Goodpasture antigen-binding protein, the kinase that phosphorylates the goodpasture antigen, is an alternatively spliced variant implicated in autoimmune pathogenesis.

The non-collagenous C-terminal domain of the alpha(3) chain of collagen IV is the autoantigen in Goodpasture disease, an autoimmune disorder described only in humans. Specific N-terminal phosphorylation is a biological feature unique to the human domain when compared with other homologous domains lacking immunopathogenic potential. We have recently cloned from a HeLa-derived cDNA library a novel serine/threonine kinase (Goodpasture antigen-binding protein (GPBP)) that phosphorylates the N-terminal region of the human domain (Raya, A. Revert, F, Navarro, S. and Saus J. (1999) J. Biol. Chem. 274, 12642-12649). We show here that the pre-mRNA of GPBP is alternatively spliced in human tissues and that the most common transcript found encodes GPBPDelta26, a molecular isoform devoid of a 26-residue serine-rich motif. Recombinantly expressed GPBPDelta26 exhibits lower activity than GPBP, due at least in part to a reduced ability of GPBPDelta26 to interact and to form very active high molecular weight aggregates. In human tissues, GPBP shows a more limited expression than GPBPDelta26 but displays a remarkable preference for the small vessels and for histological structures targeted by natural autoimmune responses including alveolar and glomerular basement membranes, the two main targets in Goodpasture disease. GPBP expression is, in turn, up-regulated in the striated muscle of a Goodpasture patient and in other autoimmune conditions including cutaneous lupus erythematosus, pemphigoid, and lichen planus.

Goodpasture (GP) 1 disease is an exclusive human disorder characterized by altered renal and pulmonary functions caused by deposits of autoantibodies along the glomerular and alveolar basement membranes. The pathogenic antibodies arise against the non-collagenous C-terminal (NC1) region of the ␣3 chain of collagen IV (␣3(IV)NC1), also called the GP antigen (for review see Ref. 1). Attention has been addressed to unique biological features of this domain compared with related domains from human or other species, since only the human ␣3(IV)NC1 has been implicated in spontaneous autoimmune pathogenesis. The human ␣3(IV)NC1 contains a unique N terminus (2,3) that Goodpasture antigen-binding protein (GPBP), a novel nonconventional protein kinase, binds and phosphorylates in vitro (Ref. 4 and this report). This fact, along with the finding that GPBP is preferably expressed in tissues and cells that are targets in common spontaneous autoimmune diseases prompted the idea that GPBP might be important in human autoimmune pathogenesis (4).
Here we report the existence of two isoforms of GPBP that are generated by alternative splicing of a 78-bp exon that encodes a 26-residue serine-rich motif. The presence of the 26-residue motif in the polypeptide chain results in a more active molecular species that shows preferential expression in tissue structures targeted by autoimmune responses, and the expression of which is up-regulated during autoimmune pathogenesis.
Oligonucleotides-The following oligonucleotides were synthesized from Life Technologies, Inc. Plasmid Construction, Expression, and Purification of Recombinant Proteins-The plasmid pHIL-FLAG-n4Ј, used for recombinant expression of FLAG-tagged GPBP in Pichia pastoris, has been described elsewhere (4). The sequence coding for the 78-bp exon was deleted by site-directed mutagenesis using ON-GPBP-⌬26 to generate the plasmid pHIL-FLAG-n4Ј⌬26. Expression and affinity purification of the recombinant proteins, rGPBP and rGPBP⌬26 and human ␣3(IV)NC1, was done as in Ref. 4.
HPLC Gel Filtration-Samples of 250 l were injected into a gel filtration PE-TSK-G4000SW HPLC column equilibrated with 50 mM Tris-HCl, pH 7.5, 150 mM NaCl. The material was eluted from the column at 0.5 ml/min and monitored at 220 nm, and fractions were collected every minute.
In Vitro Phosphorylation Assays-The autophosphorylation, transphosphorylation, and in-blot renaturation studies were performed as in Ref. 4.
