Suitability and reliability of in vitro epithelial cell models for nasal and pulmonary drug permeability determination : doctoral thesis

Sibinovska, Nadica (author), Kristan, Katja (mentor), Žakelj, Simon (comentor)

The use of suitable and reliable in vitro models of the nasal and pulmonary epithelial barriers is essential for evaluation of permeability properties of drug candidates and formulations intended for nasal or pulmonary drug administration. The in vitro models have to adequately replicate the barrier properties of the nasal and the pulmonary mucosa, and it is desirable to observe similar expression of tight junction proteins and drug transporters, as well as similar transepithelial electrical resistance (TEER) values. In addition, validation of the in vitro models is needed prior to utilizing them for high-throughput permeability assessment of drug candidates. Excised tissue, primary cell cultures, and immortalized cell lines are among the available in vitro models for this purpose, with each model having its own benefits and disadvantages. However, the immortalized cell lines are usually regarded as a reasonable choice during the early research phase, since the use of primary cells or excised mucosa, although resembling the epithelial barriers in vivo more closely, are not suitable for high-throughput assessment and can introduce variability. The RPMI 2650 and the Calu-3 cell lines are of human origin (nasal and tracheo-bronchial origin, respectively), and have been proposed as possibly useful models of the nasal (both the RPMI 2650 and the Calu-3 cells) and the airway (the Calu-3 cell line) epithelia. A review of the anatomy and physiology of the nasal cavity and the tracheo-bronchial epithelium, as well as the available in vivo and in vitro models of the respective epithelial barriers is presented in the introduction of the thesis. Moreover, detailed overview of the published data on characterization of the RPMI 2650 and the Calu-3 cells has been given, as these two cell lines have been subject of extensive investigation in the past decade, with regard to the optimal cell culturing conditions, morphology and ultrastructural characterization, as well as gene, protein and functional expression of uptake and efflux drug transporters. The finding that the air-liquid (A-L) culturing interface enables more closer resemblance of the cultured immortalized cells to the in vivo nasal and airway epithelia has shaped all further research regarding the RPMI 2650 and Calu-3 cell lines. However, evaluation of the suitability of these cell lines as models of the nasal and airway epithelial barrier for drug permeability prediction using large enough set of model drugs has not been made yet. The aim of our research was to explore the suitability of the RPMI 2650 and the Calu-3 cell lines cultured at the A-L and liquid-liquid (L-L) interface as models of the nasal and the airway epithelial barriers for drug permeability determination, considering the most recent International Council for Harmonization of Technical Requirements for Pharmaceuticals for Human Use (ICH) and US Food and Drug Administration (FDA) guidelines on showing suitability of in vitro permeability methods for drug permeability classification. Moreover, after determining which culturing interface resulted in obtaining a more suitable model for drug permeability assessment, we investigated the applicability of the respective models for evaluation of the in vitro permeation of intranasally administered drugs from different marketed formulations. The first part of the thesis (Chapter 1) is focused on assessing two RPMI 2650 cell models (cultured at the A-L and L-L interface) for nasal drug permeability prediction by adopting the regulatory guidelines on demonstrating in vitro permeability method suitability. Firstly, the cell layer integrity of the two cell models was investigated through conducting permeability assays with several zero permeability markers (fluorescein-isothiocyanate dextrans (FDs) with high molecular weight), and lower permeability values were obtained for the A-L RPMI 2650 model, indicating that the cell multilayers were less leakier when cultured at the more physiologically relevant A-L interface. Negligible functional expression of the P-glycoprotein (P-gp) and Breast Cancer Resistance Protein (BCRP) efflux transporters in both models was shown by bidirectional transport studies including appropriate substrates and inhibitors of drug transporters. Additionally, bidirectional drug transport studies were carried out in order to assess the permeability of 23 model drugs form the low, moderate and high permeability category. The A-L RPMI 2650 model was able to clearly differentiate between the highly permeable drugs (Biopharmaceutics Classification System (BCS) 1 and 2) and drugs with moderate and low permeability designation (BCS 3 and 4). On the other hand, no such clear difference in the permeability could be made with the L-L model. Moreover, the model drug permeability for 12 model drugs determined in the A-L RPMI 2650 cell model correlated well (Pearson correlation coefficient (r) = 0.96) with the fully differentiated nasal epithelial model (MucilAir). We confirmed that the A-L RPMI 2650 cell model is a promising model of the nasal epithelial barrier, used for nasal drug permeability prediction. Good, but slightly lower correlations between the drug permeability of 22 model drugs in the investigated nasal cell models and those determined in the intestinal models (Caco-2 cells and isolated rat jejunum) were also