Breeding Cell Penetrating Peptides: Optimization of Cellular Uptake

Subscriber access provided by University of Winnipeg Library

Article

Breeding Cell Penetrating Peptides: Optimization of Cellular Uptake by a Function-Driven Evolutionary Process Thorsten Krause, Niels Röckendorf, Nail El-Sourani, Katrin Ramaker, Maik Henkel, Sascha Hauke, Markus Borschbach, and Andreas Frey Bioconjugate Chem., Just Accepted Manuscript • DOI: 10.1021/acs.bioconjchem.8b00583 • Publication Date (Web): 31 Oct 2018 Downloaded from http://pubs.acs.org on November 4, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Bioconjugate Chemistry

Breeding Cell Penetrating Peptides: Optimization of Cellular Uptake by a Function-Driven Evolutionary Process

Thorsten Krause†§, Niels Röckendorf†§, Nail El-Sourani‡#, Katrin Ramaker†, Maik Henkel†&, Sascha Hauke‡¶, Markus Borschbach‡*, Andreas Frey†* †Department

of Mucosal Immunology and Diagnostics, Priority Area Asthma and

Allergy, Research Center Borstel, 23845 Borstel, Germany; Member of Leibniz Health Technologies ‡Faculty

of Computer Science, FHDW University of Applied Sciences, 51465

Bergisch Gladbach, Germany

Present address: #

T-Systems International GmbH, Telekom Open IoT Labs, Essen, Germany

¶

Landshut University of Applied Sciences, Landshut, Germany

&JPT

*

Peptide Technologies GmbH, Berlin, Germany

Corresponding authors:

ACS Paragon Plus Environment

1

Bioconjugate Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 49

For (bio)chemical part: Andreas Frey, Department of Mucosal Immunology and Diagnostics, Research Center Borstel, Parkallee 22, 23845 Borstel, Germany. Email: [email protected] For computational part: Markus Borschbach, Faculty of Computer Science, FHDW, University of Applied Sciences, Hauptstr. 2, 51465 Bergisch Gladbach, Germany. Email: [email protected] §

These two authors contributed equally


2

Page 3 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


ABSTRACT: In nature, building block-based biopolymers can adapt to functional and environmental demands by recombination and mutation of the monomer sequence. We present here an analogous, artificial evolutionary optimization process which we have applied to improve the functionality of cell penetrating peptide molecules. The "evolution" consisted of repeated rounds of in silico peptide sequence alterations using a genetic algorithm followed by in vitro peptide synthesis, experimental analysis and ranking according to their "fitness", i.e. their ability to carry the cargo carboxyfluorescein into cultured cells. The genetic algorithm-based optimization method was customized and adapted from former successful applications in the lab to realize an early convergence and a minimum number of in vitro and in silico processing steps by configured settings derived from empirical in silico simulation. We started out with 20 “lead peptides” which we had previously identified as top performers regarding their ability to enter cultured cells. Ten breeding rounds comprising 240 peptides each yielded a peptide population of which the top 10 candidates displayed a


3


Page 4 of 49

sixfold (median values) increase in its cell penetration capability compared to the top 10 lead peptides, and two consensus sequences emerged which represent local fitness optima. In addition, the cell penetrating potential could be proven independently of the carboxyfluorescein cargo in an alternative setting. Our results demonstrate that we have established a powerful optimization technology that can be used to further improve peptides with known functionality and adapt them to specific applications.


4

Page 5 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


INTRODUCTION: A broad variety of small, usually cationic or amphipathic peptides, collectively termed cell penetrating peptides (CPPs), are able to permeate cellular membranes and transport molecular cargo into living cells1. Besides protein derived sequences e.g. of viral origin2 and chimeric fusion peptide constructs3, rationally designed model peptides4–6 are described in the literature. Most of these CPPs bear a positive net charge under physiological conditions7,8, due to frequent presence of arginyl and lysyl side chains, or show a characteristic distribution of cationic and hydrophobic residues in helical wheel notation4,5. The majority of CPPs is built up from naturally occurring L-amino acids, but some rationally designed all-D- or mixed L-/D-peptides9 also have been reported. The mode of cellular entry of the peptides varies depending on their class, their concentration, the duration of the peptide-cell-interaction as well as on environmental factors like temperature and composition of the surrounding liquid10,11. Many peptides are taken up via endocytosis, but others seem to be internalized in a non-endocytotic manner12, and they vary in their subsequent


5


Page 6 of 49

cellular localization, ranging from the nucleus to the cytosol7,12,13. The nature of the cargo to be delivered, e. g. small molecule drugs, proteins, nucleic acids or nanomaterials, also plays a role in this process1,10, as does the cell type to be loaded. As the uptake of CPPs has been investigated in many different cellular systems – with HeLa cells, CHO cells and 3T3 fibroblasts as prominent examples8 – and under a wide variety of cultivation conditions, the comparability of different studies is rather poor8,14. Comparative experimental studies of CPPs under well-defined conditions are rare15 and usually deal with a limited number of CPP candidates with specified cargoes only, therefore general conclusions on structure-activity-relationships remain difficult. Although several successful attempts to predict CPPs have been reported16–19, no universal way to design CPPs for particular cell types or cargos exists, and in most cases the functionality of the predicted CPPs was demonstrated under specific conditions only20. Moreover, the number of factors potentially influencing CPP performance is huge and often neither measureable nor controllable, making a “theoretical” CPP prediction for a given application highly complex and often impossible19.


6

Page 7 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


An “experimental” approach, while strictly speaking only being valid for the delivery of the specifically tested cargo into the specifically investigated cells, carries a higher potential to identify an "optimal" CPP peptide for a desired application. In line with this notion, we recently performed a comparative study of known CPPs8,21 where we tested a total of 474 different CPPs and compared their capability to transport the cargo molecule carboxyfluorescein (FAM) into confluent HeLa cells. As a result, we could establish a ranking of those peptides according to their suitability for this specific task. Starting out with this ranking of known CPPs we now wanted to identify novel peptide sequence motifs with even better cell penetrating properties. Rather than relying on a mere screening approach using peptide libraries for this task – which would require a huge sequence space to be covered - we decided on a function-driven approach combined with a computational procedure22. As sequence information of linear peptides constitutes an ideal string type input for computer-based optimization processes, such as genetic algorithms, peptides are perfect compounds for computer-assisted molecular improvement strategies.


