Therapeutic potential of NADPH oxidase 1/4 inhibitors

Abstract

The NADPH oxidase (NOX) family of enzymes produces ROS as their sole function and is becoming recognized as key modulators of signal transduction pathways with a physiological role under acute stress and a pathological role after excessive activation under chronic stress. The seven isoforms differ in their regulation, tissue and subcellular localization and ROS products. The most studied are NOX1, 2 and 4. Genetic deletion of NOX1 and 4, in contrast to NOX2, has revealed no significant spontaneous pathologies and a pathogenic relevance of both NOX1 and 4 across multiple organs in a wide range of diseases and in particular inflammatory and fibrotic diseases. This has stimulated interest in NOX inhibitors for therapeutic application. GKT136901 and GKT137831 are two structurally related compounds demonstrating a preferential inhibition of NOX1 and 4 that have suitable properties for in vivo studies and have consequently been evaluated across a range of disease models and compared with gene deletion. In contrast to gene deletion, these inhibitors do not completely suppress ROS production, maintaining some basal level of ROS. Despite this and consistent with most gene deletion studies, these inhibitors are well tolerated and slow or prevent disease progression in a range of models of chronic inflammatory and fibrotic diseases by modulating common signal transduction pathways. Clinical trials in patients with GKT137831 have demonstrated excellent tolerability and reduction of various markers of chronic inflammation. NOX1/4 inhibition may provide a safe and effective therapeutic strategy for a range of inflammatory and fibrotic diseases.

Linked Articles

This article is part of a themed section on Redox Biology and Oxidative Stress in Health and Disease. To view the other articles in this section visit http://onlinelibrary.wiley.com/doi/10.1111/bph.v174.12/issuetoc

Abbreviations

ARDS

acute respiratory distress syndrome

DHE

dihydroethidium

DKD

diabetic kidney disease

DPI

diphenyleneidonium

DUOX

dual oxidase

EC

endothelial cell

FSGS

focal segmental glomerulosclerosis

GIT

gastrointestinal tract

HCV

hepatitis C virus

HIF

hypoxia inducible factor

HPAEC

human pulmonary artery endothelial cells

I/R

ischaemia/reperfusion

IBD

inflammatory bowel disease

IPF

idiopathic pulmonary fibrosis

KD

knock‐down

KO

knock‐out

MCAO

middle cerebral artery occlusion

OHDA

hydroxydopamine

MS

multiple sclerosis

NASH

non‐alcoholic steatohepatitis

NOX

NADPH oxidase

PD

Parkinson's disease

PTM

post‐translational modification

α‐SMA

α‐smooth muscle actin

SMC

smooth muscle cells

XPC

xeroderma pigmentosum C

Tables of Links

TARGETS
Other protein targets a	Enzymes e
Bax	Abl
Bcl‐2	Akt
Notch1	Caspase 3
TNF	FAK
Nuclear hormone receptors b	FLT3
PPARα	HIF‐1α
PPARγ	HIF‐2α
Catalytic receptors c	MAPK
DEP‐1	mTOR
TLR4	Myeloperoxidase
VEGFR	PKC
Voltage‐gated ion channels d	PTEN
TRPV1 channels	Smad2/3
	Xanthine oxidase

LIGANDS
AngII, angiotensin II	NADPH
bFGF	Nicotine
Bone morphogenetic protein 4	NO
β‐Catenin	Paraquat
CTGF	PDGF
DPI, diphenyleneiodonium chloride	Rac1
Fresolimumab	Rac2
GTP	TGFβ1
Hyaluronan	Thioridazine
IL‐β1	VEGF
Imatinib	Wnt

These Tables list key protein targets and ligands in this article which are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY (Southan et al., 2016) and are permanently archived in the Concise Guide to PHARMACOLOGY 2015/16 (^a,b,c,d,eAlexander et al., 2015b, 2015c, 2015a, 2015e, 2015d).

Introduction

The NADPH oxidases (NOX) are the only enzymes known to generate reactive oxygen species (ROS) as their sole function (Bedard and Krause, 2007; Altenhofer et al., 2012). The family has seven members: NOX1–5 and two dual oxidases (DUOX), DUOX1 and DUOX2 (Table 1). The NOX1–5 isoforms have six transmembrane‐spanning α‐helices with cytosolic N‐ and C‐termini. NOX generate superoxide radicals (O₂ ^.) or hydrogen peroxide (H₂O₂) via a NADPH‐dependent one‐ or two‐electron reduction of oxygen. Previously perceived as deleterious agents, ROS are now recognized as important second messengers able to modulate the activity, abundance and subcellular localization of a broad range of target proteins, through reversible oxidation effects, such as sulfenylation of conserved cysteine residues. Accordingly, NOX‐mediated cysteine oxidation represents an important type of post‐translational modification (PTM) of proteins. A pathogenic role for cysteine sulfenylation has now been shown for a rapidly expanding number of target proteins (Reddie et al., 2008; Leonard et al., 2009; Nakanishi et al., 2014; Peters et al., 2014; Wani et al., 2014; Yang et al., 2014). Therefore, NOX inhibition represents a novel therapeutic platform of broad potential, akin to other therapeutic classes which target other PTMs, such as phosphorylation, ubiquitination or acetylation.

Table 1.

Tissue expression, distribution and regulation of NOX/DUOX

Enzyme	Site of expression	Regulatory subunits	ROS produced
NOX1	Colon, vascular smooth muscle, endothelium, placenta, prostate, uterus, skin, osteoclasts, retinal pericytes	p22phox, NOXO1, NADPH oxidase activator, Rac1	Superoxide
NOX2	Phagocytes, β lymphocytes, cardiomyocytes, hepatocytes, vascular smooth muscle, fibroblasts, skeletal muscle, neurons, lung, carotid body, kidney	p22phox, p47phox, p67phox, Rac1, Rac2	Superoxide
NOX3	Fetal tissue, inner ear, neurons	p22phox, NOXO1, Rac1,	Superoxide
NOX4	Kidney, vascular smooth muscle, endothelium, osteoclasts, fibroblasts, keratinocytes, cardiomyocytes, bone, ovary, pancreas, eye, skeletal muscle, testis	p22phox (constitutively active)	Superoxide/H2O2
NOX5	Lymphoid tissue, testis, ovary, prostate, pancreas, spleen	Regulated by calcium and phosphorylation*	Superoxide
DUOX1/2	Thyroid, lung, salivary glands, GIT, prostate, pancreas	Regulated by calcium and phosphorylation**	H2O2

^* Jagnandan et al., 2007,

^** Rigutto et al., 2009

The NOX isoforms differ in their regulation, tissue and subcellular localization and even ROS products (Table 1) (Lambeth, 2004; Brown and Griendling, 2009; Takac et al., 2011). NOX2 was the first NOX identified, and its important role in host defense was revealed by human loss of function mutations (Dinauer et al., 1987; Volpp et al., 1988; Clark et al., 1989) and confirmed by knockout (KO) mouse studies (Pollock et al., 1995; Schappi et al., 2008; Deffert et al., 2012). Because of this essential role, NOX2 was not initially considered as an attractive therapeutic target, although there is recent interest in NOX2 as a target for CNS and other diseases. With the discovery of the other six NOX isoforms, a broad role has emerged as tissue‐specific amplifiers of signal transduction pathways, activated under conditions of acute and chronic stress. NOX1 (Gavazzi et al., 2006) and NOX4 KO mice (Carnesecchi et al., 2011) have not generally revealed significant spontaneous pathologies but have demonstrated a pathogenic role for NOX1 and NOX4 across a broad range of diseases. It has also been suggested that basal levels of NOX4 activity may be protective under specific clinical settings, including acute ischaemia or cardiac pressure overload (Zhang et al., 2010; Babelova et al., 2012). The NOX probably play an important physiological role in acute stress responses and a pathological role after over‐activation, under chronic stress conditions. Here, we review the evidence from KO, molecular knockdown (KD) and pharmacological inhibition for the roles of NOX1 and 4 in health and disease.

Inhibitors

A number of compounds, including interfering peptides, have been described as NOX inhibitors and have been recently reviewed in detail (Gatto et al., 2013, Altenhofer et al., 2015) (Table 2). Compounds more recently published include thioridazine, an N‐substituted phenothiazine, with non‐isoform selective NOX activity (Seredenina et al., 2015); GLX351322, a NOX4‐selective inhibitor (Anvari et al., 2015); and GSK2795039, a direct NOX2 inhibitor (Hirano et al., 2015) (Table 2). Two of the older compounds that have been widely used as NOX inhibitors are diphenyleneidonium (DPI) and apocynin. DPI is a general inhibitor of all flavoproteins and shows no selectivity within the NOX isoforms. Apocynin is not a direct inhibitor of the NOX and probably inhibits by a scavenging mechanism. NOX assays are very difficult to perform as NOXs are only active when embedded in membranes along with their specific regulatory subunits. Moreover, measuring ROS is difficult and prone to many artifacts. For most published compounds, data have not been published showing inhibition across the NOX isoforms in assay systems appropriate to assess direct NOX activity or the relative potency for the various NOX isoforms. Many of the molecules that have been described to date are not direct NOX inhibitors but inhibit NOX pathway activity upstream or downstream of NOX, have no specificity within the NOX family and/or have other non‐NOX targets (Table 2). In addition, many of the published compounds do not have suitable pharmacokinetic properties for in vivo use, and some are toxic at doses required to produce a therapeutic effect, probably because of their off‐target effects.

