Generation of mouse monoclonal antibody

FABP4 mAbs were generated by immunization of 7-week-old female FABP4 knockout mice with full length human recombinant FABP4 protein as previously described [20]. Briefly, 50 µg protein emulsified with complete Freund’s adjuvant (Cat. #F5506-10ML from MilliporeSigma) was subcutaneously (s.c.) injected into the back of the mice. Mice were boosted with 25 µg protein mixed with incomplete Freund’s adjuvant (IFA, Cat. #P9622-10 × 1ML from MilliporeSigma) by s.c. at day 14 and day 28. The final boosting was conducted at day 50 with 25 µg protein mixed with IFA by intravenous injection. Blood from immunized mice was collected for measurement of anti-FABP4 antibodies by ELISA. Mice with a high titer of anti-FABP4 antibodies were selected for splenocyte collection and fusion with Sp2/0 myeloma cells (ATCC). Hybridoma generation was performed using the ClonalCellTM-HY kit (Cat. #03800 from STEMCELL Technologies). Of note, compared to conventional hybridoma selection and cloning, this method uses a methylcellulose-based semi-solid medium, which increases the diversity of clones that can be easily identified and isolated, enabling hybridoma selection and cloning to complete in a single step [21]. On day 12 after fusion, a total of 1248 single clones in the semi-solid medium were collected and cultured for another 4–7 days. The supernatants were screened for reactivity to human recombinant FABP4 or FABP5 by ELISA. A total of 141 positive clones to FABP4 were selected for further screening with other sources of human or mouse FABP4 protein (Cat. #10,009,549 from Cayman Chemical). Finally, 25 positive clones with specific reactivity to both human and mouse FABP4, but not FABP5, were identified. These which were able to produce ample ascites and with better affinity to FABP4 were selected for antibody production.

Antibody purification from ascites

Ascites of the selected clones of hybridoma cells were developed in 8-week-old female Balb/c mice (n = 6–7 mice/group). Briefly, 0.5 ml pristane was intraperitoneally (i.p.) injected into each mouse. After pristane priming for 7–10 days, 5 × 106 hybridoma cells in 400 \(\:\text\) PBS were i.p. injected into each mouse. Ascites developed 5–7 days after hybridoma cell injection were harvested from the second week using the 19-gauge needles. Cellular components in the ascites were removed by centrifugation at 2000 rpm for 15 min. The monoclonal antibody purification was performed as previously described [22, 23]. In our procedure, ascites was diluted by adding 4 volumes of 60mM acetate buffer with a final pH of 4.5. Caprylic acid (Cat. #C2875-10ML from MilliporeSigma) was added slowly to the ascites with continuous stirring to ensure thorough mixing. The final concentration of caprylic acid in the ascites was 25 µl/ml. The mixture was stirred for 30 min and then centrifuged for 30 min at 10,000 g. The supernatant was collected and mixed with 1/10 volume 10x PBS after nylon mesh filtration. After pH adjustment to 7.4, ammonium sulfate (0.277 g/ml) was slowly added to the solution at 4 °C and stirred for additional 30 min before centrifugation for 30 min at 5000 g. The precipitated antibody was resuspended in small volume of PBS. The purity and quantification of different monoclonal antibodies were determined by SDS-PAGE analysis and BCA quantification, respectively. Antibodies with purity > 85% were used for further applications.

Evaluation of antibody therapeutic efficacy using breast cancer mouse modes

Mouse models of breast cancer were used to evaluate the potential therapeutic efficacy of different clones of the purified FABP4 antibodies, as described above. Mouse experiments were performed according to the approved procedures by the Institutional Animal Care and Use Committee (IACUC) at the University of Iowa. To investigate the effect of antibody produced by hybridoma cells, 0.5 ml pristine was injected intraperitoneally into the Balb/c mice on day 0 to prime the antibody production. On day 7, 12G2 hybridoma cells (5 × 106 cells in 100 µl PBS /mouse) and control Sp2/0 cells (5 × 106 cells in 100 µl PBS/mouse) were injected intraperitoneally into separate groups of mice. MMT tumor cells (1 × 106 cells in 100 µl PBS/mouse) were orthotopically injected into the mammary fats pads to monitor tumor growth. To further test the efficacy of purified 12G2 antibody from ascites, C57BL/6-derived mammary tumor cells E0771 (5 × 105 cells in 100 µl PBS/mouse) and highly aggressive Balb/c-derived 4T1 tumor cells (1 × 105 cells in 100 µl PBS/mouse) were orthotopically injected into mammary fats of C57BL/6 mice or Balb/c mice (8–10 weeks old), respectively. After tumor injection, mice were randomly divided into several groups and treated with different clones of purified antibodies (5 mg-30 mg/kg, twice/week). Mice treated with the same volume of PBS were used as controls. When the tumors were palpable, the length and width of the tumors were measured by a caliper every three days. The volume of the tumors was calculated using the formula of 0.5 x length x width [24] as described before [25, 26]. To monitor antibody efficacy against human breast cancer, MCF7 cells were utilized in a xenograft mouse model. Matrigel (Cat. # 354,262 from Corning) was mixed with MCF7 cells in PBS (Matrigel/PBS = 1:1) and injected (3 × 106 cells in 100 µl mixture/mouse) using a 23G needle into mammary fat pads of SCID mice. Tumor growth was monitored in antibody- or PBS-treated mice similarly to the method described above.

