Highly cationic cell-penetrating peptides affect the barrier integrity and facilitates mannitol permeation in a human stem cell-based blood-brain barrier model

The blood-brain barrier (BBB) allows passive permeation of only a limited number of, primarily lipophilic, low-molecular weight drugs that obey the so-called “ rule of CNS likeness ” . Therefore, novel strategies to facilitate drug delivery across the BBB are needed. Cell-penetrating peptides (CPPs) enable delivery of various therapeutic cargoes into cells and may potentially serve as shuttles for delivery of brain-specific drugs across the BBB. The CPPs Tat 47 – 57 and penetratin are prototypical cationic CPPs, whereas apidaecin and oncocin belong to the group of proline-rich cationic antimicrobial peptides displaying CPP-like properties. The aim of the present study was to investigate the potential of Tat 47 – 57 , penetratin, apidaecin, and oncocin for interaction with and permeation of the BBB in vitro . We also studied whether the CPPs facilitated permeation of the paracellular flux marker mannitol as well as the transcellular flux marker propranolol. The peptides were labelled with the fluorophore 6-TAMRA (T) for visualization and quantification purposes. CPP membrane-adherence, membrane-embedding, and cellular uptake as well as barrier-permeation were evaluated in murine brain capillary endothelial cells (bEND3) and human induced pluripotent stem cell-derived (Bioni-010c) brain capillary endothelial-like monolayers. The cationic and the proline-rich cationic CPPs were taken up into the Bioni-010c monolayers. T-Tat 47 – 57 , T-apidaecin, and T-oncocin also permeated Bioni-010c monolayers, whereas T-penetratin did not. However, both T-Tat 47 – 57 and T-penetratin affected the barrier integrity to a degree that facilitated permeation of 14 C-mannitol. These results may therefore pave the way for future CPP-mediated brain delivery of small drugs that do not obey the “ rule of CNS likeness ” .


Introduction
There is an unmet need for new medicines to treat brain diseases such as stroke, Alzheimer's disease, and Parkinson's disease.New brain drug candidates are in development; however, the majority of these fail due to their inability to cross the blood-brain barrier (BBB).The BBB allows passive permeation of small lipophilic drugs of which the most obey the so-called "rule of CNS-likeness" dictating a restricted physicochemical space (< 400 g/mol, < 3 hydrogen bond donors, and < 7 hydrogen bond acceptors) (Goldberg, 2011).Furthermore, they should not be substrates for efflux via P-glycoprotein.Several strategies are being pursued to facilitate BBB permeation of drugs that do not obey the "rule of CNS likeness".One strategy involves the use of cell-penetrating peptides (CPPs) as shuttles for drug delivery across the BBB.CPPs are typically short peptides capable of permeating cell membranes (Copolovici et al., 2014), and they have been exploited as shuttles for drug delivery across various biological barriers (Kristensen et al., 2016), including the BBB (Kristensen and Brodin, 2017).
Penetratin originates from the third helix of the Antennapedia homeodomain and was in 1994 the first CPP to be described (Derossi et al., 1994).Shortly thereafter, the peptide Tat 47-57 , derived from the HIV-1 transactivator of transcription, was discovered to possess capacity for translocation into cells (Derossi et al., 1996).In 1999, a pioneer study demonstrated BBB-permeating properties of Tat 47-57 in mice (Schwarze et al., 1999).Subsequently, additional in vivo studies including Tat 47-57 as brain drug shuttle have followed (Aarts et al., 2002;Bach et al., 2012).Penetratin is less explored for drug delivery across the BBB, but have shown potential for drug delivery across epithelial barriers (Kamei and Takeda-Morishita, 2015;Kristensen et al., 2015;Nielsen et al., 2014).The mechanisms by which Tat 47-57 and penetratin permeate biological barriers is not fully understood.Some studies point towards a transcellular route (Kamei et al., 2013a;Lim et al., 2015), whereas others suggest involvement of the paracellular route (M Kristensen et al., 2020;Lindgren et al., 2004;Diedrichsen et al., 2021).Tat 47-57 and penetratin belong to the group of cationic CPPs, and they are the most studied peptides in the CPP family.Short proline-rich antimicrobial peptides (AMPs) comprise another, but less studied, class of peptides, which have shown potential in cell uptake (Yamashita et al., 2016).Moreover, one study established that the proline-rich peptides apidaecin and oncocin may permeate the BBB in mice (Stalmans et al., 2014).Apidaecin and oncocin are AMPs, but due to their cell-penetrating potential, they may be considered as closely related to the traditional CPP family.
The aim of the present study was to investigate membrane association, cellular uptake, and barrier permeation of Tat 47-57 , penetratin, apidaecin, and oncocin, and to elucidate whether these peptides may serve as future shuttles for drug delivery across the BBB.We demonstrated that highly cationic CPPs displayed better membrane adherence and embedding as compared to the proline-rich AMPs.On the other hand, we observed that too high degree of membrane adherence and/or embedding may hinder the cationic CPPs from barrier permeation.Nevertheless, both of the cationic CPPs, but not the proline-rich AMPs, affected the barrier integrity, which facilitated permeation of 14 Cmannitol, which does not obey the "rule of CNS-likeness".
Human induced pluripotent stem cell line Bioni-010c: The Bioni010-C cells were maintained in matrigel coated 6-well plates in mTesR1 medium at 37 • C, 5% CO 2 , and 95% O 2 and passaged with Versene or Accutase at 80% confluency for maintainence or differentiation, rescpectively.Differentiation of the Bioni-010c cells into brain capillary endothelial-like cells was done as earlier decribed by Goldeman et al. (2020).Briefly, Bioni010-C cells were seeded onto Matrigel coated 6-well plates at a seeding density of 10 5 cells pr.well and cultured for three days in mTesR1 medium (day − 3 -day 0).Medium was changed every day.After three days of culture in mTesR1, the medium was changed to undifferentiated medium consisting of DMEM/F12, 1x MEM non-essential amino acids, 1 mM L-glutamine, 0.1 mM β-Mercaptoethanol and 20% KOSR) on day 0. The cells were cultivated for six days in undifferentiated medium with medium change every day (day 0 -day 6).On day 6, the medium was changed to endothelial cell medium +/+ P. Frøslev et al.
(hESFM, 1% PDS, 20 ng/mL bfGF, 10 μM retinoic acid and cultured for additional two days.On day 8, the cells were detached from the culture plate using Accutase and seeded for experimental use.For transport studies, the cells were seeded onto collagen coated (400 µg/mL) and fibronectin coated (100 µg/mL) polyester Transwell® inserts (model 3460, 1.12 cm2, pore size 0.4 μm) at a seeding density of 8.9 * 10 5 cells/cm 2 in a 12-well Transwell® plate.For cell viability studies, the cells were seeded into the wells of a 96-well plate (8.9 * 10 5 cells/cm 2 ).On day 9, the medium was changed to endothelial cell medium (hESFM) supplemented with 1% PDS.On day 10, the cells were ready for experimental use.

