1
Designing cross-linked hyaluronic acid hydrogels
with tunable physical properties
Research performed at the NWO Institute AMOLF: Physics of Functional Complex Matter
(Amsterdam)
Experimental work performed: 28-11-2016 – 30-06-2017
Date of publication thesis: 06-10-2020
Daily Supervision: Dr. Galja Pletikapic & Drs. Federica Burla (AMOLF)
First Supervisor: Prof. dr. Gijsje Koenderink (AMOLF, since 1-9-2019 TU Delft)
UvA Supervisor: Prof. dr. Peter Bolhuis (UvA)
Master Thesis of Alexandra Wolters
Master track: Chemistry, joint degree UvA/VU
EC: 42
2
Abstract
Hyaluronic acid (HA) is a disaccharide-based polymer that is an important component of the extracellular matrix (ECM) of connective tissues. In solution, HA chains form physically entangled networks. HA can be chemically modified to increase its mechanical and (bio)chemical stability and to tune its physical and chemical properties. It is also possible to create stimuli-responsive gels, which can sense changes in the surrounding environment and react to them. A promising path is the use of ssDNA molecules as crosslinkers, which would cause the gel to respond to the addition of the complementary ssDNA strand by forming alfa-helixes and thus changing the physical properties of the gel. In this thesis carbodiimide chemistry for amidation at the carbonic acid group and a Michael-addition using thiolated HA are employed to successfully synthesize HA hydrogels with PEG crosslinkers. Fluorescence recovery after photobleaching (FRAP) and Rheology experiments are employed to probe the physical properties of these hydrogels on al local and macroscopic scale. Both synthesis methods yielded hydrogels that exhibited elastic properties in rheology frequency sweep experiments. Crosslinking with the Michael-addition using thiolated HA and PEG crosslinker yielded hydrogels with an inhomogeneous structure, including pores with sizes between 15 and 27 nm. Carbodiimide chemistry was employed on HA with different molecular weights. The reaction proceeded satisfactory in PBS buffer at physiological pH. These gels were shown to absorb up to 86 times their own weight in water. The carbodiimide reagent was found to influence the physical properties of HA even without the addition of crosslinker. The carbodiimide chemistry was selected to perform initial experiments to synthesize HA hydrogels with responsive DNA crosslinkers and proved not effective under the used conditions.
3
Contents
1. Introduction 4
2. Theoretical Background 6
2.1 Chemical crosslinking of Hyaluronic Acid 6
2.2 Analysis methods 11
3. Experimental section 15
4. Results & discussion 18
4.1 Entangled Hyaluronic Acid 19
4.2 Crosslinked Hyaluronic Acid 22
5. Conclusion 31
Supplementary information 33
4
1. Introduction
Hyaluronic acid (HA) is a glycosaminoglycan that consists of a recurring disaccharide of N-acetyl-d-glucosamine and d-glucuronic acid (figure 1).1 It is an important component of the extracellular matrix (ECM) of connective tissues and is also present in the vitreous eye fluid and synovial fluid of joints. It has a number of roles in connective tissues. HA helps swelling, binds water, acts as a lubricant and is involved in the regulation of cell growth, migration, differentiation and wound healing. HA molecules are linear polymer chains with lengths that can be in the micrometer range. In solution, HA chains form physically entangled networks that have poor mechanical strength and weak resistance to biochemical degradation. However, HA can be chemically modified to increase its mechanical and (bio)chemical stability and to tune its physical and chemical properties, which opens up new directions in cell studies and tissue engineering.
When hyaluronic acid chains are crosslinked, they can form a hydrogel. A hydrogel is a substance with both solid-like and fluid-like properties.2 It can absorb a large amount of water, while still holding a solid-like structure. The solvent that is held within the pores of the gel allows for its fluid-like properties. The exact properties of a hydrogel are typically defined by the type of crosslinker and polymer used, as well as by the degree of crosslinking. Because of their high water content and tunable chemical and mechanical properties, hydrogels are particularly interesting for applications in tissue engineering and controlled drug release systems. Studies have shown the high potential for hydrogels in mimicking the extracellular matrix (ECM) of living cells.3 The ECM consists of proteins (i.e. collagen), proteoglycans and other soluble molecules, that together form a complex system.
Hydrogels can be divided into two classes: stimuli-responsive and non-responsive.2 Non-responsive gels swell and absorb water by being exposed to it, without mechanical of chemical stimuli, while stimuli-responsive gels can sense changes in the surrounding environment and react to them. Examples of these external stimuli include changes in pH, temperature and ionic strength. These stimuli can then induce swelling or shrinking of the geld. Furthermore, gels can be created synthetically that respond to the presence of specific molecules, such as antibodies, oligonucleotides and enzymes. The success of an extracellular matrix substitute is determined by, among others, its biocompatibility, biodegradability and physical stability, which in turn are determined by the used materials.1 The biocompatibility and biodegradability of hyaluronic acid, combined with the mechanical stability and stimuli-responsiveness that can be added through crosslinking, make it an ideal candidate for mimicking the extracellular matrix. Having complete control over the ECM creates a perfect environment to study cellular mechanoresponse: the study of living cells inside extracellular matrices that mimic the natural tissue environment. Having control is important, as changes in the structural and mechanical properties of the matrix are responsible for triggering the mechanoresponse inside cells.
One possibility to create stimuli-responsive hydrogels is to employ single-stranded (ss)DNA molecules as crosslinkers.4 The gel will then be responsive specifically to the complementary ssDNA strand: upon mixing, the ssDNA strands will bind together via hydrogen bonds and form double stranded (ds)DNA. Because the dsDNA double helix is more rigid and therefore much shorter in length than a single stranded oligonucleotide, the transformation will cause the hydrogel to shrink. Alternatively, hydrogels can be crosslinked with dsDNA, which will break up when the temperature or pH is changed (figure 2) to effectuate the opposite response. Such biocompatible and switchable
5
hydrogels can be useful for various applications, including tunable reconstituted model systems for the extracellular matrix.
The fabrication of responsive DNA-crosslinked hydrogels has been successful for polyacrylamide (PA) gels [4]. However, it has not yet been realized with HA as the polymer. There are multiple advantages of HA over PA, but the most important one is that PA gels are only suitable for 2D cell culturing, while HA gels can encapsulate cells in 3D.5 To this end, we focus on using HA. The aims of this research are to investigate how the properties of HA hydrogels can be influenced through crosslinking reactions, and to search for a protocol to create DNA-responsive HA hydrogels.
This report explores HA hydrogels in a model system with PEG-based molecules as crosslinkers, discussing different reaction types and conditions and different molecular weights of HA. The viscoelastic properties of the hydrogels are investigated using shear rheology. In addition, FRAP (fluorescent recovery after photobleaching) is used to measure solute diffusivity and to estimate the porosity of the network by using fluorescent dextrans of different sizes. Furthermore, the results of some initial DNA-based crosslinking experiments are described, and future steps and experiments are proposed.
Figure 1 Chemical structure of hyaluronic acid (HA).
Figure 2 Principles of designing switchable networks by using DNA strands according to 4 different methods. (A) Contraction of a HA gel crosslinked with ssDNA upon the addition of the complementary ssDNA strand. (B) Expansion of a HA gel crosslinked with ssDNA with stem-loop intramolecular base-pairing upon the addition of the complementary ssDNA strand. (C) and (D) Dissociation of crosslinks in a HA network, in which crosslinks are formed by intermolecular bonds between partially complementary ssDNA molecules, by the addition of fully complementary ssDNA. Figure taken from ref 6.
6
2. Theoretical Background
This chapter discusses the underlying theoretical background of the chemical reactions and analytical methods that are used. Firstly, the reactivity of hyaluronic acid will be discussed and specifically the three applied reaction types are further explained. Secondly, the theoretical background of the analytical methods is explained. Discussed methods are fluorescent recovery after photobleaching (FRAP), rheology and swelling experiments.
