Expert Review of Proteomics Downloaded from informahealthcare.com by Washington University Library on 01/12/15 For personal use only.
Special Report
How has urinary proteomics contributed to the discovery of early biomarkers of acute kidney injury? Expert Rev. Proteomics 11(4), 415–424 (2014)
Jorien De Loor*1, Kris Gevaert2,3, Eric Hoste4,5 and Evelyne Meyer1 1 Ghent University, Department of Pharmacology, Toxicology and Biochemistry, B-9820 Merelbeke, Belgium 2 VIB, Department of Medical Protein Research, B-9000 Ghent, Belgium 3 Ghent University, Department of Biochemistry, B-9000 Ghent, Belgium 4 Ghent University, Department of Intensive Care Medicine, Ghent University Hospital, B-9000 Ghent, Belgium 5 Research Foundation-Flanders, B-1000 Brussels, Belgium *Author for correspondence: Tel.: +32 9264 7356 Fax: +32 9264 7497 [email protected]
informahealthcare.com
In the past decade, analysis of the urinary proteome (urinary proteomics) has intensified in response to the need for novel biomarkers that support early diagnosis of kidney diseases. In particular, this also applies to acute kidney injury, which is a heterogeneous complex syndrome with a still-increasing incidence at the intensive care unit. Unfortunately, this major need remains largely unmet to date. The current report aims to explain why attempts to implement urinary proteomic-discovered acute kidney injury diagnostic candidates in the intensive care unit setting have not yet led to success. Subsequently, some key notes are provided that should enhance the chance of translating selected urinary proteomic candidates to valuable tools for the nephrologist and intensivist in the near future. KEYWORDS: acute kidney injury • animal model • diagnostic biomarker • early detection • intensive care unit • urinary proteomics
‘Kidney attack’: a novel nomenclature to sensitize clinicians for acute kidney injury & non-traditional urinary biomarkers
Acute kidney injury (AKI) encompasses a syndrome from minor kidney injury or functional impairment relative to physiological demands, to acute renal failure where kidney function needs to be replaced by extracorporeal renal replacement therapy [1]. In the intensive care unit (ICU), the reported incidence fluctuates between 30 and 60% [2–7]. Importantly, both less severe and severe AKI are independently associated with an increased risk for worse outcomes such as increased length of stay, morbidity and both ICU and long-term mortality [2,7–9]. Overall, this implies a major societal burden of AKI. Currently, the diagnosis of AKI is based on markers of decreased glomerular filtration rate (GFR), that is, serum creatinine (SCr) and urine output (UO), and therefore rather late [10]. In addition, these markers have important limitations, for example, diuretics and obstructed urinary catheters blunt the
10.1586/14789450.2014.932252
diagnostic value of UO. SCr may be false low as a consequence of decreased production or after dilution. Moreover, this gold standard only provides a good estimation of GFR when steady-state equilibrium has been achieved. Earlier diagnosis of damage to the kidney, before decreases of GFR occur, may help to prevent further deterioration and open avenues for innovative therapeutic studies by intervening in early pathophysiological events. Chertow et al. [11] calculated that an increase in SCr of 50% during hospitalization is associated with a 4.4-fold increase in the odds of in-hospital death and nearly e4000 (US$5510) in excess hospital costs. An increase in SCr of 100% was associated with a 6.5-fold increase in the odds of inhospital death and nearly e6500 (US$8999) in excess hospital costs. As the annual incidence of AKI in high-income countries, which collectively have a population of about 1 billion people, is 0.2% [12], annually 2 million people in those countries will develop AKI. When combined, the excess hospital costs [11] for the care of AKI patients in high-income countries thus approach
2014 Informa UK Ltd
ISSN 1478-9450
415
Expert Review of Proteomics Downloaded from informahealthcare.com by Washington University Library on 01/12/15 For personal use only.
