Complex alternative splicing of the hKLK3 gene coding for the tumor marker PSA (prostate-specific-antigen) Nathalie Heuze ´ -Vourc’h, Vale ´ rie Leblond and Yves Courty Laboratoire d’Enzymologie et Chimie des Prote ´ ines, EMI-U 0010, Universite ´ F. Rabelais, Tours, France PSA (prostate-specific antigen), the most useful serum marker for prostate cancer, is encoded by the hKLK3 gene and is present in the serum as a mixture of several molecular species. This work was performed to identify the hKLK3 transcripts in order to determine how many proteins resembling PSA are synthesized from the hKLK3 gene and secreted in blood. Combined Northern blotting, molecular cloning and database searching showed that the hKLK3 gene produces at least 15 transcripts ranging in size from 0.7 to 6.1 kb. Polysomal distribution analysis revealed that the transcripts shorter than 3.1 kb are efficiently translated in prostate cell line. A total of 12 hKLK3 transcripts have been completely or partially cloned. They result from alternative splicing or/and alternative polyadenylation involving com- plex regulation. They code for eight proteins: PSA, a trun- cated form of PSA (PSA-Tr), five PSA variants (PSA-RPs) and one protein (PSA-LM) unrelated to PSA. Using a spe- cific antibody, we detected the PSA-RP2 variant in prostate tissue. All the variants share the same signal peptide and could contribute to the diversity of hKLK3 proteins in prostate fluid and blood. Keywords: alternative mRNA; PSA variant; tumor marker; prostate cancer. PSA (prostate-specific antigen) is encoded by the hKLK3 gene, which belongs to the tissue kallikrein gene family located at chromosome locus 19q13.3–19q13.4 [1,2]. PSA (also named hK3) is a serine protease abundantly produced by human prostate epithelial cells. This protein is secreted into the lumina of prostate ducts and is present at very high concentrations in the seminal plasma (reviewed in [3]). PSA hydrolyses semenogelins I and II, resulting in liquefaction of the seminal plasma clot after ejaculation [4]. Although it seems modulating the proliferation of normal and malig- nant cells and the angiogenesis [5–8], the role of PSA in prostate pathologies remains unclear. PSA is presently considered to be the best available marker of prostate tumors, and is widely used for screening, diagnosing and monitoring prostate cancer (PCa) [9,10]. Nevertheless, concentrations of PSA below 10 ngÆmL )1 do not distinguish between Pca and benign prostatic hyper- plasia (BPH). Various molecular forms of PSA are present in the serum, some of them being complexed with serine- protease inhibitors while the others are uncomplexed or free [11]. It is important to identify each of the free forms, as the proportions of some of them differ in BPH and in cancer [12]. It has been recently demonstrated that some of the free forms are produced by proteolysis of proPSA [13] or mature PSA [14]. Some of the others could be produced by alternative splicing [15]. Alternative splicing is the most widely mechanism used to enhance protein diversity, and could affect the product of over 35% of human genes. Multiple hKLK3 transcripts were detected by Northern blot analysis [16,17], but most investigations have focused on PSA produced from the major 1.6 kb mRNA. This work was performed to identify the numerous hKLK3 transcripts, and then determine how many proteins resembling PSA can be synthesized by the hKLK3 gene. This report describes the complete or partial characterization of 12 hKLK3 transcripts produced by multiple splicing or polyadenylation. They code for at least eight proteins. Some of them are variants of PSA and appear to be good candidates for identifying free circulating species. Materials and methods Samples and RNA isolation The LNCaP cell line (American Type Culture Collection, ATCC CRL-1740) was derived from human metastatic adenocarcinoma of the prostate. Cells were grown in RPMI-1640 (Life Technologies SARL, Cergy Pontoise, France) supplemented with 5% (w/v) fetal bovine serum, 100 UÆmL penicillin/streptomycin, 2 m M glutamine in the presence of the synthetic androgen R1881 (0.1 n M ;NEN- Dupont, Les Ulis, France) [15,18]. Tumor specimens were obtained with informed consent from patients undergoing transurethral prostatectomy. Total and poly(A) RNA were prepared as previously described [18]. Normal prostate total RNA was from BD Clontech (Palo Alto, CA, USA). Correspondence to Y. Courty, Laboratoire d’Enzymologie et Chimie des Prote ´ ines, EMI-U 0010, 2 bis bd Tonnelle ´ , 37032 Tours, France. Fax: + 33 2 47 36 60 46, Tel.: + 33 2 47 36 60 50, E-mail: courty@univ-tours.fr Abbreviations: BPH, benign prostate hyperplasia; hK or hKLK, human kallikrein; Pca, prostate cancer; pISE, putative intron splicing enhancer; PSA, prostate-specific antigen; PSA-LM, PSA-linked molecule; PSA-RP, PSA-related protein; PSA-Tr, PSA truncated; CAPS, [cyclohexylamino]-1-propanesulfonic acid. (Received 18 October 2002, revised 6 December 2002, accepted 11 December 2002) Eur. J. Biochem. 