Salcaprozate Sodium (SNAC) Enhances Permeability of Octreotide Across Isolated Rat and Human Intestinal Epithelial Mucosae in Ussing Chambers

Octreotide is approved as a one-month injectable for treatment of acromegaly and neuroendocrine tumours. Oral delivery of the octapeptide is a challenge due mainly to low intestinal epithelial permeability. The intestinal permeation enhancer (PE) salcaprozate sodium (SNAC) has Generally Regarded As Safe (GRAS) status and is a component of an approved oral peptide formulation. The purpose of the study was to examine the capacity of salcaprozate sodium (SNAC), to increase its permeability across isolated rat intestinal mucosae from five regions and across human colonic mucosae mounted in Ussing chambers. Apical-side buffers were Kreb's-Henseleit (KH), fasted simulated intestinal fluid (FaSSIF-V2), rat simulated intestinal fluid (rSIF), and colonic simulated intestinal fluid (FaSSCoF). The basal apparent permeability coefficient (Papp) of [3H]-octreotide was equally low across rat intestinal regional mucosae in KH, rSIF, and FaSSIF-V2. Apical addition of 20 mM SNAC increased the Papp across rat tissue in KH: colon (by 3.2-fold) > ileum (3.4-fold) > upper jejunum (2.3-fold) > duodenum (1.4-fold) > stomach (1.4-fold). 20 mM and 40 mM SNAC also increased the Papp by 1.5-fold and 2.1-fold respectively across human colonic mucosae in KH. Transepithelial electrical resistance (TEER) values were reduced in the presence in SNAC especially in colonic regions. LC-MS/MS analysis of permeated unlabelled octreotide across human colonic mucosae in the presence of SNAC indicated that [3H]-octreotide remained intact. No gross damage was caused to rat or human mucosae by SNAC. Attenuation of the effects of SNAC was seen in rat jejunal mucosae incubated with FaSSIF-V2 and rSIF, and also to some extent in human colonic mucosae using FaSSCoF, suggesting interaction between SNAC with buffer components. In conclusion, SNAC showed potential as an intestinal permeation enhancer for octreotide, but in vivo efficacy may be attenuated by interactions with GI luminal fluid contents.


