This article is dedicated to the loving memory of Monica Bartow Aboody.
Author contributions: D.Z.: conception and design, collection and assembly of data, data analysis and interpretation, and manuscript writing; J.N.: conception and design, data analysis and interpretation, and manuscript writing; A.J.A.: data analysis and interpretation and manuscript editing; E.G. and M.Z.M.: technical support, contributed to experimental design; M.G., M.D.P., and M.G.: collection and assembly of data and manuscript editing; C.A.G. and S.U.K.: provision of study material; K.S.A.: conception and design, data analysis, financial support, and editing and final approval of manuscript.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLSEXPRESS Nov 14, 2011.
Metastasis to multiple organs is the primary cause of mortality in breast cancer patients. The poor prognosis for patients with metastatic breast cancer and toxic side effects of currently available treatments necessitate the development of effective tumor-selective therapies. Neural stem cells (NSCs) possess inherent tumor tropic properties that enable them to overcome many obstacles of drug delivery that limit effective chemotherapy strategies for breast cancer. We report that increased NSC tropism to breast tumor cell lines is strongly correlated with the invasiveness of cancer cells. Interleukin 6 (IL-6) was identified as a major cytokine mediating NSC tropism to invasive breast cancer cells. We show for the first time in a preclinical mouse model of metastatic human breast cancer that NSCs preferentially target tumor metastases in multiple organs, including liver, lung, lymph nodes, and femur, versus the primary intramammary fat pad tumor. For proof-of-concept of stem cell-mediated breast cancer therapy, NSCs were genetically modified to secrete rabbit carboxylesterase (rCE), an enzyme that activates the CPT-11 prodrug to SN-38, a potent topoisomerase I inhibitor, to effect tumor-localized chemotherapy. In vitro data demonstrate that exposure of breast cancer cells to conditioned media from rCE-secreting NSCs (NSC.rCE) increased their sensitivity to CPT-11 by 200-fold. In vivo, treatment of tumor-bearing mice with NSC.rCE cells in combination with CPT-11 resulted in reduction of metastatic tumor burden in lung and lymph nodes. These data suggest that NSC-mediated enzyme/prodrug therapy may be more effective and less toxic than currently available chemotherapy strategies for breast cancer metastases. STEM CELLS 2012; 30:314–325.
Breast cancer is a major cause of death of women in the Western world, who have a 10% lifetime risk of developing the disease . In the United States, the mortality rate is second only to lung cancer among female cancer patients . Although many breast tumors can be detected at early stages and removed with surgery and/or treated with radiation and chemotherapy, current therapies cannot prevent or eradicate breast cancer after metastases have spread to distant organs. In addition, breast cancer patients with bone or brain metastases have especially poor prognosis because these sites are particularly difficult to access with chemotherapy [3, 4]. The poor prognosis for patients with breast tumor metastases and the toxic side effects of currently available treatments necessitate the development of effective tumor-selective therapies.
Ideally, breast cancer therapeutics would be tumor selective and concentrated specifically at sites of metastases, thereby minimizing the side effects to healthy tissues. Aboody et al. and Benedetti et al. [5, 6] have demonstrated in animal models that neural stem cells (NSCs) have an inherent ability to seek out and infiltrate invasive glioma, even when injected into the bloodstream. Other studies have shown the safety, feasibility, and therapeutic efficacy of using NSCs to track invasive tumor cells and distant tumor foci and to deliver therapeutic gene products to tumor cells, including medulloblastoma, melanoma brain metastases, and disseminated neuroblastoma [7–10]. These NSC-based strategies may provide effective antitumor response, overcoming obstacles faced by current gene therapy strategies, for example, insufficient drug penetration or dissemination of therapeutic virus to tumor.
In this study, we used the v-myc-immortalized clonal, human HB1.F3.CD NSC line, which is genetically and functionally stable, nontumorigenic, and minimally immunogenic [11, 12]. We now show in a preclinical mouse model of metastatic human breast cancer, that NSCs preferentially target tumor metastases in multiple organs, including liver, lung, lymph nodes, and femur, versus the primary intramammary fat pad tumor. Data suggest that interleukin 6 (IL-6) is a major chemokine involved in this NSC-tumor tropism. We also investigated a novel NSC-mediated enzyme/prodrug strategy to target metastatic breast tumors. NSCs were engineered to secrete rabbit carboxylesterase (rCE), an enzyme that activates the prodrug CPT-11 (irinotecan) to the active drug SN-38, a 1,000-fold more potent topoisomerase I inhibitor that preferentially kills dividing cells, thereby concentrating chemotherapy specifically to tumor sites. CPT-11 has recently been approved for therapy of colon adenocarcinoma  and is currently in phase III trials for non-small cell lung cancer . Preclinical studies and early clinical trials suggest that CPT-11 has antitumor activity against breast cancer but is not sufficiently effective to result in cure . In vivo, human esterases convert only approximately 0.5%–5% of CPT-11 to SN-38 . rCE converts CPT-11 to SN-38 100–1,000 fold more efficiently than human enzyme and has been engineered to be secreted by cells [17, 18]. Therefore, we modified our NSCs to secrete rCE in combination with CPT-11 as a therapeutic strategy for metastatic breast cancer. We anticipate that the greatest benefit of such therapy will be to patients with advanced metastatic disease.
