A Novel Molecule Integrating Therapeutic and Diagnostic Activities Reveals Multiple Aspects of Stem Cell-Based Therapy

Abstract

Stem cells are promising therapeutic delivery vehicles; however pre-clinical and clinical applications of stem cell-based therapy would benefit significantly from the ability to simultaneously determine therapeutic efficacy and pharmacokinetics of therapies delivered by engineered stem cells. In this study, we engineered and screened numerous fusion variants that contained therapeutic (TRAIL) and diagnostic (luciferase) domains designed to allow simultaneous investigation of multiple events in stem cell-based therapy in vivo. When various stem cell lines were engineered with the optimized molecule, SRL_OL₂TR, diagnostic imaging showed marked differences in the levels and duration of secretion between stem cell lines, while the therapeutic activity of the molecule showed the different secretion levels translated to significant variability in tumor cell killing. in vivo, simultaneous diagnostic and therapeutic monitoring revealed that stem cell-based delivery significantly improved pharmacokinetics and anti-tumor effectiveness of the therapy compared to intravenous or intratumoral delivery. As treatment for highly malignant brain tumor xenografts, tracking SRL_OL₂TR showed stable stem cell-mediated delivery significantly regressed peripheral and intracranial tumors. Together, the integrated diagnostic and therapeutic properties of SRL_OL₂TR answer critical questions necessary for successful utilization of stem cells as novel therapeutic vehicles.

Introduction

Recently, stem cell-based therapies have gained increasing acceptance as viable and effective treatment options for a variety of previously incurable diseases [1, 2]. Because of their unique ability to regenerate damaged tissue, stem cells have been applied successfully for the treatment of numerous degenerative disease including cystic fibrosis and myocardial infarction. In contrast to these diseases, the regenerative properties of stem cells alone are inadequate for treating many acquired diseases such as cancer, AIDS, and vascular hyper-tension. For such diseases that are resistant to other therapeutic strategies, stem cells offer potential as effective treatment options when utilized as vehicles for the delivery of therapeutic proteins. We [3, 4] and others [1, 5-7] have shown that stem cells can be efficiently engineered with DNA-encoding secreted therapeutic proteins. Coupled with their unique ability to selectively home to sites of damage or malignancy [3, 8, 9], engineered stem cells are ideal vehicles for the delivery of therapeutic proteins to reduce tumor burden or rescue diseased tissue, while retaining the potential to regenerate damaged tissue after success of the therapy. Additionally, the capacity for stable secretion by engineered cells may overcome limitations of short drug half-life and dose-limiting nonspecific toxicities that plague i.v. or systemic infusion of purified therapies. However, vital unanswered questions remain that are crucial for effective cell-based treatment, including differences in secretion levels between cell lines, duration of delivery, and systemic distribution of therapeutic proteins delivered by stem cells, as well as how differences in these parameters affect the therapeutic potential of different cell-based treatments. Currently no methods exist that allow serial noninvasive tracking of proteins secreted by engineered stem cells to simultaneously monitor delivery, fate of the engineered cells, and therapeutic efficacy in vivo.

The development of molecules to most effectively answer these questions requires a unique multifunctional molecule containing both a diagnostic domain for monitoring pharmacokinetics of stem cell delivery and a therapeutic domain for disease treatment. Although multifunctional molecules are emerging, most are based on nanoparticles or synthetic compounds [10] and are unable to allow monitoring of secreted proteins from cells by noninvasive imaging. To permit permanent integration into the stem cell genome, we engineered viral vectors encoding a novel fusion variant containing two domains optimized for cell-based applications. The first domain consisted of light-emitting luciferase proteins to permit direct extracellular visualization and monitoring of levels, time of delivery, and localization of stem cell-delivered proteins by simple bioluminescent imaging. The second domain was composed of a secreted variant of the pro-apoptotic protein tumor necrosis factor-related apoptosis-inducing ligand (S-TRAIL), a promising anti-cancer protein currently under clinical investigation with high specificity for malignant cells [11-13]. Despite initial promise, TRAIL represents a classic example of a molecule that has lacked clinical applicability due to short half-life and inefficient delivery, particularly to brain tumors where obstacles such as the blood–brain barrier further impair the ability of many therapeutics to reach the tumor mass [14-16].

We now report the engineering, screening, and application of new multifunctional molecules designed specifically to allow simultaneous determination of multiple events in stem cell-based therapy noninvasively in vivo. Utilizing the novel multifunctional molecule as a foundation, our results reveal that different stem cell lines secrete therapeutic proteins at significantly different levels, leading to varying anti-tumor effectiveness. Yet improved pharmacokinetics and anti-tumor efficacy can be achieved when the stem cell line with optimal delivery kinetics is utilized. We anticipate this approach will provide a platform for the creation of additional molecules for characterizing numerous therapeutic stem cell lines in a variety of different disease models, ultimately improving the effectiveness of cell-based therapies.

