d‐Polyarginine Lipopeptides as Intestinal Permeation Enhancers

An estimated 285 million people were living with diabetes in 2010, and this number is expected to reach 440 million by 2030. Current treatment of this disease involves the intradermal injection of insulin analogues. Many alternative administration routes have been proposed, the oral route being the most widely studied. One of the most interesting approaches for insulin delivery is the use of permeation enhancers to increase its transport across the gastrointestinal tract (GIT). Cell‐penetrating peptides (CPPs) are a remarkable example of this family of compounds. Another alternative is the use of medium‐chain fatty acids (MCFAs) to temporally disrupt the tight junctions of the GIT, thereby allowing greater drug transport. A combination of both strategies can provide a synergistic way to increase drug transport through the GIT. In this study we evaluated the complexation of insulin glulisine, an insulin analogue administered subcutaneously or intravenously in clinical practice, with a well‐known CPP modified with the MCFA lauric acid. We prepared several formulations, examined their stability, and tested the best candidates in an intestinal cell‐based model. In particular, two compounds (C12‐r4 and C12‐r6) were found to significantly increase the transport of insulin, and therefore show promise as a new delivery system worthy of further evaluation.


Introduction
Recent research in the fieldsofb iomedicinea nd pharmacology have led to promisings trategies to treat and cure several diseases. However,d espite these huge breakthroughs, effective treatment for many others is still elusive. Although therapeutic agents achievet he desired purpose, they can cause long-term side effects. Ac lear example is the administration of insulin and its analogues,w hich provide the most convenient and effective treatments for diabetes mellitus but require multiple subcutaneous injections per day.T his administration hasm ultiple side effects, including pain, swelling and redness at the site of injection. [1] The development of other ways to administer insulin, such oral delivery,w ould provideapainless and friendlier delivery route for this protein.
Peptides are now essential tools in pharmaceutical research owing to high specificity toward their targets, low immunogenic response, and relatively affordable price. [2] Among these molecules, peptidee nhancers are widely used in biomedicine to improvet he transport of therapeutic agents across biological barriers. [3] Regarding their mechanism of action, peptide enhancers are classified into three main groups.The first comprises transcellular enhancers, mostly CPPs (Figure 1(1)) short peptides that interact with the plasma membrane, thus causing their cell uptake. Given their properties, transcellular enhancers can serve as ad rug delivery system,i ncreasing the absorption of the cargo into the desired tissue or cell type. The second group is formed by paracellular enhancers (Figure 1(2)), which interact with tight junctions( TJs). These structures are closely associated areas between two cells responsible for cell-cell adhesiona nd provide high impermeability toward substances. TJ modulators derive mainly from TJ proteins or toxins and in-An estimated 285 millionp eople were living with diabetes in 2010, and this number is expected to reach 440 millionb y 2030. Current treatment of this disease involves the intradermal injection of insulin analogues. Many alternative administration routes have been proposed, the oral route being the most widely studied. One of the mosti nterestinga pproaches for insulin delivery is the use of permeatione nhancers to increase its transport across the gastrointestinal tract (GIT). Cell-penetrating peptides (CPPs) are ar emarkable example of this family of compounds. Another alternative is the use of medium-chain fatty acids (MCFAs) to temporally disrupt the tight junctionso f the GIT,t hereby allowing greater drug transport. Ac ombina-tion of both strategies can provide as ynergistic way to increase drug transport through the GIT.I nt his study we evaluated the complexation of insulin glulisine, an insulin analogue administered subcutaneously or intravenously in clinicalp ractice, with aw ell-known CPP modifiedw ith the MCFAl auric acid. We prepared several formulations, examined their stability,a nd tested the best candidates in an intestinal cell-based model.I np articular, two compounds (C 12 -r 4 and C 12 -r 6 )w ere found to significantly increaset he transporto fi nsulin, and therefore show promise as an ew delivery system worthy of furtherevaluation. Figure 1. Schematic representation of the intestinal membranea nd the different transport mechanisms displayed by:1)transcellular peptide enhancers, 2) paracellular enhancers (TJ modulators), and 3) targeting