Administration of the carbohydrate ligand (BG34-200) via intravenous injection can induce antitumor effects in solid cancers by modulating CD11b⁺ cells

We recently developed a carbohydrate molecule that is safe and isolated from oats. This molecule has an unusual capacity to modulate myeloid-derived cells.15 16 Through our research, we found that the β-(1,3)-(1,4)-glucans (BG34) derived from oats with a weight average molecular weight (Mw) of ~200 kDa (BG34-200) can induce antitumor immune responses by modulating the surface marker CD11b (CR3; CD11b/CD18; Mac-1) of myeloid cells.15–17 In previous studies, we optimized the administration route, dosage, and schedule of BG34-200 and evaluated its antitumor effects using mouse models of melanoma.15 18

Building on these findings, we expanded our research to assess the in vivo effects of BG34-200 in multiple murine tumor models using the optimized treatment conditions. These models included metastatic melanoma (B16F10, figure 1A), metastatic osteosarcoma (K7M2, figure 1B), transgenic tamoxifen-inducible KPC pancreatic ductal adenocarcinoma (PDAC) (KRASLSL.G12D/+; TP53LSL.R172H/+; Pdx1-Cre) (figure 1C–E), and an orthotopic KPC model (figure 1F). We chose these models because they allow solid tumors to grow in immunocompetent mice, which mimic key aspects of human diseases.19–22 Our results showed that the intravenous administration of BG34-200 at 50 mg/kg effectively induced antitumor effects in all of the models tested (figure 1A–F). However, when we tested CD11b knockout mice (B6.129S4-Itgamtm1Myd/J, CD11b^−/−) with the same tumor burdens, we did not observe any antitumor effects (figure 1A and F).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

BG34-200 intravenous administration induces antitumor effects in multiple solid cancer models. Kaplan-Meier overall survival curves of mice bearing (A) B16F10 melanoma, (B) K7M2 osteosarcoma, (C–E) spontaneous KPC tumors and (F) orthotopic KPCY tumors. For A–C, n=9. (D) Cross sections of T2-weighted MR image of the spontaneous KPC mice. The red lines indicate tracing of the region of interest (ROI) for the tumor volume measurement. Four out of 20 slices of MR scans were presented as examples. (E) The areas of ROIs were summed to calculate volumes of spontaneous KPC tumors. n=5. (F) Mean fluorescence intensity (MFI) of KPCY tumors in wild type (WT) mice or CD11b^−/− mice. n=9. For E and F, **, p<0.01; ***, p<0.001; ns, not significant. PBS, Phosphate-Buffered Saline.

These data, combined with our previous findings,15–18 suggest that BG34-200 induces antitumor effects in multiple solid cancers by modulating CD11b⁺ cells.

BG34-200 mediates direct, multisite, and multivalent binding to CD11b

In order to understand how BG34-200 interacts with integrin CD11b, we conducted overlapping peptide microarrays to determine which peptide motifs are involved in their binding (figure 2A). Both linear and conformational overlapping peptides were used, and microarrays were performed on BG34 of 50, 200, and 500 kDa for comparison, as BG34-200 can induce antitumor effects but the BG34-50 and BG34-500 cannot.15 16 After incubating with biotin-conjugated BG34s or non-targeted glucan control, the peptides were stained with a streptavidin DyLight680 and control antibody, which generated heat maps of all the peptides (online supplemental figure S1A,B). The microarrays were also performed at three different concentrations of BG34 molecules (Online supplemental figure s1C), allowing us to eliminate noise signals and determine dose-dependent signals representing BG34 bound peptides.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

BG34-mediated multisite and multivalent binding to CD11b. (A) Schematics of CD11b overlapping peptides for microarrays. (B) Mean fluorescence intensity (MFI) of BG34 samples that bind to the peptides, as determined by microarray scanning and PepSlide analyzer. The conformational peptide microarrays were prestained with streptavidin DyLight680 for background control, then incubated with biotin-conjugated BG34-50, BG34-200, or BG34-500 at 100 µg/mL, followed by staining with streptavidin DyLight680. The MFI of DyLight680 signals were measured, calculated, and presented as the binding strength between BG34 and peptides. (C) Three-dimensional structures of CD11b (in teal) and the five peptides (in yellow), as determined by computational simulation method (AlphaFold: AF-P11215-F1), in combination with the crystallographic structure (PDB: 1IDO in magenta); the serine and threonine residues (in hot pink), the transmembrane region (in orange), followed by the intracellular domain (in red). (D) Fold increase or decrease of the binding strength between BG34 samples and the mutated (MT) peptides, as determined by amino acid substitution microarray assays. The binding strength between BG34-200 samples and wild type (WT) peptides were normalized to be 100%. Amino acid substitution of three peptides (¹⁵⁸S-K¹⁷⁰, ¹⁷³V-K¹⁸², and ³⁸⁰L-K³⁸⁶) are shown. Every amino acid in these sequences was substituted by all 20 amino acids. The fold changes are quantified by measuring the increase or decrease of binding affinity to MT peptide relative to the binding affinity to the WT peptides (in red). The essential conserved sequences are framed by purple rectangles. BG34-200 showed general preference for amino acids K and R (in green), and general aversion to aromatic amino acids W and Y (in blue). (E) Amino acid sequence alignment of human (black) and murine (red) CD11b peptide residues ¹⁰¹Q-L³⁰⁰. BG34 bound peptide sequences were underlined black for human and red for mice. (F) Binding affinity plots of BG34 samples at different concentrations. Plots of three different molecular weights (Mws) are shown. (G) Binding affinity plots of BG34 samples at different Mws. BG34 samples at 0.2 µM are shown. For F and G, the binding affinities were determined through the overlapping peptide microarrays which measured the MFI of BG34 samples that binds to the peptides. Slope=MFI/(concentration×Mw).

