C57BL/6 mice (Mus musculus) aged 6–8 weeks old were purchased from Taconic (Hudson, New York) and housed and maintained in specific pathogen-free conditions. Each experiment involved the use of 4–23 mice per experimental group, and experimental group sizes are noted in the figure legends. As previously published,19–24 the B16GP33 cell line was established by transfecting B16F10, a poorly immunogenic melanoma cell line arising in C57BL/6 mice, with a plasmid encoding GP33, a class I major histocompatibility (MHC)-restricted lymphocytic choriomeningitis virus (LCMV) glycoprotein. Tumor inoculations were performed using B16GP33 melanoma and Hepa1-6 hepatocellular carcinoma (ATCC, Manassas, Virginia). B16GP33 cell lines were maintained by culturing with 200 µg/mL G418 in RPMI-1640 medium (Gibco, Life Technologies, Grand Island, New York) with 10% fetal bovine serum (HyClone, GE Healthcare Life Sciences), 2 mM L-glutamine (Gibco, Life Technologies), 100 U/mL penicillin (Gibco, Life Technologies) and 100 µg/mL streptomycin (Gibco, Life Technologies), and Hepa 1–6 cell lines were maintained by culturing in Dulbecco’s modified Eagle’s medium (DMEM) (Gibco, Life Technologies) with 10% fetal bovine serum.

Mice were inoculated by subcutaneous flank injection with 10⁶ B16GP33 or 4×10⁶ Hepa1-6 cells suspended in phosphate buffered saline (PBS). Flank tumors were monitored and measured with electronic calipers every 3 days, and tumor volumes were calculated according to the following formula: volume=long dimension × short dimension × (short dimension/2). Pulmonary tumors were established by inoculating mice by tail vein injection with 2×10⁵ B16GP33 cells suspended in PBS. Pulmonary metastases were quantified after euthanasia by counting the number of pigmented tumors identified along the surfaces of all lobes. Interobserver variability was mitigated by averaging the counts of three independent observers.

Tumors were treated with histotripsy ablation, irradiation, radiofrequency ablation, or sham therapy. All procedures were performed following induction of isoflurane anesthesia. Histotripsy tumor ablation was performed by placing mice into a custom-built, eight-element, 1 MHz focused ultrasound therapy transducer. A field-programmable gate array development board (DE0-Nano, Terasic Technology, Taiwan) controlled the therapy transducer to generate one to two cycles of histotripsy pulses. An L40-8/12 20 MHz imaging probe (Ultrasonix, Vancouver, Canada) was coaxially aligned with the therapy transducer to visualize the histotripsy bubble cloud and provide real-time targeting feedback. Using a motorized positioning system, the tumor was scanned to label treatment boundaries in three orthogonal dimensions. Boundary coordinates were used to generate a three-dimensional ellipsoid consisting of uniformly spaced treatment locations (spacing ranged from 0.4 to 0.7 mm in the axial, lateral, and elevational dimensions) to encompass the tumor while avoiding injury to surrounding tissues. At each location, 50 histotripsy pulses at 100 Hz pulse repetition frequency (PRF) were delivered to generate an estimated −30 MPa peak negative pressure at the focus, and all treatment locations were traversed by moving the therapy transducer using the motorized positioning system. Histotripsy duration ranged from 4 to 15 min, depending on tumor volume and treatment location. Tumor irradiation was performed by positioning the mouse in radiation chambers with lead body shielding placed over the mouse such that only the tumor was exposed to 15 Gy unfractionated radiation. Radiofrequency ablation was performed by sterilizing the tumor site and passing an 18-gauge hollow needle into the tumor, through which a radiofrequency probe was passed for administration of thermal ablation at 90°C to encompass the tumor while avoiding injury to surrounding tissues. Sham therapy consisted of isoflurane anesthesia alone.

