Loss of genomic material from chromosomal band 13q14.3 is the most common genetic imbalance in B-cell chronic lymphocytic leukemia (B-CLL) and mantle cell lymphoma, pointing to the involvement of this region in a tumor suppressor mechanism. From the minimally deleted region, 3 candidate genes have been isolated, RFP2, BCMS, and BCMSUN. DNA sequence analyses have failed to detect small mutations in any of these genes, suggesting a different pathomechanism, most likely haploinsufficiency. We, therefore, tested B-CLL patients for epigenetic aberrations by measuring expression of genes from 13q14.3 and methylation of their promotor region.RB1, CLLD7, KPNA3, CLLD6, andRFP2 were down-regulated in B-CLL patients as compared with B cells of healthy donors, with RFP2 showing the most pronounced loss of expression. To test whether this loss of gene expression is associated with methylation of CpG islands in the respective promotor regions, we performed methylation-sensitive quantitative polymerase chain reaction analyses and bisulfite sequencing on DNA from B-CLL patients. No difference in the methylation patterns could be detected in any CpG island of the minimally deleted region. Down-regulation of genes within chromosomal band 13q14.3 in B-CLL is in line with the concept of haploinsufficiency, but this tumor-specific phenomenon is not associated with DNA methylation.
Introduction
B-cell chronic lymphocytic leukemia (B-CLL), the most frequent leukemia in the Western world, is cytogenetically well characterized.1,2 There is a significant overlap between the chromosomal alterations in B-CLL and mantle cell lymphoma (MCL).3 The most frequent genomic imbalance is loss of genetic material from chromosomal band 13q14, detected in more than 50% of B-CLL patients and almost 70% of MCL patients by fluorescence in situ hybridization (FISH).3-5 It is, therefore, very likely that this genomic region is involved in the pathogenesis of B-CLL and MCL and harbors a tumor suppressor mechanism. In the search for this tumor suppressor mechanism, several contigs of physically overlapping DNA fragments have been constructed,5-9 and 3 candidate tumor suppressor genes were isolated: RFP2 (also termed Leu5, CAR), BCMSUN (DLeu2,t4, cDNA 1B4), and BCMS (Dleu1, t5, cDNA 170C-70)5,7,9-12 (Figure1 for overview ). Extensive mutation analyses of the 3 candidate genes as well as the noncoding regions failed to detect small genetic aberrations in B-CLL patients10,13 (D.M. and S.W., unpublished results, September 2001). Therefore, it has been considered that the pathomechanism involving chromosomal band 13q14.3 in B-CLL and MCL is based on haploinsufficiency and possibly is of epigenetic nature; ie, it is not reflected in the primary sequence. Such a mechanism has been shown, for example, in the CDNK2B/p15gene in hematologic malignancies14-16: in many cases, no point mutations are found in the CDKNB2/p15 gene, but gene function is lost through genomic deletion or hypermethylation of both alleles. The RB1 gene localized at chromosomal band 13q14.1-q14.2 can be inactivated in retinoblastoma by genomic deletion, point mutation, or methylation.17 Silencing through methylation has been detected for several other genes in a variety of malignancies (for reviews, see Jones & Laird18 and Baylin & Herman19). To test whether a similar mechanism of gene repression with concomitant methylation is responsible for the proposed tumor suppressor pathomechanism at chromosomal band 13q14.3 in B-CLL, we analyzed the expression of the genes localized in this region and the methylation of the respective CpG islands.
Overview of the genes and CpG islands localized in the critical region at chromosomal band 13q14.3 in B-CLL.
Several genes and CpG islands localized at chromosomal band 13q14.3 in the vicinity of the minimally deleted region were tested for expression and methylation. Arrowheads indicate direction of genes. Diagram is not drawn to scale.
Overview of the genes and CpG islands localized in the critical region at chromosomal band 13q14.3 in B-CLL.
Several genes and CpG islands localized at chromosomal band 13q14.3 in the vicinity of the minimally deleted region were tested for expression and methylation. Arrowheads indicate direction of genes. Diagram is not drawn to scale.