established. Utilizing the same concept, we investigated the suitability of two Calu-3 cell models (A-L and L-L) for prediction of drug permeability across the airway epithelia in Chapter 2. It was shown that the two models have high barrier integrity, reflected through the very low permeability of the tested FDs with high molecular weight. Bidirectional transport studies using ATP-binding cassette (ABC) transporter substrates and inhibitors were carried out, and functional activity of P-gp, but not of BCRP was revealed. Moreover, the permeability of 22 model drugs belonging to different (low, moderate or high) permeability categories was assessed by bidirectional permeability assays, and asymmetric permeability (i.e. efflux ratio (ER) > 2) was obtained for several low and moderately permeable model drugs with both the A-L and the L-L Calu-3 model. Regardless of the cell culturing interface, the permeability of the low permeable model drugs could be readily distinguished from the permeability of highly permeable compounds, with an obvious difference in the apparent permeability coefficient (Papp) values being within the range of two orders of magnitude. Another observation from our study was that the obtained Papp values for the model drugs with different permeability properties tested across the A-L and the L-L models are generally within the same order of magnitude, with the Papp values of most of the drugs being lower for the L-L Calu-3 model. Moreover, since the Papp values determined with the two Calu-3 cell models had the same order of magnitude as the Papp values determined with the Caco-2 model, and the established correlation between the cell models was very high (r = 0.93 for the A-L Calu-3 vs. Caco-2 and r = 0.92 for the L-L Calu-3 vs. Caco-2), we were able to conclude that the Calu-3 and Caco-2 cell lines distinguish between drugs with different permeability properties in a similar way. The potential of the Calu-3 cells to be used as a Doctoral thesis Abstract 3 model of the nasal epithelium, in spite of the different anatomical source, was demonstrated by the obtained excellent correlation with MucilAir for 11 model drugs (r = 0.97 for the A-L Calu-3 vs. MucilAir), as well as by the good correlation obtained with the RPMI 2650 cell line (r = 0.95 for the A-L Calu-3 vs. A-L RPMI 2650). Although we could not claim that when it comes to the A-L and L-L Calu-3 cells, one cell model enables better differentiation between drugs belonging to different permeability categories than another, we decided to further utilize only the A-L RPMI 2650 and the A-L Calu-3 models for permeability assays of intranasally administered formulations, due to the observed presence of mucus and higher gene expression of drug transporters in the A-L Calu-3 cells, and TEER values of this model closely matching the reported TEER for rabbit airway epithelia. In Chapter 3, we utilized three different in vitro methods for nasal spray evaluation: the RPMI 2650 and Calu-3 cells cultured at the A-L interface, Transwell polycarbonate membranes with different pore size and lipid-oil-lipid tri-layer membrane in the parallel artificial membrane permeability assay (PAMPA) system. The cell lines were implemented for permeability assays of two first-generation corticosteroids, while the other two methods were additionally utilized for in vitro permeation assessment of two-second generation corticosteroids. We showed that the in vitro results for the drug permeation correlated with the results of pharmacokinetic studies of different formulations of the investigated intranasal corticosteroids and correctly predicted (non)equivalence of the nasal sprays. Thus, the three in vitro methods have potential to predict the results of bioequivalence testing of nasal spray products. In Chapter 4, we employed the A-L RPMI 2650 and A-L Calu-3 cell models for evaluation of in vitro permeation of intranasally administered drugs having local and systemic effect from solution- and suspension-based formulations, with the aim to elucidate the effects of formulations on drug permeability. The cell models were shown to be sufficiently discriminative and revealed differences in the in vitro drug permeation comparable to the in vivo bioavailability, while in only one case they showed much higher observed differences between formulations in vitro than they actually exist in vivo. Good correlation with published bioavailability data was obtained for a limited number of drugs incorporated in solution-based formulations. The two cell models were shown to be suitable for evaluation of different nasal formulations and able to detect the influence of the formulation composition on the permeability of the active drug. This research has shown that the RPMI 2650 and the Calu-3 cell lines are suitable models of the nasal and the airway epithelial barrier for drug permeability prediction. The A-L RPMI 2650 model allows better differentiation between drugs with different permeability characteristics in comparison with the RPMI 2650 cells grown at L-L interface, while the A-L and L-L Calu-3 cell models differentiate between drugs with low, moderate and high permeability designation in a similar manner. The A-L RPMI 2650 and A-L Calu-3 models are suitable for evaluation of solution- and suspension-based formulations for intranasal administration and can provide valuable information about the influence of the formulation on the permeability of the active drug. The in vitro models have the potential to correctly predict the outcome of bioequivalence testing for generic nasal formulations.

