Identification of a Metabolizing Enzyme in Human Kidney by Proteomic Correlation Profiling

Molecular identification of endogenous enzymes and biologically active substances from complex biological sources remains a challenging task, and although traditional biochemical purification is sometimes regarded as outdated, it remains one of the most powerful methodologies for this purpose. While biochemical purification usually requires large amounts of starting material and many separation steps, we developed an advanced method named “proteomic correlation profiling” in our previous study. In proteomic correlation profiling, we first fractionated biological material by column chromatography, and then calculated each protein''s correlation coefficient between the enzyme activity profile and protein abundance profile determined by proteomics technology toward fractions. Thereafter, we could choose possible candidates for the enzyme among proteins with a high correlation value by domain predictions using informatics tools. Ultimately, this streamlined procedure requires fewer purification steps and reduces starting materials dramatically due to low required purity compared with conventional approaches. To demonstrate the generality of this approach, we have now applied an improved workflow of proteomic correlation profiling to a drug metabolizing enzyme and successfully identified alkaline phosphatase, tissue-nonspecific isozyme (ALPL) as a phosphatase of CS-0777 phosphate (CS-0777-P), a selective sphingosine 1-phosphate receptor 1 modulator with potential benefits in the treatment of autoimmune diseases including multiple sclerosis, from human kidney extract. We identified ALPL as a candidate protein only by the 200-fold purification and only from 1 g of human kidney. The identification of ALPL as CS-0777-P phosphatase was strongly supported by a recombinant protein, and contribution of the enzyme in human kidney extract was validated by immunodepletion and a specific inhibitor. This approach can be applied to any kind of enzyme class and biologically active substance; therefore, we believe that we have provided a fast and practical option by combination of traditional biochemistry and state-of-the-art proteomic technology.Molecular identification for an enzyme reaction or biologically active substance in an organism is challenging, although molecular biological methodologies such as expression cloning (1), recombinant protein panel (2) and RNAi screening (3) have been introduced recently as alternative approaches. Conventional biochemical purification has provided a number of successes and thus still remains a powerful, though labor-intensive strategy.In the traditional protein purification, it had been necessary to purify an individual protein nearly to homogeneity at a microgram amount so that the purified protein could be analyzed by N-terminal amino acid sequencing. Protein identification by mass spectrometry subsequently revolutionized this technology by enabling identification of proteins at much lower abundances: individual proteins could then be associated with specific activities as soon as a band in SDS-PAGE could be observed, even when the purified protein was far from homogeneity (4–6). Although this streamlined the workflow by reducing the required starting materials as well as the separation steps for protein purification, a faster and more generalized approach from smaller starting material has still been desired because some proteins are physiochemically difficult for example in solubilization and stability. To solve these problems, we devised a proteomic correlation profiling methodology (7).The basic concept of proteomic correlation profiling was originally developed by Andersen et al. (8). They quantitatively profiled hundreds of proteins across several centrifugation fractions by mass spectrometry and identified centrosomal proteins by calculating the correlation of these protein expression profiles with already known centrosomal proteins. In the following study, Foster et al. applied this strategy to map more than 1400 proteins to ten subcellular locations (9). Although these studies used centrifugation as a separation method and a known marker profile as a standard for correlation, we extended this concept to use chromatography as a separation method and kinase activity as a basis for comparison; our approach successfully identified a kinase responsible for phosphorylation of peptide substrates just after one step chromatography, and was termed proteomic correlation profiling (7). Independently, Kuromitsu et al. reported identification of an active substance in the serum response element-dependent luciferase assay from interstitial cystitis urine after three-step chromatography by a similar concept (10). In theory, this general proteomic correlation profiling strategy can be adapted to any kind of separation method and activity profile but no other example has been reported thus far, therefore, actual