Extracellular vesicle-based anti-HOXB7 CD8⁺ T cell-specific vaccination strengthens antitumor effects induced by vaccination against Her2/neu

Abstract

We previously developed an innovative strategy to induce CD8⁺ T lymphocyte-immunity through in vivo engineering of extracellular vesicles (EVs). This approach relies on intramuscular injection of DNA expressing antigens of interest fused at a biologically-inactive HIV-1 Nef protein mutant (Nef^mut). Nef^mut is very efficiently incorporated into EVs, thus conveying large amounts of fusion proteins into EVs released by transfected cells. This platform proved successful against highly immunogenic tumor-specific antigens. Here, we tested whether antigen-specific CD8⁺ T cell immune responses induced by engineered EVs can counteract the growth of tumors expressing two “self” tumor-associated antigens (TAAs): HOXB7 and Her2/neu. FVB/N mice were injected with DNA vectors expressing Nef^mut fused to HOXB7 or Her2/neu, singly and in combination, before subcutaneous implantation of breast carcinoma cells co-expressing HOXB7 and Her2/neu. All mice immunized with the combination vaccine remained tumor-free, whereas groups vaccinated with single Nef^mut-fused antigens were only partly protected, with stronger antitumor effects in Her2/neu-immunized mice. Double-vaccinated mice also controlled tumor growth upon a later tumor cell re-challenge. Importantly, co-vaccination also contained tumors in a therapeutic immunization setting. These results showed the efficacy of EV-based vaccination against two TAAs, and represent the first demonstration that HOXB7 may be targeted in multi-antigen immunotherapy strategies.

Introduction

Immunotherapy represents the last frontier in the fight against cancer. Unfortunately, the tumor immunoediting process [1] that occurs in response to several immunotherapeutic strategies targeting a single tumor antigen often leads to tumor escape. This shortcoming is expected to be overcome, for instance, by targeting different tumor antigens simultaneously.

Anticancer immunotherapy seeks to help the immune system to destroy tumor cells through the selective recognition of tumor antigens not expressed by healthy cells. Amongst these potential targets are, for instance, tumor-specific antigens (TSAs) like Human Papilloma Virus (HPV)-E6 and -E7 in cervical tumors, or alpha-fetoprotein in both germ cell tumors and hepatocellular carcinoma. Many TSAs can be encoded also by non-canonical mRNAs arising from epigenetic changes and splicing aberrations (neo-antigens) [2]. Generally, TSAs are not subject to immunologic tolerance, as such, they are recognized by the immune system as “non-self” products. On the other hand, tumor-associated antigens (TAAs), which can be upregulated in cancer cells, are expressed by healthy cells also, as in the case of Her2/neu, MUC-1, EGFR, and melanoma-associated antigens, and are subjected to immunologic tolerance.

Both reactivation and ex novo generation of T cell-driven antitumor activity are the ultimate goals of several anticancer immunotherapies, including those based on immune checkpoint inhibitors (ICIs) [3], chimeric antigen receptor (CAR)-T cells [4], and mRNA-based vaccines [5]. Adverse events, lack of specificity, tumor escape, and huge costs call for new solutions able to render anticancer immunotherapy approaches more safe, effective, and affordable.

The homeobox (HOX) genes encode transcriptional factors controlling the coordinated expression of several genes during the embryonic development [6]. They are clustered in four distinct super-families (A to D) and are overexpressed in several forms of cancer. In particular, HOXB7, a 27 kDa component of the HOXB-subfamily, was found amplified in melanoma [7], leukemia [8], as well as in oral [9], breast [10], colorectal [11], and gastric [12] cancers. In addition, a direct correlation between increased expression of HOXB7 and poor prognosis was formally established [13]. Hence, to contain many forms of tumor, the induction of a strong immune pressure against HOXB7 is expected to be of great utility as part of novel anti-cancer immunotherapies.

Her2/neu is a 185 kilodalton transmembrane, tyrosine kinase receptor belonging to the epidermal growth factor receptor family [14]. Her2/neu has not a known ligand but initiates the intracellular signaling upon formation of heterodimers with other components of the Her family engaging the respective ligand. In humans, both overexpression and genetic alterations of Her2/neu can associate with breast cancer, ovarian, and gastric cancers.

All healthy cells constitutively release extracellular vesicles (EVs), i.e., double-layered lipid vesicles from 50 to 1000–2000 nm of diameter [15]. Their physiological functions relate to the cell-to-cell transfer of proteins, peptides, and both short- and long nucleic acids. We previously found that a biologically inactive HIV-1 Nef mutant, called Nef^mut, incorporates into EVs at quite high levels also when a heterologous protein is fused to its C-terminus [16]. The entry of Nef^mut-engineered EVs into professional antigen-presenting cells (APCs) results in cross-presentation of the EV-incorporated antigens, and activation of antigen-specific CD8⁺ T lymphocytes [17]. In mice, the CD8⁺ T cell immunogenicity of Nef^mut-based EVs was assessed in a first instance by injecting the in vitro engineered EVs [18], and afterward by engineering EVs spontaneously released by muscle cells upon injection of DNA vectors expressing Nef^mut-based fusion products [19].

