Address correspondence and reprint requests to Isobel A. Scarisbrick, Department Physical Medicine & Rehabilitation, 428 Guggenheim Building, Mayo Clinic Rochester, 200 First St., SW., Rochester, MN 55905, USA. E-mail: Scarisbrick.Isobel@Mayo.edu
Kallikrein 6 is a serine protease expressed abundantly in normal adult human and rodent CNS, and therein is regulated by injury. In the case of CNS demyelinating disease, K6 expression in CNS occurs additionally in perivascular and parenchymal inflammatory cells suggesting a role in pathogenesis. Herein we describe two unique transcripts that occur within the human and mouse K6 genes that differ in their 5′-untranslated regions. These transcripts have identical translation initiation sites in exon 3, are expressed in a tissue-specific fashion and are differentially regulated in response to CNS injury. While the human and mouse 5′-transcripts differ in sequence they are identical in genomic organization and tissue-specific expression. The most 5′-transcript, designated transcript 1, includes exon 1–7, and was detectable in all CNS regions, but not in any non-CNS tissues examined (spleen, thymus, liver, kidney, pancreas, submandibular gland and peripheral nerve). In contrast, transcript 2 lacks exon 1, but contains a unique sequence at the 5′-end of exon 2, designated exon 2A. Transcript 2 was expressed both in CNS and in each peripheral tissue. In a murine model of human CNS demyelinating inflammatory disease induced by Theiler's picornovirus, mouse K6 transcript 1 was up-regulated in brain and spinal cord at acute and more chronic phases of CNS inflammation and demyelination, while overall transcript 2 expression was not significantly altered. However, in isolated splenocyte cultures, transcript 2 was up-regulated two-fold by cellular activation. Tissue-specific expression patterns and differential regulation in CNS disease indicates that each K6 5′-transcript is probably regulated by unique promoter elements and may serve as a molecular target to treat inflammatory demyelinating disease.
Given the dense expression of K6 in CNS and its broad substrate specificity we have sought to understand its activity in nervous system disease. The potential role of K6 in CNS demyelinating disease is of particular interest as it is normally expressed abundantly by myelin producing oligodendroglia, however, in the case of CNS demyelination K6 is expressed additionally by perivascular and parenchymal inflammatory cells (Scarisbrick et al. 2002). We have generated several additional lines of evidence, which indicate K6 is a key player within proteolytic cascades mediating inflammatory cell extravasation and demyelination. This includes studies demonstrating that components of the blood–brain-barrier and myelin proteins serve as K6 substrates (Blaber et al. 2002) and that excess K6 results in a ‘dying back’ of oligodendroglial processes (Scarisbrick et al. 2002). Moreover, we have demonstrated recently that K6-function-blocking antibodies decrease Th1 responses and result in an attenuation of clinical and histological disease in murine models of multiple sclerosis (Blaber et al. 2004).
To further our understanding of the molecular mechanisms by which K6 expression may be regulated, we have turned our attention to two different 5′-transcripts described previously within the murine K6 gene termed BSSP and BSP (Meier et al. 1999; Matsui et al. 2000). These studies are driven by the hypothesis that alternative promoters of K6 may be differentially regulated within the neural and inflammatory arms governing demyelinating disease and therefore may represent unique molecular targets for development of novel therapeutic approaches. We demonstrate for the first time herein that 5′ murine K6 transcripts are expressed in a tissue specific fashion and are differentially regulated within the CNS in a murine viral model of multiple sclerosis. Importantly, we also demonstrate that homologous 5′-transcripts are generated from the human K6 gene and that these are expressed in a parallel tissue specific fashion. Cumulatively, these data point to differential promoter usage and suggest that each variant may represent a unique molecular target for the development of therapeutic strategies to treat neural pathogenesis and inflammation characterizing CNS inflammatory demyelinating disease.
Materials and methods
All studies were carried out using adult female SJL/J (H-2S) mice obtained from The Jackson Laboratory (Bar Harbor, MA, USA). Handling of animals conformed to the guidelines of both the National Institutes of Health and the Institutional Animal Care and Use Committee at the Mayo Clinic. Unless otherwise indicated all reagents were obtained from Sigma (St. Louis, MO, USA).