2.1 Chemical Crosslinking of Hyaluronic acid
Hyaluronic Acid (fig 1) has different chemical groups that can be chemically functionalized to change the properties of the material. Chemical functionalization can be divided into two different types: conjugation and crosslinking.7 These types of functionalization are based on the same chemical reaction types, but differ in that with conjugation a compound is grafted to a single HA chain, while with crosslinking two different HA chains, or two sections of the same HA chain, are bound together (see figure 3). In this research, only covalent chemical reactions of HA are considered.
Figure 3 Types of chemical functionalization.
The three relevant functional groups for crosslinking are the carboxyl group, the amide group, and the hydroxyl groups (figure 4). The circled hydroxyl group is as primary alcohol sterically the least hindered and can therefore be expected to be most reactive. The circled hydroxyl and carboxylic acid groups are most commonly used for functionalization.1,7,8 Typical reactions include amidation, esterification and oxidation for the carboxyl groups, and ether formation and esterification for the hydroxyl groups. Some of these reactions are performed in water, while others may require an organic solvent such as dimethylsulfoxide (DMSO) due to the use of reagents that are easily hydrolyzed. The natural form of HA is as a sodium salt and it is thus soluble in water. In order to dissolve it in organic solvents, HA has to be transformed into its tetrabutylammonium (TBA) salt. Unfortunately, this transformation adds extra steps to the protocol, and also increases the chances of the HA chains breaking up due to the acidic treatment required to obtain the TBA salt.9 However, some reactions in water are also pH-dependent and have to be performed under basic or acidic conditions, introducing the same problem of unwanted chain hydrolysis.10
7
Figure 4 Reactive functional groups on the HA monomer with potential for crosslinking, from left to right and top to bottom: carboxylic acid, hydroxyl and amide.
In this research project, one of the goals is to stay as close to physiological conditions as possible, which means the use of water or aqueous buffer is preferred over organic solvents. Furthermore, the protocol should be relatively simple, because it is to be applied in a biophysical lab setting, and should thus consist of few steps and include a limited amount of reactants, which also points to the use of aqueous reactions. Finally, the functionalization should be easily adjustable in order to be able to produce different products with the same protocol. With all requirements in mind, the following three reactions were chosen for further investigation:
• Modification of OH: etherification using epoxides
• Modification of COOH: amidation using carbodiimide chemistry • Reaction via thiolated HA: michael-addition
These reactions are further explained below. Modification of OH: etherification using epoxides
The first method for crosslinking HA was epoxide etherification at the hydroxyl group, published in 1964.11 Currently, butanediol-diglycidyl ether (BDDE) is frequently used as the crosslinking epoxide in this reaction. The protocol for the reaction between HA and BDDE was patented in 1986 by Mälson and Lindqvist12 and improved by Piron and Tholin in 2002.13 The use of BDDE is advantageous because HA-BDDE has not shown any cytotoxicity, nor have its metabolic derivatives.14
The reaction is an epoxide ring opening with base, of which the mechanism is shown in figure 5. The nucleophilic hydroxyl group attacks on the least substituted side of the epoxide (SN2 reaction), upon which the epoxide opens up and an ether bond is formed. The solution pH during the reaction should be higher than 12, well above the pKa value of the hydroxyl groups (~10), so that all hydroxyl groups are deprotonated. Deprotonated hydroxyl groups are more reactive than deprotonated carboxylic acid groups (pKa ~4.5) and therefore the formation of ester bonds is ensured at high pH. However, at pH lower than 10, the deprotonated carboxylic acid group is predominant and the competing ester bond formation is promoted.7 Successful HA modification has been reported, even though it has been shown that HA is subject to degradation at pH values higher than 11.10
8
Figure 6 pKa values15 of hyaluronic acid functional groups.
Modification of COOH: amidation using carbodiimide chemistry
In 1971 Danishefsky and Siskovic reported the first conversion of the carboxyl groups of hyaluronic acid using carbodiimides.16 They reacted glycine methylester with HA in the presence of the carbodiimide EDC at a pH of 4.75 and reported an amidation at 38.6% of the carboxyl groups. However, in 1991, Kuo et al. showed that no amidation took place at the carboxyl site.17 Instead, the N-acylurea byproduct was formed. They performed the reaction at the same pH of 4.75 as previous reports.
Nakajima and Ikada studied the reaction mechanism of the transformation in more detail.18 First, the carboxylic acid group of HA is activated by reacting it with EDC, which forms an O-acylisourea. Second, this intermediate undergoes nucleophilic attack by the amine to form the amide bond. However, the O-acylisourea intermediate is highly reactive and can also react with water, which causes a rearrangement to the stable N-acylisourea byproduct. In this form HA does no longer react with the amine and the preferred reaction is prevented. Problematic is that at a pH of 4.75, necessary for the activation of HA with EDC, the amine is less nucleophilic. Therefore the quick reaction with water takes precedence. At basic pH the amine is more nucleophilic, but EDC is rapidly hydrolyzed, becomes inactive, and no reaction takes place at all. These characteristics illustrate the obstacles for this method of functionalizing HA.
In 1999 Bulpitt and Aeschlimann reported a solution to these problems.19 N-hydroxysuccinimide (NHS) or 1-hydroxybenzotriazole (HOBt) could be added to the reaction with EDC to capture the O-accylisourea intermediate, preventing the transformation to the stable N-acyisourea byproduct. NHS or HOBt reacts with the reactive intermediate to form a more stable intermediate that can then react with the amine. Using this method, multiple primary and secondary amines were successfully coupled to HA.
Functionalization of HA using EDC has several significant advantages. The reaction can be carried out with many different types of amines, so many different products are accessible. In addition, EDC is a so-called zero-length crosslinker, which means none of the original reactant can be found in the final product.18 Only HA and the coupled amine remain after the reaction is completed and byproducts are dialyzed out. Furthermore, no evidence has been reported that the HA chain is cleaved by the use of EDC, leaving it and its superior viscoelastic properties intact. Finally, the reaction can be performed in water with the HA sodium salt without any previous transformation.
However, there are some side notes to these advantages. If the reaction with the amine is not 100% efficient, the sites where the crosslinker is not connected can still contain NHS or HOBt attached to the carboxyl site. Furthermore, reagents have to be added in excess amounts because of the hydrolysis of EDC, and the amine being in protonated state at the reaction pH. The reaction has been performed in DMSO (Bulpitt & Aeschlimann, 1999; Schneider et al., 2007) with high reported degrees of substitution of 60-80%, suggesting that EDC hydrolysis can be minimized this way.19,20 However, HA
9
is soluble in DMSO only after exchanging the sodium ions for tetrabutylammonium ions through cation exchange21, making the transformation more complicated than in water.
Figure 7 Strategy for the coupling of amines to HA utilizing EDC and NHS or HOBt as proposed by Bulpitt and Aeschlimann.19
Reaction via thiolated HA: michael-addition
The Michael addition is a conjugate addition to an α,β-unsaturated carbonyl.15,22,23 The nucleophilic attack on the β-carbon of the unsaturated bond yields an enolate, which is protonated to form the product. The Michael addition is a highly effective reaction that proceeds under relatively uncomplicated reaction conditions, although generally a catalyst is necessary. In this research project, specifically the Thiol-Michael reaction is employed. The HA functionalized with thiol groups reacts via a Michael addition with PEG functionalized with acrylate groups (see figure 8), in which the Thiol acts as a nucleophile and attacks the acrylate group to form a covalent bond.
A potential disadvantage of this reaction is that the thiol groups can oxidize to form disulfide bonds (see figure 9), crosslinking the HA gel without incorporating the PEGDA crosslinkers. This reaction proceeds under physiological condition and can be induced by exposure to air.24 In 2003, Shu et. al. described the crosslinking of thiolated HA with PEGDA in PBS buffer, without the addition of an additional catalyst.25 The reaction was carried out under Nitrogen atmosphere to prevent oxidation and gel formation took place within 10 minutes.
There are a few routes to synthesize thiolated HA, including the already described carbodiimide chemistry.24 Because the purpose of this study is to develop a relatively simple protocol, ready-made HA with thiol functionality is purchased.