The Nef^mut-based strategy was successfully applied against transplantable tumors expressing the TSAs HPV-E6 and -E7, whose growth was abolished in the vast majority of vaccinated mice [20]. On the other hand, tumor growth was significantly delayed, but not arrested, when transgenic mice spontaneously developing mammary tumors in view of the MMTV-regulated overexpression of Her2/neu were injected with a DNA vector expressing the Nef^mut-Her2/neu fusion product [21]. We assumed that tumor escape was due to immunoediting events driven by the processes of elimination, equilibrium, and escaping (“the three Es”) [22], and favored by the intrinsic genetic instability of breast cancer cells. Hence, we looked for additional specific antigens in breast cancer cells whose targeting would be instrumental to hinder the tumor escape.

Here, the antitumor effect of double vaccination against HOXB7 and Her2/neu through the EV-based platform was analyzed in a model of subcutaneous tumor implantation of HOXB7-Her2/neu co-expressing cells. Our results hold promise for a possible translation into the clinic.

Materials and methods

Molecular constructs, cell cultures, and transfection

The DNA vectors pTargeT/Nef^mut, pcDNA3.1/Nef^mut-Her2/neuECD (encoding a Nef^mut-fused protein encompassing the sole extracellular domain of Her2/neu, hereinafter Nef^mut-Her2/neu), pTargetT/Nef^mut-HOXB7, and pTargeT/Nef^mut-HOXB7_tr (encoding a truncated isoform of HOXB7 fused with Nef^mut) were already described [21, 23]. The HOXB7 expressing vector (LB7SNeo) was a gift from A. Carè.

Both human embryonic kidney (HEK) 293T cells (ATCC, CRL-11268) and human HLA-A.02 MCF-7 cells (ATCC, HTB-22) were grown in DMEM (Gibco, Thermo Fisher Scientific, Waltham, MA, USA) plus 10% heat-inactivated fetal bovine serum (FBS, Gibco, Thermo Fisher Scientific).

Murine tumor cells, named 305 cells, obtained from spontaneous mammary gland tumors emerging in HOXB7-Her2/neu-double transgenic FVB/N mice [24], were kindly provided by H. Chen, and grown in DMEM/F12 medium (Sigma-Aldrich, St. Louis, MO, USA) containing l-glutamine (2 mM), supplemented with 10% FBS, penicillin (100 U/mL), streptomycin (0.1 mg/mL), human insulin (10 μg/mL) and fungizone (250 ng/mL).

Human monocytes were isolated from peripheral blood mononuclear cells (PBMCs) of HLA-A.02 healthy donors by immune magnetic-based procedures. Human immature dendritic cells (iDCs) were obtained by culturing monocytes for 5 to 7 days in the presence of 500 units/mL IL-4 (R&D System, Minneapolis, MN, USA) and 30 ng/mL GM-CSF (Serotec, Bio-Rad, Hercules, CA, USA). iDC were matured by overnight treatment with 10 ng/mL lipopolysaccharide (LPS). Both iDC and mature (m)DC were phenotyped for CD1a, CD14, CD80, CD83, and CD86 marker expression by FACS analysis, with BD Bioscience (Franklin Lakes, NJ, USA) antibodies (Abs).

To obtain HOXB7-stably expressing cells, MCF-7 cells were transfected with LB7SNeo vector by Lipofectamine 2000 (Invitrogen, Thermo Fisher Scientific). After 24 h, medium was replaced, and G418 (1.5 mg/mL) added as selection agent for establishing stable clones.

All cell lines have been routinely tested for mycoplasma contamination.

EV isolation and quantification

HEK293T cells were Lipofectamine2000-transfected with vectors expressing Nef^mut‐based fusion proteins. Cells were washed 24 h later and reseeded in medium supplemented with EV‐deprived FBS. Supernatants were harvested from 48 to 72 h after transfection. EVs were recovered through differential centrifugations at 500 × g for 10 min, then at 10,000 × g for 30 min, followed by supernatant 0.22 μm pore size filtration, and ultracentrifugation at 70,000 × g for 1 h. Pelleted vesicles were resuspended in PBS, and ultracentrifuged again at 70,000 × g for 1 h. EV-containing pellets resuspended in 1/100 of the initial volume. EVs were enumerated by nanoparticle tracking analysis (NTA) by Nanosight NS300 with the NTA software (Malvern Panalytical Ltd, Henderson, NV, USA), through a 488 nm laser.

