T-cell Homing Therapy for Reducing Regulatory T Cells and Preserving Effector T-cell Function in Large Solid Tumors

Many seminal studies have indicated that cytotoxic T cells inhibit the initiation and growth of malignant solid tumors (1, 2). Tumor-infiltrating immune effector cells, particularly CD8⁺ T and natural killer (NK) cells, are associated with a promising prognosis and prolonged survival in both the early and late stages of malignancy (3, 4). Sato and colleagues reported that patients with colorectal cancer who had a high density of infiltrated T cytotoxic and memory cells in their tumors had a much lower tumor recurrence rate and a prolonged survival time (5). The expression of costimulatory receptor CD28 on infiltrated T cells reduces the requirement for T-cell receptor signaling during T-cell activation and rescues exhausted CD8⁺ T cells (6, 7). Tumor-infiltrated NKG2D⁺CD8⁺ T cells recognize and eliminate tumor cells that exhibit NKG2D ligands (8, 9). In the meanwhile, regulatory T cells (Treg) and immune checkpoint regulators (such as PD-1) in the tumor mass create an immunosuppressive environment that hampers the antitumor response from immune effector cells in advanced stages of cancer (10). As the balance between immune activation (CD28 and NKG2D) and immune suppression (Tregs and PD-1) may represent the effect of immune surveillance, the ratios of immune-stimulatory and immune-inhibitory signals in the tumor microenvironment are better predictors of immunotherapy outcomes (5, 11).

The use of T-cell immunotherapy has grown rapidly over the past decade; its success has spread from nonsolid cancers to melanoma and, gradually, to other solid tumors. Nevertheless, the majority of patients with advanced stage of cancers were associated with T-cell exhaustion (12), and only a very limited number of patients with a solid tumor experience a response to T-cell immunotherapy. These patients have been found to have strong intratumoral infiltration of T cells (13); high levels of cytotoxic T-cell effector molecules, such as IFNγ (14); and high ratios of CD8/CD4 T cells in tumors (5). These parameters could serve as markers for evaluating the efficacy of T-cell immunotherapy.

Infusion of autologous tumor-infiltrating lymphocytes (TIL) has been a remarkable breakthrough in the treatment of patients with refractory melanoma. In practice, however, the response rates are only about 50%, including a 10% to 15% complete response rate (15–17). Major challenges in TIL therapy include the reduced tumor penetration ability of TILs after reinfusion and the immunosuppressive tumor microenvironment. In recent clinical trials, patients were infused with 1.5–2 × 10¹¹ TILs to ensure sufficient tumor-targeting TILs and successful tumor remission (16, 17). However, transferring such large numbers of TILs into patients with cancer can cause off-target adverse effects. Approaches are needed that enable TILs to be delivered into large solid tumors more efficiently and therefore reduce the number of T cells that must be infused.

The loss of tumor-homing characteristics during ex vivo culture is one critical reason that TILs cannot reach tumor sites; thus, new therapies use T cells that have been engineered with receptors that recognize tumor antigens (such as CD19); this innovation is known as chimeric antigen receptor (CAR) T cell therapy. CAR T-cell therapy has had substantial success in treating hematologic malignancies, but efficacy in treating large solid tumors is compromised. Because of their heterogeneity, solid tumor cells lack common antigens. Furthermore, host conditioning often prevents T cells from entering the tumor stroma. Caruana and colleagues made a fundamental breakthrough in CAR T-cell therapy by introducing the enzyme heparanase, which degrades the extracellular matrix to allow T cells to penetrate tumors (18). This discovery opens the door to use CAR T-cell therapy for solid tumors.

In our previous work, we found that the combination of systemic IL12 expression plus low-dose doxorubicin facilitated accumulation of endogenous CD8⁺ T cells in solid tumors (up to 6 mm in diameter) in mice in vivo (9). In this study, we hypothesized that IL12 plus doxorubicin would also allow ex vivo–expanded T cells, TILs, and CAR T-cells to penetrate large solid tumors. We found that this treatment not only boosted NKG2D⁺CD8⁺ T-cell infiltration into large solid tumors but also synergistically upregulated the production of the T-cell–attracting chemokines CXCL9 and CXCL10, promoting accumulation of T cells to the tumor microenvironment and enhancing the effector functions of infiltrated T cells by increasing the ratios of stimulation to regulatory factors.

We purchased 6- to 8-week-old BALB/C, C57BL/6J, and NSG mice from The Jackson Laboratory. The mouse care and handling procedures were approved by the Institutional Animal Care and Use Committee of The University of Texas MD Anderson Cancer Center (Houston, TX).