MATERIALS AND METHODS
Neural Stem Cell Line
The human HB1.F3.C1 NSC line is v-myc-immortalized clonal, stable nontumorigenic, and minimally immunogenic [11, 12]. This NSC line has been further genetically modified to express the Escherichia coli cytosine deaminase (CD) gene, which converts the prodrug 5-fluorocytosine (5-FC) to the active chemotherapeutic drug, 5-fluorouracil (5-FU) . HB1.F3.CD NSCs were grown in a humidified atmosphere (37°C, 6% CO2) in Dulbecco's modified Eagle's medium (DMEM), supplemented with 2 mM L-glutamine and 10% fetal bovine serum (FBS). Use of the HB1.F3.CD NSCs was approved by City of Hope Institutional Review Board and Stem Cell Research Oversight Committee.
Breast Cancer Cell Lines
Highly invasive (MDA-MB-231, MDA-MB-468, and Hs578T) and less-invasive (MCF-7, T47D, ZR75, and SKBR3) human breast cancer cell lines were obtained from ATCC (Manassas, VA) and were maintained in their respective ATCC-suggested media at 37°C, 6% CO2. MDA-MB-231-luc-D3H2LN (from here on MDA-MB-231-luc) and MCF-7-luc-F5 (from here on MCF-7-luc) were purchased from Caliper (Mountain View, CA) and cultured in Eagle's Minimum Essential Medium (ATCC, Manassas, VA), supplemented with 10% FBS at 37°C, 6% CO2.
Breast Cancer Cell-Conditioned Media
To obtain conditioned media, breast cancer cells were grown in 100-mm Petri dishes to approximately 80% confluence. Cells were rinsed three times with serum-free medium and cultured for an additional 24 hours in serum-free DMEM. The conditioned media were cleared of cells by centrifugation and stored frozen at −80°C until used.
Matrigel Invasion Assay of Breast Cancer Cell Lines
In vitro invasion assays were performed in 24-well Transwell units with polycarbonate filters (8-μm pore size) coated on the upper surface with Matrigel (3 mg/ml, BD Biosciences, San Jose, CA). Cells (8 × 104) in serum-free medium (0.4 ml per well) were placed in the upper chamber of the Transwell unit and allowed to invade for 24 hours. The lower chamber of the Transwell unit was filled with 10% FBS. Invaded cells on the bottom surface of the membrane were quantified using CyQuant GR fluorescent dye (Millipore, Temecula, CA) and a fluorescence microplate reader (Spectramax Plus-384 and GeminiXS, Molecular Devices, Sunnyvale, CA). Assays were performed in triplicate. The percentage of invasive cells = (total number of cells on bottom surface/total number of cells on both upper and bottom surface) × 100.
Modified Boyden Chamber Cell Migration Assay
All in vitro cell migration experiments were performed using modified 96-well Boyden chambers (Neuro Probe, Gaithersburg, MD). Lower wells were filled with breast cancer cell-conditioned media or serum-free media as a control. HB1.F3.CD cells (1.5 × 105) were added to the top chambers. After 4 hours of migration at 37°C, the number of migrated NSCs was quantified by CyQuant GR fluorescence assay. A standard curve for fluorescence intensity was generated using defined numbers of NSCs. The percentage of migrated cells = (total number of cells on bottom surface/total number of cells on both upper and bottom surface) × 100. To investigate the function of IL-6 and IL-8 in the migration of NSCs to breast cancer cells, IL-6 (206-IL/CF, R&D, Minneapolis, MN), IL-8 (14-8089-61, eBioscience, San Diego, CA), or the function-blocking antibodies against IL-6 (ab11449, Abcam, Cambridge, MA) and IL-8 (MAB 208, R&D, Minneapolis, MN) were added to breast cancer cell-conditioned media and incubated for 4 hours, followed by NSC migration assay. IL-6 receptor (IL-6R) neutralizing antibody (ab34351, Abcam, Cambridge, MA) was incubated with NSCs for 12 hours, followed by cell migration assay.
Total RNA was extracted using the RNeasy Mini Kit (Qiagen, Valencia, CA). Random primers, ThermoScript reverse transcriptase (Invitrogen, Carlsbad, CA), and 1 μg of total RNA were used for cDNA synthesis. Relative quantification of IL-6, IL-8, or IL-6R mRNA expression was performed with the SYBR Green PCR Master Mix using an ABI PRISM 7700 thermal cycler (Applied Biosystems, Carlsbad, CA). The master mixture (25 μl) contained 0.5 μl of cDNA and 1 μl of a 5 μM solution of each primer. Reaction conditions were as follows: 3 minutes at 95°C, followed by 40 cycles of 10 seconds each at 95°C, and 15 seconds at 60°C. Primer sequences were as follows: human IL-6, forward, 5′-GCCCAGCTATGAACTCCTTCT-3′; reverse, 5′-GAAGGCAG CAGGCAACAC-3′; human IL-8, forward, 5′-GGGTGGAAAGG TTTGGAGTATGT-3′; reverse, 5′-CTTGGCAGCCTTCCTGATT TC-3′; human IL-6R, forward, 5′-GTACCACTGCCCACATTCCT-3′; reverse, 5′-CAGCTTCCACGTCTTCTTGA-3′. Samples were normalized to human β-actin as an endogenous reference. Relative quantification of target gene expression was calculated by the comparative threshold cycle (Ct) method.
