Nitrilase and Fhit homologs are encoded as fusion proteins in Drosophila melanogaster and Caenorhabditis elegans

Abstract

The tumor suppressor gene FHIT encompasses the common human chromosomal fragile site at 3p14.2 and numerous cancer cell biallelic deletions. To study Fhit function we cloned and characterized FHIT genes from Drosophila melanogaster and Caenorhabditis elegans. Both genes code for fusion proteins in which the Fhit domain is fused with a novel domain showing homology to bacterial and plant nitrilases; the D. melanogaster fusion protein exhibited diadenosine triphosphate (ApppA) hydrolase activity expected of an authentic Fhit homolog. In human and mouse, the nitrilase homologs and Fhit are encoded by two different genes: FHIT and NIT1, localized on chromosomes 3 and 1 in human, and 14 and 1 in mouse, respectively. We cloned and characterized human and murine NIT1 genes and determined their exon-intron structure, patterns of expression, and alternative processing of their mRNAs. The tissue specificity of expression of murine Fhit and Nit1 genes was nearly identical. Because fusion proteins with dual or triple enzymatic activities have been found to carry out specific steps in a given biochemical or biosynthetic pathway, we postulate that Fhit and Nit1 likewise collaborate in a biochemical or cellular pathway in mammalian cells.

The human FHIT gene at chromosome 3p14.2, spanning the constitutive chromosomal fragile site FRA3B, often is rearranged in the most common forms of human cancer and is a candidate tumor suppressor gene (1). The human FHIT gene, which is greater than 1 megabase in size, encodes an mRNA of 1.1 kilobases and a protein of 147 amino acids.

The rearrangements most commonly seen are deletions within the gene (1–4). These deletions, often occurring independently in both alleles and resulting in inactivation, have been reported in tumor-derived cell lines and primary tumors of lung (2), head and neck (3), stomach, colon (1), and other organs. In cell lines derived from several tumor types, DNA rearrangements in the FHIT locus correlated with RNA and/or Fhit protein alterations (4).

Because the inactivation of the FHIT gene by point mutations has not been demonstrated conclusively and several reports showed the amplification of aberrant-sized FHIT reverse transcription–PCR (RT-PCR) products from normal cell RNA (5, 6), a number of investigators have suggested that the FHIT gene may not be a tumor suppressor gene. On the other hand, it recently was reported that re-expression of Fhit in lung, stomach, and kidney tumor cell lines lacking endogenous protein suppressed tumorigenicity in vivo in four of four cancer cell lines (7). This finding suggests that FHIT is indeed a tumor suppressor gene. The same report suggested that Fhit enzymatic activity is not required for its tumor suppressor function (7).

Fhit protein is a member of the histidine triad (HIT) superfamily of nucleotide binding proteins (8) and is similar to the Schizosaccharomyces pombe diadenosine tetraphosphate (Ap₄A) hydrolase. Barnes et al. (9) reported that, in vitro, Fhit has diadenosine triphosphate (ApppA) hydrolase enzymatic activity.

Neither the in vivo function of Fhit nor the mechanism of its tumor suppressor activity is known. Nonetheless, genetic, biochemical, and crystallographic analysis suggest that the enzyme-substrate complex is the active form that signals for tumor suppression (10). One approach to investigate function is to study Fhit in model organisms such as Drosophila melanogaster and Caenorhabditis elegans.

Here we describe the isolation and characterization of the Fhit gene in these organisms. Fhit occurs in a fusion protein, NitFhit, in D. melanogaster and C. elegans although FHIT and NIT1 are separate genes in mammalian cells. The human and mouse NIT1 genes are members of an uncharacterized mammalian gene family with homology to bacterial and plant nitrilases, enzymes that cleave nitriles and organic amides to the corresponding carboxylic acids plus ammonia (11).

MATERIALS AND METHODS

Genomic and cDNA Clones.