Production of chimeric and humanized anti-FABP4 antibodies

For chimeric antibodies, anti-FABP4 hybridoma clones (e.g., 12G2, 6H10) were sequenced and DNA sequences of the VH and VL regions were identified. Recombinant chimeric antibodies consisting of mouse VH and VL and human IgG1 constant regions were expressed and purified in HEK293 cells. For humanized antibody production, parental VH and VL sequences were run through a CDR grafting algorithm to transfer the CDRs from the original framework onto the most matched human germline sequences. To ensure that no highly undesirable sequence liabilities were introduced into the humanized sequences, identified high-risk motifs were removed through mutagenesis. A total of 16 antibody variants composed of different pairings of 4 humanized heavy chains and 4 humanized light chains were generated using HEK293 mammalian cells. All chimeric and humanized antibodies were made by Absolute Antibody (United Kington) with high purify and low endotoxin (< 0.05EU/mg) for in vivo studies.

Aldehyde dehydrogenase (ALDH) assay

The ALDEFLUOR kit (Cat. #01700 from STEMCELL technologies) was used to detect ALDH activity for both tumor tissues and tumor cell lines. To obtain single cells from tumor tissues, E0771 and MCF7 tumors were removed from euthanized mice and mechanically dissociated into smaller fragments. These fragments were then treated with 6 ml tri-enzyme solution containing 0.5 mg/ml collagenase type 2 (Cat. #LS004177 from Worthington Biochemical), 0.2 mg/ml hyaluronidase (Cat. #0215127590 from MP Biomedicals), and 0.02 mg/ml DNase I (Cat. #E1009-A from ZYMO research) in RPMI-1640 medium containing 5% FBS and incubated at 37 °C for 45 min on an orbital shaker at speed of 50 rpm. Following enzymatic digestion, the cell suspensions were collected by vortexing, filtration, removal of tri-enzyme solution, and two washes with cold 1x PBS. The detection of ALDH activity from both single- cell suspensions derived from tumor tissues and tumor cell lines were followed the protocol provided in the ALDEFLUOR kit.

Tumor migration, tumor invasion, and limiting dilution assays

To assess the blocking activity of anti-FABP4 antibodies, the molar ratio of antibody to FABP4 antigen was set at 1: 2. Individual antibodies (1 µg/ml) and human FABP4 protein (200ng/ml) were mixed for 15 min before performing following assay. FABP4 and PBS alone were served as controls. (1) wound-healing migration was performed to assess whether antibodies were able to inhibit FABP4-mediated tumor cell migration. To induce a linear wound in the cellular monolayer, the confluent cells were mechanically scratched using a 200 µL plastic pipette tip in a six well-plate containing 2.5 ml cell culture medium. Subsequently, the scratched monolayer was carefully washed with pre-warmed 1 x PBS to eliminate any debris. Following a 96-hour incubation period at 37 °C, the migration of cells towards the wound site was captured using light microscopy, and the migration area was quantified using Image J software. (2) For tumor invasion assay, MCF-7 cells were cultured to form spheres in hanging drops of culture medium on the lid of cell culture dishes as previously described [16]. Briefly, following a seven-day incubation period, the spheroids from the lid were transferred into an equivalent volume and combined with rat tail type I collagen (Cat. #A1048301 from Fisher Scientific), reaching a final concentration of 1.7 mg/ml. This mixture was then embedded in a 24-well plate to establish a 3D culture system. FABP4/antibody mixture, FABP4 protein and control PBS were added into 1 ml cell culture medium, respectively. Quantitative analyses were performed by measuring the maximal invasive distance (longest distance from the spheroid radius) and the invaded area (total invaded area minus the spheroid area) using Image J software. (3) For in vitro limiting dilution assay (LDA), tumor cells were serially diluted to obtain cell concentration at a range of 1000, 500, 250, 125, 62, 31, 16, 8 cells and seeded into an ultra-low attachment 96-well plate containing 200 µL cell culture medium, exposed to FABP4-antibody mixture, FABP4 protein and control PBS for a duration of 48–96 h. Subsequently, cell spheres were determined using microscope and calculated the cancer cells initiating frequency and significance using online software (Extreme limiting dilution analysis, ELDA @ http://bioinf.wehi.edu.au/software/elda/index.html) following the methodology outlined by Hu and Smyth [27].

