Simultaneous Extraction of RNA and Metabolites ... - ACS Publications

Subscriber access provided by Kaohsiung Medical University

Article

Simultaneous Extraction of RNA and Metabolites from Single Kidney Tissue Specimens for Combined Transcriptomic and Metabolomic Profiling Patrick Leuthold, Matthias Schwab, Ute Hofmann, Stefan Winter, Steffen Rausch, Michael N. Pollak, Jörg Hennenlotter, Jens Bedke, Elke Schaeffeler, and Mathias Haag J. Proteome Res., Just Accepted Manuscript • DOI: 10.1021/acs.jproteome.8b00199 • Publication Date (Web): 09 Aug 2018 Downloaded from http://pubs.acs.org on August 10, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 35 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

Simultaneous Extraction of RNA and Metabolites from Single Kidney Tissue Specimens for Combined Transcriptomic and Metabolomic Profiling Patrick Leuthold1, Matthias Schwab1,4,5, Ute Hofmann1, Stefan Winter1, Steffen Rausch1,2,Michael N. Pollak3, Jörg Hennenlotter2, Jens Bedke2, Elke Schaeffeler1+ , Mathias Haag1+*

(1) Dr. Margarete Fischer-Bosch-Institute of Clinical Pharmacology, Stuttgart, Germany and University of Tübingen, Tübingen, Germany (2) Department of Urology, University Hospital Tübingen, Tübingen, Germany (3) Jewish General Hospital, Montreal, QC, Canada (4) Department of Clinical Pharmacology, University Hospital Tübingen, Tübingen, Germany (5) Department of Pharmacy and Biochemistry, University of Tübingen, Tübingen, Germany

+Elke Schaeffeler and Mathias Haag contributed equally.

*Corresponding Author: Dr. rer. nat. Mathias Haag Dr. Margarete Fischer-Bosch-Institute of Clinical Pharmacology Auerbachstr. 112 70376 Stuttgart, Germany Phone +49 (0)711 / 8101-5429 Fax +49 (0)711 / 85 92 95 Mail: [email protected]

1 ACS Paragon Plus Environment

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 35

ABSTRACT Tissue analysis represents a powerful tool for the investigation of disease pathophysiology. However, the heterogeneous nature of tissue samples, in particular of neoplastic, may affect the outcome of such analysis and hence obscure interpretation of results. Thus comprehensive isolation and extraction of transcripts and metabolites from an identical tissue specimen would minimize variations and enable the economic use of biopsy material which is usually available in limited amounts. Here we demonstrate a fast and simple protocol for combined transcriptomics and metabolomics analysis in homogenates prepared from one single tissue sample. Metabolites were recovered by protein precipitation from lysates originally prepared for RNA isolation and were analyzed by LC-QTOF-MS after HILIC and RPLC separation, respectively. Strikingly, although ion suppression was observed, over 80% of the 2885 detected metabolic features could be extracted and analyzed with high reproducibility (CV ≤ 20%). Moreover fold changes of different tumor and nontumor kidney tissues were correlated to an established metabolomics protocol and revealed a strong correlation (rp ≥ 0.75). In order to demonstrate the feasibility of the combined analysis of RNA and metabolites, the protocol was applied to kidney tissue of metformin treated mice to investigate drug induced alterations.

Keywords: RNA, metabolites, lipids, metabolomics, transcriptomics, mass spectrometry, combined transcriptomics and metabolomics 2 ACS Paragon Plus Environment

Page 3 of 35 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


INTRODUCTION Merging tissue-derived transcriptomic and metabolic data has great potential to yield insight into localized pathophysiological processes and to reveal new connections between gene expression signatures and associated metabolites. For example, the joint analysis of transcripts and metabolites has enabled to uncover an increase in gene-metabolite coupling in breast cancer and hepatocellular tumors 1. Furthermore, it was shown that metabolite levels can be predicted from transcriptome data. Another study showed that, through integrative transcriptome and metabolome profiling, novel functional roles for genes can be uncovered which determines tissue-specific metabolism 2. By applying systematic metabotyping via H NMR spectroscopy and genome-wide gene expression analysis in white adipose tissue from type 2 diabetic rats, the authors revealed specific genetic polymorphisms that are associated with cellular metabolites such as glucose and 3-hydroxybutyrate. Moreover, also with respect to the therapeutic mechanism of putative anticancer drugs (e.g. metformin), a combined metabolomic and transcriptomic study has recently revealed time-dependent effects on the proliferation of colon cancer cells implicating that energy metabolism may be one of the main targets of the drug 3. In nephrology research, disease phenotyping by metabolomics has been applied increasingly in particular to support kidney precision medicine 4. More specifically, regarding clear cell renal cell carcinoma (ccRCC), the integration of transcriptomic and metabolomic data revealed, that changes in metabolite and RNA expression levels are often asynchronous in ccRCC

5,6

.

