British journal of pharmacology 2015 volume 172 part 4

It is also important to consider drug–target relationship, any functional correlates which may be driving the structural changes, the interplay between the cardiac and vascular systems,

M J Cross1, B R Berridge2, P J M Clements3, L Cove-Smith4, T L Force5*,

P Hoffmann6, M Holbrook7, A R Lyon8, H R Mellor9, A A Norris1,

M Pirmohamed10, J D Tugwood4, J E Sidaway11and B K Park1

1MRC Centre for Drug Safety Science, Department of Molecular and Clinical Pharmacology,

University of Liverpool, Liverpool, UK,2Safety Assessment, GlaxoSmithKline, Research Triangle

Park, NC, USA,3David Jack Centre for Research & Development, GlaxoSmithKline, Ware, Herts,

UK,4Clinical & Experimental Pharmacology, Cancer Research UK Manchester Institute,

University of Manchester, Manchester, UK,5Center for Translational Medicine and Cardiology

Division, Temple University School of Medicine, Philadelphia, PA, USA,6Preclinical Safety,

Novartis Pharm Corp, East Hanover, NJ, USA,7Safety Pharmacology, Covance Laboratories, Ltd.,

Harrogate, North Yorkshire, UK,8NIHR Cardiovascular Biomedical Research Unit, Royal

Brompton Hospital and Imperial College, London, UK,9Drug Safety Evaluation, Vertex

Pharmaceuticals (Europe), Ltd., Abingdon, Oxfordshire, UK,10The Wolfson Centre for

Personalised Medicine, Institute of Translational Medicine, University of Liverpool, Liverpool, UK,

and11Innovative Medicines, AstraZeneca R&D, Macclesfield, UK

m.j.cross@liv.ac.uk

-*Current address: VanderbiltUniversity School of Medicine,Cardiology Division, Nashville,

serine/threonine kinases as well as several classes of non-oncology agents A workshop organized by the Medical ResearchCouncil Centre for Drug Safety Science (University of Liverpool) on 5 September 2013 and attended by industry, academiaand regulatory representatives, was designed to gain a better understanding of the gaps in the field of structural

cardiotoxicity that can be addressed through collaborative efforts Specific recommendations from the workshop for futurecollaborative activities included: greater efforts to identify predictive (i) preclinical; and (ii) clinical biomarkers of early

cardiovascular injury; (iii) improved understanding of comparative physiology/pathophysiology and the clinical predictivity of

current preclinical in vivo models; (iv) the identification and use of a set of cardiotoxic reference compounds for comparative

profiling in improved animal and human cellular models; (v) more sharing of data (through publication/consortia

arrangements) on target-related toxicities; (vi) strategies to develop cardio-protective agents; and (vii) closer interactionsbetween preclinical scientists and clinicians to help ensure best translational efforts

Abbreviations

ABPI, Association for British Pharmaceutical Industry; hESC-CM, human embryonic stem cell-derived cardiomyocyte;

HF, heart failure; LVD, left ventricular dysfunction; LVEF, left ventricular ejection fraction; SCD, sudden cardiac death;TKI, tyrosine kinase inhibitor

British Journal of Pharmacology (2015) 172 957–974 957

© 2014 The British Pharmacological Society

Cardiovascular toxicities are observed with therapeutic

agents used in the treatment of both cardiovascular and

non-cardiovascular diseases and affect all components of the CVS

Cardiovascular adverse reactions can occur after acute or

chronic treatment and can affect function (e.g alteration of

the mechanical function of the myocardium) and/or

struc-ture (e.g morphological damage or loss of cellular/subcellular

components of the heart) or vasculature They remain a

major cause of drug attrition during preclinical and clinical

development, and drug withdrawals from the marketplace

This was highlighted in a scientific workshop on

cardiovas-cular toxicity, which covered a wide range of potential

liabili-ties, held at the Medical Research Council (MRC) Centre for

Drug Safety Sciences (CDSS) in January 2010 (Laverty et al.,

2011) A recommendation from the workshop was to

recon-vene to discuss in greater detail a small number of selected

cardiovascular liabilities

On 5 September 2013, a workshop was hosted by the

MRC CDSS (http://www.liv.ac.uk/drug-safety), University of

Liverpool, in conjunction with the Association of the British

Pharmaceutical Industry (ABPI) and the Medicines and

Healthcare Products Regulatory Agency It discussed current

challenges in determining and understanding ‘Structural

Car-diotoxicity of Medicines’ as a major and emerging issue in the

development of new therapies – particularly oncology agents

The key aims of the workshop were to identify those areas of

cardiovascular safety testing where our knowledge and standing should be further strengthened and to recommendareas where collaborative efforts should be focused The work-shop was attended by representatives from pharmaceuticaland biotechnology companies, contract research organiza-tions, regulatory agencies and academia

under-Drug-induced structural cardiac damage is associated withchanges in multiple cardiac cell types leading to cardiac fibro-sis and cardiomyopathy (deterioration of the function of themyocardium due to injury) and subsequently, heart failure(HF) Structural cardiotoxicity is a concern with several classes

of anti-cancer agents as any gain in life expectancy fromtherapeutic intervention might be countered by increasedmorbidity and mortality due to a variety of cardiovascularproblems, including: heart muscle injury with cardiomyopa-thy and HF, complications of coronary artery disease leading

to myocardial ischaemia, arrhythmias, hypertension and

thromboembolism (Stortecky and Suter, 2010; Berardi et al., 2013) Patnaik et al (2011) showed that after 8–9 years fol-

lowing initiation of drug treatment, mortality in breast cancerpatients as a result of cardiovascular toxicity overtakes risk ofdeath from breast cancer recurrence The incidence of struc-tural cardiotoxicity depends on a number of different factorsrelated to therapy (e.g type of drug, dose administered duringeach cycle, cumulative dose, schedule of administration,route of administration, combination of other cardiotoxicdrugs or association with radiotherapy covering the heart inthe therapeutic field) and also to the patient phenotype based

Tables of Links

TARGETS

Other protein targetsa Catalytic receptorsd Enzymese

Angiotensin receptors ErbB2 (HER2) ILK

These Tables list key protein targets and ligands/inhibitors 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 (Pawson et al., 2014) and are

permanently archived in the Concise Guide to PHARMACOLOGY 2013/14 (a,b,c,d,e Alexander et al., 2013a,b,c,d,e).

958 British Journal of Pharmacology (2015) 172 957–974

on pre-existing cardiovascular risk (e.g age, presence of

‘tra-ditional’ cardiovascular risk factors, underlying cardiac

dys-function and any prior exposure to cardiotoxic chemotherapy

or radiotherapy) Cardiotoxic effects can occur immediately

following drug administration, or they may not manifest

themselves until months or, in some examples, many years

after the patient has been treated There are a number of

clinical challenges in managing cancer drug-treated patients

and it is crucial that appropriate risk stratification based on

previous drug exposure and patient phenotype is addressed

Significant challenges to preclinical assessment are also

appar-ent as attempts to model the clinical factors, highlighted

earlier, may have dubious translatable value Although the use

of traditional cancer therapies such as anthracyclines (Von

Hoff et al., 1979) and radiation (Boivin et al., 1992) have long

been associated with cardiac complications (up to a 20% risk

of HF after 20 years following a period of treatment with

chemotherapy and radiotherapy) (Hooning et al., 2007) other

agents such as cyclophosphamide, 5-fluorouracil and

pacli-taxel are known to cause cardiac injury as well, albeit at lower

rates than anthracyclines Dosing regimens and newer agents,

plus pre-screening to exclude patients with reduced cardiac

function at baseline, have helped to reduce risk in current

patients receiving anthracyclines, but in contemporary

studies the rate of left ventricular dysfunction (LVD) is still

between 5–20% (Shakir and Rasul, 2009) depending upon the

definition applied and follow-up duration

The more recently introduced ‘targeted therapies’, which

inhibit various PKs have also been associated with

cardioicity, through both on-target and off-target effects The

tox-icity of these agents is through different molecular and

cellular mechanisms of cardiotoxicity to those caused by

anthracyclines (Force and Kolaja, 2011; Mellor et al., 2011).

However, in many cases, the adverse clinical cardiac events

observed were not anticipated based on preclinical evaluation

of these compounds It is therefore important to identify new

models/techniques, which can better predict adverse clinical

outcomes with these agents

We set out to address the following points at the

workshop:

• What pathologies come under the banner of

‘cardiovascu-lar toxicity’ and how well do we understand these as

indi-vidual pathologies and components of a syndrome?

• What is the prevalence of the recognized individual

pathologies, and how well do we understand their

pathogenesis?

• How good is our mechanistic understanding of cancer

therapeutics-induced structural cardiotoxicity?

• How well do we understand patient susceptibilities to

car-diotoxicity and do we need animal models of ‘disease’

and/or ‘physiological challenge’?

• How translatable are animal pathologies to the relevant

human pathology?

• How can we share data on target-driven toxicities more

efficiently to avoid a repetition of unnecessary animal

research and preclinical toxicology studies?

• Can functional changes predict specific pathologies (and

vice versa)?

• Can we identify/improve in vitro assays to model and

predict specific animal/human pathologies?

This publication incorporates the key issues highlightedduring the workshop along with the gap analysis and identi-fies key areas where a concerted effort could make a realdifference by reducing cardiovascular liabilities of newmedicines

Clinical definitions of cardiovascular toxicity related to oncology therapies

Drug-induced cardiovascular toxicity may develop acutely orsubacutely during or after a treatment period and effectsmay include disorders such as myocardial dysfunction,ischaemia, hypotension, hypertension, QT-interval prolonga-tion, arrhythmias and thromboembolism Chronic conse-quences of cardiomyocyte insult may manifest as an ‘early’cardiomyopathy within the first year after treatment cessa-tion or as a ‘later’ cardiomyopathy, occurring more than 1year afterwards; these probably represent a continuous spec-trum of the same pathophysiology, with dose and coexistingrisk factors determining the rate of progression of cardiacdysfunction Clinical presentation late in the course of the HFprogression represents the most problematic type of injury.The most common initial feature of chronic cardiotoxicity isasymptomatic systolic LVD; left untreated, this may progress

to congestive HF This initial dysfunction may not be cally apparent (i.e asymptomatic) for many years because ofattempted normalization of function by compensatorymechanisms, as seen following other forms of cardiac injurysuch as acute myocardial infarction The incidence of chroniccardiotoxicity is influenced by a number of factors such ascumulative dose of chemotherapy administered, age ofpatient, cardiovascular disease history and prior radiationtherapy, and can range from 5 to 65% of patients treated with

clini-anthracyclines (Dolci et al., 2008).

Anthracyclines produce a dose-related cardiac tion, defined as type I cardiotoxicity (Ewer and Lippman,2005), characterized by cardiomyocyte ultrastructural abnor-malities, (vacuoles, myofibrillar disarray and dropout, necro-sis), contractile abnormalities (dilated cardiomyopathy) and

dysfunc-subsequent clinically evident dysfunction (Billingham et al.,

1978) Some elements are initially reversible, but over timethe burden of fibrosis and myocyte loss to apoptosis rendersthe dysfunction currently irreversible and more refractory tocurrent HF therapy In contrast, cardiac dysfunction not asso-ciated with ultrastructural change, described as type II,typically manifests as an asymptomatic decrease in left ven-tricular ejection fraction (LVEF) (expressed as a percentage ofthe total amount of blood in the left ventricle that is ejected

in each heartbeat, with a range of 55–70% in healthy viduals) and less often by clinical HF (Ewer and Lippman,2005) Agents such as trastuzumab (Herceptin®, Genentech/Roche, San Francisco, CA, USA) and the low molecular weighttyrosine kinase inhibitors (TKIs) for example sunitinib(Sutent®, Pfizer, New York, NY, USA), imatinib (Gleevec®,Novartis, Basel, Switzerland), lapatinib (Tykerb®, GlaxoSmith-Kline, London, UK) and sorafenib (Nexavar®, Bayer, Lev-erkusen, Germany) (see Table 1) are believed to cause type IIcardiac dysfunction (Ewer and Ewer, 2010), although thecellular mechanism may be very drug-specific rather than a

indi-British Journal of Pharmacology (2015) 172 957–974 959

Preclinical cardiac findings

Hypertension Inlyta®FDA Pharm

ReviewInlyta®PrescribingInformationBevacizumab

Avastin®PrescribingInformationCabozantinib

(Cometriq®

)

Ret, Met,VEGFR1/2/3,Kit, trkB,FLT3, Axl,TIE2

Metastaticmedullarythyroid cancer

Cardiac inflammation noted in

a single female dog whenadministered for a 6 monthperiod

Hypertension Cometriq®

PrescribingInformation

Crizotinib

(Xalkori®)

ALK, c-Met(HGFR), andROS

ALK-positiveNSCLC

Dose-dependent inhibition ofthe hERG current, decrease

in HR and increase in leftventricular end-diastolicpressure in dogs,myonecrosis in rats

QT-interval prolongation,bradycardia

Xalkori®FDA PharmReview

Xalkori®PrescribingInformation

Dabrafenib

(Tafinlar®)

B-Raf MM Adverse cardiovascular effects

in dogs consisting ofcoronary arterialdegeneration/necrosis andhaemorrhage, as well ascardiac atrio-ventricularvalve

hypertrophy/haemorrhage

QT-interval prolongation,decreased LVEF

Taflinar®PrescribingInformation

Dasatinib

(Sprycel®)

Bcr-Abl, Srcfamily, Kit,PDGFRβ,EphA2

CML, ALL QT prolongation, increased BP

Vascular and cardiac fibrosis,cardiac hypertrophy,myocardial necrosis,haemorrhage of the valves,ventricle and atrium andcardiac inflammation

QT-interval prolongation,

HF, pericardial andpleural effusion,pulmonary hypertension

Brave et al (2008) Montani et al (2012)

Sprycel®PrescribingInformation

mesylate

(Gleevec®)

Bcr-Abl, PDGFRαandβ, Kit CML, ALL, GIST,MDS/MPD,

ASM, HES,CEL, DFSP

Reversible hypertrophy in rats

Decrease in arterial BP aftersingle i.v dose in rats Noeffect on the rate of beating

or force of contraction inthe isolated atria of guineapigs

Decreased LVEF, LVD, rarefrequency of HF

Kerkelä et al (2006)

Gleevec®FDA PharmReview

Gleevec®PrescribingInformation

Lapatinib

(Tykerb®)

EGFR (ErbB1),HER-2 (ErbB2)

HER-2+ vebreast cancer

Dose-responsive increase in BP

in dog Focal fibrosis andmyocyte degeneration in ratand dog No QT changes inrat and dog

Decreased LVEF, HF,asymptomatic cardiacevents, QT-intervalprolongation

Perez et al (2008)

Tykerb®FDA PharmReview

Tykerb®PrescribingInformationNilotinib

(Tasigna®

)

Bcr-Abl, PDGFRαandβ, Kit CML QT-interval prolongation QT-interval prolongation,sudden death (possibly

ventricularrepolarization related)Ischaemia, peripheralischemia

Kantarjian et al.

(2007)Tefferi (2013)

Weisberg et al.

(2005)Tasigna®

PrescribingInformation

960 British Journal of Pharmacology (2015) 172 957–974

Preclinical cardiac findings

RCC Acute increase in BP after

dosing and decreased heartrate from 75 min to 24.5 hpost-dose in monkeys

Cardiac dysfunction(congestive HF anddecreased LVEF), QTprolongation, 2 cases ofTorsades de Pointes inclinical programme,hypertension

Motzer et al (2013)

Votrient®FDA PharmReview

Votrient®PrescribingInformation

Pertuzumab

(Perjeta®)

HER-2 (ErbB2) Breast cancer None reported Decreased LVEF, HF Perjeta®Prescribing

InformationPonatinib

(Iclusig®)

Bcr-Abl, Bcr-Abl

T315I, VEGFR,PDGFR, FGFR,Eph, Srcfamily kinases,Kit, Ret, TIE2and FLT3

CML, Phchromosome-positive ALL

Inhibition of hERG current indose-dependent manner,systolic heart murmurs andmyocardial necrosis inmonkeys

HF, myocardial ischaemia,peripheral ischaemia(stroke, peripheralvascular disease)

Iclusig®

FDA PharmReview

Iclusig®PrescribingInformation

Regorafenib

(Stivarga®)

VEGFR1/2/3,

BCR-Abl,B-Raf, B-Raf(V600E), Kit,PDGFRα/β,Ret, FGFR1/2,TIE2 andEphA2

CRC Dose-dependent increase in

the finding of thickening ofthe atrio-ventricular valve inrats at 6 months

Hypertension, myocardialischaemia and infarction

Stivarga®PrescribingInformation

Sorafenib

(Nexavar®)

Raf-1 (c-Raf),

b-Raf ,VEGFR1, 2 &

3, PDGFRfamily, Kit

HCC, RCC hERG K-current and Ca-inward

current inhibition No ECG,

BP or heart rate changesobserved in 12 month dogstudy

Autolysis, degeneration andinflammation in 3 month ratstudy An increase in CKlevels with haemorrhageand congestion of the heart

in one animal in 12 monthdog study

QT-interval prolongation,sudden death (possiblyventricular

repolarization related),

HF (cardiomyopathy),coronary vasospasm,arterial thrombosis

Choueiri et al (2010) Escudier et al (2009) Naib et al (2011) Schmidinger et al.

(2008)

Uraizee et al (2011) Veronese et al.

(2006)Nexavar®FDA PharmReview

Nexavar®

PrescribingInformationSunitinib

RCC, GIST Potent hERG channel block

and QT-intervalprolongation and HRreduction at dosesequivalent to human clinicalexposures Multiple ECHOparameter changes inprimate including reductions

in the ratio of right atrial toaortic diameter, LVEF timeand LV area

Histopathological findingsincluded capillaryproliferation, myocardialvacuolization and pericardialinflammation

QT-interval prolongation,decreased LVEF, LVD,

HF, increased BP, CHFlinked to cardiovascularco-morbidities, arterialthrombosis

Bello et al (2009) Choueiri et al (2010) Chu et al (2007) Faivre et al (2006) Telli et al (2008)

Sutent®FDA PharmReview

Sutent®PrescribingInformation

Trametinib

(Mekinist®)

MEK1, MEK2,

MEK1 kinase,MEK2 kinase

MM Inhibition of hERG channel,

cardiomyopathy (decreasedLVEF, increased heartweight) in mice

Cardiomyopathy (cardiacfailure, LVD, ordecreased LVEF)

Mekinist®FDA PharmReview

Mekinist®PrescribingInformation

British Journal of Pharmacology (2015) 172 957–974 961

class effect, reflecting both on target and off-target toxicity.

Typical features in this setting include lack of an obvious

dose–relationship, increase in toxicity when given

concur-rently with anthracyclines, some reversibility after stopping

treatment and restoration of normal cardiac function with

appropriate medical management (Perik et al., 2007; Slamon

et al., 2011) Many of these observations derived from cardiac

safety analyses in oncology trials where patients were

typi-cally younger and screened to exclude those with

pre-existing cardiovascular disease, thereby preselected as more

resistant to cardiotoxicity from that class of drugs

Cardiotox-icity rates tend to be higher in clinical practice compared

with those reported in oncology trials (Farolfi et al., 2013),

reversibility is less common, and duration of treatment, and

therefore dose, may contribute to a cumulative risk

Early detection of subclinical cardiac dysfunction could

lead to the identification, drug-intervention (e.g ACE

inhibi-tor and β-blocker) and prevention of late adverse cardiac

events and this is ultimately the goal for both cardiologists

and oncologists Collection of endomyocardial biopsies to

identify histopathological evidence of myofibrillar loss is an

inaccurate and impractical form of monitoring heart damage

The use of LVEF as the only parameter of cardiac function is

increasingly viewed, by the cardiology community, as an

inadequate measure to predict and monitor cardiac damage

Nevertheless, LVEF is measured routinely as 2-D

echocardiog-raphy (Echo) is the methodology of choice for frequent

moni-toring and is cost-effective and has fairly widespread

availability, despite its known higher method variability

Other screening modalities, such as cardiac MRI, are gaining

in popularity, because of low inter-scan variability and ability

to offer virtual histology, which is capable of detecting signs

of fibrosis when combined with the contrast agent

gado-linium (Tandri et al., 2005).

Although there is no clear definition of cardiotoxicity, apractical and easily applicable definition was created by apanel of investigators involved in the clinical development

of trastuzumab (Seidman et al., 2002), which considered

chemotherapy-induced cardiotoxicity as either a 5% declinefrom baseline LVEF to less than 55% overall with accompa-nying signs or symptoms of HF, or asymptomatic decrease inLVEF in the range of equal to or greater than 10% to less than55% Other trials have used 50% as the threshold to definecardiotoxicity, but given the potential variability of Echo-based LVEF measurements as discussed earlier, in reality, thisdepends upon the low limit of normal (LLN) for a healthypopulation for individual centres, and therefore current guid-ance is to determine LLN and interpret the guidance usinglocal cut-off values The extent to which this is practised inthe real world has yet to be clarified

Medical management of anthracycline-induced HF isbased largely on the use of agents used to treat HF, includingACE inhibitors,β-blockers and aldosterone antagonists, withloop diuretics for decreasing fluid retention In a recent study

of patients with an anthracycline-induced decrease in LVEF≤45%, treatment with enalapril and carvedilol resulted in nor-

malization of LVEF in 42% of patients (Cardinale et al., 2010).

These responders had a higher LVEF after HF treatment pared with partial responders (whose LVEF increased>10%,but did not normalize) and nonresponders who were mostresistant to HF treatment and failed to improve ventricularfunction A striking observation was that the patients in theresponder group all had their ‘rescue’ HF therapy initiatedwithin 4 months of the chemotherapy, whereas if it was

Preclinical cardiac findings

Decreased LVEF, HF,increased risk if prior orconcurrent

anthracycline treatment

Seidman et al.