Cellular viability
The cellular viability of bEND3 cells and Bioni-010c cells grown on the bottom of a 96-well plate was evaluated after 1 h of incubation with peptide at 37 • C under orbital shaking (90 rpm).Cellular metabolic activity was assessed using a commercially available sodium 3´-[1-(phenylaminocarbonyl)− 3,4-tetrazolium]-bis (4-methoxy-6-nitro) benzene sulfonic acid hydrate (XTT) assay as directed by the manufacturer.Here 5, 25, or 50 µM T-Tat 47-57 or T-penetratin, or 50, 100, or 200 µM Tapidaecin or T-oncocin were applied to the cells.After peptide incubation, the cells were washed twice with 37 • C Hanks Balanced Salt Solution (HBSS) supplemented with 10 mM HEPES and 0.05% bovine serum albumin (BSA) and adjusted to pH 7.4 (hHBSS).Then 100 µL hHBSS (at 37 • C) were applied to the cells, followed by 100 µL XTT labeling mixture (electron-coupling reagent:XTT labeling reagent 1:50) for 1 h at 37 • C with horizontal shaking (90 rpm).Finally, 100 µL from each well were transferred to a new 96-well plate, and then absorbance was measured at 466 nm by using a SPECTROstar Nano plate reader (BMG Labtech, Ortenberg, Germany).The relative cellular viability was calculated by the following equation: Where A sample is the measured absorbance of the test sample, A Triton X-100 is the absorbance from cells incubated with 2% Triton X-100 (dead cell control), and A hHBSS is the absorbance from cells incubated with hHBSS (live cell control).

In vitro peptide plasma membrane adherence, membrane embedding, and cellular uptake in bEND3 cells
Plasma membrane adherence, membrane embedding, and cellular uptake of T-Tat 47-57 , T-penetratin, T-apidaecin, and T-oncocin were evaluated in bEND3 cells grown at the bottom of a 12-well plate following 1 h incubation with peptide (50 µM) at 37 • C with horizontal shaking (90 rpm) by using a modified protocol originally developed by Trier et al. (2014).After incubation, the cells were washed twice with ice-cold hHBSS.To isolate the plasma membrane-adhered fraction, the cells were placed on ice and incubated with 300 µL ice-cold acetic acid buffer (15 μM sodium acetate trihydrate, 10 μM glucose, 1 μM ethylenediaminetetraacetic acid, 1.3 μM MgSO 4 heptahydrate, 5 μM KCl, 100 μM NaCl,100 μM acetic acid, 1% BSA at pH 3) for 5 min.Then 100 µL samples were withdrawn for analysis, and the cells were washed twice with ice-cold hHBSS.To isolate the fraction containing cell-internalized peptide, the cells were kept at -80 • C for 1 h.Then 150 µL ice-cold hHBSS were added, and the cells were scraped off the bottom of the wells before centrifugation for 20 min at 4 • C with 13.000 rpm in a Sigma 1-15 K centrifuge (DJB Labcare, Newport Pagnell, England).Samples of 100 µL supernatant were withdrawn for analysis.The pellet was first resuspended in 25 µL ice-cold acetic acid buffer before addition of 75 µL 96% ethanol.Additionally 100 µL ice-cold acetic acid buffer was added followed by 20 min centrifugation at 4 • C and 13.000 rpm. 100 µL supernatant was used for quantification of membrane-embedded peptide.All samples were analyzed in black clear-bottom 96-well plates using a NovoStar fluorescence plate reader (BMG Labtech, Offenburg, Germany) with excitation/emission set to 485/590 nm and gain settings according to Table S1.The molar amounts of membrane-adhered peptide, membrane-embedded peptide, and peptide taken up by the cells were calculated by using standard curves freshly prepared in acetic acid buffer or hHBSS in a concentration range from 80 nm to 2.5 µM.