RNA Ligase Mediated Rapid Amplification of cDNA Ends (RLM-RACE, Ambion, Austin, TX, USA) was used to obtain the full-length 5′-ends of human and mouse K6 transcripts found in CNS and non-CNS tissues. The 5′-end of hK6 or mK6 was amplified from human or murine brain stem, spinal cord or spleen total RNA using gene specific nested primers (hK6: 5′-CTCCGGGGATTCTTGAGTCG-3′, downstream; 5′-GACTCAGCACCACCATCAGC-3′, upstream; mK6: 5′-CAGAGGCAGA-GAAAGTCAGC-3′, downstream; 5′-CCAGCATCTTCATTGTCAGC-3′, upstream). Murine total RNA was isolated from SJL/J female mice using RNA STAT-60 (Tel.Test, Inc. Friendswood, TX, USA) and human total RNA samples were obtained commercially (Ambion, or BD Biosciences, Palo Alto, CA, USA). Gel fragments were cloned into pCR 2.1-TOPO (Invitrogen, Carlsbad, CA, USA) and sequenced on a 3730 XL DNA Analyser (Applied Biosystems, Foster City, CA, USA). Comparison with reported K6 5′-ends was made by searching the NCBI database using the Basic Local Alignment Search Tool (BLAST). Two distinct transcripts were identified in the case of both human and mouse K6. The most 5′-transcript in each case contained exon 1 and was designated transcript 1. The second transcript was designated transcript 2 and was shown to lack exon 1, but to contain a unique sequence at the 5′-end of exon 2, designated exon 2A.
As a first step to determining potential promoter elements regulating K6 expression, we have examined the 5′ 500 base pairs (bp) of nucleotide sequence upstream from the start of exon 1 in human (AF149289, bp 1501–2000) and mouse K6 (AB032402, bp 1209–1709), for known cis-acting regulatory elements using matinspector (Genomatix Software, Munich, Germany). This region was chosen as it is known that regulatory sequences are most frequently found just upstream of the transcription initiation site. The possible regulatory elements further upstream in the promoter, in the 3′-region, within intronic regions or elsewhere within the kallikrein gene complex were not examined as part of this study.
Tissue specific expression of K6 splice variants
The expression patterns of K6 transcripts 1 and 2 in mouse and human tissues were determined both by conventional PCR techniques and by real-time RT-PCR using primers specific to each transcript. To PCR amplify the two main 5′-transcripts in the hK6 gene, forward primers at the 5′-end of exon 1 (5′-TCGGCAGGCAGCACACAGAGGG-3′) or exon 2A (5′-AGAACCAGCCTCTTCCAGGGAGGC-3′) were used in conjunction with a reverse primer within exon 3 (5′-GCCATGAAGAAGCTGATGGTGGTG-3′) or exon 5 (5′-GATGTGGCAGCTGGTGGTGGTGTTGGC-3′). The forward primer specific for mK6 transcript 1 was (5′-CCACAGAGGGGCTTAC-3′) and for mK6 transcript 2 was (5′-GTTTCTGGAGACAGTCTGC-3′). A reverse primer in mK6 exon 3 common to both transcripts 1 and 2 (5′-GTGCTTGGTTCTTGCTAAA-3′) was used in each case. To control for loading parallel examination of the housekeeping gene glyceraldehyde phosphate 3-dehydrogenase (GAPDH) was assessed in each RNA sample using (5′-ACCACCATGGAGAAGGC-3′) as the forward and (5′-GGCATGGACTGTGGTCATGA-3′) as the reverse primer (Blaber et al. 2004).
The expression of mK6 transcripts 1 and 2 was examined in a wide variety of CNS regions and non-CNS peripheral tissues obtained from normal adult SJL mice (killed with 35 mg/kg, sodium pentobarbital). Whole brain, or select brain regions including the hippocampus, spinal cord, brain stem and corpus callosum were dissected on ice and snap-frozen in liquid nitrogen. Peripheral tissues were collected in a parallel fashion and included the brachial nerve plexus (peripheral nerve), spleen, thymus, kidney, liver, pancreas and submandibular gland. Organ samples were combined from 3–5 mice and total RNA isolated using RNA STAT-60. Samples of human total RNA from brain, hippocampus, spinal cord, brain stem, corpus callosum, spleen and kidney were obtained from commercial sources as outlined above. All purified RNA was quantified spectrophotometrically. Total RNA (0.5 µg) was subject to RT-PCR using the Light Cycler-RNA Amplification Kit SYBR Green I in a Roche Light Cycler apparatus (Roche Diagnostics, Mannheim, Germany) using the species and transcript-specific primers described. Standard curves in each PCR reaction, correlating copy number with amplification cycle at the crossing point, were generated using transcript specific cDNA clones of known copy number. To control for loading, copy number in each case was normalized to that for GAPDH. For comparison between different tissues, the relative level of transcript expression in each RNA sample was expressed as a percentage of that observed in the spinal cord. Amplification of each gene specific fragment was confirmed both by melting curve analysis and agarose gel electrophoresis.