Introduction
Acromegaly is a hormonal disorder that occurs due to the presence of a pituitary adenoma, resulting in chronic overproduction of growth hormone. It is an orphan disease with an annual prevalence of 30-60 cases/million people (Capatina and Wass, 2015). Treatment includes surgery and therapeutic strategies to improve co-morbidities and quality of life (Lim and Fleseriu, 2017). Peptides are highly potent and selective, and generate lower toxicity compared to small molecules, making them attractive candidates for disease treatment (Lau and Dunn, 2018). FDA-approved somatostatin peptide analogues for acromegaly include octreotide and lanreotide. The octapeptide, octreotide (molecular weight (MW) 1019 Da) is a first-generation synthetic analogue with a longer circulating t½ than native somatostatin (Pless, 2005;Yang and Keating, 2010). Injectable formulations of octreotide have been developed for acromegaly treatment over the last 30 years: Sandostatin ® and Sandostatin ® LAR depot (Novartis, Geneva, Switzerland) (Fattah and Brayden, 2017). According to a recent extensive survey however, a majority of patients with acromegaly stated that "an oral therapy and/or a treatment to avoid injection" would be preferable to current injectable analogues (Strasburger et al., 2016), whose administration is especially pain-inducing due to use of low gauge needles. A once-daily oral formulation of octreotide would therefore be attractive to patients and would improve compliance. The main challenges to address for oral delivery of peptides, however, relates to their metabolic instability in the gastrointestinal (GI) tract and very low permeability across intestinal epithelia Brown et al., 2020;Drucker, 2020).
Octreotide is cyclic and contains D-amino acids and a disulphide bridge, which confer some stability against intestinal peptidases compared to typical linear peptide candidates (Wang et al., 2015). Nonetheless, the hydrophilicity and high MW of octreotide result in low intestinal epithelial permeability and this is the main reason for its low oral bioavailability. Some delivery technologies have improved oral octreotide delivery in preclinical and clinical studies (Fattah and Brayden, 2017). An oral octreotide formulation (Mycappsa®, Chiasma Pharma, Jesusalem, Israel) uses Transient Permeability Enhancer™ (TPE™), developed by Chiasma Pharma (Jerusalem, Israel), recently received the approval of the US FDA for an octreotide capsule using Transient Permeation Enhancement™ technology for oral use in acromegaly patients (Melmed et al., 2015;"MYCAPSSA," 2020).
This technology involves formulating octreotide in an oily suspension with intestinal permeation enhancers (PE). Briefly, the capsule consists of the medium chain fatty acid PE, sodium caprylate (C 8 ) and other excipients, combined with octreotide in an oily suspension, which is then encapsulated in a pH-dependent polymeric enteric coating (Ben-Shlomo et al., 2017). In rat intestinal instillation studies, C 8 and the other excipients in TPE™ induced a transient opening of small intestinal tight junctions, a likely basis for enhancing the paracellular permeability of octreotide (Tuvia et al., 2014). Examination of the pharmacokinetics of octreotide in the TPE™ formulation from a Phase I study indicated that the oral bioavailability of Mycappsa™ was only ~0.25% (Tuvia et al., 2014). Co-formulation of octreotide with other PEs has been investigated to some extent, but none of these studies e progressed to clinical trials (Fricker et al., 1996;Fricker and Drewe, 1995;van der Merwe et al., 2004). Overall, This would suggest that there is an opportunity to further increase the oral bioavailability of octreotide using alternative formulations with more efficacious PEs than C8.
The 2019 FDA-approval of oral semaglutide (Rybelsus ® , Novo Nordisk, Copenhagen, Denmark) is likely to increase interest in oral formulations of other suitable peptide candidates (Bucheit et al., 2019). Rybelsus ® was the first glucagon-like peptide-1 (GLP-1) receptor agonist to be approved by FDA for Type 2 diabetics by the oral route and it is administered once-a-day. Even though its oral bioavailability in humans is 0.4 -1.0% ("RYBELSUS®," 2020), this is still higher than Mycappsa®The non-enteric coated tablet consists of either 7m or 14 mg semaglutide with the PE, 300 mg of salcaprozate sodium (SNAC). SNAC prevents semaglutide degradation by pepsin through modifying the pH around the tablet and by promoting permeability of monomeric semaglutide across the stomach mucosa (Buckley et al., 2018). SNAC was discovered as a lead PE in the Eligen ® technology by Emisphere (Roseland, NJ, USA), whose researchers hypothesized that it could chaperone poorly-permeable macromolecules across plasma membranes via a transcellular mode of action (Park et al., 2011). Over the past 20 years, several other mechanisms of action of SNAC were proposed on the basis that is it surfactant-like and is structurally related to salicylic acid and medium chain fatty acids Twarog et al., 2019). From a regulatory standpoint, SNAC has generally regarded as safe status (GRAS) and has also been marketed as an oral vitamin B 12 supplement (Eligen ® -vitamin B 12 , Emisphere, USA) under Medical Food rules for several years (Castelli et al., 2011).
The main aim of this study was therefore to examine the potential of SNAC to promote intestinal permeability of octreotide across rat and human tissue ex vivo. We investigated the effects of SNAC on the apparent permeability coefficient (P app ) of [ 3 H]-octreotide across isolated rat intestinal tissue mucosae from five regions and from colonic human mucosae mounted in Ussing chambers. A second aim was to develop a liquid chromatography-mass spectrometry (LC-MS/MS) method for octreotide detection to provide evidence that there was no radiolabel degradation during ex vivo fluxes of [ 3 H]-octreotide. The third aim was to assess the P app of [ 3 H]-octreotide in biorelevant simulated buffers bathing the apical sides of isolated rat and human tissue mucosae in order to determine if potential components of luminal fluids could interfere with SNAC performance. The results show that SNAC acts as a PE for the peptide with increasing efficacy in rat intestinal mucosae in the proximal-distal region direction as well as in human colonic mucosae in physiological Krebs-Henseleit (KH) buffer. However, efficacy was attenuated in biorelevant buffers, suggesting that SNAC performance in vivo may be over-predicted from standard Ussing chamber studies in KH buffer.

Media
Composition of buffers are shown in Table 1. KH buffer was prepared in deionized H 2 O at pH 7.4 with carbogen bubbling. Rat simulated intestinal fluid (rSIF) was prepared as prescribed (Berghausen et al., 2016). It was prepared by dissolving sodium taurocholate, sodium cholate hydrate, sodium chenodeoxycholic acid, and NaCl in 50 ml deionized H 2 O and stirred for approximately 30 min at room temperature. Lecithin, sodium oleate, and glyceryl monooleate were added and the mixture was sonicated for 10 min and left stirred overnight to yield a clear solution. Malenic acid and 40 ml deionized H 2 O were added and the pH was adjusted to 6.0 with 1N NaOH. It was stirred until a clear solution was obtained and finally the buffer volume was made up to 100 ml with deionized H 2 O. FaSSIF-V2 (pH 6.5) and FaSSCoF (pH 7.8) were made up according to the manufacturer's protocol (Biorelevant Ltd). The buffers were used within 48 h of preparation and were stored at room temperature.

