current trends in the development of new efficient and relevant models.
Models for drug absorption from the small intestine: where are we and
where are we going?
Pierre-Andre´ Billat
1, Emilie Roger
1, Se´bastien Faure
1and Fre´de´ric Lagarce
1,21MINT,UNIVAngers,INSERM1066,CNRS6021,Universite´ BretagneLoire,France
2PharmacyDepartment,AngersUniversityHospital,Angers,France
The small intestine is a complex organ with movements, flora, mucus and flows. Despite this, the most widely used absorption models consider the organ a cylindrical monoepithelial tube. This review presents the recent evolution of models to take into consideration the complex nature of gut physiology. The most commonly encountered issues are ethical (in vivo models) and differences in drug transport as a result of a modified expression of drug transporters or metabolic enzymes compared with human (in vitro and in vivo models). Finally, this review discusses the way forward to reach an ideal equilibrium between reproducibility,
predictability and efficiency for predicting permeability. The features of an ideal model are listed as a guideline for future development.
Introduction
Theoralrouteisthemostcommonandpracticalwaytoadministerdrugstothebody;evenif certainproblemsremain,especiallyforanticanceragents[1].Unfortunately,notalldrugs are goodcandidates for oral administration. The Biopharmaceutical Classification System (BCS) proposedby Amidonetal.in1995[2]showsthatsolubilityandpermeabilitycanbeusedto determinewhetheradrugisagoodcandidatefororaladministration.Similarly,theRuleofFive (Ro5;seeGlossaryforlistofabbreviationsusedinthisreview)proposedbyLipinskietal.in1997 [3]isalsoaquickwaytoassessthesuitabilityofadrugfortheoralroute.TheBCSandRo5are related tochemicalproperties ofdrugs. They arewaysto predictagoodabsorption process, helpingtoreachabetterbioavailabilityofthedrugandoftenalowinterpatientvariability,which meansreliability.Thesepositivefeaturesarecoinedinthetermdrugability,whichreflectsthefact thatthedrugisagoodcandidatefortheoralroute.Then,theactivedrugmustbeformulatedto obtainanoraldosageform.Intheearlystagesofdrugdevelopment,invitroand/orinvivomodels areextensivelyusedtodeterminethebestformulationofthedrugproduct.Ideally,thosemodels mustbeeasytoimplement,relevant,simple,costeffective,accurateandcompatiblewithhigh- throughputscreening.Someofthosefeaturesaredifficulttoobtainaltogether.Asamatteroffact, thecomplexityoftheabsorptionprocessmakesitimpossibleforthemodelstoberelevantandto
ReviewsFOUNDATIONREVIEW
Pierre-Andre´ Billat receivedhisMScinclinical andpreclinical pharmacologyfromParis DescartesUniversity,his PharmDfromtheLimoges UniversityandhisPhDin pharmacologyfrom LimogesUniversity.He
joinedtheAngersInsermUnitU1066in2015,initially focusingonthedevelopmentofanewdrugabsorption modelforthenanomedicinesconceivedinthelab.He iscurrentlyanon-tenureassistantprofessorwith appointmentsinphysiology,pharmacologyand pharmacokinetics.Hiscurrentresearchcontinuesto developabsorptionmodels:cultureofintestinal organoids,organoidgeneticcharacterizationand integrationoforganoidswithmicrofluidicsystems.
Fre´de´ricLagarce receivedhisPhDin2004, andisprofessorof Pharmaceutical Technologyand Biopharmaceutics(since 2012)attheUniversityof Angers,France.Beingalso ahospitalpharmacist,his
researchstudiesaretranslational(frombenchto bedside)andmainlyfocusedoncancertherapy, especiallyonbioavailabilityenhancementbyplayingon theinteractionsbetweendrugproducts(mainly nanosystems)andlivingtissues.Thisfieldinvolves biologicalbarriercrossingstudiesbutalsostability assessmentoftheactivemoietiesandovercomingthe acquiredresistancestodrugs.Inhospital,heisin chargeofthepharmaceuticaltechnologyandquality controlunit.Findingnewanswerstomedicalneeds usinginnovativedrugformulationsiswhatdriveshim everydaytowork.
Correspondingauthor:Lagarce,F. (frederic.lagarce@univ-angers.fr)
1359-6446/ß2017ElsevierLtd.Allrightsreserved.
http://dx.doi.org/10.1016/j.drudis.2017.01.007 www.drugdiscoverytoday.com 1
remainsimple.Thesemodelsmustalsobesuitableto assessthe absorptionofnewformulationssuchasnanomedicines. Besides predictingtheextentofdrugsabsorbed,modelsarealsousedto exploretheverydifferentbarrierstocrossandthecomplexmech- anismsofthistransportprocess.Modelsarealsousedtostudythe stability andthe behaviorofthe formulation.Alotistherefore expectedoftheabsorptionmodels,whichiswhymanytechniques areusedtoconstructthesemodelsfrominsilicomodels,basedon mathematical analysis and on chemistry properties, to in vivo models,oftenbasedonmolecularimaging.Invitromodelsbased
oncellculturesarea goodcompromise betweensimplicityand relevance and are therefore widely used. In the present study, usingthephysiologyandmolecularbiologyofthegastrointestinal tractasastartingpoint,wewouldliketoproposeacriticalanalysis ofthemostwidelyusedinvivo,invitroanddynamicabsorption models.Subsequently,thecurrenttrendsinthedevelopmentof new,efficientandrelevantmodelswillbeexploredtoproposethe crucialpointstoconsiderinthewayofinnovationinthisfield.
Thefeaturesoftheidealmodeltostudydrugabsorptionwillbe presentedasaconclusiontothiswork.
Where are we?
Physiologyandpharmacology ofthesmallintestine
Physiologyandmotilityofthesmallintestine.Themain drugabsorptionstepsoccurinthesmallintestine.Thisorganisa complextubethatcanbedividedintothreedifferentpartswith differentabsorptivecapabilities.Thefirstpart,theduodenum,is 25–30cmlongandthepassageofthedrugsthroughthispartis relativelyquick,resultinginapoornetabsorptionofdrugs.The secondpartisthejejunum.Unliketheduodenum,thejejunumis characterizedby a highlyactiveperistalsis, favoringabsorption.
Finally,drugsthathavenotbeenabsorbedinthejejunummight beabsorbedintheileum(giventhatthesiteofabsorptionmainly depends on the physicochemical properties of the drug). The ileum reveals fewer villi than the jejunum but has a similar absorptioncapability.Mucusisanotherfundamentalpartofthe gastrointestinaltract,actingasamechanicalandphysicalbarrier, butalso performingthe role ofa niche forthe microbial cells.
Microbioticfloraareanessentialpartofthegut.Itisnowclearthat thegutmicrobiomeplaysakeypartindigestionbutalsointhe productionofenzymesandvitaminsandintheregulationofthe immunesystem.Despitetheincreasinginterestinthetopic(the number of papers on Pubmed with the keyword ‘microbiota’
reached 5028 in 2015 vs 245 10years earlier), the microbiota, its metabolic activities and its interactions with the digestive epithelium are still not fully understood. In each part of the intestinaltube,thecommensalfloracompositionandthemucus bilayershow a highvariability.The maincharacteristics ofthe smallintestinearesummarizedinTable1.
Impact of drug transporters. Drug absorption is mainly subject to significant transporters, even if some drugs can be absorbedbypassivediffusion,and totheeffectsofmetabolism.
GLOSSARY
ABCATP-bindingcassette
APIactivepharmaceuticalingredient BCSbiopharmaceuticalclassificationsystem CATcompartmentalabsorptionandtransitmodel CYPcytochromeP450
DMPC1,2-dimyristoyl-sn-glycero-3-phosphocholine EGFepitheliumgrowthfactor
FAEhumanfollicle-associatedepithelium GSTglutathioneS-transferase
MATEmultidrugandtoxiccompoundextrusiontransporters MDmoleculardynamics
MDCKMadin–Darbycaninekidney MGSmetagenomicshotgunsequencing MMmolecularmodeling
NATN-acetyltransferase OATorganicanionstransporters
OATPorganicanionstransportingpolypeptides OCTorganiccationstransporters
OCTNzwitteriontransporters P-gpP-glycoprotein
PAMPAparallelartificialmembrane PB-PKphysiology-basedpharmacokinetics PEPTpeptidetransporters
PVPAphospholipidvesicle-basedpermeationassay QSARquantitativestructure–activityrelationship Ro5RuleofFive
SLCsolutecarrierfamily SULTsulfotransferase
TEERtransepithelialelectricalresistance TMDtransmembranedomains
UGTUDP-glucuronosyltransferase
TABLE1
Comparisonofthephysiologicalparametersindifferentpartsofthegut.
