Comparative Urine Protein Phenotyping Using Mass Spectrometric Immunoassay Urban A. Kiernan,† Kemmons A. Tubbs,† Dobrin Nedelkov,† Eric E. Niederkofler,† Elizabeth McConnell,‡ and Randall W. Nelson*,† Intrinsic Bioprobes, Inc., 625 South Smith Road, Suite 22, Tempe, Arizona 85281, and Arizona State University, Tempe, Arizona 85287 Received October 1, 2002
Reported here, human urine samples were analyzed for β-2-microglobulin (β2m), transthyretin (TTR), cystatin C, urine protein 1 (UP1), retinol binding protein (RBP), albumin, transferrin, and human neutrophil defensin peptides (HNP) using mass spectrometric immunoassay (MSIA). MSIA is a unique analytical technique, which allows for the generation of distinct protein profiles of specific target proteins from each subject, which may be subsequently used in comparative protein expression profiling between all subjects. Comparative profiling allows for the rapid identification of variations within individual protein expression profiles. Although the majority of analyses performed in this study revealed homology between study participants, roughly one-quarter showed variation in the protein profiles. Some of these observed variants included a point mutation in TTR, absence of wild-type RBP, monomeric forms UP1, a novel β2m glycated end product and altered HNP ratios. MSIA has been previously used in the analysis of blood proteins, but this study shows how MSIA easily transitions to the analysis of urine samples. This study displays how qualitative urine protein differentiation is readily achievable with MSIA and is useful in identifying proteomic differences between subjects that might be otherwise overlooked with other analytical techniques due to complexity of the resulting data or insufficient sensitivity. Keywords: proteomics • protein variations • MALDI-TOF • urine • biomarker discovery
Introduction Urine is an easily accessible biological fluid that has lately become more intensely studied in the quest to identify protein and peptide biomarkers that may potentially be used to assess kidney function and identify the presence of disease in the individual. Many small proteins and peptides freely pass though the glomerulus, where they are then either catabolized within the tubular cells of the kidney or are excreted in the urine.1 Abnormalities in kidney function, and the presence of disease, often result in variations in urine protein excretion rate and content, both of which have been historically monitored via enzyme-linked immunosorbent assays (ELISA).2,3 This, along with the fact that the acquisition of urine is normally a noninvasive procedure, makes it an ideal biological fluid for human proteomics studies. The field of proteomics is developing new technologies and methodologies toward the analysis of proteins from a variety of biological fluids, including urine. A common proteomic approach to analyzing urine proteins involves 2-dimensional polyacrylamide gel electrophoresis (2D-PAGE) for protein separation. Even though this method is capable of separating * To whom correspondence should be addressed. Tel: (480) 804-1778. Fax: (480) 804-0778. E-mail:
[email protected]. † Intrinsic Bioprobes, Inc.. ‡ Arizona State University. 10.1021/pr025574c CCC: $25.00
2003 American Chemical Society
hundreds to thousands of proteins in a single analysis, it is not without weakness. 2D-PAGE has in the past had poor results in the analysis of peptides due to their high mobilities.4 The identification of separated proteins, as well as the detection of low-abundance proteins has also been historically problematic. A more recent innovation, which incorporates 2D-PAGE with mass spectrometry (2DE/MS), provides more accurate results, but requires enzymatic digestion of the isolated proteins.5 Although able to accurately identify genes from which proteolytic fragments originate,6 the 2DE/MS approaches are not readily able to yield information on the full-length protein and oftentimes subtle details on the analyte (e.g., the presence of point mutations and post-translational modifications) are missed. Because this information can be lost, analysis of intact proteins is becoming more widely accepted as a mean of proteomic analysis.7 A proteomics technology that has great potential in the area of intact urine protein analyses is the mass spectrometric immunoassay (MSIA). MSIA combines the selectivity of an immunoassay with the sensitivity, resolution, and mass accuracy of matrix assisted laser desorption/ionization time-offlight mass spectrometry (MALDI-TOF MS). A major advantage of mass spectrometric detection over other conventional immunoassay schemes is the ability to discriminate between protein variants in a single assay.8 Because retrieved species Journal of Proteome Research 2003, 2, 191-197
191
Published on Web 01/16/2003
research articles
Kiernan et al.