peptides. [4] [a] Dr.J.Garcia, . clude peptides with the ability to transiently open TJs. The third and final group comprises targetingp eptides derived predominantly from phage display.T hese specific peptide enhancers direct the transporto famacromolecular agent into a specific tissue or cell type (Figure 1(3)). Awide range CPPs have been described, including those derived from venoms [5] and viruses, [6] and synthetically designed peptides such as oligoarginines. [7] Many research groups have addressed their use to transport drugs across biological barriers. Examples include arginine-richpeptides like HIV-1 TATpeptide, [8] non-natural oligoarginines such as d-octaarginine( r 8 ), [9] and amphipathic peptides such as the Drosophila antennapedia homeodomain (penetratin), among others. [10] CPPs can be used in severalwayst opromote the absorption of therapeutica gents. They can be covalently conjugated or electrostatically bound [11] to the biotherapeutic, [8] or in combination with nanoformulations. [11] Many attempts with several CPPs have been made to deliver insulin into the bloodstream by crossing the gastrointestinal barriera sa na lternative to the commons ubcutaneous administration method.O ne of the first studies in the field involved the use of ac hemically synthesized TATcovalently bound to fluorescently labelled insulin (insulin-FITC). [8] The use of TAT/insulin-FITC caused a6 -t o8fold increase in insulin transport across aC aco-2c ell monolayer (the gold-standard cellular model to simulatet he epithelial cell layer) relative to insulin-FITC alone. Af ew years later,M orishita and colleagues testedt he capacity of 10 distinct CPPs electrostatically bound to insulin to enhancec ell membrane permeation. In this regard, l-penetratin and l-pVec, followed by d-octaarginine (r 8 ), showed the best performance with regard to increasing insulinb ioavailability and low toxic effects. [12] In another study,r 8 was co-administered as ap hysical mixture (electrostatically bound) with various peptide drugs. [9] They found that the peptide drugs with highert ransport through in situ ileum loop weret hose with negative charges. Peptide drugs with an eutral charge or positive charged were not able to cross the ileum membrane. These results demonstrated that the binding affinity (electrostatic)b etween drugs and CPPs is crucial for drug absorption in the intestine. Moreover,t he bindingr atio between insulin andr 8 was ak ey factor as an increase in the ratio of CPP bound led to enhanced intestinal absorption of insulin.
Here we examined peptides, in particular CPPs, as penetration enhancers. The selected CPPs werem odifiedw ith ac ertain MCFA, reported as ap enetration enhancer. [13] The unmodified CPP weren ot included in this work because theyh ave already been widely studied. [14] Fatty acid based compoundsa re ag reat source of absorption enhancers. Despitet he potential of these compounds, some concerns have been raised regarding their toxicity. [15] Sodiumc aprate (C 10 )i np articular has been extensively studied for in vitro and in vivo studies. [16] This MCFA, as well as its homologues( C 8 ,C 12 ,C 14 ,C 16 ,e tc,), cause cytotoxicity in ac oncentration-and time-dependent manner. In spite of this, severalr eports have studied the safe concentration range in which these compounds can be used as permeatione nhancers in vitro. [17] More importantly,f ine control of the concentrationu sed has allowed the use of C10 in clinical trials. [18] Therefore, modulation of the dose of MCFAs can enhance the intestinal absorption of biotherapeutics without causingr emarkablec ytotoxicity.I na ddition, the combined effect of TJ modulator peptides and MCFAs had ag reater effect on paracellular transport, where C 14 covalently bound to the TJ modulator peptide protected it from degradation and aggregation. [19] In 2015, Zhang et al. reported as ynergistic effect when using amphiphilic lipopeptide-insulin complexes relative to r 8 alone. [20] In that case, the use of stearic acid and incorporation of glutamic acid and tryptophan increased the stabilityo ft he complex and, therefore, the transport of insulin.
Although lipopeptidesh ave the potential to increasei nsulin bioavailability through the intestinal tract, they have several drawbacks. One of the main problems is the low peptides tability in the gastrointestinal environment, which makes them susceptible to enzymatic degradation. [21] Furthermore, it has been observed that the complexes formed between CPPs and insulin show instability across the intestinal tract, possibly because of the high ionic strength in the intestinal media. In addition, at ac ertain molar ratio, insoluble aggregates are observed. [12] In this regard,m any strategies, such as the use of polymer coatings, [22] have been tested to preserve the stability of these complexes during intestinala bsorption.
