The potential of porcine ex vivo platform for intestinal permeability screening of FcRn-targeted drugs

Abstract

Conventionally, the intestinal permeability of drugs is evaluated using cell monolayer models that lack morphological, physiological and architectural features, as well as realistic neonatal Fc receptor (FcRn) expression. In addition, it is time-consuming, expensive and excessive to use a large number of mice for large-scale screening of FcRn-targeted candidates. For preclinical validation, it is critical to use suitable models that mimic the human intestine; the porcine ex vivo model is widely used for intestinal permeability studies, due to its physiological and anatomical similarities to humans. This study intended to analyze the potential to measure the intestinal permeability of FcRn-targeted substances using a porcine ex vivo platform, which is able to analyze 96 samples at the same time. In addition, the platform allows the screening of FcRn-targeting substances for transmucosal delivery, taking into consideration (cross-species) receptor-ligand binding kinetics. After analyzing the morphology of the porcine tissue, the FcRn expression across the gastrointestinal tract was verified. By studying the stomach, duodenum and jejunum, it was demonstrated that FcRn expression is maintained for up to 7 days. When evaluating the duodenum permeability of free engineered human albumin variants, it was shown that the variant with the mutation K573P (KP) is more efficiently transported. Given this, the porcine ex vivo platform was revealed to be a potential model for the screening of FcRn-targeted oral drug formulations.

Graphical Abstract

This study intended to analyze the FcRn potential in intestinal permeability analysis, using the porcine ex vivo platform.

1. Introduction

After the development of new pharmaceutical and nutritional products for oral administration, there is the need to understand their permeability in the small intestine before proceeding to further studies. A suitable model with reliable predictability is crucial and of high interest for understanding the effects of substances in humans. Currently, the Caco-2 cell line, derived from a human colorectal adenocarcinoma, is the gold standard model to validate the intestinal permeability of oral drugs. However, it is known that this 2 dimensional model has some drawbacks: it lacks morphological, cellular heterogeneity and thus physiological features of a human intestine, the expression levels of drug transporters and metabolizing enzymes are different from in vivo observations, and the tight junctions of the monolayer are more efficient at preventing leakage [1]. These differences are reflected in the correlation of drug permeability data with humans: the (Spearman’s rank) correlation coefficient between the human GI tract and the Caco-2 cell line has been calculated to be around 0.3 [2]. Besides Caco-2 cells, human colon carcinoma cells (T84) and human colorectal adenocarcinoma cells (HT29) also express the neonatal Fc receptor (FcRn). In addition, a Madin-Darby canine kidney (MDCK) cell line has been modified to overexpress human FcRn (hFcRn): the MDCK-hFcRn cell line [3]. The MDCK cell line, which has a faster culture time, is also used for intestinal permeability studies; however, it has been shown that using the MDCK and Caco-2 cell lines can lead to different permeability values for the same substance, due to heterogeneity of the cell lines as well as variations in the metabolic and transport properties of the cells [4]. In addition, the level of FcRn expression is higher in the MDCK cell line compared to human intestinal cell models [3]. This shows that current widely used models are not able to provide accurate and reliable results because they do not represent all the intestinal cells types, including the intestinal layers that are responsible for the absorption process. The use of these models can then lead to misleading results in the first stage of drug development, which be time and resource consuming in subsequent phases of drug validation. In this sense, ex vivo models can be an effective alternative because they can mimick the biology, architecture, and mechanics of the human intestine as well as FcRn expression.

FcRn has been highly investigated as a target of delivery systems, since it is expressed by mucosal epithelial cell layers, which helps oral delivery drugs to overcome the intestinal barrier and enhance their bioavailability. FcRn enables its ligands, IgG and albumin, to escape lysosomal degradation and extends their half-life through the recycling pathway [5–7]. Given the recent active development of engineered IgG and albumin variants [8–10], as well as FcRn binding peptides (FcBPs) [11, 12] and their conjugation to nanosystems as a gateway for more efficient delivery [13–17], there is a need for a tool to validate and screen FcRn-targeted candidates. In addition, a suitable model for preclinical validation will reduce in vivo testing, which is time-consuming, expensive and excessive when using a large number of mice for large-scale screening of FcRn-targeted candidates.

