WHAT IS ALREADY KNOWN ON THIS TOPIC

Adaptive cellular therapy (ACT) has been successful in the treatment of hemopoietic malignancies; however, the complex immunosuppressive tumor microenvironment of solid cancers actively inhibits T-cell trafficking into the tumors, which greatly limits the broad application of ACT.

HOW THIS STUDY MIGHT AFFECT RESEARCH PRACTICE OR POLICYBackground

Adoptive cellular therapy (ACT) is a promising immunotherapy approach that involves modification and infusion of a patient’s own immune cells to enhance their ability to recognize and destroy cancer cells.1 2 ACT can be classified as chimeric antigen receptor (CAR) T-cell (CAR-T), T-cell receptor-T cell (TCR-T), and tumor-infiltrating lymphocytes (TILs) therapies.3 4 CAR-T therapy has shown remarkable success in treating certain types of hematopoietic cancers, particularly acute lymphoblastic leukemia and diffuse large B-cell lymphoma.5 6 ACT has shown great promise in the treatment of certain solid tumors. Notably, TIL therapy has been associated with profound responses in patients with melanoma, with reported response rates as high as 50%.7 Recently, two ACTs for solid tumors have received Food and Drug Administration approval: TIL for advanced melanoma8 and MAGE-A4 TCR therapy for synovial sarcoma.9 Beyond these, TIL therapy has demonstrated modest success in various solid organ malignancies, including cervical cancer,10 HPV-associated cancers,11 cholangiocarcinoma,12 breast cancer,13 lung cancer,14 and colon cancer,15 all of which highlight the broad potential of ACT. However, the application of ACT in solid tumors remains constrained by several well-known challenges to CAR T-cell therapy, ranging from the lack of suitable target antigens, and limited number of TILs for expansion, to the poor quality of expanded T cells that fail to reach the tumor.2

In addition to T-cell intrinsic factors, the complex and dynamic immunosuppressive tumor microenvironment (TME) characteristic of solid tumors actively limits T-cell trafficking and functions, posing further challenges for ACT.16 In solid cancers, the immature and abnormal tumor vasculature is detrimental to both effector T-cell infiltration and functions in tumor tissues. Blood vessels within the tumor reduce the expression of adhesion molecules such as intercellular adhesion molecule 1 (ICAM1) and vascular cell adhesion molecule 1 (VCAM1), which are essential for T-cell extravasation.17 18 Furthermore, dysfunctional tumor vasculature often leads to hypoxia, which limits the effecter function of infiltrating T cells by recruiting immunosuppressive cells into tumors, including CD4+regulatory T (Treg) cells and myeloid-derived suppressor cells (MDSCs).19 20 Supporting that, targeting angiogenic factors to normalize blood vessels has been broadly explored to enhance T-cell infiltration and improve the efficacy of T cell-mediated immunotherapy.10 21 22

Our previous study described a CD93/IGFBP7 axis that contributes to tumor vascular abnormality, and CD93 blockade with a blocking monoclonal antibody (mAb) normalizes tumor vascular functions and improves immune checkpoint blockade (ICB) therapy.23 Here, we assessed the capacity of CD93 blockade in promoting ACT in solid tumors. We found that increased effector T-cell infiltration by anti-CD93 occurred selectively in the tumor model. In a clinical-relevant CAR-T therapy setting, we were able to observe a synergistic antitumor effect when combining CD93 blockade with CAR-T transfer.

ResultsCD93 blockade improves adaptive transgenic T-cell therapy in mouse melanoma

We investigated whether CD93 blockade promotes ACT in a B16-OVA/OT-1 mouse tumor model, which is a well-established preclinical mouse model. In this model, the B16 melanoma cell line is modified to express the model antigen ovalbumin (OVA), which allows for the use of OT-1 transgenic T cells that recognize OVA.24 Once tumors became palpable, we intravenously transferred black B6 mice with pre-activated OT-1 T cells, followed by anti-CD93 mAb treatment twice a week (figure 1A). In this model, the transfer of 5 million pre-activated OT-1 T cells alone partially attenuated B16-OVA tumor progression; control mice all died within 21 days whereas only 40% of mice transferred with OT-1 succumbed to tumor progression. The inclusion of anti-CD93 mAb was able to further improve the antitumor effect of OT-1 cytotoxic T lymphocyte (CTL) transfer (figure 1B). As a result, about half of the mice in the combinatory group survived over 30 days after tumor inoculation, with two mice becoming tumor-free (figure 1C).