Materials and methods
RNA isolation, reverse transcription, and real-time polymerase chain reaction
RNA and DNA were extracted from lymphocytes of B-CLL patients or sorted peripheral blood B-cells of healthy donors (CD19 antibody-coupled beads; Miltenyi, Bergisch Gladbach, Germany) with Trizol Reagent (Gibco BRL, Karlsruhe, Germany), and the RNA was reverse transcribed with the GeneAmp RT-PCR Kit (Applied Biosystems, Weiterstadt, Germany), as described previously.20Real-time polymerase chain reaction (PCR) was carried out in an ABI 7700 Taqman by using the SYBR-green-PCR-kit (both Applied Biosystems). Primer sequences are shown in Table 1. The cluster and treeview software used for visualization of PCR results was obtained fromhttp://rana.stanford.edu/software.21
Bisulfite conversion, real-time PCR, and sequencing
The real-time PCR quantification uses the selective conversion of nonmethylated cytosine to uracil by treatment with bisulfite, leading to different sequences of the products depending on the methylation status of the substrate DNA. This difference can be detected by a PCR that is specific for converted DNA only. For quantification, PCR efficiencies of methylation-sensitive primers spanning CpG doublets with methylation-insensitive primers not spanning CpG doublets were compared. PCR efficiency should be low with the use of nonconverted DNA, high with converted DNA, and intermediate with converted DNA pretreated with SSS1-methylase that specifically methylates CG nucleotides. Two methylation-sensitive primer pairs were designed for each CpG island analyzed. DNA was converted as described previously,22 except for using the Geneclean Kit for desalting (Qbiogene, Heidelberg, Germany). As control, 2 to 4 μg DNA was methylated with 8 U SSS1-methylase with 0.2 mM S-adenosyl-methionine (both NEB, Frankfurt, Germany) as suggested by the manufacturer for 105 minutes at 37°C. Real-time PCR was carried out by using the SYBR-green-PCR-kit and the ABI 7700 Sequence Detector (both Applied Biosystems). Bisulfite sequencing was performed with clones derived from PCR products by using the TOPO-cloning Kit (Invitrogen, Karlsruhe, Germany). Sequencing reactions were performed by using the BigDye-kit and the ABI 377 Sequencer (both Applied Biosystems). Primer sequences are presented in Table1.
Results
Expression of the genes BCL2, RB1, CHC1L, KPNA3, RFP2, BCMSUN, and BCMS (Figure 1) was measured by using real-time quantitative PCR in patients with tumors that are biallelic for the critical region at chromosomal band 13q14.3 (dis = disomic; n = 6) or patients with loss of one allele (del; n = 16) assessed by FISH. As internal standards, 12 housekeeping genes (18SrRNA, RPLP0, ACTB, PPI, GAPD, PGK, B2M, GUSB, HPRT1, TBP, TFRC, andLMNB1) were tested for expression in peripheral blood of a healthy donor and 3 B-CLL patients. Expression of 18SrRNA and GAPD varied substantially between the patients (Figure2). In contrast, Lamin B1 (LMNB1; GDB ID 512 284), phosphoglycerokinase (PGK; CAA 23 835), and cyclophilin (PPI; EC5.2.1.8) showed similar expression in all 4 patients (Figure 2) and were, therefore, suitable as internal controls to normalize samples. Expression of most genes at chromosomal band 13q14.3 varied substantially among B-CLL cases (Figure3). BCMS and BCMSUN(exons B and 4, 5, and 7) are not significantly down-regulated in B-CLL compared with sorted B cells of healthy donors. In contrast,RB1, CLLD7, KPNA3, CLLD6, and RFP2 are significantly down-regulated in B-CLL lymphocytes with loss of one allele at chromosomal band 13q14.3 (Figure 3B). Expression ofRFP2 showed the most pronounced down-regulation (more than 10-fold less on average compared with control B cells). Of the genes analyzed that are localized in the vicinity of the critical region,RFP2 is the only gene that is down-regulated in all patients disomic at chromosomal band 13q14.3 (more than 4-fold down-regulation in B-CLL patients as compared with sorted B cells,P < .0001; Figure 3B). Of the 3 genes, which are localized within the minimally deleted region, only expression ofRFP2 is down-regulated in B-CLL patients. The correlation of genetic aberrations with aberrant expression patterns in B-CLL points at a pathomechanism in this region that is mediated by down-regulation of gene expression. In addition, all investigated genes localized near the minimally deleted region are significantly down-regulated in B-CLL lymphocytes with loss of one allele at chromosomal band 13q14.3, suggesting haploinsufficiency of this region as a pathomechanism for B-CLL.