Sham-ablated tumors were explanted and placed in 1.5 mL tubes and subjected to five cycles of rapid freezing in liquid nitrogen for 2 min, followed by rapid thawing in a 60°C water bath for 2 min. Non-ablated and in vivo histotripsy-ablated tumors were liquefied consistently and exposed to 10⁶ CD8+ T cells harvested from the spleen of C57BL/6 mice 8 days after LCMV infection (at which time 5%–10% of circulating CD8+ T cells are specific for GP33) in the presence of 10 U interleukin (IL)-2 and brefeldin A at 37°C for 5 hours. CD8+ T cells were also exposed to media alone as negative controls and to 0.01 µg/mL GP33 peptide alone as positive controls.

Checkpoint inhibition immunotherapy was performed by treating tumor-bearing mice with intraperitoneal injections of 200 µg anti-cytotoxic T lymphocyte-associated protein-4 (CTLA-4) mAb (Bio X Cell, West Lebanon, New Hampshire) on days 6, 9 and 12 after B16GP33 tumor inoculation or days 3, 6, 9 and 12 after Hepa1-6 tumor inoculation.

Tumors, tumor-draining lymph nodes (TDLN), non-tumor-draining lymph nodes (NTDLN), and spleens were harvested and processed into single-cell suspensions on sterile fine mesh screens. Tumor infiltrating lymphocytes (TIL) were isolated from dissociated tumors using Histopaque (Sigma Aldrich, St Louis, Missouri) centrifugation. Lymphocytes from each sample were stained with allophycocyanin (APC)-labeled MHC class I (D^d) tetramers loaded with MHC class I-restricted GP33 peptide (National Institutes of Health Tetramer Core Facility, Atlanta, Georgia), Pacific Blue-labeled anti-CD8 mAb, BV510-labeled anti-CD4 mAb, PerCP/Cy5.5-labeled anti-CD19 mAb, PE/Cy7-labeled anti-CD11c mAb, AF488-labeled anti-Ly6G mAb, BV650-labeled anti-NK1.1 mAb, and PE-labeled anti-CD25 mAb. For intracellular staining, cells were fixed, permeabilized, and stained with PE CF594-labeled anti-FoxP3 mAb or PE-labeled anti-IFNγ mAb. All antibodies were purchased from BioLegend (San Diego, California), except for anti-CD8 (BD Biosciences, San Diego, California) and anti-Ly6G and anti-FoxP3 (eBioscience, San Diego, California). Fluorescence data were acquired on a BD LSRFortessa Flow Cytometer (BD Biosciences) and analyzed using FlowJo software (Tree Star, Ashland, Oregon). All flow cytometry data are shown following gating on lymphocytes gate.

Tumors were washed with PBS and preserved in 10% formalin for 24–48 hours and in 70% ethanol until paraffin embedding. Sections of 4.5 µm thickness were deparaffinized in xylene for 6 min, then washed in 50% xylene in ethanol and 100% ethanol, and rehydrated by sequential transfer to 95%, 70%, and 50% ethanol, and ultimately to distilled water for 3 min each before staining. Samples were washed in PBS, blocked in 1% bovine serum albumin for 30 min, and incubated with rabbit polyclonal anti-CD8a antibody to CD8a antibody (synaptic system) at 4°C overnight. Samples were then washed and blocked in goat serum, incubated with biotin-conjugated goat antirabbit IgG for 45 min, washed for 30 min in PBS, then incubated with horseradish peroxidase (HRP)-conjugated avidin. Samples were developed using the CN/DAB Substrate Kit (Thermo Fisher, Waltham, Massachusetts) for 5 min as per vendor instructions, then counterstained with hematoxylin for 10 s, and washed and mounted. Samples were visualized using a 10× or 20× objective on a Nikon Eclipse E400 microscope, and images were captured using a DXM1200F camera using the software provided by the vendor (Nikon, Melville, New York). At least 10 images from each tumor sample were obtained for each magnification. Cells were counted using ImageJ software after appropriate background subtraction.