Figure 1

CD93 blockade facilitates OT-1 CTL therapy in tumors. (A–G) C57BL/6J (CD45.2) mice with subcutaneous B16-OVA tumor were adoptively transferred with OT-1 CTL (CD45.1, purple arrow), with or without the treatment of anti-CD93 mAb (yellow arrow). (A) Schema of experiment. (B) Tumor growth (two-way ANOVA test with Dunnett’s post hoc tests, n=5) and (C) mouse survival (Kaplan-Meier curves with log-rank Mantel-Cox tests, n=10) were monitored over time. (D) Tumor weight (from B) was measured at day 15 (one-way ANOVA test with Tukey post hoc tests, n=5). (E–G) Intratumoral OT-1 T cells in B16-OVA tumors under the treatment of control or anti-CD93 were assessed by flow cytometry. (Unpaired Student’s t-test, n=8–10). (H–J) Mice with established B16F10 tumors were treated with pmel-1 CTL (purple arrow) with or without anti-CD93 mAb (yellow arrow). (H) Schema of experiment. (I) Tumor growth (two-way ANOVA test with Dunnett’s post hoc tests, n=10) and (J) mouse survival (Kaplan-Meier curves with log-rank Mantel-Cox tests, n=10) were monitored over time. ANOVA, analysis of variance; CTL, cytotoxic T lymphocyte; ip, intraperitoneal; iv, intravenous; mAb, monoclonal antibody; OVA, ovalbumin; sc, subcutaneous; TIL, tumor-infiltrating lymphocyte.

We collected tumor tissues to analyze immune cell composition after 15 days of tumor inoculation. Control tumors reached over 900 mg, and tumors transferred with OT-1 were about 300 mg whereas the average tumor weight in the combined group was less than 100 mg (figure 1D). The treatment of anti-CD93 led to a significant increase in transferred CD45.1+ OT-1 T cells within the CD8+T cell compartment of tumors, and the numbers of intratumoral OT-1 T cells doubled on anti-CD93 treatment (figure 1E, online supplemental figure S1). In addition, OT-1 T cells within anti-CD93 treated tumors exhibited more effector T-cell phenotype (CD44+CD62L−) (figure 1F), and they contained more granzyme B-producing cells proportionally (figure 1G). Analysis of host immune cells (CD45.2+) within tumors also confirmed that anti-CD93 therapy triggers a favorable TME (online supplemental figure S2). Similarly, the treatment of an IGFBP7 mAb that blocks the CD93-IGFBP7 interaction23 was able to enhance the antitumor effect by transferring OT-1 CTL (online supplemental figure S3A–C). Together, our results supported the role of the CD93-IGFBP7 axis in limiting antitumor T-cell response.

OT-1 TCR transgenic T cells have a far stronger affinity to recognize chicken ovalbumin antigen while those expanded TILs in the clinic typically have only weak or moderate binding affinity to tumor antigens.25 Therefore, we used pmel-1 transgenic T cells for transfer in B16F10 tumors to better assess the impact of CD93 blockade on ACT. Pmel-1 CD8+ T cells express a transgenic TCR that weakly recognizes gp10025-33 in the context of H-2Db expressed by B16F10 melanoma cells.26 We sublethally irradiated B16F10 tumor-bearing mice right before T-cell transfer to make room for subsequently transferred pmel-1 CTL (figure 1H). Consistent with what we observed in the B16-OVA/OT-1 model, anti-CD93 greatly improved the antitumor effect by transferring pmel-1 T cells (figure 1I) and thereby significantly elongated the survival of mice transferred with pmel-1 CTL (figure 1J).

