Cytotoxic Engineered Induced Neural Stem Cells as an Intravenous Therapy for Primary Non–Small Cell Lung Cancer and Triple-Negative Breast Cancer

Lung and breast cancer are the most common forms of cancer in the United States, together accounting for an estimated 29% of all new cancer cases diagnosed in 2018; these cancers are among the deadliest, accounting for nearly one-third of all cancer deaths (1). There remains a significant need for more effective systemic treatments to improve the care for patients suffering from these cancers. Emerging data suggest cell therapies may represent a novel and effective new treatment option.

Neural stem cells (NSC) have emerged as promising anticancer drug delivery vehicles due to their innate tumor tropism (2–7). Several studies have shown genetically engineered NSCs home selectively to brain cancer, allowing them to deliver anticancer gene products directly to local and distant tumor foci, significantly reducing tumor volumes and markedly extending survival (2, 7–10). Not limited to the brain, new studies have demonstrated that intravenously-infused tumoricidal NSCs extravasate from vessels to populate tumor foci and significantly reduce human cancer in orthotopic mouse models of breast cancer, lung cancer, and a variety of other tumor types (2, 5–11). These studies demonstrate the potential of tumor-homing NSC therapy for the treatment of aggressive extracranial cancers; as these treatments move forward, the ideal source of NSCs remains in question.

Selecting the most effective NSC carrier will be critical to maximizing persistence, tumor targeting, and ultimately tumor kill in human patient trials. Allogeneic NSCs have shown success in preclinical testing (8, 11–14) and safety in early-stage human clinical trials for gliomas (2, 9, 15). However, the high potential of allogeneic stem cell transplants for immune-mediated rejection (16) is likely to limit their residence time at tumors and ability to deliver therapeutic agents directly to peripheral cancer foci. In contrast, personalized NSC therapy holds the potential to avoid rapid immune clearance (16). This holds the potential to maximize treatment response in patients by increasing the cell carriers that reach tumor foci, increasing cell carrier residence time, and delivering a more durable exposure for the therapeutic window to induce tumor kill. We recently discovered that transdifferentiation, a process that directly converts somatic cells into other adult cell types, can be used to generate tumor-homing drug carriers that regress tumor xenografts, known as human induced NSC (hiNSC; refs. 17, 18). Using SOX2 as a single transcriptional factor, we transformed human fibroblasts into hiNSCs within 1 week of culture (17); we have shown this process is effective in transdifferentiating patient fibroblasts to be used in the treatment of patient-derived intracranial tumors (19). Although the fibroblasts show random movement, the hiNSCs following transdifferentiation show significant directional migration to a tumor (17). The hiNSCs express NSC markers SOX2 and Nestin and the neuronal marker TUJ-1 with minimal expression of the astrocyte marker GFAP or pluripotency markers NANOG or OCT4 (17). In vivo, hiNSCs maintained expression of the NSC marker Nestin whereas expression of neuronal markers or pluripotency markers was virtually absent (17, 19). In the clinic, hiNSC therapy will be a patient-specific approach. In a recent study, we utilized canine models to demonstrate the feasibility of autologous iNSC generation and implantation on a scale that more closely mirrors human patients (20). The migratory hiNSCs are also genetically engineered to secrete cytotoxic proteins such as TNFα apoptosis inducing ligand (hiNSC-TRAIL), enabling the cells to track down and kill infiltrating cancer cells after infusion. Although one initial study tested the ability of NSCs to reduce peripheral tumor burden in lung cancer (4), the therapeutic potential of hiNSCs for orthotopic breast and lung tumors has not yet been fully explored, despite the prevalence of these diseases. Moreover, no studies have investigated the efficacy of hiNSCs administered intravenously, a much less invasive technique where the ease of infusion and redosing provides significant clinical advantages over intratumoral injections.

Herein we provide the first investigation into the intravenous infusion of rapidly transdifferentiated hiNSCs for the treatment of primary lung and breast cancer. Utilizing kinetic bioluminescence imaging, we show hiNSCs migrate rapidly to both human triple-negative breast cancer (TNBC) and non–small cell lung cancer (NSCLC) in co-culture assays. Furthermore, intravenously-infused hiNSCs populated both tumor types in orthotopically-established mouse models of cancer and persisted at the tumor site for 14 days postinfusion. Exploring the antitumor efficacy of hiNSC therapy, we show that hiNSC-TRAIL induced dose-dependent killing in co-culture assays. In vivo, serial bioluminescence imaging showed that intravenous hiNSC-TRAIL therapy reduced tumor burden while extensive toxicity testing showed the treatment did not induce marked toxicity to healthy tissue. Overall, these results suggest intravenously-infused hiNSC-TRAIL is a safe and effective potential treatment for two of the most common types of cancer.

