Abstract
Background/Aim: Patients with hypoxic bladder cancer benefit from hypoxia modification added to radiotherapy, but no biomarkers exist to identify patients with hypoxic tumours. We, herein, aimed to implement oxygen-enhanced MRI (OE-MRI) in xenografts derived from muscle-invasive bladder cancer (MIBC) for future hypoxia biomarker discovery work; and generate gene expression data for future biomarker discovery. Materials and Methods: The flanks of female CD-1 nude mice inoculated with HT1376 MIBC cells. Mice with small (300 mm3) or large (700 mm3) tumours were imaged, breathing air then 100% O2, 1 h post injection with pimonidazole in an Agilant 7T 16cm bore magnet interfaced to a Bruker Avance III console with a T2-TurboRARE sequence using a dynamic MPRAGE acquisition. Dynamic Spoiled Gradient Recalled Echo images were acquired for 5 min, with 0.1mmol/kg Gd-DOTA (Dotarem, Guerbet, UK) injected after 60 s (1 ml/min). Voxel size and field of view of dynamic contrast enhanced (DCE)-MRI and OE-MRI scans were matched. The voxels considered as perfused with significant post-contrast enhancement (p<0.05) in DCE-MRI scans and tissue were further split into pOxyE (normoxic) and pOxyR (hypoxic) regions. Tumours harvested in liquid N2, sectioned, RNA was extracted and transcriptomes analysed using Clariom S microarrays. Results: Imaged hypoxic regions were greater in the larger versus smaller tumour. Expression of known hypoxia-inducible genes and a 24 gene bladder cancer hypoxia score were higher in pimonidazole-high versus -low regions: CA9 (p=0.012) and SLC2A1 (p=0.012) demonstrating expected transcriptomic behaviour. Conclusion: OE-MRI was successfully implemented in MIBC-derived xenografts. Transcriptomic data derived from hypoxic and non-hypoxic xenograft regions will be useful for future studies.
Hypoxia is a feature of most solid tumours (1) associated with genomic instability and treatment resistance, particularly to radiotherapy. Improved responses can be achieved by using hypoxia modification during radiotherapy (2). To date, there are no biomarkers in clinical use that identify patients with hypoxic tumours. Hypoxia scores derived from the collective expression of multiple genes can predict benefit from hypoxia modification (3-6). Medical imaging techniques can identify regions within tumours for additional treatment dose and are suitable for serial measurements (7). Two magnetic resonance imaging (MRI) techniques, oxygen-enhanced MRI (OE-MRI) and Dynamic Contrast Enhanced (DCE-MRI) when combined enable the non-invasive mapping of hypoxia in tumours (8). DCE-MRI identify regions within tumours that are perfused whilst OE-MRI show regions that utilise oxygen.
The aim of the study was to assist hypoxia target discovery work by: a) implementing OE-MRI/DCE-MRI in a xenograft model of bladder cancer and (b) using laser dissected tissue from pimonidazole demarcated regions in the xenograft model to produce a publicly available gene expression database.
Materials and Methods
Figure 1 shows a schematic of the xenograft model preparation, imaging, and RNA analysis.
Schematic of the experimental design for the preparation of the cell line-derived xenografts, imaging, and transcriptional profiling.
Bladder cancer cell culture. HT1376 (ATCC, Virginia) MIBC cells were cultured in Eagle’s Minimum Essential Medium (EMEM) (Sigma-Aldrich, St. Louis, MO, USA) supplemented with 10% Foetal Bovine Serum (FBS) (Gibco, Thermo-Fisher Scientific, Waltham, MA, USA). The cells were screened for mycoplasma and Murine Hepatitis Virus (MHV) prior to implantation into mice.
Animal handling procedures. The experiment was conducted on 8-week-old female CD-1 nude mice (n=2), body weight 20 g purchased from (Charles River Laboratories, Wilmington, MA, USA). The mice, housed in cages under an Individually Ventilated Caging (IVC) system, were allowed to acclimatise for 7 days prior to implantation. The mice were fed ad libitum with standard mouse diet and water during the experiment. All experiments were carried out under UK Animals (Scientific Procedures) Act 1986, UK Home Office regulations.