Live imaging of peptide distribution in bEND3 cells
Peptide cellular distribution was evaluated in bEND3 cells grown at the bottom of 8-well Ibidi chambers after 1 h incubation with 50 µM T-Tat 47-57 , T-penetratin, T-apidaecin, or T-oncocin at 37 • C with horizontal shaking (90 rpm).After incubation with peptide, the cells were washed twice with 37 • C hHBSS.Cellular peptide distribution was visualized using a Zeiss 510 laser-scanning confocal microscope (Carl Zeiss, Jena, Germany) with single wavelength excitation at 543 nm.

In vitro transport across Bioni-010c monolayers and cellular uptake
Peptide transport, membrane adherence, and cellular uptake were evaluated by using Bioni-010c monolayers grown on Transwell® inserts.The cells were equilibrated to room temperature before measurement of the transendothelial electrical resistance (TEER) with a Millicell-ERS Volt-Ohm meter (Fisher scientific, Loughborough, England).The cell media were replaced with 0.5 and 1 mL hHBSS at the luminal and abluminal compartment, respectively, and the cells were left to equilibrate for 20 min at room temperature prior to TEER measurement.Before initiation of the transport experiment, the cells were left to equilibrate for 20 min at 37 • C with orbital shaking at 90 rpm.T-Tat 47-57 , T-penetratin, T-apidaecin, T-oncocin, or the 6-mer non-CPP control peptide T-NRGTEWD were spiked into the luminal volume reaching a final 50 µM test concentration.T-Tat 47-57 , T-penetratin, Tapidaecin, and T-oncocin were applied alone or in the presence of 3 Hpropranolol (1 µCi/mL) and 14 C-mannitol (0.66 µCi/mL).Samples of 100 µL were withdrawn from the abluminal chamber after 15,30,45,60,90,120, and 180 min.Samples containing only peptide were transferred to a clear-bottom 96-well plate.Samples containing 3 Hpropranolol and 14 C-mannitol were transferred to scintillation vials before mixing with 2 mL Ultima Gold 241 TM (Perkin Elmer, Waltham, MA, USA).All sampling volumes were replaced with 37 • C hHBSS.Following incubation with peptide for 3 h, the cells were equilibrated to room temperature before TEER measurement.The monolayers were washed twice with 37 • C hHBSS.Two filters from each Bioni-010c passage were subjected to quantification of membrane-adhered peptide and cellular uptake, whereas one filter was subjected to immunocytochemistry.To isolate membrane-adhered peptide, the cells were placed on ice and incubated for 5 min with 300 µL ice-cold acetic acid buffer.Subsequently, 100 µL samples were collected in black clearbottom 96-well plates for analysis.The cells were washed twice with ice-cold hHBSS before the filters were cut out of the inserts and placed in 200 µL ice-cold lysis buffer (10 mM Tris-HCl, 0.25 M sucrose, 1 mM EDTA, 1 mM ethylene-bis(oxyethylenenitrilo)tetraacetic acid tetrasodium (EGTA), 2% Tergitol™, Complete® inhibitor tablet) for 10 min prior to homogenization using a pellet pestle (Merch, St. Louis).To isolate peptide taken up by the cells, the cell lysates were spun down at 4 • C with 13.000 rpm.Then 100 µL supernatants were collected in black clear-bottom 96-well plates for analysis.Fluorescence of the 6-TAMRA label was measured in the withdrawn samples by using a NovoStar fluorescence plate reader (BMG Labtech, Offenburg, Germany) with excitation/emission set to 485/590 nm and gain settings according to Table S1.Samples containing 3 H-propranolol and 14 C-mannitol were analyzed by using a Packard Tri-Carb 2100 TR scintillation counter (Packard, Dreieich, Germany).The molar amounts of transported peptide, membrane-adhered peptide, and peptide taken up by the cells were calculated by using standard curves (freshly prepared in hHBSS, acetic acid buffer, or lysis buffer, respectively) covering concentrations P. Frøslev et al. ranging from 80 nm to 2.5 µM.The apparent permeability (P app ) was calculated by the following equation: Where dQ/dt is the steady state flux, A is the area of the filter insert (1.12 cm 2 ), and C0 is the initial donor concentration in the luminal chamber.