NSCLC cell lines A549 and NCI-H460 and hTERT-immortalized human fibroblast line NHF1 were obtained from University of North Carolina Tissue Culture Facility. TNBC cell line MDA-MB-231-Br was obtained via MTA (Toshi Yoneda, PhD). All cell lines were cultured in DMEM (Gibco) supplemented with 10% (v/v) heat inactivated FBS (Millipore), 100 Units/mL penicillin (Gibco) and 100 μg/mL streptomycin (Gibco), and 0.01% (v/v) Plasmocin (Invivogen). Cell lines were treated for mycoplasma with 0.05% (v/v) Plasmocure (Invivogen) for 2 weeks and were tested for mycoplasma with the MycoAlert Mycoplasma Detection Kit (Lonza).

A total of 5 × 10⁴ hTERT-immortalized NHF1s were seeded in 6-well plates and transduced with LV-GFP-TRAIL, LV-GFP-Nano Luciferase (NLuc; ref. 21), or LV-mC-Firefly Luciferase (FLuc) with 8 μg/mL polybrene (Thermo Fisher Scientific) for 12 hours in DMEM supplemented with 10% FBS, 1% penicillin/streptomycin, and 0.01% Plasmocin. After initial transduction, cells were transduced with a cocktail of LV-SOX2 and LV-rtTA with polybrene for 12 hours in supplemented DMEM. A week prior to use of cells, 1 × 10⁶ transduced cells were seeded per T175 flask in supplemented DMEM. Cells were cultured in STEMdiff Neural Induction Media (STEMCELL Technologies) with 2 μg/mL doxycycline (Sigma), changing media every other day for 5 days. Cells were detached using Accutase Cell Detachment Solution (STEMCELL Technologies) at room temperature and strained using 100 μmol/L Falcon Cell Strainers (Thermo Fisher Scientific). SOX2 expression was verified by staining. Briefly, cells were fixed with 2.5% formalin for 20 minutes and rinsed with PBS. Following incubation in blocking buffer for 1 hour at room temperature, cells were incubated with 1:100 rabbit anti-SOX2 antibody (Abcam) for 1 day at 4°C. After washing thrice with PBS, cells were incubated with 1:1,000 goat anti-rabbit cross-adsorbed secondary antibody, Alexa Fluor 568 (Thermo Fisher Scientific) for at least 1 hour. Cells were washed thrice with PBS and incubated in Hoechst for 20 minutes. Cells were imaged using an EVOS FL Auto Cell Imaging System (Life Technologies).

A total of 5 × 10³ cancer cells (MDA-MB231-Br or A549) labeled with stable expression of FLuc were seeded in 24-well plates in 10% FBS, 1% penicillin/streptomycin, and 0.01% Plasmocin DMEM. The following day, wells were seeded with varying numbers of hiNSC-TRAIL cells, from 0 to 10,000 cells/well. After 1, 3, and 7 days, viability of the cancer cells was assessed by adding 0.225 μg/mL XenoLight D-luciferin (PerkinElmer) and measuring luminescence using a SynergyH1 microplate reader (BioTek) at 1-second per well measurement (N = 4) or IVIS Kinetic (Caliper Life Sciences; N = 4–8). Viability was determined by dividing the luminescent signal of these wells by their untreated counterparts at each time point. To test viability of hiNSCs after exposure to recombinant human TRAIL (Sigma), we incubated hiNSCs expressing FLuc with 1 to 500 ng/mL TRAIL for 48 hours before assessing luminescence on SynergyH1 microplate reader as described above.