Special Report
De Loor, Gevaert, Hoste & Meyer
nearly e8 billion (US$11 billion) per year. Despite these convincing facts, both non-expert clinicians and the public are not as acquainted with AKI as they are with heart attack or stroke. This is illustrated by data from Aitken et al. [13] who retrospectively showed that in their cohort (n = 1577) nearly 25% of AKI patients remained unrecognized, while a delay in AKI diagnosis of >12 h occurred in 20% of AKI patients. In addition, the ‘AKI: Adding Insult to Injury’ prospective cohort study (n = 700) of the National Confidential Enquiry into Patient Outcome and Death (NCEPOD) showed that showed that only 50% of hospitalized patients who died from/with AKI had received good care [14]. These and other observations prompted Kellum et al. to introduce the term ‘kidney attack’ [15] in order to get this syndrome the attention that it deserves. In most patients, the early phase of AKI is accompanied with few or even no kidney symptoms. As stated, overt (laboratory) signs may also be lacking. The quest for novel diagnostic injury AKI biomarkers or a ‘renal troponin’ [16], referring to the cardiac troponin necrosis biomarkers in acute coronary syndrome, inspired Goldstein and Lakhmir to define a ‘renal angina’ syndrome [17]. Angina pectoris – literally strapping of the chest – is the hallmark clinical symptom of coronary ischemia prompting immediate further investigation to rule in or out acute myocardial infarction. How can this apply to AKI that usually does not hurt? The authors defined different renal angina syndromes [17], taking into account identifiable AKI risk factors together with the clinical context to sensitize the diagnostic thresholds of the traditional functional AKI parameters SCr and UO [1]. In other words, within the appropriate clinical context, renal angina can be diagnosed when functional AKI parameters tend to diagnostic criteria in patients at increased risk for AKI. Thus, like unstable angina is a phase in acute coronary syndrome, renal angina is a phase in the AKI spectrum. Not surprisingly, in its latest redefinition [18], AKI was compared with the diagnostic work-up of acute coronary syndrome [19,20], which is familiar to clinicians. The key is the reliability on renal troponin biomarkers to diagnose AKI in the absence of a positive SCr criterion, UO criterion or both and the incorporation of the renal angina phase. Or translated to the acute coronary syndrome setting: the reliability on cardiac troponin biomarkers to diagnose acute myocardial infarction in the absence of ST-elevation on electrocardiography in patients with unstable angina. In this special report, a critical comment on urinary proteomics in the setting of AKI will be given, covering both technical aspects and hurdles in clinical study design and interpretation, as the novel (yet to be) discovered urinary biomarkers are predestined to partly found future AKI diagnosis. Likely, AKI will need a ‘renal troponin panel’ to increase the accuracy of its diagnosis rather than just one ‘renal troponin’. Although this report will focus on diagnostic AKI biomarkers, it should be emphasized that the context of use of biomarkers is much broader. Also, it should be noted that causal and sensitizing AKI risk factors, or exposures and susceptibilities, form the pillars of the individual 416
AKI risk assessment [1]. However, these risk factors and biomarkers are sometimes mixed up, which can be rather confusing. The essential difference between both is explained in the scheme on FIGURE 1, providing also an overview of different types of biomarkers relevant to AKI. In the intensive care setting, sepsis and cardiac surgery – especially with cardiopulmonary bypass (CPB) – are very important exposures with a considerable different pathophysiology of AKI. Septic shock (47.5%) and major surgery (34.3%) were identified as the top contributing factors of AKI in ICU patients [21], with cardiovascular surgery as the most common reason for ICU admission (23.2%). Therefore, this special report will focus on sepsis-induced (SI) and cardiac surgery-associated (CSA) AKI. Urinary proteomics & AKI: analytical pitfalls, methodological considerations & interpretation of results when searching for a ‘renal troponin’
At first sight, urine is the patient’s specimen of choice for proteome studies for identifying diagnostic and prognostic AKI biomarkers. Changes in the urinary proteome should indeed reflect changes in kidney function, urine collection is noninvasive and urine is typically available in large volumes. Further, urinary proteins appear quite stable due to low proteolytic activities in urine. However, proteomics on urine samples does face major obstacles as the overall protein concentration is low, while the salt concentration is high, hence sample concentration steps are needed, which should be efficient and reproducible [22]. Furthermore, the protein concentration of urine fluctuates, which is a clear obstacle when studying patients with varying degrees of proteinuria. In addition, proteins and their fragments identified in urine may originate from damage to tissues other than renal tissue, which might lead to inappropriate protein markers for AKI [23]. Jia and colleagues recently analyzed the urinary proteome collected from perfused healthy rat kidneys by combining protein separation by Sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE) with Liquid chromatography (LC)-mass spectrometry (MS)/MS analysis on trypsin-digested gel slices. Close to 1000 proteins with kidney origin and analogs in humans were identified [24]. Interestingly, 128 and 297 of these proteins were not found in normal human urine and plasma, respectively, making them potential biomarkers with zero background in urine and with ideally minor plasma-derived contribution under pathological conditions. The 57 proteins that possess both these advantages are thus possibly the primary pool of future protein biomarkers for renal diseases such as AKI. Although the literature on AKI proteomics is rather scarce, several approaches have been taken to identify AKI-related protein changes, especially in humans. Indeed, there are simply no adequate preclinical models for human AKI [25], as is also the case for human sepsis [26]. This does not preclude the use of animal models, though. After all, a good validation is pivotal for all candidate biomarkers. A selected overview to illustrate Expert Rev. Proteomics 11(4), (2014)
Expert Review of Proteomics Downloaded from informahealthcare.com by Washington University Library on 01/12/15 For personal use only.
