Development of BRD4 inhibitors as anti-inflammatory agents and antidotes for arsenicals

Abstract

Arsenicals belong to the class of chemical warfare agents known as vesicants, which are highly reactive, toxic and cause robust inflammatory response. Cutaneous exposure to arsenicals causes a wide range of systemic organ damage, beginning with cutaneous injuries, and later manifest multi-organ damage and death. Thus, the development of suitable antidotes that can effectively block injury following exposure to these agents is of great importance. Bromodomain 4 (BRD4), a member of the bromodomain and extra terminal domain (BET) family, plays crucial role in regulating transcription of inflammatory, proliferation and cell cycle genes. In this context, the development of potent small molecule inhibitors of BRD4 could serve as potential antidotes for arsenicals. Herein, we describe the synthesis and biological evaluation of a series of compounds.

Graphical Abstract

Arsenicals are a series of highly toxic arsenic compounds that are built around a chloroarsine molecule (Figure 1). Some arsenicals have been developed as organic compounds and used as chemical warfare agents including Lewisite, methyldichloroarsine, ethyldichloroarsine, diphenylchloroarsine, and diphenylcyanoarsine (Figure 1).^1–3 As vesicants, these arsenicals cause painful irritation and blistering of the skin, eyes, and respiratory system and may manifest delayed deleterious effects to human health.^4,5 We have recently developed animal models to define their pathogenesis.⁶ Since these chemicals are highly toxic and cannot be used under usual laboratory conditions for investigating their mechanism of action. We also developed a surrogate arsenical, phenylarsine oxide (PAO), which is less toxic and can be used under safety regulation available in laboratories.⁷ We showed that PAO manifests similar pathophysiological alterations in the skin and other organs that are caused by other warfare arsenicals, however, at different dose regimens.^7–9 Thus, similar to lewisite, it also causes an early robust inflammatory response in the skin, which follows blistering and severe cutaneous injury.⁷ Although efforts to cease the use of these agents have been put forth by many countries, stockpiles of chemical weapons still exist and have recently been used against civilian populations.¹⁰ Their accidental exposure has also been reported.^11,12 The only approved medical countermeasure against arsenical-induced toxicity is 2,3-dimercapto-1-propanol or British Anti-Lewisite (BAL), which works topically and systemically as a chelating agent of arsenic.⁴ However, its related toxicity and low therapeutic index are considerable limitations that demand the development of more effective and less toxic antidotes.¹³

Figure 1.

Examples of arsenicals use as chemical warfare agents

The bromodomain-containing protein 4 (BRD4) is a member of the bromodomain and extra-terminal (BET) protein family. As such, its structure consists of two bromodomains that are arranged in a conserved bundle of four α-helices and connected by two loop segments. Each bromodomain comprises an active acetyl-lysine-binding pocket allowing for the specific recognition of ε-N-acetylated lysine residues on target proteins, including histones.¹⁴ Role of BRD4 has been shown in various pathophysiological processes. In this regard, BRD4 was particularly shown to play key role in cell proliferation and inflammation.¹⁵ More recently, its involvement in complex inflammatory disorders such as sepsis and COVID-19 are revealed.^16,17 BRD4 has been shown to play an important role in the regulation of inflammatory genes.^18,19 As a result, inhibiting the activity of BRD4 has served as a strategy to suppress inflammation,^20–22 which also can be measured by a decrease of interleukin-6 (IL-6), cytokine produced in response to tissue damage and inflammation.²³ The inflammatory response caused by exposure to arsenicals is highly complex and is associated with induction of various cytokines and chemokines including IL-6.^6,24,25 In light of these observations, it becomes apparent that developing BRD4 inhibitors to attenuate arsenical-induced inflammation could pave the way for the development of a new class of arsenical antidotes. BRD4 is a druggable target and a wide range of its inhibitors have recently been synthesized. Some of these molecules have moved for clinical trials against Myelofibrosis, Multiple Myeloma, Neuroblastoma, Prostate Adenocarcinoma, Acute myeloid leukemia, Non-Hodgkin Lymphoma, etc.²⁶