Cytokine Antibody Array and Enzyme-Linked Immunosorbent Assay (ELISA)
The Cytokine Antibody Array V (Supporting Information Table 1) was used to measure cytokine protein levels in conditioned media from breast cancer cells according to the manufacturer's instructions (RayBiotech, Norcross, GA). Cytokine array membranes were blocked by incubation (room temperature, 30 minutes) with blocking buffer and then incubated (4°C, overnight) with conditioned media (1 ml) derived from highly invasive (MDA-MB-231-luc, MDA-MB-231, MDA-MB-468, and Hs578T) and less-invasive (MCF-7-luc, MCF-7, T47D, ZR75, and SKBR3) breast cancer cell lines. One Cytokine Antibody Array V was used for each cell line. Membranes were washed three times with wash buffer I and two times with wash buffer II at room temperature for 5 minutes per wash and then incubated (room temperature, 2 hours) with biotin-conjugated antibodies. Finally, the membranes were washed, incubated with horse radish peroxidase–conjugated streptavidin at room temperature for 2 hours and then with detection buffer for 2 minutes, and exposed to Kodak X-ray film for 30 seconds (Sigma-Aldrich, St. Louis, MO). The amount of human IL-6 and IL-8 protein in conditioned media derived from highly invasive and less-invasive breast cancer cell lines was measured using RayBio human IL-6 and IL-8 ELISA kits according to the manufacturer's protocol (RayBiotech).
siRNA Knockdown of Human IL-6R
IL-6R siRNA was purchased from Santa Cruz Biotechnology (sc-35663, Santa Cruz, CA). HPRT-S1 siRNA, which served as positive control, and scrambled siRNA, which served as negative control, were purchased from IDT, Inc. (TriFECTa Dicer-substrate RNAi Kit, San Diego, CA). HB1.F3.CD NSCs were transfected with IL-6R siRNA using X-tremeGENE siRNA transfection reagent (Roche, South San Francisco, CA) according to the manufacturer's instructions. Gene knockdown was confirmed by RT-PCR 24 hours after siRNA transfections, and flow cytometry assay, as well as Western blotting 48 hours after siRNA transfections.
Flow Cytometry Assay
NSCs with or without siRNA knockdown were resuspended in ice-cold staining buffer (PBS supplemented with 2% FBS) and incubated with Fc-blocking solution for 5 minutes at 4°C. Subsequently, cells were stained with mouse monoclonal antibodies against human IL-6R (1:200 dilution, ab27404, Abcam) at 4°C for 60 minutes. Cells were then reacted with fluorescein isothiocyanate (FITC)–conjugated anti-mouse secondary antibodies at a dilution of 1:1,000 (Promega, Madison, WI) at 4°C for 30 minutes, followed by two washes in staining buffer. Samples were analyzed on a Guava EasyCyte flow cytometer (Guava Biotechnologies, Hayward, CA), and the data were processed with Guava software.
Western Blot Analysis
HB1.F3.CD NSCs with or without siRNA knockdown were lysed for 20 minutes on ice in RIPA lysis buffer (Sigma-Aldrich, St. Louis, MO). Protein concentrations were determined by Bradford assay (Bio-Rad, Hercules, CA). Protein (45 μg) was subjected to immunoblotting using mouse monoclonal antibodies against IL-6R (1:500 dilution, ab27404, Abcam). Horseradish peroxidase-conjugated anti-mouse secondary antibodies (Promega, Madison, WI) were used at a dilution of 1:2,000, and membranes were visualized with Western Lightening Enhanced Chemiluminescence System (NEL100001EA, PerkinElmer Life Sciences, San Jose, CA).
Establishment of Breast Tumor Metastatic Model in Mice
Mice were housed in a vivarium accredited by the American Association for Accreditation of Laboratory Animal Care. All animal protocols were approved by the City of Hope Institutional Animal Care and Use Committee. Human MDA-MB-231-luc breast carcinoma cells that were genetically modified to express the firefly luciferase bioluminescent reporter gene were purchased from Caliper . The firefly luciferase reporter gene allows for repeated noninvasive quantitative in vivo monitoring of tumor burden by the Xenogen IVIS-100 imaging system (Caliper, Hopkinton, MA) as animals are followed for long-term survival. Female athymic nude/nude mice (8–10 weeks old; Charles River, Wilmington, MA) were injected bilaterally with 2 × 106 MDA-MB-231-luc cells in Matrigel (50 μl) into the fourth mammary fat pads. Xenogen imaging was performed every week to monitor tumor burden. Primary tumors were shielded from the Xenogen camera detector using black paper to permit detection of low-intensity signals from secondary metastases. Once metastases were detected 4–5 weeks after tumor implantation, chloromethylbenzamido-1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate (CM-DiI)-labeled HB1.F3.CD NSCs were injected into the tail vein. After 4 days of injection, the mice were euthanized by CO2 inhalation and transcardially perfused with ice-cold PBS followed by 4% paraformaldehyde (PFA). The primary tumors and other tissues, including brain, bone, lung, lymph node, and liver (potential target tissues for breast tumor metastasis) were harvested. Mice destined for ex vivo Xenogen imaging to identify signals from tumor metastases were euthanized in the same manner without transcardial perfusion. All primary tumors and tissues were dissected into two pieces, one for extraction of DNA and the other for fixation in 4% PFA for immunohistochemistry.