Urinary proteomics in acute kidney injury
Special Report
• Which patients are developing AKI, what is their prognosis, which AKI patients need individual therapy? • Classification of biomarkers relevant to AKI: • Diagnostic AKI biomarkers – To early detect AKI, often referred to as ‘predict’; the imperfect functional gold standard SCr lags behind early injury biomarkers – To confirm the renal nature of the observed azotemia with late injury biomarkers • Outcome AKI biomarkers – To make a prognosis, i.e., will a patient progress to a worse AKI stage, need RRT, become a CKD patient,...? • Predictive AKI biomarkers – To identify AKI patients likely to respond well to a specific therapy, i.e., which patient will benefit the most from RRT, specific pathophysiology-directed therapies,...?
AKI risk factors
• Which individuals will have either a low, a moderate or a high risk of developing AKI when hospitalized? • Exposures – Always necessary to cause AKI – Sometimes sufficient to cause AKI • Susceptibilities – Neither sufficient to cause AKI – Neither necessary to cause AKI
AKI biomarkers
Preventive intervention and therapy Supportive intervention and therapy Exposure containment and prevention Follow-up
Figure 1. The gear wheels ‘AKI risk factors’ and ‘AKI biomarkers’ set into motion a good AKI management; only when both these gear wheels work well, the management gear wheel will turn around smoothly. AKI: Acute kidney injury; CIN: Contrast-induced nephropathy; CKD: Chronic kidney disease; RRT: Renal replacement therapy; SCr: Serum creatinine.
the techniques used is presented below. The databases Web of ScienceTM and PubMed.gov were searched with the keywords proteomic* and kidney. Original articles were considered relevant for this manuscript when they described a urinary proteomic study to discover AKI risk factors or biomarkers in ICU, sepsis or cardiac surgery patients or in ischemia/reperfusion (I/R) or sepsis animal models. Diagnostic AKI biomarker studies were given preference. The surface-enhanced laser desorption/ionization time-offlight mass spectrometry (SELDI-TOF-MS) approach was used in many of these AKI biomarker studies. In essence, SELDIMS is a profiling technique which reads out an m/z profile of peptides and (small) proteins spotted on a surface-coated chip. By changing the surface coating, different m/z profiles can be obtained, allowing for versatile data generation. SELDI-MS informahealthcare.com
requires few sample preparation steps and allows for rapid and automated data generation, and several algorithms are available for comparing SELDI-MS profiles and pointing to distinctive features. These SELDI-MS features likely contributed to its initial success for profiling patients’ specimens such as urine by non-experts in the field of mass spectrometry. Nevertheless, a disadvantage of SELDI-MS is that from a given m/z value, which is typically measured with low resolution and low accuracy, the actual identity of the corresponding peptide or protein is not revealed. This implies that further analysis, such as SDS-PAGE followed by in-gel protein digestion and LC-MS/MS analysis, is desired to identify the protein that gave rise to the distinctive m/z feature. The latter information can then be used to further validate such a possible biomarker using for instance ELISA. Nguyen and 417
Expert Review of Proteomics Downloaded from informahealthcare.com by Washington University Library on 01/12/15 For personal use only.