Region Average length(m)[92]
Absorbingsurface area(m2)[92]
pH[93] Residence
time[93]
Mucus
thickness(mm)[94]a
Bacterial
cells(cells/ml)[95]
Duodenum 0.25–0.30 0.09 5.7–6.8 40min FAL:163
LAL:15439
101–104
Aerobesandfacultativeanaerobes
Jejunum 3 60 6.6–7.0 2–3h FAL:152
LAL:1085
104–108
Facultativeanaerobesandaerobes
Ileum 3 60 7.0–7.3 3–4h FAL:298
LAL:44747
104–108
Facultativeanaerobesandaerobes
Colon 1.5 0.3 5.7(caecum)to6.6 16.6–19.0h FAL:11651
LAL:714109
1010–1012
Facultativeaerobestostrict anaerobicbacteria(mainlyClostridia) Abbreviations:LAL,looselyadherentlayer;FAL,firmlyattachedlayer.
aStudyperformedinrats.
2 www.drugdiscoverytoday.com
ReviewsFOUNDATIONREVIEW
ThefateofthesedrugsisrepresentedschematicallyinFig.1a.The paracellulartransport,whichisrelatedtomoleculesbelow3.6A˚ or 250Da, occursthroughtight junctionsbetween epithelialcells alongtheintestinalmucosaand showsanimportant intra-and inter-individual variability[4]. However,it hasbeen previously shownthatbiggerstructuressuchasnanoparticlescancrossthe tightjunction structures [5]. Nevertheless, several current drug classescrosstheepithelialcellsbytranscellulartransport,eitherby passivetranscellulartransport(onlysmalllipophilicdrugs)and/or bycarrier-mediatedtransport(antivirals,penicillins,statins,etc.) [6].
Roleoftransportersintheabsorptionofdrugs.Trans- portersareusuallyclassedintotwocategories.Thefirstcategoryis the solute carrier (SLC) family, which mediates the uptake of drugs.TheATP-bindingcassette(ABC)familyisthesecondcate- gory and gathers effluxtransporters (Fig. 1b).Most ofthe SLC transportersaresecondaryactivetransporters,forwhichtransport isdrivenbyvariousenergy-couplingmechanisms[7].Thiscatego- ryisdivided intotheSLCO superfamily(formerSLC21),which gatherstheorganicanionstransportingpolypeptides(OATP),the SLC22 superfamily, which gathers organic cation transporters (OCT)andzwitteriontransporters(OCTN),theSLC47superfami- ly,whichgathersthenucleotidetransporters(ENTandCNT),and peptidetransporters(PEPT).
Few members of the SLCO superfamily are found in the intestine. Although OATP2A1 and OATP4A1 are ubiquitous andtransportmainlyprostaglandinsandbilesalts,respectively, onlyOATP1A2canhavearoleintransportingdrugs.Situatedat theapicalsideoftheenterocytes,itssubstratesarebilesalts, thyroid hormones and xenobiotics (antibiotics, anticancer drugs,antifungals,b-blockers,statins).
Inthe SLC22superfamily, all themembers sharea common structure:12a-helical transmembrane domains(TMDs).This familyactivelyparticipatesin smallintestinalabsorptionbut alsoinhepaticandrenalexcretionofdrugs.Intheintestine,the most common SLC22 transporters are OCT1, OCT2, OCT3, OCTN1,OCTN2andOctn3.
- OCT1isawell-knowntransporterofmetformin,quinidine and type 1 cations (such as dopamine, choline and N1- methylnicotinamide) by sodium-independent transport.
However, its location in enterocytes is still unclear [8,9].
OCT3 mediatestheuptake ofhistamine, epinephrineand norepinephrineand cationicdrugs togetherwithOCT1in theintestine.
- TheOCTNfamilyincludesthreetransporters.OCTN1and OCTN2 havebeen foundin humans, whereasOctn3has been found only in mice. OCTN1 and 2 uptakeorganic cationsandzwitterionsbyNa+-dependentor-independent transport. The most common substrates are oxaliplatin, gabapentin, verapamil, doxorubicin and quinine for OCTN1 and oxaliplatin, ipratropium and tiotropium for OCTN2.
Nucleosidetransportersareamajorconcerninthedevelopment ofanticancerandantiviraldrugsbecausetheytransportnucleo- sidesandalargevarietyofnucleoside-deriveddrugs.Threetrans- porters of this family are commonly studied in the intestine:
CNT1, ENT1 and ENT2. Like the nucleoside CNT1 transporter, locatedintheapicalmembranesofpolarcells,ENT1transporters
are locatedpredominantly on the apicalside, whereasENT2 is presentontheapicalandbasolateralsidesinCaco-2cells.
PEPT1 encoded by the SLC15A1 gene is responsible for the influxofdi-andtri-peptidesinenterocytes.Itcanalsotransport peptide-likedrugs(i.e.,angiotensin-convertingenzymeinhibitors, b-lactam antibiotics) and drugs coupled to amino acids (i.e., valganciclovirorvalacyclovir).
Once inside the cell the drug must be transported to the basolateral sidetoreachthe bloodcirculation.Inparallel,some transporterscanalsoeffluxdrugsontheapicalside,thusregulat- ingtheintracellularconcentrationofxenobioticsanddecreasing theabsorptionrate.EffluxtransportersareATP-dependentpumps andareresponsibleforawidenumberofdrugsand/ormetabolite transport.Todate,therearesevensubfamiliesofABCgathering51 transporters.Among thosetransporters,fourareresponsiblefor theeliminationofdrugsfromcellsintothelumen:P-glycoprotein (P-gp), Multidrug resistance protein 2/3 (MDR2/3), Multidrug resistance-associatedprotein2(MRP2)andbreastcancerresistance protein (BCRP). These transporters reduce the uptake of their substrates and arelocated at the apical side ofthe enterocyte.
Bycontrast,fivetransportersareresponsiblefortheeffluxofthe drugstowardthebloodandtheliver:MRP1,MRP3,MRP4,MRP5 andMRP6.Theyarepreferentiallylocatedatthebasolateralsideof theenterocyte.
TheABCB1geneencodesforP-gp,themostwell-knownefflux pump. Itpumps the xenobioticsfrom thecell back intothe lumen. Current recommendations for testing MDR1 during drugdevelopmentarebasedonitsroleinintestinalabsorption.
Moreover,P-gphasaroleinmodulatingCYP3A4expression, thuscontributingtopharmacologicalresistance[10].
TheABCB4geneencodesfortheMDR2/3protein.Smithetal.
reported an increased directional transport of several MDR1 P-gp substrates, such as digoxin, paclitaxel and vinblastine, throughcellsexpressingABCB4[11].
The ABCC2 gene encodesthe MRP2efflux protein. MRP2is mainlylocatedintheliver,inkidneyandintestine,supporting a major function in the elimination and detoxification of xenobiotics,andparticularlyglutathioneconjugates.
BCRPexhibitsbroadsubstratespecificity witha considerable substrate overlap with ABCC1 and ABCB1. BCRP is highly expressed in the small intestine, colon, blood–brain barrier, placentaandliver.
ABCC1andABCC3encodefortwobasolateraleffluxtranspor- ters. The main roles of these transporters are the efflux of xenobiotic andendogenousmetabolitesandthe transportof inflammatorymediators.
ABCC4 and ABCC5encode for MRP4and MRP5, whichare ubiquitouseffluxtransporters.Theytransportmainlynucleo- tideanalogssuchasantiviralsandanticancerdrugs.
NotmuchisknownaboutABCC6.Thisgeneisexpressedinthe duodenum,colonandliver,butitssubstrates(endogenousand exogenous)arenotknown.