Table 2.

Known NOX inhibitors

Inhibitors	NOX isoform	Mechanism and other targets	References
Apocynin	All	ROS scavenger inhibits downstream of all isoforms and therefore inhibits effects of all	(Altenhofer et al., 2015)
		No in vitro data showing direct NOX inhibition published
DPI	All	Non‐reversible inhibitor of flavoproteins, including NOX, xanthine oxidase, NOS, mitochondrial electron transport chain	(Altenhofer et al., 2015)
GKT136901	1/4	Preferential inhibition for NOX1/4 over other NOX isoforms	(Laleu et al., 2010; Sedeek et al., 2010; Schildknecht et al., 2014)
		Direct interaction with peroxynitrite, but not NO, superoxide or hydroxyl radicals
GKT137831	1/4	Preferential inhibition for NOX1/4 over other NOX isoforms	(Gaggini et al., 2011; Aoyama et al., 2012)
ML171 and thioridazine	1	Part of the phenothiazine family. Potent antagonists of serotonergic, muscarinic, histaminergic and dopaminergic receptors	(Gianni et al., 2010)
	1–5	ML171 is a false NOX inhibitor, potentially scavenger. Thioridazine behaves as a non‐competitive inhibitor of NOX2	(Seredenina et al., 2015)
GLX351322	4	Unclear	(Anvari et al., 2015)
GSK2795039	2	NADPH competitive	(Hirano et al., 2015)
		Selective over other isoforms
VCC202273	4	Unclear	(Barman et al., 2014;
			Borbely et al., 2010)
VCC588646	4	Unclear	(Barman et al., 2014;
			Borbely et al., 2010)

Two low MW inhibitors belonging to a structural class of pyrazolopyridine diones, GKT136901 (Laleu et al., 2010; Sedeek et al., 2010) and GKT137831 (Aoyama et al., 2012) have been described as preferential direct inhibitors of NOX1 and NOX4 and with suitable pharmacokinetic and tolerability profiles for use in in vivo studies (Table 2). Consequently, these two compounds have been the most extensively utilized in animal models of disease and compared with genetic deletion of the NOX isoforms (Table 3). Both compounds show the highest potency against NOX1 and NOX4 compared with other isoforms when evaluated in assays performed with cell‐free assays using membranes prepared from cells heterologously overexpressing a specific NOX enzyme isoform along with its specific regulatory proteins. In these assays, K_i (nM) for GKT136901 was NOX1 = 160 ± 10, NOX2 = 1530 ± 90 and NOX4 = 16 ± 5. E_max was only 60% for NOX2 and >90% for NOX1 and NOX4. For GKT137831, K_i (nM) was NOX1 = 110 ± 30, NOX2 = 1750 ± 700, NOX4 = 140 ± 40 and NOX5 = 410 ± 100. K_i values for NOX3, DUOX1 and DUOX2 under the same type of assay condition are not published. GKT136901 has been reported to inhibit ROS activity attributed to the DUOXs at μM concentrations using cell lines that endogenously express both DUOXs (Strengert et al., 2014). Caution needs to be exercised, making comparisons of potency with different concentrations of target protein, different methods for measuring ROS, the presence or absence of membranes and using cell‐free versus cell‐based assays.

Table 3.

Summary of genetic and pharmacological evidence for roles of NOX1 or NOX4 in pathogenesis of diseases