7


Page 8 of 49

With that in mind, we chose a refined peptide optimization methodology that mimics the Darwinian evolution23 to specifically enhance the transport-capacity of the CPPs we had investigated previously. As our goal was to provide the proof of concept for our procedure, we endeavored to keep the variables at a minimum and purposefully used only one cell line (HeLa cells) and one cargo (FAM) in the experimental determination of the respective peptides' cell penetrating activity.


8

Page 9 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


RESULTS AND DISCUSSION: In a former study, we have synthesized 474 linear cell penetrating L-peptides compiled from the literature and equipped them with a FAM-residue as model cargo. These peptide constructs were utilized for comparative cell uptake experiments21. This way we were able to rank the majority of at that time known CPPs8 according to their "fitness", which is defined in our system as their ability to carry FAM as traceable cargo into living HeLa cells, and which was computed from the fluorescence intensities measured in the internalization experiments. This ranked list of CPPs from the literature established the basis of the present optimization process which combines a genetic algorithm mimicking Darwinian evolution with a biological assay defining each candidate’s actual fitness for the predefined task. The 20 CPP candidates with the highest fitness values, i.e. the peptides resulting in the highest fluorescence intensities in our previous uptake studies, were selected as "leads" for a molecular optimization process by manual inspection and were used by the evolutionary tool as the starting point to generate the initial population23. The results of our


9


Page 10 of 49

preceding ranking study suggest that CPPs of a sequence length of 17-24 amino acids are most suitable for intracellular delivery, so we wanted to concentrate in our evolutionary process on a sequence length in this size range. As the genetic algorithm for the optimization of CPPs requires a fixed sequence length of peptides and cannot handle any variations thereof during one evolutionary process, we had to settle at the beginning of our study on one sequence length. We decided on 21 amino acids since the 20 CPPs chosen as lead peptides for the optimization had to be least modified (Table 1) when uniformly standardized to this length by addition of C- and/or N-terminal alanine residues (Table 1; green color) or by deletion of an N- or C-terminal amino acid segment (Table 1; red color). The fitness value of each of these modified CPPs was again determined in a standardized internalization experiment and utilized, together with the respective peptide sequences, to start the evolutionary optimization process23.


10

Page 11 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Table 1: Peptide sequences and fitness values expressed as relative fluorescence units (RFU) of the 20 lead peptides used in the evolutionary optimization process. Peptide Sequence * WLRRIKAWLRRIKALNRQLGVAA VKRKKKPALWKTLLKKVLKAA AAKTVLLRKLLKLLVRKIAAA AKKKKKKKKKKKKKKKKKKKA AAAKLALKLALKALKAALKAA RQARRNRRRALWKTLLKKVLKA KLALKLALKALKAALKLAAAA AALLKKRKVVRLIKFLLKAAA LIRLWSHLIHIWFQNRRLKWKKK LNSAGYLLGKINLKALAALAKKIL ACWKKKKKKKKKKKKKKKKKK AAYTAIAWVKAFIRKLRKAAA APKKKRKVALWKTLLKKVLKA GLWRALWRALRSLWKLKRKVA AGLWRALWRGLRSLWKKKRKV GLWRALWRGLRSLWKLKRKVA AKALAKALAKLWKALAKAAAA AKLAAALLKKWKKLAAALLAA GLFKALLKLLKSLWKLLLKAA AAKLALKLALKAWKAALKLAA

RFU 1011.42 925.42 855.56 632.01 478.40 450.97 440.41 435.30 425.60 422.13 417.94 401.31 390.81 384.64 375.17 347.66 345.58 335.89 326.66 309.34 *: green: alanine residues added to original sequence, red: amino acid residues deleted from original sequence In the development of evolution-based computational approaches, it is customary to use surrogate model systems in order to cut down the number of fitness functions necessary to lead to the desired results. For such an


11


Page 12 of 49

approach, different fitness function selection modi, that can be e.g. indicatorbased and aspect-based,24 can be chosen to ultimately simplify a multiobjective evolutionary algorithm. For our computational optimization procedure we decided on an evolutionary algorithm which suggests new peptide sequences based on a single fitness criterion only. The setup of the evolutionary algorithm parameters is determined by empirical verification in an in silico model system based on synthetic fitness landscapes using established evolutionary platforms25– 27.

In order to be able to handle the experimental in vitro determination of the

fitness values in a reasonable manner, the parameters for the evolutionary process are chosen in a way that convergence to an optimum can be reached in a minimal number of generations, and using a well manageable population size (i.e. number of peptides in each generation). The respective parameter setups are validated by configured settings derived from empirical simulation based on computed fitness landscapes. The structural diversity in the population of the lead peptides was evaluated by analyzing the frequency distribution of the amino acid building blocks at the


12

Page 13 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


respective sequence positions (Fig. 1). The map shows a frequent occurrence of lysine (K) at the positions 15 and 17 to 19. Due to conversion of all peptides into 21mers, both terminal ends are rich in alanine. Therefore, the sequence space covered by the population of these peptides is slightly reduced, leading to a decreased search space for optimized peptides and to potential limitations in fitness gain. At the same time, the pre-selection allowed us to start out at an already advanced level of cell penetrating abilities and increases our chances that the evolutionary fitness gain will further enhance this peptide characteristic within a minimal number of generations.

