Elastin-Like Peptide Dendrimers: Design, Synthesis, and Applications

Home , Peptide synthesis

Mingjun Zhou

Dissertation submitted to the faculty of the Virginia Polytechnic Institute and State University in

partial fulfillment of the requirements for the degree of

Doctor of Philosophy

Chemistry Department

John B. Matson, Committee Chair

Kevin J. Edgar

Webster L. Santos

Tijana Z. Grove

April 30, 2019

Blacksburg, VA

Keywords: Elastin-Like Peptide, Dendrimer, Hydrogel, H2S, Peptide-Polymer conjugate

Elastin-Like Peptide Dendrimers: Design, Synthesis, and Applications

Mingjun Zhou

ABSTRACT

Elastin like peptides (ELPs)—derived from the protein elastin—are widely used as thermoresponsive components in biomaterials due to their LCST (lower critical solution temperature) behavior at a characteristic transition temperature (Tt). While linear ELPs have been well investigated, few reports focused on branched ELPs. Using lysine (Lys, with an additional side-chain amine) as branching units, ELP dendrimers were synthesized by solid-phase peptide synthesis (SPPS) with up to 155 amino acid residues. A secondary structure change with decreasing ratio of random coil and increasing ratio of β-turn upon heating, which is typical of linear ELPs, was confirmed by circular dichroism spectroscopy for all peptides. Conformational change did not show evident dependence on topology, while a higher Tt was observed for dendritic peptides than for their linear control peptides with the same number of GLPGL repeats. Variable- temperature small-angle X-ray scattering (SAXS) measurements showed a size increase and fractal dimension upon heating, even below the Tt. These results were further confirmed by cryogenic transmission electron microscopy (cryo-TEM), and micro differential scanning calorimetry

(micro-DSC), revealing the presence of aggregates below the Tt. These results indicated the presence of a pre-coacervation step in the LCST phase transition of the ELP dendrimers.

We further prepared hydrogels by crosslinking hyaluronic acid (HA) with ELP dendrimers. We invesigated their physical properties with scanning electron microscopy (SEM), swelling tests,

SAXS, and model drug loading/release experiments. Most of the HA_denELP hydrogels retained transparent upon gelation, but after lyophilization and reswelling remained opaque for days. This reswelling process was carefully investigated with time-course SAXS studies, and was attributed

to forming pre-coacervates in the gelation step, which slowly reswelled during rehydration. We then prepared hydrogels with H2S-releasing aroylthiooxime (SATO) groups and showed human neutrophil elastase-responsive H2S-releasing properties with potential applications in treating chronic diseases with recurring inflammation.

Furthermore, we prepared a series of wedge-shaped triblock polyethylene glycol (PEG)-ELP dendrimer-C16 (palmitic acid) conjugate amphiphiles with adjustable Tts. Various techniques were used to investigate their hierarchical structures. The triblock PEG-peptide-C16 conjugate amphiphiles were thermoresponsive and showed a morphology change from small micelles to large aggregates. However, the hydrophilic shell and strong tendency for micelle formation limited the thermoresponsive assembly, leading to slow turbidity change in the LCST transition. The secondary structure was twisted from conventional β-sheet, and the thermoresponsive trend observed in typical ELP systems was not observed, either. Variable temperature NMR showed evidence for coherent dehydration of the PEG and ELP segments, probably due to the relatively short blocks. Utilizing the micelles with hydrophobic cavity, we were able to encapsulate hydrophobic drugs, with promising applications for localized drug release in hyperthermia.

Elastin-Like Peptide Dendrimers: Design, Synthesis, and Applications

Mingjun Zhou

GENERAL AUDIENCE ABSTRACT

Elastin like peptides (ELPs) are similar to the protein elastin in terms of amino acid sequence.

They are used widely as thermoresponsive (change in properties at different temperatures) components in biomaterials due to their abnormally lower solubility at higher temperatures. While linear ELPs have been thoroughly investigated, few investigations in ELP dendrimers have been studied. Dendrimers are molecules that branch in a controlled way to achieve sphere-like structures with rich surface functionalities. We synthesized the ELP dendrimers by using lysine amino acids as branching units. A protein secondary structure change, typical of ELPs, was observed for all peptide dendrimers. Secondary structure transitions showed no dependence on the molecular branching/linear structures, but a higher transition temperature (Tt) was observed for dendritic peptides than for their linear control peptides with the same number of amino acids. Various techniques confirmed the existence of aggregates below the Tts, which was never reported before.

We further fabricated hydrogels that mimic the native extracellular matrix, by connecting hyaluronic acid (HA) with ELP dendrimers. Interestingly, most of the hydrogels studied retained transparent upon gelation, but after freeze-drying and addition of water remained opaque for days.

This phenomenon was attributed to forming of small aggregates in the gelation step, which resulted in slow reswelling. We then prepared hydrogels with H2S-releasing groups, which showed human neutrophil elastase-responsive H2S-releasing properties with potential applications in treating chronic diseases with recurring inflammation.

We then prepared a series of wedge-shaped triblock poly (ethylene glycol) (PEG)- ELP dendrimer- alkyl chain molecules. The triblock molecules were thermoresponsive and showed a change from small spheres to large aggregates. However, the hydrophilic shell limited the thermoresponsive assembly, leading to slow turbidity change in the LCST transition. We found evidence of coherent assembly of the PEG and ELP parts, probably due to the relatively short polymer chains. Utilizing the micelles with hydrophobic cavity, we were able to encapsulate hydrophobic drugs, with promising applications for localized drug release for cancer treatment.

Acknowledgements

Firstly I want to thank my advisor Dr. John B. Matson for his consistent support and help. So many times, I wondered whether I could make it to finish my PhD without him.

Then I want to thank my committee members Dr. Tijana Grove, Dr. Kevin Edgar, Dr. Webster

Santos, and Dr. S. Richard Turner for their patience and time along the way, as well as Dr. Michael

Schulz for his kind help.

Thanks to our Israel collaborators Dr. Ronit Bitton, Yulia Shmidov and Yotam Navon for their hard work and contributions to our projects, as well as hosting us during the 2015 visit.

Thanks to Dr. Keith Ray and Dr. Richard Helm for their years of patient support on MALDI-TOF and not charging us.

Thanks to Greg LeBlanc from CEM for his help with the maintenance of our peptide synthesizer.

Thanks to Yin, Yun, Scott, Jeff, Jenn, Kyle, Chad, Kuljeet, Jeff 2, Mohammed, Ryan, Kearsley,

Sam, Jackie, Zhao, Sara B/W and Anastasia for their daily help and being good lab mates, and

Mariann as well as Hazel for the happy party time.

Thanks to Xiuli, Chengzhe, Junyi, Xi, Jennifer, Yumin and Xiaozhou for their help in my research.

Finally, I would like to thank my family and Na for supporting me through the entire period of time.

It has been a long and difficult journey; I am glad I have made it so far. Thanks to all of those who were helpful and nice to me and made my life easier, and thanks to all of those who were not and taught me what real life is.

Table of Contents

Acknowledgements...... iv

Table of contents...... v

Attribution...... viii

Chapter 1. Introduction to Dissertation...... 1

1.1. The lower critical solution temperature (LCST)...... 1

1.2. Elastin-like peptides ...... 2

1.3. Dendrimers ...... 4

1.4. Hydrogels ...... 5

1.5. Peptide-polymer conjugate...... 6

1.6. H2S-releasing materials...... 7

1.7. References...... 8

Chapter 2. Dendritic Elastin-Like Peptides: The Effect of Branching on Thermoresponsiveness

...... 14

2.1. Abstract ...... 14

2.2. Introduction ...... 15

2.3. Results and discussions ...... 17

2.4. Conclusions ...... 31

2.5. Experimental section ...... 32

2.6. References ...... 33

2.7. Appendix...... 36

Chapter 3: Multi-Scale Characterization of Thermoresponsive Dendritic Elastin-Like Peptides

vii

...... 52

3.1. Abstract……...... 52

3.2. Introduction………………...... 53

3.3 Results and discussions...... 55

3.4. Conclusions……...... 68

3.5. Experimental section…………...... 69

3.6. References…………...... 72

3.7. Appendix………………...... 76

Chapter 4: Hydrogels Composed of Hyaluronic Acid and Dendritic ELPs: Hierarchical Structure and Physical Properties………...... 84

4.1. Abstract…………………………………...... 84

4.2. Introduction………………………………...... 85

4.3. Experimental section...... 87

4.4. Results and discussions...... 90

4.5. Conclusions ...... 104

4.6. References ...... 105

4.7. Appendix ...... 108

Chapter 5: Thermoresponsive Wedge-shaped Triblock Peptide-polymer-C16 (palmitic acid)

Conjugate Amphiphiles: Phase and Drug Delivery Behavior...... 113

5.1 Abstract...... 113

5.2 Introduction ...... 114

5.3 Experimental section...... 115

5.4. Results and discussions...... 119

viii

5.5. Conclusions ...... 128

5.6. References ...... 128

5.7. Appendix ...... 131

Chapter 6: H2S Release of Covalent Peptide Hydrogels Triggered by Human Neutrophil Elastase

(HNE)…...... 138

6.1. Abstract ...... 138

6.2. Introduction ...... 139

6.3. Experimental section...... 140

6.4. Results and discussions...... 146

6.5. Conclusions……………………………...... 154

6.6. References...... 155

6.7. Appendix ...... 157

Chapter 7: Conclusions and future work...... 161

Attribution

Several collaborators were involved in several of the chapters of this dissertation. Their background and contributions are included here.

Chapter 2: Yotam Navon (graduate student, Department of Chemical Engineering and the Ilze

Katz Institute for Nanoscale Science and Technology, Ben-Gurion University of the Negev) characterized most of the peptides. Mingjun Zhou (graduate student, Department of Chemistry,

Virginia Tech) contributed to synthesize all of the peptides along with the MS, and characterize the linear control peptide, as well as the logD calculation and discussion (with Dr. John Matson).

John Matson (Ph.D., Department of Chemistry, and Macromolecules Innovation Institute (MII) at

Virginia Tech) and Ronit Bitton (Department of Chemical Engineering and the Ilze Katz Institute for Nanoscale Science and Technology, Ben-Gurion University of the Negev) are the advisors providing guidance, and discussion/writing of the chapter.

Chapter 3: Mingjun Zhou (graduate student, Department of Chemistry, Virginia Tech) contributed to synthesize all of the peptides along with the MS. Yulia Shmidov (graduate student,

Department of Chemical Engineering and the Ilze Katz Institute for Nanoscale Science and

Technology, Ben-Gurion University of the Negev) contributed to characterize all of the peptides.

John Matson (Ph.D., Department of Chemistry, and MII at Virginia Tech) and Ronit Bitton

(Department of Chemical Engineering and the Ilze Katz Institute for Nanoscale Science and

Technology, Ben-Gurion University of the Negev) are the advisors providing guidance. Everyone contributed in the discussion and completion of the chapter.

Chapter 4: Yulia Shmidov (graduate student, Department of Chemical Engineering and the Ilze

Katz Institute for Nanoscale Science and Technology, Ben-Gurion University of the Negev)

x contributed to characterize all of the peptides and hydrogels. Mingjun Zhou (graduate student,

Department of Chemistry, Virginia Tech) contributed to synthesize all of the peptides along with the MS. John Matson (Ph.D., Department of Chemistry, and MII at Virginia Tech) and Ronit

Bitton (Department of Chemical Engineering and the Ilze Katz Institute for Nanoscale Science and

Technology, Ben-Gurion University of the Negev) are the advisors providing guidance. Everyone contributed in the discussion and completion of the chapter.

Chapter 5: Mingjun Zhou (graduate student, Department of Chemistry, Virginia Tech) contributed to design and synthesize all of the samples along with the MS, most of other characterization and chapter writing. Yulia Shmidov (graduate student, Department of Chemical

Engineering and the Ilze Katz Institute for Nanoscale Science and Technology, Ben-Gurion

University of the Negev) contributed to cryo-TEM and SAXS results. John Matson (Ph.D.,

Department of Chemistry, and MII at Virginia Tech) is the advisor providing guidance.

Chapter 6: Mingjun Zhou (graduate student, Department of Chemistry, Virginia Tech) contributed to design (with the help of Yun) and synthesize the samples along with most of the characterization and chapter writing. Yun Qian (graduate student, MII at Virginia Tech) contributed to the calibration curve, cell study and chapter writing in the corresponding sections, including part of introduction. John Matson (Ph.D., Department of Chemistry, and MII at Virginia

Tech) is the advisor providing guidance.

Chapter 1: Introduction to Dissertation

1.1 The lower critical solution temperature (LCST)

Smart materials that are responsive to external stimuli, including pH, salt, solvent, temperature, light, electric/magnetic field, have attracted broad research interest due to their significance in fundamental science as well a variety of applications.1 Thermoresponsive materials are a unique class of materials that go through structure/property change in response to temperature change. In general, thermoresponsiveness includes LCST and the upper critical solution temperature (UCST).

As shown in Figure 1.1, LCST-type polymers become insoluble above a transition temperature

(Tt), while UCST-type polymers become insoluble below a transition temperature.

Figure 1.1 Temperature vs. polymer volume fraction (ϕ). Schematic illustration of phase diagrams for polymer solution (a) lower critical solution temperature (LCST) behavior and (b) upper critical solution temperature (UCST) behavior.2

The phase transition occurs as a result of the competition between the hydrophobic interaction of polymer chains and the hydrogen bonding among polymers and surrounding polar solvents (mostly water). Take LCST in aqueous systems as example. Below the LCST, the polymer chains are

1 soluble and stabilized by hydrogen bonding of polar groups with water; at the same time water molecules around hydrophobic groups are aligned in high order. Above the LCST, due to the fact that hydrogen bonding is weakened, hydrophobic interactions between polymer chains prevail, leading to coacervates and the release of water molecules. This process is entropy-driven due to releasing small molecules. Because of its close relationship with the hydrophobic/hydrophilic groups, the LCST is sensitive to the polymer hydrophobicity/concentration, as well as the ionic concentrations. A straightforward example is with the extra hydrophobic methyl group in the repeating unit, poly(propylene oxide) (PPO) generally has a lower LCST than poly(ethylene glycol) (PEG).2 Generally, due to the interest for biological applications, thermoresponsiveness has been thoroughly investigated in aqueous systems, but it may apply to organic solvents as well.2

Over the past decades, interest in these materials has increased due to their various potential applications, including drug delivery,3,4 sensing,5 regenerative medicine,6 protein separation7 and others.8 Typical LCST polymers are based on poly(N-isopropylacrylamide) (PNIPAM), PEG,

PPO, poly(N,N-diethylacrylamide) (PDEAM), poly(methyl vinyl ether)(MVE) and poly(N- vinylcaprolactam)(PNVCl).1 Similar to synthetic polymers, elastin-like peptides (ELPs) are another series of biopolymer that exhibit LCST.

1.2 Elastin-like peptides

ELPs are artificial biopolymers from the hydrophobic part of tropoelastin,9-11 which is composed of Xaa–Pro–Gly–Xaa–Gly (XPGXG) pentameric peptide repeats, where X, the guest residue, can be any amino acid but proline.12 This repeating pentapeptide structure imparts LCST behavior as discussed above, but UCST ELPs are also reported.13 Peptides bearing the XPGXG repeating sequence are typically soluble below a characteristic LCST, also known as the inverse transition temperature (Tt), and aggregate into large clusters above the Tt. This aggregation is a result of a

2 change in ELP secondary structure. ELPs exist primarily in a random coil conformation below the

Tt and more type II β-turns (also known as β-spirals) at higher temperature (Figure 1.2). The transformation is generally reversible and can take place in a temperature window as narrow as

3 C.9-11,14 However, although there is a well-defined LCST of macroscopic turbidity in most case of ELPs, the secondary structure changes over a very broad temperature range, even lower than

LCST.15

Many intrinsic and extrinsic factors influence the Tt of a given ELP, including the specific ELP sequence, the molecular weight of the polypeptide (number of pentapeptide repeats), as well as

ELP concentration, solution pH, and ionic strength.16-18 For example, increasing ELP

12,19,20 concentration in solution typically leads to lower Tts, as does increasing ELP molecular weights.19-21 Similarly, the characteristics of the guest residue (X in XPGXG) also have an evident effect on Tt: hydrophobic guest residues (Trp and Tyr being the most effective) decrease the Tt, while hydrophilic guest residues elevate it.11,20

Figure 1.2 A typical thermoresponsive secondary structure of ELP.15

Over the past decade ELPs have been designed to create complex nanostructures and materials such as spherical22,23 and cylindrical micelles24, fibers25, vesicles,26,27 and gels.28 In all cases,

3 however, the basic ELP component is a linear peptide. Recently several groups have begun studying ELPs in combination with branched polymer architectures. For example, Higashi et al. synthesized a peptide/dendrimer hybrid that contained thermoresponsive oligo(ELP)s at the periphery of poly(amidoamine) (PAMAM) dendrimers.29 In a subsequent report, Kojima et al.30 studied the effect of the generation number of the PAMAM dendrimers on Tt. These dendrimers were then utilized to create dual stimuli-responsive nanoparticles by loading photo thermogenic

AuNPs into their core.31 Ghoorchian et al. reported the synthesis of an amphiphilic ELP trimer consisting of three (GVGVP)n arms as hydrophobic segments, connected with a compact charged domain as the hydrophilic head group. Temperature changes drove micelle formation above the Tt of the ELP.24

To the best of our literature search, branched ELPs that do not contain other polymeric or peptide species have not been investigated. In fact, aside from the reports above, little is known about the effect of the molecular architecture on the thermoresponsive behavior of ELPs. Dendritic, hyperbranched, star, and other highly branched structures are important in biomaterials because they are often used as crosslinkers to make polymeric hydrogels.32 Moreover, globular and compact topologies such as dendrimers have very different physical properties (solubility, viscosity) from their linear counterparts. This topological difference may lead to useful properties in biomaterials, including substantially improved resistance to proteolysis of branched peptides compared to linear ones.33 Fundamental studies evaluating the effect of branching on ELP transitions may enable the design of new thermoresponsive materials.

1.3 Dendrimers

Branching structures are widely used in nature over many length scales. Examples include neural

/vascular systems, as well as tree branches, roots, and leaf veins. Inspired by branching structures,

4 chemists have developed highly branched dendritic polymers (dendrimers) 34,35. Characterized by their spherical structure and rich surface functionality with exponential branching, dendrimers have attracted lots of research interest aiming to apply them as catalysts 36,37, drug delivery systmes

38,39, and precisely controlled nano-objects 40-42.

Dendrimers with static branching structures, mostly polyamidoamine (PAMAM) dendrimers, have been widely studied 43-45. However, not many responsive dendrimers have been reported 46-48, while responsive polymers have attracted lots of research interest, with external stimuli including temperature, pH, light 49,50. Among these systems, thermoresponsive polymers have been thoroughly investigated, with applications including gene/drug delivery 51-53, tissue engineering

54,55, bio-separation 56-58, as well as sensors 59-61. Till now, most research on thermoresponsive dendrimers focused on connecting peripheral thermoresponsive components onto dendrimer cores

31,62-65. Dendrimers being thermoresponsive in the entire structure, which are more challenging to synthesize, may enable the design of new smart materials. Therefore, we were motivated to obtain insights into the phase transition of the thermoresponsive dendrimers responsive to changes in the self-assembly of the branching units throughout. Besides, with rich surface functionality, dendrimers are good candidates as crosslinkers of hydrogels.

1.4 Hydrogels

Hydrogels are three-dimensional networks of hydrophilic polymers with a large content of water, characterized by a higher storage modulus than loss modulus. In order to achieve gelation, the formation of a network with infinite high molecular weight is required. The crosslinking process generally includes physical/covalent crosslinking. Covalent crosslinking is to form the network via covalent bond from components with high functionality. Gel point calculation provides useful guidance for hydrogel design: both high number of functionality and high reaction conversion

5 favor gelation.66 On the other hand physical crosslinking is mostly achieved with weak interactions, such as hydrogen bonding, hydrophobic interaction, pi-pi stacking, and electrostatic interaction.

Similar gelation rules apply to physical hydrogels as well, for example, trivalent cations form stiffer hydrogels than divalent cations.67

With certain degrees of flexibility, hydrogels are similar to human tissues, with broad applications including drug delivery, extracellular matrix, tissue engineering, cartridge repair.68

Furthermore, by adopting responsive components, hydrogels can obsess tunable properties (e.g. gel-sol transition) with external stimuli, including temperature, salt concentration, electric/magnetic fields, pH, pressure, light.69 Heilshorn et al. reported hydrogels consisting of hyaluronic acid (HA), a major component of the ECM, and thermoresponsive linear ELPs. These materials showed thermal-induced stiffening due to secondary physical crosslinking above the Tt’s of the ELPs and were investigated in applications including cartilage regeneration and stem cell delivery.70,71

1.5 Peptide-polymer conjugates

Peptide-polymer conjugates have been widely used to combine the advantages of both components, as well as to overcome their disadvantages, with various synthetic strategies.72-74 Hybrid materials made by conjugating polymers, especially PEG, to peptides has shown improvement over the original peptides in terms of stability, solubility, circulation half-life, and immunogenicity.75

Besides their potential for therapeutic applications, it’s of fundamental significance to explore their self-assembly behavior, e.g. hierarchical/novel structures, and the corresponding driving force.74

Aspects of interest include the interaction between the two building blocks, the structure change upon conjugation, energetic contribution of phase behavior, as well as the micro-phase separation and amphiphilicity.74,76-79

1.6 H2S-releasing materials

H2S is best known as a rotten-egg smelling toxic gas. However, with research progress in understanding the signaling mechanisms, H2S has joined CO and NO as small signaling molecules

80 (gasotransmitters). As a gasotransmitter, H2S is produced endogenously by three mammalian enzymes: cystathionine β-synthase (CBS), cystathionine γ-lyase (CSE), and 3- mercaptosulfurtransferase (3MST). By regulating the physiological pathways, H2S has shown great potential for disease treatments with therapeutic effects including angiogenesis, antioxidant, anti-apoptotic, anti-inflammatory properties.81 Inflammation often accompanies injury, bacteria invasion, and even worse, recurs associated with chronic diseases such as cancer, Alzheimer’s,

82 arthritis, diabetes, stroke and heart problems. H2S treats inflammation by increasing the blood flow via blood vessel dilation,83 phosphodiesterase inhibition,84 and cytoprotective effects85

+ 86 through opening of KATP channels and partially K conductance. Besides, H2S plays a key role as a mediator of inflammation, upregulating the adherence and migration of leukocytes which

87 occur in the early stage inflammation. Thus, H2S has been used to treat chronic diseases with recurring inflammation, such as lung disease,88 ulcers89 and edemas.90

This dissertation aimed to investigate the topology effect on the phase behavior of ELPs, as well as hydrogels/ELP-polymer conjugates prepared from it. In Chapter 2, we were able to explore the topology effect on both of the secondary structure and macroscopic LCST. In chapter 3, we managed to synthesize large ELP dendrimers at different generations, and found an intermediate state below the LCST where there are small aggregates growing in size but the solution remains clear. In chapter 4, we further prepared hydrogels by crosslinking HA with ELP dendrimers, and found that the precoacervates played an important role in the thermoresponsiveness of the hydrogel

7 materials. In chapter 5 we employed the peptide dendrimers to fabricate a series of triblock PEG-

ELP-C16 conjugate amphiphiles, and obtained important insights into their phase behavior. In chapter 6, we prepared an HNE responsive H2S-releasing hydrogel, with potential for treating chronic diseases with recurring inflammation. Finally in chapter 7, I briefly summarized all of the work with future work and interest.

1.7 References

(1) Kim, Y.-J.; Matsunaga, Y. T. J. Mater. Chem. B 2017, 5, 4307.

(2) Gandhi, A.; Paul, A.; Sen, S. O.; Sen, K. K. Asian J. Pharm. 2015, 10, 99.

(3) Schmaljohann, D. Adv. Drug Deliver. Rev. 2006, 58, 1655.

(4) MacEwan, S. R.; Chilkoti, A. J. Control. Release 2014, 190, 314.

(5) Ryu, S.; Yoo, I.; Song, S.; Yoon, B.; Kim, J.-M. J. Am. Chem. Soc. 2009, 131, 3800.

(6) Nettles, D. L.; Chilkoti, A.; Setton, L. A. Adv. Drug Deliver. Rev. 2010, 62, 1479.

(7) Nagase, K.; Kobayashi, J.; Kikuchi, A.; Akiyama, Y.; Kanazawa, H.; Okano, T.

Biomaterials 2011, 32, 619.

(8) Schmaljohann, D. Adv. Drug. Deliver. Rev. 2006, 58, 1655.

(9) Wise, S. G.; Yeo, G. C.; Hiob, M. A.; Rnjak-Kovacina, J.; Kaplan, D. L.; Ng, M. K. C.;

Weiss, A. S. Acta. Biomater. 2014, 10, 1532.

(10) Wise, S. G.; Weiss, A. S. Int. J. Biochem. Cell Biol. 2009, 41, 494.

(11) Urry, D. W.; Pattanaik, A. Ann. N. Y. Acad. Sci. 1997, 831, 32.

(12) Urry, D. W.; Trapane, T.; Prasad, K. Biopolymers 1985, 24, 2345.

(13) Roberts, S.; Dzuricky, M.; Chilkoti, A. FEBS letters 2015, 589, 2477.

(14) Yeo, G. C.; Keeley, F. W.; Weiss, A. S. Adv. Colloid. Interfac. 2011, 167, 94.

(15) Nuhn, H.; Klok, H.-A. Biomacromolecules 2008, 9, 2755.

(16) Reguera, J.; Urry, D. W.; Parker, T. M.; McPherson, D. T.; Rodríguez-Cabello, J. C.

Biomacromolecules 2007, 8, 354.

(17) Li, N. K.; Quiroz, F. G.; Hall, C. K.; Chilkoti, A.; Yingling, Y. G. Biomacromolecules

2014, 15, 3522.

(18) Zhang, Y.; Trabbic-Carlson, K.; Albertorio, F.; Chilkoti, A.; Cremer, P. S.

Biomacromolecules 2006, 7, 2192.

(19) Meyer, D. E.; Chilkoti, A. Biomacromolecules 2004, 5, 846.

(20) Girotti, A.; Reguera, J.; Arias, F. J.; Alonso, M.; Testera, A. M.; Rodríguez-Cabello, J. C.

Macromolecules 2004, 37, 3396.

(21) McDaniel, J. R.; Radford, D. C.; Chilkoti, A. Biomacromolecules 2013, 14, 2866.

(22) Lee, T.; Cooper, A.; Apkarian, R.; Conticello, V. Adv. Mater. 2000, 12, 1105.

(23) Kim, W.; Thévenot, J.; Ibarboure, E.; Lecommandoux, S.; Chaikof, E. L. Angew. Chem.

Int. Edit. 2010, 49, 4257.

(24) Ghoorchian, A.; Vandemark, K.; Freeman, K.; Kambow, S.; Holland, N. B.; Streletzky,

K. A. J. Phys. Chem. B 2013, 117, 8865.

(25) Aluri, S.; Pastuszka, M. K.; Moses, A. S.; MacKay, J. A. Biomacromolecules 2012, 13,

2645.