(2002)Herceptin®FDAPharm ReviewHerceptin®PrescribingInformationVandetanib

(Caprelsa®)

EGFRs, VEGFRs,Ret, Brk, TIE2,Eph receptorsand Src

Medullarythyroid cancer

Inhibition of hERG channel,increase in BP inrats,increased QTc and BP

in dogs

QT-interval prolongation,Torsades de Pointes,acute cardiac failure,hypertension,

Scheffel et al (2013)

Caprelsa®PrescribingInformationVemurafenib

(Zelboraf®)

a/b/c-Raf andb-Raf

MM Inhibition of hERG channel,

increase in incidence of AVblock in dogs, increasedheart weight in rats

QT-interval prolongation Zelboraf®FDA Pharm

ReviewZelboraf®PrescribingInformationZiv-aflibercept

(Zaltrap®)

InformationALL, acute lymphocytic leukaemia; AP, action potential; ASM, aggressive systemic mastocytosis; AV, atrioventricular; CEL, chronic eosinophilicleukaemia; CK, creatinine kinase; CRC, colorectal cancer; DFSP, dermatofibrosarcoma protuberans; GIST, gastrointestinal stromal tumour;HCC, hepatocellular carcinoma; HES, hypereosinophilic syndrome; HR, heart rate; MDS, myelodysplastic syndrome; MM, metastaticmelanoma; MPD, myeloproliferative disorder; NSCLC, non-small cell lung cancer; RCC, renal cell carcinoma

962 British Journal of Pharmacology (2015) 172 957–974

initiated beyond 4 months then response was considerably

less This is particularly important in light of recent data that

indicates that only 31% of patients receiving chemotherapy

with an asymptomatic decrease in LVEF receive an ACE

inhibitor or angiotensin receptor blocker, 35% receive a

β-blocker and 42% are referred for cardiology consultation

(Yoon et al., 2010) This emphasizes the importance of

appro-priate communication between the oncologist and

cardiolo-gist and highlights that early detection and treatment of

cardiac injury is critical to a successful outcome

Drug-induced myocardial injury:

pathogenesis and manifestations

For preclinical structural cardiotoxicity, in the absence of a

clear target-driven liability, the underlying molecular

mecha-nism(s) are rarely known It is crucial therefore to build an

understanding of the pathogenesis of the toxicity, the

moni-torability and safety margin based on efficacious exposures,

all to inform assessment of the potential risk to man

Under-standing the temporal progression of the lesion provides

valuable information in understanding the pathogenesis It is

also important to consider drug–target relationship, any

functional correlates which may be driving the structural

changes, the interplay between the cardiac and vascular

systems, translational relevance to patients and to recognize

that results of a repeat-dose general toxicity study (mainly

macroscopic and histological data at the end of the treatment

period) provide only a static picture of a process that may be

temporally dynamic The best understanding therefore comes

from integrating all relevant pieces of data together to reveal

a wider picture Ultimately, a better understanding of the

pathogenesis may help with the development of a risk

miti-gation strategy to include a biomarker component for clinical

use Although an understanding of pathogenesis can help inthe management of liabilities, the identification of defined(and ideally common) molecular mechanisms of structuralcardiotoxicity are required to aid in better drug design.Cardiac cell injury is a continuum (Figure 1) as in manyother organs and normally progresses from degeneration,necrosis, responding inflammatory changes and eventuallyfibrosis, which can be considered the repair process although

it does not result in functional contractile tissue The ultimateimpact of cellular injury on myocardial contractile function

is highly dependent on the number and distribution of cellsinvolved

A non-lethal cell injury, generally considered tion’, can be characterized by cytoplasmic vacuolation ofcardiac myocytes and may be caused by lipid accumulation,mitochondrial swelling or dilation of sarcoplasmic reticulum.Although a non-lethal injury suggests there is an opportunityfor some reversibility of the condition, this can only beviewed in the context of the tissue; the low inherent regen-erative capability of the heart suggests that any sort of adap-tive response mounted to the injury may become a source ofsubclinical or occult change in cardiovascular function Thiscondition may predispose to an impaired ability over time tocope with stressors such as hypertension or treatment withcardiotoxic agents leading potentially to the development ofHF

‘degenera-A lethal injury to the myocardium results in myocellularnecrosis characterized by a terminal irreversible stage of cellinjury (cardiomyocyte death), loss of membrane integrity andrelease of cytosolic proteins (potential biomarkers of cardiacinjury such as troponin), in which the adverse morphologicalchange is temporally progressive and includes an inflamma-tory cell infiltrate and regions of myocardium replaced byfibrosis Extensive fibrosis can affect myocardial complianceand contractility and play a direct role in the development

of chronic progressive cardiomyopathies It is important

Figure 1

Schematic representation of the morphological continuum of myocardial injury and repair

British Journal of Pharmacology (2015) 172 957–974 963

however to recognize the difference between extensive

regional myocardial necrosis (i.e infarct) and the multifocal

lesions often seen in response to drug-induced injury

To demonstrate the usefulness of various modalities for

characterizing the pathogenesis of cardiovascular injury,

Casartelli et al (2011) reported an integrated and

longitudi-nal study to investigate the onset, progression, and

revers-ibility of an off-target cardiac lesion caused by a neurokinin

(NK)-1 receptor antagonist (casopitant) in dogs, after

long-term (6 months) administration with inlong-termittent data

col-lection Transmission electron microscopy examination

revealed changes in cardiac cells with multi-lamellar bodies

in sarcoplasm (associated with a progressive impairment and

perturbation of cardiomyocytes) after only 6 weeks of

treat-ment and this coincided with an initial rise in cardiac

tro-ponin I (cTnI) After 20–26 weeks, some necrotic myofibres,

filled with multi-lamellar bodies, were also observed at a

time when light microscopy observations were first made

The most informative picture of cardiac changes was

obtained by integrating cTnI alterations (as a biomarker of

cardiac damage) with EM findings as these changes preceded

evidence of injury at the light microscopic level Elevations

were also seen in N-terminal pro-brain natriuretic peptide

(NT-pro BNP), a biomarker for the onset and evolution of

cardiac hypertrophy, which started to increase after 2 weeks

of treatment, preceding most, if not all, the anatomical and

functional (ECG) changes Thus, the integration of different

investigative tools (in addition to the standard regulatory

requirement of histopathology) provided early evidence of

cardiac cell injury and a means of accurately characterizing

the onset and progression of the lesion with a clear

trans-latability to a clinical setting (i.e cTnI and NT-pro-BNP

increases) Early recognition of important liabilities

facili-tates early decision-making around progression or strategies

for mitigating risk in patients It is however unfeasible to

perform EM routinely as part of a high-volume toxicity

screening process and measurement of circulating

predict-able and specific biomarkers remains the goal

Preclinical cardiotoxicity (by light microscopy) was not

apparent during the development studies with the Abelson

murine leukemia viral oncogene homologue 1 (c-Ab1), PDGF

and stem cell factor receptor (CD117) (c-Kit) inhibitor,

imatinib although clinical findings suggestive of decreased

LVEF and HF in patients without previous heart disease were

reported after launch (Kerkelä et al., 2006) Taking a more

translational and innovative approach, Kerkelä et al (2006)

demonstrated, using transmission EM, that mitochondrial

abnormalities and accumulation of membrane whorls in

both vacuoles and the sarco-(endo-)plasmic reticulum of

human and mouse cardiomyocytes, in vitro, were suggestive

of the clinical presentation of toxic myopathy Similarly, the

cardiac dysfunction (LVD, LVEF and HF) seen in patients with

the multi-targeted receptor tyrosine kinase inhibitor (TKI)

sunitinib (Chu et al., 2007), was subsequently ascribed, using

transmission EM, to depletion of coronary microvascular

pericytes resulting in changes such as increased endothelial

permeability in the coronary microvasculature Pericyte loss

is not a feature of cardiotoxicity reported by other agents in

the TKI class and this observation indicates that injury to

non-contractile elements can still progress to

cardiomyopa-thy These findings also highlight the need for in vitro screens,

which reconstitute different cellular components to aid inspecific liability identification (see section ‘Novel human cell-based models to predict cardiac microvascular toxicity’).Although there is currently significant cancer biomarkertrial activity, most is related to prognostics and pharmaco-dynamics with relatively few studies in the area of safetybiomarkers To date, little has been done to assist generalpractitioners in identifying cardiovascular adverse effects ofcancer treatments, although there is a general awarenessregarding growing late toxicity from cancer treatment aspatient survival increases Investigators at the CancerResearch UK Manchester Institute (University of Manches-ter), in collaboration with scientists at AstraZeneca (AlderleyPark, UK), have initiated a study to identify predictive bio-markers of safety in rodents in one of the few studies tocapture functional change, biomarkers and histology in alongitudinal fashion Rats showed a significant reduction inLVEF after 43 days during an 8 week continual dosing studywith doxorubicin (iv), which continued after cessation ofdosing with no evidence of reversibility although the point

at which irreversibility occurred was not determined

(Cove-Smith et al., 2014) Overt histological changes were

observed after 29 days dosing although EM changes chondrial damage and myocyte ultrastructural changes)occurred after a single dose to rats Functional decline(decrease of LVEF and diastolic dysfunction measured byE/A ratio) preceded the rise in cTnI and histological damage(light microscopy) However, despite the incrementaldecline in systolic function, the LVEF remained above thenormal clinical threshold of 55% until the end of study It

(mito-is env(mito-isaged that a future panel of biomarkers will helpdetermine when cardiac damage presents initially andresolves

Mechanisms of structural cardiotoxicity

Understanding the mechanisms behind the thies that arise as a result of targeted cancer therapies anddeveloping strategies to treat these complications are impor-tant for the cardiovascular care of the cancer patient as well as

cardiomyopa-to enable future development of non-cardiocardiomyopa-toxic drugs thermore, an understanding of these cardiomyopathies mayalso have implications for more common types of HF andmay provide unexpected insights into the biology of theheart After many years of little advance in the understanding

Fur-of the mechanism Fur-of toxicity Fur-of anthracyclines, recent novelfindings point to a role for topoisomerase II β in inducingDNA damage in cardiomyocytes, mitochondrial dysfunctionand generation of reactive oxygen species leading to cardio-

toxicity (Zhang et al., 2012).

There are currently over 100 TKIs in discovery or

devel-opment (Broekman et al., 2011) Approximately 100 genes

have been implicated in driving cancers with ∼50 beingpotential anti-cancer targets and a proportion are also likely

to play an important role in cardiomyocyte homeostasis (seeTable 2) Off-target toxicity is also a major issue as ATP com-petitive inhibitors demonstrate significant kinase promiscu-ity leading to undesirable off-target effects This lack of

964 British Journal of Pharmacology (2015) 172 957–974

Table 2

Kinase/phosphatase conditional knockout mouse models associated with cardiovascular functional effects (adapted from Mellor et al., 2011)

Protein Signalling role

Knockout animal model Effect on cardiac function Reference

PTEN Lipid phosphatase

Negative regulator ofPI3-kinase signalling

Muscle-specific PTENknockout mouse

Basal hypertrophyMild reduction in contractilityReduced hypertrophy inresponse to pressure overloadcompared with wt

Crackower et al (2002) Oudit et al (2008)

AMPK Serine/threonine kinase

Activated by increase inAMP: ATP

Acts to preserve/generateATP

Heterozygous AMPKα2knockout mouse

Mild reduction in contractilityWorsened hypertrophy inresponse to pressure overloadcompared with wt

Severe dilated cardiomyopathy

HF and premature death

Kontaridis et al (2008) Princen et al (2009)

ERB2 Receptor tyrosine kinase

Co-receptor inneuregulin/EGRFsignalling

Ventricularmyocyte-specific ERB2knockout mouse

Severe dilated cardiomyopathyDecreased contractility

HF and sudden death

Crone et al (2002) Ozcelik et al (2002)

PDK1 AGC serine/threonine

kinaseActivates AKT and p70S6K

Muscle-specific PDK1knockout mouseTamoxifen-inducibleheart-specific PDK1knockout mouse

Apoptotic death ofcardiomyocytesImpaired LV contractilitySevere and lethal HF

Ito et al (2009) Mora et al (2003)

Pim1 Serine/threonine kinase

Acts downstream of AKT

to block apoptosisInduction and stabilization

of c-myc

Cardiac-specific Pim-1dominant-negative inmouse

Progressive dilationReduced contractilityIncreased LVEDPDecreased LVDPAlterations in Ca2 +handling

Muraski et al (2008)

Raf-1 (c-Raf) Serine/threonine kinase

Involved in the ERKsignalling pathway

Cardiac-specificRaf-1 knockout mouse

Reduced contractilityIncreased heart sizeDecreased posterior wallthickness

Yamaguchi et al (2004)

ILK Serine/threonine kinase

Phosphorylates Akt andGSK-3β

Muscle-specific ILKknockout mouse

Increased heart sizeDilated cardiomypathyCardiac fibrosisSudden death

White et al (2006)

AK1 Kinase/phosphotransferase

Adenine nucleotidehomeostasis

AK1 knockout mouse Reduced contractility – coronary

flow relationshipRecovery of flow after I/R wascompromised

Cardiac hypertrophyreduced fractional shortening

LV and septal wall thinningLethal cardiomyopathy

Braz et al (2003)

ERK5 MAPK (serine/threonine

kinase) phosphorylatesMEF2C, Sap1a, p90RSK

ERK5 knockout mouseERK5−/− cardiomyocyteknockout

Embryonically fatal at E9.5–10.5Defective cardiac development,heart looping, angiogenesisand vascular maturationMice develop normally but havereduced cardiac hypertrophicremodelling

Regan et al (2002) Kimura et al (2010)

AMPK, AMP-activated protein kinase; ATF-2, activating transcription factor 2; GSK-3β, glycogen synthase kinase 3 β; ILK, integrin-linkedkinase; I/R, ischaemia/reperfusion; LVDP, left ventricular diastolic pressure; LVEDP, left ventricular end-diastolic pressure; MEF2, myocyteenhancer factor 2; wt, wild type; PDK1, 3-phosphoinositide-dependent PK-1; PTEN, phosphatase and tensin homolog; Shp2, src homology

2 region

British Journal of Pharmacology (2015) 172 957–974 965

selectivity is not limited to kinases, but includes non-kinase

targets, which also bind ATP This general problem and

appar-ent ‘class effect’ has been observed for a number of approved

kinase inhibitors, as summarized in Table 1 See Force and

Kolaja (2011) for a comprehensive review of kinase cardiac

biology and potential mechanistic links to cardiotoxicity

Mechanisms of functional and/or structural

cardiotoxic-ity may fall into several major categories:

(a) Electrophysiological perturbations, mediated via ion

channel inhibition (Na, K or Ca channel interactions;

e.g hERG/KCNH2) and represents a significant cause of

early compound attrition as a result of the

implementa-tion of early screening strategies)

(b) Cytotoxicity [molecular inactivation of cell processes,

altered energetics, oxidative stress/free radical

genera-tion, may be primary (target) or secondary (off-target)

pharmacology related]

(c) Primary pharmacology (an undesirable target-mediated

activity and a common preclinical and clinical

mecha-nism, e.g hyper-pharmacology of cardio- and vasoactive

drugs)

The workshop focused largely on cytotoxic and

pharma-cological mechanisms of drug-induced on- and off-target

car-diotoxicity as primary mediators of structural injuries In the

pharmaceutical industry, to identify potential safety

liabili-ties early in drug development, initiation of new discovery

programmes includes a review of the published target biology

information This enables identification of potential

toxico-logical issues because of primary pharmacology, planning of

hypothesis-based experiments to confirm or refute potential

issues and generation of toxicological data to support or

reject target validity Secondary pharmacology screening for

compound interactions with key cardiovascular homeostatic

proteins and receptors is also becoming increasingly

impor-tant for identifying off-target liabilities associated with a

par-ticular compound or series

Case examples of pharmacological

mechanisms (on-target and off-target)

of structural cardiotoxicity

On-target

A number of companies have pursued activin receptor-like

kinase 5 (ALK5 also known as TGFBR1) as an oncology and

fibrosis target It has previously been shown that the TGF-β

superfamily signalling pathways play a key role in cardiac

development, and that ablation of ALK5 in the endocardium

of mice results in defects in epithelial-mesenchymal

transfor-mation and an early stage of cardiac valve fortransfor-mation

(Sridurongrit et al., 2008) Nevertheless, a role for ALK5 in the

developed heart was poorly understood, but a potential role

in cardiac valve homeostasis represented a safety concern As

a result, safety scientists at AstraZeneca (Anderton et al.,

2011) undertook an acute (5–8 days) study in rodents using

selective inhibitors of ALK5 to assess this potential target

liability Importantly, early ALK5 inhibitors were available

from different chemical scaffolds, enabling clear separation oftarget/off-target effects A comprehensive evaluation of theheart was performed to assess all four heart valves in eachanimal Histopathological heart valve lesions were observed

in all animals, in all heart valves and from two distinctchemical series Valves were distorted with severe haemor-rhage, fibrin deposition and neutrophil infiltration andvalvular interstitial cells were enlarged with increased cyto-plasm Immunohistochemistry analysis revealed the heartvalve, but not the myocardium was positive for ALK5 expres-sion The compounds were inactive against 5-HT receptors,previously implicated in drug-induced valvulopathy As aresult of these findings, the project was terminated in thediscovery phase because of unacceptable target-related toxic-ity No safety margin was expected, the lesion was considered

to be un-monitorable and there was no defined hypothesis tosupport humans being different from rodents in respect ofthe ALK5 inhibitor-mediated pathology Anecdotally, thisexperience was shared by at least two other pharmaceuticalcompanies, but not published The publication by Anderton

et al (2011) alerted other organizations working in this area

or considering initiating discovery efforts on this target to thesafety implications This represents an excellent example ofthe benefit of sharing adverse target safety information inorder to reduce further animal experimentation, resource andindustry attrition, and is a position encouraged greatly by all

of the workshop representatives

Off-target

A safety concern relating to compound promiscuity is thatoff-target pharmacological activity unrelated to the primarydrug action might be associated with adverse cardiac effects

An example is the c-Met inhibitor (PF-04254644), whichleads to myocardial pathological changes in rats within 6 hafter a single dose, resulting in replacement fibrosis after 7

days (Hu et al., 2012) Myocardial EM changes (necrosis of

myofibres, intra-mitochondrial densities and lipid tion) were detected at very early time points post-dosing(within 2 h) These time points were coincident with peakelevations of serum troponin and an associated functionalincrease in heart rate and BP within 2–7 h post-dose As otherc-Met inhibitors currently in clinical use are not associatedwith adverse cardiac effects, it was clear this represented anoff-target effect of the compound Wide ligand binding pro-filing revealed that PF-04254644 is a potent inhibitor of PDE3

deposi-and also 2, 5, 10 deposi-and 11 (Aguirre et al., 2010) It is well

recognized that inhibition of multiple PDEs leads to increasedheart rate, contractility and sheer stress force and may result

in secondary myocardial degeneration (Larson et al., 1996).

These observations enabled the identification of alternativec-Met inhibitors without off-target PDE activity and the asso-ciated cardiovascular liability

New approaches to the mechanistic understanding of cardio-protection

Dexrazoxane is the only clinically approved cardioprotectiveagent used to reduce the cardiotoxicity associated with

anthracyclines such doxorubicin (Lipshultz et al., 2004) The

966 British Journal of Pharmacology (2015) 172 957–974

cardioprotective effect was initially thought to be due to its

ability to chelate iron and reduce the number of metal ions

complexed with anthracycline, leading to decreased

forma-tion of superoxide radicals (Sterba et al., 2013) However,

more recent data suggest that dexrazoxane antagonizes

doxorubicin-induced DNA damage through interference with

topoisomerase IIβ (Lyu et al., 2007; Ky et al., 2013) Despite its

cardio-protective effect, there is persistent concern that

dexra-zoxane may cause myelosuppression, reduce the anti-cancer

effectiveness of anthracyclines and promote secondary

malig-nancies in patients Both the latter concerns were addressed

and refuted by randomized clinical trials and their

meta-analyses (Jones, 2008) A variety of different iron-chelators

and antioxidants, such as vitamin E and N-acetylcysteine,

have been studied in animal models and clinical trials

However, these agents failed to provide cardioprotection

against anthracycline chemotherapy (Sterba et al., 2013).