Figure 1. Distribution plot of amino acid frequencies at all sequence positions. High incidence of an amino acid at a


13


Page 14 of 49

certain sequence position is colored in red, equal distribution of amino acids is colored in purple. Color codes display how often a certain amino acid is found at the respective sequence position: █ 0; █ 1; █ 2; █ 3; █ 4; █ 5; █ 6; █ 7; █ 8; █ 9; █ 10; █ 11; █ 12; █ 14; █ 16.

The evolutionary process was started by submitting the 20 lead peptides to a software application that is based on an evolutionary algorithm23. Using the sequence information of the lead peptides as input, this software generates a population of filial peptide sequences by applying recombination, crossover and point mutation operations within an evolutionary operator setup27 derived from computational simulated synthetic landscapes. In accordance with the output of this in silico process, 240 filial peptides, each equipped with an N-terminal FAM modification, were synthesized. Cellular uptake of the filial peptides was determined in a HeLa cell assay as described previously21. Based on the results of these uptake experiments, the peptides were ranked according to their fitness, which was defined as the "fluorescence brightness value" that the cells yielded after incubation with the respective peptide.


14

Page 15 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


The 24 top-ranked peptides from this assay were selected as "parents" for the next breeding round and their sequences were processed in the evolutionary algorithm software26 to again yield 240 filial peptides which then were synthesized and tested in the same set-up. After following this procedure through ten rounds of the evolutionary optimization, the process yielded ten generations of stepwise improved cell penetrating peptides (see Supporting Information Table ST1 for peptide sequences). In order to ensure that increased cell penetration was truly due to the desired CPP action and not caused by or associated with cell membrane damage, we checked a selection of 20 wellperforming peptide candidates of each generation for any potential toxicity in a cell viability assay (see Supporting Information Figure S1). We found no correlation between peptide fitness and cell viability, thereby establishing that the CPP uptake in our system is not dependent on toxic or cell membrane damaging effects. Figure 2 depicts the relative cell penetrating fitness - in terms of fluorescence values obtained - of the top ten peptides from each generation in a boxplot


15


Page 16 of 49

style chart. In toto, within a ten-round evolutionary optimization process we could achieve a sixfold increase in CPP activity compared to the lead peptides (top ten of each generation; median values), with the best performing candidate of the evolved peptides having a fourfold higher cell uptake rate than the best candidate from the literature.

Figure 2. Increase of CPP fitness values in

the

course

of

evolutionary

optimization. The fitness of the 10 bestperforming generation fluorescence

peptides is

from

depicted units

in

(RFU).

each relative As

a

reference, "Lit. Pool" gives the values obtained in our assay system with 474 CPPs from literature sources (for details see

21).

(Boxplot: median, interquartile range, whiskers min to max).

As can be seen, the evolution starts out with a relatively steep increase in the first generations but levels out over time such that no further rise in


16

Page 17 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


median fluorescence is observed in generation eight to ten. This finding allows the conclusion that a local fitness optimum for our sequence pool has been reached at this point, which is supported by a low diversity in the sequence information of the best breeding peptides from these generations. The reshaping of the sequence space occurring during the evolutionary process was assessed by compiling the frequency distributions of the amino acids at the respective sequence positions for the best performing peptides from all generations (Fig. 3). In progression of the molecular optimization process, consensus motifs emerged, marked by an increasing incidence of related sequences with enhanced cell penetrating abilities. At the beginning of the process a generally elevated level of K and L amino acid building blocks at many sequence positions can be observed, largely characteristic for the original lead peptides. In generation 3, arginine becomes increasingly prominent at sequence positions 7, 10 and 11, and tryptophan is found to be frequent at position 8. This is not due to the formation of one early consensus sequence, but can be explained by the predominance of a limited number of sequence


17


Page 18 of 49

motifs within the fittest peptides. The top 24 breeding peptides of generation 4 are composed of six different motifs derived largely or partly from the original lead peptides. In the following generations the number of different sequence motifs within the top 24 breeding peptides steadily declines. At generation 6, the consensus sequence AXLXKTRKVVRLIKFLLKAXA (motif I), which is present in 18 of the 24 fittest peptides, has emerged. The core motif of this consensus is derived from one of the top 10 original lead peptides (E165, rationally designed28). As a second consensus sequence, VKNXXRXWLRKKKKKKKKKWK (motif II), is shared by six of the 24 top candidates in generation 6. This motif, which is not present in the original lead peptides, becomes more and more dominant in the next generations and almost eliminates the formerly prevailing sequence motifs in generation 10. Indeed, from generation 6 onwards, the evolutionary process was basically limited to a struggle between those two consensus sequence motifs. For that reason we conclude that two subpopulations with comparable fitness were formed during the optimization process. Hence, candidate peptides from both sequence clusters were


18

Page 19 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


evaluated in further assays to examine the features of the respective peptides in more detail.

Figure 3. Frequency distributions of amino acid residues in the 24 top peptides selected for breeding in the respective generations. The sequence positions 121 are plotted versus the respective amino acid residues (one-letter-code). Purple-blue staining indicates a uniform occurrence of all amino acids, yellowred reveals frequent occurrence of a certain amino acid at the respective sequence position. Color codes show how often a certain amino acid is found at the respective sequence position: █ 0; █ 1; █ 2; █ 3; █ 4; █ 5; █ 6; █ 7; █ 8; █ 9; █ 10; █ 11; █ 12; █ 14; █ 16.


19



Page 20 of 49

20

Page 21 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


As visual analysis of the cells at low microscopic resolution gave the impression of differences in the fluorescence pattern between the two subpopulations of CPPs, we wanted to resolve the intracellular distribution of CPPs by high resolution fluorescence imaging. For these experiments, one FAM conjugated peptide of each subpopulation was selected (ARLAKTRKVVRLIKFLLKATA for motif I, HKDRKRWWLWKKKKKKKKKNK for motif II) and purified by HPLC. Sets of pre-confluent HeLa cells were exposed each to one of these two purified CPP-FAM conjugates, cellular morphology was visualized in transmission light microscopy, fluorescence measurements were done in a pseudo-confocal process, and images acquired for a single position in different channels were stitched to obtain a multicolor image. As can be seen in Figure 4A, the peptide ARLAKTRKVVRLIKFLLKATA (motif I) has delivered its FAM cargo uniformly to the majority of the cells. The fluorescent conjugate appears to be located rather homogeneously in the cytosol of cells, with subcellular compartments being less prominently stained. In contrast, the CPP from the consensus motif II exhibits a more


21


Page 22 of 49

inhomogeneous cell uptake pattern, with some cells being efficiently penetrated and strongly stained, while others remain nearly unaffected by this delivery vehicle (Fig. 4B).