270, 706–714 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03425.x Polysomal RNA preparation LNCaP cells (5 · 10 8 ) were cultured for four days in the conditions described above and centrifuged at 1000 g for 10 min at room temperature. Cells collected on ice were diluted in 1 mL of a buffer (50 m M Tris/HCl pH 8.0, 250 m M KCl, 5 m M MgCl 2 ) containing 250 m M sucrose, 2m M dithiothreitol, and 3 mg of yeast total RNA (Roche Diagnostics, Meylan, France). The cells were dounce homo- genized with 10 strokes of a type B pestle and centrifuged at 10 000 g for 10 min at 2 °C. The supernatant was supple- mented with 2% Triton X100 and 0.2 mgÆmL )1 of heparin and incubated on ice for 10 min. The supernatant was layered on top of a 10–50% sucrose gradient and centri- fuged at 40 000 r.p.m. for 50 min at 2 °CinanL5 ultracentrifuge (Beckman) equipped with an SW41 rotor. Samples of 500 lL were collected from the sucrose gradient, and 250 m M EDTA and 0.5% SDS were added to each fraction. RNA was then purified using 500 lL phenol/ chloroform (1 : 1, v/v) and ethanol precipitated. The pellets were dissolved in DEPC-treated H 2 O and stored at )70 °C. Spectrophotometric RNA quantification was performed on an aliquot of each sample. Probes and hybridization A 42-base 5¢-biotinylated oligonucleotide corresponding to a part of exon 2 of the hKLK3 gene (position 1760–1801, EMBL X14810) was used as template to synthesize an antisense [a 32 P]-labeled probe using the Klenow fragment of Escherichia coli DNA polymerase I. After heat denatura- tion, the biotinylated unlabeled strand was captured using Streptavidin MagneSphereÒ Paramagnetic Particles (Promega Corp., Madison, WI, USA). The labeled strand was recovered for Northern blot hybridization. Various hKLK3 gene fragments were obtained from the LNCaP cDNA library by PCR amplification using Pro-HA DNA polymerase (1.25 U, Eurogentec, Seraing, Belgium). The PCR reactions involved heating at 94 °Cfor2minand 30 cycles of 94 °C for 30 s, annealing temperature (Table 1) for 30 s and 75 °C for 1.5 min. The resulting fragments were purified using the WizardÒ PCR preps DNA purifi- cation system (Promega Corp.) and 50–100 ng were labeled with [a 32 P]dCTP by random priming. Northern blot hybridization was performed overnight at 68 °C with the QuikHyb Hybridization solution (Strata- gene,LaJolla,CA,USA).Blotswerewashedat68°Cfor 2 · 30minin2· NaCl/Cit, 0.1% SDS and 2 · 20 min in 0.1 · NaCl/Cit, 0.1% SDS. Membranes were then exposed to Kodak AR X-ray film at )70 °C using intensifying screens from 4 h to 6 days. Rapid amplification of cDNA ends and DNA sequencing The hKLK3 cDNA clones were obtained by 5¢ and/or 3¢ rapid amplification of cDNA ends (RACE) using the Marathon cDNA Amplification Kit (BD Clontech) Mara- thon cDNAs were generated from LNCaP poly(A) RNA [15], and from tissular poly(A) RNA according to the manufacturer’s instructions. RACE-PCR was carried out with an hKLK3-specific primer (K3-PCR2: 5¢-CAC CCGGAGAGCTGTGTCACC-3¢) based on a sequence just downstream of the transcription initiation site of the hKLK3 gene and the Marathon adaptor primer 1 (AP1) using the Expand Long Template PCR System (Roche Diagnostics). The thermocycling protocol was: denatura- tion at 94 °C for 2 min; 5 cycles of denaturation at 94 °Cfor 30 s, annealing and elongation at 72 °C for 3.30 min; 5 cycles of denaturation at 94 °C for 30 s, annealing and elongation at 70 °C for 3.30 min; 25 cycles, 94 °Cfor30s, Table 1. Primers used for PCR. Localization of the primers (intron/exon) refers to the structure of the major transcript. Primer pair Localization Primer sequence (5¢fi3¢) PCR product size (bp) Annealing temperature for PCR (°C) Probes K3-540 Intron 1 AACCCAGCACCCCAGCCCAGACAG 610 65 K3-1100 CAGAGAGCTGGCAGTTGTGGCTGG K3-2016 Intron 2 CATGGCTGCCTGGGTCTCCATCTG 286 65 K3-2300 CAGGCCCATCCCGTGCGTGTG K3-3500 Exon 3 AAGCGTGATCTTGCTGGGTCGGCA 287 65 K3-3757 CACTCCTCTGGTTCAATGCTGCCC K3-int3 Intron 3 GAGTGTACGCCTGGGCCAGATG 118 55 K3-0.7rev2 TGGACCCCACCTGGTTCCTTGG K3-5016 Intron 4 TGCCGATGGTCCTCCAT 346 50 K3-1.7rev CTATCTTTCAGACCTGGACAGGC RT-PCR K3-PCR2 with Exon 1 CACCCGGAGAGCTGTGTCACC K3-1.5 Exon 5 GGGGTTGGCCACGATGGTGTC 798 (PSA) 68 K3-0.7rev2 Intron 3 TGGACCCCACCTGGTTCCTTGG 622 (PSA-RP2) 68 K3-5055 Intron 4 GACACCTCCTCTCCAGGGCAC 731 (PSA-RP1) 68 K3-MU1 Exon3alt CTCCTCTGGTTCAATGCTGGAG 352 (PSA-RP4) 68 SSI (with K3-1.5) Exon2-Exon3 CTGCCCACTGCATCAGGAAGC 463 (PSA-RP3) 68 K3-Ex1 with Exon 1 TTACCACCTGCACCC 532 (PSA-RP5) 51 K3-sp5 Exon4-Intron4 GGTCAAGAACTCCTCTG 53 Ó FEBS 2003 Complex splicing of hKLK3 (Eur. J. Biochem. 270) 707 68 °C for an initial duration of 3.30 min and an automatic increment of 20 s at each cycle. The cDNA encoding PSA- RP3 was obtained using a 5¢-anda3¢-RACE performed with the following primer pairs: AP1 and SSI-rev (5¢- TGGAGTCATCACCTGGCTTCC-3¢), and AP1 and SSI (5¢-CTGCCCACTGCATCAGGAAGC-3¢). Amplified products were cloned into a pCR 3.1 vector and trans- formed TOP10F¢ competent cells (Invitrogen, Breda, the Netherlands). DNA was sequenced on both strands with an automated sequencer (ABI prism DNA 377 sequencer, Perkin Elmer). Expression analysis of splice variants Expression of splice variants was analyzed in prostate samples by RT-PCR. cDNA was synthesized from 5 lg total RNA using SuperScript II reverse transcriptase (Invitrogen) according to the manufacturer’s instructions. PCR was performed using specific primers (Table 1) with the following cycling conditions: 94 °Cfor3minand35 cycles at 94 °C for 30 s, 68 °Cor55°C for 30 s and 72 °C for 75 s. The products were electrophoresed on 1% (w/v) agarose gels and visualized by ethidium bromide staining. DNA corresponding to the major PCR product was extracted from the agarose gel and sequenced. Production of polyclonal peptide antibodies and protein analysis A PSA-RP2 oligopeptide corresponding to amino acids 165–180 of the putative prepro PSA-RP2 was synthesized and purified by high-performance liquid chromatography. The peptide was conjugated with BSA and used to immunize rabbits. The anti PSA-RP2 Ig was purified by a recombinant PSA-RP2 peptide-affinity column. Cancer prostate tissue (100 mg) was pulverized in liquid nitrogen to a fine powder, 1.5 mL of TRIzol reagent (Life Technologies SARL) added and the proteins extrac- ted according to the manufacturer’s