Preparation of dissected rat intestinal tissue for Ussing chamber studies
Male Wistar rats (250-300 g, from Charles River Staffron Walden, UK and the UCD Biomedical Facility, Ireland) were used to study permeability. Rats were housed and used according to the National Institute of Health directive: "Principle of Laboratory Animal Care" (NIH publication number 1985). Approval of this project was granted by the UCD Research Ethics Committee for animal experimentation (AREC approval number 14-28-Brayden). Rats were euthanized by stunning and cervical dislocation. After laparotomy, resected duodenum (up to 10 cm from pylorus), upper jejunum (< 11 cm from pylorus), ileum (up to 10 cm above the caecum) and colon were placed in carbogenated KH buffer and dissected according to previous descriptions (Cuthbert and Margolius, 1982). Musclestripped stomach mucosae was prepared by a blistering technique (Hopkins et al., 2002).
Dissected tissues were mounted between two halves of chambers and maintained in 5 ml temperature-controlled carbogenated KH buffer on each side and with an exposed window area of 0.63 cm 2 . Baseline potential difference (PD, mV) and short circuit current (Isc, µA.cm -2 ) were measured using a DVC 4000 Voltage Clamp (WPI, Hichin, UK) by using a  to switch between open circuit-and voltage clamping to zero, respectively. Transepithelial electrical resistance, TEER (Ω.cm 2 ) was calculated by Ohm's Law (Equation 1).

Preparation of dissected human colonic mucosae for Ussing chamber studies
Human colonic specimens were sourced from surgical colon resections. Ethical approval for this study (Protocol 190/2012) was obtained from the St. Vincent's University Hospital Institutional Review Board, and it included informed patient consent. Resections were obtained from patients with suspected colorectal carcinoma. After resection, fresh normal human colonic tissue specimens were obtained from above the diseased margin as designated by a pathologist. Specimens were incubated in ice-cold phosphate buffered saline (PBS).
When the tissue arrived at laboratory within 60 min of excision, it was transferred into fresh carbogenated ice-cold KH buffer and kept on ice. Subsequently, the tissue was transferred to a dissection bowl containing carbogenated ice-cold KH buffer ( Figure 1). The tissue was pinned with the luminal side down on a dissection board. The circular and longitudinal smooth muscle of human colonic tissue was dissected from the basolateral side of mucosae (Kisser et al., 2017) using size 5 watchmaker forceps. Stripped mucosae were mounted in pre-equilibrated Ussing chambers (WPI) as described above within 2-3 h of resection.

Fluxes of octreotide across isolated rat and human intestinal mucosae
The apparent permeability (P app ) values of either [ 3 H]-octreotide (0.5μCi/ml) or unlabelled octreotide (10μM) were assessed across intestinal tissues from rat and human in the presence of SNAC. [ 3 H]-octreotide and octreotide with or without SNAC (20 mM and 40 mM) were added to the apical compartment bathing mucosae and 0.2 ml samples were taken from basolateral compartment every 20 min over 120 min. Each basolateral sample was replaced an equal volume of KH buffer. Some studies were also carried out in the presence of biorelevant media, rSIF, FaSSIF-V2, and FaSSCoF, which were added to apical compartment only, with KH retained on the basolateral side otherwise. These investigations were carried out in order to better model in vivo GI conditions that may affect drug and PE solubility and permeability. Human FaSSIF-V2 is now being widely used by researchers to incubate isolated rat intestinal mucosae in order to determine permeability and solubility in biorelevant buffers (Berghausen et al., 2016;Forner et al., 2017;Wuyts et al., 2015). Given that the bile salt concentrations are 3 mM in FaSSIF-V2 and 25 mM in rSIF, differences in in vitro permeability of octreotide might be expected as bile salts are also recognised as efficacious PEs (Berghausen et al., 2016;Brown et al., 2020). Therefore, we used both rSIF (to mimic rat luminal contents in vivo), as well as FaSSIF-V2 (to model those of the human small intestine). Previous studies demonstrated compatibility between rat isolated small intestinal tissue mucosae using FaSSIF-V2 in the apical compartment of Ussing chambers (Forner et al., 2017;Wuyts et al., 2015). Fed-state simulated intestinal fluid (FeSSIF) was not used as it is too damaging to isolated rat jejunal mucosa (Kristin et al., 2017)  where dQ/dt was the transport rate across the epithelium (mol/s), A was the surface area of the cell monolayer (0.63 cm 2 ), and C 0 was the initial concentration of [ 3 H]-octreotide or octreotide in the apical donor compartment (mol.ml -1 ) (Kisser et al., 2017).