CD93 blockade promotes T-cell migration leading to increased intratumoral T-cell infiltration

Increased T-cell infiltration in tumors can be caused by improved effector T-cell homing or enhanced T-cell division.27 Using the B16-OVA/OT-1 model, we investigated the mechanism by which anti-CD93 improved antitumor response by ACT. To do that, we labeled CD8+OT-1 T cells (CD45.1+) with Carboxyfluorescein succinimidyl ester (CFSE) and transferred them into mice (CD45.2+) with established B16-OVA tumors. 4 days later, we isolated tumor-draining lymph nodes (dLNs) to assess OT-1 T-cell proliferation based on CFSE dilution (online supplemental figure S4A). As shown in online supplemental figure S4B, we did not observe any change in divisions of transferred OT-1 T cells in response to anti-CD93 treatment. This implied that CD93 blockade does not affect CD8+T cell priming. We then transferred pre-activated OT-1 T cells into tumor-bearing mice which were treated with BrdU, enabling us to track cell proliferation at different time points in vivo (figure 2A). When we enumerated transferred OT-1 T cells 4 or 7 days post-transfer, we found a selective increase of OT-1 T cells in tumors and dLNs under the treatment of anti-CD93, but not in non-dLNs (figure 2B). Assessment of BrdU incorporation indicated that anti-CD93 did not affect transferred OT-1 T-cell proliferation either in the tumors or dLNs (figure 2C). Therefore, increased OT-1 CTL accumulation in the tumors by CD93 blockade is likely mediated by improved T-cell homing.

Figure 2

CD93 blockade promotes effector T-cell homing to the tumor microenvironment. (A–C) C57BL/6J (CD45.2) mice were subcutaneously inoculated with B16-OVA tumor cells. Once the tumor became palpable, half of the mice were treated with anti-mCD93 mAb every 3 days. On day 0, each mouse intravenously received 1×107 CD45.1+ pre-activated OT-1 CTL. (A) Schema of experiment. (B) Transferred OT-1 T cells in tumors, draining lymph nodes, and non-draining lymph nodes were determined by flow cytometry (representative flow plots are from day 4. Unpaired Student’s t-test, n=5). (C) On 3 and 6 days of OT-1 transfer, mice were injected with BrdU to assess OT-1 T-cell proliferation by flow cytometry (representative flow plots are from day 4. Unpaired Student’s t-test, n=5). (D) B16-OVA tumor-bearing mice were treated with control or anti-CD93 antibodies. Tumor tissues collected on day 14 were stained and quantified for vascular ICAM1 and VCAM1, unpaired Student’s t-test, n=5. (E–H) B16-OVA tumor-bearing mice were treated with control or anti-CD93 or anti-CD93 plus ICAM1 and VCAM1 neutralizing antibodies. (E) Treatment schedule. (F) Tumor growth was monitored over time (two-way ANOVA test with Dunnett’s post hoc tests, n=5). (G) Tumor weight at day 14 is indicated, one-way ANOVA test with Tukey post hoc tests, n=5. (H) Densities of intratumoral immune cells were analyzed by flow cytometry, one-way ANOVA test with Tukey post hoc tests, n=5. Scale bar 50 µm. ANOVA, analysis of variance; CTL, cytotoxic T lymphocyte; dLNs, tumor-draining lymph nodes; ip, intraperitoneal; iv, intravenous; LN, lymph node; mAb, monoclonal antibody; ndLN, non-draining lymph nodes; OVA, ovalbumin; sc, subcutaneous; TIL, tumor-infiltrating lymphocyte.

Effector T-cell infiltration into tumors is regulated by intratumoral chemokines and adhesion molecules expressed by endothelial cells (ECs).17 28 We found that the treatment of either anti-CD93 or anti-IGFBP7 led to significant upregulation of ICAM1 and VCAM1 in tumor-associated ECs (figure 2D, online supplemental figure S5). Interestingly, increased infiltrating T cells by anti-CD93 primarily located around these mature tumor vessels (online supplemental figure S6). We treated mice bearing B16-OVA tumors with neutralizing mAbs for ICAM1 and VCAM1 to further assess their involvement in anti-CD93-promoted TIL infiltration (figure 2E). As expected, CD93 blockade greatly inhibited the progression of subcutaneous B16-OVA melanoma. The addition of ICAM1 and VCAM1 mAbs greatly reduced the antitumor effect by anti-CD93 (figure 2F,G). Enumerating intratumoral immune cells further demonstrated that blockade of ICAM1 and VCAM1 completely abrogated the increased intratumoral T cells by anti-CD93 treatment (figure 2H). Together, our findings support our hypothesis that CD93 blockade promotes intratumoral T-cell infiltration via adhesion molecule upregulation on tumor vasculature.