conditions. Recom- binant PSA-RP2 was from a cytosolic extract of CHO cells (Chinese hamster ovary cell line, ATCC CCL61) stably transformed with an expression vector containing the entire sequences encoding prepro-PSA-RP2 [18]. Proteins were separated by SDS/PAGE on a 12% gel under reducing conditions and electrotransferred to a poly(vinylidene difluoride) membrane (Millipore Corp., Bedford, MA, USA) in a [cyclohexylamino]-1-propane- sulfonic acid (CAPS) buffer (Sigma-Aldrich Corp., St Louis, MI, USA) [18]. ECL Western analyses were carried out following the supplier’s instructions (Amersham Life Sciences, Les Ullis, France) using the anti-RP2 antibody described earlier. The second antibody was peroxidase-conjugated mouse antirabbit immunoglobulins (Sigma-Aldrich Corp.). Results Expression pattern of the hKLK3 gene in prostate Prostate tissue contains a major 1.6-kb transcript (K3a) that encodes hK3/PSA (Fig. 1A). Several other hybridization bands in the 6.1–0.6 kb range were detected with the exon 2 probe, albeit at much lower levels. Poly(A) RNA gave a similar pattern (not shown). The larger RNA bands could correspond to incompletely processed mRNA. This is supported by the hybridization pattern obtained with intronic probes (Fig. 1B). In addition to the larger tran- scripts, these probes revealed faint bands (transcripts K3k and K3e, Fig. 3) in the 1.4–1.65 kb range, plus the 0.9 kb band (transcript K3f) previously detected with the exonic probe. Thus, retention of multiple intronic sequences occurs in prostate tissue. In contrast, several bands shorter than the major mRNA were not detected with the intron-specific probes. This suggests that their varying lengths arise from alternative splicing or from the use of different poly(A) signals which shorten the exons. Sequence analysis of the K3b and K3 h transcripts (Fig. 3) supports this interpret- ation. The expression of hKLK3 in tissues and in LNCaP cell line differs in two major points. No short processed transcripts were found in the LNCaP cells (not shown) and the transcripts in the 1.9–2.1 kb range were less abundant (Fig. 1C). RNAs were purified from LNCaP polyribosomes to analyze the association of hKLK3 transcripts with ribo- somal and nonribosomal fractions, corresponding to trans- lationally active and inactive mRNA, respectively. As shown in Fig. 2, the major transcript (K3a) encoding hK3/PSA was mainly associated with fractions containing polyribosomes. Similar distribution was found for the Fig. 1. Expression of the hKLK3 gene in prostate tissue (A and B) and in LNCaP cells (C). Total RNA was analyzed by Northern blotting and hybridized with probes derived from exon 2 (A), or from introns 1, 3 or 4 (B and C). Autoradiography was performed for 4 h (A) or 4 days (B and C). The positions of the probes used are given in Fig. 3. The sizes of the bands are indicated. The correspondence between the bands and the cloned transcripts (lower case) was based on the length of these transcripts determined by molecular sequencing (Fig. 3), plus a poly(A) tail of about 200–250 bp and on their ability or not to hybridize with the probes. 708 N. Heuze ´ -Vourc’h et al. (Eur. J. Biochem. 270) Ó FEBS 2003 transcripts corresponding to the 0.9 (transcript K3f), 1.65 (transcript K3e; not shown), 2.1 (transcript K3c) and 3.1 kb bands, suggesting that these mRNA are efficiently transla- ted in LNCaP cells. In contrast, the transcripts larger than 3.1 kb were mainly detected in the low density fractions containing free, monosomal and small polysomal RNA and would be thus poorly translated. Structure of hKLK3 transcripts in the prostate As the molecular cloning of the major transcript (K3a, Fig. 3) encoding PSA, various alternative hKLK3 mRNAs have been described [15,16,18–20]. Figure 3 shows their schematic structure. The K3c-d transcripts retain part of the intron 4 while the K3e-f transcripts retain the intron 3 [15,16,18]. In 2000, Tanaka et al. [19] described a partial copy of a new hKLK3 transcript (K3g) with an alternative splicing site at the beginning of the exon 3. We obtained the 3¢ lacking part of this mRNA by 3¢ RACE-PCR. As shown in Fig. 3, the 3¢ end of this transcript (K3g) was identical to the 3¢ end of the major transcript (K3a). Finally, two transcripts with intronic sequences adjacent to the first exon were recently described [20]. The former one is a transcript containing the entire sequence of intron 1 (K3j, Fig. 3) while the second one derived from an alternative splicing within intron 1 (K3k). We amplified hKLK3 cDNAs by RACE-PCR to examine the structure of short processed transcripts. PCR products were fractionated on an agarose gel then cloned. The clones YC140405.00, YC171105.00 Fig. 2. Polysomal distribution of the hKLK3 transcripts. Polysomes were fractionated on a sucrose gradient. Aliquots (20 lg) of total RNA from each fraction were hybridized to a probe derived from exon 3; autoradiography was performed for 6 days. (T) Total RNA from prostate tissue. The bands corresponding to cloned transcripts (lower case) were arrowed. K3-Ex1 K 3 - sp5 K 3 -P C R2 SSI K 3 -1 .5 K3-1.5 K 3 - 5055 K3-0.7rev2 K3-PCR2 K 3 -P C R2 K 3 -P C R2 K3-MU1 hK3/PSA hK3/PSA PSA-Tr PSA-LM PSA-LM PSA-RP4 PSA-RP3 PSA-RP2 PSA-RP2 PSA-RP1 PSA-RP1 PSA-RP5 aa 69 261 238 180 218 220 104 227 a b (1) k j l (6) h (3) g (2) f e d c i (4) nt 1460 860 1 90 2 1701 1627 709 1 3 2 0 850 >1040 > 583 > 1 9 4 5 > 1130 1000 bp START Ser189 Intron-2 probe Exon-2 probe His 41 Exon-3 probe Asp 96 Intron-3 probe Intron-4 probeIntron-1 probe 2 4 3 1 Fig. 3. Compilation of the hKLK3 transcripts. Intron numbers and position of the DNA probes used for hybridization are given in