Special Report
De Loor, Gevaert, Hoste & Meyer
colleagues reported on the use of SELDI-MS to predict acute renal failure [27]. Urine samples of 60 children who underwent CPB surgery were analyzed at two different time points post-surgery. Of these patients, 15 developed acute renal failure and SELDI-MS profiles pointed to increased signals of four molecules in their urine samples, allowing the researchers to distinguish with high sensitivity and specificity – albeit on a very small patient population – acute renal failure. An analogous study was described by Ho and colleagues, who amongst others identified hepcidin-25 as highly present in post-CPB surgery urine samples of adult patients that did not develop AKI [28]. The group of Devarajan reported on the identification of an unidentified m/z signal that was only present in the SELDI-MS urinary proteome profile of pediatric AKI patients prior to elective cardiac catheterization, thus identifying a possible AKI risk factor [29]. In 2010, the same group also identified the best three performing peaks of their former study (Nguyen et al. [27]) and subsequently validated a1-microglobulin, a1-acid glycoprotein and albumin as both early diagnostic and potentially prognostic urinary AKI biomarkers in children undergoing CPB [30]. Compared to SELDI-MS, 2D electrophoresis of proteins (e.g., 2D-DIGE) is expected to provide a more comprehensive view on the actual proteome profile of analyzed samples. Although it is most effective for protein separation, this technology suffers from recognized weaknesses such as poor coverage of low abundance proteins as well as very hydrophobic proteins. On the other hand, it performs well in separating protein isoforms such as those caused by alternative splicing. In 2010, Aregger and colleagues examined 2D-DIGE urinary profiles of patients prior to and following CPB surgery and distinguished patients that developed AKI post-surgery [31]. In the urine samples of these patients, a modified form of urinary albumin was found at increased levels, whereas an adrenomedullin-binding protein fragment and the zinc-a2-glycoprotein were found at decreased levels. Decreased excretion of the latter protein was further validated by Western blot analysis and ELISA. A recent study by the same group also used 2D-DIGE for urinary proteomics of critically ill patients with AKI and compared the obtained proteome profiles between patients with early recovery to those with either late or no recovery [32]. A total of eight prognostic candidates were identified from a mean of 1090 protein spots. Subsequent ELISA analysis hereof showed that the insulinlike growth factor (IGF)-binding protein 7 (IGFBP-7) had the highest predictive power for renal recovery. Very recently, 2D-PAGE, followed by SDS-PAGE and trypsin digestion on cut-out spots prior to matrix-assisted laser desorption/ionization-MS, was used [33]. This study identified hemojuvelin, a key regulator of the above-mentioned hepcidin-25, as an early diagnostic AKI biomarker in adult cardiac surgery patients. Instead of analyzing the total urinary protein content by 2DDIGE, Zhou and colleagues enriched exosomes in urine samples obtained from a rat AKI model and used matrix-assisted 418
laser desorption/ionization-TOF/TOF-MS and LC-MS/MS to identify several proteins that were present at different levels in these exosomes. Interestingly, Western blotting failed to validate the majority of these proteins except for fetuin-A, which was present at significantly higher levels and for several days following the onset in human AKI patients [34]. In 2007, Vanhoutte et al. already suggested that the presumed ubiquitin protein observed in a creatinine-corrected master pool from cardiac surgery patients originated from urinary exosomes [23]. By focusing on urinary exosomes, the complexity of the proteome sample is reduced as the most abundant urinary proteins are not found in these exosomes. However, efficient and reproducible enrichment of exosomes from body fluids such as urine remains not straightforward and lengthy till date, and this might impede future use of exosome-specific biomarkers. Gel-free, mass spectrometry-driven proteome analysis overcomes a number of the disadvantages listed for 2D-DIGEbased proteomics [35]. Such gel-free technologies are based on chromatographic separation of peptides in one or more dimensions prior to LC-MS/MS analysis and differ in the nature and number of separations employed. Recently, our group has applied post-metabolic labeling, peptide pre-fractionation by reverse-phase HPLC to LC-MS/MS analysis to identify protein markers for SI-AKI in a novel mouse model [36]. Amongst others, the chitinase 3-like (CHI3L) proteins 1 and 3 were only found in septic mice with severe AKI. The human homologue CHI3L1 was further increasingly excreted in urine from septic patients with AKI compared to septic patients without AKI. Capillary electrophoresis (CE) has been proposed as an alternative approach to LC-based urinary peptide separation prior to MS [37]. The apparent delayed application of CE-MS within the urinary proteomics field can be attributed to its major inherent limitation – i.e., the very limited volumes (