In this study,f our distinct lipopeptides( Figure 2) were used to form complexes with insulin glulisine (commerciallyk nown as Apidra). Glulisine is an ew generation insulin analogue characterized by its rapid onset of action. We hypothesized that self-aggregation occurs as ar esult of the amphipathic nature of lipopeptides, which would lead to micelle formation. Lipopeptide micellization may affect the structure of CPPs and insulin andt hus cause precipitation of the complex.
In this regard, the criticalm icellec oncentration (CMC) of our lipopeptidesw as determined by isothermal titrationc alorimetry (ITC). In addition, the hydrodynamic properties of the complexes, such as size and z-potential, were measured by dynamic light scattering (DLS) in order to better understand their behavior in responset ov ariationsi nt he pH, ionic strength or composition of the medium. Moreover,w es creened various molar ratios of the four lipopeptidesa nd glulisine in order to optimize the binding efficiency and stabilityo ft he complexes. Finally,t he optimized complexes were assayed in the Caco-2/ HT-29 transport model, and the amount of glulisine transported across the cells was determined using various analytical techniques. Figure 2. Structures of the four lipopeptides used:C 12 -r 4 (n = 1), C 12 -r 6 (n = 2), C 12 -r 8 (n = 3), C 12 -r 12 (n = 5). ChemMedChem 2018ChemMedChem , 13,2045ChemMedChem -2052 www.chemmedchem.org 2018 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim

Determination of criticalm icelle concentration
Lipopeptide aggregation behavior in solution is ac rucial factor to study with respect to stabilityo ft he lipopeptide-glulisine complex.W eak complex stability caused by lipopeptide selfaggregation could lead to decreased glulisine transport across the intestinal barrier.W et herefore explored whether the formation of micelles triggerscomplex aggregation.
ITC is ac alorimetric high-precision technique that can be used to determine thermodynamic parametes associated to micelle formation. [23] In our case titratione xperiments of each lipopeptide into HBSS were performed to obtain the CMC (Supporting Information, FigureS1). As an example, variousi nteractions were observed in the lipopeptide C 12 -r 6 ( Figure S1B): peaks 1t o1 4s howed thermal effects produced by exothermic interactions, while peaks 15 from 24 showedt he process derived from endothermic interactions. The concentration obtained from the transition of peak 14 to 15 corresponds to the CMC. Corrected heat, in kJ mol À1 ,c orrespond to each injection of lipopeptidesi nto HBSS, and the heat rate is represented inversely( Figure S2). CMC and other thermodynamic parameters were calculated for all the lipopeptides. [24] The resultsf or each lipopeptide are shown in Ta ble1.
All the CMCs determined were above the maximum concentration used to form the complexes (0.48 mm was the maximum concentration used for C 12 -r 4 ). With these results, the hypothesis that concentrations above the CMC would trigger complex precipitation was discarded. Thus, lipopeptides were in their non-micellar form and electrostatici nteractions with glulisine were not altered by micellation processes.
Bile salts act as as urfactant, emulsifying dietary fats into micelles and thus facilitatingt heir digestion. As the lipopeptide structurec ontains af atty acid chain, the bile salts may cause early micellization.T os tudy variations in the lipopeptide CMCs caused by pH variation or the presence of bile salts, we prepared af asted simulated intestinal medium( FaSSIF). Again, ITC was performed to study C 12 -r 4 in FaSSIF mediuma tt he same concentration studied previously in HBSS ( Figure S3). Our results showedt hat all the titrations were exothermic and,i n contrastt ow hat was expected, the CMC was not reached in this case. Therefore, we assumedt hat an increasei nC MC for all the lipopeptides would be observed in FaSSIF.T he interac-tion of the lipopeptides with the bile salts of the media could explain this phenomenon.

Evaluation of the formulation of physicochemical properties
We studied the effect of factors such as the preparationo fg lulisine stock solution,m olar ratio between the lipopeptide and glulisine and ionic strength of the buffer on complex size, charge andt endency to aggregate.

Procedure to dissolve glulisine
Aw ell reported procedure to solubilize insulin for the formation of complexes with positivelyc harged peptides is the addition of HCl (0.1 m), followed by the addition of the desired buffer,a nd finally pH adjustment with NaOH (0.1 m). [9,20] Insulin and peptides are then mixed to form the complexes at as pecific molarratio.
In our hands, this methodology resultedi nh ighly polydisperse samples regarding particles ize (measured by DLS), thereby indicating that glulisine was not well dissolveda nd aggregation wast aking place. The same resultsw ere obtained when insulinw as first dissolved with NaOH (0.1 m), followed by pH adjustment with HCl (0.1 m).
We found that the most efficient approacht os olubilize glulisine was with aN aOH solution (0.01 m). The complexes with the lipopeptide weret hen formed.N ext, the desired buffer was added and the pH was adjusted. In addition, the pH achieved when dissolving glulisine in NaOH 0.01 m conferred the complexes extra negative charges, thus increasing electrostatic interaction efficiency with the positivelyc harged lipopeptides and resulting in samples with greater monodispersity.

Lipopeptide/glulisine molar ratio
Particle size and z-potential were measured after preparing the formulations. The effect of ar ange of molar ratios between lipopeptides and glulisine was tested on these two parameters. Lipopeptide concentration ranged from 15 mm to 480 mm while glulisine concentration remained constant at 15 mm.T he results are given in the Supporting Information (Table S1). On the one hand, for both C 12 -r 12 and C 12 -r 8 ,amolar ratio higher than 4:1r esulted in nanometric particles with an average size of 200 nm ( Figure 3A,B, respectively). However,m olarr atio of 1:1r esulted in aggregation and, consequently,p article size exceeded 2 mmi nb oth cases. On the other hand, C 12 -r 6 andC 12r 4 showed ad ifferent tendency.F or complexes formed by C 12r 6 /glulisine, molar ratios rangingfrom 8:1to1:1 resulted in particle aggregation ( Figure 3C). Nevertheless, molarr atios of 16:1 and 12:1 resulted in nanometric particles. In the case of C 12 -r 4 , completely different behavior was observed, asm olar ratios of 1:1a nd 32:1 resulted in particles with ad iameter slightly over 200 nm (Figure3D). However,m olar ratios ranging from 4:1t o 16:1 yielded particlea ggregation.
The z-potential of the nanocomplexes is another parameter indicating particlea ggregation. Lipopeptides containinga greater number of arginine residues in their sequence (C 12 -r 12 and C 12 -r 8 )y ieldedh igher z-potential values at lower molar ratios comparedw ith C 12 -r 6 andC 12 -r 4 .T hus,a tm olar ratios of 4:1a nd 8:1, C 12 -r 12 and C 12 -r 8 formed complexes with enough positive surfacec harget ob es table ( Figure 3A,B). Nevertheless, in both cases, am olar ratio of 1:1g ave al ower z-potential and, as ar esult,larger particle size. The z-potentiali sakey indicator of particle stability. [25] In many cases, values close to zero indicatep oor stabilitya sa result of weak repulsion between charges. This concept is reflected in Figure 3D,w here an almostz ero z-potential (between + 10 and À10 mV) correlatesw ith particlea ggregation. Otherwise, valuesh igher than AE 10 mV,such those corresponding to molar ratios of 32:1 and 1:1f or C 12 -r 4 resulted in nanometric complexes. Only one exception was observed for C 12 -r 12 at molar ratio of 1:1. In this case, aggregation takes place at high z-potential.S everals tudies outline the importance of the CPP concentration for the insulin associatione fficiency. [9,12] Thus at low concentrations, such as the one corresponding to 1:1, C 12 -r 12 would not bind properly to glulisine, thus causing instability and aggregation.

Ionic strength
Given that ionic strength plays akey role in the colloidals tability of electrostaticf ormulations, we studied the physicochemical properties of complexes in ar ange of buffer solutions, analysing both particle size and z-potential.F irst, HBBS was selected because it is the most commonly used buffer for in vitro intestinal models. Results showed that the high amount of salts have as trong effect on the size of the complexes ( Figure 4A).