• Language: English
• Work type: Dissertation
• Typology: 2.08 - Doctoral Dissertation
• Organization: FFA - Faculty of Pharmacy
• Place of publishing: Ljubljana
• Publisher: [N. Sibinovska]
• Year: 2021
• Number of pages: 185 str.
• PID: 20.500.12556/RUL-137087
• UDC: 537.63:615.015(043.3)
• COBISS.SI-ID: 80440579
• Publication date in RUL: 01.06.2022

Secondary language

• Language: Slovenian
• Title: Ustreznost in zanesljivost in vitro epitelijskih celičnih modelov za določanje nazalne in pulmonalne permeabilnosti zdravilnih učinkovin : doktorska disertacija

Abstract

Uporaba ustreznih in zanesljivih in vitro modelov nosnih in pljučnih epitelijskih barier je ključnega pomena pri vrednotenju permeabilnostnih lastnosti potencialnih zdravilnih učinkovin in formulacij, ki so namenjene nazalni in pulmunarni aplikaciji. In vitro modeli morajo ustrezno posnemati barierne lastnosti nosne in dihalne sluznice, zaželena pa je tudi podobna ekspresija proteinov, ki tvorijo tesne stike in prenašalcev zdravilnih učinkovin kot tudi podobne vrednosti za transepitelijsko električno upornost (TEER). Pred uporabo za visoko zmogljivo testiranje permeabilnosti zdravilnih učinkovin je potrebna tudi validacija in vitro modelov. In vitro modeli, ki se uporabljajo v ta namen, so izrezana tkiva, primarne celične kulture in nesmrtne celične linije, pri čemer ima vsak model svoje prednosti in slabosti. Vendar pa so nesmrtne celične linije v zgodnji fazi raziskav običajno model prvega izbora, saj uporaba primarnih celic ali izrezane sluznice, kljub temu da bolje posnemata epitelijske bariere in vivo, ni primerna za visoko zmogljivo testiranje permeabilnosti zdravilnih učinkovin, prav tako lahko pride do vpliva na variabilnost. RPMI 2650 in Calu-3 celični liniji sta človeškega izvora (nosnega oz. traheo-bronhialnega izvora) in predstavljata potencialni celični model nosnega (RPMI 2650 in Calu-3) in dihalnega (celična linija Calu-3) epitelija. V uvodu smo predstavili pregled anatomije in fiziologije nosne votline in traheo-bronhialnega epitelija ter znane in vivo in in vitro modele epitelijskih barier. Prav tako smo podali podroben pregled že objavljenih podatkov o karakterizaciji celic RPMI 2650 in Calu-3, saj sta bili ti dve celični liniji v zadnjem desetletju predmet obsežne raziskave glede optimalnih pogojev gojenja celic, morfologije in karakterizacije ultrastrukture, pa tudi glede genskega, proteinskega in funkcionalnega izražanja prenašalcev zdravilnih učinkovin. Ugotovitev, da rast na meji zrak-tekočina (A-L) omogoča večjo podobnost gojenih nesmrtnih celic z in vivo nosnim in dihalnim epitelijem, je narekovala vse nadaljnje raziskave glede celičnih linij RPMI 2650 in Calu-3. Vendar vrednotenje ustreznosti teh celičnih linij kot modelov nosnega in dihalnega epitelija za napovedovanje permeabilnosti zdravilnih učinkovin z uporabo dovolj velikega nabora vzorcev še ni bilo narejeno. Cilj naše raziskave je bil raziskati ustreznost RPMI 2650 in Calu-3 celičnih linij, gojenih na meji A-L in L-L (tekočina-tekočina), kot modelov nosnega in dihalnega epitelija za določanje permeabilnosti zdravilnih učinkovin, pri čemer smo upoštevali najnovejše ICH (International Council for Harmonization of Technical Requirements for Pharmaceuticals for Human Use ) in FDA (US Food and Drug Administration) smernice za ustreznost in vitro permeabilnostnih metod za razvrstitev zdravilnih učinkovin glede na njihovo permeabilnost. Ko smo določili, kateri pogoj rasti vodi do primernejšega modela za testiranje permeabilnosti zdravilnih učinkovin, smo raziskali uporabnost posameznih modelov za vrednotenje in vitro permeacije intranazalno apliciranih zdravilnih učinkovin iz različnih formulacij, ki so na trgu. V prvem delu disertacije (Poglavje 1) smo v skladu