A combined Globo H-OCT4-T7 immunotherapy restrains colorectal tumor growth

Background: Malignant tumors express multiple common markers that might constitute targets for tumor immunotherapy. Methods: Here we performed antigen epitope screening and molecular docking to design a new combination vaccine based on covalently linking Globo H and Oct 4 with a TLR7 agonist. We mainly use ow cytometry to analyze the changes in lymphocyte differentiation after the vaccine is applied. Use Elisa technology to detect the antibodies produced by the vaccine and the changes in lymphokines. In order to supplement the immunological data, RNA sequencing was performed on untreated and vaccine-treated tumors to dene the interference of the vaccine on the genome-wide gene expression prole. Results: Our results demonstrated that the Globo H-OCT4-T7 conjugate vaccine effectively induced cellular uptake, maturation and antigen presentation of BMDCs. The combination vaccine enhanced cytokine production, CTLs activity elicited high levels of IgG. Importantly, the tumor immune microenvironment was signicantly improved in tumor-bearing mice treated by Globo H-OCT4-T7 conjugate vaccine, demonstrated by increased the ratio of M1 to M2 TAMs, reduced number of Tregs in PBMCs and promoted the inltration of tumor-specic IFN γ-producing CD8+ T cells. Prophylactic vaccination of Globo H-OCT4-T7 signicantly reduced tumor growth by >80 % in both CT26 transplanted mice and in a human PDX mouse model when exposing mice to the combined vaccine compared to control. Our novel, combination vaccine is safe and effective at preventing tumor growth in mice. Conclusions: This combination vaccine is more effective than the single vaccine, and thus has the potential to prevent OCT4 and Globo H from expressing cancer.

system cannot effectively recognize it, resulting in strong immune escape ability of CSCs and high clinical recurrence rate of tumor patients. At present, the preventive vaccines against CSCs are generally aimed at a single target, such as ALDH (aldehyde dehydrogenase) (18), CD44 (19), etc., while, multi-target CSCs vaccines are rarely reported. CSCs express different markers in different periods. For a single target CSCs preventive vaccine, it can only target CSCs in a speci c period, and cannot effectively eliminate the variant CSCs. There is still a need for new therapy scheme for CSC possessing targeted, sustained and systemic therapeutic effects, to be able to ght against metastasis and recurrence, while limiting offtarget effects. We hope to improve the killing power of CSCs in different periods by preparing multi-target vaccine. The concept of multi-target vaccination is to limit the escape ability of tumors by targeting more potential redundancies.
OCT4 (octamer-binding transcription factor 4) is a member of the mammalian POU transcription factor family; it is a key regulator of embryonic stem cell self-renewal and pluripotency (20,21). OCT4 is highly expressed in breast, testicular and bladder cancers, germ cell tumors and cancer stem cells (22)(23)(24)(25). Tumors with high OCT4 expression are associated with advanced disease progression, high risk of metastasis, and short cancer-related survival time (26). OCT4-speci c T cells are found at a high frequency (80%) in T cells of PBMC isolated from healthy donors compared to newly diagnosed patients with germ cell tumors (35%), suggesting that reducing the OCT4 immune response might lead to germ cell tumorigenesis and tumor development (27,28). Because OCT4 is highly expressed in tumor cells, but it is either absent or very lowly expressed in many normal cells, it might constitute a potential tumor stem cell biomarker and an ideal target for tumor therapy. Indeed, previous studies have reported that vaccines targeting OCT4 can have signi cant therapeutic effects (29,30).
Globo H is a hexasaccharide glycosphingolipid that was originally discovered in the human MCF-7 breast cancer cell line (31,32). Globo H is overexpressed in most malignancies and its high levels might be predictive of poor patient outcomes (33)(34)(35). Globo H is generally not expressed in normal tissues, with only weak expression at the secretion boundary and other intraluminal sites that are not accessible to the immune system (36,37). As such, Globo H has also become an attractive vaccine target for epithelial tumors. Synthetic Globo H, which binds to the keyhole limpet hemocyanin (KLH), has been used as a therapeutic vaccine in clinical trials to elicit an antibody response (38)(39)(40)(41) and additional studies are ongoing to elucidate the effectiveness of Globo H immunotherapy in cancer patients.
Adjuvant is an effective substance to enhance immunogenicity and immune protection of vaccines (42,43). TLRs are a family of proteins expressed in various cells, including immune cells (dendritic cells [DCs], macrophages, B cells and T cells). TLR signaling is activated by ligands and contributes to the host immune defenses (44,45). Of the 13 TLRs, only TLR7/8 recognizes GU-rich short single-stranded RNA (ssRNA) and synthetic small molecules, such as nucleoside analogs and imidazoquinolin, which have been reported to have immunostimulatory properties and have been used in many tumor vaccines (46). This ability creates an opportunity to screen and modify highly potent TLR7 agonists to enhance the immune response. In addition, TLR7/8 agonists not only activate antigen-presenting cells (APCs) but also promote the activation of T cells and natural killer cells (NK) (47,48). Activation of TLR-7 results in the involvement of MyD88, MAL, IRAKs, and IKKα, which promote activation of IRF and production of type 1 IFNs and pro-in ammatory cytokines (49)(50)(51)(52). Because most tumor cells express high levels of tumorspeci c antigens, TLR7 agonists have attracted widespread interest in tumor immunotherapy (53).
Because of tumor heterogeneity, vaccines containing a combination of antigens may improve e cacy and cover a wider range of cancer targets Here, we introduced a novel TLR7 agonist called SZU-101 (54) and demonstrated that OCT4 antigenic peptides are covalently linked to Globo H polysaccharides and TLR7 agonists (Globo H-OCT4-T7), which have a surprising anti-tumor effect on colorectal cancer(CRC) by inducing a tumor-speci c immune response. The mechanism by which this combination vaccine promotes an immune response was investigated.