Due to the physiological and anatomical similarities to humans, the porcine model has increasingle been applied to GI permeability evaluation. These similarities are reflected in the (Spearman’s rank) correlation coefficient of drug permeability between ex vivo porcine models and humans, which, in a previous study, was calculated to be 0.9 [2]. In addition, the pig is considered a suitable model to perform pharmacokinetic (PK) studies of human serum albumin (HSA), since porcine serum albumin (PSA) provides minimal competition for binding to FcRn [18]. Plus, it has been used as a potential model for transepithelial delivery of FcRn-targeted proteins [19].

In short, this work intends to evaluate an ex vivo platform, using porcine tissue, for screening the intestinal permeability of FcRn-targeted drugs, while also considering (cross-species) receptor-ligand binding kinetics.

2. Experimental section

2.1. Tissue dissection and cultivation

All animal procedures were conducted in accordance with protocols approved by the Massachusetts Institute of Technology Committee on Animal Care in compliance with federal, state, local and institutional regulations, including the Animal Welfare Act and Public Health Service Policy. Small sections of intestinal tissue were isolated from freshly procured intact GI tracts from pigs from selected local slaughterhouses. A section of the tissue was cut out of the GI tract, dissected longitudinally and the outer muscle layer and serosa were removed. For standard transport experiments, duodenum tissue was used. The tissue was washed in a series of saline solutions under sterile conditions. The tissue was then either mounted on the porcine ex vivo platform or maintained in culture. For cultivation, the media advanced DMEM/F-12 (Life Technologies, 12634028) was used and supplemented with 10% fetal bovine serum (FBS) and 5% antibiotic–antimycotic solution (Thermo Fisher Scientific, 15240062). Tissue was cultured in cell culture medium in a sealed polystyrene container that was then stored in an incubator kept at 37°C (no supplement gas was provided). A set of experiments was conducted with fixed time points as well as longer-term experiments monitoring the tissue daily, with analyses performed at varying time points.

2.2. Immunohistochemistry

Sections of the porcine gastrointestinal (GI) tract were put into cassettes, fixed with 10% (v/w) formalin for 24 h at 4°C and then placed in storage in 70% ethanol at 4°C. Paraffin sections (5 μm) on glass slides were dried at 70°C for 30 min, deparaffinized and underwent heat-induced epitope retrieval (HIER) at 97°C for 20 min in citrate buffer at a pH of 6. These procedures were completely automated by using a ThermoFisher IHC Autostainer 360. Briefly, for immunohistochemistry (IHC), endogenous peroxidase blocking was done for 10 min followed by the protein block for 30 min. Primary antibody (1:100 anti-FCGRT, abcam, ab193148) was incubated for 1 h and the secondary antibody (1:2000 goat anti-rabbit IgG, abcam, ab205718) for 30 min. Both antibodies were diluted with Tris Buffered Saline with Tween 20 (TBST) and incubated at room temperature. Finally, an incubation of 5 min with 3,3′-Diaminobenzidine (DAB) was done. The stained cells were then mounted on a cover slide using ConsulMount (Thermo Fisher, 9990440). Representative images using Koehler illumination were obtained with an Olympus Microfire digital camera (M/N S97809) attached to an Olympus BX60 microscope.

2.3. RT–PCR analysis

The FcRn expression of ex vivo cultured porcine-derived intestinal tissue was determined by reverse transcription polymerase chain reaction (RT–PCR) analysis and detected by agarose electrophoresis. Initially, the tissue was removed from fresh GI, frozen in liquid nitrogen and stored at −80°C. Then, total RNA from each tissue sample was extracted and purified with Quick-RNA plus (Zymo Research) followed by reverse transcription into complementary DNA using a High-Capacity cDNA Reverse Transcription Kit (ThermoFisher Scientific). Target genes were amplified using the reported primers (GAATTGGGCCCCGACAAT and CTGCTCACCCTTCCACTCC) [20] and Platinum^™ SuperFi^™ DNA Polymerase (ThermoFisher Scientific ref.: 12351010). The PCR program was optimized and consisted of an enzyme activation step (98°C, 10min), forty cycles of denaturation (98°C, 30 sec), annealing extension (62°C, 1 min), extension (72°C, 45 sec), followed by an extension at 72°C for 10 min. The expression level of FcRn was measured with 1.5% agarose gel electrophoreses, with SYBR Safe DNA gel stain (Thermofisher, S33102) and 100 bp DNA Ladder (New England BioLab, N3231S) and detected using the Bio-Rad Image suite.