Materials and Methods

Cell Lines and Cell Culture

U251, Gli36-EGFRvIII, Gli36-EGFRvIII-FD human glioma cells, and 293T cells were grown as described previously [8]. Primary mouse neural stem cells (Stem Cell Technologies, Vancouver, BC, Canada, http://www.stemcell.com) were grown as previously described [17] in Neurocult NSC Basal media (Stem Cell Technologies) supplemented with proliferation supplements (Stem Cell Technologies) and EGF (20 ng/ml, R&D Systems, Minneapolis, MN, http://www.rndsystems.com). Human neural stem cells [9] and mouse mesenchymal stem cells were grown as previously described [3].

Enzyme-Linked Immunosorbent Assay, Viability, In Vitro Bioluminescence Imaging

1, 2, 4, and 14 days after transduction or co-culture plating, media from 293T, mNSC, hNSC, or mMSC was collected and assayed for levels of fusion variants by addition of coelenterazine as described [9]. Enzyme-linked immunosorbent assay (ELISA) was performed on the collected media as described previously [18]. To assess changes in viability of treated Gli36-EGFRvIII or U251 glioma cells, in vitro bioluminescence imaging was performed as described previously [9, 18, 19].

Bioluminescent Imaging of S-TRAIL Fusion Activity in vivo

To determine the pharmacokinetics of luciferase fusions released from tumors, U251 glioma cells transduced with luciferase fusions alone or with green fluorescent protein (GFP)-Fluc were implanted subcutaneously in SCID mice. Release of luciferase fusions was visualized by GpLuc or RLuc imaging as described previously [9]. Changes in tumor volume were monitored by Fluc imaging as described previously [9].

To investigate the pharmacokinetics of luciferase fusions delivered by mNSC, Gli36-EGFRvIII-FD cells were implanted subcutaneously in mice. 24 hours later, mice were injected with 2 mg of D-luciferin and FLuc imaging was performed to identify tumor location. 24 hours after FLuc imaging, mNSC expressing SRL_OL₂TR were implanted around established tumors. RLuc_O imaging was performed, as described above, 24 hours post-injection. In a subset of mice, animals were sacrificed 1 hour after injection of coelenterazine, and ex vivo imaging was performed on extracted lungs, liver, kidney, blood, tumor, and urine. The tissue was weighed and data were expressed relative to tissue weight. In a separate set of mice, FLuc imaging was performed 48 hours post-mNSC injection to determine the effects of SRL_OL₂TR on growth of established tumors.

Intracranial mNSC Survival and Tumor Progression

To assess the survival of transduced mNSC in the brain, 1 × 10⁶ mNSC transduced with LV-GFP-FLuc were mixed with Gli36-EGFRvIII (0.1 × 10⁶) and stereotactically implanted (from bregma, ML: 2.5 mm, SI: 2 mm) (n = 4 in each case) in mice as described previously [9]. FLuc imaging was performed 2, 6, 9, and 12 days post-injection by giving mice intraperitoneal injection of 2 mg of D-luciferin and collecting photon emission over 5 minutes with a cooled charge-coupled device camera. Images were processed as described previously [9].

To determine the effects of SRL_OL₂TR on intracranial gliomas, Gli36-EGFRvIII-FD and mNSC expressing SRL_OL₂TR were harvested at 80% confluency and a mix of glioma cells (5 × 10⁵) and transduced mNSC (1 × 10⁶) (n = 4 in each case) was implanted stereotactically (from bregma, ML: 2.5 mm, SI: 2 mm) (n = 4 in each case). Mice were injected intraperitoneally with 4.5 mg/mouse of D-luciferin on days 1, 3, 6, 9, 13, and 21 and imaged as described above. To monitor SRL_OL₂TR from mNSC, the same mice were imaged for RLuc_O activity on days 2, 6, 9, and 12 by injecting 100 μg of coelenterazine intravenously, and 5 minutes later photon emission was determined over 7 minutes. All images were processed as described previously [9].

Supplemental Online Data

Supplemental online data include four additional figures and expanded descriptions of vector generation, cell culture, in vitro assays, and in vitro and in vivo bioluminescence imaging.

Results

Engineering and Screening of Optimized Therapeutic and Diagnostic Luciferase-S-TRAIL Fusions

To create novel molecules that can both be easily visualized for serial monitoring of cell-based pharmacokinetics and retain potent anti-tumor function, we generated numerous lentiviral vectors (LV) encoding fusions between S-TRAIL, which we and others have shown to induce apoptosis specifically in tumor cells [4, 8, 20-22], and various luciferases (supplemental online Fig. 1A; Fig. 1; Table 1). To select the molecule with optimized diagnostic and therapeutic activity, 293T were transduced with LV encoding each construct at multiplicity of infection (MOI) 1 (supplemental online Fig. 1B), and conditioned media from transfected cells was screened for light emission, S-TRAIL concentration, and anti-tumor efficacy. To first select the appropriate molecular composition and orientation, direct fusions of firefly luciferase (FLuc), Renilla luciferase (RLuc), or Gaussia princeps luciferase (GpLuc) to the C-terminus of S-TRAIL were screened (Fig. 1, constructs 1-3). Of these fusions, only TRAIL-GpLuc (TRGp, construct 3) showed light emission by bioluminescence imaging on media from transduced cells; however, no S-TRAIL was detected in the conditioned media and no killing of Gli36-EGFRvIII human cancer cells was observed from TRGp-containing media (Fig. 1, construct 3; Table 1). In contrast, when GpLuc was fused to the N-terminus of S-TRAIL (GpTR, construct 4), detectable levels of light emission, S-TRAIL concentration, and reduction in tumor cell viability were observed (Fig. 1, Table 1).

