Identification of two alternatively spliced forms of Human Tubulointerstitial Nephritis Antigen (TIN-Ag)

Abstract

Tubulointerstitial nephritis antigen (TIN-Ag) is a recently described basement membrane glycoprotein reactive with autoantibodies in some forms of immunologically mediated human tubulointerstitial nephritis. This report presents the complete cDNA and predicted amino acid sequences of two human TIN-Ag mRNA species referred to as TIN1 and TIN2. Translation through the open reading frames of these clones indicates the presence of a signal peptide and putative prepropeptide. TIN1 additionally contains a characteristic laminin-like epidermal growth factor (EGF) motif and significant homology within the carboxy terminus with the cysteine proteinase family of enzymes. The EGF motif bears important similarities in the positions of cysteines with two motifs in the propeptide of von Willebrand factor. The EGF motif and part of the region that is homologous with the cysteine proteinase family are removed from the TIN2 cDNA. However, the rest of the sequence is identical in these two forms, indicating an alternatively spliced TIN-Ag mRNA product. Both forms contain putative calcium-binding sites. Secondary structure predictions strongly suggest differences between TIN1 and TIN2 leading to the hypothesis that these two forms of TIN-Ag may exhibit differences in their function. Expression studies with appropriate probes demonstrate expression mainly in the kidney and in the intestinal epithelium and lack of expression in other tissues. In the kidney, both TIN1 and TIN2 transcripts are detected, however, TIN1 appears to be the predominant form.

Basement membranes are thin extracellular matrices with a ubiquitous occurrence in the body (^1,2,3). The major constituent elements of basement membranes are type IV collagen, laminin, entactin/nidogen, and perlecan (^4,5,6,7,8). These elements assemble and form a supramolecular mesh-like scaffold that confers mechanical stability to the tissue.

Many important biologic functions have been ascribed to basement membranes. The precise temporal and spatial expression of basement membrane proteins contributes significantly to induction of tissue differentiation and determination of cellular phenotype (^9,10,11,12). In addition, they are crucially involved in cell adhesion, migration, and invasion, and act as selective barriers in the filtration of macromolecules (^{13,14,15,16,17,18,19,20,21,22,23,24}). Several basement membrane macromolecules have additionally been described in association with various structural abnormalities and autoimmune disorders (^25,26,27). One such autoimmune disorder, primary immunologically mediated tubulointerstitial nephritis (TIN), is characterized by abnormalities of the renal tubules, infiltration of the interstitial compartment with mononuclear and polymorphonuclear leukocytes, and the linear deposition of IgG along tubular basement membrane (TBM) (^28,29,30).

Investigations into the molecular origin of the TBM component(s) reactive with sera from patients with anti-TBM nephritis have resulted in the description of several nephritogenic tubulointerstitial antigens (^{31,32,33,34,35}). TIN-Ag is a recently described basement membrane macromolecule. It is the primary target antigen of the TBM in some forms of primary TIN, and anti-glomerular basement membrane nephritis (^36,37,38,39).

The deduced amino acid sequence of a recently identified cDNA encoding the rabbit TIN-Ag (⁴⁰) reveals several interesting structural motifs, including a signal peptide, prepropeptide, and a characteristic epidermal growth factor (EGF)-like motif common to basement membrane molecules in the amino terminus. Additionally, significant homology to the cysteine proteinase family of enzymes exists in the carboxy terminus, indicating conserved structure elements and functional capabilities (⁴¹).

Further knowledge especially of the structure of human TIN-Ag is essential to our understanding of the normal physiologic contributions of this unique macromolecule and in delineating its role in the pathophysiology associated with the basement membranes. In this report, we present the predicted nucleotide and deduced amino acid sequence of human TIN-Ag; discuss important structural motifs present in TIN-Ag sequence; compare the two alternatively spliced forms of TIN-Ag message and discuss putative structural differences; and examine the expression pattern of these alternatively spliced forms in several tissues, with emphasis in the kidney.