Validation of internal standards for expression analysis with real-time PCR in B-CLL.
cDNAs from whole RNA, healthy donor peripheral blood (PB), healthy donor-sorted B cells, and peripheral blood from 3 B-CLL patients were measured with the PE endogenous control plate on an Applied Biosystems 7700 Taqman. The cycle number, when the fluorescence from a PCR reaction reaches a set threshold value, corresponds to the amount of transcript present in the cDNA used as template (Ct value). The average of Ct values from 12 different housekeeping genes was used to normalize all tested patients (norm. Ct values). In contrast to 18S-rRNA andGAPD, PGK, PPI, and LMNB1are expressed at a similar rate in peripheral blood of healthy donors and B-CLL patients and were used as internal standards.
Validation of internal standards for expression analysis with real-time PCR in B-CLL.
cDNAs from whole RNA, healthy donor peripheral blood (PB), healthy donor-sorted B cells, and peripheral blood from 3 B-CLL patients were measured with the PE endogenous control plate on an Applied Biosystems 7700 Taqman. The cycle number, when the fluorescence from a PCR reaction reaches a set threshold value, corresponds to the amount of transcript present in the cDNA used as template (Ct value). The average of Ct values from 12 different housekeeping genes was used to normalize all tested patients (norm. Ct values). In contrast to 18S-rRNA andGAPD, PGK, PPI, and LMNB1are expressed at a similar rate in peripheral blood of healthy donors and B-CLL patients and were used as internal standards.
Analysis of expression of genes from the critical region with real-time PCR.
(A) Expression of genes localized at chromosomal band 13q14.3(RB1, CLLD7, KPNA3, CLLD6, RFP2, BCMSUN, andBCMS) and BCL2 was analyzed by quantitative real-time PCR together with a set of housekeeping genes (PGK, CYC, and LAMIN B1). Depicted is the logarithm of the ratio of the amount of the respective gene-specific mRNA to the average of the amount of mRNAs of the 3 housekeeping genes in peripheral blood lymphocytes (PBL; n = 4) and CD19+ peripheral blood lymphocytes (B cells; n = 9) from healthy donors, and in B-CLL patients with tumors biallelic for the critical region in chromosomal band 13q14.3 (disomic; n = 6), loss of one allele (deleted; n = 21), and loss of both alleles (biallelically deleted; n = 1). Interesting genes were either measured with 2 different sets of primers or on 2 different cDNAs prepared from the same patients or both. The color coding gives fold expression as compared with the average of sorted B cells; gray = not done. (B) A double-sided Studentt test shows either significant differences between the patient groups (dis, del) and sorted B cells or no significant differences.
Analysis of expression of genes from the critical region with real-time PCR.
(A) Expression of genes localized at chromosomal band 13q14.3(RB1, CLLD7, KPNA3, CLLD6, RFP2, BCMSUN, andBCMS) and BCL2 was analyzed by quantitative real-time PCR together with a set of housekeeping genes (PGK, CYC, and LAMIN B1). Depicted is the logarithm of the ratio of the amount of the respective gene-specific mRNA to the average of the amount of mRNAs of the 3 housekeeping genes in peripheral blood lymphocytes (PBL; n = 4) and CD19+ peripheral blood lymphocytes (B cells; n = 9) from healthy donors, and in B-CLL patients with tumors biallelic for the critical region in chromosomal band 13q14.3 (disomic; n = 6), loss of one allele (deleted; n = 21), and loss of both alleles (biallelically deleted; n = 1). Interesting genes were either measured with 2 different sets of primers or on 2 different cDNAs prepared from the same patients or both. The color coding gives fold expression as compared with the average of sorted B cells; gray = not done. (B) A double-sided Studentt test shows either significant differences between the patient groups (dis, del) and sorted B cells or no significant differences.
As loss of gene expression in neoplasia is often associated with DNA methylation, we investigated the methylation pattern of the CpG islands localized in the minimally deleted region and associated withKPNA3, both localized in chromosomal band 13q14.3 as well as the CpG island associated with BCL2, serving as control.23 Of the methods for the detection of methylated CpG islands,24,25 we chose a coupled bisulfite quantitative PCR method that allows screening of a larger number of patients and CpG islands26 (Figure4A). By using nonconverted DNA, bisulfite-converted DNA, and converted DNA pretreated with methylase, methylation sensitivity of primers specific for the genes localized in chromosomal band 13q14.3 could be shown (Figure 4B). Primers specific for the ACTB and MYOD1 genes are methylation insensitive and were used as internal controls.26
Validation of measuring DNA methylation with real-time PCR.