Organ and diseases	NOX isoform implicated	Genetic and chemical intervention		Observed effects	Signalling pathways implicated	References
		Genetic/molecular	Chemical
Lung
ARDS	NOX1	NOX1‐deficient mice	GKT136901	Reduced hyperoxia‐induced epithelial/EC death	MAPK and STAT3	(Carnesecchi et al., 2009; Carnesecchi et al., 2014)
			GKT136901	Decreased nicotine‐induced epithelial cell death	Bcl‐2/Bax modulation	(Zanetti et al., 2014)
	NOX4	NOX4 siRNA		Decreased hyperoxia‐induced endothelial migration and tube formation		(Pendyala et al., 2009)
		NOX4 siRNA		Reduced LPS‐induced inflammatory and angiogenesis gene expression	TLR4/NF‐κB pathway	(Park et al., 2004; Park et al., 2006)
Asthma	NOX4	NOX4 siRNA	DPI	Reduced human airway SMC hyper‐contractibility		(Sutcliffe et al., 2012)
			Apocynin
Fibrosis/hypertension	NOX4	NOX4 siRNA	GKT137831	Reduced TGF‐β‐induced fibrotic response in bleomycin (BLM) model	Smad2/3, α‐SMA	(Hecker et al., 2009; Hecker et al., 2014; Jarman et al., 2014)
				Decreased senescence and apoptosis resistance in aged mice exposed to BLM
		NOX4‐deficient mice	GKT136901	Prevention of TGF‐β‐induced epithelial cell death	Caspase 3 modulation	(Carnesecchi et al., 2011)
		NOX4 siRNA	GKT137831	Reduced hypoxia‐induced SMC and endothelial proliferation	PPARγ regulation	(Lu et al., 2010; Green et al., 2012)
Liver
NASH	NOX1 and NOX4	NOX4‐ and NOX1‐deficient mice	GKT137831	Attenuated liver injury induced by CCl4 or bile duct ligation, fibrosis and inflammation	TGF‐β/NF‐κB signalling/Hedgehog/TLR4/PDGF	(Aoyama et al., 2012; Jiang et al., 2012; Sancho et al., 2012; Lan et al., 2015)
Steatohepatitis	NOX4	NOX4‐deficient mice	GKT137831	Decreased liver fibrosis and inflammation in mice with diet‐induced steatohepatitis	PKR and protein kinase RNA‐like endoplasmic reticulum kinase‐mediated stress signalling	(Bettaieb et al., 2015)
Kidney
Nephropathy	NOX4		GKT137831	Reduced renal fibrosis and injury in OVE26 2and AKITA mice	TGF‐β and HIF‐1α signalling	(Gorin et al., 2015; You et al., 2016)
			GKT136901	Reduced oxidative damage in diabetic mice	ERK1/2 pathway	Sedeek et al., 2010, 2013)
		NOX4‐deficient mice	GKT137831	Decreased streptozotocin (STZ)‐induced renal inflammation and fibrosis in ApoE KO mice	NF‐κB and TGF‐β signalling	(Jha et al., 2014)
		NOX4‐deficient mice		Increased progression of kidney fibrosis in ApoE KO/STZ and unilateral urinary obstruction models	HIF‐1α and Nrf2 pathways	(Babelova et al., 2012; Nlandu Khodo et al., 2012)
Bone
Osteoporosis	NOX4	NOX4‐deficient mice	GKT137831	Reduced ovariectomy‐induced osteoporosis	Osteoclast death	(Goettsch et al., 2013)
Osteoarthritis	NOX4	NOX4 overexpressing chondrocytes		Increased IL‐1β induced ROS production and MMP activity	IL‐1β, MMPs	(Grange et al., 2006)
GIT
Colitis	NOX1	NOX1‐deficient mice associated with IL‐10 deficiency		Spontaneous colitis	Endoplasmic reticulum stress	(Treton et al., 2014)
					IL‐10
IBD	NOX1	NOX‐1 deficiency (children)		Development of IBD	Mucosal defense	(Hayes et al., 2015)
Ileocolitis	NOX1	NOX‐1 deficient associated with glutathione peroxidase‐1 and ‐2 deficiencies		Decreased symptoms of ileocolitis	TNF modulation	(Esworthy et al., 2013)
Wound healing	NOX1	NOX1‐deficient mice associated with annexin A1 deficiency		Defects in intestinal mucosal wound repair	Phosphatase and tensin homolog/focal adhesion kinase	(Leoni et al., 2013)
Skin
XPC	NOX1	NOX1‐deficient mice	GKT137831	Rescue of premature skin aging features	DNA damage/repair pathways	(Hosseini et al., 2015)
Scleroderma	NOX4	NOX4 siRNA	GKT137831	Reduction of ROS and collagen production	TGF‐β	(Piera‐Velazquez et al., 2015)
Cardiovascular system
Hypertension	NOX1	NOX1‐deficient mice		Protection against AngII‐induced blood pressure increase and medial hypertrophy	NO bioavailability	(Matsuno et al., 2005; Gavazzi et al., 2006)
	NOX4		GKT136901	Reduction of vascular remodelling and right ventricular hypertrophy	TGF‐β	(Barman et al., 2014)
Atherosclerosis	NOX1	NOX1‐deficient mice and NOX1 siRNA		Inhibition of intimal hyperplasia	p38‐MAPK	(Lee et al., 2008)
		NOX1‐ and ApoE‐ deficient mice under HFD		Reduction of lesions in aortic arch	Cellular proliferation and collagen production	(Sheehan et al., 2011)
		NOX1‐ and ApoE‐ deficient mice under HFD		Increased plasma lipids and enhance atherosclerosis		(Sobey et al., 2015)
		NOX1‐ and diabetic ApoE‐ deficient mice	GKT137831	Anti‐atherosclerotic effect	Immune‐mediated inflammation	(Gray et al., 2013; Di Marco et al., 2014)
			GKT136901	Anti‐atherosclerotic effect	CD44–hyaluronan interaction	(Vendrov et al., 2010)
	NOX4	NOX4‐ and ApoE‐deficient mice		Increased atherosclerosis	Immune‐mediated inflammation	(Schürmann et al., 2015)
Diabetic endothelial dysfunction‐	NOX1	NOX1‐deficient diabetic mice		Protection against endothelial dysfunction induced by diabetes	Rac1	(Vecchione et al., 2006)
I/R	NOX4	NOX4‐deficient mice (cardiac and systemic)		Attenuation of infarct size in I/R model	HIF‐1α/PPARα	(Matsushima et al., 2013)
LVH	NOX4	NOX4‐deficient mice		Reduction of LVH		(Lassegue et al., 2012)
Cardiac remodelling and fibrosis	NOX4	NOX4‐deficient mice	GKT137831	Attenuation of interstitial fibrosis and cardiac remodelling	Akt–mTOR / NF‐κB	(Kuroda et al., 2010; Zhao et al., 2015)
Angiogenesis	NOX1	NOX1 siRNA	GKT136901	Decreased EC migration and tube formation	PPARα	(Garrido‐Urbani et al., 2011)
		NOX1‐deficient mice	GKT137831	Reduced angiogenesis induced by a matrigel plug containing VEGF and basic FGF (bFGF)	VEGF and bFGF pathways	(Garrido‐Urbani et al., 2011)
Eye
Ischaemic retinopathy	NOX1	NOX1‐deficient mice	GKT137831	Reduced oxygen‐induced retinal neovascularization, vascular leakage and neuroglial inflammation		(Wilkinson‐Berka et al., 2013; Wilkinson‐Berka et al., 2014; Deliyanti and Wilkinson‐Berka, 2015)
Diabetic retinopathy	NOX4	NOX4 siRNA		Decreased VEGF‐induced EC proliferation and angiogenesis in oxygen‐induced retinopathy	VEGFR2/STAT3	(Al‐Shabrawey et al., 2008; Li et al., 2015; Vogel et al., 2015)
		NOX4‐deficient mice
	NOX4	NOX4 siRNA		Decreased insulin‐induced EC proliferation	VEGF/VEGFR pathway	(Meng et al., 2012)
Brain
Stroke	NOX4	NOX4‐deficient mice	VAS2870	Protect ischaemic brain injury in preventive and therapeutic mode		(Kleinschnitz et al., 2010)
	NOX1	NOX1 shRNA		Protect ischaemic brain injury (reduced infarction size and neuronal death) in MCAO rat model		(Choi et al., 2015)
Amyotrophic lateral sclerosis (ALS)	NOX1	NOX1‐deficient mice		SOD1G93A ALS mice crossed with NOX1 KO mice survived 23 days longer than SOD1G93A mice		(Marden et al., 2007)
PD	NOX1		GKT137831 apocynin	Inhibition of 6‐OHDA‐induced N27 cell death	MMP3/Rac1	(Choi et al., 2014)
	NOX1	NOX1 shRNA		Protection of dopaminergic neuronal cell death in 6‐OHDA‐ and paraquat‐induced rat model of PD and reduced α‐synuclein aggregation in the paraquat‐induced model		(Choi et al., 2012; Cristovao et al., 2012)
	NOX1		Apocynin	Protection of dopaminergic neuronal cell death in paraquat‐induced mouse model of PD		(Cristovao et al., 2009)
Nerves
Inflammatory pain	NOX1	NOX1‐deficient mice		Attenuation of inflammatory pain induced by formalin, acid acetic and carrageenan	PKC‐ε/TRPV1 pathway	(Ibi et al., 2008)
	NOX1	NOX1‐deficient mice		Increased morphine analgesia and suppression of morphine tolerance	GTPase/PKC/TRPV1 pathways	(Zachariou et al., 2003; Ibi et al., 2008)
Neuropathic pain	NOX4	NOX4‐deficient mice		Decreased peripheral demyelinization	Unknown	(Kallenborn‐Gerhardt et al., 2013)
Multi‐organ: cancer
Tumour angiogenesis	NOX1	NOX1 overexpression		Increased tumourigenicity		(Arbiser et al., 2002)
	NOX1	NOX1‐deficient mice	GKT136901	Reduced tumour angiogenesis	PPARα, NF‐κB	(Garrido‐Urbani et al., 2011)
				Reduced tumour growth
RAS‐induced DNA damage and senescence	NOX1 and 4	NOX1 and NOX4 siRNA		Inhibition of Ras‐induced damage response in fibroblasts	Ras/MAPK	(Kodama et al., 2013)
		NOX1 and NOX4 overexpression		Induction of senescence
	NOX4	NOX4 siRNA		Inhibition of Ras‐induced damage response in thyroid epithelial cells		(Weyemi et al., 2012)
Pancreatic cancer	NOX4	NOX4 siRNA	DPI	Blocked TGF‐β mediated EMT in pancreatic cancer cells	TGF‐β/protein‐tyrosine phosphatase 1B (PTP1B)/N‐cadherin	(Hiraga et al., 2013)
	NOX1–4	p22phox siRNA		Decreased cell proliferation, colony formation, glucose uptake and lactate formation		(Lu et al., 2012)
		p22phox shRNA	DPI	Inhibited growth of subcutaneously xenografted Panc‐1 cells
Hepatocellular carcinoma	NOX1	NOX1 shRNA		Reduced metabolic remodelling in HepG2 cells	Glucose and glutamine metabolism, lipid, protein and nucleotide synthesis	(Bertram et al., 2015)
	NOX4	NOX4‐deficient mice		Enhanced early hepatocarcinoma development in a xenograft model of subcutaneous tumour growth		(Crosas‐Molist et al., 2014)
Renal cell carcinoma (RCC)	NOX4	NOX4 shRNA		Reduced nuclear accumulation of HIF‐2α in von Hippel–Lindau (VHL)‐deficient RCC cells	HIF‐2α	(Gregg et al., 2014)
				Blocked cell branching, invasion colony formation and growth in a xenograft model
	NOX4	NOX4 siRNA		Reduced expression of HIF target genes, VEGF, TGF‐β and Glut1 in VHL‐deficient RCC cells		(Maranchie and Zhan, 2005)
	NOX4	NOX4 shRNA	DPI	Enhanced cytotoxicity of cisplatin, vincristine, and etoposide in two RCC cell lines	Bcl‐XL and Bcl‐2	(Chang et al., 2012)
Glioblastoma	NOX4	NOX4 siRNA	DPI or shikonin	Inhibition of NOX4‐induced cell proliferation and migration	PTP1B and Akt/HIF‐1α pathways	(Mondol et al., 2014; Gupta et al., 2015)
	NOX4	NOX4 siRNA		Reduced resistance to radiation in cell lines	HIF‐1α pathways	(Hsieh et al., 2012; Li et al., 2014)
		NOX4 shRNA		Reduced resistance to radiation in a xenograft model		(Hsieh et al., 2012)
Prostate cancer			DPI and apocynin	Reduce androgen induced radiation resistance		(Lu et al., 2010)
Melanoma	NOX1, NOX4	siRNA	DPI, VAS2870, apocynin	Inhibited invasion		(Liu‐Smith et al., 2014)
				Induced cell death
Non‐small cell lung cancer	NOX1	NOX1 overexpression	ML171	Increased metastasis	TLR4	(Liu et al., 2015)
				Decreased metastasis
Breast cancer	NOX4	Dominant negative NOX4		Blocked TGF‐β‐mediated EMT	TGF‐β, EMT markers	(Boudreau et al., 2014)
Acute myeloid leukemia (AML)	NOX4	NOX4 siRNA	DPI	Reduced DNA damage in mutant Fms‐like tyrosine kinase 3 (FLT3)‐expressing AML cells	FLT3	(Stanicka et al., 2015)
	NOX4	NOX4 siRNA	Schisandrin B	Reduced ROS and increase protein tyrosine phosphatase (DEP)‐1 activity and reduced growth of FLT3 internal tandem duplication (FLT3ITD) transformed cells	FLT3/DEP‐1 PTP/STAT5	Jayavelu et al., 2015
		NOX4 shRNA	GKT137831	Slowed disease development in vivo after i.v. injection of the FLT3ITD‐transformed cells in syngeneic mice
Chronic myeloid leukemia (CML)	All NOX		DPI and apocynin	Enhanced anti‐proliferative effects of BCR‐ABL inhibitors in two CML cell lines		(Sanchez‐Sanchez et al., 2014)
				DPI slowed tumour growth in vivo to a similar extent to imatinib, and combination was more efficacious than either drug alone

Neither GKT136901 nor GKT137831 was active in a counterscreen using xanthine oxidase using the same readout and conditions as in the NOX assays (Ki > 100 μM) (Sedeek et al., 2010; Aoyama et al., 2012). This rules out a general scavenging mechanism. In contrast, DPI showed high and similar potency across all the NOX isoforms as well as for xanthine oxidase (K_i = 10 to 70 nM), consistent with its mechanism of action as a general inhibitor of flavoproteins. Neither GKT136901 nor GKT137831 showed any potent activity against a broad panel of targets, such as GPCRs, kinases, ion channels and other enzymes including myeloperoxidases and other ROS‐producing and redox‐sensitive enzymes (Sedeek et al., 2010; Aoyama et al., 2012).