2-Methoxy-N-(3-methyl-2-oxo-1, 2, 3,4-tetrahydroquinazolin-6-yl)benzenesulfonamide (PFI-1, 1) is a potent and highly selective BET bromodomain inhibitor with an IC₅₀ value of 220 nM against BRD4 in the AlphaScreen Assay.²⁷ Crystallographic studies have revealed that the binding mode of PFI-1 to the BRD4 active site is achieved through the quinazolinone moiety.²⁷ The quinazolinone carbonyl and nitrogen form hydrogen bonding with the conserved asparagine N140. Furthermore, the quinazolinone moiety forms hydrophobic contacts at a hydrophobic region surrounded by residues such as V87, L92, L94, and I146 (Figure 2). Based on these studies, a series of PFI-1 analogs were synthesized by modifying the quinazolinone core, 2-methoxybenzene moiety and sulfonamide linker, and tested in the BRD4 inhibitory assay. Furthermore, active compounds were tested for their ability to inhibit IL6 activity and our results are presented here.

Figure 2.

(a) Structure of PFI-1. (b) Co-crystal structure of PFI-1 and BRD4 (PDB ID: 4E96). Black and orange dashed lines indicate hydrogen bonds and hydrophobic contacts, respectively. Residues that are important for ligand binding are colored in purple. Direction that faces solvent is indicated by blue shaded area. Co-crystallized water molecules adjacent to the ligand are displayed. Hydrogen atoms have been added by Schrödinger Small Molecule Drug Discovery Suite.

In an effort to optimize the drug-like properties of the quinazolinone scaffold while retaining or improving potency, a series of analogs of the quinazolinone scaffold had been proposed. According to ligand-based in silico predictions, the parent compound 1 had moderate solubility (>10 μM) with up to three metabolic labile spots (Figure 3). Based on co-crystal structures and computational docking results, the left-hand side 2-methoxybenzene moiety, which largely resided in the solvent side, had flexible binding orientations and did not form fixed interactions at the binding site. Namely, its effect on the BRD4 inhibition potency may not be directly predicted from co-crystals or docking studies. Thus, various modifications on that moiety were experimentally explored to see whether they would retain the potency. Since the 2-methoxyl group on that moiety was a metabolic labile spot, structural modifications were focused on substituting the 2-methoxyl group to improve metabolic stability (e.g., 5a to 5n). In contrast, the right-hand side quinazolinone core was predicted to form hydrophobic contact and hydrogen bonds that were critical for ligand binding. Consistent with the docking results, analogs with the quinazolinone core shrinking to a benzimidazolinone while retaining the hydrogen bonds with residue N140 (21a and 21b) were still active in the BRD4 and IL-6 assays. Moreover, the best analog (5i) of the series was successfully co-crystallized with BRD4, which displayed the same binding pose as predicted by computational docking (for details, see section ‘Binding of compound 5i to BRD4’).

Figure 3.

Brief summary of in silico predictions. Text in blue: conclusion from ligand-based solubility predictions. Text in red: conclusion from ligand-based metabolic stability predictions. Text in purple: conclusion from computational docking studies of BRD4.

Investigation of the left-hand side of PFI-1 started with the synthesis of target compounds 5a-5n as outlined in Scheme 1. Starting from the commercially available 3-methyl-1,4-dihydroquinazolin-2-one 2, treatment with potassium nitrate in the presence of concentrated sulfuric acid gave 3-methyl-6-nitro-1,4-dihydroquinazolin-2-one 3 (Scheme 1a).²⁸ Reduction of the nitro group to the corresponding amine by hydrogenation in the presence of Raney Nickel generated the intermediate 4,²⁸ which was subsequently treated with the corresponding sulfonyl or carbonyl chlorides to afford the quinazolinone derivatives 5a-5l.²⁹ Target compound 5m, the N-methylated analogue of compound 5a, was obtained by methylation of intermediate 3 to give compound 6,³⁰ which was subjected to hydrogenation to afford compound 7,²⁸ and finally reacted with 2-fluoro-5-(trifluoromethyl)benzenesulfonyl chloride.²⁹ Target compound 5n was prepared in two steps by the reaction of 4-(bromomethyl)benzenesulfonyl fluoride 8 with 1-ethylpiperazine to give the sulfonyl fluoride 9, followed by treatment with intermediate quinazolinone 4 (Scheme 1b).³¹

Scheme 1.

Synthesis of quinazolinone derivatives 5a-5n.