the genomic DNA (grey). The variants (a to l) were classified according to their encoded protein (PSA to PSA-Tr). Numbers in exponent denote the new or earlier described variants for which new data are given in the text. The length in nucleotides (nt) of the cloned sequences, without the poly(A) tail, is shown at the left of the figure while the amino-acid (aa) number of the predicted prepro proteins is mentioned at the right. Exons are shown by boxes and introns by the connecting lines, the lacking sequences of some transcripts are mentioned by dotted lines. Filled boxes represent the coding sequences. Arrows in shaded boxes correspond to the position and direction of PCR primers used in the expression experiment. The positions of the codons corresponding to the residues of the catalytic triad are indicated. Ó FEBS 2003 Complex splicing of hKLK3 (Eur. J. Biochem. 270) 709 and YC100405 corresponded to 3 novel variants. The only difference between the YC140405.00 sequence (transcript K3b, accession no AJ459783; Fig. 3) and the major mRNA (K3a) was the length of the 3¢ untranslated sequence. This sequence was 586 nucleotides shorter in the K3b transcript. Sequence analysis of the variant K3h corresponding to the clone YC171105.00 revealed an additional intron inside exon 3 (accession no AJ459782, Fig. 3). The clone YC100405 was a partial copy of a new variant (K3i, accession no AJ512346) retaining intron 4. Another partial copy of a new alternative transcript was identified by screening of an EST database with the hKLK3 genomic sequence. This new transcript retained intron 2 sequences (K3l, Fig. 3, accession no BE840537). Expression of alternatively spliced hKLK3 transcripts To determine whether alternatively spliced transcripts are expressed in normal and pathological conditions, RT-PCR was performed using total RNA from normal, BPH and cancer specimens (Fig. 4). PCR primers were designed from distant constitutive or alternative exons (Table 1 and Fig. 3) and led to amplification of different size products from the targeted transcripts and other putative transcripts with intervening sequences. All PCRs performed on each tissue specimen gave a major product, which displayed both the expected size (Fig. 4, Table 1) and DNA sequence (not shown). Additional faint bands were also observed, sug- gesting amplification of longer transcripts containing inter- vening sequences. This experiment indicates that all the splicing isoforms tested are expressed in normal, BPH and cancerous prostate tissues. However, it was not possible to determine whether the malignant transformation alters the production of alternatively spliced transcripts, as the method used was not quantitative. Fig. 4. Multiple alternative transcripts in the human prostate. Total RNA from normal prostate (N), BPH or cancer was reverse-tran- scribed. cDNAs were amplified by PCR using the primers given in Table 1. The resulting PCR products were separated on agarose gel and visualized by ethidium bromide. C: control without cDNA. From 0.1 to 1 kb, the increment of the DNA ladder was 100 bp. Table 2. Exon-intron boundaries of the hKLK3 gene. Exon and intron numbers refer to the numbers given in Fig. 3. Letters at exon or intron numbers indicate the variant exon or intron found in the referred transcript while the (¢) symbol indicates an additional exon or intron. Exon sequences are in uppercase and introns in lowercase. Residues that are identical with the consensus sequences are in bold or underlined. M ¼ Aor C, Y ¼ CorU,R¼ AorG,N¼ any. Exon No. (transcript) Size (bp) Intron No. Size (bp) 5¢ donor seq. MAGguragu Branch site ynyuray 3¢ acceptor seq. (y (n) ryagG) 1 (K3a-h, l) 87 1 1239 UUGgugaga cccugau cccccucugcagGCG 1j (K3j) 1485 2 1639 GAAgugagu uccucau cuuccuccccagCAA 1 k (K3k) 510 1k 815 GCUgugagu cccugau cccccucugcagGCG 2 (K3a-f, h- k) 160 2 1639 GAAgugagu uccucau cuuccuccccagCAA 2 (K3g) 160 2 g 1768 GAAgugagu uccugaa auuccu cagGCC 3 (K3a-d,i- k) 287 3 143 AGUguacgc ccacaac cccguagUCU 3 g (K3g) 248 3 143 AGUguacgc ccacaac cccguagUCU 3 h (K3h) 147 3¢ 123 CAGccacga cacuggggac ggggcagCAU 3’h (K3h) 17 3 143 AGUguacgc ccacaac cccguagUCU 4 (K3ab,e,gh,jk) 137 4 1375 UGCgugagu gacugac cccuuagGGU 4 (K3cd) 137 4c 933 UGCgugagu ccuccac cccacagUGG Poly(A) signal AAUAAA Poly(A) cleavage site Ca n (< 10)… yguguuyy Ca n (< 10)… u rich 5 (K3a,e,g) 794 AAUAAA Gu ugugugac 5b 207 AGUAAA Caggccaagacucaag 5c 1236 AAUAAA Gu ugugugac 5d 1034 CAUAAC Gu ugugugac 3f (K3f) 465 AAGAAA Cc uguuauuu 5 h 306 AAGAAA Ca aguguuuc 4i >521 5 (K3j,k) >36 710 N. Heuze ´ -Vourc’h et al. (Eur. J. Biochem. 270) Ó FEBS 2003 Exon/intron structure analysis We examined the intron/exon boundaries of the hKLK3 gene to see if the sequence signals required for premRNA splicing was preserved (Table 2). While the AG dinucleotide imme- diately preceding the exon was always present in the hKLK3 acceptor sequences, one donor sequence (intron 3¢,K3h) lacked the well-conserved GU dinucleotide. There were also several mismatches in the exonic part of the consensus, with great base variations at each position. Analysis of the hKLK3 gene using a search algorithm (http://www.fruitfly.org/ seq_tools/splice.html) revealed 21 potential donor and 36 potential acceptor sequences (including the sites for splicing of introns 1, 2 and 4). One found 13 nucleotides upstream of the variant donor site of intron 3¢, was a canonical 5¢ splice site. This algorithm did not detect the alternative acceptor sequence used for transcripts k3c-d and, the real sites defining the boundaries of introns 3 and 3¢. This suggests that the splice sites are not optimal, a property often found for retained introns [21]. Putative