Antitumor CD4⁺ T Helper 1 Cells Target and Control the Outgrowth of Disseminated Cancer Cells Open Access

Metastatic spread in patients with breast cancer is the major driver of cancer-related death for these patients. It is widely believed that the dissemination of cancer cells from a clinically latent stage of invasive breast tumors manifests metastasis (1, 2). However, a large cohort of breast cancer studies have demonstrated that preinvasive and primary tumors also release disseminated cancer cells (DCC) as an early hidden event, potentially forming metastases following successful treatment of the primary tumor (3–5). The presence of DCCs in the bone marrow (BM) of patients with early-stage disease is a predictor of subsequent local recurrence and the development of metastases (5). Studies from spontaneous mouse breast cancer models have demonstrated that early DCCs can reach distant organs and then proliferate to form metastases (6–8). DCCs remain dormant for prolonged periods of time before developing clinically apparent metastasis in patients (9). DCCs express gene signatures that differ from circulating tumor cells (CTC; refs. 10, 11). The heterogeneity and stemness properties of DCCs can promote their resistance to adjuvant or neoadjuvant therapies and may cause cancer recurrence and incurable metastasis (9, 10).

An immunosuppressive tumor microenvironment in the target organs can permit DCCs to evade immune surveillance, render dormancy, and then proliferate to form metastases (9, 12, 13). A granulocytic subset of myeloid-derived suppressor cells has been shown to suppress antitumor immunity and promote metastasis of DCCs (14). Similarly, macrophages have been reported to induce early cancer cell dissemination and DCC-mediated metastasis development (15). Activated hepatic stellate cells have been observed to control NK cell–mediated dormancy of DCCs and DCC-driven liver metastasis (16). In addition, neutrophils can suppress NK cell–mediated clearance of DCCs and enhance metastasis formation (17). Neutrophil extracellular traps induced by sustained lung inflammation can awaken dormant DCCs to grow into an aggressive metastasis (18). Neutrophils can promote metastasis initiation in the lung by triggering the tumorigenic potential of DCCs (19). Intratumoral priming of CD4⁺ T helper 1 (Th1) cells during the early stage of tumor progression is critical for optimal antitumor immunity (20), and these cells reinforce targeted therapy to completely eradicate tumors (21). Nevertheless, the role of CD4⁺ Th1 cells on DCCs is unknown, and this study addresses whether these cells can modulate DCCs.

The BALB-HER2/neu transgenic (BALB-neuT) mouse strain was a gift from Dr. Shari Pilon-Thomas, Moffitt Cancer Center. The breeding pair of BALB-neuT mice was obtained from Dr. Federica Cavallo, University of Torino. Mice at the age of 3 to 4 weeks were screened for hemizygosity (neuT⁺/neuT⁻), and only the positive littermates were used for all the experiments. Female BALB/c mice (RRID:MGI:2683685) and C57BL/6 mice (RRID:IMSR_JAX:000664) were purchased from Charles River Laboratories at 6 to 8 weeks of age. The NOD/SCID-γ^−/− (NSG) immunodeficient mouse strain was a kind gift from Dr. Shari Pilon-Thomas, Moffitt Cancer Center. B6.129S7-Ifng^tm1Ts/J IFN-γ knockout mice and B6.129S2-Cd4^tm1Mak/J CD4 knockout mice were purchased from The Jackson Laboratory. All mice were housed at the Animal Research Facility of the Moffitt Cancer Center. All animal studies were reviewed and approved by the Institutional Animal Care and Use Committee at the University of South Florida (protocol ID: A4100-01). All experiments were conducted in accordance with the NIH guidelines and recommendations for the care and use of laboratory animals. All efforts were made to minimize the suffering of the experimental mice.

The mouse TUBO cell line (a kind gift from Dr. Wei Zen Wei, Wayne State University) was cloned from a spontaneous primary mammary carcinoma of BALB-neuT mice (obtained in 2018). The mouse triple-negative breast cancer (TNBC) cell line 4T1 was purchased from ATCC in 2018 (CRL-2539, RRID:CVCL_0125). JIMT-1, a human breast cancer cell line, was obtained from the DSMZ in 2018 (Cat. No. ACC589). The human breast cancer cell lines BT474 (Cat. No. HTB20) and HCC1954 (Cat. No. CRL-2338) were purchased from ATCC in 2018. The mouse B16F10 melanoma cell line (RRID:CVCL_0159) was a kind gift from Dr. Shari Pilon-Thomas, Moffitt Cancer Center, in 2021. All cell lines used in the study tested negative for mycoplasma contamination using a mycoplasma kit (PlasmoTest, Cat. No. rep-pt1, InvivoGen). All cell lines were authenticated by the vendors and providers and were not listed as misidentified cell lines in the International Cell Line Authentication Committee database. Mammary glands with early tumor lesions were collected from BALB-neuT mice between 14 and 16 weeks, and single-cell suspensions were prepared as previously described (21). These early mammary tumor lesion cells were also used for in vitro experiments. All cells were cultured in complete culture medium (CCM) consisting of RPMI 1640 growth medium (Cat. No. MT-10-040-CM, Corning) supplemented with 10% heat-inactivated FBS (Cat. No. MT35010CV, Fisher Scientific), 2 mmol/L L-glutamine (Cat. No. 25005CI, Fisher Scientific), 1 mmol/L sodium pyruvate (Cat. No. MT-25-000-C1, Corning), 0.1 mmol/L nonessential amino acids (Cat. No. 25-025-CI, Corning), 100 U/mL penicillin and 100 mg/mL streptomycin (Cat. No. MT-30-002-CI, Corning), 50 mg/mL gentamicin (Cat. No. 15750-060, Gibco), 0.5 mg/mL fungizone (Cat. No. 15290018, Gibco), and 0.05 mmol/L 2-mercaptoethanol (Cat. No. 21985-023, Invitrogen). Cells were cultured in a humidified incubator with 5% CO₂ at 37°C. Cultured cells with early passages between 1 and 3 were stored in liquid nitrogen. For all experiments, freshly thawed cells were cultured for about 1 week with regular splits as needed to prevent overconfluence. The total number of passages for all cell lines was limited to six passages. Cells were treated with or without IFN-γ (10 ng/mL; Cat. No. 485-MI-100, R&D Systems) for 24 to 72 hours.

Immature dendritic cells (DC) were prepared from naïve female BALB/c mice or C57BL/6 mice at 6 to 8 weeks of age, as previously described (22). For tumor antigen–targeting DC1 preparation, immune DCs were cultured in CCM at 2 × 10⁶ cells/mL in the presence of 50 ng/mL GM-CSF (Cat. No. 415-ML-050, R&D Systems) and 10 ng/mL IL-4 (Cat. No. 404-ML-050, R&D Systems) and incubated in a 5% CO₂ incubator at 37°C overnight. Next, cells were pulsed with tumor antigen–derived immunogenic MHC class II–binding multiepitope peptides at a concentration of 10 μg/mL (pulsed with each peptide individually) and incubated for 4 to 6 hours. Toll-like receptor agonists lipopolysaccharide at 20 ng/mL (Cat. No. L4391-1MG, Sigma-Aldrich) and CpG ODN1826 at 10 ng/mL (Cat. No. NC9685794, Fisher Scientific) were added to the cells and incubated overnight for DC1 polarization. Individual peptide-pulsed DC1 was collected, pooled and washed with PBS, and used for experiments.

BALB-neuT mice were monitored for spontaneous microinvasion in their mammary glands at 8 weeks of age by MRI. The DC1 pulsed with MHC class II–binding HER2 peptides (HER2-DC1) was injected into the intramammary gland facilitated by ultrasound guidance (1 × 10⁶ cells in 50 μL of PBS/mouse). BALB-neuT mice received HER2-DC1 intramammary gland injection into their fifth mammary gland once weekly for 6 weeks. Another group of BALB-neuT mice received subcutaneous delivery of HER2-DC1 once weekly for 6 weeks. DC1 processing, pulsing with MHC class II–binding multiepitope peptides p5 (ELAAWCRWGFLLALLPPGIAG; Cat. No. 4108540, Bachem), p435 (IRGRILHDGAYSLTLQGLGIH; Cat. No. 4108541, Bachem), and p1209 (SPPHPSPAFSPAFDNLYYWDQ; Cat. No. 4108542, Bachem) from the rat HER2/neu oncogene and vaccine preparation was carried out as described above (“Tumor antigen–targeting type 1 conventional DC1 preparation”). Untreated control BALB-neuT mice were administered sterile PBS injected into the mammary gland. The spontaneous tumor growth development was continuously monitored by MRI at various intervals until 24 weeks of age.

MRI was performed using a 7-Tesla horizontal MRI scanner (Bruker BioSpin) with a 35 mm Litzcage coil (Doty Scientific). In preparation for imaging, experimental BALB-neuT mice were anesthetized using 2% isoflurane supplied in an induction chamber along with 1.5-L/minute oxygen delivery and transferred for scanning. BALB-neuT mice were secured in a mouse coil chamber and placed in the scanner, maintaining the same ventilation supply through a nose cone. The respiration rate of the mice was maintained between 40 and 60 breaths per minute while scanning. The body core temperature was maintained at 37°C with magnetic resonance–compatible Small Rodent Heater System (SAII, SA Instruments) and monitored over time. Anatomical T2-weighted coronal images were acquired for each mouse with a TurboRARE sequence. Typical acquisition parameters were as follows: 1.2 mm slice thickness, 75 × 35 mm² field of view), 512 × 256 matrix size, and 19 slices using echo time/repetition time = 4,513/38 ms. MRI and region of interest analysis for the experimental groups were performed in a blinded fashion.

