Motif Analysis of Phosphosites Discloses a Potential Prominent Role of the Golgi Casein Kinase (GCK) in the Generation of Human Plasma Phospho-Proteome Mauro Salvi,† Luca Cesaro,† Elena Tibaldi,† and Lorenzo A. Pinna*,†,‡ Department of Biological Chemistry, University of Padova, V.le G. Colombo 3, 35131 Padova, Italy, and Venetian Institute for Molecular Medicine, via Orus 2, 35129 Padova, Italy Received January 21, 2010
By comparing the recurrent features of sequences surrounding 86 Ser/Thr residues phosphorylated in peptides from human plasma collected from literature with those generated from the whole human phosphoproteome, and from repertoires of validated substrates of the acidophilic protein kinases CK2 and Golgi casein kinase (GCK), the following conclusions can be drawn: (i) the contribution of Prodirected and basophilic kinases to the plasma phosphoproteome is negligible, if any, while the contribution of acidophilic kinases is by far predominant; (ii) the plasma weblogo profile is closely reminiscent of that generated by GCK in its substrates, while it neatly differentiates from that generated by CK2; (iii) 58 plasma phosphosites out of 86 display the canonical consensus for GCK (S/T-x-E/pS), while that for CK2 (S/T-x-x-E/D/pS) is found in 15 peptides, all of which also conform to the GCK signature. These observations, in conjunction with a very similar situation disclosed by analyzing the phosphopeptides of the human cerebrospinal fluid collected from literature, support the view that GCK may play a major role in the phosphorylation of proteins secreted into body fluids. Nearly all aspects of life are under the control of protein phosphorylation, affecting a substantial proportion of individual cell proteomes (30% or more),1 and catalyzed by a very large family of enzymes, protein kinases, numbering over 500 in human, collectively referred to as the human “kinome”.2 Now-a-days, protein kinases are held as signaling molecules par excellence and reversible protein phosphorylation is looked at as a fundamental mechanism of integrated regulation within and among cells. However, the first phosphoprotein to be discovered, already in the 19th century, was casein,3 a component of milk secreted by the lactating mammary gland and devoid of any apparent regulatory relevance. In more recent times, casein was frequently used as an artificial substrate for monitoring in vitro the activity of diverse protein kinases. Some of these were consequently denoted by the misnomer “Casein kinases” although it was clearly shown that they have no physiological relatedness with casein (for an historical overview, see ref 4). Somewhat paradoxically in fact, in the present post genomic era, the gene(s) encoding for the protein kinase responsible for the biosynthetic phosphorylation of casein within the lactating mammary gland (i.e., the genuine, bona fide casein kinase(s)) are still unknown and their activity has to be ascribed to the category of “orphan” enzymes.5 Despite this limit, bona fide casein kinase activity, primarily located in the Golgi apparatus of the lactating mammary gland and thereafter * To whom correspondence should be addressed. Prof. L.A. Pinna, Department of Biological Chemistry V.le G. Colombo 3, 35131, Padova, Italy. Phone: +39 0498276108. Fax: +39 0498073310. E-mail:
[email protected]. † University of Padova. ‡ Venetian Institute for Molecular Medicine. 10.1021/pr100058r
2010 American Chemical Society
referred to as “Golgi enriched fraction casein kinase” (GEFCK),6 or more shortly Golgi casein kinase (GCK),7 underwent a thorough biochemical characterization. By inspecting the sequences of casein fractions, it was first predicted that the enzyme committed to their phosphorylation recognizes triplets where serine is two positions upstream from either Glutamic acid or another phospho-serine (S-x-E/pS).8 This keen inference was later confirmed by in vitro studies with synthetic peptide substrates.9,10 Intriguingly, this motif was also found in early studies in other phosphoproteins that are not secreted by the lactating mammary gland, such as fibrinogen,11 pepsin,12 ovalbumin,13 and ACTH,14 and was shown to be similar, but definitely distinct, to the consensus sequence of CK2 (S/T-xx-E/D/pS), a ubiquitous and highly pleiotropic protein kinase termed in early times “casein kinase-2 (or -II)”.9,10 This latter observation made possible the generation of peptide substrates which neatly discriminate among “casein kinases” in general, and between CK2 and GCK in particular, ending up with a peptide reproducing one of the sites phosphorylated in β-casein (KKIEKFQSEEQQQ). This allows the detection and quantitative evaluation of GCK with absolute specificity.15 Phosphorylation of this peptide (β[28-40]), and unique insensitivity to extremely high (>100 µM) concentrations of staurosporine, a broad specificity and very potent inhibitor of other protein kinases, are held as hallmarks of GCK activity. These two features have been exploited to show that GCK is not a dedicated enzyme just committed to casein phosphorylation in the lactating mammary gland but a ubiquitous enzyme detectable in the Golgi apparatus of many other tissues as well7 and responsible for the phosphorylation of a wide variety of protein substrates, often implicated in crucial biological functions, such as several Journal of Proteome Research 2010, 9, 3335–3338 3335 Published on Web 05/07/2010
letters
Salvi et al.