Although GKT136901 does not inhibit xanthine oxidase activity using amplex red and HRP to measure H₂O₂ generation, it has been reported to inhibit the reaction of exogenously added H₂O₂ with amplex red and HRP in the μM range (Altenhofer et al., 2015). The reason for this reported difference between enzymatically generated and exogenously added H₂O₂ is unclear. A reducing activity of GKT136901 under these specific assay conditions may lead to apparent HRP inhibition (Dikalov and Harrison, 2014). Interestingly, this phenomenon has also been reported for a structurally unrelated NOX2 inhibitor, GSK2795039, when HRP and amplex red are used to measure ROS but not with other ROS readouts, and has been attributed to a reducing activity of GSK2795039 (Hirano et al., 2015).

A scavenging activity for peroxynitrite by GKT136901 at sub‐μM concentrations has been reported (Schildknecht et al., 2014). This was specific for peroxynitrite as GKT136901 did not interact with •NO, O₂ ^. or hydroxyl radicals (•OH). The lack of activity of GKT136901 in the xanthine oxidase counterassay also rules out a general scavenging activity. Whether the peroxynitrite scavenging observed in vitro contributes to any efficacy in vivo is not clear. A recent publication (Hirschhäuser et al., 2015) also shows no general ROS scavenging activity of GKT136901 in mitochondrial fractions from mouse kidney, but a small inhibition of ROS generation was observed in cardiac mitochondria from both wild‐type and NOX4 KO mice. This was observed at μM concentrations and could represent inhibition of another NOX isoform or possibly some scavenging mechanism in cardiac but not kidney mitochondria.

The inhibition of NOX2 observed in vitro in the μM range in the membrane assays, with GKT136901 and GKT137831, is probably not of relevance for their efficacy in vivo. GKT137831 did not significantly inhibit a highly specific NOX2‐driven response (neutrophil oxidative bursts) up to 100 μM, as measured by flow cytometry in human whole blood (Aoyama et al., 2012). It also did not inhibit innate microbial bacterial killing in vitro or in vivo (when used at a concentration of up to 20 μM or administered at 100 mg·kg⁻¹ p.o. respectively). In contrast to what is observed in NOX2 KO mice and patients with mutations in the NOX2 promoter, there was no immunosuppression or increased susceptibility to infection observed in mice treated with either of these NOX inhibitors.

Both GKT136901 and GKT137831 have good pharmacokinetic properties and are p.o. active after oral administration once or twice daily (Laleu et al., 2010; Gaggini et al., 2011; Gorin et al., 2015). GKT137831 has a longer biological half like than GKT136901 as it forms an active metabolite, GKT138184, in both rodents (Gorin et al., 2015) and patients (Wiesel et al., 2012). Doses of GKT137831 required to inhibit NADPH‐dependent ROS generation in tissues in mouse models of disease, range from 5 to 60 mg·kg⁻¹ p.o. once daily, with significant inhibition with doses from 10 mg·kg⁻¹·day⁻¹ and above (Table 3). ROS in tissues is very difficult to measure, prone to artifacts and challenging to prove that it is NOX specific (Rezende et al., 2016). In studies with GKT137831, ROS has been measured by several different methods including lucigenin, amplex red, DHE and DHE with HPLC. All have shown that GKT137831 decreased ROS in a dose‐dependent manner but not completely. Efficacy on disease progression in various tissues has been observed in doses from 10 mg·kg⁻¹ once daily. Many studies have been performed with a single high dose of 60 mg·kg⁻¹.

GKT137831 is the only NOX inhibitor that has progressed through preclinical development and into clinical trials (Wiesel et al., 2012; http://www.genkyotex.com/Genkyotex/assets/File/PRESS%20RELEASE%20Ph2%2009‐09‐15%20FINAL.pdf).

Diseases by organ

Lung

ROS produced by NOX are crucial contributors to the pathophysiology of acute respiratory distress syndrome (ARDS). NOX1 is highly expressed in lung epithelial and endothelial cells (ECs) and has been shown to mediate alveolo‐capillary barrier dysfunction, one of the classical features of the acute stage of ARDS. NOX1 KD mice exposed to hyperoxia, a model of acute lung injury, showed reduced pulmonary oedema, accumulation of protein‐rich fluid into the interstitium and alveolo‐capillary damage (Carnesecchi et al., 2009; Carnesecchi et al., 2014). NOX1 was shown to participate in hyperoxia‐induced epithelial and EC death through activation of both MAPK and STAT3 pathways. These results were confirmed with the use of NOX1 KD and GKT136901.

NOX4 has also been shown to contribute to endothelial dysfunction. NOX4 specific siRNA reduced hyperoxia‐induced endothelial migration and tube formation in human pulmonary artery ECs (HPAECs) (Pendyala et al., 2009). NOX4 expressed in epithelial type II cells is reported to mediate cell death induced by TGF‐β (Carnesecchi et al., 2011); inhibition of NOX4 with GKT136901 as well as permanent deletion of NOX4 led to decreased death of type II epithelial cells, induced by TGF‐β. NOX4 participated in LPS/toll‐like receptor 4 (TLR4)‐induced ROS generation and expression of inflammatory and angiogenesis genes through NF‐κB activation in human aortic ECs (Park et al., 2004; Park et al., 2006). NOX4 participates also in the persistence of pulmonary oedema in ARDS through the modulation of epithelial sodium channel trafficking driven by TGF‐β. Genetic deletion of NOX4 protected mice from bleomycin‐induced lung fluid imbalance (Peters and al., 2014).

Nicotine is one of the most important inducing factors in the pathogenesis of lung pathologies including ARDS. Epithelial cells exposed to nicotine exhibit features of apoptosis and show high levels of oxidative stress. NOX1 is a crucial factor in apoptosis of lung epithelial cells exposed to nicotine. Inhibition of NOX1 with GKT136901, as well as its deletion, decreased nicotine‐induced ROS production and apoptosis by modulating (Bcl‐2)/Bax expression (Zanetti et al., 2014).

The NOX enzymes, and in particular NOX4, are implicated in the pathogenesis of lung fibrosis as well as in its progression (Crestani et al., 2011; Hecker et al., 2012). NOX4 is highly expressed in lung fibroblasts from patients suffering from idiopathic pulmonary fibrosis (IPF) (Hecker et al., 2009; Amara et al., 2010; Hecker et al., 2014) and in type II epithelial cells (Carnesecchi et al., 2011). Intratracheal instillation of NOX4 siRNA decreased TGF‐β‐dependent fibrotic responses in mice exposed to bleomycin challenge. NOX4 KD decreased ROS production, differentiation of fibroblasts into myofibroblasts and accumulation of extracellular matrix through the regulation of TGF‐β/Smad2/3 signalling (Hecker et al., 2009). In addition, RNAi‐mediated KD of NOX4 in fibroblasts isolated from patients with IPF reduced Smad2/3 activation and the expression of α‐smooth muscle actin (α‐SMA) and procollagen Iα1. Treatment with GKT137831 attenuated the progression of lung fibrosis following bleomycin exposure by modulating TGF‐β‐specific signalling pathways (Jarman et al., 2014).

The apoptosis of alveolar epithelial cells is an early event in the development of fibrosis. This phenotype was shown to be decreased in NOX4 KD mice exposed to bleomycin (Carnesecchi et al., 2011). In addition, inhibition of NOX4 with GKT136901 inhibited the apoptosis of type II epithelial cells induced by TGF‐β (Carnesecchi et al., 2011). The risk of developing pulmonary fibrosis is known to increase with age and NOX4 participates in age‐associated impaired resolution of fibrosis (Hecker et al., 2014). The therapeutic treatment of aged mice, with GKT137831, as well as with an NOX4 siRNA, leads to the resolution of lung fibrosis following bleomycin challenge. The senescence and apoptosis resistance of myofibroblasts that are associated with the persistence of fibrosis, as well as a decreased Nrf2 antioxidant response, were related to a persistent NOX4 expression and activity. Indeed, IPF lung fibroblasts treated with GKT137831 and NOX4‐deficient fibroblasts isolated from aged mice showed a pattern of reduced senescence and decreased apoptosis resistance.