CD4⁺ and CD8⁺ T-cell depletion experiments were performed using anti-mouse CD4 (Cat. No. BP0003-1, clone GK1.5, Bio X Cell, RRID:AB_1107636) and anti-mouse CD8α (Cat. No. BE0061, clone 2.43, Bio X Cell), respectively. BALB-neuT mice at 7 weeks of age were monitored for spontaneous mammary carcinoma by MRI and confirmed to show no evidence of microinvasion in their mammary glands. These mice were then injected with 300 μg of anti-CD4 or anti-CD8 intraperitoneally twice weekly until the experimental endpoint. HER2-DC1 delivery was carried out as described above (“HER2-DC1 intramammary gland delivery in BALB-neuT mice”). Untreated control BALB-neuT mice were injected with rat IgG2b isotype (Cat. No. BP0090, clone LTF-2, Bio X Cell) intraperitoneally. Experimental BALB-neuT mice were examined for mammary carcinoma development by MRI periodically until 24 weeks of age.

BALB/c mice were orthotopically implanted with 4T1 cells (50,000 cells) into the mammary fat pad (MFP), and microscopic tumor growth was verified on day 3 by ultrasound. Briefly, mice were anesthetized using 2% isoflurane with an oxygen supply and monitored for microscopic tumors via the Vevo 2100 ultrasound system (FUJIFILM VisualSonics Inc.) with enhanced abdominal measurement package in the B-mode and 3D mode settings. Mice were also monitored for physiological status such as respiration, blood pressure, body temperature, and electrocardiogram during ultrasound. Next, experimental mice were randomized from different cages and divided into various groups as indicated. Mice bearing microscopic tumors in the MFP were administered intratumoral HER3-DC1 twice a week for 3 weeks. Intratumoral HER3-DC1 injection was performed as previously described (21). HER3-DC1 was prepared by pulsing DC1 with MHC class II–binding HER3 multiepitope peptides as follows: extracellular domain (ECD) p12 (CEVVMGNLEIVLTGH), ECD p81 (SWPPHMHNFSVFSNL), ECD p84 (TTIGGRSLYNRGFSL), and ECD p91 (AGRIYISANRQLCYH) and intracellular domain (ICD) p38 (VADFGVADLLPPDDK), ICD p41 (QLLYSEAKTPIKWMA), ICD p52 (VPDLLEKGERLAQPQ), ICD p86 (GCLASESSEGHVTGS), and ICD p89 (EAELQEKVSMCRSRS; ref. 22). All these MHC class II–binding HER3 multiepitope peptides were purchased from GenScript. Mouse tumors were measured twice weekly using digital calipers. At the endpoint, the experimental mice were sacrificed, and the primary tumors and distant organs (lung and liver) were collected for molecular analysis and spontaneous metastases detection. In another experiment, 4T1 tumor–bearing mice were treated with intratumoral HER3-DC1, subcutaneous HER3-DC1, intratumoral immature HER3-iDC, unpulsed DC1, and intratumoral HER2-DC1 (DC1 targeting irrelevant tumor antigen). These mice were examined for primary tumor growth and spontaneous metastasis.

For the CD4⁺ and CD8⁺ T-cell depletion experiment, mice were injected with anti-CD4 or anti-CD8 intraperitoneally twice a week beginning 3 days prior to 4T1 cell inoculation in the MFP and continuing until the endpoint. The rat IgG2b isotype control antibody was injected into nondepleted mice. For in vivo IFN-γ neutralization, mice were injected intraperitoneally with 200 μg of anti-mouse IFN-γ (Cat. No. BE0055, clone XMG1.2, Bio X Cell) or rat IgG1 isotype control, anti–horseradish peroxidase (Cat. No. EB0088, clone HRPN, Bio X Cell), every third day until the endpoint. The HER3-DC1 treatment was performed as detailed above. Each experiment was performed 3 times.

C57BL/6 mice or IFN-γ knockout mice or CD4 knockout mice were subcutaneously implanted with B16-F10 cells (1 × 10⁶ cells; ref. 23) in the left flank. On day 3, mice bearing microscopic tumors were administered intratumoral HER3-DC1, and these injections continued twice a week for 3 weeks. For the high tumor burden spontaneous metastasis model, intratumoral HER3-DC1 treatment was started after primary tumors reached approximately 70 to 100 mm² in size. Mice were monitored, and tumors were measured periodically. Lungs were collected at the endpoint for spontaneous metastasis detection.

Experimental BALB-neuT mice at week 16 were sacrificed by euthanasia. The mammary glands, positive or negative for tumors, were collected separately and processed into single-cell suspensions as previously described (21, 24). Cells of 1 × 10⁶ were stained with LIVE/DEAD Zombie near IR (Cat. No. 423106, BioLegend) for 30 minutes in the dark at room temperature. Cells were then surface-stained with lymphoid immune cell phenotyping antibodies including anti-CD45 BUV 395 (Cat. No. 564279, clone 30-F11, BD Biosciences), anti-CD45 FITC (Cat. No. 501129405, clone 30-F11, Fisher Scientific) or CD45 PE/Cyanine7 (Cat. No. 103114, clone 30-F11, BioLegend), anti-CD3 APC (Cat. No. 553066, clone 145-2C11, BD Biosciences), anti-CD3 FITC (Cat. No. 553062, clone 145-2C11, BD Biosciences) or anti-CD3 BUV395 (Cat. No. 563565, clone 145-2C11, BD Biosciences), anti-CD4 PerCP-Cy5.5 (Cat. No. 550954, clone RM4-5, BD Biosciences) or anti-CD4 BUV805 (Cat. No. 612900, clone GK1.5, BD Biosciences), anti-CD8 Pacific Blue (Cat. No. 558106, clone 53–6.7, BD Biosciences) or anti-CD8 PerCP-Cy5.5 (Cat. No. 100734, clone 53-6.7, BioLegend), anti-CD62L BUV737 (Cat. No. 612833, clone MEL-14, BD Biosciences), anti-CD44 FITC (Cat. No. 553133, clone IM7, BD Biosciences) or anti-CD44 Pacific Blue (Cat. No. 103020, clone IM7, BioLegend), anti-CD49b (DX-5 pan NK) PE (Cat. No. 108908, Clone DX5, BioLegend) or anti-CD49b (DX-5 pan NK) APC (Cat. No. 560628, clone DX5, BD Biosciences), anti-CD19 PE/Cyanine7 (Cat. No. 115520, clone 6D5, BioLegend) or CD19 BV510 (Cat. No. 115546, clone 6D5, BioLegend), anti-CD183 (CXCR3) PE/Cyanine7 (Cat. No. 126516, clone CXCR3-173, BioLegend), anti-CD197 (CCR7) BV786 (Cat. No. 564355, clone 4B12, BD Biosciences), anti-CD279 (PD1) BV605 (Cat. No. 135220, clone 29F.1A12, BioLegend), anti-LAG3 APC (Cat. No. 125210, clone C9B7W, BioLegend), and CD1d tetramer PE (NIH Tetramer Core Facility) for 30 minutes on ice in the dark and washed with FACS buffer. For intracellular FOXP3 staining, cells were first surface-stained with anti-CD45 PE/Cyanine7, anti-CD3 FITC, and anti-CD4 BUV805. Next, intracellular FOXP3 staining was performed with True-Nuclear Transcription Factor Buffer Set (Cat. No. 424401, BioLegend) following the manufacturer’s protocol and stained for anti-FOXP3 PE (Cat. No. 12-5773-82, clone FJK-16s, Thermo Fisher Scientific). Data were acquired on an LSRII flow cytometer (BD Biosciences) using FACSDiva software version 9.0 (BD Biosciences; RRID:SCR_013311). FACS data were analyzed with FlowJo software version 10.8.1 (RRID:SCR_008520). Gating strategy for identifying various immune effector cells and immune profiling was performed as previously described (21, 22).

CD4⁺ T cells were isolated from tumors and BM of control and intratumoral HER3-DC1–treated 4T1 tumor–bearing mice. Isolated CD4⁺ T cells were then cocultured with HER3-DC1 (DC1 targeting TNBC tumor antigen) or HER2-DC1 (DC1 targeting irrelevant tumor antigen; 10:1 ratio) overnight. eBioscience Protein Transport Inhibitor Cocktail containing brefeldin A and monensin (Cat. No. 00498003, Thermo Fisher Scientific) was added to inhibit intracellular protein transport for 6 hours. Cells were harvested and surface-stained for anti-CD3, anti-CD4, and anti-CD8 as described above. Intracellular IFN-γ staining was performed using the eBioscience Intracellular Fixation & Permeabilization Buffer kit (Cat. No. 88882400, Thermo Fisher Scientific) following the manufacturer’s protocol and stained for anti–IFN-γ PE (Cat. No. 554412, clone XMG1.2, BD Biosciences). Data acquisition was performed as described above (“Immune cell phenotyping by flow cytometry”).

The HER2-DC1–mediated anti-HER2 Th1 response in BALB-neuT mice was examined using the Simple Plex mouse IFN-γ assay kit (Cat. No. SPCKB-MP-001945, Bio-Techne). Spleens and tumor-draining lymph nodes (TDLN) were collected from untreated and HER2-DC1–treated BALB-neuT mice at week 16, and single-cell suspensions were prepared as described previously (21, 24). Splenocytes were stimulated with or without MHC class II–binding rat HER2/neu peptides p5, p435, and p1209 individually for 72 hours. TDLNs were cocultured with or without HER2-DC1 (DC1 individually pulsed with p5, p435, and p1209) at a 1:10 ratio for 72 hours. Following incubation, culture supernatants were centrifuged at 1,000 rpm for 5 minutes. Supernatants were collected and measured for IFN-γ using a Simple Plex mouse IFN-γ assay kit and run on an Ella automated ELISA instrument (Bio-Techne).