Northern Hybridization Studies-Pre-made Northern blots (CLON-TECH) representing multiple human tissues or human-derived tumor cell lines were probed with a cDNA containing the 78-bp exon present only in GPBP or with a 261-bp cDNA fragment representing the flanking regions (104-bp 5Ј and 157-bp 3Ј) of the 78-bp exon common to both isoforms. The corresponding cDNA fragments were obtained by PCR using the pair of primers ON-GPBP-56m and ON-GPBP-57c using GPBP as a template or with primers ON-GPBP-22m and ON-GPBP-20c using GPBP⌬26 as a template. The resulting products were randomlabeled and hybridized following the manufacturer's instructions.
Antibodies-Polyclonal antibodies were raised in chicken against a synthetic peptide (GPBPpep1) representing the sequence coded by the 78-bp exon (Genosys). Egg yolks were diluted 1:10 in water, and the pH was adjusted to 5.0. After 6 h at 4°C, the solution was clarified by centrifugation (25 min at 10,000 ؋ g at 4°C), and the antibodies were precipitated by adding 20% (w/v) of sodium sulfate and centrifugation at 20,000 ϫ g for 20 min. The pellets were dissolved in phosphatebuffered saline (1 ml per yolk) and used for immunohistochemical studies. The production of antibodies against GPBP/GPBP⌬26 has been previously reported (4).
Immunochemical Techniques-SDS-PAGE and Western blotting were performed under reducing conditions as in Ref. 3. For far Western purposes, human recombinant ␣3(IV)NC1 (4) was analyzed by SDS-PAGE under reducing conditions, transferred to Immobilon P (Millipore), renatured in Tris-buffered saline in the presence of Tween 20 (0.05%), and probed for 1 h at 37°C with 30 g/ml of either rGPBP or rGPBP⌬26 in the same buffer. Bound material was detected using GPBP/GPBP⌬26-specific monoclonal antibodies (monoclonal antibody 14) and an immunoperoxidase-based conjugate.
Immunohistochemical studies were done on formalin-fixed paraffinembedded or frozen human biopsies fixed with cold acetone using an immunoperoxidase-based method (6) or standard procedures for indirect immunofluorescence. Skin biopsies from 6 healthy, 11 non-autoimmune dermatitis or vasculitis, 2 lichen planus, 5 CLE, 1 pemphigus, and 1 pemphigoid were independently analyzed using one of the chemical or immunochemical approaches indicated. In addition to clinical diagnosis, all the patients had histological diagnosis confirmation except for the pemphigus.
Sedimentation Velocity and Equilibrium-Determination of sedimentation velocities were performed in an Optima XL-A analytical ultracentrifuge (Beckman Instruments Inc.), equipped with a VIS-US scanner, using a Ti60 rotor and double sector cells of Epon-charcoal of 12-mm optical path length. Samples of ϳ400 l were centrifuged at 30,000 rpm and 20°C, and radial scans at 220 nm were taken every 5 min. The sedimentation coefficients were obtained from the rate of movement of the solute boundary using the program XLAVEL (supplied by Beckman).
Sedimentation equilibrium experiments were done as described above for velocity experiments using samples of 70 l and centrifugation at 8,000 rpm. The experimental concentration gradients at equilibrium were analyzed using the program EQASSOC (Beckman) to determine the corresponding weight average molecular mass. A partial specific volume of 0.711 cm 3 /g for GPBP and 0.729 cm 3 /g for GPBP⌬26 were calculated from the corresponding amino acid compositions. These studies were performed at the Analytical Ultracentrifugation facilities of the Centro de Investigaciones Biológicas (Consejo Superior de Investigaciones Científicas, Madrid, Spain).