24-well plates were prepared by incubating the empty wells with 10 μg/mL purified mouse laminin (EMD Millipore) in PBS for 12 hours at 37°C. Following incubation, media was aspirated and wells air-dried. Two-chamber cell culture inserts (Ibidi) were placed in each well. 1 × 10⁴ hiNSC-mC-FLuc were seeded on the left side; 2 × 10⁴ A549-GFP-NLuc or MDA-MB231-Br-GFP-NLuc in 80 μL were seeded on the right side of inserts (0.5 mm separation). Control wells (N = 4) had 1 × 10⁴ hiNSC-mC-FLuc were seeded on the left side with nothing in the right insert. Both hiNSCs and tumor cells were cultured in STEMdiff Neural Induction Media (STEMCELL Technologies). 24 hours after seeding cells, inserts were removed, and wells were filled with STEMdiff Neural Induction with 2 μg/mL doxycycline (Sigma) and 2% FBS. Cells were imaged at 10x magnification every 2 hours for 96 hours with EVOS FL Auto Cell Imaging System (Life Technologies). Ten random cells from each beacon were selected and individually tracked over time (ImageJ). Cell tracking data collected in Fiji was uploaded into R for statistical analysis and generation of rose and summary violin plots. Number of cells migrated was determined by manually counting the cells that passed a starting line designated by the leading hiNSC edge at 0 hour.

To follow tumor volume or hiNSC-mC-FLuc distribution and persistence, serial bioluminescent imaging (BLI) was performed as previously described (3, 17). Mice were administered D-luciferin (3 mg per mouse in 200 μL of PBS) via intraperitoneal injection. Fifteen minutes following injection, photon emission was measured using Ami HT (Spectral Instruments Imaging) or IVIS Kinetic (Caliper Life Sciences). Luminescence was quantified through analysis with Aura (Spectral Instruments Imaging).

Following anesthetization with 5% isoflurane, mice were perfused via intracardiac puncture with 10 mL PBS. Lungs were inflated with 10% neutral buffered formalin (NBF). Lungs, liver, kidneys, and spleen were incubated in 10% NBF for 72 hours before storage in 70% ethanol. Organs were embedded in paraffin, sectioned, and stained with hematoxylin and eosin (H&E) by UNC Animal Histopathology Core and UNC Translational Pathology Laboratory.

For the lung tumor model, A549-GFP-NLuc cells were prepared by washing cells in suspension with PBS and preparing 1.5 × 10⁶ cells in 50 μL 1:1 (v/v) Matrigel matrix (Corning) in PBS. Following anesthetization with isoflurane, female athymic nude mice were placed in a lateral decubitus position. Cells were injected into intercostal space of left lung. Tumor growth was monitored using BLI by intravenous injection of furimazine (Promega; 1:20 v/v in PBS). Tumors grew for 2 weeks. For the breast tumor model, MDA-MB231-Br-GFP-NLuc were prepared as above and brought to a concentration of 2 million cells in 60 μL 1:1 (v/v) Matrigel matrix (Corning) in PBS. Cells were injected into mammary fat pad. Tumors grew for 10 days. 1 × 10⁶ hiNSC-mC-FLuc cells suspended in 200 μL were intravenously injected. hiNSC-mC-FLuc was followed by BLI following intraperitoneal injection of D-luciferin. hiNSC-mC-FLuc cells were also injected as described above in mice without tumors. Mice were sacrificed at 3, 7, and 14 days after hiNSC-mC-FLuc injection (N = 2 per time point). Lungs and fat pad tumors were processed as described above with the following exception: no perfusion was performed prior to removal of lungs to avoid disruption of cells for mice with lung tumors. Lungs and fat pad tumors were incubated in 10% NBF for 24 hours prior to storage in 30% sucrose at 4°C. Samples were frozen in optimal cutting temperature media. Using a cryostat, 10 μm organ sections were collected, stained with Hoechst (Thermo Fisher Scientific), and mounted with ProLong Gold Antifade Mountant (Invitrogen). Fluorescent images were collected using Olympus FV3000RS confocal microscope, and fluorescent signal was quantified using ImageJ. All images were optimized for brightness/color contrast.

Lung H460-mC-FLuc [5 × 10⁴ cells in 30 μL 2:1 Matrigel (v/v) in PBS] or breast MDA-MB231-Br-mC-FLuc were implanted as described above in female athymic nude mice. Tumor volumes were monitored using BLI. Maximum tolerated dose of 1 × 10⁶ hiNSC-TRAIL cells suspended in 200 μL PBS were injected intravenously 4 and 8 days after H460 tumor implantation or 5 days after MDA-MB231-Br implantation. The orthotopic H460 tumor study was completed twice to increase N, and multiple unpaired t tests were used to determine no significant difference between controls and hiNSC-TRAIL groups of the first and second studies. Mice without visible thoracic tumors with a 2-second exposure via BLI by Day 14 were excluded. BLI values for individual days were excluded if a value dropped by an order of magnitude from previous time point and then increased by about an order of magnitude by the next time point, presumably because of a poor intraperitoneal injection of luciferin. N = 15 was used for controls, and N = 16 was used for treatment mice with H460 tumors. N = 5 for treated mice and N = 4 for control mice with MDA-MB231-Br tumors. Grubbs' test was used to identify outliers, and these mice were excluded. Mice were sacrificed when they lost >20% of their highest body weight, exhibited signs of distress, or if a visible tumor exceeded 2 cm in length. Survival is reported as median with 95% confidence intervals (CI) of median. Survival data reported for the H460 model are from the second study. The study was ended after 100 days for the MDA-MB231-Br model.

