A Selective, Hydrogel‐Based Prodrug Delivery System Efficiently Activates a Suicide Gene to Remove Undifferentiated Human Stem Cells Within Neural Grafts

The directed differentiation of human pluripotent stem cells (hPSCs) into defined populations has advanced regenerative medicine, especially for Parkinson's disease where clinical trials are underway. Despite this, tumorigenic risks associated with incompletely patterned and/or quiescent proliferative cells within grafts remain. Addressing this, donor stem cells carrying the suicide gene, thymidine kinase (activated by the prodrug ganciclovir, GCV), are employed to enable the programmed ablation of proliferative cells within neural grafts. However, coinciding the short half‐life of GCV with the short S‐phase of neural progenitors is a key challenge. To overcome this, a smart hydrogel delivery matrix is fabricatedto prolong GCV presentation. Following matrix embedment, GCV retains its functionality, demonstrated by ablation of hPSCs and proliferating neural progenitors in vitro. A prolonged GCV release is measured by mass spectrometry following the injection of a GCV‐functionalized hydrogel into mouse brains. Compared to suboptimal, daily systemic GCV injections, the intracerebral delivery of the functionalized hydrogel, as a “one‐off treatment”, reduce proliferative cells in both hPSC‐derived teratomas and neural grafts, without affecting the graft's functional unit (i.e., neurons). It is demonstrated that a functionalized biomaterial can enhance prodrug delivery and address safety concerns associated with the use of hPSCs for brain repair.


Introduction
In both preclinical and clinical studies, the transplantation of neural progenitors has been shown to effectively replace neural circuitry lost to disease or injury.Among the most advanced of these are for Parkinson's disease (PD), where fetal-derived dopaminergic (DA) progenitors have shown structural integration in the brain and alleviation of motor deficits. [1]More recently, research has shifted to human pluripotent stem cells (hPSCs) as the superior donor sourceproviding control over standardization and greater availability.However, even though differentiation protocols are increasingly refined, there are still technical challenges.Additionally, while the process yields large volumes of donor cells with high proportions of the correctly specified progenitors, the pluripotent/proliferative origin of the cells (responsible for neural overgrowth and generation of unwanted cell types and functions) continues to raise concerns for regulatory groups.
While cell sorting strategies have been heavily investigated to improve the purity of the donor material and reduce risks of neural overgrowth/tumors within grafts in PD model, [2] the technique does not account for cells that evade the sorting process, progenitors that fail to terminally differentiate following engraftment, or quiescent stem cells that emerge post-engraftment.To overcome these challenges, suicide genes have been introduced into the donor cells to enable the removal of unwanted cells before or after transplantation.[5] The suicide mechanism is activated by the enzymatic triphosphorylation of the FDA-approved prodrug ganciclovir (GCV) into a toxic form that can integrate into replicating DNA during S-phase and induce apoptosis in proliferating cells.Utilizing this system, we recently demonstrated that timely activation of the "suicide switch" could reduce proliferative cell number and control neural graft size while improving the DA neuron purity and retaining functionality of the graft in an animal model of PD. [3] Nonetheless, the efficiency of the system was hindered by the short half-life of the systemically administered GCV (<2 h in mice after intraperitoneal injection), [6] coupled with the relatively short length of S-phase in proliferating human neural progenitors. [3,7]he safety challenges of drug administration, inclusive of GCV administration, include high doses, repeated dosages, low bioavailability, short half-lives, poor target selectivity, and offtarget systemic toxicities.In this regard, hydrogel-based (i.e., highly hydrated) biomaterials can be utilized to encapsulate a drug, forming a spatially confined reservoir for its local and sustained release in vivo, while acting as a screen to reduce offtarget adverse effects. [8]GCV is most commonly used to treat cytomegalovirus infections, namely retinitis. [9]Attempts to improve intravitreal administration of GCV (and its analogs) have been made using hydrogels, notably a thermosensitive synthetic polymer, [10] and later a naturally derived collagen hydrogel containing silica nanoparticles. [11]Both of these systems demonstrated the capacity to improve the bioavailability and half-life of the drug.
Here, we sought to examine whether an injectable biomaterial may similarly aid in targeted and sustained GCV delivery within neural grafts by employing a shear reversible, self-assembling peptide (SAP) hydrogel.Previous studies have shown nanofibrous structure formation in SAP hydrogels, with subsequent modifications to enable biological signaling by the addition of a second component or the decoration of the core sequencesee review. [12]These systems however, lack an inherent delivery mechanism beyond steric hinderance of drug conjugates.Here, in a key technological advance, we utilized the peptide sequence isoleucine-lysine-valine-alanine-valine (IKVAV) of the 1 subunit of laminin -the brain's major extracellular matrix (ECM) protein essential for cell adhesion and neuronal differentiation [23] -not only as decoration or for biochemical function, but also as the structure-forming unit.Hence, no "extras" are required to make the hydrogel, avoiding off-target effects while offering the simplicity required to speed regulatory approval and ultimately clinical translation.As a result, we have developed a low molecular weight molecule to perform a triumvirate of functions: our advanced hydrogel can uniquely and powerfully determine i) scaffolding microstructures that form a network capable of mimicking key features of the ECM in the brain to support neural grafts, [13,14] ii) biochemical signaling by presenting the IKVAV peptide on the surface of the nanofibrous structure, and iii) compound entrapment via interactions with the gel for its prolonged and controlled release. [15,16]To our knowledge, our recently pioneered biomaterial is the only SAP hydrogel capable of this.Moreover, the underlying physicochemical structure means these hydrogels present with inherent biocompatibility, tuneable properties, and capacity to flow and gelate in situ following in vivo administration.
Uniquely for the application in this study, we hypothesized that the GCV can interact with the structures of the SAP hydrogel in a supramolecular fashion via complementary, yet reversible, interactions to bind and release the prodrug over time.Here, for the first time we demonstrated sustained delivery of GCV from an optimized SAP hydrogel, and retained prodrug function when utilized in vitro or in vivo to ablate proliferative pluripotent stem cells (PSCs) or proliferating neural progenitors.This drug delivery approach is of clinical importance as it provides a "one-off treatment" to address safety concerns associated with the use of PSCs in regenerative medicine.