Target compounds 11a and 11b, which are reverse sulfonamides of PFI-1 (1) and the quinazolinone derivative 5a, respectively, were also synthesized from the commercially available 3-methyl-1,4-dihydroquinazolin-2-one 2 (Scheme 2).²⁹ Reaction with chlorosulfonic acid afforded the sulfonyl chloride 10, which was further reacted with the corresponding anilines to give the desired compounds 11a and 11b.²⁹

Scheme 2.

Synthesis of reverse quinazolinone sulfonamide derivatives 11a-11b.

Compounds 16a and 16b are 2,4-dioxo-quinazoline analogues of PFI-1 (1) and 5a, respectively. Their preparation started with the amidation of 2-amino-5-nitro benzoic acid 12 in presence of HATU to give compound 13,³² which upon treatment with CDI gave the cyclized intermediate 14 (Scheme 3).³³ Reduction of the nitro group in the presence of iron and acetic acid gave the amine 15,³⁴ which was reacted with the corresponding sulfonyl chlorides to give the desired compounds 16a and 16b.²⁹

Scheme 3.

Synthesis of 2,4-dioxo-quinazoline derivatives 16a-16b.

Finally, the benzimidazolinones 21a and 21b, analogous to PFI-1 and 5a, respectively, were synthesized in four steps from starting reagent 17 (Scheme 4). Regioselective N-methylation,³⁵ cyclization in the presence of CDI,³⁶ acid-catalyzed nitro reduction,³⁴ and reaction with the corresponding sulfonyl chlorides²⁹ yielded the desired targets 21a and 21b.

Scheme 4.

Synthesis of benzimidazolinone derivatives 21a-21b.

The compounds (5a-5n, 11a-11b, 16a-16b and 21a-21b) were screened for their ability to inhibit BRD4 and IC₅₀ results are summarized in Tables 1. JQ1, a known BRD4 inhibitor, was used as the positive control for this screening. Furthermore, active compounds from the primary assay (BRD4 Alpha Screen assay) were evaluated in vitro for their potential inhibitory activity against IL-6 (ELISA assay) and was served as the secondary assay for this study (Table 1). Additionally, selected compounds were tested for metabolic stability in rodent and human liver microsomes and for solubility (Table 2). 5i, A compound with excellent in vitro activity (BRD4 and IL6), and ADME properties advanced to pharmacokinetic (PK) studies (Table 3) and in vivo studies.

Table 1.

IC₅₀ values of compounds 5a-5n, 11a-11b, 16a-16b and 21a-21b in the BRD4 and IL-6 assays

Compounds	BRD4 IC50 (μM)	IL-6 IC50 (μM)	Compounds	BRD4 IC50 (μM)	IL-6 IC50 (μM)
5a	0.17	2.16	5l	9.60	>50
5b	0.38	4.15	5m	17.81	9.60
5c	0.38	4.95	5n	4.98	>50
5d	0.86	16.20	11a	6.23	5.91
5e	5.76	36.28	11b	>40	N.T.
5f	24.5	>50	16a	3.60	5.45
5g	14.22	14.39	16b	>40	N.T.
5h	0.22	0.48	21a	0.60	3.09
5i	0.33	0.83	21b	2.40	9.82
5j	0.44	1.94	PFI-1 (1)	0.041	0.43
5k	3.48	16.33	JQ-1	0.009	0.56

N.T. = not tested

Table 2.

ADME properties of compounds 5h and 5i

Compounds	Microsomal Stability* (minutes)		Solubility† (μM)	log D
	MLM‡	HLM§
5h	6.8	39.6	65.4	1.5
5i	15.4	88.8	69.9	2.5

^*Diclofenac was used as a positive control,

^†Estradiol (haloperidol) was used as a control standard and data were normalized,

^‡HLM = Human Liver Microsome,

^§MLM = Mouse Liver Microsome

Table 3:

PK results of 5i

Compound	IP administration
	t1/2* (h)	tmax† (h)	Cmax‡ (ng/mL)	AUClast§ (h*ng/mL)
5i	3.71	1.17	858	4438

^*t_½ = half-life,

^†t_max = time of peak plasma concentration,

^‡C_max = maximum concentration,

^§AUC_last = Area under the curve from the time of dosing to the last measurable concentration.