Based on these data, we identified five peptides with different binding strengths to BG34s (figure 2B). Three of the five peptide species (S158-K170, V173-K182, and E238-247) shared a primary sequence of C-terminal K and were mapped to the consensus motifs located at the CD11b I (or A) ectodomain. The other two peptides (L380-K386 and Q485-C494) were located outside of the I domain, as determined by examining their crystallographic structure (PDB: 1IDO) using the computational simulation method (AlphaFold: AF-P11215-F1) (figure 2C). The database and simulation methods also allowed us to generate secondary structures of these peptides, which revealed that all five peptides were all located at the solvent-exposed surface of CD11b.

To verify the accuracy of these peptide sequences, we used peptide substitution microarrays. In this method, we replaced each amino acid (AA) in each of the five wild type (WT) peptides with all 20 AAs. The WT peptides and their variants were then synthesized and cyclized onto microarray chips that eventually contained 1300 different peptides printed in triplicates, framed by 134 control peptides (HA, YPYDVPDYAG). After the biotin-tagged BG34 molecules and controls were incubated with the peptide microarrays, they were stained with streptavidin Dylight680 and control antibody to measure the intensity of each peptide spot. This generated intensity plots and heat maps, enabling us to identify the mutated (MT) peptide that interacted with the BG34 molecules (online supplemental figure 2). Using this information, we produced AA plots for each WT peptide, which reflected the AA preferences at each given position, helping us determine the conserved and variable AA positions.

Based on the data analyzed, we have observed that the AA positions of L380-K386 and Q485-C494 display variable characteristics, indicating that they may be non-specific sequences (data not shown). On the other hand, the ¹⁵⁸S-K¹⁷⁰, ¹⁷³V-K¹⁸² and ²³⁸E-K²⁴⁷ peptides contain AA positions that are essential and highly conserved, such as the ¹⁶⁵F-K¹⁷⁰, ¹⁷⁹Q-K¹⁸², and ²⁴⁴N-K²⁴⁷ (indicated by purple rectangle frames). Substituting any of these AAs with others resulted in partial or complete loss of BG34-200 binding in these three sequences. Of particular interest is the fact that all AA positions in these peptides have a strong preference for K or R (indicated in green), and almost all the AA positions show an aversion to aromatic AAs Y and W (indicated in blue) (figure 2D).

In the ¹⁶⁵F-K¹⁷⁰ region, K170 showed significant sequence conservation as even substituting it with the strongest basic AA R led to a complete loss of binding. R168 and R169 appeared to be less conserved than K170, as substituting them with K resulted in slightly lower spot intensities and substituting them with any other AAs resulted in a 60% loss of binding. F165 and M168 were more tolerant of conservative exchange with K and R, as all other substitutions resulted in approximately 50% loss of binding.

In the ¹⁷⁴Q-K¹⁸² region, Q174, K181, and K182 were highly conserved, as substitutions with all other AAs led to a ~20–50% loss of binding. However, L175 appeared to be variable. In the ²⁴⁴N-K²⁴⁷ region, all the AA positions were tolerant of substitutions with K and R, but substitutions with any other AAs resulted in 50–100% loss of binding. Taken together, these results suggest that BG34 is capable of directly engaging multiple sites in the I domain of CD11b at specific peptide epitopes.

Moreover, we conducted a comparison between the human and mouse CD11b. These two proteins have a high level of similarity, with a 78.7% DNA sequence similarity and 74.8% direct AA identity.23 By analyzing the three peptide regions in both human and mouse CD11b, we found that the BG34 bound regions retained a high degree of homology (figure 2E). Using substitution microarray analysis, we discovered that BG34 molecules had a lower affinity for mouse CD11b at ¹⁶⁵FQKMK¹⁷⁰ and ²⁴⁴NGARE²⁴⁷ compared with the corresponding human CD11b at the same position. This is because the replacement of ¹⁶⁶RR¹⁶⁷ (human) with ¹⁶⁶QK¹⁶⁷ (mouse) resulted in a decrease of mean fluorescence intensity (MFI) by ~20%, and the substitution of ²⁴⁷K (human) with ²⁴⁷E (mouse) resulted in a decrease of MFI by~30%.

However, BG34 molecules had a higher affinity for mouse CD11b at ¹⁷⁹QFKK¹⁸² than for the corresponding human residues ¹⁷⁹QLKK¹⁸², as the substitution of 180L (human) with 180F (mouse) led to an increase of MFI by ~60%. Despite the differences in BG34 binding affinities to the human and mouse peptides, replacing these AA residues (¹⁶⁶RR¹⁶⁷/¹⁶⁶QK¹⁶⁷, ²⁴⁷K/²⁴⁷E, ¹⁸⁰L/¹⁸⁰F) with other AAs all resulted in a significant loss of binding (figure 2D). These findings suggest the three peptides are highly conserved in their sequences with respect to BG34 binding, and that the interaction between BG34 and CD11b occurs in a specific manner.

Furthermore, we conducted an analysis to understand how the Mw of BG34 affects its binding to CD11b using peptide substitution microarray data. Our results showed that all variations of BG34 could bind to the three peptides in a concentration-dependent manner, but with differing binding strength (figure 2F). Specifically, BG34-500 exhibited significantly stronger binding compared with BG34-50 and BG34-200.

Polysaccharides, or carbohydrate polymers, contain multiple copies of ligands or binding units that interact with their target, resulting in a multivalent interaction that enhances affinity. This phenomenon is seen in endocytic receptors such as CD11b and other molecules that play a crucial role in the first line of defense.24 This natural effect is known as the multivalent interaction through glycoside cluster effect, in which the clustering of lectin binding sites and carbohydrate recognition sites enhances affinity. This effect can be assessed by examining if the affinity enhancement effect by increasing Mw is greater than the affinity enhancement effect by increasing concentrations.24 25

Based on this knowledge, we used the microarray assay data and generated the binding affinity plots against Mw at a molar concentration of 0.2 µM (figure 2G). This concentration was selected to avoid any unwanted chain overlap or interpenetration caused by entanglement concentrations in BG34.26 27 Our results revealed that the affinity enhancement effect through the increase of Mw at the non-entanglement concentration is substantially greater than the affinity enhancement effect through an increase in concentration. This enhancement is indicated by the slopes shown in figure 2G. Therefore, these results suggest that BG34-CD11b engagement occurs through multivalent binding.

