A major obstacle to the success rate of chimeric antigen receptor (CAR-) T-cell therapy against solid tumors is the microenvironment antagonistic to T cells that solid tumors create. Conventional checkpoint blockade can silence lymphocyte antisurvival pathways activated by tumors, but because they are systemic, these treatments disrupt immune homeostasis and induce autoimmune side effects. Thus, new technologies are required to remodel the tumor milieu without causing systemic toxicities. Here, we demonstrate that targeted nanocarriers that deliver a combination of immune-modulatory agents can remove protumor cell populations and simultaneously stimulate antitumor effector cells. We administered repeated infusions of lipid nanoparticles coated with the tumor-targeting peptide iRGD and loaded with a combination of a PI3K inhibitor to inhibit immune-suppressive tumor cells and an α-GalCer agonist of therapeutic T cells to synergistically sway the tumor microenvironment of solid tumors from suppressive to stimulatory. This treatment created a therapeutic window of 2 weeks, enabling tumor-specific CAR-T cells to home to the lesion, undergo robust expansion, and trigger tumor regression. CAR-T cells administered outside this therapeutic window had no curative effect. The lipid nanoparticles we used are easy to manufacture in substantial amounts, and we demonstrate that repeated infusions of them are safe. Our technology may therefore provide a practical and low-cost strategy to potentiate many cancer immunotherapies used to treat solid tumors, including T-cell therapy, vaccines, and BITE platforms.
Significance: A new nanotechnology approach can promote T-cell therapy for solid tumors. Cancer Res; 78(13); 3718–30. ©2018 AACR.
The potential of immunotherapy as a cancer treatment option is evident from the positive outcomes many patients with leukemia show in response to adoptive cell transfer using autologous T cells genetically modified to express disease-specific chimeric antigen receptors (CAR; refs. 1–3). However, the vast majority of cancers, in particular the more common solid malignancies (such as those occurring in the breast, colon, and lung), fail to respond significantly to CAR-T-cell infusions (4–7). This is because solid cancers present formidable barriers to adoptive cell transfer, especially by suppressing T-cell functions via the inhibitory milieu they surround themselves with (8, 9).
To combat immunosuppression of T-cell therapy, many clinical trials are focused on disabling checkpoint blockades (10, 11). This is not surprising, as several antibodies targeting checkpoint molecules (such as PD-1, PD-L1, and CTLA-4) have already been approved by the FDA for the treatment of certain types of cancer, and preclinical studies have demonstrated increased CAR-T-cell potency when these are coadministered with them (10, 12). However, the tumor microenvironment comprises a complex network of heterogeneous cell types that express a variety of different immune inhibitory receptors, and it has become clear that blocking one pathway simply promotes the others, along with compensatory cellular mechanisms that ultimately enable tumors to develop resistance (13, 14). Moreover, the systemic autoimmune toxicity produced by these broad-acting treatments, as well as their high costs, limits widespread use of this therapy (15).
Biotechnology could solve this problem by making available inexpensive nanoparticle reagents that deliver rationally chosen combinations of immunomodulatory drugs into the tumor microenvironment without inducing adverse systemic side effects (illustrated in Fig. 1). In the research described here, we designed lipid nanoparticles containing a potent drug cocktail that can block suppressor cells within the tumor microenvironment and simultaneously stimulate key antitumor immune cells. Using the in vivo mouse 4T1 syngeneic breast cancer model (16, 17), we found that when administered at the optimal time and frequency, these drug nanocarriers effectively reverse the immune-hostile cancer environment and thereby create a therapeutic window of vulnerability to T cell–mediated cancer suppression. We establish that infusing tumor-specific CAR-T cells during this time frame results in disease clearance in half of the treated animals and more than doubled the survival of the others, as (in contrast to conventional CAR-T-cell therapy) infused T cells were able to effectively infiltrate tumor lesions, undergo robust expansion, and ultimately clear malignant cells. These findings were confirmed in a genetically engineered mouse model of human glioma (18), which is a tumor type notoriously resistant to many currently available immunotherapies (19, 20). We found that nanoparticle preconditioning doubled the overall survival compared with conventional CAR-T-cell therapy only.