Immunocytochemistry on Bioni-010c monolayers
After each transport study, one filter per Bioni-010c batch was subjected to immunocytochemistry.The cells were fixated with 4% paraformaledehyde in phosphate-buffered saline (PBS) for 10 min at room temperature with horizontal shaking (90 rpm).Thereafter the cells were washed twice with PBS, permeabilized with 0.1% Triton X-100 in PBS for 5 min, and blocked with 2% BSA in PBS for 30 min at room temperature with horizontal shaking (90 rpm).The filters were separated from the plastic inserts and cut into smaller pieces before incubation with an antibody directed against zonula occludens-1 (ZO-1) (1:100 in PBS with 2% BSA, ThermoFischer, catalog no 617,300) overnight at 4 • C with horizontal shaking (90 rpm).The filter pieces were washed thrice in PBS with 2% BSA before incubation with a secondary antibody conjugated to Alexa-488 (1:100 in PBS with 2% BSA, Life Technologies, catalog no.AV1008) for 30 min at room temperature with horizontal shaking (90 rpm).The filter pieces were washed thrice in PBS with 2% BSA before mounting on object glasses under coverslips.ZO-1 staining and 6-TAMRA fluorescence were visualized by using single-wavelength excitation at 488 nm and 543 nm, respectively, using a Zeiss 510 laserscanning confocal microscope (Carl Zeiss, Jena, Germany).

Real-time barrier integrity measurements on Bioni-010c monolayers
The integrity of the Bioni-010c monolayers cultured on transwell inserts was monitored in real-time by using a CellZScope device (NanoAnalytics, Münster, Germany).The transwell inserts were transferred to the CellZScope and the cell media were replaced with 37 • C hHBSS before the cells were left to equilibrate for 2 h at 37 • C with horizontal shaking (90 rpm).T-Tat 47-57 , T-penetratin, T-apidaecin, or Toncocin were spiked into the luminal volume reaching final 1, 5, 10, or 25 µM test concentrations, and then the cells were subjected to incubation for 3 h at 37 • C with horizontal shaking (90 rpm).The peptide solutions were gently removed and replaced with fresh 37 • C hHBSS before the CellZScope was transferred to an incubator (5% CO 2 , 37 • C) for 15 h to allow the cells to recover.

Data analysis and statistics
Calculations were performed by using Microsoft Office Excel.Graph presentations and statistical analyses were performed using GraphPad Prism version 8 (GraphPad, La Jolla, CA, USA).Two-way analysis of variance (ANOVA) with Dunnett multiple comparison post-test and paired t-test were applied, and P-values > 0.05 were considered significant.Data are presented as mean ± standard deviation (SD) with "N" representing technical replicates and "n" representing biological replicates.Drawing of peptide structures and Mw calculations were performed using ChemDraw professional 16.0 (PerkinElmer, Waltham, Massachusetts, USA).Helical wheel projections were obtained using NetWheels (University of Brasilia, Brasilia, Brasil).

Physico-chemical characteristics of the peptides used in the study
Tat 47-57 and penetratin belong to the group of cationic CPPs, whereas apidaecin and oncocin are classified as proline-rich CPPs (Fig. 1A).The four CPPs all carry a net positive charge at physiological pH, and were for the present study N-terminally conjugated to the fluorophore 6-TAMRA (T) for quantification and visualization purposes.We note that, TAMRA-labeling of the CPPs influences on their physicochemical properties when compared to unlabeled CPPs.These changes, however, will apply to all the CPPs and therefore allow us to do direct comparisons throughout the experiments.Full molecular structures of Tat 47-57 , penetratin, apidaecin, and oncocin are displayed in the supplementary Figs.S1-S4.The CPPs were ranged according to their relative hydrophobicity, based on their retention time during HPLC analysis (Fig. S5), revealing apidaecin as the most hydrophobic and Tat 47-57 as the least hydrophobic (Fig. 1B) of the four CPPs.All CPPs were run on the same column applying the same method with identical mobile phases during HPLC analysis.Neither, of the CPPs display primary amphipathicity as evident by the lack of hydrophilic and hydrophobic patches in the primary amino acid sequence (Fig. 1A).Helical wheel projections were drawn to visualize the spatial orientation of the amino acid side-chains upon α-helical formation (Fig. 1C), i.e. whether the CPPs display secondary amphipathicity.Tat 47-57 , with just two hydrophobic amino acids (tyrosine and glycine), do not display any amphipathic character.Penetratin displays some amphipathic character, as evident by the presence of the hydrophobic patch FWI(W), and the hydrophilic patches RKKQR and KNQK.Apidaecin and oncocin display some degree of amphipathic character, though to a lesser extent when compared to penetratin.