branch sites with adjacent polypyrimidine tracts were found 20–30 nucleotides upstream of all the acceptor sites (Table 2). It has been known for some time that many alternatively spliced exons, small exons or exons with weak splice sites rely upon the activity of enhancers for their inclusion in mRNA [22]. As several splicing events affect the region surrounding intron 3, we searched for putative regulatory signals (Fig. 5). Intron 3 is studded with G triplets and quadruplets. It has been suggested that G triplets enhance splicing efficiency and help to determine exon–intron borders [23,24]. Two G triplets and one G quadruplet belong to a 22 nucleotide duplicated element that we termed pISE (putative intron splicing enhancer). Each pISE copy (Fig. 5) also contains two short sequences, GGGUCUG and GAGGA, related to known splicing enhancers [25,26]. The first short sequence is similar to the consensus GGGGCUG of the intron splicing enhancer found down- stream of the microexon of the chicken cardiac troponin T gene. In this gene, the enhancer binds the bridging splicing factor SF1 and increases recognition of the upstream microexon of 7 nucleotides [25]. There is also an alternative microexon of 17 nucleotides upstream of the pISE in hKLK3. The GAGGA motif is present in intron 3 and in the 17 nucleotide microexon. In the latter, it lies downstream of a sequence motif similar to the (U)GGACCNG consensus sequence of an exonic splicing enhancer [26]. Another upstream sequence (UGGACCUG) fits the same consensus motif. Two other exonic enhancer sequence motifs (UCCUC and CCACCC) previously identified by in vitro selection of randomized RNA sequences [27] were found in exon 3. Structure of hKLK3 proteins The predicted amino-acid sequences of proteins encoded by the alternatively spliced mRNAs are shown in Fig. 6. The <−−−−−− PISE −−−−> <−−−−−− PISE −−−−> Fig. 5. Sequences of the region surrounding intron 3. The sequences of several transcripts were aligned with the premRNA sequence derived from the hKLK3 gene sequence. The dotted lines correspond to the intervening sequences. The putative regulatory signals are indicated in colour. The dinucleotide of the donor (red) and acceptor (green) splice site signals are highlighted, as are the putative branch points (grey). Nucleotides of the polypyrimidine tracts are in red. The G stretches are highlighted in yellow while the nucleotide sequences of putative splicing enhancers are in blue. Ó FEBS 2003 Complex splicing of hKLK3 (Eur. J. Biochem. 270) 711 conservation of the N-terminal part of PSA, including the scretion signal peptide and the propeptide, suggests that all the PSA-RPs (PSA-related proteins) were synthesized as prepro proteins. While PSA-RP1, PSA-RP2 and PSA-RP5 differ from PSA at the C-terminal region. PSA-RP3 and PSA-RP4 are shorter than PSA due to in frame deletions. In PSA-RP3, the deletion results in the loss of asparagine-45 that is the binding site for the carbohydrate chain in PSA [19]. Forty-two amino acids, including one cysteine residue and the aspartate residue-96 of the catalytic triad, are deleted in PSA-RP4. The K3l transcript (from the EST database) contains a premature stop codon located at the beginning of the retained intron 2. It might encode a truncated form of prepro PSA (PSA-Tr, PSA-truncated). The transcripts K3j and k encode a protein (PSA-LM [20]), sharing only the signal peptide with PSA due to the creation of a novel ORF by the retention of intron 1 sequences. Although recombinant PSA-RP2 has been produced in a heterologous eukaryotic cell system [18], there has been no report on expression of this variant in prostate. Therefore, polyclonal antibodies were raised against a peptide corres- ponding to the C-terminal sequence of PSA-RP2. As shown in Fig. 7, these antibodies recognized recombinant PSA- RP2 but not PSA purified from seminal fluid. Moreover, a protein with a molecular mass similar to that of recombi- nant PSA-RP2 was detected in a protein extract from a cancerous prostate tissue (Fig. 7), revealing production of PSA-RP2 in vivo. Discussion We have used Northern blotting, molecular cloning and a database search to show that the hKLK3 gene produces at least 15 transcripts, of 0.7 to 6.1 kb, in prostate. Thus, the expression and splicing of the hKLK3 gene is more complex than previously thought [17]. All transcripts larger than the major mRNA encoding hK3/PSA contain intronic sequences. Their polysomal distribution indicates that the 2.1 (K3c, PSA-RP1) and 3.1 kb transcripts are mature mRNAs efficiently translated in LNCaP cells, whereas the largest transcripts seem to be weakly translated. As the large hKLK3 transcripts retaining introns were detected in the cytosolic fraction, it is unlikely that they are splicing intermediates. These transcripts might be either aberrant (poorly spliced with nonsense codon) or coding transcripts. The presence of a premature stop codon in the k3l transcript corresponding to PSA-Tr supports the hypothesis of aberrant hKLK3 transcripts. Degradation of aberrant transcripts is thought to occur in the cytoplasm via the mRNA surveillance system that depends upon translation [28–30]. This could explain both cytoplasmic localization and association with ribosomes of the poorly spliced hKLK3 transcripts. Further investigations are required to determine whether aberrant hKLK3 transcripts significantly accumu- late before degradation. Alternatively, the larger ones could be coding transcripts. An unusual feature of the hKLK3 gene is that the open reading frame continues in intron 1 resulting in the PSA-LM protein [20]. As the large transcripts hybridized with the intron 1 probe, they might