Cardioprotective/pro-survival mechanisms exist in the

heart [e.g ErbB2 (Her2)-induced pro-survival signalling]

(Force and Wang, 2013) and the blockade of this receptor on

cardiomyocytes by trastuzumab (Herceptin) may explain in

part the cardio-toxicity of this agent Neuregulin-1β is an

agonist at the ErbB2 receptor and has been shown to offer

cardio-protective effects (Fukazawa et al., 2003) as a result of

activation of cell survival pathways in cardiomyocytes

However, there would clearly be issues in successful

manage-ment of the concomitant administration of a neuregulin-1β

receptor agonist and trastuzumab in order to achieve

maximum therapeutic value with minimal adverse effects

This paradox has recently been addressed by using a modified

bivalent neuregulin-1β ligand, which promotes

cardioprotec-tion, via ErbB4 homodimers, but minimizes proneoplastic

potential in cancer cells (Jay et al., 2013) Recent preclinical

studies have also determined a cardio-protective role of Cdk

4/6 inhibitors in anthracycline-induced cardiotoxicity

sug-gesting a new potential treatment option (McClendon et al.,

2012) The area of cardio-protection is currently under-served

and there is an urgent need for better protection strategies in

order to allow maximally effective therapies to be

adminis-tered with minimized risk of cardiotoxicity

Predicting structural effects from

functional changes

A UK pharmaceutical consortium in collaboration with ABPI

carried out an analysis of single-dose Good Laboratory

Prac-tice (GLP) safety pharmacology (cardiovascular function;

heart rate, BP and QT-interval) data from telemetered dogs

and 28 day repeat-dose toxicology (morphology) data from

rodents in an attempt to understand if there are any trends in

the acute study that predict the longer-term outcome

(Milliken et al., 2012) Relationships between these data sets

has been largely undetermined and there is increasing

inter-est in utilizing data such as this for predictive associations, for

earlier detection of compounds with risk for cardiovascular

pathologies One hundred thirty-five compounds (all low

molecular weight) were included in the data set and when

normalized for Cmax exposure concordance the number of

compounds for inclusion in the data set was 126

Cardiovas-cular histopathology changes were seen in 15 compounds ofwhich the highest incidence of pathology (eight compounds)occurred in the myocardium and presented with degenerativeand/or inflammatory events Six of the eight compoundswith pathologies showed a peak increase>40 bpm and dura-tion ≥5 h There were no instances of vessel pathologywithout a change in systemic BP and if compounds showed

no change in HR, there was a high degree of assurance thatthere would be no morphology findings (at least up to thetime of investigational new drug enabling studies in the samespecies) This initial analysis reinforces a growing generalimpression that cardiovascular injury has a haemodynamicdriver element and suggests that functional changes observed

in acute single-dose studies could prompt moving (sub)chronic toxicology studies forward, to identify cardiovascularliabilities earlier in development These findings suggest itmay be worth including more end points in future toxicitystudies (e.g telemetered animals and/or pathology readouts

in safety pharmacology studies), although issues in ability need addressing given that an ABPI-sponsored Animal

translat-Model Framework (Valentin et al., 2009) analysis showed a

weak relationship between dogs and humans in terms ofhaemodynamic changes In general, dogs are hypersensitivefor haemodynamic changes and rodents appear hypersensi-tive for vascular injury-raising issues of comparativephysiology/pathophysiology In terms of future studies, itwould be valuable to assess the panel of compounds used in

this analysis in relevant in vitro pharmacodynamic screens

(e.g rodent, dog, human cardiomyocytes) to understandbetter the role of functional changes (cause or effect) in struc-tural cardiovascular toxicity

Better predictability in man

Although LVEF is an important prognostic factor in dilatedcardiomyopathy, effective risk stratification remains chal-lenging, particularly with respect to sudden cardiac death(SCD) as most patients who experience SCD do not haveseverely reduced LVEF, and many patients with significantimpairment of LVEF may still be at low risk for SCD (Dagresand Hindricks, 2013) Identification of better independentprognostic factors is necessary to enable clinicians to moreaccurately stratify risk in patients with dilated cardio-myopathy and to implement early (preventative) medicalmanagement Measurement of cardio-specific biomarkersrepresents a valid diagnostic approach for early identi-fication, assessment and monitoring of cardiotoxicity

Cardinale et al (2004) showed that early detectable levels of

circulating cTnI after high-dose anthracycline therapy predicted both occurrence and severity of LVEFimpairment A higher rate of major cardiac events (andmorbidity) within the first year of follow-up was also noted

chemo-in patients who demonstrated elevated levels of cTnI formore than a month after the last chemotherapy adminis-tration Equally important, they reported a high negativepredictive value for cTnI (99%), which identified low-riskpatients who would most likely not encounter subsequent

cardiac complications Cardinale et al (2010) also reported

an elevation in cTnI in 72% of patients who subsequentlydeveloped trastuzumab-induced cardiotoxicity compared

British Journal of Pharmacology (2015) 172 957–974 967

with only 7% of patients who showed normal cTnI A

number of studies suggest that cTnI determination is able to

predict the occurrence of clinically significant LVD, at least

3 months in advance (Cardinale et al., 2010; Sawaya et al.,

2012)

An ongoing biomarker study at the Cancer Research UK

Manchester Institute (University of Manchester) in a cohort

of 30 patients with lymphoma and breast cancer, treated with

anthrayclines, is measuring both circulating (cTnI, ILs, TNF,

fatty-acid-binding protein, myeloperoxidase, MMPs,

NT-pro-BNP) and imaging (cardiac MRI sequences including

volu-metric analysis, T1 mapping, T2 mapping, late gadolinium

enhancement and myocardial strain) biomarkers of

cardio-toxicity in an attempt to correlate biomarkers with clinical

data Although the data are immature, it has shown that LVEF

steadily decreases over the time course; however, no

circulat-ing biomarker data are available yet (Cove-Smith et al., 2014).

In summary, strategies are urgently required to detect

cardiotoxicity in patients by focusing on pathways common

to all or many cardiotoxicities This in turn would be

expected to lead to the identification of novel biomarkers,

allowing earlier detection and potential intervention Novel

potential robust approaches to address this issue include

measurement of circulating miRNAs (De Rosa et al., 2014)

and the use of metabolomics to identify metabolic signatures

in the circulation which are indicative of injury These may

help to detect very early changes of cardiotoxicity (Roberts

et al., 2012).

New clinical directions

A cardio-oncology day-case assessment service at the Royal

Brompton Hospital in London has been established recently,

acting as a ‘one-stop shop’, in which serum biomarkers,

resting and stress echocardiography, cardiac MRI, research

(e.g genomics), and a clinical review are carried out at

base-line and during the treatment phase in patients at risk or who

have potential cardiotoxicity detected on current surveillance

strategies Given the limitations in the sensitivity of

bio-marker (particularly Echo) strategies without appropriate

baseline measurements, this approach offers a means of

monitoring any changes as they occur and with a higher

degree of certainty A key element of this approach is a close

working relationship between cardiology and oncology

col-leagues, to balance the risks and benefits of the cancer

treat-ment with any cardiotoxicity detected It is critical to

remember that this increasing problem of cardiotoxicity has

arisen out of the remarkable success of modern cancer

treat-ment, and it would be paradoxical and clinically

inappropri-ate to withhold potentially life-saving cancer treatments

because of the detection of subclinical cardiotoxicity using

more and more sensitive biomarkers and imaging

technolo-gies Instead the philosophy of early detection is to stratify

cardioprotection to those in need, and equally to reassure

those without evidence of cardiac injury, given the negative

predictive value is the most powerful This initiative has also

led to the development of a UK-wide consortium of seven

cardiac and eight oncology centres that offers coordinated

assessment of patients and follows efforts in other European

countries and North America to better coordinate oncology care

cardio-Preclinical considerations – bridging back to man

A key challenge for the pharmaceutical industry is to not onlydetect (or ideally predict) preclinical drug-induced changes incardiac structure or function, but also to understand therelevance of these cardiotoxic effects to humans and, morespecifically, the potential impact within the clinical settingfor which the drug is intended In order to avoid cardiotox-icity emerging at the later stages in the drug developmentpathway, it is vital that the potential for structural cardiovas-cular toxicity, associated with either a target or a molecule,can be identified as early as possible Failure to do so results indrug attrition, limits the availability of new treatmentoptions for patients and is extremely costly financially Thecardiac electrophysiological effects of a molecule can be

detected routinely through a combination of in silico, in vitro

approaches (utilizing cell lines overexpressing certain ion

channels) and supported via short-duration in vivo studies.

However, the assessment of structural cardiovascular toxicity

currently involves longer-term repeat-dose in vivo toxicity

models (which are cost and labour intensive) primarilyfocussed on histopathological end points In most chronictoxicology studies (e.g GLP regulatory studies), left ventricu-lar function, electrocardiogram and biomarkers of cardiac

injury are not routinely measured (Mellor et al., 2011) It is

clear that the area of structural cardiotoxicity could alsobenefit enormously from a more predictive and integrated

tiered approach (e.g in silico/in vitro/in vivo).

Novel in vivo models of

structural cardiotoxicity

One in vivo approach to providing better strategies for

pre-dicting structural cardiotoxicity involves the measurement

of cardiomyocyte loss after drug treatment in zebrafish

(Cheng et al., 2011) The most common gross morphological

defect induced by cardiotoxic drugs was an associated ing of the heart chambers If the ventricle stops beating,blood often pools and clots in the chamber, but this does notimmediately kill the fish as there is sufficient oxygenationacquired through passive diffusion Rates of pericardialoedema are noted, as well as the presence of thrombi andtheir location Most thrombi occur at the yolk sac just poste-rior to the entrance to the atrium Drugs known to haveadverse cardiotoxic effects in humans also demonstrate car-diotoxicity in zebrafish Because zebrafish can survive in theabsence of cardiac output and in the presence of major vas-cular defects for several days, abnormalities can be studiedthat would be rapidly fatal in mammals In this model,sorafenib was shown to reduce the numbers of cardiomyo-

swell-cytes (Cheng et al., 2011) and both sorafenib and sunitinib

caused a marked reduction in total myocyte number perheart, contractile dysfunction and ventricular dilatation

968 British Journal of Pharmacology (2015) 172 957–974

Novel human cell-based models to

predict structural cardiotoxicity

Currently, in vitro screening strategies are predominately

focused on identifying functional cardiotoxicity through the

detection of ECG abnormalities and QT-interval prolongation

by ion channel screening and measurement of cardiac action

potentials and changes in Echo through cardiomyocyte

con-tractility abnormalities (Pollard et al., 2010; Harmer et al.,

2012) Predictive in vitro assays to detect structural cardiac

toxicity in man have been lacking This has mainly been due

to the apparent complexity and diversity of the underlying

mechanisms and resulting pathologies and a lack of

repro-ducible human cardiomyocyte cells Recent advances in the

development and production of human embryonic stem

cell-derived cardiomyocytes (hESC-CMs) has facilitated the

devel-opment of human in vitro cell models of cardiotoxicity

(Jonsson et al., 2009) Pointon et al (2013) utilized hESC-CMs

and the rat myoblastic H9c2 cell line to phenotypically

profile a panel of structural cardiotoxins (34 clinical and

AstraZeneca internal drug candidates with known in vivo

structural cardiotoxic liabilities and 32 clinical and

AstraZen-eca internal drug candidate non-structural cardiotoxins)

High content screening was carried out using live-cell

fluorescent imaging of mitochondrial membrane potential,

endoplasmic reticulum integrity, Ca2+ mobilization, and

membrane permeability combined with an assessment of cell

viability, via ATP content Results showed that the hESC-CM

screen shows greater sensitivity and specificity compared

with the immortalized rat cardiomyocyte cell line (H9C2)

Data suggested that hESC-CMs in combination with imaging

parameters and assessment of cell viability were able to

predict the in vivo (animal) outcome (structural

cardiotoxic-ity) with an overall sensitivity and specificity of 74%

Impor-tantly, this study also demonstrated that a beating phenotype

is important for the development of structural toxicity,

high-lighting the intricate relationship between cardiac functional

and structural toxicity

The challenge for in vitro cell models with acute drug

treatment is the direct predictability of the chronic forms of

toxicity, which may take years to manifest in patients because

of cumulative dosing with drugs such as anthracyclines The

development of 3-D stem cell-based cardiac microtissues/

spheroids may offer the potential for repeated drug dosing

and cumulative effects with more predictive chronic toxicity

potential (Thavandiran et al., 2013).

Novel human cell-based models to

predict cardiac microvascular toxicity

The cardiac myocardium is composed of cardiomyocytes,

which constitute approximately 30% of the total cells, and

non-myocytes (fibroblasts, endothelial), which constitute

approximately 70% of the total cells (Brutsaert, 2003)

Car-diomyocytes generate the contractile force while fibroblasts

secrete extracellular matrix and paracrine factors Endothelial

cells line the coronary vasculature and allow delivery, via the

bloodstream, of the free fatty acids and oxygen required to

meet the high metabolic demands of the contractile

myo-cytes (Brutsaert, 2003; Tirziu et al., 2010) Endothelial cells

also form a barrier to thrombus formation and regulate theadherence of immune cells There is a growing awareness thatcardiotoxic anti-cancer drugs can also adversely affect cardiacvascular function Tubulin binding drugs, such as vincristine,have been shown to adversely affect rat cardiac microvascular

endothelial cells (Mikaelian et al., 2010) while doxorubicin

has recently been shown to affect VEGF signalling in rat

cardiac microvascular endothelial cells (Chiusa et al., 2012).

Recent data have shown that sunitinib (Sutent), which targets

a range of different receptor tyrosine kinases [VEGF receptors,Kit, PDGF receptors and Fms-like tyrosine kinase-3 (FLT3)]can adversely affect cardiac pericytes resulting in cardiac tox-

icity (Chintalgattu et al., 2013) Collectively, these data show

that drug-induced cardiac toxicity can have a multicellularcomponent This has profound implications for the develop-

ment of in vitro preclinical cardiovascular toxicity screens

within the pharmaceutical industry, which are currentlyfocused solely on detecting adverse effects in cardiomyocytes.There is a need for multicellular models that reconstitutecardiac physiology to allow researchers to simultaneouslyinvestigate structural changes and functional changes in dif-ferent cardiac cell types

Conclusions and recommendations

Although significant progress has been made in better standing the pathogenesis and mechanisms of structural car-diotoxicity caused by a number of novel therapeutics, the

under-ultimate goal is to identify more comprehensive in vivo screens and, in particular, in vitro assays, which are reliable

predictors of clinical cardiotoxicity, as achieved with mentation of hERG-centred screening to identify risks ofdrug-induced QT-interval prolongation Efforts to address thefollowing issues are encouraged and supported:

imple-(a) Standard histopathological (light microscopic) ment may not identify early changes in cardiovascularmorphology and without the aid of additional, sensitive,evaluative tools such as transmission EM or measure-

assess-ment of appropriate early biomarkers (Casartelli et al.,

2011), longer-term studies (with associated patient risks,increased costs and lengthening of project timelines)may be required to determine any liabilities Althoughimpractical to propose routine EM in early toxicitystudies, there have been encouraging signs of early detec-tion of cardiac toxicities using translatable biomarkers(e.g cTnI) and further research in this area is expected toprovide significant benefits More studies are requiredalso on mechanistic understandings and functional cor-relates that can act as better translational biomarkers(e.g serological, functional, histopathological, imaging)

An industry consortia approach has demonstrated trendstowards a relationship between functional changes inacute, single-dose studies in dogs and cardiovascularpathologies seen in (28 day) repeat-dose toxicologystudies Cell-based studies using cardiomyocytes

(Pointon et al., 2013) have shed light also on

relation-ships between a beating phenotype and mitochondrial

British Journal of Pharmacology (2015) 172 957–974 969

toxicity and calcium disruption as common mechanisms

for structural toxicity

(b) The identification and prediction of drug-induced

struc-tural cardiotoxicity in patients is complicated by the fact

that in many cases where the left ventricle is

mechani-cally disadvantaged, there may be no clinimechani-cally apparent

manifestations for many years because of an

optimiza-tion of funcoptimiza-tion by compensatory mechanisms Thus,

better independent prognostic factors are needed to

enable clinicians to more accurately stratify risk in

patients with cardiomyopathy and to implement early

preventative drug treatment Although there have been

major advances in the area of functional assessment

clinically, it is currently widely accepted that change of

function is preceded by damage to the myocardium,

which can be detected by biomarkers thus allowing a

more rapid and cheaper way to monitor patients on a

regular basis Indeed, a number of studies suggest that

cTnI determination is able to predict the occurrence of a

clinically significant LVD, at least 3 months in advance

(Cardinale et al., 2010; Sawaya et al., 2012) As it is now

becoming apparent from ultrastructural assessment

that toxicities can affect a range of different cardiac cell

types, it is important also therefore to identify early

markers which help determine the specific liability in

patients

(c) The choice of translatable animal models (and in vitro

screens) remains highly challenging as toxicities in

patients are ‘personalized’ as a result of a specific

predis-position brought about by specific prior drug treatments

or pre-existing disease(s) The use of young, healthy and

drug-nạve animals for toxicity studies likely reflects

poorly in terms of translatability to patients whereas

animals carrying a background disease, are stressed by

prior drug treatment, or are genetically modified, may

reflect better the clinical condition However,

differenti-ating toxicity from natural worsening of disease can be a

significant challenge in both preclinical and clinical

set-tings A key challenge is to know which ‘compromised’

animal models to apply and in which disease and

thera-peutic settings Successful case studies and further

research is required to inform use of these models and to

better understand the comparative cardiovascular

physiology/pathophysiology to help address preclinical

species to human translatability It is inappropriate to

consider rodent models as truly predictive and they

should be regarded as tools to detect putative liabilities;

nevertheless, many compounds have been discarded

because of rodent toxicity findings, which may or may

not have translated to clinical adverse effects Although

standard preclinical safety studies have prevented any

phase 1 catastrophes from occurring they have not

yielded much insight into later stage toxicities

(d) No predictive in vitro approaches to detect structural

car-diovascular toxicity have been described, in part because

of the challenge of recreating complex cardiovascular

physiology in an isolated set-up, as well as an incomplete

understanding of the mechanisms of toxicity to decide

the best end points Promising findings have been

pub-lished recently, however, by Pointon et al (2013) using

hESC-CMs to phenotypically profile a panel of structural

and non-structural cardiotoxins There is a need,however, to generate cell models that reconstitute mul-ticellular cardiac physiology more accurately andco-culture of cardiomyocytes, cardiac fibroblasts andcardiac endothelial cells may achieve this goal A well-characterized set of structural cardiotoxins is nowrequired by the scientific community to help compare

and contrast the different in vitro and in vivo models (and

species) An extensive and rigorous comparative study todetermine if agents that are cardiotoxic after repeatdosing in rodents and dogs also elicit changes in human(as well as rodent/dog) isolated or co-cultured cardio-myocytes would help the identification of more predic-tive models and may lead to the development of atemplate for future screening strategies and a risk assess-ment plan

(e) Pre-competitive data sharing on targets, which are sically associated with toxicity, is considered a key goal

intrin-in order to avoid potentially costly and wasted efforts bythe industry The publication by AstraZeneca scientists

(Anderton et al., 2011) showing that inhibiting a

poten-tial oncology target (ALK5) results in histopathologicalheart valve lesions in rodents benefited a significantnumber of companies who were already engaged insimilar programmes or were considering starting chemi-cal efforts on the target We encourage industry to followthe lines taken by AstraZeneca and publish their findings

on target toxicities at earliest possible opportunities toreduce needless animal usage and for broader scientificbenefit

(f) A strategy to mitigate the risks of cardiotoxicity inpatients is the development of cardio-protective agentsfor co-administration Currently, the only approvedclinical cardio-protective drug is dexazoxane, targetinganthracycline-induced cardiotoxicity New approachesbased on different mechanisms (e.g neuregulin-1 andcdk 4/6 inhibitors) are currently under early investiga-tion This area is ripe for further development of moreeffective agents and ultimately will allow maximallyeffective therapies to be administered with minimizedrisk of cardiotoxicity

(g) Cardio-oncology patients represent a unique cohort asthe timing of the insult is known and there is someunderstanding of the mechanism of insult induced bythe drug Much can be learned from this population,which may be applicable to heart disease from othercauses However, closer interactions between cardiolo-gists and oncologists, along the lines of the newlyestablished UK cardio-oncology services and networksare encouraged to better address the issue of earlyassessment/identification Greater interaction is requiredalso between preclinical and clinical safety scientists toensure that any gaps occurring in development of newagents are quickly identified A more iterative approach

is developing within the industry such that any clinicalfindings trigger preclinical modelling to better under-stand the liability and its mechanism Equally, if a car-diovascular safety signal is seen preclinically it isessential that this is followed with special attention inthe clinic and a strategy is implemented for clinicalmonitoring

970 British Journal of Pharmacology (2015) 172 957–974

Conflict of interest

Some authors of this paper are employed in the

pharmaceu-tical industry or serve as consultants to the pharmaceupharmaceu-tical

industry However, the subjects presented in the paper do not

advocate or support purchase of any of the products offered

by the respective organization

References

Aguirre SA, Heyen JR, Collette W, Bobrowski W, Blasi ER (2010)

Cardiovascular effects in rats following exposure to a receptor

tyrosine kinase inhibitor Toxicol Pathol 38: 416–428

Alexander SPH, Benson HE, Faccenda E, Pawson AJ, Sharman JL,

McGrath JC et al (2013a) The Concise Guide to PHARMACOLOGY

2013/14: Overview Br J Pharmacol 170: 1449–1458

Alexander SPH, Benson HE, Faccenda E, Pawson AJ, Sharman JL,

Spedding M et al (2013b) The Concise Guide to PHARMACOLOGY

2013/14: G protein-coupled receptors Br J Pharmacol 170:

1459–1581

Alexander SPH, Benson HE, Faccenda E, Pawson AJ, Sharman JL,

Spedding M et al (2013c) The Concise Guide to PHARMACOLOGY

2013/14: Ion channels Br J Pharmacol 170: 1607–1651

Alexander SPH, Benson HE, Faccenda E, Pawson AJ, Sharman JL,

Spedding M et al (2013d) The Concise Guide to PHARMACOLOGY

2013/14: Catalytic receptors Br J Pharmacol 170: 1676–1705

Alexander SPH, Benson HE, Faccenda E, Pawson AJ, Sharman JL,

Spedding M et al (2013e) The Concise Guide to PHARMACOLOGY

2013/14: Enzymes Br J Pharmacol 170: 1797–1867

Anderton MJ, Mellor HR, Bell A, Sadler C, Pass M, Powell S et al.

(2011) Induction of heart valve lesions by small-molecule ALK5

inhibitors Toxicol Pathol 39: 916–924

Bello CL, Mulay M, Huang X, Patyna S, Dinolfo M, Levine S et al.