Standard protocols were performed for immunohistochemical staining of cryosectioned tissues (10-μm-thick sections). Anti-luc rabbit polyclonal antibody (ab21176, Abcam) was used at 1:500 dilution (overnight incubation at 4°C). Anti-human IL-6 mouse monoclonal antibody (ab11449, Abcam) was used at 1:300 dilution. Secondary staining was performed using biotinylated anti-mouse antibodies or anti-rabbit antibodies (2 μg/ml in blocking buffer) (Vector Laboratories, Burlingame, CA), followed by detection with Avidin-FITC (1 μg/ml). Fluorescence microscopy was performed using a Nikon TE2000-U fluorescence microscope at 488 nm (FITC) and 568 nm (CM-DiI). Images were captured and processed using a SPOT RT Slider digital camera and software (Diagnostic Instruments, Sterling Heights, MI). The number of CM-DiI-labeled NSCs in tumor metastases was counted in five tissue sections taken from each of five mice. The sizes of the areas of metastatic sites (mm2) were measured using NIH ImageJ software.
Hemi-Nested PCR for V-Myc Gene in NSCs and Single-Step PCR for Luc Gene in Breast Cancer Metastases
DNA was extracted from breast cancer metastatic tissue of mice that had received injections of NSCs. The following PCR was performed for detection of the v-myc gene to detect the HB1.F3.CD cells and the luc gene to detect breast cancer metastases in mouse tissues: (a) Hemi-nested PCR for v-myc. Template DNA (250 ng) was added to 50 μl of a reaction mixture containing 45 μl Platinum PCR supermix high fidelity (12532, Invitrogen, Carlsbad, CA) and 400 nM each primer. A total of 35 cycles at 94°C for 30 seconds each, 56°C for 30 seconds, and 68°C for 45 seconds were performed. The first-round PCR primers were: forward: 5′-CCTTTGT TGATTTCGCCAAT-3′; reverse: 5′-GCGAGCTTCTCCGACACC ACC-3′. After the first-round of PCR, 4 μl of the final product was transferred to the second-round reaction mixture and reamplified using new pairs of primers under the following conditions: 40 cycles at 94°C for 30 seconds each, 56°C for 30 seconds, and 68°C for 30 seconds. The second-round PCR primers were: forward: 5′-TCACAGCCAGATATCCAGCAGCTT-3′; reverse: 5′-AGTTCTCCTCCTCCTCCTCG-3′. The size of the v-myc amplicon in second-round PCR was 198 bp. Positive-control reactions consisted of genomic DNA from HB1.F3.CD NSCs and DNA isolated from the brain of a nontumor-bearing mouse injected with HB1.F3.CD NSCs, while negative controls contained genomic DNA from the brain of a naïve mouse. (b) Single-step PCR for the luc gene. The PCR reaction was performed similarly as described for the first-round PCR of v-myc gene. The thermocycle parameters were: 40 cycles at 94°C for 30 seconds each, 58°C for 20 seconds, and 68°C for 30 seconds. The PCR primers used for detection of the luc gene were: forward, 5′-TTGCCTTCACTGATGC TCAC-3′; reverse, 5′-CCAATGAACAGGGCTCCTAA-3′. The size of the luc amplicon was 178 bp.
Conditioned Media Derived from NSC.rCE Cells
HB1.F3.CD cells were cultured for 48 hours and then exposed to adenovirus with rCE (AdCMVrCE) or adenovirus without rCE (AdCMV) at a multiplicity of infection (MOI) of 0, 20, 40, or 60 for 24 hours, after which the media were replaced with fresh media. After 4 days, the conditioned media were cleared of cells by centrifugation and stored frozen at −80°C until used.
Determination of rCE Activity in Conditioned Media Derived from NSC.rCE Cells
The carboxylesterase activity in conditioned medium derived from NSC.rCE cells was quantified as described previously . Briefly, rCE-catalyzed conversion of the general esterase substrate o-nitrophenyl acetate (o-NPA) to o-nitrophenol (o-NP) was measured spectrophotometrically at 420 nm. One unit represents the amount of enzyme required to convert 1 nmol of o-NPA to o-NP per minute.
Sulforhodamine B Assay for Measurement of Biomass
MCF-7 and MDA-MB-231 cells were cultured in 96-well plates overnight at 37°C, 6% CO2. CPT-11 or SN-38 was added to cell cultures in triplicates at various concentrations (0–100,000 nM). Cells were incubated for 72 hours, and then cell biomass was measured as an indicator of cell survival using Sulforhodamine B (SRB) assay . The SRB reagent was purchased from Sigma-Aldrich. Data are expressed as percentage survival relative to cells cultured in the absence of drugs.
Therapeutic Effect of NSC.rCE Cells in Combination with CPT-11 In Vivo
Mice (nu/nu) were injected in mammary fat pad with 1 × 106 MDA-MB-231-luc cancer cells. After 5 days, 2 × 106 NSC.rCE cells were injected into the tail vein of tumor-bearing mice, followed by CPT-11 injection into tail vein 3 days after injection of NSC.rCE cells (defined as one treatment cycle). Mice received four treatment cycles of NSC.rCEs and CPT-11 on four consecutive weeks. The dose of CPT-11 was 1.5 mg/kg per day i.v. in 50 μl PBS. Mice were randomly divided into four groups for treatment (eight mice per group, total of 32 mice): group (1) NSC.rCE + CPT-11; (2) NSC.rCE only; (3): CPT-11 only; (4) tumor only, no treatment. The mice were euthanized when the primary tumor in the fat pad became ulcerated, and organs were harvested 30–40 days post-tumor implantation. Tissue sections were analyzed for metastases by luc immuhistochemistry and bright-field imaging using an Automated Cellular Imaging System II (Clarient, Inc., San Juan Capistrano, CA). The size (area) of metastases was measured using Image J. The area of metastases was expressed as % of total area of the tissue section. n = 4 mice/group (four sections/organ per mouse); sections were 100 μm apart.