Cell specific expression of mK6 transcripts
To examine cellular specificity of mK6 transcript 1 and 2 expression within the brain, RNA obtained from cell lines corresponding to murine neurons, oligodendroglia and macrophages was examined by RT-PCR. Two neuronal cell lines were examined, an immortalized murine motoneuron cell line [NSC 34, generously provided by Dr Neil Cashman (Cashman et al. 1992)] and the mouse neuroblastoma Neuro-2a cell line (CCL-131, ATCC). Both neural cell lines were grown in Dulbecco's Minimal Essential Media (DMEM) containing 10% fetal calf serum (FCS) on untreated tissue culture plastic. A mouse oligodendrocyte precursor cell line (OliNeu) was a kind gift from Dr Jacqueline Trotter and was grown on poly-L-lysine coated cell culture plastic in DMEM supplemented with 5% FCS, Horse serum (1%) and the B-27 and N2 supplements (Gibco, Invitrogen Corp., Rockville, MD, USA) (Jung et al. 1995). An immortalized murine macrophage cell line (IC21, TIB-186 from ATCC) was grown on untreated tissue culture plastic in RPMI containing 10% FCS, 1.5 g/L sodium bicarbonate, 4.5 g/L glucose, 10 mm Hepes and 1.0 mm sodium pyruvate. All media contained 50 U/mL penicillin/streptomycin and 2 mm glutamine. Cells were lysed and total RNA isolated using RNA STAT-60.
Regulation of mK6 5′-UTR expression in immune cells by cellular activation
To determine changes in mK6 transcript 1 or 2 expression in response to immune cell activation transcript levels were determined in resting and activated splenocytes, using real-time RT-PCR. Splenocytes were isolated from normal SJL mice and cultured for a period of 72 h in the presence of 10 µg/mL Concanavalin A (ConA), 10 µg/mL lipolysaccharide (LPS), or PBS alone, or were grown on tissue culture plastic coated with 10 µg/mL CD3-antibody (CD3-Ab) (Pharmingen, San Diego, CA, USA). For each experiment, splenocyte cultures were prepared from combined spleen homogenates obtained from 4 mice, and were grown in complete Click's media at 3 × 106 cells/mL in a total volume of 12 mL. Each culture condition was examined in duplicate, with each experiment being repeated twice. Total RNA was extracted from non-cultured and cultured splenocytes using RNA STAT-60 and mK6 transcripts quantified by RT-PCR as described above. Changes in gene expression were reported as percent change relative to PBS treated controls.
Regulation of K6 splice variants in TMEV-induced demyelinating disease
TMEV infection was initiated in 8-week-old female SJL/J mice by intracerebral injection of 2 × 106 plaque-forming units of the Daniel's strain of TMEV, in a 10 µL volume. To evaluate any changes in the level of mK6 transcript 1 or 2 in response to TMEV induced disease, the brain and spinal cord were obtained from mice killed using 35 mg/kg sodium pentobarbital), prior to infection, or at 7, 21, 60, 90 and 180 days postinfection (d.p.i.). The brain stem was included in samples of spinal cord. Tissues obtained from infected and control mice were snap frozen in liquid nitrogen and RNA isolated as above. mK6 transcript 1 and 2 were quantified in five individual animals per time point by real time RT-PCR.
Where data were normally distributed, the significance of differences in transcript expression was determined by one-way anova with Student–Newman–Keul's post hoc test. When data were not normally distributed anova on Ranks with Dunn's post hoc test were performed. Statistical significance was set a priori at p≤ 0.05.
Organization and structure of K6 5′-transcripts
The organization of mouse and human K6 genes and the sequence of the 5′-end of each transcript is shown in Fig. 1. RACE PCR revealed two unique transcripts present in RNA isolated from either murine or human sources that differ in their 5′-untranslated regions (UTR), designated transcript 1 and 2. The general organization of exons contributing to each transcript in mouse and human was identical, although they differed in sequence. In each case, 5′ RACE reactions indicated that transcript 1 included exons 1, 2 and 3, while in transcript 2, exon 1 was absent. In both mouse and human transcript 2, the 3′-end of intron 1 was found at the 5′-end of exon 2, followed by exon 3. We have designated this unique sequence as exon 2A and renamed exon 2 as exon 2b.