Figure 1. Weblogo analysis. (A) Weblogo of 54 nonredundant GCK-phosphosites collected from literature and reported in Table S1 (Supporting Information). (B) Web logo of CK2-validated phosphosites. Reprinted from ref 18. (C) Weblogo of the entire human S/T phosphoproteome based on 9526 nonredundant S/T phosphosites from Phospho-Elm (http://phospho.elm.eu.org).28 (D) Weblogo of human plasma S/T phosphoproteome; 86 nonredundant phosphosites have been collected from ref 19 (serum phosphoproteome) and ref20 (plasma phosphoproteome). (E) Weblogo of CamKIIR-validated phosphosites; 137 nonredundant phosphosites have been collected from phosphosite (http://www.phosphosite.org/) and phosphoElm database (http://phospho.elm.eu.org). (F) Weblogo of 28 plasma phosphosites which do not display the canonical consensus of GCK (S/T-x-E/pS) extracted from the human plasma S/T phosphoproteome; because of the limited sequence data this weblogo representation suffers from an underestimation of the entropy that is partially ameliorated by the incorporation of small sample correction.17 (G) Weblogo of human cerebrospinal fluid S/T phosphoproteome; 92 nonredundant phosphosites have been collected from ref 23. Weblogo were generated using WebLogo 2.8.2 (http://weblogo.berkeley.edu/logo.cgi). In Weblogo, each column of the alignment is represented by a stack of letters (denoting amino acids) where the height of each letter is proportional to the observed frequency of the corresponding amino acid. The total height of each stack is proportional to the sequence conservation at that position and is expressed in bits. The maximal height of a stack is 4.32 bits that is equivalent to the occurrence of only one amino acid on a given position.17
chaperonines, osteopontin, PRP-1, aquaporin-2, p115.16 In recent years, an increasingly long repertoire of GCK substrates has become available, sharing the typical consensus of this kinase (S-x-E/pS). For many of these, the capability of being phosphorylated in vitro by GCK purified from lactating mammary gland through a reaction which is unaffected by 100 µM staurosporine, while being out-competed by the specific (β[28-40]) peptide substrate, was demonstrated. The sequences of 54 GCK putative phosphosites are on display in Table 1 of the Supporting Information. This database was exploited to construct a weblogo, that is, a graphical representation that provides a more precise and integrated description of sequence similarity than mere consensus sequence, and can rapidly disclose important features otherwise difficult to perceive.17 The weblogo extracted from GCK attributed phosphosites is shown in Figure 1A, where it is compared to that previously 3336
Journal of Proteome Research • Vol. 9, No. 6, 2010
calculated for phosphosites generated by CK218 (Figure 1B). It can be seen that the two logos neatly differentiate for the positions where acidic residues are selected by the two kinases: the only position where an outstanding selection of gluamic acid (but not of aspartic acid) is evident with GCK is n + 2, whereas CK2 displays a marked positive selection of either Glu or Asp at all downstream positions with a remarkable preference for positions n + 3 and n + 1. With GCK moreover selection of seryl residues partially displaying the SxS motif is more pronounced than in the case of CK2. As expected, both the GCK and CK2 weblogo lack the stigmata of proline directed and basophilic protein kinases which are instead very evident in the weblogo extracted from the whole phosphoproteome (Figure 1C): these mainly consist in the prevalence of Pro at position n + 1 and or Arg at several positions upstream, with special reference to n - 2, n - 3 and n - 5. Also to note however is the remarkable contribution of