Formalin-fixed paraffin-embedded tumor sections were deparaffinized and rehydrated as above. Heat-mediated antigen retrieval was performed in Tris buffer containing 1% Triton X (for CD8 staining) or citrate buffer (for calreticulin (CRT) staining) in a boiling water bath for 20 min. Sections were blocked in 5% bovine serum albumin (BSA) in PBS for 30 min, then primary antibody staining was performed with a primary rabbit IgG antibody against CD8, or rabbit IgG against CRT directly conjugated to Alexa 647 and rabbit IgG against ERp72 directly conjugated to Alexa 555 (Abcam, Cambridge, England) overnight at 4°C overnight. Secondary antibody staining (for CD8 staining) was performed with Alexa 488-labelled goat antirabbit IgG for 1 hour. Sections were washed and mounted using Prolong Gold Antifade Mountant with 4’6-diamidino-2-phenylindone (DAPI) (Thermo Fisher), and a minimum of five overlapping optical fields from two independent samples were captured. Images were acquired using a Keyence BZ-800 microscope (Keyence, Osaka, Japan) under 10× objective with a 1× digital zoom, or under 40× objective with a 1× digital zoom. Images were imported to ImageJ and the threshold was adjusted using the color threshold tool.

Serum levels of HMGB1 were measured by ELISA (Tecan, Männedorf, Switzerland), according to the manufacturer’s instructions. Intratumoral levels of HMGB1 were measured by immunofluorescence using tumor samples snap-frozen in liquid nitrogen. Tumors were prepared on round cover glass and fixed in 4% paraformaldehyde (PFA) at room temperature for 15 min after PBS washing. Slides were permeabilized in 0.3% Triton X-100 for 10 min followed by blocking solution (5% BSA in PBS) for 30 min at room temperature. Samples were stained in antibody reaction buffer (1% BSA and 0.3% Triton X-100 in PBS, pH 7.4) with a primary antibody for HMGB1 overnight at 4°C, followed by secondary antibodies for 1 hour at room temperature. F-actin was stained with rhodamine phalloidin (Thermo Fisher). Hoechst 33342 (Thermo Fisher) was used for nuclear staining. All slides were scanned under the same conditions for magnification, exposure time, lamp intensity, and camera gain. Confocal images were acquired using an Olympus FluoView 1000 microscope (Olympus, Tokyo, Japan). All images were quantitated as previously described with HMGB1 staining intensity normalized to nuclei using MetaMorph software (Molecular Devices, Downingtown, Pennsylvania).25

Experimental data were analyzed using SPSS V.23 statistical software. Two groups of independent samples were compared using the two-tailed Student’s t-test. Analysis of variance was applied to assess the statistical significance in tumor growth kinetics. Differences in survival between groups were assessed using the log-rank test. Significance was defined as p<0.05, and error bars represent SEM.

We examined the effect of histotripsy on intratumoral lymphocyte infiltration. Ten days after inoculation with subcutaneous flank injections of poorly immunogenic B16GP33 melanoma (a B16F10 melanoma cell line engineered to stably express very low levels of the LCMV peptide GP3319–24), C57BL/6 mice were treated with sham or histotripsy ablation. Histotripsy ablation significantly inhibited tumor growth without producing significant alterations in body weight (figure 1A,B). On days 3 and 10 after sham or histotripsy ablation, mice were euthanized and tumors were harvested for flow cytometric analysis. As shown in figure 1C–F a significant upregulation of intratumoral natural killer (NK) cells, dendritic cells (DCs), neutrophils, B cells, CD4+ T cells and CD8+ T cells was observed following histotripsy ablation. In addition, histotripsy was also associated with significant intratumoral infiltration of melanoma antigen-specific CD8+ T cells (figure 1G).

We compared the magnitude of histotripsy-induced intratumoral T cell infiltration with that of radiation and thermal ablation. Ten days after inoculation with subcutaneous B16GP33 melanoma tumors, C57BL/6 mice were treated with no therapy or single unfractionated treatments of 15 Gy external beam radiation, radiofrequency ablation, or histotripsy. As shown in figure 1H, flow cytometric analysis of tumor-infiltrating lymphocytes isolated 10 days after treatment demonstrated that histotripsy alone resulted in significant increases in intratumoral CD8+ T cell populations.