Obtaininganexhaustivelistofsubstratesforeachtransporter wouldbeworthwhile,butlaborious.However,transportersseem tobemoreclass-specificratherthandrug-specific.Assuch,drugs arecommonlygroupedintoclasseswithsimilarphysicochemical properties, which renders the screening of hypothetical drug transporterseasier(Table2).
www.drugdiscoverytoday.com 3 ReviewsFOUNDATIONREVIEW
(a)
(b)
(c)
Apical
Influx transporters/
facilitated diffusion
Efflux transporters/
facilitated diffusion Influx transporters/
facilitated diffusion Efflux transporters/
facilitated diffusion
Paracellular route
Drug resistance/
detoxification Absorption Parent drug Metabolite
CNT1 ENT1/2 OATP1A2 OCTN1/2 PEPT1 OCT1 ? OCT3 ?
Influx transporters
Efflux transporters ENT1/2
P-gp MDR2/3
MRP2 BCRP
Microflora enzymes
Phase I enzymes Phase II enzymes
Reduction hydrolysis
CYP3A CYP2C Other CYP
GSTP1 GSTPA1/2
NAT1/2 SULT UGT1A UGT2B
ENT1/2 OCT 1?
OCT 3?
ENT1/2 MRP1 MRP3-6
Absorption transporters Drug resistance/
detoxification transporters Efflux transporters/
facilitated diffusion Efflux transporters/
facilitated diffusion Metabolism
Basolateral
Drug Discovery Today
FIGURE1
Fateofdrugsincontactwithintestinalwall.(a)Mainpathwaysfordrugabsorption.(b)Maintransportersinvolvedindrugabsorption.(c)Mainenzymaticpaths involvedduringdrugabsorption.
4 www.drugdiscoverytoday.com
ReviewsFOUNDATIONREVIEW
TABLE2
Influxandeffluxtransportersofthegut.
Nomenclature Genename Common abbreviation
Function Intestinal
localization
Substrates Inhibitors
Organicanions transporting polypeptide1A2
SLCO1A2 OATP1A2 Uptakeofbileacids, thyroidhormones andPGE2
Apical Antibiotics,anticancerdrugs, antifungals,b-blockers,statins
Naringin[96]
Organiccation transporter1
SLC22A1 OCT1 Uptakeoforganic cations
Notclear Metformin,quinidine, dopamine,choline,PGE2, acyclovir,N1-methylnicotinamide andtype1cations
Clonidine[97]
Organiccation transporter3
SLC22A3 OCT3 Uptakeoforganic cations
Notclear Amantadine,atropine, epinephrine,histamine, metformin,norepinephrine andcationicdrugs
Quinine
Organiccation/carnitine transporter1
SLC22A4 OCTN1 Uptakeoforganic cationsand
zwitterions,L-carnitine
Apical Verapamil,pyrilamine, oxaliplatin,gabapentin, doxorubicin,quinine, organiccationsand zwitterions
–
Organiccation/carnitine transporter2
SLC22A5 OCTN2 Uptakeoforganic cationsandzwitterions
Apical Verapamil,pyrilamine, L-carnitine,oxaliplatin, ipratropium,tiotropium, organiccationsand zwitterions
–
Concentrativenucleotide transporter1
SLC28A1 CNT1 Uptakeof
1nucleoside/1Na+
Apical Nucleotides Nucleotidesanalogs
–
Equilibrativenucleotide transporter1
SLC29A1 ENT1 Uptake/Effluxof nucleotides
Apical Nucleotides Nucleotidesanalogs
KF24345, NBMPR[98]
Equilibrativenucleotide transporter2
SLC29A2 ENT2 Uptake/Effluxof nucleotides
Apical/
basolateral
Nucleotides Nucleotidesanalogs
KF24345, NBMPR[98]
Peptidetransporter1 SLC15A1 PEPT1 Uptakeofdiand tripeptides/2H+
Apical Diandtripeptides Peptide-likedrugs
Drugscoupledtoaminoacids, cephalosporins,penicillins
Lys[Z(NO2)]-Pro (GTPL4498)[99]
4-AMBA[100]
Glycylsarcosine[101]
P-glycoprotein ABCB1 MDR1,P-gp Effluxofhydrophobic, amphipathicorcationic molecules,steroid hormones,bilesalts
Apical >200xenobiotics (anticancerdrugs, digoxin,etc.)
Valspodar[102]
MDR2/3 ABCB4 PGY3 Effluxof
phosphatidylcholine
Apical Ivermectine,daunorubicin, digoxin,paclitaxel,vinblastine
–
MRP2 ABCC2 cMOAT Effluxofbilesalts Apical Organicanions,glutathione andconjugates,anticancer drugs(methotrexate),etoposide, sartans,bromosulfophthalein, nucleotideanalogs
MK-571(specific toforMRPgroup)
BCRP ABCG2 BCRP Effluxofporphyrins,
flavonoids,estrones andbileacids
Apical OverlapwithP-gpsubstrates Ko143[103]
MRP1 ABCC1 MRP1 Transportofhydrophobic
drugs,estrogens andprostaglandins
Basolateral Antivirals,anticancerdrugs, quinolones,glucuronide conjugate
MK571
MRP3 ABCC3 MRP3 Transportofglutathione
conjugates
Basolateral Organicanions,anticancer drugs,glucuronideconjugate
MK571
MRP4 ABCC4 MRP4 Nucleotides,
prostaglandins
Basolateral Cephalosporins,antivirals, anticancerdrugs
MK571
MRP5 ABCC5 MRP5 Cyclicnucleotides,
folates
Basolateral Statins,antivirals, anticancerdrugs
MK571
MRP6 ABCC6 MRP6 Notclear Basolateral ? MK571
Abbreviations:BCRP,BreastCancerResistanceProtein;MDR,MultiDrugResistanceprotein;MRP,MultidrugResistance-associatedProtein;NMBR,6-S-[(4-Nitrophenyl)methyl]-6-thioinosine;
PGE2,prostaglandinE2.
www.drugdiscoverytoday.com 5 ReviewsFOUNDATIONREVIEW
Impact of intracellular and microflora metabolism.
Drug metabolism takes place in the intracellular milieu and dependsontwoclassesofenzymes.PhaseIenzymesareresponsi- ble for the functionalization of the drugs (i.e., hydroxylation, amination,etc.).PhaseIIenzymestendtoconjugatethemetab- olitesbyglycosylation,glucuronidation,transmethylationoracet- ylation and sulfoconjugation (Fig. 1c). Phase I enzymes are cytochromes.Owingto their broadspecificity,highabundance intheintestineandpowerfulcapacityforoxidizingxenobiotics, cytochrome P450 (CYP) proteins are the most often studied.
CYP3AandCYP2CrepresentthemajorintestinalCYPs,account- ingfor80%and18%,respectively,oftotalimmunoquantified CYPs [12].PhaseIImetabolismimplies UDP-glucuronosyltrans- ferases (UGTs), glutathione S-transferases (GSTs), N-acetyltrans- ferases(NATs)andsulfotransferases(SULTs).
UGTsareasuperfamilyofenzymesthatcatalyzetheglucuro- nidationofendogenousandexogenousmolecules.Amongthe21 differentUGTproteinsthathavebeenidentifiedinhumans,ten are expressed in the small intestine: UGT1A1, 1A3, 1A4, 1A6, 1A10,2B4,2B7,2B10,2B11and2B15[13].Eachenzymeencoded byaUGTgenerevealsauniquebutusuallyoverlappingsubstrate specificity, tissue localization and regulation [14]. Human UGT1A1isthemosthighlyexpressedUGTinthesmallintestine, withactivitiesevengreaterthanintheliver[15].UGT1Aenzymes conjugateendogenousandexogenoussubstrates.
GSTsoccurinthreecellularcompartmentsandcanbedivided intocytosolic GSTs,mitochondrialGSTsand microsomalGSTs.
The most expressedGSTsinthe humanenterocytesareGSTP1, GSTA1andGSTA2[16].GSTA1andGSTA2catalyzetheconjuga- tionofglutathionewithelectrophileswhereasGSTP1inactivates toxicandcarcinogeniccompoundsbyconjugationofglutathione.