CD93 blockade selectively facilitates effector T-cell penetration in the tumor setting

The CD93/IGFBP7 pathway is upregulated in tumor vasculature in response to hypoxia.23 However, in normal tissues, blood vessels function properly and the expression of CD93 and IGFBP7 is either weak or undetectable,23 raising the question of whether the promotion of T-cell infiltration by CD93 blockade is uniquely confined to the tumor setting. In a T-cell transferred diabetic mouse model, the infusion of a large number of OT-1 T cells into Rip-OVA mice causes islet damage and thereby diabetes.29 Anti-CD93 treatment in this model did not affect serum glucose level (figure 3A) or the onset of diabetes (figure 3B). Consistently, there was no difference in transferred OT-1 cells between control and anti-CD93-treated groups, in either the spleen or pancreas (figure 3C,D). In contrast to this though, the same anti-CD93 treatment had a profound impact when OT-1 T cells were transferred into 10-week-old Rip-TAG-OVA mice, which were generated by crossing Rip-TAG2 mice with Rip-OVA mice. Similar to Rip-TAG2 mice, these double-transgenic mice develop OVA-expressing insulinoma in the pancreas at 9 weeks old.30 31 Administration of anti-CD93 mAb had no impact on OT-1 T-cell infiltration in normal pancreases of Rip-OVA mice; however, the same treatment greatly increased transferred OT-1 T-cell penetration in the pancreatic tumors of Rip-TAG-OVA mice (figure 3D). Consistently, blood vessels within tumors from Rip-TAG-OVA mice expressed high levels of CD93, which was in sharp contrast to a negligible level of CD93 in normal pancreatic blood vessels from Rip-OVA mice (figure 3E). Normal pancreatic blood vessels in Rip-OVA mice constitutively express a high level of adhesion molecule ICAM1. However, in Rip-TAG-OVA tumors, ICAM1 in intratumoral blood vessels is reduced or even lost while VCAM1 became detectible in some tumor vasculature (figure 3F). The treatment of anti-CD93 in Rip-OVA mice had no impact on the expressions of these two adhesion molecules of pancreatic blood vessels; in contrast, the same treatment in Rip-TAG-OVA mice was able to increase the expressions of both ICAM1 and VCAM1 in tumor vasculature (figure 3F), which was consistent with what we observed in subcutaneous B16-OVA tumors (figure 2D). Together, our results supported that CD93 blockade selectively promotes T-cell infiltration specifically within tumors characterized by dysfunctional or immature vasculature.

Figure 3

CD93 blockade promotes effector T-cell infiltration selectively in the tumor microenvironment. (A–C) Rip-OVA mice transferred with OT-1 T cells (1×107) were treated with or without anti-CD93 mAb. Serum glucose (A) and the onset of diabetes (B, Kaplan-Meier curves with log-rank Mantel-Cox tests, n=4–5) were monitored over time. (C) 15 days after transfer, the percentages of OT-1 T cells were determined by flow cytometry (unpaired Student’s t-test, n=4–5). (D–F) Rip-OVA or Rip-TAG-OVA mice were transferred with pre-activated OT-1 cells before being treated with control or anti-CD93 mAb. Pancreatic tissues were collected on day 5 for sections and IF staining analysis. (D) Transferred OT-1 cells in the pancreases were determined by CD45.1 staining (unpaired Student’s t-test, n=5). (E) CD93 expression in pancreatic blood vessels from RIP-OVA and Rip-TAG-OVA mice was determined by IF staining. (F) The expressions of ICAM1 and VCAM1 in blood vessels (CD31) in the pancreas were quantified (unpaired Student’s t-test, n=5). Scale bar 50 µm. White dash lines indicate islets. I, islets, IF, immunofluorescence; mAbs, monoclonal antibodies; N, normal; OVA, ovalbumin; T, tumor.