GCK in the Generation of Human Plasma Phospho-Proteome CK2 to the whole phosphoproteome weblogo, revealed by frequency of acidic residues, especially at positions n + 3 and n + 1. In contrast no substantial contribution of GCK to the whole phosphoproteome can be inferred from Figure 1C, with the caveat that it could be masked by the overwhelming contribution of CK2. This scenario is strikingly subverted if the weblogo of a plasma phosphoproteome, composed of nonredundant serum phosphoproteome phosphosites (collected from ref 19) and plasma phosphoproteome phosphosites (collected from ref 20) is considered instead of that of the whole phosphoproteome. The former one, based on a repertoire of 86 phosphopeptides derived from 44 phospho-proteins, is shown in Figure 1D. It clearly reveals no appreciable contributions from Pro-directed and basophilic kinases, while being closely reminiscent of that of GCK phosphosites rather than displaying the signature of CK2. Indeed an overrepresentation of the acidophilic SXE motifs in the plasma phosphoproteome was already observed in ref 20. Although in fact acidic residues are significantly selected at various positions other than n + 2, their prominence is negligible as compared to that of Glu at this position. Also to note is that Glu is by far preferred over Asp, as it happens in the GCK weblogo, whereas in the CK2 weblogo the two residues are selected to nearly the same extent (Figure 1, compare panel D with A and B). On the other hand, any significant contribution by CaMKIIR, another kinase whose weblogo displays an acidic stack at position n + 2, has to be ruled out since the most remarkable hallmark of this kinase, an Arg highly selected at position n - 3 (Figure 1E, showing the weblogo of CaMKIIR-substrates), is entirely lacking in the plasma phosphoproteome weblogo (Figure 1D). To make our analysis more stringent we have concentrated on those plasma phospho-sites which conform to the GCK consensus, pS/T-x-E/pS. These are 58 altogether, out of 86 phosphopeptides (67%), and they are listed in Table S2 of Supporting Information. Forty-three of these display only the GCK consensus but not that of CK2 (S/T-x-x-E/D), while the remaining 15 fulfill both requirements for being considered potential substrates of either GCK and CK2. We have also generated the weblogo of the 28 plasma phosphosites which do not display the canonical consensus of GCK (S-x-E/pS). As shown in panel F of Figure 1, the only notable feature of this weblogo is a moderate general selection of acidic residues, either upstream or downstream from the phosphoresidue. While selection of acidic residues at positions n-3 and n-4 may reflect the signature of CK1 in its non primed substrates, (E/D)n-X-X-S/T,21 the somewhat remote downstream acidic residues are intriguingly reminiscent of a non canonical consensus sequence recognized by GCK in the salivary PRP-1 protein, SEQFIDEE.22 This may suggest that also a substantial proportion of plasma phosphosites lacking the canonical GCK signature (S-x-E/pS) could nevertheless be generated by this kinase. Pertinent to this may be the observation that the great majority of the plasma phosphopeptides belong to secreted proteins whose trafficking trough the Golgi apparatus makes conceivable the implication of GCK in theirs phosphorylation prior to secretion into the extracellular lumen. Consistent with this, the weblogo generated from a repertoire of phosphopeptides recently identified in the human cerebrospinal fluid23 (Figure 1G) is strikingly similar to that of the plasma phosphopeptides. Whether phosphorylation by GCK is a general reaction inherent to the secretory machinery or it reflects specific
letters
regulatory devices remains an open question. The latter hypothesis, however, would be in accordance with a number of observations: for example, bone morphogenic protein 15 (BMP-15) and growth differentiation factor 9 (GDF-9) display opposite signaling properties depending on whether they are phosphorylated by GCK or not prior to secretion.24,25 In a similar vein, it has been recently noted that some neurological diseases correlate with defective phosphorylation of putative GCK substrates rather than to their impaired secretion.26,27 Increasingly growing evidence that GCK plays a crucial and pleiotropic role in the regulation of a large number of secreted proteins underscores the compelling need of getting at last genetic and structural information about this elusive orphan protein kinase.