The engagement of BG34-200-CD11b occurs on the activated I domain, specifically at peptide residues that are distinct from other known epitopes

Research has demonstrated that the functional capacity of CD11b in myeloid and myeloid derived cells is influenced by a conformational change induced by environmental stimuli, rather than by upregulation or downregulation.28 29 When responding to stimuli, CD11b can shift from a low-affinity state to a high-affinity state. In contrast to the ‘closed’ conformation of the low-affinity state, the ‘open’ conformation of the high-affinity state is characterized by an unfolded I ectodomain (¹²⁸S-L³²¹), which is also known as the A domain (figure 3A). Although there is limited data linking the structure and function of the I domain, experimental results suggest that the I domain contains multiple functional epitopes that can interact with various ligands, triggering multiple cell functions.12–14

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

BG34-200-CD11b engagement occurs in the activated I domain at the distinguished peptide residues and mediated distinguished function. (A) Illustration of the CD11b ‘close’ and ‘open’ conformation in the I domain, highlighting functional epitopes that engage with corresponding ligands to induce cellular responses. (B) FACS analysis of surface expression of WT.1 and VIM12 in THP-1 cells treated with PBS or PMA. (C) (1) Frequency of cells displaying positive stain to BG34-200-AF647 after pretreatment with PBS or anti-CBRM1/5-FITC (CBRM1/5) antibodies; and (2) Frequency of cells displaying positive stain to CBRM1/5-FITC after pretreatment with PBS or BG34-200, as determined by FACS. (D) Cytokine production in cells receiving different treatments, as determined by ELISA assays. (E) Intracellular ROS production in cells receiving different treatments, as determined by ROS assays. For C, D, and E, PMA-treated THP-1 cells were used; *, p<0.5; **, p<0.01; ns, not significant. Unstained, singled stained, and non-specifically stained cells served as controls to establish positive versus negative gating in (B and C). CTL, control; FACS, fluorescence-activated cell analysis; ICAM-1, intercellular adhesion molecule I; IL, interleukin; PBS, phosphate-buffered Saline; PMA, phorbol 12-myristate-13-acetate; ROS, reactive oxidative species; TNF, tumor necrosis factor; WT, wild type.

In figure 3A, the CBRM1/5 epitope is shown to be one of the well-known epitopes found on the activated I domain. This binding epitope has been characterized as the ¹⁴⁷P/¹⁴⁸H/¹⁵¹R in the loop before and at the beginning of the α-helix 1, as well as ²⁰⁰K/²⁰³T/²⁰⁶L in the loop connecting α-helices 3 and 4. These six residues are adjacent in the structure of the I domain and have side chains that are well-exposed to the solvent.30 The engagement of CBRM1/5 with the ICAM-1 (intercellular adhesion molecule I) or fibrinogen ligands plays a critical role in monocyte adhesion.30 31 An ADH503 compound, also known as GB1275, was reported to be able to stabilize the CBRM1/5-ICAM-1 engagement by binding to the allosteric pocket (³¹⁷F-G³³⁶) in the I domain. This results in the upregulation of pro-inflammatory cytokines interleukin (IL)-1β, IL-6, and tumor necrosis factor (TNF)-α.31 The linear peptide residues at ²³²R-A²⁴⁵ are also of the well-known epitopes on the I domain. Its engagement with iC3b-opsonized cells or targets plays an important role in mediating intracellular reactive oxidative species (ROS) responses in monocytes and neutrophils, inducing a cytotoxic effect on target cells.32 Zymosan, a yeast derived β-glucan, enhances the (²³²R-A²⁴⁵)–iC3b engagement and engagement-triggered ROS activity by binding to the non-membrane-activation related AA residues.33

To investigate the potential impact of BG34-200-CD11b engagement on the function of known epitopes, we used a human monocyte model: THP-1 cells treated with PMA (phorbol 12-myristate-13-acetate) for in vitro assays. At low concentrations (<20 ng/mL), PMA treatment can polarize THP-1 cells towards pro-inflammatory activation, as evidenced by the high-level expression of pro-IL-1β, without the need for additional stimulation by cytokines or bacterial components.34 After PMA treatment, THP-1 cells were positive for both the FITC-tagged anti-CD11b-VIM12 (VIM12-FITC) and the PE-tagged anti-CD11b-WT.1 (WT.1-PE) stains. Since the VIM12 clones bind to non-membrane-activation-related epitopes and the WT.1 clones bind to the epitopes located at the I domain,31 these results suggested that PMA treatment leads to the generation of cells with an activated CD11b in the ‘open’ conformation (figure 3B).

After examining the cells, we observed that both WT.1-PE⁺ cells that were prestained anti-CBRM1/5-FITC monoclonal antibodies (CBRM1/5-FITC) and those that were not showed the same level of positivity when further stained by AF647-tagged BG34-200 (BG34-200-AF647) (figure 3CI). Our previous studies characterized BG34-200-AF647, which demonstrated high stability against different pH levels and various enzymes.15 17 In addition, when we prestained the WT.1-PE⁺ cells with or without BG34-200-AF647, we observed the same level of positivity when further stained by CBRM1/5-FITC (figure 3CII).

Moreover, we found that the cells treated with ADH503 in the presence of ICAM-1-coated beads exhibited enhanced upregulation of IL-1β and IL-6, compared with the cells receiving only ICAM-1 beads. In contrast, cells treated with BG34-200 in the presence or absence of ICAM-1-coated beads showed an upregulation in the TNF-α production (figure 3D).