Important advantages of the Michael addition for crosslinking are the mild conditions and the versatility of the reaction. There are many functional groups that can be coupled to thiol groups using the Michael addition. One of the most efficient couplings is that between thiols and maleimide (figure 10).26 The wide availability of maleimide functionalized polymers makes this an attractive route towards HA hydrogels. These molecules are commonly synthesized from maleic anhydride and the amine of the molecule to be coupled to maleimide, via a ring-coupling followed by a cyclization step at high temperature.27 However, this synthesis is beyond the scope of this research.
10
Figure 8 Simplified mechanism for the crosslinking Michael reaction of thiolated HA and PEGDA.
Figure 9 Unwanted S-S bond formation under influence of oxygen.
Figure 10 Structural formula of a maleimide functional group
Design of a DNA-functionalized HA hydrogel: using ssDNA as crosslinker
In order to develop a switchable HA network, or responsive hydrogel, design A from figure 2 was selected. This approach showed promising results in polyAAm hydrogels, with only 0.002 equivalents of hydrogel crosslinker, relative to the number of monomers.4 Hydrogels crosslinked with ssDNA molecules were shown to shrink in size upon addition of the complementary ssDNA strand. In this research we attempt to employ the most effective DNA crosslinker design, of 29 adenine nucleotides, in a HA hydrogel. The complementary strand is composed of 29 Thymine nucleotides. As control, an ssDNA strand of 29 Cytosine nucleotides is used.
For the synthesis of these hydrogels, both the Michael addition and the carbodiimide-mediated reaction could be used. An exact protocol for the attachment of ssDNA to HA using EDC could not be found. However, EDC has been used to couple dsDNA28 as well as ssDNA29 to carboxyl groups. For ssDNA there are potential side reactions between EDC and thymidine residues or at the guanosine N-1 site, which could lead to unfavorable by-products.30 However, the ssDNA in this research only contains adenine nucleotides, avoiding this issue. The Michael addition, more specifically the reaction between maleimide and thiol functional groups, has also shown to be an efficient way to couple different kinds of biomolecules, including DNA.31
11
2.2 Analysis Methods
Fluorescence Recovery After Photobleaching
Fluorescence Recovery After Photobleaching (FRAP) is a technique used to measure diffusion coefficients of fluorescent macromolecules in complex environments, such as crosslinked hydrogels.32 It is important to have a measure of diffusion through the pores of the hydrogel network. In the first place, diffusion constants can be a measure for the degree of crosslinking in a hydrogel. Furthermore, the suitability of a hydrogel for cell encapsulation is in part dependent on its porosity, which determines the diffusion time of metabolites through its polymer matrix.
To measure diffusion in a hydrogel using FRAP, fluorescently labeled molecules need to be dissolved in the sample. A typical FRAP experiment consists of three phases (see figure 11).33,34 In the first phase, the fluorescence of the region of interest (ROI) in the sample is measured with a low-intensity laser beam. The ROI typically ranges from a few micrometers to several dozen. In the second phase, the fluorescent molecules in the ROI are bleached with a short and high-intensity laser pulse. This creates a decrease in the fluorescence in the ROI, and a difference in concentration of fluorescent molecules between the ROI and the rest of the sample. In the third phase, bleached fluorophores will diffuse out of the ROI, while unbleached, still fluorescent molecules will diffuse into the ROI, restoring the fluorescence. A typical FRAP recovery curve is shown in figure 11. This process is measured using a low-intensity laser beam. Two important results can be found from this experiment. By fitting the recovery curve with a suitable model, the diffusion constant D can be found. Furthermore, if a fraction of the fluorophores in the ROI is immobilized by steric effects or interactions with the surrounding hydrogel, the fluorescence will not fully recover to the value before photobleaching. Thus, the results show information about the fraction of immobilized molecules in a sample, on the timescale of the experiment, as well as about the diffusivity of the mobile fraction.
Figure 11 The steps in a typical frap experiment with the corresponding FRAP curve. For t<0, the pre-bleach fluorescence is measured. At t=0 the ROI is bleached with a strong laser pulse. For t>0 the fluorescence is measured until it reaches a steady-state value. Figure taken from ref33.
12 Rheology
Rheology is an appropriate method for the characterization of the mechanical properties of hydrogels because it is quick, sensitive, applicable on relatively small sample sizes (starting from 40 L) and reveals information on the degree of crosslinking and molecular weight.35
Hyaluronic acid based hydrogels are generally viscoelastic materials, exhibiting both a viscous and an elastic response, depending on the time scale and magnitude of the applied deformation. The viscoelastic properties can be measured using small amplitude dynamic rheology, which probes a material under “at rest” conditions and does not influence its microstructure. Dynamic rheology tests, or oscillatory tests, subject the material to a sinusoidal shear strain γ:
γ = γ0 sin(ωt) (1)
where γ0 is the strain amplitude, ω the radial oscillation frequency and t the time. The applied strain results in a mechanical response of the material, called shear stress τ. For viscoelastic materials, there is a phase difference between γ and τ, given by the phase shift angle δ. This phase shift angle can vary between 0 for ideally elastic (or solid) materials and 90˚ for ideally viscous (liquid) materials. The dynamic shear moduli G’ and G’’ characterize the viscoelasticity of a material. The elastic (or storage) modulus G’ is defined as follows:
G’ = σ0 cos(δ)/γ0 (2)
where the component of the stress (σ0) in phase with the strain is divided by the strain. The viscous (or loss) modulus G’’ is given by the component of the stress that has a 90° phase difference with the strain, divided by the strain:
G’’ = σ0 sin (δ)/γ0 (3)
The ratio of G’’ and G’ gives the loss tangent:
tan δ = G’’/G’ (4)
For a perfect liquid, δ = 90°, tan δ = 1, and G’ = 0, while G’’ is finite. Here G’’, the loss modulus, represents the amount of energy that is lost during each oscillation cycle. In a perfect liquid no energy is stored via elasticity, and therefore G’, the storage modulus, is zero. For a perfect solid, δ = 0°, tan δ = 0, and G’’ = 0, while G’ is finite. No energy is lost, all transferred energy is stored in de material through its elasticity. A higher value for G’ means that a material is stiffer. For viscoelastic materials, the phase difference δ lies somewhere between 0° and 90°, and G’ and G’’ describe the stiffness and viscosity of the material.
The above calculations are based on a strain-controlled system. This means that the input is a preset strain, and the output is the measured stress. Conversely, stress-controlled measurements, in which the stress is preset and the resulting strain is measured, are also possible. In the linear elastic regime of small deformations, the outcome of strain/stress controlled measurements is the same. When the stress/strain exceeds a critical value, we enter the nonlinear viscoelastic regime where the hydrogel structure is altered by the applied shear.
13 Frequency sweep
The viscoelastic behavior of hydrogels is dependent on the timescale, or frequency, of the deformation.36 This can be probed using oscillatory frequency sweeps. In these measurements, the gel is exposed to oscillations with a small deformation, well below the critical strain γc , of varying frequency. The dependence of G’ and G’’ on the frequency gives information about the structure of the sample. If G’ is independent of the frequency, the material is solid-like. For a viscoelastic liquid, both G’ and G’’ are dependent on the frequency, with the elastic modulus dominating at high frequencies and the viscous modulus at low frequencies. The more strongly dependent on the frequency G’ is, the more fluid-like the material. When the timescale of the deformation is quick, the material reacts elastically, but when the deformation is slow, the material has time to relax the applied stress by crosslinker unbinding or polymer disentanglement and the viscous response dominates. For typical frequency sweep results, see figure 12.
Figure 12 Typical frequency sweep profile for a viscoelastic solid, a gel and a viscoelastic liquid.37
Crosslinking reaction kinetics
When crosslinked, the properties of hyaluronic acid change from a viscous liquid to a more gel-like structure. This reaction can be monitored in the rheometer over time. From the results of the frequency sweeps a frequency can be selected at which the crosslinking reaction can be followed, using a small strain amplitude within the linear regime. Initially, in the viscous liquid state, G’’ will dominate over G’. When the crosslinks start forming, G’ will start to dominate and will keep increasing until the reaction is finished. When the reaction is complete, both moduli will be constant. This measurement can give valuable information on the crosslinking kinetics and total reaction time.