To test the toxicity of either carrier hiNSC-mC-FLuc or cytotoxic hiNSC-TRAIL, 1 × 10⁶ hiNSC-mC-FLuc or 1 × 10⁶ hiNSC-TRAIL cells in 200 μL PBS were injected intravenously into female athymic nude mice (N = 4 per time point). 3, 7, 14, and 28 days following injection, 700 μL of blood was collected via intercostal puncture, and organs were harvested following perfusion with PBS. Blood was collected in K₂ EDTA Microtainer (BD Biosciences) and in clot activator Microtainer (BD Biosciences), from which serum was extracted. Blood and serum samples were analyzed by the UNC Animal Histopathology Core.

All experimental mouse protocols were previously approved by the Institutional Animal Care and Use Committees at University of North Carolina–Chapel Hill.

Data were analyzed using GraphPad Prism and R. Multiple unpaired t tests without correction for multiple comparisons were used to compare two groups. Multiple groups were analyzed by one-way ANOVA, followed by Dunnett post hoc test. Survival analysis was conducted using a log-rank test. Unless otherwise specified, all values are expressed as mean ± SEM, and differences were considered significant when P < 0.05.

One of the most unique and beneficial attributes of hiNSC therapy is the ability of the cells to actively seek out and populate local and distant tumors. This ability has been proven extensively in mouse models of brain cancer where hiNSC are directly infused into brain tissue (17). However, it is unknown whether SOX2-expressing hiNSCs (Supplementary Fig. S1) exhibit similar homing capability to extracranial tumors or whether their tumor-homing capacity remains intact when infused intravenously into the blood stream. To answer these questions, we first investigated the tumoritropic properties of hiNSCs to TNBC and NSCLC cells using our mixed culture models where hiNSC migration is tracked in real-time. hiNSC-mC-FLuc cells were seeded 500 μmol/L apart from A549-GFP-FLuc lung human cancer cells or MDA-MB231-Br-GFP-FLuc human TNBC cells in two-chamber cell culture inserts (Fig. 1A). Twenty-four hours after seeding, the culture inserts were removed and kinetic high-resolution images were captured every 2 hours for 96 hours to track the movement of hiNSC towards the cancer cells. Analysis of time-lapse images show that hiNSCs directionally migrated towards both A549 and MDA-MB231-Br cells (Fig. 1B). Quantification further supported kinetic imaging results, showing a mean of 14 and 29 cells migrated past a starting line designated by the leading edge of the hiNSCs at time 0 towards A549 and MDA-MB231-Br, respectively, at 96 hours compared with an average of only three cells in control wells where hiNSCs were plated without nearby tumor cells (Fig. 1C). Single-cell migratory path analysis of directionality showed the mean directionality index was 0.300 ± 0.016 and 0.352 ± 0.022 for migrating hiNSCs towards A549 and MDA-MB231-Br compared with 0.166 ± 0.017 in control wells. The hiNSCs seeded adjacent to either A549 or MDA-MB231-Br showed farther mean final cell displacement (180 ± 11 or 205 ± 14 μmol/L, respectively) than hiNSCs seeded without nearby tumor cells (112 ± 13 μmol/L; Fig. 1D–F). Directional migration of hiNSCs was also observed toward H460 NSCLC cells (Supplementary Figs. S2A–S2C) while non-transdifferentiated fibroblasts did not show enhanced tropism toward H460 cells over the control (Supplementary Figs. S3A and S3B). In addition, we did not observe significant directional movement of hiNSCs toward nontumor cell lines (Supplementary Fig. S4). These data suggest hiNSCs possess tumor homing capability toward TNBC and NSCLC cell lines in vitro.