Materials and Methods

Cloning of TIN1 cDNA from Total RNA

Total RNA was isolated according to well-established protocols (^42,43) from primary cultures of human tubular epithelial cells (kind gift of Dr. Youngki Kim, Department of Pediatrics, Division of Nephrology, University of Minnesota School of Medicine). The RNA was reverse-transcribed into cDNA using random hexamers. The following components were added to the reaction tube containing 2 μg (9 μl) of human tubular epithelial cell total RNA: 4 μl of 5× reverse transcriptase buffer, 1 μl of RNase inhibitor (Rnasin; Promega, Madison, WI), 2 μl of deoxynucleotides (dATP, dCTP, dGTP, dTTP), 2 μl of random primers (Promega), and 2 μl of avian myeloblastoma virus reverse transcriptase. The reaction was incubated at 42°C for 60 min, and finally at 90°C for 5 min.

PCR

PCR was performed using the GeneAmp DNA Amplification Reagent kit (Perkin-Elmer Cetus, Norwalk, CT) with primers designed against the human TIN2 cDNA. Primer sets were designed using Oligo and Amplify software programs (NBI Biotechnology, Plymouth, MN). Primers used in the amplification of TIN1 cDNA are as follows: Primer 1 (sense) 5′-CAG AGA ATG TGG ACC GGA TAT AAG TTT-3′ encompassed bases —6 to 21 in the human TIN2 cDNA; Primer 2 (sense) 5′-GCC ATT TTC CAA GGG CAA TAC TGT-3′ encompassed bases 148 to 171 in TIN2; Primer 3 (sense) 5′-TTC TGA TTG CTG TCC TGA TTA C-3′ encompassed bases 273 to 291 in TIN2; Primer 4 (sense) 5′-GGC TGT GTC ACT GAG TTC TAT GC-3′ encompassed bases 205 to 227 of TIN2; Primer 5 (antisense) 5′-ACA CTA TTG CCC TTG GAA AAT GGA-3′ encompassed bases 262 to 286 in the rabbit clone; Primer 6 (antisense) 5′-CTT TCG ATA TTT TTC TGA TTC T-3′ encompassed bases 774 to 795 in TIN2; Primer 7 (antisense) 5′-CCA TCA GAC CTG CTT GCC AT-3′ encompassed bases 538 to 558 in TIN2; Primer 8 (antisense) 5′-AGG GGT TGC TTA AAT GC-3′ encompassed bases 1034 to 1050 in TIN2. To facilitate the isolation of TIN1, cDNA clones of several PCR amplification reactions were performed using the following primer combinations: [Primer 1 + Primer 5], [Primer 1 + Primer 7], [Primer 1 + Primer 6], [Primer 2 + Primer 7], [Primer 2 + Primer 8], [Primer 3 + Primer 6], and [Primer 4 + Primer 7].

PCR was performed in a total volume of 100 μl using GeneAmp kit reagents (Perkin-Elmer Cetus). Components were added to 52 μl of diethyl pyrocarbonate-treated water containing 2 μl of reverse-transcribed human tubular epithelial cell experimental template as follows: 10 μl of 10× PCR buffer; 200 μM each of 16-μl dNTPs (dATP, dGTP, dCTP, dTTP); 2 μl/0.5 μM sense primer; 2 μl/0.5 μM antisense primer; and 16 μl of 25 mM MgCl₂ solution was added to a 2 mM final concentration. Taq DNA polymerase (2.5 U) was added before the first heat denaturation step. PCR was conducted on all samples ([Primer 1 + Primer 5], [Primer 1 + Primer 7], [Primer 1 + Primer 6], [Primer 2 + Primer 7], [Primer 2 + Primer 8], [Primer 3 + Primer 6], and [Primer 4 + Primer 7]) using a Perkin-Elmer Cetus thermocycler according to the following cycling parameters: heat denaturation at 94°C for 3 min, first five cycles, 94°C for 1 min 20 s, 50°C for 60 s, 72°C for 2 min; subsequent (³⁵) cycles, 94°C for 1 min 20 s, 55°C for 60 s, 72°C for 2 min.

