Characterization of a Novel Type of Serine/Threonine Kinase That Specifically Phosphorylates the Human Goodpasture Antigen*

Goodpasture disease is an autoimmune disorder that occurs naturally only in humans. Also exclusive to humans is the phosphorylation process that targets the unique N-terminal region of the Goodpasture antigen. Here we report the molecular cloning of GPBP (Goodpasture antigen-bindingprotein), a previously unknown 624-residue polypeptide. Although the predicted sequence does not meet the conventional structural requirements for a protein kinase, its recombinant counterpart specifically binds to and phosphorylates the exclusive N-terminal region of the human Goodpasture antigen in vitro. This novel kinase is widely expressed in human tissues but shows preferential expression in the histological structures that are targets of common autoimmune responses. The work presented in this report highlights a novel gene to be explored in human autoimmunity.


Isolation and Characterization of cDNA Clones
Encoding Human GPBP Several human -gt11 cDNA expression libraries (eye, fetal and adult lung, kidney, and HeLa S3, from CLONTECH) were probed for cDNAs encoding proteins interacting with GPpep1. Nitrocellulose filters (Millipore) prepared following standard immunoscreening procedures were blocked and incubated with 1-10 nmol per ml of GPpep1 at 37°C. Specifically bound GPpep1 was detected using M3/1A monoclonal antibodies (7). A single clone was identified in the HeLa-derived library (HeLa1). Specificity of fusion protein binding was confirmed by similarly binding recombinant eukaryotic human GP antigen. The EcoRI cDNA insert of HeLa1 (0.5 kb) was used to screen further the same library and to isolate overlapping cDNAs. The largest cDNA (2.4 kb) containing the entire cDNA of HeLa1 (n4Ј) was fully sequenced.

Northern and Southern Blots
Pre-made Northern and Southern blots (CLONTECH) were probed with HeLa1 cDNA following the manufacturers' instructions.

Plasmid Construction, Expression, and Purification of Recombinant Proteins
GPBP-derived Material-The original -gt11 HeLa1 clone was expressed as a lysogen in Escherichia coli Y1089 (8). The corresponding ␤-galactosidase-derived fusion protein containing the N-terminal 150 residues of GPBP was purified from the cell lysate using an APTGagarose column (Roche Molecular Biochemicals). The EcoRI 2.4-kb fragment of n4Ј was subcloned in Bluescribe M13ϩ vector (Stratagene) * This work was supported in part by Grants SAL91/0513, SAF94/ 1051, and SAF97/0065 from the Plan Nacional IϩD of the Comisión Interministerial de Ciencia Tecnología, Grant 93/0343 from Fondo de Investigaciones Sanitarias, and Grants GV-3166/95 and GV-C-VS-21-118-96 from la Direcció General d'Ensenyaments Universitaris i Investigació (Comunitat Valenciana, Spain) (to J. S.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF136450.
§ Supported by a postdoctoral fellowship sponsored by Fundación Valenciana de Investigaciones Biomédicas, by Exploraciones Radiológicas Especiales S.A., and by Bancaixa.
¶ Recipient of a fellowship from the Consellería de Educació i Ciència de la Comunitat Valenciana.
(BS-n4Ј), amplified, and used for subsequent cloning . A DNA fragment  containing from 5Ј to 3Ј, an EcoRI restriction site, a standard Kozak  consensus for translation initiation, a region coding for a tag peptide  sequence (FLAG, DYKDDDDK), and the sequence coding for the first  11 residues of GPBP including the predicted Met i and a BanII restriction site, was obtained by hybridizing ON-GPBP-54m and ON-GPBP-55c and extending with modified T  or Triton X-100, 5 mM EDTA, and 1 mM phenylmethylsulfonyl fluoride in TBS) with or without 0.1% SDS depending whether the lysate was to be used for SDS-PAGE or FLAG purification, respectively. For FLAG purification, the lysate of four to six 175-cm 2 culture dishes was diluted up to 50 ml with lysis buffer and purified as above. For stable expression, the cells were similarly transfected with pc-n4Ј and selected for 3 weeks with 800 g/ml G418. For bacterial recombinant expression, the 2.0-kb EcoRI cDNA fragment of pHIL-FLAG-n4Ј was cloned in-frame downstream of the glutathione S-transferase (GST)-encoding cDNA of pGEX-5x-1 (Amersham Pharmacia Biotech). The resulting construct was used to express GST-GPBP fusion protein in DH5␣ (9). GP Antigen-derived Material-Human recombinant GP antigen (rGP) was produced in 293 cells using the pRc/CMV-BM40 expression vector containing the ␣3-specific cDNA between ON-HNC-B-N-14m and ON-HNC-B-N-16c. The expression vector is a pRc/CMV (Invitrogen)derived vector provided by Billy G. Hudson that contains cDNA encoding an initiation Met, a BM40 signal peptide followed by a tag peptide sequence (FLAG), and a polylinker cloning site. To obtain ␣3-specific cDNA, a polymerase chain reaction was performed using the oligonucleotides above and a plasmid containing the previously reported ␣3(IV) cDNA sequence (3) as template (clone C2). For stable expression of rGP, 293 cells were transfected with the resulting construct (f␣3VLC) and selected with 400 g/ml G418. The harvested rGP was purified using an anti-FLAG M2 column.
All the constructs were verified by restriction mapping and nucleotide sequencing.

