MSACL 2018 US Abstract

Topic: Small Molecules / Tox

Spatially Resolved Profiling of Polymyxin B Induced Acute Kidney Injury

Andreas Dannhorn (Presenter)
Imperial College London

Authorship: Andreas Dannhorn (1), Gregory Hamm (2), John G. Swales (2), Paolo Inglese (1), James McKenzie (1), Renata Soares (1), Richard J. A. Goodwin (2), Zoltán Takáts (1)
(1) Department of Surgery and Cancer, Sir Alexander Fleming Building, Imperial College, London SW7 2AZ, UK, (2) Drug Safety & Metabolism, AstraZeneca, Cambridge, CB4 0WG, UK

Short Abstract

The global problem of advancing antimicrobial resistance led to a renewed interest in Polymyxin antibiotics. These antibiotics are commonly used as a last resort in cases of uncontrolled infection with gram-negative bacteria. Polymyxin efficacy is linked to known neuro- and nephrotoxicity in the clinic. We applied multimodal Mass Spectrometry Imaging (MSI) to investigate Polymyxin induced acute kidney injury in a rat model. We were able to correlate drug accumulation in the organ with dose dependent effects it caused to the tissue metabolome. MSI analysis revealed drug induced phospholipidoses after receiving a low dose of PMB, whist a high dose induced acute kidney injury.

Long Abstract


The global problem of advancing antimicrobial resistance led to a renewed interest in Polymyxin antibiotics. These antibiotics are commonly used as a last resort in cases of uncontrolled infection caused by gram-negative bacteria. Polymyxin efficacy is linked to known neuro- and nephrotoxicity in the clinic. Polymyxins accumulate in the proximal tubule epithelium and their detergent like behaviour leads to necrosis of the cells causing subsequent acute kidney injury with impaired kidney functionality. The sample set for this study consisted of kidney samples receiving a low dose of Polymyxin B1 (PMB1) of 8 mg/kg/day or a high dose of 25 mg/kg/day. The study was designed with 2 control groups; one vehicle control group and a group receiving a dosing of 22 mg/kg/day with Polymyxin B nonapeptide (PMBN). PMBN has the polar head group of Polymyxin B but does not contain the lipophilic chain. Without the lipophilic moiety Polymyxin B becomes bacteriostatic, losing any bactericidal effect. Loss of antibiotic efficacy is linked with a corresponding loss of neuro-and nephrotoxicity, outlining the link between efficacy of the drug class and its toxic effects observed in the clinic. For all dosing groups samples were collected on day 3 to study the immediate effects of the administered drugs or on day 24 to allow recovery.

A previously reported study on a similar sample set demonstrated successful mapping of the renal distribution of PMB using MALDI-MSI [1]. The proposed methodology was successfully adapted to map the distribution of PMB and PMBN in the current sample set. The localization of the drugs was subsequently correlated with the spatial changes in the tissue metabolome. A spatially resolved untargeted approach using DESI-MSI was used to identify dose dependent changes in the tissue metabolome. Compounds showing significant change in abundance upon dosing were identified and a targeted DESI-MSI approach was used to specifically acquire these metabolites with increased sensitivity and specificity.


Kidney samples were cut into sections of 10 µm thickness and thaw-mounted on either Superfrost® slides for DESI-MSI or ITO-coated slides for MALDI-MSI experiments. Tissue sections for the different time points and dosings as well as control tissue sections were randomly mounted adjacent on the same slide. MSI data for all tissue sections were acquired in the same experiment to reduce inter-experiment variability.

To determine the spatial distribution of the drugs in the tissue sections MALDI-MSI experiments were performed on a rapifleX Tissuetyper equipped with a smartbeam 3D laser operated at a repetition rate of 10 kHz. As previously reported, 2,5-Dihydroxybenzoic acid (DHB) was used as MALDI matrix to ionize and detect the drugs in positive ion mode. The matrix was applied using an automated spray system (TM-sprayer). To determine the drug distribution across full tissue sections the spatial resolution was set to 50 µm. High spatial resolution images with 10 µm resolution were acquired to identify tissue fine structures across the organ showing drug accumulation.

The untargeted metabolomics approach was performed on a Q-Exactive mass spectrometer. The instrument was equipped with a 2D DESI stage and a home-build DESI sprayer. The data was acquired in positive and negative ion mode for the mass-range between m/z 100m/z 1000.The instrument was operated with a mass resolution of 70000 (peak-width at half height at m/z 200). The spectral information for regions of interest were extracted for all tissues within the kidney (cortex, outer medulla, inner medulla, and calyx) and subjected to univariate statistical analysis to identify all features with statistically different abundance compared to the controls. These features were subsequently identified by accurate mass and their identity confirmed by DESI-MS/MS analysis performed on the Q-Exactive operated under the same conditions as used for the untargeted approach.