Materials And Methods
Epitope prediction and molecular docking The OCT4 protein sequence was obtained from the National Center for Biotechnology Information (NCBI; http://www.ncbi.nlm.nih.gov). Using the epitope prediction program IEDB (http://www.iedb.org/), the OCT4 sequence was analyzed and peptides with high a nity for MHC class I was identi ed. The peptide sequence was: EAAGSPFSGGPVSFPLAPGPHFGTPGYGSPHF. The e cacy of this peptide fragment at suppressing mouse teratocarcinoma cell growth has been reported by our groups recently (Cancer Biol Med 2020. doi: 10.20892/j.issn.2095-3941). Discovery studio software (55) was used to design the optimal combination of Globo H-OCT4-T7 vaccines by combining energy and scoring functions.
Experimental animals, cell lines and human colorectal cancer specimens Female BALB/c mice (4-6 weeks old) and female BALB/c nude mice (4-6 weeks old) were purchased from the Medical Laboratory Animal Center (Guangzhou, Guangdong Province, China). All animal treatments and experimental protocols for this study were performed with the approval of the Laboratory Animal Welfare and Ethics Committee, School of Medicine, Shenzhen University (Permit No. AEWC-20190003). All mice were housed under a 12 h light/dark cycle, at 23 ± 1°C, with 39%-43% relative humidity. Water and food were provided ad libitum. CT26 cells (mouse colorectal cancer) and F9 (mouse teratocarcinoma cells) were maintained in Dulbecco's modi ed Eagle's medium (DMEM) with 10% fetal bovine serum (FBS) and 100 U/mL penicillin-streptomycin (all from Hyclone Laboratories, Inc., South Logan, UT, USA). Colorectal carcinoma samples (n=16) were obtained from patients at Shenzhen People's Hospital in China. Data collected for each patient included their age, sex and tumor samples (Tables 1). All clinicopathological information was obtained by reviewing pathology reports. This study and the data obtained from clinical samples were approved by the Ethics Committee of Shenzhen People's Hospital.

Vaccine preparation and administration
The OCT4 peptides were synthetized at ChinaPeptides Co., Ltd. (Shanghai, China). The Globo H in the form of lyophilized powder was bought from Oligotech (Crolles, French). EDC (#22980, Thermo Fisher Scienti c) and NHS (#24500, Thermo Fisher Scienti c) were used to activate carboxylic acid as an active ester. The molar ratio of Globo H, OCT4 peptides, TLR7 agonist, EDC and NHS was 1:1:2.5:5:5 and this mixture was incubated at 4°C overnight. The nal conjugate Globo H-OCT4-T7 was stable for at least 6 months at -20℃ at concentration of 1.5 µM.