Bisulfite treatment changes cytosine to uracil only if it is not methylated (A). Differences in methylation are thereby converted into differences in sequence that can be detected by quantitative real-time PCR. Because the bisulfite conversion itself is not quantitative, primers have to distinguish between converted and nonconverted DNA (examples shown in B). In addition, primers used as internal standards (ACTB and MYOD1; Eads et al26) have to be independent of methylation, whereas primers localized in CpG islands have to be methylation sensitive and should give no signal on DNA methylated with SSS1-methylase (B).
Validation of measuring DNA methylation with real-time PCR.
Bisulfite treatment changes cytosine to uracil only if it is not methylated (A). Differences in methylation are thereby converted into differences in sequence that can be detected by quantitative real-time PCR. Because the bisulfite conversion itself is not quantitative, primers have to distinguish between converted and nonconverted DNA (examples shown in B). In addition, primers used as internal standards (ACTB and MYOD1; Eads et al26) have to be independent of methylation, whereas primers localized in CpG islands have to be methylation sensitive and should give no signal on DNA methylated with SSS1-methylase (B).
In 3 patient samples, 2 CpG islands were subjected to bisulfite sequencing to allow a detailed verification of the PCR approach.22
By using the coupled bisulfite quantitative PCR method, no significant differences in methylation could be detected in B-CLL patients (n = 23) compared with sorted B cells (n = 11) at any of the CpG islands tested (see Figure 5 for overview; results in Figure 6). Even though the variability of DNA methylation is higher among B-CLL patients compared with the sorted B cells, no methylation patterns specific for B-CLL patients could be detected. Therefore, DNA methylation in the critical region localized in chromosomal band 13q14.3 is not different in B-CLL lymphocytes compared with sorted B cells (Figure 6B).
![Fig. 5. Overview of several CpG islands from the critical region that was tested for DNA methylation. / CpG doublets localized in the promotor regions of the BCL2, KPNA3, RFP2, BCMS, and BCMSUN genes are depicted as vertical lines in the genomic sequence (base pair [bp] in numbers). In addition, Msp1 restriction sites (5′-CCGG-3′) of the genomic region of the 5′-ends of the BCMS and BCMSUN genes are shown at a lower magnification. Arrows indicate localization and direction of genes; exons are shown as black or gray boxes. Primer pairs that were used in quantification of methylation with real-time PCR are shown as short black lines A to K. Dotted black lines show regions that were bisulfite sequenced.](https://ash.silverchair-cdn.com/ash/content_public/journal/blood/99/11/10.1182_blood.v99.11.4116/6/m_h81122614005.jpeg?Expires=1733879936&Signature=g1SslCAcdKzrwTPuw-gVZFofJo4avArCiwUZLFOQ2uZHRLnGrP0plZnvbAx5M6pzhzf7surPBfTGdpA8mEz4v1j6rQMP~pYGRz05DYnPGC7fWPugid9xtwSUiqiDuCnAFilOVwGHGXQODtJncChfkiJFYwjkcbBgMXZlnyAgCbuVphcfv8bS5GlvTRQscGK5YjO9UHJ4ZiMawT~Q82NcaF53GwHMyZAsIEKRuJLXvrSbJR53SAGcLqK25-EvCqiKELoUgh6SMx0tv-3cAQvRBa8Me4XnvKAeMC48--8VIU1tsYt7CmwmenPrazTrkJ7PVaBQGzSTnPf9KBRWsy3phw__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA)
Overview of several CpG islands from the critical region that was tested for DNA methylation.
CpG doublets localized in the promotor regions of the BCL2, KPNA3, RFP2, BCMS, and BCMSUN genes are depicted as vertical lines in the genomic sequence (base pair [bp] in numbers). In addition, Msp1 restriction sites (5′-CCGG-3′) of the genomic region of the 5′-ends of the BCMS and BCMSUN genes are shown at a lower magnification. Arrows indicate localization and direction of genes; exons are shown as black or gray boxes. Primer pairs that were used in quantification of methylation with real-time PCR are shown as short black lines A to K. Dotted black lines show regions that were bisulfite sequenced.