For the complexes formed with C 12 -r 12 and C 12 -r 8 ,t he increase in the ionic strengthl ed to aggregation, thereby modifying particles ize from around 200 nm to more than 1000 nm at molar ratios (CPP/glulisine) of 8:1( Figure 4A)a nd 4:1 (Table S1).  In the case of C 12 -r 6 andC 12 -r 4 ,t here was also an increasei n the particle size but it was less than that observed for the previous complexes ( Figure 4A). In all cases,t he particles ize increasedw hen HBSS buffer was used, compared with complexes formed in aqueous solution.
Regarding the z-potential, lower values were registeredi n all the complexes assayed ( Figure 4B and Table S1), except for the molar ratio 1:1o fC 12 -r 4 ,w hich led to an increase in this parameter (Table S1).
The high ionic strength of the buffer has been reported to affect electrostatic and hydrophobic interactions. [26] Therefore, we hypothesized that HBSS decreases the repulsive electrostatic lipopeptide-glulisine interactions. There is as trongc orrelation betweenp rotein solubility and protein-protein interactions. In this regard, adecrease in electrostatic repulsion results in ad ecrease in protein solubility. [27] Complexes present positive charges (conferred by arginine residues)a nd negative charges( glulisine), while HBSS containsb othc ations and anions in solution.C onsequently,t hese ions interactw ith both lipopeptidesa nd glulisine, thus hindering the formation of the complex, as reflected by an increase in their size.
As previously mentioned, z-potential is used as ap arameter to measurep article stability in solution. The decreases in z-potential brought about by the salts are the main cause of complex aggregation owing to weak positive-positive repulsion between surface charges. However,e ven with this decrease in zpotentiala nd the evident aggregation, no visible precipitation of the complexes was observed after 2hin HBSS. In summary, the selection of an optimal ionics trength is ad ifficult issue that mustb ea ddressed by achieving an equilibrium between the deleterious effect of high ionic strength on particles ize and z-potential and the minimal salt content required to ensurec ell survival during transport assays.A lthough the use of polymers to protect the complexes could be useful,t his approach would introduce other drawbacks, such as potential polymer toxicity or long degradation times.

Caco-2/HT-29 transportassay
Caco-2c ells are the gold-standard to simulatet he intestinal barriera st hey recapitulate the morphological and functional characteristics of mature enterocytes. [28] Their ability to form a monolayer and express TJs are crucial properties that make them ideal as am odel to study glulisine transport across the intestinal epithelium. In addition, these cells express microvilli, enzymesa nd transporters that are unique to enterocytes. However,t hey presents omel imitations, such as the lack of mucus secretion.T herefore, to work with am ore realistic intestinal environment, we used co-cultures with mucus-secreting HT-29 cells. [29] The optimized formulationso ft he amphiphilic lipopeptides and glulisine were assayed in this cellular model as drug delivery system acrosst he Caco-2/HT-29 cell monolayer.T ransepithelial electrical resistance (TEER) is aq uality indicator of cell monolayer integrity as it measures the electrical resistance that this monolayer offers against electrical current. [30] In transport studies, TEER values werem easured at 0, 2, and 24 hi no rder to monitor membrane disruption.
Ac ompletely different scenario was observed for C 12 -r 6. In this case, TEER decreased after 2h at the higherm olar ratios tested ( Figure S4B);h owever,n one of the values fell below 75 %. At 24 h, only the molar ratio of 4:1m aintained TEER around1 00 %, the rest of the formulations maintained the de-creasedT EER constant.
Analysis of the TEER measurements in the Caco-2/HT-29 transport assay revealed that the complexes formed with longero ligoarginine (C 12 -r 12 and C 12 -r 8 )had amore pronounced effect on the cell monolayer resistance compared with the others. This finding was reflectedi nadecrease in TEER, with no apparent recovery at 24 h, thus indicating cell monolayer disruption.I nc ontrast, the effect of C 12 -r 6 andC 12 -r 4 on the cell monolayer was not as harmful as that of the richer arginine peptides. This observation was reflected in ar eduction in TEER, which was not below 75 %i na ny case and was recovered or maintained in most of the cases.
Comparing the same lipopeptidesb ut different molarr atios, in all cases TEER decreased when the molar ratio increased. This effect can be attributed to the higher number of MCFAs present in the complex.