Special Report
How has urinary proteomics contributed to the discovery of early biomarkers of acute kidney injury? Expert Rev. Proteomics 11(4), 415–424 (2014)
Jorien De Loor*1, Kris Gevaert2,3, Eric Hoste4,5 and Evelyne Meyer1 1 Ghent University, Department of Pharmacology, Toxicology and Biochemistry, B-9820 Merelbeke, Belgium 2 VIB, Department of Medical Protein Research, B-9000 Ghent, Belgium 3 Ghent University, Department of Biochemistry, B-9000 Ghent, Belgium 4 Ghent University, Department of Intensive Care Medicine, Ghent University Hospital, B-9000 Ghent, Belgium 5 Research Foundation-Flanders, B-1000 Brussels, Belgium *Author for correspondence: Tel.: +32 9264 7356 Fax: +32 9264 7497 [email protected]
informahealthcare.com
In the past decade, analysis of the urinary proteome (urinary proteomics) has intensified in response to the need for novel biomarkers that support early diagnosis of kidney diseases. In particular, this also applies to acute kidney injury, which is a heterogeneous complex syndrome with a still-increasing incidence at the intensive care unit. Unfortunately, this major need remains largely unmet to date. The current report aims to explain why attempts to implement urinary proteomic-discovered acute kidney injury diagnostic candidates in the intensive care unit setting have not yet led to success. Subsequently, some key notes are provided that should enhance the chance of translating selected urinary proteomic candidates to valuable tools for the nephrologist and intensivist in the near future. KEYWORDS: acute kidney injury • animal model • diagnostic biomarker • early detection • intensive care unit • urinary proteomics
‘Kidney attack’: a novel nomenclature to sensitize clinicians for acute kidney injury & non-traditional urinary biomarkers
Acute kidney injury (AKI) encompasses a syndrome from minor kidney injury or functional impairment relative to physiological demands, to acute renal failure where kidney function needs to be replaced by extracorporeal renal replacement therapy [1]. In the intensive care unit (ICU), the reported incidence fluctuates between 30 and 60% [2–7]. Importantly, both less severe and severe AKI are independently associated with an increased risk for worse outcomes such as increased length of stay, morbidity and both ICU and long-term mortality [2,7–9]. Overall, this implies a major societal burden of AKI. Currently, the diagnosis of AKI is based on markers of decreased glomerular filtration rate (GFR), that is, serum creatinine (SCr) and urine output (UO), and therefore rather late [10]. In addition, these markers have important limitations, for example, diuretics and obstructed urinary catheters blunt the
10.1586/14789450.2014.932252
diagnostic value of UO. SCr may be false low as a consequence of decreased production or after dilution. Moreover, this gold standard only provides a good estimation of GFR when steady-state equilibrium has been achieved. Earlier diagnosis of damage to the kidney, before decreases of GFR occur, may help to prevent further deterioration and open avenues for innovative therapeutic studies by intervening in early pathophysiological events. Chertow et al. [11] calculated that an increase in SCr of 50% during hospitalization is associated with a 4.4-fold increase in the odds of in-hospital death and nearly e4000 (US$5510) in excess hospital costs. An increase in SCr of 100% was associated with a 6.5-fold increase in the odds of inhospital death and nearly e6500 (US$8999) in excess hospital costs. As the annual incidence of AKI in high-income countries, which collectively have a population of about 1 billion people, is 0.2% [12], annually 2 million people in those countries will develop AKI. When combined, the excess hospital costs [11] for the care of AKI patients in high-income countries thus approach
2014 Informa UK Ltd
ISSN 1478-9450
415
Expert Review of Proteomics Downloaded from informahealthcare.com by Washington University Library on 01/12/15 For personal use only.