Western blot analysis

Western blot (WB) analyses of cell lysates and EVs were conducted as reported [20], and filters revealed using 1/500-diluted anti-Nef (MA171503 Invitrogen, Thermo Fisher Scientific), anti-Alix (PA552873 Invitrogen, Thermo Fisher Scientific), and anti‐β‐actin-HRP mAbs (5125 Cell Signaling, Danvers, MA, USA).

WB analysis of MCF-7 cell clones was performed by lysing cells in 2x SDS-PAGE loading buffer. Fifty μg of cell proteins were resolved in 10% SDS-PAGE and blotted. Filters were revealed using 1/1000-diluted rabbit anti-HOXB7 (PA5-116315 Invitrogen, Thermo Fisher Scientific), and 1/500-diluted mouse anti-β-Actin-HRP Abs (AC-74 Sigma-Aldrich). Filters were analyzed by a Chemi-Doc XRS apparatus (Bio-Rad) and signals quantified by Image Lab software version 6.1.

Cross-priming, activation-induced degranulation, and trogocytosis assays

iDCs were challenged by equal amounts of engineered EVs uploading either Nef^mut or Nef^mut-HOXB7_tr isolated from supernatants of HEK293T transfected cells. After overnight incubation, iDCs were matured by LPS for 24 h. mDCs were washed and co-cultured with autologous peripheral blood lymphocytes (PBLs) in a 1:10 cell ratio. A week later, the stimulation procedure was repeated, and, after an additional week, PBLs were recovered for downstream assays.

Activation-induced degranulation was evaluated by measuring CD107a surface expression as described [25]. Briefly, 2 × 10⁵ PBLs recovered from cross-primed cultures were co-cultivated for 5 h with the HLA-matched HOXB7 expressing MCF-7 target cells, in the presence of PE-conjugated anti-CD107a (BD Bioscience) and 0.7 µg/mL monensin (GolgiStop, BD Bioscience). Cell cultures were then labeled with FITC-conjugated anti-human CD3 and APC-Cy7-conjugated anti-human CD8 Abs (both from BD Bioscience). Finally, CD8⁺ T cell populations were analyzed by FACS for the detection of CD107a-related fluorescence.

For trogocytosis assay, HOXB7 expressing MCF-7 cells were labeled with the fluorescent lipophilic dye CM-Dil (Molecular Probes, Thermo Fisher Scientific) according to the manufacturer’s instructions. After resuspension in complete medium, labeled target cells were cultured with primed PBLs (5 × 10⁵ per well in 200 μL of total volume) in U-shaped 96-well plates at a 1:5 ratio. After 5 h at 37 °C, cells were washed twice in PBS containing 0.5 mM EDTA to ensure cell dissociation, resuspended in PBS supplemented with 0.5% bovine serum albumin, and stained with FITC-conjugated anti-human CD3 and APC-Cy7-conjugated anti-human CD8 Abs (both from BD Bioscience).

In both assays, dead cells were stained with Aqua LIVE/DEAD dye (BD Bioscience). Samples were acquired by a CytoFLEX LX (Beckman Coulter, Brea, CA, USA) flow cytometer and analyzed using the Kaluza software (Beckman Coulter).

Animals and authorizations

Six to eight-week-old female FVB/N mice were purchased from Envigo RMS s.r.l. (San Pietro al Natisone, UD, Italy) and hosted at the Central Animal Facility of the Istituto Superiore di Sanità. The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Italian Ministry of Health, authorization 78/2020, released on February 11th, 2020.

Mouse immunization and tumor implantation

For all studies, animals were randomly randomized into vaccination and control groups, as planned, and investigators assessing tumor size were blinded to mouse group allocation throughout the follow-up. In the preventive vaccination setting, isoflurane-anesthetized mice were injected intramuscularly (i.m.) with 10 μg of DNA in 30 μL of sterile, 0.9% saline solution. DNA inoculation was immediately followed by electroporation at the site of injection with an Agilpulse (BTX, Holliston, MA, USA) device, as previously described [20]. Mice were vaccinated into both quadriceps, twice, 2 weeks apart. Two weeks after the last immunization, 2 × 10⁵ 305 tumor cells resuspended in 100 µl of PBS were injected subcutaneously (s.c.) into mouse flank and tumor growth followed up for 30 days. Mice were sacrificed at the end of experiment by cervical dislocation. Afterward, spleens and tumors, when developed, were explanted. Re-challenge procedures were carried out on already vaccinated mice with similar procedures, but injecting 2 × 10⁶ 305 tumor cells. For the therapeutic vaccination setting, mice were inoculated s.c. with 2 × 10⁶ 305 tumor cells in 100 µl of PBS into the flank. As soon as tumors became palpable (6 days after tumor cell implantation), animals were vaccinated twice as described above. Tumor growth was monitored for 39 days after implantation, then mice were suppressed as above, and both tumors and spleens were explanted.