Furthermore, we discovered that those which were subjected to zymosan treatment while exposed to iC3b-coated beads displayed a heightened intracellular activity in ROS. This was evidenced by the measurement of hydroxyl, peroxyl, and other ROS within the cells. On the other hand, cells that underwent BG34-200 treatment and were stimulated with iC3b beads exhibited a slight decrease in ROS activity, as depicted in figure 3E.

Based on the data we collected, it appears that the engagement of BG34-200 with CD11b did not have an impact on the I domain engagement at the allosteric pocket or the CBRM1/5 epitope. This was supported by the observation that the BG34-200 treatment did not affect the availability of peptide residues to interact with their correspondent ligands. Moreover, BG34-200 engagement had the potential to induce distinct cellular functions, as it resulted in TNF-α upregulation. This contrasted with the ADH503-stablized CBRM1/5 engagement with its ligands, which resulted in IL-1β and IL-6 upregulation. Of particular interest, our findings suggest that BG34-200 engagement exhibited higher binding affinity than I domain engagement with the iC3b targets at the epitope ²³²R-A²⁴⁵. This was implied by the slight reduction in the iC3b molecule-triggered ROS activity in cells following BG34-200 treatment.

Taken together, our findings suggest that BG34-200 acts as a specific carbohydrate ligand that engages with activated CD11b at the previously unreported peptide residues, which mediate different functions from the other known ligands.

The intravenous administration of BG34-200-AF647 had a direct impact on the tumor-associated inflammatory monocyte present in the peripheral blood

After demonstrating that BG34-200 could specifically engage with ‘open’ CD11b, we aimed to determine whether this engagement could impact any myeloid derived cell function in a specific and/or selective manner, following its intravenous administration. To investigate this, we conducted a fluorescence tracking study using BG34-200-AF647 in an xenograft KPC mouse model, which enabled us to track and monitor the kinetics, migration, and distribution of this compound.

Our findings revealed that the BG34-200-AF647 compound was completely cleared from the plasma in approximately 24 hours, as evidenced by the time-dependent decrease of AF647 fluorescence signals in the plasma samples (figure 4I). The BG34-200-AF647⁺ cells were detected in the circulating white blood cells (WBCs) around 4 hours, peaked around 10–72 hours, and returned to baseline levels after 72 hours (figure 4AII). However, in tumor-free mice, we did not observe the plasma clearance of BG34-200-AF647 compound (figure 4AIII) nor the presence of the BG34-200-AF647⁺ cells in the WBCs throughout the experiment within 72 hours (figure 4AIV). These results suggest that the tumor-associated cells were responsible for mediating the plasma clearance of BG34-200.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

BG34-200-AF647 intravenous administration shows a direct impact on the tumor-associated inflammatory monocytes (TAIMs) in the peripheral blood. (A) Kinetic distribution of BG34-200-AF647 in the peripheral blood of KPC tumor-bearing mice (I and II) and tumor-free mice (III and IV). (I) and (III): MFI of BG34-200-AF647 in plasma samples of mice, as determined by colorimetric analysis. Tumor-free plasma sample spiked with BG34-200-AF647 (0.1 mg/mL) served as positive control. MFI of plasma sample from individual mouse was graphed as mean±SD. **, p < 0.01. (II) and (IV): Frequency of BG34-200-AF647⁺ cells in the WBCs, as determined by FACS analysis. (B) Surface expression of Ly6C and Ly6G by the CD11b⁺BG34-200-AF647⁺ cells in the peripheral blood in the KPC tumor-bearing mice at 10 hours after BG34-200-AF647 intravenous administration, as determined by FACS. (C) Frequency of TAIMs and the BG34-200-AF647⁺ TAIMs in the peripheral blood in the KPC, B16F10, K7M2 tumor-bearing and tumor-free mice. (D) Plasma CCL2 concentrations in mice bearing different solid tumors with PBS or BG34-200-AF647 treatment, as determined by ELISA assays. Tumor-free mouse plasma served as controls. (E) MFI of BG34-200-AF647 in tumors, tumor draining lymph nodes (TdLNs), and different organs, as determined by colorimetric analysis. MFI of samples from mice were graphed as mean±SD. Two of the three data points represent pooled samples from two mice and one data point represents sample from one mouse. (F and G) Surface expression of CD11c, MHC II, and CCR2 by the KPC tumor-infiltrating CD11b⁺BG34-200-AF647⁺ cells (F) and circulating CD11b⁺BG34-200-AF647⁺ cells (G), as determined by FACS. For A, at 2, 10, 24 and 72 hours after BG34-200-AF647 intravenous injection, blood samples were collected and processed to collect white blood cells. The cells were then analyzed by FACS to determine the frequency of the BG34-200-AF647⁺ cells. Frequency of cells from individual mouse were graphed as mean±SD. For B, C, D, and G, the plasma and WBCs were collected at 10 hours after BG34-200-AF647 intravenous administration. For F, the tumor-infiltrating cells were collected at the 24 hours after BG34-200-AF647 intravenous administration. Mice bearing day-5 KPC tumors were intravenously administered with PBS or AF647-tagged BG34-200 (BG34-200-AF647) at 50 mg/kg. At 20 min, 2, 4, 10, 24 and 72 hours after BG34-200-AF647 injection, mice were sacrificed to harvest blood, tissue, organs, and tumor samples. n=5. BM, bone marrow; CTL, control; FACS, fluorescence-activated cell analysis; IV, intravenous; MFI, mean fluorescence intensity; PBS, Phosphate-Buffered Saline; WBCs, white blood cells.

Based on these kinetic data, we decided to investigate the BG34-200-AF647⁺ cells at 10 hours after intravenous administration in tumor-bearing mice. The BG34-200-AF647⁺ cells were found to be positive for CD11b-AF405 and Ly6C-AF488, but negative for Ly6G-PE stain (figure 4B). These findings suggest that the tumor-associated cells responsible for mediating the plasma clearance of BG34-200 were likely monocytic rather than granulocytic CD11b⁺ cells, as they were Ly6C⁺Ly6G^- in phenotype (figure 4B).