One of the major complications in IPF and in airway remodelling during chronic lung diseases, such as asthma, is pulmonary hypertension. One strain of NOX1‐deficient mice has been reported to spontaneously exhibit a pulmonary hypertension phenotype in normoxia (Iwata et al., 2014). The number of pulmonary smooth muscle cells (SMCs) was significantly increased in NOX1‐deficient mice, and this was correlated with decreased apoptosis, mainly due to the regulation of the K_v1.5 ion channel and intracellular potassium levels. However, other NOX1‐deficient mice did not exhibit this phenotype, and the number of Human Pulmonary Artery Endothelial Cells (HPAECs) and the subsequent blood pressure were identical to wild‐type mice (Gavazzi et al., 2006).

NOX4 could play a role in the development of pulmonary hypertension as NOX4 has been detected in the media and intima of thickened pulmonary arteries of IPF patients and its presence was correlated with α‐SMA expression (Pache et al., 2011). In addition, NOX4 mRNA was up‐regulated in asthmatic airway SMC compared with healthy control and was shown to contribute to the hyper‐contractibility of human asthmatic airway SMC, responsible for airflow obstruction and airway responsiveness (Sutcliffe et al., 2012). Another study demonstrated that chronic exposure to hypoxia increased NOX4 expression and SMC proliferation, which was inhibited by siRNA NOX4 (Lu et al., 2010). A reduction of PPARγ, an important regulator of SMC/EC growth and proliferation, was also observed in human pulmonary artery SMC and ECs under hypoxic conditions (Lu et al., 2010). Mouse, as well as human, pulmonary artery EC and SMC treated with GKT137831 exhibited a reduction of hypoxia‐induced H₂O₂ release, proliferation, TGF‐β expression and restoration of PPARγ expression (Green et al., 2012). In addition, treatment with GKT137831 attenuated hypoxia‐induced right ventricular hypertrophy and pulmonary artery remodelling in vivo.

In summary, many studies indicate that inhibition of NOX1 and 4 could be a novel therapeutic treatment for a range of acute and chronic respiratory diseases.

Liver

In most types of chronic liver disease, liver fibrosis, characterized by the accumulation of extracellular matrix proteins and the loss of hepatocytes, occurs as a result of the wound healing response to hepatic injury. The accumulation of fibrotic tissue distorts the hepatic architecture with eventual development of nodules of regenerating hepatocytes (cirrhosis). This can result in portal hypertension, liver failure and hepatocellular carcinoma (Paik et al., 2014).

ROS stimulate the production of type I collagen by hepatic stellate cells (HSCs) (Maher et al., 1994; Nieto et al., 2000), and NOX1 and NOX4 appear to be the primary sources of ROS. NOX1‐deficient HSCs have less ROS generation than NOX2 KO (Paik et al., 2011). Elevated liver fibrosis was reported in mice possessing an SOD1 mutation imparting enhanced NOX activity. NOX1/4 inhibition with GKT137831 attenuated liver fibrosis and ROS production in this model (Aoyama et al., 2012). NOX4 transcript and protein levels have also been shown to be elevated during fibrogenesis in HSCs, with HSC activation attenuated in NOX4 KO models (Jiang et al., 2012).

Pharmacological inhibition or genetic deletion of NOX1 or 4 prevents fibrosis in several models of fibrotic liver disease. Non‐alcoholic fatty liver disease can progress to non‐alcoholic steatohepatitis (NASH), and activation of NOX1 and NOX4 has been proposed as a key mechanism in the hepatocyte injury, inflammation and fibrosis characteristic of NASH (Basaranoglu et al., 2013; Takaki et al., 2013; Lan et al., 2015). NOX1 and NOX4 mRNA and protein levels are increased in human livers with cirrhosis and NASH compared with normal controls (Bettaieb et al., 2015; Lan et al., 2015). In murine models of steatohepatitis or NASH, global or hepatocyte‐specific NOX4 gene deletion and GKT137831 significantly reduced liver inflammation and fibrosis (Bettaieb et al., 2015). Similarly, in mouse models of fibrosis induced by bile duct ligation or carbon tetrachloride (CCl₄), fibrosis progression was accompanied by elevated expression of NOX4. NOX4 KO or inhibition using GKT137831 decreased ROS production, attenuated fibrosis and decreased hepatocyte apoptosis (Jiang et al., 2012; Sancho et al., 2012). Deficiency of either NOX4 or NOX1 and treatment with the NOX1/4 inhibitor GKT137831 reduced ROS production, hepatocyte apoptosis and liver inflammation and fibrosis (Jiang et al., 2012; Sancho et al., 2012; Lan et al., 2015). While the role of NOX4 as an important effector of TGF‐β‐mediated fibrogenesis has been well characterized, its role and that of NOX1 have been recently extended to additional fibrogenic pathways including the hedgehog, TLR4 and PDGF pathways and to the crosstalk between inflammatory and fibrogenic pathways through the down‐regulation of the TGF‐β pseudo‐receptor BAMBI (Lan et al., 2015). Separately, increased expression of NOX1 and NOX4 has been reported in hepatocytes expressing hepatitis C virus (HCV) cDNA constructs, core protein or viral RNA. Importantly, NOX4 gene silencing or an inactive NOX4 dominant negative reduced ROS production in HCV‐transfected hepatocytes (Boudreau et al., 2009; de Mochel et al., 2010).

Thus overall, these studies indicate that inhibition of both NOX1 and NOX4 may prove to be a novel and effective approach for the treatment of liver fibrosis of multiple causes.

Kidney

NOX4 seems to be the NOX isoform that plays the most important role in fibrotic kidney diseases, particularly diabetic kidney disease (DKD) and focal segmental glomerulosclerosis (FSGS). Inducible NOX4 transgenic mouse with specific expression of NOX4 in podocytes reproduced all the glomerular features of DKD, characterized by glomerular hypertrophy, mesangial matrix expansion, podocytes dropout and glomerular basement membrane thickening (You et al., 2016). In OVE26 mice and AKITA mice, two models of type 1 diabetes mellitus with progressive renal disease, administration of GKT137831 to animals with established kidney disease reduced fibrosis, glomerular hypertrophy, mesangial matrix expansion, urinary albumin excretion and podocyte loss (Gorin et al., 2015; You et al., 2016). In addition, in a long‐term model of apolipoprotein E (ApoE) KO streptozotocin mice, NOX4 KO, but not NOX1 KO, conferred renal protection from glomerular injury. Moreover, NOX1 and NOX4 inhibitions were renoprotective in a mouse model of type 2 diabetes (Sedeek et al., 2010; Sedeek et al., 2013). In cultured podocytes, NOX4 KO reduced ROS production and down‐regulated pro‐inflammatory and pro‐fibrotic markers of DKD. In addition, administration of GKT137831 reproduced the effects of NOX4 KO (Jha et al., 2014). These results strongly support the therapeutic potential of NOX4 inhibition for the treatment of fibrotic renal diseases, such as DKD and FSGS.

Nevertheless, there are also some conflicting data suggesting that NOX4 has a renoprotective role in diabetes (Babelova et al., 2012). However, mice used in this study have a genetic background resistant to the development of renal disease (Gurley et al., 2006). Possibly because of this strain effect and the short observation period, minimal kidney disease developed in wild‐type or NOX4 KO mice made diabetic with streptozotocin. Slight renal hypertrophy and glomerular hyperfiltration developed to a similar extent in WT and NOX4 KO mice (as shown by the significantly lower cystatin C levels and the trend towards increased creatinine clearance). However, albuminuria was not significantly elevated when corrected for creatinine clearance. Accordingly, NOX4 gene deletion did not worsen the renal phenotype in diabetic mice. No protective effect could be expected in the absence of kidney disease in WT mice. In fact, the authors could not detect NOX4 protein in the glomeruli of normal or diabetic WT mice, which is in agreement with the mouse strain and short observation period used in the study. In contrast, NOX4 gene deletion and treatment with GKT137831 were markedly protective in the study by Jha et al., where a susceptible mouse strain and a longer observation period were used, leading to increased renal expression of NOX4.

Another study has indicated that complete NOX4 deficiency may be detrimental following unilateral urinary obstruction (Nlandu Khodo et al., 2012). Because of the aggressive phenotype of this model and lack of pathophysiological relevance to DKD or FSGS, its relevance as a model of human chronic kidney disease is questionable. In addition, the authors demonstrated that deleterious effects observed in NOX4 KO mice were primarily due to a compensatory increase in NOX2 expression (Nlandu Khodo et al., 2012), an event not observed following pharmacological inhibition of NOX4.