Acknowledgment. This work was supported by University of Padova (Progetto Ateneo 2008 to M.S.) and by European Commission (PRO-KINASERESEARCH 503467), the Italian Cystic Fibrosis Research Foundation (Grant FFC#4/2009) with the contribution of “Banca Popolare di Verona e Novara” and “Fondazione Giorgio Zanotto”, and by AIRC to L.A.P. Supporting Information Available: Supplementary Tables 1 and 2 and references. This material is available free of charge via the Internet at http://pubs.acs.org. References (1) Ubersax, J. A.; Ferrell, J. E., Jr. Mechanisms of specificity in protein phosphorylation. Nat. Rev. Mol. Cell. Biol. 2007, 8 (7), 530–541. (2) Manning, G.; Whyte, D. B.; Martinez, R.; Hunter, T.; Sudarsanam, S. The protein kinase complement of the human genome. Science 2002, 298 (5600), 1912–1934. (3) Hammarsten, O. Zur Frage ob Caseı´n ein einheitlicher Stoff sei. Hoppe-Seyler’s Z. Physiol. Chem. 1883, 227–273. (4) Pinna, L. A. A historical view of protein kinase CK2. Cell. Mol. Biol. Res. 1994, 40 (5-6), 383–390. (5) Lespinet, O.; Labedan, B. Puzzling over orphan enzymes. Cell. Mol. Life Sci. 2006, 63 (5), 517–523. (6) Moore, A.; Boulton, A. P.; Heid, H. W.; Jarasch, E. D.; Craig, R. K. Purification and tissue-specific expression of casein kinase from the lactating guinea-pig mammary gland. Eur. J. Biochem. 1985, 152 (3), 729–737. (7) Lasa, M.; Marin, O.; Pinna, L. A. Rat liver Golgi apparatus contains a protein kinase similar to the casein kinase of lactating mammary gland. Eur. J. Biochem. 1997, 243 (3), 719–725. (8) Mercier, J. C. Phosphorylation of caseins. Present evidence for an amino acid triplet code post-translationally recognized by specific kinases. Biochimie 1981, 63, 1–17. (9) Meggio, F.; Boulton, A. P.; Marchiori, F.; Borin, G.; Lennon, D. P.; Calderan, A.; Pinna, L. A. Substrate-specificity determinants for a membrane-bound casein kinase of lactating mammary gland. A study with synthetic peptides. Eur. J. Biochem. 1988, 177 (2), 281– 284. (10) Meggio, F.; Perich, J. W.; Meyer, H. E.; Hoffmann-Posorske, E.; Lennon, D. P.; Johns, R. B.; Pinna, L. A. Synthetic fragments of beta-casein as model substrates for liver and mammary gland casein kinases. Eur. J. Biochem. 1989, 186 (3), 459–464. (11) Blomba¨ck, B.; Blomba¨ck, M.; Edman, P.; Hessel, B. Amino-Acid Sequence and the Occurrence of Phosphorus in Human Fibrinopeptides. Nature 1962, 193, 883–884. (12) Tang, J.; Sepulveda, P.; Marciniszyn, J. Jr.; Chen, K. C.; Huang, W. Y.; Tao, N.; Liu, D.; Lanier, J. P. Amino-acid sequence of porcine pepsin. Proc. Natl. Acad. Sci. U.S.A. 1973, 70 (12), 3437–3439. (13) Henderson, J. Y.; Moir, A. J.; Fothergill, L. A.; Fothergill, J. E. Sequences of sixteen phosphoserine peptides from ovalbumins of eight species. Eur. J. Biochem. 1981, 114 (2), 439–450. (14) Browne, C. A.; Bennett, H. P. J.; Solomon, S. Isolation and characterization of corticotropin- and melanotropin-related peptides from the neurointermediary lobe of the rat pituitary by reversed-phase liquid chromatography. Biochemistry 1981, 20 (16), 4538–4546. (15) Lasa-Benito, M.; Marin, O.; Meggio, F.; Pinna, L. A. Golgi apparatus mammary gland casein kinase: monitoring by a specific peptide