Splenocyte isolation from immunized mice

Spleen cell isolation was conducted as described [20] and, upon resuspension in RPMI complete medium containing 50 µM 2-mercaptoethanol and 10% FBS, fresh splenocytes were evaluated by IFN-γ EliSpot and ICS assays.

Intracellular cytokine staining (ICS) and flow cytometry analysis

Splenocytes were cultured at 1 × 10⁷/mL in RPMI 1640 medium, 10% FCS, 50 µM 2-mercaptoethanol (Sigma-Aldrich), and in the presence of 5 μg/mL of (301–309) PYNYLSTEV and (417–426) PDSLRDLSVF [26] Her2/neu peptides, or a HOXB7 15-mer peptide pool (1 μg/mL for each peptide). Cells were cultured for 6 days, adding fresh medium with peptides at day 3. Cells were then cultured without peptides for additional 2 days. Before ICS, cells were re-stimulated overnight with the same or unrelated H2^q CD8 + T (GenScript) peptides, and 1 µg/mL brefeldin A (BD Biosciences) was added. Positive controls were conducted by adding 10 ng/mL PMA (Sigma-Aldrich) plus 1 µg/mL ionomycin (Sigma-Aldrich). After overnight stimulation, cells were collected and ICS was performed as already described [27]. Staining for cell surface markers was performed upon incubation for 1 h at 4 °C with 1/100 diluted of the following anti-mouse Abs: FITC-conjugated anti-CD3 (clone 17A2, cat. 555274, BD Biosciences), APC-Cy7-conjugated anti-CD8a (clone 53-6.7, cat. 557654, BD Biosciences), PerCPCy5.5-conjugated anti-CD4 (clone RM4-5, cat. 561115, BD Biosciences), and BUV395-conjugated anti-CD44 (clone IM7, cat. 740215, BD Biosciences) in PBS with 2% FBS. For cytokine staining cells were fixed and permeabilized using the Cytofix/Cytoperm kit (BD Biosciences) according to the manufacturer’s recommendations. Thereafter, cells were labeled for 1 h at 4 °C with 1/50 diluted of the following Abs: PE-Cy7-conjugated anti-IFN-γ (eBioscience, Thermo Fisher, clone XMG1.2, cat. 25-7311-82), PE-conjugated anti-IL-2 (Invitrogen/eBioscience Thermo Fisher, clone JES6-5H4, cat. 12-7021-82), and BV421 rat anti-TNF-α (BD Biosciences, clone MP6-XT22, cat. 563387). Samples were then acquired by a CytoFLEX LX (Beckman Coulter) flow cytometer and analyzed using Kaluza software (Beckman Coulter). Gating strategy was as follows (Supplementary Fig. 1): live cells as assessed by LIVE/DEAD dye vs. FSC-A, singlet cells from FSC-A vs. FSC-H (singlet 1) and SSC-A vs. SSC-W (singlet 2), CD3⁺ cells from CD3-FITC vs. SSC-A, CD8⁺, or CD4+ cells from CD8-APC-Cy7 vs. CD4-PerCP. CD3⁺/CD8⁺ cell population was gated against CD44⁺ cells, and, to detect IFN-γ CD8⁺ T lymphocytes, the population of cells positive for both CD8 and CD44 was analyzed for APC-Cy7, to detect percentage of IFN-γ positive cells. Next, to identify polyfunctional CD8⁺ T lymphocytes, the population of cells positive for both CD8 and CD44 was analyzed for PE-Cy7, PE, and BV421 to detect simultaneous changes in IFN-γ, IL-2, and TNF-α production, respectively.

Statistical analysis

The sample size was calculated based on results from our previous studies, using G*power, in order to obtain 80% power to detect the expected effect size, assuming a 5% significance level and a two-sided test. The minimum number of animals was used for this study to comply with the 3Rs of ethical research and EU directive (Directive 2010/63/EU, 2010).

When appropriate, data are presented as mean ± standard error (SE). The Kruskal–Wallis test, or the Friedman test (nonparametric repeated measures ANOVA), followed by Dunn’s multiple comparisons test, or two-tailed Wilcoxon matched pairs test were conducted, as indicated. Correlation analyses were performed by the two-tailed Spearman rank correlation test. p < 0.05 was considered significant, p < 0.01 was considered very significant. No animals were excluded from the analyses. Statistical analyses were performed using either GraphPad Prism 9 or GraphPad InStat.

Results

HOXB7-engineered EVs prime antigen-specific, human CD8⁺ T cytotoxic lymphocytes

We first sought to establish whether the treatment of professional APCs with HOXB7-engineered EVs can lead to the generation of HOXB7-specific CD8⁺ T lymphocytes. As previously observed, in the context of a Nef^mut-based fusion product, the 22 amino acid truncation of HOXB7 at its C-terminal DNA binding domain strongly favors its EV incorporation [23]. On this basis, large-scale preparations of EVs engineered with either Nef^mut alone (control) or fused with truncated HOXB7 (HOXB7_tr) were produced and analyzed by western blot, compared to EVs engineered with Nef^mut fused with full-length HOXB7 [23] (Fig. 1A and Supplementary Fig. 2), and by nanoparticle tracking analysis (Supplementary Fig. 3). The equivalent of 2 × 10⁸ particles from preparations of HOXB7_tr-engineered and control EVs were then used to challenge 10⁶ iDCs differentiated from monocytes isolated from PBMCs of HLA-A.02 healthy donors.