Overall, the majority of studies using genetic and pharmacological inhibition indicate that NOX4 is an attractive therapeutic target for the treatment of kidney diseases, particularly DKD. However, the possibility that total loss of NOX4 activity, while unlikely to be achieved through pharmacological inhibition, may be detrimental in specific clinical settings cannot be ruled out. Accordingly, renal function should be carefully monitored in clinical trials assessing the safety and efficacy of NOX4 inhibitors in patients with kidney disease.

Bone

NOX enzymes are highly expressed on osteoclasts, the macrophage‐like cells that are responsible for bone resorption. Osteoclasts play an important role in bone regeneration, but increased activity of these cells is associated with bone diseases such as osteoporosis and osteoarthritis (Bedard and Krause, 2007). NOX enzymes are associated with osteoclast activity, where increased bone resorption is accompanied by elevations in NOX2 and NOX4 expression (Bedard and Krause, 2007). A single nucleotide polymorphism of NOX4 (rs11018628) was recently associated with elevated NOX4 mRNA expression as well as reduced bone density and increased circulating markers of bone turnover in middle‐aged women (Goettsch et al., 2013). Moreover, human bone obtained from patients with increased osteoclast activity exhibited increased NOX4 expression (Goettsch et al., 2013). In addition, the absence of NOX4 in a murine model of ovariectomy‐induced osteoporosis resulted in higher bone density and reduced osteoclast numbers, with attenuated osteoclastogenesis. These findings suggest that NOX4 may represent a target for the treatment of osteoporosis. Supporting this, chemical inhibition of NOX4 using GKT137831 or acute genetic KD of the enzyme mitigated the loss of trabecular bone in a murine model of ovariectomy‐induced osteoporosis (Goettsch et al., 2013).

NOX4 has also been implicated in the pathogenesis of osteoarthritis. The pro‐inflammatory cytokine IL‐1β induced NOX4‐mediated ROS production (Grange et al., 2006). Also, NOX4 is 100‐fold up‐regulated in a chondrocyte model of pseudoachondroplasia, which is characterized by abnormal joint architecture, joint erosion and osteoarthritis (Coustry et al., 2012). Therefore, NOX4 may also represent a therapeutic target in osteoarthritis.

Gastrointestinal tract

NOX1 is the major isoform expressed in the gastrointestinal tract (GIT) (Adare et al., 2015) and is highly expressed in the gastric epithelium (Teshima et al., 2000). Helicobacter pylori LPS activates NOX1 and increases expression of NOX1 and its cytosolic subunit NADPH oxidase organizer (NOXO1) (Kawahara et al., 2001; Kawahara et al., 2005). NOX1 is also highly expressed in the colon and follows a gradient with low levels in the proximal colon and high levels in the distal colon (Kikuchi et al., 2000; Bates et al., 2002; Szanto et al., 2005). Both proliferation and differentiation of progenitor colon cells are controlled by NOX1 through the modulation of the balance between NOTCH1 and Wnt/β‐catenin signalling pathways (Coant et al., 2010). Oxidative stress is involved in the pathogenesis of inflammatory bowel disease (IBD) (McKenzie et al., 1996), and because NOX1 is highly expressed in the colon, it could be a key driver of this condition. In situ hybridization studies of bowel biopsies from patients with Crohn's disease or ulcerative colitis demonstrated expression of NOX1 in lesional lymphocytes (Szanto et al., 2005). However, NOX1 deficiency has been reported to increase susceptibility to IBD in patients (O'Neill et al., 2015), suggesting that NOX1 has a protective role, not a pathological role. A basal level of NOX1‐derived ROS may be required for normal bowel function.

To date, there are no studies using pharmacological inhibitors of NOX1. Studies with selective NOX1 inhibitors may help to better understand its physiological versus pathological roles in the GIT.

Skin

Keratinocytes generate ROS in response to UV light (Beak et al., 2004; Wang and Kochevar, 2005) and to phorbol esters (Steinbrenner et al., 2005). NOX1, NOX2 and NOX4 isoforms have been detected at the mRNA level in keratinocytes (Chamulitrat et al., 2004).

Oxidative stress and NOX have been implicated in several skin diseases, especially fibrotic diseases (Wan and Evans, 1999; De Felice et al., 2009; Wagner et al., 2012; Babalola et al., 2014; Grygiel‐Gorniak and Puszczewicz, 2014). NOX4 has been proposed as a potential target for systemic sclerosis (scleroderma) (Dooley et al., 2012; Bohm et al., 2014; Spadoni et al., 2015). For example, expression of NOX4 is increased in human sclerodermic fibroblasts (Piera‐Velazquez et al., 2015; Rice et al., 2015). TGF‐β‐associated genes including NOX4 and connective tissue growth factor are down‐regulated after treatment with fresolimumab, a high‐affinity neutralizing antibody that targets all three TGF‐β isoforms (Rice et al., 2015). Treatment with GKT137831 reduces fibroblast collagen and fibronectin production in sclerodermic context (Piera‐Velazquez et al., 2015).

NOX1 has been implicated in skin aging. Activation of NOX1 is responsible for ROS generation in human fibroblasts with KD of xeroderma pigmentosum C (XPC), a protein involved in DNA repair mechanisms. NOX1 down‐regulation completely blocked both cytoplasmic and mitochondrial ROS generation in XPC KD human fibroblasts (Rezvani et al., 2011). NOX1/4 inhibition with GKT137831 blocked premature skin aging in XPC KO mice (Hosseini et al., 2015).

In conclusion, it could be beneficial to inhibit NOX1 as well as NOX4 in skin aging and fibrotic diseases.

Cardiovascular system

Many studies implicate NOX enzymes in cardiovascular physiology and pathology (Cave et al., 2006; Lassegue and Griendling, 2009; Brandes et al., 2010). The observation that statins have beneficial effects partly through a NOX inhibitory action (Bokoch and Prossnitz, 1992; Nakagami et al., 2003) reinforces the therapeutic potential of NOX inhibition for cardiovascular diseases. NOX1 is expressed mainly in vascular SMC although expression in EC and fibroblasts has also been reported (Sorescu et al., 2002). NOX4 is expressed in all cardiovascular cell types (Peng et al., 2005; Sirker et al., 2011). Individual cell types can co‐express more than one NOX isoform, indicating distinct roles and functions of NOX isoforms (Briones et al., 2011).

NOX1 plays a central role in vascular hypertension. NOX1 is overexpressed in transgenic hypertensive rats overexpressing the renin 2 gene (Wingler et al., 2001), in stroke‐prone spontaneously hypertensive rats (Akasaki et al., 2006) and in response to angiotensin II (AngII) (Dikalova et al., 2005). NOX1 KO mice are protected against AngII‐induced increase in blood pressure and medial hypertrophy (Matsuno et al., 2005; Gavazzi et al., 2006). NOX1 is also one of the main isoforms involved in aortic medial hypertrophy (Katsuyama et al., 2002).

NOX4 is also involved in hypertension (Paravicini et al., 2004) but could also exert beneficial effects on vasodilator function and blood pressure during cardiovascular disease (Ray et al., 2011; Schroder et al., 2012). Pharmacological inhibition of ROS production with VCC202273, VCC588646 and GKT136901 reduced vascular remodelling as well as ameliorating right ventricular hypertrophy in rats (Barman et al., 2014).

Many aspects of atherosclerosis are influenced by NOX‐derived ROS (Lambeth, 2007). Oscillatory sheer stress (OS), which occurs in regions of arteries associated with atherogenesis, provokes overexpression of NOX1 and NOX4 (Hwang et al., 2003; Sorescu et al., 2004). Bone morphogenic protein 4 mediated the OS‐induced NOX1 by promoting oxidative stress and monocyte adhesion (Jo et al., 2006). Balloon injury increases NOX1 expression, leading to restenosis and atherosclerosis (Szocs et al., 2002). One of the regulatory subunits of NOX1, NADPH oxidase activator, is up‐regulated in aortas and lesions of ApoE KO mice (Niu et al., 2010). In vivo and in vitro deletions of NOX1 inhibit intimal hyperplasia and formation of macrophage‐derived foam cells (Lee et al., 2008; Lee et al., 2009). NOX1 gene deletion decreases ROS and lesion area in aortic arch in ApoE KO mice maintained in high‐fat diet (HFD) condition (Sheehan et al., 2011). In contrast, another study reported that NOX1 gene deletion increases plasma lipids and enhances HFD‐induced atherosclerosis (Sobey et al., 2015). NOX4 expression is also increased in atherosclerosis lesions (Lee et al., 2010). A recent study indicates that NOX4 has endogenous anti‐atherosclerotic functions (Schürmann et al., 2015). NOX1 but not NOX4 gene deletion demonstrated anti‐atherosclerotic effects in diabetic ApoE KO mice associated with reduced ROS production (Gray et al., 2013). These findings are in agreement with the beneficial effect GKT137831 or GKT136901 in the same disease model (Vendrov et al., 2010; Di Marco et al., 2014). NOX enzymes, especially the Ras‐related C3 botulinum toxin (Rac)‐regulated isoforms NOX1 and NOX2 (Wendt et al., 2005), play a role in the endothelial dysfunction induced by diabetes. A dominant negative inhibition of Rac1 (and therefore inhibition of NOX1) protected EC against endothelial dysfunction in diabetic mice (Vecchione et al., 2006). Moreover, Rac1 inhibition has been proposed as the mechanism responsible for cardiovascular protective effects of statins (Vecchione et al., 2007).