(26) Santoso, S.; Hwang, W.; Hartman, H.; Zhang, S. Nano Lett. 2002, 2, 687.

(27) Martín, L.; Castro, E.; Ribeiro, A.; Alonso, M.; Rodríguez-Cabello, J. C.

Biomacromolecules 2012, 13, 293.

(28) Tu, Y.; Wise, S. G.; Weiss, A. S. Micron 2010, 41, 268.

(29) Koga, T.; Iimura, M.; Higashi, N. Macromol. Biosci. 2012, 12, 1043.

(30) Kojima, C.; Irie, K.; Tada, T.; Tanaka, N. Biopolymers 2014, 101, 603.

(31) Fukushima, D.; Sk, U. H.; Sakamoto, Y.; Nakase, I.; Kojima, C. Colloid. Surface B 2015,

132, 155.

(32) Peppas, N. A.; Keys, K. B.; Torres-Lugo, M.; Lowman, A. M. J. Control. Release 1999,

62, 81.

(33) Bracci, L.; Falciani, C.; Lelli, B.; Lozzi, L.; Runci, Y.; Pini, A.; De Montis, M. G.;

Tagliamonte, A.; Neri, P. J. Biol. Chem. 2003, 278, 46590.

(34) Bosman, A. W.; Janssen, H. M.; Meijer, E. W. Chem. Rev. 1999, 99, 1665.

(35) Caminade, A.-M.; Turrin, C.-O.; Laurent, R. Dendrimers : Towards Catalytic, Material and Biomedical Uses; John Wiley & Sons, Ltd, 2011.

(36) Peng, M.; Chen, Y.; Tan, R.; Zheng, W.; Yin, D. RSC Adv. 2013, 3, 20684.

(37) Astruc, D.; Chardac, F. Chem. Rev. 2001, 101, 2991.

(38) Martinho, N.; Florindo, H.; Silva, L.; Brocchini, S.; Zloh, M.; Barata, T. Molecules 2014,

19, 20424.

(39) Caminade, A.-M.; Turrin, C.-O. J. Mater. Chem. B 2014, 2, 4055.

(40) Woo-Dong, J.; Nishiyama, N.; Kataoka, K. Supramol. Chem. 2007, 19, 309.

(41) Maleki, B.; Sheikh, S. RSC Adv. 2015, 5, 42997.

(42) Yang, Y.; Wang, Y.; Li, H.; Li, W.; Wu, L. Chem.-Eur. J. 2010, 16, 8062.

(43) Ma, J.; Ma, X.; Deng, S.; Li, F.; Hu, A. J. Polym. Sci. A Polym. Chem. 2011, 49, 1368.

(44) Maiti, P. K.; Çaǧın, T.; Wang, G.; Goddard, W. A. Macromolecules 2004, 37, 6236.

(45) Carlmark, A.; Hawker, C.; Hult, A.; Malkoch, M. Chem. Soc. Rev. 2009, 38, 352.

(46) Wang, H.; Huang, Q.; Chang, H.; Xiao, J.; Cheng, Y. Biomater. Sci. 2016, 4, 375.

(47) Hu, W.; Qiu, L.; Cheng, L.; Hu, Q.; Liu, Y.; Hu, Z.; Chen, D.; Cheng, L. Acta Biomater.

2016, 36, 241.

(48) Chen, L.; Cao, W.; Grishkewich, N.; Berry, R. M.; Tam, K. C. J. Colloid Interf. Sci. 2015,

450, 101.

(49) Stuart, M. A. C.; Huck, W. T. S.; Genzer, J.; Muller, M.; Ober, C.; Stamm, M.;

Sukhorukov, G. B.; Szleifer, I.; Tsukruk, V. V.; Urban, M.; Winnik, F.; Zauscher, S.; Luzinov, I.;

Minko, S. Nat. Mater. 2010, 9, 101.

(50) Yan, X. Z.; Wang, F.; Zheng, B.; Huang, F. H. Chem. Soc. Rev. 2012, 41, 6042.

(51) Laga, R.; Janoušková, O.; Ulbrich, K.; Pola, R.; Blažková, J.; Filippov, S. K.; Etrych, T.;

Pechar, M. Biomacromolecules 2015, 16, 2493.

(52) Sedláček, O.; Černoch, P.; Kučka, J.; Konefal, R.; Štěpánek, P.; Vetrík, M.; Lodge, T. P.;

Hrubý, M. Langmuir 2016, 32, 6115.

(53) Twaites, B. R.; de las Heras Alarcón, C.; Lavigne, M.; Saulnier, A.; Pennadam, S. S.;

Cunliffe, D.; Górecki, D. C.; Alexander, C. J. Control. Release 2005, 108, 472.

(54) Ashraf, S.; Park, H. K.; Park, H.; Lee, S. H. Macromol. Res. 2016, 24, 297.

(55) Cortez-Lemus, N. A.; Licea-Claverie, A. Prog. Polym. Sci. 2016, 53, 1.

(56) Nagase, K.; Kobayashi, J.; Kikuchi, A.; Akiyama, Y.; Kanazawa, H.; Okano, T.

Biomacromolecules 2014, 15, 1204.

(57) Nagase, K.; Kobayashi, J.; Kikuchi, A.; Akiyama, Y.; Kanazawa, H.; Okano, T. RSC Adv.

2015, 5, 66155.

(58) Roy, D.; Brooks, W. L. A.; Sumerlin, B. S. Chem. Soc. Rev. 2013, 42, 7214.

(59) Ward, M. A.; Georgiou, T. K. Polymers 2011, 3, 1215.

(60) Liu, Y.; Xu, L.; Liu, J. S.; Liu, X. Y. Anal. Methods 2015, 7, 10151.

(61) Liu, L. X.; Li, W.; Yan, J. T.; Zhang, A. F. J. Polym. Sci. Pol. Chem. 2014, 52, 1706.

(62) Li, W.; Zhang, A.; Feldman, K.; Walde, P.; Schlüter, A. D. Macromolecules 2008, 41,

3659.

(63) Bekhradnia, S.; Zhu, K. Z.; Knudsen, K. D.; Sande, S. A.; Nystrom, B. J. Mater. Sci.

2014, 49, 6102.

(64) Zheng, Q.; Pan, C.-y. Eur. Polym. J. 2006, 42, 807.

(65) Kojima, C.; Irie, K.; Tada, T.; Tanaka, N. Biopolymers 2014, 101, 603.

(66) Prashantha, M. A. B.; Premachandra, J. K.; Amarasinghe, A. D. U. S Ind. Eng. Chem.

Res.2008, 47, 8555.

(67) Martin, N.; Youssef, G. J. Mech. Behav. Biomed. 2018, 85, 194.

(68) Caló, E.; Khutoryanskiy, V. V. Eur. Polym. J. 2015, 65, 252.

(69) Koetting, M. C.; Peters, J. T.; Steichen, S. D.; Peppas, N. A. Mater. Sci. Eng. R Rep. 2015,

93, 1.

(70) Wang, H.; Cai, L.; Paul, A.; Enejder, A.; Heilshorn, S. C. Biomacromolecules 2014, 15,

3421.

(71) Zhu, D.; Wang, H.; Trinh, P.; Heilshorn, S. C.; Yang, F. Biomaterials 2017, 127, 132.

(72) Gauthier, M. A.; Klok, H.-A. Chem. Comm. 2008, 0, 2591.

(73) Wilson, P. Macromolecular Chem. Phys. 2017, 218, 1600595.

(74) Shu, J. Y.; Panganiban, B.; Xu, T. Annu. Rev. Phys. Chem. 2013, 64, 631.

(75) Pasut, G.; Veronese, F. M. Prog. Polym. Sci. 2007, 32, 933.

(76) Shu, J. Y.; Huang, Y.-J.; Tan, C.; Presley, A. D.; Chang, J.; Xu, T. Biomacromolecules

2010, 11, 1443.

(77) Vandermeulen, G. W. M.; Tziatzios, C.; Duncan, R.; Klok, H.-A. Macromolecules 2005,

38, 761.

(78) Vandermeulen, G. W. M.; Tziatzios, C.; Klok, H.-A. Macromolecules 2003, 36, 4107.

(79) Kühnle, R. I.; Gebauer, D.; Börner, H. G. Soft Matter 2011, 7, 9616.

(80) Powell, C. R.; Dillon, K. M.; Matson, J. B. Biochem. Pharmacol. 2018, 149, 110.

(81) Kanagy, N. L.; Szabo, C.; Papapetropoulos, A. Am. J. of Physiol-Cell Ph. 2017, 312,

C537.

(82) Aimetti, A. A.; Tibbitt, M. W.; Anseth, K. S. Biomacromolecules 2009, 10, 1484.

(83) Zhao, W.; Zhang, J.; Lu, Y.; Wang, R. The EMBO journal 2001, 20, 6008.

(84) Bucci, M.; Papapetropoulos, A.; Vellecco, V.; Zhou, Z.; Pyriochou, A.; Roussos, C.;

Roviezzo, F.; Brancaleone, V.; Cirino, G. Arter. Thromb. Vasc. Biol. 2010, 30, 1998.

(85) Qian, Y.; Matson, J. B. Adv. Drug. Deliver. Rev. 2017, 110-111, 137.

(86) Zhao, W.; Zhang, J.; Lu, Y.; Wang, R. EMBO J. 2001, 20, 6008.

(87) Bhatia, M. Scientifica 2012, Article ID 159680, 12 pages.

(88) Wang, Z.; Yin, X.; Gao, L.; Feng, S.; Song, K.; Li, L.; Lu, Y.; Shen, H. SpringerPlus

2016, 5, 1084.

(89) Wallace, J. L.; Caliendo, G.; Santagada, V.; Cirino, G.; Fiorucci, S. Gastroenterology

2007, 132, 261.

(90) Zhang, H. X.; Liu, S. J.; Tang, X. L.; Duan, G. L.; Ni, X.; Zhu, X. Y.; Liu, Y. J.; Wang,

C. N. Cell. Physiol. Biochem. 2016, 40, 1603.

Chapter 2: Dendritic Elastin-Like Peptides: The Effect of Branching on

Thermoresponsiveness

“Reprinted (adapted) with permission from Novon, Y.; Zhou, M; Matson, J. B.; Ronit Bitton, R.

Authors

Yotam Navon, [a] Mingjun Zhou, [b] John B. Matson,* [b] Ronit Bitton* [a]

[a] Department of Chemical Engineering, Ben-Gurion University of the Negev, Beer-Sheva, 84105

(Israel)

[b] Department of Chemistry and Macromolecules Innovation Institute, Virginia Tech, Blacksburg,

VA 24061 (U.S.A.)

2.1 Abstract

Elastin-like peptides (ELPs) have been used widely to confer thermoresponsive characteristics onto various materials, but to this point, mostly linear ELPs have been studied. A class of linear and dendritic (branched) ELPs based on the GLPGL pentamer is reported here. The effect of peptide topology on the transition temperature (Tt) was examined, using circular dichroism to study the peptide secondary structure transition and turbidity to measure the macroscopic phase transition (coacervation). Secondary structure transitions showed no dependence on topology, but a higher Tt was observed for dendritic peptides than for linear peptides with the same number of

GLPGL repeats. The data support a phase transition model that consists of two neighboring

14 processes: a secondary structure transition, related to intramolecular interactions, followed by coacervation, associated with intermolecular interactions.

2.2 Introduction

Thermoresponsive biomaterials are a unique class of materials that are capable of changing their structures and properties in response to a change in temperature. Over the past decade, interest in these materials has increased due to their various potential applications, including drug delivery,2,3 sensing,4 regenerative medicine,5 protein separation6 and others. A class of widely explored thermoresponsive biomolecules is the elastin-like peptides (ELPs). ELPs are artificial biopolymers derived from the hydrophobic domain of tropoelastin,7-9 which is composed of Xaa–Pro–Gly–

Xaa–Gly (XPGXG) pentapeptide repeats, where X, the guest residue, can be any amino acid other than proline.10 This repeating pentapeptide structure imparts a type of phase transition behavior called a lower critical solution temperature (LCST). Peptides bearing the XPGXG repeating sequence are typically soluble below a characteristic cloud point temperature, also known as the inverse transition temperature (Tt), and aggregate into large clusters above the Tt. This aggregation is a result of a change in ELP secondary structure—ELPs exist primarily in a random coil conformation below the Tt and as type II β-turns (also known as β-spirals) above the Tt. The transformation is reversible and can take place in a temperature window as narrow as 3 C.7-9,11

Many intrinsic and extrinsic factors influence the Tt of a given ELP, including the specific ELP sequence, the molecular weight (MW) of the polypeptide (number of pentapeptide repeats), as well as ELP concentration, solution pH, and ionic strength.12-14 For example, increasing ELP

10,15,16 15-17 concentration in solution typically leads to lower Tts, as does increasing ELP MWs.

Similarly, the characteristics of the guest residue (X in XPGXG) also have a marked effect on

Tt: hydrophobic guest residues (Trp and Tyr being the most effective) decrease the Tt, while hydrophilic guest residues elevate it.9,16

Over the past decade ELPs have been designed to create complex nanostructures and materials such as spherical19,20 and cylindrical micelles21, fibers22, vesicles,23,24 and gels.25 In all cases, however, the basic ELP component is a linear peptide. Recently several groups have begun studying ELPs in combination with branched polymer architectures. For example, Higashi et al. synthesized a peptide/dendrimer hybrid that contained thermoresponsive oligo(ELP)s at the periphery of poly(amidoamine) (PAMAM) dendrimers.26 In a subsequent report, Kojima et al.27 studied the effect of the generation number of the PAMAM dendrimers on Tt. These dendrimers were then utilized to create dual stimuli-responsive nanoparticles by loading photo thermogenic

AuNPs into their core.28 Ghoorchian et al. reported the synthesis of an amphiphilic ELP trimer consisting of three (GVGVP)n arms as hydrophobic segments, connected with a compact charged domain as the hydrophilic head group. Temperature changes drove micelle formation above the Tt of the ELP.21

To the best of our knowledge, branched ELPs that do not contain other polymeric or peptidic species have not been investigated. In fact, aside from the reports above, little is known about the effect of the molecular architecture on the thermoresponsive behavior of ELPs. Dendritic, hyperbranched, star, and other highly branched structures are important in biomaterials because they are often used as crosslinkers to make polymeric hydrogels.29 Moreover, globular and compact topologies such as dendrimers have very different physical properties (solubility, viscosity.) from their linear counterparts. This topological difference may lead to useful properties in biomaterials, including substantially improved resistance to proteolysis of branched peptides compared to linear ones.30 Fundamental studies evaluating the effect of branching on ELP

16 transitions may enable the design of new thermoresponsive materials with transitions in peptide secondary structure and solubility. Here we present a series of linear and branched ELPs based on the repeating sequence GLPGL. The effects of molecular weight, salt concentration, and dendrimer generation number on Tts and secondary structure were measured to explore the structure-property relationships of branched ELPs.

2.3 Results and discussions

Synthesis of Branched/Dendritic ELPs

Because many ELPs are made with large numbers of repeating pentamers, and high molecular weight leads to low Tts, most ELP-based materials that employ the VPGVG sequence have Tts in the physiological range. Here we targeted dendrimers based on short peptides (5-30mers), which we expected would have high Tts (if any at all) if we used the VPGVG sequence. To ensure that

Tts were in a measurable range, we chose the GLPGL sequence, which has a substantially lower

18 Tt than VPGVG. Synthesis of the branched ELPs was carried out using Fmoc-based solid-phase peptide synthesis (SPPS) on a microwave-assisted peptide synthesizer.

Fmoc-Lys(Mtt)-OH was used as the branching unit. Scheme 2.1 shows the synthetic strategy for the preparation of the pentamer GLPGL dendrons. Larger dendrons were prepared similarly, adding (GLPGL)n (n = 2-6) as desired. Both capped (acetylated N-terminus) and uncapped (free

N-terminus) peptides were prepared. First, the sequence H-K(Mtt)GLPGL was added to the resin using Fmoc chemistry and HBTU as the coupling agent. The resin was then treated with dilute

TFA in CH2Cl2 to selectively deprotect the Mtt group. Acetylation using acetic anhydride followed by TFA cleavage of the peptide from the resin afforded crude capped G1-5. Uncapped G1-5 was prepared in the same way by eliminating the acetylation step. Crude peptides were purified by

Scheme 2.1. Synthesis of capped, dendritic ELPs by SPPS.a

aDendrons are labeled GX-Y as follows: GX represents the dendrimer generation number (G1, G2, or G3) and Y represents the number of residues in the repeating unit (5–30 in multiples of 5).

Uncapped peptides (see below) do not contain acetyl groups on free amines and are denoted GX-

YU.

18 preparative scale HPLC. Second generation dendrimers were prepared by adding in a second

KGLPGL sequence after the Mtt deprotection step. In this case two amino acids were added in each coupling step—one to each branch. The 3rd generation dendrimer was prepared similarly, adding four amino acids in each coupling step. All 2nd and 3rd generation dendrimers were capped, cleaved and purified in the same manner as the 1st generation dendrimers.

Determination of Transition Temperatures

First, we measured the Tts of these branched ELPs in PBS solution (pH 7.4, 5 mM) and at higher salt concentrations by adding NaCl. Figure 2.1 plots the measured Tt vs. added NaCl concentration

(0–3.5 M added salt). Tt decreased linearly with increasing NaCl concentration for all peptides, as is typical of ELPs. In order to quantify the effect of salt concentration and peptide MW on Tt, δTt values (i.e., the slope of the line in a plot of Tt vs. added NaCl concentration) were calculated

(Table 2.1). A larger value of δTt indicates higher sensitivity towards NaCl addition. The apparent

0 Tt (Tt ) is the transition temperature in PBS without added NaCl as extrapolated from the linear fit

0 of the data points. The existence of large aggregates at temperatures higher than Tt was also verified by DLS (Figure S2.7). For some peptides Tts were not observed. This was either due to a lack of solubility in aqueous solution (typical of high MW peptides), or a condition in which the peptide remained in solution at all temperatures (typical of low MW peptides).

To assess the hydrophobicity of each peptide, we calculated the distribution coefficient (logD) for each peptide. LogD is similar to the more widely used partition coefficient (logP), which is defined as the log of the ratio of compound concentrations in 1-octanol versus water. LogP values below

0 indicate high solubility in water, while values greater than 0 are characteristic of more hydrophobic compounds. In contrast to logP, logD considers the concentration contribution from ionized species.85 This distinction is irrelevant for the capped peptides presented in Table 2.1, but

19 becomes important for the uncapped peptides that contain free amines (discussed below).

Calculations were based on a combination of the residue addition model and fragment addition model, which were generated empirically from known logD values for a wide variety of short peptides.31 Our results show that peptides with logD values in the approximate range of 1–6 have

0 measurable Tt s. Those with logD values below ~1 are too soluble to exhibit a Tt at any NaCl concentration studied, while those with logD values above ~6 are insoluble in water. Note that the methods used here are derived from and typically used for small peptides, so calculated logD values for larger peptides may have substantial error.

Figure 2.1. Tt dependence on salt concentration for linear and branched capped peptides (5 mg mL-1 in 5 mM PBS).

For the linear peptides with measurable Tt values (G1-10, G1-15, G1-20, G1-25, and G1-30),

0 increasing the number of amino acid residues led to a decrease in Tt . This trend is typical of

ELPs12,15 and is supported by the logD calculations—each additional GLPGL pentamer increases

0 the hydrophobicity of the peptide, resulting in a lower Tt . Second generation dendrimers followed

0 a similar trend: G2-5 and G2-10 show Tt values of 93 °C and 30 °C, respectively. Unfortunately,

0 capped G2-15, G2-20, and G3-5 did not dissolve in PBS at any temperature, so no Tt values were obtained for these peptides.

Table 2.1. Summary of data for capped peptides.

0 Peptide Tt [°C] –δTt [°C/MNaCl] Npp LogD MW[Da]

G1-5 N/A[a] 1 -0.1±0.4 624.4 G1-10 94±4.4 21±1.7 2 1.2±0.4 1061.6 G1-15 69±1.3 21±0.7 3 2.6±0.4 1498.9 G1-20 51±1.3 22±1.3 4 3.9±0.5 1936.2 G1-25 38±1.0 16±1.1 5 5.2±0.5 2373.5 G1-30 23±0.4 13±0.6 6 6.5±0.5 2810.7

G2-5 93±5.9 30±3.2 3 2.2±0.6 1755.1 G2-10 30±0.5 16±0.5 6 6.1±0.7 3066.9 G2-15 N/A[b] 9 10.0±0.7 4378.7

G2-20 N/A[b] 12 13.9±0.8 5690.5

G3-5 N/A[b] 7 6.6±0.9 4016.5

G1-15LC 81±4.1 25±2.2 3 2.2±0.6 1755.1

G1-30LC 28±0.6 14±0.7 6 6.1±0.7 3066.9

0 -1 Abbreviations: Tt : transition temperature at 5 mg mL in PBS with no added NaCl (obtained from linear extrapolation at higher NaCl concentrations); –δTt: the change in transition temperature per mole of added NaCl; Npp: number of pentapeptide GLPGL repeats; MW: molecular weight. [a]

-1 0 Tt was not observed. [b] The peptide is not soluble at 5 mg mL . Confidence intervals for Tt and

–δTt were calculated according to least square algorithm by IGOR software. Confidence intervals for logD values were calculated according to literature values for individual amino acid residues.31

a d

b e

c f

Figure 2.2 Left column: Temperature/wavelength CD spectra of selected capped peptides: (A)

G1-15, (B) G2-5 and (C) G3-5. Right Column: Temperature dependence of the mean molar residue ellipticity at 218 nm. G1 series (D, E) G2 and G3 series (F). All samples were measured at 0.5 mg mL-1 in 2.5 mM PBS. Black lines in A, B, and C represent 218 nm wavelength.

Conformational Transitions

In order to ensure the observed Tt was accompanied by the random coil to β-turn transition typical for ELPs, CD measurements of all soluble peptides at various temperatures were performed (Fig

2.2; Fig S2.4-S2.6).

All the obtained CD profiles showed typical ELP spectra with the existence of a broad minimum in the region of 225-230 nm and a sharper maximum near 218 nm. Both peaks decreased in intensity with increasing temperature, and an isodichroic point at 205-210 nm was observed for all peptides. Additionally, increasing temperature also shifted the peak at 218 nm to lower wavelengths. These characteristics are typical of ELPs and have been observed by several groups for other short ELPs with slight variation in the pentameric unit sequence and solution conditions.32

The existence of an isodichroic point indicates a transition between two distinct local states,33,34 in this case from a random coil to a type II β-turn.

The Effect of Branching on Tt

In order to examine the effect of branching on the phase transition behavior of the peptides, a comparison between peptides with the same number of GLPLG repeats and different topologies was performed. Figure 2.3 shows a comparison of Tt values for linear and branched peptides with

3 pentamers (G1-15 and G2-5) and 6 pentamers (G1-30 to G2-10) at various salt concentrations.

In both cases the Tt values of the branched ELPs were higher than those of their linear analogues at all measured salt concentrations.

A possible explanation of the higher Tt observed for the branched peptides could be the presence of additional lysine residues. In order to measure the contribution of additional lysine residues to the Tt, two additional linear control (LC) peptides were synthesized: Ac-[K(Ac)GLPGL]3-NH2

(G1-15LC) and Ac-[K(Ac)GLPGLGLPGL]3-NH2 (G1-30LC), both with capped N-termini and

Figure 2.3 Tt values of peptides with 3 (a) or 6 (b) pentamers at various NaCl concentrations. In both series the dendritic (G2) peptides have higher Tts than the linear (G1) peptides. capped lysine ε-amines. The calculated logD values for the LC peptides are identical to those of

0 the corresponding dendritic peptides. For both LC peptides, the measured Tt and –δTt values fell between those of their linear and dendritic analogs (Table 2.1). For example, for the peptides

0 containing 3 pentamer repeats, G1-15, G2-5 and G1-15LC, the Tt values were 69, 93, and 81°C,

0 respectively. For the peptides containing 6 pentamer repeats, G1-30, G2-10 and G1-30LC, Tt values were 23, 30, and 28°C. Taken together, these data suggest that branching is an important

0 factor in accounting for the difference in Tt between linear and dendritic ELPs.

Based on the Tt data, we expected that the secondary structure change would also occur at a higher temperature for the dendritic ELPs compared with their linear counterparts. Interestingly, the CD spectra revealed no significant change in the secondary structure transition between the two topologies, as the molar ellipticity value at 218 nm decreased in the same manner for both branched and linear peptides (Figure 2.4).

Figure 2.4 Temperature-dependence of the mean molar residue ellipticity at 218 nm of linear and branched peptides with 3 and 6 pentameric units.

Table 2.2 shows a summary of data for the uncapped peptides. Similar to the capped peptides,

0 increasing the number of pentapeptide repeats decreased the Tt , suggesting that uncapped free amines did not affect the general transition behavior of the peptides. Interestingly, the δ(Tt) values for the uncapped peptides were higher than those of their capped analogues (Figure 2.5), implying that uncapped peptides were more sensitive to the addition of salt compared to capped ones. This is likely due to the fact that all of the uncapped peptides have multiple free amines that are charged at neutral pH.

Unfortunately, the peptide design makes it impossible to make a direct comparison of Tt values between uncapped peptides of the same number of pentamer repeats but different generations, as was done for the capped peptides. This is because the number of free amines is not the same between uncapped peptides of different generations (e.g., G1-15U has two free amines, while G2-

5U has four), and protonated amines are hydrophilic entities that drastically influence the Tt values of these small peptides.

Table 2.2 Summary of data for uncapped peptides.

0 Peptide Tt [°C] –δTt [°C/MNaCl] Npp LogD MW[Da]

G1-5U N/A[a] 1 -4.2±0.3 582.4 G1-10U N/A[a] 2 -2.9±0.4 1019.6 G1-15U 153±12 38±4.2 3 -1.6±0.4 1456.9 G1-20U 101±2.4 38±1.2 4 -0.3±0.4 1894.2 G1-25U 71±1.9 36±2.1 5 1.0±0.5 2373.4 G1-30U 50±1.8 27±1.9 6 2.3±0.5 2852.7

G2-5U N/A[a] 3 -6.1±0.5 1713.1 G2-10U N/A[a] 6 -2.2±0.6 3024.9 G2-15U 42±3.0 26±3.2 9 1.7±0.7 4336.7 G2-20U 36±0.7 26±1.1 12 5.7±0.7 5648.5

G3-5U 150±4.4 43±1.7 7 -10.0±0.7 3974.5

0 molecular weight. [a] Tt was not observed. Confidence intervals for Tt and –δTt were calculated according to least square algorithm by IGOR software. Confidence intervals for logD values were calculated according to literature values for individual amino acid residues.31

a b

Figure 2.5 Dependence of Tt on salt concentration for linear (a) and branched (b) uncapped peptides (5 mg mL-1 in 5 mM PBS).

