Use of FLOSEAL® as a scaffold and its impact on induced neural stem cell phenotype, persistence, and efficacy

Abstract Induced neural stem cells (iNSCs) have emerged as a promising therapeutic platform for glioblastoma (GBM). iNSCs have the innate ability to home to tumor foci, making them ideal carriers for antitumor payloads. However, the in vivo persistence of iNSCs limits their therapeutic potential. We hypothesized that by encapsulating iNSCs in the FDA‐approved, hemostatic matrix FLOSEAL®, we could increase their persistence and, as a result, therapeutic durability. Encapsulated iNSCs persisted for 95 days, whereas iNSCs injected into the brain parenchyma persisted only 2 weeks in mice. Two orthotopic GBM tumor models were used to test the efficacy of encapsulated iNSCs. In the GBM8 tumor model, mice that received therapeutic iNSCs encapsulated in FLOSEAL® survived 30 to 60 days longer than mice that received nonencapsulated cells. However, the U87 tumor model showed no significant differences in survival between these two groups, likely due to the more solid and dense nature of the tumor. Interestingly, the interaction of iNSCs with FLOSEAL® appears to downregulate some markers of proliferation, anti‐apoptosis, migration, and therapy which could also play a role in treatment efficacy and durability. Our results demonstrate that while FLOSEAL® significantly improves iNSC persistence, this alone is insufficient to enhance therapeutic durability.


| INTRODUCTION
Glioblastoma (GBM) is an aggressive, stage IV brain cancer and is the most common malignant brain tumor in adults. 1,2 Current standard of care includes tumor resection, radiation therapy, chemotherapy, with alternating electric field therapy as the most recent clinical advancement for patients. 3,4 However, complete tumor resection is often unachievable, as GBM is characterized by highly migratory cells that disperse far from the primary tumor mass, often into the contralateral hemisphere. 3,5 The aggressive, infiltrative nature of GBM results in a high mortality rate and a median patient survival of just 15 months. 1 To combat the migratory nature of GBM, neural stem cells (NSCs) have been investigated as drug delivery vehicles due to their innate tumor-tropism. Several preclinical studies have investigated the persistence, migration, and efficacy of immortalized NSCs bearing a range of therapeutic agents in mice. [6][7][8] Moreover, immortalized, allogeneic NSC therapy has entered human clinical trials for the treatment of GBM (NCT02015819,  NCT03072134,  NCT01172964, NCT02055196, and NCT02192359). [9][10][11][12][13] While NSC therapy shows promise, harvesting a sufficient quantity of autologous NSCs is challenging, and immortalized, allogeneic cells pose an immunogenic risk and have potential for unrestrained proliferation. 14, 15 We have improved upon NSC therapy by developing a rapid, singletranscription factor reprogramming that allows for the direct conversion of fibroblasts to NSCs, known as induced neural stem cells (iNSCs). Here, fibroblasts are isolated from patient skin and stably engineered with lentiviral constructs encoding for the NSC transcription factor SOX2 and the cytotoxic protein TNFα-related apoptosisinducing ligand (TRAIL). Transduced fibroblasts are then cultured in transdifferentiation media to produce therapeutic, tumor-homing iNSCs. 16 While iNSCs are efficacious, persistence in the tumor resection cavity remains a limiting factor. When iNSCs are administered to the resection cavity in saline, more than 50% of iNSCs are cleared by day 10, and nearly all iNSCs are cleared by day 25 postimplantation. 16 To address this, previous studies have demonstrated that seeding mesenchymal stem cells on both TISSEEL ® , a fibrin product, and poly(l-lactic acid) significantly improve cell persistence compared to cells injected in saline. 17,18 Additionally, increased persistence of iNSCs has been observed when seeded on Gelfoam ® , a gelatin matrix, compared to injection in saline. 19 However, the optimal characteristics of an iNSC delivery matrix and the relationship between efficacy and increased persistence remain unknown.
Herein, we investigated the use of the FDA-approved, hemostatic agent, FLOSEAL ® , a gelatin and thrombin mixture, as a delivery matrix and its impact on iNSC persistence and efficacy. When subjected to an area with active bleeding, the thrombin component of FLOSEAL ® polymerizes with circulating fibrinogen to form fibrin. 20,21 We postulated that this rapid polymerization would rapidly encapsulate iNSCs, and the gelatin granules would swell to create a physical barrier, thus providing ample protection from the post-surgical immune response in the resection cavity. We hypothesized that the iNSCs will be able to migrate from the scaffold as FLOSEAL ® degrades. It is expected that significant degradation which allows for cell migration will not be observed until the immune response has become less severe. We theorized that the 6-to 8-week resorption timeframe of FLOSEAL ® would drastically improve iNSC persistence. In this study, we demonstrate that a FLOSEAL ® -based transplant significantly improves iNSC persistence compared to Gelfoam ® , TISSEEL ® , and saline injection.
The marked increase in persistence lead to improved survival outcomes compared to control-treated animals using two unique GBM xenograft models, but extensions over therapeutic cells delivered without a scaffold were modest. These results indicate that persistence alone is inadequate as a predictive marker for therapeutic efficacy, and further research is needed to develop the optimal iNSC delivery matrix.