Most studies focus on the role of GST in pathogenic, mainly cancerous,tissue.AlthoughGSTsarefoundinenterocytes,their behaviorinthistissuehasbeenpoorlyexplored.NAT1andNAT2 canbefoundintheintestine.Theycatalyzetheacetylconjugation fromacetyl-CoAtovariousarylamine andhydrazine substrates.
There is anincreasinginterest in SULTs, becausethey seem to contributesignificantlytodrugclearance[17].Theycatalyzethe transferofasulfategrouptoseveralpharmacologicallyimportant endo-andxeno-biotics.
Thehumanmicrofloracontainsobligateanaerobes(i.e.,Bacter- iodes, Clostridium, Lactobaccillus, Bifidobacterium, Eschericia), to- gether with a variety of yeasts and other microorganisms, forming a complex ecosystem of thousands of species. About 90%ofthe intestinalmicrobiotais composedofthe Bacteroides andFirmicutes(i.e.,Clostridium,Bacillus)families.Themicrobiota canmetabolizedrugs,thusdecreasingorfavoringtheabsorption ofdrugsand/ortheirmetabolites.Althoughmostofthesemecha- nismsremainunknown,thisknowledge iscrucialfornewdrug development.Forexample,somestudieshaverevealedthat the microflora mediates the reduction [18] and hydrolysis [19] of drugs,butalsotheremovalofsuccinategroups,dihydroxylation, (de)acetylation, proteolysis, (de)conjugation and N-demethyla- tion[20],thusactivatingprodrugssuchaslovastatinorinactivat- ingdrugssuchasdigoxin[21,22].Inparallel,floracaninfluence cellbehavior,forexamplebyincreasingtheactivityofCYPphase IIenzymesinthegutorevenintheliver[23,24].Toillustrate,in germ-freemice,mRNAofCyp3ahasdecreasedby87%compared
withcontrolmicewithnormalflora,GSTsfrom32%to66%and SULTsfrom52%to68%,thusleadingtoasignificantdecreasein drugmetabolism[25].Itisclearthatfloracansignificantlycon- tribute to drug metabolism in beneficial and deleterious ways.
Somebacteriacanalsoreduceorinducedrug-relatedtoxicity.For example,bacteriaexpressingb-glucuronidase(i.e.,Escherichiacoli) increase significantly the number of ulcers in mice receiving nonsteroidalanti-inflammatory drugs (NSAIDs) [26,27] and en- hancetheformationoftoxiccompoundsinhumansundergoing chemotherapy[28].Thedevelopmentofnew-generationsequenc- ingtechniquestoreplacetheclassicalculturetechniques,which failedatidentifyingmostofthespecies,makesitpossibletostudy theminorspeciesthatcanalsohavearoleindrugabsorptionand metabolism.Currently,microbioticfloraareexploredbytargeting thehighlyconserved16SribosomalRNAgenesequencesormeta- genomicshotgunsequencing(MGS)[29].Combiningthemodern, highlysensitiveanalyticalapproacheswith16SrRNAandmeta- genomicdatamakesitpossibletodetectandquantifythemetab- olitesthatarederivedormodifiedfromthegutmicrobiotaandto identifythespeciesinvolved.
At this stage, it is easy to understand that drug absorption dependson severalsteps:reaching the apicalside ofthe enter- ocyte,transportbyseveraldifferentproteinsandmetabolismby manymoreofenzymesbeforereachingtheliver.Suchacomplex pathwayinvolvesrisksrelatingtodrug–druginteractionsoreven food–drug interactions, for example when the transporters or enzymesaresaturated,thusrenderingtheaccuratepredictionof drugabsorptionimpossibleamongindividuals.Drugpermeability couldalsobeaffectedbygeneticpolymorphismsamongprevious genesanddiseasestates.
Currentmodels
In vivo models. In vivo models are of great interest because currently these are the only models to potentially contain all the physiological parameters. Among these models, intestinal perfusion is the most common experiment used to study the invivodrugpermeabilityandintestinal metabolismin different regionsoftheintestine.
Animal intestinal perfusion. Briefly, intestinal perfusion consistsof:(i)exposing thesmall intestineandligatingthe part of interest for perfusion; (ii) rinsing the cannulated segment;(iii)perfusingthesolutionofinterestandcollecting theperfusate.Fig.2illustratestheinvivoperfusionsystem.In thissystem,themesentericveinis cannulatedandbloodis collected. During this experiment, special care should be takentomaintainanintactbloodsupply.Finally,theanimal must be euthanized. Another variation is the closed loop model.Inthismodel,thegutremainsintheanimalandeach extremityofthepartofinterestisligatured.Thedrugisthen injectedintheisolated part.Attheendoftheexperiment, drug concentration is measured inside the gut and the absorption rate is deducted from the initial concentration.
Itpresentsseveraladvantagesoverthepreviousmethod,such as the exploration of the effects of inhibitors or drug interactions,andcanbeusedtostudydrugsthatarepoorly permeable.Becausethevariationinthequantityisextremely insignificant,thismethodrequiresananalyticalmethodwith averylowlimitofquantification.
6 www.drugdiscoverytoday.com
ReviewsFOUNDATIONREVIEW
Althoughthepermeabilityofpassivelyabsorbeddrugscorrelates wellwithhumandata[30],thisisnotasclearfordrugsabsorbed by active transport. To specifically study the impact of a transporter, however, several knockout models have been developedin recent years. The firstmodel was developedby Schinkel et al. using mice without P-gp (Mdr1a / ) [31].
Subsequently,severalothermodelsappeared,suchasBcrp1 / mice[32]andOct1 / mice[33].Despitethisattempt,ithasbeen observedthatgenesofthesamefamilyoftheknocked-outgene are overexpressed in a mechanism of compensation, thus obtaining puzzling results [34]. To conclude, this method is relativelyeasytosetup,quickandcheap,whichinpartexplains itsnotoriety.
Loc-I-GutTM.Amethodtodetermineintestinalpermeabilityin humans has been developed by Lennerna¨s et al. [35]. This systemisbasedonajejunalperfusionsystemcomposedofa multichanneltubewithtwoinflatableballoons(Loc-I-GutTM).
Theobtainedpermeabilityresultscanbefurtherusedasagold standardtocomparedifferentmodels.However,permeability studiesusing volunteersarelimitedbecauseof ethicalissues and cost.Nevertheless, thisisperhapsthe mostrealisticand predictivepermeabilitymodeleverdeveloped.
Pharmacokineticstudy. Individualstake the drug orallyand venousbloodissampledatdifferenttimespost-dose.Although
it seems like one of the easiest, most reliable and simplest methods to study drug absorption, a pharmacokinetic study canhaveseveraldrawbacks.Theobserveddrugconcentrations areextremelyvariable–inter-individuallyandintra-individual- ly.Thiscanbepartiallyexplainedbydifferentratesofgastric emptying,differencesingutandlivermetabolism,differences in transporterexpressionand inelimination. Apharmacoki- neticstudyofpopulationcanovercomethesevariabilitiesbut wouldinvolvealargenumberofindividuals.
Although drugs absorbed by passive transport show a good correlation between animals and humans [36], there are huge discrepanciesbetweenthedifferentanimalmodelsandhumans in termsofmetabolism, drugtransport, floraandof coursethe surfaceareaofthegastrointestinaltract.Forexample,whileusing intestinalperfusionforfivepassivelytransporteddrugsinhumans, theapparentpermeabilitywas3.6-timeshigherthanobservedin rats but results were similar to those of mice [37]. Moreover, although moderatecorrelation (r2>0.56)wasfound inthe ex- pressionlevelsoftransportersintheduodenumofhumansand rats,nocorrelationwasfoundintheexpressionofmetabolizing enzymesbetweentheratandhumanintestine[30].Thisexplains why a reliable scalingfrom animal modelsto humans is often absent.Anothercommonlimitationoftheinvivoapproachisthat itisnotsuitableforhigh-throughputscreening,itpresentsalow sensibilityandrecoveryofdrugs,whichmakesanalysisbymass spectrometryindispensable.Withregardtotheseadvantagesand drawbacks, these models might bepreferentially used to study activelytransporteddrugs.Drugsthatareabsorbedpassivelycan bestudiedusinglessexpensiveandsimplermodels,suchasexvivo andinvitromodels.