(2009) Electrocardiographic characterization of the QTc interval in

patients with advanced solid tumors:

pharmacokinetic–pharmacodynamic evaluation of sunitinib Clin

Cancer Res 15: 7045–7052

Berardi R, Caramanti M, Savini A, Chiorrini S, Pierantoni C, Onofri

A et al (2013) State of the art for cardiotoxicity due to

chemotherapy and to targeted therapies: a literature review Crit

Rev Oncol Hematol 88: 75–86

Billingham ME, Mason JW, Bristow MR, Daniels JR (1978)

Anthracycline cardiomyopathy monitored by morphologic changes

Cancer Treat Rep 62: 865–872

Boivin F, Hutchison GB, Lubin JH, Mauch P (1992) Coronary

artery disease mortality in patients treated for Hodgkin’s disease

Cancer 69: 1241–1247

Brave M, Goodman V, Kaminskas E, Farrell A, Timmer W, Pope S

et al (2008) Sprycel for chronic myeloid leukemia and Philadelphia

chromosome-positive acute lymphoblastic leukemia resistant to or

intolerant of imatinib mesylate Clin Cancer Res 14: 352–359

Braz JC, Bueno OF, Liang Q, Wilkins BJ, Dai YS, Parsons S (2003)

Targeted inhibition of p38 MAPK promotes hypertrophic

cardiomyopathy through upregulation of calcineurin-NFAT

signaling J Clin Invest 111: 1475–1486

Broekman F, Giovannetti E, Peters GJ (2011) Tyrosine kinase

inhibitors: multi-targeted or single-targeted? World J Clin Oncol 2:

80–93

Brutsaert DL (2003) Cardiac endothelial-myocardial signaling: itsrole in cardiac growth, contractile performance, and rhythmicity.Physiol Rev 83: 59–115

Cardinale D, Sandri MT, Colombo A, Colombo N, Boeri M,

Lamantia G et al (2004) Prognostic value of troponin I in cardiac

risk stratification of cancer patients undergoing high-dosechemotherapy Circulation 109: 2749–2754

Cardinale D, Colombo A, Lamantia G, Colombo N, Civelli M,

De Giacomi G et al (2010) Anthracycline-induced cardiomyopathy

– clinical relevance and response to pharmacologic therapy J AmColl Cardiol 55: 213–220

Casartelli A, Lanzoni A, Comelli R, Crivellente F, Defazio R,

Dorigatti R et al (2011) A novel and integrated approach for the

identification and characterization of drug-induced cardiac toxicity

in the dog Toxicol Pathol 39: 361–371

Chen XL, Lei YH, Liu CF, Yang QF, Zuo PY, Liu CY et al (2013).

Angiogenesis inhibitor bevacizumab increases the risk of ischemicheart disease associated with chemotherapy: a meta-analysis PLoSONE 8: e66721

Cheng H, Kari G, Dicker AP, Rodeck U, Koch WJ, Force T (2011)

A novel preclinical strategy for identifying cardiotoxic kinaseinhibitors and mechanisms of cardiotoxicity Circ Res 109:

1401–1409

Chintalgattu V, Rees ML, Culver JC, Goel A, Jiffar T, Zhang J et al.

(2013) Coronary microvascular pericytes are the cellular target ofsunitinib malate-induced cardiotoxicity Sci Transl Med 5:

187ra69

Chiusa M, Hool SL, Truetsch P, Djafarzadeh S, Jakob SM, Seifriz F

et al (2012) Cancer therapy modulates VEGF signaling and

viability in adult rat cardiac microvascular endothelial cells andcardiomyocytes J Mol Cell Cardiol 52: 1164–1175

Choueiri TK, Schutz FAB, Je Y, Rosenberg JE, Bellmunt J (2010).Risk of arterial thromboembolic events with sunitinib andsorafenib: a systematic review and meta-analysis of clinical trials

J Clin Oncol 28: 2280–2285

Choueiri TK, Mayer EL, Je Y, Rosenberg JE, Rosenberg JE, Nguyen

PL et al (2011) Congestive heart failure risk in patients with

breast cancer treated with bevacizumab J Clin Oncol 29:

632–638

Chu TF, Rupnick MA, Kerkela R, Dallabrida SM, Zurakowski D,

Nguyen L et al (2007) Cardiotoxicity associated with tyrosine

kinase inhibitor sunitinib Lancet 370: 2011–2019

Cove-Smith L, Woodhouse N, Hargreaves A, Kirk J, Smith S, Price

SA et al (2014) An integrated characterisation of serological,

pathological and functional events in doxorubicin-inducedcardiotoxicity Toxicol Sci 140: 3–15

Crackower MA, Oudit GY, Kozieradzki I, Sarao R, Sun H,

Sasaki T et al (2002) Regulation of myocardial contractility

and cell size by distinct PI3K-PTEN signaling pathways Cell 110:737–749

Crone SA, Zhao YY, Fan L, Gu Y, Minamisawa S, Liu Y et al (2002).

ErbB2 is essential in the prevention of dilated cardiomyopathy NatMed 8: 459–465

Dagres N, Hindricks G (2013) Risk stratification after myocardialinfarction: is left ventricular ejection fraction enough to preventsudden cardiac death Eur Heart J 34: 1964–1971

De Rosa S, Curcio A, Indolfi C (2014) Emerging role of microRNAs

in cardiovascular diseases Circ J 78: 567–575

British Journal of Pharmacology (2015) 172 957–974 971

Dolci A, Dominici R, Cardinale D, Sandri MT, Panteghini M (2008).

Biochemical markers for prediction of chemotherapy-induced

cardiotoxicity Am J Clin Pathol 130: 688–695

Dzeja PP, Bast P, Pucar D, Wieringa B, Terzic A (2007) Defective

metabolic signaling in adenylate kinase AK1 gene knock-out hearts

compromises post-ischemic coronary reflow J Biol Chem 282:

31366–31372

Escudier B, Eisen T, Stadler WM, Szczylik C, Oudard S, Staehler M

et al (2009) Sorafenib for treatment of renal cell carcinoma: final

efficacy and safety results of the phase III treatment approaches

in renal cancer global evaluation trial J Clin Oncol 27:

3312–3318

Ewer MS, Ewer SM (2010) Cardiotoxicity of anticancer treatments:

what the cardiologist needs to know Nat Rev Cardiol 7:

564–575

Ewer MS, Lippman SM (2005) Type II chemotherapy-related cardiac

dysfunction: time to recognize a new entity J Clin Oncol 23:

2900–2902

Faivre S, Delbaldo C, Vera K, Robert C, Lozahic S, Lassau N et al.

(2006) Safety, pharmacokinetic and antitumor activity of SU11248,

a novel oral multitarget tyrosine kinase inhibitor, in patients with

cancer J Clin Oncol 24: 25–35

Farolfi A, Melegari E, Aquilina M, Scarpi E, Ibrahim T, Maltoni R

et al (2013) Trastuzumab-induced cardiotoxicity in early breast

cancer patients: a retrospective study of possible risk and protective

factors Heart 99: 634–639

Force T, Kolaja KL (2011) Cardiotoxicity of kinase inhibitors: the

prediction and translation of preclinical models to clinical

outcomes Nat Rev Drug Discov 10: 111–126

Force T, Wang Y (2013) Mechanism-based engineering against

anthracycline cardiotoxicity Circulation 128: 98–100

Fukazawa R, Miller TA, Kuramochi Y, Frantza S, Kima Y-D,

Marchionnic MA et al (2003) Neuregulin-1 protects venricular

myocytes from anthracycline-induced apoptosis via

erbB4-dependent activation of PI3-kinase/Akt J Mol Cell Cardiol

35: 1473–1479

Harmer AR, Abi-Gerges N, Morton MJ, Pullen GF, Valentin JP,

Pollard CE (2012) Validation of an in vitro contractility assay

using canine ventricular myocytes Toxicol Appl Pharmacol 260:

162–172

Hooning MJ, Botma A, Aleman BMP, Baaijens MHA, Bartelink H,

Klijn JGM et al (2007) Long-term risk of cardiovascular disease

in 10-year survivors of breast cancer J Natl Cancer Inst 99:

365–375

Hu W, Hirakawa B, Jessen B, Lee M, Aguirre S (2012) A tyrosine

kinase inhibitor-induced myocardial degeneration in rats through

off-target phosphodiesterase inhibition J Appl Toxicol 32:

1008–1020

Ito K, Akazawa H, Tamagawa M, Furukawa K, Ogawa W, Yasuda N

et al (2009) PDK1 coordinates survival pathways and

beta-adrenergic response in the heart Proc Natl Acad Sci U S A 106:

8689–8694

Jay SM, Murthy AC, Hawkins JF, Wortzel JR, Steinhauser ML,

Alvarez LM et al (2013) An engineered bivalent neuregulin protects

against doxorubicin-induced cardiotoxicity with reduced

proneoplastic potential Circulation 128: 152–161

Jones RL (2008) Utility of dexrazoxane for the reduction of

anthracycline-induced cardiotoxicity Expert Rev Cardiovasc Ther 6:

1311–1317

Jonsson MKB, van Veen TAB, Goumans MJ, Vos MA, Duker G,

Sartipy P et al (2009) Improvement of cardiac efficacy and safety

models in drug discovery by the use of stem cell-derivedcardiomyocytes Expert Opin Drug Discov 4: 357–372

Kantarjian HM, Giles F, Gattermann N, Bhalla K, Alimena G,

Palandri F et al (2007) Nilotinib (formerly AMN107), a highly

selective BCR-ABL tyrosine kinase inhibitor, is effective in patientswith Philadelphia chromosome-positive chronic myelogenousleukemia in chronic phase following imatinib resistance andintolerance Blood 110: 3540–3546

Kerkelä R, Grazette L, Yacobi R, Iliescu C, Patten R, Beahm C et al.

(2006) Cardiotoxicity of the cancer therapeutic agent imatinibmesylate Nat Med 12: 908–916

Kimura TE, Jin J, Zi M, Prehar S, Liu W, Oceandy D et al (2010).

Targeted deletion of the extracellular signal-regulated protein kinase

5 attenuates hypertrophic response and promotes pressureoverload-induced apoptosis in the heart Circ Res 106: 961–970.Kontaridis MI, Yang W, Bence KK, Cullen D, Wang B, Bodyak N

et al (2008) Deletion of Ptpn11 (Shp2) in cardiomyocytes causes

dilated cardiomyopathy via effects on the extracellularsignal-regulated kinase/mitogen-activated protein kinase and RhoAsignaling pathways Circulation 117: 1423–1435

Ky B, Vejpongsa P, Yeh ET, Force T, Moslehi JJ (2013) Emergingparadigms in cardiomyopathies associated with cancer therapies.Circ Res 113: 754–764

Larson JL, Pino MV, Geiger LE, Simeone CR (1996) The toxicity ofrepeated exposures to rolipram, a type IV phosphodiesteraseinhibitor in rats Pharmacol Toxicol 78: 44–49

Laverty HG, Benson C, Cartwright E, Cross M, Garland C,

Hammond T et al (2011) How can we improve our understanding

of cardiovascular safety liabilities to develop safer medicines? Br JPharmacol 163: 675–693

Lipshultz SE, Rifai N, Dalton VM, Levy DE, Silverman LB, Lipsitz SR

et al (2004) The effect of dexrazoxane on myocardial injury in

doxorubicin-treated children with acute lymphoblastic leukemia

N Engl J Med 351: 145–153

Lyu YL, Kerrigan JE, Lin CP, Azarova AM, Tsai YC, Ban Y et al.

(2007) Topoisomerase IIbeta mediated DNA double-strand breaks:implications in doxorubicin cardiotoxicity and prevention bydexrazoxane Cancer Res 67: 8839–8846

McClendon AK, Dean JL, Rivadeneira DB, Yu JE, Reed CA, Gao E

et al (2012) CDK4/6 inhibition antagonizes the cytotoxic response

to anthracycline therapy Cell Cycle 11: 2747–2755

Mellor HR, Bell AR, Valentin JP, Roberts RRA (2011) Cardiotoxicityassociated with targeting kinase pathways in cancer Toxicol Sci120: 14–32

Mikaelian I, Buness A, de Vera-Mudry MC, Kanwal C, Coluccio D,

Rasmussen E et al (2010) Primary endothelial damage is the

mechanism of cardiotoxicity of tubulin-binding drugs Toxicol Sci117: 144–151

Milliken P, Aylott M, Clements A, Edmunds N, Engle S, Ewart L

et al (2012) A cross-company initiative assessing relationships

between cardiovascular functional measurements and repeat-dosecardiovascular toxicity Toxicologist 126: 491 51st Annual Meeting

of the Society of Toxicology Poster #2275

Montani D, Bergot E, Günther S, Savale L, Bergeron A, Bourdin A

et al (2012) Pulmonary arterial hypertension in patients treated by

dasatinib Circulation 125: 2128–2137

Mora A, Davies AM, Bertrand L, Sharif I, Budas GR, Jovanovic´ S

et al (2003) Deficiency of PDK1 in cardiac muscle results in heart

failure and increased sensitivity to hypoxia EMBO J 22: 4666–4676

972 British Journal of Pharmacology (2015) 172 957–974

Motzer RJ, Hutson TE, Cella D, Reeves J, Hawkins R, Guo J et al.

(2013) Pazopanib versus sunitinib in metastatic renal-cell

carcinoma N Engl J Med 369: 722–731

Muraski JA, Fischer KM, Wu W, Cottage CT, Quijada P, Mason M

et al (2008) Pim-1 kinase antagonizes aspects of myocardial

hypertrophy and compensation to pathological pressure overload

Proc Natl Acad Sci U S A 105: 13889–13894

Naib T, Steingart RM, Chen CL (2011) Sorafenib-associated

multivessel coronary artery vasospasm Herz 36: 348–351

Oudit GY, Kassiri Z, Zhou J, Liu QC, Liu PP, Backx PH et al (2008).

Loss of PTEN attenuates the development of pathological

hypertrophy and heart failure in response to biomechanical stress

Cardiovasc Res 78: 505–514

Ozcelik C, Erdmann B, Pilz B, Wettschureck N, Britsch S, Hübner N

et al (2002) Conditional mutation of the ErbB2 (HER2) receptor in

cardiomyocytes leads to dilated cardiomyopathy Proc Natl Acad Sci

U S A 99: 8880–8885

Patnaik JL, Byers T, DiGuiseppi C, Dabelea D, Denberg TD (2011)

Cardiovascular disease competes with breast cancer as the leading

cause of death for older females diagnosed with breast cancer: a

retrospective cohort study Breast Cancer Res 13: R64

Pawson AJ, Sharman JL, Benson HE, Faccenda E, Alexander SP,

Buneman OP et al.; NC-IUPHAR (2014) The IUPHAR/BPS Guide to

PHARMACOLOGY: an expert-driven knowledgebase of drug targets

and their ligands Nucl Acids Res 42 (Database Issue):

D1098–D1106

Perez EA, Koehler M, Byrne J, Preston AJ, Rappold E, Ewer MS

(2008) Cardiac safety of lapatinib: pooled analysis of 3689 patients

enrolled in clinical trials Mayo Clin Proc 83: 679–686

Perik PJ, de Korte MA, van Veldhuisen DJ, Gietema JA, Sleijfer DT,

de Vries EGE (2007) Cardiotoxicity associated with the use of

trastuzumab in breast cancer patients Expert Rev Anticancer Ther

7: 1763–1771

Pointon A, Abi-Gerges N, Cross MJ, Sidaway JE (2013) Phenotypic

profiling of structural cardiotoxins in vitro reveals dependency on

multiple mechanisms of toxicity Toxicol Sci 132: 317–326

Pollard CE, Abi-Gerges N, Bridgland-Taylor MH, Easter A,

Hammond TG, Valentin JP (2010) An introduction to QT interval

prolongation and non-clinical approaches to assessing and reducing

risk Br J Pharmacol 159: 12–21

Princen F, Bard E, Sheikh F, Zhang SS, Wang J, Zago WM et al.

(2009) Deletion of Shp2 tyrosine phosphatase in muscle leads to

dilated cardiomyopathy, insulin resistance, and premature death

Mol Cell Biol 29: 378–388

Regan CP, Li W, Boucher DM, Spatz S, Su MS, Kuida K (2002) Erk5

null mice display multiple extraembryonic vascular and embryonic

cardiovascular defects Proc Natl Acad Sci U S A 99: 9248–9253

Roberts LD, Souza AL, Gerszten RE, Clish CB (2012) Targeted

metabolomics Curr Protoc Mol Biol 98: 30.2.1–30.2.24 Chapter 30:

Unit 30.2

Sawaya H, Sebag IA, Plana JC, Januzzi JL, Ky B, Tan TC (2012)

Assessment of echocardiography and biomarkers for the extended

prediction of cardiotoxicity in patients treated with anthracyclines,

taxanes, and trastuzumab Circ Cardiovasc Imaging 5: 596–603

Scheffel RS, Dora JM, Siqueira DR, Burttet LM, Cerski MR, Maia AL

(2013) Toxic cardiomyopathy leading to fatal acute cardiac failure

related to vandetanib: a case report with histopathological analysis

Eur J Endocrinol 168: K51–K54

Schmidinger M, Zielinski CC, Vogl UM, Bojic A, Bojic M, Schukro C

et al (2008) Cardiac toxicity of sunitinib and sorafenib in patients

with metastatic renal cell carcinoma J Clin Oncol 26: 5204–5212

Seidman AJ, Hudis C, Pierri MK, Shak S, Paton V, Ashby M et al.

(2002) Cardiac dysfunction in the trastuzumab clinical trialsexperience J Clin Oncol 20: 1215–1221

Shakir DK, Rasul KI (2009) Chemotherapy-inducedcardiomyopathy: pathogenesis, monitoring and management

J Clin Med Res 1: 8–12

Slamon D, Eiermann W, Robert N, Pienkowski T, Martin M, Press

M et al (2011) Adjuvant trastuzumab in HER2-positive breast

cancer N Engl J Med 365: 1273–1283

Sridurongrit S, Larsson J, Schwartz R, Ruiz-Lozano P, Kaartinen V(2008) Signaling via the TGF-beta type I receptor Alk5 in heartdevelopment Dev Biol 322: 208–218

Sterba M, Popelova O, Vavrova A, Jirkovsky E, Kovarikova P, Gersl

V et al (2013) Oxidative stress, redox signalling and metal

chelation in anthracycline cardiotoxicity and pharmacologicalcardioprotection Antioxid Redox Signal 18: 899–929

Stortecky S, Suter TM (2010) Insights into cardiovascularside-effects of modern anticancer therapeutics Curr Opin Oncol 22:312–317

Tandri H, Saranathan M, Rodriguez ER, Martinez C, Bomma C,

Nasir K et al (2005) Noninvasive detection of myocardial fibrosis

in arrhythmogenic right ventricular cardiomyopathy usingdelayed-enhancement magnetic resonance imaging J Am CollCardiol 45: 98–103

Tefferi A (2013) Nilotinib treatment-associated acceleratedatherosclerosis: when is the risk justified? Leukemia 27: 1939–1940.Telli ML, Witteles RM, Fisher GA, Srinivas S (2008) Cardiotoxicityassociated with the cancer therapeutic agent sunitinib malate AnnOncol 19: 1613–1618

Thavandiran N, Dubois N, Mikryukov A, Masse S, Beca B, Simmons

CA et al (2013) Design and formulation of functional pluripotent

stem-cell-derived cardiac microtissues Proc Natl Acad Sci U S A110: E4698–E4707

Tirziu D, Giordano FJ, Simons M (2010) Cell communications inthe heart Circulation 122: 928–937

Uraizee I, Cheng S, Moslehi J (2011) Reversible cardiomyopathyassociated with sunitinib and sorafenib N Engl J Med 365:1649–1650

Valentin J-P, Bialecki R, Ewart L, Hammond T, Leishmann D,

Lindgren S et al (2009) A framework to assess the translation of

safety pharmacology data to humans J Pharmacol Toxicol Methods60: 152–158

Veronese ML, Mosenkis A, Flaherty KT, Gallagher M, Stevenson JP,

Townsend RR et al (2006) Mechanisms of hypertension associated

with BAY 43-9006 J Clin Oncol 24: 1363–1369

Von Hoff DD, Lavard MW, Basa P, Davis HL, Von Hoff AL,

Rozencweig M et al (1979) Risk factors for doxorubicin-induced

congestive heart failure Ann Intern Med 91: 710–717

Weisberg E, Manley PW, Breitenstein W, Brüggen J, Cowan-Jacob

SW, Ray A et al (2005) Characterization of AMN107, a selective

inhibitor of native and mutant Bcr-Abl Cancer Cell 7:

129–141

White DE, Coutu P, Shi YF, Tardif J-C, Nattel S, St Arnaud R et al.

(2006) Targeted ablation of ILK from the murine heart results indilated cardiomyopathy and spontaneous heart failure Genes Dev20: 2355–2360

Yamaguchi O, Watanabe T, Nishida K, Kashiwase K, Higuchi Y,

Takeda T et al (2004) Cardiac-specific disruption of the c-raf-1 gene

induces cardiac dysfunction and apoptosis J Clin Invest 114:937–943

British Journal of Pharmacology (2015) 172 957–974 973

Yoon GJ, Telli ML, Kao DP, Matsuda KY, Carlson RW, Witteles RM

(2010) Left ventricular dysfunction in patients receiving

cardiotoxic cancer therapies: are clinicians responding optimally?

J Am Coll Cardiol 56: 1644–1650

Zhang DY, Wang Y, Lau CP, Tse HF, Li GR (2008) Both EGFR

kinase and Src-related tyrosine kinases regulate human

ether-a-go-go-related gene potassium channels Cell Signal 20:1815–1821

Zhang S, Liu X, Bawa-Khalfe T, Lu LS, Lyu YL, Liu LF et al (2012).

Identification of the molecular basis of doxorubicin-inducedcardiotoxicity Nat Med 18: 1639–1642

974 British Journal of Pharmacology (2015) 172 957–974

RESEARCH PAPER

Hydrogen sulphide protects

against NSAID-enteropathy

through modulation of bile

and the microbiota

Rory W Blackler1, Jean-Paul Motta2, Anna Manko1,

Matthew Workentine1, Premysl Bercik1, Michael G Surette1 and

John L Wallace2,3

1

Department of Medicine, McMaster University, Hamilton, ON, Canada,2

Department of Physiology and Pharmacology, University of Calgary, Calgary, AB, Canada, and3Department of

Pharmacology and Toxicology, University of Toronto, Toronto, ON, Canada

Correspondence

Dr John L Wallace, Department

of Physiology and Pharmacology,University of Calgary, 3330Hospital Drive NW, Calgary,Alberta, T2N 4N1, Canada

BACKGROUND AND PURPOSE

Hydrogen sulphide is an important mediator of gastrointestinal mucosal defence The use of non-steroidal anti-inflammatorydrugs (NSAIDs) is significantly limited by their toxicity in the gastrointestinal tract Particularly concerning is the lack ofeffective preventative or curative treatments for NSAID-induced intestinal damage and bleeding We evaluated the ability of ahydrogen sulphide donor to protect against NSAID-induced enteropathy

EXPERIMENTAL APPROACH

Intestinal ulceration and bleeding were induced in Wistar rats by oral administration of naproxen The effects of suppression

of endogenous hydrogen sulphide synthesis or administration of a hydrogen sulphide donor (diallyl disulphide) on

naproxen-induced enteropathy was examined Effects of diallyl disulphide on small intestinal inflammation and intestinal

microbiota were also assessed Bile collected after in vivo naproxen and diallyl disulphide administration was evaluated for cytotoxicity in vitro using cultured intestinal epithelial cells.