RESULTS
Identification of Two Spliced GPBP Variants-To characterize GPBP in human tissues, we performed RT-PCR on total RNA using specific oligonucleotides that flank the full open reading frame of GPBP. A single cDNA fragment displaying a lower size than expected was obtained from skeletal muscle (Fig. 1A), and also from kidney, lung, skin, and adrenal gland (not shown). We fully sequenced the 2.2-kb cDNA from human muscle (GenBank TM accession number AF232930), and we FIG. 1. GPBP⌬26 is a splicing variant of GPBP. A, total RNA from normal skeletal muscle was retrotranscribed using primer 53c and subsequently subjected to PCR with primers 11m-53c (lane 2) or the nested primers 15m-62c (lane 4). Control amplifications of a plasmid containing GPBP cDNA using the same pairs of primers are shown in lanes 1 and 3. Numbers on the left and right refer to molecular weight in base pairs. B, the region missing in the muscle transcript was identified, and its nucleotide cDNA sequence (lowercase) and deduced amino acid sequence (uppercase) were determined. The genomic DNA comprising the cDNA region of interest was sequenced, and its structure is drawn in C. The location and relative sizes of the 78-bp exon spliced out in GPBP⌬26 (black box), adjacent exons (gray boxes), and introns (lines) are represented. The size of both intron and exons is given, and the nucleotide sequence of intron-exon boundaries is presented with the consensus for 5Ј and 3Ј splice sites shown in boldface.
found it identical to HeLa-derived material except for the absence of a 78-bp fragment (positions 1519 -1596) that encodes a 26-residue motif (371-396 amino acids) (Fig. 1B). We therefore named this more common isoform of GPBP as GPBP⌬26. By combining nested PCR re-amplifications and endonuclease restriction mapping, we determined that the RT-PCR products obtained in every human tissue tested corresponded to the same molecular species (not shown).
To investigate whether the 78-bp fragment represents an exon skipped during pre-mRNA processing, we used this cDNA fragment to probe a human-derived genomic library, and we isolated an ϳ14-kb clone. By combining Southern blot hybridization and PCR, the genomic clone was characterized, and a contiguous DNA fragment of 12,482-bp was fully sequenced (GenBank TM accession number AF232935). The sequence contained from 5Ј to 3Ј, 767 bp of intron sequence, a 93-bp exon, an 818-bp intron, the 78-bp exon of interest, a 9650-bp intron, a 96-bp exon, and 980 bp of intron sequence (Fig. 1C). The exon/ intron boundaries determined by comparing the corresponding DNA and cDNA sequences meet the canonical consensus for 5Ј and 3Ј splice sites ( Fig. 1C) (7), thus confirming the exon nature of the 78-bp sequence.
We assessed the relative expression of GPBP and GPBP⌬26 by Northern blot analysis (Fig. 2). Both isoforms were preferably expressed in striated muscle (skeletal and heart) and brain and poorly expressed in placenta, lung, and liver. However, in kidney and pancreas as well as in the cancer cell lines, the molecular species containing the 78-bp exon (GPBP) were relatively expressed at much lower levels than those devoid of the exon (GPBP⌬26). Note that in Fig. 2 we used more permissive conditions to enhance hybridization of the GPBP-specific probe since no signal was detectable when washing at stringent conditions (not shown).
All the above indicate that GPBP is expressed at low levels in normal human tissues, and therefore the initial failure to detect GPBP by RT-PCR in different human tissues can be attributed to a preferential amplification of the more abundant GPBP⌬26. Accordingly, the cDNA of GPBP was obtained from human tissues (skeletal muscle, lung, kidney, skin, and adrenal gland) when 78-bp exon-specific primers were used in the PCR (not shown). Finally, quantitative RT-PCR studies on total RNA from human skeletal muscle reveal that the molecular species containing the 78-bp exon (GPBP) represent less than 10% of the molecular species devoid of it (GPBP⌬26) (not shown). Thus in the Northern blot studies shown in Fig. 2, the major transcript detected when using the cDNA probe that represents both isoforms was GPBP⌬26.
