Discovery and Qualification of Candidate Urinary Biomarkers of

Subscriber access provided by YORK UNIV

Article

Discovery and qualification of candidate urinary biomarkers of disease activity in lupus nephritis Veronica G. Anania, Kebing Yu, Francesco Pingitore, Qingling Li, Christopher M. Rose, Peter Liu, Wendy Sandoval, Ann E Herman, Jennie R. Lill, and W. Rodney Mathews J. Proteome Res., Just Accepted Manuscript • DOI: 10.1021/acs.jproteome.8b00874 • Publication Date (Web): 10 Dec 2018 Downloaded from http://pubs.acs.org on December 11, 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 38 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

Discovery and qualification of candidate urinary biomarkers of disease activity in lupus nephritis

Veronica G. Anania1*, Kebing Yu2, Francesco Pingitore1, Qingling Li1, Christopher M. Rose2, Peter Liu2, Wendy Sandoval2, Ann E. Herman1, Jennie R. Lill2, and W. Rodney Mathews1

1

Department of OMNI Biomarker Development, 2Department of Microchemistry, Proteomics,

and Lipidomics, Genentech, Inc., South San Francisco, CA.

*Correspondence to be addressed to [email protected]

Running Title: LN Urinary Proteome Profiling

Abbreviations: LN – Lupus nephritis, HC – healthy control, SLE - systemic lupus erythematosus, DDA - data-dependent acquisition, DIA - data-independent acquisition, MRM - multiple reaction monitoring, UPCR - urinary protein to creatinine ratio, ANA - antinuclear antibody, TMT – tandem mass tags, SDS-PAGE – sodium dodecyl sulfate polyacrylamide gel electrophoresis, FA – formic acid, DTT- dithiothreitol, IAA – iodoacetamide, PSM – peptide spectral match, GBM – glomerular basement membrane, FDR – false discovery rate, CV – coefficient of variance, TFRE – transferrin, A2MG – alpha-2-macroglobulin, A1AT – alpha-1-anti-trypsin, CERU – ceruloplamsin, HPT – haptoglobin, AFAM – afamin, ICAM-1 – intercellular adhesion molecule 1, VIME – vimentin, and FASP – filter assisted sample preparation.

ACS Paragon Plus Environment

1

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 38

Summary Lupus nephritis (LN) is a severe clinical manifestation of systemic lupus erythematosus (SLE) associated with significant morbidity and mortality. Assessment of severity and activity of renal involvement in SLE requires a kidney biopsy, an invasive procedure with limited prognostic value. Non-invasive biomarkers are needed to inform treatment decisions and to monitor disease activity. Proteinuria is associated with disease progression in LN, however, the composition of the LN urinary proteome remains incompletely characterized. To address this, we profiled LN urine samples using complementary mass spectrometry-based methods: SDSPAGE gel fractionation, chemical labeling using tandem mass tags (TMT), and data-independent acquisition (DIA). Combining results from these approaches yielded quantitative information on 2,573 unique proteins in urine from LN patients. A multiple reaction monitoring (MRM) method was established to confirm eight proteins in an independent cohort of LN patients, and seven proteins (transferrin, alpha-2-macroglobulin, haptoglobin, afamin, alpha-1-antitrypsin, vimentin, and ceruloplasmin) were confirmed to be elevated in LN urine compared to HC. In this study, we demonstrate that deep mass spectrometry profiling of a small number of patient samples can identify high quality biomarkers that replicate in an independent LN disease cohort. These biomarkers are being used to inform clinical biomarker strategies to support longitudinal and interventional studies focused on evaluating disease progression and treatment efficacy of novel LN therapeutics.


2

Page 3 of 38 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 Systemic lupus erythematosus (SLE) is a heterogeneous chronic disease with significant morbidity and mortality. Lupus nephritis (LN) is a severe manifestation of SLE, affecting up to 50% of adults with SLE and 80% of pediatric SLE patients 1. In LN, accumulation of immune complexes in the glomeruli leads to renal damage 2-4. These deposits can accumulate in different regions of the glomeruli including the subepithelial, subendothelial, and mesangial regions 2, 4. This results in lymphocyte and immune cell recruitment causing prolonged inflammation and damage to the glomerular basement membrane (GBM) 2, 5-9. The GBM separates the vasculature from the urinary space and deterioration of this barrier results in protein leakage from blood into urine 2. Thus, a common characteristic in LN is high levels of protein in urine (proteinuria) 10-12. Currently the gold standard method to assess disease activity in LN is percutaneous kidney biopsy, which is invasive and has limited prognostic value 13. In the context of clinical trials, kidney biopsies are difficult to obtain and therefore clinicians rely on less sensitive metrics like urinary protein to creatinine ratio (UPCR) and serum complement levels to assess disease activity. Recent retrospective studies have shown that baseline UPCR and reduction in UPCR during induction therapy are significantly associated with renal outcome 14. Although UPCR has modest utility as a pharmacodynamic biomarker of disease activity, it has several limitations that confound its ability to inform real-time clinical decisions. For instance, some patients require years to repair their GBM and resolve proteinuria despite having low immunological disease activity 11-12. These patients are indistinguishable from patients experiencing high levels of acute disease activity by UPCR, and sole reliance on UPCR could