examples where the method can be applied to other enzyme classes are required to prove its generality.Multiple sclerosis is the most common autoimmune disorder of the central nerve system in which the fatty myelin sheaths around the axons of the brain and spinal cord are damaged, leading to demyelination and scarring (11, 12). Until recently, the standard treatments for multiple sclerosis such as interferon beta, glatiramer acetate, mitoxantrone, and natalizumab would often cause severe adverse events (13, 14), providing an opportunity for development of less dangerous treatments for this disease. However, in 2010, Food and Drug Administration approved fingolimod (Gilenya; chemical structure in Fig. 1) as the first oral medicine, and recommended this as a first-line treatment for relapsing-remitting multiple sclerosis, opening up a new therapeutic approach to the disease (15).Open in a separate windowFig. 1.The chemical structures of CS-0777, fingolimod and their phosphorylated derivatives.Sphingosine 1-phosphate receptor 1 (S1P1)¹ modulators are emerging as a new class of drugs with potential therapeutic application in multiple sclerosis (15), and fingolimod is a nonselective sphingosine 1-phosphate (S1P) receptor modulator (16–18, 21, 22). Given its structural similarity to sphingosine, fingolimod is phosphorylated in vivo by sphingosine kinase, in particular sphingosine kinase 2 (SPHK2) (19, 20), and the fingolimod-phosphate (fingolimod-P, Fig. 1) binds to and activates four G protein-coupled S1P receptors (21, 22). By this mechanism, fingolimod-P induces internalization of S1P1 on lymphocytes, blocking the ability of the receptor to support lymphocyte egress and recirculation through secondary lymphoid organs. This suppresses immune responses and is presumably the main immunomodulatory mode of action of fingolimod.CS-0777 (Fig. 1) is a novel selective S1P1 modulator (23). Although the immunomodulatory effects are supposed to be mainly mediated by S1P1, some lines of evidence suggest that the agonist activity on S1P receptor 3 (S1P3) could cause acute toxicity and cardiovascular deregulation, including bradycardia in rodents (24, 25). Thus, CS-0777 was designed to have more selectivity on S1P1 over S1P3 in contrast to fingolimod-P which has potent agonistic activity for S1P3, S1P4, and S1P5 in vitro (22). Like fingolimod, CS-0777 is also a prodrug phosphorylated in vivo, and the phosphorylated CS-0777 (CS-0777-P, Fig. 1) agonizes S1P1 with more than 300-fold selectivity relative to S1P3 whereas CS-0777-P has weaker effects on S1P5 and no activity on S1P2 (23). CS-0777 showed immunosuppressive activity in mouse and rat models of experimental autoimmune encephalitis, animal models for multiple sclerosis. In healthy volunteers, single oral doses of CS-0777 caused marked, dose-dependent decreases in numbers of circulating lymphocytes, including marked and reversible decreases in circulating T and B cells (26). Furthermore, in multiple sclerosis patients, single oral doses of CS-0777 caused dose-dependent decreases in circulating lymphocytes, with a slightly greater suppression of CD4+ versus CD8+ T cells. Therefore, CS-0777 would alter immune responses solely through activation of S1P1 without S1P3 modulation in humans, which could circumvent a bradycardia adverse effect, although the relationships associating selectivity of S1P1 to S1P3 with bradycardia in humans are not fully understood (12).Orally administrated CS-0777 is phosphorylated and rapidly reaches equilibrium with CS-0777-P as in the case of fingolimod (22), suggesting that the high kinase activity in blood is balanced by phosphatases. Therefore, identification of a phosphatase, the inactivating enzyme of an active metabolite, as well as identification of a kinase, the activating enzyme of a prodrug, are critical to fully understand the mechanism of action at the molecular level for both CS-0777 and fingolimod. Sphingosine kinase 2 (SPHK2) was identified as the major kinase of fingolimod (21, 28, 29) and lipid phosphate phosphatase 3 (LPP3) was reported to be a phosphatase for fingolimod-P dephosphorylation (30), although contribution of LPP3 in vivo has not been fully studied. In our previous work, we have identified CS-0777 kinases in human blood as fructosamine 3-kinase-related protein (FN3K-RP) and fructosamine 3-kinase (FN3K) (6), whereas the phosphatase of CS-0777-P had not been identified thus far.In this study, we have successfully identified alkaline phosphatase, tissue-nonspecific isozyme (ALPL) as the major CS-0777-P phosphatase candidate in the human kidney by proteomic correlation profiling. According to available information, this is the first report applying proteomic correlation profiling to enzyme classes other than kinases; similarly, we believe this to be first application of proteomic correlation profiling to human tissue extract, which therefore has opened up wide usage of proteomic correlation profiling for all types of enzyme identification.