One million plaques of a mouse genomic library (bacteriophage library from strain SVJ129, Stratagene) and 100,000 plaques of a D. melanogaster genomic library were screened with corresponding cDNA probes. Clones were purified and DNA was isolated (12). Sequencing was carried out by using Perkin—Elmer thermal cyclers and Applied Biosystems 377 automated DNA sequencers. DNA pools from a human (bacterial artificial chromosome) library (Research Genetics, Huntsville, AL) were screened by PCR with NIT1 primers (TCTGAAACTGCAGTCTGACCTCA and CAGGCACAGCTCCCCTCACTT) according to the supplier’s protocol. The DNA from the positive clone, 31K11, has been isolated by using standard procedures (12) and sequenced. Chromosomal localization of the human NIT1 gene was determined by using a radiation hybrid mapping panel (Research Genetics) according to the supplier’s protocol and with the same primers as above. To map murine Nit1 gene, Southern blot analysis of genomic DNA from progeny of a (AEJ/Gn-a bp^H/a bp^H × Mus spretus)F₁ × AEJ/Gn-a bp^H/a bp^H backcross was performed by using a full-length murine Nit cDNA probe (13). This probe detected a unique 2.0-kb DraI fragment in AEJ DNA and a unique 0.75-kb fragment in M. spretus DNA. Segregation of these fragments were followed in 180 N2 offspring of the backcross.

Additional Mit markers (D1Mit34, D1Mit35, and D1Mit209) were typed from DNA of 92 mice by using PCR consisting of an initial denaturation of 4 min at 94°C followed by 40 cycles of 94°C for 30 sec, 55°C for 30 sec, and 72°C for 30 sec. Linkage analysis was performed by using the computer program spretus madness: part deux. Human and mouse NIT1 expressed sequence tag (EST) clones were purchased from Research Genetics. The sequences of human and murine NIT1 genes and cDNAs and D. melanogaster and C. elegans NitFhit cDNAs have been deposited in GenBank.

In Situ Hybridization.

D. melanogaster polytene chromosome spreads were prepared from salivary glands of third-instar larvae as described (14). NitFhit DNA fragments were labeled with digoxigenin-11-dUTP by using a random-primed DNA labeling kit (Boehringer Mannheim) and were used as probes for the chromosomal in situ hybridization. Hybridization was for 20 hr at 37°C in hybridization buffer: 50% formamide, 2× standard saline citrate (SSC), 10% dextran sulfate, and 400 mg/ml of salmon sperm DNA. Antidigoxigenin-fluorescein antibodies (Boehringer Mannheim) were used for detection of hybridizing regions. DNA was counterstained with Hoechst 33258 (Sigma). The slides were analyzed by fluorescence microscopy. For in situ hybridization, embryos were fixed and processed as described previously (15), except that single-stranded RNA probes were used. Full-length NitFhit cDNA was cloned into BluescriptII KS+ vector and used to synthesize antisense RNA probes with the Genius 4 kit (Boehringer Mannheim).

RT-PCR, Northern, and Rapid Amplification of cDNA Ends (RACE) Analysis.

Human and mouse multiple tissue Northern blots (CLONTECH) were hybridized with corresponding NIT1 cDNA probes and washed by using the supplier’s protocol. For the HeLa cell line, total RNA was isolated from 1–5 × 10⁸ cells by using Trizol reagent (GIBCO/BRL). D. melanogaster poly(A)⁺ RNA was purchased from CLONTECH. Three micrograms of poly(A)⁺ RNA or 15 μg of total RNA were electrophoresed in 0.8% agarose in a borate buffer containing formaldehyde, transferred to HybondN⁺ membrane (Amersham) by using standard procedures and hybridized as described above. For RT-PCR, 200 ng of poly(A)⁺ RNA or 3 μg of total RNA were treated with DNaseI (amplification grade, GIBCO/BRL) following the manufacturer’s protocol. DNaseI-treated RNA was used in RT reactions as follows: 10 nM each dNTP, 100 pmol random hexamers [oligo(dT) priming was used in some cases], DNaseI-treated RNA, and 200 units of murine leukemia virus reverse transcriptase (GIBCO/BRL), in a total volume of 20 μl were incubated at 42° for 1 hr followed by the addition of 10 μg of RNase A and incubation at 37° for 30 min. One microliter of the reaction was used for each PCR. PCRs were carried out under standard conditions by using 10 pmol of each gene-specific primer and 25–35 cycles of: 95° 30", 55–60° 30", 72° 1′. Products were separated on 1.5% agarose gels and sometimes isolated and sequenced or cloned and sequenced. Oligo(dT)-primed double-stranded cDNA was synthesized by using procedures and reagents from the Marathon RACE cDNA amplification kit (CLONTECH); the cDNA was ligated to Marathon adapters (CLONTECH). 3′ and 5′ RACE products were generated by long PCR using gene-specific primers and the AP1 primer (CLONTECH). To increase the specificity of the procedure, the second PCR was carried out by using nested gene-specific primers and the AP2 primer (CLONTECH). PCRs were performed according to the Marathon protocol by using the Expand long template PCR system (Boehringer Mannheim) and 30 cycles of 94° 30′′, 60° 30′′, 68° 4′. RACE products were electrophoresed, identified by hybridization, and sequenced. Degenerate Fhit primers were: GTNGTNCCNGGNCAYGTNGT and ACRTGNACRTGYTTNACNGTYTGNGC. D. melanogaster Fhit RACE and RT-PCR primers were: GCGCCTTTGTGGCCTCGACTG and CGGTGGCGGAAGTTGTCTGGT. C. elegans Fhit RACE and RT-PCR primers were: GTGGCGGCTGCTCAAACTGG and TCGCGACGATGAACAAGTCGG. Human NIT1 RT-PCR primers were: GCCCTCCGGATCGGACCCT (exon1); GACCTACTCCCTATCCCGTC (exon 1a); GCTGCGAAGTGCACAGCTAAG and AAACTGAAGCCTCTTTCCTCTGAC (exon 1c); TGGGCTTCATCACCAGGCCT and CTGGGCTGAGCACAAAGTACTG (exon 2); GCTTGTCTGGCGTCGATGTTA (exon 3).