Recombinant Expression and Functional Characterization of GPBP⌬26 -To investigate whether the absence of the 26-residue serine-rich motif would affect the biochemical properties of GPBP, we expressed and purified both isoforms (rGPBP and rGPBP⌬26), and we assessed auto-and trans-phosphorylation activities along with their ability to bind recombinant ␣3(IV)NC1 (Fig. 3). SDS-PAGE and Western blot analysis revealed that purified materials contained a major single polypeptide and several related products, which number, relative amounts, and molecular weight in rGPBP⌬26 varied with respect to rGPBP (Fig. 3A). However, these differences including the molecular weight displayed by the major polypeptide product (89 kDa for GPBP and 84 kDa for GPBP⌬26) could not only be attributed to the mere presence or absence of the 26-residue, thus revealing the existence of important structural differences between GPBP and GPBP⌬26. These structural differences were responsible at least in part for the higher in-solution activity displayed by rGPBP, which was more efficient in both autophosphorylation and phosphorylating the N-terminal region of the human ␣3(IV)NC1 domain (Fig. 3B). The phosphate transfer activities of rGPBP and rGPBP⌬26 were, however, very similar after SDS-PAGE and in situ renaturation studies (Fig. 3C), indicating that the structural features that depend on the presence or absence of the 26-residue motif could not be efficiently renatured. The higher activity of GPBP was also confirmed by showing higher binding of this isoform to recombinant human ␣3(IV)NC1 domain in specific far Western studies 2 (Fig. 3D). Renaturation studies further show that kinase activity resides at the major polypeptides representing the proposed open reading frames and is not detectable at the derived products.
rGPBP and rGPBP-26 Exist as Very Active High Molecular Weight Aggregates-Gel filtration analysis of rGPBP or rG-PBP⌬26 yielded two chromatographic peaks (I and II) (Fig. 4), both displaying higher molecular weight than expected for the major individual molecular species as determined on SDS-PAGE studies (89 and 84 kDa, respectively). The bulk of the FIG. 2. Differential expression of GPBP and GPBP⌬26. Fragments representing the 78-bp exon (GPBP) or flanking sequences common to both isoforms (GPBP/GPBP⌬26) were 32 P-labeled and used to hybridize human tissue and tumor cell line Northern blots (CLONTECH). The membranes were first hybridized with GPBP-specific probe, stripped, and probed again with GPBP/GPBP⌬26-specific 32 P-labeled material. Washing conditions were less stringent for GPBP-specific probe (0.1% SSPE, 37°C or 55°C) than for the GPBP/GPBP⌬26 (0.1% SSPE, 68°C) to increase GPBP and GPBP⌬26 signals, respectively. No detectable signal was obtained for the GPBP-specific probe when the washing program was done at 68°C (not shown). The numbers denote the position and the sizes in kb of the RNA markers used. recombinant material eluted as a single peak between the 158and 669-kDa molecular weight markers (peak II), whereas limited amounts of rGPBP and only traces of rGPBP⌬26 eluted in peak I (Ͼ1000 kDa). Aliquots of fractions representing each chromatographic profile were subjected to SDS-PAGE and Coomassie Blue-stained or incubated in the presence of [␥-32 P]ATP and analyzed by immunoblot and autoradiography. The material in every chromatographic peak contained the primary polypeptide and minor related products (not shown in Fig. 4 composite), indicating that the primary polypeptide associates to form high molecular weight aggregates mainly stabilized by non-covalent bonds. The kinase activity also exhibited two peaks coinciding with the chromatographic profiles; however, the material eluted in peak I showed a much higher specific activity than the material present in peak II, indicating that these high molecular weight aggregates contained a much more active form of the kinase. In the study shown, equal volumes of rGPBP fractions number 13 and 20 exhibited comparable phosphorylating activity, despite that the protein content was ϳ20 times lower in fraction 13 as estimated by Western blot and Coomassie Blue staining (Fig. 4A). The specific activities of rGPBP and rGPBP⌬26 at peak II are also different and consistent with the studies shown for the whole material, thus supporting that the presence of the 26-residue serine-rich motif renders a more active kinase. These results also suggest that both rGPBP and rGPBP⌬26 exist as oligomers under native conditions and that both high molecular weight aggregate formation and specific activity are greatly dependent on the presence of the 26-residue serine-rich motif. Analytical centrifugation analysis of rGPBP revealed that peak I contained large aggregates (over 10 7 Da). Peak II of rGPBP contained a homogeneous population of 220 Ϯ 10 kDa aggregates likely representing trimers with a sedimentation coefficient of 11 S. Peak II of rGPBP⌬26, however, consisted of a more heterogeneous population that likely contains several oligomeric species. The main population (ϳ80%) displayed a weight average molecular mass of 310 Ϯ 10 kDa and a coefficient of sedimentation of 14 S.