Special Report
De Loor, Gevaert, Hoste & Meyer
nearly e8 billion (US$11 billion) per year. Despite these convincing facts, both non-expert clinicians and the public are not as acquainted with AKI as they are with heart attack or stroke. This is illustrated by data from Aitken et al. [13] who retrospectively showed that in their cohort (n = 1577) nearly 25% of AKI patients remained unrecognized, while a delay in AKI diagnosis of >12 h occurred in 20% of AKI patients. In addition, the ‘AKI: Adding Insult to Injury’ prospective cohort study (n = 700) of the National Confidential Enquiry into Patient Outcome and Death (NCEPOD) showed that showed that only 50% of hospitalized patients who died from/with AKI had received good care [14]. These and other observations prompted Kellum et al. to introduce the term ‘kidney attack’ [15] in order to get this syndrome the attention that it deserves. In most patients, the early phase of AKI is accompanied with few or even no kidney symptoms. As stated, overt (laboratory) signs may also be lacking. The quest for novel diagnostic injury AKI biomarkers or a ‘renal troponin’ [16], referring to the cardiac troponin necrosis biomarkers in acute coronary syndrome, inspired Goldstein and Lakhmir to define a ‘renal angina’ syndrome [17]. Angina pectoris – literally strapping of the chest – is the hallmark clinical symptom of coronary ischemia prompting immediate further investigation to rule in or out acute myocardial infarction. How can this apply to AKI that usually does not hurt? The authors defined different renal angina syndromes [17], taking into account identifiable AKI risk factors together with the clinical context to sensitize the diagnostic thresholds of the traditional functional AKI parameters SCr and UO [1]. In other words, within the appropriate clinical context, renal angina can be diagnosed when functional AKI parameters tend to diagnostic criteria in patients at increased risk for AKI. Thus, like unstable angina is a phase in acute coronary syndrome, renal angina is a phase in the AKI spectrum. Not surprisingly, in its latest redefinition [18], AKI was compared with the diagnostic work-up of acute coronary syndrome [19,20], which is familiar to clinicians. The key is the reliability on renal troponin biomarkers to diagnose AKI in the absence of a positive SCr criterion, UO criterion or both and the incorporation of the renal angina phase. Or translated to the acute coronary syndrome setting: the reliability on cardiac troponin biomarkers to diagnose acute myocardial infarction in the absence of ST-elevation on electrocardiography in patients with unstable angina. In this special report, a critical comment on urinary proteomics in the setting of AKI will be given, covering both technical aspects and hurdles in clinical study design and interpretation, as the novel (yet to be) discovered urinary biomarkers are predestined to partly found future AKI diagnosis. Likely, AKI will need a ‘renal troponin panel’ to increase the accuracy of its diagnosis rather than just one ‘renal troponin’. Although this report will focus on diagnostic AKI biomarkers, it should be emphasized that the context of use of biomarkers is much broader. Also, it should be noted that causal and sensitizing AKI risk factors, or exposures and susceptibilities, form the pillars of the individual 416
AKI risk assessment [1]. However, these risk factors and biomarkers are sometimes mixed up, which can be rather confusing. The essential difference between both is explained in the scheme on FIGURE 1, providing also an overview of different types of biomarkers relevant to AKI. In the intensive care setting, sepsis and cardiac surgery – especially with cardiopulmonary bypass (CPB) – are very important exposures with a considerable different pathophysiology of AKI. Septic shock (47.5%) and major surgery (34.3%) were identified as the top contributing factors of AKI in ICU patients [21], with cardiovascular surgery as the most common reason for ICU admission (23.2%). Therefore, this special report will focus on sepsis-induced (SI) and cardiac surgery-associated (CSA) AKI. Urinary proteomics & AKI: analytical pitfalls, methodological considerations & interpretation of results when searching for a ‘renal troponin’
At first sight, urine is the patient’s specimen of choice for proteome studies for identifying diagnostic and prognostic AKI biomarkers. Changes in the urinary proteome should indeed reflect changes in kidney function, urine collection is noninvasive and urine is typically available in large volumes. Further, urinary proteins appear quite stable due to low proteolytic activities in urine. However, proteomics on urine samples does face major obstacles as the overall protein concentration is low, while the salt concentration is high, hence sample concentration steps are needed, which should be efficient and reproducible [22]. Furthermore, the protein concentration of urine fluctuates, which is a clear obstacle when studying patients with varying degrees of proteinuria. In addition, proteins and their fragments identified in urine may originate from damage to tissues other than renal tissue, which might lead to inappropriate protein markers for AKI [23]. Jia and colleagues recently analyzed the urinary proteome collected from perfused healthy rat kidneys by combining protein separation by Sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE) with Liquid chromatography (LC)-mass spectrometry (MS)/MS analysis on trypsin-digested gel slices. Close to 1000 proteins with kidney origin and analogs in humans were identified [24]. Interestingly, 128 and 297 of these proteins were not found in normal human urine and plasma, respectively, making them potential biomarkers with zero background in urine and with ideally minor plasma-derived contribution under pathological conditions. The 57 proteins that possess both these advantages are thus possibly the primary pool of future protein biomarkers for renal diseases such as AKI. Although the literature on AKI proteomics is rather scarce, several approaches have been taken to identify AKI-related protein changes, especially in humans. Indeed, there are simply no adequate preclinical models for human AKI [25], as is also the case for human sepsis [26]. This does not preclude the use of animal models, though. After all, a good validation is pivotal for all candidate biomarkers. A selected overview to illustrate Expert Rev. Proteomics 11(4), (2014)
Expert Review of Proteomics Downloaded from informahealthcare.com by Washington University Library on 01/12/15 For personal use only.
