Genetically modified mouse model and NP137 administration
Table of Contents
Floxed homozygous Pten (C;129S4-Ptentm1Hwu/J, hereafter called Pten f/f) Cre:ER (B6.Cg-Tg(CAG-CRE/Esr1* 5Amc/J) mice were obtained from the Jackson Laboratory. Cre:ER+/− Pten f/f mice were bred in a mixed background (C57BL6; 129S4) by crossing Pten f/f and Cre:ER+/− mice. To obtain mice carrying both Pten floxed alleles (Pten f/f) and a single Cre:ER (Cre:ER+/−), Cre:ER+/− Pten f/+ mice were backcrossed with Pten f/f mice. To induce deletion of floxed alleles, tamoxifen (Sigma-Aldrich) was dissolved in 100% ethanol at 100 mg ml–1. Tamoxifen solution was emulsified in corn oil (Sigma-Aldrich) at 10 mg ml–1 by vortexing. To induce Pten deletion, adult mice (4–5 weeks old) were given a single intraperitoneal injection of 0.5 mg of tamoxifen emulsion (30–35 μg mg–1 body weight). Three weeks after tamoxifen injection, mice were treated via intraperitoneal injection of 100 µl of NP137, or its isotypic control NP001 diluted in PBS, at 10 mg kg–1 every 2 days. Animal care and housing were in accordance with institutional European guidelines from the CEEA local Ethical committee of Lleida University concerning Pten mouse experiments. General behaviour and weight were monitored three times per week and animals were killed in the event of strong alteration or weight loss under 20%. All mice were always killed before terminal tumour progression, and endometrial tissues were analysed blind by a pathologist.
Endometrial samples collected in the Biomedical Research institute of Lerida (Spain) were frozen and sent to the Cancer Research Center of Lyon (France) in dry ice. Samples were cryoground to obtain tumour powder, which was processed for total RNA extraction using the Nucleospin RNA Plus kit (Machery-Nagel) according to the manufacturer’s instructions. Expression of mRNA was measured using a NanoDrop1000 (Themo Scientific). RNA was retrotranscribed using the T100 ThermoCycler (Bio-Rad) and the iScript cDNA Synthesis Kit (Bio-Rad) according to the manufacturer’s instructions. RT–qPCR was performed using LC480 qPCR (Roche) and OneGreen Fast qPCR Premix (Ozyme) according to the manufacturers’ instructions.
Patient analysis: patient microbiopsy RNA was extracted with the RNA easy FFPE kit (Qiagen). The RNA-seq library was produced from 100 ng of RNA with the Illumina TruSeq Exome kit (RNA Library Prep for Enrichment & TruSeq RNA Enrichment) according to the manufacturer’s instructions and then sequenced with an Illumina NovaSeq 6000. FASTQ files were then processed with STAR (v.2.7.10a). Briefly, FASTQ files were mapped to the human reference genome (gencode.v.27) and aligned reads were converted for counting with STAR. The quality of FASTQ files was also checked, by FATSQC (v.0.11.9). RNA-seq analysis was performed with R (v.4.0.3) and the DESeq2 package (v.1.30.1). log2-Transformed transcripts per million were calculated, and we performed EMT score calculation as previously described20.
Murine endometrial cancer model: tumours were collected, snap-frozen and cryoground. RNA was extracted with a standard kit (Macherey-Nagel). The RNA-seq library was produced from RNA with Illumina TruSeq Stranded Total RNA Library Prep Human/Mouse/Rat, according to the manufacturer’s instructions, then sequenced with Illumina NOVASeq. FASTQ files were processed as described above, except for the reference genome which was Mus_musculus.GRCm38 (GENECODE release 25). Analysis was done with DESeq2 (v.1.30.1) and ggplot (v.3.1.3) packages in R (v.4.0.3). EMT scoring and heatmaps were done with log2 fragments per kilobase exon per million mapped reads values.
Endometrial metastasis biopsies were dissociated for single-cell RNA-seq using the Tumor Dissociation kit by Miltenyi Biotec (no. 130–095-929). Briefly, the biopsy was placed in RPMI medium in a Petri dish on ice and cut into small pieces (2–4 mm) after removal of necrotic tissue. Pieces were then infused with the RPMI/enzyme mix (Miltenyi Biotec), transferred to a gentleMACS C tube containing RPMI/enzyme mix, attached to the sleeve of the gentleMACS Octo Dissociator and run using the programme 37_h-TDK1. After completion of the programme the cells were spun down at 300g for 7 min at 4 °C, resuspended in RPMI, passed through a 70 μm strainer and centrifugation was repeated. The cell pellet was treated with 500 µl of ACK solution for 5 min at room temperature and lysis then stopped with 5 ml of RPMI/10% FBS. After centrifugation, cells were resuspended in 100 µl of RPMI. The number of live cells was determined with a Luna-FL Dual fluorescence cell counter (Logos Biosystems) to obtain an expected cell recovery population of 10,000 cells per channel, loaded on a 10x G chip and run on the Chromium Controller system (10x Genomics) according to manufacturer’s instructions. Single-cell RNA-seq libraries were generated with the Chromium Single Cell 3′ v.3.1 kit (10x Genomics, no. PN-1000121) and sequenced on the NovaSeq 6000 platform (Illumina) to obtain around 50,000 reads per cell.