Materials and Methods
The murine 4T1 breast cancer cell line (ATCC, cat no. CRL-2539) was cultured in complete RPMI 1640 medium containing 10% heat-inactivated fetal bovine serum (FBS), 2 mmol/L l-glutamine, 1.5 g/L sodium bicarbonate, 4.5 g/L glucose, 1.0 mmol/L sodium pyruvate, 0.05 mmol/L 2-mercaptoethanol, and 10 mmol/L HEPES buffer (pH 7.2–7.5). To generate a 4T1 tumor cell line that expresses murine ROR1 (4T1-ROR1), we transfected cells with the ROR1 lentiviral vector (pLenti-GIII-CMV-GFP-2A-Puro) marketed by abmgood.com (cat. no. LV536399). Tumor cells were cultured in complete medium containing 10 μg/mL puromycin (Invivogen) for three weeks. The Platinum-E (Plat-E) retroviral packaging cell line (Cell Biolabs, cat no. RV-101) was cultured in DMEM containing 10% FBS, 2 mmol/L glutamate, 100 U/mL penicillin, and 100 μg/mL streptomycin. To ensure high retroviral titer, only Plat-E cells passaged no more than 5 times were used for gene transfer. Cell line authentication was not performed. All cell lines tested negative for Mycoplasma using a DNA-based PCR test (DDC Medical).
Retroviral vectors and virus production
The anti-ROR1-28z CAR and the anti-EGFRvIII-28z CAR were custom-designed by Creative Biolabs. Each consists of a single-chain antibody-targeting mouse ROR1 or EGFRvIII linked via a c-myc tag to a synthetic receptor skeleton comprised of the CD8 hinge, the CD28 transmembrane and signaling domains, and the signaling domain from CD3ζ. This construct was then cloned by VectorBuilder into their MMLV Retrovirus Gene Expression Vector. SFG-CBR-luc vectors (which express click beetle luciferase) were kindly provided by Dr. Michel Sadelain (Memorial Sloan-Kettering Cancer Center, New York, New York). To generate retroviral particles encoding the ROR1-28z CAR or CBR, we transfected the Plat-E retroviral packaging cell line according to the manufacturer's instructions and harvested retroviral supernatant 48 hours later.
Preparation of tumor-targeting lymphocytes
To generate ROR1-specific (CAR-transduced) T cells, spleens of BALB/cJ mice were macerated over a filter and resuspended in ACK lysing buffer (Biosource). To generate EGFRvIII-specific CAR-T cells, we used spleens from Ntv-a; Ink4a-Arf−/−; floxed PTEN EGFRvIII transgenic mice. CD8+ T cells were immunomagnetically selected using the EasySep mouse CD8+ T-cell isolation kit (Stemcell Technologies). We then activated the murine T cells with anti-mouse CD3 (10 μg/mL), anti-mouse CD28 (0.1 μg/mL), and mIL2 (19 IU/mL), and transduced them 24 hours later with the Plat-E retroviral supernatant on Retronectin-coated plates. Following spinoculation (3,000 × g, 2 hours, 32°C), T cells were harvested and resuspended in complete RPMI with IL2 (10 IU/mL) and mouse T Activator Beads (Thermofisher; 0.3 × 106 beads per 1 × 106 T cells). Following a second spinoculation in retroviral supernatant the next day, the cells were cultured for 24 hours in the presence of mouse IL15 (50 ng/mL) before using them as a tumor therapeutic.
We measured in vitro cytotoxic activity of T cells using standard 51Cr release assays as described elsewhere (21). Briefly, 4T1 or 4T1-ROR1 cells were labeled with 51Cr for 1 hour at 37°C, washed with RPMI containing 10% FBS, and resuspended in the same medium at a concentration of 1 × 105 tumor cells/mL. T cells were added to the suspensions at varying effector-to-target cell ratios in 96-well plates (final volume, 200 μL) and incubated for 4 hours at 37°C, then 30 μL of supernatant from each well was transferred into Lumaplate-96 microplates for analysis with a Top Count NXT microplate scintillation counter (Packard Bioscience).
Liposome synthesis and characterization
Solutions of egg phosphatidylcholine, cholesterol, 18:0 PEG(2000)-PE, and DSPE-PEG-maleimide (65:30:4:1 molar ratio, 10 mg total lipid; Avanti Polar Lipids) in chloroform were combined in a 20-mL glass vial, then 50 μg 7DW8-5 (Funakoshi) and 1.0 mg PI-3065 (ChemShuttle) were dissolved in the lipid solution. The chloroform was removed via a rotary evaporator, and then residual solvent was eliminated under high vacuum overnight. The resulting lipid film was hydrated in 5 mL PBS using a vortex device for 1 minute, and then sonicated for 5 minutes to produce a liposome suspension.