Figure 1.

Engineering and screening of multiple S-TRAIL and luciferase fusions in vitro. (A): Schematic representations of lentiviral transfer vectors bearing IRES-GFP cassettes and encoding various fusions between secreted variant of the pro-apoptotic protein tumor necrosis factor-related apoptosis-inducing ligand (S-TRAIL) and different luciferase proteins. Direct fusion variants: (1) TRAIL-Rluc, (2) TRAIL-Fluc, (3) TRAIL-GpLuc, (4) GpLuc-TRAIL. Variants to test intramolecular spacing: (5) GpLuc-linker 1-TRAIL, (6) GpLuc-linker 2-TRAIL. Variants to test modification of secretion sequence: (7) SGpLuc-Linker 2-TRAIL, (8) SRluc_O-linker 2-TRAIL. To screen the various fusion molecules, 293T cells were transduced with lentiviral vectors encoding the designated fusion variant. Bioluminescence imaging and enzyme-linked immunosorbent assay were performed on conditioned medium from the transduced cells to determine diagnostic luciferase activity or concentration of S-TRAIL, respectively. Therapeutic activity of each variant was determined by luciferase-based assay on human Gli36-EGFRvIII cells 24 hours after incubation with equal volumes of conditioned media from lentiviral transduced 293T cells. Abbreviations: GpL₁TR, GpLuc-linker 1-TRAIL; GpL₂TR, GpLuc-linker 2-TRAIL; GpTR, GpLuc-TRAIL; SGpL₂TR, SGpLuc-Linker 2-TRAIL; SRL_OL₂TR, SRluc_O-linker 2-TRAIL; TRFL, TRAIL-Fluc; TRGp, TRAIL-GpLuc; TRRL, TRAIL-Rluc.

Table 1.

Composition and acitivity of luciferase and S-TRAIL fusion proteins: − (no activity); ++++ (highest activity)

Construct Name	Fusion	Diagnostic	Therapeutic
		Extracellular In Vitro Activity	In Vivo Activity	In Vitro Activity	In Vivo Activity
TRRL	S-TRAIL-RLuc	−	N/A	−	N/A
TRFL	S-TRAIL-FLuc	−	N/A	−	N/A
TRGp	S-TRAIL-GpLuc	++	N/A	−	N/A
GpTR	GpLuc-S-TRAIL	+	N/A	+	N/A
GpL1TR	GpLuc-L1-S-TRAIL	++	N/A	+	N/A
GpL2TR	GpLuc-L2-S-TRAIL	+++	N/A	+++	N/A
SGpL2TR	S-GpLuc-L2-S-TRAIL	++++	+	++++	++++
SRLOL2TR	S-RLucO-L2-S-TRAIL	+++	++++	+++	++++

As GpTR showed maximal diagnostic and therapeutic activity, we next investigated the effects of intramolecular spacing by engineering GpLuc and S-TRAIL fusions separated by two different linkers. As shown in Figure 1, increased intramolecular spacing increased the function of both GpLuc and S-TRAIL as the fusion variant containing linker-2 (GpL₂TR, construct 6) showed markedly increased diagnostic and therapeutic activity compared to the GpTR direct fusion.

Lastly, to maximize delivery by optimizing extracellular secretion, we engineered two additional fusion variants in which the endogenous secretion sequence was replaced by the signal sequence from Flt3 ligand, a sequence known to induce highly efficient protein secretion [21] (Fig. 1). The first variant was based on GpL₂TR (SGpL₂TR, construct 7). In the second (SRLuc_OL₂TR, construct 8), GpLuc was replaced by an alternative form of Renilla luciferase (Rluc_O) optimized for extracellular light production and suggested to have increased in vivo luciferase activity compared to GpLuc [23]. Our results showed that the modified secretion sequence in SGpL₂TR led to over a 1.5-fold increase in photon emission compared to that in GpL₂TR (Fig. 1, Table 1). Similarly, greater S-TRAIL concentration and tumor cell killing were observed by SGpL₂TR. Although SRLuc_OL₂TR demonstrated significant diagnostic and therapeutic activity, in vitro levels of bioluminescence signal, S-TRAIL concentration, and tumor cell killing were less than those observed for SGpL₂TR yet greater than or equal to those for GpL₂TR (Fig. 1, Table 1).