Moreover in order to study metabolic reprogramming in kidney cancer, Wettersten et al. showed a combined metabolomics and proteomics approach of grade-dependent RCC samples 7. However these studies are based on transcriptomics/proteomics and metabolomics analyses on separate tissue pieces and a systematic study, e.g. providing RNA and metabolite extraction from a single tissue piece is still lacking. Since heterogeneity in renal cancer is a major concern, potentially resulting in clinically-relevant consequences like treatment decision

8

, novel 3

ACS Paragon Plus Environment


Page 4 of 35

approaches that enable to capture comprehensive molecular information from a single tissue sample are required. The isolation of both RNA and metabolites in human tissue can be challenging as such material is often available in limited amounts. Moreover, as tissue samples, in particular tumorous material, are heterogeneous in nature, tumor purity (i.e. extend of stroma and lymphocyte infiltration) can affect the outcome of genomic 9,10 as well as metabolomics analyses. This aspect is even more critical if integration of data from different omics disciplines is envisioned in a system-wide approach hence raising the demand to obtain “multi-omics data” from a single piece of tissue rather than from replicate sample aliquots. Such a combined analysis offers several advantages including economic use of biological material and minimization of inconsistencies introduced through the analysis of different sample aliquots. As a consequence more reliable functional transcriptome-metabolome relationships may be revealed. Regarding RNA isolation, one of the most applied protocols is the single-step method from Chomczynski et al. using reagents containing guanidinum thiocyanate and phenol-chloroform mixtures

11,12

, which has been successfully commercialized. Adaptions of such commercially

available DNA and RNA purification kits led to the development of innovative protocols for the simultaneous isolation of DNA, RNA, miRNA and proteins from a single tissue specimen

13

.

However, to our knowledge, the simultaneous analysis of metabolites and RNA in lysates obtained from commercially available nucleic acid purification kits has not been considered previously. We therefore aimed to establish a protocol for the combined analysis of transcripts and metabolites. Importantly, the protocol can be used without modifications for RNA purification and thus enables the recovery of high quality transcripts. Metabolites can be recovered from tissue lysates after protein precipitation in a highly reproducible manner. Besides sample reproducibility,

analytical

reproducibility

is

not

compromised

by

constituents

of

the 4

ACS Paragon Plus Environment

Page 5 of 35 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


homogenization buffer and only moderate ion suppression was observed. Furthermore the novel protocol was compared to a recently published nontargeted metabolomics method

14

regarding

its capability to separate tumor and nontumor kidney tissue samples based on unsupervised statistical analysis. As proof of concept the protocol was applied to kidney tissue of metformintreated mice in order to demonstrate the strength of combined transcriptomics/metabolomics analyses to narrow down specific drug-induced pathway alterations.



Page 6 of 35

MATERIALS AND METHODS Materials Ultra LC-MS grade acetonitrile (ACN) and methanol (MeOH) were purchased from Carl Roth GmbH & Co KG (Karlsruhe, Germany). Pure water was generated from a Milli-Q system (Millipore, Billerica, MA, USA). 5mM purine dissolved in acetonitrile and 2.5 mM HP-0921 [Hexakis(1H,1H,3H-tetrafluoropropoxy)phosphazene] dissolved in acetonitrile was purchased from Agilent Technologies (Waldbronn, Germany). Lysing buffer was obtained from the mirVanaTM miRNA Isolation Kit (Ambion, Life Technologies, Darmstadt, Germany). Tissue samples Porcine kidney was obtained as commercially available fresh food product. Human kidney samples were obtained after surgery from the University of Tübingen, informed written consent was provided by each subject and the use of the tissue was approved by the ethics committee of the University of Tübingen, Germany. The histological evaluation of the human tissue sections, was performed at the Department of Pathology, University Hospital Tübingen, Germany. Mice samples were obtained from the McGill University and the use of the tissue was approved by the ethics committee of the McGill University, Canada. Three healthy male mice (Mus musculus, strain C57BL/6) were administrated with 200mg/kg metformin via intraperitoneal injection and the kidney was removed 1.5 h after the administration. Kidney tissue of three control healthy mice was obtained in parallel. Upon resection, all kidney tissue samples were snap-frozen in liquid nitrogen and were stored at -80 °C until sample homogenization. Kidney tissue samples were prepared and analyzed in four different batches for the independent assessment of reproducibility, method comparability and application. Batch A was used for the reproducibility assessment and contained technical replicates of porcine kidney (n=10). Batch B1 was used for method comparability and contained human tumor, diseased and benign kidney 6 ACS Paragon Plus Environment