A solution of iRGD-SH (3 molar equivalents with respect to maleimide, GL Biochem) and TCEP (30 equivalents, Thermo Fisher) was prepared in PBS to yield 5 mmol/L iRGD-SH. This solution was incubated at room temperature for 5 minutes and then added to the liposome suspension. After rotary mixing for 1 hour, the suspension was dialyzed (Slide-A-Lyzer Dialysis Cassette, 20K MWCO, Thermo Fisher) against water for 2 hours. Sucrose was added to the dialyzed liposome suspension to a final concentration of 2.9% before it was lyophilized for storage at −20°C. For application, the liposomes were resuspended to their original concentration in sterile PBS. The hydrodynamic radius of the particles was measured with a Nanosite (Malvern), and their zeta potential was determined using dynamic light scattering detected with a Zetapals instrument (Brookhaven Instrument Corporation).
Measurements of drug encapsulation efficiencies and drug-release kinetics
Encapsulation efficiency of PI-3065. Rehydrated liposomes were passed through a desalting column (Thermo Scientific Zeba spin column, 7K MWCO) to remove free PI-3065, followed by disruption with 1% Triton X-100. Encapsulated PI-3065 concentration was measured by HPLC. The total concentration of PI-3065 was obtained by repeating the procedure using unfiltered liposomes.
In vitro release of PI-3065 from liposomes. Rehydrated liposomes (1.8 mL) were placed in a dialysis device (Thermo Scientific Slide-A-Lyzer MINI, 3.5K MWCO, 2 mL) and then immersed in 45 mL water at 37°C for 70 hours. Dialyzed material (200 μL) was removed at predetermined time points and treated with 1% Triton X-100. PI-3065 concentration was measured by HPLC.
HPLC analysis of PI-3065. The HPLC system consisted of an Agilent 1260 Infinity Quaternary Pump equipped with a UV detector and a Poroshell 120 C18 column, 4.6 × 50 mm, 2.7 μm (Agilent Technologies). Sample (20 μL) was applied to the column, which was equilibrated with 90% buffer A/10% buffer B (buffer A: water + 0.1% phosphoric acid, buffer B: acetonitrile + 0.1% phosphoric acid). Elution was performed at 1 mL/minute using a linear gradient to 5% buffer A/95% buffer B over 8 minutes, followed by a 2 minute equilibration back to the starting mobile phase ratio. PI-3065 was detected at 254 nm.
Mice and tumor models
Female BALB/cJ mice (4–6 weeks old) were obtained from The Jackson Laboratory (stock no. 000651) and housed in the animal facility of Fred Hutchinson Cancer Research Center. They were used in the context of an animal protocol approved by our Institutional Animal Care and Use Committee. After 1 × 106 4T1-ROR1 tumor cells were transplanted into the mammary gland, the disease was allowed to establish for 10 days before starting treatments. As indicated, drug-loaded liposomes, nonencapsulated drugs, and anti-ROR1 CAR-T cells were administered intravenously with the exception of experiments described in Fig. 2, where CAR-T cells were injected directly into tumor lesions. Progression of tumor growth following therapy was determined by caliper measurements.
Nestin-specific, Cre-recombinase–responsive EGFRvIII mice were generated by crossing Nestin-Tv-a Ink4a/Arf−/− Pten−/− mice with Cre-responsive Stopfl/fl EGFRvIII mice (previously described; ref. 22). Following nine generations of crosses, we genotyped mice for Nestin-Tv-a Ink4a/arf−/− Pten−/− Stopfl/fl EGFRvIII. These mice were then used for generating EGFRvIII-overexpressing gliomas. The initiation of PDGF-driven gliomas with RCAS has been previously described (23).
In vivo bioluminescence T-cell imaging
We visualized T cells using d-luciferin (Xenogen) in PBS (15 mg/mL) as a substrate for CBR-luc. Bioluminescence images were collected with a Xenogen IVIS Spectrum Imaging System (Xenogen). Living Image software version 4.3.1 (Xenogen) was used to acquire (and subsequently quantitate) the data 10 minutes after intraperitoneal injection of the probe into animals anesthetized with 150 mg/kg of 2% isoflurane (Forane, Baxter Healthcare). Acquisition times ranged from 10 seconds to 5 minutes.