To create the transplant tumor mouse models, Lewis lung carcinoma (LLC) tumor cells (1.5 × 10⁵ per mouse) were subcutaneously inoculated into C57BL/6 and NSG mice. 4T1 tumor cells (1.5 × 10⁵ per mouse) were orthotopically inoculated into BABL/C mice via the mammary fat pad. 4T1 primary tumors were surgically removed on day 18 after inoculation. Lungs from mice in all treatment groups were intratracheally injected with India ink (15% India ink, 85% water, 3 drops NH4OH/100 ml) and then washed in Feket solution (300 mL 70% EtOH, 30 mL 37% formaldehyde, 5 mL glacial acetic acid) overnight. The tumor cell lines LLC and 4T1 were obtained from ATCC. Mel2549 (tumor-derived human melanoma) cells and autologous TILs, which were expanded and maintained in Chantale Bernatchez's laboratory at The University of Texas MD Anderson Cancer Center (Houston, TX; ref. 19), were inoculated subcutaneously (3 × 10⁶ per mouse in 30 μL of PBS) into NSG mice. These mice also received an intravenous injection of autologous TILs (5 × 10⁶) the day after each treatment. Nalm6 human lymphoblasts were mixed with Matrigel (1:1 in PBS) and inoculated subcutaneously (5 × 10⁶ per mouse in 50 μL of solution). CD19-28ζ CAR T cells were expanded and maintained in Gianpietro Dotti's laboratory at The University of North Carolina at Chapel Hill (Chapel Hill, NC; ref. 18). Murine T cells were enriched from splenocytes using an EasySep kit and cultured in RPMI 1640 medium supplemented with 10% FBS and 1% penicillin/streptomycin, and treated with Mycoplasma removal agent from Bio-Rad to ensure that cells were Mycoplasma negative before inoculation.

All the tumor cell lines were characterized by DNA fingerprinting within 6 months of initiating the experiments at The University of Texas MD Anderson Cancer Center's Characterized Cell Line Core Facility (Houston, TX), and treated with Mycoplasma removal agent from Bio-Rad to ensure that cells were Mycoplasma negative before inoculation.

Tumor-bearing mice were treated with control DNA (10 μg/mouse), control DNA plus doxorubicin (1 mg/kg), mouse or human IL12-encoding DNA (10 μg/mouse), or mouse or human IL12-encoding DNA plus doxorubicin, followed by electroporation as described previously (20). Tumor volume was calculated using the formula V = (π/8) × (a × b²), where V = tumor volume in cubic centimeters, a = maximum tumor diameter, and b = diameter at 90°.

Sham siRNA or CXCL9 and CXCL10 siRNA (500 pmol) was injected intratumorally, followed by electroporation twice weekly. The electroporation was performed using the following parameters: two 50-ms pulses of 150 V/cm with a 100-ms interval between pulses. The primer sequences for CXCL9 and CXCL10 were as follows:

• mCXCL9 siRNA: GUUUGUAAGCACGAACUUUA[dT][dT]

• mCXCL10 siRNA: CCAAUAGUAACAAUUGCUA[dT][dT]

Our lab generated both mouse and human IL12 constructs and confirmed by sequence analyses. DNA was prepared by using the endotoxin-free Mega preparation kit from QIAGEN following the manufacturer's instructions. Doxorubicin (Bedford Laboratories) was purchased from the pharmacy at The University of Texas MD Anderson Cancer Center (Houston, TX).

Frozen tumor sections were sequentially fixed with cold acetone, acetone plus chloroform (1:1), and acetone. Tissue sections were blocked with blocking buffer (5% normal horse serum and 1% normal goat serum in PBS) and then incubated with primary antibody overnight at 4°C and secondary antibody for 1 hour at room temperature. For IHC staining, the secondary antibody was biotin-conjugated, the sections were treated with ABC reagent, and the nuclei counterstained with hematoxylin (Sigma-Aldrich). Tumor sections were mounted with Cytoseal mounting medium (Life Technologies). Quantifications of immunohistochemistry were assessed by three low-power fields (LPF; 20×) per slide in a random manner. For immunofluorescence staining, tumor sections were mounted in antifade with DAPI fluorescence mounting medium. Slides were visualized under a Nikon Eclipse Ti fluorescence microscope.

A sample of each tumor from each treatment group was dissociated with Liberase TM (Sigma-Aldrich) enzyme cocktail. Cells were sequentially incubated with primary and secondary antibodies for 30 minutes each at 4°C. Stained cells were analyzed on an Attune acoustic focusing cytometer (Applied Biosystems). Flow cytometry data were analyzed by the FlowJo software program (BD Biosciences).

Alexa Fluor 488–conjugated anti-mouse CD45, anti-human CD45, violet 421–conjugated anti-mouse CD4 and CD8, anti-human CD4 and CD8, and phycoerythrin-conjugated anti-mouse or anti-human CD28, 41BB, NKG2D, CD39, PD-1, LAG3, TIM3, and forkhead box P3 (FOXP3), as well as isotype control antibodies, were purchased from BioLegend. Rabbit anti-mouse and human CD3 and rabbit anti-human CD8 antibodies were purchased from Abcam. Biotin anti-mouse NKG2D, mouse anti-human NKG2D, rabbit anti-human FOXP3, and anti-human T-bet antibodies were purchased from R&D Systems. Anti-mouse FOXP3 antibody was purchased from Thermo Fisher Scientific. Horseradish peroxidase–conjugated anti-rabbit IgG was purchased from Cell Signaling Technology.