In human patients, the majority of cell therapies are infused intravenously for the treatment of peripheral tumor types (22). As we found hiNSCs migrate to both TNBC and NSCLC in co-cultures, we sought to explore the ability of hiNSCs to target both tumor types in vivo following intravenous infusion and determine the residence time of the cells at the tumor. We first established either A549-GFP-NLuc orthotopic lung tumors or MDA-MB231-Br-GFP-NLuc fat pad tumors in athymic nude mice (Fig. 2A and B). Once tumor establishment was confirmed by serial imaging, 1 × 10⁶ hiNSC-mC-FLuc cells were administered intravenously into tumor-bearing mice, and hiNSC distribution was monitored by firefly luciferase BLI. In animals with orthotopic lung cancer xenografts, analysis of BLI signal showed hiNSCs were detectable at tumor foci in the thorax shortly after infusion. Thoracic BLI signal gradually declined over 11 days in mice both with and without tumors (Fig. 2C; Supplementary Fig. S5). To visualize hiNSCs at levels below the limit of detection by BLI, we performed fluorescent microscopic analysis of tissue sections from a subset of mice sacrificed at each time point. Our high-resolution imaging showed hiNSC-mC-FLuc were present and co-localized with NSCLC GFP+ tumor within 3 days postinjection and could still be detected 14 days postinfusion (Fig. 2C and E). In mice bearing TNBC tumors, fluorescent analysis showed hiNSCs were detected at the tumor by day 3, reached peak accumulation at day 7, and gradually declined through day 14 (Fig. 2D and F). Fluorescent analysis was used to characterize migration of hiNSCs toward TNBC tumors because the BLI signal of these cells was below the limit of detection. In both tumor models, hiNSCs were observed to colocalize to tumors within 3 days and persisted for at least 14 days.

To take advantage of the hiNSCs' tumor-homing migratory capabilities, we produced a line of hiNSC-TRAIL to test the ability of hiNSCs to deliver therapeutics to tumors. We performed co-culture assays with ratios ranging from 10:1 to 1:5 of tumor cells (NSCLC H460 or TNBC MDA-MB231-Br) to hiNSC-TRAIL cells for up to 7 days (Fig. 3A). Cell viability assays and summary graphs showed hiNSC-TRAIL cells significantly reduced both H460 and MDA-MB231-Br tumor cells in a dose- and time-dependent manner. Viability as a percentage of control is presented in Fig. 3B and C, and raw luminescent values are presented in Supplementary Figs. S6A and S6B. To next investigate the kinetics of the tumor-killing by hiNSC-TRAIL, we performed real-time serial imaging of tumor cell viability when incubated with hiNSC-TRAIL cells (Fig. 3D–G). Analysis of kinetic killing curves showed a left-shift in the curve with increasing concentrations of hiNSC-TRAIL cells for both tumor cell lines, indicating tumor killing in a rapid and dose-dependent manner. At a 1:1 ratio of tumor: hiNSC-TRAIL, we found the therapy required 42 and 18 hours to induce 50% reduction in tumor cell signal of H460 and MDA-MB231-Br, respectively. Despite initial differences in the rate of tumor kill, fluorescent images captured 96 hours posttreatment showed hiNSC-TRAIL therapy was able to reduce tumor cell signal by >98% relative to nontreated cells at the highest dose. Taken together, these data suggest hiNSC-TRAIL therapy induces killing of both TNBC and NSCLC cells. Importantly, decreasing the tumor: hiNSC-TRAIL ratio not only increased the percentage of tumor cells killed, but also the rate at which the tumor cells died.