| Impact of FLOSEAL ® on iNSC gene expression
Previous studies have shown that a material's physiochemical properties can influence gene expression, particularly as it relates to markers of differentiation, proliferation, and migration. [22][23][24][25] To understand FLOSEAL ® 's transcriptomic impact, therapeutic iNSCs were encapsulated in FLOSEAL ® and cultured up to two weeks in the matrix using transwell inserts, which allow for scaffold hydration and nutrient exchange but prevents dissolution of the scaffold in liquid media, to study how the material influenced iNSC gene expression (Figure 2a).
Cell migration from the scaffold was considered negligible due to the lack of chemoattractant present in the culture system. Gene expression of iNSCs in scaffolds was compared to both nontransdifferentiated fibroblasts (Supporting Figure S1) and to day 0 iNSCs (Figure 2b-g). "Day 0 iNSCs" denotes fibroblasts that have been transduced and transdifferentiated to become iNSCs but not placed into a scaffold. NSC, differentiation, proliferation, pluripotency, migration, therapy, and anti-apoptosis markers were monitored; specific genes were selected based on bulk and single-cell RNA sequencing conducted previously by our group. 26,27 The differentiation markers, namely GFAP, TUBB3, and VMAC, were found to remain fairly constant in their expression levels over time, but downregulated compared to the day 0 iNSCs (Figure 2b). Of the NSC markers, To further confirm our findings, we opted to repeat this experiment using three unique batches of iNSCs for the day 14 time point.
In sharp contrast to the findings presented in Figure 2b, GFAP was upregulated while TUBB3 and VMAC remained downregulated ( Figure 3a). Similar to the first gene expression experiment, NESTIN was the only NSC marker to be upregulated, and similar trends were observed for NANOG (Figure 3b,c). The proliferation, anti-apoptosis, and therapy markers remained downregulated as well (Figure 3d,e). As for the migration genes, similar trends were observed wherein SOX2, STC1, VCAM-1, FLT-1, and CXCR4 were upregulated; however, P2RX7 was slightly downregulated in this experiment ( Figure 3f).

| In vivo iNSC persistence
Next, we sought to compare the persistence of nontherapeutic iNSCs using four different implantation techniques: direct injection or

| In vivo iNSC efficacy
After observing significantly improved iNSC persistence with the aid of FLOSEAL ® , we sought to determine if increased persistence correlated to enhanced therapeutic durability and improved survival in mice. We first tested therapeutic durability using the GBM line GBM8. This tumor line was selected for its ability to mimic the invasive nature of tumors seen in the clinic. Three days after implantation, tumors were resected to reflect clinical procedures and treatments were administered into the resection cavity. Tumor size was moni- We next tested a second GBM tumor model using the U87 cell line to mimic the primary nonmigratory tumor mass seen clinically.
Seven days after tumor implantation, the tumors were resected, and iNSC therapies were administered into the resection cavity