Swelling experiments
Crosslinked hyaluronic acid hydrogels are able to swell and absorb large amounts of water. The amount of water absorbed at equilibrium depends on the properties of the hydrogel. Furthermore, the ionic strength of the aqueous solution strongly influences the ability to swell.11 It has also been mentioned that gels lose their ability to restore to their original volume after being freeze dried11, although most papers do not report this issue.
14
The equilibrium swelling ratio (Q) is defined as the ratio of the weight of the swollen gel (Ws) to the weight of the dried gel (Wd):
Q = Ws/Wd (5)
It has been reported that differences in crosslinking densities due to the use of different types of crosslinkers or different amounts of crosslinker can be observed using the swelling ratio.38,39 The swelling ratio of a hydrogel decreases with increasing rigidity of the gel. A high crosslinker density could increase the deformability of the gel and thus decrease the amount of water that can be absorbed by the gel. The crosslink density does not influence the pore size, unless it causes inhomogeneities or bundling of polymers, changing the microstructure of the gel.
15
3. Experimental Section
All chemicals were obtained from Sigma-Aldrich, unless otherwise mentioned. Hyaluronic acid was obtained by fermentation of Streptococcus Equii bacteria from Sigma-Aldrich (MW between 1.1–1.6 MDa) and from Lifecore Biomedical (MW ~ 100 kDa). Diamino PEG has a molecular weight of 2000 Da. Oligonucleotides were obtained from biomers.net. The used crosslinker (29 adenine nucleotides) was amino C6 functionalized on both the 3’and 5’end. Due to the different molecular weights of HA all molar quantities have been calculated for the amount of monomer, and thus reactive groups in the HA samples.
Hydrogel synthesis via etherification with epoxides
HA (1 eq monomer, final conc. 20 mg/ml) was dissolved in 1 wt% aqueous NaOH solution, to which BDDE (0.9 eq) was added. Reaction mixture was left overnight at RT to react. Visually, the resulting mixture did not look gel-like, and less viscous than HA and BDDE combined in solution in water, without NaOH.
Hydrogel synthesis carbodiimide chemistry HMW HA
Protocol prod 3
Dissolve HA (10 mg, 1 eq) in 5 ml MES buffer (pH 5.25, 10 mM), add 25 mg diaminoPEG (0.5 eq) and mix until dissolved. Dissolve 12 mg EDC (3 eq) and 10 mg HOBt (3 eq) in 1:1 DMSO:H2O, and add this to the HA-PEG solution. Mix well and leave overnight at RT, stirring. Dialyze in water (MWCO=10000 Da) for 48 hours at RT. Yield was obtained by lyophilization.
Protocol prod 4 & 5
Dissolve HA (10 mg, 1 eq) and diaminoPEG (25 mg, 0.5 eq) in 1.8 mL 1X PBS buffer for prod 4, or in 1.8 mL MES buffer (pH 5.25, 10 mM) for prod 5, and mix until dissolved. Dissolve 12 mg EDC (3 eq) and 10 mg HOBt (3 eq) in 200 uL 1:1 DMSO:H2O, and add this to the HA-PEG solution. Mix well and leave overnight at RT, stirring. Dialyze in water (MWCO=10000 Da) for 48 hours at RT. Yield was obtained by lyophilization. HA PEG diamine Reaction conditions Expected mass of product Measured mass of product Yield Mg of PEG reacted Fraction of PEG reacted Prod 3 10 mg 25 mg ~2 mg/ml in MES 35 mg 17.6 mg 50 % 7.6 0.31 Prod 4 10 mg 25 mg 5 mg/ml HA in PBS 35 mg 20.9 mg 60 % 10.9 0.44 Prod 5 10 mg 25 mg 5 mg/ml HA in MES 35 mg 25.08 mg 72% 15.1 0.60
16 LMW HA
Protocol prod 10
Dissolve LMW HA (10 mg, 1 eq) and diaminoPEG (25 mg, 0.5 eq) in 1.8 mL 1X PBS buffer. Dissolve 12 mg EDC (3 eq) and 10 mg HOBt (3 eq) in 200 uL 1:1 DMSO:H2O, and add this to the HA-PEG solution. Mix well and leave overnight at RT, stirring. Dialyze in water (MWCO=10000 Da) overnight at RT. Yield (according to lyophilization) = 15.8 mg (45%)
Protocol prod 11
Dissolve LMW HA (19.8 mg, 1 eq) and diaminoPEG (54.2 mg, 0.5 eq) in 0.5 mL MES buffer (pH 5.25, 10 mM). Dissolve 24 mg EDC (3 eq) and 24 mg HOBt (3 eq) in 200 uL 1:1 DMSO:H2O, and add this to the HA-PEG solution. Mix well and leave overnight at RT, stirring. The formed gel was very dense, to the point where it could not be transferred into the dialysis frame. Add in total 2.8 mL MQ water to dilute. Dialyze in water (MWCO=10000 Da) overnight at RT. Quite some gel was lost because it was difficult to handle. Yield (according to lyophilization) = 40.8 mg (55%).
Hydrogel Synthesis Michael Addition
Crosslinked HA samples using Michael addition were synthesized using commercially available thiolated hyaluronic acid (Glycosil, 2B Scientific, 240 kDa) and poly(ethylene glycol) diacrylate crosslink (Extralink PEGDA, 2B Scientific, 3.4 kDa). Glycosil was dissolved in water to obtain a concentration of 10 mg/ml in 1x PBS buffer and Extralink was dissolved in water to obtain a 10 mg/ml concentration in 1x PBS buffer. Crosslinking was performed by adding PEGDA to ThiolHA and vortexing the mixture. Crosslinking was performed with different amounts of crosslinker, namely 25%, 50%, 75% and 100%, with 0.5 mol PEGDA per 1 mol of HA monomer for 100% crosslinker.
Hydrogel synthesis with DNA crosslinker
Hydrogels were synthesized according to the above described carbodiimide protocol, with a HA (MW between 1.1–1.6 MDa) concentration of 6.0 mg/ml, 0.36 eq diaminoPEG crosslinker and 0.02 eq DNA crosslinker (29 adenine nucleotides, amino-functionalized) (total 75% crosslinker relative to HA monomers, of which 5% DNA). No hydrogels were formed. Reaction mixture was liquid at RT after the reaction time. Simultaneous control experiments with 0.38 eq diaminoPEG and no DNA did form hydrogels.
Shear Rheology
Rheological measurements were carried out on a stress-controlled Anton Paar MCR 501 rheometer, with a stainless steel cone-plate system with a diameter of 30 mm. The plates were set to a gap of 55 µm and experiments were performed at a constant temperature of 22°C, set by a Peltier system. To minimize the evaporation of water, mineral oil was applied around the samples during all measurements. Frequency sweep measurements were performed for an oscillation frequency range from 0.01 to 10 Hz with a constant strain amplitude of 0.5%.
FRAP
FITC dextrans (70, 500 and 2000 kDa) were mixed in vigorously by vortexing with hydrogel samples at a concentration of 0.1 mg/ml in PBS (1X). 23 µl samples were transferred using a syringe to a cylindrical well in a silicone mold of 2.5 mm diameter on a glass slide and sealed with a glass coverslip. All samples were stored in the dark to prevent bleaching and loss of fluorescence. FRAP measurements were
17
performed using a Nikon A1 confocal microscope with a ROI diameter of 20, unless mentioned otherwise, and the power of the laser on 60% for the bleaching event. On each sample FRAP was performed at RT and repeated at least 3 times on different spots in the sample.
Swelling experiments
Hydrogels were prepared according to the carbodiimide synthesis method as described above, in PBS and with a HMW HA (MW between 1.1–1.6 MDa) concentration of 6.67 mg/ml and 5 mg/ml, and with different amounts of diaminoPEG crosslinker (0.5 eq, 0.38 eq, 0.25 eq, 0.13 eq), at RT. After the overnight reaction, the samples were left at RT to swell in PBS buffer for 72 hours. Afterwards, gels were filtrated through a paper filter (quickly to prevent excess drainage), weighed, freeze-dried and weighed again to investigate the water content of the gels.