Journal of Proteome Research • Vol. 9, No. 6, 2010 3337
letters (16)
(17) (18)
(19) (20)
(21) (22)
3338
substrate and definition of specificity determinants. FEBS Lett. 1996, 382 (1), 149–152. Tibaldi, E.; Arrigoni, G.; Brunati, A. M.; James, P.; Pinna, L. A. Analysis of a sub-proteome which co-purifies with and is phosphorylated by the Golgi casein kinase. Cell. Mol. Life Sci. 2006, 63 (3), 378–389. Crooks, G. E.; Hon, G.; Chandonia, J. M.; Brenner, S. E. WebLogo: a sequence logo generator. Genome Res. 2004, 14 (6), 1188–1190. Salvi, M.; Sarno, S.; Cesaro, L.; Nakamura, H.; Pinna, L. A. Extraordinary pleiotropy of protein kinase CK2 revealed by weblogo phosphoproteome analysis. Biochim. Biophys. Acta 2009, 1793 (5), 847–859. Zhou, W.; Ross, M. M.; Tessitore, A.; Ornstein, D.; Vanmeter, A.; Liotta, L. A.; Petricoin, E. F., 3rd. An initial characterization of the serum phosphoproteome. J. Proteome Res. 2009, 8 (12), 5523–5531. Carrascal, M.; Gay, M.; Ovelleiro, D.; Casas, V.; Gelpiı´, E.; Abian, J. Characterization of the Human Plasma Phosphoproteome Using Linear Ion Trap Mass Spectrometry and Multiple Search Engines. J. Proteome Res. 2009, 9 (2), 876–884. Pinna, L. A.; Ruzzene, M. How do protein kinases recognize their substrates. Biochim. Biophys. Acta 1996, 1314 (3), 191–225. Brunati, A. M.; Marin, O.; Bisinella, A.; Salviati, A.; Pinna, L. A. Novel consensus sequence for the Golgi apparatus casein kinase, revealed using proline-rich protein-1 (PRP1)-derived peptide substrates. Biochem. J. 2000, 351 (3), 765–768.
Journal of Proteome Research • Vol. 9, No. 6, 2010
Salvi et al. (23) Bahl, J. M.; Jensen, S. S.; Larsen, M. R.; Heegaard, N. H. Characterization of the human cerebrospinal fluid phosphoproteome by titanium dioxide affinity chromatography and mass spectrometry. Anal. Chem. 2008, 80 (16), 6308–6316. (24) McMahon, H. E.; Sharma, S.; Shimasaki, S. Phosphorylation of bone morphogenetic protein-15 and growth and differentiation factor-9 plays a critical role in determining agonistic or antagonistic functions. Endocrinology 2008, 149 (2), 812–817. (25) Tibaldi, E.; Arrigoni, G.; Martinez, H. M.; Inagaki, K.; Shimasaki, S.; Pinna, L. A. Golgi apparatus casein kinase phosphorylates bioactive Ser-6 of bone morphogenetic protein 15 and growth and differentiation factor 9. FEBS Lett. 2010, 584 (4), 801–805. (26) Castagnola, M.; Messana, I.; Inzitari, R.; Fanali, C.; Cabras, T.; Morelli, A.; Pecoraro, A. M.; Neri, G.; Torrioli, M. G.; Gurrieri, F. Hypo-phosphorylation of salivary peptidome as a clue to the molecular pathogenesis of autism spectrum disorders. J. Proteome Res. 2008, 7 (12), 5327–5332. (27) Miller, G. Neuroscience. Getting a read on Rett syndrome. Science 2006, 314 (5805), 1536–1537. (28) Diella, F.; Gould, C. M.; Chica, C.; Via, A.; Gibson, T. J. Phospho. ELM: a database of phosphorylation sites--update 2008. Nucleic Acids Res. 2008, 36 (Database issue), D240–244.
PR100058R