GPBP and GPBP⌬26 Self-interact in a Yeast Two-hybrid System-To assess the physiological relevance of the self-aggregation and to dig into the role of the 26-residue motif, we performed comparative studies using a two-hybrid interaction system in yeast. In this type of studies the polypeptides whose interaction is under study are expressed as a part of a fusion protein containing either the activation or the binding domains of the transcriptional factor GAL4. An effective interaction between the polypeptides under study would result in the reconstitution of the transcriptional activator and the subsequent expression of the two reporter genes lacZ and his3 genes, allowing colony color detection and growth in a His-defective medium, respectively. We estimated the intensity of interactions by the growth rate in histidine-defective medium in the presence of different concentrations of a competitive inhibitor of the his3 gene product (3-aminotriazole) and a quantitative colorimetric liquid ␤-galactosidase assay. A representative experiment is presented in Fig. 5. When assaying GPBP⌬26 for self-interaction, a significant induction of the reporter genes  4. rGPBP and rGPBP⌬26 form very active high molecular weight aggregates. About 300 g of rGPBP (A) or rGPBP⌬26 (B) were subjected to HPLC gel filtration. Arrowheads and numbers, respectively, indicate the elution profile and molecular mass of molecular weight standards used. Larger aggregates eluted in the void volume (I), and the bulk of the material eluted in the separation range of the column as a second peak between 669 and 158 kDa standards (II). Fifteen microliters of the indicated fractions were subjected to SDS-PAGE and Coomassie Blue-stained (Coomassie). Five microliters of the same fractions were in vitro phosphorylated, and the reaction was stopped by boiling in SDS sample buffer, analyzed by SDS-PAGE, transferred, and autoradiographed for 1 or 2 h (Kinase assay), and then blotted using anti-FLAG antibodies (Western). was observed, whereas no expression was detectable when each fusion protein was individually expressed (not shown) or expressed with the partner empty plasmid (Fig. 5A). The insertion of the 26-residue motif in the polypeptide resulted in a notable increase in polypeptide interaction allowing significant growth in plates containing 1 mM 3-aminotriazole (Fig. 5A). We arrived to similar conclusions when the interactions for the different plasmid combinations were measured in a ␤-galactosidase liquid assay (Fig. 5B). These results indicate that GPBP⌬26 self-associates in vivo and that the insertion of the 26 residues into the polypeptide chain yields a more interactive molecular species.
GPBP Is Highly Expressed in Human but Not in Bovine or in Murine Glomerulus and Alveolus-By using antibodies that recognize both GPBP and GPBP⌬26, we have shown preferential expression of these proteins in human cells and tissues targeted by common autoimmune responses (4). To investigate specifically the expression of GPBP, we raised polyclonal antibodies against a synthetic peptide representing the 26-residue motif characteristic of this isoform (GPBPpep1), and we used them for immunohistochemical studies on formalin-fixed paraffin-embedded human tissues. In some tissues, these antibodies showed more specificity for the autoimmune targets (e.g. the biliary ducts and the Langerhans islets) than the antibodies recognizing both isoforms (not shown). Nevertheless, the most remarkable finding was the presence of linear deposits of antibodies around the small vessels (Fig. 6, A and B), suggesting that GPBP associates with endothelial basement membranes. Consequently, the anti-GPBPpep1 antibodies displayed a strong association with glomerular (Fig. 6, C and D) and alveolar (not shown) basement membranes and were deposited in human glomerulus closely resembling GP autoantibodies (Fig.  6, E and F). These findings further support the initial observation that GPBP is expressed in tissue structures targeted by FIG. 6. GPBP is expressed associated with endothelial and glomerular basement membranes. Formalin-fixed paraffin-embedded sections of human control muscle (A and B) or renal cortex (C and D) were probed with GPBP-specific antibodies and stained using a peroxidase-based immunoconjugate and hematoxylin. Magnification was ϫ 40 in A and C and ϫ 100 with immersion in B and D. E and F, frozen section of human renal cortex were probed with GPBP-specific antibodies (E) or with GP autoantibodies (F). Control sera (chicken preimmune and normal human) did not display significant binding in parallel studies (not shown).