Exvivoandinvitromodels.Asinthecaseoftheabove- mentionedinvivomodels,evertedintestinalsactechniquesare used todeterminedrugpermeability. Theintestineisremoved fromdeadanimalsandcutintosmallsegments,alsomakingit possible to evaluate the permeabilityin different partsof the intestine.Thesesegmentsaresutured atone end,filledwitha drug solution, sutured at the other end and immersed in an oxygenated mediumat 378C. Thismodel presentstheadvan- tageofa relativelylargesurface forpermeabilityandthepres- enceofamucuslayer.However,onelimitingparameterofthis model is tissueviability [38]. Moreover,no invivo correlation has been established through this method but results from evertedintestinalsacmodelshavebeenconsistentwithinvivo findings[39].Asforinvivomethods,themajordrawbackisthe poorscreeningrate.Nevertheless,thesemodelsshould bepre- ferredtoinvivomodelsifanestheticdrugsmightinterferewith theanalysis.
Cellularmodelshavebeenwidelyusedtostudydrugperme- abilityinthesmallintestine.Althoughmostcellularmodelsare simple monocellular layers, complex models have also been developed, forexample in including a liver-likecompartment composed ofmicrosomes[40].First,Caco-2cellsrepresentthe referencemodelinthepredictionofdrugpermeabilityandare routinelyusedforstudyingenterocytetransepithelialdrugtrans- portforthepassivetranscellularroute,paracellularroute,carri- er-mediated route and transcytosis [41]. Caco-2 cell lines, derivedfrom ahumancolorectal carcinoma,are cultivatedon semipermeablefilters(Transwell1system)for21–23days.After Infusion pump
drug in
Jugular vein blood supply
Blood sampling Ligatures
Drug out 37°C physiological
bath
Drug Discovery Today
FIGURE2
Invivoperfusion:exampleoftherat.
www.drugdiscoverytoday.com 7 ReviewsFOUNDATIONREVIEW
differentiation,thecellsformapolarizedmonolayerwithapical andbasolateral sidesdisplayinga brush border,microvilliand tightjunctions,andexpressingP-gpandseveralrelevantefflux transportersandenzymes[42].SubclonesofCaco-2cells(TC-7) are alsoused andhavedifferentlevelsof transportersanden- zymeexpressions,closertohumanlevelsthanclassicCaco-2,are also used [43]. Although a good correlation between in vitro permeability (Papp) of drugs and their in vivo bioavailability wasfound[44,45],theCaco-2cellmodelisnotperfect.Indeed, paracellulartransportislowerthantheinvivopermeability[41].
Consequently, alternative cells were put forward [IEC-18 (rat smallintestinecells)and2/4/A1(fetalratintestinecells),which display higher paracellular permeability]; however these cells containfewcarrier-mediatedtransportsystemscomparedwith Caco-2 cells [46,47]. Moreover, Caco-2 cells express a low amount of CYP3A enzymes, which are major metabolizing enzymes for many drugs. In this way, CYP3A4-transfected Caco-2 cells with higherlevelsofCYP3A4 are developed[48].
Moreover,otherdisadvantages ofthis modelinclude thelong differentiationperiod,thewidevariationwithpassagenumber ofcellsandtheinter-andintra-laboratoryvariability.Asaresult, twoothercellmodels arealsousedforpermeability transport:
Madin–Darbycaninekidney(MDCK)IandIIderivefromcanine kidney cells; and Lewis lung carcinoma-porcine kidney 1 (LLCPK1) derived from pig kidney epithelial cell line. MDCK cellsexhibitashorterculturetime(3–5days)andlowertranse- pithelialelectricalresistance(TEER)valuescomparedwithCaco- 2cells(MDCKvaluesaremuchclosertotheinvivoTEERofthe smallintestine).TheMDCKmodelalsopresentspolarizedcells, withbrushborderandtightjunctions,butthismodelexpresses some transporters, such as P-gp, with lower levels compared with Caco-2 cells [49]. Nevertheless, MDCK cells transfected with thehuman MDR1gene (MDCK II) have beendeveloped toexpressP-gp.
Anotherlimitationofthesecellularmodelsistheabsenceofa mucus layer and M cells, which also constitute the intestinal barrier.Inthisway,co-culturemodelscomprisingdifferentcells havebeenproposedtorepresenttheheterogeneityoftheintesti- nalepithelium.First,HT-29-HorHT-29-MTXcellsthataremucus- secretingcellshavebeenco-cultivatedwithCaco-2cells.Different methodsofculture(cellculturetime,Caco-2/HT-29ratio,culture medium,timeofHT-29addition,etc.)havebeendeveloped,and representative models in terms of mucus layer, P-gp-mediated efflux expression and paracellular permeability have been obtained [44–46]. Moreover, these co-culture models make it possible to evaluate absorption enhancers and mucoadhesive systems onthe permeability of drugs [42].Moreover, Raji cells have been added to Caco-2 cells to take into account human follicle-associated epithelium (FAE), which represents lessthan 1%ofthetotalintestinalsurfacebutshowsanimpressivepropen- sitytotranscytosesmallinertparticlessuchasnanoparticlesand largemoleculesfromthelumentothelymphocytes.Finally,triple co-culturecellmodelswithCaco-2,HT-29andRaji Bcellshave beenputforwardtoobtainamorephysiological,functionaland reproducibleinvitromodel[50,51].Antunes etal.demonstrated thatthistripleco-culturecellmodelisthemostefficientmodelin predicting insulinpermeability,ascompared withpermeability valuesobtainedfromexvivoexperiments[50].
Toenhancetherelevanceofmodelsothermodificationshave been put forward to improve correlations between absorbed drugsinvitroandinvivo.Forexample,3Dinvitromodelshave beenproposedtoreproduceintestinalvilli[52]andepithelial–
stromal interactions [53]. Moreover, studies have been per- formed with simulated gastrointestinal fluid apposed on the apicalsideto mimicluminalconditionsinthegastrointestinal tract, but no effect of simulated media was demonstrated as comparedwithaclassicalmedium[54,55].Themaincharacter- isticsofeachmodelaresummarizedinTable3.Fromthesedata, twocellularmodelsseem ofgreatinterestowingto theirsimi- laritywithhumanpermeability.First,theCaco-2cellcloneTC-7 improves significantly the initial Caco-2 model in terms of transporterexpression andmetabolism,but stillshows impor- tantinter- andintra-laboratoryvariations. Second,the 2/4/A1 modeldemonstratesthepotentialtomakereliablepermeability predictions–evenmorereliablethanCaco-2TC-7cells.Unfor- tunately,therearescarcedataavailableconcerningthismodel.
Moreover,Caco-2TC-7cellsareeasiertogrowthan2/4/A1cells, whichrequire an overexpression of the antiapoptotic protein Bcl-2tobemaintainedinculture.Moreover,passivetransportis more suitable for IEC-18 cells than Caco-2 cells, whereas no carrier transport can be determined in IEC-18 in comparison toCaco-2cells[46].Todate,noinvivocorrelationwiththeIEC- 18cellmodelhasbeenpublished.
Consequently, although all of these cellular models show goodormoderatecorrelationswithhumanpassivelyabsorbed drugpermeability, correlationswith activelytransporteddrugs arevariableandmainlylow.Thisloweractivetransportobtained withtheCaco-2 cellmodels couldbe explained either bythe underexpressionofcarrier-mediatedtransportersinCaco-2cells whencomparedtoinvivoorbythesaturationofthecarriers[56].
However,evenifinvivocorrelationisslowintheCaco-2cells, this is an interesting model to determine the drug transport mechanism,andidentifyingtherelevantcarrierusedandactive transportmechanismneedtobeextensivelystudied[57].With regardtotheseadvantagesanddrawbacks,thesemodelsmight be used to study several passively transported drugs and to predict, in some but not all cases, carrier-mediated transport ofdrugs.