Figure 2.6 shows the CD profiles of selected uncapped peptides. The secondary structure transition was observed for the uncapped peptides as a decrease in the signal at 218 nm with increasing temperature. Overall, the same trends were observed for the uncapped peptides as for the capped peptides.

a b

c a b

Figure 2.6 Temperature/wavelength CD spectra of uncapped peptides: (A) G1-30U, (B) G2-15U and (C) G3-5U. Measurements were taken at 0.5–1 mg mL-1 in PBS (2.5 mM).

Thermodynamic Characterization of Conformational Changes

In order to evaluate the thermodynamic parameters of the system, a Van’t Hoff plot of lnK versus

1/T was generated from the CD temperature scans at 218 nm35 for a series of uncapped peptides

(Figure 2.7). Uncapped peptides were chosen for this analysis because their macroscopic Tts are higher than the capped peptides; therefore, the formation of large coacervates, which disrupts the

CD signal at this wavelength range, is less likely to occur during the process. The free energy of the transition from random coil (unfolded) to β-turn (folded) was fitted to a two-state model,

K=[U]/[F], where K is the equilibrium constant for the unfolded (U) and folded (F) states. The

Van’t Hoff plot was constructed based on the Gibbs free energy equation using the following relations:

∆퐺 = ∆퐻 − T∆푆 푈 → 퐹 푈→ 퐹 푈 → 퐹 (Equation.2.1)

∆퐺푈→퐹 = −푅푇푙푛퐾푒푞 (Equation.2.2)

[푈 ] (Equation.2.3) 퐾 = = ([휃]표푏푠-[휃]푈)/([휃]퐹 − [휃]표푏푠) 푒푞 [퐹]

∆ 퐻 1 ∆ 푆 (Equation.2.4) 푙푛퐾 = ∙ − 푅 푇 푅 obs Here ΔGU→F is the free energy of the transition into the folded form at higher temperatures, [θ] is the experimentally observed ellipticity measurement at 218 nm, and [θ]F/[θ]U represents the values for the fitted minimum and maximum points of the transition (high-temperature folded form, [θ]F, and low-temperature unfolded form, [θ]U). The reversibility of the transition process was examined by carrying out two consecutive reverse temperature scans from 20–60 °C. The transition was found to be fully reversible for all peptides examined.

Figure 2.7 Van’t Hoff plot as constructed from the CD data at 218 nm, which were fitted to the two-state model introduced in the text for linear (a) and branched (b) uncapped peptides. The solid lines represent a linear fit to the data points.

Fitting of the CD data to the Van't Hoff equation allowed calculation of ΔH and ΔS values for the transition over the examined temperature range. The thermodynamic parameters obtained from the

Van’t Hoff plots are presented in Table 2.3. All peptides studied exhibited positive ΔH and ΔS

29 values, supporting the model of an entropy-driven folding transition. The values of enthalpy and entropy for the transition did not vary significantly with the change in peptide length or branching, suggesting that the secondary structure transition is a property of the single pentameric unit.

Similar values were found by Rees et. al.35 for short ELPs GVG(VPGVG)n where n varied from

3 to 5.

Table 2.3 Thermodynamic parameters obtained from the Van’t Hoff plot.

-1 -1 -1 Peptide ΔH U→F [kcal mol ] ΔS U→F [cal mol K ]

G1-5U 25.9±2.0 21.0±0.7 G1-10U 23.7±2.5 19.3±0.8 G1-15U 26.4±3.5 21.5±1.1 G1-20U 25.4±2.8 20.6±0.9 G2-5U 25.6±2.9 20.9±0.9 G2-15U 28.9±5.0 23.4±1.7 G3-5U 24.7±3.7 20.1±1.2

The phase transition behavior of ELPs is often described as a combination of two neighboring processes: 1) a secondary structure transition that takes place within each peptide molecule, which is referred to as a microscopic change associated with intramolecular interactions of the peptide; and 2) a coacervation event (i.e., macroscopic change), which occurs between peptide molecules and involves intermolecular interactions. The results of the data presented here are summarized as a schematic illustration in Scheme 2.2. The data suggest that peptide topology (i.e., branching) has little to no effect on the secondary structure transition process. Coacervation, however, occurs at higher temperatures for branched peptides compared with similar linear peptides. Taken together, these results suggest that a branched topology possesses steric hindrance not present in linear

0 ELPs, which results in an elevated Tt , thus affecting the macroscopic process of the phase transition.

Scheme 2.2 Phase transition behavior of linear and branched peptides, illustrating the initial secondary structure transition (an intramolecular process) followed by coacervation (an intermolecular process).

2.4 Conclusions

We have investigated the effect of molecular architecture on the thermoresponsiveness of ELPs.

A novel class of dendritic ELPs based on the GLPGL pentamer was synthesized, and transition temperatures were compared to linear analogues. Turbidity and CD measurements suggest that the peptide topology affects the temperature of the macroscopic coacervation process. In contrast, peptide topology has little effect on the microscopic ELP secondary structure transition. These findings can be useful for the design of complex ELP-based thermoresponsive materials.

2.5 Experimental section

Synthesis

Rink amide MBHA resin (100-200 mesh, 0.48 mmol g-1) and Fmoc-protected amino acids were purchased from ChemPep, Inc. All other solvents and reagents were purchased from commercial sources and used as received. All of the peptides were synthesized using a CEM Liberty 1 automated microwave peptide synthesizer. Purification by preparative-scale reverse phase high performance liquid chromatography (RP-HPLC) was carried out on an Agilent Technologies 1260

Infinity HPLC system, eluting with a gradient of 2% to 90% ACN in milliQ H2O on an Agilent

PLRP-S column (100Å particle size, 25 x 150 mm) and monitoring at 220 nm. 0.1% TFA was added to each mobile phase. Fractions were analysed by mass spectrometry, including an Advion

ExpressIon Compact Mass Spectrometer, and a Matrix-Assisted Laser Desorption Ionization– tandem Time Of Flight Mass Spectrometer (4800 MALDI TOF/TOF; AB Sciex). Product- containing fractions were combined, concentrated in vacuo to remove ACN, and lyophilized

(LabConco).

Turbidity measurements

Turbidity measurements were carried out using a quartz capillary. 200μL of sample was added to the quartz capillary, which was then placed in a heating block (Thermo scientific, U.S.A.). The

-1 temperature was increased at a rate of 1 °C min and the Tt was defined as the temperature at which the solution turned from clear to turbid.

UV-vis

Transition temperature determination was carried out using a UV-vis spectrophotometer

(Varioskan LUX multimode microplate reader, Thermo scientific, U.S.A.) with temperature control unit at a wavelength of 350 nm. Samples were heated from 25 °C to 45 °C at a heating rate

32 of 1 °C min-1 and the samples were left for 1 min at each temperature to equilibrate prior to measurement. The optical density was recorded at each temperature with intervals of 1 °C between measurements. The instrument reading was normalized with extrapolation to give a value of zero for a clear solution and value of 1 for a turbid solution. The Tt was defined as the temperature at the onset of turbidity.

Tt was measured by both methods; however, the difference in the Tt obtained by the two methods was insignificant (< 2 °C). Hence, the Tt values reported here refer to the values obtained from turbidity measurements.

Circular Dichroism

Measurements were carried out using a spectropolarimeter (J-715, Jasco, U.S.A.) equipped with a

-1 PTC-348WI temperature control unit under N2 flow (80 mL min ). Spectra were recorded from

260 to 190 nm at a scan speed of 50 nm min-1, a step size of 0.5 nm, a bandwidth of 1 nm, and a response time of 1 s. For each sample three scans were averaged. Far UV-vis quartz cells (Starna, type 21/Q/1) were used, with a path length of 0.1 cm. Unless indicated otherwise, all peptides were dissolved in phosphate buffer (2.5 mM; pH 7.4) and measured at a concentration of 0.5 or 1 mg mL-1. The heating rate was 1 °C min-1. The samples were held 5 min at each temperature to equilibrate. The data were analyzed using Jasco spectra analysis software version 1.53. Three measurements were averaged using moving average technique. The CD values were converted into mean molar residue ellipticity (MMRE; [θ], in mdeg ·cm2 dmole -1res -1).

2.6 References

(1) Schmaljohann, D. Adv. Drug Deliver. Rev. 2006, 58, 1655.

(2) MacEwan, S. R.; Chilkoti, A. J. Control. Release 2014, 190, 314.

(3) Ryu, S.; Yoo, I.; Song, S.; Yoon, B.; Kim, J.-M. J. Am. Chem. Soc. 2009, 131, 3800.

(4) Nettles, D. L.; Chilkoti, A.; Setton, L. A. Adv. Drug Deliver. Rev. 2010, 62, 1479.

(5) Nagase, K.; Kobayashi, J.; Kikuchi, A.; Akiyama, Y.; Kanazawa, H.; Okano, T.

Biomaterials 2011, 32, 619.

(6) Wise, S. G.; Yeo, G. C.; Hiob, M. A.; Rnjak-Kovacina, J.; Kaplan, D. L.; Ng, M. K. C.;

Weiss, A. S. Acta. Biomater. 2014, 10, 1532.

(7) Wise, S. G.; Weiss, A. S. Int. J. Biochem. Cell Biol. 2009, 41, 494.

(8) Urry, D. W.; Pattanaik, A. Ann. N. Y. Acad. Sci. 1997, 831, 32.

(9) Urry, D. W.; Trapane, T.; Prasad, K. Biopolymers 1985, 24, 2345.

(10) Yeo, G. C.; Keeley, F. W.; Weiss, A. S. Adv. Colloid. Interfac. 2011, 167, 94.

(11) Reguera, J.; Urry, D. W.; Parker, T. M.; McPherson, D. T.; Rodríguez-Cabello, J. C.

Biomacromolecules 2007, 8, 354.

(12) Li, N. K.; Quiroz, F. G.; Hall, C. K.; Chilkoti, A.; Yingling, Y. G. Biomacromolecules

2014, 15, 3522.

(13) Zhang, Y.; Trabbic-Carlson, K.; Albertorio, F.; Chilkoti, A.; Cremer, P. S.

Biomacromolecules 2006, 7, 2192.

(14) Meyer, D. E.; Chilkoti, A. Biomacromolecules 2004, 5, 846.

(15) Girotti, A.; Reguera, J.; Arias, F. J.; Alonso, M.; Testera, A. M.; Rodríguez-Cabello, J. C.

Macromolecules 2004, 37, 3396.

(16) McDaniel, J. R.; Radford, D. C.; Chilkoti, A. Biomacromolecules 2013, 14, 2866.

(17) Lee, T.; Cooper, A.; Apkarian, R.; Conticello, V. Adv. Mater. 2000, 12, 1105.

(18) Kim, W.; Thévenot, J.; Ibarboure, E.; Lecommandoux, S.; Chaikof, E. L. Angew. Chem.

Int. Edit. 2010, 49, 4257.

(19) Ghoorchian, A.; Vandemark, K.; Freeman, K.; Kambow, S.; Holland, N. B.; Streletzky,

K. A. J. Phys. Chem. B 2013, 117, 8865.

(20) Aluri, S.; Pastuszka, M. K.; Moses, A. S.; MacKay, J. A. Biomacromolecules 2012, 13,

2645.

(21) Santoso, S.; Hwang, W.; Hartman, H.; Zhang, S. Nano Lett. 2002, 2, 687.

(22) Martín, L.; Castro, E.; Ribeiro, A.; Alonso, M.; Rodríguez-Cabello, J. C.

Biomacromolecules 2012, 13, 293.

(23) Tu, Y.; Wise, S. G.; Weiss, A. S. Micron 2010, 41, 268.

(24) Koga, T.; Iimura, M.; Higashi, N. Macromol. Biosci. 2012, 12, 1043.

(25) Kojima, C.; Irie, K.; Tada, T.; Tanaka, N. Biopolymers 2014, 101, 603.

(26) Fukushima, D.; Sk, U. H.; Sakamoto, Y.; Nakase, I.; Kojima, C. Colloid. Surface B 2015,

132, 155.

(27) Peppas, N. A.; Keys, K. B.; Torres-Lugo, M.; Lowman, A. M. J. Control. Release 1999,

62, 81.

(28) Bracci, L.; Falciani, C.; Lelli, B.; Lozzi, L.; Runci, Y.; Pini, A.; De Montis, M. G.;

Tagliamonte, A.; Neri, P. J. Biol. Chem. 2003, 278, 46590.

(29) Nuhn, H.; Klok, H.-A. Biomacromolecules 2008, 9, 2755.

(30) Kah, M.; Brown, C. D. Chemosphere 2008, 72, 1401.

(31) Tao, P.; Wang, R.; Lai, L. J. Mol. Model. 1999, 5, 189.

(32) Yamaoka, T.; Tamura, T.; Seto, Y.; Tada, T.; Kunugi, S.; Tirrell, D. A.

Biomacromolecules 2003, 4, 1680.

(33) Kypr, J.; Vorlickova, M. Gen. Physiol. Biophys 1986, 5, 415.

(34) Holtzer, M. E.; Holtzer, A. Biopolymers 1992, 32, 1675.

(35) Reiersen, H.; Clarke, A. R.; Rees, A. R. J. Mol. Biol. 1998, 283, 255.

2.7 Appendix

Synthesis of Branched/Dendritic ELPs

MS spectra of capped (Figure S2.1) and uncapped (Figure S2.2) peptides.

Figure S2.1 The MS spectra of capped peptides: (a) G1-5 obtained by ESI-MS, and (b) G1-10, (c)

G1-15, (d)G1-20, (e)G1-25, (f) G1-30, (g) G2-5, (h) G2-10, (i) G2-15, (j) G2-20, (k) G3-5, (l) Ac-

[K(Ac)GLPGL]3-NH2, (m) Ac-[K(Ac)GLPGLGLPGL]3-NH2 obtained by MALDI-TOF MS.

Figure S2.2 MS spectra of uncapped peptides: (a) G1-5U, (b) G1-10U obtained with ESI-MS, and

(k) G3-5U obtained with MALDI-TOF MS.

LogD calculations

Because of their high molecular weights, typical software packages were not capable of calculating logP or logD values for several peptides. Thus, a traditional residue addition model was used for most residues, and a fragment addition model was incorporated to calculate contributions from

44 branching lysine residues.1 Each lysine has a reported logD contribution of -2.27. Notably, this value refers to an internal lysine residue with a free ε-amine.

Figure S2.3. G2-5 is used as example to clarify the logD calculation. Different components are shown in different colors.

Capped G2-5 is used here as an example (Figure S2.3). The residues in red are three (GLPGL) repeating units, each of which has a logD contribution of (0.80(2) – 0.22(2) + 0.15 = 1.31). The

C- and N-termini in black contribute –1.18 to the logD value as has been reported. Each of the lysine residues in blue has capped ε amines. The logD contribution for each of these residues was calculated as follows from the sum of logD contributions for a regular (uncapped) amine plus the logD contribution from uncapped amines to capped amines, which is also the difference between the blocked and unblocked peptides.

– 2.27 (general lysine residue)

+ 2.07 (comes from –1.18 + 3.25, which is conversion from uncapped to capped amine)

= –0.20

The remaining fragments (in green), combine to make another capped lysine with the logD contribution of (–0.20). The logD of uncapped peptides was analyzed and calculated in a similar way. The logD of peptides was calculated according to: logD(capped) = 1.31(# of GLPGL) – 1.18 (blocked peptide) – 0.20*(# of capped lysine residues) logD(uncapped)= 1.31(# of GLPGL) – 3.25 (unblocked peptide) – 2.27*(# of free lysine residues)

ChemSketch was also used to predict logP values for some capped peptides, but G2-15, G2-20, and G3-5 were above the 255-atom limit in this software. Table S2.1 shows a comparison between the logD calculated from the equation above and the logP obtained from ChemSketch. Considering all of the amines were capped, and all C-termini were uncharged amides, logD and logP of the capped peptides are assumed to be very close. Therefore, we believe that logP predictions in

Chemsketch should be similar to our logD predictions. As listed in Table S2.1, the logD/P values obtained from two different methods are reasonably close.

Table S2.1. Summary of the calculated logD and and logP obtained with ChemSketch

G1-5 G1-10 G1-15 G1-20 G1-25 G1-30 G2-5 G2-10 logD - 1.2±0.4 2.6±0.4 3.9±0.5 5.2±0.5 6.5±0.5 2.2±0.6 6.1±0.7

(this paper) 0.1±0.4 logP - 0.7±1.0 2.2±1.0 3.8±1.1 5.4±1.1 7.0±1.2 0.3±1.1 5.1±1.2

(Chemsketch) 0.9±0.9

Circular Dichroism measurements

The following figures show circular dichroism data of linear (Figure S2.4) and branched (Figure

S2.5) capped peptides and of linear and branched uncapped peptides (Figure S2.6). Measurements were carried out in the same conditions as described in the main text.

a b c

d e f

Figure S2.4. Temperature/wave length scan of linear capped peptides. G1-5 (a), G1-10 (b), G1-

15 (c), G1-20 (d), G1-25(e), G1-30(f). Measurements were taken at 0.5-1 mg/mL in PBS (2.5 mM).

a b c

Figure S2.5. Temperature/wavelength scan of branched capped peptides. G2-5 (a), G2-15 (b), G3-

5 (c). Measurements were taken at 0.5-1 mg/mL in PBS (2.5 mM).

a b c

d e

g h i

Figure S2.6. CD spectra of uncapped peptides at different temperatures. G1-5U (a), G1-10U(b),

G1-15U(c), G1-20U(d), G1-25U(e), G1-30U(f) G2-15U (g), G2-20U (h), G3-5U (i).

Measurements were taken at 0.5-1 mg/mL in PBS (2.5 mM).

Van’t Hoff endpoint determination

The endpoints of the transition, [θ]F and [θ]U, represent the values for the fitted minimum and maximum points of the transition (high-temperature folded form, [θ]F, and low-temperature

48 unfolded form, [θ]U). They were chosen as the initial and the final points of the transition in the inspected temperature range where [θ]F and [θ]U were the values of the mean molar residue ellipticity at 80 °C and 20 °C, respectively.

DLS measurements

DLS measurements were performed using a Zetasizer Nano-ZS (Malvern, UK). Samples were dissolved in PBS (pH 7.4) to a final concentration of 5 mg/mL. Samples were equilibrated for 5 min at the measuring temperature prior to data collection. Correlograms were collected at 173° for at least 10 runs of 10 s. The recorded correlograms were analyzed with the CONTIN procedure using the software provided with the instrument.

Figure S2.7: Temperature dependence of the average scattered intensity. Solid lines are drawn as guides to the eye to emphasize the sharp increase in intensity (attributed to coacervate formation) above the peptide's Tt.

Figure S2.8: Size distribution of G1-15 (A), G1-20 (B), G1-25 (C) and G2-10 (D) in PBS

0 measurements. The temperatures above which large aggregates are detected correspond to the Tt determined by UV-Vis and turbidity measurements.

References

(1) Tao, P.; Wang R.; Lai, Lu. J. Mol. Model. 1999, 5, 189.

Chapter 3: Multi-Scale Characterization of Thermoresponsive Dendritic

Elastin-Like Peptides

“Reprinted (adapted) with permission from Zhou, M; Shmidov, Y.; Matson, J. B.; Ronit Bitton, R.

Authors

Mingjun Zhou, [a] Yulia Shmidov, [b] John B. Matson, * [a] Ronit Bitton* [b]

[a] Department of Chemistry and Macromolecules Innovation Institute, Virginia Tech,

Blacksburg, VA 24061 (U.S.A.)

[b] Department of Chemical Engineering, Ben-Gurion University of the Negev, Beer-Sheva, 84105

(Israel)

3.1 Abstract

Elastin like peptides (ELPs)—polypeptides based on the protein elastin—are used widely as thermoresponsive components in biomaterials due to the presence of a sharp soluble-to-insoluble phase change at a characteristic transition temperature (Tt). While linear ELPs have been thoroughly studied, few investigations into branched ELPs have been carried out. Using lysine amino acids as branching and terminal units with 1–3 pentameric repeats between each branch,

ELP dendrimers were prepared by solid-phase peptide synthesis with molecular weights as high as 14 kDa. A conformation change from random coil to β-turn upon heating through the Tt, typical of ELPs, was observed by circular dichroism spectroscopy for all peptides. The high molecular weights of these peptides enabled the use of characterization techniques typically reserved for

52 polymers. Variable-temperature small-angle X-ray scattering measurements in dilute solution revealed an increase in size and fractal dimension upon heating, even well below the Tt. These results were corroborated by cryogenic transmission electron microscopy, which confirmed the presence of aggregates below the Tt, and micro differential scanning calorimetry, which showed a broad endothermic peak below the Tt. These results collectively indicate the presence of a pre- coacervation step in the phase transition of ELP dendrimers.

3.2 Introduction

Branching structures are widely adopted in nature across many length scales. Examples include neural processes, the mammalian vascular system, as well as tree branches, root systems, and leaf veins. Nature-inspired branching systems appear widely in human designs (e.g., art and architecture), analyses (e.g., factor trees, heat maps), and organization (e.g., family trees, company organizational structures). Chemists have also been inspired by branching structures, with highly branched dendritic polymers (dendrimers) being perhaps the most thoroughly studied example 1,2.

Characterized by their globular structure and high surface functionality imparted by exponential branching, dendrimers have attracted much research interest with goals of applying these unique polymers as catalysts 3,4, drug delivery vehicles 5,6, and nano-objects with precisely controlled structures 7-9.

Dendrimers with static branching structures, including Frechét-type dendrimers and polyamidoamine (PAMAM) dendrimers among many others, have been widely investigated 10-12.

In contrast, responsive dendrimers have been less thoroughly studied 13-15. This is surprising given the large amount of research effort devoted toward responsive polymers, with stimuli including temperature, pH, light, and others 16,17. Among these, thermoresponsive polymers have gained the most research attention, with potential applications primarily in drug/gene delivery 18-20, tissue

53 engineering 21,22, bioseparation 23-25, and sensors 26-28. Thus far, most reports on dendrimers with thermoresponsive components have focused on attaching peripheral thermoresponsive moieties to static dendrimer cores 29-33. Dendrimers with thermoresponsive units built into the entire structure, which are synthetically more challenging, may allow the benefits of branching to be applied broadly in new smart materials. Therefore, it remains interesting and useful to understand how the

3D structures of thermoresponsive dendrimers transform in response to changes in the conformation of branching units throughout.

Elastin is a thermoresponsive extracellular matrix protein that maintains the structure and persistent elasticity of many parts of the human body, including skin, blood vessels, and connective tissues 34,35. Elastin like peptides (ELPs) are biomacromolecules derived from the hydrophobic domains of elastin and are widely used as thermoresponsive units 36,37. With the sequence Xaa-

Pro-Gly-Xaa-Gly (XPGXG), ELPs exhibit lower critical solution temperature (LCST) behavior

(X represents different amino acids, which can be changed to tune the transition temperature). The

LCST behavior of linear and star-shaped ELPs has been thoroughly investigated, including factors such as hydrophobicity balance, salt/peptide concentration, chain length 38-40.

Evidence suggests that the phase transition in ELPs from soluble peptide to insoluble precipitate proceeds through a two-step process of a secondary structure transition from random coil to type

II β-turn followed by coacervation 33,41. Initial studies from our groups showed that while these two steps also govern the phase transition of branched/dendritic ELPs, the branching structure in the dendrimers slows down the coacervation process, leading to a higher transition temperature

(Tt). Furthermore, the change in peptide secondary structure starts well below the Tt, and there are no significant differences in entropy/enthalpy changes for the secondary structure transition upon heating between the dendrimers and their linear counterparts.

Herein we discuss an extension of our previously developed synthetic methods to make ELP dendrimers with molecular weights approaching those of typical polymers, with some exceeding

14,000 Da. This enables the use of characterization techniques suitable for small peptides along with other techniques typically reserved for polymers and nanostructures to evaluate how changes in peptide secondary structure affect the 3-dimensional structure of the ELP dendrimers. Using a combination of turbidity studies, circular dichroism (CD), micro differential scanning calorimetry

(micro-DSC), cryogenic transmission electron microscopy (cryo-TEM) and small-angle X-ray scattering (SAXS), we sought to gain insight into the process of coacervation and precipitation in large ELP-based dendrimers.

3.3 Results and Discussion

Design and synthesis of ELP dendrimers

Scheme 3.1 Synthetic route for preparation of ELP dendrimers using SPPS.a

55 aCleavage steps for DG2-5 and DG3-5 have been omitted from the figure for clarity. ELP dendrimers with 10mer and 15mer repeats were prepared in a similar fashion.

Scheme 3.1 shows the synthetic route used to produce the ELP dendrimers. Compared to our previous report 41, peptides with higher branching generation and MW were synthesized to enable characterization using previously inaccessible techniques. Using the divergent synthesis strategy, in which mass and branch number increases exponentially, the synthesis of these peptide dendrimers was less time consuming than their linear counterparts, with satisfactory yield maintained (Figure S3.1). The notation “DGx-y” designates a peptide dendrimer of xth generation with y amino acid residues between the branching units. For example, DG3-5 stands for the 3rd generation peptide dendrimer, with the pentameric repeating unit GLPGL between branching Lys resides (Figure 3.1). We synthesized the peptides on a microwave-assisted synthesizer, and Fmoc-

Lys (Mtt)-OH was used as the branching unit. After deprotection of the Fmoc group, the Mtt protecting group was removed using dilute trifluoroacetic acid (TFA) to reveal two amines. Each generation was grown through the repeated procedure of Fmoc-Lys(Mtt)-OH addition, removal of both protecting groups, and conventional solid-phase peptide synthesis (SPPS) to add (GLPGL)n repeats (n = 1–3) onto each free amine. We used the sequence GLPGL as the repeating pentamer because ELPs with Xaa = Leu have lower LCSTs compared with traditional Xaa = Val sequences, with good solubility retained 42. The crude products were purified by preparative HPLC and lyophilized to afford dry powders for storage and analysis. Peptide purity was confirmed by LC-

MS (Figure S3.2) with purity >90% for all peptides except for DG3-10, which was too large to analyze by this technique. Prior to LC-MS analysis, ELP dendrimers were acetylated by addition of acetic anhydride in order to reduce the complexity of the chromatograms resulting from differing retention of peptides in different ionization states. Analysis of peptide DG3-10 was

56 performed using gel electrophoresis because its molecular weight is close to that of a small protein, over 14kDa (Figure S3.3).

Figure 3.1 Chemical structure of DG3-5 highlighting the differences between G1, G2, and G3

ELP dendrimers. In the case of the DGX-5 dendrimers, each has a single pentamer between branching points. DGX-10 and DGX-15 dendrimers have two or three pentamers between branching points, respectively.

Characterization of the thermal properties of ELP dendrimers

Secondary Structure

ELPs undergo a characteristic conformational transition from random coil to -turn structure upon heating 39. To verify this transition in the ELP dendrimers, circular dichroism (CD) spectra of all peptides were obtained at various temperatures. As expected, at temperatures below the LCST all profiles showed minima at 200 nm and maxima near 218 nm (Figure 3.2A, Figure S3.4) typical of a random-coil conformation 43. Upon heating, the peak intensity decreased (Figure 3.2B), and an isodichroic point was observed at 210 nm, indicating a transition from random-coil to turn 44.