Generation and stimulation of bone marrow derived dendritic cells (BMDCs) in vitro
BMDCs were isolated from hind limb bones of mice as previously described (56,57). Brie y, both distal bone ends were excised and the marrow cells were ushed using RPMI 1640 (Gibco, Grand Island, NY). The red blood cells were lysed and the remaining cells were centrifuged at 450 x g for 10 min. Then, 2 10 6 /mL cells were aliquoted in RPMI 1640 supplemented with 10% FBS, 2 mmol/L L-glutamine, 1 mmol/L sodium pyruvate, 100 U/mL penicillin, 100 mg/mL streptomycin, 20 ng/mL GM-CSF (PeproTech) and 10 ng/mL IL-4 (PeproTech) and incubated at 37°C. The cells were cultured in six well plates in complete medium with cytokines at 4´10 6 cells/2 mL/well. On the fth day of culturing, OCT4-T7, Globo H-T7 or Globo H-OCT4-T7 were added to a nal concentration of 10 mg/mL, and the cells were cultured for two more days. On day 7, most of the non-adherent cells had acquired typical dendritic morphology, and these cells were used as the source of DCs in subsequent experiments. For ow cytometry, the cells were cultured for 7 days and then rinsed three times with PBS and brie y trypsinized to form a single cell suspension. The cells were stained for 30 min on ice with the following uorescence-labeled antibodies: Anti-mouse MHCII APC (eBioscience: 17-5321-82), Anti-mouse CD86 PE (clone: GL1), Anti-mouse CD80 FITC (clone: 16-10A1). Finally, the cells were washed twice with cold PBS and analyzed on ow cytometry (BD FACSAria II cell sorter, BD Biosciences). Tumor protection in BALB/c mice Female BALB/c mice (6-10 mice per group) were vaccinated with untreated group (PBS), T7 (0.012 μmol), OCT4-T7 (0.012 μmol), Globo H-T7 (0.012 μmol) and Globo H-OCT4-T7 (0.012 μmol) via intraperitoneal (i.p.) injection on days 0, 14 and 21. To assess the memory response, a booster dose was given on day 28. Five days after the nal vaccination, the mice were challenged with 1´ 10 5 CT26 cells (100 μl of cell mixture in 50% Matrigel; #354234, Corning) that were injected subcutaneously (s.c.) into the lower ank of the right back. Then, the tumor size was measured periodically with calipers. Starting from day 0 after cell inoculation, the tumor volume was measured every 2 days. The tumor size was calculated using the formula: 0.5 × length × width 2 (cm 3 ) (58). Mice were euthanized when the tumor size was > 1.5 cm 3 . Mouse blood samples were collected from retro-orbital tissues.
Enzyme-linked immunosorbent assay (ELISA) Sera samples were collected at 3-5 days after four doses of the vaccine had been administered. An ELISA for multiple cytokines in peripheral blood was performed using a Ready-SET-GO! ELISA kit (eBioscience, Thermo Fisher Scienti c), according to the manufacturer's instructions.

Cytotoxicity assay
Splenocytes and lymph nodes (LN) of vaccinated mice were harvested for evaluating the cellular immune responses 4-5 days after the nal vaccination. The cytotoxicity activity of the vaccinated mice was analyzed using a CytoTox96 Non-Radioactive Cytotoxicity Assay Kit (G1780, Promega, Madison, WI, USA). Brie y, splenocytes from the vaccinated mice were mixed at an effector: target (E: T) ratio of 12.5:1, 25:1, 50:1 with 5 ´ 10 3 CT26 cells. Cytotoxic T lymphocyte (CTL) activity was evaluated by analyzing the released LDH. The cytolysis rate (%) was calculated based on the equation: cytotoxicity (%) = (Experimental release spontaneous release)/ (maximum release spontaneous release) ´ 100.