| DISCUSSION
Survival rates among GBM patients remain extremely poor, which may be attributed to the inadequate standard-of-care treatment regimen. 30 Currently, a major factor in patient survival is extent of surgical resection of the bulk GBM tumor-still, local tumor recurrence almost always occurs. 31 Novel treatment strategies for GBM are being developed, but often fail before or during clinical trials. 32 The efficacy of targeted therapeutics remains low due to a lack of known, unique GBM receptors. Immunotherapies are difficult to implement for GBM treatment, as overstimulation is known to cause toxicity, and the blood-brain barrier often prevents the accumulation of recruited immune cells within the tumor. 33 Additionally, the immune evasive nature of GBM can minimize efficacy. Repetition of treatments is also undesirable-repeated surgery, chemotherapy, and radiation therapy can be particularly harmful to the patient. 32 The ideal treatment would be administered once yet remain effective until complete tumor eradication is achieved. As such, NSC therapy is a potential alternative to traditional GBM treatment-cells may be administered to the patient at the time of tumor resection, constitutively produce tumoricidal drugs, and remain in the brain while migrating to invasive tumor foci.
Although NSC therapy is a promising new approach for brain cancer treatment, cell persistence remains a challenge to prolonged therapeutic durability, and therefore total elimination of GBM. Our group previously shown the therapeutic promise of iNSC-mediated delivery of TRAIL, 16,18,28 a cytotoxic peptide that initiates apoptosis through interactions with death receptors 4 (DR4) and 5 (DR5), 34 which are highly expressed on GBM tumor cells. 35 Although TRAIL has been investigated as a cancer therapeutic in numerous studies, 36 Previous work by our group has demonstrated the benefit of delivery matrices to increase NSC persistence in the tumor resection cavity. [16][17][18]40 While Gelfoam ® , TISEEL ® , and HySTEM™ each statistically improved NSC persistence compared to direct inject controls, iNSCs still failed to persist beyond 28 days. [16][17][18]40 In tumors such as GBM where recurrence is inevitable, there is a strong need to have the iNSCs persist long enough to migrate vast distances across the brain while still constantly producing their therapeutic payload. To achieve this goal, we sought to combine the best features of previous materials, specifically taking into account biocompatibility, cell binding sites, ease of fabrication, and handleability. Based on these criteria, we selected FLOSEAL ® as our candidate matrix.
As an FDA-approved hemostatic product, FLOSEAL ® 's intrinsic features make it a desirable scaffold material. the rapid reaction of thrombin with fibrinogen completely wrapped the iNSCs in a fibrin web with insufficient time to adhere to gelatin particles. This is a possible explanation for the low seeding efficiency of iNSCs in FLOSEAL ® , which is also reflected by the lower BLI signals exhibited the FLOSEAL ® group in Figure 4.
Furthermore, we observed consistent iNSC density across scaffolds. Because the total volume of gelatin and thrombin (86 mg and 500 μl, respectively) presented here produces enough material to make six scaffolds, we were initially concerned the gelatin particles would be compressed during ejection from the syringe, such that the thrombin and cell solution would be ejected first, therefore producing scaffolds that were initially more liquid and had a higher cell density than those ejected last. Although the first three scaffolds plated were more liquid in nature, no statistical differences were observed in the cell densities, thus ensuring consistent dosing between scaffolds.
Given the volume of literature on the impact of material properties on stem cell proliferation, differentiation, and migration, we sought to understand the impact of FLOSEAL ® on iNSC gene expression. In accordance with previous studies, upregulation NESTIN and SOX2 were observed compared to fibroblasts. 16  were conducted using less than desired cell doses, but despite this, promising trends were observed in therapeutic durability and survival.
Based on these results, there is a clear need for a delivery matrix that allows for an initial burst release of iNSCs to combat tumor cells in the immediate vicinity of the resection cavity and sustained release of iNSCs to support therapeutic durability. Future studies will explore how a combinatory burst release and sustained release of iNSCs impacts therapeutic durability as well as the impact of the immune system on iNSC persistence and treatment efficacy.

| CONCLUSIONS
In this study, we show the impact FLOSEAL ® has on iNSC gene expression, persistence, and efficacy. While encapsulating iNSCs in FLOSEAL ® produced the longest persistence to date, only some mice showed a corresponding increase in survival. Moreover, culturing iNSCs in FLOSEAL ® most notably impacted proliferation, anti-apoptosis, and migration gene expression. These data serve as the framework for future scaffold optimization studies as iNSCs advance toward human clinical trials.

| Transduction
Transduction was performed to produce cells expressing optical reporters and therapeutic proteins. Fibroblasts were transduced by incubating the cells with 8 μg/ml polybrene and the lentiviral cocktail for 24 h at 37 C/5% CO 2 . The next day, the virus-containing media was aspirated and replaced with fresh standard culture media. "Nontherapeutic cells" denotes NHF1 cells transduced with lentiviruses encoding eGFP-Fluc, SOX2, and rtTA. The eGFP-Fluc plasmid construct contained a puromycin-resistance gene to allow for selection of cells.
"Therapeutic cells" denotes NHF1 cells transduced with eGFP-TRAIL, Fluc, SOX2, and rtTA lentiviruses. GBM8 and U87 cells were transduced using lentiviral mCh and Fluc. All lentiviruses were purchased from the Duke Viral Vector Core.