To investigate the efficacy of intravenous hiNSC therapy for lung and breast cancer, we tested the efficacy of hiNSC-TRAIL therapy against primary NSCLC and TNBC in vivo. Orthotopic H460-mC-FLuc or MDA-MB231-Br-mC-FLuc tumors were established in the lungs or fat pad of nude mice, respectively. One week after tumor implant, 1 × 10⁶ hiNSC-TRAIL cells or control hiNSC-GFP cells were injected intravenously (Fig. 4A and B). In mice bearing orthotopic H460 tumors established in the lungs, mice treated with hiNSC-TRAIL showed a significant reduction in tumor volume days 7 and 14 after the first dose of intravenous hiNSCs; the mean tumor volumes of mice treated with hiNSC-TRAIL were 33.1% and 29.8% of the volumes of control-treated animals 7 and 14 days posttreatment, respectively (Fig. 4C and E; Supplementary Fig. S7A). The highly aggressive nature of this model led to rapid tumor rebound, although we did observe a modest trend toward improvement in overall median survival (49 days with 95% CI, 24–104 days for mice treated with hiNSC-TRAIL; 35.5 days with 95% CI, 22–82 days for control mice; Fig. 4G). hiNSC-TRAIL therapy induced strong tumor kill in TNBC tumors. Serial imaging showed mice with human TNBC infused with hiNSC-TRAIL therapy had a mean tumor volume of 28.2% and 7.3% compared with control-treated animals 21 and 36 days posttreatment, respectively (Fig. 4D and F; Supplementary Figs. S7B–S7D). For both tumor models, we observed significant reductions in tumor volumes in mice treated with intravenous hiNSC-TRAIL based on multiple unpaired t-tests without correction for multiple comparisons. Given the large number of time points, when a two-stage setup was used to correct for multiple comparisons, the adjusted P values were >0.05. A strong though non-significant trend toward increased median survival was observed (100 days with 95% CI, 64–100 days for TNBC mice treated with hiNSC-TRAIL; 44 days with 95% CI, 44–100 days for control mice; Fig. 4H).

To determine the safety of hiNSC therapy, we had previously characterized both immortalized fibroblasts (17) and nonimmortalized patient fibroblasts (19), in which we observed no oncolytic transformation. To further investigate the safety of hiNSC therapy, we infused a single maximum tolerated dose of either nontherapeutic hiNSC-mC-FLuc or therapeutic hiNSC-TRAIL to evaluate the impact of both the cell carriers and the carrier + drug, respectively. On subsequent days over a 28-day time-course postinfusion, lung, kidney, liver, and spleen were harvested, and histology, hematology, and blood chemistry values were evaluated and compared with blank, noninfused mice. H&E histochemistry of major organs demonstrated all examined organs were within normal limits regardless of whether the mice received the carrier + drug or the carrier alone. Rare, scattered cell clusters with large nuclei were identified within alveolar septal walls of lungs harvested from early time points in the postinfusion and carrier + drug groups; these are presumed to be the infused hiNSCs. The cell clusters were no longer observed in either Day 28 sample. Nonspecific lymphocytic infiltration in liver samples was moderately increased in hiNSC-infused mice compared with blank, noninfused mice. Spleen samples in both groups showed moderately increased follicle size with occasional formation of germinal centers at each time point with the most activity seen in mice 3 days after infusion with the carrier. We speculate that hepatic lymphocytic and splenic responses are likely due to murine immune responses against human hiNSC infusion. Minimal interstitial nephritis was occasionally observed in the kidneys of mice infused with the carrier (Fig. 5). Hematology and blood chemistry values showed no significant sustained alterations in hiNSC-infused mice compared with blank mice (Table 1). Mice infused with carrier only showed mild, transient increases in creatinine that resolved within 28 days of infusion. Minor transient decreases in BUN and creatinine were observed in mice infused with carrier + drug that each resolved. Taken together, the pathology, hematology, and clinical chemistry values indicate no significant toxic effects of either carrier or carrier + drug following intravenous infusion. TRAIL itself is only toxic to hiNSCs or fibroblasts at concentrations (Supplementary Figs. S8A and S8B) well above documented levels achieved through production by hiNSCs, about 2 fg TRAIL/cell/hour (19) except if hiNSC-TRAIL outnumbers tumor cells 50:1 (Supplementary Fig. S8C). TUNEL stains indicated no significant apoptosis in healthy tissue following intravenous infusion of hiNSC-TRAIL (Supplementary Figs. S9A–S9D).