Characterizing Optimized Luciferase-S-TRAIL Fusions In Vivo

As the results above suggest, SGpL₂TR and SRLuc_OL₂TR have the greatest diagnostic and therapeutic activities in vitro, we next investigated their activity in vivo. To first confirm proper expression of the proteins, we performed Western blot analysis on lysates from transduced cells using an anti-TRAIL antibody. As shown in Figure 2A, both SGpL₂TR and SRLuc_OL₂TR were detected at high levels and at the anticipated size. For in vivo characterization of SGpL₂TR and SRLuc_OL₂TR, we utilized U251 cancer cells that have been previously shown to have a slow rate of TRAIL-induced cell death [18, 20], making this line ideal for longitudinal assessment of therapeutic protein release and anti-tumor efficacy. To track changes in tumor volume, U251 cells were first transduced with LV vectors encoding GFP-FLuc, followed by transduction with SGpL₂TR, SRL_OL₂TR, or control virus (Fig. 2B; supplemental online Fig. 1). Following subcutaneous implantation, GpLuc or RLuc_O imaging was performed to track fusion proteins and Fluc imaging was performed to monitor changes in tumor volume. As shown in Figure 2C–2D, SGpL₂TR or SRLuc_OL₂TR imaging on days 1, 3, and 15 demonstrated stable secretion of the luciferase-S-TRAIL fusion proteins from the infected cells when expressed relative to FLuc photon emission (Fig. 2C). This stable delivery coincided with a gradual yet significant reduction in tumor volume determined by FLuc imaging on days 2, 7, and 15 (Fig. 2D). Interestingly, in contrast to the results observed in vitro, tumors transduced with SRL_OL₂TR showed greater photon emission than SGpL₂TR-transduced tumor in vivo. The enhanced photon emission by RLuc_O-containing fusion was not due to differences in TRAIL-induced reduction in tumor volume, as further side-by-side in vivo comparison using nontherapeutic forms of SGpLuc and SRLuc_O confirmed greater light emission from RLuc_O tumors (Fig. 2E, supplemental online Fig. 2A–2D). As both SGpL₂TR and SRL_OL₂TR display similar anti-tumor activity yet SRL_OL₂TR shows enhanced light emission allowing improved in vivo detection (data summarized in Table 1), all further studies utilized SRL_OL₂TR. Together, these results demonstrate the engineering and optimization of SRL_OL₂TR to follow delivery and therapeutic efficacy of the multifunctional protein in vitro and in vivo. Additionally, they demonstrate the importance of molecular organization, intermolecular spacing, optimized secretion, and in vivo activity in developing multifunctional molecules.

Figure 2.

Screening S-TRAIL and luciferase fusion variants in vivo. (A): Western blot analysis of lysates from 293T cells transduced with LV demonstrating expression of SGpL₂TR and SRL_OL₂TR. (B): Representative green fluorescent protein (GFP) photomicrograph (large micrograph, 4×; inset, 10×) of human U251 glioma cells co-transduced with equal MOI of lentiviral vectors (LV) encoding SGpL₂TR, SRL_OL₂TR, or control virus and GFP-FLuc. (C–D): U251 glioma cells co-expressing GFP-FLuc and SGpL₂TR, SRL_OL₂TR, or control virus were implanted subcutaneously in mice and imaged on days 1, 3, and 15 to monitor secretion of TRAIL fusion proteins (GpLuc or RLuc_O intensities, (C)) and on days 2, 7, and 15 to follow changes in tumor volume (FLuc intensities, (D)). (E): Representative images and summary data of similar experiments as those described in (C and D) instead using nontherapeutic SGpLuc or SRLuc_O. Subcutaneous tumors were imaged on days 1, 5, and 10 for FLuc intensities to determine tumor volume or to monitor secretion of SGpLuc or SRLuc_O by coelenterazine injection. Representative day 10 images are shown. In all panels, *, p < .05 versus control. Abbreviations: SGpL₂TR, SGpLuc-Linker 2-TRAIL; SRL_OL₂TR, SRluc_O-linker 2-TRAIL; S-TRAIL, secreted variant of the pro-apoptotic protein tumor necrosis factor-related apoptosis-inducing ligand.

SRL_OL₂TR Reveals Stem Cell Lines Exhibit Different Secretion Kinetics That Effect Cancer Cell Killing

Next, we investigated if different stem cell lines exhibited different secretion levels and duration of delivery that influence anti-tumor efficacy. To this end, three different stem cell lines, (1) primary mouse neural stem cells (mNSC), (2) human neural stem cells (hNSC), and (3) primary mouse mesenchymal stem cells (mMSC), were transduced with increasing MOI of LV encoding SRL_OL₂TR (Fig. 3A–3B). For transduction of mNSC, cells were first seeded on coated tissue culture plates to establish a monolayer that ensured efficient LV-mediated transduction (Fig. 3A). As shown in Figure 3B, hNSC and mNSC were robustly transduced at low MOI, whereas mMSC required MOI of 8 to reach the same percentage of transduction as hNSC or mNSC. Serial bioluminescence imaging performed on media from equally transduced cells showed that, despite early robust secretion of SRL_OL₂TR by hNSC (supplemental online Fig. 3A), levels rapidly declined and were nearly absent by 48 hours (Fig. 3C) due to TRAIL-induced death of hNSC (data not shown). In contrast, mNSC and mMSC retained stable secretion of SRL_OL₂TR through 14 days, although mNSC secretion of SRL_OL₂TR was markedly greater than levels detected from mMSC (supplemental online Fig. 3A).