Characterization of antibody/antigen binding

The binding of anti-FABP4 antibodies with FABP4 was evaluated by ELISA and BIAcore assays. For ELISA, FABP4 protein or biotinylated FABP4 epitope peptides (Mimotopes) was diluted with 1 x PBS and coated either to a non-coated 96-well plate or to a streptavidin-coated 96-well plate at a final volume of 100 µL overnight. After blocking with 5% BSA at room temperature for 1 h, anti-FABP4 antibodies (e.g., chimeric, humanized) were diluted with 5% BSA solution and added into indicated wells. The plate was washed three times using 200 µL of 1 x PBS containing 0.5% Tween-20 and incubated with 100 µL of secondary antibody solution containing goat anti-human IgG conjugated with HRP (Cat. #A18805 from Invitrogen) at a dilution of 1:10000 in 5% BSA solution for 1 h. Color development was performed by adding 100 µL of substrate solution and incubating for 5 min at room temperature before the reaction was stopped by 100 µL of 2 N sulfuric acid. OD value was acquired using a BioTek Synergy LX Multimode Reader. Binding affinity measurement was performed by ProteoGenix (France). Briefly, human FABP4 (10 µg/ml) was immobilized on CM5 sensor chip of BIAcore 8 K using maleimide EDC/NHS coupling. A stable cell pool of 12G2 V9 antibody (s-V9) was produced using Chinese hamster ovary (CHO) cells (ProteoGenix, France). Antibody (s-V9) at a defined concentration (ranging from 0.156 to 2.5µM) was flowed over CM5 chip and response captured over time, showing the progress of the interaction and association/dissociation cycle. After different concentrations are successively tested, the kinetics parameters and affinity are calculated using the BIA-evaluation software.

Immunophenotype analysis by flow cytometry

Immune phenotypes were performed using multi-color staining panel designed by improved version of full spectrum viewer in Cytek Cloud. Signle-cell suspension of the primary tumor was resuspended in 1 x PBS containing 0.5% FBS and kept in the ice all the time. Cells were pre-incubated with anti-mouse CD16/CD32 antibody (Cat. # 101,302 from Biolegend) for 5 min to block Fc receptors. Surface staining was prepared using the following antibodies: Zombie-violet (Cat. #423,108, Biolegend), anti-mouse CD45 (Cat. #103,116, Biolegend), anti-mouse CD11b (Cat. #612,800, BD), anti-mouse CD3 (Cat. #100,355, Biolegend), anti-mouse CD4 (Cat. #740,208, BD), anti-mouse CD8 (Cat. #752,642, BD), anti-mouse F4/80 (Cat. #123,120, Biolegend), anti-mouse MHCII (Cat. #107,604, Biolegend), anti-mouse Ly6G (Cat. #127,664, Biolegend), anti-mouse NK1.1 (Cat. #108,736, Biolegend), anti-mouse B220 (Cat. #103,232, Biolegend), and anti-mouse CD11c (Cat. #117,349, Biolegend). The intracellular cytokines staining with anti-mouse IL-6 (Cat. #504,508, Biolegend) and anti-mouse TNFα (Cat. #506,338, Biolegend) were fixed and permeabilized using True-Nuclear transcription factor buffer set (Cat. #424,401 from Biolegend) according to the manufacturer’s introduction. All samples were acquired with an Cytek Aurora instrument. Data were analyzed with FlowJo (BD).

