Comparative transport analysis of cell penetrating peptides and Lysosomal sequences for selective tropism towards RPE cells

Cell penetrating peptides are typically nonspecific, targeting multiple cell types without discrimination. However, subsets of Cell penetrating peptides (CPP) have been found, which show a ‘homing’ capacity or increased likelihood of internalizing into specific cell types and subcellular locations. Therapeutics intended to be delivered to tissues with a high degree of cellular diversity, such as the intraocular space, would benefit from delivery using CPP that can discriminate across multiple cell types. Lysosomal storage diseases in the retinal pigment epithelium (RPE) can impair cargo clearance, leading to RPE atrophy and blindness. Characterizing CPP for their capacity to effectively deliver cargo to the lysosomes of different cell types may expand treatment options for lysosomal storage disorders. We developed a combinatorial library of CPP and lysosomal sorting signals, applied to ARPE19 and B3 corneal lens cells, for the purpose of determining cell line specificity and internal targeting. Several candidate classes of CPP were found to have as much as 4 times the internalization efficiency in ARPE19 compared to B3. Follow-up cargo transport studies were also performed, which demonstrate effective internalization and lysosomal targeting in ARPE19 cells.


Introduction
The abnormal accumulation of material in the lysosome is the cause of a broad class of diseases termed lysosomal storage disorders.These disorders are typically caused by de cient activity in lysosomal enzymes, such as seen in mutations of GBA1 leading to Gaucher's disease which is characterized by accumulation of glucocerebroside lipids. 1 The accumulation of metabolic byproducts over long periods is also a factor in promoting these diseases, such as seen in lipofuscin formation in aged cells and the accumulation of bisretinoid molecules in the RPE cell layer of the eye which is associated with the onset and progression of Stargardt's disease and AMD 2,3 .Enzyme replacement therapy is one therapeutic modality for removing accumulated lysosomal material.In Guacher's disease treatments, and in some experimental methods for treating macular degeneration, this is accomplished through the delivery of therapeutic proteins via mannose-receptor mediated endocytosis 4 .However, the di culties of protein glycoengineering, as well as the ine ciency and non-speci city of glycan delivery system has driven research into targeted enzyme delivery systems (EDSs) [5][6][7][8] .CPP have become popular EDSs for therapeutic enzymes since their discovery in 1988, when it was observed that the trans-activator protein (Tat) of HIV-1 allowed for translocation across cell membranes into the cytoplasm [9][10][11][12] .Besides the capacity for translocation across membranes, CPP can perform a variety of novel functions due to the diversity in their sequence, electrostatic charge, hydrophobicity, and structure.Certain CPP have demonstrated a "Homing" capacity, de ned as the ability to preferentially penetrate speci c cell types such as cancer cell lines, which may enable more e cient delivery and lower off target toxicity [13][14][15] .CPP have also been found which deliver cargos preferentially to subcellular locations, with successful delivery to the nucleus, endoplasmic reticulum, and lysosomes.Due to these features, the application of CPP for site directed biologics has been rapidly expanding in the eld of oncology, where numerous CPP-conjugates have entered clinical development over the last decade 16-18 19 .As CPP research continues to evolve, their application as EDSs may improve the e cacy, safety, and scope of enzyme replacement therapeutics.
Cell line and subcellular targeting CPP may serve as useful EDSs in delivery of therapeutic enzymes to the posterior segment of the retina.The retina of the eye is composed of layers of specialized epithelial and neural cells that serve as a physical barrier between the posterior segment and the intraocular space.
Therapeutic enzymes intended for delivery to the cells at the posterior of the eye, such as the retinal pigment epithelium (RPE) are required to pass by this complex layers of cells as they diffuse across the retina 20,21 .Cell line speci c CPP may serve as ideal EDSs for targeted delivery to the RPE, which may open new avenues for therapeutic delivery of novel enzymes for the treatment of Stargardt's disease and AMD.
The focus of this study is to characterize the features of CPP that show higher internalization to the RPE, and to determine whether the inclusion of lysosomal sorting signals would improve localization of a uorescent cargo to lysosomes.Based on the membrane characteristics of RPE cells and known cellspeci c surface receptors, we identi ed 50 candidate CPPs that we anticipated to show greater internalization into RPE cells 28,29 .To achieve lysosomal targeting, we coupled these 50 candidates with 9 lysosomal targeting motifs, resulting in a library of 500 CPP-LYS constructs.These CPP-LYS constructs were tested in ARPE-19 and B3 cells for their ability to enter the cell and hone to the lysosome, resulting in several candidate sequences with a preference for AREP19 localization that could be applied to future in vivo studies.