Figure 3.

Imaging of SRL_OL₂TR reveals differences in stem cell secretion and cancer cell killing. (A): Representative images of mNSC, hNSC, and mMSC transduced with LV encoding SRL_OL₂TR. (B): Summary data demonstrating differences in transduction efficiency between mNSC, hNSC, and mMSC 24 hours post-transduction with increasing MOI of LV-SRL_OL₂TR. Green fluorescent protein (GFP)-positive cells were counted and expressed as a ratio of total cell number for each stem cell type. (C): Photon emission from mNSC, hNSC, and mMSC transduced with LV-SRL_OL₂TR were assayed at days 0, 2, 7, and 14 post-transduction. (D and E): Representative images and summary graphs demonstrating the effects of different stem cell lines secreting SRL_OL₂TR co-cultured at increasing stem cell to tumor cell ratios with Gli36-EGFRvIII (D) or U251 human cancer cells (E). After 24 hours of co-culture, levels of SRL_OL₂TR secretion by the stem cells were visualized by RLuc_O bioluminescence imaging and tumor cell killing was visualized by Fluc bioluminescence imaging and quantified using a luminometer. (F): Western blot analysis of cell lysates from mNSC or Gli36-EGFRvIII tumor cells demonstrating the expression of DR4 in each cell line. (G): Immunocytochemical analysis of undifferentiated mNSC stained with an antibody against NSC marker Nestin (a), or following 10 days of differentiation using antibodies against glial fibrillary acidic protein (GFAP) (b) or Olig-2 (c). (H): Representative photomicrographs demonstrating the migration of transduced mNSC towards gliomas over time. GFP-expressing mNSC were implanted 1 mm lateral to established Gli36-EGFRvIIIFD intracranial gliomas. On days 2 (a), 5 (b), and 10 (c) post-mNSC, implantation mice were sacrificed, brains were removed and sectioned, and both mNSC and glioma volumes were visualized using fluorescence confocal microscopy. Panels a and b: 4x magnification; Panel c: 10x magnification. (I and J): Human Gli36-EGFRvIII glioma cells were incubated with conditioned media from mNSC transduced with control vector, SRL_OL₂TR, or purified S-TRAIL and caspase-3/7 activity (I), cleaved caspase-8 levels (J), and cleaved PARP levels (J) were determined by luciferase-based caspase 3/7 assay (I) and Western blot analysis (J). In all panels, *, p < .05 versus control. Abbreviations: hNSC, human neural stem cells; mMSC, primary mouse mesenchymal stem cells; mNSC, primary mouse neural stem cells; SRLuc_OL₂TR, SRluc_O-linker 2-TRAIL.

To determine if differences in secretion between the three stem cell lines translated to differences in anti-tumor cell efficacy, we next performed co-culture studies using different ratios of stem cells expressing SRL_OL₂TR and glioma lines with different sensitivities to TRAIL-mediated apoptosis (Gli36-EGFRvIII, highly TRAIL sensitive; U251, less TRAIL sensitive), both engineered to express mCherryFLuc. The imaging of hNSC secretion showed extremely low levels of SRL_OL₂TR in the media at the time point assayed, and this translated to minimal effects on either Gli36-EGFRvII (Fig. 3D) or U251 (Fig. 3E) tumor cell viability. Imaging of media from mNSC showed robust levels of SRL_OL₂TR that increased as the stem cell/tumor cell ratio increased, with a ratio of 0.04/1 leading to detectable amounts of SRL_OL₂TR. Importantly, the efficient secretion of SRL_OL₂TR lead to stem cell/tumor cell ratios as low as 0.1/1 and 0.2/1, significantly decreasing the viability of Gli36-EGFRvII (Fig. 3D) or U251 (Fig. 3E) cells, respectively, in a dose-dependent manner. Imaging of the third stem cell line, mMSC, showed markedly lower levels of SRL_OL₂TR compared to mNSC, suggesting significantly lower delivery of therapeutic proteins (Fig. 3D–3E). Similarly, Gli36-EGFRvIII cell viability was decreased but required 10-fold more therapeutic mMSC to induce killing compared to mNSCs (Fig. 3D), whereas U251 cells required a mMSC/tumor ratio of 3/1 before a significant difference in tumor cell viability was detected (Fig. 3E). Together, these results show that imaging of SRL_OL₂TR revealed marked differences in the delivery of therapeutics by hNSC, mNSC, and mMSC, which resulted in significant differences in the ability of each line to induce tumor cell killing in human cancer cell lines both highly sensitive and less sensitive to TRAIL. Additionally, mNSC secreted the highest levels of SRL_OL₂TR and induced the largest decrease in cancer cell viability at the lowest stem cell/cancer cell ratio.