Page 7 of 35 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


tissue samples from patients who underwent radical nephrectomy suffering from different forms of kidney diseases such as clear cell renal cell carcinoma (ccRCC, tumor area), oncocytoma (OC, tumor area) and urothelial cell carcinoma (UCC, benign area). For each kidney tissue entity, technical replicates were collected (n=3). Batch B2 was also used for method comparability and represents a subset of a sample batch that was prepared and analyzed previously

14

and contained amongst others technical replicates (n=6) of the same donor

specimens (ccRCC, OC and benign UCC kidney tissue) as described for Batch B1. Batch C was used for method application and consisted of kidney tissues of three control and, three metformin treated mice. Kidney tissue homogenization and metabolite extraction The following procedure describes the sample preparation for the batches A, B1 and C. Batch B2 was prepared as described

14

. Frozen kidney samples of approximately 4-50 mg were

transferred to 2ml homogenization tubes containing 1.4 mm ceramic spheres (Lysing Matrix D, MP Biomedicals, Heidelberg, Germany) prefilled with 600 µL of cold (4°C) lysing buffer derived from the mirVanaTM miRNA Isolation Kit. Exact buffer composition is a company secret, however commercialized buffers are adapted from widely used protocols

11,12

that follow the principle of

RNA isolation by using an acidic solution containing guanidinium thiocyanate, sodium acetate, phenol and chloroform. Samples were homogenized in a FastPrep-24TM instrument (MP Biomedicals, Heidelberg, Germany) at 7°C. Homogenization was achieved within three cycles (6.5 m/s for 20 seconds) followed by weight determination of the homogenate. Samples were centrifuged for 10 minutes (min) at 9391 x g and 4°C (Centrifuge type: 5424R, Eppendorf, Germany) and the supernatants were separated into two aliquots of 300 µL in 1.5 mL polypropylene tubes (Eppendorf, Germany). One aliquot was stored at -20°C as backup or for transcriptome analysis. The other aliquot was pre-normalized according to weight (Batch A and B1) or total RNA amount (Batch C) by addition of cold (4°C) lysing buffer to achieve the same 7 ACS Paragon Plus Environment


Page 8 of 35

solvent/tissue ratio for each sample in an analysis-batch. Samples were vortexed for 30 seconds and from each sample an aliquot was transferred to a 1.5 mL polypropylene tube, followed by the addition of acetonitrile (4 times the volume of the aliquot). The samples were vortexed for 30 seconds, incubated for 1h at -20°C for protein precipitation and centrifuged for 10 min at 21130 x g at 4°C. For quality control (QC) sample preparation, aliquots of 100 µL from the supernatant of each sample were pooled in a new 1.5 mL polypropylene tube and vortexed for 30 seconds. QC dilution samples were prepared by further diluting the QC pool by 2, 3, 5 and 11 fold with acetonitrile. A blank sample, consisting of only lysing buffer and another blank sample of which the lysing buffer was substituted with water were also processed with the samples. All samples, including blank and QC samples were transferred into 2 mL glass vials containing 250 µL glass inserts with polymer feets (Agilent Technologies, Waldbronn, Germany) and were covered with pre-slit polytetrafluoroethylene (PTFE)/silicone screw caps (Agilent Technologies, Waldbronn, Germany). LC-MS analysis Samples were analyzed by using a 1290 Infinity UHPLC System coupled to a 6550 iFunnel QTOF-MS from Agilent Technologies equipped with a Dual Agilent Jet Stream electrospray source. In brief, each sample was analyzed separately on hydrophilic interaction liquid chromatography (HILIC) (Acquity UPLC BEH Amide Column, 1.7 µm, 2.1 mm x 150 mm; Waters, Eschborn, Germany) and on reversed phase liquid chromatography (RPLC) (Acquity UPLC BEH, 1.7 µm, 2.1 mm x 100 mm; Waters, Eschborn, Germany) with temperature (auto sampler and column), solvent and gradient settings as previously described