Fig. 1. HOXB7-engineered EVs characterization and CTL activity.

A Western blot analysis of EV lysates from cultures of HEK293T cells transiently transfected with either Nef^mut, Nef^mut-HOXB7, or Nef^mut-HOXB7_tr DNA vectors. Nef-based products were detected in EVs using anti-Nef Ab. Alix served as marker for EVs. Molecular marker is given in kilodaltons (kDa). The results are representative of three independent experiments. B, C CD107a and trogocytosis analyses in cross-primed PBLs. B CD107a FACS analysis on PBLs recovered after co-cultivation with DCs treated with either control or HOXB7_tr-engineered EVs. After isolation, PBLs were co-cultivated for 5 h with either parental or HOXB7-stably transfected MCF-7 cells. On the left, representative dot plots are reported. On the right, shown are the results obtained from three independent experiments carried out with PBLs isolated from different healthy donors, and calculated after subtraction of values detected in PBLs co-cultivated with parental MCF-7. *p < 0.05. C Trogocytosis analysis with cross-primed CD8⁺ T lymphocytes. Shown is the FACS analysis of PBLs isolated from co-cultivations with DCs treated with either control or HOXB7-engineered EVs. PBLs were then co-cultivated with either parental or HOXB7-stably transfected, CM-Dil-labeled MCF-7 cells. On the left, representative dot plots are reported. On the right, shown are the results obtained from three independent experiments carried out with PBLs isolated from different healthy donors, and calculated after subtraction of values detected in PBLs co-cultivated with parental MCF-7 cells. *p < 0.05.

Twenty-four hours later, iDC were LPS-matured before being co-cultured with autologous PBLs for 7 days. Then, lymphocytes were isolated and subjected to a second stimulation cycle by adding fresh mDC previously challenged with engineered EVs. After additional seven days, lymphocytes were recovered and put in co-culture with either parental HLA-A.02 MCF-7 cells or a derived cell clone stably transfected with an HOXB7-expressing vector (Supplementary Fig. 4). The generation of HOXB7-specific CD8⁺ T lymphocytes was assessed by both CD107a and trogocytosis assays, both aimed at evaluating the presence of antigen-specific CD8⁺ T cytotoxic lymphocytes (CTLs) within the bulk of PBLs isolated from the co-cultures. In both instances, percentages of HOXB7-specific cytotoxic CD8⁺ T lymphocytes significantly above the control conditions, i.e., co-cultures with DCs treated with Nef^mut EVs, have been detected (Fig. 1B, C).

These results indicated that HOXB7_tr engineered EVs have the intrinsic ability to elicit a CTL-specific immune response.

Antigen-specific CD8⁺ T cell immune response in mice vaccinated against HOXB7

We already demonstrated the induction of Her2/neuspecific CD8⁺ T lymphocyte responses in transgenic mice injected with Nef^mut-Her2/neu DNA vector [21]. Here, we first verified whether the injection in mice of the Nef^mut-HOXB7_tr DNA vector was able to elicit a detectable HOXB7-specific T-cell immune response. To this end, FVB/N mice were injected with either Nef^mut (N = 3 mice) or Nef^mut-HOXB7_tr (N = 6 mice) DNA vectors twice, two weeks apart. Fifteen days after the last immunization, splenocytes were isolated and tested by IFN-γ EliSpot assay after overnight incubation with a pool of HOXB7-specific 15-mers. A number of spots significantly higher in splenocytes from mice injected with Nef^mut-HOXB7_tr DNA compared to control cultures was observed (Supplementary Fig. 5).

These data highlighted that the EV-based vaccination strategy is able to break the immunologic tolerance towards HOXB7.

Cumulative antitumor effects of anti-HOXB7 and anti-Her2/neu vaccinations in double-injected mice

Next, we were interested in assessing the antitumor efficacy of vaccination with HOXB7 and Her2/neu-based vectors. To this aim, FVB/N mice were immunized twice, two weeks apart, with vectors expressing either Nef^mut (N = 6) or Nef^mut-HOXB7_tr and Nef^mut/Her2-ECD either singly or in combination (N = 10). Three weeks after the last immunization, animals were implanted subcutaneously with 2 × 10⁵ syngeneic 305 cells, stably expressing both HOXB7 and Her2/neu (Supplementary Fig. 6). As control, an additional group of naïve mice (N = 10) was also inoculated with cells. Mice were monitored for tumor growth for the following 30 days (Fig. 2A). Tumors did not develop in 3 out of 10 mice vaccinated with the HOXB7 expressing vector, and in 7 out of 10 mice vaccinated against Her2/neu. In both HOXB7- and Her2/neu- vaccination groups, the remainder mice presented with delayed tumor development and contained tumor growth compared to control mice (Fig. 2A, inset). Notably, all 10 mice in the combination vaccine group were completely protected from tumor development. All 16 control mice (naïve and Nef^mut) developed tumor masses. Representative images are shown in supplementary Fig. 7.