HT1376 cell line derived xenograft (CDX) model. Using a 25-gauge needle, HT1367 cells (107 cells in 100 μl of phosphate buffered saline) were subcutaneously inoculated into the flanks of female CD nude mice on the lower quadrant, 1 cm from the tail base. This was counted as day zero. Pre-injection sites were swabbed with chlorohexidine wipes and Betadine (iodine) post swabs for sterilisation. The injection was done under isoflurane anaesthesia. Mice were sedated using 3-5% isoflurane for around 5 minutes in total (knockdown and chip implant (ID microchipped to distinguish mice) and cell inoculation. After subcutaneous injection, mice were allowed to fully recover in a warming box if needed and observed every 10-15 min for 1h. They were then returned to their cages and provided with food and water during the study period.
Tumour measurement and monitoring. Tumour volume measurements and weight were recorded every other day using a digital calliper and scale. The tumour volume was calculated using the formula: H×W×L where H is the tumour height, W is the tumour width and L is the tumour length, in mm. Mice were euthanized by a schedule 1 technique when the tumour reached 300 mm3 (small tumour) and 700 mm3 (large tumour) in size.
MRI and pathology correlation. One mouse was imaged at tumour size 300 mm3 and the other at 700 mm3. Mice had intraperitoneal (ip) injections with the hypoxia biomarker pimonidazole (Hypoxyprobe-1-pimo) (60 mg/kg) (Chemicon Ltd, Feltham, UK) immediately before MRI imaging. The mice were then placed in an Agilant 7T 16cm bore magnet MRI interfaced to a Bruker Avance III. Animals were scanned using T2-TurboRARE, OE-MRI and DCE-MRI. OE-MRI scans were acquired using a dynamic MPRAGE acquisition: TR/TE=6,000/3.10 ms, TI=2,000 ms, voxel size=0.234×0.234×0.5 mm, 30 slices with total duration of 32 min. The OE-MRI acquisition was carried out with the mouse breathing medical air then switched from medical air to 100% O2 midway through for the remainder of the scanning session. Following OE-MRI, DCE-MRI was performed to determine perfused volume pixel (voxels). Dynamic Spoiled Gradient-Recalled (SPGR) images were acquired for 5 min 52 s (flip angle=30°, TR/TE=10.2/1.9 ms), with 0.1 mmol/kg Gd-DOTA (Dotarem, Guerbet, UK) injected after 60 s at 1ml/min via a tail vein cannular, under anaesthesia, during the scan for contrast enrichment. Voxel size and field of view (FOV) of DCE-MRI and OE-MRI scans were matched. The voxels with significant post-contrast enhancement (p<0.05) in DCE-MRI scans were classified as perfused. From the perfused voxels, tissue was further split into two sub-regions pOxyE (normoxia) and pOxyR (hypoxia) based on post-contrast enhancement on OE-MRI (p<0.05). At the end of the study, mice were euthanised by schedule 1 method and tumours cut in half and frozen in liquid nitrogen.
DCE-MRI image processing and analysis. DCE-MRI (OE-MRI) images were analysed using MATLAB® version R2017a.
Laser capture microdissection (LCM), IHC and histology. Sections were obtained from across the middle of the tumour using a cryostat. Two×4 μm sections were cut for immunohistochemistry (IHC) to localise pimonidazole, and hematoxilyn and eosin (H&E) staining to identify necrotic regions. Further sections (10 μm) were prepared for microdissection using laser capture microdissection (LCM) from pimonidazole-high and pimonidazole-low regions The pimonidazole (Hypoxyprobe Mab-1) antibody was used at a dilution of 1/50 (1.2 μg/ml) to localise pimonidazole in one of the 4 μm sections (Figure 2). Necrotic regions were identified and excluded using the H&E-stained sections.
(A) Regions of oxygen enhancement (i) in blue, perfusion (ii) in red and a Montage (iii) of i and ii illustrating regions of perfusion and oxygen enhancement in white. (B) Montage of images derived by OE-MRI and DCE- MRI for the small tumor and large tumor. Top image in each panel reflects the right side of the tumor and bottom image reflects the left side of the tumor. pOxyE: Perfused normoxic areas (yellow); pOxyR: perfused hypoxic areas (blue); not perfused areas identified (grey).
RNA extraction, sequencing, and Clariom S microarray. RNA from xenograft tissue was extracted following the manufacturer’s protocol (Thermo-Fisher, UK). Transcriptomes were generated using Clariom S Affymetrix transcriptome arrays. RNA extracted from the 6 micro-dissected samples were subject to Clariom S analysis using the Clariom S pico HT human assay (Thermo Fisher Scientific). Sample hybridisation on Clariom S arrays was carried out by Yourgene Health (Manchester, UK). Batches of CEL files were GC SST (Signal Space Transformation with probe Guanine cytosine count correction) RNA normalised using Affymetrix.