Cell Culture and DNA Transfection
Human 293 cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. Transfections were performed using the calcium phosphate precipitation method of the Profection Mammalian Transfection Systems (Promega). Stably transfected cells were selected by their resistance to G418. Foci of surviving cells were isolated, cloned, and amplified.

Antibody Production
Polyclonal Antibodies Against the N-terminal Region of GPBP-Cells expressing HeLa1 -gt11 as a lysogen were lysed by sonication in the presence of Laemmli sample buffer and subjected to electrophoresis in a 7.5% acrylamide preparative gel. The gel was stained with Coomassie Blue, and the band containing the fusion protein of interest was excised and used for rabbit immunization (10). The antiserum was tested for reactivity using APTG affinity purified antigen. To obtain affinity purified antibodies, the antiserum was diluted 1:5 with TBS and loaded onto a Sepharose 4B column containing covalently bound affinity purified antigen. The bound material was eluted and, unless otherwise indicated, used in the immunochemical studies.
Monoclonal Antibodies against GPBP-Monoclonal antibodies were produced essentially as previously reported (7) using GST-GPBP. The supernatants of individual clones were analyzed for antibodies against rGPBP.

In Vitro Phosphorylation Assays
About 200 ng of rGPBP were incubated overnight at 30°C in 25 mM ␤-glycerol phosphate (pH 7.0), 0.5 mM EDTA, 0.5 mM EGTA, 8 mM MgCl 2 , 5 mM MnCl 2 , 1 mM dithiothreitol, and 0.132 M [␥-32 P]ATP, in the presence or absence of 0.5-1 g of protein substrates or 10 nmol of synthetic peptides, in a total volume of 50 l.

In Vivo Phosphorylation Assays
Individual wells of a 24-well dish were seeded with normal or with stably pc-n4Ј-transfected 293 cells. When the cells were grown to the desired density, a number of wells of the normal 293 cells were transfected with pc-FLAG-n4Ј. After 12 h the culture medium was removed, 20 Ci/well of H 3 32 PO 4 in 100 l of phosphate-free Dulbecco's modified Eagle's medium added, and incubation continued for 4 h. The cells were lysed with 300 l/well of TBS containing 1% Triton X-100, 2 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 50 mM NaF, and 0.2 mM vanadate and extracted with specific antibodies and protein A-Sepharose. When anti-GPBP serum was used, the lysate was precleared using preimmune serum and protein A-Sepharose.

Nucleotide Sequence Analysis
The cDNA sequences were performed by the dideoxy chain termination method using ␣-35 S-dATP, modified T 7 DNA polymerase (Amersham Pharmacia Biotech), and universal or GPBP-specific primers (8 -10).

P-Phosphoamino Acid Analysis
Immunopurified rGPBP or high performance liquid chromatography gel filtration fractions therefrom containing the material of interest were phosphorylated, hydrolyzed, and analyzed in one-(4) or twodimensional thin layer chromatography (12). When performing twodimensional analysis, the buffer for the first dimension was formic acid:acetic acid:water (1:3.1:35.9) (pH 1.9), and the buffer for the second dimension was acetic acid:pyridine:water (2:0.2:37.8) (pH 3.5). Amino acids were revealed with ninhydrin and 32 P-phosphoamino acids by autoradiography.