To measure potential in vivo toxicities of repeated liposome infusions, we injected mice intravenously with a total of nine doses of 7 × 1013 liposomes carrying PI-3065 and 7DW8-5 (3 doses × 3 weeks) or with phosphate-buffered saline as a control. To measure effects of the nanocarriers per se, a third cohort of mice was injected with empty liposomes. Each experimental group consisted of 8 mice. Twenty-four hours after the final liposome infusion, the animals were anesthetized, and blood was collected by retro-orbital bleed using a heparinized microcapillary tube. Blood was collected into microcontainers containing ethylenediaminetetraacetic acid for basic analysis, which included a white blood cell count with differential, a red blood cell count, hemoglobin, hematocrit, and a platelet count. Blood was also collected into serum separator tubes for serum chemistry and cytokine profiles performed by Ani Lytics, Inc. Animals were then euthanized with carbon dioxide to retrieve organs, which were washed with deionized water and fixed in 4% paraformaldehyde. The tissues were processed routinely, and tissue sections were stained with hematoxylin and eosin before evaluation by Dr. Smitha Pillai, MVSc, PhD, DACVP, a board-certified staff pathologist, in a blinded fashion.
All flow cytometry antibodies were purchased from eBioscience, including their anti-mouse FoxP3 staining set to detect regulatory T cells. We used CD1d tetramer from Proimmune to stain mouse iNKT cells. We used a mouse tumor dissociation kit from Miltenyi Biotec to generate single-cell suspensions from 4T1-ROR1 tumors. Subsequently, erythrocytes and dead cells were removed by Percoll density gradient. Signals were acquired using a FACSCanto Flow Cytometer (BD Biosciences).
Pairwise differences in bioluminescent signals, or cell numbers per tumor weights, were analyzed at selected time points using the Wilcoxon rank-sum test, and we characterized survival data using the log-rank test. A P value less than 0.05 was considered statistically significant. We treated 8 animals per group, which provided 80% power to detect an effect size of 1 standard deviation between groups, based on a t test with a two-sided significance level of 0.05. With the exception of toxicity studies, investigators conducting the experiments were not blinded. All statistical analyses were performed using GraphPad Prism software version 6.0.
Experiments and handling of mice were conducted following federal, state, and local guidelines under an Institutional Animal Care and Use Committee protocol and with approval from the Fred Hutchinson Cancer Research Center Institutional Animal Care and Use Committee.
Therapeutic limitations of conventional CAR-T-cell therapy
In our first set of experiments, we examined why injections of cancer-specific CAR-T cells fail to eradicate solid tumors, using an orthotopic 4T1 mouse mammary tumor model that recapitulates the immunosuppressive microenvironment and metastatic pattern typified by human breast cancer (16, 17). To create therapeutic lymphocytes against these tumors, we transduced mouse T cells with a retrovirus encoding an ROR1-28z CAR specific for the tyrosine kinase-like orphan receptor ROR1 (24), which is strongly expressed at the cell surface of breast cancers (particularly triple-negative breast cancer; refs. 25, 26). This CAR is composed of an anti-mouse ROR1 scFv fused to the murine CD8 transmembrane region, murine CD28 signal transduction domain, and murine CD3 cytoplasmic domains (Fig. 2A). We confirmed successful transduction of mouse T lymphocytes with ROR1-28z transgenes (Fig. 2B), which enables them to destroy 4T1 tumor cells existing in culture (Fig. 2C). To determine whether these CAR-T cells can persist and proliferate in the tumor microenvironment in vivo and identify how tumors adapt to these attacks, we locally injected 1 × 106 ROR1-28z CAR-T cells into established 4T1 tumors (thereby bypassing the problem of tumor targeting) and compared cellular compositions of the tumor microenvironment with that of untreated 4T1 tumors over time. We found that ROR-28z CAR-T cells did not thrive in 4T1 tumors and expanded only 1.65-fold over a 12-day period (Fig. 2D, bottom) before tumors reached endpoint criteria based on their size. Tumor growth was associated with an increased accumulation of cells known to inhibit tumor-specific T-cell responses in patients with cancer, including tumor-associated macrophages (TAM; ref. 27), myeloid-derived suppressor cells (MDSC; ref. 28), and regulatory T cells (Treg; ref. 29). Notably, CAR-T-cell treatment was associated with higher numbers of these suppressor cells (in particular populations of monocytic- and granulocytic-MDSCs; Fig. 2D, top) compared with controls, indicating that tumors can amplify their populations as a strategy to evade T-cell attacks.
We next examined the therapeutic benefit provided by a clinically representative dose of ROR1-28z CAR-T cells (10 × 106) administered systemically rather than locally. In order to track and quantify the in vivo migration and accumulation of the transferred T cells in relation to 4T1 tumors, we included vectors for click beetle luciferase (CBR-luc; ref. 30) in the plasmid. The results establish that, although intravenously infused T cells accumulate at high levels in the spleen and the liver, they inefficiently traffic to tumor sites (Fig. 3A–C), yielding a modest 6-day survival advantage compared with untreated control animals (Fig. 3D and E).