18
4. Results & discussion
Hyaluronic acid in solution is in ‘entangled’ form. It generally forms a very viscous gel-like substance. By contrast, crosslinked HA gels are solid-like because of the presence of permanent, chemical bonds that fix the HA chains in place. Probe particle diffusivity in entangled HA and crosslinked HA solutions was probed with FRAP and rheology experiments were performed to characterize the viscoelastic properties of the gels.
FRAP
FRAP was performed on different concentrations of entangled HA with fluorescent FITC Dextrans of different sizes. To measure the diffusion of these fluorescent macromolecules through the gel, the dextrans were mixed in with entangled HA. For crosslinked samples, the dextrans were mixed in before the crosslinking process to incorporate them in the matrix. All samples were kept in the dark until measuring to prevent photobleaching by ambient light. FRAP recovery curves were normalized by setting the average fluorescence of the pre-bleach measuring points equal to 1 and the bleach-time fluorescence to 0. The normalized curves were fitted to a single-exponential decay function:
𝑦 = 𝑦 + 𝐴𝑒 (6)
where y0 is the total recovery, or mobile fraction, and t is the time constant. From the two fitting parameters, the immobile fraction and the half-life t1/2 can be calculated as follows:
𝑖𝑚𝑚𝑜𝑏𝑖𝑙𝑒 𝑓𝑟𝑎𝑐𝑡𝑖𝑜𝑛 = 1 − 𝑦 (7)
𝑡 / = 𝑡 𝑙𝑛(2) (8)
The immobile fraction represents the fraction of dextran molecules that are immobilized within the sample. Based on the t1/2, the diffusion constant of the dextran tracers can be estimated using the Soumpasis equation.40
𝐷 = 0.224𝑟
𝑡 (9)
Here, Rn is the radius of the laser bleach spot (10 µm for standard experiments). According to this equation, faster diffusion corresponds to a lower t1/2.
Rheology
Frequency sweeps with a rheometer were performed to measure the viscosity and elasticity of the gels. Entangled gels have a characteristic relaxation time, where G’ and G” cross, with the elastic modulus dominating at high frequencies and the viscous modulus at low frequencies. For a crosslinked gel, G’ and G’’ are parallel. The rheometer can probe this behaviour over a range from 0.01 to 10 Hz.
19
4.1 Entangled Hyaluronic Acid
We first consider diffusion of dextran tracers in entangled HA solutions. In entangled HA gels the chains can move separately. Their movement is hindered by the other chains but not constrained. Therefore diffusivity of the dextran tracers through the spaces between the chains as well as diffusion of the chains themselves both contribute to the total diffusion. Figure 13 shows that dextrans diffuse progressively more slowly as the concentration of HA is raised from 2 to 8 mg/ml. It also shows that a larger dextran (2000 kDa) diffuses slower than a smaller dextran (500 kDa). This is also supported by figure 16. Rheology measurements shown in figure 15 and table 2 show that denser HA solutions have a higher viscosity. This result is consistent with the FRAP results because the Stokes-Einstein relationship predicts an inverse relation between the diffusion coefficient and the solution viscosity.
Figure 14 and 17 show that the immobile fractions are close to zero, as expected since entangled networks are viscous at long times. It is unclear whether the nonzero values represent true immobilization, for instance due to sticking, or represent an artefact such as sample inhomogeneities or bleaching. The large error bars can point to inhomogeneity. Bleaching by the laser is unlikely, because the laser was used at very low intensity. However, the samples could inadvertently have been exposed to light.
To probe the viscoelastic properties of the solutions, we performed oscillatory shear rheometry applying an oscillatory strain with a small amplitude (0.5%) and varying frequency. The resulting frequency sweep profiles (figure 15) show that entangled HA solutions are more viscous than elastic, and that the elastic modulus increases with a higher concentration. Table 2 shows the complex viscosities that were calculated from G’ and G”. Table 3 shows the apparent viscosity as seen by the fluorescent dextran particles, calculated from FRAP results from figure 16. The apparent viscosities reported by FRAP are 3 to 8 times larger than the viscosity of water (1.00 ⋅ 10-3 at 20°C41), whereas the
macroscopic viscosity, as measured by rheology frequency sweeps, is 103 to 104 times bigger than that of water. This large discrepancy suggests that the motions of FITC dextran molecules report on the local viscosity between the polymers rather than the macroscopic viscosity, which is consistent with their size.
Figure 13 FRAP recovery times measured for dextran tracers of 500 and 2000 kDa molecular weight in entangled HA solutions of 0, 2, 5 and 8 mg/ml. 0 5 10 15 20 25 30 0 mg/ml 2 mg/ml 5 mg/ml 8 mg/ml t1/2 (s ) Conc. HA (mg/ml) 2000 kda 500 kDa
20
Figure 14 Immobile fraction measured by FRAP for dextrans of 500 and 2000 kDa molecular weight in entangled HA solutions of 0, 2, 5 and 8 mg/ml. Concentration Complex viscosity (Pa*s) 5 mg/ml 1.04 10 mg/ml 4.80 20 mg/ml 10.2
Table 2 : Complex viscosity of entangled HA solutions at different concentrations from small amplitude oscillatory shear rheology at an oscillation frequency of 1 Hz.
Size Dextran (kDa) Viscosity (Pa*s)
70 3.31 ⋅ 10-3
500 4.76 ⋅ 10-3
2000 8.23 ⋅ 10-3
Table 3 Apparent viscosity of 5 mg/ml HA calculated from FRAP results from figure 4, using the Stokes-Einstein equation, with the diffusion coefficient estimated following the formula explained above and the stokes radii according to the product information sheet (Table S1).
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0 mg/ml 2 mg/ml 5 mg/ml 8 mg/ml im m ob ile fr ac tio n Conc HA (mg/ml) 2000 kda 500 kDa
21
Figure 15 Frequency sweeps of HA solutions at different concentrations of 5, 10 and 20 mg/ml, showing the elastic moduli G’ (circles) and viscous moduli G’’ (diamonds).
Figure 16 FRAP results for t1/2 and immobile fraction obtained for 70, 500 and 2000 kDa FITC Dextrans in entangled 5 mg/ml HA solution.
Figures S1 and S2 show that the values of t1/2 as well as of the immobile fraction can depend on the bleaching spot size used for the FRAP measurements. In the calculation for the diffusion coefficient (figure 5) the spot size is taken into account, yet still some slight differences can be seen. This could be due to three-dimensional effects.42 The calculation for the diffusion takes the radius of the bleaching spot in account, but not the three dimensional effects within the sample. It is possible that these effects scale differently across different bleach spot sizes. Thus, it is important to use the same bleaching spot size throughout the different experiments, for the results to be comparable. The different spot sizes do yield similar trends: with increased HA concentration, the diffusion constant decreases. The maximal hindrance at 8 mg/ml is about 50%.
0.01 0.10 1.00 10.00 100.00 1000.00 0.01 0.1 1 10 M od ul us (P a) f (Hz) G' 10 mg/ml G" 10 mg/ml G' 5 mg/ml G" 5 mg/ml G' 20 mg/ml G" 20 mg/ml 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0 5 10 15 20 25
70 kda 500 kda 2000 kda
t 1 /2 im m ob ile fr ac tio n
MW of FITC Dextran (kDa)
thalf imm frac
22
Figure 17 Diffusion coefficients of 500 kDa dextran in entangled HA solutions measured from FRAP measurements with different bleach area diameters (10, 20, 35 m), calculated according to Kang et al.43
4.2 Crosslinked Hyaluronic Acid
In crosslinked HA gels the chains are constrained by chemical bonds. Therefore diffusivity through the pores of the matrix is the major contributor to the total diffusion of the dextran tracers. Rheology frequency sweeps will mainly measure the elastic modulus of the crosslinked gel network.