The sequences of both mouse and human 5′ RACE products matched sequence within genomic clones available at the Genebank database with accession numbers AB032402 for mK6 and AF149289 for hK6. In the case of mouse transcript 1, however, 5′ RACE revealed an additional 52 bp within the 5′-end of exon 1 not previously described (Fig. 1b) such that transcription initiation occurred at bp 1710 of AB032402. With the exception of the addition of 52 nucleotides to the 5′-end, and a G to A bp change in exon 1 the RACE sequence for mK6 transcript 1 corresponded 100% with mRNA for mouse serine protease BSP (AB015206). The organization of mK6 transcript 1 was 88 bp of exon 1, followed by 156 bp of exon 2b and 61 bp of exon 3. Using 5′ RACE PCR, we showed that mouse transcript 2 contained 60 bp of unique sequence at the 5′-end of exon 2, which corresponded to the sequence at the tail of intron 1, with transcription initiation at bp 2654 of AB032402. This unique sequence was designated exon 2A and was followed by exon 2B and exon 3 in an identical fashion to that observed in transcript 1. mK6 transcript 2 was identical in sequence to mouse mRNA for serine protease BSSP (Y18723).
The 5′-ends of hK6 transcripts 1 and 2 are shown in Fig. 1c, and their sequence can be found within genomic clone AF149289 described by Yousef et al. (1999), and correspond at 100% with D78203 and AY318867 for human neurosin and AY318869 for human kallikrein 6 precursor mRNA, respectively. hK6 transcript 1 contained 185 bp of exon 1 with transcription initiation at bp 2001 of genomic clone AF149289, followed by exons 2b and 3. Parallel to mK6, the 5′-end of hK6 transcript 2 corresponded to 176 bp seen at the 3′-end of the first intron of the hK6 gene, which we have designated exon 2A. Transcription initiation of Exon 2A in hK6 transcript 2 was at bp 2908 of genomic clone AF149289, and was followed by 52 bp of exon 2, designated exon 2b, and by 48 bp of exon 3 as seen in transcript 1. Additional minor variants of transcript 2 have been reported recently in which transcription initiation begins at bp 2894 or 2910 of genomic clone AF149289 and which were cloned from human testes or mammary tissue, respectively (Pampalakis et al. 2004). In both mouse and human, exon 3 was followed by exons 4–7 as described previously (Yousef et al. 1999; Matsui et al. 2000). The expression patterns of K6 variants lacking exon 3 (AY318870) or exon 4 (AY318868), which encode putative non-functional proteins, or those with different 3′-ends (AY457039) were not evaluated in the present study. In the tissues examined, however, variants lacking exons 3 or 4 were found to form only a very small fraction of total hK6 mRNA as judged by gel electrophoresis of PCR products.
Examination of 500 bp of nucleotide sequence upstream from the start of each transcript using matinspector revealed 10-cis-acting regulatory sequences common to both mouse and human K6 in addition to a number of unique regulatory sequences of particular interest in relation to both regulation of K6 by steroid hormones and its potential role in inflammatory disease (Table 1).
Table 1. Potential cis-acting regulatory elements in the human and mouse K6 promoter
Cis-acting regulatory elements hK6/mK6
The 500 bp immediately 5′ to the transcription initiation site of the human (AF149289, bp 1501– 2000) and mouse K6 (AB032402, bp 1209– 1709) genes was examined using matinspector (Gentomatix) for cis-acting regulatory sequences. Shown are the 10 elements common to both the human (hK6) and mouse (mK6) promoter regions (+/+) as well as elements found in either human (+/–) or mouse (–/+) of interest in terms of potential mechanisms regulating K6 expression in CNS inflammatory disease. Capital letters indicate core sequences used by the MatInspector program.
Expression of K6 splice variants in CNS and peripheral tissues
Use of primer sets specific to transcript 1 or 2 of the mouse and human K6 genes demonstrated tissue restricted expression patterns in our initial semi-quantitative PCR-analysis (Fig. 2). In both mouse and human, transcript 1 was detectable only in RNA isolated from CNS, while transcript 2 was additionally detected in immune organs, including thymus and spleen. Transcript 1 was not detectable in these same tissues nor was transcript 1 mRNA induced in murine splenocytes by cellular activation.