2.4. PLGA-PEG NPs production

Engineered albumin-conjugated and insulin-encapsulated poly(lactic-co-glycolic)-poly(ethylene glycol) (PLGA-PEG) nanoparticles (NPs) were produced by following the procedures previously described [17]. Briefly, NPs were produced through a modified solvent emulsification-evaporation method based on a w/o/w double emulsion technique. It used 100 mg of polymer, in which 95% was PLGA (PURASORB^® PDLG 5004A, in a 50/50 molar ratio of DL-lactide and Glycolide, kindly provided by Corbion) and 5% PLGA-PEG-MAL (Akina, PolyScitech). To load the NPs, 75 mg/mL insulin solution (Sigma Aldrich) in 0.1 M of HCl was added to the polymeric solution. Finally, the conjugation was done through the covalent thioether bond between cysteine 34 (Cys34) from DI of albumin and the maleimide group of PLGA-PEG-MAL on the surface of the NP, in a molar ratio 1:0.1 of PLGA-PEG-MAL:Albumin [17]. Wild type (WT) albumin and engineered albumin variants, namely K500A/K510Q (KAHQ) and K573P (KP), were generously provided by Jan Terje Andersen from Oslo University Hospital. PSA was obtained from Sigma Aldrich.

2.5. Transport experiments using the ex vivo platform

For assembly of the ex vivo platform, freshly isolated intestinal tissue from the duodenum was prepared according to the above-described tissue dissection procedure and mounted on the manufactured interface designed with a generic 96-well plate (96-well plates, clear bottom, Corning), as previously described [2]. Then, 100 μL of Hanks’ Balanced Salt solution (HBSS), pH 7.4, was added to the bottom plate facing the basolateral surface. The duodenum was identified as the region of the small intestine approximately 1 m after the stomach. Duodenum tissue was then carefully moved over the bottom magnetic plate with the apical side facing upwards. Afterwards, the top magnetic plate was mounted, and 90 μL of 189.83 μg/mL samples was added, at the apical side. Samples were prepared at acidic pH, using HBSS buffered to a pH of 6.0 by addition of MES buffer (Sigma Aldrich). After 4 h of incubation, samples were collected from the basolateral side, with a syringe, at the bottom magnetic plate and quantified by ELISA. All experiments, including sample incubation, were performed at room temperature. Results were expressed in relative transport and apparent permeability (Papp), using the following formulas:

$$\text{Relative}\text{Transport}=\frac{\text{Amount}\text{of}\text{sample}}{\text{Mean}\text{amount}\text{of}\text{control}}$$ Equation (1)

$$\mathit{Papp}=\frac{\mathrm{\Delta }\mathrm{Q}}{\mathrm{A}\times \mathrm{C}0\times \mathrm{\Delta }\mathrm{t}}\mathrm{*}100$$ Equation (2)

where C0 is the initial concentration in the apical compartment (mg/mL), A is the surface area of the insert $\left({\mathrm{c}\mathrm{m}}^2\right)$, $\mathrm{\Delta }\mathrm{t}$ is the time during which the experiment occurred (seconds) and $\mathrm{\Delta }\mathrm{Q}$ is the amount of compound detected in the basolateral side (mg).

2.6. Albumin quantification

The quantification was done using a two-way anti-albumin enzyme-linked immunosorbent assay (ELISA). Briefly, 96 wells plates (Costar) were coated with 100 μL of a polyclonal anti-albumin antibody produced in goat (Sigma Aldrich, A1151) diluted 1:2000 in PBS by incubation overnight at 4°C. The wells were blocked with 200 μL PBSTM solution, composed by PBS, 0.005% Tween 20 and 4% skimmed milk (Acumedia), for 1 h at room temperature (RT) and then washed three times with 200 μL PBST (PBS, 0.005% Tween 20). Samples were titrated 1:2 in PBSTM, and 100 μL of each dilution was subsequently added in triplicates to wells and incubated for 1 h at RT. The wells were washed as before. Alkaline phosphatase (AP)-conjugated polyclonal anti-albumin from goat (Bethyl Laboratories, A80229AP) was diluted 1:4000 in PBSTM and 100 μL was added per well and incubated for 1 h at RT. After washing as before, visualization was performed by adding 100 μL of AP substrate (4-nitrophenyl phosphate disodium salt hexahydrate substrate) (Sigma Aldrich) diluted in diethanolamine buffer. The absorbance was measured at 405 nm using the sunrise spectrophotometer (TECAN).