The transwell migration assay was performed using Corning 12 mm Transwell with 3-μm pore permeable polycarbonate membrane inserts (Cat. No. 07-200-157, Fisher Scientific). Anti-HER2 CD4⁺ Th1 cells were cocultured with HER2-DC1 or unpulsed DC1 in the presence or absence of IFN-γ neutralizing antibody (Cat. No. BE0055, clone XMG1.2, Bio X Cell) or isotype control antibody (rat IgG1 isotype control, anti–horseradish peroxidase; Cat. No. EB0088, clone HRPN, Bio X Cell) in the top chamber of the transwell plate. The DCCs from control BALB-neuT mice or HER2⁺ TUBO cells were seeded in the bottom chamber of the transwell plate in the presence or absence of IFN-γ neutralizing antibody (Cat. No. BE0055, Bio X Cell) or isotype control antibody (Cat. No. EB0088, Bio X Cell). After 72 hours of coculture, the top chamber was removed, and the bottom chamber containing DCCs or HER2⁺ TUBO cells was assessed for senescence-associated β-galactosidase (SA-β-gal) activity using a cellular senescence assay kit (Cat. No. KAA002, Fisher Scientific) per the manufacturer’s instructions.

The collected tumor tissue, lung, liver, and brain were fixed with 10% formalin for 48 to 72 hours and then transferred to 70% ethanol. Tissue paraffin embedding, slide sectioning, and hematoxylin and eosin staining were performed as previously described (25). The IHC double staining for anti-HER2 (Cat. No. 4290, Cell Signaling Technology) and anti-cytokeratin 8/18 (Cat. No. MA5-32118, Thermo Scientific) or anti-HER3 (Cat. No. 12708, Cell Signaling Technology) and anti-Pan cytokeratin (Cat. No. NB600-579, Novus Biologicals) was performed using the Ventana Discovery XT automated system (Ventana Medical Systems) as per the manufacturer’s instructions. All stained slides were scanned and analyzed using the Leica Aperio AT2 scanner (Leica Biosystems) at the Microscopy Core Facility of the Moffitt Cancer Center.

BM was collected from the femurs and tibiae of experimental BALB-neuT mice at week 16 or experimental BALB/c mice bearing 4T1 tumors or naïve BALB/c mice as described previously (7). Briefly, the BM was flushed with PBS using a 26-G needle, and then the red blood cells were lysed with ammonium-chloride-potassium (ACK) lysis buffer and resuspended in RPMI 1640 (Cat. No. MT-10-040-CM, Corning). Cells were placed on cell culture chambers coated with poly-D-lysine (P6407, Sigma-Aldrich) (Cat. No. 80841, Ibidi) and incubated for 2 hours at 37°C in a CO₂ incubator. The attached cells were fixed using ice-cold 4% paraformaldehyde and permeabilized with 0.2% Triton X. Slides were blocked with 5% BSA to reduce background staining and then incubated with anti-HER2 (Cat. No. BE0277, clone: 7.16.4, Bio X Cell), anti-cytokeratin 8/18 (Cat. No. ab53280, Abcam), and anti-Ki-67 (Cat. No. AF7649, R&D Systems). Goat anti-mouse FITC (Cat. No. NC9352374, Fisher Scientific), goat anti-rabbit AF594 (Cat. No. 8889S, Cell Signaling Technology), and donkey anti-sheep AF647 (Cat. No. ab150179, Abcam) were used as secondary antibodies. Stained slides were mounted in VECTASHIELD antifade mounting medium with DAPI (Cat. No. H-1200-10, Vector Laboratories). Slides were observed using a ZEISS Z2 imager upright fluorescence microscope (Carl Zeiss, AG) attached to an ORCA- Flash4.0 V3 CMOS camera (Hamamatsu Photonics K.K.), X-Cite FIRE LED light source (Excelitas Technologies Corp.), and lenses: 20×/0.7 NA, 40×/1.3 NA, and 63×/1.4 NA. Images were acquired using ZEISS Zen 2 Blue software (Carl Zeiss, AG). For flow cytometry, single-cell suspensions of BM or lungs (1 × 10⁶ cells) were incubated with LIVE/DEAD Zombie near IR as specified above. Cells were stained with anti-CD45 BUV395 (Cat. No. 564279, clone 30-F11, BD Biosciences), anti-HER2 APC (Cat. No. FAB6744A-100UG, R&D Systems) or anti-EpCAM PE (Cat. No. 118206, BioLegend), anti-cytokeratin 8/18 FITC (Cat. No. ab52459, Abcam), and anti-Ki-67 BV786 (Cat. No. 563756, BD Biosciences) or anti-Ki-67 PE (Cat. No. NB110-89717PE, Fisher Scientific) for 30 minutes at 4°C. Cells were run on an LSRII flow cytometer, and the data were analyzed with FlowJo software version 10.8.1 (RRID:SCR_008520). In addition, indirect flow cytometry staining on DCCs was also performed using the primary and secondary antibodies mentioned above.

A pool of BM from experimental BALB-neuT mice (n = 8–10 BALB-neuT mice) was collected, and a single-cell suspension was prepared as specified above. CD45⁺ immune cells were depleted using the EasySep mouse CD45 positive selection kit (Cat. No. 18945, STEMCELL Technologies) following the manufacturer’s instructions. Ep-CAM⁺ DCCs were then isolated from the CD45^– cells using the MojoSort mouse CD326 (Ep-CAM) selection kit (Cat. No. 480142, BioLegend) following the manufacturer’s instructions. The isolated DCCs were cultured in the presence of 30 ng/mL EGF (Cat. No. 315-09, PeproTech) and 20 ng/mL FGF-basic (Cat. No. 450-33, PeproTech) at 37°C in a CO₂ incubator.

An equal number of isolated viable DCCs (2.5 × 10⁵) from control and HER2-DC1–treated BALB-neuT mice between 16 and 18 weeks, in vitro expanded DCCs (2.5 × 10⁵) from control BALB-neuT mice, or TUBO cells (2.5 × 10⁵) were resuspended in PBS, plus an equal volume of Cultrex BME (Cat. No. 3432-005-01, R&D Systems), and mixed with EGF (5 μg; Cat. No. 315-09, PeproTech) and FGF-basic (5 μg; Cat. No. 450-33, PeproTech). Cells were then injected subcutaneously into the NSG mice as previously described (26). Upon appearance of palpable tumors, the tumor growth in NSG mice was measured once a week with a vernier caliper. After monitoring tumor growth for 3 months, mice were euthanized and lung, liver, and brain were collected to examine micrometastases.

Cell lysates or tissue lysates were prepared using RIPA buffer (Cat. No. 20-188, Millipore) containing a protease inhibitor cocktail (Cat. No. P8340-1ML, Sigma-Aldrich) and phosphatase inhibitors (Cat. No. A32957, Pierce). The protein concentration was determined using the Bradford protein assay (Cat. No. 5000006, Bio-Rad). Twenty to forty micrograms of total protein were resolved in 4% to 12% SDS-PAGE gels and transferred onto a PVDF membrane (Cat. No. IPVH00010, Millipore) using the eBlot L1 wet transfer system (GenScript). Subsequently, the membranes were blocked with 5% BSA in TBST and then immunoblotted with primary and secondary antibodies. The membranes were visualized using the Pierce ECL Western Blotting Substrate (Cat. No. 32106, Thermo Fisher Scientific). Bands were visualized and imaged using the Odyssey Fc imaging system (LI-COR Biosciences). The primary antibodies used were anti-HER2 (Cat. No. 2165S, Cell Signaling Technology, or ab131490, Abcam), anti-pHER2 Tyr1221/1222 (Cat. No. 2243S, Cell Signaling Technology), anti-Wnt4 (ab91226, Abcam), anti-NR2F1/COUP-TFI (Cat. No. 6364S, Cell Signaling Technology), anti-Twist (ab50887, Abcam), and anti-β actin (Cat. No. 4967S, Cell Signaling Technology). The secondary antibodies used were goat anti-rabbit IgG HRP (Cat. No. 7074S, Cell Signaling Technology) or goat anti-mouse IgG HRP (Cat. No. 1721011, Bio-Rad).

A mammosphere formation assay was performed as previously described (8). DCCs, TUBO cells, or JIMT-1 cells were passed through a 25-G needle 10 times and filtered through a 70-μm cell strainer. The enriched single cells were plated in Nunclon Sphera 6-Well Plates (Cat. No. 174932, Thermo Fisher Scientific) or Corning Costar ultralow attachment plates (Cat. No. 7200601, Fisher Scientific). Cells were cultured in PromoCell 3D Tumorsphere Medium XF (Cat. No. C-28070, Sigma-Aldrich) and supplemented with 1:50 B27 Plus Supplement (Cat. No. A3582801, Thermo Fisher Scientific) and 20 ng/mL EGF. Cells were then treated with various treatment conditions as specified above. Cells were incubated for 14 days, and mammospheres were counted using a bright-field microscope.

DCCs from BM of experimental BALB-neuT mice or in vitro expanded DCCs treated with various treatment conditions were assessed for SA-β-gal activity using a cellular senescence assay kit (Cat. No. KAA002, Fisher Scientific) per the manufacturer’s instructions. Cells were visualized and imaged using a bright-field microscope. The SA-β-gal–positive (blue) cells or SA-β-gal–negative cells were counted, and the percentage of SA-β-gal–positive cells was determined.