3


Page 4 of 38

result in use of toxic immunosuppressant drugs in patients with limited potential for benefit. UPCR also lacks the sensitivity required to enable early intervention prior to significant damage to the GBM and delayed therapeutic intervention increases risk of terminal renal failure 15. Therefore, more sensitive biomarkers reflective of immunological disease activity are needed to longitudinally and noninvasively differentiate between true disease activity and lingering, unresolved tissue damage. Given the ease of urine sample collection and the proximity of urine to the site of disease activity, there have been significant efforts to identify urinary biomarkers that have diagnostic, prognostic, or pharmacodynamic utility in LN. Urinary proteins including cytokines, autoantibodies, adhesion molecules and serum proteins have been identified as potential disease activity biomarkers in cross-sectional studies 16-26, however, few candidate biomarkers have been validated in independent cohorts and none have been successfully employed in the context of a clinical trial 24. Advances in both mass spectrometry instrumentation and analytical tools have enabled increased detection of low abundance analytes within matrices that have a high dynamic range of protein concentrations such as proteinuric urine from LN patients 27. In order to deeply profile the LN urinary proteome, we applied three complementary proteomic profiling approaches to identify biomarkers associated with LN disease activity. To confirm a subset of these findings, a targeted multiple reaction monitoring (MRM) method was used to assess levels of eight proteins in an independent cohort of LN patients and healthy controls (HC). The findings presented herein provide peptide-level results on more proteins in the LN urinary


4

Page 5 of 38 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


proteome than any study to date, and findings from these discovery analyses will be used to explore new biomarker hypotheses to inform clinical development of novel LN therapeutics. Experimental Procedures Patients and Sample Collection All LN urine samples were baseline samples collected as part of a clinical study in patients with active LN 28. Briefly, first morning void urine samples were collected using the clean capture approach and samples were centrifuged to remove cellular debris within 4 hr of collection. HC urine samples were collected in a similar manner. Urine supernatants were stored at -80°C prior to analysis. Eligible LN patients were 16–75 years of age and had a diagnosis of SLE according to the revised American College of Rheumatology criteria 29. They were required to have a history of antinuclear antibody (ANA) positivity, class III or class IV (± class V) LN according to the 2003 International Society of Nephrology/Renal Pathology Society (ISN/RPS) criteria supported by renal biopsy (within 12 months), and proteinuria (urine protein to creatinine (UPCR) ratio >1.0). Complement C3, Complement C4, anti-dsDNA antibody titers, glomerular filtration rate (GFR) and UPCR were measured as part of the LUNAR clinical study 28. LUNAR was a phase III clinical trail that assessed the safety and efficacy of Rituximab, an antiCD20 biologic, in a randomized, double-blind placebo-controlled study in patients with LN treated concomitantly with mycophenolate mofetil and corticosteroids. The study protocol was approved by institutional review boards and ethics committees, and the participants provided written informed consent 28. Mass Spectrometry Raw Data


5


Page 6 of 38

All mass spectrometry raw datasets have been deposited to the MASSIVE database (http://massive.ucsd.edu/) and can be downloaded by the identifier MSV000081953. Skyline data files for the targeted MRM method development (peptide stability assessment and trypsin digestion efficiency assessment) as well as the results from the LN and HC MRM analysis are publicly available on Panorama (https://panoramaweb.org/project/gRED%20-%20OMNIBD/LN_urinary_biomarkers_2018/begin.view?). SDS-PAGE Fractionation Analysis Protein concentration of urine samples was assessed using a Bradford Assay according to the manufacturer’s instructions (Sigma). Protein was precipitated using chloroform and methanol from each urine sample and 20 µg of protein was separated by SDS-PAGE on a NuPAGE 10% Bis-Tris protein gel (Thermo) and stained with SimplyBlue SafeStain (Thermo). Proteins were digested in-gel and peptides were eluted, dried down, and reconstituted (detailed information available in the supplemental information). Samples were injected and analyzed on-line via nanospray ionization into a hybrid Linear Trap Quadropole (LTQ)-Orbitrap mass spectrometer (Thermo Fisher Scientific). Data was collected using data-dependent acquisition (DDA) with the parent ion analyzed in the Orbitrap and the top 8 most abundant ions selected for fragmentation and analysis in the LTQ over a 40 min gradient. Peptide identification: ReAdW (v.4.3.1) was used to convert raw data into peaks and MS/MS spectra were searched using Mascot (v.2.3.02) against a concatenated target-decoy database comprised of human protein sequences (UniProt Dec. 2016) and known contaminants. MS/MS spectra were searched by Mascot selecting the enzyme trypsin with up to two missed cleavages. Detailed information on search parameters and allowed peptide