22

Page 23 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 4. Analysis of the intracellular distribution of optimized FAM-CPPs. Preconfluent HeLa cells were used to achieve a better resolution of cellular architecture. The cells were incubated for 30 min with 10 µM of a FAM-carrying representative of the first consensus motif (A, C) or of the second consensus motif (B, D). Cell nuclei were stained with Hoechst 33258 dye. An overlay of the fluorescence images and the bright field images of the cells is shown in A and B. Figures C and D show the mere fluorescence images to provide a better impression of the subcellular distribution of the fluorescent dye.

All in all, these observations confirm the notion that our optimization process produced two independent CPP motifs with comparable fitness values but possibly different cell penetration characteristics. This is also reflected in the helical wheel projections of the two consensus motifs (Supporting Information Figure S2). From these it can be concluded that motif I belongs to the group of amphipathic CPPs which is represented by for example transportan-like or KLA derived peptides. Motif II belongs to the group of cationic CPPs which is represented by short strands of arginine- and/or lysine-rich peptides below a sequence length of 30 amino acids. In the case of the peptides evolved here,


23


Page 24 of 49

it is mainly lysine which is highly abundant. We ascribe this fact to the already high presence of lysine in the population of lead peptides (Figure 1). In the progress of the optimization, the introduction of arginine (instead of lysine) obviously does not provide an evolutionary benefit that would be sufficient for the transition to this amino acid by random mutation in the final populations. In an independent set of experiments we aimed to analyze the relevance of the FAM moiety for the cell penetrating capability of our evolved CPPs. For this purpose, we developed a method to detect CPPs within HeLa cells independently of the presence of a fluorophore. A selection of original lead peptides, top peptides of generations 5 and 10 as well as various control peptides (Table 2) were equipped during solid phase peptide synthesis with two different tags, namely biotin and 2,4-dichlorophenoxyacetic acid (2,4-D)29,30. HeLa cells were then incubated with these dually labeled CPPs, washed and trypsinized to detach the cells from the plate support and simultaneously degrade any externally attached peptide material31. Subsequently, cells were lysed to release the cellular contents. For analysis, an anti-2,4-D antibody


24

Page 25 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(provided by M. Fránek, Veterinary Research Institute, Brno, CZ) and a horseradish peroxidase (HRP)-labeled streptavidin (SA) were used in an ELISAbased assay system to first capture and then detect the CPPs present in the HeLa cell lysate. Quantification of CPPs was achieved by comparison with a standard curve established with the pure double-labeled peptides. As this ELISA can - due to its set-up - only measure CPPs containing both, the biotin and the 2,4-D tag, intracellular proteolytic degradation of a peptide containing one tag on either end might compromise the assay outcome and lead to an underestimation of translocated peptide material. To rule out such an effect of intracellular peptide degradation, we additionally made use of CPPs labeled with both tags at one end of the peptides interconnected in a protease stable manner.

Table 2. Peptides from different generations of the evolutionary optimization process selected for FAM-independent uptake analysis. Fitness values (RFU) obtained in the FAM-uptake studies are given as reference. CPP ID Lead I Lead II Lead III GA5 I

Description Top 5, lead peptides Top 5, lead peptides Top 5, lead peptides Top 5, generation 5

Sequence KTVLLRKLLKLLVRKI VKRKKKPALWKTLLKKVLKA WLRRIKAWLRRIKALNRQLGVAA VKNEKRFWLRKKKKKKKKKWK


Fitness 855 (RFU) 925 1011 2246

25


GA5 II GA5 III GA10 I GA10 II GA10 III Control I Control II Control III

Top 5, generation 5 Top 5, generation 5 Top 5, generation 10 Top 5, generation 10 Top 5, generation 10 Bottom 5, generation 6 GA5 I, scrambled version good CPP generation 2-5,

Page 26 of 49

2713 2884 4006 3257 3459 50 n.d. n.d.

ANKLKKRKVVRLIKFLLGAAA ARLAKTRKVVRLIKFLLKATA HKDRKRWWLWKKKKKKKKKNK TKNEKPFWLRKKKKKKKKKWK HKDRKRWWLRGKKKKRKKKQK VKNNKRFKVVRLIKPYLGAAA VKRWKLKKKEKKKWKKNFKRK VAKAKLRLTLAIKKLAFARVL

scrambled

The results are summarized in Figure 5 and on the whole confirm the FAMCPP analyses. In particular, the CPPs of generation 10 are capable to translocate the 2,4-D/biotin tags into HeLa cells to a considerably larger extent than the original lead peptides. Comparison of the experiments utilizing the twoside-tagged, potentially protease-sensitive CPP constructs (blue bars in figure 5) with the one-side-tagged constructs (red bars in figure 5) shows that, for some sequence

motifs,

the

degradation-independent

one-side-tagged

construct

is

found in a markedly higher concentration than its two-side-tagged counterpart.

Figure 5. FAM-independent uptake analysis

of

CPPs.

CPPs

were

equipped either with a 2,4-D tag at the N-terminus and a biotin tag at the C-terminus (blue bars) or both tags

at


the

C-terminus

of

the

26

Page 27 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


peptide bridged via a proteolysis-resistant linker (red bars). Concentration of CPPs in cell lysates was determined via sandwich ELISA detecting both tags. (Mean values plus standard deviation of three independent experiments).