KEY RESULTS

Diallyl disulphide co-administration dose-dependently reduced the severity of naproxen-induced small intestinal damage,inflammation and bleeding Diallyl disulphide administration attenuated naproxen-induced increases in the cytotoxicity of bile

on cultured enterocytes, and prevented or reversed naproxen-induced changes in the intestinal microbiota

CONCLUSIONS AND IMPLICATIONS

Hydrogen sulphide protects against NSAID-enteropathy in rats, in part reducing the cytotoxicity of bile and preventingNSAID-induced dysbiosis

Abbreviations

brain heart infusion; CBA, Columbia blood agar; CFU, colony-forming units; DADS, diallyl disulphide; DGGE,

denaturing gradient gel electrophoresis; GI, gastrointestinal; MPO, myeloperoxidase; NSAID, non-steroidal

anti-inflammatory drugs; PPI, proton pump inhibitor; TLR, Toll-like receptor; UPGMA, unweighted-pair-group methodwith arithmetic mean

992 British Journal of Pharmacology (2015) 172 992–1004 © 2014 The British Pharmacological Society

Non-steroidal anti-inflammatory drugs (NSAIDs) are among

the most widely used drugs for treating the symptoms of

inflammatory conditions, most notably osteoarthritis and

rheumatoid arthritis In such conditions, NSAIDs are taken

chronically and have the ability to cause significant

ulcera-tion and bleeding in the gastrointestinal (GI) tract Therapies

designed to limit NSAID-induced GI injury have focused

almost exclusively on gastroduodenal injury, often ignoring

the serious small intestinal damage that can also occur

(Wallace, 2013b) This is a concern because it is now clear that

NSAID-enteropathy occurs more frequently than

gastroduo-denal injury (Lanas et al., 2009), and it can also be more

dangerous, given that there is a poor correlation of symptoms

with the injury, the damage is more difficult to detect and

there are no proven effective preventative or curative

treat-ments for NSAID-enteropathy (Lanas et al., 2009; Wallace,

2013a) The most common approach to reduce

NSAID-induced gastroduodenal injury is via suppression of gastric

acid secretion, usually through co-administration of a proton

pump inhibitor (PPI) (Scheiman et al., 2006) However, there

is no evidence or rationale to support the notion that

sup-pression of gastric acid secretion would have any benefit in

terms of reducing damage or bleeding distal to the ligament

of Treitz (i.e beyond the proximal duodenum) On the

con-trary, there is emerging evidence that PPIs and histamine H2

receptor antagonists exacerbate the small intestinal damage

and bleeding caused by NSAIDs (Zhao and Encinosa, 2008;

Lanas et al., 2009; Wallace et al., 2011; Blackler et al., 2012;

Satoh et al., 2012) A recent cross-sectional study by

Watanabe et al (2013) highlighted this problem, identifying

PPI usage as the greatest independent risk factor for severe

ulceration and bleeding in patients with rheumatoid arthritis

who were being treated with NSAIDs, with use of histamine

H2 receptor antagonists also being a significant risk factor

Studies in rodents demonstrated that the exacerbation of

NSAID-enteropathy by PPIs is due to a significant shift in the

intestinal microbiota, with a marked decrease in intestinal

colonization by Bifidobacteria (Wallace et al., 2011).

The damaging effects of NSAIDs in the upper GI tract are

directly related to their ability to suppress mucosal synthesis

of PGs (Wallace, 2008) Suppression of PG synthesis renders

the mucosa susceptible to damage induced by luminal agents

such as acid, digestive enzymes, bacteria, bile, and sometimes

by the NSAIDs themselves (Wallace, 2013a) However, it is

now clear that hydrogen sulphide (H2S), an endogenousgaseous mediator, also plays a pivotal role in mucosal defence

and in promoting repair of mucosal injury (Wallace et al., 2007b; 2009; Wallace, 2010; Wallace et al., 2012) Inhibition

of H2S synthesis renders the gastric mucosa more susceptible

to NSAID-induced ulceration (Wallace et al., 2010), whereas

co-administration of H2S donors reduces the severity of

NSAID-induced damage (Fiorucci et al., 2005; Wallace, 2010; Wallace et al., 2014) and promotes healing (Wallace et al.,

2007b) The protective actions of H2S against NSAID-inducedgastric injury are at least in part due to inhibitory effects onNSAID-induced leukocyte adherence to the vascular endothe-

lium (Zanardo et al., 2006), which is an early and critical

event in the pathogenesis of NSAID-induced gastropathy

(Wallace et al., 1990) Moreover, the vasorelaxant effects of

H2S also contribute to mucosal protection by preventing thereduction in gastric blood flow caused by NSAIDs (Fiorucci

et al., 2005; Mard et al., 2012).

The observations of protective effects of H2S in the GItract have prompted the design of a new class of NSAIDs thatrelease H2S These H2S-releasing NSAIDs have been shown toproduce negligible gastroduodenal damage compared withtheir respective parent drugs in healthy rats, in rats withimpaired mucosal defence, and at doses many times greaterthan those required for anti-inflammatory effects (Wallace

et al., 2007a; 2010; Blackler et al., 2012) In addition to

sparing the gastric mucosa, H2S-releasing NSAIDs also cause

negligible damage in the small intestine (Wallace et al.,

2007a; 2010), even when co-administered with PPIs and/oraspirin, which significantly enhance the intestinal-damagingeffects of conventional and COX-2-selective NSAIDs (Wallace

et al., 2011; Blackler et al., 2012).

The pathogenesis of NSAID-enteropathy is distinct fromthat of NSAID-gastropathy (Wallace, 2012) Several studieshave suggested critical roles for bile and for enterohepaticcirculation of NSAIDs in the pathogenesis of NSAID-

enteropathy (Wax et al., 1970; Bjarnason et al., 1993; Seitz

and Boelsterli, 1998) There is also a wealth of evidence thatthe enteric flora contributes to the pathogenesis of NSAID-enteropathy, but it is unclear if the bacteria play a primaryrole in ulcer formation, or merely exacerbate injury once it

has occurred (Uejima et al., 1996; Hagiwara et al., 2004) In

the present study, we have examined the possibility that

a garlic-derived H2S donor, diallyl disulphide (DADS)

(Benavides et al., 2007), could prevent NSAID-induced

enter-opathy in a rat model We also attempted to determine if

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 (Pawson et al., 2014) and are permanently archived in the Concise Guide to PHARMACOLOGY 2013/14 (Alexander et al., 2013).

British Journal of Pharmacology (2015) 172 992–1004 993

effects on enteric flora, bile and/or enterohepatic circulation

of NSAIDs might explain any observed beneficial effects of

this H2S donor

Methods

Animals

All animal care and experimental procedures complied with

the guidelines of the Canadian Council of Animal Care and

were approved by the Animal Care Committee of the Faculty

of Health Sciences at McMaster University All studies

involv-ing animals are reported in accordance with the ARRIVE

guidelines for reporting experiments involving animals

(Kilkenny et al., 2010; McGrath et al., 2010) A total of 148

animals were used in the experiments described here

Effects of a H2S donor on naproxen-induced

enteropathy

Rats (n≥ 6 per group) were treated orally, twice daily, with

naproxen (20 mg·kg−1) or vehicle (DMSO and 1%

carboxym-ethylcellulose; 5:95 ratio) for 4.5 days (nine administrations

in total) Three hours after the final administration of drug or

vehicle, a blood sample was drawn from the tail vein for

measurement of haematocrit (Reuter et al., 1997) The rats

were then anaesthetized with sodium pentobarbital and

blood was drawn from the aorta for measurement (byELISA) of

whole-blood thromboxane (TX) B2¬synthesis, as an index of

systemic COX-1 activity (Wallace et al., 1998) The small

intestine was then evaluated for haemorrhagic damage,

which involved measuring the area, in mm2, of all

hemor-rhagic lesions The damage areas were summed for each rat

to give the ‘intestinal damage score’ (Wallace et al., 2011).

These evaluations were performed without knowledge if the

treatments

Immediately prior to each administration of naproxen or

vehicle, rats were treated with DADS (10, 30 or 60 mmol·kg−1

p.o.) or an equivalent volume of vehicle (1%

carboxymeth-ylcellulose) Damage was assessed and samples were taken, as

described earlier These doses of DADS were selected after

completion of a preliminary dose-ranging study

A series of experiments was performed, using the same

protocol as described earlier, in which a H2S-releasing

deriva-tive of naproxen (ATB-346) was administered at a dose of

20 mg kg−1twice daily The effects of this drug in producing

intestinal damage and in altering cytotoxicity of bile (see

later) were examined

Effects of inhibition of H2S synthesis

Rats (n= 6 per group) were treated orally, twice daily, with a

lower dose of naproxen (10 mg·kg−1) for 4.5 days Previous

studies have demonstrated that this dose of naproxen

signifi-cantly reduced inflammation in a rat-adjuvant arthritis

model and suppressed systemic and small intestinal COX-1

and COX-2 activity (Blackler et al., 2012), but elicited a low

level of damage in the small intestine The rats were also

treated twice daily with an inhibitor of cystathionineγ-lyase

(β-cyano-L-alanine (BCA); 50 mg·kg−1 i.p.; Kawabata et al.,

2007), or with vehicle (PBS) immediately prior to naproxen

administration Three hours after the final dose, the smallintestine was evaluated for damage and samples were col-lected, as described above

PG synthesis

Three hours after the final dose of naproxen, samples ofjejunum and of the corpus region of the stomach were col-lected for the measurement of PGE2 synthesis, as described

previously (Wallace et al., 2000) Briefly, the samples were

excised, weighed, and added to a tube containing 1 mL ofsodium phosphate buffer (10 mmol·L−1; pH 7.4) Using scis-sors, the tissue sample was minced for 30 s then placed in ashaking water bath (37°C) for 20 min The samples werecentrifuged (9000× g) for 30 s and the supernatants were col-lected The concentrations of PGE2in the supernatants weredetermined byELISA

Intestinal inflammation

Intestinal inflammation was assessed in jejunal samples bythe measurement of myeloperoxidase (MPO) activity, a quan-titative index of granulocyte infiltration (Boughton-Smith

et al., 1988), and by histology Samples of jejunal tissues were

collected, fixed and processed by routine techniques for light

microscopy (haematoxylin and eosin staining) (Wallace et al.,

2009)

Intestinal epithelial cell culture (IEC)

Rat IEC-6 and human intestinal epithelial (HT-29) cells wereobtained from American Type Culture Centre (Manassas, VA,USA) IEC-6 cells are a non-transformed, homogenous popu-lation of epithelial-like cells that remain in an undifferenti-ated state, and thus, retain some features consistent with

intestinal crypt cells (Quaroni et al., 1979) Cultures were

maintained in DMEM containing 5% (v·v−1) FBS, 4 mmol·L−1glutamine, 50 U·mL−1penicillin and 50μg·mL−1streptomycin(complete medium) at 37°C and 5% (v·v−1) CO2 Subculturewas carried out at confluence and cells between passages 17and 20 were used for bile cytotoxicity assays Prior to theassays, cells were seeded at 5× 104 cells per well in 24-wellplates and allowed to grow for 1–2 days post-confluence Bilecytotoxicity assays were also conducted using HT-29 epithe-lial cells, which were cultured as previously described (Jobin

et al., 1998).

Collection of bile

One hour after the final administration of drug or vehicle,rats were anaesthetized with sodium pentobarbital A lapa-rotomy was performed and the bile duct was cannulated with

a polyethylene cannula (PE-10; Clay Adams, Parsipany, NJ,USA) Bile was collected for 30 min The bile was stored at

−80°C until use in the cytotoxicity assay

Bile cytotoxicity assay

Bile samples were diluted with Dulbecco’s PBS (DPBS) (pH7.4) immediately prior to incubation with IEC-6 cells Dilu-tions (1:3–1:12) that fell within the physiological range ofconcentrations of bile acids present in the small intestine ofrats (Dietschy, 1968) were assessed for their cytotoxic effects.Cells were washed with warm DPBS prior to bile application.Solutions of bile were added to IEC-6 cells for 3 h at 37°C and

994 British Journal of Pharmacology (2015) 172 992–1004

5% (v·v−1) CO2 Following the incubation period, the cells

were centrifuged at 250× g for 5 min and the supernatants

collected for lactate dehydrogenase measurement, using a

Cytoscan-LDH Cytotoxicity Assay Kit (G-Biosciences, St

Louis, MO, USA) Additional experiments were performed in

a similar manner, but using HT-29 cells

Biliary naproxen levels

Concentrations of naproxen in bile (using coded samples)

were measured by liquid chromatography/mass

spectrom-etry, as described previously (Blackler et al., 2012) These

measurements were carried out by Nucro-Technics

(Scarbor-ough, ON, Canada)

Intestinal bacterial growth

Samples of jejunum (∼200 mg; with the luminal contents

preserved) from rats treated with vehicle or naproxen, and

co-treated with vehicle or DADS, were collected under sterile

conditions and homogenized in PBS Homogenates were kept

on ice until serially diluted and plated onto Columbia blood

agar (CBA, with 5% sheep blood) or brain heart infusion

(BHI) agar and incubated for 48 h under either aerobic or

anaerobic conditions Plates containing between 20 and 200

colony-forming units (CFU) were analysed to determine

bac-terial numbers, and the results expressed as CFU per gram of

tissue

DNA extraction and polymerase chain

reaction–denaturing gradient gel

electrophoresis (DGGE)

Bacterial DNA was extracted from caecal content samples as

previously described (Park et al., 2013) DNA concentrations

were determined spectrophotometrically using a NanoDrop

2000 (Thermo Scientific, Wilmington, DE, USA) The

hyper-variable V3 region of the bacterial 16s ribosomal RNA gene

was amplified using PCR with universal bacterial primers

(HDA-1 and HDA-2) (Mobixlab, McMaster University Core

Facility, Hamilton, ON, Canada) DGGE was performed using

a DCode universal mutation system (Bio-Rad Laboratories,

Mississauga, ON, Canada) Electrophoresis was conducted at

130 V at 60°C for 4.5 h Gels were stained with SYBR Green

(Molecular Probes, Eugene, OR, USA) and viewed by UV

tran-sillumination A scanned image of an electrophoretic gel was

used to measure the staining intensity of the fragments using

Quantity One software (version 4-2; Bio-Rad Laboratories)

The intensity of fragments is expressed as a proportion (%) of

the sum of all fragments in the same lane of the gel

Simi-larities among bacterial profiles were determined using the

Dice coefficient, and the Ward and majority

unweighted-pair-group method with arithmetic mean (UPGMA) algorithms

Construction of majority UPGMA trees was based on a

resa-mpling strategy of 200 permutations UPGMA trees are

dis-played using a multidimensional scaling, which positions the

entry nodes so that they occupy the best possible distance to

each other to reflect the distances in the similarity/distance

matrix

Data analysis

Results are shown as means± SEM The data presented in this

manuscript were analyzed by one-way ANOVA followed by

Dunnett’s or Bonferroni post hoc tests, with the exception of

the data presented in Figure 1, which were analysed using aStudent’s t-test

in marginal intestinal damage Co-treatment with BCA significantlyworsened naproxen-induced intestinal erosions Panel B: ratsco-treated with BCA and naproxen had significantly reduced haema-tocrit compared with rats treated with vehicle and naproxen PanelC: treatment with BCA twice a day did not significantly changeintestinal MPO activity Results are shown as mean± SEM (n ≥ 6 per group) *P < 0.05, **P < 0.01, significantly different from vehicle; unpaired, two-tailed Student’s t-test.

British Journal of Pharmacology (2015) 172 992–1004 995

the PGE2 and TXB2 ELISAkits were purchased from Cayman

Chemical (Ann Arbor, MI, USA) Columbia and BHI agar

media plates were purchased from Becton-Dickinson

(Missis-sauga, ON, CA) DMEM, FBS, penicillin and streptomycin

were purchased from Life Technologies, Inc (Burlington, ON,

CA)

Results

Suppression of H2S synthesis exacerbated

naproxen-induced ulceration and bleeding

Administration of naproxen at 10 mg·kg−1 resulted in a

low level of intestinal damage (Figure 1A) However,

co-administration of an inhibitor of H2S synthesis, BCA, resulted

in a significant increase (P< 0.05) in the severity of

naproxen-induced intestinal damage (Figure 1A) and a small, but

sig-nificant decrease in haematocrit (Figure 1B) Jejunal

granulocyte infiltration (MPO activity) in naproxen-treated

rats was not affected by BCA co-treatment (Figure 1C)

DADS dose-dependently reduced enteropathy

and bleeding

Administration of naproxen (20 mg·kg−1) twice daily for

4.5 days resulted in severe intestinal ulceration and

bleeding (Figure 2A) Rats treated with naproxen exhibited

significant weight loss (∼10%), and blood was evident in the

intestinal lumen Co-administration of DADS with naproxen

resulted in a dose-dependent reduction in the extent of

intestinal damage (Figure 2A) Naproxen treatment resulted

in a 35% decrease in haematocrit (P< 0.001), whereas rats

treated with DADS at doses of 30 or 60 mmol·kg−1 did not

exhibit a significant change in haematocrit (Figure 2B)

Co-administration of DADS (30 or 60 mmol·kg−1) also

signifi-cantly reduced weight loss in naproxen-treated rats (P< 0.01;

Figure 2C)

Effects of DADS on suppression of

COX activity

Naproxen administration profoundly suppressed systemic

COX-1 activity (whole-blood TX synthesis; by 99%)

(Figure 3A) and intestinal PGE2synthesis (by 64%) after twice

daily dosing for 4.5 days A similar degree of suppression of

TX synthesis was observed in rats co-treated with DADS and

naproxen However, naproxen-treated rats that were

co-treated with DADS at 30 or 60 mmol·kg−1 exhibited an

increase (∼1.8-fold) in intestinal PGE2 synthesis, compared

with naproxen-treated rats (P < 0.05) (Figure 3B) Similar

effects were observed in gastric tissue (Supporting

Informa-tion Fig S1) Naproxen significantly inhibited gastric PGE2

synthesis (by 81%), compared with vehicle-treated rats

However, co-treatment with DADS at 30 or 60 mmol·kg−1

resulted in significantly elevated gastric PGE2synthesis,

com-pared with that in naproxen-treated rats (P< 0.05)

DADS dose-dependently reduced

intestinal inflammation

Naproxen administration resulted in a significant

(approxi-mately fourfold) increase in jejunal granulocyte infiltration

Figure 2

Dose-dependent reduction of naproxen-induced intestinal ulceration

by DADS Rats were co-treated, twice daily, with naproxen(20 mg·kg−1) and vehicle or DADS (10, 30, or 60 mmol·kg−1) for 4.5days Panel A: naproxen-induced small intestinal damage was signifi-cantly reduced by co-treatment with DADS at doses of 30 and

60 mmol·kg−1kg−1 Panel B: naproxen administration caused cant bleeding compared with vehicle treatment, but co-treatmentwith DADS at doses of 30 or 60 mmol·kg−1significantly reduced thenaproxen-induced decrease in haematocrit Panel C: weight losscaused by naproxen administration was significantly reduced byco-treatment with DADS at doses of 30 or 60 mmol·kg−1kg−1 Resultsare shown as mean± SEM (n ≥ 6 per group) ***P < 0.001, signifi-

signifi-cantly different from vehicle;ψP< 0.05,ψψP< 0.01, (ψψψP< 0.001,significantly different from naproxen alone; one-wayANOVAfollowed

by Dunnett’s and Bonferroni post hoc tests.