2.7. Statistical analysis

Statistical analysis was performed using the GraphPad Prism Software v. 6.0 (GraphPad Software Inc). Statistical significance of the difference between the two groups was evaluated by the Student’s t-test. Differences between groups were compared using one-way analysis of variance (ANOVA) Dunnett’s multiple comparison test. Results are expressed as mean ± standard deviation and geometric means with 95% confidence intervals. The level of significance was set at probabilities of *P < 0.05, **P < 0.005, and #P < 0.0001.

3. Results and discussion

3.1. FcRn expression across GI tract

Initially, the viability and morphology of the GI tract from the porcine ex vivo model was evaluated by hematoxylin and eosin (H&E) staining to assure the maintenance of tissue integrity (Figure S1). After this, FcRn expression was confirmed by RT-PCR, after collecting the tissue, extracting the RNA and amplifying the cDNA (Figure 1). As expected, the porcine receptor is broadly expressed throughout the GI tract, as previously confirmed [19, 20].

Figure 1 |. Confirmation of FcRn gene expression across GI tract.

(A) Schematic illustration of the studied porcine organs; (B) Relative band intensity quantified through the results from C, using ImageJ; (C) Before performing the RT-PCR, RNA was extracted from fresh tissue of different sections of GI tract and cDNA was amplified. Results shown were performed with one tissue batch.

Subsequent to this analysis, only stomach, duodenum and jejunum sections were considered for further studies, due to their acidic pH; studies have shown that the duodenum has the most favorable pH for FcRn binding [21, 22]. As described, FcRn binds to its ligands in a pH-dependent manner. Briefly, the receptor binds its ligands at acidic pH (6.5–5.0) and releases them upon neutral pH. This is an important feature, since the receptor helps ligands escape lysosomal degradation and enables successful transcytosis to the basolateral side [9, 10]. Results from IHC staining showed that FcRn is expressed in stomach, duodenum and jejunum cells; this is consistent with the PCR results, which showed more intense expression in villi epithelial cells, especially in the duodenum (Figure 2 A). Since the intestinal epithelium contains enterocytes responsible for absorption, this evidence suggests that FcRn plays a role in the transportation of albumin and IgG across mucosal surfaces [23]. These results are then in line with published data [19].

Figure 2 |. Evaluation of FcRn expression over time, of stomach, duodenum and jejunum.

(A) Evaluation of FcRn expression by IHC of fresh porcine tissue. Control samples were only treated with the secondary antibody, to check non-specific binding. Brown indicates the antibody signal, while blue is the counterstain. (B) FcRn expression after 7 days of porcine ex vivo incubation or freshly isolated at day zero. Before the RT-PCR, RNA was extracted from fresh tissue and cDNA was amplified. Water and porcine blood were included as negative and positive controls, respectively. Results shown were performed with one tissue batch.

Since there are no available antibodies for porcine FcRn (pFcRn), hFcRn, which has a similar genomic structure, was used. According to literature, the pFcRn gene has 80% sequence identity to the corresponding human gene. Both genes have the single N-linked glycosylation site (NVSV) in the α domain, as well as both di-leucine- and tryptophan-based endocytosis signals present in the cytoplasmic tail. At the moment, there is no agreement about the specific amino acid differences in the FcRn structure between human and porcine species, which represents different polymorphisms [19]. However, it is known that the porcine receptor has a type 1 membrane protein with an N-terminal signal peptide and a single transmembrane domain. According to literature, pFcRn is expressed in the heart, lungs, kidneys, intestines, muscle, brain, as well as in endothelial cells and monocytes; it is also highly expressed in the liver, spleen, thymus, and lymphoid tissue, and expression is maintained in the adult stage, similar to humans [19, 20].