While detection of peptides GA10 I and GA10 II was rather unaffected by the different positioning of tags, peptides GA5 I of generation 5 and GA10 III of generation 10 are detectable inside the cells only in form of their degradationindependent constructs. Obviously, these two peptides do not display an inferior cell penetrating ability compared to the other generation 10 top breeding peptides but rather exhibit a higher susceptibility to proteolytic degradation which leads to “invisibility” in the ELISA when one tag is placed at each end of the peptides. However, apart from the issue of potential intracellular degradation, our FAM-independent assay set-up with the ELISA-detectable cargo substantiates an additional issue which needs to be considered for CPP evolution. Our results show that, while e.g. the CPPs from generation 10 which had performed excellently in the FAM-based uptake assay (GA10 I, II & III) also yielded high uptake rates in the dual-tag system, others, such as CPPs


27


Page 28 of 49

GA5 II and GA5 III which represented top candidates of generation 5 in the FAM-CPP analyses, performed only poorly in the dual-tag ELISA system, independent of tag localization. From this it can be concluded that CPPs evolved with one cargo (in our case FAM) are not always optimal for other applications (in our case with a 2,4D/biotin double tag freight) as alternative cargos will impose new molecular properties on CPPs1,19. Hence, the molecular features of the entire peptide conjugate used are important within an optimization process, and, consequently, universally suitable transport peptides are probably difficult to obtain. Even though, we could clearly demonstrate that it is feasible to tailor CPPs for defined cargoes with our evolutionary optimization procedure leading to specific transporters for the intended molecules and applications.


28

Page 29 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


EXPERIMENTAL PROCEDURES: Algorithm for peptide evolution: As in silico part for the molecular optimization process, an evolutionary algorithm23 was applied with the following settings which were tested in in silico performed simulations25,32,33: To process the fitness values, a linear selection method for recombination of the lead/parent peptides was chosen. Single fracture sites were set in the peptides from the parent population, two parent peptide fragments were recombined, one filial peptide randomly selected from each recombination event was used. The sequence strings ensuing from this process were subsequently mutated by single amino acid substitutions (mutation rate 12 %, choice of mutated amino acid: random). Sites of recombination and mutation were scattered over all 21 sequence positions in a Gaussian distribution, such that the majority of recombination and mutation events took place in the central region of the sequences. The population size of each generation was set to 240 peptides each; as mating pool for the next generation the 24 top candidates from the bioassay were chosen by manual inspection.


29


Page 30 of 49

The evolutionary operator setup was derived from computational simulation based on synthetic fitness functions. A single objective optimization was chosen because it had proven more efficient than many- and multi-objective cases, and an inclusion of more fitness goals would have slowed down the optimization process32,34. With that in mind, we have handled the criterion "cellular uptake" which in fact is a complex criterion that includes a variety of parameters - in this optimization process as a single objective criterion. Thus, our formally single objective optimization in fact is taking into account more than one objective, simply in a fixed weighted metric, consequently the different objectives are not optimized independently here. In the optimization process performed in this manner, all evolutionary generated solutions are feasible and by default encompass other - inherent - optimization constraints that could and were not experimentally accessed. Peptide synthesis: Peptides were synthesized in 96-well microplates on an amide resin by Fmoc solid phase synthesis techniques using standard amino acid building blocks as described before21 (see Supporting Information for


30

Page 31 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


details). Labels (Biotin (Fmoc-Glu(biotinyl-PEG)-OH); FAM (5(6)carboxyfluorescein); 2,4-D (6-(11-(2-(2,4-dichlorophenoxy)acetylamino)undecanoylamino)-2-(9H-fluoren-9-yl-methoxycarbonylamino)-hexanoic acid) )30 were attached to the peptides at the C- and/or N-terminus using standard coupling conditions. After synthesis and deprotection, the peptides were cleaved off the synthesis resin, precipitated, washed and dried in vacuo. The dried peptide conjugates were dissolved in 1 ml acetonitrile (ACN)/water (1:1 (v/v)) and their respective concentrations were determined photometrically or gravimetrically (see Supporting Information for details). Peptides were diluted to 50 µM stock solutions with 20% ethanol / 80% water. Peptide purification: Selected peptides were dissolved in water/ACN and purified by reverse phase chromatography (RP-HPLC) (LiChrospher 100 RP-18 column, 10 µm diameter, 250 x10 mm) on a Bischoff HPLC system (Leonberg, Germany) using optimized gradients typically ranging from 20% ACN to 40% ACN in water within 30 min. Peptide-containing fractions were collected, pooled,


31


Page 32 of 49

dried in vacuo and stored at 4°C. For usage, lyophilized peptides were dissolved in ACN/water (1:1 (v/v)). Cell Culture: The human cervix carcinoma (HeLa) cell line (I.A.Z., Munich, Germany) was cultured as described previously21 in DMEM/Ham’s F-12 (1:1) supplemented with 2 mM L-glutamine, 10% fetal bovine serum, and 1% penicillin/streptomycin (subsequently called complete cell culture (CCC) medium) at 37 °C and 5% CO2. For sub-culturing, cells were trypsinized and passaged to new culture flasks. After a maximum of 20 passages cells were discarded and replaced by freshly thawed HeLa cells to minimize cell age dependent effects. For testing uptake efficiency of CPPs, cells were trypsinized and subsequently seeded in 96-well µ-plates (ibidi, Munich, Germany) at a concentration of 6 x 104 cells per well. Cells were incubated for 48 h at 37 °C and 5% CO2, after which time they had reached confluence and experiments were performed. Uptake experiments with FAM-labeled CPPs: 10 µM FAM-CPP working solutions were prepared in 96-well polypropylene plates by diluting the 50 µM