All data were analyzed with GraphPad Prism 4.0 software (GraphPad). Data are presented as mean ± SE. Statistical differences between two groups were evaluated using Student's t test. p values less than 0.05 were considered significant.
NSC-Tumor Tropism Correlates with Invasiveness of Breast Cancer Cells In Vitro
To investigate NSC tropism to breast cancer cells, we used in vitro Boyden chamber assays to compare the migration of NSCs to breast cancer cell lines with varying degrees of invasiveness. NSCs displayed threefold to fourfold greater migration to highly invasive breast cancer cells (MDA-MB-231-luc, MDA-MB-231, MDA-MB-468, and Hs578T), as compared to migration to less-invasive breast cancer cells (MCF-7, MCF-7-ffLuc, T47D, ZR75, and SKBR3) (Fig. 1A, 1B). In addition, enhanced NSC-tumor tropism strongly correlated with the invasiveness of breast cancer cells in vitro (Fig. 1C). MDA-MB-231-luc breast cancer cells displayed the greatest invasiveness and elicited the greatest tropism of NSCs. Therefore, we used this cancer cell line to establish a mammary fat pad metastatic model in nu/nu mice.
IL-6 Is a Major Cytokine That Mediates Tropism of NSCs to Breast Cancer Cells
To identify the cytokines responsible for the attraction of NSCs to breast cancer cells, cytokine levels in breast cancer cell-conditioned media were analyzed by cytokine antibody arrays. IL-6 and IL-8 were most highly expressed in invasive breast cancer cells as compared to less-invasive cancer cells (Fig. 2A, Supporting Information Table 1). The cytokine array data for IL-6 and IL-8 were confirmed by quantitative real-time PCR and ELISA (Fig. 2B–2E).
To further investigate the function of IL-6 and IL-8 in the tropism of NSCs to breast cancer cells, we added these cytokines, singly, or in combination, to conditioned media derived from breast cancer cells with low invasiveness. Addition of IL-6 resulted in increased NSC migration, while addition of IL-8 did not show such an effect (Fig. 3A). Addition of both IL-6 and IL-8 did not affect NSC migration any more than did IL-6 alone (Fig. 3A). The amounts of purified IL-6 and IL-8 added to NSC migration assays (0–100 ng/ml IL-6 or IL-8) (Fig. 3A) were comparable to the levels detected by ELISA in cancer cell-conditioned media (0–3.2 ng/ml of IL-6, and 0–65 ng/ml of IL-8; Fig. 2D, 2E). Next, we added function-blocking antibodies against IL-6 and IL-8 to breast cancer cell-conditioned media in NSC migration assays. IL-6 neutralizing antibody (50 μg/ml) resulted in approximately 50% inhibition of NSC migration to breast cancer cell-conditioned media, but the IL-8 neutralizing antibody did not show a significant effect (Fig. 3B). Therefore, we focused on studying the role of the IL-6R, present on NSCs, on the migration of NSCs to breast cancer cells. Both IL-6R neutralizing antibody and IL-6R siRNA resulted in decreased migration of NSCs to breast cancer cell-conditioned media derived from highly invasive MDA-MB-231 cells. However, such an effect was not observed for NSC migration to cells with low invasive potential (MCF-7), which express low levels of IL-6 (Fig. 3C, 3D). Knockdown of IL-6R in NSCs by siRNA was confirmed using real-time PCR, flow cytometry, and Western blotting (Supporting Information Fig. 1). Taken together, these data suggest that IL-6 is an important cytokine that mediates the selective, tumor-directed migration of NSCs, but other factors may also be involved in NSC-tumor tropism.
Human NSCs Target Primary Tumor and Metastatic Tumor Foci in Mouse Model of Metastatic Human Breast Cancer
To determine whether NSCs can target primary and metastatic breast tumor foci in vivo, we performed initial experiments using human MDA-MB-231-luc cells to establish a mammary fat pad model in nu/nu mice. In vivo Xenogen imaging showed that following MDA-MB-231-luc engraftment into the mammary fat pad, tumor was detected at the primary site at 1 week postimplantation, and lymph node metastases were detected at 3 weeks postimplantation (Fig. 4A, 4B). At the end of the experiment (week 5), ex vivo Xenogen imaging of harvested organs showed that metastases had formed in the lymph nodes in 60% (3/5) of the mice, lungs in 40% (2/5) of the mice, and femurs in 40% (2/5) of the mice. No Xenogen signal was observed in the liver or brain of any mice, suggesting that metastases were absent or below the level of detection (Fig. 4C). Furthermore, immunohistochemistry with an anti-luc antibody revealed a higher detection rate of breast cancer metastases in lymph nodes (100% of the mice), lung (100%), femur (80%), and liver (40%) than did Xenogen imaging, although, again, no metastases were detected in the brain (Fig. 5A, green). These data suggest that our breast cancer xenograft model recapitulates advanced breast cancer in human patients, including development of metastases in the major target organs, such as lymph nodes, lung, bone, and liver.