Mel2549 and Nalm6 tumor cells (1 × 10⁵) were treated with vehicle or doxorubicin (100 nmol/L) in 600 μL of culture medium for 72 hours. The conditioned media were then collected and placed on the bottom chambers of a Boyden transwell chamber system. Human TILs or CAR T cells (1 × 10⁴) were labeled with violet tracker (Invitrogen), resuspended in 100 μL of chemotaxis buffer (RPMI1640/0.5% BSA), and placed in the top chambers with 5-μm pores (Corning) for 1.5 hours at 37°C. Cells that migrated to the bottom chamber were collected, washed once with PBS, and counted by FACS.

Tumor lysates were collected from tumor-bearing mice that underwent the indicated treatments at the indicated time points. The levels of IL12 and IFNγ were measured by using ELISA Ready-SET-Go! kits (eBioscience). The levels of human CXCL9 and CXCL10 in xenograft tumors were measured using ELISA kits (LSBio).

RNA was isolated from tumors with TRIzol reagent (Invitrogen). Residual genomic DNA was removed from total RNA using the TURBO DNA-free kit (Life Technologies).

Two micrograms of RNA were used for cDNA synthesis with the High-Capacity RNA-to-cDNA kit (Life Technologies). The relative gene expression levels were determined by using the RT-qPCR and the SYBR Green labeling method in a StepOnePlus real-time PCR system (Life Technologies). The reaction contained 2 μL of cDNA, 12.5 μL of SYBR Green PCR master mix (Life Technologies), and 200 μmol/L primer in a total volume of 25 μL. The PCR cycling conditions were as follows: 40 cycles of 15 seconds at 95°C and 60 seconds at 60°C. All samples were run in duplicate. The C_t value of each sample was acquired, and the relative level of gene expression was calculated using the ΔC_t method, which was normalized to the endogenous control, GAPDH. Data are expressed as n-fold relative to the control. The primer sequences for human genes were as follows:

• mCXCL9: F: AGCAGTGTGGAGTTCGAGGAA, R: GGTGCTGATGCAGGAGCAT

• mCXCL10: F: GACGGTCCGCTGCAACTG, R: GGTGCTGATGCAGGAGCAT

• mGAPDH: F: CCAGCCTCGTCCCGTAGAC, R: CGCCCAATACGGCCAAA

• hCXCL9: F: AGGTCAGCCAAAAGAAAAAG, R: TGAAGTGGTCTCTTATGTAGTC

• hCXCL10: F: AGCTCTACTGAGGTGCTATGT, R: GTACCCTTGGAAGATGGGAAAG

• hGAPDH: F: ATGGAAATCCCATCACCATCTT, R: CATCGCCCCACTTGATTTTG

The directly measured outcomes were analyzed by using the two-sided Student t test to compare two treatment groups or one-way ANOVA to compare more than two treatment groups. The statistical significance of each comparison was determined using GraphPad software. A P value < 0.05 indicated statistical significance. *, P < 0.05; **, P < 0.01; ***, P < 0.005; ****, P < 0.001.

Systemic immune therapy in mice has generally been effective only in small tumors (<5 mm in diameter) because the infused T cells do not penetrate into large tumors. Our previous discovery that treatment with a combination of IL12 plus doxorubicin accumulated endogenous NKG2D⁺CD8⁺ T cells into small solid tumors (up to 6 mm in diameter; ref. 9) prompted us to test whether this same treatment enriches exogenous NKG2D⁺ T cells in large solid tumors. We inoculated C57BL/6 and immunodeficient Nod/SCID/gamma (NSG) mice with LLC murine lung cancer cells to establish solid tumors and intentionally waited until the tumors reached 8 to 10 mm in diameter before beginning treatment with control DNA, mouse IL12 plasmid DNA, doxorubicin, or mouse IL12 plasmid DNA plus doxorubicin using a well-established method. This treatment was followed by adoptive intravenous transfer of exogenous T lymphocytes isolated from the tumor-bearing immunocompetent C57BL/6 mice (21). The treatments were repeated 1 week later. Both IL12 and doxorubicin were administered systematically to treat inaccessible tumors such as metastatic tumors. Tumors were collected 4 days after the second treatment to evaluate T-lymphocyte penetration of tumors. Doxorubicin alone greatly increased the numbers of CD4⁺ T lymphocytes in periphery tumors compared with the control DNA treatment (Fig. 1A; Supplementary Fig. S1A), whereas IL12 plus doxorubicin allowed much greater penetration of CD8⁺ T lymphocytes into tumors (Fig. 1B).