Array power tools. Microarray analytical tools are available on the Thermo-Fisher website (9). The log2 summarised gene level expression values generated were batch corrected using the comBat function from the Bioconductor package sva (10). Hypoxia scores were calculated as the median expression of the 24 signature genes (3).
Data sharing. All transcriptomic data from this study are deposited in the GEO repository (Geo{at}ncbi.nlm.nih.gov) under accession GSE262610.
Results
Identification of oxygen-rich and perfused regions. Figure 2A shows representative slices from OE-MRI (A) and DCE-MRI (B) images of a small and a large bladder cancer xenograft. The montage map (C) identifies regions that are perfused and demonstrate enhanced signal during 100% O2 breathing (p-OxyE) in white or no enhancement (p-OxyR) in red. Figure 2B shows montage plots of the perfusion (DCE-MRI) and oxygen-enhancement (OE-MRI) maps in slices from left to middle and from right to middle of the small and large tumour. Regions that are perfused normoxic (pOxy-E) are shown in yellow and regions that are perfused hypoxic (pOxy-R) are shown in blue.
Table I shows the overall levels of normoxia (P-OxyE) and hypoxia (P-OxyR) in the small and large tumour and the proportion of tumours that are necrotic. Both hypoxia and necrosis are higher in the larger compared with the smaller tumour.
Percentages of the small and large tumour that were perfused.
Identification of non-necrotic hypoxic and non-hypoxic regions. Figure 3 shows adjacent tumour sections from frozen xenografts subjected to H&E staining (A) and IHC using a pimonidazole antibody (B). Pimonidazole-positive and - negative regions excluding necrotic regions were micro-dissected from contiguous sections to the H&E and pimonidazole-guide slides. Figure 3C shows regions stained with the pimonidazole-antibody in sections immediately adjacent to the upper and lower sections micro-dissected for RNA extraction. The locality of pimonidazole-staining is similar in both indicating that the regions of hypoxia will be consistent though the micro-dissected series of sections. Six regions were micro-dissected, three from pimonidazole-positive and three from pimonidazole-negative regions. The quantity and quality (RIN) of the RNA extracted from micro-dissected tissue from six regions demonstrating pimonidazole-negative (three samples) and pimonidazole-positive (three samples) staining are shown in Table II. RIN >1.4 is considered as a minimal RNA integrity level (9) RIN values were 7.0 and 7.6 for Pimonidazole-rich and Pimonidazole-low samples respectively. These samples were considered acceptable for transcriptomic analysis using Clariom S.
Identification of necrotic and hypoxic/non-hypoxic regions in adjacent tissue sections. (A) H&E-stained section of xenograft showing region of necrosis and (B) adjacent tumor section subject to immunohistochemistry with pimonidazole antibody showing pimonidazole-positive (hypoxic) and pimonidazole-negative (non-hypoxic) regions. (C) Sections above and below micro-dissected series of sections. RNA extraction from xenografted tumours.
Concentration and quality of RNA extracted from tissue micro-dissected contiguous sections from 3 pimonidazole-positive and 3 pimonidazole-negative regions.
Expression of genes encoding cell surface proteins associated with hypoxia. The expression of known hypoxia associated genes, CA9, SLC2A1 (Glut-1) and VEGFA, shown in Figure 4, were increased in expression in pimonidazole-positive compared with pimonidazole-negative regions.
Expression of genes associated with cell surface proteins of hypoxia. Expression of (A) CA9, (B) SLC2A1 and (C) VEGFA in pimonidazole-positive and pimonidazole-negative regions. *p<0.05, ***p<0.001. Pimo: Pimonidazole.
Hypoxia score (HS) generated using the West 24 gene hypoxia signature. Hypoxia scores (HS) were generated using a 24-gene expression bladder cancer hypoxia signature (3). Figure 5 shows the HS of each (A) individual sample and (A) mean of the pimonidazole-positive (n=3) and pimonidazole-negative (n=3) samples. HS was higher (p=0.0052) in the pimonidazole-positive than in pimonidazole-negative samples.
Hypoxia scores (HS) generated from RNA extracted from pimonidazole-positive and pimonidazole-negative samples. (A) The HS in samples from each pimonidazole-positive and pimonidazole-negative sample. (B) Mean and SD of HS differences (**p<0.0052) between pimonidazole-positive and pimonidazole-negative. Pimo: Pimonidazole.