Both the cationic-and proline-rich cell-penetrating peptides were taken up by brain capillary endothelial cells, but differed in their degree of cell membrane adherence and embedding
The viability of brain capillary endothelial cells was assessed after incubation with T-Tat 47-57 , T-penetratin, T-apidaecin, or T-oncocin using the mouse-derived bEND3 cell line and the commercially available XTT assay for measuring mitochondrial activity (Fig. 2A).The CPPs did not decrease the cell viability as compared to the buffer control.Next, the ability of the CPPs to adhere to -and embed within the bEND3 cell membranes as well as to internalize into the cells was evaluated (Fig 2B)  S6) .T-apidacin and T-oncocin -membrane adherence, -membrane embedding, and -cellular uptake did not differ from that of the 6-TAMRA-labelled 7-mer control peptide NRGTEWD.NRGTEWD does not belong to the class of CPPs, but was originally described as a brain homing peptide when surface-displayed on virus capsids (Körbelin et al., 2016).Live cell microscopy was applied to study the overall mechanism of CPP cellular internalization, which was indicated to involve direct membrane translocation or endocytosis (Fig. 2C).T-Tat 47-57 appeared to internalize into the cells via both direct membrane translocation and endocytosis, as judged by the appearance of diffuse cytosolic T-Tat 47-57 distribution and dot-like intracellular vesicular structures indicating T-Tat 47-57 uptake via an endocytic mechanism.T-penetratin appeared to mainly internalize into the bEND3 cells via direct membrane translocation, whereas T-apidaecin and T-oncocin cell internalization seemed to be driven by endocytosis.Subjecting the peptides to SDS-PAGE did not point to degradation during incubation in the experimental buffer (Fig. S7).

Bioni-010c cells differentiated into brain capillary endothelial-like cells generate tight monolayers suitable for barrier transport studies
The bEND3 cells do not form tight endothelial monolayers suitable for transport studies.Therefore, the human induced pluripotent stem cell line Bioni-010c was introduced to investigate the potential of the CPPs to permeate a tight cell monolayer, which moreover represents the human BBB.The Bioni-010c cells were differentiated into brain capillary endothelial-like cells (Goldeman et al., 2020) using a previously established protocol (Stebbins et al., 2016) (Fig 3A).Transport studies investigating the luminal to abluminal transport of the CPPs as well as their potential effects on the barrier integrity were performed on transwell®-cultured Bioni-010c-derived cell monolayers (Fig. 3B).The monolayers displayed high initial TEER in cell media (8282 ± 1725 Ω × cm 2 ) as well as after 3 h incubation in buffer (4621 ± 2332 Ω × cm 2 ) or cell media (5108 ± 2522 Ω × cm 2 ) (Fig. 3C).Immunofluorescence staining of the tight junction associated ZO-1 protein revealed a strict localization to the junctional zones (Fig. 1D).

Cationic CPP interaction with and/or internalization into Bioni-010c monolayers leads to a dramatic decrease in monolayer integrity
The ability of T-Tat 47-57 , T-penetratin, T-apidaecin, and T-oncocin to internalize into Bioni-010c cells, and to permeate monolayers hereof, was evaluated following incubation for 3 h.The cells were fixated prior to visualization of CPP uptake via 6-TAMRA fluorescence by using confocal microscopy.Staining of the peripheral tight junction-associated ZO-1 protein was included for evaluation of potential effects on the monolayer morphology (Fig. 4A).However, no conclusions may be drawn regarding the mechanism of cellular uptake of the CPPs as the fixation procedure affects the intracellular distribution of internalized CPPs (Richard et al., 2003).
The cationic T-Tat 47-57 and T-penetratin efficiently internalized into The barrier integrity was evaluated by TEER measurements before and after each transport study, demonstrating initial mean TEER values ranging from 3854 ± 1056 Ω × cm 2 to 5340 ± 1233 Ω × cm 2 (Fig. 4D).Incubation for 3 h with the highly cationic CPPs resulted in significant effects on the Bioni-010c barrier integrity, with mean TEER values of 219 ± 79 Ω × cm 2 and 125 ± 94 Ω × cm 2 after incubation with T-Tat 47-57 and T-penetratin, respectively.Neither T-apidaecin nor Toncocin affected the barrier integrity as compared to the slight drop in TEER observed for the buffer control.The cell viability of the filtergrown Bioni-010c cells was evaluated after each transport study by using the XTT assay (Fig. 4E).Incubation with T-Tat 47-57 slightly decreased the cell viability as compared to the buffer-treated cells, whereas neither of the remaining CPPs appeared to affect the Bioni-010c cell viability.Moreover, the effects on cell viability were also evaluated by using Bioni-010c cells grown at the bottom of plastic wells followed by incubation with the CPPs (Fig. S8).In that setup, T-Tat 47-57 did not give rise to any adverse effects as compared to buffer-incubated cells (Fig. S8A); neither did T-penetratin (Fig. S8B), T-apidaecin (Fig. S8C), or T-oncocin (Fig. S8D); the latter two tested in concentrations up to 200 µM.

