Final Report Summary - PROTOCLIN (IMPROVEMENT OF TOOLS AND PORTABILITY OF MASS SPECTROMETRY-BASED CLINICAL PROTEOMICS AS APPLIED TO CHRONIC KIDNEY DISEASE)
The main objective of the PROTOCLIN project was the development of a mass-spectrometry (MS) based system to assess chronic kidney disease (CKD) and predict the progression of developmental renal disease which is the first cause of CKD in children. Central to the project was the advancement of the CE-MS (capillary electrophoresis-mass spectrometry) technology in both biomarker discovery and as a tool for diagnostic applications in the context of CKD, This was addressed via two main specific aims: a) establishing the appropriate tools for CE-MS portability and b) evaluating CE-MS combination with MALDI TOF MS, in the form of an integrated MS diagnostic platform. The project was built on previous results that indicated that peptides in urine detected by CE-MS can inform about CKD with high accuracy.
To reach these objectives the following steps were followed:
To enable inter-platform and –laboratory comparative analysis, two standard urine samples, reflecting the healthy male and female population, were generated and characterized in detail (Mischak et al., 2010). These samples were of exceptional value in the subsequent comparison of different analytical platforms, and also in establishing transferability of the analytical platforms between the laboratories.
Evaluation of the combination of CE-MS with MALDI MS in the form of an integrated diagnostic platform, combining the high-throughput capabilities of MALDI MS with the high specificity of CE-MS involved multiple steps of analysis; The different ionization means employed by the (CE)-MS (electron spray) and MALDI-MS platforms are expected to generate potential deviations in amplitude measurements of respective peptides. In addition, the MALDI-MS technique may not be able to detect all CE-MS biomarkers and vice-versa ; As such, various sample fractionation approaches had to be tested to obtain an optimal match between the CE-MS and MALDI spectra. A case-control study (CKD patients in comparison to healthy controls) was performed, using both platforms (Molin et al., 2012).
Complimentarity of the two platforms could be established, nevertheless and as expected, CE-MS provided increased peptide resolution and better disease prediction accuracy rates compared to the MALDI MS. In support of our hypothesis, we demonstrated that several of the CKD discriminatory peptides detected by CE-MS could also be detected by MALDI MS. To enable transferring the findings (specific peptide biomarkers) from CE-MS to MALDI MS, a correlation map of the data from these two platforms was established enabling comparing data, and transferring results from one platform to the other with good confidence.(Molin et al, 2012)
Collectively these results provided the basis for follow-up studies targeting an optimization in peptide quantification by MALDI MS. This was attempted in two ways; A) development of an algorithm to adjust for the non-linear ionization of peptides in complex samples, - likely a result of signal suppression by competition for charge-. As this “quality” of MALDI prohibits semi-quantitative analysis, we developed an appropriate software solution allowing assessment of multiple sample dilutions and identification of the dilution where linear response is observed, hence enables semi-quantification by MALDI-MS (Albalat et al., 2013). The validity of this approach was demonstrated in a study on patients with CKD (Albalat et al., 2013). B) In parallel and as an alternative approach to quantify CE-MS based biomarkers by MALDI MS we aimed at developing MALDI MS-based immunoassays employing antibodies that capture specific CKD biomarkers. In the course of the project it became evident that the most relevant urinary peptide biomarkers in CKD appear to be fragments of collagen type I and of alpha-1-antitrypsin. Commercially available antibodies against these specific peptides were tested. Unfortunately, the selectivity/specificity of the antibodies was consistently too low to be of use in this context. Based on these experiments, and also exchange of information with other laboratories the conclusion was drawn that this approach is in general not feasible, as a result of the too low selectivity of antibodies.
To further evaluate the suitability of CE-MS compared to other mass-spectrometry methods for use in clinical proteomics, CE-MS was also compared to liquid chromatography (LC)-MS using the above described standard urine sample (Mischak et al., 2010). In this comparison it the complementarity of the two platforms in peptide identification could be clearly shown, with LC-MS in general providing higher coverage, attributed to higher protein load, while CE-MS was found to be of higher reproducibility attributed to calibration of the peptide migration time and elimination of carry over effects (Mullen et al 2012).
Besides the development of the aforementioned tools allowing for CE-MS data cross-talk with MALDI-MS but also LC-MS technologies, and to ensure transferability of the results, researchers from all partner laboratories were exchanged and trained in all available mass spectrometry platforms. Standard protocols and operating procedures for urine handling and processing were established in all laboratories. As a further part of these efforts, the CE-MS platform was characterized in detail (Mischak et al., 2013) for its analytical performance, (reproducibility, repeatability, detection limits, freeze-thaw effects etc|) to further set the basis for comparison between sites.