Urinary proteomics in acute kidney injury
Special Report
• Which patients are developing AKI, what is their prognosis, which AKI patients need individual therapy? • Classification of biomarkers relevant to AKI: • Diagnostic AKI biomarkers – To early detect AKI, often referred to as ‘predict’; the imperfect functional gold standard SCr lags behind early injury biomarkers – To confirm the renal nature of the observed azotemia with late injury biomarkers • Outcome AKI biomarkers – To make a prognosis, i.e., will a patient progress to a worse AKI stage, need RRT, become a CKD patient,...? • Predictive AKI biomarkers – To identify AKI patients likely to respond well to a specific therapy, i.e., which patient will benefit the most from RRT, specific pathophysiology-directed therapies,...?
AKI risk factors
• Which individuals will have either a low, a moderate or a high risk of developing AKI when hospitalized? • Exposures – Always necessary to cause AKI – Sometimes sufficient to cause AKI • Susceptibilities – Neither sufficient to cause AKI – Neither necessary to cause AKI
AKI biomarkers
Preventive intervention and therapy Supportive intervention and therapy Exposure containment and prevention Follow-up
Figure 1. The gear wheels ‘AKI risk factors’ and ‘AKI biomarkers’ set into motion a good AKI management; only when both these gear wheels work well, the management gear wheel will turn around smoothly. AKI: Acute kidney injury; CIN: Contrast-induced nephropathy; CKD: Chronic kidney disease; RRT: Renal replacement therapy; SCr: Serum creatinine.
the techniques used is presented below. The databases Web of ScienceTM and PubMed.gov were searched with the keywords proteomic* and kidney. Original articles were considered relevant for this manuscript when they described a urinary proteomic study to discover AKI risk factors or biomarkers in ICU, sepsis or cardiac surgery patients or in ischemia/reperfusion (I/R) or sepsis animal models. Diagnostic AKI biomarker studies were given preference. The surface-enhanced laser desorption/ionization time-offlight mass spectrometry (SELDI-TOF-MS) approach was used in many of these AKI biomarker studies. In essence, SELDIMS is a profiling technique which reads out an m/z profile of peptides and (small) proteins spotted on a surface-coated chip. By changing the surface coating, different m/z profiles can be obtained, allowing for versatile data generation. SELDI-MS informahealthcare.com
requires few sample preparation steps and allows for rapid and automated data generation, and several algorithms are available for comparing SELDI-MS profiles and pointing to distinctive features. These SELDI-MS features likely contributed to its initial success for profiling patients’ specimens such as urine by non-experts in the field of mass spectrometry. Nevertheless, a disadvantage of SELDI-MS is that from a given m/z value, which is typically measured with low resolution and low accuracy, the actual identity of the corresponding peptide or protein is not revealed. This implies that further analysis, such as SDS-PAGE followed by in-gel protein digestion and LC-MS/MS analysis, is desired to identify the protein that gave rise to the distinctive m/z feature. The latter information can then be used to further validate such a possible biomarker using for instance ELISA. Nguyen and 417
Expert Review of Proteomics Downloaded from informahealthcare.com by Washington University Library on 01/12/15 For personal use only.