T-Tat 47-57 and T-penetratin affect the barrier integrity to a degree that facilitates permeation of the paracellular flux marker mannitol
Finally, we investigated the kinetics of the barrier-modulating effects of T-Tat 47-57 and T-penetratin, as well as whether these highly cationic CPPs improve barrier permeation of 3 H-propranolol and 14 C-mannitol included as molecules representing a small hydrophobic and a hydrophilic compound, respectively (Fig. 5). 3 H-propranolol (259 g/mol, hydrogen bond donors: 2, hydrogen bond acceptors: 3), but not 14 Cmannitol (182 g/mol, hydrogen bond donors: 6, hydrogen bond acceptors: 6), obeys the "rule of CNS-likeness".Bioni-010c cell monolayers were transferred to a CellZScope device enabling real-time TEER monitoring (Fig. 5A).The monolayers were incubated with T-Tat 47-57 or T-penetratin for 3 h, where after they were allowed to recover for 15 h in fresh buffer.Incubation with T-Tat 47-57 and T-penetratin resulted in a 50% and 90% drop in TEER, respectively, and that already at the first measurement (25 min time-point).The TEER stayed low during the 3 h incubation with T-Tat 47-57 or T-penetratin.The barrier integrity gradually recovered upon CPP removal and application of fresh buffer, and appeared fully restored after approximately 7 h.The barrier-modulating effects of both T-Tat 47-57 (Fig. S9A) and T-penetratin (Fig. S9B) were concentration-dependent. Application of 1 µM CPP did not decrease the TEER of the Bioni-010c monolayers as observed upon incubation with the higher CPP concentrations, and therefore it reflects the effect of pure buffer.
Next, we applied 3 H-propranolol and 14 C-mannitol together with T-Tat 47-57 or T-penetratin to elucidate whether the cell uptake (Fig. 4A, B) and/or barrier effects (Figs.4D, 5A) exerted by the CPPs facilitate barrier permeation of a small hydrophobic compound and a small hydrophilic compound, respectively (5B-D).The barrier permeation of 3 Hpropranolol in the presence of T-Tat 47-57 or T-penetratin did not differ from that observed in the presence of the buffer control (Fig. 5B).In contrast, a significant increase in 14 C-mannitol permeation was observed upon application together with T-Tat 47-57 or T-penetratin as compared to that seen in the presence of the buffer control (Fig. 5C).The effect on 14 C-mannitol permeation was observed after the 90 min time-point in the presence of T-penetratin, whereas the effect of T-Tat 47-57 was not observed until the 180 min time-point.The permeability-enhancing effects of T-Tat 47-57 and T-penetratin for uptake of 14 C-manniol (but not of 3 H-propranolol) were also reflected in the calculated apparent permeabilities (Fig. 5D).The 14 C-mannitol permeability was enhanced 3.8 fold and 10.5 fold in the presence of T-Tat 47-57 and T-penetratin, respectively, as compared to that found in buffer; with the effect of T-penetratin being significantly higher in comparison to the effect of T-Tat 47-57 (p < 0.0001).