By studying the durability of FcRn expression of cultured tissue, it was observed that expression is maintained for at least 7 days in the duodenum (Figure 2 B). These results are in line with previous characterization of the ex vivo platform, when evaluating the expression of other receptors. In a previous study by the authors, it was verified that the complex GI architecture, mucus layer and the expression of several drug transporters is observed for at least 7 days when the tissue is maintained ex vivo in culture [2].

3.2. Transcytosis of engineered albumin variants using porcine ex vivo platform

Next, porcine duodenum was used (Figure 3 A) to evaluate the ability of the ex vivo platform to screen the intestinal permeability of FcRn-targeted drugs. For this analysis, a porcine ex vivo platform, previously developed by the authors [2], was used (Figure 3 B). In addition, engineered human albumin variants with different affinities to FcRn were used in order to evaluate their transcytosis profile. For example, K573P (KP) has a single a.a. substitution in the C-terminal domain III, which extends its serum half-life in hFcRn-expressing mice and non-human primates [10, 24]. In contrast, the K500A/H464Q (KAHQ) variant exhibits a lower binding affinity towards the receptor [10, 24]. These mutations are in the C-terminal domain III of albumin, which is the FcRn binding site, resulting in different binding affinities to the receptor. The transcytosis of these variants was previously tested by us in the MDCK-hFcRn cell line. In this previous study, it was demonstrated that KP was 2.5 times more efficiently transported than WT. In addition, when conjugated to polymeric nanoparticles (NPs), the NPs-KP demonstrated a transport rate that was 2 times higher than the NPs-WT [17]. Here, for the permeability studies, PSA was included as a control. It has previously been reported that HSA has 2.9 times more affinity than PSA towards pFcRn, which does not result in competitive binding [18].

Figure 3 |. Engineered albumin showed enhanced transcytosis capacity in the porcine ex vivo platform.

(A) Representation showing porcine duodenum, the tissue used to study the intestinal permeability, using the porcine ex vivo platform. (B) Schematic illustration of the cross-section of the interface design in a 96-well plate format. 100 μL of HBSS, pH 7.4, was added to the plate from the bottom, mimicking the basolateral side. Next, the tissue was put on the plate, facing the apical side upwards. Finally, the magnetic plate from the top was added, the tissue was pressed to avoid leakage, and samples, prepared in HBSS, pH 6, were added. (C) Relative amounts of naked albumin variants (WT, KAHQ, KP and PSA) detected in media collected at the basolateral side 4 h after adding the variants to the apical side in the porcine ex vivo platform (n=6–10); (D) Relative amounts of NPs functionalized with albumin variants (n=4–10) collected at basolateral side, as described previously. The amounts were quantified by two-way anti-albumin ELISA, where WT albumin was set to 1, as a control. These results were obtained from 3 different pigs. Shown are the arithmetic means ± SD. ∗P < 0.05, ∗∗P < 0.005, and #P < 0.0001, comparing with WT or NPs-WT by one-way ANOVA Dunnett’s multiple comparison test.

As shown in Figure 3 C, the KP human albumin variant was 2.4 times more efficiently transported compared to WT. These results confirm previously reported data related to pFcRn binding affinity, which showed that the Kd (10⁻³/s) values for HSA-WT, HSA-KP and PSA are around 15, 2.6 and 53, respectively [18]. Kd is the constant of dissociation, meaning that the higher the affinity, the lower the dissociation. The albumin quantification was done with an ELISA using human albumin antibodies, which may under reprsent PSA levels. However, it is known that PSA and HSA have 75.1% sequence identity [18]. Despite this, the results are consistent with published data: PSA was transported by pFcRn at a lower rate compared to WT (Table 1). The calculated Papp values of the free albumin variants showed that the rate of transport in this ex vivo platform was approximately 56% the rate of transport in the MDCK-hFcRn in vitro model of previously reported studies [17].

Table 1 |.

Apparent permeability coefficient calculated after 4 h of incubation in ex vivo porcine, based on the albumin amount quantified by ELISA. These results were obtained from 3 different pigs.