A highly relevant issue especially in biomarker analysis in urine is quantitation of compounds (biomarkers) and enabling comparative analysis. These issues require normalization of data. Since concentration varies depending on e.g. fluid intake, exercise, etc., quantitation relative to a stable reference standard appears the most appropriate approach. To serve this requirement, an software solution has been developed that allows normalization of CE-MS signals and enables semi-quantitation of MS signals in reference to internal standards, which can be freely chosen (Albalat et al., 2013).
In parallel, we also set the basis for the investigation of fetal urine as a source of biomarkers for developmental fetal disease. The basis for this study stemmed from the hypothesis and preliminary data supporting that fetal urine (FU) and/or amniotic fluid (AF) holds information on pathological changes in kidney development, and may ultimately give guidance in pregnancy counselling. We have therefore developed a protocol that enables the reproducible analysis of FU and AF. Relevant FU and AF peptide databases based on CE_-MS analysis were established. Even more importantly, the data obtained demonstrated that fetal urine can be employed as a source of biomarkers, and can inform about kidney development (Klein et al., 2013). Specific peptides in the fetal urine could be identified being predictive of disease, and they could be placed in context with relevant biological processes. These peptides were validated in a prospective study where their prognostic value for severe kidney malfunction, significantly superior to the currently used biomarkers, could be shown. In contrast to the fetal urine, the use of amniotic fluid was not found to be of any added value as a source of biomarkers, likely probably since amniotic fluid is the product of fetal urine and maternal plasma, in highly variable fractions.
Discussions with regulatory agencies in the course of the project (see also below) have clearly demonstrated a reluctance to accept urinary proteomic biomarkers for clinical implementation. The need for more clear connection of the biomarkers to renal pathophysiology was emphasized. To ease linking the identified biomarkers to relevant biological processes in CKD, KUPKB and KUPNetViz, a web-application (www.kupkb.org) to enable easier access to omics datasets in the context of CKD have been established (Moulos et al., 2013; Klein et al., 2012). Since a link of urinary peptides to molecular events in the tissue may exist via the activity of the proteases generating these urinary peptides, PROTEASIX, a tool enabling mapping of proteases onto c- and n-termini of peptides was developed (Klein J. et al., 2013). In addition and on a positive note, the CE-MS CKD biomarkers, as quantified using the optimized tools developed within PROTOCLIN, were recently approved to be used as a patient stratification means and guide for therapy in a clinical trial involving early detection of diabetic nephropathy (www.eu-priority.org).
To reach these objectives the following steps were followed:
To enable inter-platform and –laboratory comparative analysis, two standard urine samples, reflecting the healthy male and female population, were generated and characterized in detail (Mischak et al., 2010). These samples were of exceptional value in the subsequent comparison of different analytical platforms, and also in establishing transferability of the analytical platforms between the laboratories.
Evaluation of the combination of CE-MS with MALDI MS in the form of an integrated diagnostic platform, combining the high-throughput capabilities of MALDI MS with the high specificity of CE-MS involved multiple steps of analysis; The different ionization means employed by the (CE)-MS (electron spray) and MALDI-MS platforms are expected to generate potential deviations in amplitude measurements of respective peptides. In addition, the MALDI-MS technique may not be able to detect all CE-MS biomarkers and vice-versa ; As such, various sample fractionation approaches had to be tested to obtain an optimal match between the CE-MS and MALDI spectra. A case-control study (CKD patients in comparison to healthy controls) was performed, using both platforms (Molin et al., 2012).