Discussion
The BBB is a major hindrance for successful brain drug delivery.Interestingly, CPPs have shown potential as shuttles for brain drug delivery (Aarts et al., 2002;Bach et al., 2012;Lim et al., 2015).With the present study, we have investigated the potential of the highly cationic CPPs T-Tat 47-57 and T-penetratin as well as the proline-rich T-apidaecin and T-oncocin for permeating a stem cell-based human BBB model.We also studied whether the CPPs facilitate barrier permeation of 14 C-mannitol, which does not obey the "rule of CNS-likeness", and/or further improve barrier permeation of 3 H-propranolol that obeys this rule.
The highly cationic CPPs displayed better cell membrane adherence -and embedding as well as cellular uptake when compared to the proline-rich CPPs (Figs. 2B and 4B).Both T-apidaecin and T-oncocin carry a substantial net positive charge (+6) at the experimental pH (7.4) (Fig. 1A), which facilitates electrostatic interactions with the negatively charged phospholipid head groups of the plasma membrane.T-Tat 47-57 and T-penetratin possess a higher positive charge (+9 and +8, respectively; which confers a higher charge per residue) than the proline-rich CPPs.Thus, an increased charge per residue conferred by a higher content of arginine and lysine (displaying cationic amino-and guanidino-functionalized side chains), appears to improve especially cell membrane adherence but also cellular uptake in bEND3 cells (Fig. 2B) and Bioni-010c cells (Fig. 4B).Cell membrane adherence and embedding as well as cellular uptake properties did not differ among the two proline-rich CPPs, but did differ significantly for T-Tat 47-57 versus Tpenetratin (Fig. 2B).T-Penetratin carries two tryptophan residues, which adds hydrophobicity to the peptide, and therefore confer improved membrane-adhering and -embedding properties as compared to T-Tat 47-57 , which does not contain amino acids with hydrophobic side chains.That observation is in agreement with earlier studies, demonstrating that the presence of tryptophan residues, as well as their specific positioning within a CPP sequence, potentially enhances the cellular Representative confocal laser-scanning microscopy images of fixated Bioni-010c monolayers.Peptide uptake was visualized via 6-TAMRA fluorescence (red), while the junctional zones were visualized via zonula occludens-1 (ZO-1) immunostaining (green).Maximal z-stack projection, scale bar: 20 µm.(B) Peptide membrane adherence -and cellular uptake in Bioni-010c cells.(C) Calculated apparent peptide permeability (P app ) based on flux curves.ND: Not detected.(D) Transendothelial electrical resistance (TEER) measurements across Bioni-010c monolayers before (0 min) and after (180 min) a transport experiment.(E) Bioni-010c cell viability depicted as percentage relative to buffer control with Triton-X100 included as control for 100% cell death.Data are presented as mean ± SD (N = 3, n = 3); *: p < 0.05 (paired t-test) (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.).uptake (Kamei et al., 2013b;Kristensen et al., 2015;Rydberg et al., 2012aRydberg et al., , 2012b)).Ranging the four CPPs according to hydrophobicity, revealed that T-apidaecin was more hydrophobic than T-penetratin, and that T-Tat 47-57 was the least hydrophobic (Fig. C).Thus, both charge and hydrophobicity, and the proper balance between these, are important factors for cell membrane interaction and embedding of CPPs as well as for their propensity toward cellular uptake.In addition to charge and hydrophobicity, an amphipathic nature may promote cellular uptake of CPPs.The hydrophobic and hydrophilic patches present in the projected α-helical structures of penetratin, apidaecin, and oncocin, do not completely distribute on one or the other side of the helix (Fig. 1D).Therefore, only minor degree of secondary amphipathicity is expected if these CPPs adopt an α-helical conformation.Ideal helical structure formation of apidaecin and oncocin is moreover questioned, due to their large number of proline residues, which likely destabilizes α-helical conformations (Kim and Kang, 1999).The limited degree of penetratin amphipathicity supports an earlier study (Bahnsen et al., 2013), which demonstrated that penetratin adopts an α-helical conformation in the vicinity of lipid membranes.This is in contrast to a more recent study, which showed that penetratin does not adopt a well-defined α-helical structure (Kristensen et al., 2020).Amphipathic characteristics are, however, not a prerequisite for cellular uptake of CPPs as also observed in the present study where T-Tat 47-57 exhibited excellent cell-penetrating properties like most polyarginines (Verdurmen et al., 2011), which both do not display any amphipathicity.
T-penetratin was superior to both T-Tat 47-57 and the proline-rich CPPs with respect to cell membrane adherence -and embedding as well as cellular uptake (Figs.2B and 4B).In contrast, T-penetratin did not permeate the Bioni-010c monolayers (Fig. 2C).In that context, T-Tat 47-57 was superior to the remaining CPPs.Thus, some CPPs may display a too high degree of plasma membrane adherence -and embedding that in fact reduces their ability to permeate the BBB (e.g., as seen for T-penetratin), whereas others display too little degree of membrane adherence -and embedding for efficient barrier permeation (e.g. as seen for T-apidaecin and T-oncocin).It should, however, be noted that the cell-penetrating -and barrier permeating properties of CPPs may depend on multiple experimental factors such as its concentration and the specific cell type.
Incubation with T-penetrain and T-Tat 47-57 had dramatic and immediate, yet reversible, effects on the Bioni-010c barrier integrity (Figs.2D and 5A).We recently demonstrated that T-Tat 47-57 affects the integrity of a primary bovine BBB model (Kristensen et al., 2020) in a similar manner as observed for T-Tat 47-57 and T-penetratin in the present study.Staining of the tight junction-associated protein ZO-1, after incubation with CPP, revealed that T-penetratin appears to affect the Bioni-010c monolayer morphology (Fig. 4A).This was not the case for T-Tat 47-57 despite the fact that comparable degrees of TEER lowering was observed for both of the highly cationic CPPs (Fig. 4D).An early study investigating effects on tight junctions in cultured brain endothelial cells after incubation with HIV-transactivator of transcription protein-derived Tat 1-72 , did not reveal any effects on monolayer morphology upon ZO-1 staining (Andra et al., 2003).On the other hand, that study documented disruption of the intercellular tight junction zones upon Claudin-1 -and Claudin-5 staining.Therefore, no conclusions can be drawn from the present study on whether T-Tat 47-57 affects the morphology of the Bioni-010c monolayers as observed for T-penetratin via ZO-1 staining.
We recently also investigated whether the T-Tat 47-57 -mediated decrease in barrier integrity would facilitate permeation of 3000-5000 Da dextran across a primary bovine BBB model (Kristensen et al., 2020).Only a minor, but insignificant, increase in dextran permeation was observed.In contrast, another study demonstrated that the CPP octaarginine facilitated brain delivery of insulin (5808 Da) when co-administered to rats (Kamei et al., 2018).On the other hand, we did demonstrate that covalent conjugation of T-Tat 47-57 to a therapeutic peptide (NA-1 (Aarts et al., 2002)) gave rise to improved barrier permeation of the peptide cargo as compared to that of the cargo itself (M Kristensen et al., 2020).In the present study, we turned our attention to small compounds, and investigated whether the T-Tat 47-57 -and T-penetratin-mediated decrease in barrier integrity facilitates permeation of small compounds (Fig. 5).Thus, 3 H-propranolol and 14 C-mannitol were included as model compounds with good and poor BBB-permeating properties, respectively.They both have MWs below 400 Da, but 14 C-mannitol exceeds both the number of hydrogen-bond donors -and acceptors to obey the "rule of CNS-likeness", whereas 3 H-propranolol fully obeys that rule.Both T-Tat 47-57 and T-penetratin significantly improved 14 C-mannitol permeation across the Bioni-010c monolayers, whereas no effect on the 3 H-propranolol permeation was observed (Fig. 5B-D).The mechanisms by which CPPs mediate permeation of covalently conjugated or co-administered cargo molecules across biological barriers are not fully understood.With this study, we demonstrated that T-Tat 47-57 and T-penetratin mediated opening of the paracellular space thereby facilitating enhanced barrier permeation of the small hydrophilic compound 14 C-mannitol.Some of the 14 C-mannitol may also be endocytosed together with T-Tat 47-57 , and then in principle subjected to exocytosis at the basolateral membrane; i. e. involving CPP-mediated barrier permeation via adsorptive-mediated transcytosis as earlier suggested (Hervé et al., 2008).However, we did not observe any improved 3 H-propranolol permeation when applied together with T-Tat 47-57 .In addition, neither 3 H-propranolol nor 14 C-mannitol permeation was improved in the presence of T-apidaecin or T-oncocin (Fig. 5D), which displayed endocytic uptake (Fig. 2C) but no effects on barrier integrity (Fig. 4D).Thus, we expect limited degree of transcytosis across the BBB for compounds applied together with CPPs taken up by endocytosis, such as T-Tat 47-57 , T-apidaecin, and T-oncocin.The fate of the CPP itself upon endocytic cellular uptake is moreover questioned; a fraction is likely segregated to lysosomes for degradation (Kristensen et al., 2020), whereas some may experience transcytosis.