Modifications of the 2-OMe-benzene ring of PFI-1 afforded compounds with a wide range of IC₅₀ values. While the hit compound PFI-1 remained the most potent in the BRD4 assay (IC₅₀ = 41 nM), compounds 5a-5d and 5h-5j also displayed excellent inhibitory activities with sub-micromolar IC₅₀ values (0.17–0.86 μM). Subtle variations on the benzene ring, as in 5a (2-F, 5-CF₃; IC₅₀ = 0.17 μM), 5c (2,3,5-triF; IC₅₀ = 0.38 μM), 5i (3-Me, 4-F; IC₅₀ = 0.33 μM), and 5j (4-NH₂; IC₅₀ = 0.44 μM), appeared to minimally disrupt the inhibitory potency in the BRD4 assay. Furthermore, introducing a short aliphatic chain between the benzene ring and the sulfonyl group (5d; IC₅₀ = 0.86 μM) or simply fusing a heterocyclic ring to the benzene ring (5b; IC₅₀ = 0.38 μM/5h; IC₅₀ = 0.22 μM) did not make any change in the activity. Indeed, as the 2-OMe benzene ring of PFI-1 resides in the hydrophobic region, minimal substitutions or those favoring pi-pi stacking are predicted to be well tolerated. As a result, 5k (IC₅₀ = 3.48 μM), 5l (IC₅₀ = 9.60 μM) and 5n (IC₅₀ = 4.98 μM), which all bear hydrophilic substitutions, only displayed moderate inhibitory activity.

When the sulfonamide linker in compound 5d (IC₅₀ = 0.86 μM) was replaced by an amide in compound 5e, we observed around seven-fold decrease in activity (IC₅₀ = 5.76 μM). The carbonyl group in the amide moiety thus appears to be of lesser efficacy, which was further observed in compound 5f (IC₅₀ = 24.5 μM) and compound 5g (IC₅₀ = 14.22 μM). For compounds 11a and 11b, sulfonamide linker in PFI-1 (IC₅₀ = 0.041 μM) and 5a (IC₅₀ = 0.17 μM) was replaced with a reverse sulfonamide linker and both compounds showed decreased activity (11a, IC₅₀ = 6.23 μM and 11b. IC₅₀ > 40 μM) by >150-fold which may have been due to the disruption of the interactions of the sulfonyl group in the BRD4 active site.

Following these modifications of the left-hand side aryl ring and the sulfonamide linker, compound 5a displayed the most effective inhibitory activity in the BRD4 assay. PFI-1 and compound 5a were thus chosen to further investigate variations of the quinazolinone moiety. When the NH group on the quinazolinone ring in compound 5a (IC₅₀ = 0.17 μM) was converted to N-Me group as in compound 5m (IC₅₀ = 17.81 μM), a 100-fold increase in IC₅₀ value was observed, resulting in a significant loss of inhibitory activity. This is in accordance with the crystallographic data, which revealed that the NH group on the quinazolinone ring participates as a donor in the hydrogen bonding in the Kac site of BRD4.

Modifications of the quinazolinone moiety in PFI-1 and compound 5a by adding a carbonyl group at the C4-position to form the 2,4-dioxo-quinazoline derivatives 16a and 16b, respectively, or by completely removing the carbon atom at the C4-position to give the benzimidazolinone derivatives 21a and 21b, respectively, did not enhance the BRD4 inhibitory activity, although compound 21a still showed excellent activity. We observed a good BRD4 inhibitory activity for compound 16a (IC₅₀ = 3.60 μM) and a complete loss of activity for compound 16b (IC₅₀ >40 μM). Compound 21a (IC₅₀ = 0.60 μM) exhibited excellent inhibitory activity with sub-micro molar IC₅₀ values while compound 21b (IC₅₀ = 2.40 μM) displayed good activity.

The selected BRD4 inhibitors were also evaluated in the cell-based IL-6 assay (Table 1) and the observation made in the BRD4 assay translated well into the IL-6 assay. PFI-1 and compounds 5a-5c, 5h-5j, and 21a, which all displayed excellent inhibitory activities with sub-molar IC₅₀ values (0.041–0.60 μM) in BRD4 assay, also showed good to excellent efficacy in the IL-6 assay (IC₅₀ = 0.43–4.95 μM). The only exception is compound 5d, which had an excellent IC₅₀ value of 0.86 μM in BRD4 assay but a moderate efficacy in the IL-6 assay with IC₅₀ value of 16.2 μM. Also, all other synthesized compounds that had IC₅₀ values >1μM in BRD4 assay only had moderate to poor efficacy in the IL-6 assay.