996 British Journal of Pharmacology (2015) 172 992–1004

(MPO activity), compared with vehicle-treated rats (P <

0.001) However, naproxen-induced granulocyte infiltration

was prevented when rats were co-treated with DADS at doses

of 30 or 60 mmol·kg−1(Figure 4A) Histological examination

of jejunal sections from naproxen-treated rats confirmed the

extensive macroscopic erosions and granulocyte infiltration

Naproxen-treated rats exhibited a complete loss of mucosal

architecture, granulocyte infiltration, and extensive

subepi-thelial oedema, compared with vehicle-treated rats (Figure 4B

and C respectively) However, mucosal structure was largely

intact when naproxen-treated rats were co-treated with DADS

at doses of 30 or 60 mmol·kg−1, with similar appearance tovehicle-treated rats (Figure 4D)

Cytotoxic effects of bile were enhanced

by naproxen

When tested at dilutions of 1:6 or 1:12, bile collected fromrats given naproxen (20 mg·kg−1) over 4.5 days exhibited sig-nificantly increased cytotoxic effects on IEC-6 intestinal epi-thelial cells, compared with bile from rats given vehicle.Results for the 1:6 dilutions are shown in Figure 5 Thus,exposure of the cells to a 1:6 dilution of bile from naproxen-treated rats for 3 h resulted in 58% cytotoxicity, compared

with 26% cytotoxicity (P< 0.001) for bile from vehicle-treatedrats (Figure 5) Similar results were also observed for the 1:12bile dilution (data not shown)

DADS reduced naproxen-enhanced bile toxicity

The enhancement of the cytotoxicity of bile by naproxen wasdose-dependently reduced in rats co-treated with DADS(Figure 5) Indeed, bile collected from rats co-treated withnaproxen and DADS at 60 mmol·kg−1 was not significantlydifferent, in terms of cytotoxicity, from the bile collectedfrom rats treated only with vehicle Similar results were alsoobserved for the 1:12 bile dilution and in a series of experi-ments evaluating bile cytotoxicity on cultured HT-29 cells(Supporting Information Fig S2)

The severity of naproxen-induced intestinal damagecorrelates well with the concentrations of naproxen in thebile after administration of naproxen to rats (J L Wallace,unpubl data) To explore whether DADS co-administrationreduced biliary concentrations of naproxen, we measured theconcentrations of naproxen in bile from rats treated withnaproxen (20 mg·kg−1) alone or co-treated with DADS Theconcentration of naproxen in the bile of naproxen-treatedrats did not differ significantly when DADS was co-administered at doses of 10, 30 or 60 mmol·kg−1, suggestingthat DADS co-administration did not significantly alter theenterohepatic recirculation of naproxen (Supporting Infor-mation Fig S3)

DADS administration altered the composition

of the microbiota

We examined whether DADS administration would alter thecomposition of the intestinal microbiota DGGE analysis ofcaecal contents demonstrated that treatment with DADScaused a marked shift in the composition of the microbiota.DGGE analysis was performed to compare the microbial com-position in rats treated with vehicle, naproxen (20 mg·kg−1)plus vehicle, or naproxen plus DADS at a protective(30 mmol·kg−1) and a non-protective (10 mmol·kg−1) dose.Naproxen administration caused caecal dysbiosis in ratsand co-treatment with a non-protective dose of DADS(10 mmol·kg−1) did not correct this shift, as analysed by theDice coefficient and Ward algorithm to determine similarities(Figure 6A) Interestingly, the microbiota of naproxen-treated, rats co-treated with a protective dose of DADS(30 mmol·kg−1), clustered with that of rats not treated withnaproxen Construction of a UPGMA tree demonstrated

Figure 3

DADS did not prevent systemic COX inhibition by naproxen

Nap-roxen administration significantly suppressed (by 99%) whole-blood

synthesis of TXB2 (panel A), and this was not affected by

co-administration of DADS (10, 30 or 60 mmol·kg−1) Three hours

after the final dose, naproxen administration also significantly

inhib-ited (by 64%) intestinal PGE2 synthesis compared with

vehicle-treated rats (panel B) However, co-treatment with DADS at 30 and

60 mmol·kg−1increased intestinal PGE2synthesis in naproxen-treated

rats to levels comparable with vehicle-treated rats Results are shown

as mean± SEM (n ≥ 6 per group) *P < 0.05, **P < 0.01, significantly

different from vehicle; one-wayANOVA followed by Dunnett’s and

Bonferroni post hoc tests.

British Journal of Pharmacology (2015) 172 992–1004 997

similar clustering and each branch had 100% re-sampling

support (Figure 6B) The total number of aerobes in the

jejunum of rats treated with naproxen or naproxen plus

DADS (10, 30 or 60 mmol·kg−1), was not significantly

differ-ent from that in vehicle-treated rats (whether plated on CBA

or BHI media) (Figure 6C)

To further investigate if DADS administration alone could

shift the composition of the microbiota in rats, an additional

experiment was conducted in which rats were treated with

vehicle or DADS (10 or 30 mmol·kg−1) Similar to the results

mentioned earlier, the total number of aerobes and anaerobes

in the jejunum of rats treated with DADS at doses of 10 or

30 mmol·kg−1was not significantly different from that in

vehicle-treated rats (whether plated on CBA or BHI media)

(Supporting Information Fig S3C) DGGE analysis of the

caecal microbiota demonstrated that treatment of rats with

DADS at a non-protective dose (10 mmol·kg−1) did not cause

a shift in the microbiota, compared with treatment with

vehicle However, the microbiota of rats treated with DADS at

a protective dose (30 mmol·kg−1) resulted in a distinct

clus-tering of the microbiota, compared with that in rats treated

with vehicle (Supporting Information Fig S4A and B)

We also analysed the taxonomic composition of the

caecal microbiota via deep sequencing of 16S rRNA

(Illumina, San Diego, CA, USA) Administration of DADS

at 30 mmol.kg−1, but not at 10 mmol.kg−1, significantly

decreased multiple Clostridiales families, such as

Ruminococ-caceae and Eubacteriaceae, and decreased EnterococRuminococ-caceae as

compared with vehicle-treated rats (Supporting InformationFig S5) Interestingly, DADS at the higher dose increased the

abundance of Mucispirillum, a type of bacterium that

colo-nizes the mucus layer of the mammalian tract

Lack of intestinal damage or alteration of bile cytotoxicity with an H2S-releasing naproxen derivative

Twice daily administration of ATB-346 for 4.5 days at a doseequimolar to naproxen at 20 mg·kg−1 did not cause signifi-cant damage to the small intestine of rats (scores of 0 in allsix rats) When HT-29 cells were exposed to a 1:6 or 1:12dilution of bile collected from ATB-346-treated rats, theextent of cell death observed was significantly reduced com-pared with that observed when the cells were exposed tothe same dilution of bile from naproxen-treated rats (Sup-porting Information Fig S6) ATB-346 did not significantlychange the cytotoxicity of bile as compared with vehicletreatment

Figure 4

DADS dose-dependently prevented naproxen-induced mucosal inflammation and structural damage Panel A: naproxen administration cantly increased intestinal MPO activity compared with vehicle-treated rats However, co-treatment with DADS at doses of 30 or 60 mmol·kg−1significantly diminished the naproxen-induced increase in MPO activity Panel B: loss of mucosal structure in the intestine after naproxentreatment Mucosal structure remained intact when naproxen-treated rats were co-administered DADS (30 mmol·kg−1) (panel D), with a similarappearance to tissue from vehicle-treated rats (panel C) Results are shown as mean± SEM (n ≥ 6 per group) ***P < 0.001, significantly different

signifi-from vehicle;ψψψP< 0.001, significantly different from naproxen alone; one-wayANOVAfollowed by Dunnett’s and Bonferroni post hoc tests Scale

bar, 100μm (applicable to each panel)

998 British Journal of Pharmacology (2015) 172 992–1004

NSAID-induced enteropathy is a significant clinical concern

because of the widespread use of this class of drugs,

particu-larly among the elderly, who have an increased propensity to

develop GI ulcers (Wallace, 2013b) In recent years,

consid-erable evidence has been provided for important roles of H2S

as a mediator of mucosal defence throughout the GI tract

Endogenous H2S synthesis is up-regulated at sites of mucosal

injury and contributes significantly to healing of the injury

(Wallace et al., 2007b; Wallace et al., 2009; Flannigan et al.,

2013) Administration of H2S donors has been shown to

accelerate the healing of gastric and colonic ulcers, and to

reduce mucosal inflammation (Fiorucci et al., 2007; Wallace

et al., 2007b) In the present study, a garlic-derived H2S donor

(DADS) was shown to dose-dependently reduce the severity

of ulceration and bleeding in the small intestine following

administration of naproxen, one of the most prescribed

NSAIDs Furthermore, DADS administration led to

signifi-cant changes in the intestinal microbiota and to the

cyto-toxicity of bile that could account, at least in part, for the

protective effects of this H2S donor against

NSAID-enteropathy Suppression of endogenous H2S synthesis

resulted in a significant exacerbation of naproxen-induced

intestinal ulceration and bleeding The intestinal-sparing

effect of DADs and reduced cytotoxicity of bile were also

observed with an H2S-releasing derivative of naproxen 346), which suppressed PG and TX synthesis as effectively asnaproxen Indeed, with a number of H2S-releasing derivatives

(ATB-of NSAIDs, we have consistently observed a greatly reducedcapacity for inducing gastrointestinal damage, despite com-parable suppression of COX-1 and COX-2 activity as seen

with the parent NSAID (Wallace et al., 2007a; 2010; Elsheikh

et al., 2014) These observations support the hypothesis that

the beneficial effects of DADS were attributable to the H2S it

can release (Benavides et al., 2007) Moreover, there does not

appear to be any tachyphylaxis to the beneficial effects of H2S

in the GI tract, as GI safety was observed even after 2 weeks

of daily administration of these compounds (Elsheikh et al.,

2014)

While the mechanism underlying the gastroduodenaldamage caused by NSAIDs is clearly related to the ability ofthese drugs to suppress COX-1 and -2 activity, the mecha-nism for NSAID-enteropathy is less clear and is likely to bemore complex (Wallace, 2012) COX inhibition contributes

to the injury that develops in the intestine following NSAIDadministration, but at least three other interrelated factorsappear to be more important: bile, enteric bacteria and theenterohepatic circulation of the NSAID The latter is clearfrom evidence that NSAIDs that do not undergo enterohe-patic recirculation do not cause significant intestinal damage

(Wax et al., 1970; Reuter et al., 1997; Somasundaram et al.,

1997) After absorption, NSAIDs can be glucuronidated in theliver, and then excreted into bile As shown in the presentstudy, bile containing NSAIDs or NSAID-glucuronides aremore damaging to intestinal epithelial cells than bile fromrats that were not treated with an NSAID This may be due, atleast in part, to NSAID-induced changes in the enteric flora,leading to increased concentrations of more cytotoxic, sec-ondary bile acids (Hofmann, 1999; Martinez-Augustin andSanchez de Medina, 2008) Re-absorption of the NSAIDs inthe ileum can only occur if the NSAID is deconjugated fromthe glucuronide, which requires the activity of bacterialβ-glucuronidase It has recently been demonstrated that aninhibitor of bacterial β-glucuronidase prevented NSAID-

induced intestinal damage in mice (Saitta et al., 2014),

con-sistent with observations that prevention of enterohepatic

circulation of NSAIDs (Wax et al., 1970), or performing

studies in germ-free animals (Robert and Asano, 1977; Uejima

et al., 1996), resulted in prevention of intestinal damage.

Thus, bacteria may contribute to NSAID-enteropathy throughtheir critical role in enterohepatic circulation, as well as in theconversion of primary to secondary bile acids A third mecha-nism through which bacteria can contribute to NSAID-enteropathy is through their colonization of the initial

lesions that form after NSAID administration (Elliott et al.,

1998), and in the case of Gram-negative bacteria, via

activa-tion of Toll-like receptor 4 (TLR-4) Watanabe et al (2008)

demonstrated that activation of TLR-4 contributed cantly to the intestinal injury that developed in mice and ratsafter administration of indomethacin Studies involving pre-treatment with antibiotics have been less informative inestablishing the contribution of enteric bacteria to NSAID-enteropathy, as it is difficult to separate a primary (preventa-tive) effect from a secondary effect such as post-injury

signifi-acceleration of healing (Kent et al., 1969; Yamada et al.,

1993)

Figure 5

DADS dose-dependently reduced naproxen-induced bile

cytotoxic-ity Bile collected from rats treated with naproxen (20 mg·kg−1) twice

daily for 4.5 days was significantly more cytotoxic to cultured rat

IEC-6 cells than bile collected from vehicle-treated rats Co-treatment

with DADS at 30 mmol·kg−1 significantly reduced the

naproxen-induced increase in cytotoxicity of bile Co-treatment with DADS at

60 mmol·kg−1further reduced naproxen-induced bile cytotoxicity, to

a level similar to that of bile from vehicle-treated rats Data shown are

from the 1:6 dilutions of bile samples, and are expressed as the mean

± SEM of at least six rats per group ***P < 0.001, significantly

different from vehicle;ψP< 0.05,ψψψP< 0.001, significantly different

from naproxen alone; one-wayANOVA followed by Dunnett’s and

Bonferroni post hoc tests.

British Journal of Pharmacology (2015) 172 992–1004 999

In the present study, treatment with DADS resulted in a

reduction of the in vitro cytotoxicity of bile to the same level

as in rats not treated with an NSAID, triggered marked

changes in the intestinal microbiota, but did not alter

entero-hepatic recirculation of naproxen (i.e the levels of naproxen

in bile were unaltered by DADS treatment) This suggests that

DADS did not significantly change bacterial deconjugation of

naproxen-glucuronide, a necessary step for the NSAID to be

re-absorbed in the ileum We cannot rule out the possibility

that treatment with DADS may have altered the levels of

naproxen-glucuronide in bile In a previous study, we

observed that the ratio of biliary naproxen-glucuronide :

nap-roxen after administration of an H2S-releasing naproxen

derivative decreased over time with repeated administration

of this compound (Blackler et al., 2012) Concentrations of

naproxen in the bile of rats treated with naproxen

(20μg·kg−1) were∼3 μg·mL−1 This concentration of naproxen

(∼1.2 μM) is well below the concentration required to

signifi-cantly increase the cytotoxicity of the bile in our in vitro assay

(only at concentrations of naproxen of >10 μM was an

increase in the cytotoxicity of bile observed)

Another mechanism through which NSAIDs have beensuggested to cause small intestinal ulceration is through theirability to uncouple oxidative phosphorylation, leading to

death of epithelial cells (Somasundaram et al., 1997) If this is

the case, it raises the intriguing possibility that the protectiveeffects of H2S could be related to the ability of this gaseousmediator to act as an electron donor in mitochondrial respi-

ration (Goubern et al., 2007) H2S has been shown capable ofrescuing mitochondrial function during hypoxia and anoxia,

by virtue of this action, contributing to its cytoprotective

effects in the GI tract and elsewhere (Elrod et al., 2007; Kimura et al., 2010; Campolo et al., 2013) Epithelial cells in

the GI tract have been reported to be the most efficient cells

at using H2S as a mitochondrial energy source (Mimoun et al.,

2012)

Treatment with DADS also resulted in a statistically nificant increase in intestinal PG synthesis As expected, nap-roxen markedly inhibited intestinal PGE2 synthesis andwhole-blood TX synthesis (the latter is almost entirely

sig-derived from platelets) (Wallace et al., 1998) In rats

pre-treated with DADS at the two higher doses, which were

Figure 6

Co-treatment with DADS prevented naproxen-induced dysbiosis Panel A: DGGE analysis revealed that naproxen (20 mg·kg−1) administration torats caused dysbiosis of the caecal microbiota, with distinct clustering from vehicle-treated rats Co-treatment with DADS at 30 mmol·kg−1shiftedthe microbiota of naproxen-treated rats back to being similar to that of vehicle-treated rats Using a resampling technique (majority UPGMAalgorithm), the dendrogram clustering observed in Panel A was confirmed, indicating a robust difference in microbiota composition betweengroups (panel B) Panel C: The total number of aerobes in the jejunum did not significantly differ in rats treated with vehicle, naproxen, ornaproxen plus DADS (10, 30 or 60 mmol·kg−1) twice daily for 4.5 days Results in panel C are from samples plated on CBA and shown as mean

± SEM (n ≤ 5 per group) The data were analysed by a one-wayANOVAfollowed by Dunnett’s multiple comparison test

1000 British Journal of Pharmacology (2015) 172 992–1004

protective against intestinal injury, the levels of PGE2

synthe-sis were significantly greater than in rats pretreated with

vehicle However, the differences in intestinal PGE2synthesis

between the groups treated with ‘protective’ doses of DADS

and the group treated with a non-protective dose of DADS

were negligible, suggesting that altered PGE2 synthesis was

unlikely to have contributed significantly to the protective

effects of DADS

As mentioned earlier, the pathogenesis of

enteropathy is more complicated than that of

NSAID-gastropathy The multifactorial aspect of NSAID-enteropathy

may explain why doses of DADS in the mmol·kg−1range were

required to observe a protective effect against small intestinal

damage, whereas doses of DADS in theμmol·kg−1range were

effective in preventing naproxen-induced gastric damage

(Wallace et al., 2010) However, the ability of an H2S donor to

protect against NSAID-induced intestinal damage may also

depend on the nature of the donor, and the manner in which

H2S is delivered by the donor relative to the delivery and

absorption of the NSAID We observed that ATB-346, an H2

S-releasing derivative of naproxen, at a dose equivalent to a

20 mg·kg−1 dose of naproxen, did not produce significant

small intestinal damage, consistent with previous

observa-tions (Wallace et al., 2010; Blackler et al., 2012) This effect

was seen even though, at this dose, ATB-346 would deliver

less than 1% of the H2S that is released by the protective doses

of DADS Interestingly, as was the case with DADS

co-administration, the cytotoxicity of bile from rats treated

with ATB-346 was significantly reduced compared with bile

from rats treated with naproxen

Changes in the microbiota induced by administration of

DADS may have contributed significantly to the ability of this

H2S donor to reduce the severity of naproxen-induced

intes-tinal damage As outlined earlier, enteric bacteria can

contrib-ute to the pathogenesis of NSAID-enteropathy in several

ways, affecting both the cytotoxicity of bile, the

enterohe-patic circulation of the NSAID and the ability of ulcers to

heal In this study, we have demonstrated that naproxen

administration also caused significant changes in the enteric

microflora, as has been reported previously (Uejima et al.,

1996; Reuter et al., 1997; Hagiwara et al., 2004; Watanabe

et al., 2008) Co-administration of ‘protective’ doses of DADS

with naproxen resulted in a normalization of the microbiota

(i.e to be similar to that in rats not receiving naproxen) Of

course, this latter effect could simply be a consequence of the

prevention of naproxen-induced intestinal damage A causal

relationship between DADS-induced changes in the

micro-biota and reduced intestinal damage has not been

estab-lished However, it is noteworthy that administration of

DADS alone resulted in significant changes in the enteric

microflora, including causing a marked decrease in multiple

Clostridiales families, and a substantial increase in abundance

of Mucispirillum These effects were observed with a dose of

DADS that was effective in reducing naproxen-induced

enter-opathy (30 mmol·kg−1), but not at a dose that was ineffective

in reducing naproxen-induced enteropathy (10 mmol·kg−1)

In summary, the present study extends previous

observa-tions that H2S has protective effects in the GI tract, with the

demonstration that DADS could substantially reduce the

severity of intestinal ulceration and bleeding induced by

repeated administration of an NSAID The pathogenesis of

NSAID-enteropathy is complicated, involving cytotoxiceffects of bile, changes in intestinal microbiota and entero-hepatic circulation of the NSAID The present study providesevidence that administration of an H2S donor can signifi-cantly affect two of these factors: reducing the cytotoxicity ofbile and significantly altering the enteric microbiota Thesetwo effects may be related, as changes in enteric bacteria canlead to altered bile metabolism, and in turn to an alteration ofcytotoxicity

Acknowledgments

This work was supported by a grant from the Canadian tutes of Health Research The authors are grateful to WebbMcKnight for technical assistance

Insti-Author contributions

R W B., J P M., A M and M W performed the experiments;

R W B., J P M., P B and J L W designed the experiments;

R W B., J P M., A M., M W., P B., M G S and J L W.analysed the data; and R W B., M W and J L W wrote thepaper

Conflict of interest

Dr Wallace is Founder and one of the Directors of AntibeTherapeutics, Inc., a company developing novel anti-inflammatory drugs

References

Alexander SPH, Benson HE, Faccenda E, Pawson AJ, Sharman JL,

Spedding M et al (2013) The Concise Guide to PHARMACOLOGY

2013/14: Enzymes Br J Pharmacol 170: 1797–1867

Benavides GA, Squadrito GL, Mills RW, Patel HD, Isbell S, Patel RP

et al (2007) Hydrogen sulfide mediates the vasoactivity of garlic.

Proc Natl Acad Sci U S A 104: 17977–17982

Bjarnason I, Hayllar J, MacPherson AJ, Russell AS (1993) Sideeffects of nonsteroidal anti-inflammatory drugs on the small andlarge intestine in humans Gastroenterology 104: 1832–1847.Blackler R, Syer S, Bolla M, Ongini E, Wallace JL (2012)

Gastrointestinal-sparing effects of novel NSAIDs in rats withcompromised mucosal defence PLoS ONE 7: e35196

Boughton-Smith NK, Wallace JL, Whittle BJ (1988) Relationshipbetween arachidonic acid metabolism, myeloperoxidase activityand leukocyte infiltration in a rat model of inflammatory boweldisease Agents Actions 25: 115–123

Campolo M, Esposito E, Ahmad A, Di Paola R, Wallace JL,Cuzzocrea S (2013) A hydrogen sulfide-releasing cyclooxygenaseinhibitor markedly accelerates recovery from experimental spinalcord injury FASEB J 27: 4489–4499

British Journal of Pharmacology (2015) 172 992–1004 1001

Dietschy JM (1968) Mechanisms for the intestinal absorption of

bile acids J Lipid Res 9: 297–309

Elliott SN, Buret A, McKnight W, Miller MJ, Wallace JL (1998)

Bacteria rapidly colonize and modulate healing of gastric ulcers in

rats Am J Physiol 275: G425–G432

Elrod JW, Calvert JW, Morrison J, Doeller JE, Kraus DW, Tao L et al.

(2007) Hydrogen sulfide attenuates myocardial

ischemia-reperfusion injury by preservation of mitochondrial

function Proc Natl Acad Sci USA 104: 15560–15565

Elsheikh W, Blackler RW, Flannigan KL, Wallace JL (2014)

Enhanced chemopreventative effects of a hydrogen sulfide-releasing

NSAID (ATB-346) in experimental colorectal cancer Nitric Oxide

41: 131–137

Fiorucci S, Antonelli E, Distrutti E, Rizzo G, Medcarelli A, Orlandi S

et al (2005) Inhibition of hydrogen sulfide generation contributes

to gastric injury caused by anti-inflammatory nonsteroidal drugs

Gastroenterology 129: 1210–1224

Fiorucci S, Orlandi S, Mencarelli A, Caliendo G, Santagada V,

Distrutti E et al (2007) Enhanced activity of a hydrogen

sulphide-releasing derivative of mesalamine (ATB-429) in a mouse

model of colitis Br J Pharmacol 150: 996–1002

Flannigan KL, Ferraz JG, Wang R, Wallace JL (2013) Enhanced

synthesis and diminished degradation of hydrogen sulfide in

experimental colitis: a site-specific, pro-resolution mechanism PLoS

ONE 8: e71962

Goubern M, Andriamihaja M, Nubel T, Blachier F, Bouillaud F

(2007) Sulfide, the first inorganic substrate for human cells FASEB

J 21: 1699–1706

Hagiwara M, Kataoka K, Arimochi H, Kuwahara T, Ohnishi Y

(2004) Role of unbalanced growth of gram-negative bacteria in

ileal ulcer formation in rats treated with a nonsteroidal

anti-inflammatory drug J Med Invest 51: 43–51

Hofmann AF (1999) The continuing importance of bile acids in

liver and intestinal disease Arch Intern Med 159: 2647–2658

Jobin C, Morteau O, Hans DS, Sartor RB (1998) Specific NF-kappaB

blockade selectively inhibits tumour necrosis factor-alpha-induced

COX-2 but not constitutive COX-1 gene expression in HT-29 cells

Immunology 95: 537–543

Kawabata A, Ishiki T, Nagasawa K, Yoshida S, Maeda Y, Takahashi T

et al (2007) Hydrogen sulfide as a novel nociceptive messenger.