are detected at precise molecular masses, mass-shifted variants of a protein (i.e., post-translational modifications or point mutations) are readily detected in a single assay. This approach is contrary to conventional immunoassays, where each protein variant would require an individual, monospecific assay, which obviously requires an a priori knowledge of the variants under investigation (i.e., a monospecific assay must be constructed for each variant after the variant is either discovered or hypothesized). Conversely, the MSIA approach is able to discover as yet unidentified variants of proteins through the use of pan antibodies towards the protein of interest. Thus, MSIA, when applied to the routine screening of known protein variants (i.e., wild-type), holds much potential in the discovery and identification of variants resulting from post-translational modifications, splicing variations or point mutations. Moreover, MSIA in the form of an affinity pipettor tip is capable of analyzing very low abundance proteins via a repetitive pipetting action. The flowing action of the pipettor tip concentrates and purifies the target protein prior to MALDITOF MS analysis allowing for routine analyses of protein targets in the picomolar and sub-picomolar range.9 Once in the mass spectrometer, different forms of the same protein are readily distinguishable by measurable alterations in molecular mass, thus allowing for protein isoform identification. The general utility of MSIA is not that of generating global gene product expression profiles, as done by many other proteomics technologies, but is that of targeting a specific protein and analyzing all endogenous forms of the intact target protein from within a specific human biological fluid. In this manner, the exact form(s) of a proteinsrather than those presumed from a genomic database searchscan be determined within an individual, and slight changes in structure discerned for (ultimately) correlation with disease. Previously, MSIA has been demonstrated in the high throughput quantification and characterization of various human plasma proteins10-12 but still has been largely unexploited in the field of urine protein analyses, with the only reported application being the quantitation of β-2-microglobulin (β2m)13 and the comparative profiling of retinol binding protein (RBP).14 Described here is the development of further urine-based MSIA assays targeting transthyretin (TTR), cystatin C (CYSC), urine protein 1 (UP1), albumin (ALB), transferrin (TRFE), and human neutrophil defensins (HNP), and a brief study illustrating their use in profiling these proteins between individuals.
Experimental Section Study Subjects. Urine samples were collected from 5 unrelated male subjects, ages ranging from 26 to 79. Four subjects, ages 26-68, were healthy study participants, whereas one individual, age 79, was diagnosed with pancreatic cancer. Urine samples were obtained via protocols approved through Intrinsic Bioprobes Inc.’s Internal Review Board (IRB). The individuals had read and signed an Informed Consent form. Sample Preparation. Urine samples, 25 mL mid-stream voids, from five individuals were collected. The urine was collected directly into sterile urine collection cups that were pretreated with 50 µL of the protease inhibitor cocktail consisting of AEBSF (100 mM); aprotin (80 µM); bestatin (5 mM); E-64 (1.5 mM); leupeptin (2 mM); pepstatin A (1 mM) to prevent any enzymatic breakdown or modification. Samples were collected and stored at -70°C until ready for analysis. Samples were thawed in a warm water bath (37 °C) just prior to analysis. 192
Journal of Proteome Research • Vol. 2, No. 2, 2003
Figure 1. Mass spectrum of urine diluted 1:20. Small peptides were present, whereas target proteins were not observed.
Sample Analysis. Each sample was combined 1:1 (v/v) with 2M ammonium acetate (to adjust the pH to ∼6.8-7.2) and poured into an individual poly(vinyl chloride) solution basin prior to analysis. Each sample was individually addressed with individual MSIA-Tips (Intrinsic Bioprobes, Inc.) derivatized with anti-β2m, anti-TTR, anti-CYSC, anti-UP1, anti-RBP, anti-ALB, anti-TRFE, or carboxylic acid surface (cation exchange for HNP analysis). All eight analyses were run in parallel through each sample with MSIA-Tips loaded onto an octapette. The manual incubation of each urine sample consisted of 300 cycles (150 µL of sample) through each MSIA-Tip. After incubation, tips were thoroughly rinsed using HBS buffer (10 cycles, 150 µL), doubly distilled water (5 cycles, 150 µL), 20% acetoniltrile/1M ammonium acetate wash (10 cycles, 150 µL) and finally with doubly distilled water (15 cycles, 150 µL). Retained species were eluted by drawing 4 µL of MALDI matrix solution (saturated aqueous solution of sinapic acid (SA), in 33% (v/v) acetonitrile, 0.4% (v/v) trifluoroacetic acid) into each tip and depositing directly onto a 96-well formatted hydrophobic/hydrophilic contrasting MALDI-TOF target.10 Because of the larger sample volumes of urine and the number of iterations used, the time spent to run the assays were ∼20minutes/sample. MALDI-TOF mass spectrometry and data analysis were performed on all samples as described previously.10 Acquired mass spectra all had mass accuracy within 0.01%, which was sufficient to correctly identify target proteins.