For DG2-15 and DG3-10, the signal reached a plateau above 30 °C (Figure 3.2B). This was likely

57 not a result of a sudden halt to the conformational change, but rather due to the presence of aggregates, which disrupt the CD signal even at low concentrations 45.

Figure 3.2. A) CD spectra of DG3-5 at various temperatures, demonstrating the conformational change from random coil to β–turn upon heating. B) Temperature dependence of the mean molar residue ellipticity at 218 nm of selected dendrimers. All samples were measured at 0.5 mg mL−1 in 2.5 mM PBS.

Coacervation

The phase transitions of dendrimers in pure aqueous solutions were examined by turbidity measurements, but none of the solutions became turbid even after heating above 80 °C, indicating the Tt is beyond our detection limit. Kosmotropic salts, which are typically used to salt out proteins from water, lower the Tt of ELPs due to several effects. Kosmotropic anions polarize the water molecules involved in hydrogen bonding to the amide and thereby weaken the hydrogen bonding

58 of the water to the ELP. The cost of hydrating hydrophobic regions of ELPs (i.e., Leu residues) increases with higher salt concentration as well 46,47. In addition, salt screens the positive charges of the Lys side-chain amines at the periphery of the ELP dendrimers and thus weakens the charge- dipole interaction between ELP dendrimers and water molecules 47. Because salts such as NaCl

48 lower the Tt of ELPs , the turbidity experiments were repeated in phosphate buffered saline (PBS) at pH = 7.4 with various amounts of added NaCl. The PBS concentration was kept constant

(0.01 M) while the NaCl concentration was increased in 0.25 M intervals until turbidity was observed. Figure 3.3 shows the dependence of Tt on NaCl concentration for all dendrimers.

Figure 3.3 Tt of the ELP dendrimers in PBS (pH = 7.4) with various amounts of added NaCl. The

NaCl concentration is the total concentration, including 0.137 M NaCl from PBS.

0 In our previous report on branched ELPs, we calculated a Tt value for each peptide, which was obtained by linear extrapolation of the Tt vs. [NaCl] graph to zero salt concentration. Here we observed that the trend deviated from linearity outside of a narrow range for each dendrimer,

0 making Tt calculations unreliable. Instead, we chose to compare the amount of salt required for each peptide dendrimer to have a Tt at 37 °C (Table 3.1), which we refer to as the critical salt

59 concentration (CSC). In all cases, this was in the linear (or near-linear) range of the plot. Variable- temperature UV-vis spectroscopy measurements of DG2-10, DG2-15 and DG3-10 (Figure S3.5) in pH = 7.4 solutions with salt concentration equal to the CSC show an upturn in the O.D at ~ 37 o C (Table 3.1), confirming our CSC results. The Tt of ELP dendrimers in salty solution was inversely correlated to the peptide concentration (Figure S3.6), which is compatible with the general trend described in literature for linear ELPs 49. It should also be noted that all turbid solutions became clear after cooling, indicating that the phase transition/coacervation is reversible.

Table 3.1. Summary of data for the dendrimers

a Peptide Npls MW CNaCl [M] required name [Da]b for coacervation

at 37 °C (CSC)c

DG1-5 2 1276 _____d

DG2-5 4 3537 3.39

DG1-10 4 2150 3.25

DG3-5 6 8060 2.24

DG2-10 8 6161 0.89

DG3-10 12 14181 0.55

DG2-15 12 8784 0.24

a Npls indicates the number of pentamers in the longest linear segment (Figure 3.1). b MW indicates exact mass. c Total NaCl concentration required to achieve turbidity at 37 °C, calculated from best fits to lines determined from the data shown in Figure 3.3. d For DG1-5 no turbidity was observed in the examined salt range.

We first aimed to determine whether any correlations existed between CSC and the molecular structure of the ELP dendrimers. Contrary to linear ELPs, in which molecular weight (MW) and peptide length can be treated as the same parameter 42,50, for ELP dendrimers they are two separate

0 variables. In linear ELPs, increasing the MW of a peptide decreases its Tt and its CSC. For the

ELP dendrimers studied here, however, there is no correlation between the MW of the peptide and its CSC. For example, DG3-5 has higher molecular weight than DG2-10, but it has a higher CSC.

By analyzing the data in Table 3.1, we can correlate CSC to molecular structure based on two factors: the number of pentamer units (GLPGL) in the longest linear peptide sequence (Npls) and the generation number. An increase in Npls clearly results in a decrease in CSC. This is consistent with linear ELPs, where peptide length is the only variable between peptides of the same structure.

When the Npls of several dendrimers is similar, CSC is correlated with generation number, with higher generation ELP dendrimers showing higher CSC values. This can be attributed to the fact that generation number is a result of two contributions: the number of terminal amines and the number of branches. At pH=7.4 most of the terminal amines would be expected to be protonated, and hydrophilic components such as ammonium salts elevate the CSC. The number of terminal amines increases with each dendrimer generation, so CSC would be expected to increase with

0 increasing dendrimer generation. Branching also elevates the Tt (equivalent to elevating the CSC), probably due to steric effects, as we reported previously 41.

Taken together, these data demonstrate that Npls is the most important factor in determining CSC in these ELP dendrimers. This result is supported by recent molecular dynamics studies, which indicate that solvent accessible surface area is a key parameter influencing ELP Tts. The authors conclude that ELPs with greater solvent accessible surface area have lower Tts (i.e., lower CSCs) due to a more significant dehydration process 51,52. A lower generation number ELP dendrimer

61 with longer segments between branches (e.g., DG2-15) would therefore be expected to have a larger solvent accessible surface area than a higher generation number ELP dendrimer with shorter segments between branches (e.g., DG3-5). These two ELP dendrimers have similar molecular weights, but DG2-15 has a much lower CSC than DG3-5 (0.24 M vs. 2.24 M), consistent with the conclusions from the molecular dynamics studies.

Figure 3.4 Small angle scattering curves of 1 wt.% dendrimers solution at 20 °C. Curves are offset for sake of clarity.

Nano Structure

The high MWs of these dendrimers allowed us to employ characterization techniques typically used for polymers and nanoparticles to evaluate how changes in peptide secondary structure affect the 3-dimensional structure of the dendrimer. First, SAXS measurements were conducted on 1 wt.% solutions of dendrimers in aqueous PBS solution. Log-log plots of the scattering patterns of the dendrimer solutions at 20 oC are presented in Figure 3.4, and the curves resemble typical patterns for low generation dendrimers or star-like molecules (i.e., a plateau in the mid-q regime and an upturn in the low-q regime) 53,54.

The scattering of dendrimers in solution can be analyzed using the empirical multiple level unified exponential/power-law fit method developed by Beaucage 55, described by Equation 3.1:

N BerfqR[(/6)] 3Pi IqBkgdGq()exp(/=+−+ 3) R 22 ig i ,  ig i , Pi i=1 q

Where G is a Guinier scaling factor, Rg is radius of gyration, and B is a prefactor specific to the type of power-law scattering. B is defined according to the regime in which the exponent P falls

(which is practically a power low of the scattering curve in the high-q regime). Generally, the value of P is defined as P < 3 for mass fractals, 3 < P < 4 for surface fractals, and P > 4 for diffuse interfaces.

The upturn in the scattering at low q is not fully understood, although it typically indicates the presence of larger structures, most likely aggregates. This phenomenon has been seen for dendritic structures and has been attributed to weak attractive interactions between molecules 56. A two- level model (i.e., two structural elements) seems the most appropriate to describe these scattering curves. However, though this model can be fitted to the data, the obtained Rg values of the aggregates were sensitive to our initial guesses and not reliable. Therefore, we chose to use a one- level model to describe the scattering from the smaller features, i.e., the ELP dendrimers. SAXS measurements were taken for each dendrimer at various temperatures, and the best fits to the one- level model are presented as solid lines in Figure 3.5. As DG1-5 and DG1-10 are not branched, their scattering curves are not fitted to this model. The Rg values of the dendrimers obtained from the best fits are also presented in Figure 3.5.

SAXS measurements of 1 wt.% solutions of ELP dendrimer in 0.01 M PBS (pH = 7.4) at 20 °C,

40 °C, and 50 °C were conducted in order to investigate how temperature influences the dendrimer sizes in solution. As coacervation is considered to be a sharp transition process, we expected to see differences in the scattering curves only for the peptides which become turbid upon heating in

PBS with no added salt and only between temperatures that are lower and higher than the Tt.

Consistent with our expectations, for ELP dendrimers DG1-5, DG1-10, DG2-5 and DG3-5, which

Figure 3.5. SAXS curves of: A) DG1-5, B) DG1-10, C) DG2-5, D) DG3-5, E) DG2-10, F) DG3-

10, G) DG2-15. Solid black and yellow lines represent fits to the model described by Equation 1.

All samples were measured at 1 wt.% ELP dendrimers in 0.01 M PBS (pH = 7.4) solution at 20

°C, 40 °C and 50 °C. 64 do not become turbid during heating under these conditions, SAXS patterns show no significant changes upon heating (Figure 3.5A-D). In contrast, ELP dendrimers DG3-10 and DG2-15 did not behave as expected. These peptides become turbid upon heating in PBS solution without added salt (DG3-10 at 65 °C and DG2-15 at 43 °C), and significant changes in the SAXS patterns are observed upon heating (Figure 3.5F and 3.5G). Remarkably, a continuous change occurs at temperatures well below the Tt for both peptides. Similarly, there are clear changes in the SAXS pattern of DG2-10 even though it does not become turbid under these conditions (Figure 3.5E).

To understand the changes occurring in the ELP dendrimers DG2-10, DG2-15, and DG3-10 during heating, we carefully examined the SAXS patterns of all peptides. For all ELP dendrimers, we evaluated the Porod exponent (P), which is taken from the high q region and provides information about the surfaces of the particles. All ELP dendrimers showed 2

4.0 at 50 °C, which indicates presence of surface fractals (indicative of phase separation). For

DG3-10, P increases from 2.8 to >4 over this temperature range, indicating diffuse interfaces, which correlates to a phase transition in progress 57. For DG2-10, P also increases from 2.0 to 2.7 over this temperature range.

We also evaluated the changes in Rg for each dendrimer at different temperatures based on fits to the Beaucage model (note that for DG2-15 at 50 C, the Beaucage model could not be fitted to this turbid solution since the scattering from the interpartical interactions (structure factor) overlaps with the scattering of the dendrimers (form factor)) 58. From Figure 3.5E-G it can easily be seen that with increasing temperature a larger Rg is obtained for the same dendrimer. This is surprising because during heating the conformation of the peptide changes from random coil to β-

65 turn, which should lead to a more compressed structure. The fact that Rg becomes larger rather than smaller upon heating suggests that there is continuous aggregation process, i.e., a pre- coacervation involving small numbers of peptides prior to the coacervation, which is visible as turbidity. Similar aggregation associated with dehydration prior to coacervation was observed for

33 a PAMAM dendrimer with linear VPGVG units on its surface . We note that this increase in Rg upon heating is not exclusive to dendrimers; it was also observed for the linear KK(GLPGL)8KK peptide, which has the same Npls as DG2-10 (Figure S3.7).

Figure 3.6 Cryo-TEM images of ELP dendrimer DG3-10 (5 mg mL-1 in PBS at pH = 7.4) at 25°C

(A) and 50°C (B) and peptide DG2-15 (10 mg mL-1 in PBS at pH = 7.4) at 25°C (C) and 50°C (D).

In order to verify the existence of large aggregates below the Tt, cryo-TEM images of DG3-10 and

DG2-15 were taken at 25 oC and 50 oC (Figure 3.6). Consistent with the SAXS data, large

o aggregates are visible for DG3-10 at 50 C, even though it is below Tt, while none are visible at 25

°C. In cryo-TEM images of peptide DG2-15 (Tt = 43 °C), smaller aggregates are already seen at

25 oC, and larger ones appear at 50 oC, validating the existence of a pre-coacervation phase change.

The complementary characteristics of SAXS and cryo-TEM allow us to portray the entire system—SAXS shows scattering primarily from single dendrimers and small aggregates while cryo-TEM gives us information about large aggregates.

Figure 3.7 Micro DSC of peptide DG2-15 (10 mg mL-1 in PBS at pH = 7.4) with a heating rate 60 oC h-1. Samples were allowed to equilibrate for 15 min before each scan.

This aggregation well below the Tt of peptides DG2-15 and DG3-10 is likely associated with dehydration, which triggers coacervation of ELPs. This is supported by the CD data, which show that the ordered β-turn character of the dendrimers increases with increasing temperature. Thus, the secondary structure transition might be a consequence of hydrophobic interactions that are driven by positive entropies of (partial) dehydration 44,59. Micro-DSC measurements of DG2-15

(Figure 3.7) indeed show that in addition to the small endothermic peaks characteristic of sedimentation (i.e., coacervation), which start at ~46 °C, another endothermic peak exists at 23 °C

67 corresponding to the dehydration process. A clear dehydration peak could be obtained only for

DG2-15 (the ELP dendrimer with the lowest CSC and highest Npls), emphasizing the importance of solvent accessible surface area. Recently oligo(ethylene glycol) decorated dendrimers were reported to exhibit a “sub-LCST” below the traditional LCST with an drastic aggregate size increase 60,61. Although the transition was rather sharp compared to the gradual change over a broad temperature range in our system, this phenomenon was also presumed to be related to the molecular hydration state. Importantly, the reported drastic change in guest molecule encapsulation stability indicated potential drug release applications with such transitions. It is also worth noting that experiments were all conducted at pH = 7.4; this pre-coacervation phenomenon may be quite different at different pH values.

3.4 Conclusions

We have synthesized amine-terminated ELP dendrimers and investigated the effect of molecular architecture on their thermoresponsiveness using a variety of characterization techniques from molecular to macroscale. Turbidity measurements revealed a correlation between the number of pentamer units in the longest linear peptide sequence and CSC, consistent with computational studies that established solvent accessible surface area as an important variable influencing the Tt of ELPs. SAXS and cryo-TEM were also employed to investigate the process of the phase transition, exposing a pre-coacervation step well below the Tt of these ELP dendrimers. Micro-

DSC measurements confirmed these results through the presence of a broad endothermic peak below Tt, likely resulting from the dehydration process before sedimentation. The highly branched nature of these ELP dendrimers may enable their preferred use in applications where linear ELPs have drawbacks. For example, branching is a well-known strategy for promoting proteolytic resistance in peptides, potentially enabling ELP dendrimers to survive longer in vitro and in vivo

62 than their linear counterparts . Additionally, the Tt in ELP dendrimers appears to be determined primarily by the longest linear peptide sequence, allowing for a wide variety of ELPs to be prepared with similar Tts but variable topologies and degrees of branching. Incorporation of terminal amines on the periphery of the dendrimer makes them chemically reactive and thus easily modifiable for various applications. Lastly, the pre-coacervation step observed here may enable controlled drug delivery from ELP dendrimers while avoiding safety concerns resulting from insolubility above the Tt. Overall, we expect that ELP dendrimers may find utility in a variety of complex materials, enabling applications in tissue engineering and drug delivery.

3.5 Experimental Section

Synthesis: Rink amide MBHA resin (100-200 mesh, 0.48 mmol g-1) and Fmoc-protected amino acids were purchased from ChemPep, Inc. All other solvents and reagents were purchased from commercial sources and used as received. All of the peptides were synthesized using a CEM

Liberty 1 microwave-assisted automated peptide synthesizer. Purification by preparative-scale reverse phase high performance liquid chromatography (RP-HPLC) was carried out on an Agilent

Technologies 1260 Infinity HPLC system, eluting with a gradient of 2% to 90% ACN in MilliQ

H2O on an Agilent PLRP-S column (100Å particle size, 25 x 150 mm) and monitoring at 220 nm.

0.1% TFA was added to each mobile phase. Fractions were analyzed by mass spectrometry, including a single-quad electrospray ionization mass spectrometer (Advion ExpressIon Compact

Mass Spectrometer) and a matrix-assisted laser desorption ionization–tandem time of flight mass spectrometer (4800 MALDI TOF/TOF; AB Sciex). Product-containing fractions were combined, concentrated in vacuo to remove ACN, and lyophilized (LabConco).

Circular dichroism spectroscopy (CD): Measurements were carried out using a spectropolarimeter

(J-715, Jasco, U.S.A.) equipped with a PTC-348WI temperature control unit under N2 flow (80

69 mL min−1). Spectra were recorded from 260 to 190 nm at a scan speed of 50 nm min−1, a step size of 0.5 nm, a bandwidth of 1 nm, and a response time of 1 s. For each sample four scans were averaged. Far UV−vis quartz cells (Starna, type 21/Q/1) were used with a path length of 0.1 cm.

All peptides were dissolved in phosphate buffer (2.5 mM; pH 7.4) and measured at a concentration of 0.5 mg mL−1. The heating rate was 1 °C min−1. The data were analyzed using Jasco spectra analysis software version 1.53. Four measurements were averaged using a moving average technique. The CD values were converted into mean molar residue ellipticity (MMRE; [θ], in mdeg cm2 dmole −1 res −1).

Turbidity measurements: 200 µL of 5 mg mL-1 peptide solution in PBS buffer (from Sigma

Aldrich, 10 mM phosphate, 2.7 mM KCl, 137 mM NaCl, pH = 7.4) with different added salt

(NaCl) concentrations were added to the quartz capillary, which was then placed in a heating block

(Thermo scientific, U.S.A.). The Tt was defined as the temperature at which the solution turned from clear to turbid.

UV-vis spectroscopy: Transition temperature determination was confirmed a using UV−vis spectrophotometer (BioTek Synergy Mx) with temperature control unit at a wavelength of 350 nm. Samples were heated from 24 to 46 °C. At each temperature the sample was shaken for 5 s and then held in the unit for 10 s prior to measurement. The optical density was recorded at each temperature with intervals of 1–2 °C between measurements using Gen5 software.

Small-angle X-ray scattering (SAXS): SAXS experiments were performed at the BM29 beamline at the European synchrotron radiation facility [ESRF] in Grenoble, France. An energy of 12.5 keV corresponding to a wavelength of 0.998 Å−1 was selected. The scattering intensity was recorded using a Pilatus 1 M detector, in the interval 0.004 < q < 0.5 Å−1. Ten frames with 2 s exposure time were recorded for each sample. Measurements were performed in flow mode where samples were

70 pushed through the capillary at a constant flow rate. The dedicated beamline software BsxCuBe was used for data collection and processing. The scattering spectra of the solvent were subtracted from the corresponding solution data using the Irena package for analysis of small-angle scattering data 63. Data analysis was based on fitting the scattering curve to an appropriate model by software provided by NIST (NIST SANS analysis version 7.0 on IGOR) 64.

Cryo-TEM: Dispersions were characterized via direct imaging using cryo-TEM. Samples were prepared by depositing a droplet of 2−4 μL on a TEM grid (300 mesh Cu Lacey grid; Ted Pella

Ltd.). Ultrathin films (100−150 nm) were formed as most of the solution was removed by blotting.

The specimen was vitrified by rapid plunging into liquid ethane precooled with liquid nitrogen at controlled temperature and relative humidity (Leica EM GP). The vitrified samples were transferred to a cryo-specimen holder (Gatan model 626) and examined at -181 °C in low-dose mode. Imaging was carried out using FEI Tecnai 12 G2 TEM operated at 120 keV equipped with a Gatan 794 CCD camera.

Differential scanning calorimetry (DSC): DSC experiments were performed on a VP-DSC

(MicroCal, USA) calorimeter. Dendrimer concentrations used in the experiments were 1.14 mM

(10 mg mL-1), and a PBS solution (0.01 M, pH = 7.4) served as a blank. Heating scans were run at a rate of 1 °C min–1. Data analysis was performed using MicroCal Origin 6.0 software.

Gel electrophoresis: 6 μL of 2.5 mg mL-1 peptide aqueous solution was mixed with 5 μL of sample buffer (100 mM Tris, pH 6.8, 2% SDS, 5% β-mercaptoethanol, 15% glycerol and traces of bromophenol blue), prior to heating for 3 min at 100 oC. After cooling, the sample mixture was loaded into the gels (5% stacking; 15% separating). SDS-PAGE was performed with Tris-Glycine

SDS buffer at 150V. A mixture of standard proteins (10-250kDa, BioLabs) was used as molecular weight marker.

Peptide acetylation: 15 mg of peptide was dissolved in 1mL of capping cocktail (acetic anhydride/N,N-diisopropylethylamine/DMF 40:10:100 v/v/v), and allowed to react for 50 min. 1 mL of water was added to the reaction mixture, prior to removing DMF with sep-pak C18 cartridges (acetonitrile/H2O as eluent). Then the sample was dried with lyophilization.

3.6 References

(1) Bosman, A. W.; Janssen, H. M.; Meijer, E. W. Chem. Rev. 1999, 99, 1665.

(2) Caminade, A.-M.; Turrin, C.-O.; Laurent, R. Dendrimers : Towards Catalytic, Material and Biomedical Uses; John Wiley & Sons, Ltd, 2011.

(3) Peng, M.; Chen, Y.; Tan, R.; Zheng, W.; Yin, D. RSC Adv. 2013, 3, 20684.

(4) Astruc, D.; Chardac, F. Chem. Rev. 2001, 101, 2991.

(5) Martinho, N.; Florindo, H.; Silva, L.; Brocchini, S.; Zloh, M.; Barata, T. Molecules 2014,

19, 20424.

(6) Caminade, A.-M.; Turrin, C.-O. J. Mater. Chem. B 2014, 2, 4055.

(7) Woo-Dong, J.; Nishiyama, N.; Kataoka, K. Supramol. Chem. 2007, 19, 309.

(8) Maleki, B.; Sheikh, S. RSC Adv. 2015, 5, 42997.

(9) Yang, Y.; Wang, Y.; Li, H.; Li, W.; Wu, L. Chem.-Eur. J. 2010, 16, 8062.

(10) Ma, J.; Ma, X.; Deng, S.; Li, F.; Hu, A. J. Polym. Sci. A Polym. Chem. 2011, 49, 1368.

(11) Maiti, P. K.; Çaǧın, T.; Wang, G.; Goddard, W. A. Macromolecules 2004, 37, 6236.

(12) Carlmark, A.; Hawker, C.; Hult, A.; Malkoch, M. Chem. Soc. Rev. 2009, 38, 352.

(13) Wang, H.; Huang, Q.; Chang, H.; Xiao, J.; Cheng, Y. Biomater. Sci. 2016, 4, 375.

(14) Hu, W.; Qiu, L.; Cheng, L.; Hu, Q.; Liu, Y.; Hu, Z.; Chen, D.; Cheng, L. Acta Biomater.

2016, 36, 241.

(15) Chen, L.; Cao, W.; Grishkewich, N.; Berry, R. M.; Tam, K. C. J. Colloid Interf. Sci. 2015,

450, 101.

(16) Stuart, M. A. C.; Huck, W. T. S.; Genzer, J.; Muller, M.; Ober, C.; Stamm, M.;

Sukhorukov, G. B.; Szleifer, I.; Tsukruk, V. V.; Urban, M.; Winnik, F.; Zauscher, S.; Luzinov, I.;

Minko, S. Nat. Mater. 2010, 9, 101.

(17) Yan, X. Z.; Wang, F.; Zheng, B.; Huang, F. H. Chem. Soc. Rev. 2012, 41, 6042.

(18) Laga, R.; Janoušková, O.; Ulbrich, K.; Pola, R.; Blažková, J.; Filippov, S. K.; Etrych, T.;

Pechar, M. Biomacromolecules 2015, 16, 2493.

(19) Sedláček, O.; Černoch, P.; Kučka, J.; Konefal, R.; Štěpánek, P.; Vetrík, M.; Lodge, T. P.;

Hrubý, M. Langmuir 2016, 32, 6115.

(20) Twaites, B. R.; de las Heras Alarcón, C.; Lavigne, M.; Saulnier, A.; Pennadam, S. S.;

Cunliffe, D.; Górecki, D. C.; Alexander, C. J. Control. Release 2005, 108, 472.

(21) Ashraf, S.; Park, H. K.; Park, H.; Lee, S. H. Macromol. Res. 2016, 24, 297.

(22) Cortez-Lemus, N. A.; Licea-Claverie, A. Prog. Polym. Sci. 2016, 53, 1.

(23) Nagase, K.; Kobayashi, J.; Kikuchi, A.; Akiyama, Y.; Kanazawa, H.; Okano, T.

Biomacromolecules 2014, 15, 1204.

(24) Nagase, K.; Kobayashi, J.; Kikuchi, A.; Akiyama, Y.; Kanazawa, H.; Okano, T. RSC Adv.

2015, 5, 66155.

(25) Roy, D.; Brooks, W. L. A.; Sumerlin, B. S. Chem. Soc. Rev. 2013, 42, 7214.

(26) Ward, M. A.; Georgiou, T. K. Polymers 2011, 3, 1215.

(27) Liu, Y.; Xu, L.; Liu, J. S.; Liu, X. Y. Anal. Methods 2015, 7, 10151.

(28) Liu, L. X.; Li, W.; Yan, J. T.; Zhang, A. F. J. Polym. Sci. Pol. Chem. 2014, 52, 1706.

(29) Li, W.; Zhang, A.; Feldman, K.; Walde, P.; Schlüter, A. D. Macromolecules 2008, 41,

3659.

(30) Fukushima, D.; Sk, U. H.; Sakamoto, Y.; Nakase, I.; Kojima, C. Colloid. Surface B 2015,

132, 155.

(31) Bekhradnia, S.; Zhu, K. Z.; Knudsen, K. D.; Sande, S. A.; Nystrom, B. J. Mater. Sci.

2014, 49, 6102.

(32) Zheng, Q.; Pan, C.-y. Eur. Polym. J. 2006, 42, 807.

(33) Kojima, C.; Irie, K.; Tada, T.; Tanaka, N. Biopolymers 2014, 101, 603.

(34) Navon, Y.; Bitton, R. Isr. J. Chem. 2016, 56, 581.

(35) Bidwell Iii, G. L.; Raucher, D. Adv. Drug. Deliver. Rev. 2010, 62, 1486.

(36) Hassouneh, W.; MacEwan, S. R.; Chilkoti, A. Fusions of elastin-like polypeptides to pharmaceutical proteins; Elsevier Academic Press Inc: San Diego, 2012; Vol. 502.

(37) van Eldijk, M. B.; McGann, C. L.; Kiick, K. L.; van Hest, J. C. M. Elastomeric

Polypeptides; Springer-Verlag Berlin: Berlin, 2012; Vol. 310.

(38) Ghoorchian, A.; Holland, N. B. Biomacromolecules 2011, 12, 4022.

(39) Nuhn, H.; Klok, H.-A. Biomacromolecules 2008, 9, 2755.

(40) Quiroz, F. G.; Chilkoti, A. Nat. Mater. 2015, 14, 1164.

(41) Navon, Y.; Zhou, M.; Matson, J. B.; Bitton, R. Biomacromolecules 2016, 17, 262.

(42) Urry, D. W.; Trapane, T.; Prasad, K. Biopolymers 1985, 24, 2345.

(43) Reiersen, H.; Clarke, A. R.; Rees, A. R. J. Mol. Biol. 1998, 283, 255.

(44) Nicolini, C.; Ravindra, R.; Ludolph, B.; Winter, R. Biophys. J. 2004, 86, 1385.