Uptake differences between cell types
Comparative internalization e ciency was measured by comparing green channel uorescence of treated APRE19 cells to B3 cells(Supplemental Table 2).The ratio of ARPE19 green uorescence to B3 cell uorescence was mostly equivalent between cell lines (Fig. 1) with ~ 13.5% of the CPP's having a ratio ≥ 3.0 (favoring ARPE19 internalization) and 4.0% having a ratio ≤ 0.3 (discriminatory against ARPE19 internalization).The in uence of biophysical characteristics on CPP uptake e ciency between cell lines was signi cant for the overall degree of internalization but did not signi cantly contribute to improving the ratio of internalization between ARPE19 cells compared to B3 cells.Net charge, hydropathy, and sequence length were found to have a negligible impact on internalization e ciency between cell lines, but Net charge, hydropathy were associated with increased uptake.Sequence length did not signi cantly impact the rate of CPP uptake and did not contribute to cell line uptake speci city (Fig. 2).
CPP Class and sequence speci city contributed strongly to both internalization e cacy and cell line speci city (Fig. 3).CPP with the Membrane targeting, metabotropic, and surface receptor general classes averaged a ratio greater than 1.5 (2.4,2.1, and 1.7 ARPE19/B3 cell internalization e cacy ratio respectively).Speci c CPP sequences selected for their capacity to selectively target membranes based on their lipid composition showed the most signi cant cell line speci city.The AAXH class (X = 1 through 13) of CPP was enriched for ARPE19 targeting, with AA10, AA9H, and AA4H demonstrating a mean uptake ratio of 4.6 ± 2.1, 4.3 ± 1.2, and 4.1 ± 1.5 respectively (Fig. 4).Similarly, the cell surface receptor speci c peptides CD36-1 and CD36-3 demonstrated a high mean uptake ratio, preferring ARPE19 to B3 cells at a ratio of 2.7 ± 0.8 and 2.5 ± 0.3 respectively, but it was characteristic of this class that the internalization rate overall was low in comparison to standard CPP.

Confocal Microscopy Analysis of Lysosomal Localization
3 Representative CPP sequences and their lysosomal sorting counterparts (TAT (YGRKKRRQRRR), Penetratin (RQIKIWFQNRRMKWKK), Nicastrin (RLPRCVRSTARLARALSPAF) were selected for higher magni cation confocal microscopy imaging due to desirable features observed in the lower magni cation imaging and their relevance in the CPP literature.Each CPP was compared between ARPE19 cells and B3 cells as in the screening study.For the Nicastrin and Penetratin series, the inclusion of a lysosomal targeting tag increased overall green uorescence and demonstrated and preferentially localized within lysosomes but we also observe a general decrease of red uorescence from Lysotracker in ARPE19 cells, but not B3 cells.Lysotracker is a pH sensitive dye, so this may indicate deacidi cation of the lysosomes of the RPE due to membrane disruption, or a reaction speci c to the ARPE19 cells in response to a foreign material entering the lysosome.The inclusion of a lysosomal targeting tag in the TAT series of CPP did not reduce distribution through the cytoplasm, but appears to promote higher association with the lysosome.This may re ect a mechanical aspect of the TAT peptide itself, as this sequence is known for escaping endosomes.Like Nicastrin and Penetratin, Lysosome uorescence was dimmed in ARPE19(Fig.5), but mostly unaffected in B3 cells (Fig. 6).B3 cells display a higher degree of uorescence overlap than ARPE19, which may indicate differences due to the composition of their membrane or endosome.