These results were supported by immunofluorescence staining of tumor sections with CD8 (T-cell marker) and NKG2D (activating receptor) showing that large numbers of NKG2D⁺CD8⁺ T cells were localized in tumors after treatment with IL12 plus doxorubicin, whereas only a few NKG2D⁻CD8⁺ T cells were found in tumors of mice in the control groups treated with doxorubicin, IL12 DNA alone, or control DNA (Fig. 1C). Most of the CD8⁺ T cells recruited into tumors by IL12 plus doxorubicin treatment were NKG2D⁺CD8⁺ T cells (Fig. 1C). These results agree with the reported accumulation of endogenous NGK2D⁺CD8⁺ T cells in solid tumors induced by IL12 plus doxorubicin, suggesting that this treatment also promoted infiltration of the exogenous murine NKG2D⁺CD8⁺ T-lymphocyte subpopulation in solid tumors.

Next, we sought to assess whether our systemic therapeutic approach could delay the metastatic diseases, the primary cause of cancer-related death. Metastatic mouse breast carcinoma cells 4T1 were orthotopically inoculated into BALB/C mice, and the primary tumors were surgically removed on day 18 after inoculation to establish lung metastases. The tumor-bearing mice were treated with the same four groups of treatments as the LLC mice, followed by infusion of T cells that were isolated from 4T1 tumor–bearing BALB/C mice. Treatment with IL12 plus doxorubicin followed by T-cell infusion not only inhibited the primary tumor progression but significantly prolonged the survival time of 4T1 tumor–bearing mice, with 2 of 6 long-term survivors (Fig. 1D). Also, this treatment dramatically reduced the numbers and sizes of lung metastases (Fig. 1E). Mice treated with IL12 plus doxorubicin plus adoptive T-cell transfer showed much smaller pulmonary tumor nodules (less than 100 μm in diameter) than mice in other treatment groups (Fig. 1F). Intriguingly, a few NKG2D⁻CD8⁺ T cells accumulated in lung metastatic tumors after treatment with control DNA or control DNA plus doxorubicin, whereas small numbers of NKG2D⁺CD8⁺ T cells infiltrated into the periphery of lung nodules in mice treated with IL12. Strikingly, IL12 plus doxorubicin led to influx of NKG2D⁺CD8⁺ T cells throughout the tumor nodules, suggesting that this treatment facilitates effector T-cell penetration into metastatic tumors (Fig. 1G).

Although infusion of TILs into patients with melanoma can result in tumor remission, about 50% of patients experience little response to the treatment. The capacity of TILs to penetrate a tumor is often correlated with the outcome of TIL transfer. To determine whether the systemic administration of IL12 plus doxorubicin boosts the penetration of autologous human melanoma–specific TILs as it did in mouse tumors, xenograft tumors were initiated in NSG mice via injection of patient-derived Mel2549 melanoma cells; the tumor-bearing mice were then divided into the same four treatment groups: control DNA, control DNA plus doxorubicin, human IL12 DNA, human IL12 DNA plus doxorubicin. Ex vivo–expanded autologous TILs were transferred intravenously as adoptive therapy 1 day after each treatment. Because TILs often lose their tumor-homing ability after ex vivo expansion and stimulation, most studies infuse a large number (1 × 10⁷) of T cells, but we transferred only 5 × 10⁶ T cells, with the assumption that fewer T cells would be required because of their increased penetration into the solid tumors after IL12 plus doxorubicin treatment. We detected CD4⁺ T cells in the periphery tumor areas after control groups (control DNA, control DNA plus doxorubicin, and IL12 DNA) of treatments (Supplementary Fig. S1B) but rarely any CD4⁺ T cells in tumors after IL12 plus doxorubicin treatment, suggesting that CD4⁺ T cells may not play a critical role in our treatments. Although doxorubicin alone slightly increased the TIL infiltration into tumors, the absolute numbers of infiltrated CD8⁺ TILs did not differ significantly between the group treated with IL12 DNA and that treated with control DNA. In contrast, IL12 plus doxorubicin remarkably increased the accumulation of infused CD8⁺ TILs in the large melanoma tumors (Fig. 2A).

To validate the distribution of infused TILs in tumors, we examined tumor sections from the peritumor and intratumor areas. All control groups showed TIL accumulation only in the peritumoral areas. In contrast, IL12 plus doxorubicin led to 6-fold increase of TIL infiltration into both peritumoral and intratumoral areas (Fig. 2B). Significantly, only treatment with IL12 plus doxorubicin increased NKG2D⁺CD8⁺ TIL cell numbers in the melanoma tumors (Fig. 2C).

As infused T cells penetrated and distributed into large solid tumors after IL12 plus doxorubicin treatment, we sought to determine the antitumor efficacy of this combinational therapy. Among the four treatments, only IL12 plus doxorubicin significantly hampered the progression of melanoma tumors (Fig. 2D). This inhibition of tumor progression also significantly prolonged survival; 2 of the 5 mice that received IL12 plus doxorubicin had particularly long-term survival (Fig. 2E). In contrast, the control groups (control DNA, doxorubicin, or IL12 DNA) experienced no substantial improvement in tumor volume or survival duration (Fig. 2D and E).

