Injectable bottlebrush hydrogels with tissue-mimetic mechanical properties

Injectable hydrogels are desired in many biomedical applications due to their minimally invasive deployment to the body and their ability to introduce drugs. However, current injectables suffer from mechanical mismatch with tissue, fragility, water expulsion, and high viscosity. To address these issues, we design brush-like macromolecules that concurrently provide softness, firmness, strength, fluidity, and swellability. The synthesized linear-bottlebrush-linear (LBL) copolymers facilitate improved injectability as the compact conformation of bottlebrush blocks results in low solution viscosity, while the thermoresponsive linear blocks permit prompt gelation at 37°C. The resulting hydrogels mimic the deformation response of supersoft tissues such as adipose and brain while withstanding deformations of 700% and precluding water expulsion upon gelation. Given their low cytotoxicity and mild inflammation in vivo, the developed materials will have vital implications for reconstructive surgery, tissue engineering, and drug delivery applications.


Fig. S5.
NMR of representative LBL, system 1 sample. Degree of polymerization for backbone and linear blocks was determined via 600MHz NMR Bruker at UNC core facility (LBL -system 1).

Fig. S9.
Cooling cycle and gel reversibility (LBL). Temperature sweep of samples with various degree of polymerization of back bone and linear block under cooling cycle. X-axis shows temperature varied from 40 to 20 ° at the rate of 1 (° /min) and y-axis represents the storage ( ′ − ) and loss ( ′′ − ) modulus in Pa. The transition point from gel to liquid state is highlighted with yellow filled circles. The experiment is done at three different concentrations of 5 (purple), 10 (blue) and 20 (green) wt.%. Samples are represented with [ , ].  (Table 1), synthesized with three step combination of RAFT and ATRP (LBoBL -system 2) -X-axis shows temperature varied from 25 to 40 ° at the rate of 1 (°/min) and y-axis represents the storage ( ′ − ℎ ) and loss ( ′′ − ) modulus in Pa.

Fig. S11.
The gelation process is fully reversible upon heating-cooling cycles (LBL). Sample: 5 wt% solution of =550, =0.125 triblock. Temperature from 20 ° to 40 ° at 1 Hz. Inset: complex viscosity at 1 Hz over 75 minutes upon three consecutive heating-cooling cycles. Sample is heated rapidly to 45 ° (closely following the instrument temperature overshoot to about 48 °) followed by prompt cooling.

Fig. S12.
Strain endurance (forward cycle, LBL system 1). Strain sweeps of samples with different composition and concentration at physiological temperature of 37 (°C). Polymer is initially at gel state then shows liquid-like behavior under higher strains. Measurements were done at three frequencies of 0.1, 1 and 10 Hz. Strain applied from 10 −2 to 10 4 % (x-axis is limited to [1, 1000]%) and y-axis represents the storage ( ′ − ) and loss ( ′′ − ℎ ) modulus in Pa. The transition point from gel to flow state occurs where loss moduli take over the storage moduli.

Fig. S13.
Strain endurance (reverse cycle, LBL system 1). Strain sweeps of samples with different composition and concentration at physiological temperature of 37 (°C). Polymer is initially at gel state then shows liquid-like behavior under higher strains. Measurements were done at three frequencies of 0.1, 1 and 10 Hz. Strain applied from 10 4 to 10 −2 % (x-axis is limited to [1, 1000]%) and y-axis represents the storage ( ′ − ) and loss ( ′′ − ℎ ) modulus in Pa. The transition point from gel to flow state occurs where loss moduli take over the storage moduli.