FIG. 5. Self-association of GPBP and GPBP⌬26 using the yeast two-hybrid system. A, transformants for the indicated combinations (1-6) of plasmids (AD and BD refer to plasmid-expressing activation and binding domains of GAL4, respectively) were selected in tryptophan/ leucine-deficient medium. Independent transformants were streaked either onto tryptophan/leucine-deficient medium (ϪTrp, ϪLeu) or onto tryptophan/leucine/histidine-deficient plates (ϪTrp, ϪLeu, ϪHis) in the presence or absence of 1 mM 3-aminotriazole (3-AT) to assess interaction. The picture was taken 3 days after streaking. B, the bars represent mean values in ␤-galactosidase arbitrary units of four independent ␤-galactosidase liquid assays.
natural autoimmune responses (4) and suggest that GPBP expression confers vulnerability to autoimmune attack.
To assess further this hypothesis, we investigated the expression of GPBP and GPBP⌬26 in the kidney of mammals that naturally do not undergo GP disease, and we compared their expression in human renal cortex (Fig. 7). GPBP-specific antibodies failed to stain the vessels and the glomerulus of either bovine or murine specimens (compare Fig. 7, A with B  and C). However, antibodies specific for GPBP and GPBP⌬26 stained renal cortex in all three species, although with different distributions and intensities (Fig. 7, D-F). In bovine renal cortex, GPBP⌬26 was relatively less expressed than in human but showed similar tissue distribution. In murine samples, however, GPBP⌬26 displayed a tissue distribution closely re-  3 and 4, respectively) were analyzed by RT-PCR as in Fig. 8. Similar amounts of GPBP⌬26 products were loaded into the gel to show the relatively higher expression of GPBP in lupus erythematosusaffected skin. A total of three independent patients with CLE were analyzed using either immunohistochemistry or RT-PCR approaches. sembling that of GPBP in human samples. Similar results were obtained when studying lung samples from the three different species (not shown). To rule out that the differences in the antibody detection were due to primary structure divergence rather than to a differential expression, we determined by cDNA sequencing the corresponding primary structure in these two species (GenBank TM accession numbers AF232931 and AF232932). Bovine and mouse GPBP displayed an overall identity with human material of 97.9 and 96.6%, respectively. Furthermore, the mouse 26-residue motif was identical to human, whereas the bovine motif differed only in one residue. Finally, we successfully amplified GPBP cDNA from mouse or bovine kidney total RNA using 78-bp exon-specific oligonucleotides (not shown), indicating that in these species GPBP is also expressed in the kidney but at levels that are not detectable by the immunochemical techniques.
GPBP Is Highly Expressed in Several Autoimmune Conditions-We analyzed several tissues from different GP patients by specific RT-PCR to assess GPBP/GPBP⌬26 mRNA levels. As in control kidneys, the major expressed isoform in GP kidneys was GPBP⌬26 (not shown). However, GPBP was preferentially expressed in the skeletal muscle of one patient, whereas GPBP⌬26 was the only isoform detected in a control human muscle (Fig. 8A). Quantitative RT-PCR studies supported these findings, and thus in the patient muscle the 78-bp exon-containing mRNAs were approximately 10 times more abundant than GPBP⌬26, whereas these molecular species represented less than one-tenth of the GPBP⌬26 mRNAs in control muscle (not shown).
Since we did not have a kidney sample from this particular patient, we could not assess GPBP/GPBP⌬26 expression in the corresponding target organ. For similar reasons, we could not assess GPBP/GPBP⌬26 expression in the muscle of the patients whose kidneys were analyzed. Nevertheless, the increased levels of GPBP in a GP patient suggest that GPBP/ GPBP⌬26 expression is altered during GP pathogenesis and that augmented GPBP expression has a pathogenic significance in autoimmunity.