To better understand the tumor-associated CD11b⁺ monocytic cells, we implemented a complete separation strategy of CD11b⁺ cells. We analyzed the cells 10 hours after the intravenous administration of BG34-200-AF647 to the KPC, B16F10, and K7M2 mice. This approach divided the Lin (CD3, CD19, CD20, CD56)⁻CD11b⁺ cells into Ly6G⁻ and Ly6G⁺ cells and then analyze the cell expression of CD62L and Ly6C. As a result, we could distinguish granulocytic-Myeloid-Derived Suppressor Cells (MDSCs, Lin⁻CD11b⁺Ly6G⁺CD62L⁻Ly6C⁺) and monocytic MDSCs (Lin⁻CD11b⁺Ly6G⁻Ly6C⁺CD62L⁻CCR2⁻). Moreover, positive CCR2 expression by the Lin⁻CD11b⁺Ly6G⁻Ly6C⁺CD62L⁺ cells allowed us to identify inflammatory monocytes (Lin⁻CD11b⁺Ly6G⁻Ly6C⁺CD62L⁺CCR2⁺) (online supplemental figure s3A,B). Since murine anti-CD11b antibodies targeting different peptide residues within the I domain were not available, in this study we could only use anti-CD11b antibodies that bound to non-membrane-activation related sites. Based on these analyses, we found that tumor-associated inflammatory monocytes (TAIMs) directly interacted with the BG34-200-AF647 in the peripheral blood in the three mouse models of solid cancers, as almost all the TAIMs were BG34-200-AF647⁺ in the tumor-bearing mice (figure 4C).

According to literature, healthy individuals typically have only 1–5% of activated CD11b in the ‘open’ conformation in their peripheral blood, whereas patients with tumor burdens may have 10–30%.35 Based on this information, we believe that our findings are supported, demonstrating that BG34-200 can directly, effectively, and specifically target tumor-associated CD11b⁺ inflammatory monocytes by engaging with the activated CD11b in the open conformation I domain.

Since the tumor stroma could produce CCL2 chemokines that attract CCR2⁺ inflammatory monocytes to the TME,36–39 we characterized the CCL2 concentrations in the plasma and found a positive correlation between the frequencies of TAIMs and the CCL2 plasma concentration in the three mouse models (figure 4C and D). Next, we wanted to track and analyze the tumor-infiltrating BG34-200-AF647⁺ cells. Based on the kinetic biodistribution data, we determined to investigate the tumor-infiltrating cells 24 hours after the intravenous administration (figure 4E). Interestingly, the tumor-infiltrating BG34-200-AF647⁺ cells expressed CD11b, CD11c and MHC II but were negative for CCR2, suggesting a phenotype of monocyte-derived dendritic cells (mo-DCs) (figure 4F). However, we found that the circulating CD11b⁺BG34-200-AF647⁺ cells were positive for CCR2 but negative for CD11c and MHC II, suggesting that these cells in the peripheral blood were not differentiated (figure 4G).

Multiparameter FACS and t-SNE analysis reveals a correlation between the direct impact of BG34-200 on circulating TAIMs and the tumor-infiltrating mo-DCs, and the observed antitumor effects

To investigate the correlation between the antitumor response and the impact of BG34-200 impact on TAIMs, we used a multiparameter fluorescence-activated cell analysis (FACS) analysis to characterize myeloid cell subsets in blood and tumor samples from untreated and treated groups. We applied a validated 15-parameter flow panel to characterize the cells, and statistical t-distributed stochastic neighbor embedding (t-SNE) analysis to identify distinct cell clusters within the samples.

In our multiparameter FACS setting, we opted to use 15 markers due to their expression across different subsets of myeloid cells. These markers are valuable in characterizing cell phenotype, function, and tissue localization, making them a useful tool for understanding cell subsets within a heterogeneous sample (figure 5A). However, defining specific cell subsets based on cellular markers alone can be challenging. To overcome this challenge, we employed statistical t-SNE analysis to identify distinct cell clusters within our samples by integrating the FACS data with tumor size data to provide additional context for cell identity and function (figure 5B–E).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

Multiparameter FACS and t-SNE analysis reveals a correlation between the direct impact of BG34-200 on circulating TAIMs and the tumor-infiltrating mo-DCs, and the observed antitumor effects. (A) Description of 15 markers used to analyze myeloid derived cells. (B) Illustration of experimental design for the schedule, sample collection, and labeling. Each sample label represents an average of three data points obtained from three individual mouse. (C) PhenoGraph analysis of all the samples identified 27 distinct cell clusters that can be spatially organized by t-SNE 1 and 2. (D) X-shift analysis of the 27 cell clusters (X-axis), characterized by phenotype color barcode to define the average expression of fluorescent markers (Y-axis). (E) Grayscale PhenoGraph analyses of the cell clusters in blood sample B3 (PBS group) and B8 (BG34-200 group), as well as tumor sample T3 (PBS group) and T8 (BG34-200 group). The selected clusters are highlighted by orange contour lines. (F) Illustration of plotting the frequency of a specific cell cluster in both the PBS and BG34-200 groups against tumor size on the same graph. (G) Cell frequency of each selected cluster in blood or tumor samples from both the PBS and BG34-200 groups against tumor size. ns, not significant; **, p<0.01. FACS, fluorescence-activated cell analysis; IV, intravenous; mo-DCs, monocyte-derived dendritic cells; TAIMs, tumor-associated inflammatory monocytes; t-SNE, t-distributed stochastic neighbor embedding.