Targeting the CD93 pathway improves CAR-T therapy in a preclinical mouse tumor model

We further evaluated the impact of anti-CD93 on CAR-T therapy in solid tumors (figure 4A). We subcutaneously implanted mice with B16-mCD19 tumor cells, which were transduced to express surface mouse CD19 antigen. 6 days later, mice were sublethally irradiated before receiving 0.2 million mCD19 CAR-T cells, which were generated and expanded in vitro to recognize mouse CD19 antigen.32 Without treatment, all mice succumbed to tumor progression within 26 days. The infusion of either CAR-T cells or anti-CD93 alone modestly slowed down tumor progression, as shown by tumor growth and mouse survival (figure 4B,C). The combination of anti-CD93 and CAR-T was able to further inhibit tumor progression; as a result, the combination therapy was able to extend the survival of most mice over 35 days, with 20% of mice eventually becoming tumor-free (figure 4B,C).

Figure 4

CD93 blockade improves CAR-T therapy in a preclinical mouse tumor model. B16-mCD19 melanoma cells were subcutaneously inoculated into C57BL/6J mice. After irradiation, mice were randomized into four different treatment groups based on tumor sizes. Half of the mice received 2×105 mCD19 CAR-T cells intravenously (purple arrow). Anti-mCD93 mAb or control treatments (yellow arrow) were given every 3 days right after the CAR-T transfer. (A) Schema of experiment. (B) Tumor growth (two-way analysis of variance test with Dunnett’s post hoc tests, n=5) and (C) mouse survival (Kaplan-Meier curves with log-rank Mantel-Cox tests, n=6–9) were monitored over time. On day 17, CAR-T cells in tumors (D) or spleens (E) were determined by flow cytometry (unpaired Student’s t-test, n=5). The percentages of granzyme B-expressing cells in CD8+ (F) and CD4+ (G) CAR-T populations were compared between control and anti-CD93-treated groups (unpaired Student’s t-test, n=5). CAR, chimeric antigen receptor; ip, intraperitoneal; iv, intravenous; mAb, monoclonal antibody; sc, subcutaneous; TIL, tumor-infiltrating lymphocyte.

10 days after CAR-T transfer, we sacrificed mice to quantify transferred CAR-T cells in response to CD93 blockade by flow cytometry (online supplemental figure S7). The sublethal irradiation procedure largely abrogated host T-cell immune response, as CAR-T cells consisted of 80–90% of total CD3+T cells within tumors (online supplemental figure S7). Anti-CD93 selectively increased the numbers and percentages of CAR-T cells (CD45.1) in tumors (figure 4D), but not in spleens (figure 4E). Granzyme B-expressing cells in CD8+ (figure 4F) and CD4+CAR T cells (figure 4G) were both found to be increased in tumors treated with anti-CD93 mAb. Therefore, CD93 blockade improves both the infiltration and function of CAR-T cells in solid tumors.

Anti-CD93 and ACT synergistically promote tumor vascular maturation

We assessed vascular changes in B16-mCD19 tumors under the treatments of anti-CD93 together with or without CAR-T transfer. Effector T cells in solid tumors are known to secrete cytokines like interferon-γ to promote tumor vascular maturation.33 In the CD19 CAR-T cell transfer model, the treatment of anti-CD93 or CAR-T cell transfer alone was able to increase pericyte coverage on tumor vasculature, as assessed by αSMA immunofluorescent staining (figure 5A,B). The combination therapy was able to further improve the maturation of tumor vasculature. As a result, virtually all intratumoral blood vessels in the combination group were covered with αSMA-expressive pericytes (figure 5A,B).