Fabrication of Self-Assembling Peptide Hydrogels & Ganciclovir Functionalization
SAP hydrogels, presenting part of the 1 subunit of laminin epitope (IKVAV), were fabricated using solid-phase peptide synthesis, as previously described. [17]Aspartate amino acid (D) residues were added to the N (and/or C)-terminus of the peptide to lower the pK a .This altered pK a enables spontaneous self-assembling of the peptides to occur under physiological conditions (pH 7.4).The resultant Fmoc-SAP hydrogels -i) Fmoc-DDIKVAV, ii) Fmoc-DIKVAVD, and iii) a 1:1 mix of Fmoc-DIKVAVD + Fmoc-DDIKVAV -were gelated to a final peptide concentration of 15 mg mL −1 .In brief, 10 mg of Fmoc-SAP powder was suspended in 100 μL deionized water (DI H 2 O) and fully dissolved by minimal addition of sodium hydroxide (NaOH, 0.5 m) while vortexing.The self-assembly was then induced by a pHswitching method using the dropwise addition of hydrochloric acid (HCl, 0.1 m) while vortexing.PBS was then added to achieve the final peptide concentration.For preparing the mixture of Fmoc-DIKVAVD+Fmoc-DDIKVAV, the Fmoc-DDIKVAV and Fmoc-DIKVAVD hydrogels were mixed at a 1:1 ratio after gelation.

Initial Comparison of Hydrogels
For initial comparison of the composite hydrogels, maximal GCV (dissolved in PBS) was loaded into the hydrogels after gelation only by adding the GCV solution in the hydrogel.The resultant GCV concentrations were 4.5 mg mL −1 in the Fmoc-DDIKVAV gel, 8 mg mL −1 in Fmoc-DIKVAVD and 4.5 mg mL −1 within the 1:1 mixture of Fmoc-DDIKVAV + Fmoc-DIKVAVD gel.

Optimized Hydrogel Fabrication for In Vitro and In Vivo Experiments
In an effort to further increase the GCV concentration, GCV was added to Fmoc-DIKVAVD during as well as after gelation.In brief, the GCV solution was added to the Fmoc-DIKVAVD hydrogel during the gelation by adding 16.5 mg of GCV to 18 mg of Fmoc-DIKVAV in the suspension step (DI H 2 O addition) and self-assembled in physiological condition.After gelation, another 11 mg of GCV was added to the resultant composite hydrogel.HCl was added dropwise to reach physiological pH and gel formation to the final volume of 1380 μL.The total concentration of GCV in the final Fmoc-DIKVAVD hydrogel after fabrication was 20 mg mL −1 .Fmoc-DIKVAVD hydrogel (without drug) at 13 mg mL −1 , matching the final peptide concentration as the composite hydrogel incorporating GCV, was also used for in vitro and in vivo experiments.

Transmission Electron Microscopy (TEM)
Negative stain TEM analysis was conducted using a Hitachi H7100FA with a tungsten filament fitting at 100 kV.Formvarcoated copper grids were glow-discharged at 15 mA for 60 s, immersed in the 20 μL of hydrogel droplet sample for ≈30 s, and then immersed in DI H 2 O twice for washing.The grids were then briefly placed in one drop of uranyl acetate contrast-enhancing agent solution (UA, 2%, 20 μL), stained and blotted, followed by immersion in a second drop of UA for 30 s and an immersion in DI H 2 O as a final wash.Excess liquid was blotted off after every step using Whatman paper.Before TEM imaging, the grids were left to dry for >2 h.

Fourier Transform Infrared (FTIR) and Circular Dichroism (CD) Spectroscopy
FTIR spectroscopy was conducted to confirm anti-parallel -sheet formation in the different libraries of prepared SAP hydrogel samples (≈20 μL) using an Alpha Platinum Attenuated Total Reflectance FTIR (Bruker Optics, Germany).CD analysis was performed to further determine the secondary structure of the fabricated hydrogels using a Chirascan CD Spectrometer.Here, 10 μL of each hydrogel sample was diluted 1:50 in DI H 2 O and filled gently in a quartz cuvette with a 1 mm pathlength.Air bubbles were carefully removed.Chirascan software was used to analyze and smooth the data post-acquisition.DI H 2 O was considered as the baseline and subtracted from the hydrogel samples in both CD and FTIR spectra.

Zeta Potential Measurements for Surface Charge
The -potential of Fmoc-DDIKVAV, Fmoc-DIKVAVD, and their mix (1:1) was measured using a Zetasizer (Nano ZS (Redbadge), ZEN3600, Malvern Instruments, Malvern, UK).Each hydrogel was resuspended in a total solution of 1 mL of DI H 2 O (dilution 1:100) and then loaded into a capillary zeta cell with a syringe, removing any air bubbles.Measurements were analyzed with the Zetasizer Marvel Software (Malvern Instruments).

Rheology
To confirm viscoelastic behavior of the fabricated hydrogel, a Kinexus Pro+ Rheometer (Malvern) was used to conduct the rheological analysis.The test was performed using flat plate geometry (20 mm with solvent trap, Upper Geometry: PU20 SR1351 SS, Lower Geometry: PLS55 C0177 SS) with gap distance fixed at 0.2 mm.Frequency sweeps were performed from 0.1-100 Hz at a constant oscillatory strain (shear) of 0.1% at 37 °C on ≈80 μL of undiluted hydrogels.Furthermore, to determine shear-thinning behavior of the hydrogel, an oscillatory rheological experiment with cone plate configuration was used.Here, ≈250 μL of Fmoc-DIKVAVD was placed on the center of the sample holder and the test was run with multiple frequency sweeps between 0.1-100 Hz with a shear strain of 0.1% at 37 ˚C.Continuous step-strain measurement was applied to the hydrogel, in which the oscillatory strain changes from 0.1% to 100% after 1 min at 1 Hz frequency.This cycle repeated 4 times and the waiting time decreased in each cycle.