Overview of several CpG islands from the critical region that was tested for DNA methylation.
CpG doublets localized in the promotor regions of the BCL2, KPNA3, RFP2, BCMS, and BCMSUN genes are depicted as vertical lines in the genomic sequence (base pair [bp] in numbers). In addition, Msp1 restriction sites (5′-CCGG-3′) of the genomic region of the 5′-ends of the BCMS and BCMSUN genes are shown at a lower magnification. Arrows indicate localization and direction of genes; exons are shown as black or gray boxes. Primer pairs that were used in quantification of methylation with real-time PCR are shown as short black lines A to K. Dotted black lines show regions that were bisulfite sequenced.
DNA methylation of CpG islands from the critical region as measured with real-time PCR.
(A) Methylation of CpG islands localized at chromosomal band 13q14.3 and in the BCL2 promotor was assessed with quantitative real-time PCR by using primers specific for bisulfite-converted and nonmethylated DNA from peripheral blood lymphocytes (PBL; n = 7), CD-19+ peripheral blood lymphocytes (B cells; n = 11), B-CLL patients with tumors biallelic for the critical region in 13q14.3 (disomic; n = 9), loss of one allele (deleted; n = 14), or loss of both alleles (biallelically deleted; n = 1). Depicted are the ratios of 13q14.3- orBCL2-specific primer pairs and 2 methylation-insensitive primer pairs that are specific for bisulfite-converted DNA (Eads et al26). The color coding gives fold nonmethylated DNA (green = methylated) as compared with the average of sorted B cells; gray = not done. (B) A double-sided Student t test shows no significant differences between the patient groups (disomic or deleted) and sorted B cells.
DNA methylation of CpG islands from the critical region as measured with real-time PCR.
(A) Methylation of CpG islands localized at chromosomal band 13q14.3 and in the BCL2 promotor was assessed with quantitative real-time PCR by using primers specific for bisulfite-converted and nonmethylated DNA from peripheral blood lymphocytes (PBL; n = 7), CD-19+ peripheral blood lymphocytes (B cells; n = 11), B-CLL patients with tumors biallelic for the critical region in 13q14.3 (disomic; n = 9), loss of one allele (deleted; n = 14), or loss of both alleles (biallelically deleted; n = 1). Depicted are the ratios of 13q14.3- orBCL2-specific primer pairs and 2 methylation-insensitive primer pairs that are specific for bisulfite-converted DNA (Eads et al26). The color coding gives fold nonmethylated DNA (green = methylated) as compared with the average of sorted B cells; gray = not done. (B) A double-sided Student t test shows no significant differences between the patient groups (disomic or deleted) and sorted B cells.
To verify this result and to rule out inhomogeneous methylation patterns that cannot be detected in the PCR-based approach, the CpG island associated with the RFP2 gene was completely bisulfite sequenced in 3 B-CLL patients and partially in theBCMS-associated CpG island (Figure 7 and data not shown). Only clones carrying converted DNA inserts were analyzed for each PCR product, and only 2 methylated cytosines could be detected in the critical region at chromosomal band 13q14.3 (Figure7), verifying that DNA-methylation patterns of CpG islands localized in chromosomal band 13q14.3 do not differ between normal B cells and B-CLL lymphocytes.
Bisulfite sequencing of the CpG island associated with the
RFP2 gene in 3 B-CLL patients. DNA was bisulfite-converted, PCR-amplified, and cloned, and single clones were sequenced. (A) In the genomic sequence of the 5′ end of theRFP2 gene, the transcription start of RFP2 is shown as an arrow, CpG doublets are depicted as open circles and numbers are bp. (B) Sequences of clones are shown as lines and methylated cytosines as vertical lines.
Bisulfite sequencing of the CpG island associated with the
RFP2 gene in 3 B-CLL patients. DNA was bisulfite-converted, PCR-amplified, and cloned, and single clones were sequenced. (A) In the genomic sequence of the 5′ end of theRFP2 gene, the transcription start of RFP2 is shown as an arrow, CpG doublets are depicted as open circles and numbers are bp. (B) Sequences of clones are shown as lines and methylated cytosines as vertical lines.