Detailed quantitative analysis of K6 transcript 1 and 2 expression in a wide range of human and mouse CNS regions and peripheral tissues by quantitative real-time RT-PCR confirmed our initial qualitative observations indicating that K6 5′-transcripts 1 and 2 are expressed in a tissue restricted manner (Figs 3 and 4). In both mouse and human K6 transcript 1 was expressed in all CNS regions examined, but remarkably not in any peripheral tissues including peripheral nerve, spleen, thymus, kidney, liver, pancreas and submandibular gland (Figs 2,3 and 4). K6 transcript 1 levels in both mouse and human CNS differed significantly across the neuraxis (anova, p ≤ 0.001), being most abundant in the brain stem and spinal cord, followed by samples of total brain, corpus callosum and hippocampus (Figs 3a and 4a). In human brain, K6 transcript 1 mRNA was approximately three-fold higher in brain stem compared to samples of whole brain, in mouse this difference was approximately 18-fold. In both mouse and human RNA samples, levels of K6 transcript 1 were significantly higher in brain stem compared to corpus callosum, total brain and hippocampus. Also, levels of mouse and human K6 transcript 1 were significantly higher in spinal cord compared to the hippocampus (Dunn's Method, p < 0.05).
In contrast to K6 transcript 1, transcript 2 was readily detected in all human and mouse RNA samples examined whether derived from CNS or peripheral tissue sources. In general, transcript 2 exhibited a more uniform distribution across the different brain regions examined compared to transcript 1. In mouse, transcript 2 levels were highest in spleen and corpus callosum, followed by spinal cord, submandibular gland, pancreas, liver, thymus, brain stem, hippocampus, total brain and peripheral nerve. Levels of K6 transcript 2 were significantly higher in the spleen compared to samples of total brain, hippocampus, peripheral nerve and kidney (anovap = 0.003, Student–Newman–Keul's, p < 0.05). Levels were ≈ two-fold higher in spleen, corpus callosum and spinal cord compared to RNA samples of whole brain. In human tissue samples examined, transcript 2 mRNA was most abundant in corpus callosum and brain stem, followed by spinal cord, total brain, hippocampus, spleen and kidney. These differences were statistically significant between corpus callosum compared to kidney and spleen and between brain stem and kidney (anova on Ranks, p < 0.001, Dunn's Method p < 0.05). Overall, both mouse and human transcript 2 levels were ≈ 1.5-fold higher in corpus callosum and brain stem relative to samples of total brain.
As an initial approach to address the identity of cells expressing K6 transcript 1 and 2 in whole brain and spinal cord, we examined expression of each in murine neuronal, oligodendroglial and macrophage (microglial) cell lines (Fig. 5). Transcript 2 mRNA was readily detected in all three cell types, while transcript 1 was only detected in oligodendroglia. The oligodendroglial cell line examined expressed transcript 1 at ≈ 15% of the level detected in the spinal cord. Transcript 2 was detected in each neuronal cell line examined with levels reaching 170% of that detected in the spinal cord in the NSc 34 motoneuron cell line but only 12% of the spinal cord level in N-2 A cells. In both the oligodendroglial cell line and IC21 macrophages the level of transcript 2 was just less than 50% of the level detectable in the spinal cord.
Regulation of K6 splice variants in activated immune cells
Given the abundant expression of K6 by inflammatory cells in CNS demyelinating disease (Scarisbrick et al. 2002; Blaber et al. 2004) we determined whether immune cell activation alters mK6 transcript 1 or 2 expression levels. mK6 transcripts were examined in resting and activated mouse splenocytes by quantitative real-time RT-PCR. Transcript 2 was readily detected in RNA obtained from all splenocyte preparations and was up-regulated by two- to 2.5-fold in response to 72 h exposure to Con A or CD3 receptor cross-linking (Fig. 6). In contrast, transcript 1 was not detectable in the same RNA samples, even following cellular activation (Fig. 2).
Regulation of K6 splice variants in inflammatory demyelinating disease
To determine the potential role of K6 in the pathogenesis of viral induced demyelinating disease, and the relative contributions of transcripts 1 and 2, we quantified the expression of each in the brain and spinal cord of control mice and mice at 7, 21, 60, 90 and 180 day post-TMEV infection (d.p.i.) (Fig. 7). Primers specific to each transcript produced strikingly different patterns of K6 expression across the different time points examined. Most notably statistically significant changes in gene expression in response to TMEV-induced disease were observed only in the case of K6 transcript 1 and increases in expression were observed in both the brain and spinal cord (anova, p ≤ 0.001). Minor elevations in transcript 2 mRNA seen acutely in brain did not reach the level of statistical significance.