To investigate the expression of GPBP and GPBP⌬26 in autoimmune pathogenesis, we studied cutaneous autoimmune processes and compared with control samples representing normal skin or non-autoimmune dermatitis (Fig. 8). A variable but limited expression of GPBP restricted to the most peripheral strata of the epidermis was observed in control skins (Fig. 8, B and E). However, keratinocytes expanding from basal to corneum strata expressed abundant GPBP in skin affected by either CLE (Fig. 8, C and F), lichen planus (Fig. 8, D and G), or pemphigoid (not shown). GPBP was expressed in bleb structures at the cell surface (Fig. 8, F and G) previously identified as apoptotic bodies in cultured keratinocytes upon UV irradiation (8). In contrast, antibodies recognizing both GPBP/ GPBP⌬26 yielded a major diffuse cytosolic pattern through the whole epidermis in both autoimmune-affected or control samples (not shown).
The differences in GPBP expression found between control and autoimmune-affected epidermis were also evident when the comparative immunochemical studies were done using samples representing non-affected or affected skin regions of individual CLE patients (Fig. 9, A-C). Furthermore, the relative content of GPBP in skin extracts representing CLE-affected regions was significantly augmented in comparison with non-affected regions of the same patient or with control samples (normal and non-autoimmune dermatitis) as determined by specific RT-PCR studies (Fig. 9D).

DISCUSSION
Alternative pre-mRNA splicing is a fundamental mechanism for differential gene expression that has been reported to regulate the tissue distribution, the intracellular localization, and the activity of different protein kinases (9 -12). Closely resembling GPBP, B-Raf exits as different spliced variants in which the presence of specific exons results in more interactive, efficient, and oncogenic kinases (13).
Although it is evident that rGPBP⌬26 still bears the uncharacterized catalytic domain of this novel kinase, both binding and phosphorylating activities are greatly reduced when compared with rGPBP. Gel filtration and two-hybrid experiments provide some insights into the mechanisms that underlie such a reduced phosphate transfer activity. About 1-2% of rGPBP is organized in very high molecular weight aggregates that display about one-third of the phosphorylating activity of rGPBP, indicating that molecular aggregation renders a more efficient quaternary structure. Consistently rGPBP⌬26, with virtually no peak I material, displays a reduced kinase activity. However, aggregation does not seem to be the only mechanism by which the 26-residue motif increases the specific activity since the rGPBP⌬26 material present in peak II also shows a reduced phosphorylating activity when compared with homologous fractions of rGPBP. Far Western and two-hybrid studies suggest that rGPBP-derived aggregates also display higher specific activities because the insertion of the 26-residue motif renders a more interactive and strengthened quaternary structure. Conformational changes induced by the presence of an exon-encoded motif that alters the activation status of the kinase have been proposed for the linker domain of the Src protein (14) and exons 8b and 10 of B-Raf (13). Alternatively, the 26-residue motif may provide the structural requirements for specific phosphorylation (Ser or Tyr) necessary for full activation of the kinase. In any event, the similar activity displayed by each individual isoform in the in-blot renaturation studies indicates the existence of a common catalytic domain and suggests that GPBP⌬26 and GPBP represent different strategies to regulate its activity.
The primary structure of the ␣3(IV)NC1 lends itself to a complex folding process that yields multiple conformers (conformational isomers). Non-assembled conformers are metastable structures specifically activated by phosphorylation for supramolecular aggregation and likely quaternary structure formation. 3 Furthermore, phosphorylation is a signal that the cell uses to derive the folding of the ␣3(IV)NC1 domain toward non-minimum energy ends. 3 Finally, the ␣3(IV)NC1 domain of GP patients shows conformational alterations that are specifically recognized by the pathogenic autoantibodies, suggesting that altered conformers elicited the autoimmune response. 3 In this scenario, three features of the human system place the conformational process of the ␣3(IV)NC1 domain in a vulnerable condition that, in turn, predispose humans for GP disease. 1) The N terminus of the human ␣3(IV)NC1 contains a phosphorylatable motif shared by cAMP-dependent protein kinase and GPBP (2)(3)(4). 2) The human ␣3(IV) gene generates multiple products by alternative exon splicing (5,15). Exon skipping generates alternative products with divergent C-terminal ends that up-regulate the cAMP-dependent protein kinase phosphorylation of the primary ␣3(IV)NC1 product. 4 3) GPBP is expressed and associated with the two major targets of the GP autoantibodies, glomerular and alveolar basement membranes. So far, all the GP kidneys studied expressed higher levels of the alternative product (5), and an augmented expression of GPBP has been found in a GP patient. Both conditions are expected to increase the phosphorylation of the ␣3(IV)NC1 domain and therefore to influence the corresponding conformational process.