Figure 5

Anti-CD93 and T-cell transfer synergistically promote tumor vascular maturation. B16-mCD19 tumor tissues from mice under the therapy of anti-CD93 and CAR-T were stained and analyzed for tumor vascular maturation. (A, B) Pericyte coverage on tumor vasculature was assessed by αSMA staining. The expressions of adhesion molecule ICAM1 (C, D) and VCAM1 (E, F) on tumor vasculature were quantified. Tumor blood vessels were identified by CD31 staining. One-way analysis of variance test with Tukey post hoc tests, n=5. Scale bar 50 µm. CAR, chimeric antigen receptor.

We further confirmed the synergistic impact of the combinatory treatment on the expression of adhesion molecules that are crucial for T-cell extravasation. The transfer of either mCD19 CAR-T cells or the administration of anti-CD93 was able to increase the percentages of ICAM1-expressing and VCAM1-expressing ECs within the tumors (figure 5C–5F). The combinatory treatment further increased the percentages of blood vessels positive for these two adhesion molecules that support T-cell extravasation (figure 5C–5F).

Discussion

Our study explores the potential of CD93 blockade to enhance ACT for solid tumors. By targeting CD93, we demonstrated that vascular normalization significantly enhances T-cell infiltration, thereby improving ACT efficacy in a tumor-setting-specific manner. This approach represents a promising strategy for overcoming key challenges in the treatment of solid tumors, particularly by addressing the issue of poor immune cell infiltration.

One of the main hurdles for ACT in solid cancers is that few transferred T cells reach the tumor site to execute their antitumor function successfully. Intratumoral blood vessels are detrimental to T-cell extravasation as they downregulate the expression of ICAM1, an adhesion molecule that is indispensable for T-cell extravasation.34 Further, hypoxia in solid tumors causes the accumulation of immunosuppressive MDSCs and CD4+Treg cells, which further limits effector T-cell functions within the tumors. We have found that the treatment of anti-CD93 promoted vascular function to alleviate hypoxia in tumors, which led to reduced MDSCs and improved T-cell functions within the tumors.23 Here we found that vascular normalization by anti-CD93 or IGFBP7 mAb increased the expression of ICAM1 and VCAM1. Supporting this, the inclusion of mAbs for these two adhesion molecules completely abolished the effect of anti-CD93 on intratumoral T-cell infiltration. It remains to be unveiled how CD93 blockade impacts the downstream signaling pathways which lead to vascular maturation in tumors. Nevertheless, increased T-cell infiltration by anti-CD93 further promoted tumor vascular maturation, which triggers a positive feedback loop in improving vascular functions.

Our study also reveals a selective impact of CD93 blockade on effector T-cell infiltration in the TME. We used an autoimmune diabetic mouse model, in which healthy Rip-OVA transgenic mice were transferred with large quantities of OT-1 cells. The expression of CD93 in normal pancreatic blood vessels is almost undetectable while CD93 is highly expressed in tumor-associated vasculature in the pancreas. It is not surprizing to see a selective impact of CD93 blockade in TIL infiltration. Furthermore, normal pancreatic blood vessels constitutively express a high level of ICAM1, which tends to be easier for activated OT-1 T cells to transport, compared with the pancreas with insulinoma. The selective effect of anti-CD93 on T-cell infiltration in the tumor setting would suggest that CD93 blockade is a safer approach compared with vascular endothelial growth factor (VEGF) inhibition in promoting ACT in solid cancers. Besides targeting CD93, IGFBP7 blockade could be a valuable approach for promoting ACT in cancer, although it may be less effective in inhibiting tumor progression likely due to the presence of additional ligands for CD93, namely MMRN2.17 35 However, the selective expression of IGFBP7 in the TME would support its blockade as an interesting alternative to facilitating ACT.

Several limitations and future directions must be considered to fully understand the translational potential of CD93 blockade in ACT therapy. Our study is limited by the exclusive use of implanted mouse melanoma models. Whether CD93 blockade can enhance ACT in other solid tumors, such as pancreatic cancer with dense stroma, requires further investigation. In addition, the models employed in our study have not fully recapitulated clinic settings. As a result, we are unable to assess the potential toxicity of CD93 blockade in ACT for clinical usage. We used expanded transgenic T cells in B16 tumors and used CD19 antigen in the B16/CAR-T tumor model. While these models provided useful insights into the antitumor efficacy of CD93 blockade, they lack the complexity and heterogeneity characteristic of human tumors. Future studies with CD93 blockade would be performed in ACT models with expanded TILs and clinically relevant antigens.