In the brain, K6 transcript 1 mRNA was elevated acutely, that is by 7 d.p.i., and this elevation persisted throughout the remainder of the disease course examined. The highest levels of expression of K6 transcript 1 in brain occurred at 7 and 21 d.p.i. when levels were increased 2.5-fold compared to controls (Student–Newman–Keul's, p < 0.05). In more chronically infected mice, that is from 60 to 180 d.p.i., transcript 1 remained significantly elevated over controls, ≈ 1.7-fold, a level which was significantly reduced from that observed at more acute phases of disease pathogenesis (Student–Newman–Keul's, p < 0.05). Paralleling changes in transcript 1 expression within brain, at acute stages of disease in spinal cord a two-fold increase in K6 transcript 1 mRNA levels was observed (Student–Newman–Keul's, p < 0.05). In contrast to the decline in transcript 1 expression observed in brain by 60 d.p.i. relative to more acute stages, in the cord K6 transcript 1 mRNA levels remained significantly elevated out to the 90 day time-point, but by 180 d.p.i. had returned to control levels.
As K6 is a trypsin-like serine protease abundantly expressed in CNS and implicated in neuropathogenesis, delineation of K6 expression patterns and regulation is crucial to understanding potential pathogenic mechanisms. We demonstrate two unique variants that differ in their 5′-UTR and are transcribed from the human and mouse K6 genes. We show that these transcripts are differentially expressed across regions of the CNS, between CNS and non-CNS tissues, within different cell types, and in response to TMEV-induced CNS inflammatory demyelinating disease. These findings suggest distinct promoter elements are responsible for tissue-restricted expression and regulation of K6 mRNA and lend further support to the hypothesis that this degradative enzyme plays an integral role in CNS inflammatory demyelinating disease.
Multiple K6 transcripts
The different mouse and human 5′-UTRs are the result of alternative transcription initiation sites. As each transcript has the same open reading frame however, they probably utilize the same translation initiation codon located in exon 3, and therefore the 5′ differences described herein would not be predicted to alter the protein generated. Comparison of the different mouse and human 5′ UTR indicates considerable divergence in sequence but a high degree of conservation in the pattern of intron/exon usage. In each case, transcript 1 includes exons 1–7 while transcript 2 lacks exon 1, with transcription initiation occurring within the first intron of the K6 gene, which we term exon 2A, followed by exons 2–7. In the case of human K6 transcript 2, a recent report by Pampalakis et al. (2004), indicates that in addition to the transcription start site described herein, there are at least two additional sites which begin 14 bp upstream or 2 bp downstream from that described in the present manuscript. Combined with multiple EST's or partial sequences displaying homology to transcript 2 (GenBank accession nos BC015525, BG468256, BM763868, BM763657 and BM761360), it is likely that several minor variants of this transcript exist which will be important to examine in terms of tissue distribution and regulation in CNS inflammatory disease in future studies.
The significance of multiple K6 transcripts which differ in their 5′-UTR may relate in part to a level of regulatory control which is directly related to, or conferred by, the different 5′-ends. For example, generation of multiple 5′-transcripts from a single gene is known to result from alternative promoter usage (Landry et al. 2003), which in the case of K6 appears to result in unique tissue and injury specific patterns of expression. Additionally, 5′-UTRs have been demonstrated to affect translational efficiency, mRNA stability, subcellular localization and/or may themselves contain unique promoter elements (Landry et al. 2003). Thus, in addition to regulation of K6 by cellular activity, autolysis, pro-peptide cleavage and endogenous serpin inhibitors (Scarisbrick et al. 1997; Bernett et al. 2002; Blaber et al. 2002; Magklara et al. 2003; Blaber et al. 2004), multiple K6 5′-UTRs are expected to participate in regulating the availability, and ultimately the activity, of this unique and degradative enzyme.
The generation of multiple mRNA transcripts from the same gene is common among the kallikreins. In this family of 15 genes, examples of multiple transcripts arising from alternative splicing, retained intronic segments, exon deletion or utilization of alternative transcription initiation sites have all been described (Clements 1997). Notably, like K3 (prostate specific antigen) levels of several of the newly identified kallikreins are altered in biological fluids serving as biomarkers or possibly having prognostic value in certain types of cancer (Diamandis et al. 2000a). Several reports indicate that kallikrein splice variants may also be useful diagnostically. Distinct forms of K3 (Henttu et al. 1990; Baffa et al. 1996), K4 (Dong et al. 2001), K5, K7 (Dong et al. 2003) and K11 (Nakamura et al. 2001) are expressed in malignant tumor cells. In the case of K5, the presence of different 5′-UTRs has been demonstrated in ovarian cancer cells pointing to selective use of different promoters in a tumor specific fashion (Dong et al. 2003). As altered levels of K6 have been reported in breast and ovarian carcinoma (Diamandis et al. 2000c; Diamandis et al. 2003; Sauter et al. 2004) examination of the different K6 5′-UTRs in malignancy will be of considerable interest.