Fig. S14.
Dynamic crosslinking of tissue-mimetic elastomers (LBL system 1). Compound 2 is the PEGbrush bbCTA (Scheme 1 -system 1). The triblock is functionalized with copolymerization of HEMA (compound 7, <5 mol%) during the second step of RAFT and is modified via postpolymerization. The LBL was under reflux for 8 hours with excess AIBN to remove CTA end groups and cap both chain ends. Compound 7 is treated with maleimide-isocyanate (8) and furfuryl isocyanate (9) Diels-Alder adducts in two separate reactions to yield 10a and 10b, respectively. The two polymers with functionalized moieties are then mixed to afford linear domain crosslinked bottlebrush elastomer (11). The dashed box schematically shows the phase separation of Ldomains (and magnification on a single domain) at dry state followed by DA crosslinking among linear chains. Black, green and blue strands show the brush backbone, PNIPAM linear chains and functional groups, respectively.  UV crosslinking of LBL, system 1, hydrogels. UV cured hydrogels upon phase separation (gelation) above LCST. Base, EGM, UV-1 and UV-2 represent for different uniaxial stress-strain tests on PNIPAM-bbPEG-PNIPAM, RAFT end group removed PNIPAM-bbPEG-PNIPAM, 3.5% and 7% crosslinking, respectively. To investigate the crosslinking effect on network structure, we conduct comparative tensile tests and SAXS measurements of physically and chemically crosslinked elastomers with the same LBL composition. The obtained stress-strain curves closely match at moderate elongation < 2, suggesting that the chemical crosslinking inside the L-domains do not perturb the overall network topology ( fig. S18A). The small increase in stiffness is attributed to the decrease in the L-domain size due to introduction of the F and M groups to linear blocks ( fig. S18B), resulting in a higher number of crosslink junctions at the same LBL composition.

Mechanics of dry elastomers (LBL). (A)
The true stress-elongation response of PNIPAM-bbPEG-PNIPAM LBL elastomers strongly depends on the B-block degree of polymerization ( ) at same L-block volume fraction ≅ . . The LBL elastomers with =274 and 880 closely match the deformation response of porcine aorta and skin tissues, respectively (hollow symbols). (B) Effect of the L-block volume fraction on the stress-elongation response at constant backbone length with = . The increase in enhances both the strain-stiffening behavior and strength ( ).