Humoral immune response
In order to assess humoral responses to anti-OCT4 IgG and anti-Globo H IgG levels, ELISA was performed to measure antigen-speci c total IgG, according to previously described (59). Brie y, each 96-well ELISA plate (Costar, Corning, New York, USA) was coated with 2 μg synthetic OCT4 peptide and 2 μg Globo H and incubated at 4°C overnight. The plates were then washed with PBST (0.01 M PBS containing 0.05% Tween 20, pH 7.4) three times. Serum samples were detected at a different dilution for 2 h. Alkaline phosphate-conjugated detection antibody for total IgG (Sigma-Aldrich) was added and incubated for 1 h at room temperature. Then, p-NPP substrate (Millipore) and stop solution (50 μL of 3 M NaOH) were added to each well, and the optical density (OD, 405 nm) was measured using a spectrophotometer (BioTek, Winooski, VT, USA).
Tumor-reactive IgG assay To detect tumor-reactive serum IgG, CT26 and F9 cells were cultured with serum harvested from BALB/c mice immunized with different vaccines for 4 h, washed thoroughly with PBS, and stained with an GFPlabeled anti-mouse IgG antibody, before analysis by ow cytometry using a FACS Calibur ow cytometer (BD FACSAria II cell sorter, BD Biosciences).
Splenocyte and lymph node cell proliferation assay One week after the last immunization, spleen and lymph node cells were isolated from BALB/c mice and plated into 96-well plates. The lymphocytes were then co-cultured with different vaccines. After 3 days, a 3-(4, 5-dimethylthiazol-2-yl)-2, 5-diphenyltetrazolium bromide (MTT) assay was performed to test the proliferation of the lymphocytes, as previously described (60).

Analysis of tumor in ltrating immune cells
CT26-bearing mice were sacri ced on day 28. Tumors were dissociated using a mouse tumor dissociation kit according to the manufacturer's protocol (Miltenyi Biotec). The cells were stained for PE F4/80 (macrophage marker), APC CD86 (M1 marker), PE/Cy7 CD206 (M2 marker). Post-staining, the cells were washed and quanti ed by ow cytometry. The ratio of M1/M2 macrophages was calculated based on the % M1 positive cells to % M2 positive cells. Tregs were identi ed as CD4 + CD25 + Foxp3 + .

RNA sequencing
Gene expression analysis of subcutaneous tumor samples from mice exposed to different vaccine immunizations was evaluated at Novo Gene Corporation (Beijing, China). The up-regulated or downregulated genes were identi ed by ltering the RNA-seq data with the following cutoff values: 2-fold change in expression level and a false discovery rate (FDR) analog with a P value of < 0.05.

H&E and immuno uorescence staining
Tumor tissues were harvested and xed in 10% buffered formalin, dehydrated, cleared and embedded in para n according to standard procedures (61). The tissues were then sectioned at 3 μm, mounted onto polylysine-coated slides and then the sections were stained with hematoxylin and eosin (H&E). Histopathological morphology was checked under a light microscope (Olympus BX51). For immuno uorescence assays, the sections were depara nized in xylene and rehydrated in graded ethanol. Antigen retrieval was performed in citrate buffer (pH 6.0) by boiling for 30 min in a microwave and cooling down to room temperature. Then, the sections were blocked with 3% bovine serum albumin in PBS for 1 h at room temperature. Following this, the sections were stained with antibodies against mouse CD8 (14-0081-82, Invitrogen), OCT4 (701756, Invitrogen) and SSEA3 (Globo H homologs) (1:200, 14-8833-80, eBioscience) at 4°C overnight. After washing, the sections were incubated with a secondary AlexaFluor 488 goat anti-rabbit antibody (1:200, ab16669, Abcam) for 1 h, and were then washed and mounted in Fluoroshield with DAPI (F6057, Sigma-Aldrich). Images were captured under a Nikon Eclipse E400 microscope using an Olympus DP73 camera and analyzed with cellSence Entry software.
The transfer experiment The primary tumor was resected from a patient diagnosed with CSC. For the patient-derived xenograft (PDX) model, the fresh patient tumor specimen was cut into small pieces (less than 2-3 mm) and initially implanted subcutaneously in nude BALB/c mice. For the transfer experiment, PBMCs harvested from healthy people's peripheral blood were stimulated with or without Globo H-OCT4-T7 vaccines and injected (i.p.) into the PDX once a week for 3 weeks. The tumor volumes were measured every one week.

TUNEL staining
Tissue sections were dewaxed with xylene and then hydrated with a gradient of ethanol using the TUNEL FITC Apoptosis detection kit (yazyme, A111). After washing three times with PBS, it was permeabilized with proteinase K. The TUNEL reaction mixture was prepared, and after completion of the reaction, the lm was sealed with a DAPI-containing sealing liquid, and then observed under a uorescence microscope.