Physical Methods and Immunochemical Techniques
SDS-PAGE and Western blotting were performed as in Ref. 4. Immunohistochemistry studies were done on human multi-tissue control slides (Biomeda, Biogenex) using the ABC peroxidase method (13).

Computer Analysis
Homology searches were carried out against the GenBank TM and SwissProt data bases with the BLAST 2.0 (14) at the NCBI server and against the TIGR Human Gene Index data base for expressed sequence tags using the Institute for Genomic Research server. The search for functional patterns and profiles was performed against the PROSITE data base using the ProfileScan program at the Swiss Institute of Bioinformatics (15). Prediction of coiled-coil structures was done at the Swiss Institute for Experimental Cancer Research using the program Coils (16) with both 21-and 28-residue windows.

Molecular
Cloning of GPBP-To search for proteins specifically interacting with the divergent N-terminal region of the human GP antigen, a 21-residue peptide (GPpep1), encompassing this and flanking sequence, and specific monoclonal antibodies against it were combined to screen several human cDNA expression libraries. The screening of more than 5 ϫ 10 6 phages was required to identify a single HeLa-derived recombinant encoding a fusion protein specifically interacting with GPpep1 without disturbing antibody binding.
By using the cDNA insert of the original clone (HeLa1), we isolated a 2.4-kb cDNA (n4Ј) that contains 408 bp of 5Ј-untranslated sequence, an open reading frame (ORF) of 1872-bp encoding 624 residues, and 109-bp of 3Ј-untranslated sequence (Fig. 1). Other structural features are of interest. First, the predicted polypeptide has a large number of phosphorylatable (17.9%) and acidic (16%) residues with an unequal distribution along the sequence. Serine, which is the most abundant residue (9.3%), shows preference for two short regions of the protein where it comprises nearly 40% of the amino acids compared with an average of less than 7% throughout the rest of the polypeptide chain. It is also noteworthy that the more N-terminal serine-rich region consists mainly of a Ser-Xaa-Yaa repeat. Acidic residues are preferentially located at the three N-terminal quarters of the polypeptide with nearly 18% of the residues being acidic. These residues represent only 9% in the most C-terminal quarter of the polypeptide, resulting in a polypeptide chain with two electrically opposite areas. At the N terminus, the polypeptide contains a pleckstrin homology (PH) domain that has been implicated in the recruitment of many signaling proteins to the cell membrane where they exert their biological activities (17). Finally, a bipartite nuclear targeting sequence (18) exists as an integral part of a heptad repeat region that meets all the structural requirements to form a coiled-coil (16).
Protein data bank searches revealed homologies almost exclusively within the approximately 100 residues at the N-terminal region harboring the PH domain. The PH domain of the oxysterol-binding protein, which displays an overall identity of 33.5% and a similarity of 65.2% with our cloned protein, is the most similar. In addition, the Caenorhabditis elegans cosmid F25H2 (GenBank TM accession number Q93569) contains a hypothetical ORF that displays an overall identity of 26.5% and a similarity of 61% throughout the entire protein sequence indicating that similar proteins are present in lower invertebrates.
The distribution and expression of the GPBP gene in human tissues was first assessed by Northern blot analysis ( Fig. 2A). The gene is expressed as two major mRNAs species between 4.4-and 7.5-kb in length and other minor species of shorter lengths. The structural relationship between these multiple mRNA species is not known, and their relative expression varies between tissues. Striated muscle (skeletal and heart) is the tissue with highest expression, whereas lung and liver show the least.
Southern blot studies using genomic DNA from different species indicate that homologous genes exist throughout phylogeny (Fig. 2B). Consistent with the human origin of the probe, the hybridization intensities decrease in a progressive fashion as the origin of the genomic DNA moves away from humans in evolution.
Experimental Determination of the Translation Start Site-To confirm experimentally the predicted ORF, eukaryotic expression vectors containing either the 2.4-kb of cDNA of n4Ј or only the predicted ORF tagged with a FLAG sequence (Fig.   FIG. 1. Nucleotide and  3A) were used for transient expression assays in 293 cells. The corresponding extracts were analyzed by immunoblot using GPBP-or FLAG-specific antibodies. The GPBP-specific antibodies bind to a similar major polypeptide in both transfected cells, but only the polypeptide produced by the engineered construct expressed the FLAG sequence (Fig. 3B). This locates the translation start site of the n4Ј cDNA at the predicted Met and confirms the proposed primary structure. Furthermore, the recombinant polypeptides identified display a molecular mass higher than expected (80 versus 71 kDa) suggesting that GPBP undergoes post-translational modifications.
Expression and Characterization of Yeast rGPBP-Yeast expression and FLAG-based affinity purification were combined to produce rGPBP (Fig. 4A). A major polypeptide of ϳ89 kDa along with multiple related products displaying lower M r were obtained. The recombinant material was recognized by both anti-FLAG and specific antibodies guaranteeing the fidelity of the expression system. Again, however, the M r displayed by the major product was notably higher than predicted and even higher than the M r of the 293 cell-derived recombinant material, supporting the idea that GPBP undergoes important and differential post-translational modifications. Since phosphorylatable residues are the most abundant in the polypeptide chain, we investigated the existence of phosphoamino acids in the recombinant materials. By using monoclonal or polyclonal (not shown) antibodies against phosphoserine (Ser(P)), phosphothreonine (Thr(P)), and phosphotyrosine (Tyr(P)), we identified the presence of all three phosphoresidues either in yeast rGPBP (Fig. 4B) or in 293 cell-derived material (not shown). The specificity of the antibodies was further assessed by partially inhibiting their binding by the addition of 5-10 mM of the corresponding phosphoamino acid (not shown). This suggests that the phosphoresidue content varies upon the cell expression system, and the M r differences are mainly due to phosphorylation. Consistently, dephosphorylated yeast-derived material displays similar M r to the material derived from 293 cells, and phosphoamino acid content correlates with SDS-PAGE mobilities (Fig. 4C). As an in vivo measurement, the phosphorylation of rGPBP in the 293 cells was assessed (Fig. 4D). Control cells (lanes 1) and cells expressing rGPBP in stable (lanes 2) or transient (lanes 3) modes were cultured in the presence of H 3 32 PO 4 . The recombinant material specifically immunoprecipitated contains 32 P indicating that the phosphorylation of GPBP occurs in vivo and therefore is likely to be a physiological process.