Histology of tissue mucosae
Following the 2 h permeability experiments, rat and human tissues were removed from the chambers and fixed in 10% w/v buffered formalin for histology to evaluate potential epithelial damage. After 24 h, the tissues were embedded in paraffin wax. Tissue sections of 5 µm thickness were cut with a microtome (Leitz 1512; GMI, Minnesota, USA) and stained with haemotoxylin/eosin (H & E) and examined with a light microscope (BX43, Olympus, Tokyo, Japan).

Statistical analysis
Statistical analysis was carried out using Prism-8 ® software (GraphPad, San Diego, USA).
Results are given as mean ± standard error of mean (SEM). Student's t-tests (unpaired) and ANOVA were used for group comparisons with post hoc tests used for the latter. Statistical significance was considered present if p < 0.05.

SNAC increases the P app of [ 3 H]-octreotide across rat intestinal mucosae in KH
The effects of 20 mM SNAC in KH buffer were examined on the P app of [ 3 H]-octreotide across rat intestinal mucosae isolated from five regions: stomach, duodenum, upper jejunum, ileum, and colon. The basal P app of [ 3 H]-octreotide ranged from 1.3 x 10 -6 cm.s -1 to 2.5 x 10 -6 cm.s -1 across mucosae from each region, but they were not different from each other ( Figure   2A). The SNAC concentration was below its critical micellar concentration of 36 mM in KH buffer (Suppl. Figure 1). In stomach mucosae, apical addition of SNAC increased the P app to a small extent from 1.3 x 10 -6 cm.s -1 to 1.9 x 10 -6 cm.s -1 . In the presence of SNAC, the P app across duodenal mucosae was also marginally increased in the presence of SNAC from 2.1 x 10 -6 cm.s -1 to 3.0 x 10 -6 cm.s -1 , but neither trend was significant. Statistical increases in the P app were however, observed in upper jejunal, ileal, and colonic mucosae: from basal values of 1.9 x 10 -6 cm.s -1 to 4.2 x 10 -6 cm.s -1 , from 2.2 x 10 -6 cm.s -1 (basal) to 7.4 x 10 -6 cm.s -1 , and from 2.5 x 10 -6 cm.s -1 (basal) to 7.9 x 10 -6 cm.s -1 , respectively ( Figure 2A). There was therefore a pattern indicating increasing efficacy of 20 mM SNAC in KH buffer in the order of stomach < duodenum < jejunum < ileum < colon ( Figure 2A). These results indicate regional effects of SNAC as a PE for this peptide in rat intestinal tissue with increasing efficacy observed in the proximal-to-distal direction.
TEER values in the presence of [ 3 H]-octreotide gradually declined to ~40% of the initial TEER over 120 min in tissues from the different GI regions from rats ( Figure 2B-F). These changes in TEER were similar to patterns observed in untreated rat intestinal mucosae (Stuettgen and Brayden, 2020;Twarog, 2020), so there was no indication that [ 3 H]-octreotide itself might be causing TEER reductions across mucosae from the five regions. Following apical addition of 20 mM SNAC however, TEER values were further reduced across four of the regions compared to [ 3 H]-octreotide-exposed controls, except for duodenal mucosae. The times required for significant reductions in TEER when exposed to SNAC compared to the respective tissue controls ([ 3 H]-octreotide alone) were: stomach from 20 -120 min (20% -40% reduction), jejunum at 120 min (35% reduction), ileum from 80 -120 min (30% -38% reduction), and colon from 15 -120 min (44% -83% reduction) ( Figure 2B-F).