Statistical analysis
We used Graphpad Prism 8.0 to conduct statistical analyses. For lifespan, Kaplan-Meier survival was utilized and P values were calculated using the log-rank test. To compare two groups, Student's t-test was performed. One-way analysis of variance (ANOVA) with Duncan's test was conducted for comparing multiple groups. P < 0.05 was considered statistically signi cant.

Results
A Globo H-OCT4-T7 combination vaccine is stable and has low binding energy and is suitable for use as an immunotherapy vaccine We rst aimed to conjugate TLR7 agonist to peptides antigens or polysaccharide. Here, we combined the three components of Globo H-OCT4-T7 in different arrangements to produce multiple vaccine formats in different spatial conformations: type I, II III and IV. We pioneered the use of computer simulation technology to virtually screen the optimal spatial con guration based on the principle of lower bound energy. We chose C-type lectin (DC-SIGN), a receptor protein that DC cells recognize antigens, and then we used discovery studio software to perform 3D homology modeling DC-SIGN on DC cells. We molecularly docked four designed vaccine conformations with DC-SIGN, and calculated the binding energy to obtain a scoring function (Fig. 1A). The software can use a variety of scoring functions to give a score for the combination. A higher score indicates that the combination is better and more consistent with the real situation. From the scoring function, the comprehensive score of the type I vaccine was the highest (18.64) compared to the other three con gurations (Fig. 1B). We know that when two things are combined, the lower the energy, the more stable it is. We then tested the binding energy of the four vaccines. From the combination score, the type II vaccine had the lowest binding energy and the most stable structure, but these parameters did not signi cantly differ from the other three con gurations (Fig.  1C). After that the type 2 vaccine's spatial structure easily breaks the linked chemical bonds and makes synthesis di cult, we chose the type I vaccine as the subject of subsequent studies.
To construct the conjugate vaccines, we chose EDC and NHS to active carboxylic acid as active ester. We rst used them to activate the carboxyl group of the TLR7 agonist, form an amide bond attached to OCT4, and then use OCT4 as a carrier to link Globo H through an amide bond at the other end to synthesize Globo H-OCT4-T7 (Fig. 1D). We also generated OCT4-T7 (MW 4613 Da), Globo H-T7 (MW 1757 Da), Globo H-OCT4-T7 (MW 5926 Da) and analyzed all three molecules by mass spectrometry ( Figure 1E).
Globo H-OCT4-T7 promotes antigen uptake in BMDCs and induces BMDC maturation The induction of protective immune responses involves several key steps. These include antigen uptake and processing by antigen-presenting cells (APCs), activation of APCs for effective T and B cells. After immunization, APCs, especially DCs, capture and present antigens through the MHC I and MHC II pathways to activate T and B lymphocytes. Having generated our multi-marker vaccine, we next examined its e cacy on antigen uptake in BMDCs. We incubated OCT4-T7, Globo H-T7 and Globo H-OCT4-T7 with BMDCs for 24 h and quanti ed antigen uptake through surface binding by MFI of GFP and APC using ow cytometry. Cellular localization of OCT4 and Globo H was examined by confocal laser scanning microscopy ( Fig. 2A). We found a 3-fold increase of Globo H and 4-fold increase of OCT4 uptake in BMDCs co-cultured with Globo H-OCT4-T7 compared to controls (Fig. 2B&C). These results suggest that compared to OCT4-T7 and Globo H-T7, Globo H-OCT4-T7 can signi cantly increase antigen uptake of BMDCs, partly via enhanced surface binding.
To investigate the effect of Globo H-OCT4-T7 on DCs in vitro, we harvested immature BMDCs on day 5 after isolation and exposed them to Globo H-OCT4-T7 for 48 h. We then characterized DC maturation based on surface marker expression by ow cytometry. Here we found that exposure to Globo H-OCT4-T7 signi cantly enhanced (P < 0.01) the expression levels of CD80 and CD86 -surface markers that indicate DC maturation -compared with other groups (Fig. 2D). MHC II expression levels (P < 0.05) were also upregulated compared with T7 group in DCs treated with Globo H-OCT4-T7 (Fig. 2E). Taken together, these data indicate a facilitative role of Globo H-OCT4-T7 on DC maturation.
Globo H-OCT4-T7 induces cytokine production and enhances cytotoxic T lymphocyte (CTL) responses against tumors To evaluate the e cacy of Globo H-OCT4-T7 as potential cancer vaccine, mice were immunized four times. One week after the forth immunization, we collected serum and then measured IFN-γ, IL-12, IL-2 and TNF-α cytokine levels by ELISA. We found that the levels of secreted IFN-γ, IL-12, IL-2 and TNF-α were signi cantly higher in the Globo H-OCT4-T7 vaccine group compared with the other groups (Fig. 3A). To test the e cacy of T cells in mediating tumor-speci c CTL responses, we established an in vitro CTL assay. First, we co-cultured effector cells (splenocytes and lymph node cells isolated from immunized mice) with the CT26 expressing OCT4 and Globo H as target cells. We then tested target-cell lysis at the Effector: Target (E: T) cell ratios of 12.5:1, 25:1 and 50:1 (Fig. 3B). Importantly, the effector T cells isolated from mice immunized with Globo H-OCT4-T7 lysed ~50% target CT26 cells, while the lysed target CT26 cells in control group was only ~20% (Fig. 3B). This nding suggests that mice immunized with Globo H-OCT4-T7 can induce an e cient CTL response against the CT26 target cells.
Following the successful preparation and evaluation the immunogenicity of vaccines, we explored whether humoral immunity was involved in the anti-CRC effect. We found that the titer of OCT4 and Globo H-speci c antibodies in the serum was signi cantly increased (Fig. 3C). We also found that the level of tumor-speci c IgG antibodies was signi cantly increased in mice immunized with Globo H-OCT4-T7 vaccine compared to F9 cells expressing only OCT4 as a control (Fig. 3D&E). These results indicated that the anti-tumor effect induced by the Globo H-OCT4-T7 vaccine was dependent on both cellular and humoral immunity.