Finally,becausedrugpermeabilitycanalsoberelatedtopassive diffusion,onestatic model,parallelartificialmembrane(PAM- PA),wasusedtostudythistransport.Thismodelinvolvesadding amixtureofphospholipidsandorganicsolventontoaporous hydrophobicfiltersupporttoformalipidmembrane.Thereare differenttypesofPAMPAmodelsbasedonthenatureofthefilter, lipidsand transportmedia used [58]. Severalfactors influence PAMPApermeabilityperformance,suchasincubationtempera- ture,pHconditionsandlipidmembranecomposition.Indeed, the lipids in these PAMPA modelswere mainlycommercially availablelipidsorextractedfromnaturaltissues.Differentlipid compositionsare nowavailable:PC-PAMPA (phosphatidylcho- line-PAMPA),DS-PAMPA(adodecanesolutionoflecithinmix- ture-PAMPA), DOPC-PAMPA [a dodecane solution of highly purified dioleyoylphosphatidylcholine (DOPC)-PAMPA], HDM PAMP(ahexadecanesolutionofDOPC-PAMPA)andBML-PAMP (abiomimeticlipidmembranebasedonamixtureofphosphati- dylcholine, phosphatidylethanolamine, phosphatidylserine,
8 www.drugdiscoverytoday.com
ReviewsFOUNDATIONREVIEW
phosphatidylinositol, cholesterol, 1,7-octadiene membrane- PAMPA)[59].Moreover,withcellmodelscultivatedontheTrans- well1 system, a filter separates one side containing the test molecule(donor side)fromareceiverside(initiallyfreeofthe molecule).PAMPAisahigh-throughputscreeningtechnique,but makesitpossibletostudyonlypassivepermeability.Acompari- sonofDS-PAMPAwitharatinsituclose-looptechniqueshowed anacceptablecorrelation(r2=0.87)for17fluoroquinolonedrugs [60].Similarly,atpH5.5or6.5,BML-PAMPAalsodemonstrated anacceptablecorrelation(r2=0.86)formorethan25compounds [61]. Moreover, passively transported compounds, other than acidic compounds, also demonstrate a good permeability on HDM-PAMPAatpH7.4[62].
Nevertheless, this model presents limited effectiveness and applicability owing to the absence of stirring conditions, the presence of solvent and the difficultyto reach sink conditions duringtransportstudies.Consequently,otherartificialmodelsare developed to overcome these disadvantages. The phospholipid vesicle-basedpermeationassay(PVPA)isanotherartificialmem- branemodel.Thismodelconsistsofadepositionofliposomesona filter[63].LikethePAMPAmodel,PVPAcanbeusedtostudythe transportofpassivedrugsand,owingtoanexcellentcorrelation with data obtained for the Caco-2 cell model, it represents a valuable alternativetocellmodels[64].Assuch,eventhougha goodinvivocorrelationwasobtainedwithartificialmodels(PAM- PAandPVPA),thesemodelscanonlybeusedfordrugspassively transported,comparedwithcellmodelsthatmakeitpossibleto studyalltypesofdrugtransport.Insilicoapproaches(seebelow) cannowsimulatepurepassivetransport,thuslimitingtheinterest inPAMPAs.
Dynamicmodels. Dynamicmodelshavebeenputforward to avoid the limitations of static models and potentially en- hancecorrelationwithinvivostudies.Assuch,Caco-2cellshave been cultivated on permeable filters, as previously described.
Subsequently,filtersweremountedintodiffusionsystems,such as the Ussing chamber [65] or a multicompartment model (membrane bioreactor) to simulate flow-mediated transport through the biological membrane [66]. For dynamic artificial membrane models, an impregnated membrane with a lipid mixture is inserted intoa diffusion cell connectedto a donor and receiver compartments, where liquidcirculation is main- tained using a peristaltic pump. This dynamicartificial mem- brane demonstrated an excellent correlation (r2=0.95) with permeabilitydatainhumansforhighlyabsorbedhydrophobic drugs[67].Ofcourse,suchsystemsarenotusefulwhenstudying activelytransporteddrugs.
The Ussing chamber technique was also used with excised human or animal intestinal segments. Similarly, there are two diffusion compartments between the intestinal segments. The diffusionchamberscanbefilledwithaphysiologicalbuffersolu- tion (Krebs-ringer-bicarbonate buffer), which can also contain glucose,glutamate,fumarate,orwithasimulatedintestinalfluid (into the donorchamber).The system is keptat378C and the solutionisconstantlygassedwithoxygenandcarbondioxideto maintaintissueviabilityandtocreateafluidmovement[65].This modelprovidesagoodpredictionforintestinaldrugpermeability, interaction with efflux transporters and drug metabolism[68].
Moreover, using this model, permeability could be determined TABLE3 Characteristicsofmonoculturemodels. CellslinesCaco-2/TC-7[43]2/4/A1[47]IEC-18[48]MDCK[104] OriginDerivedhuman coloncellsRatfetalintestinal epithelialcellsRatfetalintestinal epithelialcellsDogkidney epithelialcells MorphologiesPolarizedmonolayerswith tightjunction,brush borderandapicalmicrovilli
Polarizedmonolayerswith tightjunction,brushborder andfewmicrovilli Polarizedmonolayerswith tightjunction,brush borderandapicalmicrovilli
Polarizedmonolayers withtightjunction,brush borderandapicalmicrovilli Paracellulartransport(TEERvalues)Underpredicted (highervaluesofTEER)Closetoinvivo (TEERvaluesclosetoinvivo)Closetoinvivo(TEERvalues closetoinvivo)Closetoinvivo (TEERvaluesclose toinvivo) Passivetranscellular transportTransportofdrugwithlow permeabilityUnderpredictedClosetoinvivoClosetoinvivoUnderpredictedbut higherthanCaco-2cells Transportofdrugwithhigh permeabilityClosetoinvivoClosetoinvivoClosetoinvivoClosetoinvivo ActivetranscellulartransportCarrierandeffluxtransporter (e.g.,P-gp,MRP-1,BCRP)HighandvariableAbsenceAbsenceLow Culturetime3weeks3–4days3weeks3–4days Abbreviations:BCRP,BreastCancerResistanceProtein;MRP,MultidrugResistance-associatedProtein;P-gp,P-glycoprotein;TEER,TransepithelialElectricalResistance.
www.drugdiscoverytoday.com 9 ReviewsFOUNDATIONREVIEW
at different parts of the intestine (jejunum, ileum and colon).
Indeed,dependingonthepartoftheintestineusedandtheorigin tissue, in vivocorrelation might bedifferent. Comparisonfrom invivoandexcisedratjejunalsegmentsshowedahighcorrelation (r2=0.95)fordrugstransportedbypassivediffusionwithhighor lowpermeability,whereasdrugstransportedviacarriersdisplayed certaindifferences[69].
However, this model reveals several drawbacks: it shows a relatively lowthroughput and recovery, and it requires several human or animal biopsies that: (i) are difficult to obtain (for humansamples);(ii)presentimportantintraindividualor inter- speciesvariabilities(foranimalvshuman)thuscontributingtoa poorreproducibility[70];(iii)haveaviabilitybeyond2h;and(iv) decrease the functionality of their activetransporters.Unfortu- nately,thereisnocomparisonbetweenthesemodelsandtheLoc- I-GutTMmodel.Althoughtheyseemeasiertosetup,presentfewer ethicalquestionsandcanbeusedtotestmoredrugthantheinvivo methods, they lose several physiological aspects partially (i.e., mucus, active transport, flux and microbiotic flora) or totally (i.e.,chymeandbloodenvironmentandperistalsis).Nevertheless, despitethese issues,the modelis,undoubtedly, aftertheLoc-I- GutTMapproach,thecurrentmostreliablemodeltosimultaneous- ly study passiveand active drug absorption in animals and in humans.
AnotherinterestingdynamicmodelistheTIMgastrointestinal modelTM(TNO).TheTIM system,originallydevelopedforfood digestion research, is a kind ofancestor of the body-on-a-chip approach [71]. It reveals several compartments mimicking the stomach, the smallintestineand the largeintestine.Moreover, some other important physiologicalparameters areeither inte- grated(suchasbodytemperature,peristalticmovements)orcan beparameterized(suchasacidity,enzymes,bilesalt,etc.).These interesting features make the TIM system very interesting for studyinggastricdigestionandabsorption.Nevertheless,theintes- tinalbarrierismadeofasemipermeablemembrane,whichmakes itimpossibletoreproduceinvivoprocessessuchasactivetransport andintestinalmetabolism.Otherdrawbacksremain:themodelis largeandcomplexinitsuse,despitethedevelopmentofasimpli- fiedtiny-TIM(whichmergestheduodenum,jejunumandileum intoasinglecompartment).