Rational design of nanocarriers to reshape the tumor milieu
In the case of conventional nanocarrier-mediated chemotherapy, it is desirable for the tumor cells to internalize the drug-loaded particle (31, 32). By contrast, immune modulators are more effective if they are slowly released from nanocarriers situated in the stroma, so they must be engineered in a way different from those designed to deliver chemotherapy or siRNA.
We decided upon a liposomal drug carrier composed of egg phosphatidylcholine, cholesterol, and DSPE–PEG–maleimide (Fig. 4A) rather than polymeric nanoparticles because liposomes tend to be nontoxic, biodegradable, and nonimmunogenic (33, 34). Furthermore, their flexible physicochemical properties facilitate manipulation in order to address various delivery considerations. To propel homing of the particles to tumors, we decorated them with iRGD peptide so they can bind to αv integrins (35). Proteolytic processing of iRGD then unmasks a second sequence motif, the cryptic C-terminal CendR motif (R/KXXR/K), which binds to neuropilin-1 and activates an endocytic bulk transport pathway through tumor tissue (36). Once activated, this pathway can take iRGD-bound and -unbound payloads deep into solid malignancies (as demonstrated in human tumor explants as well as in mouse models; refs. 35, 37–39). We confirmed the tumor-targeting effect of iRGD in a mouse model of glioma, which we use as one of our in vivo test systems (see Therapeutic studies below). Decorating liposomes with iRGD peptide improved tumor localization of systemically infused liposomes by an average of 3.2-fold, when compared with nontargeted liposomes (P = 0.0074; Supplementary Fig. S1A and S1B). To attenuate immunosuppression in the tumor microenvironment and simultaneously increase the number and activity of tumor-specific effector immune cells, we chose to combine two immunomodulator agents as nanoparticle cargo: PI-3065, a selective inhibitor of the p110δ PI3K kinase that has activity against immune-suppressive regulatory T-cell subsets and tumor-associated myeloid cells (40), and 7DW8-5, an immunostimulant-invariant natural killer T-cell (iNKT) agonist with 100 times higher potency than its parental compound, α-GalCer (41, 42). We decided to stimulate iNKT rather than T cells, as activated iNKT cells themselves orchestrate broad antitumor attacks that include several other cell types, including natural killer (NK-) cells, T cells, macrophages, neutrophils, and dendritic cells (43). Both PI-3065 and 7DW8-5 are highly hydrophobic and can be efficiently placed inside the bilayer of the liposome (Fig. 4A; Supplementary Fig. S2A–S2C).
To evaluate the advantage targeted carriers provide in comparison with freely circulating therapeutics, we treated mice bearing 4T1 tumors with triweekly intravenous infusions of iRGD-liposomes loaded with PI-3065 and 7DW8-5 (either alone or in combination) or, for comparison, systemic injections of the carrier-free drugs in equal amounts. Afterward, differences in the composition of immune suppressor cells and antitumor cells were measured, and tumor progression rates were compared. We observed that only the nanoparticle-formulated combination of PI-3065 and 7DW8-5 potently reduced the concentration of immune suppressors at the tumor site, in particular TAMs (9.4-fold), monocytic MDSCs (4.6-fold), and Tregs (4.8-fold). At the same time, this combination increased the number of antitumor immune effector cells, such as CD8+ T cells (6.2-fold) and iNKT cells (29.8-fold; Fig. 4B). As a result, the disease remained stable in such treated animals for as long as 30 days following the first nanoparticle injection, whereas tumors progressed in all other treatment cohorts. However, despite continued triweekly nanoparticle infusions, tumors eventually developed resistance and relapsed (Fig. 4C and D), displaying high numbers of monocytic MDSCs and TAMs (Fig. 4E). These data indicate that rationally designed nanocarriers can significantly improve the effect of immunomodulatory agents compared with nontargeted drugs and achieve substantial tumor stasis even in relatively nonimmunogenic cancers, such as our 4T1 model.