DCCs from the BM of BALB-neuT mice were treated with various treatment conditions for 48 hours as detailed above. Cells were collected using StemPro Accutase Cell Dissociation Reagent (Sigma-Aldrich), resuspended in PBS, and filtered through a 70-μm cell strainer. An ALDEFLUOR assay was performed using the ALDEFLUOR kit (Cat. No. 01700, STEMCELL Technologies) following the manufacturer’s instructions. For CD44⁺CD24⁺ cancer stem–cell staining, cells were incubated with LIVE/DEAD Zombie near IR and stained with anti-CD44 Pacific Blue (Cat. No. 103020, BioLegend) and anti-CD24 PE (Cat. No. 138504, BioLegend). Stained cells were acquired using an LSRII flow cytometer (BD Biosciences), and FACS data were analyzed using FlowJo software (Tree Star).

Total RNA was extracted using the RNeasy Mini Kit (Cat. No. 74104, Qiagen) following the manufacturer’s instructions. RNA was quantitated with the Qubit Fluorometer (Thermo Fisher Scientific) and screened for quality on the Agilent TapeStation 4200 (Agilent Technologies). The samples were then processed for RNA sequencing using the NuGEN Universal RNA-Seq Library Preparation Kit with NuQuant (Tecan Genomics). One hundred nanograms of RNA was used to generate cDNA and a strand-specific library following the manufacturer’s protocol. Quality control steps were performed including TapeStation size assessment and quantification using the KAPA Library Quantification Kit (Roche). The final libraries were normalized, denatured, and sequenced on the Illumina NextSeq 2000 sequencer with the P3-200 cycle reagent kit to generate approximately 100 million 105-base read pairs per sample (Illumina, Inc.).

After the initial quality assessment and adaptor trimming with cutadapt version 1.8.1 (27), the RNA sequencing reads from mice were subjected to various pre- and postalignment QC measures before being mapped against the reference genome mm10 using TopHat 2.0.13 (RRID:SCR_013035; ref. 28). Gene-level quantification was performed using HTSeq 0.6.1 (RRID:SCR_005514; ref. 29) by summing the raw counts of reads aligned to each gene’s region. For human RNA sequencing data, the reads were aligned against the human genome HG19 using STAR-2.5.3a (30). Gene expression was evaluated by calculating the read count at the gene level with RSEM (RRID:SCR_013027) and Gencode gene model version 30. Normalized gene expression was then calculated using the R/Bioconductor package DESeq2 v1.6.3 (RRID:SCR_000154; ref. 31).

Enrichment analyses on expression datasets were performed using gene set enrichment analysis (GSEA; refs. 32–35) and MetaCore (Clarivate Analytics; RRID:SCR_008125). GSEA was performed on ranked gene lists based on their differential gene expression and adjusted P values via the REACTOME database (RRID:SCR_003485; ref. 36). To examine the enrichment of specific gene networks, MetaCore-specific gene ontology analysis was performed on curated gene lists of control and treatment samples for the following cutoffs: twofold change and adjusted P value < 0.05.

Patients newly diagnosed with HER2⁺ ductal carcinoma in situ (DCIS) or early invasive breast cancer (inclusion criteria: prior to any treatment, and exclusion criteria: received treatment or a history of any previous treatment) with a high risk for residual disease or recurrence, as assessed by DCISionRT (PreludeDx), were recruited to participate in this feasibility study at the H. Lee Moffitt Cancer Center and Research Institute. Informed written consent was obtained from all participants (nine patients). The study protocol was reviewed and approved by the Institutional Review Board at the H. Lee Moffitt Cancer Center and Research Institute (protocol ID: MCC21045). BM aspirates were collected from the anterior or posterior superior iliac crest at the time of surgery (prior to any treatment), while patients were under anesthesia, and transferred into EDTA-containing collection tubes. DCC isolation was performed within 24 hours of BM aspiration. Red blood cells were lysed with ACK lysis buffer prior to DCC isolation. EasySep Direct Human CTC Enrichment Kit (Cat. No. 19657, STEMCELL Technologies) was used to deplete all immune cells, hematopoietic cells, platelets, and fibroblasts in the BM. The enriched cells were then processed to isolate epithelial cellular adhesion molecule (EpCAM)⁺ DCCs using the EasySep Human EpCAM Positive Selection Kit II (Cat. No. 17846, STEMCELL Technologies) following the manufacturer’s instructions. Immunofluorescence was then performed using anti-HER2 (Cat. No. BE0277, clone: 7.16.4, Bio X Cell) and anti-cytokeratin 8/18 (Cat. No. 4546S, Cell Signaling Technology; Cat. No. ab53280, Abcam) primary antibodies. Goat anti-mouse FITC (Cat. No. NC9352374, Fisher Scientific) and goat anti-rabbit AF594 (Cat. No. 8889S, Cell Signaling Technology) were used as secondary antibodies. Isolated patient DCCs were also stained for LIVE/DEAD Zombie near IR and anti–Pan-cytokeratin PE (Cat. No. ab52460, Abcam) and analyzed by flow cytometry as detailed above. In addition, cytospin of isolated patient DCCs was stained by Diff-Quik and examined by a cytopathologist.

The isolated patient DCCs were cultured and expanded in RPMI 1640 supplemented with 10% FBS, 30 ng/mL rhEGF (Cat. No. 236-EG-200, R&D Systems), and 20 ng/mL rhFGF-basic (Cat. No. 233-FB-025, R&D Systems) with a 5% CO₂ supply. DCCs were treated with or without IFN-γ (20 ng/mL; Cat. No. 285-IF-100, R&D Systems) for 24 to 72 hours. For the tumorigenicity assay, patient’s viable BM DCCs (30,000 cells; treated with or without IFN-γ) or JMIT-1 (3 × 10⁶) were implanted in NSG mice as detailed above and were monitored for tumor growth. In addition, a mammosphere formation assay, cellular senescence assay, and Western blot for HER2, phospho-HER2, NR2F1, and Wnt4 were performed as mentioned above. For apoptosis detection, cells were stained with Apotracker Green (Cat. No. 427402, BioLegend) following the manufacturer’s instructions. Stained cells were acquired using LSRII flow cytometer (BD Biosciences), and FACS data were analyzed using FlowJo software (Tree Star). RNA sequencing was performed as detailed above.

All data are presented as the mean ± SEM. Statistical analysis was done using GraphPad Prism 9 software (RRID:SCR_002798) and Microsoft Excel (RRID:SCR_016137). The unpaired or paired two-tailed Student t test or one-way ANOVA with Tukey multiple comparisons test was used to calculate the P values. A P value less than 0.05 was considered statistically significant.

All of the data included in this article are available in the manuscript and its supplementary files or from the corresponding author upon request. The RNA sequencing data generated in this study are available in Gene Expression Omnibus (RRID:SCR_005012) at GSE261236, GSE261237, and GSE261959.

We observed the presence of DCCs in the BM of BALB-neuT mice with early lesion mammary carcinoma (Fig. 1A). This was consistent with previous findings (7, 8). We did not detect any DCCs in the BM of naïve BALB/c mice (Fig. 1A). The EpCAM⁺ DCCs from the BM of BALB-neuT mice were isolated utilizing a two-step immunomagnetic enrichment assay, and isolation purity was verified by flow cytometry (Fig. 1B and C). The gene expression profile of DCCs from BALB-neuT mice was analyzed by RNA sequencing and revealed upregulation of various stemness and epithelial-to-mesenchymal transition (EMT) genes and downregulation of cell-cycle genes compared with early lesion tumor cells of BALB-neuT mice and metastatic TUBO cells (Fig. 1D; Supplementary Fig. S1A and S1B). Next, BM aspirates from patients with HER2⁺ DCIS and early invasive breast cancer (prior to any treatment) obtained at the time of surgery were analyzed by two-step immunomagnetic enrichment assays to isolate EpCAM⁺ DCCs (Fig. 1E). These isolated EpCAM⁺ DCCs were examined by a cytopathologist and confirmed for cancer cell phenotype (Fig. 1F). In addition, the presence of DCCs in the BM of patients with HER2⁺ breast cancer and isolation purity were further confirmed by immunofluorescence staining and flow cytometry (Fig. 1G and H). These isolated DCCs were cultured and expanded in appropriate culture conditions (Fig. 1I). The gene expression profile of isolated DCCs from patients with HER2⁺ breast cancer was examined by RNA sequencing. The DCCs of patients with HER2⁺ breast cancer had similar gene expression patterns for cancer stemness, EMT, and cell cycle as we observed in DCCs of BALB-neuT mice when compared with human HER2⁺ breast cancer cells such as JIMT-1 (grade 3 invasive, T2N1M0), BT474 (invasive), and HCC1954 (stage IIA, grade 3 invasive; Fig. 1J; Supplementary Fig. S1C–S1F). GSEA also revealed enrichment of cancer stemness and EMT gene signatures in DCCs of patients with HER2⁺ breast cancer compared with JIMT-1, BT474, and HCC1954 cells (Supplementary Fig. S1G). Taken together, these data indicate that DCCs display a different gene expression profile than primary and metastatic tumor cells in breast cancer.