14

. The injection

volume was set to 1.5-8 µL for HILIC and RPLC analysis, depending on the analysis batch or mode. Between injections in RPLC and HILIC analysis, needle wash with 95 % acetonitrile was performed. Mass spectrometric and auto MS/MS analysis settings were similar as previously described

14

, except, quadrupole band-pass for precursor isolation was set to narrow (∼1.3 m/z) 8 ACS Paragon Plus Environment

Page 9 of 35 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


for MS/MS acquisition. Cell, quadrupole, optics and funnel settings were adjusted separately for lipid analysis on RPLC and small-molecule profiling on HILIC analysis according to manufacturer’s instructions. In addition, the mass analyzer was calibrated on a daily basis immediately before starting an analytical run. Moreover, in order to provide mass correction and to assess ion suppression over the analytical gradient, two compounds (purine and HP-0921 [Hexakis(1H,1H,3H-tetrafluoropropoxy)phosphazene])

were

added

post-column

via

the

instrument internal calibrant delivery system. In positive ionization mode analysis the masses m/z 121.0506 (purine) and m/z 922.0097 (HP-0921) and for negative mode analysis the masses m/z 119.0363 (purine) and m/z 966.0007 (HP-0921) were used for dynamic mass correction. Following concentrations in acetonitrile/water (95:5, v/v) were used for the two reference masses: HILIC positive mode: 1.66 µM purine and 8.33 µM HP-0921; HILIC negative mode: 1.66 µM purine and 0.43 µM HP-0921; RPLC positive mode: 1.66 µM purine and 0.83 µM HP0921; RPLC negative mode: 0.56 µM purine and 0.83 µM HP-0921. Metabolite annotation For structural assignments of metabolomic features, mass spectra (MS1 and MS/MS) acquired in QC samples were investigated with the Mass Hunter Qualitative Analysis Software (Version B.06.00, Agilent Technologies). Features were annotated based on accurate mass (± 15 ppm) and/or fragmentation patterns (MS/MS spectrum) that matched to data derived from online databases LIPID MAPS 15, HMDB 16, METLIN 17 and MassBank 18 or literature 19,20. Quantification of Metformin by LC-MS/MS Tissue samples (20 – 30 mg) were homogenized in a total volume of 400 µl of 0.9% NaCl in a FastPrep® 24 homogenizer (MP Biomedicals, Santa Ana, USA) for 20 s at speed 6.0 using lysing matrix D. The homogenate was centrifuged for 10 min at 21130 x g. An aliquot of the supernatant was spiked with internal standard [2H6]metformin and deproteinized with a threefold 9 ACS Paragon Plus Environment


Page 10 of 35

volume of 0.1 % formic acid in acetonitrile. After an additional centrifugation step, metformin was quantified in the supernatant by LC-MS-MS using an Agilent 6460 triple quadrupole mass spectrometer (Agilent, Waldbronn, Germany) coupled to an Agilent 1200 HPLC system. Ionization mode was electrospray (ESI), polarity positive. Electrospray jetstream conditions were as follows: capillary voltage 3500 V, nozzle voltage 1000 V, drying gas flow 11 l/min nitrogen, drying gas temperature 300°C, nebulizer pressure 55 psi, sheath gas temperature 350 °C, sheath gas flow 11 l/min. HPLC separation was achieved on a Synergi Polar-RP 80A column (150×2 mm I.D., 4 µm particle size, Phenomenex, Aschaffenburg, Germany) using (A) 0.1 % formic acid in water and (B) 0.1 % formic acid in acetonitrile as mobile phases at a flow rate of 0.4 ml/min. Gradient runs were programmed as follows: 10 % B from 0 min to 2 min, linear increase to 50 % B to 5 min, then re-equilibration. The mass spectrometer was operated in the multiple reaction monitoring (MRM) mode using m/z 130.1 and 136.1 as precursor ions for metformin and [2H6]metformin, respectively, and the product ion m/z 60.1 for both compounds. Dwell time was 100 ms, fragmentor voltage was set at 60, and the collision energy at 10. Calibration samples were prepared by adding varying amounts of metformin to untreated tissue homogenate. Concentration range was from 5 pmol to 5000 pmol per 25 µl of tissue homogenate. Calibration samples were worked up as described above, and analyzed together with the unknown samples. Calibration curves based on internal standard calibration were obtained by weighted (1/x) linear regression for the peak-area ratio of the analyte to the internal standard against the amount of the analyte. The concentration in unknown samples was obtained from the regression line. Assay accuracy and precision were determined by analyzing quality controls that were prepared like the calibration samples. Isolation of RNA Total RNA from animal and human kidney samples was isolated using the mirVanaTM miRNA Isolation Kit (Ambion/Life Technologies) according to the manufacturer`s protocol. RNA quality 10 ACS Paragon Plus Environment