| iNSC production
To manufacture therapeutic and nontherapeutic iNSCs, 2 Â 10 6 trans-  Next, iNSCs were suspended in 8 μl thrombin and pipetted directly on to the fibrinogen. TISSEEL ® scaffolds were allowed to polymerize for approximately 15 min at room temperature and then kept on ice, up to 4 h. Lastly, to seed cells onto Gelfoam ® , a 3 mm diameter hole punch was used to create uniform scaffold discs. Discs were placed into a 96-well plate and 2.5 μl of the iNSC suspension was pipetted directly onto each side of the disc. iNSCs were allowed to adhere for 1 h at 37 C/5% CO 2 , and were then kept on ice until use, up to 4 h.

| Scanning electron microscopy (SEM)
A conjectural count (i.e., not accounting for loss of cells during the seeding process) of 6 Â 10 6 therapeutic iNSCs was encapsulated in FLOSEAL ® as described above. The resulting six scaffolds were polymerized and incubated at 37 C/5% CO 2 for 30 min. Following incubation, scaffolds were submerged in 10% formalin for 30 min. Samples were dehydrated using a graded ethanol series of 50%, 75%, 90%, and 100% ethanol. Next, samples were dried using a critical point drier (Tousimis Autosamdi-931), placed on aluminum stubs, and sputter coated with 6 nm of gold-palladium (Cressington Sputter Coater 108auto). The seeded scaffolds were imaged using a FEI Helios 600 Nanolab Dual Beam System microscope with a 2 kV accelerating voltage.
Each cell concentration was quantified in triplicate to create a standard curve. To quantify seeding efficiency, nontherapeutic iNSCs were encapsulated in FLOSEAL ® as described above, but not polymerized with fibrinogen. The scaffold mixture was divided into six tubes and DNA was isolated. This experiment was done in triplicate to produce a total of 18 scaffold samples.

| Quantitative reverse transcription polymerase chain reaction
Therapeutic iNSCs were produced as described above and encapsulated in FLOSEAL ® at a conjectural density of 2 Â 10 6 cells/scaffold.  Table S1 lists genes and corresponding assay IDs.

| Histology
Mice were anesthetized using 5% inhaled isoflurane. Cardiac perfusion was performed by injecting 5 ml 1X PBS followed by 5

| In vivo iNSC efficacy
Mice were anesthetized using 2.5% inhaled isoflurane and placed into a stereotaxic frame. The surgical site was prepared using 70% isopropyl alcohol and betadine. An incision was made in the skin on the head of the mouse to expose the skull. Next, using a microdrill, a craniotomy was performed in the right hemisphere, between the bregma and lambda points, on the parietal skull plate. Cold saline and Surgicel ® were used to control bleeding. Surgicel ® was removed, and the wound was closed with Vetbond (3M 1469SB). Three days after the craniotomy, mice were again anesthetized and prepared for surgery.
The wound was reopened, and using a stereotaxic auto-injector, 1 Â 10 5 U87-mCh-Fluc cells or 3 Â 10 5 GBM8-mCh-Fluc cells suspended in 3 μl of 1X PBS were infused into the brain parenchyma at stereotaxic coordinates 2.5, À0.5, À0.5 from bregma at a rate of 1 μl/min, avoiding the lateral ventricles. Cells were given 5 min to settle before slowly removing the syringe. The wound was closed with Vetbond. Seven days after implanting the U87 tumors and 3 days after implanting the GBM8 tumors, mice were anesthetized and prepared for surgery. The wound was reopened, and the tumors were resected using fluorescence guidance and a vacuum pump. Once bleeding subsided, therapeutic iNSCs were implanted into the cavity in a 1X PBS suspension or encapsulated in FLOSEAL ® . For the GBM8 efficacy study where two doses of iNSCs were tested, "FLOSEAL ® high" TRAIL denotes mice that received 1.5 Â 10 6 therapeutic iNSCs, and "FLOSEAL ® low TRAIL" denotes mice that received 6 Â 10 5 therapeutic iNSCs conjecturally. All other groups received 1 Â 10 6 therapeutic or nontherapeutic iNSCs, again noting conjecturally for the "FLOSEAL ® Control iNSC" group. In the U87 efficacy study, FLOSEAL ® TRAIL denotes mice that received 1.5 Â 10 6 therapeutic iNSCs conjecturally, and all remaining groups received 1 Â 10 6 therapeutic or nontherapeutic iNSCs again noting conjecturally for the "FLOSEAL ® Control iNSC" group. Postoperative pain was managed with 5 mg/kg of subcutaneous meloxicam 24 h after surgery. iNSC Tumor volume was monitored over time via BLI (AMI HTX, Spectral Instruments Imaging). Animals were euthanized when more than 20% of their original body weight was lost or when the animal displayed physical symptoms of pain-based dehydration, hunched position, tremors, and cold body temperature.