Pain 132: 74–81

Kent TH, Cardelli RM, Stamler FW (1969) Small intestinal

ulcersand intestinal flora in rats given indomethacin Am J Pathol

54: 237–249

Kimura Y, Goto Y, Kimura H (2010) Hydrogen sulfide increases

glutathione production and suppresses oxidative stress in

mitochondria Antioxid Redox Signal 12: 1–13

Kilkenny C, Browne W, Cuthill IC, Emerson M, Altman DG (2010)

Animal research: reporting in vivo experiments: the ARRIVE

guidelines Br J Pharmacol 160: 1577–1579

Lanas A, Garcia-Rodriguez LA, Polo-Tomas M, Ponce M,

Alonso-Abreu I, Perez-Aisa MA et al (2009) Time trends and impact

of upper and lower gastrointestinal bleeding and perforation in

clinical practice Am J Gastroenterol 104: 1633–1641

Mard SA, Neisi N, Solgi G, Hassanpour M, Darbor M, Maleki M

(2012) Gastroprotective effect of NaHS against mucosal lesions

induced by ischemia-reperfusion injury in rat Dig Dis Sci 57:

1496–1503

Martinez-Augustin O, Sanchez de Medina F (2008) Intestinal bileacid physiology and pathophysiology World J Gastroenterol 14:5630–5640

McGrath J, Drummond G, McLachlan E, Kilkenny C, Wainwright C(2010) Guidelines for reporting experiments involving animals: theARRIVE guidelines Br J Pharmacol 160: 1573–1576

Mimoun S, Andriamihaja M, Chaumontet C, Atanasiu C,

Benamouzig R, Blouin JM et al (2012) Detoxification of H2S bydifferentiated colonic epithelial cells: implication of the sulfideoxidizing unit and of the cell respiratory capacity Antioxid RedoxSignal 17: 1–10

Park AJ, Collins J, Blennerhassett PA, Ghia JE, Verdu EF, Bercik P

et al (2013) Altered colonic function and microbiota profile in a

mouse model of chronic depression Neurogastroenterol Motil 25:733–e575

Pawson AJ, Sharman JL, Benson HE, Faccenda E, Alexander SP,

Buneman OP et al.; NC-IUPHAR (2014) The IUPHAR/BPS Guide to

PHARMACOLOGY: an expert-driven knowledge base of drug targetsand their ligands Nucl Acids Res 42 (Database Issue): D1098–1106.Quaroni A, Wands J, Trelstad RL, Isselbacher KJ (1979) Epithelioidcell cultures from rat small intestine Characterization bymorphologic and immunologic criteria J Cell Biol 80: 248–265.Reuter BK, Davies NM, Wallace JL (1997) Nonsteroidalanti-inflammatory drug enteropathy in rats: role of permeability,bacteria, and enterohepatic circulation Gastroenterology 112:109–117

Robert A, Asano T (1977) Resistance of germfree rats toindomethacin-induced intestinal lesions Prostaglandins 14:333–341

Saitta KS, Zhang C, Lee KK, Fujimoto K, Redinbo MR, Boelsterli UA(2014) Bacterialβ-glucuronidase inhibition protects mice againstenteropathy induced by indomethacin, ketoprofen or diclofenac:mode of action and pharmacokinetics Xenobiotica 44: 28–35.Satoh H, Amagase K, Takeuchi K (2012) Exacerbation ofnonsteroidal anti-inflammatory drug-induced small intestinallesions by antisecretory drugs in rats: the role of intestinal motility

J Pharmacol Exp Ther 343: 270–277

Scheiman JM, Yeomans ND, Talley NJ, Vakil N, Chan FK, Tulassay Z(2006) Prevention of ulcers by esomeprazole in at-risk patientsusing non-selective NSAIDs and COX-2 inhibitors Am JGastroenterol 101: 701–710

Seitz S, Boelsterli UA (1998) Diclofenac acyl glucuronide, a majorbiliary metabolite, is directly involved in small intestinal injury inrats Gastroenterology 115: 1476–1482

Somasundaram S, Rafi S, Hayllar J, Sigthorsson G, Jacob M, Price AB

et al (1997) Mitochondrial damage: a possible mechanism of the

‘topical’ phase of NSAID induced injury to the rat intestine Gut 41:344–353

Uejima M, Kinouchi T, Kataoka K, Hiraoka I, Ohnishi Y (1996).Role of intestinal bacteria in ileal ulcer formation in rats treatedwith a nonsteroidal antiinflammatory drug Microbiol Immunol 40:553–560

Wallace JL (2008) Prostaglandins, NSAIDs, and gastric mucosalprotection; why doesn’t the stomach digest itself? Physiol Rev 88:1547–1565

Wallace JL (2010) Physiological and pathophysiological roles ofhydrogen sulfide in the gastrointestinal tract Antioxid Redox Signal12: 1125–1133

Wallace JL (2012) NSAID gastropathy and enteropathy: distinctpathogenesis likely necessitates distinct prevention strategies Br JPharmacol 165: 67–74

1002 British Journal of Pharmacology (2015) 172 992–1004

Wallace JL (2013a) Mechanisms, prevention and clinical

implications of nonsteroidal anti-inflammatoty drug-enteropathy

World J Gastroenterol 19: 1861–1876

Wallace JL (2013b) Polypharmacy of osteoarthritis: the perfect

intestinal storm Dig Dis Sci 58: 3088–3093

Wallace JL, Keenan CM, Granger DN (1990) Gastric ulceration

induced by nonsteroidal anti-inflammatory drugs is a

neutrophil-dependent process Am J Physiol 259: 462–467

Wallace JL, Bak A, McKnight W, Asfaha S, Sharkey KA,

MacNaughton WK (1998) Cyclooxygenase 1 contributes to

inflammatory responses in rats and mice: implications for

gastrointestinal toxicity Gastroenterology 115: 101–109

Wallace JL, McKnight W, Reuter BK, Vergnolle N (2000)

NSAID-induced gastric damage in rats: requirement for inhibition

of both cyclooxygenase 1 and 2 Gastroenterology 119: 706–714

Wallace JL, Caliendo G, Santagada V, Cirino G, Fiorucci S (2007a)

Gastrointestinal safety and anti-inflammatory effects of a hydrogen

sulfide-releasing diclofenac derivative in the rat Gastroenterology

132: 261–271

Wallace JL, Dicay M, McKnight W, Martin GR (2007b) Hydrogen

sulfide enhances ulcer healing FASEB J 21: 4070–4076

Wallace JL, Vong L, McKnight W, Dicay M, Martin GR (2009)

Endogenous and exogenous hydrogen sulfide promotes resolution

of colitis in rats Gastroenterology 137: 569–578

Wallace JL, Caliendo G, Santagada V, Cirino G (2010) Markedly

reduced toxicity of a hydrogen sulphide-releasing derivative of

naproxen (ATB-346) Br J Pharmacol 159: 1236–1246

Wallace JL, Syer S, Denou E, de Palma G, Vong L, McKnight W

et al (2011) Proton pump inhibitors exacerbate NSAID-induced

small intestinal injury by inducing dysbiosis Gastroenterology 141:

1314–1322

Wallace JL, Ferraz JG, Muscara MN (2012) Hydrogen sulfide: an

endogenous mediator of resolution of inflammation and injury

Antioxid Redox Signal 17: 58–67

Wallace JL, Blackler RW, Chan MV, Da Silva GJ, Elsheikh W,

Flannigan KL et al (2014) Anti-inflammatory and cytoprotective

actions of hydrogen sulfide: translation to theapeutics Antioxid

Redox Signal Epub ahead of print

Watanabe T, Higuchi K, Kobata A, Nishio H, Tanigawa T, Shiba M

et al (2008) Non-steroidal anti-inflammatory drug-induced small

intestinal damage is Toll-like receptor 4 dependent Gut 57:

181–187

Watanabe T, Tanigawa T, Nadatani Y, Nagami Y, Sugimori S,

Okazaki H et al (2013) Risk factors for severe nonsteroidal

anti-inflammatory drug-induced small intestinal damage Dig Liver

Dis 45: 390–395

Wax J, Clinger WA, Varner P, Bass P, Winder CV (1970)

Relationship of the enterohepatic cycle to ulcerogenesis in the rat

small bowel with flufenamic acid Gastroenterology 58: 772–780

Yamada T, Deitch E, Specian RD, Perry MA, Sartor RB, Grisham MB

(1993) Mechanisms of acute and chronic intestinal inflammation

induced by indomethacin Inflammation 17: 641–662

Zanardo RC, Brancaleone V, Distrutti E, Fiorucci S, Cirino G,

Wallace JL (2006) Hydrogen sulfide is an endogenous modulator of

leukocyte-mediated inflammation FASEB J 20: 2118–2120

Zhao Y, Encinosa W (2008) Hospitalizations for gastrointestinal

bleeding in 1998 and 2006: Statistical brief #65 Health Care Cost

and Utilization Project (HCUP) Statistical Briefs Agency for Health

Care Policy and Research

Supporting information

Additional Supporting Information may be found in theonline version of this article at the publisher’s web-site:http://dx.doi.org/10.1111/bph.12961

Figure S1 Effects of DADS on inhibition of gastric PGE2

synthesis by naproxen Three hours after the final dose,naproxen significantly suppressed gastric PGE2 synthesis

(***P < 0.001) However, co-treatment with DADS at 30 or

60 mmol·kg−1significantly increased gastric PGE2synthesis innaproxen-treated rats (ψP< 0.05) Results are shown as mean

± SEM (n ≥ 6 per group) The data were analysed by a one-way

ANOVAfollowed by Dunnett’s and Bonferroni post hoc tests.

Figure S2 DADS dose-dependently reduced induced bile cytotoxicity on HT-29 cultured cells Similar tothe bile cytotoxicity results from IEC-6 cell cultures, bilecollected from rats treated with naproxen (20 mg·kg−1) twicedaily for 4.5 days was significantly more cytotoxic than bile

naproxen-collected from vehicle-treated rats (***P < 0.001) Bile lected from rats co-treated with DADS at 60 mmol·kg−1wassignificantly less cytotoxic than bile from naproxen-treated

col-rats (*P< 0.05) The bile samples were diluted 1:6 in co’s PBS (DPBS) prior to being added to the cultured cells.Results are shown as mean± SEM (n = 6 per group) The data

Dulbec-were analysed by a one-way ANOVA followed by Dunnett’smultiple comparison test

Figure S3Biliary concentrations of naproxen were unaltered

by co-administration of DADS Naproxen (20 mg·kg−1) wasco-administered with vehicle or with DADS (10, 30 or

60 mmol·kg−1) twice daily for 4.5 days Bile was collected 1 hafter the final drug administration Biliary naproxen concen-trations were measured by liquid chromatography/mass spec-trometry Results are shown as mean± SEM (n = 6 per group).

The data were analysed by a one-way ANOVA followed by

Dunnett’s and Bonferroni post hoc tests.

Figure S4Treatment with DADS caused a shift in the biota composition Panel A: DGGE analysis revealed thatthere were significant differences in the composition ofcaecal microbiota in vehicle-treated and DADS-treated(30 mmol·kg−1) rats The clustering observed in the dendro-gram constructed using the Dice coefficient and Ward algo-rithm in panel A was confirmed using majority UPGMAalgorithm (panel B) Panel C: the total number of aerobes andanaerobes in the jejunum did not significantly differ in ratstreated with vehicle or with DADS (10 or 30 mmol·kg−1) twicedaily for 4.5 days Results in panel C are from samples plated

micro-on CBA and shown as mean± SEM (n ≤ 5 per group) The data

were analysed by a one-way ANOVA followed by Dunnett’smultiple comparison test

(30 mmol·kg−1) and a non-protective dose (10 mmol·kg−1) onthe rat microbiota were examined The taxonomic composi-tion of the caecal microbiota from rats treated with DADS orvehicle was evaluated via deep sequencing of 16S rRNA withIllumina Data are shown as the mean± SEM of five rats per

group *P< 0.05 versus the vehicle-treated group (one-way

ANOVAfollowed by Bonferroni test)

Figure S6ATB-346 administration did not enhance the toxic effects of bile on human intestinal epithelial (HT-29)

cyto-British Journal of Pharmacology (2015) 172 992–1004 1003

cells (n= 12 per group) Bile collected from rats treated with

naproxen (10 mg kg−1) twice daily for 4 days resulted in a

significant increase in cytotoxicity as compared with the

effects of bile from vehicle-treated rats (***P < 0.001) In

contrast, bile from rats treated with an equimolar dose of

ATB-346 exhibited significantly less cytotoxicity than bile

from naproxen-treated rats (ψψP< 0.01), and not significantly

different from bile from vehicle-treated rats Results are fromthe 1:6 and 1:12 dilutions and shown as mean± SEM Thedata were analysed by one-way ANOVA and Bonferroni test.ATB-346 is 2-(6-methoxy-napthalen-2-yl)-propionic acid4-thiocarbamoyl-phenyl ester, a H2S-releasing derivative ofnaproxen

1004 British Journal of Pharmacology (2015) 172 992–1004

M Sarwar1, C S Samuel2, R A Bathgate3, D R Stewart4 and R J Summers1,2

1Drug Discovery Biology, Monash Institute of Pharmacology,2Department of Pharmacology,

Monash University, Melbourne, Vic., Australia,3The Florey Institute of Neuroscience and Mental

Health, Melbourne, Vic., Australia, and4Corthera Inc., San Mateo, CA, USA

Correspondence

Professor Roger Summers,Monash Institute ofPharmaceutical Sciences, MonashUniversity, 399 Royal Parade,Parkville, Melbourne, Vic 3052,Australia E-mail:

BACKGROUND AND PURPOSE

In a recently conducted phase III clinical trial, RELAX-AHF, serelaxin infusion over 48 h improved short- and long-term clinicaloutcomes in patients with acute heart failure In this study we used human primary cells from the umbilical vasculature tobetter understand the signalling mechanisms activated by serelaxin

accumulation and pERK1/2, and the concentration–response curves (CRCs) were bell-shaped Similar bell-shaped CRCs forcGMP and pERK1/2 were observed in HCFs, whereas in HUASMCs, serelaxin increased cAMP, cGMP and pERK1/2 with

accumulation in HUVSMC but not HUASMC Longer term serelaxin exposure increased the expression of neuronal NOS,

CONCLUSIONS AND IMPLICATIONS

Serelaxin caused acute and chronic changes in human umbilical vascular cells that were cell background dependent

Abbreviations

AHF, acute heart failure; CRC, concentration–response curve; DEA, diethylamine NONOate; eNOS, endothelial NOS;

cell; HUASMC, human umbilical artery smooth muscle cell; HUVSMC, human umbilical vein smooth muscle cell;iNOS, inducible NOS; nNOS, neuronal NOS; pERK1/2, phosphorylated ERK 1 and 2; PI3K, phosphoinositide 3-kinase;PTX, pertussis toxin; RXFP1 receptor, relaxin family peptide receptor 1

British Journal of Pharmacology (2015) 172 1005–1019 1005

© 2014 The British Pharmacological Society

Acute heart failure (AHF) is a major global health challenge

with high morbidity and mortality that represents a great

burden on health care (Mosterd and Hoes, 2007) Along with

predictions of increasing prevalence, treatment options for

AHF have changed little over the last two decades and

con-sequently patients continue to experience high morbidity

and mortality However, in the recent phase III clinical trial

(RELAX-AHF), serelaxin (the recombinant form of human

relaxin-2) produced a moderate improvement in one of the

primary end points, dyspnoea, but also significantly reduced

patient mortality at day 180 without any notable side effects

(Teerlink et al., 2013) Further analysis of the RELAX-AHF

findings showed fewer signs of cardiac, renal and liver

damage with early administration of serelaxin, which may

contribute to the long-term survival benefit (Metra et al.,

2013)

Relaxin is a hormone that mediates cardiovascular

adap-tations observed during pregnancy and in particular systemic

and renal vasodilatation (Conrad, 2010), although these

effects are also observed in non-pregnant animals, ex vivo

studies and in animal models of cardiovascular disease

(Masini et al., 1997; 2006; Bani et al., 1998b) There are two

distinct actions of serelaxin that have been described in in

vitro and in vivo studies Rapid serelaxin-mediated responses,

observed after stimulation of serelaxin for minutes to hours

(<1 h), occur via a Gαi/PI3K/cAMP/Akt/eNOS-dependent

mechanism in human subcutaneous and rodent renal and

mesenteric arteries and also in human coronary artery and

aortic endothelial cells (McGuane et al., 2011b) Sustained

serelaxin responses, observed after stimulation of serelaxin

for days (24–48 h), are seen in rodent small renal and human

subcutaneous arteries involving MMPs (gelatinases),

endothelin receptor B (ETB receptor), VEGF and NOS

(Jeyabalan et al., 2003; McGuane et al., 2011a).

Although in vitro and in vivo studies support the potential

benefits of serelaxin in humans in cardiovascular disease,

there are knowledge gaps in our understanding of the nism of action There is little information on the cells tar-geted by serelaxin and on signal transduction mechanisms intissues relevant to the human cardiovascular system thatendogenously express the RXFP1 receptor, the cognate sere-laxin receptor However, it is clear that serelaxin affects thetone of blood vessels In rats, it was recently reported that theRXFP1 receptor is localized to endothelial and smooth musclecells, although there are marked regional variations in distri-

mecha-bution (Jelinic et al., 2014) However, very little is known on

the expression of the RXFP1 receptor and the signalling ways it activates in human arteries and veins In rats, theeffects of serelaxin on arteries have been described in detail

path-(Jeyabalan et al., 2003; Conrad et al., 2004; Conrad and

Shroff, 2011) but much less is known of effects in veins Inaddition, rat mesenteric arteries and veins both expressRXFP1 receptors, yet only the arteries show serelaxin-

mediated vascular remodelling (Jelinic et al., 2014).

The effects of serelaxin on arteries and veins could becritical for the understanding of the clinical actions of sere-laxin because nitrates, a classical therapy for heart failure,reduce congestion by causing venodilatation Serelaxin, aknown arteriodilator, also reduced congestion without

causing hypotension in RELAX-AHF (Teerlink et al., 2013),

suggesting that serelaxin could potentially have additionalvenodilator properties Therefore, we examined whether sere-laxin targets cells in both the arterial and venous vasculature

to activate vasodilator signal transduction mechanisms

In order to address these knowledge gaps, we examinedsignal transduction mechanisms in primary cells from thehuman arterial and venous umbilical vasculature and heart,including endothelial cells, smooth muscle cells and cardiacfibroblasts These were examined for RXFP1 receptor expres-sion and markers of cardiovascular function and disease:short-term effects of serelaxin (<1 h) on cAMP, cGMP andpERK1/2; and the longer term effects of serelaxin (1–48 h) onthe expression of VEGF, ETB receptors, NOS isoforms andMMP activity

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 (Pawson et al., 2014) and are

permanently archived in the Concise Guide to PHARMACOLOGY 2013/14 (a,b,c Alexander et al., 2013a,b,c).

1006 British Journal of Pharmacology (2015) 172 1005–1019

Cell culture

Primary cultures of human umbilical artery endothelial cells

(HUAECs), HUVECs, human umbilical artery smooth muscle

cells (HUASMCs), human umbilical vein smooth muscle cells

(HUVSMCs) and fetal human cardiac fibroblasts (HCFs:

pooled from fetal atria and ventricles) were obtained from

ScienCell Research Laboratories (San Diego, CA, USA)

Endothelial cells were characterized by their expression of

von Willebrand factor (Factor VIII), CD31 and by uptake of

acetylated low-density lipoprotein (DiI-Ac-LDL) Smooth

muscle cells were characterized by the presence ofα-smooth

muscle actin and desmin and fetal HCF were characterized by

the presence of fibronectin (ScienCell) Cells were maintained

in Medium 199 containing 5% FBS, penicillin (100 u·mL−1),

streptomycin (100μg·mL−1) and the relevant growth

supple-ments for optimal growth of each cell type (ScienCell,)

Endothelial cells were grown in endothelial cell growth

sup-plement [BSA 10μg·mL−1, apo-transferrin 10μg·mL−1, insulin

5μg·mL−1, EGF 10 ng·mL−1, fibroblast growth factor-2 (FGF-2)

2 ng·mL−1, VEGF 2 ng·mL−1, insulin-like growth factor-1

(IGF-1) 2 ng·mL−1, hydrocortisone 1μg·mL−1, retinoic acid

10−7M], smooth muscle cells in smooth muscle cell growth

supplement (BSA 10μg·mL−1, apo-transferrin 10μg·mL−1,

insulin 5μg·mL−1, FGF-2 2 ng·mL−1, IGF-1 2 ng·mL−1,

hydro-cortisone 1μg·mL−1) and fibroblasts in fibroblast cell growth

supplement-2 (BSA 10μg·mL−1, apo-transferrin 10μg·mL−1,

insulin 7.5μg·mL−1, EGF 2 ng·mL−1, FGF-2 2 ng·mL−1,

hydro-cortisone 1μg·mL−1) (information provided by ScienCell)

Early passage cultures (2–5) were used for all experiments

Reverse-transcription real-time

PCR (RT-qPCR)

RNA was extracted using the TRIzol® reagent (Invitrogen,

Mulgrave, Victoria, Australia) and cDNA synthesized using

the iScript cDNA synthesis kit (BioRad, Gladesville, NSW,

Australia) according to the manufacturer’s instructions

RT-PCR was performed using the Taqman Assay for RXFP1

receptors (Invitrogen) according to the manufacturer’s

instructions

Radioligand binding

Serelaxin was iodinated with Na125I (Perkin-Elmer, Glen

Waverley, Victoria, Australia) using the chloramine-T method

as described previously (van der Westhuizen et al., 2010).