We further investigated whether NSCs could target primary and metastatic breast tumor sites in mice. After 5 weeks of implantation of tumor cells, CM-DiI-labeled human HB1.F3.CD NSCs were injected into the tail vein. After 4 days of NSC injection, mice were euthanized and tissues from major organs were harvested and processed for immunohistochemistry with an anti-luc antibody. We found that CM-DiI-labeled NSCs (red) targeted metastatic foci (green) throughout the body, with an apparently higher preference for metastatic tumor sites in the lymph node, lung, liver, and femur (Fig. 5A, 5C), versus the primary tumor site. Little or no NSC localization to normal-appearing tissues in these organs was observed (Supporting Information Fig. 2). We concluded that NSCs migrated selectively to breast cancer primary tumor and metastases with preferential tropism to sites of tumor metastases, including lymph node, lung, liver, and femur.
Human NSCs Colocalize with IL-6-Expressing Breast Cancer Metastases In Vivo
Because IL-6 mediated the migration of NSCs to breast cancer cells in vitro, we further investigated the role of IL-6 in NSC targeting of breast cancer metastases in vivo. When breast tumor metastases were detected in our breast cancer xenograft model at week 5, CM-DiI-labeled NSCs (red) were injected into the tail vein. After 4 days of injection of NSCs, tissues were harvested and processed for immunohistochemistry with an anti-IL-6 antibody. Tissue sections were stained to detect the expression of human IL-6 (green) in tumor and counterstained with DAPI (blue). We observed significant targeting by NSCs (red) to IL-6-rich areas in primary and metastatic tumor sites (Fig. 5B).
To obtain metastases of sufficient size to perform real-time PCR for detection of human IL-6 mRNA, we used a tumor resection model in which the primary implanted tumor was resected. Such mice survive longer than mice without tumor resection, allowing the metastases in various organs to grow to a larger size. After 4–8 weeks of resection, organs were harvested and RNA was extracted, followed by real-time PCR using primers specific for human IL-6. Samples were normalized to human β-actin as an endogenous reference. Significantly more human IL-6 mRNA levels were detected in organ metastases, when compared with the primary tumor (p < .01, mean ± SE, n = 3 mice) (Fig. 5D). These data, together with the immunohistochemical colocalization of IL-6 and NSCs (Fig. 5B), suggest that NSCs preferentially target tumor metastases in which IL-6 is highly expressed.
PCR Evidence for NSC and Tumor Cell Colocalization
Metastasis of breast cancer to brain is generally a late feature of metastatic disease , which may explain why we did not detect micrometastases in the brain using immunohistochemistry. To address whether breast cancer metastases formed in brain and whether NSCs migrated to these tumor cells, we used a sensitive molecular approach (PCR) to detect the presence of minimal numbers of tumors cells and NSCs. We used hemi-nested PCR to detect the v-myc gene present in NSCs and single-step PCR to detect luc gene present in tumor cells. PCR for luc gene revealed a positive band in primary tumors and metastases that were detectable by immunohistochemistry, and also in brain (Fig. 5E), which suggests that only few breast cancer cells had metastasized to the brain at the time when mice were euthanized. Tumor metastases to the brain might become detectable by immunohistochemistry at a later stage of disease. PCR to detect the v-myc gene, a marker for HB1.F3.CD NSCs, indicated the presence of NSCs in the primary tumor as well as in metastases, including the lymph nodes, lung, liver, femur, and brain (Fig. 5E). The band indicating the presence of v-myc in the brain was of low intensity, but, nevertheless, it indicated that NSCs can target breast cancer micrometastases in the brain. To confirm the specificity of the PCR, we used DNA isolated from brain of a nontumor-bearing mouse that had been injected intracranially with HB1.F3.CD NSCs as positive control, and as a negative control, we used DNA from the brain of a naïve mouse. We detected an intense v-myc band in the positive control brain but not in the negative control (data not shown). Taken together, these data indicate that NSCs localize to hard-to-treat breast cancer metastases, including brain and bone, and therefore may be useful as drug delivery vehicles for treatment of advanced breast cancer.
Therapeutic Effect of NSC.rCE Cells in Combination with CPT-11 In Vitro and In Vivo
For therapeutic studies, NSCs were genetically modified to secrete rCE, which activates the prodrug CPT-11 to the active drug SN-38, to provide tumor-localized therapeutic drug delivery. Human HB1.F3.CD NSCs were transduced with replication-deficient adenovirus to express rCE (NSC.rCE), and expression of functionally active, secreted rCE was quantified by assessing the CE activity in conditioned medium. First, we determined that conditioned media harvested from transduced cells readily converted the CE substrate o-NPA to o-NP, with an activity of approximately 300, 400, and 500 nmol/min per ml 5 days after transduction at MOI of 20, 40, 60, respectively (Supporting Information Fig. 3). Next, we determined the IC50 cell killing activity (and/or growth inhibition) of CPT-11 and SN-38 against MDA-MB-231 breast cancer cells. The IC50 values of CPT-11 and SN-38 were 25,000 and 100 nM, respectively (Fig. 6A). These data demonstrate that MDA-MB-231 breast cancer cells are approximately 250-fold more sensitive to SN-38 than CPT-11.