We then lyophilized the samples and concentrate them 10 fold. After resuspension, samples were analysed by UPLC and transported glulisine was calculated by the following equation[Eq. (1)]: This cellular model allowed us to identify two formulations that can improvet he transport of glulisine through the intestinal barrier. C 12 -r 6 at am olar ratio of 4:1a nd C 12 -r 4 at am olar ratio of 1:1s ignificantly increased glulisine transport across the Caco-2/HT-29 monolayer.I nt he case of the C 12 -r 12 and C 12 -r 8 formulations ( Figure 5A), the amount of glulisine detected was similar to the control. In contrast, for C 12 -r 6 formulations (Figure 5B), as light increase in glulisine transport was observed in all the conditions assayede xcept for the molar ratio of 12:1. Remarkably,t he molar ratio of 4:1i ncreased the transport of glulisine by 30 %. Similar resultsw ere observed for C 12 -r 4 formulations ( Figure 5C). As light increase in glulisine transport was detected for molar ratios of 32:1 and 4:1. However,t he 1:1m olar ratio significantly increased glulisine transport up to 40 %. These results highlight the potentialo ft hese formulations. Althought he use of Caco-2/HT-29 monolayerst oe valuate intestinal permeability is widely used, [31] direct correlation with in vivo transport cannot be done. The information extracted from this model should be considered only as qualitative indication of transport. [32] These resultsr eveal that the transport of glulisine through the Caco-2/HT-29 monolayer is inverselyp roportionalt ot he number of argininer esidues in the lipopeptides used in the formulations. Furthermore, in our hands, as the molar ratio increased, glulisine transport decreased.
Ar ecent study highlighted the importance of negative and neutrals urfacec harge particlesf or diffusivityt hrough porcine intestinal mucus. [33] On the basis of finding on negative mucus glycoproteins such as the mucin [34] secreted by HT-29 cells, [35] electrostatic binding between mucin and complexes could be the plausible cause of the poor transport of the more polar positively charged formulations (those containing C 12 -r 12 and C 12 -r 8 ). In contrast, neutral and negative formulationss uch those corresponding to C 12 -r 6 and C 12 -r 4 would be electrostatically trapped to al esser extent in the mucus and could promote the transport of glulisine across the cell monolayer more efficiently.I fw ec onsider the particle size of these complexes, C 12 -r 6 :g lulisine 3715 nm, C 12 -r 4 :g lulisine 992 nm, ad isaggregation process that results in size reduction upon interaction with mucus can be proposed. These phenomenah as to be further evaluated.

Conclusions
Here we studied complexes formed by amphiphilic lipopeptides and glulisine as potential permeation enhancers. Several parameters weree xamined and optimized. As general trend, the size and homogeneity of the complexes were strongly affected by changes in pH, molar ratio and ionic strength. These changes can be causedb yt he self-assembly of the lipopeptides which can occur at concentrations below the CMC duet o their amphipathic character.T wo of the formulations tested, namely positivelyc harged C 12 -r 6 at am olar ratio of 4:1a nd negativelyc harged C 12 -r 4 at am olar ratio of 1:1, enhanced the passage of glulisine through the Caco-2/HT-29 model approximately a3 0a nd 40 %, respectively.T hese promising formulations need to be further evaluated.

Experimental Section
Lipopeptide synthesis:L ipopeptides were synthesized by Solidphase Peptide Synthesis (SPPS) following the Fmoc/tBu strategy.H -Rink Amide-ChemMatrix resin was used to obtain C-terminal amidation. Peptide elongation was performed manually using PyAOP and DIEA as coupling reagents. After the introduction of each amino acid, the Kaiser test [36] was used to ensure high coupling efficiency.F moc deprotection was performed by the addition of 20 % piperidine in DMF.L auric acid was coupled to the N-terminus using the previous strategy.T he peptides were deprotected and cleaved from the resin using the following mixture:T FA/TIS/H 2 O (95:2.5:2.5) for 4-5 h. Peptides were purified by RP-HPLC at semipreparative scale (Sunfire C 18 column (150 10 mm 5 mm, 100 , Waters), flow rate 6.6 mL min À1 using acetonitrile (0.1 %T FA)a nd H 2 O( 0.1 %T FA)a nd characterized by UPLC and UPLC-MS (Acquity UPLC BEH C 18 column (50 2.1 mm 1.7 mm, Waters) coupled to a PDA Acquity detector and SQ detector 2, flow rate 0.6 mL min À1 using acetonitrile (0.036 %T FA)a nd H 2 O( 0.045 %T FA)). All peptides were obtained with high purity (> 95 %) and stored lyophilized at À20 8C.