Except when specifically mentioned, all analyses were performed with R/Bioconductor packages, R v.4.2.2 (2022-11-10 r83330) (https://cran.r-project.org/; http://www.bioconductor.org/) in a Linux environment (x86_64-pc-linux-gnu (64-bit)).
Filtered barcoded matrices from single-cell RNA-seq data were imported into R using the Seurat package (v.4.1.1). Doublets were detected with DoubletFinder (v.2.0.3) and filtered out, together with cells showing a low number of features (nFeature_RNA < 500) or a high percentage of mitochondrial genes (above 25%). Seurat functions were used for normalization (SCTransform), merging, dimensional reduction and clustering. Initial cell type identification was based on consensus from several automated cell annotation packages (SciBet’ v.1.0, SingleR v.1.10.0 and scType (https://github.com/IanevskiAleksandr/sc-type/blob/master/README.md)). T cell subtypes were visually inspected and manually curated after further annotation with ProjecTILs (v.3.0.3) and the python implementation of CellTypist (v.1.3.0). EMT signature scores were calculated by three methods (Seurat’s AddModuleScore function, UCell (v.2.2.0) and AUCell (v.1.18.1)) and using two different gene lists (MsigDb Hallmark EMT pathway and the PanCancer EMT signature20). Pathway enrichment analyses were performed with the ‘escape’ package (v.1.6.0). Additional visualizations were based on functions from Nebulosa (v.1.6.0), Scillus (v.0.5.0) and ggplot2 (v.3.3.6).
Spatial RNA-seq matrices and images were imported into R using the Load10X_Spatial function of the Seurat package. For each sample, only those spots with more than 1,000 features were kept for downstream preprocessing, including SCTransform normalization, dimensional reduction and clustering. Most analyses were performed independently for each sample (without merging or integration). Two strategies were used for cell type annotation: label transfer following integration of the single-cell RNA-seq data described above, and manual annotation using known markers for major cell types.
Inference of cell–cell communication was done with CellChat (v.1.6.0), for both single-cell and spatial RNA-seq data.
Histology and IHC analysis
Cre:ER+/− Pten f/f mice were euthanized by cervical dislocation after 3 weeks of NP137 treatment. Endometrial samples were collected and formalin fixed overnight at 4 °C. Tumours were paraffin embedded for further histologic analysis. Paraffin blocks were sectioned at 3 µm and dried for 1 h at 65 °C before the pretreatment procedures of deparaffinization, rehydration and epitope retrieval in the pretreatment module at 95 °C for 20 min in 50× Tris/EDTA buffer. Before staining of sections, endogenous peroxidase was blocked. Representative images were taken with a Leica DMD108 microscope.
Immunohistochemistry was performed on an automated immunostainer (Ventana discoveryXT, Roche) using the rabbit Omni map DAB Kit. Sections were incubated with specific antibodies targeting EpCAM (no. ab71916, abcam), cleaved caspase-3 (no. 9661, Cell Signaling Technologies), netrin-1 (no. CPA2389, Cohesion Biosciences), Unc5B (no. 13851S, Cell Signaling Technologies) and CD8 (no. 4SM15, eBioscience). Staining was by anti-rabbit horseradish peroxidase, visualized with 3,3′-diaminobenzidine as a chromogenic substrate and counterstained with Gill’s haematoxylin. Histological quantifications were performed with Halo software (Indica Labs).