To determine the cancer cell killing activity of conditioned media from NSC.rCE cells compared to media from NSCs alone or NSC.null cells, we cultured the MDA-MB-231 cells in such conditioned media in the presence of various concentrations of CPT-11 (0-100 μM). The IC50 cell killing activities of the media from NSC.rCE, NSC alone, NSC.null cells were 0.5, 50, and 100 μM CPT-11, respectively. These data demonstrate that the rCE secreted by NSC.rCE cells resulted in a 200-fold increase in breast cancer cell killing when compared to the NSC.null control (Fig. 6B). The data also suggest that the NSC.rCE/CPT-11 enzyme/prodrug system may be effective in killing breast cancer cells in vivo.
To investigate the in vivo therapeutic efficacy of the NSC.rCE/CPT-11 system, we used our established mammary fat pad human breast cancer xenograft model. Treatment groups were as follows: (1) NSC.rCE + CPT-11; (2) NSC.rCE only; (3) CPT-11 only; (4) tumor only, no treatment. Mice received the above treatments weekly for four consecutive weeks, a regimen similar to that used in clinical trials of breast cancer patients treated with CPT-11 . A CPT-11 dose of 1.5 mg/kg body weight of mice was chosen based on the pharmacokinetics of CPT-11 and endogenous carboxylesterase activity in nu/nu mice. Organs were harvested 30–40 days post-tumor implantation. Mice treated with NSC.rCE cells in combination with CPT-11 resulted in reduce size of tumor metastases in lung and lymph nodes compared to the control NSC.rCE only and tumor only groups (p < .05) (Fig. 6D). Representative images of lymph nodes from experimental groups stained for luc are shown in Figure 6E.
Metastasis is the leading cause of mortality in breast cancer patients. Breast cancer metastases often develop in regional lymph nodes and in distant organs, including bone, lung, liver, and brain . A major obstacle for current breast cancer therapies is ineffective delivery of drugs to specific metastatic tumor areas, especially bone and brain metastases [23, 25]. Indeed, more than 70% of patients who die as a result of breast cancer show skeletal involvement at autopsy . Therefore, designing specific therapeutic interventions for breast cancer metastases is a major goal but also a significant challenge. Furthermore, chemotherapy kills not only cancer cells but can also harm healthy tissues, causing serious undesired side effects . Therefore, doses of chemotherapeutics must be limited to prevent permanent or life-threatening damage. The use of NSCs as drug delivery vehicles may offer a major therapeutic breakthrough for treatment of metastatic breast cancer, by localizing high concentrations of chemotherapy selectively to tumor sites, thus overcoming some major limitations of currently available treatment strategies. Here, we have shown that NSCs can specifically target primary and distant metastatic breast cancer foci in bone, brain, and other tissues. This stem cell-based breast cancer therapy should be readily translatable to clinical studies, as is currently being used in an investigational phase I clinical trial in adult patients with recurrent glioblastoma (clinical trial ID # NCT01172964; http://clinicaltrials.gov/ct2/show/NCT01172964).
To our knowledge, this is the first report to demonstrate preferential NSC tropism to breast cancer metastases, including lymph nodes, lung, liver, bone, and brain, versus the primary mammary fat pad tumor in a metastatic mouse model. Previous studies have shown that NSCs or mesenchymal stem cells can migrate to breast cancer metastases in lung following i.v. injection or to breast cancer cell implants in mouse brains [27–31]. Rachakatla et al.  reported that combination treatment of human umbilical cord matrix stem cell-based interferon-β gene therapy and 5-FU (administered intraperitoneally) significantly reduced the growth of metastatic human breast cancer in the lungs of SCID mice. In addition, Joo et al.  showed that HB1.F3.CD NSCs migrated selectively to MDA-MB-435 human breast cancer cells implanted into brain, when such NSCs were injected into the hemisphere contralateral to the tumor. This NSC-mediated therapy in combination with 5-FC resulted in significantly reduced tumor volume. More recently, Seol et al.  used the same MDA-MB-435 breast cancer cell intracranial xenograft model in mice and NSC.rCE/CPT-11 therapy, which resulted in decreased tumor volume.
The intramammary fat pad breast tumor engraftment model we used, which forms both primary and metastatic tumors, closely resembles the multiple stages involved in development of malignant breast cancer in humans. The luciferase-expressing subclones of MDA-MB-231 cells (MDA-MB-231-D3H2LN-luc cells) used in our study were isolated for stable firefly luciferase expression in vitro and passaged in mice to enhance their tumorigenic and metastatic potential . We showed that the MDA-MB-231-D3H2LN-luc cell line produced multiple metastases at high frequencies in clinically relevant tissues such as lung, liver, and bone after injection into mammary fat pads. Importantly, we demonstrate that NSCs preferentially target metastases in multiple organs in mice, whereas most currently available therapies do not possess such targeting ability in human patients.