In this study, we investigated the ability of human fibroblasts that were rapidly transdifferentiated into hiNSCs to home to and kill primary NSCLC or TNBC tumors following intravenous infusion. While previous studies have investigated the use of hiNSCs to treat glioblastoma (17, 18) or other types of stem cells to treat extracranial tumors (4, 23–25), to our knowledge, this study marks the first effort to treat primary breast or lung tumors with hiNSCs. Migration of hiNSCs has been well-characterized in the brain parenchyma (18, 26), but this is the first study to investigate the distribution of intravenously-administered hiNSCs. Our in vitro results show directional migration of hiNSCs toward both TNBC and NSCLC. Given that minimal and nondirectional migration was observed in the control group, we infer the migration observed toward the cancer lines was not merely random motion or cell proliferation. hiNSCs are thought to follow chemokine gradients produced by tumor cells or surrounding areas of hypoxia (6), and we expected that an increasing number of hiNSCs would migrate toward cancer cells as the chemokine gradient is produced over time. Although the number of migrating cells toward NSCLC was not significantly different between the control and treatment groups, the number of migrating hiNSCs toward tumor cells trended toward an increase over time to tumors, and significantly more hiNSCs migrated toward TNBC cells than controls. However, counting cells that pass the starting line discounts movement of cells that start far from the starting line. To account for differences in starting position, we also report directionality and displacement. We observed significantly increased directionality indices and final displacements of hiNSCs toward both tumor lines than in the control group, suggesting hiNSCs do migrate to both NSCLC and TNBC in vitro. To determine if these in vitro results translate to an in vivo setting, we tested the migration of hiNSCs to TNBC fat pad tumors or NSCLC lung tumors. We observed colocalization of hiNSCs at both NSCLC and TNBC tumor foci as early as 3 days and as late as 14 days after intravenous infusion. These suggest the cells not only migrate to tumors quickly but also persist, albeit at low levels, for a relatively long period of time. These persistence results are consistent with our previous studies of hiNSCs in the brains of mice (17), and an autologous infusion of iNSCs likely would increase persistence even more, as our previous canine iNSC study showed persistence for more than 80 days (20). The presence of a tumor does not appear to significantly impact hiNSC persistence or overall distribution, consistent with similar studies with MSCs (27). Importantly, our results here also suggest the migratory ability of hiNSCs is not limited by a specific tumor line as we observed migration toward both NSCLC and TNBC cells. These results indicate the versatility of these therapeutic hiNSCs to target a variety of primary tumors extracranially.

Our in vitro co-culture killing assay demonstrated that both NSCLC and TNBC cell lines are sensitive to cytotoxic TRAIL. TRAIL has been shown to increase the activity of caspase 3 and 9 as well as annexin V and PI in tumor lines such as the NSCLC line H460 (28). However, the TNBC line was much more sensitive to TRAIL than the NSCLC line. Differential TRAIL resistance for various tumor lines has been widely documented (29–33), and so partial resistance to TRAIL is not unexpected. Thus, in the proceeding experiments, we anticipated that the TNBC line would respond better to hiNSC-TRAIL than the NSCLC line. Despite the partial sensitivity to TRAIL observed in our in vitro studies and the highly invasive nature of the NSCLC line, the NSCLC tumors in vivo did respond well to hiNSC-TRAIL therapy. Both the mice with TNBC tumors and those with NSCLC that were treated with hiNSC-TRAIL therapy did show a reduction in the tumor BLI signal within 2 weeks of the initial infusion. Mice with NSCLC tumors treated with hiNSC-TRAIL therapy trended toward increased survival, likely limited by the tumor line's partial TRAIL resistance. We also saw a strong trend toward increased survival of mice with TNBC tumors treated with hiNSC-TRAIL, indicating the potential of this therapy in a TRAIL-sensitive tumor model.

Although TRAIL is a useful cytotoxic protein for proof-of-concept studies and is well-tolerated in humans (34), hiNSCs are not limited to using a monotherapy with TRAIL. Given the prevalence of tumor resistance to TRAIL (25, 30, 31, 33, 35–37), potential next steps include combining TRAIL with a drug/prodrug system. Our lab has previously shown that hiNSCs can be engineered to produce the prodrug-activator thymidine kinase, which has shown good results against glioblastoma in vivo (3, 17). In this treatment paradigm, if a tumor shows resistance to TRAIL, a prodrug can be administered to initiate an additional therapy. Another option is to combine hiNSC-TRAILs with radiation therapy. Given the prevalence of radiation therapy to treat both NSCLC and TNBC (38–41), it will be key to study this interaction in the future. In this treatment paradigm, hiNSC-TRAIL would scavenge remaining tumor sites postradiation. Furthermore, radiation can sensitize tumors to TRAIL (33) and promote migration by increasing expression of hypoxic and inflammatory markers (6, 42–44).