Globo H-OCT4-T7 vaccination has therapeutic effects in a CT26 tumor model
To examine the antitumor e cacy of Globo H-OCT4-T7, we thus next evaluated the therapeutic effects of Globo H-OCT4-T7 vaccines using a protection experiment. Brie y, we vaccinated BALB/c mice with the combination vaccine following an immunization schedule (Fig. 4A), with PBS as a negative control. Then, we measured the tumor volumes every 2 days after the tumors began to grow. Compare with the PBS group, all treatment groups showed reduced tumor growth rates (Fig. 4B). As expected, the Globo H-OCT4-T7 vaccine group exhibited the best anti-tumor effects with ~5-fold smaller tumor size compared to control group, suggesting the advantages of combination therapy (Fig. 4C). Furthermore, in an overall survival analysis after the nal day of treatment, median survival analysis was also signi cantly prolonged up to 20 days in Globo H-OCT4-T7 vaccine group compared to control (Fig. 4D). These ndings indicate that muti-target therapeutic strategies combining TLR7 agonist, OCT4 and Globo H surface polysaccharide can increase the immunotherapy effects of CT26 and has a long-lasting overall response.

The Globo H-OCT4-T7 vaccine changes the proportion of macrophages in the tumor microenvironment and reduces the proportion of regulatory T cells in the PBMC
The occurrence and metastasis of tumors are closely related to the internal and external tumor microenvironment. Tumor cells can promote tumor growth and development through autocrine and paracrine changes and maintain conditions for survival and development. Tumor-associated macrophages TAMs have an important role in this process (62). These cells can switch between two major phenotypes, the anti-tumor M1 phenotype and pro-tumor M2 phenotype, depending on surrounding signals [63]. We immunized BALB/c mice four times (as described above) and after 32 days, inoculated the mice subcutaneously with 1 × 10 5 CT26 cells. After 1 week, we isolated TAM and found that the population of CD86 + F4/80 + M1 TAMs from BALB/c mice immunized with Globo H-OCT4-T7 vaccine increased compared to control (P < 0.05, Fig. 5A), while there was no difference in the population of CD206 + F4/80 + CD45 + M2 TAM cells among the other groups (Fig. 5B). The M1/M2 ratio was calculated by measuring the %M1 positive cells to %M2 positive cells. As shown in Fig. 5B, M1 macrophages were signi cantly upregulated to an M1 phenotype (P < 0.01) in comparison to control.
Meanwhile, Tregs play a key role in the maintenance of immune tolerance and participated in the inhibition of harmful immune responses to the host. Evidence suggested that Tregs were immunesuppressive lymphocytes, which helped immune escape and suppressed anti-tumor immune response (63,64). So the number of CD3 + CD4 + CD25 + Foxp3 Tregs cells in PBMCs from the tumor-bearing BALB/c mice were recorded and analyzed by Flow cytometry. We observed a signi cantly decreased CD3 + CD4 + CD25 + Foxp3 Tregs population after Globo H-OCT4-T7 vaccine treatment when compared with control and T7 groups (P < 0.05, Fig. 5C&D). These data indicate that Globo H-OCT4-T7 vaccine acted on the TAMs and enhanced macrophage polarization and Tregs mediates the tumor inhibition.