Determination of critical micelle concentration:T itration experiments were performed in aL ow-Volume Nano ITC (TAI nstruments). The sample cell was filled with the buffer.T he syringe contained ac oncentrated solution of the desired lipopeptide in HBSS. Each titration experiment consisted of 24 injections of 2 mLo fe ach lipopeptide into as ample cell (280 sec interval) with as tirring Figure 5. Representation of the relativetransport of glulisine across the Caco-2/HT-29 model comparedw ith glulisine alone. A) C 12 -r 12 and C 12 -r 8 formulations, B) C 12 -r 6 formulations, and C) C 12 -r 4 formulations. Dataa re expressed as the mean AE SD, n = 3; *p 0.05 (t-test). ChemMedChem 2018ChemMedChem , 13,2045ChemMedChem -2052 www.chemmedchem.org 2018 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim speed of 207 rpm. Af irst injection of 0.5 mLw as performed to avoid air bubbles. Sample cell and syringe samples were degassed for 15 min and centrifuged for 15 sec at 6000 rpm before each titration experiment. Fasted simulated intestinal fluid (FaSSIF) was also used to simulate the effect of bile salts. [37] The experimental data were analyzed by TA Instruments NanoAnalyze TM software. CMCs and other thermodynamic parameters were calculated by a Microsoft Excel macro kindly provided by Prof. Dr.S andro Keller. [24] Preparation of complexes:5 00 mLo fg lulisine (in 0.01 m NaOH) solution was placed in ag lass vial, and the same volume of lipopeptide aqueous solution was added while the solution was under magnetic stirring. After 10 min, pH was adjusted to 7.4 with HCl 0.1 M. For Caco-2/HT-29 transport assay,l ipopeptides and glulisine were mixed as previously explained. Concentrated HBSS was then added in order to obtain the desired lipopeptide/glulisine molar ratio in astandard HBSS solution.
Dynamic light scattering:T he particle size and z-potential of all the formulations were determined by DLS using aM alvern Zeta-Sizer (NanoZS, ZEN 3600, Malvern Instruments) upgraded with zpotential capability.T he equipment was adjusted to 3measures of 3r uns (10 sec per run) for size measurements and 3measures of 10 runs for z-potential. The temperature was set at 25 8Ca nd the scattering angle to 1738.D ata analysis was performed using Zeta-Sizer software.
Cellular transport model:F or the transport model, co-cultures of Caco-2 and HT-29 cells were prepared. Passages from 7t o1 1w ere used in both lines. The insert (PET,1mmp ore size, Falcon) was prepared as follows:3 00 mLo fc old Corning Matrigel (10 mLmL À1 in non-supplemented DMEM) was incubated in each insert, in sterile conditions, for 1h.M atrigel was then removed, and inserts were washed three times with non-supplemented DMEM medium. Caco-2( 90 %) and HT-29 (10 %) cells were added to each insert (300 000 cells per well) in 500 mLo fc omplete DMEM medium. Next, 1.5 mL of complete DMEM was added to the donor chamber of each well to each well in the donor side, and cells were grown at 37 8Ca nd 5% CO 2 for 21 days in order to obtain ac ell monolayer.M edium was replaced every other day.T EER was measured as ac ontrol of monolayer formation. Medium was changed every two days.
For transport studies, the acceptor (0.5 mL) and donor (1.5 mL) chambers were equilibrated with HBSS buffer for 30 min at 37 8C and 5% CO 2 .A fter this time, the lipopeptide complexes were incubated for 2h in the acceptor chamber at 37 8Ca nd 5% CO 2 .T o evaluate the effect of the different formulations on the cell monolayer,T EER was measured before sample addition and 2hand 24 h after sample addition.
Glulisine quantification:T ime 0 ,D onor and Acceptor samples were concentrated 10-fold and analyed by UPLC-MS. The gradient used was from 15 %t o6 5% acetonitrile in 2min. Glulisine retention time was 1.39 min. Ac alibration curve of glulisine was used to calculate the concentration transported across the monolayer.