Discussion
More than 50% of B-CLL and almost 70% of MCL patients show loss of genomic material from a critical region localized in chromosomal band 13q14.3.3,5,27 Therefore, this region is the prime candidate for containing genes that lead to the expansion of malignant B-cell clones in B-CLL and MCL patients. However, despite multiple efforts by several groups, no mutations were found in genes localized in this region in B-CLL patients5,10,13 that would fit the 2-hit model for tumor suppressor genes.28 This finding prompted us to search for epigenetic aberrations at chromosomal band 13q14.3 in B-CLL patients that could possibly identify the postulated tumor suppressor mechanism. Expression of the 3 genes located in the minimally deleted region was previously tested in B-CLL patients by Northern blot analysis: all patients showed RNA expression ofBCMS and RFP2, whereas expression of theBCMSUN gene in B-CLL patients was below the detection limit of the analysis.13 However, our expression analysis of genes localized in the critical region and the near vicinity with the more sensitive quantitative real-time PCR method showed significant down-regulation of a number of genes in B-CLL patients with loss of one allele. Only RFP2 mRNA was significantly down-regulated in all B-CLL patients, including those disomic at 13q14.
The RFP2 gene product is the homolog of the RET finger protein (RFP), which was cloned as the rearrangement partner of a tyrosine kinase, the RET protooncogene.29 The RFP protein has a tripartite motif: a RING-zinc-finger domain, which in other proteins has been shown to mediate ubiquitination (reviewed in Joazeiro & Weissman30), a B-box domain, and a coiled-coil domain, which were shown to interact with the enhancer of Polycomb (E(Pc)) and thus strongly repress gene expression.31
To further elucidate the expression pattern of the respective region within 13q13.4, the mode of RFP2 down-regulation in B-CLL patients is of particular interest. Loss of expression has been shown in a variety of genes and tumors to be associated with methylation of the respective CpG islands (for review, see Jones & Laird18 and Baylin & Herman19). In B-CLL, genome-wide hypomethylation32 and hypomethylation of theBCL2 gene, an inhibitor of apoptosis, have been reported together with up-regulation of the BCL2 protein.23 We used real-time PCR to measure methylation of a small set of CpG dinucleotides localized in chromosomal band 13q14.3 in a larger group of patients combined with qualitative screening by bisulfite sequencing of 2 CpG islands in the minimally deleted region of 3 selected patients. No consistent B-CLL–specific methylation patterns were detected with either method. However, methylation of small regions within CpG islands that escapes detection by the methylation-specific PCR cannot be strictly excluded. Furthermore, it cannot be ruled out that the 3 B-CLL patients might not represent the entire range of alterations in B-CLL, or methylation occurs outside of the sequenced DNA segments. However, the data we present strongly argue for an involvement of RFP2 in the pathomechanism of B-CLL because of the frequent loss of genetic material harboring the gene in B-CLL patients and its significant down-regulation in B-CLL patients without detectable loss of the gene as compared with the other genes localized in the vicinity of the critical region that are not down-regulated in these patients. The down-regulation ofRFP2 gene expression is independent of methylation in the minimally deleted region. It is intriguing that the majority of the genes localized in the vicinity of the minimally deleted region have homologs, which are involved in regulation of chromatin condensation, such as the homolog of the RFP2 gene,RFP,31,CHC1L,33,CLLD7 and 8,34 and KPNA3.35 Genes involved in the regulation of chromatin condensation have been found defective in acute lymphocytic leukemia (MLL alias HRX36 aliasALL137), a disease with genomic translocations near the BCMS gene,38 and in MCL (BMI-139).
Clustering of genes involved in chromatin condensation near the minimally deleted region in B-CLL suggests the existence of a functional unit localized at 13q14.3. If this is the case, loss of expression of RFP2 could be an early event, which is enhanced in B-CLL patients by later genetic loss of the remaining functional unit. For functional analysis of the pathomechanism involving the critical region localized at chromosomal band 13q14.3, it might, therefore, be worthwhile to investigate the chromatin status of this genomic region.
We thank Frank Lyko and Ruthild G. Weber for helpful discussions and Christian Korz as well as Armin Pscherer for support.
Supported by grants 01KW9935, 01KW9937, and 01SF9903/3 from the Bundesministerium für Bildung und Forschung (BMBF) and by the European Union (EU) QLG1-CT-2000-00687.
The publication costs of this article were defrayed in part by page charge payment. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 U.S.C. section 1734.