In our study, we administered a high dose of BG34-200-AF647 (150 mg/kg) B16F10 tumor-bearing mice intravenously, and collected blood (B) and tumor samples (T) from untreated PBS (Phosphate-Buffered Saline) and treated (BG34-200) groups at various time points following the intravenous administration of either PBS or BG34-200 to the B16F10 mice or tumor-free mice (figure 5B). Using our 15-parameter flow panel, we stained both the blood and tumor samples and performed t-SNE analysis, which enabled us to identify and characterize a total of 27 distinct cell subsets within all the samples (figure 5C). The t-SNE analysis allowed us to generate a heatmap that visualizes the expression patterns of multiple markers, which helped to highlight the subsets of cells that exhibit unique expression profiles (figure 5D).

To illustrate our strategy of analysis, we presented the t-SNE analysis of blood samples B3 (PBS group, day 3) and B8 (BG34-200, day 3), as well as the tumor samples T3 (PBS, day 3) and T8 (BG34-200, day 3) (figure 5E and F). From these graphs, we identified the subsets of cells that differed in frequency between the PBS and BG34-200 groups and plotted the frequency of each subset in each sample against tumor size. Specifically, we selected cluster 3, 9, 11, 12, 13 14, 18, 19, 23, 25 and 26 from the blood samples, as well as cluster 9, 13, 16, 17, 18, 20, 25 and 27 from the tumor samples (orange dash contour lines), which together accounted for over 90% of the cells in these samples (E). Next, we analyzed the cell frequencies of each of these clusters in each sample and plotted the frequencies of both the PBS and BG34-200 groups on the same graph to determine whether and how the frequency of the subsets correlated with tumor size (F). Our analysis revealed that only cluster 3 in the blood samples and cluster 27 in the tumor samples were affected by BG34-200 treatment and exhibited a significant correlation with tumor reduction (figure 5G).

Based on the phenotype analysis, it was observed that cluster 3 exhibited a phenotype of CD11b⁺CCR2⁺CXCR3⁺Ly6C⁺Ly6G⁻Siglec1^low, suggesting that it displays a phenotype of inflammatory monocytes. Importantly, this cluster 3 was found to be absent in the blood of tumor-free mice, suggesting that it is a tumor associated cell subset (online supplemental figure s4). The analysis of cluster 27 revealed that it exhibited a phenotype of F4/80⁺CD68⁺⁺⁺CD11b⁺⁺Arg⁺⁺CD86⁺⁺IDO⁻MHC II⁺⁺Siglec 1⁺CD11c⁺, suggesting that it is a macrophage-like DC subsets or mo-DCs. Since our fluorescence tracking studies have identified a similar tumor-infiltrating subset (CD11b⁺CD11c⁺MHC II⁺CCR2⁻) referred as mo-DCs (figure 4), we preferred to refer to this F4/80⁺CD68⁺⁺⁺CD11b⁺⁺Arg⁺⁺CD86⁺⁺IDO⁻MHC II⁺⁺Siglec 1⁺CD11c⁺ subset as mo-DCs instead of macrophage-like DCs. It is important to note that DCs derived from monocytes have similar morphology and phagocytic activity as macrophages; however, they also possess the ability to present antigens to T cells, which is a hallmark function of DCs. Therefore, we studied the function of these cells in the following sections.

Although the results presented in figure 5 did not directly support the cluster 3 differentiation towards cluster 27, phenotypes of these two clusters demonstrated consistency to our findings in figure 4. Our findings in figure 4 regarding the correlation between the BG34-200-AF647⁺ cells in peripheral blood and tumor tissue suggested the potential impact of BG34-200 on monocyte differentiation towards DCs, which was further investigated in the following section. Here, considering the results presented in figure 5, as well as those obtained from figure 4, it could be inferred that BG34-200 exerted a direct effect on TAIMs, leading to an increase in mo-DCs within the TME.

The intravenous administration of BG34-200 can promote the differentiation of TAIMs into mo-DCs, resulting in an increased capacity for antigen presentation

To investigate whether the direct interaction of BG34-200 with TAIMs could impact the differentiation of cells into mo-DCs, ex vivo assays were performed to examine cell function and differentiation. Initially, the CD11b^highCCR2^high cells sorted from the blood of PBS-treated mice showed significantly lower messenger RNA (mRNA) expression in PU.1 (transcriptional factor regulating cell differentiation into DCs) and MafB (transcriptional factor regulating cell differentiation into macrophages) compared with bone marrow (BM) derived inflammatory monocytes (figure 6A), indicating that TAIMs may have limited differentiation capabilities. In contrast, TAIMs sorted from the blood of BG34-200-treated mice exhibited significantly higher mRNA expression in PU.1 (transcriptional factor regulating cell differentiation into DCs) compared with cells from PBS-treated mice (figure 6B), suggesting that BG34-200 treatment could potentially convert the untreated TAIMs, the CD11b^highCCR2^highPU.1^low subsets, into the CD11b^highCCR2^highPU.1^high subsets in peripheral blood. Notably, the impact of BG34-200 treatment on PU.1 expression in TAIMs was observed in all three mouse models of solid cancers.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6