6

Page 7 of 38 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


modifications can be found in the supplemental information. Final data were filtered with peptide false-discovery-rate (FDR) and protein FDR of 1% each 30. DDA with TMT chemical labeling Sample Preparation: 3 HC and 3 LN urine samples were individually digested with trypsin. Peptides were chemically labeled and combined into one sample, fractionated, and analyzed through the Thermo Fisher Scientific Center for Multiplexed Proteomics at Harvard Medical School and prepared as previously described 31 (see supplemental methods for more information). LC-MS/MS: Fractions were analyzed with an LC-MS3 data collection strategy

32

on an

Orbitrap Fusion mass spectrometer (Thermo Fisher Scientific). The top 10 parent ions were selected for ion trap tandem mass spectrometry (ITMS2) analysis and the top ten fragment ion precursors from each MS2 scan were selected for MS3 analysis (synchronous precursor selection). More details on the instrument parameters can be found in the supplemental information. Peptide identification and quantification: ReAdW (v.4.3.1) was used to generate peaks and MS/MS spectra were searched using Mascot (v.2.3.02) against a concatenated target-decoy database comprised of human protein sequences (SwissProt Nov. 2014) and known contaminants. MS/MS spectra were searched by Mascot using trypsin with up to two missed cleavages. Detailed peptide search parameters, allowed peptide modificaitons, and information regarding TMT reporter ion quantification can be found in the supplemental information. Data Independent Acquisition (DIA) Analysis


7


Page 8 of 38

3 HC and 3 LN human urine samples were shipped to Biognosys frozen on dry ice for DIA sample analysis and spectral library building. All solvents were HPLC-grade from Sigma-Aldrich and all chemicals where not stated otherwise were obtained from Sigma-Aldrich Sample Preparation: For DIA analysis, 3 HC and 3 LN samples were processed individually. 10 µl of LN urine or 100 µl of HC urine was denatured with Biognosys’ Denature Buffer and spiked with 2 µg bovine serum albumin (Pierce) per sample as carrier protein. Samples were digested as outlined above and in the supplemental information. No depletion or fractionation was used for DIA sample analysis. LC-MS/MS: Peptides were injected onto a Thermo Scientific QExactive mass spectrometer equipped with a standard nano-electrospray source and analyzed over a 130 min gradient. A DIA method with one full range survey scan and 12 DIA variable width windows was used. More information on the instrument parameters can be found in the supplemental information. Peptide identification and quantification: DIA mass spectrometric data were analyzed using the spectral library generated in this study (see supplemental information) and the Spectronaut 8.0 software (Biognosys). The FDR at the peptide level was set to 1%, and data was filtered using row-based extraction. Peptide level results were summed together, log2 transformed, and normalized by equalizing the median protein intensity across all samples to determine protein level results. MRM Analysis Isotopically labeled and endogenous AQUA peptides with >95% purity were synthesized by Cell Signaling Technology (Protein-AQUA) (Supplemental Table 1).


8

Page 9 of 38 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


Peptide Selection: Peptides were tested for autosampler stability and trypsin digestion efficiency. See supplemental information for details on these experiments and supplemental figures 6 and 7 for results of these analyses. Sample Preparation: 20 HC and 20 LN urine samples were utilized for this experiment. Urinary protein concentrations were determined using a Bradford assay according to the manufacturer’s instructions (Sigma) and 50 µg of protein processed through single filter units (Sartorius Vivacon 500, 10,000 MWCO HY). Proteins were reduced, alkylated, and digested into peptides on filter using trypsin (Promega) for 18 h at 37°C. Peptides were eluted from the columns, dried down using a SpeedVac, and reconstituted in LC solvent. More detailed information can be found in the supplemental information. Peptide amounts were quantified with the Peptide Quant Kit (Thermo Scientific) and adjusted to a concentration of 0.15 µg/µL. Samples were then spiked with Biognosys’ HRM kit calibration peptides and AQUA peptide standards prior to mass spectrometric analyses. Isotopically labeled AQUA peptides spiked into samples were diluted in water and pooled before use to concentrations that ranged from 2–100 fmol/µL depending on the analyte to match the level of endogenous peptide observed in a pooled LN urine sample. In addition, a pool of heavy (stable-isotope labeled) and light (unlabeled) AQUA peptides were prepared and diluted serially to generate calibration curves. The heavy peptide concentration was held constant while serially diluting the light peptide in LC buffer to generate the calibration curve. Concentrations of light and heavy peptides used for the standard curve can be found in Supplemental Table 2. Samples were prepared and analyzed in 5 batches with equal numbers of healthy and LN urine samples in each batch. The Skyline