CPP synthesis and preparation
We used a three prong methodology for selecting potential CPP's/internalization signals: (i) Database and literature review to select a chemically diverse selection of popular CPP that were known to act through different internalization mechanisms 9,22 .(ii) we selected less common CPP that were known to have a selective internalization e ciency based on the relative lipid composition of the cell membrane, as the RPE membrane is enriched for phosphatidylethanolamine compared to the other cells of the eye.(iii), we selected receptor mediated internalization sequences that were known to bind to surface receptors CD-36, MERTK, and integrin, that are present in ARPE19 cells, which would be the cell culture model used experimentally.We produced a nal selection of 50 CPP sequences(Supplementary Table 1), with each of these produced alone or with the CPP sequences appended to 9 different lysosomal targeting sequences.
Each sequence was prepared via parallel peptide synthesis with a FITC uorophore attached to the N terminus of these sequences via a β-alanine linker as a model cargo.Lysosomal targeting sequences were selected from literature sources reporting on the sorting and tra cking of lysosomal proteins.
The aim of this study was to determine whether different classes of cell penetrating peptide could selectively target different lineages of cell within the eye, and to determine whether the addition of lysosomal targeting sequences could be appended to these sequences to allow for cell-line and organelle speci c targeting in a single motif.Peptides were selected from different known classes of CPP and for the potential to interact with surface proteins that are enriched on the surface of RPE cells.The CPP library was produced in conjugation with an N terminal Lysosomal targeting sequence, and with an N terminal cap of the uorophore FITC linked to the peptide with a beta-alanine linker, which prevents selfcleavage of FITC during peptide synthesis 23 .This sequence format was selected to maximize the distance of the uorophore from the CPP region of each peptide.
B3 Corneal epithelial cells were selected as a comparison cell line for our applications due to their accessibility as an eye cell line, their similar size to ARPE19 cells, and the similarity of culturing conditions to ARPE19 24 .One limitation of this study is that we did not use a photoreceptor cell line to gauge speci city and toxicity in a nerve cell or photoreceptor that would be at the same site of action are the RPE cell layer, due to di culty of applying this cell line to screening conditions, as they achieve low cell densities than can be achieved compared to ARPE19, which would complicate the comparison, and are quite fragile in culture 25 .Further studies using primary polarized RPE cells and a primary photoreceptor cell line such as 661W would provide a more ideal comparison for physiological conditions.
Our uorescence detection was performed with an I3 spectrophotometer and a Sartorius IncuCyte S3 automated uorescence cell monitoring system.The screening plates had uorescence quanti ed in both the I3 and Sartorius IncuCyte S3 to provide orthogonal uorescence screening using a different instrument than is typically used in related literature and for capturing bright eld images.We used I3 uorescence counts for the graphs displayed, as historically uorescence studies of CPP are used with raw uorescence counts on a plate reader format 26 .To improve data quality, future studies might be performed using a high throughput confocal microscope to provide more detailed images of uorescence overlap in screening and localization studies.
The most successful CPP classes in terms of their internalization ratio of ARPE19 to B3 cells were found to be membrane selective, which supports our hypothesis that the enriched phosphatidylethanolamine layer of RPE (67.5%) compared to the retina (38.6%), lens (15%), or cornea (8.5%) may provide enough of a basis to preferentially deliver CPP's and their cargo to the RPE cell layer 27,28 .Annexin A isoform derived CPP (AA1H to AA13H) have been shown previously to have lipid binding properties, with annex III, IV, V, and VI preferring phosphatidylethanolamine.In this screen, Annexin III-VII all had a mean ratio greater than 1.5 (ARPE:B3 uorescence), with AA6H being the highest of this class at a ratio of 4.7 29 .Future studies should be carried out using these classes in comparison with a variety of other cells with known compositions of cell membrane lipids, or arti cial de ned membranes to verify the mechanism of internalization to be based on the ratio of phosphatidylethanolamine.

Conclusion
The CPP developed in this screen demonstrate a preference for ARPE19 cells vs B3 corneal cells and can be used to internalize a recombinantly linked cargo protein effectively.Mechanistically, this is likely due to the membrane composition of RPE cells being enriched in phosphatidylethanolamine.As the RPE cell layer is implicated the development of macular degeneration, therapeutic fusion proteins prepared with these cell penetrating peptides may be used to deliver catabolic proteins that could be used to degrade intracellular lipofuscin in the RPE, which may delay the progression of macular degeneration.Other applications might be used for enzyme replacement therapy in other phospholipid rich cell lines.