We investigated the efficacy of intramammary gland delivery of HER2-DC1 in a BALB-neuT mouse model. These mice spontaneously develop microinvasion from 8 to 9 weeks of age, which slowly progresses from preinvasive lesions to primary invasive tumors in all 10 mammary glands and then to distant metastasis (7). cDC1 were pulsed with MHC class II–binding immunogenic HER2/neu peptides to prime a HER2 tumor antigen–specific CD4⁺ Th1 response (21, 37). BALB-neuT mice at 8 to 9 weeks of age (before onset of primary tumors) received weekly ultrasound-guided intramammary gland injection of HER2-DC1 into their fifth mammary gland for 6 weeks. Severe spontaneous mammary carcinoma progression was observed in control BALB-neuT mice and reached endpoint at 22 to 24 weeks of age (Fig. 2A and B; Supplementary Fig. S2A). HER2-DC1 delivery significantly delayed spontaneous tumor growth and reduced disease severity in BALB-neuT mice (Fig. 2A and B; Supplementary Fig. S2B). We observed micrometastases in the lung, liver, and brain of control BALB-neuT mice at 24 weeks of age (Fig. 2C–G). No evidence of micro-metastases was observed in these organs collected from HER2-DC1–treated BALB-neuT mice at 24 weeks of age (Fig. 2C–G).

Next, we examined the level of CD4⁺ T and CD8⁺ T-cell infiltration by flow cytometry. A significant increase in the frequency of CD4⁺ T cells was observed in the mammary glands of HER2-DC1–treated BALB-neuT mice compared with control group at 16 to 18 weeks of age (Fig. 2H). High intratumoral infiltration of CD4⁺ T cells after HER2-DC1 delivery in BALB-neuT mice was further confirmed by IHC (Fig. 2I; Supplementary Fig. S2C). Next, we assessed the infiltration of CD4⁺ T cells and their phenotypic status in HER2-DC1–injected mammary glands and noninjected mammary glands of BALB-neuT mice. Higher infiltration of CD4⁺ T cells, CD4⁺CD44⁺CD62L^– effector memory (EM) cells, and CD4⁺CD44^–CD62L^– effector cells was observed in both the HER2-DC1–injected mammary glands and noninjected mammary glands of BALB-neuT mice, compared with control (Supplementary Fig. S3A). CD4⁺CD44⁺CD62L⁺ central memory (CM) cells were increased only in HER2-DC1–injected mammary glands but not in noninjected mammary glands of BALB-neuT mice (Supplementary Fig. S3A). In addition, a significant increase in the accumulation of CD1d tetramer⁺ invariant NK T (iNKT) cells, NK cells, and B cells was observed in the HER2-DC1–injected mammary glands and noninjected mammary glands of BALB-neuT mice, compared with control (Supplementary Fig. S3B–S3D). The anti-HER2 Th1 immune response was evaluated in both control and HER2-DC1–treated BALB-neuT mice by coculturing TDLNs with HER2-DC1 pulsed with MHC class II HER2 peptides p5, p435, and p1209 individually. We observed significantly increased IFN-γ secretion after coculturing HER2-DC1 with TDLNs from the HER2-DC1–treated BALB-neuT mice, compared with the control group (Fig. 2J), suggesting that anti-HER2 CD4⁺ Th1 cell priming may occur in TDLNs. In addition, restimulation of splenocytes from the HER2-DC1–treated BALB-neuT mice with HER2 peptides p5, p435, and p1209 showed increased IFN-γ secretion compared with control (Fig. 2K).

We also examined the efficacy of tumor antigen–pulsed DC1 delivery on the spontaneous metastatic growth in the 4T1 tumor model. The 4T1 tumor model mimics human TNBC and has been shown to overexpress the HER3 tumor antigen (22). Overexpression of HER3 in TNBC promotes tumor growth, aggressiveness, and metastasis and is associated with poor survival in patients (22). We pulsed cDC1 with MHC class II–binding immunogenic HER3 peptides to prime an anti-HER3 CD4⁺ Th1 response (22). Intratumoral delivery of cDC1 targeting HER3 (HER3-DC1) was able to delay orthotopically implanted 4T1 primary tumor growth in the mammary gland of BALB/c mice (Fig. 2L). HER3-DC1 delivery significantly inhibited spontaneous lung and liver metastasis in these mice compared with untreated control orthotopic 4T1 tumor–bearing mice (Fig. 2M and N; Supplementary Fig. S4A and S4B). Next, we extended our study in a HER3-overexpressing B16-F10 melanoma spontaneous metastasis model and observed that intratumoral HER3-DC1 delivery significantly inhibited primary melanoma growth and spontaneous lung metastasis in C57BL/6 mice (Fig. 2O and Q; Supplementary Fig. S4C). To further test the metastasis-eradicating potential of the antitumor CD4⁺ Th1 response in a clinically relevant setting, we used a high tumor burden B16-F10 spontaneous metastasis model. As shown in Fig. 2P and Q, delivering intratumoral HER3-DC1 after primary tumors were between 70 and 100 mm² in size resulted in attenuation of larger melanoma growth and eradication of spontaneous lung metastasis. Collectively, these results suggest that cDC1 intratumoral delivery primes tumor antigen–specific antitumor CD4⁺ Th1 immunity, which only modestly controlled primary tumor growth but significantly inhibited metastasis.

We recently showed a critical role for tumor antigen–specific CD4⁺ Th1 cells in driving systemic and local antitumor immunity and eradicating tumors in HER2⁺ breast cancer (21). To further understand the interplay of CD4⁺ Th1 cells in the tumor antigen–pulsed cDC1–mediated antitumor response and metastasis prevention, a CD4⁺ T-cell depletion experiment was performed. As shown in Fig. 3A, intramammary gland delivery of HER2-DC1 failed to control spontaneous mammary carcinoma growth and severity of disease progression in BALB-neuT mice in the absence of CD4⁺ T cells. These mice also developed spontaneous micrometastasis in lung, liver, and brain, suggesting the loss of metastasis preventive potential of HER2-DC1 when CD4⁺ T cells were not present (Fig. 3B and C; Supplementary Fig. S4D). To determine whether CD8⁺ T cells had any impact on the HER2-DC1–mediated antitumor response and metastasis preventive capacity, we also depleted CD8⁺ T cells in BALB-neuT mice. Intramammary gland delivery of HER2-DC1 was able to control the tumor growth in the mammary glands of BALB-neuT mice in the absence of CD8⁺ T cells (Fig. 3A). We also observed the absence of micrometastasis in the lung, liver, and brain after HER2-DC1 intramammary gland delivery in BALB-neuT mice with intact CD4⁺ T cells and depletion of CD8⁺ T cells (Fig. 3B and C; Supplementary Fig. S4D). Similarly, the absence of CD4⁺ T cells, but not CD8⁺ T cells, abrogated the primary tumor and metastasis inhibitory potential of HER3-DC1 intratumoral delivery in the 4T1 tumor model (Fig. 3D–F). The metastasis inhibitory potential of antitumor CD4⁺ Th1 cells was further evident using a CD4 knockout mouse model, in which intratumoral activation of antitumor CD4⁺ Th1 cells failed to eradicate B16-F10 primary tumors and spontaneous lung metastasis (Fig. 3G and H).

Intratumoral delivery of tumor antigen–targeting mature DC1 was critical for controlling primary tumor growth and inhibiting spontaneous metastasis (Fig. 3I and J). This conclusion was supported by the failure of subcutaneous delivery of HER3-DC1, intratumoral delivery of HER3-immature DC, unpulsed DC1, or HER2-DC1 (DC1 targeting an irrelevant tumor antigen) to inhibit primary tumors and metastasis in the 4T1 tumor model (Fig. 3I and J). Next, we tested the specificity of the CD4⁺ Th1 response induced by intratumoral delivery of tumor antigen–targeting DC1. We observed significantly increased CD4⁺IFN-γ⁺ T cells after coculturing CD4⁺ T cells (isolated from intratumoral HER3-DC1–treated 4T1 tumors) with HER3-DC1, compared with coculture with HER2-DC1 or CD4⁺ T cells from control 4T1 tumors cocultured with HER3-DC1 or HER2-DC1 (Fig. 3K and L). This suggests that in addition to TDLNs, activation and priming of antitumor CD4⁺ Th1 cells may also occur in the tumors. The critical contribution of the antitumor CD4⁺ Th1 response was further investigated using IFN-γ neutralizing antibody treatment in a 4T1 TNBC model. The primary tumor and metastasis inhibitory capacity of HER3-DC1 intratumoral delivery was abrogated by IFN-γ neutralization in 4T1 tumor–bearing mice (Fig. 3M and N). Consistent with the TNBC spontaneous metastasis model, intratumoral delivery of HER3-DC1 failed to inhibit B16-F10 primary melanoma growth and lung metastasis in IFN-γ knockout mice (Fig. 3O and P). Collectively, these results demonstrated that the antitumor CD4⁺ Th1 cell–mediated immune response is necessary for prevention of overt metastasis.

The DCCs present in the BM of BALB-neuT mice with early lesions or primary mammary carcinoma have been reported to migrate very early and develop overt metastases (7, 8). We next investigated the efficacy of the HER2-DC1–primed CD4⁺ Th1 immune response on DCCs in a BALB-neuT mouse model. BM from control or HER2-DC1–treated BALB-neuT mice was first analyzed for CD4⁺ T-cell and CD8⁺ T-cell infiltration at 16 to 18 weeks of age by flow cytometry. As shown in Fig. 4A and Supplementary Fig. S5A, a significant increase in the number of CD4⁺ T cells with no changes in CD8⁺ T-cell infiltration was observed in the BM of HER2-DC1–treated BALB-neuT mice compared with control. Intramammary gland HER2-DC1–treated BALB-neuT mice had reduced levels of DCCs in their BM and lungs compared with control BALB-neuT mice (Fig. 4B and C). We also noted no impact on the proportion of DCCs in the BM of subcutaneous HER2-DC1–treated BALB-neuT mice compared with intramammary gland HER2-DC1 treatment (Fig. 4B). These DCCs also remained at reduced proportions in the BM of mice depleted of CD8⁺ T cells; however, this effect was abrogated in the absence of CD4⁺ T cells in BALB-neuT mice after HER2-DC1 delivery (Fig. 4B and C). We also observed a significantly increased proportion of senescent DCCs in the BM of HER2-DC1–treated BALB-neuT mice compared with control, based on SA-β-gal staining (Fig. 4D and E). The increased proportion of senescent DCCs was reversed only by the depletion of CD4⁺ T cells and not by the depletion of CD8⁺ T cells in BALB-neuT mice following HER2-DC1 delivery (Fig. 4D and E).