PCR Product Analysis

After PCR amplification, 20-μl aliquots were size-fractionated by agarose gel (1.2%/1× Tris-acetate-ethylenediaminetetra-acetic acid) electrophoresis. PCR products were visualized with an ultraviolet transilluminator subsequent to staining of nucleic acids with Syber-Green (Molecular Probes, Eugene, OR). Bands corresponding to the amplified TIN1 cDNA were isolated and gel-purified using Gene-Clean (Bio 101, Inc., Vista, CA). Twenty-five nanograms of the final experimental template from each of the amplification reactions was ligated into PCR II (Invitrogen, San Diego, CA) cloning vector by the T/A cloning method according to the manufacturer's protocol. Competent Escherichia coli cells were transformed with the ligated constructs and plated. Drug-resistant colonies were selected for the presence of ligated product on LB agar plates supplemented with 90 μg/ml ampicillin and X-GAL (Fisher Scientific, Pittsburgh, PA). Four colonies were selected from each plate, propagated in LB media containing 50 μg/ml ampicillin, and plasmids were harvested by the alkaline lysis method as described previously (⁴⁴). Three clones from each of the TIN1 cDNA amplifications were sequenced using vector-specific (M13 forward, M13 reverse) and internally derived (Primers 1 to 8) primers by the dideoxynucleotide chain termination reaction (⁴⁴).

Cloning of TIN1 cDNA from Kidney Cortex Library

TIN1 cDNA was cloned by PCR from a human kidney library (Human Kidney 5′-STRETCH PLUS cDNA library, Clontech, Palo Alto, CA). Sense (5′-TCC ATT TTC CAA GGG CAA TAC-3′), (5′-GCT ACT TTA CCT GAA ACA ACT-3′) and antisense (5′-AGT TGT TTC AGG TAA AGT AGC-3′), (5′-GGC TTG AAC TGG TCC ATT TTG CAT GAT-3′) primer sequences were derived from the rabbit TIN-Ag sequence (GenBank accession no. U24270) (⁴⁰). Primer sets were designed using Oligo and Amplify software programs (NBI Biotechnology). Database searches were performed using the Blast Network Service of the National Center for Biotechnology Information.

PCR was performed in a GeneAmp PCR system 9600 (Perkin-Elmer). PCR was performed in a total volume of 50 μl using GeneAmp kit reagents (Perkin-Elmer). Components were added to diethyl pyrocarbonate-treated water containing 0.5 μl of Human Kidney 5′-STRETCH PLUS cDNA template (Clontech) as follows: 5 μl of 10× PCR buffer; 2 μl each of dATP, dGTP, dCTP, and dTTP; 2 μl (40 pmol) each of the sense and antisense primers; 3 μl of 25 mM MgCl₂ solution; and AmpliTaq Gold DNA polymerase (2.5 U). The following cycling parameters were used: first denaturation at 95°C for 3 min, subsequent denaturation at 94°C for 1 min, annealing at 59°C for 2 min, and extension at 72°C for 3 min, followed by 35 cycles. PCR products were size-fractionated by agarose gel (1%/1× TBE) electrophoresis with SyberGreen (Molecular Probes). Bands corresponding to the amplified TIN1 cDNA were eluted using NA 45 DEAE Cellulose Membranes (Schleicher & Schuell, Keene, NH). The final experimental template was ligated into the PCR II vector (Invitrogen) and used to transform XLI-Blue supercompetent cells. Sequencing was performed using Thermo Sequenase kit (Amersham) with M13 forward, M13 reverse, and internally derived primers. In addition, automated fluorescence sequencing was performed using dye-labeled terminators in a Strectch DNA sequencer (Applied Biosystems) in the MicroChemical Facility of Institute of Human Genetics at the University of Minnesota. The clone shown was sequenced in both directions.

Cloning of TIN2 cDNA

Human kidney libraries (Clontech; Stratagene, La Jolla, CA) cloned into λgt11 were screened for the presence of cDNA encoding TIN-Ag(s) using the previously identified rabbit TIN-Ag (TIN1) cDNA. Nucleotide sequences corresponding to the rabbit TIN-Ag SphI/EcoRI cDNA fragment were random-prime-labeled with [α-³²P]dCTP and used to screen duplicate nylon filters under high-stringency conditions (6× SSC/0.5% sodium dodecyl sulfate [SDS] twice at room temperature for 30 min each; 1× SSC/0.1% SDS for 60 min at 65°C). After three rounds of selection, two positive clones were identified and subcloned into pBluescript II SK+ at the EcoRI site. DNA sequencing of the clones was performed by the dideoxynucleotide chain termination method of Sanger et al. (⁴⁴), using Life Sciences Sequenase version 2.0 (Hialeah, FL). Both strands of the TIN2 cDNA were sequenced using a combination of vector-specific and internally generated primers.