The therapeutic effect of hiNSC-TRAIL cells was limited in this model by the flow of cells through vessels. Our results indicate that intravenously-infused hiNSCs immediately travel to the lungs and persist for at least 2 weeks. Our cell therapy takes advantage of this first pass through the lungs in the case of treating orthotopic lung tumors. We anticipate we would observe a stronger therapeutic effect in the case of TNBC if the cells were able to pass through the lungs more quickly. Previous studies have indicated that smaller cells, such as bone marrow-derived mononuclear cells (BMMC) with a cell volume of 150 μm³, are more likely to pass through the lungs into arterial circulation than larger cells, such as NSCs with a cell volume of about 2,500 μm³ (45). Human T-cells, which have a volume the same order of magnitude as BMMCs (46), have been shown to distribute throughout the mouse following intravenous infusion in a similar fashion to a human (47). Thus, we anticipate the reason for the persistence of our human hiNSCs in the lungs of mice is due at least in part to the large hiNSCs being unable to pass through small mouse lung capillaries (48, 49).

Despite the lengthy persistence of hiNSCs in the lungs, they do not appear to cause significant toxicity. Following infusion of hiNSCs, there was no concerning inflammatory response observed in the lungs, liver, kidneys, or spleen. Only mild reactive changes were observed in the germinal centers of the spleen at early postinfusion time points that appear to decrease over time. Elevated creatinine levels returned to baseline 1 month after infusion. This correlates with our previous canine study in which induced neural stem cells transdifferentiated from canine fibroblasts showed no signs of significant iNSC-related tissue abnormality for greater than 84 days after injection (20). When compared with the significant toxicities associated with current chemotherapies such as taxol (50), the variations we observed from baseline in histology, hematology, and clinical chemistry values after infusion are comparably mild and transient.

Although we recognize the limitations of our current model, we still observed migration to both NSCLC and TNBC tumors following intravenous infusion of hiNSCs. To our knowledge, this is the first study to investigate the potential of any NSC to treat NSCLC following intravenous infusion. Furthermore, this is the first study to use hiNSCs to treat either NSCLC or TNBC. Mice with TNBC or NSCLC tumors showed significant reduction following intravenous hiNSC-TRAIL therapy. To further improve the therapeutic efficacy of these cells, we are currently in the process of developing a second generation of hiNSCs with more rapidly migrating cells and increased residence time. Moreover, future studies should also examine the impact of hiNSCs on patient-derived tumors. Overall, these are promising findings in the development of this cell therapy with implications to treat a variety of different cancers.

J.R. Bago reports personal fees from Falcon Therapeutics outside the submitted work and has a patent for WO2016160918A1 pending to Falcon. S.D. Hingtgen reports grants from NINDSI (R01NS099368), The Eshelman Institute for Innovation, UNC-CH, and NCI (F30CA243270) during the conduct of the study; also has a patent for WO2016160918A1 pending to Falcon. A.R. Mercer-Smith reports grants NCI (F30CA243270) during the conduct of the study. K.T. Sheets has an ownership interest in Falcon Therapeutics, Inc., which has licensed aspects of hiNSC technology from the University of North Carolina at Chapel Hill. C.K. Anders has research funding from PUMA, Lilly, Merck, Seattle Genetics, Nektar, Tesaro, and G1-Therapeutics. C.K. Anders has compensated consulting roles from Genentech, Eisai, IPSEN, and Seattle Genetics with royalties from UpToDate, and Jones & Bartlett.

A.R. Mercer-Smith: Conceptualization, formal analysis, funding acquisition, investigation, visualization, methodology, writing–original draft, writing–review and editing. W. Jiang: Conceptualization, formal analysis, investigation, visualization, methodology, writing–original draft, writing–review and editing. J.R. Bago: Conceptualization, methodology, writing–review and editing. A. Valdivia: Methodology, writing–review and editing. M. Thang: Data curation. A.S. Woodell: Software, formal analysis, writing–review and editing. S.A. Montgomery: Formal analysis, writing–review and editing. K.T. Sheets: Conceptualization, writing–review and editing. C.K. Anders: Conceptualization, resources, funding acquisition, writing–review and editing. S.D. Hingtgen: Conceptualization, supervision, funding acquisition, writing–review and editing.

The authors would like to thank the Biomedical Research Imaging Center, the Tissue Pathology Laboratory, the Animal Histopathology Core, the Histology Research Core, and the Neuroscience Microscopy Core at The University of North Carolina-Chapel Hill. Schematic figures were made using BioRender. The authors would also like to thank Dr. Chad Pecot's lab and Michael Marand. This work was supported by Eshelman Institute for Innovation and the NCI of the NIH under Award Number F30CA243270. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.