encode PSA-LM. In this case, weak association of the large hKLK3 transcripts with polysomes could be due to peculiar structures that reduce translation efficiency [31]. Numerous cis-acting sequences and trans-acting cytoplasmic proteins participating in mRNA stability, localization or translation, PSA MWVPVVFLTLSVTWIGA APLILSRIVGGWECEKHSQPWQVLVASRGRAVCGGVLVHPQWVLTAAHCIR NKSVILLGRHSLFHPEDTGQVFQVSHSFPHPLYDMSLLKNRFLRPGDDSSHD PSA-RP1 MWVPVVFLTLSVTWIGA APLILSR IVGGWECEKHSQPWQVLVASRGRAVCGGVLVHPQWVLTAAHCIR NKSVILLGRHSLFHPEDTGQVFQVSHSFPHPLYDMSLLKNRFLRPGDDSSH D PSA-RP2 MWVPVVFLTLSVTWIGA APLILSR IVGGWECEKHSQPWQVLVASRGRAVCGGVLVHPQWVLTAAHCIR NKSVILLGRHSLFHPEDTGQVFQVSHSFPHPLYDMSLLKNRFLRPGDDSSH D PSA-RP3 MWVPVVFLTLSVTWIGA APLILSR IVGGWECEKHSQPWQVLVASRGRAVCGGVLVHPQWVLTAAHCIR KPGDDSSH D PSA-RP4 MWVPVVFLTLSVTWIGA APLILSR IVGGWECEKHSQPWQVLVASRGRAVCGGVLVHPQWVLTAAHCIR NKSVILLGRHSLFHPEDTGQVFQVSHSFPHPLYDMSLLKNRFLRPGDDSS PSA-RP5 MWVPVVFLTLSVTWIGA APLILSR IVGGWECEKHSQPWQVLVASRGRAVCGGVLVHPQWVLTAAHCIR NKSVILLGRHSLFHPEDTGQVFQVSHSFPHPLYDMSLLKNRFLRPGDDSSH D PSA-Tr MWVPVVFLTLSVTWIGA APLILSR IVGGWECEKHSQPWQVLVASRGRAVCGGVLVHPQWVLTAAHCIR K PSA-LM MWVPVVFLTLSVTWIG ERGHGWGDAGEGASPDCQAEALSPPTQHPSPDRELGSFLSLPAPLQAHTPSPSILQQSSLPHQVPAPSHLPQNFLPIAQPAPCSQLLY PSA LMLLRLSEPAELTDAVKVMDLPTQEPALGTTCYASGW GSIEPEEFLTPKKLQCVDLHVISNDVCAQVHPQKVTKFMLCAGRWTGGKSTCSGD SGGPLVCNGVLQGITSWGSEPCALPERP PSA-RP1 LMLLRLSEPAELTDAVKVMDLPTQEPALGTTCYASGWGSIEPEEFLTPKKLQCVDLHVISNDVCAQVHPQKVTKFMLCAGRWTGGKSTCS WVILITELTMPALPMVLHGSLVPWRGGV PSA-RP2 LMLLRLSEPAELTDAVKVMDLPTQEPALGTTCYASGWGSIEPEE CTPGPDGAAGSPDAWV PSA-RP3 LMLLRLSEPAELTDAVKVMDLPTQEPALGTTCYASGWGSIEPEEFLTPKKLQCVDLHVISNDVCAQVHPQKVTKFMLCAGRWTGGKSTCSGD SGGPLVCNGVLQGITSWGSEPCALPERP PSA-RP4 IEPEEFLTPKKLQCVDLHVISNDVCAQVHPQKVTKFMLCAGRWTGGKSTCSGD SGGPLVCNGVLQGITSWGSEPCALPERP PSA-RP5 LMLLRLSEPAELTDAVKVMDLPTQEPALGTTCYASGWGSIEPEEFLTPKKLQCVDLHVISNDVCAQVHPQKVTKFMLCAGRWTGGKSTCS VSHPYSQDLEGKGEWGP PSA SLYTKVVHYRKWIKDTIVANP PSA-RP3 SLYTKVVHYRKWIKDTIVANP PSA-RP4 SLYTKVVHYRKWIKDTIVANP Fig. 6. Alignment of the predicted hKLK3 proteins. The signal peptide is highlighted in yellow and the propeptide in blue. The amino-acid residues of the catalytic triad are in red, while the binding site for the carbohydrate chain is in green. Sequences divergent to the PSA sequence are highlighted in grey. Fig. 7. Detection of PSA-RP2 in prostate tissue. PSA from seminal fluid (40 ng), cytosolic proteins from CHO cells expressing recom- binant PSA-RP2 (60 lg) and from a cancerous prostate tissue (250 lg) were subjected to SDS/PAGE and analyzed by Western blot using the polyclonal anti-RP2 Ig. The band corresponding to PSA-RP2 is indicatedbyanarrow. 712 N. Heuze ´ -Vourc’h et al. (Eur. J. Biochem. 270) Ó FEBS 2003 have been identified in eukaryotes. A search for cis elements within the hKLK3 gene sequence using the UTRScan computer program [32] revealed no conserved sequences involved in (de)stabilizing, locating or translating mRNAs. However, unconserved cis-acting sequences could play a regulatory role in the translation efficiency of the large hKLK3 transcripts. Our study reveals the large 3¢-UTR diversity of hKLK3 transcripts. Two well documented functions of 3¢-UTRs are mRNA stabilization and its localization in specific regions of the cytoplasm [32]. The use of different polyadenylation sites suggests that there is a post-transcriptional regulation of hKLK3 gene expression. This is supported by data indicating that the 3.1 kb transcript is more unstable than the major hKLK3 mRNA [33]. Functional analyses will be needed to assess the role of the 3¢UTR in the stability of hKLK3 mRNAs in normal and pathological prostatic cells. The process by which constitutive and alternative exons are recognized in a premRNA is complex. The early steps of spliceosome assembly involve recognition of consensus elements at both ends of the intron. Although these sequences are usually short, they are often degenerate. Nevertheless, about 99% of splice site pairs are GT-AG [34]. The alternative intron 3¢ of hKLK3 does not follow this rule, but has unusual CC-AG pairs. Recognition of this atypical site is probably related to the presence of a canonical site upstream to the variant. Indeed, Burset et al. suggested that uncanonical sites could function exclusively in association with a canonical site [34]. In the other cases, reasonably conserved signals were found at both ends of the hKLK3 introns; however, their relative strength remains to be determined. It is clear that conserved sequences near the 3¢ and 5¢ splice sites are generally insufficient for selecting true splice sites among abundance of similar sequences. Unconserved sequences commonly named splicing enhancers and silencers provide more information to specific regulatory factors that interact or interfere with the splicing machinery. We looked for putative regulatory sequences because of the complexity of the splicing events affecting the middle of the hKLK3 gene. Intron 3 contains a high concentration of G triplets; these are frequently found close to 5¢ splice sites in mammals [23,24]. This well-established splicing enhancer promotes the selection of a 5¢ splice site by recruiting U1 snRNP. Many other putative splicing enhancers were detected in the alternative exons and introns, suggesting that there is considerable information in the various segments of the hKLK3 premRNA. Some sequences also contain overlap- ping elements. We identified a 22-nucleotide repeat (pISE) which contains G triplets and an internal motif known to recruit SF1. Thus, pISE could be involved in the determination of exon-intron borders via interaction of the G sequences with U1 snRNPs and, in definition of the microexon 3¢ via recruitment of SF1 by the internal motif [25]. These observations suggest that the complex splicing of hKLK3 probably reflects the probability of occupancy of individual sites and the cross-talk between multiple interactions, as in other genes [35]. The splicings result in two short introns (3 and 3¢) and a 17 nucleotide microexon. This is unusual as the exons are typically 100–200 nucleotides in human, and the introns are much longer, averaging about 3 kb. Only about 10% of the introns are classified as short (< 134 nucleotides), while no more than 4% of vertebrate internal exons are shorter than 50 nucleotides [36]. To date, 12 hKLK3 transcripts have been cloned and sequenced. The proteins predicted from the nucleotide sequences are PSA, truncated PSA and six alternate proteins. Five predicted proteins are PSA variants (PSA- RP1 to RP5) that could be synthesized as precursors. The presence of a common signal peptide suggests that all these PSA-RPs are secreted from prostate cells. Previous recom- binant experiments [15,18] and the identification of PSA- RP1 in the spent medium of LNCaP cells [37,38] strongly support this assertion. In the present time, two PSA-RPs have been identified in prostate tissue, PSA-RP1 [37] and PSA-RP2 using immunohistochemical and Western blot analysis, respectively. Characterization of other PSA-RP variants is currently under investigation. The variation in the mRNA will result in several great changes in the amino- acid sequences that probably interfere with the protease activity of hK3/PSA. As PSA function depends on this activity, we need to know how these variants that seem to have no enzymatic activity, influence prostate physiology and pathology. By contrast to the PSA-RPs, the protein PSA-LM encoded by the transcripts containing intron 1 is quite unlike PSA. A recombinant form of this protein has been recently characterized [20]. PSA-LM has also been found in the secretory epithelial cells of prostate; however, its function remains unknown. All these observations emphasize the complexity of the protein resulting from hKLK3 gene expression. Numerous efforts are made to ameliorate the diagnostic value of the PSA assay. The major aim in this field is to enhance the discrimination of patients with BPH from those with Pca. One way would be to use additional markers. As cancer is said to alter the splicing pattern of some genes [39], some variants of PSA could be useful to improve the tumor selectivity of the PSA assay. Additional studies are required to determine the clinical values of these PSA variants. Acknowledgements We are indebted to Drs Lanson and Haillot of the Department of Urology, Hoˆ pital Bretonneau de Tours for providing human prostate tissues. We thank Mme E. Bataille ´ , Drs Gutman and Rosinski-Chupin for their assistance and O. Parkes for critically reviewing this manu- script before its submission. This work was supported by grants from the Association pour la Recherche sur le Cancer, the Ligue contre le Cancer (Comite ´ d’Indre-et-Loire) and from the Association de Recherche sur les Tumeurs de la Prostate. References 1. Yousef, G.M., Chang, A., Scorilas, A. & Diamandis, E.P. (2000) Genomic organization of the human kallikrein gene family on chromosome 19q13.3-q13.4. Biochem. Biophys. Res. Commun. 276, 125–133. 2. Diamandis, E.P., Yousef, G.M., Luo, I., Magklara, I. & Obiezu, C.V. (2000) The new human kallikrein gene family: implications in carcinogenesis. Trends Endocrinol. Metab. 11, 54–60. 3. Rittenhouse, H.G., Finlay, J.A., Mikolajczyk, S.D. & Partin, A.W. (1998) Human kallikrein 2 (hK2) and prostate-specific antigen (PSA): two closely related, but distinct, kallikreins in the prostate. Crit. Rev. Clin. Lab. Sci. 35, 275–368. Ó FEBS 2003 Complex splicing of hKLK3 (Eur. J. Biochem. 270) 713 4. Lilja, H. (1985) A kallikrein-like serine protease in prostatic fluid cleaves the predominant seminal vesicle protein. J. Clin. Invest. 76, 1899–1903. 5. Cohen, P., Peehl, D.M., Graves, H.C. & Rosenfeld, R.G. (1994) Biological effects of prostate specific antigen as an insulin-like growth factor binding protein-3 protease. J. Endocrinol. 142, 407–415. 6. Fortier, A.H., Nelson, B.J., Grella, D.K. & Holaday, J.W. (1999) Antiangiogenic activity of prostate-specific antigen. J. Natl. Cancer. Inst. 91, 1635–1640. 7. Heidtmann, H., Nettelbeck, D., Mingels, A., Ja ¨ ger, R. & Kon- termann, R. (1999) Generation of angiostatin-like fragments from plasminogen by prostate-specific antigen. Br. J. Cancer 81, 1269–1273. 8. Rehault, S., Monget, P., Mazerbourg, S., Tremblay, R., Gutman, N., Gauthier, F. & Moreau, T. (2001) Insulin-like growth factor binding proteins (IGFBPs) as potential physiological substrates for human kallikreins hK2 and hK3. Eur. J. Biochem. 268, 2960– 2968. 9. Carroll, P., Coley, C., McLeod, D., Schellhammer, P., Sweat, G., Wasson, J., Zietman, A. & Thompson, I. (2001) Prostate-specific antigen best practice policy – part II: prostate cancer staging and post-treatment follow-up. Urology 57, 225–229. 10. Carroll, P., Coley, C., McLeod, D., Schellhammer, P., Sweat, G., Wasson, J., Zietman, A. & Thompson, I. (2001) Prostate-specific antigen best practice policy – part I: early detection and diagnosis of prostate cancer. Urology 57, 217–224. 11. Lilja, H., Christensson, A., Dahlen, U., Matikainen, M T., Nilsson, O., Pettersson, K. & Lo ¨ vgren, T. (1991) Prostate-specific antigen in serum occurs predominantly in complex with a1-antichymotrypsin. Clin. Chem. 37, 1618–1625. 12. Hilz,H.,Noldus,J.,Hammerer,P.,Buck,F.,Lu ¨ ck, M. & Huland, H. (1999) Molecular heterogeneity of free PSA in sera of patients with benign and malignant prostate tumors. Eur. Urol. 36, 286–292. 