32

Page 33 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


FAM-labeled CPP stock solutions with CCC medium directly before incubation of cells. Confluent HeLa cells grown in 96-well µ-plates were treated with 100 µl 10 µM FAM-CPP working solution per well for 90 min at 37 °C and 5% CO2. CCC medium alone served as a negative control. After incubation the cells were washed twice with PBS containing calcium and magnesium (CMPBS) and covered with 100 µl CM-PBS before determining fluorescence intensities in a plate reader at an excitation wavelength of 495 nm and an emission wavelength of 530 nm. The fluorescence readouts (relative fluorescence units, RFU) obtained thereby served as fitness value input in the evolutionary process. Transmission and epifluorescence images of the confluent cell layers after incubation with the CPPs were taken using a MORE Life Cell Imaging System (FEI Munich formerly Till Photonics, Gräfelfing, Germany) equipped with a 10x/0.45 objective (planapochromat 420640-9900-000, Zeiss, Jena, Germany), a FITC filterset (ET480_40x/ET535_50m/T510LPXRXT, Chroma, Rockingham, USA) and LED (transmitted light) and oligochrome (fluorescence) light sources (FEI Munich). 12-bit images were taken with a


33


Page 34 of 49

stingray F-145B CCD camera (Allied Vision Tech., Stadtroda, Germany) using live acquisition software (V. 2.2.2, FEI Munich). To minimize errors introduced by high background fluorescence (e.g. due to improper washing), the epifluorescence imaging data were used to check all top-30-ranked peptides from generation 3 onwards for bright diffuse background fluorescence. Utilizing a rolling ball (radius 50 pixels) software algorithm35 included in the image processing software FIJI36 (Life-Line version 2013 July 15, NIH, USA) background correction was achieved. Peptides with a mean grey value below 300 in their epifluorescence images after correction were classified as false positive results and excluded from the population of lead peptides, because their fitness was judged to be overestimated in our fluorescence reader-based analyses. High Resolution Fluorescence Imaging: HeLa cells were cultivated as described above, trypsinized, and subsequently 4 x 104 cells in 400 µl CCC medium were seeded in 35 mm glass bottom petri dishes. Cells were incubated for 48 h at 37 °C and 5% CO2, during which time they stay pre-confluent.


34

Page 35 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Cells were carefully washed 2x with 1.5 ml CCC medium and subsequently treated with 400 µl staining solution (10 µM FAM-labeled peptide and 5 µM Hoechst dye 33258 in CCC medium) for 30 min at 37 °C. Cells were carefully washed 2x with 1.5 ml CM-PBS and transferred to a microscope adjusted to 37 °C. 16-bit greyscale images were acquired using the MORE Life Cell Imaging System equipped with a 40x/1.2 water immersion objective (capochromat 421767-9970-000, Zeiss), a DAPI filterset (AT350_50x/ET460_50m/T400LP, Chroma) a FITC filterset (ET480_40x/ET535_50m/T510LPXRXT, Chroma) and an Andor Clara (Andor Technology Ltd. Belfast, UK) 16-bit CCD camera. At each position, image stacks across the complete cell height were recorded, single images from each stack from different channels (DAPI, FITC and transmitted light) of each position were combined into one image using the “Stitching” feature of a FIJI plugin37. Conversion from grayscale images into green (FAM fluorescence) and blue (Hoechst 33258) pictures was performed using the respective look-up tables (LUT) of the FIJI36 software package.


35


Page 36 of 49

Uptake experiments with 2,4-D/Biotin-labeled CPPs: Confluent HeLa cells grown in 96-well plates were incubated for 90 min at 37 °C and 5% CO2 with 100 µl 10 µM 2,4-D/Biotin-CPP working solution in CCC medium. After 2x washing with CM-PBS, cells were harvested via detachment by incubation for > 10 min with 0.05% trypsin / 0.02% EDTA in PBS. Suspended cells were counted, aliquots of 1 x 105 living cells were sedimented (500 x g, 5 min) and the cell pellets were treated for 30 min at 4 °C with a cell lysis buffer (RIPA buffer (Abcam, Cambridge, UK) containing 1:100 diluted protease inhibitor cocktail (Sigma-Aldrich, Steinheim, Germany) and 1 mM PMSF). Subsequently, cell debris was sedimented (15 min at 16,100 x g,), the supernatant was harvested, aliquoted and stored at -80 °C until analysis with an ELISA-based assay system was performed. Quantification of uptake of 2,4-D-/Biotin-labeled CPPs: Cellular lysis supernatants harvested after the uptake experiments with 2,4-D-/Biotin-labeled CPPs were analyzed by a capture ELISA29 (for details see Supporting Information). Plates were coated with anti-2,4-D antibody (clone F6/C10, M.


36

Page 37 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Fránek, Veterinary Research Istitute, Brno, Czech Republic) and blocked with 1% casein before being incubated with serially diluted cell lysis supernatants. Detection of bound peptide was done with HRP-labeled streptavidin and the substrate 3,3′,5,5′-tetramethylbenzidine38, and the absorbance at 450 nm was recorded. For conversion of the absorbance values into total peptide concentrations, the assay was performed in parallel with known concentrations of the respective CPPs to establish calibration curves. Cytotoxicity test: HeLa cells were treated with 100 µl of 10 µM FAM-labeled CPPs in CCC medium as described above, but incubation time was extended to 180 min. After CPP incubation, 10 µl of 2.5 mg/ml MTT (methylthiazolyldiphenyl-tetrazolium bromide, Sigma-Aldrich) in CCC medium were added and incubation was continued for another 2 h. Thereafter, 290 µl of extraction solution (93.1% isopropanol, 6.9% formic acid) were added and the cells were treated for 15 min at RT in an ultrasound bath. After cell lysis and centrifugation (3350 x g, 5 min), 200 µl of the supernatant were transferred to a 96-well polystyrene plate (Corning, Amsterdam, NL) and the absorbance at


37


Page 38 of 49

570 nm was determined. HeLa cells incubated in CCC medium without CPP were defined as 100% viable.

ACKNOWLEDGMENTS: We thank Dr. Milán Franek, Veterinary Research Institute, Brno, Czech Republic, for providing the anti-2,4-D antibody. This work was funded by the German Ministry for Education and Research (BMBF), grant No. 13N10505.