Results and Discussion The analysis of urine directly with MALDI-TOF MS is shown in Figure 1. The spectrum was produced by diluting human urine (Individual 1) by a factor of 20 in double distilled water and serves as a control (point of reference) for the MSIA process. A dilution of urine was required in order to reduce the high salt content of the biological fluid, which would otherwise disrupt the MALDI process. Only small peptides are observed. β-2-Microglobulin (β2m). β2m is a low molecular weight protein (11 729 Da), which was identified as a light chain of the class I major histocompatability antigens.15 Found in most biological fluids, elevated levels of β2m in blood and urine can result from a number of ailments (i.e., AIDS,16 rheumatoid arthritis,17 leukemia18). Glycation of β2m, the covalent attachment of a reducing sugar, is commonly observed in individuals
Comparative Urine Protein Phenotyping
Figure 2. Mass spectrometric results of urinary β2m MSIA. Signal from wild-type β2m is consistent in all five traces, however Trace E also contains a high proportion of glycated β2m. Glycation resulted in a + 146 Da mass shift due to the covalent modification by a deoxyhexose.
with one of a number of metabolic disorders (i.e., diabetes mellitus, uremia, hypoglycemia, etc.) resulting in advanced glycation end products (AGEs).19,20 Moreover, β2m has also been pathogenically associated in many amyloid disorders including dialysis related amyloidosis (DRA).21 Analyses of the human urine samples with anti-β2m MSIA are shown in Figure 2 with wild-type (wt-)β2m seen in the mass spectra of all study participants. Inter-individual results are all very similar, as shown in Figure 2 inlet, except for Trace E in which a glycated β2m variant is present in relatively high abundance. Glycated β2m is the result of excess dietary reducing sugars present in the blood nonenzymatically attaching to free amine groups of local proteins through the “Maillard reaction” that can result in a number of possible Amidori products.22,23 The observed glycated β2m in Figure 2-Trace E has a measured mass increase of ∼146 Da, which corresponds to an imidazolone formation, an Amidori product consisting of a cyclic ring formed between a reduced sugar and an arginine residue. This β2m variant was originally observed by Niwa et al. in amyloid plaque deposits,22 but was then postulated to be artifacts of the overexposure of the insoluble β2m plaques to normal levels of blood reducing sugars. The data presented here is the first report of imidazolone-modified β2m from a human biological fluid and demonstrates that this unique form of glycation can be found on native unplaqued β2m. The individual whose results were shown in Trace E was diagnosed with pancreatic cancer, and clinical studies have shown a strong correlation between the development of the pancreatic cancer and diabetes mellitus.24,25 Hence, the presence of AGEs in this patient might be suggestive of the presence of a metabolic disease, such as diabetes mellitus, but due to a lack of symptoms, the patient was not tested at the time of sample collection. Transthyretin (TTR). TTR, the second target, is a thyroid hormone carrier found in high levels in both serum and cerebral spinal fluid. Produced mainly in the liver, TTR forms a homotetramer26,27 and is often complexed with other proteins in the transport of various biologically active compounds. Structurally, wt-TTR comprises 127 amino acids and has a MW of 13 762. Over 80 point mutations have been cataloged for TTR, with all but 10 potentially leading to severe complica-
research articles
Figure 3. Comparison of urinary TTR MSIA. Both the wild-type and a post-translationally modified (cysteinylation; ∆m ) +128 Da) forms are seen in all five traces. While, varied amounts of cysteinylated TTR are seen present in each sample, Trace B also shows the presence of a variant form of the TTR protein. Peak splitting, ∼∆m ) +30, is observed in the wild-type and the cysteinylated forms of TTR because of two forms of the protein being expressed, a wild-type and mutant, that are eventually excreted into urine.