Conclusions
We demonstrated uptake in brain endothelial cells of both highly cationic (T-Tat 47-57 , T-penetratin) and proline-rich (T-apidaecin, Toncocin) CPPs.The highly cationic CPPs displayed better plasma membrane adherence and embedding as well as increased cellular uptake as compared to the proline-rich CPPs.T-Tat 47-57 was the only peptide observed to cross the barrier in a significantly higher amount than a 6-mer control peptide not belonging to the class of CPPs.
T-penetratin surprisingly showed high membrane association and embedding, but no barrier permeation.There was no clear correlation between peptide hydrophobicity and membrane adherence -and embedding or barrier permeation properties.The cationic peptides T-Tat 47-57 and T-penetratin both affected the integrity of the BBB model to a degree that facilitated permeation of the paracellular flux marker 14 Cmannitol, but not of the transcellular flux marker 3 H-propranolol.The barrier-modulating effects were reversible.Thus, CPPs affecting the barrier properties of the brain capillary endothelium in a reversible manner may be explored to facilitate brain delivery of small hydrophilic drug compounds, which otherwise are restricted from brain entry by the BBB.

Supplementary Materials
Table S1: Overview of detector gain for peptide analysis using a NOVOstar MicroPlate Reader with excitation/emission set to 485/590 nm.

Fig. 1 .
Fig. 1.Overview and characteristics of the cell-penetrating peptides (CPPs) included in the study.(A) Table with primary sequences, molecular weight (MW), net charge, and peptide class for T-Tat 47-57 , T-penetratin, T-apidaecin, and T-oncocin.All peptides are C-terminally amidated (NH 2 ).(B) Chemical structure of the fluorophore 6-TAMRA (T) N-terminally conjugated to the peptides for quantification and visualization.(C) Hydrophobicity ranging of T-Tat 47-57 , T-penetratin, T-apidaecin, and T-oncocin based on their retention time during reverse-phase HPLC analysis on a C 18 column.(D) Helical-wheel projections of Tat 47-57 , penetratin, apidaecin, and oncocin with gray circles representing polar residues and white circles representing nonpolar residues.
. The highly cationic T-Tat 47-57 and T-penetratin adhered to the cell membrane of the bEND3 cells to a greater extent as compared to that of the proline-rich T-apidaecin and T-oncocin, and with T-penetratin being superior to T-Tat 47-57 .Moreover, T-penetratin displayed a higher degree of membrane embedding as opposed to that of T-Tat 47-57 .T-apidaecin and T-oncocin demonstrated limited ability to embed within the cell membranes of the bEND3 cells.A similar pattern was observed when quantifying the cell uptake, where T-Tat 47-57 and T-penetratin appeared to internalize into the bEND3 cells more efficiently than T-apidaecin and T-oncocin (Fig. 2B,C).Despite the superior cell-adhering and cellpenetrating abilities of T-Tat 47-57 and T-penetratin, when compared to T-apidaecin and T-oncocin, less than 1% and 3% of initially applied T-Tat 47-57 and T-penetratin, respectively, were detected in the different fractions (Fig.