Page 11 of 35 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


and integrity was determined for all samples using the Agilent 2100 Bioanalyzer (Nano LabChip Kit, Agilent Technologies). RNA Integrity Number (RIN) for all samples was above 7 and the exact values for each sample are listed in Supplemental Table S-1. Analysis of gene expression Transcriptome profiling of kidney tissue samples of three control and three metformin treated mice was performed using the Clariom S Assay (Affymetrix) according to the manufacturer`s standard procedure (Affymetrix). Data processing and statistical analysis Preprocessing of nontargeted metabolomics data was carried out by using Mass Hunter Profinder Software (version B.06.00 Agilent Technologies) using batch recursive and/or targeted feature extraction with an intensity threshold ≥ 500. Settings were similar as described previously

14

. Extracted features were exported as comma separated value files and were used

for further statistical analysis. For reproducibility assessment (batch A), extracted features were normalized using locally weighted scatterplot smoothing (LOESS) over quality control samples (QCs) and the technical replicates

21

. Coefficients of variation were calculated on the normalized

values. For each of the other batches (batch B1/B2 and C), the nontargeted and targeted feature lists were sum normalized (peak area for each feature was divided by the sum of all features per mode) and feature lists from the different analysis modes (HILIC and RPLC in positive and negative ion mode) were merged. Of note, for batch C, the sum normalized values were multiplied with 1x106. Subsequently, QC filtering (features with CV ≥ 30% over the QC samples were removed) and log2 transformation was applied. Only features detected across all samples were considered in all further analyses.



Page 12 of 35

Principal component analysis (PCA) was used to cluster ccRCC, oncocytoma and benign kidney tissue of UCC (batch B1/B2). Pearson correlation coefficients (rp) were computed between log2 fold changes obtained by lysis buffer and standard protocol (batch B1/B2). Quality control and preprocessing of microarrays was performed with Affymetrix Expression Console (Build 1.4.1.46; annotation: Affymetrix ClariomS mouse Transcript Cluster Annotation, Release 36). To be more precise, robust multi-array average

22

was applied to preprocess the

six Clariom S mouse arrays (processed data, Supplemental Table S-6). For analysis of the exploratory proof of concept experiment (batch C) differentially altered features (metabolites and transcripts) between metformin treated mice and control mice were assessed by linear modeling and related empirical Bayes moderated t-tests corrected for multiple testing by the Benjamini-Hochberg procedure

23

. P-values were

24

. For the assignment of

pathways to transcripts and metabolites the pathways glycolysis and gluconeogenesis, triacylglyceride synthesis, TCA cycle and amino acid metabolism (for urea cycle) for mus musculus were selected from WikiPathways

25,26

. Pathways were chosen on the basis of known

modes of metformin action (i.e. lowering blood glucose level by affecting gluconeogenesis

27,28

).

The GenMAPP Pathway Markup Language (gpml) files from WikiPathways were modified and adapted using PathVisio (version 3.2.4) 29,30. R statistics software (version 3.2.3) packages limma

32

, beanplot

33

31

was used for all statistical analyses using the additional

, org.Mm.eg.db

34

, data.table

35

and plyr

36

. Radial plots were

generated in R with the radial.plot function within the plotrix package 37.


Page 13 of 35 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


RESULTS AND DISCUSSION Combined transcriptomic and metabolomic analysis of tissue samples offers great potential to study localized disorders and coupled signaling systems. However, tissue heterogeneity may affect the concordance by superposition of small effects. Thus, the development of simultaneous extraction procedures for the recovery of transcripts and metabolites from a single tissue specimen represents a desired attempt. Although several protocols for the co-extraction of metabolites/RNA and proteins from plants, cells and microorganisms have been described previously,