Sequential chromogenic multiplex IHC for vimentin/pancytokeratin (panCK), as previously described, was performed on tumour sections from patients included in the NP137 clinical trial and that were collected at C1D1 and C3D1. Dewaxed 4-μm-thick, paraffin-embedded tissue sections were subjected to two successive steps of IHC on a Ventana discovery XT platform (Ventana, Roche Diagnostics) using the REDMap and DABMap detection systems according to the manufacturer’s recommendations. In a first step for vimentin expression, slides were incubated with mouse monoclonal anti-human vimentin for 1 h (mouse, clone V9, Leica, no. NCL-L-VIM-V9, dilution 1:100) and incubated with rabbit monoclonal anti-mouse secondary antibody for 20 min (clone M1gG51-4, abcam, no. 125913, dilution 1:750). The slides were then incubated with biotinylated anti-rabbit secondary antibody for 24 min (Vector Laboratories, dilution 1:200) followed by the addition of the streptavidin–alkaline phosphatase complex. Immunostaining was detected by incubation with naphthol and Fast red. Tissue sections were counterstained with Gill’s haematoxylin, dehydrated and mounted. Whole histological slides were digitized at ×20 magnification using a Hamamatsu 2.0 HT scanner. After removal of coverslips, slides were incubated in 100% ethanol until complete erasure of red colour. In a second step, to show panCK expression, the slides were incubated with mouse monoclonal anti-panCK antibody for 1 h (mouse, clone CKAE1AE3, Dako Belgium, no. M351529-2, dilution 1:150) and then with rabbit monoclonal anti-mouse secondary antibody for 20 min. The slides were then incubated with biotinylated anti-rabbit secondary antibody for 28 min (Vector Laboratories, dilution 1:200) followed by the addition of streptavidin–alkaline phosphatase. PanCK immunostaining was detected by incubation with naphthol and Fast red. The IHC slides were counterstained with Gill’s haematoxylin, dehydrated, mounted and again digitized. Image processing and analysis for Hynrid EMT score computation were performed using Visiomorph DP 2018.4 to determine vimentin and panCK co-expression in each tissue slide. Briefly, each pair of vimentin- and panCK-labelled virtual slides, which were acquired from the same tissue section, was subjected to image registration. Vimentin- and panCK-positive areas were automatically detected in the aligned virtual slides to evidence their co-expression in tumour cells. This co-expression was evaluated on whole slides at ×10 magnification to take into account potential imperfections. Manual corrections were carried out to exclude irrelevant sample parts, such as necrosis. Cell nuclei negative for both markers were also excluded, to focus only on cytoplasmic areas where colocalization could occur.
Spatial transcriptomics using Visium FFPE technology
FFPE tissue sections were placed on Visium slides and prepared according to the 10x Genomics protocols. After H&E staining, imaging and de-crosslinking steps, tissue sections were incubated with human-specific probes targeting 17,943 genes (10x Genomics, Visium Human Transcriptome Probe Set v.1.0). Probes hybridized on mRNA were captured on Visium slides and a gene xpression library prepared following the provided protocol and sequenced on an Illumina NovaSeq 6000 with 50,000 reads per spot targeted sequencing depth.
For each FFPE section, FASTQ files and histology images were processed using 10x Space Ranger v.2.0 to obtain the gene expression matrix associated with each spot.
Seurat v.4 (https://satijalab.org/seurat/) in R 4.1 was used to perform the analysis. Briefly, filtered matrices were loaded, merged per patient and spots with fewer than 1,000 detected genes were removed. Following SCTransform normalization we subset the tumoural spot according to pathologist spot identification then calculated the EMT gene set enrichment score (escape R package) with the UCell method.
NP137 clinical trial
NP137 is a first-in-human Phase I trial with a dose-escalation part followed by extension cohorts (NCT02977195) conducted in adult patients with advanced or metastatic solid tumours. The dose-escalation part was initiated using an accelerated dose titration with one patient per dose level until the occurrence of a grade 2 or higher drug-related adverse event. Following occurrence of a grade 2 NP137-related adverse event, patients were enrolled in a slower dose-escalation design with at least three patients per dose level using a modified continual reassessment method. Additional patients were enrolled in three biomarker cohorts, at dose levels that had been declared safe, and underwent paired biopsies for pharmacodynamics purposes. In the dose-escalation part, 19 patients were enrolled in seven dose levels (1–20 mg kg–1, intravenous, Q2W). No dose-limiting toxicities were observed and 11 (58%) patients had infusion-related reactions of grade 12 severity, all at doses of 4 mg kg–1 and above14. Based on available data, 14 mg kg–1 Q2W was selected as the recommended Phase 2 dose. Two extension cohorts were opened, including one in patients with hormone receptor-positive EC (enrolment closed in October 2021).
The trial was conducted according to Good Clinical Practice guidelines, the Declaration of Helsinki and relevant French and European laws and directives. All patients provided written informed consent.
Statistical analyses were performed on Prism (GraphPad Software). In the figure legends, n denotes the total number of replicates. All statistical tests were two-sided. For mouse experiments, statistical methods were not used to predetermine necessary sample size but sample size was chosen based on pilot experiments applying appropriate statistical tests that could return significant results. Survival curves were analysed using the log-rank (Mantel–Cox) test. Population ratios were analysed by chi-squared tests. qPCR expression, caspase-3 IHC and thyroid weights were compared by Mann–Whitney test. For data involving patients, gene expressions and EMT scores were compared by t-test.
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.