The intravenous administration of BG34-200 can promote the differentiation of TAIMs into mo-DCs, resulting in an increased capacity for antigen presentation. In panel (A and B): FACS sorting of CD11b⁺CCR2⁺ cells from the peripheral blood of mice with different solid tumors (KPC, B16F10, and K7M2) treated with PBS (A) or BG34-200 (B) at 10 hours following the intravenous administration, and RT-PCR analysis of the relative mRNA expression of PU.1 and MafB in the sorted cells. Cells were sorted from blood of mice bearing three different solid tumors (KPC, B16F10, and K7M2). Mouse BM derived inflammatory monocytes (PMA treatment at 10 ng/mL) served as control (CTL). The mRNA levels were expressed relative to the normalized CTL. In B, the mean value of PU.1 expression in the sorted cells from KPC, B16F10 and K7M2 mice treated with PBS is represented by the blue, green, and red dashed lines, respectively. In A and B, n=5; 88, p<0.01; ***, p<0.001. (C) FACS analysis of the frequency of CD11c⁺ cells in mature mo-DCs derived from the blood of KPC mice treated with PBS or BG34-200. (D) The percentage of proliferating OT I CD8⁺ T cells in response to DCs pulsed by irradiated B16F10-OVA cells, as determined by FACS. In C and D, the cells sorted from PBS or BG34-200 groups were cultured in DC medium supplemented with GM-CSF/IL-4 and stimulated by TNF-α to generate mature mo-DCs. The mo-DCs were then pulsed by B16F10-OVA cells at 1:4 ratio and cultured with OT I CD8⁺ T cells at CD11c⁺ cells:T cells ratio of 5:1. At the end of culture, cells were stained by anti-CD8-FITC and ki67, followed by FACS analysis to assess T-cell proliferation. BM, bone marrow; FACS, fluorescence-activated cell analysis; FSC, Forward Scatter; GM-CSF, Granulocyte Macrophage Colony-Stimulating Factor (GM-CSF); IL, interleukin; mo-DCs, monocyte-derived dendritic cells; mRNA, messenger RNA; PBS, Phosphate-Buffered Saline; PMA, phorbol 12-myristate-13-acetate; RT-PCR, real-time PCR; TAIMs, tumor-associated inflammatory monocytes; TNF, tumor necrosis factor.

Next, we discovered that culturing the cells with a DC differentiation medium supplemented with Granulocyte macrophage Colony-Stimulating Factor (GM-CSF)/IL-4 and stimulating cytokines (TNF-α) led to the generation of approximately 18.9% CD11c⁺ cells. This was a significant increase compared with the CD11c⁺ cells (~7.3%) generated from the PBS group, as shown in figure 6C. This in vitro condition was employed based on established method for culturing DCs and was supported by our published results,15 which indicated elevated concentrations of the cytokines GM-CSF, IL-4 and TNF-α in tumor tissues following systemic administration of BG34-200.

Moreover, when we pulsed the cells derived from B16F10 mouse blood with irradiated B16F10-OVA cells and co-cultured them with CD8⁺ T cells from OT I mouse spleen, we observed a substantial increase in T-cell proliferation in the cultures from the BG34-200-treated groups compared with the cultures from the PBS group (figure 6D).

Taken together, these findings in figures 4–6 suggest that the intravenous administration of BG34-200 could potentiate the untreated TAIMs, the CD11b^highCCR2⁺PU.1^low cells, leading to their transformation into the CD11b^highLy6C^highCCR2⁺PU.1^high cells in the peripheral blood. These cells exhibited a greater capability to differentiate into mo-DCs, which in turn displayed an improved ability to present antigens and stimulate antigen specific T-cell proliferation.

It is noted that in this study, we did not conduct experiment involving the injection of BG34-200-treated cells into tumor-bearing mice and investigate the secondary lymphoid organs, where the activation of the tumor specific immune response occurs. However, these studies have been conducted and are documented in our published paper.18 In that study, we successfully observed the presence of BG34-200⁺ DCs in both the tumor tissues and tumor-draining lymph nodes. Moreover, we performed injections of BG34-200-treated lymph node-derived T cells into tumor-bearing mice, which resulted in a significant improvement in antitumor efficacy.¹⁸

Intravenous administration of BG34-200 induces changes in the internalization rate of binding complexes, phagocytosis rate and intrinsic ICAM-1 status

To examine the phenotype of CD11c⁺ cells generated from groups treated with PBS and BG34-200, we gated on CD11c⁺ mature DCs derived from mouse blood treated with PBS or BG34-200, and analyzed the surface expression levels of MHC I (k^b), MHC II (I-A^b), CD86 (B7.2 co-stimulatory molecules), CD80 (B7.1), CD40, CD11b, and CD54 (intercellular adhesion molecule I, that is, ICAM-1) using corresponding antibodies at saturation concentrations (figure 7A). The two groups showed no significant difference in the percentage of surface expression of these biomarkers. However, a subset of the CD11b⁺ cells from the BG34-200-treated group exhibited lower fluorescence intensity of CD11b compared with the cells from the PBS-treated group (figure 6A, red rectangle). Previous studies have shown that certain ligand-mediated binding to CD11b can trigger internalization of the binding complex, leading to a reduction in CD11b cell surface expression.40

• Download figure
• Open in new tab
• Download powerpoint

Figure 7

BG34-200 treatment modulates function of the TAIMs-derived DCs through triggering internalization of binding complexes, phagocytosis, F-actin cytoskeletal rearrangement, and intrinsic ICAM-1 surface aggregation. (A) The per cent expression of DC-related biomarkers by matured CD11c⁺ cells derived from KPC tumor-bearing mice treated with PBS or BG34-200 treatment, as determined by FACS. Within the CD11b⁺ cells, a fraction of cells of the BG34-200-treated group exhibited lower fluorescence intensity than the cells of the PBS-treated group (red rectangles). Within CD54⁺ cells, a fraction of cells of the BG34-200-treated mice showed higher fluorescence intensity than the cells of the PBS-treated group (green rectangle frame). (B) Schematics of the ex vivo experimental setting to study the internalization rates of binding complexes and the phagocytosis rates. KPC-derived cell line TB32048 was directly injected into pancreas to create the KPC-tumor-bearing mice. For both assays, the TAIM-derived immature DCs were co-cultured with BG34 samples for 30, 60, 90, 120, 150 and 180 min. (C) Averaged MFI of the single CD11c⁺CD11b-AF488⁺BG34-AF647⁺ cells at different co-culture time, as determined by FACS. (D) Average MFI of the single CD11c-PE⁺E coli –AF488⁺ at different co-culture time, as determined by FACS. For C and D, the averaged MFI of the single cells = (Total MFI of the positive cells) / (total number of the positive cells); cells of PBS, BG34-50, BG34-200, and BG34-500 groups were examined in triplicates; Data expressed as mean±SD. (E) Stimulated emission depletion (STED) microscopic imaging (100×) of mature DCs from PBS and BG34 groups. The DCs were stained with ICAM-1-AF488 and F-actin-CF568 and fixed on non-functional glass surface. For statistical analysis, the STED imaging was performed on 50 randomly picked cells of each group. Image analysis of the four groups of 50 cells resulted in the identification of four different status (I, II, III, and IV) (left). The numbers of cells adopting each status were recorded for PBS, BG34-50, BG34-200, and BG34-500 groups to assess the effect of different treatments on ICAM-1 status and F-actin cytoskeletal structure (right). Abs, antibodies; DCs, dendritic cells; E. coli; Escherichia coli; FACS, fluorescence-activated cell analysis; GM-CSF, Granulocyte Macrophage Colony-Stimulating Factor (GM-CSF); ICAM-I, intercellular adhesion molecule I; IV, intravenous; MFI, mean fluorescence intensity; PBS, Phosphate-Buffered Saline; ROI, region of interest; TAIMs, tumor-associated inflammatory monocytes.