To explore the ability of histotripsy to release intact tumor antigens capable of stimulating CD8+ T cell responses, subcutaneous B16F10 and B16GP33 melanoma tumors were explanted immediately after sham or histotripsy ablation performed 10 days after inoculation. Sham-ablated tumors were subjected to five cycles of rapid freezing and heating (consisting of exposure to liquid nitrogen followed by exposure to a 60°C water bath). CD8+ T cells isolated from splenocytes of C57BL/6 mice 8 days after LCMV infection (at which time a small percentage of CD8+ T cell populations are specific for the LCMV peptide GP33) were exposed to media (negative control), B16F10 or B16GP33 tumors treated with freeze-heat, B16F10 or B16GP33 tumors treated with histotripsy, or 0.01 µg/mL GP33 peptide (positive control) in the presence of IL-2 and brefeldin. As shown in figure 2, activation of CD8+ T cells (as indicated by intracellular interferon gamma (IFNγ) expression) was only observed after exposure to B16GP33 melanoma tumors treated with histotripsy. The magnitude of this CD8+ T cell stimulation approached the level of stimulation seen with direct exposure to GP33 peptide (positive control), indicating that non-thermal disruption of tumors by histotripsy was capable of releasing preserved tumor antigens capable of stimulating CD8+ T cell responses. The absence of CD8+ T cell stimulation by B16F10 tumors (which do not express GP33) treated with histotripsy confirmed the antigen-specific nature of this response.

We examined the effect of histotripsy ablation on other lymphoid compartments beyond the treated tumor. Ten days after sham or histotripsy ablation of B16GP33 melanoma tumors, ipsilateral inguinal lymph nodes were harvested as TDLN, and contralateral axillary lymph nodes were harvested as NTDLN. As shown in online supplementary figure 1, flow cytometric analysis identified slightly higher numbers of CD8+ T cells specific for the weakly immunogenic tumor antigen GP33 within TDLN but not NTDLN after histotripsy, but this did not reach statistical significance. However, analysis of splenocytes did identify significantly higher numbers of circulating GP33-specific CD8+ T cells after histotripsy as compared with controls, radiation, and radiofrequency ablation (figure 3A,B). Analysis of CD4+/CD25+/FoxP3+ cells demonstrated a favorable skewing of tumor-specific CD8+ T cell:regulatory T cell ratios after histotripsy (figure 3C).

Because of the unique ability of histotripsy to trigger significant local and systemic tumor-directed CD8+ T cell responses, we evaluated the effect of histotripsy on distant tumor sites. Ten days after inoculation with bilateral subcutaneous flank B16GP33 melanoma tumors, C57BL/6 mice were treated with no therapy or with single unfractionated treatments of 15 Gy external beam radiation, radiofrequency ablation, or histotripsy ablation of unilateral tumors. Ten days after treatment, contralateral (non-treated) tumors were harvested for lymphocyte assays. As shown in figure 4A,B, flow cytometric analysis identified significantly higher levels of abscopal CD8+ T cell infiltration into distant tumor sites after contralateral histotripsy ablation, but not after radiation therapy or radiofrequency ablation. Comparison between bilateral tumors using immunohistochemistry confirmed minimal infiltration of CD8+ T cells in control mice (figure 4C). Whereas tumors treated with histotripsy ablation exhibited brisk CD8+ T cell infiltration along the periphery of ablation zones, abscopal CD8+ T cell infiltration in contralateral tumors after histotripsy ablation was qualitatively different, exhibiting a more diffuse and homogeneous pattern (figure 4D,E). CD8+ T cell infiltration on days 3, 5 and 7 after histotripsy was analyzed by immunohistochemistry. In control tumors, CD8+ T cell numbers were low and decreased over time; in histotripsy-ablated tumors, CD8+ T cell numbers increased dramatically from day 3 to day 5, and remained high on day 7. The abscopal effect in contralateral tumors after histotripsy was characterized by a more gradual increase in CD8+ T cell numbers (figure 4F). Of note, the ability of unilateral histotripsy ablation to promote distant CD8+ TIL was associated with a small but significant inhibition of contralateral tumor growth. In contrast, no abscopal inhibition of contralateral tumor growth was observed with radiation or radiofrequency ablation (figure 4G–I). Furthermore, despite the presence of a contralateral untreated tumor, survival studies indicated that the abscopal impact of unilateral histotripsy ablation was associated with a statistically significant survival advantage to mice bearing bilateral B16GP33 melanoma tumors (figure 4J).