The tumor suppressive effect of Globo H-OCT4-T7 vaccine is relied on T-cell regulation
We next aimed to identify the mechanism of action of Globo H-OCT4-T7 vaccine on tumor inhibition. Whole transcriptome RNA-seq analysis was carried out on tumors from PBS, T7, OCT4-T7, Globo H-T7 and Globo H-OCT4-T7 groups. The transcriptome heat map illustrated great differences between the treatment groups (Fig. 6A). From the transcriptome volcano map, the control group had 54 up-regulated genes and 351 down-regulated genes compared to the Globo H-OCT4-T7 group (Fig. 6B). We then analyzed the vaccine action pathway using ClueGO (Fig. 6C). Gene enrichment analysis showed that the three vaccines mainly affected the immune pathway within the tumor. Speci cally, the differentially expressed genes between the Globo H-T7 and the control groups were mainly concentrated in the regulation of cell activation, side of membrane and regulation of leukocyte activation (Fig. 6D); the differentially expressed genes between the OCT4-T7 and the control group were mainly concentrated in the positive regulation of the immune response, side of membrane and T-cell activation (Fig. 6E); and the differentially expressed genes between the Globo H-OCT4-T7 group and the control group were mainly concentrated in the regulation of leukocyte activation, the adaptive immune response, and T-cell activation (Fig. 6F). Overall, transcriptome analysis showed that the tumor suppressive effect of the Globo H-OCT4-T7 vaccine is mainly focused on T-cell regulation.

Globo H-OCT4-T7 induces antitumor immune responses
It is well established that CD8 + T cells are important effector cells in vaccine-induced anti-tumor responses (65). Interestingly, our combination vaccine received a single component effect during the induction of CD4 + , CD8 + T cells differentiation. CD8+ T cells in the spleen and LN of mice immunized with OCT4-T7 rather than Globo H-OCT4-T7 were signi cantly increased (Fig. 7A, B&C). We consider this is due to T cell proliferation. To this end, we examined cell proliferation by MTT assay and calculated the absolute number of CD8 + T cells in the spleen and LN, and found no signi cant difference between the Globo H-OCT4-T7 group and OCT4-T7 group (Fig. 7D).
Previous studies demonstrated that T-cell dysfunction or exhaustion in tumor-bearing mice hinders the production of anti-tumor immunity (66). To examine the potential role for preventing CD8 + T-cell exhaustion with our combined vaccine, we examined the production of IFN-γ, which is the signature Th1 cytokine and granzyme, which can directly kill infected cells. We found that the combined vaccine elevated the production of IFN-γ by CD8 + T cells (Fig. 7 A, B&E). These results indicate that the combined vaccine can induce antitumor immune responses and prevent CD8 + T cells from CT26-induced exhaustion. TIL contains different proportions of CD3 + CD4 + and CD3 + CD8 + T lymphocytes. Its functional status is considered to be an important indicator of positive prognosis and long-term survival of patients (67).
Here, we aimed to further assess CD8 + T cell populations by histological analyses. We inoculated nude mice subcutaneously with 1 × 10 5 CT26 cells after four rounds of immunization with the vaccine. Then, we isolated tumors when they became visible and monitored the proportion of CD8 + T cells by immuno uorescence. We found that the number of CD8 + T cells in the tumor tissue sections of mice immunized with Globo H-OCT4-T7 was signi cantly higher than those in other groups (P < 0.05, Fig. 8 A&B). Next, we examined tumor-cell apoptosis by TUNEL staining and found that the percentage of apoptotic tumor cells was signi cantly increased in tumor tissues immunized with Globo H-OCT4-T7 vaccine (P < 0.05, Fig. 8 C&D). The TUNEL assay further demonstrated that Globo H-OCT4-T7 induced extensive tumor cell apoptosis.
Globo H-OCT4-T7 vaccine can safely inhibit tumor growth in PDX mice Safety evaluation is an important aspect in the development of immunoprevention. Blood cell levels of different vaccines demonstrated no signs of change compared with control healthy mice (Table S1).
However, White blood cell count (WBC) and Lymphocytes (Lym) of T7 group showed a litter reduction compared with control (P < 0.05). The vital organs (heart, liver, spleen, lung and kidney) of the mice were taken out four times after immunization for toxicological analysis, and no signi cant difference was found between the groups, illustrating the safety of the vaccine (Fig. 9A).
We then assessed tumor samples from 16 patients by immunohistochemistry, to determine the positive rates of OCT4 and Globo H expression (Table.1). Here, we found that most patients express highly OCT4 and Globo H. Then, we isolated PBMCs from the healthy people's peripheral blood and stimulated the cells with or without vaccines before implanting immunode cient mice with patient tissues to construct PDX model (Fig. 9D). On the 50th day after inoculation, all the mice were then sacri ced, and tumor tissues were resected and weighed (Fig. 9B&C). The combination therapy signi cantly inhibited tumor growth and brought the inhibition rate up to 60% (P < 0.01). Immunohistochemical uorescence analysis showed the content of OCT4 positive cells and Globo H positive cells in the original patient tumor samples had markedly decreased compared to control (Fig. 9E&F). These data clearly indicate that Globo H-OCT4-T7 vaccine possesses an excellent therapeutic e cacy in the PDX model.