Where are we going?
Insilicomodels
Insilicoapproachesareextremelyappealingmainlybecausethey require less living material, consumable and personal material than classical approaches. However, they also require heavy computationalresourcesforthedifferentsimulations,whichare directly related to the accuracy of the desired model. Among insilicomodels,twoclassesemerge.Thefirstgroupofmechanistic modelsfocusesoninteractionsbetweendrugsandtheirreceptor, transporterordirectenvironment,throughmolecularmodeling (MM)orQSARapproaches.Asecondgroupofmodels,thephysi- ology-basedpharmacokinetics(PB-PK)models,integratesthebe- havior of a drug in different physiological compartments and pharmacokineticsmodeling.Atthisstage,onecaneasilyunder- standthatthechoiceofthemodeldependsontheobjectiveofthe study–mechanisticmodelsaremoresuitableforexplorationata smallscale(i.e.,passivediffusionoractivetransport,drug–drug
interactions,etc.)andPB-PKmodelsaremoresuitableforexplora- tionsonabiggerscale(behaviorofthedrugintissues,organsor systems).
Moleculardynamics.MMisatoolthatdescribestheposition ofthe particles ina system using classicalphysics.Briefly,MM considersa moleculetobea complexstructuremadeofatoms, (consideredasballs)andbonds(consideredassprings).Toobtain anaccuratemodel,theparameterdeterminationmustbebasedon results obtained from experiments or calculated by high-level quantummethods.High-levelquantummethodsrefertoquantic chemistry,which ismoreaccurate thanMM,butis completely unsuitableforbigstructures(>10000atoms)suchasproteinsand membranebilayers.
MM helps find the most stable structure from a given 3D structure(mainlyobtained bycrystallography)– thecalculation timedependsonthesizeandthenumberofthestudiedmolecules andrequirespowerfulcomputerhardware.Thesemethodsgivea realistic, butstatic, model.The evolution ofthe system can be estimatedusing molecular dynamics (MD) for timescales from 100nsto1ms.Thisisofmajorinterestwhilestudyingthebehavior ofdrugstowardmembranesorproteins(i.e.,transportersorcyto- chromes).
Tostudyproteins,thefirststepistoobtaintheirexperimental X-ray crystallography. Subsequently, MM simulations provide more-realisticconformations (i.e., the proteinin aqueous solu- tion).Finally,MDsimulations provideatomisticinsightsofdy- namic processes. More precisely, MD can help determine or confirmbindingsites,thedifferentconformationsoftransporters andtheeffectofphosphorylationorATPontheproteinconfor- mations,furtherpredictingtheeffect ofachangeintheamino acidsequence[72].
Drugmembranecrossingdependsonmanyparametersinclud- ing(i)size,(ii)chargeand(iii)lipophilicityofthemolecule.Evenif membranecrossingcanbeevaluatedbyparameterssuchaslogPor logD,anatomisticdescriptionisrequiredto fullydealwiththe mechanismsofaction.MDhelpsdeterminetheorientationand locationsofdrugsand eventheirmetabolitesinthe membrane [73].Acommonimportantissueisthecompositionofthemem- branebilayer.Moststudieshaveconsideredthemembranetobea singlephospholipid membranebilayer (mainly 1,2-dimyristoyl- sn-glycero-3-phosphocholine;DMPC).Previousstudieshavedem- onstratedinvitroandinsilicothatthisassumptionisfarawayfrom reality. The reason is twofold: the membrane bilayer contains differentlipids(suchastriglycerides,sphingomyelin,cholesterol, etc.)invariouspercentages(dependingonthecelltypeandside) butitisalsobecauseofthepresenceofproteinsembeddedinthe membrane that can interact with the drug of interest [74]. In conclusion,MDsimulations arecurrently capableofpredicting thebehaviorofdrugsinsimplelipidmembranes.Nevertheless,the numberofpublicationsexploringthefunctioningoftransporters usingaMDapproachisconsiderablyincreasing.Thenumberand typeofmembranecomponentsremainlimitingformakingmore- realisticpredictions.Thankstoaperpetualimprovementofcalcu- lationpower,newcompletemodelsshouldappear,pavingtheway towardahighlypredictiveinsilicopharmacology.
QSARapproach.QSARmethodsattempttoestablishquanti- tativerelationshipsbetweenthestructureofa moleculeand its activity.Briefly,QSARusesalibraryofmoleculeswithwell-known
10 www.drugdiscoverytoday.com
ReviewsFOUNDATIONREVIEW
structuresandactivitiesandrelatesabiologicaleffect onanew moleculewithunknownbiologicaleffects.QSARmethodsmight beuseful toolstopredictpassivedrugabsorptionorinteraction withregions ofinterestin influxor effluxproteinsinvolvedin active drug absorption [75]. The interest in QSAR methods is growingthankstotheEuropeanRegistration,Evaluation,Autho- rizationand RestrictionofChemicals (REACH)protocol,which stronglyincitestouseQSARmethodsratherthanlivingmodelsto evaluatechemicaltoxicity.
AsforMM,QSARapproachesareinadequatetodealwithhighly complexmolecules(mainlybecauseoflowpredictivepowerowing to a poor library and because of the difficulty to associate a combinationofseveralpharmacophoreswithaneffect).Although recentdevelopmentsinQSARapproachesmakeitpossibletostudy noncovalentfield(3D-QSAR)andtheensembleofligandconfigu- ration(4D-QSAR)andfurthereventoputforwardroughtoxicity predictions[76],theseapproachessufferfromalackofparameters thatdescribedrug–receptorinteractions.
Ensemble learning methods are powerful tools for SAR approaches owing to their unique advantages in dealing with smallsamplesizes,highdimensionalityandcomplexdatastruc- tures[77].Ensemblelearningmethodsareparticularlyadaptedto modeldrugpermeabilitywhenthesamplesizeissmallorwhenthe relationshipsbetweenpredictorsandthedependentvariablesare notclear. Briefly,ensemble learning is based on the computer choiceofthemostsuitablealgorithmstosolveacomplexproblem.
Inthiscase,ensemblelearninghelpschoosethebestalgorithmsto relateanactivitytoacomplexstructure.Ensemblelearningmeth- odshavenotyetbeenappliedtoabsorptionpredictionmodeling [78].
PB-PKmodeling. Theabovemethods sequentiallydescribe thetransportofdrugs.Wholekineticstudiesarestillunpredictable whenusingthepreviouslydescribedinsilicomethods.Bycontrast, PB-PK enable the simultaneous study of drug absorption and metabolism using realistic physiological models. These models aimatpredictingthetargettissuedose(s)fordifferentexposure situations and to evaluate the disposition of drugs within the body. The firststepis to obtain the animal PB-PKof the drug.
Briefly,afterdoseadministration,drug concentrationsaremea- suredineachorganofinterest.Then,thePB-PKapproachmodels thewholebodyasaclosedcompartmentwithseveralsubcompart- mentsrepresentinganorgan(i.e.,thegut,theliver,etc.)oratissue.
All these subcompartments are connected with mathematical equationssuchasrateconstants,clearance,amongothers.Simu- lationsprovidetheequivalenthumanmodelusingmathematical techniques,parameterizedwithknownphysiologicalfeaturesof theorgans inanimals and inhumans (bloodflow, organmass, enzymesactivity,etc.).Finally,PB-PKsimulationsprovidephysi- ological and pharmacokinetics insights into the behavior of a drug. More precisely, PB-PK can help to determine or confirm accumulationsitesandrates,andcancontributetothestudyof efficacyandtoxicity.
Amodelwidelyusedtostudydrugabsorptionisthecompart- mentalabsorptionandtransit(CAT)model.TheCATmodelviews thegastrointestinaltracttobeaseriesofcompartmentsruledby mathematical absorption equations. A more complete model (ACAT)hasbeenproposedbyAgorametal.[79].Themajorfeature isthe addition ofthe hepaticfirstpassmetabolismto the CAT
model.Inliterature,thereareseveralphysiologicallybasedmodels that consider other covariates developed to predict oral drug absorption. An exhaustive listhas been summarized by Huang etal.[80].To conclude, althoughthe CAT modelcanestimate accuratelytherateofdrugabsorptionandiseasilycoupledwith compartmental pharmacokinetics models, it seems limited to passivelytransporteddrugs.
