Integrated mapping of pharmacokinetics and pharmacodynamics in a patient-derived xenograft model of glioblastoma

Abstract

Therapeutic options for the treatment of glioblastoma remain inadequate despite concerted research efforts in drug development. Therapeutic failure can result from poor permeability of the blood-brain barrier, heterogeneous drug distribution, and development of resistance. Elucidation of relationships among such parameters could enable the development of predictive models of drug response in patients and inform drug development. Complementary analyses were applied to a glioblastoma patient-derived xenograft model in order to quantitatively map distribution and resulting cellular response to the EGFR inhibitor erlotinib. Mass spectrometry images of erlotinib were registered to histology and magnetic resonance images in order to correlate drug distribution with tumor characteristics. Phosphoproteomics and immunohistochemistry were used to assess protein signaling in response to drug, and integrated with transcriptional response using mRNA sequencing. This comprehensive dataset provides simultaneous insight into pharmacokinetics and pharmacodynamics and indicates that erlotinib delivery to intracranial tumors is insufficient to inhibit EGFR tyrosine kinase signaling.

Fig. 1

Schematic of multimodal analysis workflow for investigation into drug delivery and response. Two groups of mice were injected with patient-derived GBM12 tumor cells to establish xenografts either in the flank or brain. Mice in both groups were treated with either placebo, ultra-low dose (5 mg kg⁻¹), low dose (33 mg kg⁻¹), or high dose (100 mg kg⁻¹) erlotinib. Tissue from mice in each group was analyzed according to the workflow shown in the schematic

Fig. 2

Multimodal 3D imaging with MALDI MSI, MRI and H&E, letters ‘L’ and ‘R’ denote left and right sides of the mouse brain. a contrast-enhanced T1-weighted MR image of mouse with established intracranial tumor. b MALDI MS imaging of serial coronal sections of mouse brain (subset from tumor core only displayed, see Supplementary Figure 2 for full set of images) with established GBM tumor after treatment with 100 mg kg⁻¹ erlotinib (2 h post-dose) with corresponding H&E staining of adjacent sections (ion images of erlotinib (green, m/z 394.176 ± 0.001) and heme (red, m/z 616.177 ± 0.001) overlaid, c 3D reconstructions of MALDI ion images collected at 160 μm intervals and imaged at 100 μm spatial resolution, Erlotinib (green, m/z 394.176 ± 0.001), heme (red, m/z 616.177 ± 0.001), the active metabolite of erlotinib (M13/M14) (cyan, m/z 380.160 ± 0.001), and a tumor biomarker (white, m/z 503.949 ± 0.001), d automatic multimodal non-rigid alignment of erlotinib ion image with T2W and T1Gd MR images of one example section/plane, and regions of interest (ROI) mask image for normal brain (1), tumor associated drug exclusive of T1Gd hyperintensity (2), and T1Gd hyper intensity (3) were generated from the registered T1Gd MR and erlotinib ion images, and e boxplot of erlotinib intensity detected in each ROI, where the center line represents the median, the upper and lower bounds of the box represent the inter-quartile range and the whiskers represent ±1 standard deviation of the mean

Fig. 3

Correlated MALDI MS imaging of the drug (erlotinib) distribution and SRS imaging of the brain tissue morphology. a MALDI MS ion imaging shows the distribution of Erlotinib across the entire brain tissue section. b SRS imaging shows the chemical constituents (lipid in green; protein in blue; and heme in red) of the entire serial brain tissue section. c H&E image of another serial brain tissue section corresponding to a and b. d Enlarged overlaid view of areas marked by the large rectangle in a and b shows distribution of the drug (erlotinib in pink) and tissue morphology (lipid and protein in green and blue, respectively) of the tumor area, e H&E image corresponding to d. f Enlarged overlaid images of the tumor margin area (marked by rectangle in d) shows cancer cell invasion into the normal fatty brain structure. g H&E image corresponding to f. h Enlarged overlaid images of the brain structure (marked by smaller rectangle on a and b featuring the difference between gray (protein rich—blue) and white (lipid rich—green) matter. High concentration of erlotinib was found in the third ventricle h, with corresponding H&E image shown in i. Scalebars represent lengths as follows: a–c = 1 mm, d and e = 500 µm, f–i = 200 µm

Fig. 4

Tissue Immunofluorescence imaging of total EGFR and phosphorylated-EGFR (p-EGFR) in brain and flank tumors. a Images showing total EGFR in brain and flank tumors (red) overlaid with Hoechst (blue), treated with placebo, low (33 mg kg⁻¹) and high (100 mg kg⁻¹) dose erlotinib, and graphs (on right) showing quantitative analysis EGFR levels over whole tumor region in each treatment group. b Images showing p-EGFR in brain and flank tumors (red) overlaid with Hoechst (blue), with placebo, low (33 mg kg⁻¹) and high (100 mg kg⁻¹) dose erlotinib and graphs (on right) showing average (mean) intensity of p- EGFR levels in each treatment group. (scalebars on images represent 250 µm, error bars represent 1 standard deviation, n = 2 for brain, n = 3 for flank)

Fig. 5

In vivo phosphorylation changes in erlotinib-treated flank tumors reveal inhibition of EGFR signaling. a Hierarchical clustering (Euclidian distance) of phosphopeptide changes for GBM12 flank tumors treated with 5, 33, and 100 mg kg⁻¹ erlotinib (n = 4) and displayed relative to a vehicle-treated flank tumor measured in both of the analyzed TMT-10-plex MS runs (E1 and E2), b Phosphoprotein interaction network based on STRING for members of cluster 1 in a and represented by their gene name

Fig. 6

mRNA sequencing of flank and intracranial (IC) tumor tissues from animals dosed with erlotinib showing dose-dependent pathway regulation. a Summary of pathways regulated by different doses of erlotinib in flank tumors (n = 5), z-Scores of significant pathways relative to placebo are presented. b MALDI MS images of erlotinib in brain tissue with IC tumor and regions of interest (R1, R2, L1 and L2) used for laser capture microdissection (LCM) and subsequent mRNA sequencing, adjacent section with H&E staining and optical image of adjacent section used for LCM. c Average (mean) relative intensity of erlotinib (m/z 394.176) detected in each ROI of MALDI MS ion image. d Clustermap of transcripts from flank and IC tumors by mRNA sequencing

Fig. 7

Integrative analysis of phosphoproteomic and mRNASeq data. a Hierarchical clustering of correlation matrix for phosphoproteomic data reveals that most sites are correlated. Sub-clusters highlight known interactions as potentially novel signaling network. b Clustering of mRNA correlation matrix highlights the bi-modal response to Erlotinib, with most transcripts either decreasing or increasing on Erlotinib treatment and thus either positively or negatively correlated with other transcripts. c Clustering of phosphoproteomic and mRNASeq correlation matrix. Phosphorylation sites tend to decrease on Erlotinib treatement and are therefore positively correlated with approximately half of the transcripts and negatively correlated with the other half of the transcripts. Sub-clusters of phosphorylation sites indicate well-established signaling networks. d Line graphs showing analysis of correlation between phosphorylation sites and their transcripts indicating that activation/inhibition of the kinase may be regulating the site. Error bars represent the standard deviation