Compounds 5h (BRD4 IC₅₀ = 0.22 μM; IL-6 IC₅₀ = 0.48 μM) and 5i (BRD4 IC₅₀ = 0.33 μM; IL-6 IC₅₀ = 0.83 μM) thus appeared to be the most promising compounds with sub-micro molar IC₅₀ values in both assays. Furthermore, the ADME (Absorption, Distribution, Metabolism, and Excretion) properties (Table 2) of compound 5i compared to 5h made the former a better candidate to advance to further studies such as pharmacokinetics (PK), binding and in vivo studies.

Pharmacokinetic (PK) studies:

The compound 5i was advanced to pharmacokinetic (PK) studies to evaluate the drug-like properties in vivo. This compound was administered IP (Intraperitoneal) and the results are presented in Table 3. It exhibited a half-life (t_½) of 3.7 hours with a C_max (maximum concentration) of 858 ng/mL.

Binding of compound 5i to BRD4:

The binding of compound 5i to BRD4 was confirmed by isothermal titration calorimetry (ITC) (Figure 4a). The binding affinity (Kd) of the compound is about 1 μM. To understand the binding details of compound 5i to BRD4, the co-crystal structure of BRD4 with compound 5i was determined. Compound 5i binds to the acetyl lysine binding site on BRD4 in the structure. The quinazoline ring is sandwiched in the hydrophobic activity formed by Cys136, Phe83, Ile146, Pro82, Tyr139, Tyr97, Leu94 and Val87. Asn140 forms two hydrogen bonds with oxo and NH groups on the quinazoline ring. Sulfonamide group does not interact with any residues and is exposed to the solvent area. The fluoro methyl phenyl group forms hydrophobic interactions with Ile146, Met149 and Trp81 (Figure 4b). The location of compound 5i is similar to PFI-1 (1) in the BRD4 structure.²⁷ The phenyl ring is shifted upward due to the location of methyl in the hydrophobic packet (Figure 4c). The ITC experiment and co-crystal structure confirmed that compound 5i specifically targets BRD4.

Figure 4.

Binding affinity of 5i to BRD4 and co-crystal structure of 5i and BRD4. a) Isothermal titration calorimetry (ITC) measurement of 5i binding to BRD4. The ITC titration curves are showed in the upper panel. The reference titration curve in which BRD4 protein was titrated into buffer is above the BRD4-5i titration curve. The ITC data fitting is shown in the bottom panel. The final parameters of the analysis are shown in the box. b) The binding of compound 5i in the crystal structure of BRD4 (PDB ID 7T3F). BRD4 structure is shown in gray cartoon. Compound 5i is shown in green sticks. All residues of BRD4 involved in the binding are shown in stick. c) Comparison of PFI and compound 5i in the BRD4 crystal structure. PFI is shown in yellow stick while compound 5i is in green stick.

In vitro and in vivo evaluation of 5i in a model of arsenical-mediated cutaneous BRD4 induction and consequent inflammatory response:

To demonstrate the arsenical-induced BRD4 expression in skin keratinocytes, we treated HaCaT cells, which are immortalized human skin keratinocytes, with PAO (0.5μM, 6h). As can be seen in Fig. 5a, PAO induces nuclear accumulation of BRD4. This can also be confirmed from Western blot analysis of PAO-treated keratinocytes. Co-treatment of these cells with 5i (1μM) diminished nuclear accumulation of BRD4 as well as its expression (Figure 5a). The decrease in BRD4 was accompanied by the decrease in H4 histone acetylation at H4k5ac (Figure 5b). Furthermore, 5i co-treatment also attenuated the PAO-induced BRD4-regulated transcripts of the inflammatory genes. Data shown here, demonstrate a significant decrease in Tfe3, Nfe2l2 and Nlrp3 (Figure 5c).

Figure 5.

Effects of 5i on PAO-induced BRD4 and its inflammatory target genes in HaCaT cells. (a) Immunofluorescence staining of BRD4 (green) in HaCaT cells treated with vehicle control, PAO or PAO+5i. Yellow arrows indicate nuclear localization of BRD4. Nuclei were stained with DAPI (blue). (b) Western blot analysis of BRD4 and H4K5ac. β-actin and H4 was used as endogenous loading control for BRD4 and H4k5ac respectively. (c) RT-PCR analysis of BRD4 target inflammatory genes. p<0.05*, p<0.01** and p<0.001*** show significant levels.