CD19 CAR-T-cell therapy is known to be successful in the treatment of hematologic malignancies in which T cells target CD19 antigen on malignant B cells. However, CAR T-cell therapy has so far been largely ineffective in treating large solid tumors. To test the hypothesis that IL12 plus doxorubicin can boost CAR T-cell penetration into large solid tumors, we inoculated large numbers of CD19⁺ Nalm6 cells (5 × 10⁶) mixed with Matrigel subcutaneously into immunodeficient NSG mice to generate solid tumors. CD19 28ζ CAR T cells were delivered using the same treatment as the Mel2549 tumor model and T-cell transfer strategy used in the other models until solid CD19⁺ Nalm6 tumors reached 6 to 8 mm in diameter. The accumulation of CD4⁺ T cells in the periphery tumor areas were detected after control groups of treatments and was substantially reduced in the tumors treated with IL12 plus doxorubicin (Supplementary Fig. S1C). The accumulation of CD8⁺ T cells in large solid tumors was assessed via flow cytometry after two treatment administrations. IL12 plus doxorubicin (but not the other treatments) resulted in infiltration of a large number of CD8⁺ T cells into the solid tumors (Fig. 3A).

To gain insight into the localization of infiltrated CD8⁺ CAR T cells, we dissected the center area of each tumor (>10 mm in diameter) for whole slide scanning. Small numbers of CD8⁺ CAR T cells accumulated in the peritumoral areas following all the control treatments. In striking contrast, large numbers (more than 20-fold increase) of CD8⁺ CAR T cells accumulated in the central tumor areas of mice receiving systemic IL12 plus doxorubicin treatment (Fig. 3B). Dual staining by human CD8 and NKG2D on tumor sections (Fig. 3C) confirmed that IL12 plus doxorubicin allowed significantly greater numbers of NKG2D⁺CD8⁺ T cells to penetrate the tumor centers than the other treatments. This dramatic elevation of intratumoral infiltration by NKG2D⁺CD8⁺ CAR T cells after IL12 plus doxorubicin treatment reduced tumor volume and tumor weight by more than 5-fold, reflecting significantly enhanced antitumor efficacy (Fig. 3D and E).

The immunosuppressive tumor microenvironment contains a network of factors, including intrinsic inhibitory receptors and extrinsic regulatory immune cells, that may impede the effector functions of antitumor cytotoxic T cells. The expression levels of inhibitory receptors reflect the status of infused T cells and are often altered after infusion in response to cytokines and the tumor microenvironment. The patterns of costimulatory and coinhibitory receptors may characterize the persistence of infiltrated T cells. To test the T-cell persistence in a mouse LLC tumor model, we primed splenic T cells from LLC tumor–bearing immunocompetent mice with plate-bound CD3/CD28 antibodies and IL2 for 24 hours before infusion and assessed expression of the costimulatory receptor CD28, the exhaustion marker LAG3, and the checkpoint regulator PD-1 (Fig. 4A). These T cells were then infused into LLC tumor–bearing NSG mice following two administrations of control or systemic IL12 plus doxorubicin treatment. Only IL12 plus doxorubicin treatment maintained the expression levels of costimulatory and coinhibitory receptors in the infiltrated T cells (Fig. 4B). Along with lower numbers of infiltrated T cells, we observed significant reductions in CD28 expression and substantial upregulation of LAG3 and PD-1 in the control groups, suggesting that the infused T cells became exhausted and lost their effector functions. Intriguingly, the number of infiltrated CD8⁺ T cells, but not CD4⁺ T cells, exhibited a strong positive correlation with the ratios of CD28 expression to LAG3 and PD-1 expression (Supplementary Fig. S2), showing that the infused CD8⁺ T cells with higher ratios of costimulatory/coinhibitory receptors (S/I) were associated with high persistence states in large solid tumors.

Given that the enrichment of Tregs in tumors often compromises the efficacy of T-cell adoptive transfer (21–23), we stained LLC tumor sections with FOXP3 antibody to quantify Treg infiltration in tumors (Fig. 4C). Tumors from control groups (control DNA, control DNA plus doxorubicin, or IL12 DNA alone) showed notable Treg infiltration in clusters. In contrast, treatment with IL12 plus doxorubicin remarkably reduced the accumulation of Tregs in large solid tumors (Fig. 4C), suggesting that this treatment inhibited the infiltration of suppressive immune cells so as to create an environment “friendly” to effector immune cells.