No effects of SNAC on the P app of [ 3 H]-octreotide across rat jejunal mucosae in rSIF or FaSSIF-V2
We further investigated the permeability of [ 3 H]-octreotide across rat intestinal tissue mucosae by replacing KH buffer on the apical side with simulated intestinal fluids as indicated in section 2.2.4.
Since 20 mM SNAC induced a significant increase in the P app of [ 3 H]-octreotide across rat upper jejunal mucosae in KH buffer, we re-examined whether this effect pertained in tissues just from this region in the presence of rSIF and FaSSIF-V2 on the apical side. The basal P app for [ 3 H]-octreotide across jejunal mucosae in rSIF and FaSSIF-V2 was 1.6 x 10 -6 cm.s -1 and 1.7 x 10 -6 cm.s -1 , respectively, unchanged from that obtained in KH buffer (1.9 x 10 -6 cm.s -1 ).
However, attenuation of the permeation-enhancement effects of SNAC in KH buffer was observed in both rSIF and FaSSIF-V2. The P app of [ 3 H]-octreotide in the presence of 20 mM SNAC in rSIF and FaSSIF-V2 was 1.2 x 10 -6 cm.s -1 and 1.9 x 10 -6 cm.s -1 , respectively, not different from the control P app in these buffers. Yet these values were both lower than the P app value detected in presence of SNAC in KH buffer, (4.2 x 10 -6 cm.s -1 ) (p < 0.05) ( Figure 3A).
Since the critical micellar concentration (CMC) of SNAC in rSIF and FaSSIF-V2 was lowered to 5.6 mM and 6.3 mM, respectively, an explanation is that the free concentration was of SNAC was reduced in these buffers due to mixed micelles forming in the presence of NaTC and lecithin contributions (Suppl. Figure 1).
There was a trend of a slight gradual reduction in TEER values over 2 h in control tissues exposed to [ 3 H]-octreotide in rSIF and FaSSIF-V2 buffers. In turn, these reductions were similar to the gradual pattern for TEER reduction seen in KH buffer, indicating compatibility of jejunal tissue with these biorelevant media over this period ( Figure 3B). This result indicated compatibility of these two simulated small intestinal buffers with isolated intestinal tissue mounted in Ussing chambers (Forner et al., 2017;Jantratid et al., 2008;Wuyts et al., 2015). 20 mM SNAC did not cause additional TEER reductions in jejunal tissue in either simulated buffers, in contrast to additional reductions observed in the presence of SNAC in KH.

SNAC increases the P app of octreotide across human colonic mucosae in KH buffer
Due to difficulty in accessing sufficient human specimens from different GI regions, we were limited to investigating effects of SNAC on the P app of [ 3 H]-octreotide in human colonic mucosae, where surgeries to obtain tissue are common. Following apical addition of SNAC (20 mM and 40 mM) in KH buffer, the P app increased by 1.5-fold to 1.9 x 10 -6 cm.s -1 (NS) and by 2.1-fold to 2.7 x 10 -6 cm.s -1 (p = 0.03) respectively, compared to the control P app of 1.3 x 10 -6 cm.s -1 ( Figure 4A) for [ 3 H]-octreotide alone. Therefore, 40 mM SNAC was required to statistically achieve same effect in human colonic mucosae as seen for the rat counterpart (20 mM).
We developed an LC-MS/MS method to quantify unlabelled octreotide on the basolateral side in order to confirm the data seen with [ 3 H]-octreotide in KH buffer and to show that measurement of the radiolabel did not reflect metabolism of the link between tritium and octreotide during transit. Following addition of 40 mM SNAC, the P app of unlabelled octreotide increased to 2.3 x 10 -6 cm.s -1 compared the control basal P app of 6.2 x 10 -7 cm.s -1 , although this trend was not quite significant (p= 0.23) ( Figure 4C). The P app values for labelled and unlabelled octreotide were therefore of the same order, indicating that both sets of values were a true measure of intact peptide flux. Had tritium split from octreotide, the P app for tritium would have been orders of magnitude higher, ~0.68 x 10 -4 cm.s -1 (Twarog, 2020).
The effect of SNAC on the TEER of human colonic mucosae was investigated. The mean basal TEER was 99.6 ± 9.5 Ω.cm 2 (n=24), similar to values reported by others (Maher et al., 2009;Petersen et al., 2012). Following 120 min incubation, the mean TEER of control tissues

Attenuated effects of SNAC on the P app of octreotide across human colonic mucosae in FaSSCoF
In an attempt to better mimic the in vivo situation where the luminal side of intestinal fluid is exposed to bile salts and lecithin, we used a simulated human colonic buffer on the apical side of human colonic tissue. FaSSIF-V2 is considered suitable for the upper small intestinal region (Forner et al., 2017;Jantratid et al., 2008), so we tested the effects of SNAC on the P app of [ 3 H]-octreotide using a human colonic version, FaSSCoF. The P app of [ 3 H]-octreotide in FaSSCoF increased by 1.7-fold from 1.5 x 10 -6 cm.s -1 (mean basal) to 2.6 x 10 -6 cm.s -1 , and by 1.9-fold to 2.8 x 10 -6 cm.s -1 in the presence of 20 mM and 40 mM SNAC, respectively, but statistical significance was not reached ( Figure 5A). TEER values of control tissues exposed to [ 3 H]-octreotide in FaSSCoF were unchanged over 2 h, whereas in the presence of SNAC (20 mM, 40 mM), TEER values decreased further ( Figure 5B).