In conclusion, our study identifies a role for the CD93-IGFBP7 axis in limiting effector T-cell infiltration in tumors selectively. Vascular normalization such as CD93 blockade can be a valuable approach to improving ACT in solid tumors.

MethodsMice and tumor models

C57BL/6 (#000664) was purchased from the Jackson Laboratory (Bar Harbor, Maine, USA). C57BL/6-Tg (Ins2-TFRC/OVA) 296Wehi/WehiJ (Rip-OVA, #005431) mice and C57BL/6-Tg (TcraTcrb) 1100Mjb/J (OT-1) mice were obtained from the Jackson Laboratory and were housed at the animal facility of the University of Colorado Anschutz Medical Campus. Rip-TAG-OVA mice were bred in-house by crossing Rip-OVA females with Rip-TAG2 males, which were rederived with cryopreserve sperm obtained from the National Cancer Institute Mouse Repository. Mice aged 6–10 weeks were used for all experiments. All animal protocols and procedures conducted were approved by the Institutional Animal Care and Use Committee of the University of Colorado Anschutz Medical Campus (Protocol #00132).

B16F10 and B16-OVA mouse melanoma cell lines were maintained in the laboratory, and the B16 tumor cell transduced to express mouse CD19 (B16-mCD19) was obtained from Dr Tuoqi Wu (the University of Texas Southwestern Medical Center). All cultured cell lines were tested for Mycoplasma contamination on receipt and monitored for phenotypic consistency throughout the study by assessing morphology and growth curves regularly.

Mouse cancer cell lines were cultured in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 2% 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), and 1% penicillin-streptomycin. Tumor cells were harvested and prepared as a single-cell suspension in Hanks' Balanced Salt Solution (HBSS) buffer at the appropriate density to ensure a total injection volume of 100 µL. For all tumor models, cells were inoculated subcutaneously into the right flank at the dosages indicated in the figures. Once tumors became palpable, tumor-bearing mice were randomized into groups based on tumor size. Tumor dimensions were measured every other day using calipers to determine length and width, and tumor volume was calculated using the formula 1/2×(length×width2). Mice were euthanized if an ulcerated tumor was observed, if the tumor length exceeded 20 mm, or if the experiment reached its endpoint (tumor volume ≥1,000 mm3 was considered a terminal point in survival studies). Tumor tissues were collected and processed according to the requirements of downstream analyses (frozen or formalin-fixed paraffin-embedded (FFPE) sections, flow cytometry).

Adopted cell preparation and transfer

Pre-activated OT-1 T cells (CD45.1+) were generated by stimulating OT-1 splenocytes with OVA (257–264) peptide at a concentration of 20 ng/mL for 72 hours before being transferred intravenously. Pmel-1 T cells were pre-activated with mgp100 peptide (EGSRNQDWL) at a concentration of 1 µg/mL for 72 hours prior to being transferred. The mCD19-CAR lentivirus was produced using the Lenti-X system and an mCD19-CAR construct32 was obtained from Addgene (Plasmid #107226). CD45.1+T cells were transduced with mCD19-CAR lentivirus twice before intravenous transfer. All transferred cells were counted and analyzed by flow cytometry prior to transfer to validate activation status (OT-1 and pmel-1) or transduction efficiency (mCD19 CAR-T). Mice bearing B16F10 or B16-mCD19 tumors were irradiated with 5 Gy the day before receiving pmel-1 T cells or mCD19-CAR-T cells. Cells were prepared as a single-cell suspension in phosphate-buffered saline buffer to ensure a total injection volume of 100–200 µL for each mouse before tail vein injection, and the amount of the transferred T cells was indicated in the figures. mCD19 CAR-T cells were treated with interleukin (IL)-15 at 10 ng/mL for 48 hours and IL-7 at 10 ng/mL for 24 hours before transfer. No cytokine was used in our OT-1 or pmel-1 transgenic T-cell transfer experiments.