Discussion
The need for hypoxia biomarkers is exemplified by Spiegelberg et al. who found that the failure of phase III trials can be fully explained by lack of patient classification based on tumour hypoxia status (11). Imaging and transcriptomics have been applied to various cancer types (12) for investigations including diagnosis, prognosis, and prediction of extracellular matrix (13) as they provide complementary and potentially mechanistic information. Small animal experiments either preclinical or back translational are essential to understanding and developing novel approaches to characterising cancer. The findings here show that xenografts are convenient models for combined imaging and genomic analysis.
Combined DCE-MRI and OE-MRI developed at the University of Manchester, for the identification of hypoxia has been validated in clinical (14) as well as pre-clinical (15) work. In this study the higher proportion of hypoxia in the larger tumour compared with the smaller demonstrated using OE-MRI/DCE-MRI agrees with other studies (16-20). Necrosis was also higher in the large tumour and was evident in over half the volume in agreement with previous studies reporting on size association with necrotic fraction (18, 19). Several studies have shown that the presence of the necrotic regions in tumours associated with hypoxia (21, 22). Patients with highly necrotic muscle invasive bladder cancer were shown to benefit from carbogen-nicotinamide (CON) modified therapy (21).
Tumour hypoxia induces angiogenesis through up-regulation of VEGF expression. Other mechanisms that enable cancer cells to thrive in hypoxic conditions include up-regulation of cell surface expression of Glut-1 thus increasing glucose uptake to fuel glycolysis and CA9 to reduce the consequent intracellular acidification. Genes encoding the cell surface proteins CA9, Glut-1 and VEGF3 (CA9, SLC2A1 and VEGF3), associated with hypoxia (23, 24), were increased in expression in pimonidazole-positive regions. Clinical work has shown that both CA9 and Glut-1 co-localise with pimonidazole in patients with MIBC (23). Studies have demonstrated that overexpression of CA9 and Glut-1 protein in squamous cell Head and Neck Cancer and MIBC is associated with poor response to chemotherapy (22-24).
Glut-1, CA9 and VEGF3 protein expression can readily be detected by IHC. A general disadvantage of IHC is that it does not give an overall measure of hypoxia in the tumour. However, gene hypoxic status generated using gene expression signatures (3) demonstrate low levels of intra-tumour heterogeneity (25) suggesting that robust tumour characterisation at the gene expression level can be derived from individual samples. The 24-gene bladder cancer hypoxia signature derived from a candidate panel of hypoxia-responsive genes and refined by prognostication in bladder cancer patient cohorts has also been shown to be predictive of benefit from hypoxia modification (2). This study has demonstrated that the signature is sensitive to hypoxia in in vivo model.
This work does have the obvious limitation in sampling from just two tumours.
However multiple sampling of high and low-pimonidazole regions (n=3 of each) from one tumour demonstrated up-regulation of hypoxia-sensitive genes CA9 and GLUT-1 validating the data set for exploration of other gene-expression hypoxia biomarkers and mechanisms underlying the hypoxia-associated phenotype. Single samples from multiple mice increase the potential for genetic variability between replicates and the requirement for many more mice. Genomic analysis facilitates patient stratification and predicts patent outcome. Combining hypoxia imaging techniques with gene expression signatures has the potential to provide robust biomarkers of tumour hypoxia leading to improved patient treatment selection (26). This work has demonstrated that data from MRI based imaging techniques can be acquired alongside gene expression in xenograft models to assist is the development of biomarkers.
Acknowledgements
The University of Manchester Core Facility for preparation of sections, micro-dissection, and RNA extraction.
Footnotes
Funding
Funding from the Saudi Arabian Bureau of Culture is gratefully acknowledged.
Conflicts of Interest
The Authors declare no conflicts of interest.
Authors’ Contributions
Manuscript preparation: RS, TADS; Edited and approved manuscript; All authors, Designed study TADS, KW, CW; MRI measurements and data analysis: BD, MB; In vivo model development BT and RS; Sample preparation and transcriptomics RS and MR.
- Received April 5, 2024.
- Revision received May 8, 2024.
- Accepted May 20, 2024.
- Copyright © 2024, International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved
This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY-NC-ND) 4.0 international license (https://creativecommons.org/licenses/by-nc-nd/4.0).
References
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.