Identified features were mapped using a targeted MSI approach on a TQ mass spectrometer operated in MRM mode to increase sensitivity and specificity for the target compounds. A TQ-S equipped with a 2D stage was used for the experiments. Samples were re-analysed using this targeted approach. The spectral information for regions of interest within the different tissue types were extracted and analysed using multivariate statistics to identify the features with the biggest changes in the different tissues.


The MALDI-MSI experiments show high abundance for the Polymyxin B1 in the renal cortex, specifically co-localizing with in the proximal tubule. Overall, the abundance for Polymyxin B1 is approx. 4-fold the abundance in the outer medulla and approx. 6-fold compared to the inner medulla.

Univariate analysis performed on the untargeted DESI-MSI data reported a total of 182 significant features across all tissues and dosing in positive ion mode and 183 in negative ion mode. The biggest differences compared to the controls are within the low and high PMB1 dosing groups with substantial overlap in regulated features, but also unique features for both dose groups. PMBN dosing resulted in a lower number of unique differences.

The regulated features that could be identified can be divided into 4 classes of molecules: glyerophospholipids, lipid precursor and small molecules associated with lipid metabolism, tissue protective factors and free fatty acids.

Phospholipid precursors and molecules associated with lipid metabolism, namely choline, glycerophosphocholine, betaine, carnitine and acetylcarnitine were depleted upon dosing. Most glycerophospholidid-classes (e.g. phosphatidylethanolamines (PEs), phosphatidylcholines (PCs), phosphatidylinositols (PIs)) showed increased abundances upon dosing, indicating phospholipidosis (PLD). This assumption is supported by the increase of bismonoacylglycerophosphates (BMP) as identified biomarker for PLD [2]. Increase in free fatty acids, especially poly-unsaturated fatty acids FA20:4 and 22:6 indicate triggering of inflammation via the eicosanoid pathway resulting in depletion of tissue protective factors such as ascorbic acid, taurine and pantothenic acid.

Subsequently, the targeted MSI approach was applied to map the spatial distribution of these molecules within all samples to identify the spatially resolved changes in the tissue metabolome. All dosing groups showed increased phospholipid abundances. The abundances for most phospholipids were higher in the PMB1 low dose group compared to the high dose group. The relative abundances for BMP species, as PLD liability marker, were significantly increased. The high dose group showed high abundances for pro-inflammatory molecules associated with the eicosanoid pathway and depletion of the tissue protective molecule. Both changes were less pronounced in the low and PMBN dosed groups.

The increased abundances for PLD marker and phospholipids in the PMB1 low dose group suggest strongly drug induced PLD, whilst the upregulation in pro-inflammatory pathways in combination with increased phospholipid abundances suggests drug induced acute kidney injury. These assumptions were confirmed by histopathological evaluation of H&E stained tissue sections. Sections from the PMB1 low dose group showed mainly vacuolization as main impairment of the kidney. The high dosed samples showed single cell apoptosis of the proximal tubule epithelium and mononuclear cell infiltration, especially in the renal cortex around the proximal tubule and the glomeruli.

Conclusions & Discussion

In this study, we successfully applied spatially resolved analysis to investigate Polymyxin induced kidney injury. The combination of the different MSI techniques and platforms used enabled us to correlate the localization of the drug with the metabolomic changes in the tissue metabolome upon dosing. The changes in the tissue metabolome show a dose dependent response towards PMB1. Whilst a low dose induces PLD resulting in kidney impairment, a high dose leads drug induced acute kidney injury.

The advantage of spatially resolved analysis performed over classical analysis performed on tissue homogenates, is the possibility to track changes in the tissue metabolome for all distinct tissue types of the organ. Analysis of tissue homogenates loses all spatial information and represents an average of all tissues within the organ, the resulting data does not necessarily represent the actual changes and damage within the tissue. Furthermore, as MSI data can be visualized as images the correlation with optical methods such as H&E stained tissue sections can be easily made to complete the data set and provide greater understanding of the underlying etiology.

References & Acknowledgements:

1. Nilsson, A., et al., Investigating nephrotoxicity of polymyxin derivatives by mapping renal distribution using mass spectrometry imaging. Chem Res Toxicol, 2015. 28(9): p. 1823-30.

2. Baronas, E.T., et al., Biomarkers to monitor drug-induced phospholipidosis. Toxicol Appl Pharmacol, 2007. 218(1): p. 72-8.

Financial Disclosure

SalaryyesImperial College London
Board Memberno

IP Royalty: no

Planning to mention or discuss specific products or technology of the company(ies) listed above: