Injectable diblock copolypeptide hydrogel provides platform to maintain high local concentrations of taxol and local tumor control

Introduction Surgical resection and systemic chemotherapy with temozolomide remain the mainstay for treatment of glioblastoma. However, many patients are not candidates for surgical resection given inaccessible tumor location or poor health status. Furthermore, despite being first line treatment, temozolomide has only limited efficacy. Methods The development of injectable hydrogel-based carrier systems allows for the delivery of a wide range of chemotherapeutics that can achieve high local concentrations, thus potentially avoiding systemic side effects and wide-spread neurotoxicity. To test this modality in a realistic environment, we developed a diblock copolypeptide hydrogel (DCH) capable of carrying and releasing paclitaxel, a compound that we found to be highly potent against primary gliomasphere cells. Results The DCH produced minimal tissue reactivity and was well tolerated in the immune-competent mouse brain. Paclitaxel-loaded hydrogel induced less tissue damage, cellular inflammation and reactive astrocytes than cremaphor-taxol (typical taxol-carrier) or hydrogel alone. In a deep subcortical xenograft model, of glioblastoma in immunodeficient mice, injection of paclitaxel-loaded hydrogel led to a high local concentration of paclitaxel and led to local tumor control and improved survival. However, the tumor cells were highly migratory and were able to eventually escape the area of treatment. Conclusions These findings suggest this technology may be ultimately applicable to patients with deep-seated inoperable tumors, but as currently formulated, complete tumor eradication would be highly unlikely. Future studies should focus on targeting the migratory potential of surviving cells.


INTRODUCTION
Glioblastoma accounts for 40% of primary brain tumors and results in over 15,000 deaths a year in America alone [1]. Despite maximal therapy including surgical resection, average survival is only 20 months [3]. Some patients are not candidates for surgery due to poor health, deep or bilateral location or proximity to eloquent structures [5,6]. Without cytoreductive surgery, the average survival is four months [3].
In addition to cytoreductive surgery, current standard of care also includes temozolomide and radiation. Temozolomide treatment has a small but statistically significant improvement in survival [7]. However, it also induces mutations leading to more aggressive recurrent tumors [8].
Further, treatment with temozolomide can result in enhanced resistance by demethylatation of the MGMT promoter [9]. There are many promising chemotherapy options that show improved potency over temozolomide without the genotoxic damage. However, often these drugs are limited by their pharmacokinetics, inability to cross the blood-brain barrier or difficulties and side effects with systemic administration [10].
To address these limitations, drug releasing implants are being developed, most often composed of biodegradable polymers, that may provide improved local delivery of promising chemotherapeutics in the CNS [11]. While multiple studies have evaluated the ability of different carriers to deliver chemotherapeutic payloads to glioblastoma cells, the majority of these studies have tested them in sub-optimal conditions using either in vitro [12] or subcutaneous flank environments [13], or testing them on suboptimal, multi-passaged glioma cell lines that often do not recapitulate the migratory capacity of patient tumors [14] [15] [16]. Fully understanding the potential of these vehicles requires testing in realistic in vivo environments. To this end, we have constructed diblock copolypeptide hydrogels (DCH) capable of delivering paclitaxel to a patientderived gliomasphere model of GBM in a mouse brain. We found that paclitaxel-loaded hydrogels led to local control and improved survival in this mouse model, but surviving cells were able to migrate out of the treatment zone.

METHODS
Gliomasphere culture and in vitro proliferation assay.

Cancerous stem cells can be derived from brain tumors and propagated in vitro as
neurosphere-like gliomaspheres [17] [18]. For this study, we utilized the sample HK308, a gliomasphere line derived from a recurrent GBM in a 50-year-old male as previously described [17,18]. The culture and the original tumor sample have an amplification of the epidermal growth factor receptor gene, and express the EGFRvIII mutation. MGMT  Hydrogel Synthesis and Fabrication.
The diblock copolypeptide, K 180 A 40 was the DCH used in this study, and was synthesized using techniques described in detail previously [20]. Briefly, within an inert, dinitrogen filled glovebox environment a solution of N ε -carbobenzyloxy-L-lysine N-carboxyanhydride ((Z-L-lysine NCA) in THF solvent was polymerized upon addition of the transition metal initiator, Co(PMe 3 ) 4 .
The polymerization of the N ε -carbobenzyloxy-L-lysine block was allowed to proceed for approximately 45 minutes and was monitored for completeness by FTIR before the subsequent addition of a solution of L-alanine NCA in THF. The consumption of the second NCA was also allowed to proceed for 45 minutes before the reaction was removed from the glovebox for subsequent deprotection of the N ε -carbobenzyloxy-L-lysine residues by HBr in an acetic acid/TFA solution followed by polymer precipitation in diethyl ether. The precipitated polymer was redissolved in nonpyrogenic DI water and then dialyzed within a 2000 Da MWCO dialysis bag exhaustively against NaCl for two days and then nonpyrogenic DI water for two days before being lyophilized. Paclitaxel was loaded into the DCH by dissolving and mixing the DCH and paclitaxel in equal mass in an 80/20 Methanol/water solution and then drying under vacuum overnight to remove the solvent. The dry DCH/paclitaxel powder was resuspended in nonpyrogenic DI water at 3 wt% for subsequent in vitro and in vivo evaluation.
In vitro evaluation of release of paclitaxel from DCH.
For the in vitro paclitaxel release assay, 100 µL of 3 weight-percent (wt%) DCH loaded with paclitaxel (DCH-paclitaxel) was injected into pre-hydrated 20,000 MWCO Slide-A-Lyzer dialysis cassettes before being placed in a sink of 100 mL of 5% Fetal Bovine Serum (FBS) containing 1xPBS. A cassette contacting 100 µL of paclitaxel in DMSO was used as a control.
To analyze the amount of released paclitaxel, 50 mL of the incubation media from each time point was extracted with four separate 50 mL ethyl acetate washes. The extracted organic layer was evaporated to dryness under vacuum, resolubilized in acetonitrile and filtered with a 0.2 µm syringe filter prior to analysis by HPLC. HPLC analysis of paclitaxel concentration was performed using methods adapted from a previous report [21]. Specifically, 10 µL of release assay samples of unknown paclitaxel concentration and paclitaxel standards were run on a C18 150 x 4.6 mm I.D. Agilent column using an isocratic 50/50 Acetonitrile/Water mobile phase and detected by UV absorption detection at a wavelength of 227 nm.

Surgery and Hydrogel implantation.
Biocompatibility evaluations of DCH and Cremphor vehicle with and without paclitaxel were performed using adult male and female C57Bl6 mice aged 8-14 weeks. For gliomaspherederived hGBM cell injections, Non-obese Diabetic/ Severe Combined Immunodeficiency/Gamma Null (NSG) mice were used. All surgical procedures performed within this paper were evaluated by the UCLA animal research committee (ARC) and described within an approved ARC protocol.
Mice were anesthetized using isoflurane and a craniotomy was performed by drilling a rectangular exposure window in the bone with a high-speed dental drill. For biocompatibility evaluations, DCH or vehicle were injected stereotactically into the center of the caudate putamen nucleus using the target coordinates of 0.5 mm anterior to Bregma, 2 mm lateral to Bregma and a depth of 3.0 mm below the cortical surface. A 2 μL volume of DCH was injected using a pulled glass micropipettes ground to a beveled tip with 150-250 μm inner diameter attached via specialized connectors and high-pressure tubing to a 10 μL syringe that was mounted to a stereotaxic frame and controlled by an automated microdrive pump. For studies on hGBM cells, a 2µL suspension of 100,000 hGBM cells was injected first into the striatum using the target coordinates of 0.5 mm anterior to Bregma, 2 mm lateral to Bregma and a depth of 3.0 mm below the cortical surface. This first injection was followed immediately by a second injection of 2 µL of DCH placed directly above using the target coordinates of 0.5 mm anterior to Bregma, 2 mm lateral to Bregma and a depth of 2.0 mm below the cortical surface. Postoperative care and analgesia was administered preoperatively as well as for two days following surgery to alleviate pain.