SUPPORTING INFORMATION DESCRIPTION: Supporting Information is available: Details on experimental procedures (peptide synthesis, determination of peptide concentration, capture ELISA); correlation between CPP uptake and cytotoxicity (Figure S1); helical wheel projection of consensus peptide motifs I & II (Figure S2); Table S1 of all peptide sequences and fluorescence values


38

Page 39 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


REFERENCES (1)

Copolovici, D. M., Langel, K., Eriste, E., Langel, Ü. (2014) Cell-

Penetrating Peptides: Design, Synthesis, and Applications. ACS Nano 8, 1972– 1994. (2)

Elliott, G., O’Hare, P. (1997) Intercellular Trafficking and Protein Delivery

(3)

Myrberg, H., Zhang, L., Mae, M., Langel, U. (2008) Design of a Tumor-

(4)

Scheller, A., Oehlke, J., Wiesner, B., Dathe, M., Krause, E., Beyermann,

by a Herpesvirus Structural Protein. Cell 88, 223–233.

Homing Cell-Penetrating Peptide. Bioconjug. Chem. 19, 70–75.

M., Melzig, M., Bienert, M. (1999) Structural Requirements for Cellular Uptake of Alpha-Helical Amphipathic Peptides. J. Pept. Sci. 5, 185–194. (5)

Milletti, F. (2012) Cell-Penetrating Peptides: Classes, Origin, and Current

(6)

Lundberg, P., El-Andaloussi, S., Sütlü, T., Johansson, H., Langel, Ü.

Landscape. Drug Discov. Today 17, 850–860.

(2007) Delivery of Short Interfering RNA Using Endosomolytic Cell-Penetrating Peptides. FASEB J. 21, 2664–2671. (7)

Futaki, S., Suzuki, T., Ohashi, W., Yagami, T., Tanaka, S., Ueda, K.,

Sugiura, Y. (2001) Arginine-Rich Peptides. An Abundant Source of MembranePermeable

Peptides

Having

Potential

Delivery. J. Biol. Chem. 276, 5836–5840. (8)

as

Carriers

for

Intracellular

Protein

Gautam, A., Singh, H., Tyagi, A., Chaudhary, K., Kumar, R., Kapoor, P.,

Raghava, G. P. (2012) CPPsite: A Curated Database of Cell Penetrating Peptides. Database Oxf. 2012, bas015. (9)

Wender, P. A., Mitchell, D. J., Pattabiraman, K., Pelkey, E. T., Steinman,

L., Rothbard, J. B. (2000) The Design, Synthesis, and Evaluation of Molecules

That Enable or Enhance Cellular Uptake: Peptoid Molecular Transporters. Proc.

Natl. Acad. Sci. U.S.A. 97, 13003–13008. (10)

Bechara, C., Sagan, S. (2013) Cell-Penetrating Peptides: 20 Years Later,

Where Do We Stand? FEBS Lett. 587, 1693–1702.


39


(11)

Page 40 of 49

Madani, F., Lindberg, S., Langel, U., Futaki, S., Graslund, A. (2011)

Mechanisms

of

Cellular

Uptake

of

Cell-Penetrating

Peptides.

414729. (12)

J. Biophys.

Ter-Avetisyan, G., Tunnemann, G., Nowak, D., Nitschke, M., Herrmann,

A., Drab, M., Cardoso, M. C. (2009) Cell Entry of Arginine-Rich Peptides Is Independent of Endocytosis. J. Biol. Chem. 284, 3370–3378.

(13)

Hariton-Gazal, E., Feder, R., Mor, A., Graessmann, A., Brack-Werner, R.,

Jans, D., Gilon, C., Loyter, A. (2002) Targeting of Nonkaryophilic CellPermeable Peptides into the Nuclei of Intact Cells by Covalently Attached Nuclear Localization Signals. Biochemistry (Mosc.) 41, 9208–9214. (14)

Stalmans, S., Wynendaele, E., Bracke, N., Gevaert, B., D’Hondt, M.,

Peremans, K., Burvenich, C., De Spiegeleer, B. (2013) Chemical-Functional Diversity in Cell-Penetrating Peptides. PLoS ONE 8, e71752. (15)

Mueller,

J.,

Kretzschmar,

I.,

Volkmer,

R.,

Boisguerin,

P.

(2008)

Comparison of Cellular Uptake Using 22 CPPs in 4 Different Cell Lines.

Bioconjug. Chem. 19, 2363–2374. (16)

Lindgren, M., Hallbrink, M., Prochiantz, A., Langel, U. (2000) Cell-

(17)

Hansen, M., Kilk, K., Langel, Ü. (2008) Predicting Cell-Penetrating

(18)

Sanders, W. S., Johnston, C. I., Bridges, S. M., Burgess, S. C.,

Penetrating Peptides. Trends Pharmacol. Sci. 21, 99–103.

Peptides. Adv. Drug Deliv. Rev. 60, 572–579.

Willeford, K. O. (2011) Prediction of Cell Penetrating Peptides by Support Vector Machines. PLOS Comput. Biol. 7, e1002101. (19)

Reissmann, S. (2014) Cell Penetration: Scope and Limitations by the

(20)

Almeida, P. (2011) Tools for Predicting Binding and Insertion of CPPs

Application of Cell-Penetrating Peptides. J. Pept. Sci. 20, 760–784.

into Lipid Bilayers. In Cell-Penetrating Peptides, Langel, Ü., Ed., Methods in Molecular Biology, Humana Press, pp 81–98. (21)

Ramaker, K., Henkel, M., Krause, T., Röckendorf, N., Frey, A. (2018)

Cell Penetrating Peptides: A Comparative Transport Analysis for 474 Sequence Motifs. Drug Deliv. 25, 928–937.