Invitromodels
Cultureofhumandigestiveepithelium.Aspreviouslyseen, hugediscrepanciescanbeobservedindrugpermeabilitybetween cultured cellsandintestinalcells. Similarly,data obtainedfrom animalmodelsdonotadequatelydescribepermeabilityorabsorp- tioninhumans.Thecultureofahumandigestiveepitheliumcan help overcomethelimitationsofthesemodels. RecentlyBarker etal.isolatedstemcellsinthehumandigestiveepithelium[81].
These cells express a specific G-protein-coupled receptorcalled Lgr5.Inthe intestinalcrypts,stem cellsand Panethcellsarein closecontactandcollaborateactively.Forexample,Panethcells can secrete the epithelial growth factor (EGF) and WNT3A (a proteininvolvedinembryogenesisandoncogenesis)[82].Expos- ing these Lgr5+ cells to defined culture conditions results in perpetualstemcellness.Todate,therearenoconsensualguidelines onhowtoculturethesecells,butmostmethodsusedthefollowing growth factors: WNT-3A, R-Spondin and Noggin. WNT-3A and R-SpondinareligandsofLRP5/6-FrizzledandLgr4/5,respectively, bothofwhichactivatetheWnt–cateninpathway[83].Briefly,the Wnt pathway consists in an accumulation of b-cateninin the cytoplasm,whichfinallyentersthenucleustoactasatranscrip- tionalfactortopromotestemcellness.Nogginistheinhibitorof theBMPreceptor,whichisinvolvedincelldifferentiation.More- over,thisinhibitiontendstoleadtotheappearanceofcrypt-like structuresalongtheflanksofthevilli[84].Isolated,stemcellshave thecapacitytoself-renewanddifferentiateintoseveralspecialized intestinalcells.FromasingleLgr5+cell,Satoetal.establisheda long-termculture(>1.5years)ofintestinalepithelium[85].Con- sequently,humanstemcellgutorganoidscanbeobtainedwhen cultured inMatrigel1withsubtlechangesintheculturecondi- tions.Moreover,allthecellspresentnaturallyinthegutcanbe found inthis organoid model: Panethcells, enterocytes, enter- oendocrinecellsandgobletcells[86].
Suchepitheliumcouldprovideahighlyrelevantmodeltostudy permeability.Thismodelmightalsobeusefulfordrugscreening and tissue regeneration. Adding the chyme flux and blood fluxcouldcontributetorecreatingphysiologicalconditions.Cur- rently,thereareseverallimitations:(i)modelsarecultured3Din Matrigel1,makingdrugtransportstudiesdifficult;(ii)stemcells are commonlyobtained frompatients admittedto the surgical departmentforaboweldisease,thuslimitinginterpretation;(iii) the modeliscomplexandexpensive tosetup;and (iv)thereis little control over the morphogenesis and composition of the epithelium.
3Dmodels:organ-on-chipmodels.Several3Dmodelshave been developed, such as organ-on-chip devices or organoids.
Organoidsarestructure-likeorgansthatpresentseveraldrawbacks, whichlimitstheirusefordrugabsorptionmodeling.Giventhat theyarecellsgrownina3Dmatrix,itisdifficulttoobservethem, toinjectthedrugintothelumenwithoutalteringthemembrane
www.drugdiscoverytoday.com 11 ReviewsFOUNDATIONREVIEW
oreventoquantifythedruginthematrix.Themostpromising systemisthemicroengineeredbiomimeticsystems,whichcanbe usedtoculturekeyfunctionalunitsofhumanorgans.Microengi- neeredbiomimeticsystemsmakeitpossibletomimicepithelium–
endotheliuminterfaces,alongwithcomplexorgan-specificphys- iologicalmicroenvironmentsinasimpleandwell-controlleden- vironment (Fig. 3). For example, organ-on-chip models can reproduce gut 3D tissue architecture with the chyme and the bloodflux[87].
Microfluidic systems can generate controlled concentration gradients to be integrated with cultured intestinal cells. These biomimetic microdevices can mimic physiological gradients of drugs, oxygen, growth factors and hormones in the gut. Such modelscanoffermore-predictivemodelstostudydrugtransport.
Forexample,usingCaco-2cells,Kimetal.demonstratedthatcell geneticprofilesevolvedtowardamorereliablemodel;preliminary studieshaveshownthatsuchfluxledtheCaco-2epitheliumto form villi-like structures[88]. Moreinterestingly, theseauthors haveaddedbacteriaonthecells,thusimprovingthemodel.
Some studies have already used a microsystem approach to evaluate cell permeability[89]. These modelsare mainly made of glassor transparent polymer (i.e.,PDMS, polycarbonate and polyester),containmicrochannels,areeasytosterilizeandcom- posedofamembranesimilartotheTranswell1membrane.Cell viabilityismaintainedonthesemembranes,withdifferentculture mediumandflowrates.Similarmethodshavebeenusedtointe- grate polarized epithelium with living vascularendothelium in organ-on-chipdevicesthat reproducetissueinterfacesinorgans (mainlythelung,eye,breastandbrain)[90].Usingthisapproach, some studies showed that it is possible to generate invivo-like epithelial or endothelial tissues and to study their interactions [91].
Oneofthelimitationsofmicrosystemapproachesistheneces- sity to use different culture media, especially if more-accurate modelsmustbedeveloped(humanepitheliumandendothelium).
Moreover, their use is further restricted because they require skillsinmicrofluidicsandbiomaterials.Anotherdrawbackisthe
difficultyto performclassical cell cultureon thesedevices. For example,itisextremelychallengingto harvestorpassagecells.
Nevertheless, thesemodelsreveal severaladvantages: reproduc- ibility, highthroughput and control overphysiological factors suchasflowrates.Moreover,differenttissuescanbeused,suchas Caco-2cells, butalsoHUVECsto studythe permeabilitytothe blood or even the intestinal stem cells to reproduce a human intestinalepithelium.Suchsystemsarenotrestrictedtoaspecific organbuthavethepossibilitytointegrateseveralorgans,suchas the gut and liver, to study multiorgan interactions. This is an interestingnewchallengethatcouldpromotemore-advancedand -accuratepredictionsrelatingtodrugpermeabilityandevendrug absorption.Thisalsorevealsagreatinterestinpersonalizedmedi- cine:anintestinalbiopsygrowninsuchasystemcouldpredictthe required dose to be administrated, overcoming a part of the intraindividualvariabilityindrugresponse.Formoreinformation on organ-on-chip models, we refer the reader to the work of Skardaletal.[87].Toconclude,gut-on-chipmicrodevicesoffera morephysiologicalmodelthanclassicstaticmodelsbutarealso morecomplextosetup.
Concluding remarks
Modelingcomplexprocessessuchasdrugabsorptionisa great challenge.Infact,themorerelevantthemodelthemoredifficult it is to implement and to validate. The trend of new model developmentin pharmacology can be divided into twomain categories. First, there are the very simple, high-throughput insilicomodelsthatcanscreentheactivepharmaceuticalingre- dient(API)candidatesintheearlystagesofdevelopment.Second, therearethemoresophisticatedinvitroorexvivomodels,which canpredict(withahighaccuracy)thebioavailabilityoftheAPI and study the mechanisms of absorption and the impact of biological parameters. In silico models will continue to grow becausemathematicalprocessing ofdata becomesfasterevery day.Modelingin3Disnowperformedroutinelyandcanhelp designdrugsabletoreachahigherbioavailabilityandtodiffuseas requestedin theorganism. However, totake intoaccount the
Lumen compartment
Epithelium
Endothelium Drug
Blood compartment
Chyme-like flux Blood-like flux Porous membrane
Drug Discovery Today
FIGURE3
Proposalforacontrolledmicroenvironmenttomeasuredrugpermeabilityinagut-on-chipplatform.
12 www.drugdiscoverytoday.com
ReviewsFOUNDATIONREVIEW