38–41

these methods were not specifically tailored for human tissue analysis. One of

the most applied protocols for RNA isolation

11

was modified over the last couple of years and

was extended for the simultaneous extractions of RNA, DNA and proteins

13,42

but to our

knowledge not for metabolites. This motivated us to investigate the applicability of using commercial lysing buffer for combined RNA isolation and metabolite extraction from kidney tissue samples. For the protocol described herein tissue samples were homogenized, using lysing buffer that is usually applied for RNA isolation, and subsequently divided for separate RNA purification and metabolomics analyses. RNA quality is not compromised indicated by average RIN values above 7 (Table S-1). For metabolomics analysis, homogenized tissue in the lysis buffer was further diluted with acetonitrile followed by centrifugation and LC-MS analysis of the supernatant. For method validation, ion suppression and assay reproducibility was assessed. Assessment of ion suppression and reproducibility The lysing buffer contains several constituents (reducing agent, salts and detergent) and such components may induce matrix effects. Here we investigated the effect of ion suppression for two compounds with different molecular weights which were constantly infused post-column over the whole gradient. As expected ion suppression was observed in all analytical modes however with varying degrees (Supplemental Figure S-1 A-H). Whereas matrix effects in HILIC mode 13 ACS Paragon Plus Environment


Page 14 of 35

were generally more variable (Supplemental Figure S-1 A-D) signal reduction in RPLC mode was more evenly distributed over the gradient (Supplemental Figure S-1 E-H). Of note, in HILIC (-) mode ion enhancement was observed for the 966.0007 ion (Supplemental Figure S-1 D) in most of the retention time regions indicating a beneficial effect with respect to detection sensitivity for this compound. Next, we investigated if reproducibility of both sample preparation and analytical measurement was compromised on the basis of metabolic features measured in porcine kidney. Altogether, 2885 non-unique features, detected in different analytical modes, were considered (Table 1). Table 1: Reproducibility assessment based on metabolomic features Sample preparation Total Analytical reproducibility: Analysis reproducibility: number number of features with mode number of features with of featuresa CV ≤ 20%b CV≤ 20%c HILIC POS 265 263 (99%) 247 (93%) HILIC NEG 444 440 (99%) 422 (95%) RPLC POS 1087 1026 (94%) 902 (83%) RPLC NEG 1089 1022 (94%) 887 (81%) a

Non-unique features. Features were not assessed for redundancy between modes (i.e. detectable as different adduct within or between analysis modes). b Coefficient of variation (CV) calculated over 8 QC samples analyzed throughout the analytical batch c CV calculated over 10 individually prepared and analyzed kidney tissue samples

Of these, more than 93% and 80% showed excellent analytical and sample preparation reproducibility (CV ≤ 20%), highlighting reliable extraction and analytical performance for the majority of features (Table 1). Median CVs in positive and negative mode analysis for analytical (HILIC: 5.2% / 4.7%; RPLC: 6.8% / 6.9%) and sample preparation reproducibility (HILIC: 7.4% / 6.9%; RPLC: 12.6% / 12.9%) were all below 15% indicating excellent assay precision (Figure 1).


Page 15 of 35 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 1

Similar proportions of reproducible features could also be found in previous metabolomics studies, using conventional extraction and homogenization solvents composed of mixtures of organic solvents and water

43–45

. Of note, whereas analytical reproducibility of both HILIC and

RPLC analysis exhibited comparable CVs (median CVs ≤ 6% and ≤ 7%, respectively) sample preparation reproducibility assessed by RPLC (median CV ≤ 13%) was higher compared to those monitored by HILIC mode (median CV ≤ 8%). To assess which metabolite classes mainly contribute to the observed differences, reproducibility was further examined on the basis of structurally assigned features. To this end, CV values of 177 annotated features, covering metabolite classes of various polarities, were visualized in radial plots (Figure 2). In these plots, CVs for the indicated metabolite species and analytical modes are graphically represented as distances from the center with circles representing different CV ranges. To enable a quick visualization and differentiation between analytical and sample preparation reproducibility radials were colored in blue and green, respectively. Information about CV values used to construct the plots are provided in Supplemental Tables S2-S5



Page 16 of 35

As becomes apparent from the radial plots, the higher imprecision observed in the RPLC mode (Figure 2 C/D and Supplemental Table S-4/S-5) can be attributed to lipids that exhibit larger CVs compared to the more polar fraction of analytes captured by HILIC MS (Figure 2 A/B and Table S-2/S-3). Such a higher imprecision in the RPLC mode is likely indicative of a less reproducible extraction of features with intermediate and low polarity by the rather polar extraction solvent used for RNA isolation. Of note, no particular lipid class seems to disproportionally contribute to higher CVs as indicated by evenly distributed CVs over the individual species. Surprisingly, also very non-polar lipids such as triacylglycerols (TAG) could be extracted and analyzed by RPLC with CVs

Simultaneous Extraction of RNA and Metabolites ... - ACS Publications

Recommend Documents