The rGPBP Is a Serine/Threonine Kinase That Phosphorylates the N-terminal Region of the Human GP Antigen-Al-
though GPBP does not contain the 12 conserved structural regions required to define the classic catalytic domain for a protein kinase, the recent identification and characterization of novel non-conventional protein kinases (19 -27) encouraged the investigation of its phosphorylating activity. Addition of [␥-32 P]ATP to rGPBP either from yeast or 293 cells (not shown) in the presence of Mn 2ϩ and Mg 2ϩ resulted in the incorporation of 32 P as Ser(P) and Thr(P) in the major and related products that were recognized by both anti-FLAG and specific antibodies (Fig. 5, A and B), indicating that the affinity purified material contains a Ser/Thr protein kinase. To characterize this activity further, GPpep1, GPpep1Ala 9 (a GPpep1 mutant with Ser 9 replaced by Ala), native and recombinant human antigens, and native bovine antigen were assayed (Fig. 5C). Affinity purified rGPBP phosphorylates all human-derived material to a different extent; however, in similar conditions no appreciable 32 P incorporation was observed in the bovine-derived substrate. The lower 32 P incorporation displayed by GPpep1Ala 9 when compared with GPpep1 and the lack of phosphorylation of the bovine antigen indicates that the kinase present in rGPBP discriminates between human and bovine antigens and that Ser 9 is a target for the kinase. Although the purification system provides high quality material, the presence of contaminants with a protein kinase activity could not be ruled out. The existence of contaminants was also suggested by the presence of a FLAG-containing 40-kDa polypeptide displaying no reactivity with specific antibodies nor incorporation of 32 P in the phosphorylation assays (Figs. 4A and 5A). To identify precisely the polypeptide harboring the protein kinase activity, we performed in vitro kinase renaturation assays after SDS-PAGE and Western blot (Fig. 6). We successfully combined the use of specific antibodies (lane 1) and autoradiographic detection of in situ 32 P incorporation (lane 2), and we identified the 89-kDa rGPBP material as the primary polypeptide harboring the Ser/Thr kinase activity. The lack of 32 P incorporation in the rGPBP-derived products as well as in the 40-kDa contaminant further supports the specificity of the renaturation assays and locates the kinase activity to the 89-kDa polypeptide. Recently, it has been shown that traces of protein kinases intimately associated to a polypeptide can be released from the blot membrane, bind to, and phosphorylate the polypeptide during the labeling step (28). To assess this possibility in our system, we performed the renaturation studies using a small piece of membrane containing the 89-kDa polypeptide either alone or together with membrane pieces representing the different regions of the blot lane. We observed similar 32 P incorporation at the 89-kDa polypeptide regardless of the co-incubated pieces (not shown), indicating that if there are co-purified protein kinases in our sample they are not phosphorylating the 89-kDa polypeptide in the renaturation assays unless they co-migrate. Co-migration does not, however, appear to be a concern since rGPBP deletion mutants displaying different mobilities also have kinase activities and could similarly be in-blot renatured (not shown).
Immunohistochemical Localization of the Novel Kinase-To investigate GPBP expression in human tissues, we performed immunohistochemical studies using specific polyclonal (Fig. 7) or monoclonal antibodies (not shown). Although GPBP is widely expressed in human tissues, it shows tissue and cell specificity. In the kidney, the major expression is found at the epithelial cells of the tubules and at the mesangial cells and podocytes of the glomerulus. At the lung alveolus, the antibodies display a linear pattern suggestive of a basement membrane localization along with staining of pneumocytes. Liver shows low expression in the parenchyma but high expression in biliary ducts. The expression at the central nervous system is observed in the white matter and not in the neurons of the brain. In testis, a high expression in the spermatogonium contrasts with the lack of expression in the Sertoli cells. The adrenal gland shows a higher level of expression at