Mesh Size
Hydrogels consist of a cross-linked polymer network with open spaces, called mesh size.The mesh size allows liquid and small solute diffusion through the network.As a scaffold and drug delivery system, it was of paramount importance to calculate/measure the mesh size of the hydrogel and determine how a drug diffuses through the hydrogel.To this end, rheology was used to evaluate the hydrogel mesh size.The average mesh size (, nm) could be calculated based on the rubber elasticity theory (RET) from the following equation: where G′ is the storage modulus, N A is the Avogadro's number (6.022 × 10 23 ), R is the gas constant (8.314J/K mol) and T is Kelvin temperature.As all the rheological experiments were performed at 37 ˚C, T in the equation was considered 310 K. [18]

Surgical Procedures
All animal procedures were conducted in agreement with the Australian National Health and Medical Research Council's published Code of Practice for the Use of Animals in Research, and approval granted by The Florey Institute of Neuroscience and Mental Health Animal Ethics committee (approval number: 18-005 and 20-050).Animals were group housed on a 12:12 h light/dark cycle with ad libitum access to food and water.Surgeries were performed on 33 adult (10 week old) female athymic nude mice (Foxn1nu/Arc) under 2-5% isoflurane anesthesia, as previously described. [21]All intracerebral implants (cells or gels) were made into the striatum (coordinates: 1.0 mm anterior to bregma, 2.0 mm lateral, and 2.8 mm below the surface of the brain) using a 10 mL Hamilton syringe fitted with a glass capillary.All cell transplants were of the H1-TK cell line.

Cohort 1
Mice received a single (1 μL) injection of either SAP hydrogel (Fmoc-DIKVAVD) or the same hydrogel shear-encapsulated with GCV (20 mg mL −1 ) into the striatum for the purpose of assessing the temporal delivery of the drug in vivo (Figure 3).
For liquid chromatography-mass spectrometry (LC-MS) assessment of GCV levels in the brain, Cohort 1 animals were killed at 0, 12, 24, 48, and 72 h after gel implantation by cervical dislocation.Brains were rapidly removed, the striatum microdissected, weighed and homogenized prior to snap freezing.

Liquid Chromatography-Mass Spectrometry (LC-MS) Analysis
GCV calibration samples (ranging from 0 -1000 nm) were prepared in 0.1% formic acid (FA) in Milli-Q water and spiked with the internal standard Ganciclovir-d5 (GCV-d5), (Figure S1, Supporting Information).The striatal brain homogenates were spiked with 6.5 ng of the internal standard.The GCV was then extracted from the tissue using 100 μL of cold extraction medium (80% methanol), followed by vortexing and sonication in an ice slurry water bath for 30 min.The sample was then centrifuged at 18000 x g for 10 min.The extraction step was repeated twice more where after each time the supernatant was collected in a 1.5 mL protein low-bind microfuge tube.The supernatant was then dried down in a Speed Vac®.Finally, the dried residue was reconstituted with 50 μL of mobile phase (0.1% FA in Milli-Q water), vortexed and sonicated in ice/water bath for 10 min.Samples were then filtered through 0.2 μm spin filters (spun at 10000 x g for 5 min) and transferred to 2 mL autosampler vials.A total of 10 μL of each sample was injected into LC-MS spectrometer (Thermo Scientific Q-Exactive Focus hybrid quadrupole-Orbitrap). Chromatographic separation was achieved by an analytical Zorbax Eclipse XDB-C18 2.1 × 50 mm 1.8 μm column.Mobile phases consisted of 0.1% FA in water and 0.1% FA in 100% acetonitrile (95:5, v/v) pumped at a constant flow of 5 μL min −1 .To account for the hydrogel noise, SAP only hydrogel-injected brain samples were used as the baseline and subtracted from the SAP-GCV hydrogel-injected samples.

Microscopy and Quantification
Brightfield and fluorescent images of graft tissues were captured using a Leica DM6000 microscope.Fluorescent images of cultures were captured on a Zeiss Axio Observer Z.1 epifluorescence microscope or Invitrogen EVOS M5000 Imaging System.Human specific PSA-NCAM expression was used to delineate the graft boundaries and estimate graft volume, according to Cavalieri's principle, as previously described. [22]Total number of DAPI + , OTX2 + , FOXA2 + , HNA + , KI67 + , TH + , NEUN + , SOX9 + and CC1 + cells, either in culture or within the grafts, were counted from images captured at 20x magnification, and calculated stereologically across the rostro-caudal axis of the graft (from 1:6 series).For assessment of the host inflammatory response, the density of GFAP and IBA-1 labelling (captured at 40x magnification in the graft core, the graft/host interface and at the periphery) was estimated on color inverted images, using the color range tool on Photoshop (Adobe).Data are expressed as percentage of immunoreactive pixels (of total pixels).

Statistical Analysis
All data were presented as mean ± SEM, except where stated.Statistical tests employed (one-way ANOVA with Tukey's posthoc multiple comparison test, Multiple unpaired t-tests and Student's t-tests) are stated in the figure legends.All statistical analyses were performed using GraphPad Prism and alpha levels of p < 0.05 were considered significant (*p < 0.05, **p < 0.01, ***p < 0.001).