Western blot analysis revealed reduced protein expression of HER2, stemness/dormancy marker NR2F1, and EMT markers Wnt4 and Twist in DCCs from HER2-DC1–treated BALB-neuT mice compared with control mice (Fig. 4F). Immunofluorescence staining also demonstrated that HER2 and Ki-67 expression was reduced in DCCs from the BM of HER2-DC1–treated BALB-neuT mice (Fig. 4G). However, these changes were not apparent in the absence of CD4⁺ T cells (Fig. 4G).

Next, we investigated the specificity and functional activity of antitumor CD4⁺ Th1 cells on DCCs using a transwell migration assay. Activation and priming of anti-HER2 CD4⁺ Th1 cells (from HER2-DC1–treated BALB-neuT mice) with HER2-DC1 significantly increased the proportion of senescent DCCs (from control BALB-neuT mice) compared with anti-HER2 CD4⁺ Th1 cells activated with unpulsed DC1 (Fig. 4H; Supplementary Fig. S5B). We further confirmed the contribution of IFN-γ secreted by anti-HER2 CD4⁺ Th1 cells in inducing senescence in DCCs by IFN-γ neutralization in the coculture (Fig. 4H; Supplementary Fig. S5B). These observations were also consistent in assays in which HER2⁺ TUBO cells were cocultured with anti-HER2 CD4⁺ Th1 cells and HER2-DC1 with or without IFN-γ neutralization (Supplementary Fig. S5C and S5D).

We extended our study in the 4T1 TNBC spontaneous metastasis model and observed that intratumoral delivery of HER3-DC1 reduced the proportion of DCCs in the BM of 4T1 tumor–bearing mice when compared with control and subcutaneous HER3-DC1 delivery groups (Fig. 4I). Next, the specificity and functional status of CD4⁺ Th1 cells from the BM of control and intratumoral HER3-DC1–treated 4T1 tumor–bearing mice were examined. A significant increase in the level of CD4⁺IFN-γ⁺ T cells was observed after coculturing CD4⁺ T cells (isolated from the BM of HER3-DC1–treated 4T1 mice) with HER3-DC1 (Fig. 4J and K). However, no change was observed when these CD4⁺ T cells were cocultured with HER2-DC1 (DC1 pulsed with irrelevant tumor antigen; Fig. 4J and K), highlighting the priming and propagation potential of tumor antigen–specific antitumor CD4⁺ Th1 cells in the BM as well as their potential to target DCCs in TNBC.

To evaluate the tumorigenic and metastatic seeding potential of DCCs in the BM of control and HER2-DC1–treated BALB-neuT mice, we performed experiments with NSG mice. As the intratumoral priming of antitumor CD4⁺ Th1 cells reduced the number of DCCs in the BM, we injected the same number of viable BM DCCs from the experimental BALB-neuT mice into NSG mice to provide more specific evidence for the inhibition of DCC tumorigenesis by the CD4⁺ Th1 immune response. Primary tumor growth and metastasis in the lung and liver were observed in NSG mice following injection of the same number of viable DCCs from the untreated control BALB-neuT mice (Fig. 4L–N; Supplementary Fig. S5E). Injection of the same number of viable DCCs from the BM of HER2-DC1–treated BALB-neuT mice into NSG mice did not result in the development of primary tumors and micrometastases in distant organs, lung and liver (up to 120 days; Fig. 4L–N; Supplementary Fig. S5F). We next assessed the impact of active metastatic tumor cells isolated from the metastatic HER2⁺ tumors of control and HER2-DC1–treated BALB/c mice. Although we observed a significant delay in tumor growth in NSG mice that were subcutaneously transplanted with metastatic tumor cells of HER2-DC1–treated mice compared with control, the HER2-DC1–treated metastatic tumor cells still formed active metastasis (Supplementary Fig. S5G). Collectively, these results provide evidence that the tumor antigen–specific CD4⁺ Th1 immune response targets more specifically DCCs in distant organs and inhibits their tumorigenic and metastatic growth potential in breast carcinogenesis.

We observed significantly increased infiltration of CD4⁺ T cells in the primary tumor, TDLN, and DCC-driven distant metastasis in intratumoral HER3-DC1–treated B16-F10 spontaneous metastasis–bearing mice compared with the untreated control group (Fig. 5A–C). These results were consistent with the BALB-neuT spontaneous metastasis model (Figs. 2H, I, and 4A; Supplementary Fig. S3A). A small proportion of these tumor-infiltrating CD4⁺ T cells were CD4⁺PD1⁺ T cells (Fig. 5D), with no obvious changes observed in TDLN and the metastatic site with DCCs (Supplementary Fig. S6A and S6B). This result suggests that the tumor antigen–targeting DC1-driven CD4⁺ T cells were not at the exhaustion stage. Next, a phenotypic analysis of these infiltrating CD4⁺ T cells elicited an increase in CD4⁺CD44⁺CD62L^– EM T cells and CD4⁺CD44^–CD62L^– effector T cells in intratumoral HER3-DC1–treated B16-F10 tumors (Fig. 5E), similar to the results obtained with the BALB-neuT model (Supplementary Fig. S3A). On the other hand, TDLN of HER3-DC1–treated mice had a significant increase in CD4⁺CD44⁺CD62L⁻ EM and CD4⁺CD44⁺CD62L⁺ CM T cells (Fig. 5F; Supplementary Fig. S6C). A significant increase in CD4⁺CD44⁺CD62L^– EM, CD4⁺CD44⁺CD62L⁺ CM, and CD4⁺CD44^–CD62L^– effector T cells was detected in a DCC-driven distant metastasis site of HER3-DC1–treated mice (Fig. 5G).

We also observed an increased level of CD4⁺FOXP3⁺ regulatory T cells (Treg) in primary tumors, TDLN, and DCC-driven metastases in HER3-DC1–treated mice bearing B16-F10 spontaneous metastasis compared with the control group (Supplementary Fig. S6D–S6F). We have previously shown that cDC1 can drive the differentiation of CD4⁺FOXP3⁺ Treg into IFN-γ–producing CD4⁺FOXP3⁺T-bet⁺ Th1 cells, contributing to antitumor immunity (38). In addition, the infiltration of CD4⁺FOXP3⁺ Treg may strongly support their requirement to maintain immune tolerance and prevent chronic inflammation and autoimmunity-related tissue damages in treatment-responsive mice (completely regressed for primary tumors and distant metastasis). Further studies will characterize and assess the function of these infiltrated CD4⁺FOXP3⁺ Treg in the HER3-DC1–treated group.

As our results provide strong evidence for the enhancement of antitumor CD4⁺ T cells in primary tumors, TDLN, and DCC-driven distant metastases, we next examined the expression patterns of two important migratory chemokine receptors, CXCR3 and CCR7, on these CD4⁺ T cells. Expression of CXCR3 and CCR7 on CD4⁺ Th1 cells plays an essential role in the migration of these cells into secondary lymphoid organs and their infiltration and homing into primary tumors and distant metastatic sites (39–41). A significant increase in the infiltration of both CD4⁺CCR7⁺CXCR3⁺ Th1 cells and CD4⁺CCR7^–CXCR3⁺ Th1 cells was observed in primary tumors, TDLN, and a DCC-driven distant metastasis site from the intratumoral HER3-DC1–treated B16-F10 tumor–bearing group compared with the untreated control group (Fig. 5H–L). Collectively, these results further support the notion that antitumor CD4⁺ Th1 cells primed intratumorally can migrate into distant organs to eradicate DCC-driven metastases.

Differential gene expression analysis from the RNA sequencing experiments showed downregulation of various chemokine genes, MHC class I and MHC class II gene subsets, and CD1d genes in DCCs of BALB-neuT mice and patients with HER2⁺ breast cancer (Fig. 6A–D). We further confirmed decreased cell surface expression of MHC class I, MHC class II, and CD1d in DCCs of BALB-neuT mice (Fig. 6E–I). This observation was consistent in tumor cells from early mammary lesions of BALB-neuT mice and TUBO cells (Supplementary Fig. S7A–S7H). Together, these data suggest that DCCs become invisible to CD4⁺ T cells, CD8⁺ T cells, B cells, and NKT cells and escape immunosurveillance. We next observed that the CD4⁺ Th1 cytokine IFN-γ upregulated the expression of genes encoding CXCL9, CXCL10, CXCL11, CCL7, CXCL16, and CX3CL1 in DCCs of BALB-neuT mice and patients with HER2⁺ breast cancer (Fig. 6A and B; Supplementary Fig. S8B). These chemokines are involved in the chemoattraction of CD4⁺ Th1 cells, CD8⁺ T cells, B cells, NK cells, NKT cells, and cDC1 and drive strong antitumor immunity (42). MHC class I and class II gene subsets and antigen presentation machinery were upregulated in DCCs of BALB-neuT mice and patients with HER2⁺ breast cancer following IFN-γ treatment, compared with control DCCs (Fig. 6C and D; Supplementary Fig. S8A and S8C–S8F). We further confirmed an increase in cell surface expression of MHC class I, MHC class II, and CD1d following IFN-γ treatment in DCCs of BALB-neuT mice (Fig. 6E and G–I). Similarly, IFN-γ treatment also increased cell surface expression of MHC class I, MHC class II, and CD1d in tumor cells of BALB-neuT mice mammary lesions and TUBO cells (Supplementary Fig. S7A–S7H). The CD1d⁺ population was found at a higher frequency in DCCs of BALB-neuT mice compared with early mammary lesions and TUBO cells (Fig. 6F). IFN-γ treatment increased the percentage of the CD1d⁺ population in all three cell types (Fig. 6G and H; Supplementary Fig. S7C, S7D, S7F, and S7G), highlighting a role for the CD4⁺ Th1 response in regulating CD1d expression in DCCs and tumor cells.