Spatial transcriptomics and analysis

Fresh tumor tissues were removed and placed in a petri-dish and embedded with room temperature OCT without any bubbles on the tissue’s surface. The embedded samples were transferred into the cryo mold. The cryo mold containing OCT-embedded samples were put into the metal beaker with 2-methylbutane in a dewar of liquid nitrogen until the OCT was solidified. Sample cryosectioning, affixment to cDNA capture slide, H&E staining, tissue permeabilization, RNA capture, and cDNA synthesis were performed according to the 10 x Genomics Visium spatial transcriptomics’ methods.

The four visium libraries (two PBS tumors, and two S-V9-treated tumors) were sequenced on an Illumina NovaSeq 6000 located in the Iowa Institute of Human Genetics (IIHG) Genomics division. Paired-end reads were demultiplexed and converted from the native Illumina BCL format to fastq format using an in-house python wrapper to Illumina’s ‘bcl2fastq’ conversion utility. The data were deposited to GEO repository (GSE264099). Bioinformatic analysis was carried out by the IIHG Bioinformatics division. Fastq data were merged across lanes and the PE reads were used as input for the 10X SpaceRanger pipeline in ‘count’ mode (v1.3.1). SpaceRanger was run on the Argon High-Performance Computing (HPC) cluster at the University of Iowa using 32 cores and 128GB of RAM per sample. The reference transcriptome was specified as ‘mm10-2020-A’ and the chemistry was specified as ‘Spatial 3’ v1’. QC analysis of the four samples showed no quality problems for each sample other than an alert that “Low Fraction Reads in Spots” was detected for 3 of 4 samples. Filtered barcode matrices were used as input for downstream analysis in Seurat (v5). Four Seurat objects were created from the individual barcode matrices and quality control (QC) metrics were visualized as violin plots that included the number of genes (nFeature), number of UMI (nCount) and percentage of mitochondrial UMI (percent_mt). We filtered cells that have less than 100 features (low-quality cells). The filtered datasets were subjected to normalization, detection of variable features, scaling/centering and PCA analysis. Following this, the sample layers were integrated together using the “RPCA” method of integration available in Seurat 5 (https://satijalab.org/seurat/articles/integration_rpca). To cluster the cells, we used K-nearest neighbors (KNN) networks based on the calculated PCs. Modularity optimization was applied (Louvain method, resolution = 0.1) and a UMAP embedding was calculated. Searching for DEGs (cluster biomarkers), we found markers for every cluster compared with all remaining cells using the Wilcoxon Rank Sum test and a log2FC threshold of 0.4 and expressed in more than 30% of the cells. Cluster-wise DE analysis of the treatment effect of S-V9 vs PBS was carried out by using the “FindMarkers” function on the integrated object. Pathway analysis was carried out using g: Profiler (https://biit.cs.ut.ee/gprofiler/gost) and iPathwayGuide software (AdvaitaBio).

H&E staining

Fresh tissues were obtained after removing primary tumor from euthanized mice. Lung samples were collected from the right inferior lobe and fixed immediately in 10% neutral buffered formalin. Air microbubbles were removed by placing lungs in a vacuum chamber for 5 min, then re-fixed the lungs into the fresh 10% neutral buffered formalin for 24 h. Following serial alcohol dehydration (50%, 75%, 95%, and 100%), the samples were embedded in paraffin. The paraffin-embedded samples were sliced into 8 μm sections and stained in the DRS-601 Auto Stainer with hematoxylin and eosin (H&E) for 1 min. Slides were mounted with VectaMount® Express Mounting Medium (vector laboratories, H-5700-60), and were scanned by slide scanner (Leica Aperio GT 450) for quantification analysis. The metastatic tumor number and area was analyzed by the SlideViewer 2.7.0.191696 software.

Quantification of serum biomarkers

Serum samples were collected at the end point of tumor challenge mouse model. Samples from each mouse were divided into aliquots and preserved in an -80 C freezer for future purposes. The quantification of FABP4 (Cat. #CY-8077, MBL), IL-6 (Cat. #431,301, Biolegend) and glucose (Cat. #81,692, Crystal Chem) levels were performed separately using ELISA kits in accordance with the instructions provided by the manufacturers.

Statistical analysis

All data were presented as the mean ± SD unless notified specifically. For in vitro and in vivo experiments involving two groups, a two-tailed, unpaired Student’s t-test was performed by GraphPad Prism 9. For experiments containing more than two groups, a two-way ANOVA with multiple comparison test was used by GraphPad Prism 9. Statistical significance was defined as a p-value of less than 0.05.