Minor changes in intestinal mucosae morphology in the presence of SNAC in KH
Histology of rat and human tissue mucosae following exposure to SNAC over 120 min was assessed ( Figure 6A and B). Rat stomach mucosae exposed to KH buffer and 20 mM SNAC were structurally intact with just minor perturbation observed ( Figure 6A). In rat duodenal and jejunal mucosae, minor cell sloughing at the tip of villi was seen in control and SNACexposed tissues. More sloughing was noticeable in rat ileal and colonic mucosae where the tips of villi were absent following exposure to SNAC ( Figure 6A). As seen in Figure 6B, human colonic mucosae in KH buffer appeared structurally intact with healthy villi present, similar to that of control tissue. Human tissues exposed to 20 mM SNAC showed minor damage to villi tips, whereas 40 mM SNAC caused tip erosion.

Discussion
PEs have potential to increase the bioavailability of poorly permeable drugs and macromolecules (Maher et al., , 2016. The approval of Rybelsus ® (oral semaglutide) for the treatment of Type 2 diabetes is rightly considered a landmark for oral peptide delivery and offers encouragement to use PE-containing formulations to be used for other peptides even though the oral bioavailability is ~1% ("Rybelsus® (oral semaglutide)," 2020). Upon closer scrutiny however, rather than opening up the oral peptide field for use with PE formulations, it restricts it to niche peptides as only highly potent peptides of reasonable MW, with acceptable stability, a long t½, and a large therapeutic index can be considered as candidates at that low and variable level of bioavailability. Perhaps such candidates will emerge through the efforts of medicinal chemistry, which is focussed on producing membrane-permeable stable macrocycles of lower MW than typical linear peptides (Naylor et al., 2017). For this study, we selected octreotide because it displays several of those promising features, although its stability in the GI tract is still relatively low despite its cyclic structure (Wang et al., 2015). Moreover, an oily suspension largely based on the PE, C 8 , was recently approved by the FDA despite an oral bioavailability of ~0.25%, so there is potential to further improve its systemic delivery. We paired octreotide with SNAC as a well- Here, we demonstrated that SNAC is efficacious at increasing octreotide flux across isolated rat and human intestinal mucosae in the standard KH physiological buffer used for ion transport and drug flux studies across epithelial mucosae mounted in Ussing chambers. This was not a particular surprise as SNAC is well-known to have capacity to increase oral absorption of several co-administered poorly permeable molecules apart from semaglutide including unfractionated heparin, cromolyn, and vitamin B 12 (Leone-Bay et al., 1998Smith et al., 2016). Yet, exact mechanism of action remains controversial, recently it has been provided a specific mechanism for SNAC in respect of close association with semaglutide in a tablet in the stomach, based on buffering against acid, protection against pepsin, and presentation of semaglutide to the gastric epithelium as a monomer (Buckley et al., 2018). In that study, it was suggested that neither replacing semaglutide with another GLP-1 analogue, liraglutide, nor replacing SNAC with another Eligen™ PE analogue sufficed to increase flux across gastric epithelial monolayers. The gastric mechanism of action of SNAC in respect of semaglutide that was demonstrated in ligated dogs does not, however, rule out SNAC effects in other GI regions with other payloads where it can exert membrane perturbation due to its surfactant structure and capacity to alter a range of epithelial intracellular parameters (Twarog, 2020 (Forner et al., 2017). Peterson et al., on the other hand demonstrated regional differences in the basal P app of [ 14 C]-mannitol and FD4, with higher permeability observed in isolated rat jejunal-than colonic mucosae (Petersen et al., 2012). In the current study and that  (Fricker et al., 1992(Fricker et al., , 1991, another possible source of variation. SNAC's effects were regional-dependent in rat tissue in KH buffer, with greater efficacy as a PE at increasing the P app of [ 3 H]-octreotide in the proximal-to-distal direction. To our knowledge, this is the first study to systematically evaluate regional differences of SNAC as a PE for octreotide across rat intestinal mucosae. A recent presentation from Merck (NJ, USA) also investigated the capacity of SNAC to increase flux of octreotide across isolated rat intestinal mucosae mounted in diffusion chambers ("Novel Approaches for Enhancing the Oral Bioavailability & Absorption of Peptides," 2020). Flux was increased by 4.8% w/v SNAC, equivalent to 66 mM, more than a 3-fold higher concentration than used here. No histology was presented in the Merck study and the intestinal region was not stated.