Delineation of the full 5′-end of human and mouse K6 mRNA allowed a preliminary examination of potential promoter elements regulating K6 expression through an examination of the adjacent 5′ 500 bp of nucleotide sequence for known cis-acting regulatory sequences. This analysis revealed 10 cis-acting regulatory sequences common to both mouse and human K6 as shown in Table 1. In addition, several cis-acting regulatory elements potentially associated with the proposed role of K6 in CNS neuroinflammatory disease (Scarisbrick et al. 2002; Blaber et al. 2004) were identified. Of particular interest was the identification of immune modifying regulatory sequences such as B-cell-specific activating protein, serum responsive factor, STAT6 and STAT1. Interestingly, K6 is among the kallikreins known to be regulated by steroid hormones and several hormone related response elements were identified including sterol regulatory element binding protein 1 and 2, progesterone receptor binding site and cAMP-response element-binding protein.
Tissue specific expression
In both mouse and human, K6 transcripts 1 and 2 were associated with strikingly different tissue specific expression patterns supporting the existence of unique promoter elements. In the tissues examined, transcript 1 expression was restricted to CNS, while transcript 2 was readily detectable in both CNS and peripheral tissues. From our previous studies of K6 RNA and protein, we know this trypsin-like enzyme is preferentially expressed in brain relative to most peripheral tissues and within normal adult brain is densely produced by both neurons and glia (Scarisbrick et al. 1997, 2000, 2001, 2002). Under normal physiological conditions in human and rodent, we have further shown that K6 predominates in a subpopulation of CNS glia, that is the oligodendroglia, with a relative absence in astrocytes and microglia (Scarisbrick et al. 2000). In terms of overall distribution within the CNS, K6 mRNA is more heavily represented in the spinal cord and brain stem relative to other brain regions and to samples of whole brain (Scarisbrick et al. 1997, 2001). This pattern of expression was completely recapitulated in the expression patterns of K6 transcript 1 and 2 observed in human and mouse tissues. Within the CNS, by far the highest levels of each transcript were detected in areas containing the largest relative volume of white matter that is the brain stem, spinal cord and corpus callosum. Moreover, expression analysis of each transcript in neuronal, oligodendroglial and macrophage cell lines demonstrated that while oligodendroglia express both transcripts, in neurons and macrophages only transcript 2 was detectable. Thus elevated levels of K6 and K6 5′-transcripts in regions rich in white matter is probably a reflection of the dense expression of this enzyme by oligodendroglia.
In addition to robust expression of K6 by myelin producing cells of the CNS we have demonstrated previously, K6 protein and mRNA in corresponding cells of the peripheral nervous system, that is the Schwann cell (Scarisbrick et al. 2001). Notably, however, while oligodendroglia produce both transcript 1 and 2, in RNA preparations of peripheral nerve only transcript 2 was detectable. The selective expression of both transcripts in oligodendroglia supports the probable role(s) of K6 in oligodendroglial biology and raises important questions regarding the potentially unique role of transcript 1 in this cell type. Determining the role of each transcript in the development of the oligodendroglial lineage, myelin production, demyelination and remyelination is a goal of ongoing studies.
In addition to high levels of expression in CNS regions rich in white matter, both K6 transcripts were detected in all other brain regions examined, with transcript 2 being most uniformly distributed. Considerably lower levels of transcript 1 were observed in RNA samples of hippocampus and total brain. The abundance of transcript 2 across the neuraxis suggests it may be abundant in neurons as well as glia. Supporting this, transcript 2 was detected in both neuronal cell lines examined, that is NSc34 motoneruons and N-2 A cells, in addition to the oligodendrocyte and macrophage cell lines. While these results will need to be confirmed by analysis of primary cell lines and/or by in situ hybridization, the expression of transcript 2 in diverse cell types is indicative of possible parallel functional roles.