Northern Blot Analysis

Northern blot hybridization was performed using human multiple tissue Northern blots I and II (Clontech). The membranes were prehybridized at 42°C overnight; cDNA probes were labeled using the random primer method with ³²P-dCTP (3000 Ci/mmol; Amersham Pharmacia Biotech, Piscataway, NJ) according to the manufacturer's protocol. Unincorporated nucleotides were removed by a G-50 Sephadex spin column (Boehringer Mannheim, Indianapolis, IN). Hybridization was performed in a 40% formamide solution overnight at 42°C. Hybridized blots were washed twice at room temperature for 15 min in 2× saline-sodium phosphate-ethylenediaminetetra-acetic acid (SSPE)/0.2% SDS and a third time in 0.1× SSPE/0.2% SDS at 55°C. The membrane was then exposed to a phosphorimage screen.

Northern Probes

The full-length TIN1 cDNA was used to detect TIN1 and TIN2 transcripts. A 270-bp TIN1-specific probe was generated from TIN1 plasmid DNA by PCR, as described below. The sense primer was 5′-TTACCTGCAACAACTGATC-3′, which corresponds to nucleotides 631 to 650 of Figure 1, Figure 1b, Figure 1c. The antisense primer was 5′-TCCACGTTTTCTCAGGTAC-3′, which corresponds to nucleotides 881 to 900 of Figure 1, Figure 1b, Figure 1c. These primers were selected to specifically amplify nucleotides 631 to 900, which are spliced out in TIN2. The PCR protocol consisted of a 3-min incubation at 95°C followed by 35 cycles (1 min at 95°C, 1 min at 60°C, 1.5 min at 72°C, and final extension for 15 min at 72°C). Ten-microliter aliquots were size-fractionated by agarose gel electrophoresis with SyberGreen (Molecular Probes). Bands corresponding to the amplified 270-bp cDNA were isolated and gel-purified using spin columns (Bio-Rad, Hercules, CA). The DNA sequence was confirmed as nucleotides 631 to 900 of TIN1 by automated fluorescence DNA Sequencer Stretch in the Microchemical Facility of the University of Minnesota. For positive control, a probe specific for human β-actin was used (Clontech).

Figure 1: Nucleotide and predicted amino acid sequence of the tubulointerstitial nephritis antigen-1 (TIN1) and TIN2 cDNA. Numbers indicate the last nucleotide or amino acid in each row. The putative signal peptide and propeptide are underlined. The furin cleavage site is denoted in brackets in both TIN1 and TIN2. Presumptive alternative splice sites are bracketed, and potential N-linked glycosylation sites are underlined in both predicted sequences. Conserved putative catalytic domains are indicated in parentheses.

Figure

Figure

Results

Human kidney libraries were screened for the presence of cDNA clones corresponding to the human TIN-Ag coding sequences. Nucleotide sequences corresponding to the rabbit TIN-Ag SphI/EcoRI cDNA fragment were labeled and used for screening as described in the experimental procedures. After three rounds of selection, two positive clones were identified for further analysis. Sequence analysis of the clones identified them as TIN2 mRNA species. Next, to facilitate the recovery of TIN1 cDNA clones, primers designed against the TIN2 nucleotide sequence were used to identify positive clones through PCR analysis of reverse-transcribed human tubular epithelial cell total RNA. Finally, to confirm independently the presence of TIN1, a kidney cortex library was screened by PCR, and the resulting clones were sequenced.

