INTRODUCTION

ICBP90 (gene symbol UHRF1), which was first identified as a topoisomerase II alpha regulator, is a very important protein that is associated with cell cycle progression and is overexpressed in proliferating cells and cancer cells (1–3). ICBP90 has a binding affinity to methylated CpGs, makes a complex with HDAC1, and binds to the promoters of methylated tumor suppressor genes, suggesting that ICBP90 promotes cancer cell proliferation through suppression of these tumor suppressor genes (2). ICBP90 is also down-regulated by p53 (4), possesses ubiquitin E3 ligase activity (5), and plays a role in maintaining DNA methylation with DNMT1, keeping the genes silent (6, 7). ICBP90 has three family proteins, NIRF/UHRF2, ICBP55/ UHRF3 and ICBP87/UHRF4, that share 52.6, 59.8 and 78.1% identity with ICBP90, respectively (3).

Recently an article titled “Isolation and Characterization of a Novel Human Radiosusceptibility Gene, NP95” was published in Radiation Research (8). The authors claimed that they identified a new human homolog of mouse Np95 and the “new” gene was identical to UHRF1 but not to ICBP90. However, because the gene symbol of ICBP90 is UHRF1 (i.e., they are the same) and the ICBP90 sequence has been registered for a long time (AF129507, submitted by Hopfner et al. on Feb. 19, 1999), we are afraid that the article will create major confusion in the future. We would like to discuss the possibilities that could have led to the confusion.

MATERIALS AND METHODS

RESULTS AND DISCUSSION

Concerning the recently published article entitled “Isolation and Characterization of a Novel Human Radiosusceptibility Gene, NP95” (8), we have a strong concern that the authors did not isolate a new gene. In the discussion of the article, it is described that they could not recognize the sequence of exon F (2.6 kb) of the ICBP90 gene in the Homo sapiens chromosome 19 genomic contig, NT_011255 (GenBank), at 19p13.3, but found this exon in a BAC clone, AC112777, that includes 12p12.2-p12.1. Hopfner et al. (9) have reported that the ICBP90 gene spans approximately 35.8 kb on chromosome 19p13.3 and contains six coding exons named A to F. We carefully checked the genomic structure of the ICBP90 gene and found that exon F is actually composed of 11 exons (Fig. 1). All exons of ICBP90 are on the contig, NT_011255, which is composed of many genomic fragments, including AC027319, AC053467 and AB07561, which are discussed in the article. Exons 1 to 14 are on AC027319, exons 15 to 17 are on AB075601, and exons 16 to 18 are on AC053467. Therefore, the ICBP90 gene is localized at the same position where the “new” human NP95 homolog is located. We think that this incorrect definition of exon F (9) is one of the major sources of confusion that led Muto et al. to think that the ICBP90 gene is different from the gene that they identified.

It is also stated in their discussion that human NP95 and UHRF1 are identical in ORF and that ICBP90 and human NP95 differ in ORF by only two amino acids (Lys383 → Asn383, Ala457 → Ser457) and seven nucleotides. A difference of nine nucleotides in a long identical sequence (in this case, ICBP90's ORF is 2382 bp) does not make it a different gene. Usually these differences are polymorphisms or sequence errors. Therefore, we searched SNP databases and found 366 polymorphisms including 308 single nucleotide polymorphisms (SNPs) and 58 deletions on the gene (Table 1, Supplemental Table 1). Among these 366 polymorphisms, 337 were located in introns, six were in 3′UTR, one was in 5′UTR, and 22 were in exons (Table 1). Among the 22 polymorphisms in exons, 15 were synonymous substitutions and seven were nonsynonymous substitutions (Table 2).

We are not sure whether all nine nucleotide differences are polymorphisms, because there are no descriptions of seven of the nine different nucleotides in the article. However, the two differences (Lys383 → Asn383, Ala457 → Ser457) that are described in the article were not involved in the 366 polymorphisms. We determined the sequence by sequencing genomic DNA from several cancer cell lines and Japanese individuals and determined that codon 383 is AAG (Lys) and codon 457 is GCG (Ala); on the other hand, codon 383 is AAT (Asn) and codon 457 is TCG (Ser) in the registered ICBP90 sequence (AF129507). Therefore, AF129507 includes at least two rare polymorphisms that are not registered in the database or are sequence errors at codons 383 and 457. The other seven different nucleotides could be polymorphisms. Unoki et al. (2) used ICBP90 plasmid vectors that include Lys383 and Ala457, not Asn383 and Ser457, and three different polymorphisms compared with AF129507, because these polymorphisms were major alleles in the database. If the claim of Muto et al. (8) is right, Unoki et al. (2) did not use ICBP90 in their experiments. But no one accepts that two genes at exactly the same location in a chromosome with 99.748% amino acid identity (only two amino acids are different among 754 amino acids) are different.

Our other concern is the 5′UTR splice variations that are described in Fig. 2 (8). Hopfner et al. (9) reported three 5′UTR variations that are different from the variations in their article. We checked the EST database using BLAST and found that the three different 5′UTR variations that Hopfner et al. (9) described were registered as EST fragments, indicating that they are actually expressed in cells (Fig. 2, Supplemental Table 2). However, there are no ESTs that include exon 1 and 2 as described by Muto et al. (8). Transcripts that include these exons may be expressed at a low level, but we need expression data. At the least, they should add the 5′UTR variation that Hopfner et al. (9) identified in their article.

The authors also claimed that their newly isolated gene differs in its effect on topoisomerase II activity. However, the authors did not compare the activity of their “new” gene with ICBP90 in the same experiment. The difference between two proteins should be compared in the same experiment. The effect of ICBP90 on topoisomerase II α activation might not be as strong. In fact, Unoki et al. (2) got a similar result. They showed that ICBP90 has a stronger affinity for methylated CpG than for the CCAAT repeat at the topoisomerase promoter. However, as is well known, to get the maximum protein activity, special conditions are required (salt concentration, binding partner, etc.), and the condition that Unoki et al. (2) used might not be suitable to get the strongest activation of topoisomerase by ICBP90. Although the results of Muto et al. (8) are interesting, it is too early to conclude anything.

In conclusion, the genomic structure of ICBP90 that was defined previously was revised (Fig. 1), and the registered ICBP90 sequence (AF129507) includes at least two rare polymorphisms or sequence errors. All exons of the ICBP90 gene are on 19p13.3. ICBP90 (gene symbol UHRF1) is a human homolog of mouse Np95, as we have described before (3). Actually, because ICBP90 and mouse Np95 share 73.3–73.7% amino acid identity, the percentage is a little low compared to an ordinary homolog between human and mouse. However, because the human genome project was finished in 2003, we do not think that there is another human NP95 homolog on the human genome. If there is another one, it should share a higher percentage of amino acid identity with mouse Np95 than ICBP90. In addition, functional differences should be compared in the same experiment. Since the gene that the authors isolated and ICBP90 are identical, the authors' functional analysis data seem to suggest that ICBP90's function is the same as Unoki et al. reported (2) where they used the same sequence that the authors identified. We hope that this commentary can prevent future confusion.

SUPPLEMENTARY INFORMATION

Supplementary Table 1. Detailed information for all SNPs on the ICBP90/UHRF1 gene.  http://dx.doi.org/10.1667/RR1209.1.s1

Supplementary Table 2. ESTs that include ICBP90/ UHRF1 5′UTR variation (available August 13, 2007).  http://dx.doi.org/10.1667/RR1209.1.s2