Assessment of Laminin-Based SAP Scaffolds for Neural Compatibility and Optimal GCV Encapsulation
Three tissue-specific, GCV-functionalized SAP hydrogels were engineered and compared to identify the optimal biomaterial formulation for the delivery of GCV and its activation of the HSV-TK suicide gene within hPSC-derived grafts targeted for tissue repair.Each gel shared a core biochemical peptide motif, IKVAV, presenting part of the 1 subunit of laminin.The peptide sequences were synthesised with an aromatic Nfluorenylmethyloxycarbonyl (Fmoc) moiety, protecting the Nterminus of the peptide and providing shared electrons for - stacking interactions.Here, the - stacking embeds the Nterminus and the first two residues within the core of the assembly while only presenting the subsequent peptide sequence to the surface.Then, to ensure self-assembly of the peptides at physiological pH, two aspartate (D) residues were added to the chain.First, two were added on the epitope N-terminus (DDIK-VAV, Figure 1Ai), resulting in an adjusted pK a and affecting the pH of gelation.In this sequence, the D residues (and the Fmoc) are embedded in the assembly and not available for binding.The second hydrogel comprised of a D residue at both the N-and C-terminus (DIKVAVD, Figure 1Aii), similarly affecting pK a but fabricated a resultant gel at physiological pH and ensured the highly charged residues were presented in a way the GCV can bind.The third hydrogel employed was a 1:1 mixture ensuring self-sorted stacking of the two peptides (Fmoc-DDIKVAV+Fmoc-DIKVAVD) to space out the density of the D residues.
The self-assembly structure of these SAP hydrogels were driven by various polar and non-polar, non-covalent intramolecular interactions (e.g., hydrogen bonding, hydrophobic interactions, van der Waals, electrostatic interactions and - stacking interactions) [17,24] resulting in the formation of a stable backbone structure (Figure 1Bi,ii) as well as anti-parallel -sheets (Figure 1Ci,ii) in all three gels, confirmed by Fourier transform infrared (FTIR, major amide peak at ≈1630 cm −1 and a minor peak at 1690 cm −1 , Figure 1F) and circular dichroism (CD, Figure 1G) spectroscopies. [24]The FTIR spectra for all the fabricated hydrogels showed similar peaks, confirming that hydrogen bonding between individual peptide sequences was unaffected by the addition of D residues. [25]Furthermore, the presence of -sheets shown in CD (maximum transitions are observed <200 nm) further reinforces the FTIR data.The higher intensity of the amide peak in the 1:1 mix hydrogel in FTIR spectra was attributed to the increased number of amide bond (per unit volume). [26]-sheet motif-generated nanofibrils (Figure 1D) were capable of interacting as a matrix network of highly hydrated bundles (Figure 1E), as visualized by transmission electron microscopy (TEM, Figure 1Ii,iii).Optical images of the fabricated hydrogels further confirmed the stable 3D gel formation (Figure S2C, Supporting Information).To mimic the physical properties of the brain, substrate stiffness/elasticity are critical factors in the development of hydrogels.All three hydrogels demonstrated viscoelastic behavior (G′>G″) and mechanical properties within the range of rodent brain stiffness [27] (Figure 1H), with minor physiochemical variations between the gels due to the final peptide sequence.Here, the Fmoc-DIKVAVD exhibited the greatest stiffness (700 Pa), compared to the Fmoc-DDIKVAV and Fmoc-DDIKVAV+Fmoc-DIKVAVD gels, and selfassembled at pH of 7.3 (Figure 1J).This resulted from the different intramolecular interactions within the varying peptide sequences between each of the hydrogels.The shear-thinning properties of Fmoc-DDIKVAV hydrogel was investigated with continuous step-strain measurement (Figure S2A, Supporting Information).As the oscillatory strain increased to 100%, the loss modulus of the gel passed the storage modulus after decreasing to 0.1% in less than 5 s.This behavior is indicative of the sol-gel transformation of the SAP in response to the applied shear force.Additionally, the material fully recovered its storage modulus after the removal of the strain in between each of the 4 strain cycles, highlighting the self-healing nature of our Fmoc-DDIKVAV hydrogel.
The overall charge pattern of the hydrogels was confirmed by testing of the -potential.The greatest negative charge was observed in the Fmoc-DIKVAVD hydrogel due to the D group on the surface of the nanofibril being presented at a high density to the solvent (Figure 1K).Conversely, the lowest  -potential was observed in the Fmoc-DDIKVAV.Interestingly, the  -potential observed for 1:1 mix Fmoc-DDIKVAV+Fmoc-DIKVAVD hydrogel was the average of Fmoc-DIKVAVD and Fmoc-DDIKVAV hydrogels, suggesting the valine (V) and D groups were regularly distributed throughout the system.
The drug loading capacity and its release from the SAP hydrogel network is dependent on the mesh size of the hydrogel.As expected, the DIKVAVD hydrogel exhibited the smaller mesh size (mean r mesh = 13 nm) compared to Fmoc-DDIKVAV and 1:1 mix Fmoc-DDIKVAV+Fmoc-DIKVAVD (mean r mesh = 16 nm) (Figure S2B, Supporting Information).Consequently, in the initial comparison of GCV incorporation into the three hydrogels, both Fmoc-DDIKVAV and 1:1 mix Fmoc-DDIKVAV+Fmoc-DIKVAVD gels exhibited similar GCV encapsulation capacity (GCV amount = 4.5 mg mL −1 , Figure 1J).However, the Fmoc-DIKVAVD gel was capable of a higher GCV loading (GCV amount = 8 mg mL −1 ), demonstrating the ability for hydrogels of smaller mesh size to encapsulate more GCV small molecules within its structure.Furthermore, the Fmoc-DIKVAVD hydrogel showed a higher ablation of hPSCs carrying the suicide gene (Figure S3, Supporting Information), likely a result of the increased stiffness of the gel, generating higher nanofiber entanglement for compound entrapment.The loading capacity of the drug in the hydrogel, based on (total amount of encapsulated GCV drug/(total weight of GCV+Peptide weight) x 100), was 60.4%.
For subsequent in vitro and in vivo testing, the Fmoc-DIKVAVD SAP hydrogel was selected for its superior biochemical properties (biocompatibility, stiffness, gelation pH, and GCV loading).We have also previously shown that it takes a total of 9 months for the SAP hydrogel to be cleared post-implantation in the rodent brain. [14]To achieve a higher prodrug concentration suitable for in vivo delivery, GCV was incorporated into the hydrogel during as well as after gelation (GCV: 20 mg mL −1 ).The composite Fmoc-DIKVAVD SAP hydrogel, shear-encapsulating GCV, maintained the nanofibrous matrix formation (Figure 1L), investigated previously by us using small angle x-ray scattering, [24] and exhibited physiochemical and viscoelastic properties comparable to Fmoc-DIKVAVD SAP hydrogel alone (Figure 1M-O).