tions.28 The majority of these mutation-related disorders are caused by amyloid plaques depositing on various tissues, eventually leading to complex dysfunctions; including carpal tunnel syndrome, drussen, and familial amyloid polyneuropathy.29-32 Multiple TTR post-translational modifications have been previously detected in the plasma of healthy individuals,33 due mostly to TTR having a free cysteine residue, which commonly react with other sulfhydrils, cysteines, glutathions, etc. Figure 3 shows the result of the urinary MSIA analyses of TTR. Both wt- and post-translationally modified (PTM) forms of intact TTR are readily apparent in all five traces. This in itself is novel, due to conflicting reports regarding proximal tubular reabsorption and the presence of TTR in urine.34,33 The most abundant PTM observed is the cysteinylated (cys) form, ∆m ) +119 Da. Expanded views of the singly charged TTR signals, shown in each corresponding Figure 3 inlet, clearly show that ratios between the PTM and the wild-type forms of TTR vary between all individuals analyzed. Moreover, the TTR signal shown in Figure 3-Trace B exhibits peak splitting, which is indicative of the presence of a heterozygous point mutation. The resolution of the linear TOF MS (m/∆m ) ∼1000) is sufficient to determine the mass shift of the variant to be ∼+ 30 Da. This approximate mass shift is accurate enough to decrease number of possible variants from 80 to ∼7. The results shown in Figure 3-Trace B would normally warrant further protein characterization, i.e., enzymatic digestion; however, previous MSIA plasma-based studies involving the same individual have already determined the point mutation to be a Thr119Met substitution.11 Cystatin C (CYSC). Cystatin C (CYSC) is an extracellular cysteine protease inhibitor found in most biological fluids. Even though CYSC is freely filtered by the glomerulus, urinary levels of CYSC are a poor marker for glomerular filtration due to Journal of Proteome Research • Vol. 2, No. 2, 2003 193
research articles
Figure 4. Results of urinary CYSC MSIA analysis. Both the wildtype and hydroxylated (∆m ) +16 Da) forms of CYSC are present in all five traces. Varied amounts of hydroxylated CYSC are seen in each individual as well as multiple truncated forms of the protein. These truncations include the systematic N-terminal cleavage of S-, SSP-, and SSPG-. Extensive cleavage of CYSC, with the loss of SSPGKPPR-, SSPGKPPRL-, and SSPGKPPRLV-, are only seen in Traces D and E.
tubular reabsorption,35 but have been used as a reliable measure of proximal tubular reabsorption, which has been linked to renal failure.36 Moreover, hereditary cerebral hemorrhage with amyloidosis (HCHWA), an autosomal dominant disorder prevalent in Icelandic, Dutch, and Finish populations, is the result of a CYSC Leu68fGln variant.37 This variant of CYSC results in amyloid deposits of the walls of cerebral arteries. A number of carcinoma cell lines have also been reported to secrete CYSC, leading to investigations of its role as a possible tumor marker. CYSC also has several PTM associated with it, most notable is the hydroxylation of a Pro residue at position 336 which results in a mass shift of the wtCYSC protein by ∼+16 Da. The results of the anti-CYSC MSIA analyses are shown in Figure 4 in which very similar protein profiles are observed between all subjects. The mass spectrometric analysis was able to sufficiently resolve the wt- (13 344 Da) and the hydroxylated form (13 360 Da) of CYSC, m/∆m ) ∼1000. Varied amounts of hydroxylation are seen between each individual. Multiple N-terminally truncated forms of CYSC are also present, most notable are the S- (13 256 Da) and SSP- (13 072 Da). Hydroxylation still occurs in the S- variant (13 272 Da), but is lost with the cleavage of the P- at position 3. Further truncated forms of CYSC are observed in Traces D and E in which SSPGKPPR- (12 536 Da), SSPGKPPRL- (12 423 Da), and SSPGKPPRLV- (12 324 Da) are also present. The degradation of CYSC has been reported from a significant portion of native urine samples to date,36 but this CYSC profiling clearly shows that this catabolic process is conserved a N-terminal proteolytic process. Urine Protein 1. UP1, also known as Clara cell protein, CC10, or uteroglobin, is a biomarker for a variety of pulmonary ailments and urinary tract dysfunctions. UP1 is a small protein MW ) 7909 that is primarily secreted by Clara cells in the bronchi alveolar lining in mammalian lung tissue is an anti194
Journal of Proteome Research • Vol. 2, No. 2, 2003
Kiernan et al.