Metastasis is a complex process that requires three essential characteristics of cells: expression of invasion/metastasis-associated genes, cancer cell invasion, and homing to and proliferation at sites of metastasis. The invasive property is an early step that is necessary for metastasis . Our data showed that NSC-tumor tropism strongly correlated with the degree of invasiveness of breast cancer cells in vitro (Fig. 1). Furthermore, NSCs had high preference for metastatic tumor sites in the lung, lymph node, femur and liver, as compared to primary tumors (Fig. 5A). Because NSCs strongly localized to hard-to-treat breast cancer metastases, this result may have important potential therapeutic implications, including for the treatment of both primary tumors and metastases, possibly as an adjuvant therapy to eliminate micrometastatic disease after resection of the primary tumor. Of note, the highly invasive breast cancer cell lines that attracted large numbers of NSCs are “triple-negative” (i.e., they do not overexpress HER-2 and are negative for estrogen and progesterone receptors; Fig. 1, Supporting Information Table 2) [34, 35]. Because triple-negative breast cancer is extremely difficult to treat, efficient targeting by NSCs may be especially important for developing a stem cell-based therapy for this subtype.
Many stem cell attractants and cytokines that emanate from solid tumors, including glioma, have been identified. Agents such as stem cell factor-1, monocyte chemoattractant protein-1, and stromal cell-derived factor-1 are potent chemotactic molecules known to stimulate hematopoietic cell migration but have also been shown to stimulate NSC migration [36–39]. Furthermore, growth factors such as epidermal growth factor, hepatocyte growth factor, and vascular endothelial growth factor are also potent mediators of NSC migration to tumors [39–41]. However, the NSC-attracting cytokines produced by breast cancer cells have not yet been defined. Identification of breast cancer-produced cytokines involved in NSC targeting will be important for developing optimal NSC-based cancer therapies. In metastatic breast cancer, high IL-6 levels in the serum have been associated with a greater number of metastatic sites, poorer clinical outcome, lack of response to therapy, resistance to chemotherapy, and resistance to hormonal therapy . Yokoe et al.  showed that serum IL-6 levels in patients with progressive recurrent breast cancer who did not respond to therapy were significantly higher than levels in recurrent breast cancer patients who were stable after therapy. Here, we provide the first evidence that IL-6 produced by breast cancer cells is an important mediator in NSC targeting to invasive breast cancer cells in vitro and to metastases in vivo in mice (Fig. 3, Fig. 5B). Of the 79 cytokines investigated by antibody array, we found that IL-6 was highly expressed in most invasive breast cancer cells (Fig. 2A, 2B, 2D, and Supporting Information Table 1). In addition, our cytokine array data suggest that in addition to IL-6, other cytokines are also expressed and secreted by breast cancer cells, for example, IL-1α, fibroblast growth factor 4, oncostatin M, and vascular endothelial growth factor (Fig. 2A and Supporting Information Table 1). Further work will be needed to determine the role of these cytokines in NSC tropism to breast cancer.
In addition to in vitro studies, we also report a novel in vivo study that uses the inherent tumor tropism of NSCs transduced with adenovirus-expressing rCE to target breast tumor metastases. We observed a significant reduction in the size of metastases in the lung and lymph nodes of mice after treatment with NSC.rCE/CPT-11 comparing to tumor only and stem cell only group. Then we observed a tendency toward reduction in the size of metastases in the lung and lymph nodes as compared to CPT-11 only; however, this did not reach statistical significance. This might be due to the presence of endogenous carboxylesterase(s) in the nu/nu mice that can convert CPT-11 to SN-38, thus masking such conversion by the NSC.rCE cells. To more clearly distinguish between the therapeutic effect of NSC.rCE + CPT-11 versus CPT-11 alone, future studies will be conducted in esterase-deficient severe combined immunodeficiency (Es1e/SCID) mice, which have plasma esterase activity levels similar to those in human patients .
A major advantage of the NSC.rCE/CPT-11 therapeutic modality is the efficient distribution of NSCs and delivery of activating enzymes selectively to multiple breast cancer metastases throughout the body. CE/CPT-11 also known to have a large bystander effect should increase the radius of tumor killing around the rCE-secreting NSCs. This targeted therapy of breast cancer may increase the therapeutic index in patients, while minimizing toxicity to normal tissues, thus also improving quality of life.
Breast cancer patients with metastases have especially poor prognosis because these sites are particularly difficult to access with chemotherapy. Here, we report the first study in a preclinical model that NSCs have a preferential tropism to breast cancer metastases in multiple organs as compared to the primary tumor site. Additionally, our in vitro and in vivo data suggest that IL-6 may be a major cytokine involved in mediating NSC-tumor tropism, as demonstrated with highly invasive, IL-6-expressing breast tumor cells. In our therapeutic study, we used the NSC.rCE/CPT-11 treatment strategy in a metastatic breast cancer mouse model. NSC.rCE cells in combination with CPT-11 resulted in reduced size of tumor metastases in lung and lymph nodes. These results warrant further investigation of NSCs as vehicles for targeted delivery of anti-cancer therapies for advanced metastatic breast cancer.
We are grateful to Dr. Keely L. Walker (City of Hope) for critical reading and editing of the article. We thank Soraya Aramburo and Valerie V. Valenzuela for their help with animal experiments. We acknowledge financial support of this work from the Department of Defense-Breast Cancer Concept Award (BC087425), Rosalinde and Arthur Gilbert Foundation, STOP CANCER Foundation, and CIRM Stem Cell Research Biotechnology Training Program at CSULB (TB1-01182).
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
KA and AA are co-founders of TheraBiologics, Inc., which supports the development of NSC-mediated treatments for cancer. The authors indicate no other potential conflict of interest.