Sequence analysis of the TIN1 and TIN2 cDNA clones indicates the presence of 1428- and 1002-bp open reading frames, coding for proteins of 476 and 334 amino acids, respectively (Figure 1,Figure 1b,Figure 1c). The predicted amino-terminal region of both TIN1 and TIN2 proteins contains a putative signal peptide constituted by a charged N-terminal, central hydrophobic, and polar C-terminal region (Figure 1,Figure 1b,Figure 1c, sequence underlined) (⁴⁵). This sequence is similar in amino acid composition to that of the previously reported rabbit TIN1 signal peptide, placing the macromolecule in the extracellular/secretory pathway. Twenty-six amino acids lie between the carboxy terminus of the signal peptide and a consensus cleavage site (RXXR) for the furin family of proteases at amino acid positions 46 to 49 in TIN1 and TIN2 (Figure 1,Figure 1b,Figure 1c, brackets). Putative proteolytic cleavage at this site by furins (⁴⁶) could possibly generate the mature form of the TIN-Ag(s), as suggested by previous amino-terminal sequencing of purified 58- and 50-kD forms from rabbit TBM (³⁷).

Further analysis of the primary amino acid sequence of the TIN1 cDNA illustrates a characteristic cysteine-rich EGF-like motif (amino acids 119 to 169) and the presence of a uniquely conserved region within the carboxy-terminal half that shares significant homology with the cysteine proteinase family of molecules (⁴¹). Highly homologous domains thought to be important in the catalytic activity of cysteine proteinases are similarly conserved in the predicted C-terminal region of the human TIN1 sequence. However, a cysteine residue thought to be crucial for the conformation of the active site in cysteine proteinases is substituted by a serine residue. This substitution (at position 243 in the TIN1 protein) is conserved in the predicted amino acid sequences of both the rabbit and human TIN1 macromolecules, suggesting functional importance (related to the presence or absence of proteolytic activity) for this region. Additional evidence that is supportive of conserved structural elements resides in the spatial relationship between cysteine residues. Similar to the rabbit TIN-Ag, cysteines are clustered in the human TIN-Ag in two regions: between amino acids 57 to 154 and 240 to 351. Notably, the spatial relationship in the placement of cysteines is identical between the human TIN1 and the rabbit TIN-Ag and similar to that of the cysteine proteinases (^40,41).

The sequence identity between the human TIN1 and rabbit macromolecules of the TIN-Ag is 88% at the level of the nucleic acid and 85.6% between the primary amino acid sequences. Interestingly, the sequence identity in the region between the signal peptide and the putative furin cleavage site (predicted amino acids 20 to 49) is only 55%. This may signify a lack of selective pressure within this area if it is degraded following posttranslational modification of the molecule in vivo. Six potential N-linked glycosylation sites are present in TIN1 (Figure 1,Figure 1b,Figure 1c, underlined).

Similar analysis of the open reading frame of the TIN2 cDNA indicates the presence of a previously unidentified alternatively spliced TIN-Ag mRNA product. The TIN2 coding region contains identical sequences coding for several structural features such as the signal peptide (amino acids 1 to 19) and putative propeptide (amino acids 20 to 49). However, a region including the EGF-like motif (amino acids 119 to 169) and sequences conserved with the cysteine proteinase family (amino acids 209 to 299) are absent. Remarkably, the removed segments code for parts of the cysteine-rich regions in the TIN1 molecule. More specifically, five cysteines within the EGF-like region (amino acids 119 to 169) are removed and four cysteines are removed from the second cluster located within the carboxy terminus (amino acids 209 to 299) (Figure 1,Figure 1b,Figure 1c). Five of the six N-linked glycosylation sites that are in common with TIN1 are indicated (Figure 1,Figure 1b,Figure 1c, underlined). An interesting homology between TIN-Ag and von Willebrand factor (vWF) is found within the EGF-like motif. The homology is to sequences present in domains D1 and D2 of the pro-vWF and is shown in Figure 2. The conserved position of six cysteines within these regions many be of importance (see Discussion). These two regions of vWF propeptide share a 37 to 40% identity and a 51 to 60% homology. The alternate usage of the above-mentioned regions strongly suggests that differences in the function of the protein products resulting from TIN1 and TIN2 mRNA may exist.

Figure 2: Homology of TIN-Ag with two domains of pro-von Willebrand factor (vWF). For TIN-Ag, amino acids shown are 120 to 154. For pro-vWF-domain D1, amino acids shown are 350 to 384. For pro-vWF domain D2a, amino acids shown are 829 to 863. The CxxCxC in the center is obvious. Apart from the conservation of the positions of all six cysteines present, two glycine residues, an asparagine residue, and a tryptophan residue are conserved.