Fig. 2. Preventive antitumor effect induced by i.m. injection of Nef^mut-HOXB7 and Nef^mut-Her2/neu DNA vectors alone or in combination.

A Tumor growth curves. FVB/N mice (10 per group, 6 for the group injected with the vector expressing Nef^mut alone) were inoculated with the indicated vectors twice, 15 days apart. Thereafter, mice were challenged with 2 × 10⁵ cells, and the growth of tumor masses was followed over time. Shown are the mean tumor volumes for each group ± s.e. The numbers of mice that did not develop tumors are indicated. In the inset, the tumor growth curve in mice that developed tumors only is reported. *p < 0.05, **p < 0.01. B Detection by ICS/flow cytometry analysis of both HOXB7 and Her2/neu-specific CD8⁺ T cells in splenocytes isolated from FVB/N mice (N = 5) i.m. injected with either Nef^mut, Nef^mut-HOXB7_tr, Nef^mut-Her2/neu or Nef^mut-HOXB7_tr plus Nef^mut-Her2/neu DNA vectors. Raw data from a representative analysis of IFN-γ-expressing CD8⁺ T lymphocytes are presented. C Shown are the percentages of cells expressing IFN-γ over the total of CD8⁺/CD44⁺ T cells from each mouse injected with the indicated DNA vectors. D Percentages of cells from each mouse simultaneously expressing IFN-γ, IL-2, and TNF-α over the total of CD8⁺/CD44⁺ T cells are presented^. Shown are mean values of the absolute percentages of cytokine-expressing cells treated with specific peptides after subtraction of values detected in cultures treated with an unrelated peptide. For all panels, Kruskal–Wallis Test followed by Dunn’s Multiple Comparisons Test was conducted. Error bars, s.e. *p < 0.05, **p < 0.01.

To identify potential correlates of protection, for each mouse, the specific CD8⁺ T responses towards both HOXB7 and Her2/neu were evaluated at sacrifice. Splenocytes were isolated and tested by ICS/flow cytometry analysis to enumerate both IFN-γ-expressing and polyfunctional (i.e., co-expressing IFN-γ, IL-2, and TNF-α) antigen-specific CD8⁺ T lymphocytes (Fig. 2B–D). Overall, CD8⁺ T cell immune response against Her2/neu was stronger than that detected against HOXB7. Importantly, antigen-specific polyfunctional CD8⁺ T lymphocytes were found in both single and combined vaccine-immunized mice (Fig. 2D).

Correlation analyses showed negative correlations between tumor volumes and specific CD8⁺ T responses directed against both HOXB7 (r = −0.57, p < 0.1) and Her2 (r = −0.61, p < 0.1).

To assess the duration and strength of protective immune responses, 86 days after the first cell inoculation four tumor-free, double-vaccinated mice were re-implanted with a 305-cell inoculum ten-fold larger than the previous one. New naïve mice (N = 4) were similarly inoculated with tumor cells as a control. Tumor growth was almost abolished in vaccinated mice, over the following 39 days of observation, whereas all control mice developed tumors as expected (Fig. 3).

Fig. 3. Antitumor effect against re-implanted 305 cells.

Tumor-free FVB/N mice (N = 4) after the double vaccination and implantation of 305 cells, were re-challenged with 2 × 10⁶ 305 cells, and the growth of tumor masses was followed over time. As control, age-matched naïve mice were inoculated with the same number of cells. Shown are the mean tumor volumes for each group ± s.e. **p < 0.01.

These data demonstrate that the vaccination with either DNA vector generated an immunity able to counteract the growth of implanted HOXB7-Her2/neu expressing cells. The antitumor effect was stronger in Her2/neu- compared to HOXB7-vaccinated mice, and was maximum in mice immunized with the combination vaccine, which blocked tumor growth completely and persistently, breaking the immunological tolerance towards both TAAs.

The HOXB7-Her2/neu co-vaccination blocks tumor development in a therapeutic setting

We then investigated the antitumor efficiency of the combination vaccine in a therapeutic setting. To this end, FVB/N mice (N = 5) were subcutaneously implanted with a large (2 × 10⁶ cells) 305-cell inoculum. After 6 days, i.e., as soon as tumors became palpable, mice were injected with both DNA vectors twice two weeks apart. Over a 39-day follow-up, we observed that the growth of tumors was strongly impaired in double-vaccinated mice, compared to control animals (Fig. 4). Hence, as also assessed for two HPV-related TSAs [20], the EV-based vaccination against two TAAs efficiently counteracted the growth of already implanted tumors.

