as anticancer agents. PABA/NO showed strong antitumor activity comparable to that of cisplatin in subcutaneous A2780 human ovarian cancer xenografts in female immunodeficient mice.³¹ Modulation of NO signaling may be a promising strategy in treating glioblastoma. U87 glioma cells exposed to PABA/NO showed a strong dose-dependent growth-inhibitory effect in vitro, and a strong synergistic effect was observed after concomitant treatment with temozolomide (TMZ), but not with carboplatin (CPT).³³ In an animal model where nude rats underwent stereotactic implantation of U87 glioma cells into the right striatum, PABA/NO administration by differing routes did not lead to an extension of survival time despite decreases in tumor proliferation and increases in apoptosis.³³

JS-K has shown potent antileukemia activity in vitro using HL-60 and U937 cell lines and in vivo with HL-60 cells using a xenograft model.²⁹ JS-K has also been shown to inhibit hepatoma Hep 3B cell proliferation,³⁴ enhance arsenic- and cisplatin-induced cytolethality in arsenic-transformed rat liver cells,³⁵ and induce apoptosis in human multiple myeloma cell lines.³⁶ In addition, JS-K was shown to be effective against leukemia and renal, prostate, and brain cancer cells,^37,38 as well as both ER-positive and ER-negative breast cancer cell lines without affecting normal mammary epithelial cells.³⁹ In further evaluating the metabolic actions of JS-K in order to decipher its mechanisms of cytotoxicity, it was determined that the activating step in the metabolism of JS-K in the cell was the dearylation of the diazeniumdiolate by GSH resulting in release of two equivalents of NO.⁴⁰ This caused rapid depletion of GSH, resulting in alterations in the redox potential of the cellular environment, which led to activation of the MAPK stress signaling pathways, and induction of apoptosis.

In the androgen receptor (AR)-positive prostate cancer (PCa) cell line 22Rv1, JS-K was able to reduce the intracellular concentration of functional AR.²⁷ This was most likely due to high intracellular levels of NO as demonstrated indirectly by high levels of nitrotyrosine in JS-K–treated cells. Moreover, JS-K diminished WNT signaling in AR-positive 22Rv1 cells. In line with these observations, castration-resistant 22Rv1 cells were found to be more susceptible to the growth inhibitory effects of JS-K than the androgen dependent LNCaP cells, which do not exhibit an active WNT-signaling pathway.

Malignant gliomas exhibit overexpression and genetic polymorphisms of the GST gene which influence the malignancy of the tumor and its response to chemo- or radiotherapy.^41,42 Therefore, JS-K might be a suitable candidate for treating malignant gliomas. In human U87 glioma cells and primary glioblastoma cells, JS-K showed a dose-dependent cytotoxicity.⁴³ Cell death was partially induced by caspase-dependent apoptosis, which could be blocked by the pancaspase-inhibitors Z-VAD-FMK and Q-VD-OPH. Inhibition of GST by sulfasalazine, cGMP inhibition by ODQ (1H-[1,2,4] oxadiazolo[4,3-a]quinoxalin-1-one), and MEK1/2 inhibition by UO126 (bis[amino[(2-aminophenyl) thio]methylene]butanedinitrile) attenuated the antiproliferative effect of JS-K, suggesting the involvement of various intracellular death signaling pathways. JS-K also reduced the growth of U87 xenografts, showing reduced proliferation, increased apoptosis, and increased necrosis.⁴³

Using 18 different non-small-cell lung cancer (NSCLC) cell lines, it was shown that JS-K was most effective against a subset of NSCLC cell lines where the endogenous levels of reactive oxygen and nitrogen species (ROS/RNS) were high.⁴⁴ Treatment of lung adenocarcinoma cells with JS-K resulted in oxidative/nitrosative stress in cells with high basal levels of ROS/RNS, which, combined with the arylating properties of the compound, was reflected in glutathione depletion and alteration in cellular redox potential, mitochondrial membrane permeabilization, and cytochrome c release. JS-K treatment that was formulated in Pluronic P123 led to 75% reduction in the growth of H1703 lung adenocarcinoma cells in a xenograft model.⁴⁴ Of note, levels of peroxiredoxin 1 (PRX1) and 8-oxo-deoxyguanosine glycosylase (OGG1) also correlated with JS-K sensitivity. Taken together, JS-K may have a role in personalized therapy for lung cancers characterized by high levels of ROS/RNS. PRX1 and OGG1 proteins, which can be easily measured, could function as biomarkers for identifying tumors sensitive to this therapy.