Discussion
Non-invasive cancer prevention using anti-cancer vaccines has non-toxic potential to ensure protection due to the speci city of the immune response and its long-term memory (68). CRC as the third most commonly diagnosed and the second leading cause of cancer death worldwide (69), the treatment of CRC is still a huge clinical challenge, suggesting that identi cation of CRC-speci c targets for prevention is urgently needed. Accumulating evidence has revealed the importance of CSCs in chemotherapy and recurrence (70). Because of tumor heterogeneity, vaccines containing a combination of antigens may improve e cacy and cover a wider range of cancer targets (71). In this study, we use a novel TLR7 agonist SZU-101 as an adjuvant for the Globo H-OCT4 combination vaccines. The results demonstrated that Globo H-OCT4-T7 induced speci c cellular immune response and protection of BALB/c mice from CRCs growth, as evidenced by signi cantly enhanced lymphocyte proliferation, IFN-γ, IL-2, IL-12, TNF-α, cytotoxicity and in vivo antitumor effects, compared with control groups and single antigen vaccine.
CSCs are key drivers of cancer recurrence after therapy, but there is currently a lack of immune vaccines that target these cells (72). The identi cation of speci c markers for CSCs has been the focus of research, enabling the preparation of immune vaccines by stimulating an immune response to target and kill CSCs, thereby preventing the onset or recurrence of cancer (73,74). Synthetic Globo H binds to the key pore limpet hemocyanin (KLH) and has been used as a therapeutic vaccine in clinical trials to elicit an antibody response in breast cancer cells (75). Clinical trials are currently underway to test vaccines for speci c markers (76,77). However, the challenge remains that tumor cells differ from foreign microorganisms in that their speci c proteins are less immunogenic, and the mechanism of tumor immune escape is complicated; notably, tumor cells are not easily recognized by the immune system.
Targeting CSCs may damage tumor growth and decrease the risk of tumor recurrence and progression (78). We focused on CSCs as a target of cancer immunoprevention to achieve more e cient anti-tumor response. In this study, we illustrated that Globo H-OCT4-T7 signi cantly enhanced immune response. Elucidating the mechanism underlying these effects indicated that such combination therapy enhanced the CD8 + T cell population and enhancing CTL activity.
We propose that coupling multiple active components together in one vaccine would have better therapeutic effect compared to a single component vaccine. Compared with conventional vaccines, peptide subunit vaccines have many advantages, such as minimal toxicity, easy preparation, immediate and long-term memory and cost effectiveness (79). Indeed, we demonstrate that our dual-target tumor vaccine based on polysaccharide peptide is more effective at killing tumor cells than the equivalent single-target vaccine. Therefore, we can also explore the overall effect of different ratios of different components of the vaccine on the e cacy of the combination vaccine.

Conclusions
In summary, our study rst showed that Globo H-OCT4-T7 acted on TAMs, promoted their antigen-