13. Mikolajczyk, S.D., Marker, K.M., Millar, L.S., Kumar, A., Saedi, M.S., Payne, J.K., Evans, C.L., Gasior, C.L., Linton, H.J., Carpenter, P. & Rittenhouse, H.G. (2001) A truncated precursor form of prostate-specific antigen is a more specific serum marker of prostate cancer. Cancer Res. 61, 6958–6963. 14. Noldus, J., Chen, Z. & Stamey, T.A. (1997) Isolation and characterization of free form prostate specific antigen (f-PSA) in sera of men with prostate cancer. J. Urol. 158, 1606–1609. 15. Heuze ´ ,N.,Olayat,S.,Gutman,N.,Zani,M L.&Courty,Y. (1999) Molecular cloning and expression of an alternative hKLK3 transcript coding for a variant protein of prostate-specific antigen. Cancer Res. 59, 2820–2824. 16. Riegman, P.H.J., Klaassen, P., Van der Korput, J.A.G.M., Romijn, J.C. & Trapman, J. (1988) Molecular cloning and characterization of novel prostate antigen cDNA’s. Biochem. Biophys. Res. Comm. 155, 181–188. 17. Henttu, P., Lukkarinen, O. & Vihko, P. (1990) Expression of the gene coding for human prostate-specific antigen and related hGK-1 in benign and malignant tumors of the human prostate. Int. J. Cancer. 45, 654–660. 18. Heuze-Vourc’h, N., Leblond, V., Olayat, S., Gauthier, F. & Courty, Y. (2001) Characterization of PSA-RP2, a protein related to prostate-specific antigen and encoded by alternative hKLK3 transcripts. Eur. J. Biochem. 268, 4408–4413. 19. Tanaka, T., Isono, T., Yoshiki, T., Yuasa, T. & Okada, Y. (2000) A novel form of prostate-specific antigen transcript produced by alternative splicing. Cancer Res. 60, 56–59. 20. David, A., Mabjeesh, N., Azar, I., Biton, S., Engel, S., Bernstein, J., Romano, J., Avidor, Y., Waks, T., Eshhar, Z., Langer, S.Z., Lifschitz-Mercer, B., Matzkin, H., Rotman, G., Toporik, A., Savitsky, K. & Mintz, L. (2002) Unusual alternative splicing within the human kallikrein genes KLK2 and KLK3 gives rise to novel prostate-specific proteins. J. Biol. Chem. 277, 18084–18090. 21. Stamm,S.,Zhu,J.,Nakai,K.,Stoilov,P.,Stoss,O.&Zhang, M.Q. (2000) An alternative-exon database and its statistical analysis. DNA Cell. Biol. 19, 739–756. 22. Blencowe, B.J. (2000) Exonic splicing enhancers: mechanism of action, diversity and role in human genetic diseases. Trends Biochem. Sci. 25, 106–110. 23. McCullough, A.J. & Berget, S.M. (1997) G triplets located throughout a class of small vertebrate introns enforce intron borders and regulate splice site selection. Mol. Cell. Biol. 17, 4562–4571. 24. McCullough, A.J. & Berget, S.M. (2000) An intronic splicing enhancer binds U1 snRNPs to enhance splicing and select 5¢ splice sites. Mol. Cell. Biol. 20, 9225–9235. 25. Carlo, T., Sierra, R. & Berget, S.M. (2000) A 5¢ splice site- proximal enhancer binds SF1 and activates exon bridging of a microexon. Mol. Cell. Biol. 20, 3988–3995. 26. Schaal, T.D. & Maniatis, T. (1999) Multiple distinct splicing enhancers in the protein-coding sequences of a constitutively spliced pre-mRNA. Mol. Cell. Biol. 19, 261–273. 27. Schaal, T.D. & Maniatis, T. (1999) Selection and characterization of pre-mRNA splicing enhancers: identification of novel SR protein-specific enhancer sequences. Mol. Cell. Biol. 19, 1705– 1719. 28. Goldstrohm, A.C., Greenleaf, A.L. & Garcia-Blanco, M.A. (2001) Co-transcriptional splicing of pre-messenger RNAs: considerations for the mechanism of alternative splicing. Gene 277, 31–47. 29. Lykke-Andersen, J. (2001) mRNA quality control: marking the message for life or death. Curr. Biol. 11, R88–R91. 30. Mitchell, P. & Tollervey, D. (2001) mRNA turnover. Curr. Opin. Cell. Biol. 13, 320–325. 31. Macdonald, P. (2001) Diversity in translational regulation. Curr. Opin. Cell. Biol. 13, 326–331. 32. Pesole, G., Liuni, S., Grillo, G., Licciulli, F., Mignone, F., Gissi, C. & Saccone, C. (2002) UTRdb and UTRsite: specialized databases of sequences and functional elements of 5¢ and 3¢ untranslated regions of eukaryotic mRNAs. Nucleic Acids Res. 30, 335–340. 33. Wolf, D., Schulz, P. & Fittler, F. (1992) Transcriptional regulation of prostate kallikrein-like genes by androgen. Mol. Endocrinol. 6, 753–762. 34. Burset, M., Seledtsov, I.A. & Solovyev, V.V. (2000) Analysis of canonical and non-canonical splice sites in mammalian genomes. Nucleic Acids Res. 28, 4364–4375. 35. Smith, C.W. & Valcarcel, J. (2000) Alternative pre-mRNA splicing: the logic of combinatorial control. Trends Biochem. Sci. 25, 381–388. 36. Lim, L.P. & Burge, C.B. (2001) A computational analysis of sequence features involved in recognition of short introns. Proc. Natl Acad. Sci. USA 98, 11193–11198. 37. Kumar, A., Mikolajczyk, S.D., Hill, T.M., Millar, L.S. & Saedi, M.S. (2000) Different proportions of various prostate-specific antigen (PSA) and human kallikrein 2 (hK2) forms are present in noninduced and androgen- induced LNCaP cells. Prostate 44, 248–254. 38. Meng, F.J., Shan, A., Jin, L. & Young, C.Y.F. (2002) The expression of a variant prostate-pecific antigen in human prostate. Cancer Epidemiol. Biomarkers Prev. 11, 305–309. 39. Mercatante, D. & Kole, R. (2000) Modification of alternative splicing pathways as a potential approach to chemotherapy. Pharmacol. Ther. 85, 237–243. 714 N. Heuze ´ -Vourc’h et al. (Eur. J. Biochem. 270) Ó FEBS 2003 . Complex alternative splicing of the hKLK3 gene coding for the tumor marker PSA (prostate-specific-antigen) Nathalie. arrowed. K3-Ex1 K 3 - sp5 K 3 -P C R2 SSI K 3 -1 .5 K3-1.5 K 3 - 5055 K3-0.7rev2 K3-PCR2 K 3 -P C R2 K 3 -P C R2 K3-MU1 hK3 /PSA hK3 /PSA PSA-Tr PSA- LM PSA- LM PSA- RP4 PSA- RP3 PSA- RP2 PSA- RP2 PSA- RP1 PSA- RP1 PSA- RP5 aa 69 261 238 180 218 220 104 227 a b (1) k j l (6) h (3) g (2) f e d c i (4) nt 1460 860 1 90 2 1701 1627 709 1 3 2 0 850 >1040 >