the cortical cells versus the medullar. In the pancreas, GPBP is preferentially expressed in Langerhans islets versus the exocrine moiety, and in prostate, GPBP is expressed in the epithelial cells but not in the stroma (Fig. 7). Other locations with high expression of GPBP are striated muscle, epithelial cells of intestinal tract, and Purkinje cells of the cerebellum (not shown). In general, in the tissues where GPBP is highly expressed the staining pattern is mainly diffuse cytosolic. However in certain   FIG. 4. Characterization of rGPBP from yeast and 293 cells. A, 1 g (lane 1) or 100 ng (lanes 2 and 3) A and B) or presence of GP antigen-derived material (C). A, the mixture was subjected to reducing SDS-PAGE (10% gel) and autoradiographed. B, the mixture was subjected to 32 P-phosphoamino acid analysis by two-dimensional thin layer chromatography. The dotted circles indicate the position of ninhydrin-stained phosphoamino acids. C, the phosphorylation mixtures of the indicated GP-derived material were analyzed by SDS-PAGE (15% gel) and autoradiography (GPpep1 and GPpep1Ala 9 ) or immunoprecipitated with monoclonal antibody 17, a monoclonal antibody that specifically recognizes GP antigen from human and bovine origin, and analyzed by SDS-PAGE (12.5%) and autoradiography (rGP and GP). locations there is, in addition, an important staining reinforcement at the nucleus (spermatogonium), at the plasma membrane (pneumocyte, hepatocyte, prostate epithelial cells, white matter), or at the extracellular matrix (alveolus) (Fig. 7). DISCUSSION Our data show that GPBP is a novel non-conventional serine/ threonine kinase and present evidence that indicate that GPBP discriminates between human and bovine GP antigens and targets the exclusive human phosphorylatable region in vitro.
Although the presence of additional protein kinases in the affinity-purified rGPBP cannot completely be ruled out, several lines of evidence indicate that the 89-kDa polypeptide is the only kinase therein. First, we found no differences in auto-or trans-phosphorylation among rGPBP samples purified either in the presence of 150 mM or 0.5, 1, or 2 M salt (not shown) suggesting that rGPBP does not carry kinases intimately bound. Second, the presence of a FLAG-containing yeast-derived kinase in our samples is not a concern since material purified using GPBP-specific antibodies show no differences in phosphorylation (not shown). Third, a deletion mutant of GPBP displays reduced auto-and trans-phosphorylation activities, 3 suggesting that is the only material in the rGPBP with the ability to carry out phosphate transfer.
Immunohistochemistry studies show that GPBP is a cytosolic polypeptide also found in the nucleus, associated with the plasma membrane and likely at the extracellular matrix associated with basement membrane, indicating that it contains the structural requirements to reach all these destinations. The nuclear localization signal and the PH domain confer to it the potential to reach the nucleus and the cell membrane, respectively (17,29,30). Although GPBP does not contain the structural requirements to be exported, at the 5Ј-end untranslated region of its mRNA exists an upstream ORF of 130 residues with an in-frame stop codon at the beginning (Fig. 1). An mRNA editing process inserting a single base pair (U) would generate an operative in-frame start site and an ORF of 754 residues containing an export signal immediately downstream of the edited Met (not shown). Polyclonal antibodies against a synthetic peptide representing part of this hypothetical extra sequence display a linear vascular reactivity in human tissues suggestive of an extracellular basement membrane localization. 3 Alternatively, a splicing phenomenon could generate transcripts with additional unidentified exon(s) that would provide the structural requirements for exportation. The multiple cellular localization, the high content in Tyr(P), and the lack of tyrosine kinase activity in vitro suggest that GPBP in addition is the target of specific tyrosine kinase(s) and therefore likely involved in specific signaling cascade(s).