Etherification of HA using epoxides
The experiments functionalizing HA with BDDE under basic conditions did not yield hydrogels. Visually, the product looked more liquid then gel-like, even less viscous and more fluid then entangled HA in solution. Rheological experiments (see figure S7) showed that the crosslinked HA-BDDE product is less elastic than unmodified HA in water. FRAP experiments with FITC dextran (500 kDa) yielded typical t1/2 values of 5 seconds or less, which corresponds most to t1/2 values for solutions without any added HA (figure 13). This does not indicate the formation of a hydrogel either and could possibly point to breaking down of the HA polymers.
The lack of hydrogel formation could be caused by the high pH (>=10) of the reaction. Both literature10 and experiments (figure S7) show that HA is not stable at a high pH. Potentially the reaction could work with careful pH monitoring and adjustments, not letting the pH increase too much above 10, and neutralizing the mixture quickly after the reaction. However, these pH adjustments are complex in very viscous solutions, such as entangled HA or hydrogels, because they do not mix easily and it is difficult to take a liquid sample for pH measurements.
Because of the basic nature of the reaction conditions (pH > 10), the conditions lie far from physiological conditions. Because physiological conditions were formulated as a goal for this research, it was decided not to continue this type of reaction further.
Reaction via thiolated HA: michael-addition
The analysis of FRAP data for the Thiolated HA samples was complicated by the fact that the fluorescence recovery did not level off. Furthermore, the fluorescence recovered to a higher level than the initial pre-bleach fluorescence. An exemplary experiment is shown in figures S5 and S6. Attempts to solve this issue using a PCA/PCD (Protocatechuic Acid and Protocatechuate Dioxygenase) scavenger
0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 0 mg/ml 2 mg/ml 5 mg/ml 8 mg/ml D iff us io n co ef fic ie nt (u m 2/s ) Concentration HA (mg/ml) 10 um 20 um 35 um
23
system, commonly used in fluorescence microscopy to prevent damage to fluorophores by radical oxygen44, were unsuccessful. This could be explained by the conditions of the experiments, which were not anaerobic. Furthermore, oxidative damage would cause a decreasing fluorescence, not an increasing one. To try and limit the effect of the increasing fluorescence, all data was analysed using a ROI significantly smaller than the bleach size. This decreased the effect, but did not completely remove it. For these results, see fig S4 and S5.
Figure 18 and 19 show the t1/2 and immobile fraction obtained by FRAP for diffusion of FITC Dextrans through entangled and crosslinked HA with different amounts of crosslinker. For 70 and 500 kDa dextrans, a small difference can be seen between the different amounts of crosslinker, with a slight peak in both the t1/2 and immobile fraction at the 4:1 HA:crosslinker ratio (v:v at 10 mg/ml). Diffusion is slower through the crosslinked gel than in the entangled ThiolHA. For 2000 kDa Dextrans this effect is reversed: diffusion is faster through the crosslinked gel than in entangled thiolHA.
The immobile fraction for the 2000 kDa dextran is almost zero for entangled HA, yet almost 0.5 for crosslinked HA. This indicates the presence of pores in the crosslinked gel with a smaller size than 27 nm, which is the Stokes radius for a 2000 kDa dextran. For 70 and 500 kDa Dextrans the immobile fraction does not differ clearly between the entangled and crosslinked samples. However, there is a larger immobile fraction for 500 kDa dextrans compared to the 70 kDa dextrans. The precise immobile fractions are difficult to conclude due to the measurement issues discussed above.
The results from figure 18 and 19 indicate that the local viscosity with respect to 70 and 500 kDa dextrans is changed only slightly by the crosslinking. The molecules move slower than in entangled ThiolHA, but no significant immobile fraction is observed. However, for the 2000 kDa dextrans the crosslinking causes the formation of an immobile fraction, combined with an increase in diffusion speed. This indicates the formation of pores that are smaller than 270 Angstrom, but generally larger than 147 Angstrom (table S1). The dextran molecules are only partly immobilized, indicating an inhomogeneous gel is formed. Furthermore, the crosslinkers decrease the local viscosity with respect to the 2000 kDa dextrans.
The frequency sweeps obtained by rheometry in figure 20 show that all samples formed an elastic gel as signified by a frequency-independent elastic modulus, with the 4:1 ratio gel being the strongest.
24
Figure 18 FRAP results for the recovery half time t1/2 for FITC dextrans of different molecular weights (70, 500, 2000 kDa) in thiolHA samples in entangled and in crosslinked form. Crosslinking was performed with different HA-to-crosslinker ratios (v:v at 10 mg/ml) as indicated on the x-axis.
Figure 19 FRAP results for the immobile fraction for FITC dextrans of different molecular weights (70, 500, 2000 kDa) in thiolHA samples in entangled and in crosslinked form. Crosslinking was performed with different HA-to-crosslinker ratios (v:v at 10 mg/ml) as indicated on the x-axis. Concentration of ThiolHA is 7.1 mg/ml.
Figure 20 Frequency sweeps measuring the viscoelastic behaviour of thiolated HA samples (7.1 mg/ml) for varying amounts of crosslinker (v:v at 10 mg/ml) as indicated in the legend. Circles: G’, diamonds: G’’.
0 10 20 30 40 50 60 solvent no linker 8:1 4:1 2:1 t1/2 (s ) sample 70 kda 500 kda 2000 kda -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 solvent no linker 8:1 4:1 2:1 Im m ob ile fr ac tio n
amount of Thiol HA/Crosslinker
70 kda 500 kda 2000 kda 0.10 1.00 10.00 100.00 1000.00 0.01 0.10 1.00 10.00 M od ul us (P a) f (Hz) G' 8:1 G" 8:1 G' 2:1 G" 2:1 G' 4:1 G" 4:1
25
Carbodiimide mediated crosslinking
Table 4 shows that the reaction in MES buffer has a higher yield compared to reaction in PBS, but that the reaction in PBS buffer does proceed as well. It also shows that, comparing product 3 and 5, a higher concentration of reactants gives a higher product yield (higher degree of crosslinking). However, this has a limit as the gel becomes difficult to mix well and pipet at higher concentrations.
HA PEG diamine Reaction conditions Expected mass product Measured mass product Yield Mg of PEG reacted Fraction of PEG reacted Prod 3 10 mg 25 mg ~2 mg/ml in MES 35 mg 17.6 mg 50 % 7.6 0.31 Prod 4 10 mg 25 mg 5 mg/ml HA in PBS 35 mg 20.9 mg 60 % 10.9 0.44 Prod 5 10 mg 25 mg 5 mg/ml HA in MES 35 mg 25.08 mg 72% 15.1 0.60
Table 4 Yield for reactions in MES (prod 5) and PBS (prod 4) buffers.
Figures 21 and 22 show that the 2000 kDa dextran molecules exhibit slower diffusion and a higher immobile fraction than the 500 kDa dextrans, both in gels synthesized in PBS and in MES. According to rheology frequency sweeps, both product 4 and 5 show the profile of a crosslinked gel, not of a viscous liquid. The diffusion of the 500 kDa dextrans at 1.8 mg/ml HA concentration is comparable to diffusion in 2 mg/ml entangled HA (figure 13). However, the diffusion of 2000 kDa dextrans is significantly slower than in the entangled solution. There is a large standard deviation of the mean for the immobile fraction results (figure 22). This points to inhomogeneity of the gels on the scale (ROI = 20 m diameter) of the FRAP measurements.
Figure 24 shows that prod 4 (reaction in PBS) has higher G’ and G” than prod 5 (reaction in MES). Also, it shows that the rheological properties of the gel change significantly after lyophilization and rehydration. There are some reports of different types of hydrogels of which the structure and properties significantly changed after freeze-drying, which is posited to be due to a change in the conformation of the chains caused by lyophilization.45,46 To be sure about this, more research is necessary.