Antibodies for in vivo studies

Anti-mCD93 (7C10) and anti-mIGFBP7 (2C6) antibodies were produced in-house from hybridomas generated in the laboratory.23 Mouse ICAM1 (YN1/1.7.4, #BE0020-1) and VCAM1 (M/K-2.7, #BE0027) neutralizing monoclonal antibodies for in vivo studies were purchased from Bio X Cell. All antibodies were administered intraperitoneally according to the dosage and schedule indicated in the figures.

Immunofluorescent staining

After tissue samples were frozen in optimum cutting temperature mounting fluid, 8 µm thick frozen sections were prepared. The sections were air-dried, fixed in −20°C acetone, and either proceeded to the next staining steps or preserved at −80°C. FFPE blocks and sections were processed and prepared by the histology core at the University of Colorado Anschutz Medical Campus. FFPE sections were rehydrated and treated in Tris-EDTA buffer (pH 9.0 or pH 5.0) at 110°C for 15 min before proceeding with downstream staining steps. For immunofluorescent staining, tissues were blocked with 2.5% goat or donkey serum, then incubated with primary antibodies at the appropriate concentration overnight at 4°C. Secondary antibodies were incubated for 1 hour, followed by 4',6-diamidino-2-phenylindole (DAPI) counterstaining for 10 min, both at room temperature. After clearing and mounting, immunofluorescent images were captured using a Zeiss Axio Observer upright microscope and analyzed with 3i SlideBook software (V.6.0.23).

The following antibodies were used for immunofluorescent staining: mouse CD3 (Cell Signaling, 4443T), mouse CD31 (FFPE tissue, R&D Systems, AF3628; frozen tissue, BioLegend, clone 390), CD45.1 (BioLegend, 110702), mouse ICAM1 (Abcam, AB179707), mouse VCAM1 (Cell Signaling, 32653S), mouse Alpha-Smooth Muscle Actin mAb (1A4, eBioscience, Cat # 50-9760-82), mouse CD93 (Invitrogen, PA5-81345), and mouse CD8 (Cell Signaling, 98941T).

Flow cytometry

Single-cell suspensions were prepared by cutting tumor tissues into small pieces less than 1 mm in diameter and digesting them with 50 µg/mL Liberase (Roche Diagnostics Corporation, Indianapolis, Indiana, USA) in Roswell Park Memorial Institute (RPMI)-1640 medium for 45 min with shaking at 37°C. After digestion, single-cell suspensions were collected by passing them through a 70 µm cell strainer (Fisherbrand, Cat# 22363548). Red blood cells were removed by incubating with 2 mL ACK lysis buffer at room temperature for 3 min. Cells were then washed with a culture medium for surface and intracellular staining. LEAF anti-mouse CD16/32 (anti-FcγRIII/II receptor, Clone 93) was added for blocking prior to staining. Surface markers were stained using conjugated antibodies for 30 min at 4°C. Viability staining was performed with Ghost Dye Red 780 (Tonbo, 13–0865 T100). For intracellular staining, cells were then fixed and permeabilized with BD Fix/Perm Kit (for intracellular) following the manufacturer’s instructions. Flow cytometric analysis was conducted using a Northern Lights or Aurora cell analyzer (Cytek Biosciences), and data were analyzed with FlowJo software (V.10.9.0, Tree Star). The gating strategy for tumor-infiltrating lymphocytes is shown in online supplemental figure S1, S7. Antibodies for flow cytometry staining were obtained from BioLegend, if not specified.

Statistics

After data collection, all plots and figures were generated using GraphPad Prism software (V.9.5.0). Statistical significance (including mean±SEM) between groups was calculated using appropriate methods, as specified in the figure legends. Differences were considered significant when p values were <0.05.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.

Ethics statementsPatient consent for publication

Not applicable.

Ethics approval

Not applicable.

Acknowledgments

We thank the Animal Core facility at the National Jewish Health for sperm rederivation and the flow cytometry core at the University of Colorado Cancer Center.