These encouraging results led us to determine whether the same highly persistent CD8⁺ T cells accumulated in human xenograft large solid tumors treated with IL12 plus doxorubicin. CAR expression on CD4⁺ and CD8⁺ T cells was confirmed on CD19 28ζ CAR T cells before infusion (Fig. 5A). The expression levels of costimulatory receptors CD28, 41BB, and NKG2D, as well as inhibitory receptors CD39, PD-1, LAG3, and TIM3, were assessed on expanded CAR T cells before and after infusions plus the indicated treatments (Fig. 5B and C). On the basis of the median fluorescence intensity of each receptor on T cells, we created two heatmaps; these maps showed that after control treatments, infiltrated T cells displayed extremely low levels of costimulatory receptors CD28 and NKG2D (Fig. 5B) but increased the levels of the coinhibitory receptor LAG3 (Fig. 5C). In striking contrast, IL12 plus doxorubicin induced sustained expression of CD28 and NKG2D, with some suppression of PD1 and LAG3, in the tumor-infiltrating CD8⁺ T cells (Fig. 5B and C). The levels of these immune cell index markers suggest that IL12 plus doxorubicin treatment enhanced CD8⁺ T-cell effector function and prevented exhaustion of the infiltrated CD8⁺ T cells. We further converted the two heatmaps to better assess the S/I ratios on CAR-T cells in vitro and in vivo (Fig. 5D), and found that the control treatments dramatically reduced the S/I ratios on tumor-infiltrated T cells in vivo, whereas IL12 plus doxorubicin maintained the S/I ratios on tumor-infiltrated CD4⁺ and CD8⁺ T cells as high as that on CAR T cells before infusion.

Because it is well accepted that the Tbox transcriptional factor TBX21 (T-bet) plays a very crucial role in T-cell development and functions, serving as a marker for effector T cells, we stained tumor sections from indicated groups with anti–T-bet antibody. T-bet⁺ cells were almost absent in the tumors from the control groups but were present in significantly greater numbers in the tumors from mice treated with IL12 plus doxorubicin (Fig. 5E). Just as in LLC tumors (Fig. 4C), the infiltration of Tregs in Nalm6 tumors was dramatically reduced by treatment with IL12 plus doxorubicin (Fig. 5F). Thus, our findings reveal that treatment with IL12 plus doxorubicin produced the most powerful cytotoxic pattern of infiltrated T cells.

Recent reports indicate that a variety of chemotherapeutic agents stimulate tumor cells to release T-cell attractants CXCL9, CXC10, and CCL5 (22, 23). We therefore first tested whether doxorubicin (100 nmol/L) induces these chemokines in tumor cells in vitro. qPCR analysis showed a remarkable induction of CXCL9 and CXCL10 in LLC cells and Mel2549 cells treated with doxorubicin (Supplementary Fig. S3, data not shown). In addition, the human CAR T cells we infused into mice expressed high levels of CXCR3, the receptor for CXCL9 and CXCL10 (Fig. 6A).

In theory, the chemokine-producing tumor cells should be able to promote T-cell migration. To test this hypothesis, conditioned medium from Mel2549 or Nalm6 tumor cells treated with doxorubicin was placed at the bottom of Boyden transwell chambers. Violet cell tracker–labeled human TILs or CAR T cells were cultured in the top chambers for 1.5 hours. The numbers of T cells that migrated to the bottom chambers containing doxorubicin conditioned medium were significantly higher (5-fold for Mel2549 and 3-fold for Nalm6) than those that migrated to the bottom chambers containing vehicle control medium (Fig. 6B).

Nevertheless, in our study, treatment with low-dose (1 mg/kg) systemic doxorubicin alone was unsuccessful in attracting significant numbers of T cells to large solid tumors (Figs. 1–3), which led us to hypothesize that although doxorubicin or IL12 alone may induce the T-cell–attractant chemokines to some extent, only IL12 plus doxorubicin makes tumor cells produce adequate chemokines to massively attract effector T cells to large solid tumors. Although IL12 DNA was delivered systemically through intramuscular injection, it induced IL12 production in tumors (Fig. 6C), which stimulates tumor-infiltrated T cells to secrete the effector molecule IFNγ (Fig. 6D). Treatment with IL12 plus doxorubicin greatly augmented the intensity of infiltrated T cells, which in turn dramatically increased the expression of IFNγ in tumors (Fig. 6D). IFNγ is a well-known stimulator of chemokines CXCL9, CXCL10, and CXCL11, resulting in the synergistic induction of chemokines after treatment with IL12 plus doxorubicin. Moreover, IFNγ may initiate positive feedback to stimulate T-cell activation. In line with these data, the expression profile of T-cell–associated chemokines in mouse and human xenograft tumors after the indicated treatments demonstrated that the levels of CXCL9 and CXCL10 mRNA transcripts were increased in tumors that had been treated with doxorubicin or IL12 DNA alone, and that expression of these chemokines was further boosted in tumors from mice treated with both IL12 and doxorubicin (Supplementary Fig. S4). Consistent with induction at the mRNA level, ELISA results showed that the secretion of CXCL9 and CXCL10 in Mel2549 and Nalm6 tumors was substantially induced after treatment with IL12 plus doxorubicin (Fig. 6E). Strikingly, depletion of these chemokines from tumors impaired the recruitment of CD8⁺ T cells into tumors from mice treated with both IL12 plus doxorubicin (Fig. 6F and G), suggesting that the mechanism regulating CXCL9 and CXCL10 upregulation accounts for the induction of exogenous CD8⁺ T-cell accumulation in large solid tumors by IL12 plus doxorubicin.