Nonetheless, it corroborates some of the findings of this project. In the current study, SNAC's efficacy as a PE was weakest in stomach mucosae for this payload. In Buckley et al., rat gastric mucosae were mounted in chambers and exposed to 10-30 mM SNAC. Their results showed an increase in permeability of semaglutide during an acute 10 min exposure to SNAC followed by removal of the PE-containing buffer, with efficacy gradually declining at 30 and 60 min (Buckley et al., 2018). In our study, the gastric mucosa was exposed chronically to 20 mM SNAC over 2 h, so constant exposure to SNAC in this design would favour a sustained effect on the epithelium, even if acute exposure is a more likely scenario in vivo (Buckley et al., 2018). Irrespective, there is no doubt that the efficacy of SNAC as a PE on rat stomach mucosae is markedly different between these studies: this could be down to differences in stomach dissection technique, or that SNAC is simply more efficacious with semaglutide than with octreotide in stomach mucosa. Taken together, since the stomach appears less amenable to PEs than regions lower down the GI tract, the gastric mode of action for SNAC in the oral formulation of semaglutide is likely to be built primarily upon pHbuffering and monomer induction rather than relying on stomach permeation-enhancement. Small intestinal regional absorption has in fact been demonstrated before for octreotide.
Fricker et al., showed higher uptake of octreotide from jejunum than ileum from humans and rats (Fricker et al., 1992(Fricker et al., , 1991. Despite efficacy observed for SNAC in respect of the P app of [ 3 H]-octreotide in rat distal intestinal mucosae in the current study, it still took a concentration of 20 mM to achieve it, reflecting its low potency. This is consistent with a requirement for 300 mg in the oral semaglutide tablet, a weight ratio of 21:1 (SNAC : peptide).
formulations that act locally (Bak et al., 2018). Compared to the small intestine, the colon has a lower surface area compensated by a longer transit time (20-30 h), which permits a longer time to release the material (Jones et al., 2016). The observation of increased efficacy of SNAC in lower GI intestinal mucosae is in line with other studies of PEs with isolated rat intestinal mucosae (Maroni et al., 2012;Petersen et al., 2012;Yamamoto et al., 1997). The PE, sodium caprate (C 10 ), also demonstrated superior efficacy in isolated rat colonic mucosae compared to jejunal in respect of promoting flux of macromolecule (Maroni et al., 2012;Petersen et al., 2012;Yamamoto et al., 1997). Recently, our group also showed efficacy of another PE class, piperazine derivatives, on increasing the P app of [ 3 H]-octreotide across rat colonic mucosae (Stuettgen and Brayden, 2020). It is not known why the colon seems to be especially amenable to many PEs, but it is possible that the plasma membrane lipid composition of colonic enterocytes may be different from the small intestine. A more likely explanation is that the small intestine has adapted to coping with up to 20 mM bile salts in the fed state along with challenges from nutrients, so that it should be more resilient seems intuitive. Even if the colon is arguably the most amenable site for PE to act, it is still not a very attractive option for the systemic delivery of peptides due to the variable GI transit time to reach there with coated oral dosage forms and also due to the complexity of the colonmicrobiome interaction (Bak et al., 2018). The histological observations of just mild perturbation observed with SNAC in mucosae from different GI regions from rats are in line with effects of other surfactant-type PEs (Maher et al., 2018;McCartney et al., 2019;Stuettgen and Brayden, 2020 (Forner et al., 2017). Attenuation of the efficacy of SNAC was however seen here in FaSSIF-V2 and rSIF, and to lesser extent with FaSSCoF. The lack of effect of SNAC on the P app of [ 3 H]-octreotide seen in these buffers is likely to be due to interaction of SNAC with components of simulated fluids since basal fluxes were unchanged.
Several factors might contribute to the attenuation of SNAC's effects in the simulated fluids.