Hydrogel Encapsulation and Release of GCV Retained Prodrug Functionality to Ablate Proliferative Cells In Vitro
The ability of the GCV to activate the suicide gene after encapsulation in the hydrogel was initially assessed in vitro.Treatment of proliferative hPSCs with the unfunctionalized DIKVAVD SAP hydrogel had no impact on the cell survival (Figure 2A,B).However, adding GCV or hydrogel-entrapped GCV directly to the culture media resulted in ablation of the hPSCs, with evidence of minimal, residual hypertrophied cells undergoing GCV-induced apoptosis (Figure 2C-E).In addition, only H1 hPSCs expressing the HSV-TK suicide gene (subsequently referred to as H1-TK), not control H1 hPSC, displayed apoptotic responses to SAP-encapsulated GCV (Figure 2F).Additionally, SAP-encapsulated GCV retained comparable levels of apoptotic cell death (% viable cells in culture) when used within 24 h (fresh) or 7 days after fabrication of the functionalized hydrogel, which is crucial for targeted and sustained delivery of GCV in vivo (Figure 2G).
To demonstrate the efficacy of the SAP-encapsulated GCV to activate the suicide gene only within cells undergoing S-phase of cell cycle, mixed cultures of both post-mitotic and proliferative cells were examined.For this purpose, differentiated hPSCs were assayed at day 13 (D13) of a ventral midbrain (VM) differentiation, a critical time when cells have acquired a VM fate (reflected by the high proportion, >80%, FOXA2 + /OTX2 + cells, Figure S4, Supporting Information), and contain ≈50% proliferating cells. [3]Within 7 days, cultures treated with GCV or SAP-GCV had a significantly reduced proportion and density of KI67 + proliferative cells (Untreated: 7.6 ± 1.6% KI67 + cells, +SAP: 8.6 ± 1.6%, +GCV: 1.9 ± 1.2%, +SAP-GCV: 1.9 ± 0.4% KI67 + cells; Figure 2H-K,P,Q).Despite an evident trend, no significant difference was observed in the percentage of TUJ1 + neurons or their density -One-way ANOVA detected a group effect (F (3,8) = 4.279; p = 0.045) but no significant difference between groups in Tukey's multiple comparison (Control vs SAP-GCV: p = 0.065, other comparisons: p > 0.1; Figure 2L-O,R,S).Note, this observed reduction in TUJ1 + cells in GCV treated cultures, D) hollow nanofiber and E) hydrogel matrix formation through the entanglement of the nanofibers.F) FTIR spectroscopy of the amide I region indicating anti-parallel -sheet formation in the three fabricated hydrogels.G) CD spectra of all SAP hydrogel assemblies.H) Assessment of the viscoelastic properties of the three hydrogels showing higher G' (storage modulus) to G", (loss modulus) which was within the range of the rodent brain's (grey region), critical for in vivo biocompatibility.I) TEM images illustrating the nanofiber structure of the i) Fmoc-DDIKVAV, ii) Fmoc-DIKVAVD, and iii) Fmoc-DDIKVAV+Fmoc-DIKVAVD SAP hydrogels.J) Comparison of the three hydrogels reveals the optimal properties of the Fmoc-DIKVAVD hydrogel, based upon biocompatible stiffness and pH, as well as highest GCV loading capacity.K)  -potential of the 3 fabricated SAPs.L) High power TEM image visualizing the nanofiber matrix of the composite Fmoc-DIKVAVD loaded with GCV.M) Incorporation of GCV into the hydrogel had no effect on the physiochemical and mechanical properties of Fmoc-DIKVAVD as revealed in FTIR and N) CD spectroscopies and O) rheology.Scale bars: I) 100 nm and L) 250 nm.while not significant, likely reflects the activation of the suicide system within mitotic cells at the initiation of treatment (D13), noting analysis of TUJ1 + cells was performed at D20.The results demonstrated that shear-encapsulating GCV prodrug within SAP hydrogels has no effect on the compound's functional efficacy or stability.

SAP Hydrogel Encapsulation of GCV Prolongs Delivery in Rodent Brain
We examined the release kinetics and stability of GCV by delivering SAP-GCV in vivo into the mouse brain (Figure 3A).We isolated striatal brain tissue at defined intervals over 72 h (Figure 3B), which is equivalent to two full cell cycles of hPSCderived VM neural progenitors (≈33 h per cycle). [3]It was estimated that this duration would be sufficient to target all proliferating neural progenitors within grafts.LC-MS analysis of the brain samples indicated that the majority of GCV was released within the first 12 h after in vivo delivery (Figure 3C,D).At 12 h, 6.5% of the initial GCV concentration was detectable, with a gradual and sustained release over 48 h until no GCV was detectable at 72 h.In comparison to existing approaches, such as repeated daily intraperitoneal injection where the drug remains detectable in the blood and brain for less than 2 h [3,6] (as depicted in Figure 3E), these findings demonstrate that GCV encapsulated within the hydrogel will provide long-term delivery following implantation, coinciding with the prolonged cell cycle of human neural progenitors (Figure 3F), intended for suicide gene-induced ablation.