We previously reported that IL12 plus doxorubicin promoted endogenous T-cell infiltration into mouse tumors (9), and others found that IL12 plus cyclophosphamide induced macrophage- and T-cell–dependent tumor eradication only in mouse fibrosarcomas, not in mouse melanoma, lung, pancreatic, or colon cancers (24–27). None of these studies showed tumor regression in human tumor models, nor did in-depth characterizations of tumor-infiltrated T cells after treatment. Significantly, our study showed that treatment with IL12 plus doxorubicin successfully enriched not only murine T cells but also clinically relevant human T cells (TILs and CAR T cells) in large solid tumors, both in peritumoral areas (<3 mm from tumor margin) and in the centers of tumors 10 mm in diameter. Furthermore, this same treatment yielded high ratios of costimulatory/coinhibitory receptors in the tumors and decreased the numbers of Tregs, creating a tumor microenvironment friendly to infiltrated CD8⁺ T cells. The infiltration of highly active NKG2D⁺CD8⁺ T cells was associated with tumor remission and significantly extended survival in murine and human xenograft tumor models with large tumors (Figs. 1–3).

Although the infiltration of tumors by T lymphocytes, especially CD8⁺ cytotoxic T cells and memory T cells, has been reported to be associated with positive outcomes in terms of low relapse rates and prolonged overall survival in patients with melanoma (28, 29) or lung (30, 31), ovarian (4, 5), breast (32, 33), or colorectal cancer (34, 35), the localization of infiltrated CD8⁺ T cells, which is part of the “immunoscore,” is often neglected. In fact, the localization of infiltrated CD8⁺ T cells affects the outcomes of T-cell therapy in many cancers. Pages and colleagues observed that the presence of infiltrated CD8⁺ T cells in both the centers and invasive margins of colorectal tumors served as a marker of a better outcome (35). Sato and colleagues showed prolonged survival with CD8⁺ T cells localized in the intraepithelial, instead of the tumor stroma, of ovarian tumors (5). In melanoma metastases, the presence of CD8⁺ T cells throughout the tumor was correlated with prolonged survival (36). A recent report compared the prognostic outcomes of patients with metastatic melanoma with intratumoral versus peritumoral T-cell infiltration into the primary tumors and found that the high ratio of intratumoral to peritumoral values was strongly correlated with longer survival duration after dendritic cell therapy (13). Therefore, we studied not only the numbers of infiltrated CD8⁺ T cells but also their distribution. Treatment with IL12 or doxorubicin alone attracted CD8⁺ T cells to the margins of tumors. Only the combination of IL12 plus doxorubicin recruited CD8⁺ T cells to accumulate in much deeper areas of the tumors, reaching the center of tumors as large as 10 mm in diameter. Such deep penetration of CD8⁺ T cells into tumors led to massive tumor cell destruction and delayed tumor progression.

Besides the localization of infused T cells, another key parameter of the immune contexture is the ratio of immunostimulatory and immune-inhibitory markers, which reflects the functional orientation of infused T cells. T-cell therapy cannot only augment the infiltration of both cytotoxic T cells and Tregs but also trigger negative feedback, inducing T-cell exhaustion. In particular, long-term ex vivo expansion may result in the loss of costimulatory receptor CD28 on the CD8⁺ T-cell population, a loss associated with chronic stimulation (37, 38). In fact, high numbers of infiltrated T cells are often correlated with high levels of immunosuppressive markers (39, 40), which restrains the outcomes of CD8⁺ T-cell infiltration (41). Therefore, it is critical to maintain the effector functions of infiltrated T cells (42).

Our assessment of T-cell markers showed elevated numbers of stimulatory CD28⁺ and NKG2D⁺ T cells in large solid tumors after treatment with IL12 plus doxorubicin. CD28 signaling may reinvigorate the antitumor effect of the infiltrated CD8⁺ T cells. Besides antigen-specific cytotoxicity, the deep tumor-infiltrated NKG2D⁺CD8⁺ T cells also recognize NKG2D ligand–expressing tumor cells to induce NKG2D-mediated killing (8, 9). Significantly, the same combination treatment (IL12 plus doxorubicin) is able to restore NKG2D ligand expression in tumors (8), suggesting that this combination therapy could promote engagement between NKG2D⁺CD8⁺ T cells and NKG2D ligands on tumor cells to cause tumor rejection. NKG2D costimulation also facilitates the formation and rescue of memory CD8⁺ T cells (43).