Figure 5. Mass spectrometric results of urinary UP1 MSIA. Dimerized UP1 (MW ) 15 819) is seen in all five traces, but Traces A and B both contain large proportions of UP1 monomer conjugated to glutathion (∆m ) +305 Da).
inflammatory agent,38 but to date its physiological role is still largely unclear. The native state of UP1 is a covalently associated homodimer, which results from the disulfide cross bridging between two UP1 monomers.39 When damage occurs to the respiratory tract, plasma and urine UP1 levels increase due to increased bronchioalveolar permeability and the overloading of the tubular reabsorption process, respectively.40,41 Moreover, increased UP1 concentration in urine alone, basal levels 5-10 µg/L,41 is often an indication of proximal tubular dysfunction,42 whereas decreased UP1 plasma levels have been found in smokers,43 asthmatics,44 and schizophrenics.45 Figure 5 shows the results of the qualitative urinary UP1 MSIA analysis. Dimerized UP1 with MW ) 15 819 is present in all five traces. Although multiple charging of UP1 during the MALDI process is unable to directly differentiate between potential UP1 monomers and the +2 state of the UP1 dimer, conjugated monomers are readily identifiable. Because wt-UP1 monomer would have exposed free cysteine groups some sort of chemical modification through these reactive sulfhydryls would be expected, as seen in TTR. Closer examination in the Figure 5 inlets shows that both Traces A and B contain UP1 monomer with varied amounts of glutathion conjugate (∆m ) +305 Da) associated. Although being virtually undetectable in Traces C-E, this is the first reported incidence of UP1 monomer being detected. These results suggest that the individuals’ results shown in Traces A and B have more glutathion and/or UP1 monomer present in their systems. Whether this observation is associated with health state or significant to disease has yet to be determined. Retinol Binding Protein (RBP). RBP was the fifth urine protein target. A member of the lipocalin family, RBP has a plasma concentration level of ∼50 mg/L and serves the role of the major carrier of retinol (vitamin A) from the liver to peripheral tissues.46 With a molecular weight of 21 065, RBP is believed to escape glomerular filtration by associating with the homotetramer of transthyretin in its holo- (retinol bound) form.47 RBP has been previously reported to exist in two posttranslationally truncated versions; one missing the C-terminal Leucine (RBP-(Leu)) and the second missing two C-terminal leucines (RBP-(Leu-Leu)), which are believed to be nonfunc-
Comparative Urine Protein Phenotyping
Figure 6. MSIA results of RBP analysis. Conserved C-terminal cleavage pattern is seen in Traces A-D, with the loss of -L, -LL, -RNLL, and -RSERNLL. Trace E displays an abnormal RBP profile due to the noticeable decrease in the amount of wt-RBP present compared to the truncated forms.
tional variants of RBP due to their lower binding affinities to the transthyretin complex.48 The results of the urinary RBP analysis are shown in Figure 6. Conserved protein profiles are seen in Traces A-D, with wtRBP along with -L, -LL, -RNLL, and -RSERNLL C-terminally truncated variants. The source and function of these variants are still unknown, but have been determined to be the result of some unreported enzymatic process that occurs after the RBP is filtered from the blood.14 Interestingly, the individual with pancreatic cancer in Figure 6-Trace E displays an altered RBP profile. Most notable is the marked relative decrease in the amount of wt-RBP present. Similar results were reported with the analysis of urinary RBP of a 94-year old woman with renal failure stemming from chronic diabetes mellitus, in which wt-RBP was completely absent from the RBP protein expression profile.14 Albumin (ALB). ALB, the sixth urine protein target, is the best studied of all plasma proteins.49 At ∼66.3 kDa, ALB is considerably larger than any of the previously discussed protein targets. As a multipurpose house-keeping protein, ALB serves a multitude of functions including; the binding and transport of many metallic, organic, and biochemical compounds, antioxidant effects as well as plasma buffering.50,51 Figure 7 shows the results of the urinary ALB MSIA with the +1 and the +2 states of ALB are present in all traces. Interindividual results are all very similar, as shown in Figure 7 inlet, except for Trace E in which extensive peak broadening is observed. Since albumin participates in the transport of so many biological, inorganic, and pharmacological compounds, adduct formation with one or many of these compounds is possible. Albumin is also known to undergo glycation, like β2m, hence the observed peak broadening may be the result of the formation of an ALB-AGE’s. Transferrin (TRFE). TRFE, the seventh protein target, is a large globular glycoprotein (MW ) 79.6 kDa) used in the transport of dietary iron in human plasma.52 TRFE readily crosses the glomerular membrane, despite its large size, due to its strong cationic nature53 resulting in urine levels