(45) Urry, D. W. BBA-Biomembranes 1972, 265, 115.

(46) Deyerle, B. A.; Zhang, Y. Langmuir 2011, 27, 9203.

(47) Cho, Y.; Zhang, Y.; Christensen, T.; Sagle, L. B.; Chilkoti, A.; Cremer, P. S. The Journal of Physical Chemistry B 2008, 112, 13765.

(48) Luan, C. H.; Parker, T. M.; Prasad, K. U.; Urry, D. W. Biopolymers 1991, 31, 465.

(49) Meyer, D. E.; Chilkoti, A. Biomacromolecules 2004, 5, 846.

(50) Meyer, D. E.; Chilkoti, A. Biomacromolecules 2002, 3, 357.

(51) Li, N. K.; Quiroz, F. G.; Hall, C. K.; Chilkoti, A.; Yingling, Y. G. Biomacromolecules

2014, 15, 3522.

(52) Zhao, B.; Li, N. K.; Yingling, Y. G.; Hall, C. K. Biomacromolecules 2016, 17, 111.

(53) Prosa, T. J.; Bauer, B. J.; Amis, E. J. Macromolecules 2001, 34, 4897.

(54) Prosa, T. J.; Bauer, B. J.; Amis, E. J.; Tomalia, D. A.; Scherrenberg, R. J. Polym. Sci. Pol.

Phys. 1997, 35, 2913.

(55) Beaucage, G. J. Appl. Crystallogr. 1995, 28, 717.

(56) Pötschke, D.; Ballauff, M.; Lindner, P.; Fischer, M.; Vögtle, F. Macromolecules 1999,

32, 4079.

(57) Schmitz, G. J. Thermodynamics of Diffuse Interfaces; Springer: Berlin, Heidelberg, 2003.

(58) Beaucage, G. J. Appl. Crystallogr. 1996, 29, 134.

(59) Rousseau, R.; Schreiner, E.; Kohlmeyer, A.; Marx, D. Biophys. J. 2004, 86, 1393.

(60) Fuller, J. M.; Raghupathi, K. R.; Ramireddy, R. R.; Subrahmanyam, A. V.; Yesilyurt, V.;

Thayumanavan, S. J. Am. Chem. Soc. 2013, 135, 8947.

(61) Munkhbat, O.; Garzoni, M.; Raghupathi, K. R.; Pavan, G. M.; Thayumanavan, S.

Langmuir 2016, 32, 2874.

(62) Wynn, J. E.; Santos, W. L. Org. Biomol. Chem. 2015, 13, 5848.

(63) Ilavsky, J.; Jemian, P. R. J. Appl. Crystallogr. 2009, 42, 347.

(64) Kline, S. J. Appl. Crystallogr. 2006, 39, 895.

3.7 Appendix:

Figure S3.1. MALDI-TOF MS spectra of purified ELP dendrimers: a) DG1-5 (Yield=27%), b)

DG1-10 (Yield=20%), c) DG2-5 (Yield=15%), d) DG2-10 (Yield=17%), e) DG2-15 (Yield=8%), f) DG3-5 (Yield=10%), g) DG3-10 (Yield=5%). (Reflector mode for a-f, linear mode for g)

Figure S3.2. Analytical LCMS (220 nm UV detector) traces of purified acetylated ELP dendrimers: a) DG1-5, b) DG1-10, c) DG2-5, d) DG2-10, e) DG2-15, f) DG3-5.

Figure S3.3. Gel electrophoresis plate of DG3-10 with 15 wt.% SDS page gel. (left: marker; right:

DG3-10) 80

Figure S3.4. CD spectra of A) DG1-5, B) DG1-10, C) DG2-5, D) DG2-10, E) DG2-15, F) DG3-

10 at various temperatures, demonstrating the conformational change from random coil to β–turn upon heating. All samples were measured at 0.5 mg mL−1 in 2.5 mM PBS.

Figure S3.5. UV-vis temperature scan profiles of A) DG2-10, B) DG2-15, C) DG3-10, peptides at 350nm (5 mg∙mL-1 at PBS and CSC)

Figure S3.6. Tt dependence on peptide concentration (248-15190 ϻM) for ELP dendrimer DG3-5 in 0.01 M PBS (pH = 7.4) with 2 M NaCl. The red line represents a fit to the equation suggested by Meyer and Chilkoti for the correlation between an ELP concentration and its Tt for linear ELPs.

Figure S3.7. SAXS curves of linear peptide KK(GLPGL)8KK. Solid yellow lines represent fits to the model described by Equation 1. All samples were measured at 1 wt.% ELP in 0.01 M PBS (pH

= 7.4) solution at 20 °C, 40 °C and 50 °C.

Chapter 4: Hydrogels composed of hyaluronic acid and dendritic ELPs:

Hierarchical structure and physical properties

“Reprinted (adapted) with permission from Shmidov, Y.; Zhou, M.; Yosefi, G.; Bitton, R.; Matson, J.

Authors

Yulia Shmidov, [a] Mingjun Zhou [b] Gal Yosefi, [a], Ronit Bitton * [a] John B. Matson * [b]

[a] Department of Chemical Engineering, Ben-Gurion University of the Negev, Beer-Sheva, 84105

(Israel)

[b] Department of Chemistry and Macromolecules Innovation Institute, Virginia Tech, Blacksburg,

VA 24061 (U.S.A.)

4.1 Abstract

Hydrogels that mimic the native extracellular matrix were prepared from hyaluronic acid (HA) and amine-terminated dendritic elastin-like peptides (denELPs) of generations 1, 2, and 3 (G1, 2, and 3) as crosslinking units. The physical properties of the hydrogels were investigated by rheology, scanning electron microscopy, swelling tests, small-angle X-ray scattering (SAXS), and model drug loading and release assays. Hydrogel properties depended on the generation number of the denELP, which contained structural segments based on the repeating GLPGL pentamer.

Hydrogels with higher generation denELPs (G2 and 3) showed similar properties, but those prepared from G1 denELPs were rheologically weaker, had a larger mesh size, absorbed less model drug, and released the drug more quickly. Interestingly, most of the HA_denELP hydrogels studied here remained transparent upon gelation, but after lyophilization and addition of water retained

84 opaque, “solid-like” regions for up to 4 d during rehydration. This rehydration process was carefully evaluated through time-course SAXS studies, and the phenomenon was attributed to the formation of pre-coacervates in the gel-forming step, which slowly swelled in water during rehydration. These findings provide important insights into the behavior of ELP-based hydrogels, in which physical crosslinking of the ELP domains can be controlled to tune mechanical properties, highlighting the potential of HA_denELP hydrogels as biomaterials.

4.2 Introduction

Dendrimers are a class of theoretically monodisperse branched macromolecules with a globular structure.1,2 Regardless of the method of synthesis, the number of terminal functionalities grows exponentially with generation number, and in many cases a core–shell structure can be generated by using terminal and dendritic units with differing hydrophobicity/hydrophilicity. The core-shell structure enables both encapsulation of hydrophobic molecules and conjugation of other moieties to the surface, leading to promising biological applications, such as drug delivery,3,4 biosensors,5 medical diagnostics,6 and antiangiogenic/gene therapy.7,8 Aside from their ability to serve as soluble nanostructures, another intriguing but less explored use of dendrimers is as crosslinkers in hydrogels, which have potential applications in tissue engineering/regenerative medicine,9 biofabrication,10 and extracellular matrix (ECM) mimics.11 Despite their potential uses in biology and medicine, dendrimer-based hydrogels are typically made from synthetic dendrimers that are not bio-derived or biodegradable, most notably polyamidoamine (PAMAM).12-14

Peptides are convenient bio-derived building blocks for fabricating dendrimers, with tunable sequences and various functions.15,16 One of the more studied peptide sequences is elastin-like peptides (ELPs). Characterized by a repeating XPGXG pentamer (X can be any amino acid except proline), ELPs are derivatives of elastin, which is an ECM protein that provides elasticity in tissues

85 and organs.17 With thermoresponsiveness, namely lower critical solution temperature (LCST) behaviour,18,19 linear ELPs have been studied with potential applications including targeted drug release and tissue engineering.17,20 In previous work, we synthesized and characterized the properties of dendrimers based on ELPs, evaluating the thermal behaviour of ELP dendrimers

(denELPs) with up to 3 generations and 155 amino acid residues.21,22 LCST behaviour, which is commonly observed in linear ELPs, remained with the dendrimer topology. The branching structure in the denELPs changed the transition temperature (Tt) compared to linear counterparts, likely due to steric effects and changes in solvent-accessible surface area. Interestingly, we also found that denELPs exhibited a continuous aggregation that preceded coacervation (macroscopic aggregate formation).22 Namely, denELP coacervates did not form in a stepwise manner, but rather by a slow increase in the size of ELP aggregates.

One potential use of ELPs is in hydrogels that provide support for cells in tissue engineering and regenerative medicine applications.23-26 For example, Heilshorn et al. reported hydrogels consisting of hyaluronic acid (HA), a major component of the ECM, and linear ELPs. These materials showed thermal-induced stiffening due to secondary physical crosslinking above the Tt’s of the ELPs and were investigated in applications including cartilage regeneration and stem cell delivery.27,28 To our knowledge all previous studies on ELP-based hydrogels involved linear ELP components; we hypothesized that denELPs might provide unique properties due to their geometry and pre-coacervation tendencies.

In early experiments using denELPs as crosslinkers in hydrogels based on HA, we noticed a surprising result—the hydrogel was transparent upon formation but did not change in appearance after heating. However, opaque regions emerged after lyophilization and rehydration, disappearing after several days. Following up on these unexpected observations, we explored a series of denELP

86 crosslinkers in HA hydrogels. With denELPs of different generation numbers and Tt’s, we investigated the hydrogels’ macroscopic, microscopic, and nanoscale structures, mechanical responses at various times, and model drug loading/release properties.

4.3 Experimental section

Materials

2-(N-Morpholino)ethanesulfonic acid (MES-buffer), N-hydroxysuccinimide (NHS), phosphate buffered saline (PBS) and N-(3-dimetylaminopropyl)-N’-ethylcarbodiimide hydrochloride (EDC) were purchased from Sigma-Aldrich. Sodium hyaluronate (HA) (MW=360 kDa) was purchased from Lifecore Biomedical, LLC.

Methods

Peptide and hydrogel synthesis

The synthesis of the denELPs on a Liberty1 microwave-assisted peptide synthesizer (CEM

Corporation) was detailed in our previous reports on denELPs.21,22 HA_denELP hydrogels were prepared according to a representative procedure as follows: A stock solution of MES buffer

(0.1 M, pH 4.7) was prepared. HA (10 mg) was dissolved in the stock MES buffer (1 mL) to give a 1% w/v aqueous solution. NHS (1.2 mg, 0.38 equiv. with respect to carboxylate groups) and

EDC (3.3 mg, 0.65 equiv. with respect to carboxylate groups) were then added, and the solution was transferred to one well in a 12-well plate. After 5 min, denELP (0.2 equiv. amine groups with respect to carboxylate groups) dissolved in water (200 µL) was added to the well. The solution was left at rt, with no stirring, for 3 h and then placed at 4 °C for 16 h. Next, the hydrogels were placed at -18 °C for 24 h and then lyophilized for 48 h. All hydrogels were synthesized in 12, 48, or 96-well plates according to the required geometry of the experimental technique, with total

87 volumes of 1.2 mL for hydrogels prepared in 12-well plates, 300 µL for hydrogels prepared in 48- well plates, and 150 µL for 96-well plates.

Scanning electron microscopy (SEM)

SEM images were taken using a high-resolution field-emission gun-SEM (JEOL, JSM-7400F) equipped with an energy-dispersive X-ray spectroscopy (EDS) instrument (Noran Vantage) and a back-scattered electron (BSE) detector (AutraDet, AUTORATA YAG) operated at 2.5 keV.

Hydrogel samples for SEM were obtained in 2 different manners: 1. Fast cooling of swollen hydrogels in liquid nitrogen and then lyophilization. 2. By critical point drying (CPD) as follows: specimens were washed with water for 10 min, repeating 5 times. Dehydration in the presence of ethanol was performed in rising concentrations of ethanol in water in the following order: 25% (a single repeat/10 min), 50% (a single repeat/10 min), 70% (2 repeats/10 min), 90% (3 repeats/10 min), 95% (3 repeats/10 min), and 100% (4 repeats/10 min). Ethanol exchange with CO2 was performed by a CPD instrument. Samples were then coated with a few nm thick carbon-coating

(Emitech K575X).

Swelling tests

Dry hydrogels were placed in 1 mL of PBS (0.01 M) and taken out for weighing at different times.

The swelling ratio was then calculated using eqn (1). The hydrogels were tested in triplicate, with at least two different syntheses for each hydrogel.

Rheology

The viscoelastic properties of the hydrogel samples were determined by oscillatory rheology experiments at different times. Measurements (n=4) were carried out using an Ares Rheometric

Scientific LS strain-controlled rheometer equipped with stainless steel parallel plates (d = 8 mm).

First, strain sweep experiments were performed in order to establish the linear viscoelastic region of the hydrogels, and a strain of 1% was chosen for frequency sweep measurements.

Small angle X-ray scattering (SAXS)

Small angle X-ray scattering patterns of the hydrogels were obtained with a SAXSLAB

GANESHA 300-XL. CuKα radiation was generated by a Genix 3D Cu-source with an integrated monochromator, 3-pinhole collimation and a two-dimensional Pilatus 300 K detector. Hydrogel specimens were placed in stainless steel sample cells with entrance and exit windows made of mica. The scattering intensity q was recorded at a range of 0.012 < q < 0.6 Å−1 (corresponding to lengths of 10–300 Å). Measurements were performed under vacuum at ambient temperature. The scattering curves were corrected for counting time and sample absorption.

Drug absorption and release tests

Aqueous solutions (150 µL) of rhodamine 6G (0.428 mM) in PBS (0.01 M, pH 7.4) were added to hydrogels (synthesized in a 48-well plate) directly after synthesis. The loaded hydrogels were then lyophilized. The hydrogels were immersed in 300 μL of fresh PBS solution (0.01 M, pH 7.4).

In order to quantify the amount of released dye, hydrogels were transferred to a fresh PBS solution at specified time points. The aliquots were analysed by UV-Vis spectroscopy (UltrospecTM2100 pro), measuring the absorbance at 500 nm. The dye concentration of the aliquots was determined using a calibration curve (at 500 nm). The release percentage displayed in the graph is normalized to the total dye released by HA_DG2-5 (taken as 100%). Experiments were conducted in triplicate.

For the absorption and release experiments, aqueous solutions (300 µL) of rhodamine 6G (0.428 mM) in PBS (0.01 M, pH 7.4) were added to lyophilized hydrogels with similar geometry and size

(all were synthesized in a 96-well plate) and left at rt for 1 week to allow for swelling and dye absorbance. Then, the hydrogels were immersed in an 80 μL

(0.01 M, pH 7.4) solution. The entire amount of the PBS solution (80 µL) was removed at various time points, and replaced with fresh PBS solution (80 μL). These aliquots were then analysed by

UV-Vis spectroscopy (UltrospecTM2100 pro), measuring the absorbance at 526 nm. The dye concentration of the aliquots was determined from a calibration curve (Figure S4.1, prepared by measuring the absorbance of known concentrations of rhodamine 6G at 526 nm). The release percentage displayed in the graph is normalized to the total absorbed concentration of dye for each hydrogel. All absorption and release experiments were conducted in quadruplicate.

4.4 Results and discussions

DenELP Design

Figure 4.1 Abbreviated chemical structures of denELPs, n=1, 2, 3.

Similar to our previous work,22 denELPs were constructed by microwave-assisted solid-phase peptide synthesis (SPPS), using Lys (lysine) residues as the branching units. The notation “DGx- y” designates a denELP of xth generation with y amino acid residues between the branching units

(Figure 4.1). For example, in DG3-10, 3 stands for a 3rd generation denELP, with two pentameric repeating units GLPGL (10 amino acids) between branching Lys residues. In this work denELPs up to G3 and with up to three pentameric repeat units (15 amino acids) per arm were used. The terminal Lys residues were maintained as amines for use in the crosslinking reaction. In the series of dendrimers presented here, DG1-y dendrimers, containing a branching Lys residue in the centre

90 of the peptide, have four amine end groups (the N-terminal amine and the side-chain amine from each terminal Lys residue), while each DG2-y dendrimer has eight and each DG3-y dendrimer has sixteen amine end groups.

Hydrogel Synthesis

In order to create an HA_denELP hydrogel, we crosslinked HA with denELPs using EDC and

NHS (Figure 4.2). Although EDC chemistry is well known, the synthetic route for HA_denELP hydrogel formation needed to be established (the details of the optimized procedure are described in the experimental section). In brief, a solution of HA in MES buffer (pH 4.7) was prepared; next,

NHS and EDC were added to the HA solution, followed by addition of a solution of denELP in water. In each formulation, a carboxylate/amine ratio of 5:1 was maintained by varying the amount of denELP according to the number of amine end groups. All hydrogels were allowed to cure for

3 h at rt, followed by another 16 h at 4 oC, and then lyophilized. HA_DGx-y stands for the hydrogel formed from HA and the corresponding dendrimer. There was no hydrogel formation upon physical mixing of HA and denELPs, and no self-crosslinking occurred when the synthesis was carried out with either HA or denELPs alone. HA_denELP hydrogels were prepared using 1 wt.%

HA and 0.2-0.5 wt.% denELPs to enable expression of the dendrimers in the hydrogel structure and its physical properties. HA_denELP synthesis was designed such that the concentration of carboxylate groups and amine groups (i.e., potential crosslinks) was the same for all hydrogels.

Therefore, any differences between the hydrogels’ properties should arise from the differences in the dendrimers themselves.

As mentioned above, denELPs are thermoresponsive with LCST behaviour, so we looked for the

22 presence of coacervates during synthesis. While all denELPs used here have Tt’s above rt, we expected that the combination of increased salt concentration in the hydrogel formation reaction

91 combined with consumption of the free amines would lower the Tt, potentially leading to the appearance of coacervates during gel formation. All hydrogels were transparent upon

22 formation except for HA_DG2-15, which has the lowest Tt among the tested dendrimers.

Figure 4.2 (A) Synthetic route for preparation of hydrogel HA_DG2-10 (orange rod represents pentameric unit GLPGL). A carboxylate/amine ratio of 5:1 was maintained in all hydrogels. (B)

Hydrogels HA_DG2-10 and HA_DG2-15 prior to lyophilization.

Figure 4.2B shows a photo of the transparent HA_DG2-10 hydrogel versus the visible coacervates in the HA_DG2-15 hydrogel. Interestingly, the opaque HA_DG2-15 hydrogel was not robust, in stark contrast to the optically clear HA_DG2-10 hydrogel. We attribute this to a reduction in the number of free amines available during crosslinking, as many were likely encapsulated within the insoluble DG2-15 coacervates. Performing the synthesis of HA_DG2-15 at 4 °C resulted in a clear hydrogel (Figure S4.2).

Hydrogel characterization

Bulk Properties

For ease of handling, all hydrogels were lyophilized after formation. During rehydration in PBS solution, we noticed significant visual differences between the HA_denELP hydrogels. Figure

4.3A shows the HA_DGx-5 series (the full HA_DGx-y series is presented in Table S4.1). While

HA_DG1-5 was transparent, HA_DG2-5 and HA_DG3-5 exhibited opaque regions that did not immediately disappear upon cooling (as would be expected for ELP coacervates in solution).

Instead, the opaque regions disappeared over the course of 1-4 d (Figure 4.3A bottom row), depending on temperature. Disappearance was faster at 4 °C than at 25 °C, indicating that the thermoresponsive (LCST) nature of the denELPs affected this process (Table S4.1).

Figure 4.3 (A) HA_DGX-5 hydrogel after 10 min (upper row) and 24 h (bottom row) in PBS solution at rt. (B) HA_DGX-10 hydrogel after 4 d in PBS solution at rt (upper row) and 4 °C

(bottom row).

We speculated that the presence of the opaque regions in HA_DG2-5 and HA_DG3-5 may be attributable to the pre-coacervation tendencies of denELPs. Pre-coacervates are small ELP aggregates that are not large enough to scatter light29 and likely exist during gel formation.

Removal of water during lyophilization may cause these small aggregates to cluster due to the collapse of HA chains. Upon rehydration after lyophilization, HA chains, which absorb water quickly,30 contribute to the hydrogel swelling. The ELP-rich regions, large enough to scatter light, have lower tendency to bind water due to their hydrophobic nature, and thus their rehydration

93 is considerably slower, resulting in “solid-like” opaque regions. HA_DG1-5 resulted in a clear hydrogel with no visible opaque regions, which would be expected since DG1-5 has the highest Tt among the tested dendrimers, and hence the lowest tendency to form ELP aggregates. In other words, there are no hydrophobic regions separating the HA chains in the HA_DG1-5 hydrogel, so the re-swelling process is homogenous. The faster disappearance of the opaque regions in

HA_DG2-5 and HA_DG3-5 at lower temperature (4 °C) may be explained by the secondary structure transition in these ELPs. At low temperatures the dominant secondary structure of ELPs is random coil, while increasing temperature causes a gradual shift to β-turns.18 After initial swelling, the secondary structure of the denELP likely shifts towards more hydrophilic random coils upon cooling, enabling faster re-swelling of the HA_denELP hydrogels at 4 °C than at 25 °C.

We also examined the HA_DGX-10 series (Figure 4.3B) in a similar manner. The same trend was observed: HA_DG1-10 resulted in a homogeneous hydrogel with no visible opaque regions, similar to HA_DG1-5; in contrast, HA_DG2-10 and HA_DG3-10 both showed opaque regions after lyophilization and rehydration. Incubation of the HA_denELP hydrogels in PBS solution for

4 d at 4 °C and 25 °C (Figure 4.3B), showed that rehydration was faster at lower temperature, indicating again that the thermoresponsive nature of the denELPs affected this process. As a control, we synthesized a hydrogel with a dendrimer composed of a non-thermoresponsive, hydrophilic pentameric repeating unit GGPGG (HA_DG2-10_Control, Figure S4.3). Indeed, the resulting hydrogel was transparent upon formation, and it remained transparent after lyophilization and rehydration for 10 min in PBS solution (Figure S4.4). Generally, despite the initial presence of opaque regions, the hydrogels retained higher transparency in comparison to pure ELP hydrogels. This is likely due to the combination of hydrophilic HA with the denELPs, decreasing the formation of hydrophobic aggregates within the hydrogel.27

The water content in hydrogels is an important factor that determines several parameters, such as overall permeation of nutrients into the hydrogel in applications as biomaterials. The water uptake of the HA_denELP hydrogels was evaluated by swelling tests of lyophilized hydrogels in 0.01 M

PBS solution (physiological salt concentration) at rt. HA_DG1-5 and HA_DG1-10 were too weak to handle (Figure 4.4A) and were not evaluated. Swelling of hydrogels is highly dependent on the hydrophilicity of the network and is limited by the crosslinks, leading to an elastic network retraction force.31 Thus, when immersed in an aqueous solution, hydrogels reach an equilibrium swelling level that depends mainly on the polymer type and the crosslinking density, where a less hydrophilic polymer and higher crosslinking density leads to less water absorption. The swelling ratio (Sr) is calculated according to Equation 1, where Ms and Md are the mass of the swollen and the dried hydrogel, respectively.32

푀푠−푀푑 푆푟 = Equation (4.1) 푀푑

The swelling of all the tested hydrogels was characterized by a rapid increase of Sr within the first

5 min followed by a very slow increase of Sr over the course of several days, supporting the two- phase rehydration process we suggested above. After 5 min incubation, the Sr values of HA_DG2-

5/10 and HA_DG3-5/10 were similar and ranged from 10 to 18 (Table S4.2). These Sr’s are higher than the Sr’s of pure chemically crosslinked linear ELP hydrogels and lower than Sr’s of HA hydrogels.27,33

Hydrogel Rheology

To evaluate the mechanical properties of the HA_denELP hydrogels during the rehydration process, oscillatory rheology measurements were performed. Hydrogels for both of the HA_DGx-

5 and HA_DGx-10 series were prepared, but as in the swelling tests, HA_DG1-5 and HA_DG1-

10 were too weak to use in further studies (Figure 4.4A).

Figure 4.4 (A) HA_DGx-y after 10 min in PBS solution. (B) Frequency sweep results of

HA_DG2-10 hydrogel showing G' from 2 h to 4 d of incubation in PBS solution. (C) Frequency sweep results of HA_DG3-10 showing G' from 2 h to 4 d of incubation in PBS solution.

The storage modulus, G’, obtained from frequency sweep results of HA_DG2-10 and HA_DG3-

10 at different rehydration times (from 2 h to 4 d) are shown in Figures 4.4B and 4.4C, respectively.

Because G’ represents the elastic behavior of a material, its value should correspond to the presence and size of the solid-like regions of the HA_denELP hydrogels. Both HA_DG2-10 and

HA_DG3-10 showed G' values of ~104 Pa shortly after the addition of PBS to the lyophilized powders, when opaque regions were still present. A decrease of G’ is directly correlated to the duration of rehydration. After full rehydration (transparent hydrogels), a considerable decrease in the G’ of the hydrogels was observed, down to <200 Pa.

We also incubated fully rehydrated (transparent) HA_denELPs for 4 d at 50 °C. This did not result in reappearance of the opaque regions, and the rheological measurements showed no significant change in G’. The absence of the reappearance of opaqueness suggests that the confinement of the denELPs within the hydrogels and the presence of hydrophilic HA prevented their coacervation.27

Furthermore, frequency sweep experiments on the HA_DG2-10_Control hydrogel, which was transparent immediately after rehydration, showed no decrease in G’ upon incubation at 4 °C for

4 d (Figure S4.5).

In addition to exploring the phenomena of the opaque regions, we sought to explore the influence of denELP generation number on the resulting physical and structural properties of the hydrogels, regardless of the thermoresponsiveness of the denELPs (i.e., after full rehydration of the hydrogels). DenELP generation number had no substantial effect on G’ in any of these experiments. As can be seen from Figures 4.4B and 4.4C, the G' values after full rehydration are similar for HA_DG2-10 and HA_DG3-10, as well as for HA_DG2-5 and HA_DG3-5 (Figure

S4.6). These results are in agreement with previously published work by Sheardown, in which G1 dendrimer-crosslinked collagen hydrogels had relatively poor mechanical properties compared with the G2 and G3 dendrimer-crosslinked hydrogels, but no significant difference was found in mechanical properties of G2 and G3 dendrimer-crosslinked collagen hydrogels.34

Hydrogel Microstructure

In order to investigate the microstructural features of the hydrogel networks, we used scanning electron microscopy (SEM) on hydrogels of the HA_DGX-10 series (Figure 4.5) and the

HA_DGX-5 series (Figure S4.7).35 As expected, SEM images verified the presence of a continuous network. The general structure of each HA_denELP is a hierarchical network composed of two meshes of different size scales. The coarser mesh had an average pore size of tens of µm for all hydrogels, which makes these materials suitable for cell encapsulation. A careful examination of the finer mesh showed that the HA_DG1-10 network was more porous and less dense compared to those of HA_DG2-10 and HA_DG3-10; a rough estimation revealed a mesh size of approximately tens of nm for HA_DG2-10 and HA_DG3-10 and up to hundreds of nm for

HA_DG1-10. In agreement with rheological and swelling properties of the hydrogels, the SEM images revealed higher crosslinking density for HA_DG2-10 and HA_DG3-10 than HA_DG1-10.