The idea that common pathogenic events exist at least for some autoimmune disorders is suggested by the significant number of patients displaying more than one autoimmune disease, and also by the strong and common linkage that some of these diseases show to specific major histocompatibility complex haplotypes (31,32). The experimental observation that the autoantigen is the leading moiety in autoimmunity and that a limited number of self-components are autoantigenic (31) suggest that these components share biological features with important consequences in self/non-self recognition by the immune system. One possibility is that triggering events by altering different but specific self-components would result in abnormal antigen processing. In certain individuals expressing a particular major histocompatibility complex specificity, the abnormal peptides could be recognized by non-tolerized T cells and trigger an immune response (1).
We explored the GP antigen to identify biological features of relevance in autoimmune pathogenesis. Since the human antigen is a natural autoantigen but not the homologous counterparts from other superior mammals, and only ␣3 is involved in autoimmunity but not the remaining five ␣ chains, comparative studies among NC1 domains were a useful initial approach. These studies revealed that specific serine phosphorylation as well as pre-mRNA alternative splicing are biological hallmarks of the human versus the other species GP antigens (4,5). These two features are also associated with the biology of other autoantigens including acetylcholine receptor and myelin basic protein (4). The latter is suspected to be the major antigen in multiple sclerosis, another exclusively human autoimmune disease in which the immune system targets the white matter of the central nervous system. GP disease and multiple sclerosis are human disorders that display a strong association with the same HLA class II haplotype (HLA DRB1*1501) (32,33). This along with the recent report of death by GP disease of a multiple sclerosis patient carrying this HLA specificity (34) support the existence of common pathogenic events in these human disorders.
Phosphorylation of specific serines has been shown to change intracellular proteolysis (35)(36)(37)(38)(39)(40). Conceivably alterations in protein phosphorylation can affect processing and peptide pres-entation and thus mediate autoimmunity. GP antigen-derived peptide presentation by the HLA-DR15 depends more on processing than on preferences of relatively indiscriminate DR15 molecules (41), suggesting that if processing is influenced by abnormal phosphorylation, the resulting peptides would likely be presented by this HLA. Our more recent data indicate that in both the GP and myelin basic protein systems, the production of alternative splicing products serves to regulate the phosphorylation of specific and structurally homologous cAMPdependent protein kinase sites, 2 suggesting that this or a closely related kinase is the in vivo phosphorylating enzyme. Alterations in the degree of antigen phosphorylation, caused either by an imbalance in alternative products or by the action of an intruding kinase that deregulates phosphorylation of the same motifs, could lead to an autoimmune response in predisposed individuals. Accordingly, we found that in kidney, GP patients express relatively more alternative products than control individuals (5) and that rGPBP phosphorylates the human GP antigen at a major cAMP-dependent protein kinase phosphorylation site in an apparently unregulated fashion since alternative products do not affect antigen phosphorylation. 3 Although GPBP is ubiquitously expressed, in certain organs and tissues it shows a preference for cells and tissue structures that are a target of common autoimmune responses as follows: the Langerhans cells (type I diabetes); the white matter of the central nervous system (multiple sclerosis); the biliary ducts (primary biliary cirrhosis); the cortical cells of the adrenal gland (Addison disease); striated muscle cells (myasthenia gravis); spermatogonium (male infertility); Purkinje cells of the cerebellum (paraneoplastic cerebellar degeneration syndrome); and epithelial intestinal cells (pernicious anemia, autoimmune gastritis and enteritis). Although it is premature to draw definitive conclusions on the pathogenic involvement of GPBP, all the above observations point to this novel kinase as an attractive candidate to be considered when envisioning a model for human autoimmune disease.