IVIS Spectrum in vivo Imaging
The gliomasphere cell proliferation in vivo within the brain was monitored at discrete time intervals using non-invasive bioluminescence imaging on an IVIS® Spectrum in vivo imaging system. Mice received intraperitoneal injections of 200 μL of luciferin and were scanned 5 minutes later with a 10 second exposure. Results were listed as flux (photons/second) through the area of interest.

Animal Perfusions and Histology.
At specific predetermined times after hydrogel/cell injection, or when tumors reached a terminal size requiring euthanasia, animals were given a lethal dose of pentobarbital and a transcardial perfusion was performed with heparinized saline and 4% paraformaldehyde. Brains were excised, post-fixed in 4% paraformaldehyde for 5 hours and then stored in 30% sucrose.
Brains were sectioned along the coronal plane using a cryostat. Tissue sections (40 μM) were stained using standard fluorescent immunohistochemical techniques as previously described [22] to visualize host tissue, injected gliomaspheres cells and hydrogel deposit localization. Primary

Paclitaxel is more potent than Temozolomide
Temozolomide is available in an oral formulation, is well tolerated, and is able to cross the blood-brain barrier making it a convenient therapeutic option. However, temozolomide has only modest efficacy in vitro[23] and in vivo [7]. Given space limitations the ideal candidate chemotherapeutic agent would have both high efficacy and high potency. We tested both temozolomide and paclitaxel at various doses against our patient-derived gliomasphere line (HK308). Both of the two candidate drugs were effective at suppressing the proliferation of the gliomaspheres in a concentration dependent manner over the seven-day evaluation period ( Figure 1A). By applying growth rate (GR) inhibition calculations[24], we quantified the relative effectiveness of each drug's capacity to suppress cell growth in the endpoint assays. The GR 50 value (representing the drug concentration at which the cell proliferation rate is reduced by half), was less than 1 nM (the lowest dose tested) for paclitaxel, and was approximately 0.5 mM for temozolomide. Based on these results, paclitaxel was at least 500,000 times more potent on our human patient derived gliomasphere cell line than temozolomide and was considered to be the more appropriate candidate for further exploration in subsequent in vivo studies.

DCH-Paclitaxel maintains a high local concentration over a prolonged period
The ideal chemotherapeautic agent should maintain high local concentrations and not become diluted through diffusion. To determine in vitro drug release kinetics we prepared a 3 wt% K 180 A 40 DCH at a drug concentration of 3% w/v and placed the gel in a dialysis cassette suspended in an incubation buffer with 5% v/v fetal bovine serum (FBS) to simulate the brain environment. The incubation media was replaced at discrete time points and the concentration of paclitaxel was evaluated by HPLC following organic extraction. A solution of paclitaxel in DMSO at the same concentration was also incubated within a separate dialysis cassette over the same time course to serve as a control. The paclitaxel stayed inside the DCH matrix throughout the six week period indicating that the paclitaxel does not diffuse out of the matrix but is rather only released as the gel is degraded ( Figure 1B). After the experiment, nearly all of the paclitaxel was recovered and found to have retained its chemical signature and biological activity ( Figure 1C).