Less is known about the functions of NOX in myocardial ischaemia–reperfusion (I/R) injuries. I/R has been associated with increased circulating and myocardial levels of cytokines and NOX1, NOX2 and NOX4 (Guggilam et al., 2006). Systemic and cardiac‐specific NOX4 KO mice exhibit attenuation of infarct size after I/R (Matsushima et al., 2013). Interestingly, double NOX2 and NOX4 KO mice have larger infarct size after I/R despite the fact that double KO mice have lower oxidative stress than single KO mice. A certain amount of ROS derived from NOX may be required to protect the heart via signalling pathways involving hypoxia‐inducible factor‐1α (HIF‐1α)–PPARα‐dependent mechanisms (Matsushima et al., 2014).

During chronic heart failure resulting from hypertension or I/R, the heart adapts to increase systemic pressure load through left ventricular hypertrophy (LVH). NOX4 has been implicated in pressure overload LVH (Byrne et al., 2003). LVH is reduced in NOX4 KO and exacerbated in NOX4‐overexpressing mice (Lassegue et al., 2012). After pressure overload LVH, NOX4 KO mice develop greater hypertrophy than controls, and NOX4‐overexpressing mice are protected against LVH (Sirker et al., 2011). These results indicate that, during pressure overload‐induced LVH, NOX4 is beneficial.

Left ventricle remodelling, characterized by interstitial fibrosis and cardiomyocyte hypertrophy in response to cardiac injury, leads to heart failure. Several studies implicate NOX, especially NOX4, in cardiac remodelling and fibrosis. NOX4 is expressed in aortic adventitial fibroblasts of rabbit and mouse, but mRNAs for NOX4 were only detected in human fibroblasts (Cucoranu et al., 2005; Jiang et al., 2014). In the latter study, TGF‐β up‐regulated NOX4 mRNA expression and NOX activity and induced the expression of the myofibroblast marker, α‐SMA. Cardiac‐specific NOX4 KO mice exhibit attenuated interstitial fibrosis and better cardiac function than wild‐type mice after pressure overload (Kuroda et al., 2010). Another study using cardiac‐specific human NOX4 transgenic mice demonstrated that up‐regulation of NOX4 provokes cardiac remodelling through activating Akt–mammalian target of rapamycin (mTOR) and NF‐κB signalling pathways. Supporting a pathological role of NOX4, treatment with GKT137831 attenuated cardiac remodelling (Zhao et al., 2015).

Taken together, these many studies suggest a role for NOX1 and 4 in various cardiovascular pathologies. Conversely, complete loss of NOX1 or NOX4 may have detrimental cardiovascular effects in the context of specific cardiovascular disorders, and therefore, careful monitoring should be implemented during clinical trials assessing NOX1 and/or NOX4 inhibitors.

Eye

Ischaemic retinopathies, which are characterized by capillary degeneration and pathological neovascularization, are induced by prematurity and diabetes. Retinopathy of prematurity is the major cause of blindness in neonates and occurs when preterm infants are exposed to hyperoxic conditions following normoxia. This change in the micro‐environment leads to the release of pro‐angiogenic factors such as VEGF and increased ROS production by vascular cells, as well as activation of inflammatory and immune cells of the retina. This leads to the disruption of normal vascularization leading to vasculopathy. NOX1, 2, 4 and 5 are expressed in human retinal tissue (Wilkinson‐Berka et al., 2013). Deficiency of NOX1 but not NOX4 or NOX2, decreased retinal neovascularization, vaso‐obliteration, vascular leakage, leukostasis and microglial density, in a mouse model of oxygen‐induced retinopathy. In contrast, NOX4 KO, as well KD, was reported to reduce neo‐vascularization in the same model through the regulation of VEGF/VEGFR2 pathways (Li et al., 2015; Vogel et al., 2015). The protective effect of NOX4 deficiency could be related to the different KO constructs as well as the breeding conditionsof the mice. Nevertheless, it has been demonstrated that NOX1/4 inhibition with GKT137831 suppressed retinal neovascularization and neuroglial cell inflammation and protected from oxygen‐induced retinopathy (Wilkinson‐Berka et al., 2014; Deliyanti and Wilkinson‐Berka, 2015). Thus, taken together, all these studies suggested that both NOX1 and NOX4 may participate to the pathogenesis of retinopathy of prematurity.

NOX4 has also been shown to be involved in impaired retinal vascular permeability observed in diabetes. Indeed, NOX4 KO reduced ROS production, VEGF expression and retinal vascular permeability in db/db mice (Li et al., 2010). Similarly, a NOX4‐specific siRNA decreased VEGF expression, cell migration, proliferation and tube formation in human microvascular ECs treated with insulin (Meng et al., 2012).

Thus, inhibition of NOX1 and 4 could be a novel therapeutic treatment for ischaemic retinopathy.

Brain

NOX enzymes, particularly NOX1, 2 and 4 are expressed in the brain, and their involvement in different pathologies of the CNS has been reviewed in detail elsewhere (Nayernia et al., 2014). The main NOX isoform found in the brain is NOX2, which is implicated in neurodegenerative and psychiatric disorders. However, the other isoforms may also play a role.

NOX1 and its subunit NOXO1 are expressed in macrophages, activated microglia, astrocytes and ECs in and around active plaques of acute and relapsing multiple sclerosis (MS) (Fischer et al., 2012) and could therefore be an interesting target in this pathology. NOX1 has also been implicated in Alzheimer's disease (Cheret et al., 2008), in amyotrophic lateral sclerosis (Marden et al., 2007) and in Parkinson's disease (PD). NOX inhibition by GKT137831 or apocynin significantly reduced 6‐hydroxydopamine (6‐OHDA)‐mediated N27 dopaminergic cell death in vitro (Choi et al., 2014). Moreover, dopaminergic neurons were protected and/or α‐synuclein aggregation reduced in 6‐OHDA‐ or paraquat‐induced rodent models of PD using either apocynin or NOX1 KD (Cristovao et al., 2009; Choi et al., 2012; Cristovao et al., 2012). Thus, NOX1 may be an important driver of PD, and inhibition of this isoform offers a novel approach for the treatment of this disease. In addition, NOX1 and NOX4 have been shown to be involved in stroke. NOX4 (but not NOX1) genetic and/or pharmacological inhibition has been shown to be protective in ischaemic stroke (Kleinschnitz et al., 2010). In another study, adeno‐associated virus‐mediated NOX1 KD enhanced functional recovery after stroke in rats (Choi et al., 2015).

Nerves

Hyperalgesia is characterized by increased sensitivity to noxious stimuli. This pain hypersensitivity is caused by tissue damage and inflammation, leading to decreased activation of the TRPV1 ion channels. NOX1 has been detected in neurons, spinal cord and dorsal root ganglion and is responsible for inflammatory pain following different stimuli in vivo. NOX1 KO, but not mice lacking NOX4 (Kallenborn‐Gerhardt et al., 2013), exhibit attenuation of inflammatory pain induced by several stimuli such as formalin, acetic acid and carrageenan (Ibi et al., 2008). ROS generated by NOX1 was shown to regulate TRPV1 channel activity through PKC‐ε activation (Ibi et al., 2008). There is also evidence that NOX1 contributes to analgesia and tolerance induced by morphine (Doyle et al., 2010) and deficiency of NOX1 leads to increased morphine analgesia and suppression of morphine tolerance (Ibi et al., 2011). This occurs via the modulation of GTP and PKC activities, important for the regulation of TRPV1 channel activity (Zachariou et al., 2003; Ibi et al., 2008).

NOX4 expressed in primary afferent neurons contributes to peripheral nerve injury‐induced neuropathic pain (Kallenborn‐Gerhardt et al., 2012). Mice lacking NOX4 demonstrated decreased peripheral demyelinization and an amelioration of structural changes in the injured sciatic nerve. In addition, the persistence of neuropathic pain behavior was decreased in inducible conditional NOX4 KO mice, suggesting that selective inhibition of NOX4 could be a target in neuropathic pain that persists and in other diseases associated with demyelinization, such as Guillain–Barré syndrome, Charcot–Marie Tooth type I disease and MS.

In summary, NOX1 appears to have a critical role in inflammatory and nociceptive pain, whereas NOX4 contributes to peripheral nerve injury‐induced neuropathic pain, suggesting that NOX1 and 4 inhibition could be a novel strategy for the treatment of different types of chronic pain.