After observing the increased infiltration of CD4⁺ T cells (Fig. 4A), we then tested if intratumoral priming of antitumor CD4⁺ Th1 cells can also drive the infiltration of NKT cells, NK cells, and B cells into the BM to target DCCs in a BALB-neuT mice model. Consistent with the possibility raised by our in vitro findings, intratumoral priming of antitumor CD4⁺ Th1 cells increased the infiltration of CD1d tetramer⁺ iNKT cells, NK cells, and B cells in the BM of BALB-neuT mice compared with untreated control BALB-neuT mice (Fig. 6J–M). Collectively, these data suggested that the intratumoral delivery of tumor antigen–targeting cDC1 drives antitumor CD4⁺ Th1 responses as well as the infiltration of iNKT cells, NK cells, and B cells in the BM. These immune effector cells may collectively enhance the expression of chemokines and classical and nonclassical MHC molecules in DCCs, leading to increased immune recognition, antigen presentation, and the facilitation of strong antitumor immunity to eradicate DCCs in distant organs in breast cancer.

Stemness and self-renewing potential of DCCs have been shown to be responsible for resistance to conventional therapies, resulting in recurrence and metastasis in breast cancer (11, 43). We examined the therapeutic efficacy of the CD4⁺ Th1 cytokine IFN-γ on the stemness of DCCs isolated from untreated control BALB-neuT mice. The ALDEFLUOR assay revealed an enriched ALDH⁺ stem-cell population in DCCs of control BALB-neuT mice (Supplementary Fig. S9A). In addition, we also observed an enriched ALDH⁺ stem-cell population in tumor cells from the control BALB-neuT mice mammary lesion and TUBO cells (Supplementary Fig. S9B and S9C). The enriched ALDH⁺ stem-cell population was reduced after IFN-γ treatment of DCCs from control BALB-neuT mice, tumor cells of control BALB-neuT mice mammary lesion, and TUBO cells (Supplementary Fig. S9A–S9C). Consistent with the above finding, high frequencies of CD44⁺CD24⁺ cancer stem cells were detected among DCCs of control BALB-neuT mice, tumor cells of control BALB-neuT mice mammary lesions, and TUBO cells (Supplementary Fig. S9D–S9F). The percentage of enriched ALDH⁺ and CD44⁺CD24⁺ cancer stem cells was higher among DCCs of untreated control BALB-neuT mice compared with their early mammary lesions (Supplementary Fig. S9A, S9B, S9D, and S9E). We also noted that IFN-γ treatment greatly reduced the frequency of CD44⁺CD24⁺ cancer stem cells in all three cell types (Supplementary Fig. S9D–S9F). By providing critical culturing conditions, we observed a high number of mammospheres forming from BM DCCs of control BALB-neuT mice (Fig. 7A and B), confirming the self-renewing potential of DCCs. In addition, untreated TUBO cells also generated higher mammosphere numbers (Fig. 7A and C). IFN-γ treatment significantly reduced mammosphere numbers formed by DCCs of BALB-neuT mice and TUBO cells compared with untreated controls (Fig. 7A–C). Next, the IFN-γ–mediated inhibition of the self-renewing potential of DCCs from control BALB-neuT mice was also accompanied by the downregulation of various gene signatures, mainly related to cell adhesion (Fig. 7D), highlighting the critical role of IFN-γ in inhibiting proliferation, survival ability, and tumor-forming potential of DCCs. We further observed that treatment of DCCs from control BALB-neuT mice with IFN-γ significantly increased the number of senescent DCCs compared with untreated control DCCs (Fig. 7E and F).

We further extended our study to investigate the direct effects of IFN-γ on BM DCCs of patients with breast cancer. DCCs of patients with HER2⁺ breast cancer were able to form mammospheres (Fig. 7G and H). However, IFN-γ inhibited the self-renewing capability of DCCs from patients with HER2⁺ breast cancer, as indicated by a significant reduction in mammosphere numbers in the presence of IFN-γ (Fig. 7G and H). This result was further confirmed with JIMT-1 cells, in which we found reduced numbers of mammospheres after IFN-γ treatment (Fig. 7G and I). Western blotting revealed reduced expression and activation of HER2, Wnt4, and NR2F1 in IFN-γ–treated DCCs when compared with untreated control DCCs of patients with HER2⁺ breast cancer (Fig. 7J). Subsequently, RNA sequencing identified the downregulation of various gene signatures related to progesterone signaling and cell adhesion in IFN-γ–treated DCCs of patients with HER2⁺ breast cancer (Fig. 7K and L). We next examined the effect of IFN-γ on the senescence and apoptosis-inducing potential in DCCs from patients with HER2⁺ breast cancer. As shown in Fig. 7M–P, IFN-γ significantly increased the level of senescence and apoptosis in DCCs of patients with HER2⁺ breast cancer. Consistent with the preceding experiments, we also found that IFN-γ treatment upregulated various genes involved in apoptosis in patient DCCs (Supplementary Fig. S10A). Subcutaneous implantation of viable DCCs from patients with HER2⁺ breast cancer into NSG mice led to tumor development (Fig. 7Q). IFN-γ–treated viable DCCs of patients with HER2⁺ breast cancer failed to develop tumors in NSG mice (up to 120 days; Fig. 7Q), highlighting the ability of the CD4⁺ Th1 cytokine IFN-γ to inhibit tumorigenesis mediated by patient DCCs.

To further characterize the gene profiles of control DCCs of BALB-neuT mice and IFN-γ–treated DCCs of BALB-neuT mice, we performed RNA sequencing. The differential gene expression analysis revealed that IFN-γ treatment regulated the expression of various genes in DCCs compared with untreated control DCCs from BALB-neuT mice (Supplementary Fig. S11A). Next, pathway enrichment analysis for upregulated and downregulated datasets was performed to uncover the protein response networks. As shown in Supplementary Fig. S12A, the downregulated networks were mainly related to cell adhesion, Wnt signaling, the cell cycle, cytoplasmic microtubules, and extracellular matrix remodeling in IFN-γ–treated DCCs of BALB-neuT mice. Networks corresponding to inflammation and the immune system were highly enriched in the upregulated datasets including interferon signaling, chemotaxis, antigen presentation, NK cell cytotoxicity, lymphocyte proliferation, and apoptosis in IFN-γ–treated DCCs of BALB-neuT mice (Supplementary Fig. S12B). Next, IFN-γ treatment was found to downregulate various cancer stemness genes and EMT genes and to upregulate cell-cycle genes in DCCs from BALB-neuT mice when compared with untreated control DCCs of BALB-neuT mice (Supplementary Fig. S11B–S11D).

The differential gene expression and pathway enrichment analyses for the RNA sequencing data of patients’ DCCs showed downregulation for various gene networks that were mainly associated with progesterone signaling, TGF-β, GDF/activin signaling, cell cycle, and cell adhesion in IFN-γ–treated DCCs of patients with HER2⁺ breast cancer compared with untreated controls (Supplementary Figs. S11E and S12C). The majority of upregulated gene networks in IFN-γ–treated DCCs of patients with HER2⁺ breast cancer were involved in interferon signaling, inflammation, antigen presentation, chemotaxis, and apoptosis as compared with controls (Supplementary Figs. S11E and S12D). The GSEA curating the REACTOME database for downregulated genes revealed that cholesterol biosynthesis was the highest-ranked gene set negatively correlated with IFN-γ treatment in DCCs of patients with HER2⁺ breast cancer (Supplementary Fig. S11F and S11G). Consistent with the modulatory effects of IFN-γ in DCCs of control BALB-neuT mice, we also observed downregulation of specific cancer stemness genes, EMT genes, and cell-cycle regulatory genes after IFN-γ treatment in DCCs of patients with HER2⁺ breast cancer (Supplementary Fig. S11H–S11J). Collectively, these data provide transcriptomic evidence that the antitumor CD4⁺ Th1 response regulates cancer stemness, EMT, cell cycle, and cholesterol biosynthesis signatures to restrain DCCs’ tumorigenesis.

The IFN-stem cell-down signature (ISDS) of gene expression has been associated with good prognosis for relapse-free, distant metastasis-free, and overall survival in patients with breast cancer (44). DCCs of patients with HER2⁺ breast cancer showed downregulated expression of the ISDS (Supplementary Fig. S11K). Indeed, IFN-γ treatment upregulated the expression of 27 genes of ISDS in DCCs from the BM of patients with HER2⁺ breast cancer (Supplementary Fig. S11K). A similar effect on the upregulation of the ISDS gene signature was observed in JIMT-1 cells after IFN-γ treatment (Supplementary Fig. S10B).