DCH-Paclitaxel maintains a high local concentration over a prolonged period
To test the CNS biocompatibility of the DCH vehicle, we injected the K 180 A 40 DCH with and without paclitaxel into the caudate putamen of healthy adult mice in the absence of hGBM cells and evaluated the general foreign body response using standard immunohistochemistry markers.
For comparison, we injected an equivalent volume of 50% v/v Cremophor EL (polyoxyethylated castor oil) in ethanol with and without paclitaxel in separate cohorts of mice. Cremophor EL is used as the surfactant vehicle for paclitaxel within the standard Taxol formulation approved clinically for intravenous administration of the drug. Animals tolerated all four formulations well after injection and there were no observable adverse health events throughout the duration of the study. At one week post injection, which was previously characterized as the time of maximum foreign body response to DCH alone (without cargo molecules) [25], brain tissue was evaluated at and around the injection site using immunohistochemical markers for macrophages and activated microglia (CD68, IBA-1) and astrocyte reactivity (GFAP) (Figures 2A-D). To quantify the extent of the foreign body response, we characterized the intensity of immunohistochemical staining across a radial distance of 1 mm originating from the center of the injection ( Figure   2E,F). To establish a cumulative measure of immunostaining across this circular tissue injection zone, we calculated the area under the curve (AUC) on the intensity plots ( Figure 2F).
The Cremophor EL vehicle treated group exhibited a small, localized core of densely  (Figures 2B,F). Interestingly, in the Cremophor EL plus paclitaxel group there was the appearance of increased microglia reactivity within the preserved tissue outside the direct lesion/disrupted tissue field suggesting soluble paclitaxel had reached these tissue regions and was mildly stimulating this cell population and/or that these cells were responding to cues from cells receiving paclitaxel in the core of the injection ( Figure 2B). Notably, the injection region in the Cremophor EL plus paclitaxel group exhibited a remarkable and statistically significant absence of GFAP-positive cells, such that there was no compact border of reactive astrocytes in this group (Figure 2B,F).
Compared to the Cremophor EL treated groups, the DCH injected groups showed a milder and more focal foreign body response (Figures 2C-F), and in particular, there was significantly less CD68 immunoreactivity in the DCH and DCH plus paclitaxel groups (Figures 2E,F). The DCH-paclitaxel group also exhibited significantly less CD68 staining around the DCH boundary compared to the DCH only control (Figures 2C-F). Host neuronal viability was well preserved around the deposit margins, with only a limited radial zone of about 150 µm radially becoming depleted of NeuN positive neurons after 4 weeks of paclitaxel exposure ( Figure 3B). By contrast paclitaxel delivered using Cremophor-El caused more profuse NeuN positive neuron loss around the injection site ( Figure 3C) as well as a considerable depletion of GFAP positive astrocytes ( Figure 2B). These findings provide strong evidence that the DCH-paclitaxel system caused less damage to host tissue compared to the standard paclitaxel in Cremophor-El vehicle, while at the same time providing a local depot of high drug concentration for at least four weeks after injection.