After demonstrating ability of 5i to target BRD4 in our in vitro assay, we confirmed its in vivo potential to inhibit PAO-induced BRD4 expression, transcription activity and regulated inflammatory responses. For this the topical treatment with 5i was done 10 minutes following challenge of Ptch1^+/−/SKH1 mouse skin to PAO as shown in Figure 6a. We observed that 5i treatment attenuated PAO-induced inflammatory response dose-dependently. Significant decrease in skin bi-fold thickness, a marker of inflammation could be observed as early as 3 h following PAO challenge, which was maintained up to 6 h of observation (Figure 6b). This, in vivo decrease in cutaneous inflammation was associated with decrease in the expression of BRD4 and its H3 histone acetylation marks namely H3k9ac and H3k18ac (Figure 6c). However, H4 histone mark H4k5ac was unaffected at this time of analysis. The lower doses of 5i (0.5mg and 1.0mg) were highly effective in reducing both BRD4 expression and its histone marks as further increase in its dose did not add any significant benefit. The decrease in the BRD4-dependent inflammatory response was also confirmed by assessing the impact of 5i on the BRD4-regulated inflammatory transcripts namely Nlrp3, Nfe2l2 and Il6 (Figure 6d). A consistent and highly significant decrease in these transcription target genes suggest that our novel BRD4 inhibitor is highly effective in reducing arsenical-induced cutaneous inflammation in a BRD4-dependent manner.

Figure 6.

Effects of 5i on PAO-induced BRD4 and its inflammatory target genes in mice skin. (a) Exposure paradigm showing treatment protocol. Mouse photographs showing gross changes in the skin at 6 h fallowing PAO (upper) and PAO+5i treatment (lower). (b) % Change in skin bi-fold thickness at 3 and 6 h. (c) Western blot analysis of BRD4, H3k9ac, H3k18ac and H4k5ac at 6 h. β-actin, H3 and H4 were used as endogenous controls. Histogram showing densitometry analysis of band intensity, are shown in % change. (d) RT-PCR analysis of BRD4 target inflammatory genes. p<0.05*, p<0.01**, p<0.001*** and p<0.0001**** show significant levels. N=3/sex/group.

Using in silico studies and synthetic structural modifications, we have designed and prepared a series of compounds, of which eight displayed strong BRD4 inhibitory effects with IC₅₀ values in the sub-microlar range and two compounds exhibited IL-6 inhibition below 1 μM. Compound 5i, in particular, exhibited a potency (IC₅₀) of 0.33 μM for BRD4 and IC₅₀ = 0.83 μM in the IL-6 assay, and both the ITC and crystallographic studies confirmed a specific and tight interaction of 5i with BRD4. Furthermore, compound 5i was advanced to in vivo pharmacokinetic studies and evaluated in vivo for its ability to suppress inflammation resulting from exposure to arsenical. 5i was highly effective in reducing arsenical-induced BRD4 expression, nuclear localization and transcription of genes regulating cutaneous inflammation both in vitro in human skin keratinocytes and in vivo in arsenical challenged murine skin. Overall, our results from this study revealed that the active compounds can lead to chemically distinct scaffolds for the treatment of arsenical-induced inflammation.

Abbreviations:

BRD4

Bromodomain 4

BET

Bromodomain and extra terminal domain

IL-6

interleukin-6

IC₅₀

Half-maximal inhibitory concentration

CDI

Carbonyldiimidazole

HATU

1-[Bis(dimethylamino)methylene]-1H-1,2,3-triazolo[4,5-b]pyridinium 3-oxide hexafluorophosphate

ADME

Absorption, Distribution, Metabolism, and Excretion

HLM

Human Liver Microsome

MLM

MLM, Mouse Liver Microsome

PK

Pharmacokinetic

IP

Intraperetonial

Cmax

Maximum concentration

t_1/2

Half-life

t_max

Time of peak plasma concentration

AUC_last

Area under the curve from the time of dosing to the last measurable concentration

μM

Micromolar

ITC

Isothermal titration calorimetry

ELISA

Enzyme-linked immunoassay

PAO

phenylarsine oxide