Whole-cell competition-binding assays were performed as

described previously (Halls et al., 2005) Briefly, cells (0.8–1×

106 cells) with [125I]-serelaxin (200 pM) were allowed to

compete with unlabelled serelaxin (10 pM–0.1μM) for

90 min at room temperature Total binding was determined

by radioligand alone whereas non-specific binding was

deter-mined by 0.1μM unlabelled serelaxin

cAMP and cGMP accumulation assays

cAMP accumulation was determined as previously described

(Halls et al., 2006) Briefly, cells were plated into 24-well

plates (1× 105cells per well) and grown overnight to achieve

a confluent monolayer Prior to stimulation, cells were serum

starved in M199 medium for 4 h Where appropriate, the cells

were pre-incubated with the PI3K inhibitor wortmannin

(100 nM, 30 min), the Gαi/oinhibitor PTX (50 ng·mL−1, 18 h),the Gαsinhibitor NF449 (10μM, 30 min), the Gαi/oinhibitorNF023 (10μM, 30 min), suramin (10 μM, 30 min) or filipinIII (1μg·mL−1, 1 h) Levels of cAMP and cGMP were detectedaccording to the manufacturer’s instructions (Perkin-Elmer)

pERK1/2 assay

pERK1/2 was measured using the Surefire ERK kit (TGR Sciences, Hindmarsh, South Australia, Australia) as described

Bio-previously (van der Westhuizen et al., 2007) Briefly, cells

were plated into 24-well plates (1 × 105 cells per well) andgrown overnight to achieve a confluent monolayer Prior tostimulation, cells were serum starved in M199 medium for

4 h Where appropriate, the cells were pre-incubated withwortmannin (100 nM, 30 min) or PTX (50 ng·mL−1, 18 h).Levels of pERK1/2 were detected according to the manufac-turer’s instructions (Perkin-Elmer)

Gelatin zymography

Gelatin zymography was performed as described previously

to determine changes in levels of MMP2 (gelatinase A) andMMP9 (gelatinase B) in cultured medium samples (Chow

et al., 2012).

Western blotting

Western blotting was performed as described previously

(Chow et al., 2012) Briefly, 20–30μg of protein was separated

on a polyacrylamide gel, transferred to a PVDF membraneand probed with the following primary antibodies: anti-eNOSantibody (R&D Systems; 1:1000, overnight at 4°C), anti-nNOS antibody (R&D Systems; 1:500, overnight at 4°C), anti-iNOS antibody (Cell Signalling, Beverley, MA, USA; 1:1000,overnight at 4°C), anti-VEGF antibody (Abcam, Cambridge,

MA, USA; 1:1000, 2 h at room temperature), anti-ETBreceptorantibody (Pierce, Rockford, IL, USA; 1:1000, overnight at4°C), anti-β-actin antibody (Cell Signalling; 1:1000, 2 h atroom temperature) Membranes were then probed with sec-ondary antibodies (Alexa Fluor 647 anti-rabbit IgG antibody,Alexa Fluor 647 anti-mouse IgG antibody; 2μg·mL−1; Invitro-gen) and scanned using the Typhoon Trio (GE Healthcare,Melbourne, Victoria, Australia), and densitometry was con-ducted on each band using the ImageJ software (NIH,Bethesda, MD, USA)

ANOVAwith a Dunnett’s post hoc test for each cell type studied and statistical significance accepted at P< 0.05

filipin III and suramin were purchased from Sigma (Castle

Hill, NSW, Australia) NF023 and NF449 were purchased from

Calbiochem (Alexandria, NSW, Australia) TGF-β1 was

pur-chased from R&D Systems (Gymea, NSW, Australia)

Results

Cell surface RXFP1 receptors expression

occurs in HUVECs, HUVSMCs, HUASMCs

and HCFs, but not HUAECs

RXFP1 receptor mRNA, measured by qPCR, was present in

HUAECs, HUVECs, HUVSMCs, HUASMCs, HCFs and human

testis (positive control; Figure 1A) Cell surface RXFP1

recep-tor expression, measured by competition binding of [125

I]-serelaxin with unlabelled serelaxin, was detected in

HUASMCs, HUVECs, HUVSMCs and HCFs, but not in

HUAECs (Figure 1B) Binding affinity correlated well with

that observed in HEK cells recombinantly expressing RXFP1

receptors (Supporting Information Table S1) The lack of cellsurface expression of RXFP1 receptors in HUAECs was sup-ported by the failure of serelaxin to cause cAMP accumulation(Supporting Information Fig S1a), cGMP accumulation (Sup-porting Information Fig S1b) or pERK1/2 (Supporting Infor-mation Fig S1c) in these cells

cAMP accumulation in response to acute serelaxin administration

Serelaxin increases cAMP in vitro (Halls et al., 2009b) and

stimulates inotropy in human atrial myocardium (Dschietzig

et al., 2011) In this study, serelaxin (30 nM) increased cAMP

accumulation in HUVECs, HUVSMCs and HUASMCs with theresponse peaking by 30 min (Supporting InformationFig S2a) In HUVECs (Figure 2A) and HUVSMCs (Figure 2B),serelaxin (30 min) concentration- dependently increasedcAMP accumulation to 5% of the forskolin response respec-tively CRCs for cAMP accumulation in both cell types werebell-shaped with pEC50s of 9.1± 0.4 and 9.6 ± 0.4, respectively,for the initial part of the curve As previous studies haveshown that serelaxin-mediated cAMP accumulation in HEK-RXFP1 receptor cells involves Gαi proteins and PI3K (Halls

et al., 2009b), we investigated the effect of PTX (Gαi/otor) and wortmannin (PI3K inhibitor) on the cAMP response.Both PTX (50 ng·mL−1, 18 h) and wortmannin (100 nM,

inhibi-30 min) significantly inhibited serelaxin-mediated (inhibi-30 nM)cAMP accumulation in HUVECs (Figure 2A) and HUVSMCs(Figure 2B) In HUASMCs, however, serelaxin (30 min)increased cAMP accumulation but the response was unaf-fected by PTX (50 ng·mL−1, 18 h) or wortmannin (100 nM,

30 min) pretreatment, suggesting that the response nantly involved Gαs(Figure 2C) Most interestingly, and incontrast to HUVECs and HUVSMCs, the CRC to serelaxin inHUASMCs was sigmoidal with a pEC50of 9.0± 0.3 In HCFs,serelaxin failed to cause cAMP accumulation (Figure 2D),reflecting previous findings in rat atrial and ventricular fibro-

predomi-blasts (Samuel et al., 2004; Mosterd and Hoes, 2007).

cGMP accumulation in response to acute serelaxin stimulation

Serelaxin-mediated vasodilatation, in vitro (Bani et al., 1998a) and ex vivo (Jeyabalan et al., 2003; McGuane et al., 2011a,b),

occurs via a NOS/NO/cGMP-dependent mechanism Inhuman vascular cells, serelaxin (30 nM) time dependentlyincreased cGMP accumulation in HUVECs, HUVSMCs,HUASMCs and HCFs with the response peaking by 30 min(Supporting Information Fig S2b) Serelaxin increased cGMPaccumulation concentration-dependently in HUVECs(Figure 3A: pEC50= 8.9 ± 0.4), HUVSMCs (Figure 3B: pEC50=9.2± 0.4), HUASMCs (Figure 3C: pEC50= 9.1 ± 0.3) and HCFs(Figure 3D: pEC50 = 9.1 ± 0.3) The maximum serelaxinresponse was higher in HUVECs and HCFs (75 and 60% ofDEA) but lower in HUASMCs (40% of DEA) and HUVSMCs(25% of DEA) PTX (50 ng·mL−1, 18 h) and wortmannin(100 nM, 30 min) pretreatment significantly inhibitedserelaxin-mediated (30 nM) cGMP accumulation, confirmingthe involvement of Gαiand PI3K, but these responses werequalitatively different in different cell types In HUVECs,HUVSMCs and HCFs, serelaxin produced bell-shaped CRCswhereas in HUASMCs, as observed for cAMP, it was sigmoidal

Figure 1

The expression of RXFP1 and RXFP2 receptor mRNA and RXFP1

receptor protein in human primary umbilical vascular cells and

human primary cardiac fibroblasts qPCR (A) was utilized to show

expression levels of RXFP1 and RXFP2 receptor mRNA in HUAECs,

HUVECs, HUASMCs, HUVSMCs and HCFs relative toβ-actin (n = 2).

RXFP2 receptor mRNA was only measureable in the positive control

Cell surface RXFP1 receptor protein expression was determined by

radioligand binding (B) utilizing [125I]-serelaxin and showed specific

serelaxin binding in HEK-RXFP1 cells (n = 6), HUASMCs (n = 4),

HUVECs (n = 4), HUVSMCs (n = 3) and HCFs (n = 3), but not in

HUAECs (n= 2)

1008 British Journal of Pharmacology (2015) 172 1005–1019

ERK1/2 phosphorylation in response to acute

serelaxin stimulation

ERK1/2, a kinase that promotes cell survival and inhibits

apoptosis, is phosphorylated following treatment with

sere-laxin in HUVECs and vascular smooth muscle cells (Bani

et al., 1998b; Zhang et al., 2002; Dschietzig et al., 2003;

Masini et al., 2006), but the mechanism involved is not

known Our study is the first to show that serelaxin causes an

increase in pERK1/2 in human vascular cells and cardiac

fibroblasts in a Gαi- and PI3K-dependent manner Serelaxin

(30 nM) treatment rapidly increased pERK1/2 in HUVECs

(15% of FBS), HUVSMCs (15% of FBS) and HCFs (50% of FBS)

peaking at 10 min, and in HUASMCs (25% of FBS) at 5 min

with responses that declined to basal after 15 min, except in

HUVECs where pERK1/2 levels plateaued after 10 min

(Sup-porting Information Fig S2c) Responses in all cell types were

inhibited by PTX and wortmannin pretreatment suggesting

the involvement of Gαiand PI3K in pERK1/2 responses

(Sup-porting Information Fig S3a–d) Serelaxin concentration

dependently increased pERK1/2 in HUVECs (Supporting

Information Fig S3a, pEC50: 9.2± 0.3), HUVSMCs

(Support-ing Information Fig S3b, pEC50: 9.2± 0.4), HUASMCs

(Sup-porting Information Fig S3c, pEC50: 9.1 ± 0.4) and HCFs

(Supporting Information Fig S3d, pEC50: 9.1 ± 0.3) As in

the case of cAMP and cGMP accumulation, CRCs were

bell-shaped in HUVECs, HUVSMCs and HCFs, and sigmoidal

in HUASMCs

Role of G proteins in the regulation of serelaxin concentration–response relationships

Previous studies have suggested that differential coupling

to G proteins can explain biphasic or bell-shapedconcentration–response relationships (Baker and Hill, 2006)

As the RXFP1 receptor is known to couple to Gαs, GαO/Band

Gαi/o(Halls et al., 2006), we used pharmacological inhibitors

to selectively disrupt coupling of Gαs, GαO/B and Gαi/o toRXFP1 receptors In both HUASMC (Figure 4A) and HUVSMC(Figure 4B), pretreatment with the Gαs inhibitor NF449(10μM, 30 min) reduced the E-max and right shifted theserelaxin CRC for cAMP accumulation Similar effects wereobserved on the concentration–response relationships forcGMP accumulation (Figure 5A, B) In HUASMC, pretreat-ment with the Gαi/oinhibitor NF023 (10μM, 30 min) reducedthe E-max of cAMP (Figure 4C) and cGMP (Figure 4C) accu-mulation without affecting serelaxin potency In HUVSMC,the Gαi/oinhibitor NF023 (10μM, 30 min) reduced the E-maxfor cAMP (Figure 4D) and cGMP (Figure 5D) accumulation,but also altered the shape of the CRC from bell-shaped tosigmoidal These data suggest that Gαs and Gαi/o regulatedistinct phases of the serelaxin CRC for cAMP and cGMP

Figure 2

The effect of serelaxin on cAMP accumulation in human primary umbilical vascular cells and cardiac fibroblasts Serelaxin treatment (30 min)

increased cAMP accumulation in (A) HUVECs (n = 6), (B) HUVSMCs (n = 6) and (C) HUASMCs (n = 4), but not in (D) HCFs (n = 3) The serelaxin

CRC was bell-shaped for HUVECs and HUVSMCs but sigmoidal for HUASMCs For each cell type, the effect of PTX (50 ng·mL−1, 18 h) andwortmannin (100 nM, 30 min) pretreatment was determined after exposure to serelaxin (30 nM) for 30 min to determine the role of GαiandPI3K Statistical significance was assessed using a one-way ANOVAwith a Dunnett’s post hoc test compared with serelaxin alone: *P< 0.05

and **P< 0.01

British Journal of Pharmacology (2015) 172 1005–1019 1009

accumulation in primary human vascular cells and that the

G-protein coupling is influenced by the cellular background

This was further supported by co-incubation with NF449 and

NF023 that totally abolished serelaxin-mediated cAMP

(Figure 4E, F) and cGMP (Figure 5E, F) responses in both

HUASMC and HUVSMC As NF449 and NF023 are analogues

of suramin, a generic purine receptor antagonist, and as such

these peptides are selective antagonists for P2X1 receptors

(Soto et al., 1999; Rettinger et al., 2005), we conducted

control experiments that showed that pretreatment with

suramin (10μM) had no effect on serelaxin-mediated cAMP

(Supporting Information Fig S4a, b) and cGMP (Supporting

Information Fig S3c, d) accumulation in HUASMC and

HUVSMC Previous studies showed that Gαican be localized

to specific regions of the plasma membrane and RXFP1

recep-tor coupling to Gαi3, but not Gαsor GαO/B, is dependent on

membrane lipid rafts in HEK-RXFP1 receptor cells (Halls et al.,

2009a) Therefore, we disrupted lipid rafts in HUASMC by

filipin III pretreatment (1μg·mL−1, 1 h) that caused a drop in

E-max for cAMP (Figure 4G) and cGMP (Figure 5G)

accumu-lation without any change in serelaxin potency Pretreatment

of HUVSMC with filipin III (1μg·mL−1, 1 h) reduced E-max,

and converted the bell-shaped CRCs for cAMP (Figure 4H)

and cGMP (Figure 5H) accumulation to sigmoidal CRCs

Thus, the effects of filipin III mimicked the effects of the Gαi/o

inhibitor (NF023), suggesting that both the presence and the

location of Gαi/o critically shape serelaxin concentration–

response relationships

Changes in expression of VEGF, ETB and nNOS in human vascular cells following 48 h serelaxin administration

The effects of 48 h serelaxin treatment on protein expression

of VEGF, ETBand NOS are implicated in its vasodilator effects

(Dschietzig et al., 2003; Conrad, 2010; McGuane et al.,

2011b) Serelaxin (10 ng·mL–1; 1.68 nM) caused a twofoldincrease in nNOS expression in HUVECs, HUVSMCs andHCFs, and a threefold increase in HUASMCs (Figure 6A–D)that parallels our findings previously observed in rat renal

myofibroblasts (Mookerjee et al., 2009) Similarly, and also consistent with previous findings (Alexiou et al., 2013), sere-

laxin increased the expression of iNOS in HUVECs (∼2-foldinduction) and had no effect on the expression of eNOS (datanot shown) We were unable to detect and measure iNOS andeNOS in HUASMCs, HUVSMCs and HCFs (data not shown).Similar to previous studies in human endometrial cells

(Unemori et al., 2000), serelaxin also elevated VEGF165

(referred to as VEGF-A) expression by 1.5-fold in HUVECs,HUASMCs and HCFs (Figure 6A, C, D) VEGF-A is the mostpredominant isoform expressed in the vasculature and it has

a crucial role in angiogenesis There are four different isoforms

of VEGF-A (VEGF121, VEGF165, VEGF189, VEGF206) of whichVEGF165is the most predominant (Ferrara, 2004) AlthoughVEGF expression has been reported to be associated with

cAMP accumulation (Unemori et al., 2000), it was unaltered in

HUVSMCs (Figure 6B) despite increased cAMP accumulation

Figure 3

The effect of serelaxin on cGMP accumulation in human primary umbilical vascular cells and cardiac fibroblasts Serelaxin treatment (30 min)

increased cGMP accumulation in (A) HUVECs (n = 7), (B) HUVSMCs (n = 5), (C) HUASMCs (n = 6) and (D) HCFs (n = 5) The serelaxin CRC was

bell-shaped for HUVECs, HUVSMCs and HCFs but sigmoidal for HUASMCs PTX (50 ng·mL−1, 18 h) and wortmannin (100 nM, 30 min)pretreatment significantly inhibited serelaxin (30 nM)-mediated cGMP accumulation in each cell type Statistical significance was assessed using

a one-wayANOVAwith a Dunnett’s post hoc test compared with serelaxin alone: *P < 0.05 and **P < 0.01.

1010 British Journal of Pharmacology (2015) 172 1005–1019

Figure 4

The role of G-protein coupling in serelaxin-mediated cAMP accumulation in human primary umbilical vascular cells Pretreatment of HUASMC

(n = 6) in (A) and of HUVSMC (n = 6) in (B) with the selective Gαsinhibitor NF449 (10μM, 30 min) caused a rightward shift and reduced the

E-max of the cAMP CRC to serelaxin without modifying the shape of the curve Pretreatment of HUASMC (n = 7) in (C) and of HUVSMC (n = 6)

in (D) with the selective Gαi/oinhibitor NF023 (10μM, 30 min) reduced the E-max of the cAMP CRC to serelaxin and changed the shape of the

curve observed with HUVSMC (D) from bell-shaped to sigmoidal In both (E) HUASMC (n = 6) and (F) HUVSMC (n = 6), pretreatment with both

NF449 and NF023 completely abolished serelaxin-mediated cAMP responses, showing that the responses result entirely from RXFP1 receptors

interaction with G proteins In (G) HUASMC (n = 6) and in (H) HUVSMC (n = 6), pretreatment with filipin III (1 μg·mL−1, 1 h), which disrupts lipidrafts, mimicked the effect of the Gαi/oinhibitor NF023 – reducing E-max and converting CRCs from bell-shaped to sigmoidal in HUVSMC

British Journal of Pharmacology (2015) 172 1005–1019 1011

Figure 5

The role of G-protein coupling in serelaxin-mediated cGMP accumulation in human primary umbilical vascular cells Pretreatment of HUASMC

(n = 6) in (A) and of HUVSMC (n = 6) in (B) with the selective Gαsinhibitor NF449 (10μM, 30 min) caused a rightward shift and reduced the

E-max of the cGMP CRC to serelaxin without modifying the shape of the curve Pretreatment of HUASMC (n = 7) in (C) and of HUVSMC (n = 6)

in (D) with the selective Gαi/oinhibitor NF023 (10μM, 30 min) reduced the E-max of the cGMP CRC to serelaxin and changed the shape of the

curve observed with HUVSMC (D) from bell-shaped to sigmoidal In both (E) HUASMC (n = 6) and (F) HUVSMC (n = 6), pretreatment with both

NF449 and NF023 completely abolished serelaxin-mediated cGMP responses, showing that the responses result entirely from RXFP1 receptors

interaction with G proteins In (G) HUASMC (n = 6) and in (H) HUVSMC (n = 6), pretreatment with filipin III (1 μg·mL−1, 1 h), which disrupts lipidrafts, mimicked the effect of the Gαi/oinhibitor NF023 – reducing E-max and converting CRCs from bell-shaped to sigmoidal in HUVSMC

1012 British Journal of Pharmacology (2015) 172 1005–1019

in these cells (Figure 2B) Furthermore, in HCFs, serelaxin

increased VEGF expression but had no effect on cAMP

accu-mulation (Figure 2D) The changes may involve the ERK1/2

(Milanini et al., 1998; van der Westhuizen et al., 2007) or NO

(Dulak et al., 2000) pathways that are activated in all cell

types Furthermore, and consistent with previous studies

(Dschietzig et al., 2003), serelaxin increased ETB receptor

expression in HUVECs and HUVSMCs by two- to threefold,and in HUASMCs and HCFs by 1.5-fold (Figure 6A–D) ETB

receptor expression has been reported to be downstream of

MMP and ERK1/2 signalling (Dschietzig et al., 2003; Jeyabalan

et al., 2003; Chow et al., 2012), and these pathways were

activated in all cell types where ETBreceptor expression wasincreased (see Figure 7 and Supporting Information Fig S3)

Figure 6

Changes in the expression of nNOS, VEGF and ETBreceptors in human primary umbilical vascular cells and cardiac fibroblasts after serelaxin

(1.68 nM) exposure for 24 and 48 h In HUVECs (A), serelaxin treatment increased the expression of nNOS (n= 5), ETBreceptors (n= 6) and VEGF

(n = 7) In HUVSMCs (B), serelaxin treatment increased the expression of nNOS (n = 5) and ETB(n = 5) but not VEGF (n = 4); however, in HUASMC (C), serelaxin treatment increased the expression of nNOS (n= 6), ETBreceptors (n = 7) and VEGF (n = 5) In HCFs (D), similar to HUVECs and HUASMCs, serelaxin treatment increased the expression of nNOS (n= 5), ETBreceptors (n = 5) and VEGF (n = 5) A representative blot of each

protein andβ-actin, a loading control, is shown with the densitometry in each figure Statistical significance was assessed using a one-wayANOVA

with a Dunnett’s post hoc test compared with vehicle alone: *P < 0.05 and **P < 0.01.

British Journal of Pharmacology (2015) 172 1005–1019 1013