Special Report
De Loor, Gevaert, Hoste & Meyer
colleagues reported on the use of SELDI-MS to predict acute renal failure [27]. Urine samples of 60 children who underwent CPB surgery were analyzed at two different time points post-surgery. Of these patients, 15 developed acute renal failure and SELDI-MS profiles pointed to increased signals of four molecules in their urine samples, allowing the researchers to distinguish with high sensitivity and specificity – albeit on a very small patient population – acute renal failure. An analogous study was described by Ho and colleagues, who amongst others identified hepcidin-25 as highly present in post-CPB surgery urine samples of adult patients that did not develop AKI [28]. The group of Devarajan reported on the identification of an unidentified m/z signal that was only present in the SELDI-MS urinary proteome profile of pediatric AKI patients prior to elective cardiac catheterization, thus identifying a possible AKI risk factor [29]. In 2010, the same group also identified the best three performing peaks of their former study (Nguyen et al. [27]) and subsequently validated a1-microglobulin, a1-acid glycoprotein and albumin as both early diagnostic and potentially prognostic urinary AKI biomarkers in children undergoing CPB [30]. Compared to SELDI-MS, 2D electrophoresis of proteins (e.g., 2D-DIGE) is expected to provide a more comprehensive view on the actual proteome profile of analyzed samples. Although it is most effective for protein separation, this technology suffers from recognized weaknesses such as poor coverage of low abundance proteins as well as very hydrophobic proteins. On the other hand, it performs well in separating protein isoforms such as those caused by alternative splicing. In 2010, Aregger and colleagues examined 2D-DIGE urinary profiles of patients prior to and following CPB surgery and distinguished patients that developed AKI post-surgery [31]. In the urine samples of these patients, a modified form of urinary albumin was found at increased levels, whereas an adrenomedullin-binding protein fragment and the zinc-a2-glycoprotein were found at decreased levels. Decreased excretion of the latter protein was further validated by Western blot analysis and ELISA. A recent study by the same group also used 2D-DIGE for urinary proteomics of critically ill patients with AKI and compared the obtained proteome profiles between patients with early recovery to those with either late or no recovery [32]. A total of eight prognostic candidates were identified from a mean of 1090 protein spots. Subsequent ELISA analysis hereof showed that the insulinlike growth factor (IGF)-binding protein 7 (IGFBP-7) had the highest predictive power for renal recovery. Very recently, 2D-PAGE, followed by SDS-PAGE and trypsin digestion on cut-out spots prior to matrix-assisted laser desorption/ionization-MS, was used [33]. This study identified hemojuvelin, a key regulator of the above-mentioned hepcidin-25, as an early diagnostic AKI biomarker in adult cardiac surgery patients. Instead of analyzing the total urinary protein content by 2DDIGE, Zhou and colleagues enriched exosomes in urine samples obtained from a rat AKI model and used matrix-assisted 418
laser desorption/ionization-TOF/TOF-MS and LC-MS/MS to identify several proteins that were present at different levels in these exosomes. Interestingly, Western blotting failed to validate the majority of these proteins except for fetuin-A, which was present at significantly higher levels and for several days following the onset in human AKI patients [34]. In 2007, Vanhoutte et al. already suggested that the presumed ubiquitin protein observed in a creatinine-corrected master pool from cardiac surgery patients originated from urinary exosomes [23]. By focusing on urinary exosomes, the complexity of the proteome sample is reduced as the most abundant urinary proteins are not found in these exosomes. However, efficient and reproducible enrichment of exosomes from body fluids such as urine remains not straightforward and lengthy till date, and this might impede future use of exosome-specific biomarkers. Gel-free, mass spectrometry-driven proteome analysis overcomes a number of the disadvantages listed for 2D-DIGEbased proteomics [35]. Such gel-free technologies are based on chromatographic separation of peptides in one or more dimensions prior to LC-MS/MS analysis and differ in the nature and number of separations employed. Recently, our group has applied post-metabolic labeling, peptide pre-fractionation by reverse-phase HPLC to LC-MS/MS analysis to identify protein markers for SI-AKI in a novel mouse model [36]. Amongst others, the chitinase 3-like (CHI3L) proteins 1 and 3 were only found in septic mice with severe AKI. The human homologue CHI3L1 was further increasingly excreted in urine from septic patients with AKI compared to septic patients without AKI. Capillary electrophoresis (CE) has been proposed as an alternative approach to LC-based urinary peptide separation prior to MS [37]. The apparent delayed application of CE-MS within the urinary proteomics field can be attributed to its major inherent limitation – i.e., the very limited volumes (