Multi‐organ

Cancer

There is considerable literature linking oxidative stress and NOX‐generated ROS to cancer (Blanchetot and Boonstra, 2008; Block and Gorin, 2012; Bonner and Arbiser, 2012; Landry and Cotter, 2014; Guo and Chen, 2015; Roy et al., 2015). NOX have been implicated in genomic instability, angiogenesis, invasion, metastasis, proliferation, cell survival, epithelial–mesenchymal transition (EMT) and activation of stromal fibroblasts, essentially covering most hallmarks of cancer (Hanahan and Weinberg, 2011).

NOX isoforms and regulatory subunits show highly specific expression across panels of human cell lines (Roy et al., 2015). NOX1 is highly expressed in some colorectal cell lines and NOX4 in a wider range of tumour types including ovarian, renal, glioblastoma and melanoma. Even higher expression of the NOX isoforms and regulatory subunits is found in human tumour tissue, underscoring limitations of cell systems for studying the role of NOX in cancer. Increased expression is found in both early and late stages of tumourigenesis suggesting roles for NOX in initiation and maintenance phases (Block and Gorin, 2012).

Inflammatory cytokines and chemokines are potent stimulators of NOX expression. NOX production of ROS in premalignant conditions characterized by chronic inflammatory stress may result in DNA damage and mutations, eventually leading to malignant transformation. ROS generation has been implicated in the progression from IBD to cancer (Coussens and Werb, 2002; Itzkowitz and Yio, 2004), with NOX1 implicated in the progression of IBD to colorectal cancer and gastric tumourigenesis (Tominaga et al., 2007; Oshima et al., 2014). A positive correlation between the level of NOX1 expression and the degree of differentiation was observed in adenocarcinomas (Fukuyama et al., 2005). However, other studies using in situ hybridization and mRNA array analysis observed no difference between normal and tumour colon samples, and an antisense down‐regulation of NOX1 did not decrease proliferation of Caco‐2 colonocytes (Geiszt et al., 2003; Szanto et al., 2005).

NOX4 expression is increased in premalignant fibrotic states, such as NASH and IPF, characterized by activation of the TGF‐β pathway signalling which may lead to the development of carcinomas of the liver and lung (Roy et al., 2015). In contradiction to this, one strain of NOX4 KO mice showed an increased proliferative capacity of hepatocytes with earlier onset of tumour formation and increased tumour size (Crosas‐Molist et al., 2014).

NOX4 localized on nuclear membranes has been implicated in inducing DNA damage in leukaemia cell lines (Stanicka et al., 2015). NOX4 is localized on mitochondrial membranes and has been implicated in mitochondria DNA damage leading to mitochondrial dysfunction (Block and Gorin, 2012), which in turn may initiate a metabolic switch to aerobic glycolysis (Lu et al., 2012). NOX1 has also been implicated in metabolic remodelling in a liver cancer cell line (Bertram et al., 2015). Malignant transformation by oncogenic Ras is known to attenuate mitochondrial function and promote glycolysis and induce senescence in fibroblasts as well as tumour cells, and both NOX1 and NOX4 have been implicated in this process (Adachi et al., 2008; Weyemi et al., 2012; Calvert et al., 2013; Kodama et al., 2013). NOX4 has been implicated in TGF‐β‐mediated EMT (Hiraga et al., 2013; Boudreau et al., 2014).

Both NOX1 (Liu et al., 2015) and NOX4 (Hsieh et al., 2011; Fitzgerald et al., 2012; Mondol et al., 2014) are implicated in tumour invasion and metastasis and also angiogenesis. NOX1 overexpression induced and increased tumourigenicity of fibroblasts or tumour cell lines respectively and increased expression of VEGF, VEGF receptor and tumour vascularity (Arbiser et al., 2002). Mice deficient in NOX1 but not NOX4 demonstrated impaired angiogenesis and reduced tumour vascularization (Garrido‐Urbani et al., 2011). Similar to NOX1 deletion, GKT136901 blocked angiogenesis and was even more effective in reducing tumour vascularization. NOX4 may also play a role in tumour angiogenesis via the Akt pathway, induction of HIF‐2α and increased VEGF expression (Maranchie and Zhan, 2005; Block and Gorin, 2012).

NOX4 KD and GKT137831 treatment attenuated Fms‐like tyrosine kinase 3 internal tandem duplication‐mediated myeloproliferative‐like disease in a mouse model of acute myeloid leukemia via protein tyrosine phosphatase oxidation and inactivation (Jayavelu et al., 2015).

NOX, particularly NOX4, have also been implicated in resistance to target therapies (Liu‐Smith et al., 2014; Sanchez‐Sanchez et al., 2014), anti‐hormonal therapies (Lu et al., 2010) and cytotoxics (Chang et al., 2012) as well as radiotherapy (Hsieh et al., 2012; Li et al., 2014), indicating that NOX inhibitors might enhance the efficacy of current cancer treatments.

In summary, there are many studies supporting a key role for both NOX1 and NOX4 in carcinogenesis and tumour progression, with involvement in most of the hallmarks of cancer. Although there are many studies using specific genetic KD, little has been done using KO mice or specific inhibitors, and this warrants more extensive exploration.

Summary and discussion

As summarized in Table 3, gene deletion and KD studies support a key pathogenic role for NOX1 and NOX4 across a wide range of diseases. In particular, these NOX have been shown to be important drivers of inflammation and fibrogenesis in multiple organs. They do so by amplifying a broad range of signalling pathways involved in inflammatory and fibrotic disorders across multiple organs and diseases. Specifically, NOX1 is stimulated by and is a key enhancer of inflammatory and angiogenic pathways. NOX4 also appears to be a key enhancer of inflammation and drives angiogenesis more upstream from NOX1 via HIF pathway signalling. It also plays a key role in TGF‐β‐mediated processes such as EMT, cell senescence and fibrogenesis. NOX1 and NOX4 may also act through additional fibrogenic pathways which include PDGF and hedgehog, and seem to participate in the co‐amplification of inflammatory/immunological and fibrogenic pathways through their effects on TLR4 signalling and BAMBI expression. It is tempting to speculate that NOX1‐ and NOX4‐mediated activation of innate immunity plays a particularly important role in organs exposed to pathogens, from the environment for skin and lungs, and from the gut for liver.

The role of each NOX may depend on age and also the stage of a disease. Some conflicting data have been generated with different strains of KO mice but may indicate that there is also a protective as well as pathological role for these NOX. It may be that a basal level of NOX is necessary for normal functions in some organs. However, if levels are increased excessively and chronically, then this leads to disease states. Interestingly, pharmacological inhibition of NOX1/4 with GKT137831 decreases but does not completely suppress ROS production in tissues (Gray et al., 2013; Gorin et al., 2015; Zhao et al., 2015) and thereby may maintain a physiological basal level of ROS necessary for normal physiology.

It is also noteworthy that in studies where pharmacological inhibition with GKT136901 or GKT137831 was compared with genetic deletion using the same disease model, the beneficial effects were very similar. Also consistent with the KO mice that do not show any spontaneous pathologies, GKT137831 was extremely well tolerated in all animal studies, in toxicological studies and in clinical trials in more than 170 patients (Wiesel et al., 2012; (http://www.genkyotex.com/Genkyotex/assets/File/PRESS%20RELEASE%20Ph2%2009‐09‐15%20FINAL.pdf). In phase I studies, no dose‐limiting toxicities were observed, and fewer adverse events were observed in the GKT137831‐treated arm than in placebo after 12 weeks of treatment in a phase II trial in patients with diabetic nephropathy. Despite very promising efficacy in several different mouse models of diabetic nephropathy, there was no significant reduction in albuminuria, the primary efficacy end point of the study with GKT137831. The short duration of treatment, the advanced stage of the disease, the type of disease (type 2 and not type 1 as in the majority of animal studies), stabilization of disease by pretreatment with blockers of the renin–angiotensin system (all animal studies were with single‐agent treatment) and/or a role of NOX5 in human kidney disease might account for the lack of effect on albuminuria in this study. However, the observation that GKT137831 induced statistically significant reductions in liver enzymes and inflammatory marker levels in these patients provides encouragement for further trials in other inflammatory and fibrotic disease settings. Pharmacological inhibition of NOX1 and/or NOX4 may provide a safe yet broadly effective therapeutic approach in a range of inflammatory and fibrotic disorders including DKD, NASH, scleroderma and IPF. Over the coming years, NOX inhibitors are poised to become an important therapeutic class for the treatment of a range of diseases with high unmet medical need.

Conflict of interest

The authors declare no conflicts of interest.

Teixeira, G. , Szyndralewiez, C. , Molango, S. , Carnesecchi, S. , Heitz, F. , Wiesel, P. , and Wood, J. M. (2017) Therapeutic potential of NADPH oxidase 1/4 inhibitors. British Journal of Pharmacology, 174: 1647–1669. doi: 10.1111/bph.13532.