In thermosets and hydrogels, crosslinkers with a higher functionality generally lead to lower gel points (minimum conversion required to gel).36

Figure 4.5 SEM images of HA_denELP hydrogels at different magnification levels, revealing meshes of two different sizes: (A) and (D) HA_DG1-10; (B) and (E) HA_DG2-10; (C) and (F)

HA_DG3-10.

Hydrogel Nanostructure

In order to investigate the nanostructural features of the hydrogel networks, we used small angle

X-ray scattering (SAXS). To examine whether rehydration of the hydrogels affected their nanostructure alongside their microstructure, we selected three representative HA-denELP hydrogels: HA_DG1-10, which is transparent even at the initial stages of rehydration, HA_DG2-

10, which shows opaque regions that disappear upon rehydration over the course of 4 d, and

HA_DG2-15, which has coacervates upon initial gel formation (Figure 4.2). SAXS measurements of theses hydrogels were performed at different time points. The 1st was 2 h after rehydration in the PBS solution after lyophilization (designated as 2 h) and the 2nd was after 2 d of rehydration at

98 rt (designated as 2 d). The obtained scattering curves are presented in Figure 4.6. As expected, rehydration affected the nanostructure of these three hydrogels in different manners. The 2 h scattering curve of HA_DG2-10 exhibited a strong upturn at low q. This pronounced excess scattering at low q-values is typically attributed to local heterogeneities in the hydrogel network37 and was previously reported for cross-linked PNIPAM gels at elevated temperatures (40 °C).38

Figure 4.6 Scattering profiles of the HA_DGx-10 series at various time points after rehydration in

PBS solution at rt. (A) HA_DG1-10 (B) HA_DG2-10 and (C) HA_DG2-15.

The 2d scattering curve of HA_DG2-10 did not exhibit such a prominent upturn, indicating the disappearance of the heterogeneities. To further confirm this result, an additional SAXS measurement after 4 d was performed, which showed that the signal at low-q decreased further.

Because the decrease of the intensity in the low-q region and the disappearance of the opaque

99 regions in this hydrogel occurred simultaneously, it is reasonable to assume that the origin of the nanoscale heterogeneities in this hydrogel is the opaque regions.

The fact that the scattering curve of HA_DG1-10 remained the same after 2 d supports this assumption. For hydrogel HA_DG2-15, the scattering profile resembled the scattering profile obtained for denELPs in solution,22 indicating that true coacervates were formed during the hydrogel synthesis. The difference in the shape of the scattering profiles of HA_DG2-10 and

HA_DG2-15 is another indication that the opaque regions present after rehydration are not coacervates, but rather solid-like, ELP-rich regions.

Rhodamine 6G Encapsulation and Release

Given their slow morphological changes as they rehydrated, we envisioned that the HA_denELP hydrogels might be useful for controlling drug release. Thus, we encapsulated rhodamine 6G,39 a model drug, in the hydrogels and performed quantitative release tests in two different types of experiments. First, rhodamine 6G was added to the hydrogels directly after hydrogel synthesis but before lyophilization. The loaded hydrogels were then lyophilized, and release experiments were run on these hydrogels as they rehydrated. In the second set of experiments, lyophilized hydrogels were immersed in a solution of rhodamine 6G. Release experiments were then run after full rehydration of the hydrogels. These two types of release experiments allowed us to explore both the effects of hydrogel rehydration and dendrimer generation number on release profiles.

In the first set of experiments (release from pre-loaded hydrogels) cumulative release of rhodamine

6G from HA_DG2-5 (less opaque) and HA_DG2-10 (more opaque) was measured over 24 h

(Figure 4.7A). Both hydrogels showed an initial burst release, but after this initial period,

HA_DG2-5 released the model drug faster (70% within 1 h) than HA_DG2-10 (30% after 1 h).

After 6 h nearly all of the encapsulated dye (98%) had been released from HA_DG2-5, while

100

HA_DG2-10 had released only 70%. This difference in release rate was also revealed by examining the hydrogels after 24 h: HA_DG2-5 was colorless, while HA_DG2-10 remained a light pink color (Figure 4.7B).

Figure 4.7 (A) Cumulative release of rhodamine 6G from HA_DG2-5/10 over 24 h. (B) Images of hydrogels HA_DG2-5/10 after 24 h of release.

In the second set of experiments (release from lyophilized hydrogels after full rehydration), lyophilized hydrogels were placed in a rhodamine 6G solution for 7 d to ensure full rehydration and maximum uptake of the dye molecules by the hydrogel. This experimental setup allowed us to measure both dye absorption and release. The percentage of rhodamine 6G absorbed by

HA_DG1-10, HA_DG2-10, and HA_DG3-10 is presented in Figure 4.8A. Dye molecules present in water may or may not penetrate into the hydrogel depending on the interactions between dye molecules and polymer chains.40 HA_DG1-10 absorbed 70% of the loaded rhodamine 6G while

HA_DG2-10 and HA_DG3-10 absorbed 90% and near 100%, respectively. Rhodamine 6G is an organic molecule with positively charged amines at physiological pH (Figure S4.1A). As the experiment was performed in PBS buffer (0.01 M, pH=7.4), the positive charge of rhodamine 6G as well as the internal negative charge of the hydrogels (from unreacted carboxylate groups on

HA) are screened. We attribute the increased loading of rhodamine 6G in higher generation

101 number HA-denELPs to the higher number of ELP pentamers in DG2-10 and DG3-10 compared to DG1-10, leading to more van der Waals interactions between the dye and hydrogel.

Next, the ability of each hydrogel to release rhodamine 6G was evaluated. Figure 4.8B shows the relative amount of dye released from each hydrogel over the course of 24 h. The initial dye release

(in the first hour), which usually corresponds to dye released from the surface and the outer layer of the hydrogel, was twice as much for HA_DG1-10 as for HA_DG2-10 and four times higher than for HA_DG3-10. Over time, the release rate decreased for all three hydrogels. HA_DG1-10 released 60% of the absorbed dye after only 6 h and then plateaued. In contrast, HA_DG2-10 required 24 h to reach 60% release, while HA_DG3-10 released only 40% of the absorbed dye in this period.

Figure 4.8 (A) Absorption of rhodamine 6G by the HA_DGX-10 hydrogels after exposure of the hydrogels to a 0.428 mM solution of the dye. (B) Release of absorbed rhodamine 6G from the

HA_DGX-10 hydrogels over time relative to the absorbed amount. (C) Images of HA_DGX-10- rhodamine 6G hydrogels 24 h after initiating release studies. Asterisks indicate wells containing

102 the hydrogels. (D) Quantitative estimation of the amount of remaining dye in the hydrogels

(100%= 0.428 mM).

The final states of the hydrogels after conducting the release tests for 24 h are shown in Figure

4.8C. It should be noted that each hydrogel initially held a different amount of dye, and data were scaled to the percent release of the total encapsulated amount of dye. In other words, the amount of dye both encapsulated and released by HA_DG1-10 was lower than that of HA_DG2-10 and

HA_DG3-10; The images revealed that HA_DG3-10 had more dye remaining within the hydrogel than HA_DG2-10, while HA_DG1-10 was colorless. Quantitative estimation of the amount of remaining dye in the hydrogels (Fig 4.8D) can be calculated by eqn (2), where ABS is the percentage of absorbed dye (Fig 4.8A) and R is the fraction of the released dye normalized to the absorbed amount (Fig 4.8B).

% 푓푖푛푎푙 푑푦푒 푐표푛푡푒푛푡 = 퐴퐵푆(1 − 푅) Equation (4.2)

All of these results indicate stronger interactions (likely Van der Waals interactions) between rhodamine 6G and HA_DG3-10 compared to HA_DG2-10 and HA_DG1-10, which is consistent with the dye loading results.

Taken together, the results above indicate that two factors determine the physical properties of the

HA_denELP hydrogels: 1) the thermoresponsive nature of the ELPs; and 2) the dendritic structure of the ELP crosslinkers. The LCST behavior of the denELPs renders them with a dynamic nature, leading to the existence and disappearance of opaque, solid-like regions after lyophilization and rehydration of these hydrogels. The slow disappearance of the solid-like regions leads to gradual changes in the hierarchical structure, drug release properties, and mechanical properties of the

HA_denELP hydrogels. It is noteworthy that over the course of 4 d, the stiffness of the hydrogels

103 decreased by two orders of magnitude without applying any external stimuli to the system and without network degradation. The dendritic structure of the denELPs dictates the properties of the hydrogels in their “final” fully rehydrated state. In all of our experiments, we noticed a clear difference between the hydrogels crosslinked with G1 dendrimers and those crosslinked with G2 and G3 dendrimers, but there were no major differences between HA_DG2-y and HA_DG3-y hydrogels. In dendritic systems when functional groups or units behave differently alone versus in a dendrimer, this phenomenon is often called the multivalency effect.12,16,41,42 In the context of the

HA_denELP hydrogels described here, the difference between the cross-linking density of HA-

DG1-y hydrogels and HA_DG2/3-y hydrogels can be attributed to the same effect.

4.5 Conclusions

We prepared hydrogels from small amounts of HA (1 wt.%) and denELP crosslinkers (<0.5 wt.%) with varying generation number and arm length of the denELPs, and we investigated their hierarchical structure (macro-, micro- and nanoscale), mechanical response, and model drug loading/release properties. Upon lyophilization of the hydrogels and addition of water, we observed the formation of opaque, solid-like regions, which disappeared over several days of rehydration. This was concurrent with a gradual decrease in hydrogel stiffness due to the thermoresponsive character of the denELP crosslinker. Characterization of the hydrogels after full rehydration revealed significant differences in the mesh density of hydrogels constructed from G1 dendrimers in comparison to G2 and G3 dendrimers, demonstrating the importance of the multivalency effect in determining hydrogel properties.

These hydrogels are attractive candidates for synthetic ECM materials due to their hierarchical mesh structure, including porosity on both the nano- and microscales, as well as their transparency, which may facilitate light microscopy imaging throughout the structure. Additionally, the

104 multivalent arms of the dendrimers provide sites for attachment of biological signals or other markers. Finally, the potential to tune their physical properties demonstrates the role that dendritic peptides can play in making new biomaterials.

4.6 References

(1) Astruc, D.; Boisselier, E.; Ornelas, C. Chem. Rev. 2010, 110, 1857.

(2) Hawker, C. J.; Frechet, J. M. J. J. Am. Chem. Soc. 1990, 112, 7638.

(3) Lee, C. C.; Gillies, E. R.; Fox, M. E.; Guillaudeu, S. J.; Fréchet, J. M. J.; Dy, E. E.; Szoka,

F. C. P. Natl. Acad. Sci. USA 2006, 103, 16649.

(4) Wolinsky, J. B.; Grinstaff, M. W. Adv. Drug. Deliver. Rev. 2008, 60, 1037.

(5) Harpham, M. R.; Süzer, Ö.; Ma, C.-Q.; Bäuerle, P.; Goodson, T. J. Am. Chem. Soc. 2009,

131, 973.

(6) Rozkiewicz, D. I.; Brugman, W.; Kerkhoven, R. M.; Ravoo, B. J.; Reinhoudt, D. N. J.

Am. Chem. Soc. 2007, 129, 11593.

(7) Bermejo, J. F.; Ortega, P.; Chonco, L.; Eritja, R.; Samaniego, R.; Müllner, M.; de Jesus,

E.; de la Mata, F. J.; Flores, J. C.; Gomez, R.; Muñoz-Fernandez, A. Chem.-Eur. J. 2007, 13, 483.

(8) Hong, S.; Rattan, R.; Majoros, I. J.; Mullen, D. G.; Peters, J. L.; Shi, X.; Bielinska, A. U.;

Blanco, L.; Orr, B. G.; Baker, J. R.; Holl, M. M. B. Bioconjugate Chem. 2009, 20, 1503.

(9) Magalhães, T. M.; Guerra, R. C.; San Gil, R. A. d. S.; Valente, A. P.; Simão, R. A.; Soares,

B. G.; Mendes, T. d. C.; Pyrrho, A. d. S.; Sousa, V. P. d.; Rodrigues-Furtado, V. L. J. Nanopart.

Re. 2017, 19, 277.

(10) Bi, X.; Amie Luckanagul, J.; Allen, A.; Ramaboli, M.; Campbell, E.; West, D.;

Maturavongsadit, P.; Brummett, K.; Wang, Q. J. Biomat. Sci-Polym. E. 2015, 26, 669.

(11) Wang, Y.; Zhao, Q.; Zhang, H.; Yang, S.; Jia, X. Adv. Mater. 2014, 26, 4163.

105

(12) Ghobril, C.; Rodriguez, E. K.; Nazarian, A.; Grinstaff, M. W. Biomacromolecules 2016,

17, 1235.

(13) Holden, C. A.; Tyagi, P.; Thakur, A.; Kadam, R.; Jadhav, G.; Kompella, U. B.; Yang, H.

Nanomed.-Nanotechnol. 2012, 8, 776.

(14) Wang, J.; He, H.; Cooper, R. C.; Yang, H. ACS Appl. Mater. Inter. 2017, 9, 10494.

(15) Ghobril, C.; Charoen, K.; Rodriguez, E. K.; Nazarian, A.; Grinstaff, M. W. Angew. Chem.

Int. Edit. 2013, 52, 14070.

(16) Wathier, M.; Jung, P. J.; Carnahan, M. A.; Kim, T.; Grinstaff, M. W. J. Am. Chem. Soc.

2004, 126, 12744.

(17) Navon, Y.; Bitton, R. Isr. J. Chem. 2016, 56, 581.

(18) Nuhn, H.; Klok, H.-A. Biomacromolecules 2008, 9, 2755.

(19) Quiroz, F. G.; Chilkoti, A. Nat. Mater. 2015, 14, 1164.

(20) Lowik, D. W. P. M.; Leunissen, E. H. P.; van den Heuvel, M.; Hansen, M. B.; van Hest,

J. C. M. Chem. Soc. Rev. 2010, 39, 3394.

(21) Navon, Y.; Zhou, M.; Matson, J. B.; Bitton, R. Biomacromolecules 2016, 17, 262.

(22) Zhou, M.; Shmidov, Y.; Matson, J. B.; Bitton, R. Colloid. Surface B 2017, 153, 141.

(23) Amruthwar, S. S.; Janorkar, A. V. J. Mater. Sci. Mater. Med. 2012, 23, 2903.

(24) Annabi, N.; Mithieux, S. M.; Boughton, E. A.; Ruys, A. J.; Weiss, A. S.; Dehghani, F.

Biomaterials 2009, 30, 4550.

(25) Mahara, A.; Kiick, K. L.; Yamaoka, T. J. Biomed. Mater. Res. A 2017, 105, 1746.

(26) Trabbic-Carlson, K.; Setton, L. A.; Chilkoti, A. Biomacromolecules 2003, 4, 572.

(27) Wang, H.; Cai, L.; Paul, A.; Enejder, A.; Heilshorn, S. C. Biomacromolecules 2014, 15,

3421.

106

(28) Zhu, D.; Wang, H.; Trinh, P.; Heilshorn, S. C.; Yang, F. Biomaterials 2017, 127, 132.

(29) Fuller, J. M.; Raghupathi, K. R.; Ramireddy, R. R.; Subrahmanyam, A. V.; Yesilyurt, V.;

Thayumanavan, S. J. Am. Chem. Soc. 2013, 135, 8947.

(30) Kaczmarek, B.; Sionkowska, A.; Kozlowska, J.; Osyczka, A. M. Int. J. Biol. Macromol.

2018, 107, 247.

(31) Hoffman, A. S. Adv. Drug. Deliver. Rev. 2012, 64, 18.

(32) Martín, L.; Alonso, M.; Girotti, A.; Arias, F. J.; Rodríguez-Cabello, J. C.

Biomacromolecules 2009, 10, 3015.

(33) Lampe, K. J.; Antaris, A. L.; Heilshorn, S. C. Acta Biomaterialia 2013, 9, 5590.

(34) Duan, X.; Sheardown, H. J. Biomed. Mater. Res. A 2005, 75A, 510.

(35) Altin, H.; Kosif, I.; Sanyal, R. Macromolecules 2010, 43, 3801.

(36) Flory, P. J. J. Am. Chem. Soc. 1941, 63, 3083.

(37) Bekhradnia, S.; Zhu, K.; Knudsen, K. D.; Sande, S. A.; Nyström, B. J. Mater. Sci. 2014,

49, 6102.

(38) László, K.; Guillermo, A.; Fluerasu, A.; Moussaïd, A.; Geissler, E. Langmuir 2010, 26,

4415.

(39) Vanamudan, A.; Bandwala, K.; Pamidimukkala, P. Int. J. Biol. Macromol. 2014, 69, 506.

(40) Paulino, A. T.; Guilherme, M. R.; Reis, A. V.; Campese, G. M.; Muniz, E. C.; Nozaki, J.

J. Colloid Interf. Sci. 2006, 301, 55.

(41) Caminade, A.-M.; Ouali, A.; Laurent, R.; Turrin, C.-O.; Majoral, J.-P. Chem. Soc. Rev.

2015, 44, 3890.

(42) Tomalia, D. A. New. J. Chem. 2012, 36, 264.

107

4.7 Appendix

Peptide Synthesis

Peptides DG1-5, 2-5, 3-5, 1-10, 2-10, 3-10 and 2-15 were reported in previous work, with MS, analytical HPLC and gel electrophoresis data.1

OD= a+bC a= 0.06 ±0.01 b= 2.06±0.04 Fit Type: least square fit

Figure S4.1. A. Chemical structure of Rhodamine 6G. B. Absorbance spectrum of rhodamine 6G in 0.01M PBS. λmax=526 nm. C. Calibration plot of rhodamine 6G.

Table S4.1. Summary of the den_ELPs used in this paper and their transition behavior at 5 mg/mL in 10 mM PBS buffer at pH=7.4. (a MW indicates exact mass. b Total NaCl concentration [M]

108 required to achieve turbidity at 37 °C. c For DG1-5 no turbidity was observed in the examined salt range.)1

Figure S4.2. Synthesis of HA_DG2-15 at A. 25 °C and B. 4 °C.

Table S4.2. HA_DGx-y at 4 °C and 25 °C after 10 min and 1,4 d of incubation in PBS solution.

109

Figure S4.3. The MALDI-TOF spectrum of peptide DG2-10 control with non-thermoresponsive

GGPGG repeating sequence.1

Figure S4.4. HA_DG2-10 control after 10 min incubation at PBS solution 25 °C (no opaque regions).

Hydrogel Sr

HA_DG2-5 18.7±0.7

HA_DG3-5 18±2

HA_DG2-10 12.8±1

HA_DG3-10 10.9±0.5

110

Table S4.3. Swelling ratio of the hydrogels after 5 min of immersion at rt in PBS solution (0.01M, pH=7.4).

Figure S4.5. Frequency sweep experiment curve of HA_DG2-10 control after 2h incubation in

PBS solution (0.01M) at 23 °C and after 4d incubation in PBS solution (0.01M) at 4 °C.

Figure S4.6. Frequency sweep experiment curve of HA_DG2-5, 2-10, 3-5, and 3-10 after full rehydration.

111

Figure S4.7. SEM image of A. HA_DG1-5 B. HA_DG2-5 C. HA_DG3-5 hydrogel.

References:

1. Zhou, M.; Shmidov, Y.; Matson, J. B.; Bitton, R. Colloid. Surface B, 2017, 153, 141.

112

Chapter 5: Thermoresponsive wedge-shaped triblock peptide-polymer-C16

(palmitic acid) conjugate amphiphiles: phase and drug delivery behavior

Mingjun Zhou, Yulia Shmidov, Yin Wang, Ronit Bitton and John Matson*

Authors

Mingjun Zhou,[a] Yulia Shmidov,[b] Yin Wang, [a] Ronit Bitton[b] and John B. Matson*[a]

[a] Department of Chemistry and Macromolecules Innovation Institute, Virginia Tech, Blacksburg,

VA 24061 (U.S.A.)

[b] Department of Chemical Engineering, Ben-Gurion University of the Negev, Beer-Sheva, 84105

(Israel)

5.1 Abstract

With a combination of solid phase and solution phase peptide synthesis, we designed and synthesized a series of wedge-shaped triblock peptide-polymer-C16 (palmitic acid) conjugate amphiphiles. With a variety of techniques, we investigated their hierarchical structures and phase behavior. The amphiphiles were thermoresponsive and showed a morphology change from small micelles to large aggregates as hypothesized. The hydrophilic shell and strong tendency for micelle formation limited the thermoresponsive assembly, leading to slow turbidity change in the LCST transition. The secondary structure was twisted from conventional β-sheet, and the thermoresponsive trend observed in typical ELP systems was not observed, either. Variable temperature NMR showed evidence for coherent dehydration of the PEG and ELP segments, probably due to the relatively short blocks. Utilizing the micelles with hydrophobic cavity, we encapsulated hydrophobic drugs with potential applications in hyperthermia.

113

5.2 Introduction

Peptide-polymer conjugates have been widely used to combine the advantages of both components, as well as to overcome their disadvantages, with various synthetic strategies.1-3

Hybrid materials by conjugating polymers, especially poly(ethylene glycol) (PEG) to peptides has shown improvement over original peptides, in terms of stability, solubility, circulation half-life, and immunogenicity.4 Besides their potential for therapeutic applications, it’s of fundamental significance to explore the insights of their self-assembly behavior, e.g. hierarchical/novel structures, and the corresponding driving force.3 Topics of interest include the interaction between the two building blocks, the structure change upon conjugation, energetic contribution of phase behavior, as well as the micro-phase separation and amphiphilicity.3,5-8

Peptide amphiphiles have attracted lots of research interest in recent years. They are generally composed of three segments: a hydrophobic alkyl chain at one end, a charged hydrophilic peptide block at the other end and a beta-sheet forming peptide block in the middle. They form high aspect ratio 1-D nanofibers with micro-phase separation while the middle block stabilizes the aggregation structure with hydrogen bonding.9-11 Peptide amphiphiles have been used in a variety of applications including hemorrhage control,12 neurite formation,13 nerve repair,14 brain progenitor cell migration.15 However, most of prior research focused on linear amphiphiles. Little is known about the correlation between molecular topology and assembly behavior.

Branched molecular structure has been proved to stabilize aggregation structures with surface curvature such as spheres and columns. Compared to its linear counterpart, branched topology can better fill in the external space of the core-shell structures to minimize free volume by attaching multiple peripheral chains, while linear molecules have a strong tendency to form lamellar assemblies.16-18 We have previously reported the synthesis and thermoresponsiveness of

114 branched/dendritic elastin like peptide (ELP),19-21 which is an artificial biopolymer derived from the hydrophobic domain of tropoelastin.22

We aimed to design and investigate a triblock PEG/ELP/C16 conjugate amphiphile system, to obtain insights in the topology-assembly relationship. We hypothesized that in aqueous solution, this amphiphilic structure would self-assemble to micelles, with PEG block as the hydrophilic shell and C16 as the hydrophobic core. ELP dendrimer was assumed to strengthen the micelle by intermolecular hydrogen bonding and volume effect. Furthermore, we were interested in their thermoresponsiveness on heating, including secondary structure change, interaction between the

PEG and ELP blocks, and the morphology change. Due to the hydrophobic cavity inside the micelle, we speculated this system could be used to encapsulate hydrophobic drugs. Eventually we were interested in whether the morphology change would lead to the release of the encapsulated drug, as a potential drug delivery capsule for hyperthermia.23-25

5.3 Experimental section

Materials. Rink amide MBHA resin (0.754 mmol/g) and Fmoc-protected amino acids were purchased from P3 Biosystems and used as received. PEG mono methyl ether (CH3O-PEG-OH,

MW=750) was purchased from BeanTown Chemical. All other reagents were purchased from commercial vendors and used as received without further purification unless noted.

Synthesis of PEG carboxylate methyl ether

PEG-COOH was synthesized by oxidizing PEG-OH with Jones’ reagent. Jones reagent was prepared by mixing water (35 mL), concentrated sulfuric acid (98wt.%, 5 mL) and CrO3 (6.64 g).

For the oxidation reaction, PEG-OH (10 g, 13.3 mmol) was dissolved in acetone (50 mL). Then the polymer solution was added to the Jones reagent (32.1 mL, in ice bath) dropwise to avoid forming dimer. The mixture was stirred for another half an hour, after which the organic layer was

115 blown away with air. Remaining mixture was extracted with DCM three times, followed by drying the organic layer with anhydrous MgSO4. After rotary evaporation, the product was obtained as colorless viscous liquid. The reaction was quantitative and confirmed by thin layer chromatography.

Peptide synthesis

The coupling of palmitic acid was performed by dissolving palmitic acid (513 mg, 2 mmol), hexafluorophosphate benzotriazole tetramethyl uronium (HBTU) (683 mg, 1.8 mmol) and N,N-

Diisopropylethylamine (DIEA) (0.698 mL, 4 mmol) in DMF/DCM/NMP (7/7/7 mL), prior to reacting with resin (0.5 mmol, amine count) under shaking for 2 hours. 4-methyl trityl (Mtt) deprotection was performed by reacting with 10 mL of cocktail (4 v% TFA 5 v% TIPS and 91 v%

DCM) for 5 min and three times. After washing with DCM and DMF, the resin was neutralized with 1 mL of DIEA and 15 mL of DMF for 10 min under shaking. All peptides were prepared with solid phase peptide synthesis (SPPS) using a liberty 1 peptide synthesizer. After cleavage (95 v%

TFA, 2.5v% TIPS and 2.5 v% water, 2 h), the peptides were purified by preparative scale reverse phase high-performance liquid chromatography (RP-HPLC) on an Agilent Technologies 1260

Infinity HPLC system, eluting with a gradient of 2% to 98% ACN in Milli-Q H2O on an Agilent

PLRP-S column (100 Å particle size, 25 × 150 mm) and monitoring at 220 nm. Prior to lyophilization, fractions were analyzed with matrix-assisted laser desorption ionization tandem time of flight mass spectrometer (4800 MALDI TOF/TOF; AB Sciex, with α-Cyano-4- hydroxycinnamic acid (CHCA) as matrix). The PEG coupling was performed by dissolving PEG-

COOH (138 mg, 185 μmol), Hexafluorophosphate Azabenzotriazole Tetramethyl Uronium

(HATU) (31.6 mg, 83 μmol), peptide-C16 conjugate (15.4 μmol, amine count) and DIEA (185 μL,

370 μmol) in DMF (4 mL), followed by coupling under microwave (CEM Liberty 1 system, 25W,

116

75 oC) for 10 min. The DMF was then blown away with air. The reaction mixture was then purified with HPLC, followed by MALDI-TOF (with CHCA as matrix) analysis and lyophilization, similar to peptide purification.