Two domains of interest are present close to the amino-terminal end of TIN-Ag: amino acids 64 to 74, which form the sequence EDRDDGCVTEF, and amino acids 88 to 97, which form the sequence DRENSDCCPD. These two sequences, with the position of the amino acids aspartic acid and glutamic acid, fall in the category of consensus sequences for binding to Ca²⁺ (⁴⁷). Similar sequences are present in the rabbit TIN-Ag. In view of the fact that TIN-Ag can be extracted from basement membranes by using ethylenediaminetetra-acetic acid (A. S. Charonis, unpublished observations), TIN-Ag may constitute a Ca²⁺ binding extracellular matrix glycoprotein.

Recently, a portion of the mouse TIN-Ag, corresponding to amino acids 328 to 476 of the human sequence, has been cloned (⁴⁸). Figure 3 compares the structure of rabbit, murine, and human TIN-Ag from amino acids 428 to 474. It is important to note that in this 47-amino acid stretch, rabbit and murine exhibit a 100% identity. The human sequence is 96% identical to both, and only two amino acids are different: at position 437 a serine is substituted by a phenylalanine and at position 463 a leucine is substituted by a valine.

Figure 3: Conserved 47-amino acid peptide from carboxy terminus of TIN-Ag. ^*, Rabbit TIN-Ag sequence from Nelson et al. (J Biol Chem 270: 16265-16270, 1995). ^**, Murine partial TIN-Ag sequence from Kumar et al., (Kidney Int 52: 620-627, 1997). ^***, Human TIN-Ag sequence in the present report. Rabbit and murine sequences are identical. The human sequence differs in two amino acids, one of which is a conservative substitution.

Secondary structure predictions based on the primary structure of TIN1 and TIN2 were performed using the PHD program (Profile network from HeiDelberg) proposed by Rost and Sander (⁴⁹). In this program, the prediction is performed by a system of neural networks that uses as input evolutionary information in the form of multiple sequence alignments. Only predictions of α helices, β strands, and loops with accuracy higher than 70% were accepted. In addition, especially for α helices, predictions given by the program were accepted only if they included regions with a length longer than four residues.

The secondary structure predictions for TIN1 and TIN2 as resulted from PHD are depicted in Figure 4. The sequences of both proteins start from amino acid 50, which is the initial amino acid of the mature, extracellular protein, since TIN-Ag is processed intracellularly, first by signal peptidase and then by furins. Obviously, the major difference between TIN1 and TIN2 is the presence or absence of the amino acids equivalent to the spliced-out regions. Several cysteine residues exist in these areas; furthermore, the predicted secondary structure of these domains includes three α helices (amino acids 159 to 168, 243 to 253, and 289 to 299) and two β strands (amino acids 257 to 260 and 266 to 269). It is of interest to note that even in areas where the amino acid sequence is identical, differences in secondary structure are predicted by the program. These differences include two α-helical domains in TIN1 (amino acids 99 to 104 and 159 to 168) missing from TIN2. As expected, after the second spliced-out region, starting from amino acid 300, TIN1 and TIN2 are predicted to be very similar with the exception of one small five-residue helical domain (amino acids 383 to 387) of TIN2 that is missing from TIN1.

Figure 4: Secondary structure predictions for TIN1 and TIN2. Using the PHD program (Profile network from HeiDelberg; Rost and Sander), predictions of the secondary structure of TIN1 and TIN2 are presented. TIN1 and TIN2 are aligned and the spliced-out regions are represented as gaps. Thick lines (red), α helices; arrows (blue), β strands; thin lines (black), loops. Areas not designated are impossible to predict with a high degree of certainty according to the program used.