On the basis of these results, mNSC secreting SRL_OL₂TR were selected as the stem cell line for further investigation. Western blot analysis confirmed that mNSC did not express TRAIL death receptors (Fig. 3F). Immunocytochemistry revealed that these cells robustly expressed the NSC marker Nestin (Fig. 3Ga) and possessed the capacity to generate cells of neural lineage (Fig. 3Gb–3Gc). In addition, when transduced stem cells were implanted 1 mM adjacent to established Gli36-EGFRvIII human gliomas expressing FLuc-DsRed2 (Gli36-EGFRvIII-FD; supplemental online Fig. 3B) in the frontal lobe of mice, two distinct populations of green stem cells and red gliomas were observed 2 days post-implantation (Fig. 3 Ha). However, stem cells were observed specifically migrating towards the glioma by day 5(Fig. 3 Hb) and showed accumulation in the tumor by day 10 (Fig. 3 Hc). Demonstrating the role of the extrinsic apoptotic cascade in stem cell-secreted SRL_OL₂TR-induced cancer cell death, Western blot analysis and luciferase-based assay on Gli36-EGFRvIII cells treated with SRL_OL₂TR conditioned media revealed significant activation of caspase 3/7 (Fig. 3I) and increases in cleavage of caspase-8 and Poly(ADP-ribose) polymerase (PARP) (Fig. 3J). These results show that transduced mNSC retain all the characteristics of NSC in vitro and also migrate specifically to gliomas in mice bearing intracranial tumors.

Stem Cell-Based Delivery Significantly Improves Pharmacokinetics and Efficacy of Therapies Determined by Noninvasive Imaging of SRL_OL₂TR

In clinics, numerous chemotherapies are administered to patients via i.v. infusion or direct injection, yet these methods often lead to significant levels of the drugs accumulating in normal organs, resulting in dose-limiting toxicities. In stem cell-based delivery, the cells are typically engrafted around the tumor to provide sustained levels of therapeutic protein for direct targeting of tumor cells. We next determined if SRL_OL₂TR could be used to visualize differences between the pharmacokinetics of delivery to tumors by mNSC and i.v. or intratumoral injection of purified protein in vivo in models designed to mimic the clinical settings. First, the bioluminescence signal from both Gli36-EGFRvIII-FD and mNSC engineered with SRL_OL₂TR were shown to correlate directly with cell number (supplemental online Fig. 4A–4B). Next, mice were implanted with mNSC expressing SRL_OL₂TR around established Gli36-EGFRvIII-FD tumors to mimic clinical engraftment of cell-based therapies. RLuc_O imaging performed 24 hours after implantation of mNSC revealed a robust bioluminescence signal that co-localized with the established tumor (Fig. 4A). FLuc bioluminescence imaging showed a significant reduction in tumor volume by 48 hours (Fig. 4B). Ex vivo analysis demonstrated RLuc_O signal was present in the excised tumor but absent from other organs and tissues (Fig. 4C). Alternatively, when media containing SRL_OL₂TR was delivered by i.v. injection to mimic the systemic delivery of chemotherapies, a signal was detectable that persisted through 40 minutes but was absent by 24 hours (Fig. 4D). Furthermore, the bioluminescence signal was detected in the liver, lung, kidney, and blood as well as the tumor (Fig. 4E) and did not have any effect on tumor volume (Fig. 4F). Lastly, to determine if SRL_OL₂TR could be used to visualize differences in delivery of purified protein administered in the same manner as stem cells, media containing SRL_OL₂TR was injected directly into established tumors. As shown in Figure 4G–4I, the direct injection of media resulted initially in a robust signal present at the site injection that gradually decreased to baseline near baseline levels by 40 minutes and was entirely absent at 24 hours (Fig. 4G). Similar to i.v. infusion, the direct injection of media to the tumor led to detectable accumulation of the SRL_OL₂TR in the liver, lung, and kidney in addition to the tumor (Fig. 4H) and also had minimal effect on tumor volume (Fig. 4I). These results suggest that combining the novel SRL_OL₂TR protein and optical imaging permits elucidation of differences in pharmacokinetics, tissues distribution, and therapeutic efficacy of anti-cancer proteins delivered to tumors by engineered stem cells or i.v. injection. As SRL_OL₂TR allows visualization of therapeutic levels in real-time, we show that a single administration of engineered stem cells provides continuous sustained and localized delivery of therapeutics that attenuates tumor growth, whereas a single i.v. infusion or direct administration of media containing SRL_OL₂TR results in widespread off-target binding and significantly shortened delivery window that correlates with minimal anti-tumor effects.

Figure 4.

Delivery by engineered stem cells improves SRL_OL₂TR pharmacokinetics in vivo. (A): Representative images and summary graphs showing SRL_OL₂TR levels when delivered to tumors by engineered stem cells. Gli36-EGFRvIII-FD glioma cells were implanted subcutaneously in mice, and 24 hours later FLuc imaging was performed to demonstrate the localization of the tumor. 24 hours post-imaging, mNSC secreting SRL_OL₂TR were injected around one of the established tumors, and SRL_OL₂TR imaging was performed to visualize the secretion of SRL_OL₂TR. (B): Summary graph showing the effects on tumor volume of control mNSC or mNSC secreting SRL_OL₂TR 48 hours after implantation around established Gli36-EGFRvIII-FD tumors assessed by FLuc imaging. (C): Ex vivo analysis of biodistribution of mNSC-delivered fusion proteins assessed by RLuc_O imaging of organs removed 1-hour post injection of coelenterazine. (D–I): In vivo bioluminescence imaging of conditioned medium from LV-SRL_OL₂TR transduced cells injected into mice bearing established Gli36-EGFRvIII-FD subcutaneous tumors by i.v. infusion (D) or direct intratumoral administration (G) and analyzed at different time points after coelenterazine injection. Ex vivo bioluminescence imaging of organs and tumor tissue from mice 1-hour post-injection of media administered by i.v. infusion (E) or direct injection (H) followed by coelenterazine. Forty-eight hours after media injection, Fluc imaging was performed to determine changes in tumor volumes (F, I). In all panels, *, p < .05 versus control. Abbreviations: SRL_OL₂TR, SRluc_O-linker 2-TRAIL.