DCH-paclitaxel depots reduce bioluminescence imaging of hGBM cells, delay
tumor expansion, and confer a significant survival advantage in NSG mice. Since the DCHpaclitaxel system showed prolonged local availability of active drug in combination with a favorable foreign body response in healthy mice compared to the standard Cremophor EL vehicle, we evaluated its effects on in vivo xenotransplantation model of glioblastoma. We injected a 2 µL suspension of 100,000 cells from a patient-derived gliomasphere line (GFPpositive, Luciferase-positive) into the striatum, followed immediately, but separately, by a second injection of 2 µL of DCH placed directly above the hGBM cells ( Figure 4A). Animals received either DCH-only or DCH-paclitaxel. We used non-invasive, semi-quantitative bioluminescence imaging to estimate hGBM tumor size and follow the approximate progression of tumor growth in live animals, We first conducted a small sample pilot study (Cohort 1, n = 4 animals per group) in which bioluminescence imaging was conducted immediately after brain injections and at weekly intervals from 4 weeks onwards ( Figure 4B). In both treatment groups, there was no significant increase in the bioluminescence flux signal at 4 weeks compared to several hours after injection.
In the hGBM/DCH-only group, there was a measurable increase in the bioluminescence flux signal at 5 weeks, which continued to increase exponentially until 10 weeks when the signal plateaued as the hGBM tumors reached a terminal size and animals required euthanasia ( Figure   4B). To evaluate the effect of paclitaxel treatment on hGBM cells, we applied exponential regression analysis on the bioluminescence flux signal values across the interval of 5 to 9 weeks of exponential growth for each animal to determine (i) the extrapolated flux signal at the initial time in which tumor growth kinetics appear to begin exponential growth (A o ); and (ii) the kinetic rate constant of the exponential growth (k). Paclitaxel treatment resulted in (i) an approximately 4fold reduction in hGBM bioluminescence signal compared to the control group at 4 weeks after injection, which was just prior to the start of measurable increase, and (ii) reduced the subsequent rate of signal increase by approximately 23% (Figure 4B).
To test whether the effect of paclitaxel treatment observed in Cohort 1 ( Figure 4B) was robust, we performed an equivalent study on a second, larger group of animals (Cohort 2, n=10).
We conducted bioluminescence imaging over the time period of exponential signal increase from 5 to10 weeks after injection of hGBM cells. In this second, larger group of mice, the DCHpaclitaxel treatment also reduced the initial size of the of hGBM bioluminescence signal by approximately 4-fold, but the rate of the increase in signal was not altered (as shown by the similar gradients for the two curves in Figure 4C) and luminescence signals eventually reached similar levels to that of DCH-only controls. As expected, the high variability of luminescence signal among different animals precluded statistical evaluation and the values obtained can only be regarded as semi-quantitative estimates. Nevertheless, the clear shift to the right in the timing of the exponential increase in bioluminescence in the DCH-paclitaxel group (∆t in Figure 4C) suggested that DCH-paclitaxel treatment reduced the number of hGBM cells, but did not eliminate them and for this reason tumor expansion continued at a similar exponential rate but in a delayed manner. This suggestion was further supported by the observation that DCH-paclitaxel treatment was associated with a statistically significant 2.5 longer survival time in DCH-paclitaxel animals compared with DCH-only controls (p value = 0.0063 in log-rank test) (Figure 4D), representing a 23% extension of life in this hGBM xenotransplantation model. This prolonged survival time correlated well with the approximately 1.5-week delay in the onset and progression of the exponential increase of bioluminescence imaging (∆t, Figure 4C) in DCH-paclitaxel treated mice. Together, these data strongly suggested that our DCH-paclitaxel treatment initially reduced the hGBM tumor load and thereby delayed the expansion of hGBM growth, but eventually the hGBM cells were able to escape treatment. To quantify the number of hGBM cells in different brain locations at 5 weeks after injection we measured GFP staining intensity as a function of brain depth in serial linear units relative to the cortical surface at the injection coordinates ( Figure 5D). These measurements showed that the DCH-paclitaxel treatment significantly reduced the number of hGBM cells in the area immediately around the injection site (i.e. cortical surface to 3 mm deep) by 97% (p<0.001) ( Figure 5D,E). The number of total tumor cells across the entire depth of the brain was significantly reduced by approximately 72.3% (p<0.001) in the paclitaxel group compared to controls (Figure 5D,E). Interestingly, in ventral brain regions 1 or more mm distal from the injection site (Figure 5C), the number of tumor cells was somewhat higher in the paclitaxel group ( Figure 5D,E), suggesting that tumor cells had migrated away from the paclitaxel.

DISCUSSION
Extent of resection (EOR) has been highly correlated to improved survival in low and highgrade glioma [29][30][31][32]. However, the ability to perform near complete tumor resection is dependent on the location of the primary mass within the brain. Tumors located in deep subcortical or midline structures, in close proximity to ventricles, near large vessels or at eloquent brain regions, are often classified as inoperable or unsuitable for large volume tumor resection due to the high risk of severe postsurgical morbidity [33,34]. These unresectable tumors are associated with shorter patient survival [32,33]. The paclitaxel-loaded DCH system represents a potential approach for debulking these difficult to access tumors without causing significant postsurgical morbidity.
In this study we show that DCH can locally deliver a potent hydrophobic chemotherapeutic agent, paclitaxel, to a focal site within the brain and substantively ablate locally residing hGBM Devices using ultrasound [37] or thermo-ablation [38] are being developed to debulk the central tumor mass. These technologies leave a hole in the tumor mass that could potentially be filled with a hydrogel for extended tumor control.
One problem with delivering chemotherapeutic agents at the time of surgery is the concern for wound healing. The initial trials with gliadel were associated with an increase in wound breakdown. It is for this reason, that many systemic chemotherapies are started several weeks after the surgery to allow for the wound to heal. One advantage of minimally invasive injection techniques is that they involve smaller wounds, which are less prone to wound complications and break-down.
In conclusion, this study demonstrates the potential for using DCH to deliver a potent chemotherapeutic agent to a focal site within the brain and effectively chemically debulk a central area of tumor with a resulting survival benefit. This technology brings hope to a population of patients for whom surgical debulking is not possible.      GFP staining intensity was measured as a function of brain depth in serial linear units in the