Circular Dichroism (CD)

The CD spectra of peptide polymer conjugate from 190 nm to 250 nm (20 nm/min, data pitch 1 nm) were recorded with a Jasco J-815 spectropolarimeter (JASCO, Easton, MD, USA) using a 1 cm path length quartz UV-Vis absorption cell (Thermo Fisher Scientific, Pittsburgh, PA, USA).

The peptides were dissolved in Milli-Q water at a concentration of 0.25 mg/mL. Spectra were recorded at various temperatures with a heating speed of 3 oC/min after stabilization for 10 min.

The result was the average of triplicate runs.

Conventional transmission electron microscopy (TEM)

Samples (PEG-peptide-C16 conjugate and C16 peptides) were dissolved in 10 mM PBS buffer

(pH=7.40). Samples were allowed to age overnight. Then 10 μL of this conjugate solution was deposited onto a carbon-coated copper grid (Electron Microscopy Services, Hatfield, PA, USA) and allowed to stand for 5 min. Excessive solution was wicked away by filter paper. Then Milli-

Q water was deposited for 40 s and wicked away to remove excess salts. Finally, 10 μL of a 2 wt.

% aqueous uranyl acetate (UA) solution was deposited onto the grid for 5 min. A thin layer was formed after carefully wicking away excess UA. The sample grid was then dried at rt prior to imaging. Bright-field TEM imaging was performed on a Philips EM420 Transmission Electron

Microscope operated at an acceleration voltage of 100 kV. TEM images were recorded by a slow scan CCD camera.

Fluorescence spectroscopy

117

Fluorescence spectroscopy was performed on a Varian Cary Eclipse Fluorescence

Spectrophotometer with a scanning speed of 400 nm·min-1, data pitch of 1 nm, an integration time of 0.1s and slits of 10 nm in a 1 cm quartz cuvette.

UV-Vis spectroscopy

UV-Vis spectroscopy was recorded from 800 nm to 300 nm on a Varian Cary 100 UV-Vis spectrophotometer with a scanning speed of 400 nm·min-1 in a 1 cm quartz cuvette. Spectra were recorded at various temperatures with a heating speed of 3 oC/min and stabilized for 10 min prior to measurement.

Dynamic light scattering (DLS)

Dynamic light scattering was conducted using a Malvern Zetasizer Nano at various temperatures with a heating speed of 3 oC/min and incubated for 10 min prior to measurement. Peptide-polymer conjugate was prepared at 5 mg/mL. The calculations of the particle size distributions and distribution averages were conducted using CONTIN particle size distribution analysis routines with intensity averages. Measurements were made in triplicate and errors reflect standard deviations.

Variable temperature NMR

Variable temperature 1H NMR spectra were acquired using a 400 MHz Bruker Advance III WB

NMR spectrometer, equipped with a MIC probe and 1H 5 mm coil. For elevated temperatures, the gas flow was 400 L/h and maximum heating power was set to 10%. Before each measurement, the sample was equilibrated for more than 10 min before running experiments at the set temperature to ensure thermal equilibrium. Sufficient signal-to-noise (S/N) ratio of polymer was obtained with

32 number of scans and acquisition time of 1.5 s with 2 Hz line broadening. Relaxation delay time of 3.0 s was used.

118

Drug encapsulation

Valsartan was dissolved in ethanol to make a 1mg/mL stock solution, then it was diluted with

Milli-Q water to afford a 0.1 mg/mL solution with ethanol/water 1/9 by volume. 5 mL of such solution was used to dissolve 4 mg of PEG-F4 conjugate, and it was still cloudy. Then it was sonicated for half an hour.

The solution was then added to the dialysis tube with MWCO = 3.5 kDa, and dialyzed against

Milli-Q water 3 times over 2 d. After the 3rd dialysis, completion was confirmed by testing the water outside the tube by fluorescence spectroscopy. Then the conjugate with loading drug was tested with fluorescence spectroscopy as well (emission measurement with excitation at 300 nm and emission at 498 nm).

5.4 Results and discussion

Design and Synthesis

As shown in Scheme 5.1, the triblock PEG-peptide-C16 amphiphiles were synthesized with a combination of SPPS and solution coupling. As detailed in the experimental section, the palmitic acid and amino acids were coupled onto the resin with SPPS, and then cleaved and purified. PEG-

COOH was coupled with the peptide-C16 in solution. We firstly tried directly coupling PEG-

COOH onto the resin, since PEG is not sensitive to the cleavage process. However, the coupling conversion was low and the product couldn’t be purified, due to the increased steric hindrance on the solid substrate compared to reaction in solution. The conversion from solution coupling was much higher. As shown in Figure 5.1, after the 2nd HPLC, we were able to obtain pure product with neglectable impurity from incomplete PEG coupling. We also found the MW of the PEG had a significant influence on the coupling conversion. For example, PEG-COOH with MW of 1500 couldn’t be coupled on all of the four arms of the peptide dendrimer, while small molecules such

119 as benzoic acid could be easily coupled onto 8 arms of the peptide dendrimer. This MW effect is likely due to the steric hindrance during the coupling.

Scheme 5.1 Synthetic route of the triblock PEG-peptide-C16 conjugate amphiphile. Orange rod stands for GXPGY pentamer unit, X and Y are Leu (L), Phe (F) or Gly (G).

As shown in Figure 5.1, 12 amino acid residues at four positions in orange were changed with F and L. The molecule in Figure 5.1 was named as PEG-F4, because all four positions were F.

Besides that we synthesized PEG-L4, L3F, L2F2 and LF3 to tune the hydrophobicity of the peptide dendrimer,26 and expected to achieve different LCST of the entire conjugate.26 A non- thermoresponsive conjugate was synthesized with GGPGG pentamer unit (PEG-G4) as control.

Peptide-C16 diblock amphiphiles were named as Peptide-L4, L3F, L2F2, LF3, F4 and G4, corresponding to different peptide sequences. The MALDI-TOF spectra of all of the molecules are listed in Figure S5.1.

The molecular difference between the expected molecular and peak molecular weight values was due to HPLC purification. Due to the MW distribution of PEG, the product coupled with PEG of higher MW was more hydrophilic, resulting in shorter retention time during HPLC. This was

120 confirmed by MALDI-TOF showing that earlier fractions had higher molecular weight. During

HPLC, the fraction collection was threshold-based (absorbance at 220 nm), consequently, not all of the product was collected and used for further characterization. The MALDI-TOF spectra shown in Figure 5.1 and S5.1 were obtained from combining all of the fractions for each sample.

Turbidity study and UV-Vis

We hypothesized the PEG-peptide-C16 conjugate would show a trend in LCST with different composition of Leu and Phe, since F is more hydrophobic.26 We firstly tried using visual observance to determine the LCST. The solutions (5 mg/mL in 10 mM PBS at pH = 7.40) did turn turbid over heating in water batch. However, unlike peptide dendrimers’ evident turbidity change,19,20 the change in the current system was very subtle, making it difficult to pinpoint LCST.

As shown in Figure 5.2, a similar phenomenon was shown in UV-Vis measurement that the turbidity increase was slow and gradual over a broad temperature range, likely due to the formation of stable micelle structure. The LCST by UV-Vis was determined as the temperature range when the turbidity started to increase evidently. Besides, although the PEG-F4 with the most hydrophobic peptide segment showed the lowest LCST, the system did not show a clear relationship between peptide hydrophobicity and LCST, as in pure ELP systems.19 Besides the difficulty in the LCST determination, this likely originated from the molecular weight distribution of the PEG block. For example, PEG-L4 showed an abnormally lower LCST than PEG-L2F2, considering the latter one contains a more hydrophobic peptide block. As calculated from Figure

S5.1 (g and i), the average molecular weight of PEG attached to PEG-L4 was 701, comparing to

833 on PEG-L2F2. The decreased hydrophobicity from longer PEG segment compensated the more hydrophobic peptide block with more Phe.

121

Figure 5.1 Molecular structure of the triblock PEG-peptide-C16 conjugate amphiphile (PEG-F4), and its MALDI-TOF spectrum.

Figure 5.2 Variable temperature UV-Vis measurement of (a) PEG-LF3, (b) PEG-F4, (c) control

PEG-G4 and (d) summary of the LCST determined from UV-Vis experiment. (The lamp crossover was at 300 nm for a/b, and 350 nm for c.)

122

Morphology study with DLS and TEM

Based on the turbidity study, we chose the two samples (PEG-L4 and F4) with the lowest LCST for the following investigations. We used DLS to analyze the LCST and size change during the phase transition. As shown in Figure 5.3, we used the variable temperature volume distribution of

PEG-L4 and F4, considering the significant size increase during the LCST transition. Generally, both of them showed small aggregates in the size of tens of nm below the LCST, and large particles

(~1000 nm) above it. At ~ 20 oC over the initial UV-Vis absorbance increase, DLS showed large aggregates for both samples.

Figure 5.3 Variable temperature DLS result of (a) PEG-L4 and (b) PEG-F4.

We used conventional TEM to explore the morphologies of PEG-L4 and F4 below and above

LCST, by changing the sample/salt concentration, as shown in Figure 5.4. PEG-F4 showed small spherical/rod-like assembly below the LCST (Figure 5.4b), and large spherical aggregates (Figure

5.4c), longer fiber (Figure 5.4d) above the LCST. PEG-L4 formed spherical/elliptical morphologies below the LCST (Figure 5.4a), but we were not able to achieve clear morphologies by increasing the salt/sample concentrations.

In general, below the LCST, the diameter/short-axis length of the assembly for both of the samples was ~ 20 nm. Calculated from Chem3D, the fully extended molecular length was 11.2 nm, which indicates the structures shown in Figure 5.4a/b are monolayer micelles. Above the LCST, PEG-F4

123 formed much larger aggregates including spherical multi-layer particles in the size of 100-200 nm and fibers of hundreds of nm long, which account for the increased absorbance in the UV-Vis experiment and the large aggregates in DLS. These results indicate the PEG-peptide-C16 amphiphiles self-assemble to micelles and form large aggregates above LCST as designed.

ELPs form large aggregates above the LCST as observed previously,22 however, the PEG block made the system more complex and interesting. PEG is also thermoresponsive, but the LCST is typically higher than 100 oC, especially at low PEG MW.27 Therefore the LCST of PEG is expected to be higher than that of the ELP dendrimer block.19 But we were not sure whether it would stay hydrated above the LCST. Theoretically, the properties of block copolymers could be decoupled for each block.28 PEG-PNIPAM block copolymers were reported to exhibit individual LCST, because the LCST of PNIPAM is much lower than that of the PEG block.29

Figure 5.4 TEM images of (a) PEG-L4 at 0.25mg/mL in 10mM PBS, (b) PEG-F4 at 0.25 mg/mL in 10 mM PBS, and (c, d) PEG-F4 at 5 mg/mL in 10 mM PBS with additional 1M NaCl.

124

Variable temperature NMR and CD

Dehydration during the LCST transition can be detected with variable temperature 1H-NMR. Upon

LCST, the dehydrated proton NMR signal tends to be broad and weak, with a downfield shift, due to the breakdown of hydrogen bonding between the sample and solvent.30 Without the LCST transition and its dehydration process, the proton NMR signal would be sharper at higher temperatures due to the change in the relaxation time, without any significant change in chemical shift.

1 Figure 5.5 shows the variable temperature H NMR spectra of PEG-F4. The peak at ~3.6 ppm corresponds to the PEG ethylene protons. By zooming in the PEG region, firstly we found that the proton peaks turned evidently weaker at higher temperatures, with a downfield shift, similar to literature reports of dehydration above LCST, indicating that the PEG block dehydrated along with the ELP block above the LCST. Interestingly, there was an evident split of the PEG ethylene peak at 65 oC. The large aggregates in Figure 5.4c cannot be monolayer micelles considering that their radii are much larger than the molecular length. The PEG ethylene peak split at 65 oC is possibly related to the formation of large multi-layer aggregates. Combining the results above, we speculate that the PEG block inside the large aggregates experienced LCST transition and accompanying dehydration, corresponding to the spitted peak at the downfield region, due to its linkage to the

ELP block. In another word, the PEG and ELP blocks exhibited coherent LCST, likely due to the block segments were not long enough to behave individually. However, because of the water environment, certain amounts of PEG chains are located on the surface of the aggregates and extended to the surrounding water. This part of PEG accounts for the upfield part of the split peak, being fully hydrated without the LCST transition.

125

As expected, control PEG-G4 didn’t show a significant change, either in the chemical shift or the peak shape (Figure S5.2a), since the peptide block wasn’t hydrophobic enough to show the LCST with a GGPGG repeating unit. And PEG-L4 showed similar result to PEG-F4 (Figure S5.2b). At higher temperatures, the PEG ethylene protons showed down field shift with an evident decrease in intensity, and a split.

Figure 5.5 Variable temperature NMR of PEG-F4 at 5 mg/mL in D2O with additional 0.5M NaCl.

We further used CD to explore the secondary structure change over the temperature range to understand the thermoresponsive self-assembly at hierarchical levels. ELPs are known to show a gradual random coil to beta sheet transition at higher temperatures.22 As shown in Figure 5.6, PEG-

L4 showed a positive maximum at 204 nm and a minimum at 230 nm, indicating a twisted beta-

31 sheet secondary structure. Furthermore, similar to PEG-G4, PEG-L4 didn’t show a gradual secondary structure change as expected. This is likely because the PEG and ELP blocks were not decoupled, as seen in the turbidity study and NMR. The existence of the linked PEG chain constrained the hydrogen bonding of the ELP block, leading to distortion from conventional beta-

126 sheet; in a similar manner, the hydrophilic PEG prohibited the neighboring ELP dendrimer from rearrangement to a different secondary structure. The strong tendency for micelle formation may also accounts for the constraint in further self-assembly at different temperatures.

Figure 5.6 Variable temperature CD measurement of PEG-L4 and PEG-G4 at 0.25 mg/mL in water.

800

600

400

Intensitya.u. 200

0 290 340 390 440 Wavelength (nm)

Figure 5.7 Emission measurement of valsartan encapsulated in PEG-F4 with fluorescence spectroscopy.

Potential as drug delivery capsules

Based on the previous results, we found that the PEG-peptide-C16 conjugate amphiphiles formed small micelles at RT, while at higher temperatures they self-assembled to larger aggregates. We hypothesized the micelles could be used for the encapsulation of hydrophobic drugs, and upon

127 morphology change at higher temperatures, the drug would be released. Figure 5.7 shows the successful encapsulation of valsartan with PEG-F4, which showed better encapsulation capability compared to PEG-L4, possibly due to the ℼ-ℼ interaction between Phe with valsartan.

5.5 Conclusions

In summary, the triblock PEG-peptide-C16 conjugate amphiphiles were thermoresponsive and showed a morphology change from small micelles to large aggregates as hypothesized. However, due to the attachment of the PEG block, the amphiphiles behaved differently from conventional

ELP systems. The hydrophilic shell and strong tendency for micelle formation limited the thermoresponsive assembly, leading to slow turbidity change in the LCST transition. Secondary structure was twisted from typical β-sheet, and the thermoresponsive trend observed in typical

ELP systems was not observed, either. Variable temperature NMR showed strong evidence for coherent dehydration of the PEG and ELP segments, probably due to the relatively short blocks.

Utilizing the micelles with hydrophobic cavity, we were able to encapsulate hydrophobic drugs, and hopefully release them upon heating and morphology change. Our results exhibit significant insight in the thermoresponsiveness of the PEG-peptide conjugate systems, with promising applications for localized drug release in hyperthermia.

5.6 References

(1) Gauthier, M. A.; Klok, H.-A. Chem. Comm. 2008, 0, 2591.

(2) Wilson, P. Macromol. Chem. Phys. 2017, 218, 1600595.

(3) Shu, J. Y.; Panganiban, B.; Xu, T. Annu. Rev. Phys. Chem. 2013, 64, 631.

(4) Pasut, G.; Veronese, F. M. Prog. Polym. Sci. 2007, 32, 933.

(5) Shu, J. Y.; Huang, Y.-J.; Tan, C.; Presley, A. D.; Chang, J.; Xu, T. Biomacromolecules

2010, 11, 1443.

128

(6) Vandermeulen, G. W. M.; Tziatzios, C.; Duncan, R.; Klok, H.-A. Macromolecules 2005,

38, 761.

(7) Vandermeulen, G. W. M.; Tziatzios, C.; Klok, H.-A. Macromolecules 2003, 36, 4107.

(8) Kühnle, R. I.; Gebauer, D.; Börner, H. G. Soft Matter 2011, 7, 9616.

(9) Cui, H.; Webber, M. J.; Stupp, S. I. Pep. Sci. 2010, 94, 1.

(10) Hartgerink, J. D.; Beniash, E.; Stupp, S. I. Science 2001, 294, 1684.

(11) Palmer, L. C.; Stupp, S. I. Accounts Chem. Res. 2008, 41, 1674.

(12) Morgan, C. E.; Dombrowski, A. W.; Rubert Pérez, C. M.; Bahnson, E. S. M.; Tsihlis, N.

D.; Jiang, W.; Jiang, Q.; Vercammen, J. M.; Prakash, V. S.; Pritts, T. A.; Stupp, S. I.; Kibbe, M.

R. ACS Nano 2016, 10, 899.

(13) Dobbs, R.; Choe, S.; Kalmanek, E.; Harrington, D. A.; Stupp, S. I.; McVary, K. T.;

Podlasek, C. A. Nanomed.-Nanotechnol. 2018, 14, 2087.

(14) Greene, J. J.; McClendon, M. T.; Stephanopoulos, N.; Álvarez, Z.; Stupp, S. I.; Richter,

C.-P. J. Tissue Eng. Regen. M. 2018, 12, 1389.

(15) Motalleb, R.; Berns, E. J.; Patel, P.; Gold, J.; Stupp, S. I.; Kuhn, H. G. J. Tissue Eng.

Regen. M. 2018, 12, e2123.

(16) Kana, T.; Takuma, Y.; Takashi, K. Chem. Lett. 2008, 37, 1208.

(17) Wöhrle, T.; Wurzbach, I.; Kirres, J.; Kostidou, A.; Kapernaum, N.; Litterscheidt, J.;

Haenle, J. C.; Staffeld, P.; Baro, A.; Giesselmann, F.; Laschat, S. Chem. Rev. 2016, 116, 1139.

(18) Shimura, H.; Yoshio, M.; Kato, T. Org. Biomol. Chem. 2009, 7, 3205.

(19) Zhou, M.; Shmidov, Y.; Matson, J. B.; Bitton, R. Colloid. Surface B 2017, 153, 141.

(20) Navon, Y.; Zhou, M.; Matson, J. B.; Bitton, R. Biomacromolecules 2016, 17, 262.

(21) Shmidov, Y.; Zhou, M.; Yosefi, G.; Bitton, R.; Matson, J. B. Soft Matter 2019, 15, 917.

129

(22) Nuhn, H.; Klok, H.-A. Biomacromolecules 2008, 9, 2755.

(23) Mackay, J. A.; Chilkoti, A. Int. J. Hyperther. 2008, 24, 483.

(24) Raucher, D.; Chilkoti, A. Cancer Res. 2001, 61, 7163.

(25) Andrew Mackay, J.; Chilkoti, A. Int. J. Hyperther. 2008, 24, 483.

(26) Tao, P.; Wang, R.; Lai, L. Mol. Model. Annu. 1999, 5, 189.

(27) Yue, G.-l.; Cui, Q.-l.; Zhang, Y.-x.; Wang, E.-j.; Wu, F.-p. Chinese J. Polym. Sci. 2012,

30, 770.

(28) Chang, A. B.; Bates, C. M.; Lee, B.; Garland, C. M.; Jones, S. C.; Spencer, R. K. W.;

Matsen, M. W.; Grubbs, R. H. PNAS 2017, 114, 6462.

(29) Motokawa, R.; Morishita, K.; Koizumi, S.; Nakahira, T.; Annaka, M. Macromolecules

2005, 38, 5748.

(30) Chen, H.; Yang, Y.; Wang, Y.; Wu, L. Chem.-Eur. J. 2013, 19, 11051.

(31) Micsonai, A.; Wien, F.; Kernya, L.; Lee, Y.-H.; Goto, Y.; Réfrégiers, M.; Kardos, J. PNAS

2015, 112, E3095.

130

5.7 Appendix

131

132

133

134

135

Figure S5.1 MALDI-TOF spectra of a-f for Peptide-L4, L3F, L2F2, LF3, F4 and G4, respectively and g-l for PEG-L4, L3F, L2F2, LF3, F4 and G4, respectively. 136

Figure S5.2 Variable temperature NMR results of (a) control PEG-G4, (b) zoomed in PEG region of PEG-F4. All measurements were done at 5 mg/mL in D2O with additional 0.5 M NaCl.

137

Chapter 6: H2S release of covalent peptide hydrogels triggered by human neutrophil elastase (HNE)

Mingjun Zhou, Yun Qian, Yumeng Zhu, Ronit Bitton, and John Matson*

Authors

Mingjun Zhou,[a] Yun Qian,[a] Yin Wang, [a] Ronit Bitton[b] and John B. Matson*[a]

[a] Department of Chemistry and Macromolecules Innovation Institute, Virginia Tech, Blacksburg,

VA 24061 (U.S.A.)

[b] Department of Chemical Engineering, Ben-Gurion University of the Negev, Beer-Sheva, 84105

(Israel)

6.1 Abstract

We report a novel HNE-responsive H2S-releasing hydrogel material by covalently crosslinking biopolymers with an HNE-degradable peptide. Due to the cleavage of the peptide with HNE, a gel-sol transition was triggered to expose encapsulated S-aroylthiooxime (SATO) groups to cysteine, leading to higher H2S-releasing peaking concentration and longer peaking time, compared to the release profile without HNE. More interestingly, after consuming the SATO groups on the surface of the hydrogel, the H2S release can be resumed by adding HNE, which shows potential applications in treating chronic disease with recurring inflammation. Furthermore, the H2S-releasing behavior can be tuned by adjusting the polymer composition, leading to different crosslinking density, hydrogel stiffness, SATO loading, as well as release rate and peaking time.

Our work presents a novel synthetic strategy for H2S-releasing materials, as well as important insight into the enzyme-degradable hydrogels.

138

6.2 Introduction

As a gasotransmitter, H2S is produced endogenously by three mammalian enzymes: cystathionine

β-synthase (CBS), cystathionine γ-lyase (CSE), and 3-mercaptosulfurtransferase (3MST). By regulating the physiological pathways, H2S has shown great potential for disease treatments with therapeutic effects including angiogenesis, antioxidant, anti-apoptotic, and anti-inflammatory properties.1 Inflammation often accompanies injury, and/or bacterial invasion, and even worse, recurs associated with chronic diseases such as cancer, Alzheimer’s, arthritis, diabetes, stroke and

2 3 heart problems. H2S treats inflammation by increasing the blood flow via blood vessel dilation,

4 5 phosphodiesterase inhibition, and cytoprotective effects through opening of KATP channels and

+ 6 partially a K conductance. Besides, H2S plays a key role as a mediator of inflammation, upregulating the adherence and migration of leukocytes which occur in early stage inflammation.7

Thus, H2S has been used to treat chronic diseases with recurring inflammation, such as lung disease,8 ulcers9 and edemas.10

Enzymes are important in the pathology of many diseases, and/or metabolism as well as other biological processes.11,12 Being highly selective towards substrates under mild conditions,13 enzymes have been widely used as the trigger of responsive materials, especially for drug release.14-16 Human neutrophil elastase (HNE) is a serine protease, secreted by neutrophils responding to inflammation,17 including that associated with chronic wounds.18,19 Cleaving peptides/proteins selectively at the carbonyl side of Val/Ala, HNE has been employed as trigger for smart materials.18,20-22 By degrading the peptide/protein components of hydrogels, HNE has shown potential for localized drug delivery through triggering gel-sol transitions.2,23 Due to the link through inflammation between HNE trigger and H2S, we are interested in developing a hydrogel, which is triggered by HNE from inflammation, to release H2S to treat it. Even more

139 interestingly, it would show great therapeutic potential for chronic diseases with recurring inflammation, maintaining the capability of releasing H2S without inflammation, and releasing

H2S when inflammation happens.

We were motivated to fabricate such materials by crosslinking biocompatible polymers with

HNE-degradable peptides. Peptides are convenient biodegradable building blocks for fabricating hydrogels due to their various side chains and tunable number of functional groups. Our lab has

24,25 reported H2S-releasing peptide hydrogels based on physical crosslinking. Here we chose covalently crosslinked hydrogels due to their stability to concentration, pH, ions, as well as their tunable stiffness by adjusting the number of functional groups. We hypothesize that the designed hydrogel will release H2S in response to HNE degradation, due to gel-sol transition. We examine their H2S-releasing profiles at different HNE concentrations, as well as the correlation between

H2S release and hydrogel stiffness. Cell studies were performed to examine their toxicity and protection effect with doxorubicin.

6.3 Experimental section

Materials. Rink amide MBHA resin (0.754 mmol/g) and Fluorenylmethyloxycarbonyl (Fmoc)- protected amino acids were purchased from P3 Biosystems and used as received. HNE was purchased from Athens Research & Technology. Upon arrival, it was dissolved in water, aliquoted and lyophilized, prior to being stored at -80 oC as salt-free powder. Carboxymethyl cellulose

(CMC) (MW=90,000, DS=0.7) was purchased from Acros and used as received. HO-PEG-OH

(polyethylene glycol) (MW=600) was purchased from BeanTown Chemical. All other reagents were purchased from commercial vendors and used as received without further purification unless otherwise noted.

140

Peptide synthesis

All peptides were prepared by solid phase peptide synthesis (SPPS) using a Liberty 1 peptide synthesizer. After cleavage (97.5 v% TFA and 2.5 v% water, 2h), the peptides were purified by preparative HPLC, with ACN/H2O with 0.1 v% TFA, prior to lyophilization. Fractions were analyzed with Advion Express Ion compact mass spectrometer.

Synthesis of Di-Carboxylate PEG (HOOC-PEG-COOH)

HOOC-PEG-COOH was synthesized by oxidizing HO-PEG-OH with Jones’ reagent. Jones reagent was prepared by mixing water (35 mL), concentrated sulfuric acid (98 wt.% 5 mL) and

CrO3 (6.64 g). For the oxidation reaction, HO-PEG-OH (5g, 8.3 mmol) was dissolved in acetone

(60 mL). Then the polymer solution was added to the Jones reagent (in ice bath) dropwise to avoid forming dimer. The mixture was stirred for another half an hour, after which the organic layer was blown away with air. The remaining mixture was extracted with DCM (20 mL) three times, followed by drying the organic layer with anhydrous MgSO4. After rotary evaporation, the product was obtained as colorless viscous liquid. The reaction was quantitative and confirmed by thin layer chromatography.

Hydrogel synthesis

Hydrogel with 35% carboxymethyl cellulose (CMC) and 4-formylbenzoic acid (FBA)-VKVKVK-

NH2 was prepared as follows. All other hydrogels were prepared in a similar way.