Fig. 4. Antitumor therapeutic effect induced by i.m. injection of both Nef^mut-HOXB7_tr and Nef^mut-Her2/neu DNA vectors.

Tumor growth curves. FVB/N (N = 5) were challenged with 2 × 10⁶ 305 cells (day 0) and then, in the presence of palpable tumor masses, co-inoculated with the indicated DNA vectors. The DNA inoculations were repeated at day 20 after tumor cell implantation, and the growth of tumor mass was followed over time. Shown are the mean tumor volumes for each group ± s.e. **p < 0.01.

Taken together, the here presented data strongly support the idea that the Nef^mut-based vaccine strategy should be considered for new therapeutic approaches in humans to combat tumors co-expressing HOXB7 and Her2/neu.

Discussion

A number of anticancer immunotherapy strategies, including those based on monoclonal antibodies, CAR-T cells, and neoantigen-specific vaccines, rely on targeting selected TSAs and TAAs. Under immune pressure, intrinsic growth potential and genetic instability of tumor cells often cooperate in the selection of null tumor cells, which no longer express the tumor antigen of interest. This results in tumor escape and clinical relapse. Such undesired effects can more easily occur when a single tumor antigen is targeted, whereas an antitumor strategy designed against different tumor antigens simultaneously is expected to have a greater chance to be successful. In the case of TAAs, it should be also considered that immune response activation needs to overcome the immunologic tolerance established at the time of immune system development.

We already provided evidence that the Nef^mut-based vaccine platform can efficiently induce specific CD8⁺ T cell immune responses against two [20] or more antigens [28] used in combination. In addition, our vaccination strategy is able to overcome immune tolerance as already demonstrated in Her2/neu transgenic mice upon injection with a vector expressing Nef^mut-Her2/neu [21]. In this system, although the Her2/neu-specific immune response induced by engineered EVs was sufficient to significantly delay tumor onset, with time, the tumors still emerged most likely as a consequence of the immunoediting process.

To test whether targeting two TAAs at the same time by EV-based vaccination can actually represent an advantage over the single immunizations in terms of antitumor effect, we exploited a model of transplantable tumors based on a tumor cell line co-expressing HOXB7 and Her2/neu [24]. Over the last few years, HOXB7 gained great consideration as a TAA as it was found overexpressed in a variety of solid tumors [29]. Importantly, HOXB7 overexpression contributes to cancer progression by promoting invasion, migration, angiogenesis, as well as the TGF-β-driven epithelial-to-mesenchymal transition [29, 30]. All this evidence justified the choice of HOXB7 as a pathogenically relevant TAA.

The possibility to apply the EV-based technology to HOXB7 mostly depends on its efficiency of incorporation into EVs upon fusion at the C-terminus of Nef^mut. As we have previously shown, a C-terminal, 22 amino acid-truncated form of HOXB7 is incorporated in EVs more efficiently than its full-length isoform [23], possibly making it more suitable for our EV-based vaccine strategy. We assumed that this difference was a consequence of the loss of the DNA binding domain in the truncated form, with a possibly reduced tendency to localize into the nucleus.

In the present work, purified preparations of HOXB7_tr-engineered EVs were used in cross-priming assays to test their intrinsic ability to elicit HOXB7-specific CTLs in human PBLs. The results we obtained through both CD107a and trogocytosis assays consistently supported the idea that HOXB7-engineered EVs act as an effective inducer of HOXB7-specific CTLs, thus justifying the extension of our investigations to a suitable animal model. In addition, our data represent the proof-of-principle for the efficacy of the Nef^mut-based vaccine in overcoming the natural HOXB7-specific immunologic tolerance in humans.

Our studies in vaccinated mice demonstrated the generation of both IFN-γ-expressing and polyfunctional, HOXB7-specific CD8⁺ T lymphocytes. We found that, in the same experimental conditions, the immune response against Her2/neu was reproducibly stronger than that induced by HOXB7-expressing vector. Dissimilarities inherent to the fusion products in terms of both intrinsic immunogenicity and intracellular trafficking in professional APCs may account for these differences. Possible yet unpredictable differences in the peptide binding/recognition efficiency taking place during the immunological assay may also be considered. Nonetheless, the results reported here showed for the first time that the EV-based vaccination was able to overcome the immune tolerance towards the TAA HOXB7, in mice. At the same time, these studies also confirmed the breach of immune tolerance towards Her2/neu, which we reported previously in a different mouse model [21]. To be noticed, here we provide the proof of concept that, using our EV-based platform, combined vaccination against two TAA is possible and more efficacious than vaccination against a single protein target.