Peptide synthesis
Peptides were synthesized via automated peptide synthesis using a syro I peptide synthesis robot (Manufacturer).Peptides were assembled on a RINK-amide resin in parallel 96-well tip synthesis format using 10 mg of resin per tip.The resin was primed for synthesis by three cycles of resin incubation with 200 µL dimethylformamide (DMF) for 30 minutes.FMOC deprotection was performed at each stage of synthesis by treating the resin three times with 80 µL of a 40% piperidine/60% DMF mixture then washing the resin 8 times with 100 µL of DMF.Coupling was performed by incubating the resin with 6 molar equivalents of a preactivated mixture of each FMOC protected amino acid for 30 minutes.Coupling steps were performed twice, rst using a 2/2/1 mixture of 450 mM FMOC amino acid in DMF/ 450 mM N,N'-Diisopropylcarbodiimide in DMF/ 900 mM Oxyma in DMF, and secondly using a 2/2/1 mixture of 450 mM FMOC amino acid in DMF/430 mM (2-(1H-benzotriazol-1-yl)-1,1,3,3-tetramethyluronium hexa uorophosphate in DMF/ 1.8 M N,N-Diisopropylethylamine in N-Methyl-2-pyrrolidone.Each coupling step was performed using 100 µL of each preactivation mixture, and was followed with washing 4 times with 110 µL of DMF and a deprotection step as described previously.After the nal amino acid coupling, two additional rounds of coupling and deprotection were performed to create an N terminal Fluorescein label connected to the peptide chain via a Beta-alanine linker.The beta alanine linker coupling was performed as speci ed above for the individual amino acid building block steps, and the uorophore coupling step was performed by incubating the resin overnight in 100 µL of 3 molar equivalents of Fluorescein isothiocyanate isomer I dissolved in a mixture of DMF and 6 molar equivalents of DIPEA followed by washing the resin 8 times with 110 µL DMF.Cleavage of the peptide from the resin backbone was performed by washing the resin with Dichloromethane, then treating the resin for 2 hours with 200 µL of a mixture composed of 92.5% TFA, 2.5% TIS, 2.5% DODT, 2.5% Water.The labeled peptides were collected in a deep well 96 well plate, precipitated using cold ethyl ether and then centrifuged at 1000 RPM for 10 minutes to pellet the precipitate.Pelleted peptides were then resuspended in cold ethyl ether and centrifuged again in a similar fashion for two additional cycles to wash the peptides of residual cleavage cocktail.The peptides were then resuspended in 10 mM HCl, frozen overnight at -80°C, and lyophilized to remove water and tri uoroacetic acid.The resulting peptide powders were sealed and stored at -80°C.

Adjustment of peptide concentrations
One milliliter of 20% ethanol was added to each well of the 96 well plates containing the lyophilized peptides and the plates were then re-sealed.Sealed plates were placed until halfway submerged into a bath sonicator for 30 minutes to aid resuspension.The plates were centrifuged at 1000 RPM for 10 minutes to pellet any undissolved material.A 100 µL volume of each peptide were transferred into a clear bottom 96 well plate containing an internal standard of FITC in 80% water/20% ethanol, 10 mM HCl.Each peptide was quanti ed using FITC absorbance as a proxy for peptide concentration.Each peptide was normalized to 50 µM in a new deep well 96 well plate by adding calculated amounts of 20% ethanol and labelled peptide, then aliquoted to opaque black 96 well daughter plates that were sealed with foil and stored at -80°C.

Cell culture
Human B3 corneal epithelial cells and ARPE19 cells were obtained from ATCC.Cells were grown to con uency using DMEM at 37°C, 5% CO2.ARPE-19 and B3 cells for uptake experiments were seeded into a dark walled, clear bottom 96 well plate at 200 cells per well.All solutions were warmed to 37°C, and CPP incubations were performed at 37°C, 5% CO2.

CPP uptake
CPP working solutions were prepared by diluting the samples from the 50 µM stock plate to 10 µM in cell growth media.Cells were incubated with 100 µL of the 10 µM CPP solution for 2 hours, after which the CPP solution was aspirated from the wells, which were then washed three times with phosphate buffered saline.Cells were then incubated with 100 uL of PBS for analysis.Raw Cell uorescence values were monitored using a spectramax I3 plate reader using an excitation wavelength of 495 nm and an emission wavelength of 530 nm.An Incucyte live cell analysis microscopy system was used to extract bright eld and uorescence images of the cells as well, using a 20X objective microscope for bright eld images and a green uorescent channel with an excitation wavelength of 460 nm (passband 440,480 nm) and an emission wavelength of 524 nm (passband 504,544 nm).Incucyte software was used to determine green calibrated units and phase area.

Confocal Microscopy
Three representative sequences: TAT (YGRKKRRQRRR), Penetratin (RQIKIWFQNRRMKWKK), Nicastrin (RLPRCVRSTARLARALSPAF) were selected for comparative confocal microscopy analysis to the same CPP bearing lysosomal targeting sequences (LYSTAT-CSEWAYGRKKRRQRRR; LYSPenetratin-SLLKGRQGIYRQIKIWFQNRRMKWKK; LYSNicastrin-SLLKGRQGIYRLPRCVRSTARLARALSPAF) to assess whether Localization to the lysosome was increased by the inclusion of lysosomal targeting sequences.Representative images of the CPP that were selected to move on to cargo transport studies  Supplementary Files

Figure 3 Visual
Figure 3