There are two previous reports regarding sequence variants in the 5′-end of transcripts derived from the mouse K6 gene which were termed BSP (Matsui et al. 2000) and BSSP (Meier et al. 1999) and which correspond in part to mK6 transcript 1 and 2 described herein. In addition to determining that two different 5′-transcripts are also generated from the human K6 gene, the present studies extend initial reports of murine 5′-transcripts. First, using RACE PCR we have characterized the complete sequence of each 5′ mK6 transcript showing transcript 1 to be considerably longer at its 5′-end than previously recognized. Secondly, analysis of tissue specific expression of each transcript in previous studies employed a probe for Northern analysis covering exons downstream of, and not including the 5′ UTR. By utilizing primers specific to each unique 5′ K6 transcript in sensitive RT-PCR assays we have been able to demonstrate unique tissue specific expression patterns in both mouse and human. In the range of tissues examined, transcript 1 (BSP) expression was restricted to brain while transcript 2 (BSSP) was expressed not only in brain but also in a wide range of peripheral tissues.
Regulated expression of 5′-transcripts in demyelinating disease
The unique K6 5′-transcripts were found to be differentially regulated in response to TMEV induced CNS-demyelinating disease supporting the idea that K6 expression is regulated by unique promoter elements and that K6 is a key effector molecule mediating inflammatory pathogenesis (Scarisbrick et al. 2002; Blaber et al. 2004). Striking elevations in mK6 mRNA were observed only in the case of transcript 1 and were apparent from the earliest to more chronic time-points examined. Peak elevations in brain and spinal cord correlated with viral infection, encephalitis and acute neuronal infection (Dal Canto and Lipton 1975), while decreased expression at more chronic stages may relate to progressive demyelination and the accumulation of neurological deficits, largely due to axonal injury (McGavern et al. 1999; Sathornsumetee et al. 2000; Ure and Rodriguez 2000). In addition to CNS injury related factors, it is probable that K6 mRNA expression is regulated by cellular activity (Scarisbrick et al. 1997), such that in the case of paralysis and neuronal inactivity caused by viral infection, K6 expression becomes down-regulated.
Given the preferential expression of mK6 transcript 1 within oligodendroglia, we suspect that increased transcript 1 mRNA in the brain and spinal cord in response to TMEV infection was at least in part a result of specific increases within injured oligodendrocytes, although we cannot exclude contributions from other cell types. Notably, while virus is cleared from neurons in susceptible stains of mice (Dal Canto and Lipton 1982), virus antigen and RNA persists in oligodendrocytes (Rodriguez et al. 1983; Aubert et al. 1987), microglia (Dal Canto and Lipton 1982; Clatch et al. 1990; Lipton 1995), and astrocytes (Aubert et al. 1987). Despite the fact that overall quantification of mK6 transcript 2 levels did not reveal significant increases or decreases in gene expression over the course of disease examined, it is possible that differential cell-specific changes were induced, masking gene expression changes in tissue homogenates. For example, we know that activated immune cells, such as those seen at sites of inflammation in human and murine CNS lesions express high levels of K6 protein and mRNA (Scarisbrick et al. 2002). Moreover, results of the present study demonstrate that immune cell activation results in more than a two-fold increase in K6 transcript 2 mRNA. Hybridization of transcript specific mRNA in situ will be necessary to delineate the full scope of cell-specific changes in K6 5′-transcript expression in response to CNS demyelinating disease.
We have demonstrated previously that K6 function blocking antibodies attenuate CNS inflammation and disease in both TMEV and experimental autoimmune encephalomyelitis induced disease (Blaber et al. 2004). Together with the present results these findings strongly support the idea that K6 is a key effector molecule mediating CNS-inflammatory demyelinating disease. The development of effective strategies to modulate K6 activity is therefore of considerable importance. The presence of tissue and cell specific 5′-UTR generated from both the human and murine K6 genes represents an important potential target for regulation of K6 activity at the transcriptional level. We anticipate that future examination of the cell specific expression patterns of each 5′-transcript in the normal and injured CNS, identification of promoter elements responsible in each case and the effects on disease course of modulating transcript expression will further elucidate not only the potential physiological function of K6 but also delineate this enzyme as a therapeutic target for disease attenuation.
This research was supported by PP 0964 and RG 3367-A-2 (IAS) and RG 3406-A-2 (MB) from the National Multiple Sclerosis Society, by the Craig H. Neilsen Foundation (IAS) and by P01-NS38469 (MR) from the National Institutes of Health. GPC was the recipient of a Mayo Clinic Summer Undergraduate Research Fellowship.