Several human tissues were then studied to determine the expression of TIN-Ag message. In these studies, a full-length TIN1 probe was used, which is likely to detect both TIN1 and TIN2 transcripts. The results of these experiments are shown in Figure 5, A and B. Using the conditions described in Materials and Methods, TIN-Ag transcripts were detected only in the kidney (Figure 5A) and in the intestine (Figure 5B). Tissues including heart, brain, placenta, lung, liver, skeletal muscle, pancreas, spleen, thymus, prostate, testis, ovary, and peripheral blood leukocytes were negative. In both the kidney and the intestine, two forms with mobilities equivalent to 1.8 and 1.3 kb were detected. It is obvious that the ratio of the 1.8-kb/1.3-kb transcript is drastically different, at least between the kidney and the small intestine. As a positive control, a probe for β-actin was used to screen the blots. The screening was performed by stripping and reprobing, using the procedure detailed above.

Figure 5: (A and B) Northern blotting of human tissues with TIN-Ag probes. The full-length TIN1 probe was used. Tissues used are shown on top of each lane. Mobility markers are shown on the left, and detected mobilities are designated on the right. On the bottom, screening with β-actin, the positive control, is shown.

Because the probe used is likely to detect both TIN1 and TIN2 transcripts, we constructed by PCR a probe specific for the longest spliced-out region that should detect TIN1 mRNA, but should not hybridize with TIN2 mRNA. This probe was then used to examine TIN-Ag expression in kidney. The results of these experiments are shown in Figure 6. The probe hybridized only with the 1.8-kb transcript, confirming that this transcript codes for the full-length TIN-Ag isoform, the TIN1. Lack of recognition of the 1.3-kb transcript constitutes indirect but strong evidence that the 1.3-kb transcript may be equivalent to TIN2. Again, as a positive control the β-actin probe was used, by stripping and reprobing the blots.

Figure 6: Northern blotting with TIN1-specific probe. A 270-bp probe specific for TIN1 was used to screen human tissues. Mobility markers are shown on the left. The 1.8-kb mobility is shown on the right, where the only band is detected. Tissues are indicated on top of each lane. On the bottom, screening with β-actin, the positive control, is shown.

Discussion

Basement membranes are well-differentiated extracellular domains that serve several important functions. Their macromolecular constituents interact with each other and with cell surface receptors, thus forming a heteropolymeric meshwork that can influence cellular phenotype. We have recently focused our attention on a novel component of basement membranes, TIN-Ag. This basement membrane component is expressed mainly in the kidney cortex, but also has been detected in the basement membrane underlying the intestinal epithelium, the epidermis, and the cornea (³⁹). Recent cloning of the rabbit TIN-Ag established that this macromolecule is subject to significant posttranslational modifications, and its carboxyterminal part bears homology to the family of cysteine proteinases (^40,41).

In the present report, cloning of human TIN-Ag led to the important finding that alternative splicing of TIN-Ag primary gene transcript results in two mRNA species coding for structurally different protein products. It is well accepted that alternative splicing is a widespread mechanism for generation of protein diversity. Differential exon incorporation within mRNA allows genes to produce templates that code for protein products with differing structural features and consequently different functions (⁴³). Indeed, in the case of human TIN-Ag, the clones TIN1 and TIN2 presented in this report exhibit sequence differences that should lead to products drastically different in function for the following reasons. (1) Two long stretches of nucleotides (nucleotides 354 to 507 and 625 to 897) within TIN1 are absent from TIN2; one of these domains codes for an EGF repeat. (2) Missing nucleotides 625 to 897 code for a potential glycosylation site and several sequences that are homologous to cathepsins. (3) Both domains absent from TIN2 code for a number of cysteine residues. The selective removal of regions containing several cysteine residues suggests that altered disulfide bond formation might contribute significantly to differences between the protein products of these two splice variants.

EGF domains are a common structural feature of several basement membrane macromolecules, typically localized to the amino-terminal regions of laminin isoforms, entactin/nidogen, perlecan, and of BM40/SPARC. These regions are associated with several biologically active functions, including cell adhesion, mitogenic activity, and macromolecular interactions. In the case of vWF, it is established that the domains where the homology with TIN-Ag is found (⁵⁰) are involved in multimerization, because vWF lacking these domains can only dimerize but does not form multimers (⁵¹). Therefore, one can speculate that a similar function may be present in the product of TIN1 but absent from the protein encoded by TIN2. Another interesting feature shared by these two domains of pro-vWF and TIN-Ag is the presence of vicinal cysteines (⁵¹). This is an important structural arrangement, present in the catalytic site of an endoplasmic reticulum resident enzyme known as protein disulfide isomerase. Isomerase activity is also found in the hormones leutropin and folliotropin (⁵¹). Vicinal cysteines have been shown to play a role in the polymerization of vWF (⁵¹).