Stem Cells Lead to Sustained Delivery of SRL_OL₂TR for Treatment of Highly Malignant Intracranial Glioblastoma

Lastly, we tested the above stem cell-based approach in an intracranial glioma model, a disease where effective delivery of therapeutic agents is further limited by the blood–brain barrier. To first investigate the survival of mNSC in the context of glioma, mNSC expressing GFP-FLuc (supplemental online Fig. 4C) were mixed with Gli36-EGFRvIII human glioma cells and implanted intracranially in mice. FLuc bioluminescence imaging revealed the presence of transduced mNSC in the brain, and the levels remained constant through 15 days (Fig. 5A). After confirming intracranial survival of transduced mNSC, therapeutic mNSC engineered to express SRL_OL₂TR were implanted intracranially in severe combined immunodeficiency (SCID) mice together with Gli36-EGFRvIII-FD. RLuc_O imaging on days 2, 6, 9, and 12 showed robust and stable delivery of SRL_OL₂TR by mNSC (Fig. 5B). Serial FLuc imaging revealed that mNSC delivery of SRL_OL₂TR led to marked attenuation in glioma progression, with significant decrease in FLuc signal in SRL_OL₂TR-treated mice by day 6 (Fig. 5C). Post-mortem immunohistochemical analysis performed 4 days post-implantation confirmed the presence of GFP-expressing mNSC (Fig. 5D) and demonstrated expression of the NSC marker Nestin (Fig. 5Da, 5De, 5Di). Furthermore, mNSC did not stain positive for the astrocyte marker glial fibrillary acidic protein (GFAP) (Fig. 5Db, 5Df, 5Dj), the neuronal marker Tuj-1 (Fig. 5Dc, 5Dg, 5Dk), or the proliferation marker Ki67 (Fig. 5D-d,h,l), which showed strong staining of the highly proliferating glioma cells. By simultaneously monitoring therapeutic delivery by mNSC in the brain and intracranial glioma volumes, these results show that therapeutic stem cells secreting SRL_OL₂TR are effective anti-glioma therapies.

Figure 5.

Stem cells efficiently deliver SRL_OL₂TR to eradicate intracranial glioblastoma. (A): Representative FLuc bioluminescent images and summary data of mice implanted intracranially with mNSC transduced with LV-GFP-FLuc, mixed with Gli36-EGFRvIII, and serially imaged for 15 days. (B–D): mNSC were transduced with control vector or SRL_OL₂TR, and implanted with Gli36 EGFRvIII-FD intracranially in mice. On days 2, 6, 9, and 12 post-implantation, SRL_OL₂TR mice were injected with coelenterazine and RLuc_O imaging was performed to visualize SRL_OL₂TR secretion (B). Mice were injected with D-Luciferin and FLuc imaging was performed to visualize changes in glioma on days 1, 3, 6, 9, 13, and 21 post-implantation. (C): Representative images and summary data are shown. C = Control; T = SRL_OL₂TR. (D): Immunohisto-chemistry was performed on sections from brains containing GFP-FLuc-expressing mNSC 4 days post-implantation. Representative merged images are shown of brain sections containing mNSC (GFP) and stained with antibodies (Red) against nestin (a, e, i), glial fibrillary acidic protein (GFAP) (b, f, j), Tuj-1 (c, g, k), or Ki67 (d, h, l). Nb = normal brain; T = tumor. In all panels, *, p < .05 versus control. Abbreviations: SRL_OL₂TR, SRluc_O-linker 2-TRAIL.

Discussion

Rapid and simple means to assess pharmacokinetics and anti-tumor efficacy of therapeutic proteins delivered by engineered stem cells have the potential to markedly accelerate the development of stem cell-based therapies by providing vital answers to key questions. In this study, we have developed novel multifunctional molecules that have both diagnostic (in vivo tracking) and therapeutic (anti-tumor) properties and demonstrate their application in characterizing therapeutic delivery by engineered stem cells.