z regulativnimi smernicami ovrednotili dva celična modela RPMI 2650 (A-L in L-L) za napovedovanje nazalne permeabilnosti zdravilnih učinkovin. Najprej smo preverili integriteto celične plasti obeh celičnih modelov tako, da smo izvedli teste permeabilnosti s številnimi označevalci ničelne permeabilnosti (dekstran, konjugiran s flourescein izotiocianatom (FD) z visoko molekulsko maso). Ker smo pridobili nižje permeabilnostne vrednosti za model A-L RPMI 2650, sklepamo, da so bili celični večplastni sloji manj prepustni, če smo jih gojili na fiziološko bolj relavantni A-L meji. V obeh modelih smo s pomočjo dvosmernih transportnih študij z uporabo ustreznih prenašalnih substratov in zaviralcev prikazali zanemarljivo funkcionalno ekspresijo prenašalcev, ki pospešujejo izločanje, kot sta npr. P-gp (P-glikoprotein) in BCRP (protein odpornosti za raka dojke, angl. Breast Cancer Resistance Protein). Z namenom, da bi ocenili permeabilnost 23 modelnih zdravilnih učinkovin z nizko, srednjo in visoko permeabilnost, smo izvedli dvosmerne transportne študije. Model A-L RPMI 2650 je omogočal jasno razlikovanje med visoko permeabilnimi učinkovinami (razreda 1 in 2 po biofarmacevtskem klasifikacijskem sistem (BCS)) in učinkovinami z zmerno in nizko permeabilnostjo (BCS razreda 3 in 4), česar pa model L-L ni razlikoval. Vrednosti za permeabilnost, ki smo jih določili dvanajstim modelnim zdravilnim učinkovinam s celičnim modelom A-L RPMI 2650, so dobro korelirale (Pearsonov koeficient korelacije (r = 0.96)) z vrednostmi pri popolnoma diferenciranem modelu nosnega epitelija (MucilAir). Potrdili smo, da celični model A-L RPMI 2650 predstavlja obetaven model nosnega epitelija za napovedovanje permeabilnosti zdravilnih učinkovin za nazalno aplikacijo od modela L-L. Ugotovili smo, da sta korelaciji med vrednostmi za permeabilnosti 22 modelnih zdravilnih učinkovin v A-L RPMI 2650 modelu in tistimi, ki so bile določene v črevesnih modelih (celice Caco-2 in izolirani jejunum podgan), dobri, vendar nekoliko nižji. V Poglavju 2 smo na isti način raziskali ustreznost dveh celičnih modelov Calu-3 (A-L in L-L) za napovedovanje permeabilnosti zdravilnih učinkovin skozi epitelij dihalnih poti. Ugotovili smo, da imata oba modela visoko barierno integriteto, kar je bilo razvidno iz zelo nizkih vrednosti za permeabilnost testiranih FD z visoko molekulsko maso. Z uporabo ABC (angl. ATP-binding cassette) transportnih substratov in zaviralcev smo izvedli dvosmerne transportne študije, prav tako smo določili funkcionalno aktivnost P-gp, vendar ne BCRP. Vrednosti za permeabilnost 22 modelnih zdravilnih učinkovin z različno (nizko, srednjo in visoko) permeabilnostjo smo določili z dvosmernimi permeabilnostnimi testi, asimetrično permeabilnost (i.e. efluksno razmerje (ER) > 2) pa smo določili za več nizko in srednje permeabilnih modelnih zdravilnih učinkovin z obema celičnima modeloma Calu-3. Ne glede na način rasti smo permeabilnost nizko permeabilnih zdravilnih učinkovin zlahka ločili od permeabilnosti visoko permeabilnih zdravilnih učinkovin, saj je bila očitna razlika v Papp vrednostih, ki so bile znotraj območja dveh redov velikosti. Dodatno smo ugotovili, da so bile izračunane Papp (navidezni permeabilnostni koeficient) vrednosti modelnih zdravilnih učinkovin za oba celična modela v splošnem znotraj istega velikostnega razreda, pri čemer so Papp vrednosti pri večini zdravilnih učinkovin nižje za L-L Calu-3 model. Papp vrednosti, ki smo jih določili s Calu-3 celičnima modeloma, so imele enak red velikosti kot Papp vrednosti, ki so bile določene s Caco-2 modelom. Ker je bila korelacija med celičnimi modeli zelo visoka (r = 0.93 za A-L Calu-3 in Caco-2 ter r = 0.92 za L-L Calu-3 in Caco-2), smo zaključili, da celični liniji Calu-3 in Caco-2 ločujeta