Keratinocytes, during maturation from basal to corneum strata, undergo an apoptosis-dependent differentiation process (16,17). Cell surface bleb formation is a well-established step between nuclear condensation and cytoplasmic contraction in the apoptotic morphological cascade of many cell systems including keratinocytes (8). In the epidermis, GPBP is associated with cell surface blebs of keratinocytes suggesting that GPBP expression and apoptosis are related processes. Keratinocytes from SLE patients show a remarkably increased sensitivity to UV-induced apoptosis (8,18,19), and a premature and aberrant apoptosis of the basal keratinocytes has been reported to occur in SLE and dermatomyositis (20). Consistently, we found GPBP expressed in apoptotic bodies expanding from basal to peripheral strata in the epidermis affected by different autoimmune processes including CLE, pemphigus, pemphigoid, and lichen planus. Autoantigens and modified versions thereof are clustered at the cell surface blebs of apoptotic keratinocytes (8,18,19). Apoptotic surface blebs present autoantigens and likely release modified versions of them to circulation (18 -24). It has been suggested that the release of modified autoantigens from apoptotic bodies could be the immunizing event that mediates systemic autoimmune responses leading to SLE and scleroderma (18,24). Finally, phosphorylation has been reported to be a major modification that autoantigens undergo in apoptotic keratinocytes (19).
Although not being limited to an exact mechanism, we propose, in light of all of the above data, that GPBP/GPBP⌬26 likely represent two different strategies to regulate a kinase that is involved in the phosphorylation-dependent folding process of specific self-components. In certain tissues, the expression of the more active GPBP could serve to generate misfolded substrates as a part of an apoptotic-dependent strategy for desired cell removal, e.g. corneum stratum of the epidermis. Ectopic and/or augmented expression of GPBP could release an excess of altered antigens and engage the immune system in a more general autoimmune response as occurs in SLE or scleroderma. In other locations, e.g. the human glomerular and alveolar basement membranes, the presence of the active GPBP isoform may be required for efficient phosphorylation-mediated collagen IV assembly. An augmented expression of the alternative ␣3(IV)NC1 products alone or with increased levels of GPBP could result in the generation of ␣3(IV)NC1 conformers for which the immune system has not established a tolerance. Upon assembly, the altered conformers could engage the immune system in a tissue-restricted autoimmune response as in GP disease.
The substrate condition of the ␣3(IV)NC1 domain for GPBP (4) along with the presence of GPBP associated with the glomerular and alveolar basement membranes seem to predispose humans to undergo GP disease. Similarly, autoantigens and GPBP co-localize in the apoptotic bleb, and recombinant proteins representing autoantigens in SLE (P1 ribosomal phospho-protein and Sm-D1 small nuclear ribonucleoproteins) or in dermatomyositis (histidyl-tRNA synthetase) are in vitro substrates of GPBP (not shown). It is therefore likely that GPBP generates modified versions of the autoantigens which when released from the apoptotic bodies mediate the autoimmune responses in SLE or in dermatomyositis.
Although further studies are needed to define precisely the biological role of GPBP/GPBP⌬26, the low expression of GPBP in cancer cell lines (Fig. 2) and its high expression in the apoptotic bodies suggest that GPBP is involved in signaling pathways induced during programmed cell death. As discussed, GPBP expression could be up-regulated during autoimmune pathogenesis and mediate the immune response. Conversely, down-regulation of GPBP during cell transformation would prevent an autoimmune attack on the transformed cells during tumor growth.