Theoretical analysis of gelation temperature
The effect of the PNIPAM block DP on the gelation temperature could be rationalized on the basis of scaling arguments. At low concentrations of triblock copolymer in solution, gelation/network formation might be associated with micellization of PNIPAM-bbPEG-PNIPAM, triggered by temperature-induced condensation of linear (PNIPAM) blocks. An increase in the solution temperature T leads to aggregation of PNIPAM blocks in the cores of spherical micelles, interconnected by bridging bbPEG blocks. The number of PNIPAM blocks in the core of micelle (the aggregation number ) can be estimated using the scaling theory of polymer micelles. Due to low content of PNIPAM in the copolymer micelles are starlike, with thickness of PEG corona larger than the core size. In the vicinity of ≈ 32° for pure PNIPAM solution, the PEG monomers have positive but small second virial coefficient, which indicates that central PEG block is close to theta-solvent conditions. In this case, the aggregation number in starlike micelle is regulated by the length of the condensed PNIPAM block and also depends on the solution concentration . The scaling model of spherical micelles predicts three different regimes of micelles formed by AB diblock copolymer with linear blocks. We assimilate of PNIPAM to NB of core-forming block B, and implement theta-solvent conditions for bbPEG (soluble block A). Because side chains length in the bbPEG block is small ( = 9), in scaling terms it can be treated as linear, and /2 of bbPEG block is related to of block A as ~ −1~. In a dilute solution of micelles (below * ~1 /5 −1/2 ), the aggregation number does not depend on , and depends on the parameters of the corona (bb) block only logarithmically. With omitted logarithmic prefactor, the power law dependence for micelle aggregation number is given by ( 1) where is surface tension (surface free energy of unit area at the core/corona interface, expressed in units of ), and is volume fraction of PNIPAM monomers in the core of micelle.
At concentrations * * > > * with * * ~− 1/5 the omitted in eq 1 prefactor incorporates ratio ( / * ) indicating logarithmically weak increase in , while at > * * the scaling dependence for Q changes to ~ and micelles start to grow with increasing solution concentration, c, reaching its maximal size in melts with ~~1. Near , the volume fraction of PNIPAM monomers in the core is relatively low. Like in the case of infinitely large globule with positive third virial coefficient of monomer-monomer interactions, be specified via the second virial coefficient of binary monomer-monomer contacts, Notably, ( ) accounts for the temperature-dependent hydration of PNIPAM monomeric units. In water solution of PNIPAM macromolecules with infinite DP, is specified by ( ) = 0. However, the presence of PEG in PNIPAM-bbPEG-PNIPAM triblock copolymer can shift sign reversal of ( ) to temperature = 0 ≠ , to change ( ) to ′ ( )~− ( − 0 )/ = − , and specify ~. Here, superscript in ′ signifies the second virial coefficient of PNIPAM monomers incorporated in the triblock copolymer. The surface tension at the surface of PNIPAM core is then linked to as ~2. By substituting ~2 and ~ in eqs 1 and 2, one finds ~{ 8/5 4/5 < * * ~ 3/5 −1/5 > * * If the gelation threshold temperature, , is associated with onset of micellization in the solution, then formation of micelle with ~ 1 specifies = ( − 0 )⁄ as To quantitatively compare the experimental ( ) to eq 4, one should estimate 0 corresponding to '( 0 ) = 0. In the mean-field framework, ' ( ) can be approximated as Here, ( ) is the second virial coefficient of monomer-monomer interactions in the solution of infinitely long PNIPAM macromolecules at temperature (in the absence of PEG), while Δ ( ) is the contribution due to plausible interactions between PNIPAM and PEG monomer units. By expanding ( ) in Taylor series with retention of only linear in ( 0 − ) term one finds Notably, ( ) = 0, while the derivative ( ) is negative.
By using the condition '( 0 ) = 0, one finds If Δ ( 0 ) is positive, then 0 > , while if it is negative, then 0 < . A negative value of Δ ( 0 ) corresponds to an affective attraction/complexation between PNIPAM and PEG below . Because hydrogen bonding is energetically favorable at low temperatures, attraction could (hypothetically) arise if PNIPAM and PEG form hydrogen bonds with the same water molecule or between PEG backbone and pendant groups of PNIPAM. As it follows from Fig. 4C, gelation temperature for several experimentally investigated samples is found below of PNIPAM (32°), indicating a possible shift in ( 0 − ) < 0. To the best of our knowledge there is no theoretical model relating 0 to composition of PNIPAM-bbPEG-PNIPAM block copolymer or its concentration in the solution, and we implement 0 as an adjustable parameter. In fig. S20, we present − 0 as a function of for three solution concentrations (5, 10, and 20 wt.%) in − coordinates. The data for all three solution concentrations collapsed on a master curve if values of T0 were adjusted as 31.5 ° (5 wt.%), 29 ° (10 wt.%), and 27 ° (20 wt.%). The slope of master curve −1/2 was in good agreement with the theoretical predictions in eq. 4 for the interval of solution concentrations < * * . Therefore, gelation and micellization of PNIPAM-bbPEG-PNIPAM block copolymer could be linked due to plausible association of PNIPAM and PEG monomer units below .       ∶ degree of polymerization of linear domains (PNIPAM).
∶ volume fraction of linear chains. : strain-stiffening (or firmness) parameter acquired from fitting the true stress verses strain curves. = 3 : structural modulus acquired from the fitting equation.
: fitting intervals used to fit the stress-strain curves. 0 , and 0 , are Young's modulus from fitting stress-strain curves with the equation of state (eq S8) and experimental measurements as a slope at zero strain, respectively.
Critical strain at which storage moduli equates loss moduli extracted from figs. S12 and S13. Measurements were done at three frequencies of 0.1, 1, and 10 Hz for Forward (increase in strain from 10 -2 to 10 4 ) and Reverse strain, (%) (decrease from 10 4 to 10 -2 ). For the description of headers refer to table S1.

Entry
Mn, GPC Mw/Mn Nomenclature