We report the creation of novel fusion variants containing optical reporters (luciferase) and the cytotoxic agent TRAIL. To develop molecules with the greatest optical reporter activity and highest tumor cell toxicity, it is critical to select both the appropriate luciferase and orientation for the fusion to TRAIL (N- or C-terminus). After screening fusion proteins containing several different luciferase proteins, our results showed fusion proteins containing GpLuc produced the greatest light emission in vitro, while SRLuc_O containing fusions performed better in vivo. Furthermore, we found fusion of luciferase proteins to the N-terminus of TRAIL permitted retention of both the imaging properties of the luciferase and anti-tumor properties of TRAIL, whereas C-terminal fusions inactivated TRAIL. The inactivation of S-TRAIL is most likely due to the fact that the C-terminus of S-TRAIL contains the cell-binding domain of TRAIL [24], and therefore, the fusion of proteins to the C-terminus of TRAIL either prevents proper folding of the protein or interferes with interaction of S-TRAIL with its receptors, DR4 and DR5, and is in agreement with the structure of other TRAIL fusion proteins [25, 26]. Furthermore, previous reports have emphasized the importance of protein linkers in order to achieve optimal activity of luciferase fusion proteins [27, 28]. In agreement with these reports, we observed greater extracellular BLI signal, TRAIL levels, and cell killing by the inclusion of longer intracellular linkers. We speculate that the increased intramolecular spacing between S-TRAIL and luciferase by inclusion of linker-2 better preserves the functionality of both luciferase and S-TRAIL, thus leading to the observed increases in photon emission, S-TRAIL concentration, and cell killing. Taken together, SRL_OL₂TR combines the potent anti-tumor properties of S-TRAIL with the simple noninvasive assessment of therapeutic delivery afforded by luciferase imaging.

One of the primary challenges to achieving effective anti-tumor therapy is highly efficient delivery of the anti-tumor agent specifically to the tumor, while minimizing toxicity to nonmalignant tissue. Although simple to administer, systemic administration of therapies often leads to accumulation of the toxic compounds at high levels in the liver and kidneys, resulting in dose-limiting renal- and hepatotoxicity [14, 29]. TRAIL has been shown to have minimal cytotoxic effects on normal tissue; however, its short half-life and accumulation after systemic injection have been limitations to its potential use in clinics [30]. Because of their potential to migrate to sites of disease and integrate into the cytoarchitecture of the brain, stem cells (NSC, MSC) have received much interest for the treatment of numerous neurologic disorders [1, 31]. Previous studies from our lab and others demonstrated that NSC and human MSC migrate extensively throughout the murine brain and exhibit an inherent capacity to home to established gliomas [3, 8, 9]. These findings suggest stem cells would make ideal vehicles for effective on-site delivery of therapies to tumors, and recent studies have focused on “arming” stem cells with therapeutic proteins. Results from our lab showed that stem cells armed with S-TRAIL inhibited progression of gliomas in a xenogenic transplant model [3, 8]; however, assessing the pharmacokinetics of the molecules released by therapeutic NSC has been difficult. In this study, the combination of SRL_OL₂TR and real-time imaging demonstrated the advantages of stem cell-based delivery. Noninvasive monitoring of SRL_OL₂TR pharmacokinetics revealed a markedly increased delivery time and reduced nonspecific biodistribution that culminated in effective reduction in tumor burden. In contrast, i.v. or intratumoral injection of SRL_OL₂TR resulted in rapid clearance, widespread biodistribution, and minimal effects on the tumor. These results begin to provide insights into advantages of cell-based therapy by permitting the first real-time comparison of pharmacokinetics when therapies are delivered by stem cells or systemic administration.

The ability to serially monitor the level of therapeutic protein delivered by NSC is critical to effective cell-based therapy. Longitudinal imaging of therapies permits confirmation of the initial levels delivered, permits confirmation of whether there is a need to increase dose by injection of additional therapeutic NSC should inhibition of tumor growth not be observed, or indicates the need for re-administration if the dose begins to decline. Our studies on intracranial glioblastoma xenograft using the combination of mNSC and SRL_OL₂TR revealed robust and sustained delivery of TRAIL fusion as early as 2 days post-implantation and showed that sustained SRL_OL₂TR persisted through day 12. Importantly, the continuous delivery of SRL_OL₂TR by stem cells markedly decreased glioma burden as early as day 6. These results demonstrate the effectiveness of stem cell-mediated delivery of SRL_OL₂TR as an anti-glioma therapy, as significant tumor regression was achieved with a single administration of therapeutically engineered stem cells. In the event that SRL_OL₂TR levels decrease at time points beyond the scope of this study, the diagnostic capacity of SRL_OL₂TR would reveal these changes and stem cells could be re-administered to ensure continued suppression of tumor growth.

Conclusion

In conclusion, we have engineered novel fusion proteins with both diagnostic and therapeutic function, and utilized the novel molecule SRL_OL₂TR as a means to determine vital aspects of stem cell-based therapies. These included how differences in delivery efficiency between different cell lines affected their therapeutic application, how improved pharmacokinetics mediated by stem cell delivery influenced anti-tumor efficacy of a therapy, and how selection of the optimal therapeutic stem cell line could effectively attenuate highly malignant tumors in vivo. Using this study as a model, development of additional multifunctional molecules would improve preclinical characterization of new cell-based therapies for numerous diseases, with the ultimate goal of better patient outcomes through treatment with highly effective therapeutic stem cells.