9


Page 10 of 38

files shared on Panorama contain a representative calibration curve that was generated and used in one batch analysis. LC-MS/MS: Digested peptide was injected and analyzed on a Sciex 6500 triple quadrupole equipped with an IonDrive Turbo V Source over a 60 min gradient. The triple quadrupole was operated in scheduled MRM mode with the 3-5 most abundant transitions were selected for each peptide. See Supplemental Table 3 for peptides and transitions monitored in this method. Detailed instrument parameters can be found in the supplemental information. Peptide identification and quantification: Data files were imported into Skyline-daily v3.6.1 for analysis. Peak integration was done automatically by the software using Savitzky– Golay smoothing, and all the data were visually inspected in order to confirm correct peak assignment and integration. Peptide levels were quantified using the calibration curves generated the same day as samples were analyzed. LLOQ was determined using the calibration curve. The LLOQ was defined as the lowest calibration curve data point with CV +/- 20%. All data points below LLOQ are reported as 0. Results An overview of mass spectrometry approaches used in this study are outlined in Figure 1A. Briefly, urine samples from three LN patients and three gender matched HC samples were profiled using gel fractionation, TMT chemical labeling, and label-free DIA. A highly quantitative, multiplexed MRM assay was used to measure abundance of eight selected candidate biomarkers in an independent cohort of LN patient and HC urine samples (Figure 1B). A


10

Page 11 of 38 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


summary of patient characteristics from the discovery proteomics cohort and from the targeted MRM analysis cohort are summarized in Table 1 and Table 2, respectively. Profiling the LN proteome using SDS-PAGE fractionation with LC-MS/MS For a semi-quantitative comparison of HC and LN urinary proteomes, gel-based fractionation was utilized. Approaches such as this have utility in proteomic profiling especially when the dynamic range of protein concentrations is extremely large like in the case of LN urine. Gel fractionation of high complexity protein matrices can sequester high abundance proteins into individual bands thereby enabling higher sensitivity within other molecular weight regions of the gel. Furthermore, gel-based fractionation approaches are commonly used for urinary proteomic profiling and for biomarker discovery applications 33-38. In this study three urine samples from patients with LN were compared to three gender matched HC urine samples by gel electrophoresis (Figure 2A). In total, 4,396 unique peptides from 591 proteins were identified in HC samples compared to 2,097 unique peptides from 354 proteins identified in LN samples (Figure 2B). These results were filtered to exclude proteins with fewer than 10 total PSM detected across the six urine samples, leaving a total of 396 proteins identified in this analysis. A total of 57 proteins had more PSM identified in the LN samples than in HC samples (Figure 2C, Supplemental Table 4). Quantitative approaches to profile the LN urinary proteome In order to probe deeper into the LN urinary proteome and assess differences in urine protein abundance between LN patients and HC, two different quantitative proteomic approaches were employed: TMT chemical labeling and label-free DIA. Importantly, the same three LN and three HC samples were used across all proteomic profiling experiments.


11


Page 12 of 38

Leveraging orthogonal discovery proteomic approaches enabled a more thorough investigation into the LN proteome which increased the potential to discover novel biomarkers of disease activity in LN. Profiling the LN proteome using TMT chemical labeling For TMT chemical labeling, protein was precipitated from 3 LN and 3 HC urine samples and equal amounts of protein from each sample were digested and chemically labeled with 6plex tandem mass tags. Samples were combined together and fractionated by high pH reverse phase chromatography. 12 fractions were analyzed by LC-MS/MS operated in a Top 10 DDA mode, and results were queried using Mascot followed by TMT reporter ion quantification using Mojave 39. TMT reporter ion intensities were summed and log2 transformed for each protein to assess protein level changes. The protein level intensity distributions were median normalized to adjust for slight differences in protein input across the six samples (Figure 3A). Unsupervised hierarchical clustering of protein intensities showed a distinct expression pattern unique to disease status (Figure 3B). Across biological replicates, there was strong agreement for both LN and HC groups. Specifically, 99% of proteins detected in the HC samples had CV

Discovery and Qualification of Candidate Urinary Biomarkers of

Recommend Documents