Formulation	Papp ± SD (x10−6cm/s)	Formulation	Papp ± SD (x10−6 cm/s)
		NPs-NF	0.66 ± 0.37 #
WT	0.83 ± 0.58	NPs-WT	5.34 ± 1.44
KAHQ	0.34 ± 0.22 *	NPs-KAHQ	2.66 ± 1.49 **
KP	1.86 ± 1.12 *	NPs- KP	3.09 ± 2.95 *
PSA	0.44 ± 0.53	NPs-PSA	0.53 ± 1.14 #

Shown are the arithmetic means ± SD.

^*P < 0.05,

^**P < 0.005, and

^#P < 0.0001, comparing with WT or NPs-WT, by Student’s t-test.

After this evaluation, albumin-functionalized NPs were also tested to further study the potential of the porcine ex vivo platform (Figure 3 D). The produced NPs were previously characterized as being a monodisperse solution having particles 150 nm in size, with a zeta potential around −9 mV and 10% loaded with insulin [17]. The typical behavior pattern between variants was observed, with the exception of the NPs-KP, which might be due to some steric hindrance of this particular albumin variant, in an ex vivo environment, when conjugated to NPs. Also, it was not expected that NPs, which are larger particles, would have higher Papp values compared to free albumin. Additionally, NPs-KAHQ and NPs-PSA had lower transport rates than expected. Also, non-functionalized NPs produced a signal, which was unexpected since the surfaces of the NPs were not decorated with albumin variants. This might be due to the presence of some possible interferents from the biologic fluids contained on the tissue. Given these findings, future studies are warranted; one improvement would be to increase the number of replicates in order to validate the experiment and obtain comparable Papp values.

Porcine tissue is often used for preclinical drug validation. One example is the InTESTine^™ system, which contains porcine intestine tissue and has been demonstrated to have more comparable Papp and transepithelial electrical resistance (TEER) values to humans than the Caco-2 and Caco-2/HT29-MTX in vitro models [25]. Also, Rani Therapeutics has evaluated the PK/pharmacodynamics (PD) of recombinant human insulin delivered by an ingestible device into the porcine jejunum [26]. However, none of these studies have evaluated the potential of using porcine tissue to screen FcRn-targeted drugs in a platform, where 96 samples can be analyzed at the same time. Here, the results demonstrate that the ex vivo platform might be a promising tool for screening the intestinal permeability of FcRn-targeted drugs.

4. Conclusions

A suitable model that can reliably predict the intestinal permeability of oral drugs is crucial and of high interest for understanding the effects of substances in humans. The porcine model has been increasingly applied towards modeling the absorption of drugs via the GI tract. Using a previously developed porcine ex vivo platform, it was possible to study the potential for evaluating FcRn-targeted substances for transmucosal delivery, while taking into consideration receptor-ligand binding kinetics. In this study, it was demonstrated that the porcine ex vivo platform had the potential to be used for screening FcRn-targeted oral drug formulations, since FcRn expression was confirmed throughout the GI tract and was maintained for up to 7 days. This was validated through an intestinal permeability assay, in which it was shown that the transportation rate of engineered human albumin variants followed their characteristic pattern. Therefore, this platform has the potential to have a positive and significant impact on the development of preclinical FcRn-targeted drugs, which could reduce research time and costs by providing more reliable data.

Highlights.

• FcRn is expressed across the gastrointestinal tract;

• FcRn expression in ex vivo porcine tissue is maintained up to 7 days in culture;

• Free KP present higher permeability in porcine ex vivo platform;

• The porcine ex vivo platform was revealed to be a potential model for the screening of FcRn-targeted oral drug formulations.

Abbreviations

FcRn

neonatal Fc receptor

WT

wild type

NPs

Nanoparticles

GIT

gastrointestinal tract

PLGA-PEG

poly(lactic-co-glycolic)-poly(ethylene glycol)

MAL

maleimide

Cys

cysteine

PSA

porcine serum albumin

HSA

human serum albumin

hFcRn

human FcRn

pFcRn

porcine FcRn

MDCK-hFcRn

Madin-Darby Canine Kidney cell line over-expressing human FcRn

KAHQ

human albumin variant with the mutations K500A and K510Q

KP

human albumin variant with the mutation K573P