Guided by the therapeutic benefits achieved with PI-3065/7DW8-5 liposome injections, we next assessed whether these nanoreagents are biocompatible and safe for repeated dosing. Mice were injected with a total of nine doses of 7 × 1013 liposomes (3 doses/week × 3 weeks) or phosphate-buffered saline as a control. To measure effects of the nanocarriers per se, a third cohort of mice was injected with empty control liposomes. Mice were euthanized 24 hours after the final dose, body weight was recorded, blood was collected by retro-orbital bleed for serum chemistry, and a complete gross necropsy was performed. There was no difference in body weights between groups. The following tissues were evaluated by a board-certified staff pathologist: liver, kidney, heart, spleen, and lungs. Histopathologic evaluation revealed moderate mononuclear infiltrates in the liver and heart of PI-3065/7DW8-5 liposome-treated mice (Fig. 5), which is a mild version of the severe hepatitis reported upon conventional α-GalCer treatment (44). Cellular infiltrates in the parenchyma were occasionally associated with hepatocellular loss and hepatocytes showing high mitotic activity (up to 5 mitotic figures were present in a single 400 × field, Fig. 5). Importantly, overall liver function was normal in all PI-3065/7DW8-5 liposome-treated mice, with only minimally elevated blood levels of the liver enzyme alanine transaminase (ALT; Fig. 5). Also, kidneys of PI-3065/7DW8-5 liposome-treated mice showed slight multifocal interstitial, perivascular, and periglomerular infiltrates of mononuclear cells (Fig. 5); these lesions were subtle, and there was no associated proteinuria or necrosis. Serum chemistry revealed mild leukocytosis in the PI-3065/7DW8-5 liposome treatment group, accounted for by neutrophilia, monocytosis, and lymphocytosis, which reflects the proinflammatory condition caused by treatment with α-GalCer. We also detected a mild anemia in this group, which is likely associated with the inflammatory phenotype (Fig. 5). In “anemia of inflammation,” erythropoiesis is iron restricted by hepcidin-mediated hypoferremia, and erythrocyte production is suppressed by cytokines acting on erythroid progenitors (45). Levels of glucose were within normal limits for all mice tested (106-278 mg/dL). Hyperglycemia, a documented side effect of PI3K inhibitors (46), was not present in treated mice. All serum chemistry, blood counts, and histopathology of animals injected with empty control liposomes were normal. Taken together, these results indicate that PI-3065/7DW8-5 liposomes are biocompatible and safe for repeated dosing. The most prominent histologic lesions, mononuclear infiltrates around portal tracts, central veins, and within the parenchyma of the liver, were minor reactions that usually resolve with minimal or no clinical intervention (47).
The results described above indicate that repeated infusions of targeted PI-3065/7DW8-5 liposomes are a safe and effective method to create a tumor microenvironment where suppression of immune effector cells is minimal and functional support for them is maximal. Accordingly, we next sought to synergistically combine nanoparticle conditioning and adoptive T-cell therapy to test whether liposomes can create a therapeutic window for externally programmed tumor-specific T cells to be effective. To accomplish this, we repeated our in vivo studies in the 4T1 breast cancer model (which did not respond well to conventional infusions of CAR-T cells; Fig. 2), but this time administered liposome infusions according to the treatment regimens illustrated in Fig. 6A: (i) 9 doses of PI-3065/7DW8-5 liposomes before CAR-T-cell infusion, or (ii) CAR-T-cell administration first, followed by 9 doses of PI-3065/7DW8-5 liposomes. As in our previous experiments, we retrovirally transduced adoptively transferred CAR-T cells with luciferase to track their migration and accumulation in relation to 4T1 tumors. Serial imaging of mice revealed that ROR1-28z CAR-T cells concurrently injected with the starting dose of PI-3065/7DW8-5 liposomes showed only a modest (yet significant) improvement in their ability to expand in established 4T1 tumors (3.4-fold by day 6 after infusion; Fig. 6B and C, middle). By contrast, T cells infused after 3 weeks of PI-3065/7DW8-5 liposome preconditioning accumulated in breast tumors at 22-fold higher levels by day 6 after infusion relative to T cells injected into mice that received no preconditioning (Fig. 6B and C, right). Quantitatively similar results were obtained by dissecting tumors 6 days after T-cell transfer and assessing the frequency of infiltrating ROR1-28z CAR-T cells by flow cytometry (Fig. 6D and E). Liposome preconditioning followed by CAR-T-cell therapy eradicated 4T1 tumors in 4 of 8 treated mice (Fig. 6F–H), and the other 4 showed substantial tumor regression with a doubling (average 42-day improvement) in survival compared with conventional CAR-T-cell therapy. This regimen was safe and well tolerated (Supplementary Fig. S3A and S3B). Although PI-3065/7DW8-5 liposomes administered during and after T-cell therapy never produced complete clearance, they increased survival by 17 days compared with injecting T cells only (Fig. 6F–H). We already established in Fig. 4 that unformulated PI-3065 as monotherapy has no significant therapeutic effect in our 4T1 breast tumor model. To test whether repeated PI-3065 infusions can boost the antitumor activity of adoptively transferred CAR-T cells in this model, we administered CAR-T cells, followed by 9 doses (150 μg each) of PI-3065 (Supplementary Fig. S4A). We measured a modest increase in bioluminescent CAR-T-cell signal at the tumor site (1.4-fold higher by day 6 after T-cell infusion, P = 0.14, ns) in animals that received PI-3065 (Supplementary Fig. S4B and S4C), which translated into an average 8-day survival benefit (Supplementary Fig. S4D and S4E).