Control treatments upregulated exhaustion markers PD-1 and LAG3 on infiltrated T cells and increased Treg infiltration, whereas IL12 plus doxorubicin prevented the exhausted and suppressive T cells present in tumors. It has been reported that CD28 costimulation augmented CD8⁺ T-cell proliferation in tumors in response to impairment of PD-1 signaling (7). In this study, treatment with IL12 plus doxorubicin not only maintained high levels of CD28 but also suppressed PD-1 expression. Thus, both deep tumor infiltration and cell proliferation may account for the accumulation of great numbers of CD8⁺ T cells in the large solid tumors. Recent studies indicated that inhibitory receptors suppress the expression of IFNγ by T cells and that these T cells fail to proliferate despite antigen stimulation (44, 45). Our treatment with IL12 resulted in a long duration of IFNγ expression in tumors, which overcame the suppressive effect induced by low levels of inhibitory receptors. As a result, the exogenous source of IFNγ from IL12 treatment shifted the balance to favor CD8⁺ T-cell antitumor immunity.

Several studies have highlighted the role of chemotherapy in promoting an antitumor immune response. The most frequently proposed mechanism is that chemotherapy, including anthracyclines, causes immunogenic tumor cell death, and the dead cells provide tumor antigens to stimulate immune responses (46–48). Recent studies revealed that chemotherapy also upregulates the expression of chemokines that attract effector T cells to tumors (49). In particular, Ma and colleagues showed that intratumoral administration of high-dose (2 mmol/L) doxorubicin induced expression of T-cell–attractant chemokines CXCL9, CXCL10, and CCL5 mRNA in mouse sarcoma tumors in vivo and, in turn, recruited CD8⁺ T and γδ T cells to tumors (22). Hong and colleagues claimed that temozolomide and cisplatin boosted the production of T-cell–attracting chemokines, such as CXCL9, CXCL10, CXCL11, and CCL5, in human xenograft tumors in vitro and in vivo (23). Unlike these previous studies, we administered low-dose doxorubicin (1 mg/kg) systemically in our tumor models (20) and showed that such a low dose of doxorubicin induced the chemokines CXCL9 and CXCL10 in tumors, which was more or less expected. Surprisingly, doxorubicin in combination with IL12 significantly boosted this chemokine induction, which became the key mechanism for T-cell penetration.

In fact, IFNγ is produced by immune cells (T cells, NK cells, etc.) upon IL12 stimulation in tumors. As there are very few effector immune cells recruited to tumors after IL12 DNA–alone treatment (Figs. 1–3), despite enhanced IL12 protein accumulation in tumors (Fig. 6C), it failed to boost IFNγ, a potent inducer of T-cell–attracting chemokines (Fig. 6D; ref. 21). In contrast, in our combination therapy, doxorubicin allowed effector T cells to penetrate into tumors where, in response to accumulated IL12 (Fig. 6C), IFNγ production was greatly upregulated (Fig. 6D). Subsequently, IFNγ synergized with doxorubicin to enhance chemokine production, and chemokines, in turn, recruit more effector T cells to amplify the antitumor immune response.

The correlation between the levels of T-cell–attracting chemokines, the numbers of infiltrating T cells, and antitumor efficacy indicated that only IL12 plus doxorubicin treatment induced production of CXCL9 and CXCL10 at levels sufficient to recruit the great numbers of CD8⁺ T cells into large solid tumors needed to obstruct tumor development. However, the source of the chemokines was not limited to tumor cells, because a higher induction of CXCL9 and CXCL10 was observed in tumors from C57BL/6 mice after doxorubicin treatment (data not shown). In fact, T-cell–attracting chemokines are also secreted by macrophages, endothelial cells, and other stromal cells (50), and it remains unclear how much other cell types contribute to this induction of chemokines.

Unlike previously published strategies in which T cells had to be engineered to gain a tumor-homing ability, our approach modified the tumor microenvironment to favor T-cell accumulation in large solid tumors. Nevertheless, doxorubicin alone was inadequate at inducing the persistent infiltration of effector T cells. However, doxorubicin plus IL12 promoted the production of IFNγ, which further boosted CXCL9 and CXCL10 expression and stimulated the activation of infiltrated effector NKG2D⁺CD8⁺ T cells. These findings suggest that TIL and CAR T-cell therapy may be extended beyond hematologic malignancies and melanoma to a broad range of advanced large solid tumors.

G. Dotti reports receiving commercial research grants from Bellicum Pharmaceutical, Bluebird Bio, and Cell Medica. No potential conflicts of interest were disclosed by the other authors.

Conception and design: J. Hu, S. Li

Development of methodology: J. Hu, X. Xia, P. Hwu, G. Dotti

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J. Hu, C. Sun

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. Hu, P. Hwu, S. Li

Writing, review, and/or revision of the manuscript: J. Hu, C. Sun, P. Hwu, S. Li

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C. Bernatchez, G. Dotti

Study supervision: S. Li

The authors thank Ms. Kathryn Hale for editing this manuscript. This study was supported by NIH grant R01 CA200574 and The University of Texas MD Anderson's Genetically Engineered Mouse Facility, which is partially supported by the NIH/NCI under award number P30 CA016672.

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.