med zdravilnimi učinkovinami z različno peremabilnostjo na podoben način. Kljub drugačnemu anatomskemu izvoru predstavljajo Calu-3 celice obetaven model nosne epitelne bariere, kar smo dokazali z odlično korelacijo z MucilAir modelom za 11 modelnih zdravilnih učinkovin (r = 0.97 za A-L Calu-3 in MucilAir), pa tudi z dobro korelacijo s celično linijo RPMI 2650 (r = 0.95 za A-L Calu-3 in A-L RPMI 2650). Čeprav nismo mogli zaključiti, da eden izmed Calu-3 celičnih modelov (A-L in L-L) omogoča boljše razlikovanje med zdravilnimi učinkovinami z različno permeabilnostjo, smo se odločili, da bomo za nadaljnje permeabilnostne teste formulacij za intranazalno uporabo uporabili le modele A-L RPMI 2650 in A-L Calu-3 zaradi opažene prisotnosti mukusa, višje genske ekspresije prenašalcev zdravilnih učinkovin pri A-L Calu-3 modelu ter zaradi tega, ker se TEER vrednosti omenjenjega modela zelo dobro ujemajo z objavljenimi vrednostmi TEER-a dihalnega epitelija kuncev. V 3. poglavju smo uporabili tri različne in vitro metode za vrednotenje nosnega pršila: celice RPMI 2650 in Calu-3, gojene na način A-L, polikarbonatne membrane Transwell z različno velikostjo por in trislojno membrano lipid-olje-lipid v PAMPA sistemu (test permeabilnosti s sistemom vzporednih umetnih membran) (angl. parallel artificial membrane permeability assay). Oba celična modela smo uporabili za permeabilnostne teste dveh kortikosteroidov prve generacije, medtem ko sta bili drugi dve metodi dodatno uporabljeni za in vitro vrednotenje permeacije kortikosteroidov druge generacije. Pokazali smo, da in vitro rezultati permeacije zdravilnih učinkovin sovpadajo z rezultati farmakokinetičnih študij različnih formulacij preiskovanih intranazalnih kortikosteroidov in pravilno napovedali (ne) enakovrednost nosnih pršil. To pomeni, da so omenjene tri in vitro metode obetavne v smislu napovedovanja rezultatov bioekvivalenčnega testiranja nosnih sprejev. V poglavju 4 smo za vrednotenje in vitro permeacije intranazalno apliciranih zdravilnih učinkovin z lokalnim in sistemskim učinkom v obliki raztopin in suspenzij prav tako uporabili A-L RPMI 2650 in A-L Calu-3 celična modela, z namenom, da pojasnimo vpliv različnih formulacij na permeabilnost zdravilnih učinkovin. Celična modela sta se izkazala za dovolj diskriminatorna, prav tako smo z njima zaznali razlike v in vitro permeaciji zdravilnih učinkovin glede na in vivo biološko uporabnost, pri čemer smo pri samo enem primeru opazili večje razlike med formulacijami in vitro, kot pa dejansko obstajajo v in vivo. Za nekatere zdravilne učinkovine, ki so bile v obliki raztopin, smo dobili dobro korelacijo z objavljenimi podatki o biološki uporabnosti. Oba celična modela sta izkazala ustreznost za vrednotenje različnih nosnih formulacij in sta zaznala vpliv sestave formulacije na permeabilnost zdravilne učinkovine. Ta raziskava je pokazala, da sta celični liniji RPMI 2650 in Calu-3 primerni za napovedovanje permeabilnosti zdravilnih učinkovin kot modela nosnega in dihalnega epitelija. Model A-L RPMI 2650 omogoča boljše razlikovanje med zdravilnimi učinkovinami z različno permeabilnostjo v primerjavi z L-L modelom, medtem ko celična modela A-L in L-L Calu-3 ločita med zdravilnimi učinkovinami z nizko, zmerno in visoko permeabilnostjo na podoben način. Modela A-L RPMI 2650 in A-L Calu-3 sta primerna za vrednotenje formulacij na osnovi raztopine in suspenzije za intranazalno uporabo, s pomočjo katerih lahko pridobimo uporabne informacije o vplivu formulacije na permeabilnost zdravilne učinkovine. In vitro modeli predstavljajo obetaven sistem za pravilno napovedovanje izida bioekvivalenčnih testiranj za generične nosne formulacije.

• Keywords: permeabilne lastnosti, celična linija Calu-3, celični modeli, celični model RPMI 2650