To evaluate the therapeutic efficacy of our technology in a second clinically relevant system, we used a genetically engineered mouse model of human glioblastoma (18). Adoptive cell therapy using T cells engineered with CAR targeting an ideal molecular marker in glioblastoma, e.g., epidermal growth factor receptor type III (EGFRvIII), has demonstrated satisfactory efficacy in treating malignant brain tumors (48). To model this clinical scenario, we induced EGFRvIII+ glioma driven by RCAS-PDGF and RCAS-Cre in Ntv-a; Ink4a/Arf−/−; floxed PTEN transgenic mice (18). Following tumor initiation, these mice establish 4- to 6-mm diameter tumors, with nearly complete tumor penetrance within 17 to 21 days (Fig. 7A). To treat these lesions, we created a CAR directed against EGFRvIII by linking a single-chain antibody to a synthetic receptor skeleton comprised of the CD8 hinge, the CD28 transmembrane and signaling domains, and CD3ζ signaling domains (Fig. 7B). We transduced murine T cells using recombinant retroviruses expressing these EGFRvIII-specific CARs, which enabled them to recognize and lyse brain tumor cells isolated from PDGF-EGFRvIII tumors (Fig. 7C). We then treated mice with these CAR-expressing T cells, either as monotherapy or followed by 4 doses of PI-3065/7DW8-5 liposomes (Fig. 7D). In order to track and quantify the in vivo migration of these cells in relation to brain tumors, we included vectors encoding click beetle luciferase in the plasmid. Results based on this PDGF-EGFRvIII tumor model were similar to those from 4T1 breast tumor (Fig. 6): although EGFRvIII-specific CAR-T cells traffic to PDGF-EGFRvIII tumors, they do so very inefficiently and do not persist or expand (Fig. 7E and F), producing only a 4-day survival advantage compared with untreated controls (Fig. 7G). By contrast, CAR-T cells infused after nanoparticle preconditioning underwent significant proliferation at the tumor site (12.1-fold higher peak photon count relative to conventional T-cell therapy on day 6, P < 0.0001, Fig. 7F) and substantially reduced tumor growth (they more than doubled the survival of treated mice, Fig. 7G).
In conclusion, our results demonstrate that nanoparticle-induced remodeling of the tumor microenvironment is safe and can substantially increase the therapeutic activity of CAR-T cells. This adjuvant nanotherapy is most effective when administered repeatedly several weeks prior to T-cell infusion, thereby permitting the delivered immune modulating drugs to first complete their effects on components of the tumor microenvironment.
Solid cancers present formidable barriers against the use of adoptive cell therapies, including functional suppression of T cells and inhibition of their localization. In this project, we developed a nanoreagent that can—as a singular therapeutic—safely decrease the numbers of immunosuppressive cells in the tumor microenvironment and, at the same time, boost key mediators of antitumor immune responses. We demonstrate that repeated administration of these functionalized nanocarriers favorably preconditions the tumor milieu for a limited but defined time period, during which adoptively transferred T cells can have curative effects.
The dosing and frequency required to achieve this transient state will likely require optimization in a clinical setting and may prove to be different for various tumor types and disease stages. In our preclinical test system, we intravenously infused PI-3065/7DW8-5 liposomes every 3 days and induced tumor regression followed by stasis between 10 and 15 days after the first liposome injection (Fig. 4C and D). In a clinical scenario, the same amount of time is required for apheresis, manufacturing, and reinfusion of engineered CAR-T cells into the patient (which typically involves from 2.5 to 3 weeks); our evidence indicates this provides a period long enough to precondition patients with PI-3065/7DW8-5 liposomes. An advantage of the engineered nanoparticles described here is that, after they are configured for specificity and functionality, they can safely be lyophilized for long-term storage. This enables them to be available for “off the shelf” use as early as the day of diagnosis.