Single Intracerebral Injection of GCV-Carrying Smart Hydrogel Improves Suicide Gene Activation in hPSC-Derived Teratomas
We previously demonstrated that activation of the HSV-TK suicide gene in hPSC-derived teratomas and neural progenitor grafts could reduce tumor/graft size and the proportion of KI67+ proliferative cells. [4]To achieve this, daily intraperitoneal injections of GCV were required for 4-8 weeks, a cumbersome and suboptimal regimen that presented systemic and toxic side effects and lacked efficacy. [3]Here, we compared the efficiency of a one-off, focal delivery of SAP-GCV hydrogel to the conventional intraperitoneal systemic administration to activate the TK suicide gene in vivo.As proof-of-principle, to demonstrate sustained GCV delivery and functionality in vivo, we implanted the SAP-GCV hydrogel focally into the core of intracerebral teratomas (Figure 4A), and compared outcomes to the unfunctionalized hydrogels (as well as untreated teratomas and animals receiving daily intraperitoneal injections of GCV).Histological analysis at 3 weeks after intracerebral delivery of hPSC revealed large tumors in all animals, evident by human-specific NCAM staining (Figure 4B-E).However, grafts treated with GCV or SAP-GCV hydrogel had reduced cell density compared to untreated or SAP-treated grafts (Figure 4F-J).Only SAP-GCV-treated grafts showed significantly more pyknotic nuclei (a phenotype indicative of apoptotic cells) compared to other groups (Figure 4K-O), demonstrating the ability of hydrogel-released GCV to maintain functionality and activate suicide genes more effectively than systemic GCV administration daily.Correlatively, grafts treated with SAP-GCV displayed a significant 53% reduction in KI67 + cells, while the conventional systemic regime showed just a 32% reduction that was not significant compared to untreated teratomas (Untreated: 40.0 ± 2.2% KI67 + cells, +SAP: 37.2 ± 2.8%, +GCV: 27.3 ± 3.3% KI67 + cells, +SAP-GCV: 18.8 ± 5.8% KI67 + cells, Figure 1P-T).Here, our smart hydrogel was shown to have superior efficacy in focally delivering GCV to activate the suicide gene, resulting in the apoptotic destruction of proliferating cells in teratomas.

GCV-Delivering Scaffold Reduced Proliferative Population within Neural Grafts in Parkinsonian Mice
With evidence of improved suicide gene activation by the smart, functionalized hydrogel in teratomas, we next examined the more targeted influence of the SAP-GCV on improving neural graft outcomes in a rodent model of PD.Intranigral delivery of 6-OHDA ablated >70% of endogenous midbrain DA neurons (data not shown).Animals subsequently received grafts of H1-TK hPSC-derived VM progenitors, were treated with SAP or functionalized SAP-GCV hydrogel at 7 weeks post-transplant and histologically examined at 12 weeks (Figure 5A).Viable grafts, contained within the host striatum, were confirmed by human specific NCAM (Figure 5B,C).SAP-GCV treatment significantly reduced graft volume (VM + SAP: 0.115 ± 0.017 mm 3 , VM + SAP-GCV: 0.071 ± 0.009 mm 3 , Figure 5D) and total number of graft-derived human nuclear antigen (HNA + ) cells (VM + SAP: 21664 ± 3761 cells; VM + SAP-GCV: 11, 275 ± 1779 cells; Figure 5E).Indicative of the functionality of the SAP-GCV, these treated grafts showed a significant (49%) reduction in KI67 + proliferative cells (VM + SAP: 366 ± 61 cells, VM + SAP-GCV: 185 ± 41 cells, Figure 5F-I).Our next step was to determine which population of cells within the grafts were most affected by the SAP-GCV delivery and associated apoptotic cell death.Noting oligodendrocytes were previously shown to account for <3% of these graft in previous work, [3] we focused our attention on the SOX9 + astrocyte and NEUN + neuronal populations.A trend toward reduced total numbers of NEUN + neuronal and SOX9 + astroglial cell populations was observed (Figure 5J-M), indicative of the death of both neuronal and glial progenitors.Reflective of our previous observations, [3] and while not significantly different, the proportion of SOX9 + astrocytes decreased in SAP-GCV-treated grafts (%SOX9 + /HNA + , SAP: 35.03 ± 2.82%; SAP-GCV: 28.74 ± 3.29%) and the proportion of neurons increased (%NEUN + /HNA + , SAP: 37.15 ± 5.72%; SAP-GCV: 41.41 ± 3.82%; Figure 5N).
To protect the TH + DA neuronal population within the grafts, the functional unit of the transplant, SAP-GCV treatment was delivered 7 weeks after grafting -a period anticipated to be after cell cycle exit of VM DA progenitors. [3]Interestingly, however, a notable, but not significant, reduction in TH + DA neurons was observed, suggestive of the presence of DA progenitors at the time of suicide-activating hydrogel implantation (Figure 5O-Q).Such observations suggest that delaying SAP-GCV delivery to the graft could protect these DA progenitors, and warrants further investigation, inclusive of the impact on the graft's capacity to reverse motor deficits.Of interest, the subtle increase in the proportion of TH + DA neurons within suicide-activated grafts suggests selective ablation of non-DA neurons (%TH + /NEUN + , SAP: 25.98 ± 3.44%; SAP-GCV: 31.14 ± 6.03%; Figure 5R).
Of final importance, focal hydrogel delivery of GCV, and associated apoptotic-induced cell death within the grafts, did not exacerbate local host inflammation, confirmed by the unchanged density of GFAP + reactive astrocytes (and IBA-1 + microglia, data not shown) within the graft core (field of view 1, FOV1), at the graft-host interface (FOV2) and distal to the graft (FOV3), Figure S5 (Supporting Information).