The protein β-catenin has a central role in the Wnt signaling pathway that regulates cell–cell adhesion and may promote leukemia cell proliferation.⁴⁵ β-catenin is highly expressed in acute lymphoblastic leukemia (ALL), tumor lines of hematopoietic origin, and primary leukemia cells. However, it is undetectable in normal peripheral blood T cells. Among the leukemia cell lines, β-catenin is expressed at high levels in Jurkat T cells.⁴⁵ In this cell line, JS-K strongly inhibited its growth, reduced its proliferation, and caused G₂/M cell cycle arrest, which led to increased apoptosis.⁴⁶ JS-K reduced the transcriptional activity of β-catenin and also reduced levels of nuclear β-catenin protein, but had no appreciable effect on levels in the cytosol. The mechanism underlying these phenomena was shown to be S-nitrosylation of nuclear β-catenin.^46,47

In order to improve the selectivity for cancer cells and particularly to enhance the uncatalyzed stability of GSTπ-activated O²-(2,4-dinitrophenyl) diazeniumdiolates, novel hybrids comprising the latter and oleanolic acid (OA) have been developed.⁴⁸ The most potent of these is shown in Fig. 3F. The rationale for their development was based on observations that (1) OA imparts additional hepatic selectivity and a synergetic biological profile to the GSTπ-activated moiety because of liver-specific distribution and liver-protective effects⁴⁹; (2) galactosyl moieties are recognized by high levels of asialoglycoprotein receptors (ASGPRs) expressed in human hepatocellular carcinoma (HCC) cells, further enhancing selectivity and bioactivity⁵⁰; and (3) amino acids link the 2,4-dinitrophenyl ring with OA to avoid its hydrolysis, and the amine moiety may act as an electron-donating group to increase electron density of the 2,4-dinitrophenyl ring, thus slowing the uncatalyzed reaction with GSH. Indeed, the hybrid shown in Fig. 3F was shown to be very stable. It induced apoptosis in HepG2 cells by arresting the cell cycle at the G2/M phase, activating both the mitochondrion-mediated pathway and the MAPK pathway and enhancing the intracellular production of ROS.⁴⁸

VI. CONCLUDING REMARKS

Clearly, the full potential of the diazeniumdiolate platform in developing clinically relevant drugs has yet to be realized, although much progress has been made thanks to the pioneering work of Dr. Larry Keefer and his colleagues. In a paper published in 2011 highlighting his journey in this chemical and biological field,⁴ Dr. Keefer concluded by asking three questions that he hoped would pique interest and further discussion. One, how can it be explained that amines generally react smoothly with NO, whereas water and hydroxide apparently do not? The product of a water/hydroxide diazeniumdiolation reaction is Angeli’s salt [NaO-N(O)dNONa], an HNO donor that is synthesized by nitrating hydroxylamine. Like most other ionic diazeniumdiolates, Angeli’s salt is increasingly stabilized as pH is raised, such that if it were to be formed in 1-M solutions of sodium hydroxide in contact with 3 atm of NO, it would be spectrophotometrically detectable. But this does not happen. Why not? Two, what are the structural and stereoelectronic determinants of the partition between the NO and HNO generation pathways in the primary amine diazeniumdiolate series? And three, what are the structural and stereoelectronic determinants of hydrolysis half-life?

On September 11, 2020, after a long bout with Parkinson’s disease, Dr. Larry Kay Keefer passed away in Silver Spring, Maryland. He will be greatly missed.