26
Figure 21 FRAP results for the recovery half time t1/2 for dextrans of two molecular weights (500, 2000 kDa) embedded in
three different HA gels: the product of EDC/HOBt crosslinking in MES buffer vs PBS buffer. Prod 5 (1) has the same initial HA concentration as prod 4 (1.8 mg/ml), prod 5 (2) has the same final product concentration as product 4 (3.6 mg/ml). Product 4 was crosslinked in PBS, while product 5 was crosslinked in MES buffer.
Figure 22 FRAP results for dextrans of two molecular weights (500, 2000 kDa) embedded in gels obtained either by EDC/HOBt crosslinking in MES buffer or in PBS buffer. Prod 5 (1) has the same initial HA concentration as prod 4 (1.8 mg/ml), prod 5 (2) has the same final product concentration as product 4 (3.6 mg/ml). Product 4 was crosslinked in PBS, while product 5 was crosslinked in MES buffer.
0 5 10 15 20 25 30
prod 4 prod 5 (1) Prod 5 (2)
t1/2 (s ) sample 500 kDa 2000 kda 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18
prod 4 prod 5 (1) Prod 5 (2)
im m ob ile fr ac tio n sample 500 kda 2000 kda
27
Figure 23 Frequency sweeps showing the viscoelastic moduli of products 4 & 5. Circles: G’, diamonds: G’’.
Figure 24 Rheology measurements of the frequency-dependent shear moduli of prod 4 & 5 at 1 Hz, comparing the response of the samples before lyophilisation, after lyophilization and rehydrated in a small amount of water, and after lyophilization and diluted to the same concentration as before lyophilization. Circles: G’, diamonds: G’’.
Figure 25 shows G’ and G” values for HMW (MW between 1.1–1.6 MDa) and LMW (MW ~ 100 kDa) HA reacted with different amounts of crosslinker (molar percentages). For HMW HA, the strongest gel is formed at 75 mol% (relative to amount of monomers of HA) crosslinker, while for LMW HA, this is for 100 mol% crosslinker. The highest G’ is expected to be measured for the point where most crosslinkers are attached to the HA chains on both sides. More connected crosslinkers should lead to a stronger/more elastic gel. HMW HA already shows elastic properties at a low amount of crosslinker, while LMW HA needs a higher amount of crosslinker to form an elastic gel.
The HMW HA results show a trend where G’ is higher for a gel without crosslinker, than for a gel with 25% or even 50% crosslinker. In figure 15, G’ equals 4 Pa for 5 mg/ml entangled HA, and 22 Pa for 10 mg/ml entangled HA. In figure 25, it equals 16 Pa for 6.7 mg/mL HA. These results indicate that just the addition of EDC/HOBt to HA could have an effect on the elasticity of the material. It is possible that for a lower amount of crosslinker, few actual crosslinks are formed, but the crosslinker molecules do disturb the interactions between the HA chains. However, the experiment in figure S7 shows the
1.00E-01 1.00E+00 1.00E+01 1.00E+02 0.01 0.1 1 10 M od ul us (P a) f (Hz) G' prod 4 G" prod 4 G' prod 5
G" prod 5 - before lyphilization 1
1 10 100 1000
before lyophilization after lyophilization -concentrated
after lyophilization -same conc. as before
lyophilization M od ul us (P a) 4 G' 4 G" 5 G' 5 G"
28
formation of a gel-type substance for HA with EDC/HOBt but without crosslinker present. This points to a significant effect on the physical properties of HA, caused by the reaction with just EDC/HOBt, without added crosslinker. More research is necessary to be able to draw conclusions about this.
Figure 26 shows that the crosslinking of LMW HA is complete after 7 hours. A similar experiment (figure S8) shows the reaction of HMW HA is complete after 12 hours. The final values for G’ and G” turn out much higher for crosslinking in the rheometer than for the gels crosslinked outside of the rheometer. This is probably due to the loading procedure of gels crosslinked outside of the rheometer, in which the gel is broken apart to fit in a small pipet tip and a small space inside the rheometer. By contrast, when the gel is crosslinked in the rheometer, it is measured ‘intact’, without disturbance.
Figure 25 Shear moduli measured at 1 Hz for HA of two different molecular weights (HMW: 1.1–1.6 MDa, LMW: ~ 100 kDa), with varying amounts of PEG crosslinker.
0.10 1.00 10.00 100.00 0% 25% 50% 75% 100% M od ul us (P a)
Amount of crosslinker (% of HA crosslinking sites)
HMW vs LMW HA - G'and G" for varying amounts of
crosslinker - at 6.7 mg/ml HA
G' HMW G" HMW G' LMW G" LMW
29
Figure 26 Crosslinking process followed in the rheometer for LMW HA (30 mg/ml) with different amounts of crosslinker, in percentage relative to number of HA monomers (1 eq HA monomer with 0.5 eq crosslinker = 100%) (see legend), using small-amplitude oscillatory shear measurements at an oscillation frequency of 1 Hz. Inset shows zoom of the blue-boxed region in the main plot, to highlight the development of elasticity at short times.
Swelling experiments were performed on HMW HA crosslinked with 25%, 50%, 75% and 100% crosslinker, where 100% crosslinker corresponds to 0.5 equivalents for 1 equivalent of HA monomers. The equilibrium swelling ratio (Q), defined as the ratio of the weight of the swollen gel (Ws) to the weight of the dried gel (Wd), is reported in table 5. These results show that at a higher concentration HA during the reaction, a more absorbent hydrogel is formed. This corresponds to table 4, which shows that a higher concentration gives a higher yield. To be able to distinguish a trend in the different degrees of crosslinking, more research is needed. In any case, these HA hydrogels can absorb a large amount of water, around 52 – 86 times their own weight.
Amount of crosslinker (diaminoPEG)
Q
(HA initial conc. 7 mg/ml)
Q
(HA initial conc. 5 mg/ml)
100 % 71 52
75 % 65 60
50 % 75 59
25 % 86 68
Table 5 Q values from swelling experiments for different reaction conditions of the carbodiimide mediated crosslinking of HMW HA (initial concentrations of 7 and 5 mg/ml) with diaminoPEG. Q is the equilibrium swelling ratio, the ratio of the weight of the swollen gel (Ws) to the weight of the dried gel (Wd).
Reactions with DNA as crosslinker did not form hydrogels. It could be seen by the naked eye that the reaction mixture had turned completely liquid after the reaction time, indicating that no hydrogel was formed and the naturally high viscosity of HA was disturbed, either by intermolecular interactions or degradation of HA. The reaction was done in PBS (pH = 7.4). At this pH it is not expected for either HA or ssDNA to break down.
The DNA crosslinkers have a high molecular weight, meaning that 0.02 eq of DNA crosslinker and 1 eq of HA, means that for 1 gram of HA, 0.5 grams of DNA have to be added. This could have a
0 200 400 600 800 0 100 200 300 400 500 600 700 G' LMW 100% G'' LMW 100% G' LMW 75% G'' LMW 75% G '/G '' (P a) Time (min) -20 0 20 40 60 80 0 50 100 G ' L M W 1 00 % ( P a ) Time (min)
30
negative effect on the formation of the hydrogel and the viscosity of HA. However, the addition of 2 gram diamino PEG (approximately 0.4 eq) per 1 gram HA in a typical reaction, did not prevent the formation of hydrogels.
It is possible that an unforeseen chemical side-reaction between EDC and the DNA or directly between DNA and HA has taken place. However, the aforementioned possible known side-reactions should have been eliminated by the use of Adenosine nucleotides.
More research is needed to draw clear conclusions about this reaction. For further experiments, HA-ssDNA hydrogels could be synthesized using the Michael addition and maleimide-functionalized ssDNA.
31
Conclusion
In this research project it was attempted to create Hyaluronic Acid-based hydrogels employing carbodiimide chemistry for amidation at the carbonic acid group and a Michael-addition using thiolated HA. The properties of the resulting hydrogels were investigated using rheology to determine the viscoelastic properties of the gel network and FRAP (fluorescent recovery after photobleaching) to measure solute diffusivity and to estimate the porosity of the network. Furthermore, initial experiments were carried out to synthesize HA hydrogels with responsive DNA crosslinkers. Crosslinking of HA hydrogels was achieved using both reaction types. The carbodiimide chemistry was selected for further investigation, but did not succeed in synthesizing HA hydrogels with DNA crosslinkers.