Complimentarity of the two platforms could be established, nevertheless and as expected, CE-MS provided increased peptide resolution and better disease prediction accuracy rates compared to the MALDI MS. In support of our hypothesis, we demonstrated that several of the CKD discriminatory peptides detected by CE-MS could also be detected by MALDI MS. To enable transferring the findings (specific peptide biomarkers) from CE-MS to MALDI MS, a correlation map of the data from these two platforms was established enabling comparing data, and transferring results from one platform to the other with good confidence.(Molin et al, 2012)
Collectively these results provided the basis for follow-up studies targeting an optimization in peptide quantification by MALDI MS. This was attempted in two ways; A) development of an algorithm to adjust for the non-linear ionization of peptides in complex samples, - likely a result of signal suppression by competition for charge-. As this “quality” of MALDI prohibits semi-quantitative analysis, we developed an appropriate software solution allowing assessment of multiple sample dilutions and identification of the dilution where linear response is observed, hence enables semi-quantification by MALDI-MS (Albalat et al., 2013). The validity of this approach was demonstrated in a study on patients with CKD (Albalat et al., 2013). B) In parallel and as an alternative approach to quantify CE-MS based biomarkers by MALDI MS we aimed at developing MALDI MS-based immunoassays employing antibodies that capture specific CKD biomarkers. In the course of the project it became evident that the most relevant urinary peptide biomarkers in CKD appear to be fragments of collagen type I and of alpha-1-antitrypsin. Commercially available antibodies against these specific peptides were tested. Unfortunately, the selectivity/specificity of the antibodies was consistently too low to be of use in this context. Based on these experiments, and also exchange of information with other laboratories the conclusion was drawn that this approach is in general not feasible, as a result of the too low selectivity of antibodies.
To further evaluate the suitability of CE-MS compared to other mass-spectrometry methods for use in clinical proteomics, CE-MS was also compared to liquid chromatography (LC)-MS using the above described standard urine sample (Mischak et al., 2010). In this comparison it the complementarity of the two platforms in peptide identification could be clearly shown, with LC-MS in general providing higher coverage, attributed to higher protein load, while CE-MS was found to be of higher reproducibility attributed to calibration of the peptide migration time and elimination of carry over effects (Mullen et al 2012).
Besides the development of the aforementioned tools allowing for CE-MS data cross-talk with MALDI-MS but also LC-MS technologies, and to ensure transferability of the results, researchers from all partner laboratories were exchanged and trained in all available mass spectrometry platforms. Standard protocols and operating procedures for urine handling and processing were established in all laboratories. As a further part of these efforts, the CE-MS platform was characterized in detail (Mischak et al., 2013) for its analytical performance, (reproducibility, repeatability, detection limits, freeze-thaw effects etc|) to further set the basis for comparison between sites.
A highly relevant issue especially in biomarker analysis in urine is quantitation of compounds (biomarkers) and enabling comparative analysis. These issues require normalization of data. Since concentration varies depending on e.g. fluid intake, exercise, etc., quantitation relative to a stable reference standard appears the most appropriate approach. To serve this requirement, an software solution has been developed that allows normalization of CE-MS signals and enables semi-quantitation of MS signals in reference to internal standards, which can be freely chosen (Albalat et al., 2013).
In parallel, we also set the basis for the investigation of fetal urine as a source of biomarkers for developmental fetal disease. The basis for this study stemmed from the hypothesis and preliminary data supporting that fetal urine (FU) and/or amniotic fluid (AF) holds information on pathological changes in kidney development, and may ultimately give guidance in pregnancy counselling. We have therefore developed a protocol that enables the reproducible analysis of FU and AF. Relevant FU and AF peptide databases based on CE_-MS analysis were established. Even more importantly, the data obtained demonstrated that fetal urine can be employed as a source of biomarkers, and can inform about kidney development (Klein et al., 2013). Specific peptides in the fetal urine could be identified being predictive of disease, and they could be placed in context with relevant biological processes. These peptides were validated in a prospective study where their prognostic value for severe kidney malfunction, significantly superior to the currently used biomarkers, could be shown. In contrast to the fetal urine, the use of amniotic fluid was not found to be of any added value as a source of biomarkers, likely probably since amniotic fluid is the product of fetal urine and maternal plasma, in highly variable fractions.
Discussions with regulatory agencies in the course of the project (see also below) have clearly demonstrated a reluctance to accept urinary proteomic biomarkers for clinical implementation. The need for more clear connection of the biomarkers to renal pathophysiology was emphasized. To ease linking the identified biomarkers to relevant biological processes in CKD, KUPKB and KUPNetViz, a web-application (www.kupkb.org) to enable easier access to omics datasets in the context of CKD have been established (Moulos et al., 2013; Klein et al., 2012). Since a link of urinary peptides to molecular events in the tissue may exist via the activity of the proteases generating these urinary peptides, PROTEASIX, a tool enabling mapping of proteases onto c- and n-termini of peptides was developed (Klein J. et al., 2013). In addition and on a positive note, the CE-MS CKD biomarkers, as quantified using the optimized tools developed within PROTOCLIN, were recently approved to be used as a patient stratification means and guide for therapy in a clinical trial involving early detection of diabetic nephropathy (www.eu-priority.org).