Detection and persistence of DCCs in the BM of patients with breast cancer have been identified as a primary source for late recurrence and distant metastasis in multiple organs (5, 45, 46). Selective targeting and eradication of DCCs in the BM before succumbing to metastases can be a substantial advantage for these patients (47). Unfortunately, no such targeted therapy presently exists; instead, intense chemotherapy in the adjuvant or neoadjuvant setting is given over a period of time with the ultimate goal of preventing recurrence in patients with invasive breast cancer. Clinical studies confirm that these DCCs have not been effectively eradicated by chemotherapies (48–50). Our study provides evidence that intratumoral priming of tumor antigen–specific CD4⁺ Th1 cells enhances their migration to the BM to selectively target DCC burden and eradicate DCC metastasis in distant organs in multiple breast cancer spontaneous metastasis models (Supplementary Fig. S13). Antitumor CD4⁺ Th1 cells mediated the eradication of lung metastasis in a melanoma spontaneous metastasis model, further confirming this finding. The tumor antigen–targeting DC1-mediated activation and priming of antitumor CD4⁺ Th1 cells occurred both intratumorally and in TDLNs in our study. Increased infiltration of CD4⁺CCR7⁺CXCR3⁺ Th1 cells and CD4⁺CCR7^–CXCR3⁺ Th1 cells in primary tumors, TDLNs, and DCC-seeding distant metastatic sites following intratumoral delivery of tumor antigen–targeting DC1 further supports our finding. The selective targeting of DCCs was not as extensive when CD4⁺ T cells were absent, even in the presence of CD8⁺ T cells, suggesting a prominent role for antitumor CD4⁺ Th1 cells in targeting DCCs, but they may also require other infiltrating immune effector cells such as iNKT cells, NK cells, and B cells.

Previous reports have suggested that NR2F1, HER2, Wnt4, and Twist can regulate dormancy, stemness, and EMT transition in DCCs, which renders them more susceptible to a reversible programming switch from dormancy to proliferative growth, leading to disease relapse in patients with HER2⁺ breast cancer (8, 25, 51, 52). We demonstrated that the antitumor CD4⁺ Th1 response was highly effective in reducing NR2F1, HER2, Wnt4, and Twist in DCCs. Although radiotherapy and chemotherapy can induce senescence in cancer cells and inhibit tumorigenesis, the senescence-associated secretory phenotype and reversible senescence leave them more vulnerable to contribute to a disease relapse in patients and mouse models (53). Our study discovered an important insight that the antitumor CD4⁺ Th1 response rendered irreversible senescence in DCCs. These senescent DCCs completely failed to grow into primary tumors and overt metastasis in distant organs in an immunocompromised mouse model.

In this study, we provide evidence that the antitumor CD4⁺ Th1 cell–mediated targeting of DCCs and prevention of metastasis in distant organs was primarily driven by IFN-γ. The precise dynamics of the targeted effects of the CD4⁺ Th1 immune response on DCCs was established in the present study, and it was observed that the CD4⁺ Th1 cytokine, IFN-γ, inhibited stemness and self-renewing potential and induced apoptosis in DCCs. Increased cholesterol biosynthesis has been linked to the propagation of cancer cell stemness and self-renewal and correlated with reduced recurrence-free survival in patients with breast cancer (54). A recent study has shown that cholesterol alterations contribute to invasion and metastasis and decreased patient survival in melanoma (55, 56). Higher steady-state expression of cholesterol biosynthesis signatures was observed in DCCs. The CD4⁺ Th1 cytokine IFN-γ downregulated cholesterol biosynthesis, stemness, and EMT signatures in DCCs, and our data may provide molecular evidence for a role for the CD4⁺ Th1 response in inhibiting stemness and metastasis of DCCs. Although the total RNA sequencing findings from our study were limited to expanded DCCs of patients with breast cancer, we provide the first transcriptomic profiles for these DCCs and associated molecular changes for post–CD4⁺ Th1 response. Our ongoing clinical study will investigate the molecular aspects, heterogeneity, stemness, EMT, and cholesterol biosynthesis of isolated patient DCCs compared with their tumor lesions at a single-cell level using single-cell RNA sequencing. IFN-γ has been shown to synergize with cholesterol-inhibiting statins to effectively reduce the growth of HER2 mammary carcinoma (57). Further studies are warranted for antitumor CD4⁺ Th1 cells on the regulation of cholesterol biosynthesis in DCCs.

In this study, we utilized a well-established EpCAM marker–based strategy (11, 58, 59) to isolate DCCs from the BM of patients with breast cancer and breast cancer mouse models. A recent parallel study has provided strong evidence that EpCAM⁺ DCCs or EpCAM⁺ CTCs display similar molecular characteristics of tumor phenotypes (11). Although the findings of our study were limited to EpCAM⁺ DCCs, clinical studies have confirmed that the presence of EpCAM⁺ DCCs is associated with poor outcomes in patients with cancer (60–62). Our isolation method was more efficient in detecting and isolating DCCs in the BM in the majority of tested DCIS and patients with early invasive breast cancer (a total of 59 patients), which is a markedly higher proportion than reported previously (18, 63, 64). These observed findings will be further confirmed in our ongoing clinical study with a larger number of patients. Our study showed the tumorigenic and metastatic potential of these isolated EpCAM⁺ DCCs in an immunodeficient mouse model and that an antitumor CD4⁺ Th1 response could inhibit EpCAM⁺ DCC–driven metastasis.

Loss of MHC class I, MHC class II, and CD1d expression has been found in various solid tumors (65–68). Our study observed the downregulation of their expression in DCCs, which suggests a defect in their antigen presentation machinery and a subsequent escape mechanism from T cell– and iNKT cell–mediated elimination. The CD4⁺ Th1 cytokine IFN-γ increased MHC class I and II expressions in DCCs, and that IFN-γ increased CD1d expression in DCCs, supporting what we believe to be a novel role for the CD4⁺ Th1 response to drive iNKT-cell infiltration and NKT cell–mediated clearance of DCCs in distant organs. Intratumoral expression of certain antitumor immunostimulatory chemokines CXCL9, CXCL10, CXCL11, CXCL16, CX3CL1, and CCL7 can attract CD4⁺ Th1 cells, CD8⁺ T cells, B cells, and NK cells into the immunosuppressive tumor microenvironment (42, 69, 70). These immune effector cells may play a critical role in driving a stronger anti-DCC response. B cells express MHC class II and can present tumor antigens to CD4⁺ Th1 cells (71, 72). NK cells recognize cancer cells independent of MHC class I and can produce IFN-γ (73, 74). iNKT cells may recognize cancer cells or DCCs expressing CD1d and produce IFN-γ (75). Various findings have shown the beneficial therapeutic efficacy and clinical outcomes of triggering intratumoral chemokines by chemotherapy, radiotherapy, and anti-PD1 immunotherapy in solid tumors (42, 76). In this study, IFN-γ–treated DCCs revealed an enrichment of these immunostimulatory chemokine signatures and antigen presentation–associated signatures. Consistent with this assessment, the intratumoral priming of antitumor CD4⁺ Th1 cells also triggered the infiltration of B cells and NK cells into the BM in the breast cancer spontaneous metastasis model.

Collectively, our observations demonstrate, for what we believe to be the first time, a clear role for the antitumor CD4⁺ Th1 immune response in targeting and inducing immune recognition of DCCs and generating an amplification loop for the accumulation of CD4⁺ Th1 cells, iNKT cells, NK cells, and B cells in the BM to eliminate DCCs and prevent overt metastasis (Supplementary Fig. S13). Our data have important clinical implications and suggest that early cancer therapies driving antitumor CD4⁺ Th1 immunity may have great potential in preventing metastasis in a wide array of cancers.

G. Ramamoorthi reports a patent for PCT/US2024/013114 pending to Moffitt Cancer Center. M.C. Lee reports other support from private donors during the conduct of the study. B.J. Czerniecki reports grants from CDMRP/DOD, Pennies in Action, and CDMRP/DOD during the conduct of the study as well as other support from ImmunoRestoration outside the submitted work and has a patent for use of syringe ready type I dendritic cells pending. No disclosures were reported by the other authors.

G. Ramamoorthi: Conceptualization, formal analysis, supervision, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. M.C. Lee: Conceptualization, resources, investigation, writing–review and editing. C.M. Farrell: Investigation. C. Snyder: Investigation. S.K. Garg: Formal analysis, visualization, writing–review and editing. A.L. Aldrich: Writing–review and editing. V. Lok: Investigation. W. Dominguez-Viqueira: Formal analysis, investigation. S.K. Olson-Mcpeek: Investigation. M. Rosa: Formal analysis, investigation. N. Gautam: Investigation. S. Pilon-Thomas: Resources. L. Cen: Formal analysis. K.N. Kodumudi: Conceptualization, supervision. D. Wiener: Writing–review and editing. T. Oskarsson: Writing–review and editing. A.P. Gomes: Writing–review and editing. R.A. Gatenby: Writing–review and editing. B.J. Czerniecki: Conceptualization, resources, supervision, funding acquisition, project administration, writing–review and editing.

This work was supported by Department of Defense Award W81XWH-19-1-0675, awarded to B.J. Czerniecki; Department of Defense Award W81XWH-16-1-0385 and Pennies in Action, awarded to B.J. Czerniecki; and the Miles for Moffitt Milestone Award from Moffitt Cancer Center, awarded to B.J. Czerniecki, G. Ramamoorthi, and K.N. Kodumudi. This work was also supported in part by the Small Animal Imaging Lab, Flow Cytometry Core, Molecular Genomics Core, Biostatistics and Bioinformatics Shared Resource, Tissue Core Histology, Analytic Microscopy Core, Advanced Analytical and Digital Laboratory, and Vivarium Services Core at the Moffitt Cancer Center, an NCI-designated Comprehensive Cancer Center (P30-CA076292).