40

Page 41 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(22)

Singh, J., Ator, M. A., Jaeger, E. P., Allen, M. P., Whipple, D. A.,

Soloweij, J. E., Chowdhary, S., Treasurywala, A. M. (1996) Application of Genetic Algorithms to Combinatorial Synthesis:

A Computational Approach to

Lead Identification and Lead Optimization. J. Am. Chem. Soc. 118, 1669–1676. (23)

Röckendorf, N., Borschbach, M., Frey, A. (2012) Molecular Evolution of

Peptide

Ligands

with

Custom-Tailored

Characteristics

for

Glycostructures. PLoS Comput. Biol. 8, e1002800.

(24)

Targeting

of

Rosenthal, S., Borschbach, M. (2018) Indicator-Based Versus Aspect-

Based Selection in Multi- and Many-Objective Biochemical Optimization. In

Bioinspired Optimization Methods and Their Applications, Lecture Notes in Computer Science, Springer, Cham, pp 258–269. (25)

Rosenthal, S., Borschbach, M. (2014) Impact of Population Size and

Selection within a Customized NSGA-II for Biochemical Optimization Assessed on the Basis of the Average Cuboid Volume Indicator. Proc. of the Sixth

International Conference on Bioinformatics, Biocomputational Systems and

Biotechnologies, BIOTECHNO 2014, 1-7. (26)

Rosenthal,

S.,

Borschbach,

M.

(2014)

Impact

of

Population

Size,

Selection and Multi-Parent Recombination within a Customized NSGA-II and a Landscape Analysis for Biochemical Optimization. Int. J. Adv. Life Sci. 6, 310324. (27)

Rosenthal, S., Borschbach, M. A (2015) Concept for Real-Valued Multi-

Objective Landscape Analysis Characterizing Two Biochemical Optimization Problems.

In

Applications of Evolutionary Computation, Lecture Notes in

Computer Science, Springer, Cham, pp 897–909. (28)

Hällbrink, M., Kilk, K., Elmquist, A., Lundberg, P., Lindgren, M., Jiang, Y.,

Pooga, M., Soomets, U., Langel, Ü. (2005) Prediction of Cell-Penetrating Peptides. Int. J. Pept. Res. Ther. 11, 249–259. (29)

Gorris, H. H., Bade, S., Röckendorf, N., Albers, E., Schmidt, M. A.,

Fránek, M., Frey, A. (2009) Rapid Profiling of Peptide Stability in Proteolytic Environments. Anal. Chem. 81, 1580–1586.


41


(30)

Page 42 of 49

Bade, S., Röckendorf, N., Franek, M., Gorris, H. H., Lindner, B., Olivier,

V., Schaper, K.-J., Frey, A. (2009) Biolabeling with 2,4-Dichlorophenoxyacetic Acid Derivatives: The 2,4-D Tag. Anal. Chem. 81, 9695–9702. (31)

Burlina, F., Sagan, S., Bolbach, G., Chassaing, G. (2005) Quantification

of the Cellular Uptake of Cell-Penetrating Peptides by MALDI-TOF Mass Spectrometry. Angew. Chem. Int. Ed. 44, 4244–4247.

(32)

Rosenthal, S., El-Sourani, N., Borschbach, M. (2013) Impact of Different

Recombination

Methods

in

a

Mutation-Specific

MOEA

for

a

Biochemical

Application. In Evolutionary Computation, Machine Learning and Data Mining in

Bioinformatics, Lecture Notes in Computer Science, Springer, Berlin, Heidelberg, pp 188–199. (33)

Rosenthal, S., El-Sourani, N., Borschbach, M. (2012) Introduction of a

Mutation Specific Fast Non-Dominated Sorting GA Evolved for Biochemical Optimizations. In Simulated Evolution and Learning, Lecture Notes in Computer Science, Springer, Berlin, Heidelberg, pp 158–167. (34)

Rosenthal, S., Freisleben, B., Borschbach, M. (2018) Aggregate Selection

in Multi-Objective Biochemical Optimization via the Average Cuboid Volume Indicator. In EVOLVE - A Bridge between Probability, Set Oriented Numerics,

and

Evolutionary

Computation

VI,

(35)

Sternberg, S. R. (1983) Biomedical Image Processing. Computer 16, 22–

Advances

in

Intelligent

Systems

and

Computing, Springer, Cham, pp 3–17. 34. (36)

Schindelin, J., Arganda-Carreras, I., Frise, E., Kaynig, V., Longair, M.,

Pietzsch, T., Preibisch, S., Rueden, C., Saalfeld, S., Schmid, B., et al. (2012) Fiji: An Open-Source Platform for Biological-Image Analysis. Nat. Methods 9,

676–682. (37)

Preibisch,

S.,

Saalfeld,

S.,

Tomancak,

P.

(2009)

Globally

Optimal

Stitching of Tiled 3D Microscopic Image Acquisitions. Bioinformatics 25, 1463– 1465. (38)

Frey, A., Meckelein, B., Externest, D., Schmidt, M. A. (2000) A Stable

and Highly Sensitive 3,3’,5,5’-Tetramethylbenzidine-Based Substrate Reagent for Enzyme-Linked Immunosorbent Assays. J. Immunol. Methods 233, 47–56.


42

Page 43 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60



43


Figure 1. Distribution plot of amino acid frequencies at all sequence positions. 210x215mm (300 x 300 DPI)


Page 44 of 49

Page 45 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 2. Increase of CPP fitness values in the course of evolutionary optimization. 117x150mm (300 x 300 DPI)



Figure 3. Frequency distributions of amino acid residues in the 24 top peptides selected for breeding in the respective generations.


Page 46 of 49

Page 47 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 4. Analysis of the intracellular distribution of optimized FAM-CPPs.



Figure 5. FAM-independent uptake analysis of CPPs. 164x162mm (300 x 300 DPI)


Page 48 of 49

Page 49 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Table of Contents Graphic 73x50mm (300 x 300 DPI)


Breeding Cell Penetrating Peptides: Optimization of Cellular Uptake

Recommend Documents