Interestingly, immunogenicity differences between these two TAAs well reflected the ability of the respective immune responses to counteract the replication of tumor cells implanted after vaccination with single antigens. Most importantly, when mice were preventively vaccinated with both Nef^mut-HOXB7 and Nef^mut-Her2/neu DNA vectors, the tumor growth was totally abolished. Data from tumor-free mice reinoculated with tumor cells suggested the persistence over time of antitumor immunity, similar to what we already observed in mice vaccinated against both HPV-E6 and -E7 TSAs [20].

In this study, we also identified Her2/neu- and HOXB7-specific CD8⁺ T cell responses as correlates of protection against tumor growth. These results are in line with the complete protection from tumor development observed in the group immunized with the combination of antigens, in the preventive vaccination setting. The concomitant action of specific CD8⁺ T responses induced against both HOXB7 and Her2 proteins, which add up both in qualitative (multiple targets) and quantitative terms (increase in the number of specific activated clones) is reasonably at the basis of the higher anti-tumor efficacy found in this group. In a previous work from our group, a positive correlation between protection from infection and T cell-specific immunity generated by this vaccine platform has been identified in a preclinical model of SARS-CoV-2 respiratory infection [31]. The here presented data allows to extend the identification of correlates of protection also in a preclinical model of cancer.

The anti-HOXB7-Her2/neu combined vaccination proved to be efficacious also when administered after tumor implantation, i.e., in a therapeutic setting. Considering that, in humans, tumors are usually fought by therapeutic, rather than prophylactic, approaches, this latter result allows to consider the HOXB7-Her2/neu double EV-based vaccination protocol for a translation into the clinic. Moreover, considering that the EV-based platform can be exploited to induce CD8⁺ T cell immunity against a combination of at least up to four antigens [28], the EV-based antitumor vaccination strategy could be readily implemented by including additional tumor targets to further strengthen its overall protective effect.

The present work shows a number of limitations. The antitumor effects of the combined vaccination would have gained even more significance if applied in an orthotopic tumor model also, i.e., against tumors arising in HOXB7-Her2/neu double-transgenic mice [24]. Yet, as the results described herein for Her2 reflect what we have already published using a Her2-transgenic model [21], we assume that the same may apply for both HOXB7 alone and HOXB7-Her2/neu-combined immunization. This will be verified in future studies.

Moreover, we did not investigate a possible role, of both antibody and CD4⁺ T cell immune responses in the overall protection against tumor growth observed in vaccinated mice, even though, on the basis of our previous work, such immunity is expected to be ancillary. Finally, the contribution of the CD8⁺ T cell response against each antigen to the overall antitumoral effect was not evaluated in the therapeutic setting.

Nonetheless, the here presented data are strongly supportive of the idea that the EV-based, CD8⁺ T cell-specific vaccine platform represents a quite flexible weapon against tumors expressing known TAAs, which may deserve consideration for further development in view of possible clinical application. Our vaccination strategy can break immune tolerance towards tumor antigens of interest by combining the expression of protein vaccine targets and the exploitation of EV capability of inducing a transient inflammatory state when taken up by antigen-presenting cells. In fact, it has been recently reported that breach of CD8 peripheral tolerance can be obtained through the synergistic actions of both strong T cell receptor (TCR) signaling, as a consequence of target antigen overexpression, and local inflammation [32].

On the other hand, EV-based vaccines are not expected to induce adverse effects similar to those directly relatable to therapies based on engineered cells, such as CAR-T. In fact, in our system, immune responses are generated through vaccination, which activates T cells via the native TCR, leveraging all-natural activation and regulatory feedback mechanisms and persist due to induction of immunological memory.

Hypothetically, this vaccine platform would also represent an attractive alternative to elicit an effective CD8⁺ T cell immunity against neoantigens, considering the non-satisfactory CD8⁺ T cell immune responses observed in the recently developed peptide-[33] and mRNA-based [34–38] anticancer immunotherapies against TSAs, TAAs, and neoantigens.

Author contributions

FF, MF, and CC were responsible of conceptualization and investigation; FM, MD, EO, PL, KP, AG, and ADV performed the experiments and provided technical support; MF provided funding; FF and CC contributed to the analysis of the data; FF, MF, and CC co-wrote the manuscript; FF, FM, MD, MF, and CC discussed the results and revised the manuscript. All authors have read and agreed to the published version of the manuscript.

Funding

This work was supported by the grant PGR00810 from Ministero degli Affari Esteri e della Cooperazione Internazionale, Italy.

Data availability

The data presented in this study were generated by the authors and included in the article.

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

The authors declare that all methods were performed in accordance with the relevant guidelines and regulations and that the study on animals was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Italian Ministry of Health, authorization 78/2020, released on February 11th, 2020.

Supplementary information

The online version contains supplementary material available at 10.1038/s41417-024-00831-2.