We selected 7DW8-5 to improve the immune response to tumors because it has the strongest adjuvant effect measured for a CD1d-binding NKT cell ligand in humans (41). Originally identified by Li and colleagues out of a library of 25 synthetic analogues of α-GalCer, 7DW8-5 has been shown in multiple vaccine trials to induce a significantly higher Th1 response compared with α-GalCer (42, 49, 50). The selective p110δ inhibitor PI-3065 was chosen as a desirable comodulator to accompany 7DW8-5, based on previous work establishing it can delay 4T1 tumor growth and reduce Treg suppression in mice (40). Also, other PI3K-δ inhibitors such as CAL-101 have been reported to decrease the numbers of suppressive Treg and increase the numbers of antitumor T cells (51, 52). In our 4T1 test system, even repeated administration of unencapsulated PI-3065 did not significantly reduce immune suppressor cells (1.2-fold ↓ Tregs, P = 0.38), but required delivery by targeted liposomal nanocarriers (2.7-fold ↓ Tregs, P = 0.0051; Fig. 4B) to be effective.
But beyond the PI-3065/7DW8-5 combination described here, our treatment concept of “releasing immune brakes” while at the same time “stepping on the gas” using a targeted nanocarrier to codeliver synergistic immunomodulatory agents could be applied to other potent combinations of effectors. Promising candidates to remove suppressor cells are, for example: (i) BLZ945, an inhibitor of colony-stimulating factor 1 receptor (CSF1R) that selectively binds to it on TAMs, thereby inhibiting signal transduction pathways mediated by this receptor in TAMs (23, 53); and (ii) PLX3397, a small-molecule receptor tyrosine kinase inhibitor that binds to and inhibits phosphorylation of stem cell factor receptor (c-KIT), CSF1R, and FMS-like tyrosine kinase 3 (FLT3), thereby causing the downmodulation of TAMs (54). Drug candidates that can stimulate an adaptive immune response similar to that resulting from the α-GalCer agonist 7DW8-5 include (i) TLR-based immune adjuvants, such as monophosphoryl lipid A (MPLA), which could easily be incorporated into the liposome envelope (55), or (ii) STING agonists, which have shown dramatic effects against solid tumors in rodent models when administered locally and repeatedly (56, 57).
Clinical implementation of liposome-mediated tumor preconditioning for adoptive T-cell therapy of cancer will relies heavily on the safety of the procedure. Many liposome products are already available for human use (e.g., Doxil, Ambisome, and DepoDur), and various others are undergoing clinical testing, based on their favorable safety profile (along with the robust, rapid, and affordable manufacturing processes that are available; ref. 58). As indicated above, even repeated infusions of empty control nanocarriers did not result in significant macroscopic or microscopic lesions in our test system. Furthermore, liposomes that codelivered a potent α-GalCer agonist induced only mild inflammation (Fig. 5). Notably, these treatments did not significantly affect organ function, animal weight, or their overall well-being, which highlights the safety and clinical applicability of the approach. This is in sharp contrast to systemic administration of α-GalCer (which fails to promote antitumor immune responses; Fig. 4C and D), a treatment known to trigger liver injury and autoimmune hepatitis (44).
In summary, our work describes a new nanotechnology approach that can promote T-cell therapy for solid tumors. Implemented in the clinic as a preconditioning regimen, targeted nanocarriers that transiently reset the tumor microenvironment may improve the success of such treatments. More broadly, we believe that this platform could be explored clinically to maximize the outcomes of other immunotherapy approaches that rely on robust T-cell responses to treat solid tumors (such as personalized cancer vaccines, neoantigen-targeted methods, and bispecific antibody therapies).
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: M.T. Stephan
Development of methodology: F. Zhang, M.T. Stephan
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): F. Zhang, S.B. Stephan, C.I. Ene, T.T. Smith, E.C. Holland, M.T. Stephan
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): F. Zhang, S.B. Stephan, T.T. Smith, M.T. Stephan
Writing, review, and/or revision of the manuscript: M.T. Stephan
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C.I. Ene
Study supervision: M.T. Stephan
This work was supported by the Fred Hutchinson Cancer Research Center's Immunotherapy Initiative, with funds provided by the Bezos Family Foundation (M.T. Stephan), the NCI (RO1CA195718; M.T. Stephan and E.C. Holland), and the Immunotherapy Integrated Research Center (IIRC) at Fred Hutchinson Cancer Research Center (Collaborative Grant Accelerator; M.T. Stephan and E.C. Holland).
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.