In addition, we observed that a subset of CD54⁺ cells from the BG34-200-treated group exhibited higher fluorescence intensity of CD54 (ICAM-1) compared with the cells from the PBS-treated group (figure 6A, green rectangle). Interestingly, DCs’ intrinsic ICAM-1, which is a co-stimulatory ligand that binds to leukocyte function-associated antigen 1 (LFA-1) of T cells to induce cell action, is known to exist in two statuses on the plasma membrane.41 42 The first status is a high lateral mobility status that is incapable of T-cell priming. This is supported by findings that DCs expressing a high mobility ICAM-1 mutant were unable to induce LFA-1 affinity maturation in T cells, resulting in diminished antigen-dependent conjugate formation with T cells. The second status is a constrained clustering status that is important for antigen presentation capability. These two statuses are regulated by DC maturation-induced changes, including phagocytosis and F-actin cytoskeletal rearrangement.

Based on this information, we sought to investigate whether treatment with BG34-200 could trigger these changes. To assess the CD11b-mediated internalization of binding complexes, we co-cultured immature DCs with AF647-tagged anti-CD11b antibodies (clone 1C4) or AF647-tagged BG34 samples for varying periods of time, stained by AF488-tagged anti-CD11b antibodies, and measured the averaged MFI of CD11c⁺ cells that potentially internalized the CD11b-AF488⁺BG34-AF647⁺ complexes (figure 6B). The anti-CD11b-AF647 antibodies (1C4 clones) served as a control since they could induce internalization of the target,34 and we used them to monitor the baseline internalization of CD11b. By plotting the average MFI of the CD11c⁺CD11b-AF488⁺BG34-AF647⁺ cells versus co-culture time, we observed that BG34-200 treatment could induce immature DCs to internalize the binding complexes at a significantly higher rate than BG34 of 50 and 500 kDa, as well as the antibody control (figure 6C).

To assess the phagocytosis rate, we co-cultured immature DCs with AF488-tagged Escherichia coli microparticles for varying periods of time and then examined the average MFI of CD11c⁺ cells that phagocytosed the AF488⁺ E. coli (figure 6B). By plotting the average MFI of CD11c⁺E coli-AF488⁺ cells versus co-culture time, we observed that BG34-200 treatment could induce immature DCs to phagocytose E. coli particles at a significantly higher rate than the PBS, BG34-50 or BG34-500 treatment (figure 6D). Notably, although the BG34-500 could mediate a higher binding affinity to the same peptide residues (figure 2F), it did not appear to induce the internalization of the binding complex and phagocytosis.

To investigate the impact of BG34 treatment on the arrangement of the F-actin cytoskeleton, we generated mature DCs from four groups: PBS, BG34-50, BG34-200 and BG34-500. These DCs were cultured on non-coated (non-functional) coverslips in untreated tissue culture plates, fixed and stained with anti-ICAM-1-PE and CF568 Phalloidin. The coverslips were then mounted on premium frosted microscope slides for two-dimensional Stimulated Emission Depletion Microscopy (STED) imaging. We randomly selected 50 cells from each group for statistical analysis. The STED imaging generated stacks of images of ICAM-1 and the actin filament network at various axial positions. The resulting images revealed novel, distinct structures of ICAM-1 and filamentous actin, which could be categorized into four distinct types (figure 6E). Cells of type I and II displayed evenly distributed ICAM-1 on the cell membrane, while cells of type III and IV displayed aggregated ICAM-1.

In terms of actin cytoskeletal structures, cells of status I are distinguished by a ring-shaped structure and actin that forms ruffles at the cell perimeter. Cells of status II exhibited a dense network of long filaments at the cell perimeter, whereas cells of status III display densely organized short filaments that are evenly distributed in the cytoplasm space and on the cell surface, pointing towards the extracellular space. Finally, cells of status IV possess both a dense network of long filaments and densely organized short filaments in the cellular and intracellular spaces. Among 50 DC morphologies that we examined, the cell frequencies of status I–IV were significantly different among the four groups. While the visualization of actin dynamics in T cells during immune activation is well-established, little is known about actin dynamics in DCs. As a result, we are unable to predict DC functions based on their status I–IV. Nonetheless, our data demonstrate that BG34-200 treatment can induce a unique F-actin filament network that differs from those of the non-treated, BG34-50-treated and BG34-500-treated groups.

Taken together, these data indicate that treatment with BG34-200 can stimulate immature DCs to internalize the CD11d binding complex and phagocytose foreign particles at a different rate than the PBS control, BG34-50 and BG34-500 groups. Furthermore, the BG34-200 treatment-induced rearrangement of the F-actin cytoskeleton in mature DCs, which was accompanied by intrinsic ICAM-1 surface aggregation and an enhanced cross-antigen presentation capability. These effects were not observed in mature DCs derived from the PBS, BG34-50, and BG34-500 groups.