Given the evidence of abscopal immune stimulation, we tested the ability of local histotripsy ablation to inhibit growth of distant tumors. C57BL/6 mice were inoculated with unilateral subcutaneous injections of B16GP33 melanoma cells on day 0 and with intravenous injections of B16GP33 melanoma cells on day 3 to establish flank and pulmonary melanoma tumors, respectively. On day 10, mice were treated with or without histotripsy ablation of flank tumors; on day 20, mice were euthanized for visual quantitation of pulmonary metastases. As shown in figure 5A,B, significantly fewer pulmonary metastases were observed in mice treated with histotripsy ablation of flank tumors. Immunohistochemical analysis demonstrated smaller tumor size and marked CD8+ T cell infiltration among pulmonary metastases after flank tumor histotripsy ablation (figure 5C).

We hypothesized that the immunostimulatory effects of histotripsy could be associated with an ability to promote the release of intratumoral HMGB1. As illustrated in figure 1F, histotripsy tumor ablation was followed by an intense inflammatory infiltrate of NK cells, DCs, neutrophils and macrophages. As shown in figure 6A–C, flow cytometric analysis of splenocytes on day 10 after histotripsy demonstrated a significant increase in circulating NK cells and a trend toward higher percentages of circulating neutrophils. One day after histotripsy treatment, immunofluorescence assays identified apoptotic changes within treated tumors, as evidenced by nuclear condensation; this was associated with marked translocation of CRT from the endoplasmic reticulum to the plasma membrane as compared with untreated controls (figure 6D). On days 1 and 7 after histotripsy treatment, immunofluorescence assays demonstrated a significant extranuclear translocation of HMGB1 within the intratumoral microenvironment compared with untreated controls (figure 6E, online supplementary figure 2). By day 7 after histotripsy treatment, ELISA (figure 6F) measured significant increases in serum levels of HMGB1 compared with untreated controls.

We tested the ability of histotripsy-associated antitumor immune stimulation to work cooperatively with conventional immunotherapy. C57BL/6 mice were inoculated with bilateral subcutaneous B16GP33 melanoma tumors, then treated with sham or intraperitoneal injections of anti-CTLA-4 mAb on days 6, 9, and 12, and sham or histotripsy ablation on day 7. As shown in figure 7A, the systemic ability of checkpoint inhibition to control tumor growth was slightly stronger than the abscopal effect of histotripsy ablation. When used together, the addition of contralateral histotripsy significantly augmented the therapeutic efficacy of checkpoint inhibition. As demonstrated in figure 7B,C, checkpoint inhibition, unilateral histotripsy, and combination therapy were all associated with significant increases in CD8+ tumor-infiltrating lymphocytes in contralateral, non-ablated tumors; however, when quantified as numbers of infiltrating CD8+ T cells normalized to tumor volume, the addition of unilateral histotripsy significantly increased CD8+ TIL concentration compared with checkpoint inhibition alone. To verify that histotripsy was able to potentiate immune responses against less immunogenic tumors, C57BL/6 mice were inoculated with bilateral subcutaneous Hepa1-6 hepatocellular carcinoma tumors, then treated with sham or intraperitoneal injections of anti-CTLA-4 mAb on days 3, 6, 9 and 12, and sham or histotripsy ablation on day 10. As shown in figure 7D, contralateral histotripsy stimulated abscopal antitumor immune effects that produced a similar enhancement of checkpoint inhibition immunotherapeutic efficacy in this more poorly immunogenic tumor model.