Discussion
Despite advancements and refinements in the protocol to generate high levels of correctly patterned VM progenitors, hPSCderived VM progenitor grafts produce low yields of DA neurons and a high proportion of incorrectly specified neural and nonneural cells. [19,28]A timely activation of the HSV-TK suicide gene has resulted in enriched grafts for DA cells by removing proliferative cells destined to become non-DA cells, improving neural graft standardisation. [3]However, the short half-life of the GCV prodrug [6] to successfully ablate proliferating cells undergoing Sphase, coupled with the relatively short S-phase of neural progenitors, presented a challenge for targeting all proliferative cells. [3]ere, we engineered a smart, neural tissue-specific hydrogel to selectively modulate the delivery of the GCV prodrug focally at the graft core, and demonstrated superior outcomes to systemic administration, which currently requires cumbersome, daily systemic dosing.
Single administration of the hydrogel delivering GCV remarkably improved the duration of detectable GCV up to 48 h.The significance of this can be highlighted in the context of cell cycle dynamics for different proliferative cell populations.Human PSCs spend an estimated 8 h (of a 15 h cell cycle) in S-phase, while VM neural progenitors ≈15 h (of a 33 h cycle). [3]Consequently, the gel-based delivery prolonged presentation of GCV to these populations for the duration of their S-phase, circumventing the challenges of the short <2 h duration detectable after systemic intraperitoneal injection.The efficacy of this delivery  was evident in both PSC and neural progenitor cultures, with near complete ablation of the cells and/or KI67 + population.In vivo, despite significantly reducing the KI67 + cell numbers, residual proliferative cells remained.Such observations likely reflect suboptimal delivery of the hydrogel, noting the large size of the teratomas (≈2 mm 3 ) after PSC implantation, yet small volume of SAP-GCV administered (just 1 μL, containing 20 μg of GCV).Added to this was the small size of the drug (255 g mol −1 ) that reduced entrapment within the nanofibrous structure of the gels, resulting in shorter delivery time compared to larger drugs and proteins from previous SAP hydrogels. [15,16]ith future efforts to alter the physicochemical properties of the hydrogel, we will be able to enhance the immobilization of drug molecules, while maintaining biocompatibility and physiological modulus, as well as sustaining higher drug concentrations and delivery for longer periods. [29]Such modifications may include increasing the peptide chain length and density to modulate the mesh size of the fibrillar network, introducing branching in the fibre network, and altering the peptide sequence to increase electrostatic charge or hydrophobicity.Additionally is the prospects to engineer multifunctional composite scaffolds incorporating (e.g., combining GCV-functionalized polymers, short electrospun nanofibers and/or microspheres) as has been performed in other contexts. [30]Alternative approaches may be adopted to increase the size of the drug to enhance physical entrapment and prolong its release rate.Covalent conjugation of the drug to polypeptides, requiring spontaneous matrix network degradation to break the bond holding the drug to the scaffold, may also allow for more controlled and prolonged drug release.Despite this, the present findings show for the first time the superior efficacy of utilizing hydrogels for focal drug delivery to neural grafts, providing a platform to investigate future material modifications.
Previous studies by us and others [3,4,31] have highlighted the necessity for high GCV doses (50 mg kg −1 ), requiring repeated daily administration over several weeks that is not only cumbersome, but can lead to off-target side effects including weight loss and gastrointestinal complications. [32]Although the use of a mutated variant of the HSV-TK gene to improve sensitivity to GCV [33] and/or using a less toxic nucleoside analog are able to reduce the prodrug dosage, [34] these regimes continue to require long-term, repeated treatment.By localising GCV's release to the graft and/or brain, the present findings circumvent the need for long-term, repeated dosing and reduce off-target side effects.
The initial burst release of the GCV upon implantation of the hydrogel, a previously observed feature of SAP-based hydrogel delivery of proteins that reflects the short delay in gel reassembly in vivo, [15] may lead to concerns of cytotoxicity or increased bystander killing effect.Such events have been observed in the use of the HSV-TK suicide system to ablate cancer cells, resulting from phosphorylated GCV passing through gap junctions to adjacent heathy/desired cells that do not harbour the suicide gene and induce non-selective apoptosis. [35]The trend showing an increased proportion of post-mitotic TH + neurons within the grafts is indicative that post-mitotic populations were spared such effects.To confirm these events, a more detailed assessment of defined post-mitotic subpopulations is needed within more acute intervals following SAP-GCV delivery.Aforementioned approaches to improve the sustained delivery of GCV, through hydrogel physicochemical modifications, will also likely negate potential bystander concerns.
Based on the data presented, HSV-TK gene activation was more efficient in teratomas with our hydrogel system compared to conventional GCV delivery, where the proliferative population in neural grafts was significantly reduced, addressing our primary goal of improving the safety of hPSC-derived brain transplants.Previously, we demonstrated that the timely delivery of GCV could additionally improve graft purity, by impacting cell populations that become post-mitotic later than the desired functional (TH + DA) cell population. [3]In particular, was the effect on the non-DA neurons and astrocytes within the grafts.Here we show a similar trend (reduction in total NEUN + neurons and SOX9 + astrocytes), yet levels remain not significantly different.The discrepancies in levels of significance likely reflect the analysis of more immature grafts in the present study (12 weeks), compared to former analysis at 24 weeks, [3] where ongoing neurogenesis and gliogenesis in non-GCV treated animals likely contributed to larger neuronal/glial pools and hence more substantive differences.
Since the discovery of hPSCs and the great anticipation for their application in regenerative medicine, only now are these cells encroaching clinical testing, inclusive of ongoing trials for PD. [36]One of the most recognized and critical concerns with the use of these cells has been the risk of tumors or expansion of unwanted cell populations after in vivo delivery.While cell sorting strategies have been heavily explored, the incorporation of suicide genes is being recognized as advantageous, and in many regards a superior approach.Moving forward these suicide genecarrying cell lines will require further safety testing and validation of functionality -at the cellular level by electrophysiological assessment of the DA neurons, as well as at the neural systems level by confirmation of dopamine-induced reversal of motor deficits in rat (and non-human primate) models of PD.Additional to this will be the need for regulatory approval for the use of a genetically modified cell product, however the HSV-TK suicide gene has previously been used in several clinical trials of haematopoietic stem cell transplantation in leukemia within the EU and USA -see review. [37]iomaterials have been used extensively as drug delivery systems to improve the treatment of a plethora of disorders (see review), [12] including the delivery of drugs targeted at activating suicide genes.Most noteworthy has been the development [38] and subsequent clinical trial [39] of the intraocular polyvinyl alcohol (PVA) and ethylene vinyl acetate (EVA) suture-implanted device (Vitrasert®) that sustained GCV over several months, delaying the progression of cytomegalovirus retinitis.However, unlike the implantable device employed in their trial, our peptide-based SAP hydrogel delivering GCV is biodegradable, circumventing the need for surgical removal at the completion of the treatment dose.More recently, Bareiss et al. [11] developed silica nanoparticle encapsulating acyclovir, an analogue of GCV, to sustain and retain drug functionality, demonstrating effective ablation of HSVinfected cells.GCV is recognised as a much more potent analog compared to acyclovir, and the present work provides the first evidence of targeted, functional delivery to impact a stem cellderived graft.
While the present study highlights the benefits of this refined biomaterial-based drug delivery system to improve stem cellderived neural grafts for PD, these combined tools will have similar benefits in other neurological conditions for which cell therapy is a possibility, including stroke and epilepsy.Additional to this will be refinement in the in vivo delivery approach, inclusive of intrathecal [40] or subdural [41] injection for example, of the functionalized hydrogels.