Gelation

Peptide FBAVKVKVK-NH2 (200 mg, 720 μmol amine) was dissolved in PBS buffer (pH = 7.4,

10 mM), followed by adding PEG-2COOH (145 mg, 468 μmol carboxylate) and CMC (85 mg, pre-dissolved, 252 μmol carboxylate), and then vortexed for 2 min. EDC-Cl (275 mg, 1440 μmol) and NHS (165 mg, 1440 μmol) were then added, and the total volume was adjusted to 10 mL by

141 adding 10 mM PBS buffer (final pH = 5.0). The mixture was then vortexed for another 5 min. The hydrogel formed within 5 minutes. After reacting overnight, the hydrogel was crushed with a spatula, and 20 mL of milli Q water was added to the hydrogel. It was sonicated for 30 min, followed by vacuum filtration. This sonication-filtration process was repeated twice more (to remove EDC and NHS), prior to lyophilization.

S-Aroyl Thiohydroxylamine (THA) coupling

THA (100 mg, 654 μmol, prepared following literature procedures 34,35) and dry DMSO was added to the as-prepared hydrogel fluffy powder, followed by 2 min of sonication. Dowex (50 mg) resin and activated molecular sieves (50 beads) were added as wrapped in a filter paper pack, which was tied with a cotton thread. It was allowed to react at RT without stirring for 6h. DCM (40 mL) was then added after removing the paper pack, followed by 30-min sonication and vacuum filtration.

This DCM wash-sonication-filtration process was repeated twice more to remove excess THA and

DMSO. The final product was then dried under vacuum and obtained as white powder.

H2S-releasing test

A special glass vial with a small well attached at the bottom was used, similar to previous reports

34, 35. For the HNE-free release test, 1 mg of hydrogel powder was added in to the well, along with

100 μL of PBS (10 mM, pH = 7.40, applying to all PBS mentioned unless indicated otherwise).

The vial was then sonicated for half an hour to ensure hydration. Cysteine (2 μmol) was dissolved in PBS (10 μL) and added to the well, which was immediately covered with a Breathe-EASIER membrane (Diversified Biotech) and sealed with an O-ring. PBS (10 mL) was added to the glass vial, along with a stirring bar. The entire setup was soaked in water bath at 37 oC, which was monitored using a hot plate with a feedback temperature probe (Corning). An H2S-sensitive electrode probe was submerged into the PBS, and H2S concentration was monitored over time.

142

For the HNE-containing release test, 1 mg of hydrogel powder was added to the well, along with

PBS buffer (90 μL, 10 mM, pH = 7.56 to ensure pH = 7.40 for the final mixture). The well was then sonicated for half an hour to ensure hydration. 2.5/5 μg of HNE was firstly dissolved in sodium acetate buffer (10 μL, 50 mM, pH 5.5 with 150 mM NaCl, applying to all sodium acetate buffer mentioned unless indicated otherwise) to ensure homogenization, and then added to the well. The subsequent steps including adding cysteine solution and sealing the well were identical to the HNE-free release test.

For the HNE triggered H2S-releasing test, following similar procedures, hydrogel powder (1 mg) was added to the gel-holding well, along with PBS (100 μL) and cysteine (2 μmol). After setting

o up the apparatus similar to the procedures above, H2S release was monitored for 24 h at 37 C.

Then the 10 mL PBS buffer inside the vial was removed, along with the membrane and O-ring.

HNE (2.5 μg) in sodium acetate buffer (10 μL) was added to the well, prior to sealing with membrane and O-ring. Fresh PBS buffer (10 mL) was added into the vial, an H2S-sensitive electrode probe was submerged into the PBS, and H2S concentration was monitored over time. As a control experiment, sodium acetate buffer (10 μL) was added to the well without HNE. The apparatus was set up in a similar way as described above.

Signals from the electric probe were collected and recorded by LabScribe software (World

Precision Instrument), then converted to H2S concentration (M) by a calibration curve determined at 37 oC. All release profiles were averaged from triplicates.

All of the release results are reported as the average of triplicate experiments.

Rheology

The rheology of hydrogel samples was determined by oscillatory rheology experiments with a TA

2000 rheometer. Measurements were carried out using stainless steel parallel plate (d =8 mm) at

143

1% strain, 1mm gap and frequency from 0.1-100 Hz. The samples were cut into round disks with diameter of 13 mm, and thickness of 2 mm. All of the release results are reported as the average of triplicate experiments.

Circular Dichroism (CD)

CD spectra of FBA-VKVKVK-NH2 from 190 nm to 250 nm were recorded with a Jasco J-815 spectropolarimeter (JASCO, Easton, MD, USA) using a 1 mm path length quartz UV-Vis absorption cell (Thermo Fisher Scientific, Pittsburgh, PA, USA). The peptide was dissolved in 10 mM PBS buffer (pH = 7.40) at a concentration of 0.25 mg/mL. All of the release results are reported as the average of triplicate experiments

Peptide solution HNE degradation test

A brief HNE-peptide degradation test with peptide FBA-VKVKVK-NH2 was performed. HNE

(2.5 μg) was dissolved in sodium acetate buffer (10 μL) first, followed by the addition of peptide in PBS buffer (100 μL, 1 mg/mL). Then the mixture was soaked in water bath at 37 oC. The reaction mixture (10 μL) was analyzed at different time points with a matrix-assisted laser desorption ionization−tandem time of flight mass spectrometer (4800 MALDI TOF/TOF; AB

Sciex).

Cell culture

An adherent H9C2 line of rat embryonic cardiomyocytes (ATCC, Manassas, VA, USA) was used in this study. Complete DMEM media were prepared with DMEM supplemented with 10% fetal bovine serum (FBS), 50 IU/mL penicillin, and 50 μg/mL streptomycin. H9C2 cells were cultured with complete DMEM at 37 °C in 5% CO2-air. The cultures were passaged after 70–80

% confluence was achieved. Cells were rinsed with PBS (1X) solution three times, and then

144 released with trypsin and 0.25% EDTA solution (VWR, Radnor, PA). The suspension of released cells was centrifuged at 1000 rpm for 5 min.

Cell Viability Assays

H9C2 cells (5000 cells per well) were plated in a 96-well plate with 180 μL complete DMEM media. Doxorubincin hydrochloride (Dox, 0.116 mg) were dissolved in PBS (1 mL) to prepare

200 M of Dox stock solution. Gel (1 mg) was swellon in PBS (781 L) to make hydrogel stock suspension for varied treatments. After culturing for 24 h, cells were washed with PBS (1X) three times, and 125 μL of complete DMEM media was added. Cells were pretreated with the following groups: (a) Dox only: PBS (70 μL); (b) Gel+Cys+HNE+Dox: 664 M gel (50 μL),

25.6 mM Cys (5 μL), 1.6 μg/μL HNE (10 μL) and PBS (5 μL); (c) Gel+Cys+Dox: 664 M gel

(50 μL), 25.6 mM Cys (5 μL), and PBS (15 μL); (d) Gel+HNE+Dox: 664 M gel (50 μL), 1.6

μg/μL HNE (10 μL) and PBS (10 μL); (e) Cys+HNE+Dox: 25.6 mM Cys (5 μL), 1.6 μg/μL

HNE (10 μL) and PBS (55 μL); (f) Gel+Dox: 664 M gel (50 μL) and PBS (20 μL); (g)

Cys+Dox: 25.6 mM Cys (5 μL), and PBS (65 μL); (h) HNE+Dox: 1.6 μg/μL HNE (10 μL) and

PBS (60 μL). Cells were incubated with the treatment groups for 1 h, then 5 µL of Dox in PBS

(200 M) was added to each well to make the final Dox concentrations 5 µM. After incubation for 24 h, cells were washed three times with PBS (1X), and then treated with serum-free DMEM

(100 μL) and 10 μL Cell Counting Kit-8 solution (CCK-8, Dojindo, Rockville, MD). After incubation for another 2 h to allow for development of the CCK-8 dye, absorbance was recorded at 450 and 650 nm using a BioTek Synergy Mix plate reader (BioTek, Winooski, VT). For data analysis, multiple comparisons were done using a one-way variance analysis (ANOVA) followed by Tukey-Kramer HSD for multiple pairwise examinations. Mean values are reported together with the standard deviation of mean (n = 5, p < 0.01).

145

Elemental Analysis

Elemental Analysis result was obtained from Midwest micro lab via Schoniger combustion.

6.4 Results and discussion

Scheme 6.1 Synthetic route. Red bonds indicate HNE degradation sites.

Design and synthesis

Scheme 6.1 shows the synthetic route of the HNE-triggered H2S-releasing hydrogel. We started with crosslinking CMC with the peptide FBA-VKVKVK-NH2. However, we found that the resulting hydrogel was very robust, and there was no gel-sol transition observed after adding HNE.

We then lowered the percentage of CMC by adding HOOC-PEG-COOH, and found the lowest % of CMC required to gel was 35 mol% (carboxylate count, abbreviated as 35% CMC). After gelation, we analyzed the rheology of the FBA-hydrogel (Figure S6.2), which confirmed hydrogel formation by G’>G’’. Elemental analysis of the SATO-hydrogel shows sulfur wt.% was 1.70%

(100% yield), which is converted to the SATO loading of 0.53 mmol/g.

To confirm that the sulfur from elemental analysis was from SATO group instead of physically encapsulated THA or residual DMSO, we synthesized a control hydrogel by changing the FBA N- terminus to acetamide group (Scheme S6.1). THA couldn’t react with the acetylated hydrogel, and

146 so could not form the SATO group. Elemental analysis (sulfur) of the control Ac-hydrogel was

0.25 wt.%, which is lower than the instrument inaccuracy 0.3%. This confirms the successful formation of SATO group and the THA encapsulation was neglectable.

Effect of HNE on the H2S-releasing profile

We first investigated the degradation of peptide FBA-VKVKVK-NH2 in solution with HNE at different time points in 24 h (Figure S6.3). Degradation happened as early as 0.5 h. However, the intensity of the degradation products was weak until after 2h. The products corresponding to cleaving the two Val close to C-terminus were clear and identified. But the direct degradation product by cleaving the N-terminal Valine wasn’t observed. This coincides with literature that Val doesn’t necessarily guarantee HNE recognition.23

c (HNE) effect 1.6 no enzyme 1.4 1 enzyme

1.2 2 enzyme M)

μ 1 0.8 0.6

c[H2S] ( c[H2S] 0.4 0.2 0 0 200 400 600 800 Time (min)

Figure 6.1 H2S-releasing profile with different HNE concentrations (1enyzme corresponds to 0.17 nmol HNE in 110uL reaction mixture).

147

Table 6.1 Peak concentration and time of the H2S-releasing profile with different concentration of

HNE

35% CMC Peaking H2S c (μM) Peaking time (min)

No HNE 0.6±0.2 240±20

1 HNE 0.8±0.1 354±4

2 HNE 1.4±0.3 600±50

We then investigated the H2S release of the SATO hydrogel (35% CMC) with different concentrations of HNE, as shown in Figure 6.1. By comparison to a calibration curve (Figure

S6.3), we were able to convert the probe current to H2S concentration. The corresponding peaking time and concentration of each H2S-releasing curve is summarized in Table 6.1. Without HNE,

H2S was released with a peaking concentration of 0.6 M of H2S with a peaking time of 240 min.

Without degradation and gel-sol transition, this part of release is likely from the SATO groups from the hydrogel surface. With HNE, the peaking H2S concentration was increased to 0.8 M, with elongated peaking time of 354 min. The same trend applied to the release test when the HNE concentration was doubled (labelled as 2 HNE). We expected a burst release of H2S with HNE due to the gel-sol transition. However, this result makes sense considering the HNE degradation of the peptide crosslinker is time-consuming. The solution degradation result showed that there was still a considerable amount of intact FBA-VKVKVK-NH2 left in the reaction mixture even after 24 h

(Figure S6.3). Besides, cysteine, which is used to react with SATO groups to generate H2S, is significantly smaller than HNE, corresponding to faster diffusion rate. In previous results from our lab, the peaking times of SATO groups on small molecules are rather short,25,27 which may suggest that the HNE degradation is the rate determining step rather than the cysteine attack. This also

148 explains why the initial release rate with 2 HNE was slower than the other two: the H2S release from the surface SATO groups dominates in the beginning stage, while later on the release due to gel-sol transition prevails.

In summary, the result indicates that there were two types of SATO groups in the hydrogels: the surface SATO groups can be easily accessed by cysteine and release H2S; the interior SATO groups were encapsulated inside the hydrogel matrix. Only by cleaving the peptide crosslinker and degrading the hydrogel are the buried SATO groups are exposed to cysteine attack, leading to additional H2S release. That’s why with more enzyme, there was longer H2S-releasing peaking time and higher peaking concentration. This is further confirmed by visual observation in Figure

6.2. After 24 h release test without HNE, the hydrogel had swelled much more compared to its initial status. Conversely, with 2 HNE, most of the hydrogel was degraded.

Figure 6.2 The H2S-releasing apparatus and photos taken before the release test (left), after the release test without HNE for 24h (middle) and after the release test with 2 HNE for 24h (right).

149

To further confirm that the difference in the curves in Figure 6.1 originated from HNE degradation, we synthesized a control hydrogel crosslinked by a non-HNE degradable peptide FBA-GKGKGK-

NH2, with the rest of the synthetic route unchanged. As shown in Figure 6.3, with HNE, the release concentration was even slightly higher than the one without HNE, which is likely from instrument error (considering big error bar in Table 6.1). This supports the HNE degradation effect in Figure

6.1.

0.4 with HNE Without HNE 0.3

0.2

S)[μM] 2

c (H c 0.1

0 0 200 400 600 Time (min)

Figure 6.3 H2S-releasing profile of hydrogel crosslinked by non-HNE degradable FBA-

GKGKGK-NH2.

Potential for treating chronic diseases

The results above show that H2S release is responsive to HNE, but not perfectly so. Based on the discussion of surface/interior SATO groups, we were interested in the H2S-releasing behavior after consuming the surface SATO groups. Thus, we designed a release test in which we reacted the surface SATO groups without HNE for 24h, then adding HNE, as shown in Figure 6.4.

The hydrogel was first tested for H2S release without HNE (blue curve). It showed a low release concentration, similar to the result above. After 24 h, the H2S concentration is evidently decreased,

150 nearly to a plateau, indicating most of the surface SATO groups are consumed. After that, HNE in acetate buffer was added to the reaction mixture, to trigger the gel-sol transition and expose the interior SATO groups to cysteine. As shown in the red curve, the H2S release resumed. The control black curve confirmed that the change in H2S release wasn’t from the addition of the acetate buffer.

Besides the understanding of the surface/interior SATO groups, this experiment was also designed to mimic the process to treat chronic diseases with recurring inflammation. The blue curve was a mimic of the process with the hydrogel loaded in human body when there was no inflammation and HNE generated. Even after the surface SATO groups are consumed, the hydrogel still maintains the capability to release H2S when triggered. When inflammation happens with HNE, the hydrogel can resume H2S release to treat the disease. Thus, this material shows great potential for treating chronic diseases with recurring inflammation.

1.2

1 First 24h without HNE Add HNE in acetate buffer 0.8 Add acetate buffer as control

0.6

S] S] (μM) 2

c[H 0.4

0.2

0 0 500 1000 1500 2000 2500 Time (min)

Figure 6.4 The H2S-releasing profile from surface SATO groups (blue curve without HNE) and interior SATO groups (red curve with HNE). The black curve is a control by adding only acetate buffer.

Tuning H2S release by changing the polymer composition

151

Based on the understanding of hydrogel design, we speculated that varying polymer composition could be used to tune the H2S release. As shown in Figure 6.5, we prepared two more hydrogels with 55 and 75 mol% (carboxylate count) CMC. A Frequency sweep experiment showed an evident trend that higher % CMC led to higher G’ (at 5Hz, G’ values of hydrogels with 35, 55, and

75 % CMC were 1586, 1930, 3370 Pa, respectively), indicating a higher crosslinking density.28

Elemental analysis showed sulfur wt.% was 1.70, 1.35 and 0.44% for hydrogels with 35, 55 and

75% CMC, respectively. The release profiles are shown in Figure 6.4. With more CMC, the hydrogel H2S release was slower, and the concentration decrease over time was less evident as well.

Our explanation is that these differences were initially from the change in the number of functional groups. The average number of carboxylates per polymer chain was 182 for CMC and 2 for PEG.

Both Carother’s and Flory-Stockmayer’s equations show that a higher number of functional groups favors gelation.28 Thus higher % of CMC leads to a higher crosslinking density, as well as a stiffer hydrogel. This further affected the SATO formation because with a higher crosslinking density, the aldehyde in the hydrogel was more difficult to access by THA, resulting in lower SATO loading and sulfur wt.%. For the H2S release, with a higher crosslinking density, less of SATO groups are located on the surface of the hydrogel, leading to slower release at the beginning; while to expose the interior SATO groups, the HNE needs to degrade more crosslinks to trigger the gel- sol transition. This is likely the reason that the HNE wasn’t able to evidently elevate the H2S release over time for the hydrogel with 75% CMC. On the other hand, because the HNE couldn’t effectively degrade the hydrogel, the H2S release in the gray curve was considerably lower as well.

152

% of CMC effect 0.9 35% cmc 0.8 55% cmc 0.7 75% cmc 0.6

0.5 S] S] (μM)

2 0.4

c[H 0.3 0.2 0.1 0 0 200 400 600 800 Time (min)

Figure 6.5 Release profiles of hydrogels with various % of CMC and 1 HNE.

Figure 6.6 Cell viability of H9C2 cardiomyocytes with different treatments (1 h) followed by exposure to Dox for 24 h. * p < 0.01 vs Dox.

Cell study

As shown in Figure 6.6, treatments Gel+Cys+HNE+Dox and Gel+Cys+Dox displayed cell viabilities of 91% and 81%, respectively. These results indicated the CMC-SATO gel can protect

153

H9C2 from Dox treatment. Statistical differences were found between these two treatments, suggesting that the HNE improved the H2S release and enhanced the protection. Other control treatments were also applied following the same procedure. However, the viabilities of these control groups were not statistically different from Dox only, suggesting that these controls had no protective effects against Dox (5 M).

Scheme 6.2. A schematic demonstration of the role of HNE in the H2S release of the hydrogel.

6.5 Conclusions

In this study, we have designed and synthesized an HNE-responsive H2S-releasing hydrogel.

Without enzyme, cysteine can only react with SATO groups on the surface of the hydrogel to release H2S. In contrast, due to the gel-sol transition triggered by HNE degrading the peptide crosslinker, interior SATO groups are exposed to cysteine. (Scheme 6.2) Therefore, with more

HNE, the release profile showed a longer peaking time and higher peaking concentration. After consuming most of the surface SATO groups, the H2S release resumed after adding HNE.

Furthermore, the H2S-releasing behavior can be tuned by adjusting the polymer composition, 154 leading to different crosslinking density, hydrogel stiffness, SATO loading, as well as release rate and peaking time. In a cell study our system demonstrated good viability, and protection effect against doxorubicin. With all these findings, our system has provided unique insight in hydrogel drug release, enzyme degradation, as well great potential as treatment to chronic diseases with recurring inflammation.

6.6 References

(1) Kanagy, N. L.; Szabo, C.; Papapetropoulos, A. Am. J. Physio-Cell Ph. 2017, 312, C537.

(2) Aimetti, A. A.; Tibbitt, M. W.; Anseth, K. S. Biomacromolecules 2009, 10, 1484.

(3) Zhao, W.; Zhang, J.; Lu, Y.; Wang, R. EMBO J. 2001, 20, 6008.

(4) Bucci, M.; Papapetropoulos, A.; Vellecco, V.; Zhou, Z.; Pyriochou, A.; Roussos, C.;

Roviezzo, F.; Brancaleone, V.; Cirino, G. Arter. Thromb. Vasc. Biol. 2010, 30, 1998.

(5) Qian, Y.; Matson, J. B. Adv. Drug. Deliver. Rev. 2017, 110-111, 137.

(6) Zhao, W.; Zhang, J.; Lu, Y.; Wang, R. EMBO J. 2001, 20, 6008.

(7) Bhatia, M. Scientifica 2012, Article ID 159680, 12 pages.

(8) Wang, Z.; Yin, X.; Gao, L.; Feng, S.; Song, K.; Li, L.; Lu, Y.; Shen, H. SpringerPlus

2016, 5, 1084.

(9) Wallace, J. L.; Caliendo, G.; Santagada, V.; Cirino, G.; Fiorucci, S. Gastroenterology

2007, 132, 261.

(10) Zhang, H. X.; Liu, S. J.; Tang, X. L.; Duan, G. L.; Ni, X.; Zhu, X. Y.; Liu, Y. J.; Wang,

C. N. Cell. Physiol. Biochem. 2016, 40, 1603.

(11) Hu, Q.; Katti, P. S.; Gu, Z. Nanoscale 2014, 6, 12273.

(12) Owen, C. A.; Campbell, E. J. J. Leukocyte Biol. 1999, 65, 137.

(13) Lee, S.; Park, K.; Kim, K.; Choi, K.; Kwon, I. C. Chem. Comm. 2008, 4250.

155

(14) Ulijn, R. V. J. Mater. Chem. 2006, 16, 2217.

(15) Zelzer, M.; Todd, S. J.; Hirst, A. R.; McDonald, T. O.; Ulijn, R. V. Biomater. Sci. 2013,

1, 11.

(16) Zelzer, M.; Ulijn, R. V. In Smart Polymers and their Applications; Aguilar, M. R., San

Román, J., Eds.; Woodhead Publishing: 2014, p 166.

(17) Stein, R. L.; Strimpler, A. M. Biochemistry 1987, 26, 2238.

(18) Döring, G. Am. J. Resp. Crit. Care 1994, 150, S114.

(19) Allan, E. K.; Holyoake, T. L.; Jørgensen, H. G. Leukemia 2006, 20, 2054.

(20) Demkow, U.; van Overveld, F. J. Eur. J. Med. Res. 2010, 15 Suppl 2, 27.

(21) Ohbayashi, H. Expert. Opin. Inv. Drug. 2002, 11, 965.

(22) Pak, C. C.; Erukulla, R. K.; Ahl, P. L.; Janoff, A. S.; Meers, P. BBA-Biomembranes 1999,

1419, 111.

(23) Kamarun, D.; Zheng, X.; Milanesi, L.; Hunter, C. A.; Krause, S. Electrochim. Acta 2009,

54, 4985.

(24) Qian, Y.; Kaur, K.; Foster, J. C.; Matson, J. B. Biomacromolecules 2019, 20, 1077.

(25) Carter, J. M.; Qian, Y.; Foster, J. C.; Matson, J. B. Chem. Comm. 2015, 51, 13131.

(26) Wang, Y.; Kaur, K.; Scannelli, S. J.; Bitton, R.; Matson, J. B. J. Am. Chem. Soc. 2018,

140, 14945.

(27) Foster, J. C.; Powell, C. R.; Radzinski, S. C.; Matson, J. B. Org. Lett. 2014, 16, 1558.

(28) Shmidov, Y.; Zhou, M.; Yosefi, G.; Bitton, R.; Matson, J. B. Soft Matter 2019, 15, 917.

156

6.7 Appendix

Figure S6.1 ESI-MS of (a) FBA-VKVKVK-NH2, (b) FBA-GKGKGK-NH2, (c) H-VKVKVK-

NH2 and (d) Ac-VKVKVK-NH2.

157

1000000 G' G'' 100000 35% CMC

) 10000 Pa 1000 100

Modulus( 10 1 0.1 1 10 100 Frequency (Hz)

Figure S6.2 FBA-hydrogel with 35 % CMC (left) and the frequency sweep result.

Scheme S6.1 The synthetic route of the control Ac-hydrogel.

158

Figure S6.3 Degradation profile of peptide FBA-VKVKVK-NH2 with HNE in solution with

MALDI-TOF. The MW of products cleaved at position 1, 2, and 3 is expected to be 599.5, 476.3 and 703.4, respectively. Peak 600.5 was never observed, which is likely associated with peak

568.1.

159

100 Calibration Curve (10 nA) 80 y = 2.93x 60 R² = 0.9602

40 Voltage (V) Voltage

0 0 10 20 30 40 Concentration (uM)

Figure S6.4 Calibration curve of probe voltage and H2S concentration.

-5

-10

mdeg -15

-20

-25 190 200 210 220 230 240 250 Wavelength (nm)

Figure S6.5 CD spectrum of peptide FBA-VKVKVK-NH2 at the concentration of 0.25 mg/mL in

10 mM PBS (pH=7.4). 160

Chapter 7. Conclusions and Future Work

We designed and synthesized a series of ELP dendrimers, optimized the synthetic conditions, and explored the limit of synthesizing ELP dendrimers with solid phase peptide synthesis (SPPS). We then explored their thermoresponsiveness, in terms of secondary structure change, macroscopic

LCST, as well as self-assembly behavior during the transition. By using them as crosslinkers, we were able to investigate their thermoresponsiveness in the gel state, as well as their potential as drug releasing vehicles. Through conjugating the ELP dendrimers into multi-block amphiphiles, we investigated their thermoresponsiveness in a complex system, and obtained insights into the self-assembly over the transition.

Chapter 2, 3 and 4

These three chapters compose of a complete story of expanding the research from synthesizing

ELP dendrimers to synthesizing larger ELP dendrimers and eventually preparing hydrogels with thermoresponsive ELP dendrimer crosslinkers.

For future work, in chapter 2, we would like to measure the LogD values directly with HPLC, and compare the results with our calculations. Besides, when the dendrimers were used as crosslinkers, the first-generation moiety of the dendrimers wasn’t likely to contribute to the gels’ thermoresponsiveness, since it isn’t between any two amine groups. This was overcome in chapter

3. In chapter 3, we would like to conduct a more thorough comparison in the yields between ELP dendrimers and their linear control peptides. We only had a semi-quantitative evaluation. In chapter 4, the as-prepared hydrogels showed very slow reswelling. I suspected it was related to the topology that due to the amine groups were located on the surface of the dendrimers, leading to higher accessibility compared to the amine groups in linear ELPs. We would like to conduct such a control experiment and investigate this possible effect. A real drug would be more convincing in

161 the Rhodamine 6G loading/releasing experiments. It would be interesting to compare the loading/releasing results of hydrophobic and hydrophilic drugs, which would provide a better understanding of the hydrogel network. Besides, we would like to get a quantitative evaluation of the conversion of the EDC coupling, possibly with MALDI-TOF and NMR in a solution system.

In chapter 5, during the study, I realized that the number of branches were likely related to the size of the aggregates. The wedge shape of the molecular structure (4-arms) was intended for micelle formation. However, what I didn’t realize during the experiment design was lowering the number of branches (2 arms or linear) would likely lead to bigger micelles, due to smaller cone angles if each molecule was considered as a cone in a sphere. Besides, bigger micelles would probably result in better drug encapsulation. And I would like to conduct a more thorough study about the

LCST dependence on the PEG molecular weight, by conjugating a PEG of different molecular weight, or collecting and analyzing different fractions from the HPLC.

In chapter 6, I would like to evaluate the H2S-releasing behavior related to the hydrogel mesh size.

It would definitely improve the reproducibility, and provide new insights, as well as tune the release behavior. Furthermore, I would like to explore of potential of utilizing the system to run the enzyme-triggered H2S release in multiple cycles, to better mimic the chronic diseases with recurring inflammation.

162