It has been reported previously (⁴⁰) that rabbit TIN-Ag exhibits a serine residue at amino acid position 243. The same residue is present in the deduced amino acid sequence of TIN1 clone. All members of the family of cysteine proteinases contain a cysteine residue in this position that is important for their function, because it participates in the formation of their active site. Despite this difference, the carboxy-terminal half of TIN1 and TIN2 retains a 30% identity at the amino acid level to cathepsin B. At present, the structural data do not support the possibility that TIN-Ag may retain some form of proteolytic activity. Although it has been reported that TIN-Ag has a concentration-dependent inhibitory effect on laminin polymerization and promotes dissociation of preformed laminin polymers (⁵⁰), it is not yet clear whether this results from a putative proteolytic activity or from a direct interaction. Within the region of the similarity to cathepsin B lies the stretch of 47 amino acids shown in Figure 3, which is extremely well conserved among the three species shown. It is possible that this region may serve a function different from proteolysis but still quite important in evolution.

The finding of two domains that represent a consensus sequence for Ca²⁺ binding is of potentially great significance. Recent data from several lines of investigation suggest that certain extracellular microenvironments may exist where Ca²⁺ concentrations differ and that these differences may be physiologically important (⁴⁷). Some extracellular matrix proteins contain calcium-binding domains, and it is possible that in some cases Ca²⁺ may not be used just for structure stabilization, but might serve a more important regulatory function. The amino acid sequence of TIN-Ag both in rabbit and in human contains two consensus sequences for Ca²⁺ binding. Furthermore, expression of TIN-Ag is detected almost exclusively in tissues that are heavily involved with Ca²⁺ absorption, as is the case for the proximal tubule and the intestine. More work needs to be done in this area, to confirm and quantify the ability of TIN-Ag to associate with Ca²⁺ and to assess the functional significance of any such interaction.

Northern blotting experiments confirmed the restricted expression of TIN-Ag. Both TIN1 and TIN2 were detected in kidney and intestine. An important finding is the different ratio of expression of TIN1/TIN2 in these two tissues; however, we do not yet know the functional significance of this pattern. Furthermore, we cannot exclude the possibility that TIN-Ag may be expressed in lower amounts in other tissues, but factors such as abundance and stability of the message might account for the negative results.

Western blotting of isolated TBM from several species has established the presence of two TIN-Ag forms (58 and 50 kD), both sharing complete identity within the amino-terminal region (^37,38). We have previously verified that a full-length rabbit TIN-Ag clone (which is equivalent to human TIN1) codes for the 58-kD form (⁴⁰). It is possible that TIN2 may code for the 50-kD form; however, we do not have at present direct evidence that this is the case. Our sequence analysis indicates that TIN1 and TIN2 contain complete nucleotide and predicted amino acid sequence identity with the exception of the alternatively spliced regions. Furthermore, despite extensive attempts using several primers at many locations, no other products were amplified using the PCR technique on human tubular epithelial cell RNA. However, it should be noted that other alternatively spliced forms may exist in other tissues where TIN-Ag might be expressed.

Previous studies have established that TIN-Ag interacts in a specific and saturable manner with type IV collagen and laminin and that it can influence the extent of laminin polymerization (⁵²); that TIN-Ag promotes cell adhesion for several cell types and that it is a strong adhesive component (^52,53); and that at least for proximal tubule epithelial cells, integrins α3β1 and αvβ3 can act as receptors for TIN-Ag (⁵³). The finding of this study that two alternatively spliced messages exist raises several important questions regarding the differences that these two protein forms of TIN-Ag may exhibit in their functional properties as well as the important common features. Addressing these issues will be critical in understanding the role of TIN-Ag in the pathophysiology of basement membranes.

Acknowledgments

These studies were supported by National Institutes of Health (NIH) Grant DK-36007, Karatheodori to Dr. Charonis, and NIH Grant AI-0704 to Dr. Michael.