For entangled HA in PBS buffer, it was found that solutions with a higher concentration of HA have a higher viscosity. Entangled HA solutions are more viscous than elastic up to high concentrations, although the elastic modulus does increase with higher concentration. On the timescale of the FRAP experiments, the measured immobile fractions of dextran probes are close to zero, as expected. It is important to use the same bleaching spot size throughout the different experiments, for the results to be comparable.
The etherification of HA using epoxides did not produce hydrogels in our experiments. Because of the non-physiological reaction conditions and complex pH management, this reaction is not advised for further investigation for the specific purpose of crosslinking HA with DNA. For different type of reactions, better pH management could lead to better results.
For the crosslinking of HA via Michael addition of thiolated HA and acrylated PEG molecules, hydrogels were formed that exhibited elastic behaviour in rheology frequency sweeps. It was found that for 70 and 500 kDa dextran probes, diffusion is slower in the crosslinked gels than in the entangled ThiolHA solutions. For 2000 kDa Dextrans this effect is reversed: diffusion is faster in the crosslinked gels than in the entangled thiolHA solutions. Under all conditions, the 70 kDa dextrans had the highest, and 2000 kDa the lowest diffusion. Only for 2000 kDa dextrans a significant increase in immobile fraction was seen from entangled solution to crosslinked gel, not for 70 and 500 kDa dextrans. Possibly, the formed hydrogels are inhomogeneous. The immobile fraction results indicate the presence of pores in the crosslinked gel with a smaller size than 27 nm, but generally larger than 15 nm. The 2000 kDa molecules could be trapped in these pores, while the molecules outside the pores are able to diffuse relatively freely. For smaller dextrans, the crosslinks slow down the diffusion slightly, but do not cause significant immobility.
The crosslinking of HA via the carbodiimide-mediated reaction with PEG-diamines yielded hydrogels that exhibited elastic behaviour in rheology frequency sweeps. The reaction in MES buffer had a higher yield, but the reaction in PBS buffer, thus under physiological conditions, proceeds satisfactory as well. A higher concentration of reactants gives a higher yield and a more absorbent gel. An influence of EDC/HOBt, without the addition of crosslinker, was observed on the physical properties of HA. More research into this effect is necessary. HMW HA exhibits elastic properties at a significantly lower amount of crosslinker than LMW HA. The HA hydrogels crosslinked via the carbodiimide-mediated reaction were shown to be able to absorb up to 86 times their own weight in water. It was also shown that the rheological properties of the gel change significantly after lyophilization and rehydration, which indicates there are still challenges in work-up of the synthesized gels.
32
Reactions with amidated ssDNA as crosslinker did not form hydrogels, and possibly degraded the structure of HA. More research is needed to draw clear conclusions about this reaction.
The carbodiimide mediated reaction and Michael addition both provide viable reaction paths for the synthesis of HA hydrogels. The Michael addition is quick and easy, proceeds under physiological pH, but the functionalization of HA with thiol groups is complex and time-consuming. Thiolated HA can also be purchased, but this limits the number of available molecular weights of HA. Furthermore, maleimide-functionalized DNA is much more expensive than amine-functionalized DNA. The carbodiimide mediated reaction is very versatile since it proceeds from the water-soluble HA sodium salt and an amine-functionalized crosslinker in one reaction step. However, the effect of the carbodiimide on HA without added crosslinker has to be investigated in more depth, and when reacting with DNA, no hydrogel is formed and HA appears to be degraded. Thus, for a proof of concept of being able to create a responsive hydrogel based on HA and ssDNA, I would advise to use purchased thiolated HA with maleimide-functionalized ssDNA. For a more versatile and cost-effective protocol the carbodiimide reaction could be an option, provided a solution is found for the issues encountered in this research.
33
Supplementary Information
Figure S1: Investigation of the effect of the FRAP bleach spot size (see legend) on recovery half time results, for 500 kDa FITC-dextran probes in entangled HA solutions at different concentrations.
Figure S2: Investigation of the effect of the FRAP bleach spot size (see legend) on immobile fraction results, for 500 kDa FITC-dextran probes in entangled HA solutions at different concentrations.
Figure S3: Recovery half time t1/2 for FITC-dextran probes of different sizes (see legend) in thiolHA (240 kDa, 7.1 mg/mL) in entangled and in crosslinked form, with different HA-to-crosslinker ratios (v:v at 10 mg/ml). Alternative analysis, using a ROI of 7.3 um placed within a bleach spot of 20 m.
0 10 20 30 0 mg/ml 2 mg/ml 5 mg/ml 8 mg/ml t1/2 (s ) concentration HA (mg/ml)
FRAP t1/2 spot size investigation, 500 kDa
dextran
10 um 20 um 35 um -0.02 0 0.02 0.04 0.06 0.08 0.1 0 mg/ml 2 mg/ml 5 mg/ml 8 mg/mlFRAP imm frac spot size investigation, 500
kDa dextran
10 um 20 um 35 um 0 20 40 60 80 no linker 8:1 4:1 2:1 t1/2 (s )amount of thiolHA : crosslinker
FRAP t1/2 entangled vs crosslinked ThiolHA, alt. method
34
Figure S4: FRAP results for the immobile fraction of FITC-dextran probes of different sizes (see legend) in thiolHA in entangled and in crosslinked form, with different HA-to-crosslinker ratios. Alternative analysis, using a ROI of 7.3 m placed within a bleach spot of 20 m (see figure S5 for placement of the ROI).
t = 0 / bleach time
t = full recovery
Figure S5: Confocal images taken from a typical FRAP experiment on crosslinked Thiol HA samples. It can be seen that in the area around the ROI, the fluorescence unexpectedly increases relative to the pre-bleach intensity. This increase in fluorescence distorts the recovery curve (See Fig. S6).
-0.1 0 0.1 0.2 0.3 0.4 0.5 no linker 8:1 4:1 2:1
FRAP imm frac entangled vs crosslinked ThiolHA, alt. method
70 kda 500 kda 2000 kda
0 0.2 0.4 0.6 0.8 1 1.2 -100 0 100 200 300 400 500 600 Fl uo re sc en ce Time (s)
FRAP of 70 kDa FITC dextran in ThiolHA:Extralink
2:1 hydrogel
35
Figure S6: FRAP recovery curve of experiment corresponding to the confocal images in Figure S5, showing that the final fluorescence intensity is larger than the pre-bleach intensity.
Figure S7 Frequency sweep comparing HA-BDDE crosslinking product (20 mg/ml HA) with entangled HA (10 and 20 mg/ml) in water (see legend).
0.00 0.00 0.01 0.10 1.00 10.00 100.00 1,000.00 0.01 0.10 1.00 10.00 M od ul us (P a) f (Hz)
Frequency sweep HA - BDDE crosslinking product
G' HA (20 mg/ml) G'' HA (20 mg/ml) G' HA-BDDE (20 mg/ml) G'' HA-BDDE (20 mg/ml) G' HA (20 mg/ml) pH>10 G'' HA (20 mg/ml) pH>10
36
Table S1: Stokes radii and radii of gyration (in units of Angstroms) for FITC Dextran molecules of different molecular weights (in Daltons) as provided by Sigma-Aldrich47.
Figure S8: Crosslinking process followed in the rheometer for LMW HA (30 mg/ml) using small-amplitude oscillatory shear measurements at an oscillation frequency of 1 Hz. Inset: zoom of the boxed region in the main plot. The graph shows two separate crosslinking reacions. #1 = 75% CL (0.38 eq CL for 1 eq of HA monomer), #2 = 0% CL (no crosslinker added, with EDC&HOBt).
-100 0 100 200 300 400 500 600 700 800 0 5 10 15 20 25 G' #1 G'' #1 G' #2 G'' #2 G '/G '' (P a ) Time (min) 0 20 40 2 4 G ' # 1 ( P a) Time (min)