The initial study showing efficacy of SNAC was carried out in rat and human intestinal tissue in KH buffer at pH 7.4, whereas both FaSSIF-V2 and rSIF were used at pH < 6.5. Even with even a small change in pH, different dissolution profiles can be obtained (Berben et al., 2019;Falavigna et al., 2019). A difference in pH might affect the solubility of SNAC (Cumming and Martin, 2000). SNAC is the sodium salt of a weak acid with a pKa of 5.01. At low pH, the solubility of SNAC decreases and the SNAC precipitates as the SNAC free acid form (Cumming and Martin, 2000). Changes in pH could also alter the degree of ionization of octreotide, but this does not seem relevant, otherwise a decrease in the basal P app would have been detected. It is also possible that a change in pH could affect the interaction between SNAC and octreotide. A decrease in pH can also alter rheology of mucus, which might increase the barrier for the peptide or PE to reach the epithelium (Berben et al., 2019;Falavigna et al., 2019).
SNAC might also be entrapped in micelles by key components of biorelevant fluids, bile salts and lecithin, thereby reducing its free concentration. Both sodium taurocholate (NaTC) and lecithin form colloidal structures in simulated intestinal fluids, while FaSSIF consists of mixed-micelles of different sizes (40-100 nm). Gradauer and colleagues also saw similar attenuation in efficacy of the non-ionic surfactant PE, tetradecyl maltoside (Gradauer et al., 2015). In that study, the enhancement capacity of the maltoside to increase permeability of FD4 was reduced in FaSSIF-V2 across Caco-2 monolayers and in rat intestinal instillations.
The authors provided evidence for entrapment of the maltoside at concentrations above its CMC in mixed micelles comprising NaTC/lecithin. Whether this occurs for SNAC is unclear as its CMC in KH buffer is 36 mM, above the 20 mM concentration that proved efficacious in rat tissue, but lower than the 40 mM concentration required for human colonic mucosae.
However, the CMC of SNAC in rSIF and FaSSIF was 5.6 mM and 6.3 mM, respectively, so it is likely that mixed micelles form in these buffers that lead to reduction in its free concentration, thereby providing a plausible explanation for the lack of effect.
Evaluation of permeability and PE effects using simulated colonic fluid with human mucosa on Ussing chambers has to our knowledge never been conducted before. To date, only dissolution of molecules has been determined in this new buffer, so it needs verification to confirm its suitability for predicting permeability in human colonic mucosae (Augustijns et al., 2014;Vertzoni et al., 2019Vertzoni et al., , 2010. However, incubating human colonic mucosae in KH buffer is well-established and yields correlation with fraction absorbed across human colon in vivo (Lennernäs, 2007 (Hess et al., 2005). Thus, the reduction in TEER induced by SNAC in the mucosae of the current study reveals very little on electrophysiology changes, permeability pathways, or the mechanism of action. It is more likely to simply reflect the mild perturbation seen in the histology images.
In conclusion, the effect of SNAC as a PE for octreotide permeability across in human and rat intestinal membrane was evaluated. Regional differences were found for the SNAC enhancement effect with the best efficacy seen in rat colon, which was confirmed for human colon. Moreover, the influence of rSIF and FaSSIF-V2 on the permeability of octreotide in presence of SNAC was investigated which resulted in attenuation of the PE effect, likely related to entrapment in micellar components of simulated fluid containing NaTC and lecithin, leading to a reduction in the free concentration that can access the plasma membrane. Finally, our results showed compatibility of simulated fluids mapped to intestinal regional mucosae in Ussing chambers. Use of biorelevant compatible buffers for ex vivo fluxes which mimic the lumen contents of intestinal regions have the potential to offer more complex mechanistic insights for how PEs act and more accurate predictions of in vivo outcomes. Whether SNAC can be formulated in tablets with octreotide to compete with the very low oral bioavailability in humans offered by TPE™ will require initial assessment in large animal studies.    unpaired t-test) comparing P app in the presence of SNAC against regional controls exposed to  rSIF, and FaSSIF-V2 in rat jejunal mucosae. * P < 0.05, one-way ANOVA followed by Dunnett's post hoc analysis comparing the P app in rSIF and FaSSIF-V2 in the presence of 20 mM SNAC to that seen in KH buffer. B. Percentile of initial TEER in the controls and in the presence of 20 mM SNAC in different buffers, * P < 0.05, two-way ANOVA followed by

Authorship contribution statement
Bonferroni post hoc analysis to compare reduction in TEER in tissue exposed to SNAC versus the control from individual regions. Data represent Mean ± SEM (n= 3-7). The dotted line indicates when a significant TEER reduction was seen at particular time point compared to the corresponding control of [ 3 H]-octreotide alone.