Figure 1 .
Figure 1.Generating the optimal hydrogel for biocompatibility and GCV encapsulation.A) Schematic of Fmoc-DDIKVAV and Fmoc-DIKVAVD peptide structure and B) peptide assembly driven by - stacking interactions and C) polar and non-polar, non-covalent intramolecular interactions to form a

Figure 2 .
Figure 2. Hydrogel-encapsulated and -delivered GCV retains functional capacity to ablate proliferative cells in culture.A) Representative image of an untreated H1-TK hPSC culture and following exposure to B) a DIKVAVD SAP hydrogel, C) GCV or D) SAP hydrogel-encapsulated GCV.E) Exposure of cultures to GCV (alone or via SAP) resulted in ablation of the proliferative hPSCs.F) Only exposure of hPSCs carrying suicide gene (H1-TK), but not control H1 hPSCs, to SAP-GCV eliminated the viable (proliferative) cells in culture, demonstrating selectivity of the SAP-GCV for the transgene.G) Cultures treated with fresh or 7 days old SAP-GCV both ablated proliferative hPSCs, elucidating the stability of the prodrug within the gel.H-K) Representative images of proliferating KI67 + cells within day 20 VM cultures, and following treatment with SAP, GCV or SAP-GCV.L-O) Photomicrographs illustrating TUJ1 + neurons within the VM cultures.P,Q) Treatment of VM cultures with GCV (alone or via SAP encapsulation) significantly decreased the proportion and density of KI67 + proliferative cells.R,S) Proportions of TUJ1 + neurons remained unaffected by SAP or GCV treatment.Data represent mean ± SEM. n > 3 independent cultures.E,G,P-S) One-way ANOVA, F) Student's t-test.*p < 0.05, **p < 0.01, ***p < 0.001.Scale bars: A-D,H-O) 100 μm.

Figure 3 .
Figure 3.In vivo temporal release profile of GCV from smart hydrogel.A) Schematic diagram of in vivo study design and B) the timeline for isolation of brain striatum extracts for LC-MS analysis.C) Overlayed representative liquid chromatography traces illustrating GCV concentration-time curves after a single SAP-GCV injection in the striatum at T = 0 (red), 12 (blue), 24 (green), 48 (yellow), and 72 h (black) post-implantation.C') Enlargement of the GCV peak at 12-72 h.D) Quantification of GCV concentration in striatal brain samples over time (and percentage remaining of initial GCV concentration).E) Graphical representation of the cell cycle kinetics of neural progenitors (green) with relatively short time spent in S-phase (shaded region).Daily systemic administration of GCV (red), with a short half-life, often misses the S-phase window to activate the apoptotic mechanism.F) Representation of sustained GCV delivery using a SAP hydrogel to capture all neural progenitors in vitro or in vivo during their S-phase.Data represent mean ± SEM.D) n = 4-5 brain extracts/time point.RT retention time; NL normalised level; GNL global normalised level; m/z mass to charge ratio.

Figure 4 .
Figure 4. Comparison of focal and sustained GCV delivery via hydrogel with standard systemic delivery in hPSC-derived teratoma grafts.A) Schematic diagram of in vivo study design, illustrating hPSC-derived teratoma grafts either i) receiving no treatment after 2 weeks, ii) a single intracerebral injection of SAP, iii) or SAP-GCV, or iv) daily intraperitoneal injections of GCV.B-E) Representative coronal sections of hPSC-derived teratoma grafts untreated, injected with SAP hydrogel, receiving systemic GCV, or implanted with SAP-GCV hydrogel.F-I) Photomicrographs showing DAPI + cell density, J) quantified within teratoma grafts.K-N) Representative images of teratoma grafts O) and quantitative assessment of pyknotic nuclei, noting the significant increase in dying cells only with SAP-GCV treatment.P-S) Representative photomicrographs and T) quantification of the proportion of proliferating KI67 + cells in teratoma grafts.Only SAP-GCV treated teratomas (and not systemic GCV treatment) showed a significant decrease in KI67 + cells.Data represent mean ± SEM, n = 4-5 grafts per group, one-way ANOVA.*p < 0.05, **p < 0.01.Scale bars: B-E) 1 mm, F-I,P-S) 100 μm, and K-N) 100 μm.

Figure 5 .
Figure 5. SAP hydrogel delivering GCV reduces proliferative populations in VM neural grafts.A) Schematic diagram of neural progenitor grafting study in a 6-OHDA PD model.B,C) Representative photomicrograph illustrating a coronal overview of the hPSC-derived VM progenitor graft treated with SAP or SAP-GCV hydrogel.D-G) Quantitative assessment of D) graft volume, E) human nuclear antigen (HNA + ) cells, as well as F) total number and G) proportion of KI67 + proliferative cells within SAP and SAP-GCV treated VM grafts.H,I) Representative images depicting fewer HNA + (red) and KI67 + cells (green) in grafts receiving SAP-GCV hydrogel compared to SAP alone.H',I') High power images illustrating H,I.J,K) Sample images of neuronal NEUN + neurons (red) and SOX9 + astrocytes (green) within grafts, and J',K') higher magnification showing colocalisation with HNA + (blue).L,M) Quantification of the total number and N) proportion of NEUN + and SOX9 + cells within grafts.O,P) Images representing graft-derived TH + dopaminergic neurons (green) within the striatum, colocalising with NEUN (red).O',P') Higher magnification from (O,P).Q) Quantification of total TH + neurons and R) their proportion out of the NEUN + neuronal pool.Fine dotted line in J,K,O,P delineate the graft.Data represent mean ± SEM, n = 7-10 grafts per group.Student's t-test.*p < 0.05.Scale bars: B,C) 1 mm, H-K,N,O) 100 μm and H'-K',N',O') 50 μm.