Protein Expression and Enzymatic Characterization.

The NitFhit cDNA was amplified with primers TGACGTCGACATATGTCAACTCTAGTTAATACCACG and TGGGTACCTCGACTAGCTTATGTCC, digested with NdeI and KpnI, and cloned into plasmid pSGA02 (16) as a NdeI–KpnI fragment. Escherichia coli strain SG100 (16) transformants were grown in Luria–Bertani medium with 100 μg/ml of ampicillin and 15 μg/ml of chloramphenicol at 15°C. When the culture reached an optical density (600 nm) of 0.25, isopropyl β-d-thiogalactoside was added to a final concentration of 200 μM. NitFhit protein was purified from inclusion bodies as described (17). Briefly, the cell pellet from a 1-liter culture was resuspended in 50 ml of 20 mM Tris⋅HCl (pH 7.5), 20% sucrose, 1 mM EDTA and repelleted. Outer cell walls were lysed by resuspension in ice water. Spheroplasts were pelleted, resuspended in 140 mM NaCl, 2.7 mM KCl, 12 mM Na⋅P04 (pH 7.3), 5 mM EDTA, 500 mM phenylmethylsulfonyl fluoride, 1 μg/ml of leupeptin, and 20 μg/ml of aprotinin, and sonicated. The resulting inclusion body preparation was washed and solubilized in 5 M guanidinium hydrochloride, 50 mM Tris⋅HCl (pH 8.0), 5 mM EDTA. Soluble NitFhit protein was added dropwise to 250 ml of 50 mM Tris⋅HCl (pH 8.0), 1 mM DTT, 20% glycerol at 40°C. After a 14-hr incubation, the 13-kg supernatant was concentrated 100-fold with a Centricon filter. A 1-liter culture yielded approximately 200 μg of partially purified, soluble NitFhit. ApppA hydrolase activity was assayed at 30°C in 20 μl of 50 mM Na⋅Hepes (pH 7.5), 10% glycerol, 0.5 mM MnCl2, 4 mM ApppA, 1 μM NitFhit. TLC plates were developed as described (9).

RESULTS

Cloning and Characterization of D. melanogaster and C. elegans Fhit Homologs.

To obtain D. melanogaster Fhit sequences, degenerate primers were designed to amplify sequences in the conserved regions of exons 5 and 7 of human FHIT. RT-PCR experiments with these primers and D. melanogaster RNA resulted in an ≈200-bp product whose translated sequence predicts a protein that showed ≈50% identity to human Fhit protein. This amplified sequence was used to design specific D. melanogaster Fhit primers. 5′ and 3′ RACE with these primers yielded an ≈1.5-kb full-length cDNA [including polyadenylation signal and poly(A) tail] encoding a 460-aa protein with a 145-aa C-terminal portion homologous to human Fhit (40% identity and 47% similarity) and a 315-aa N-terminal extension (Fig. 1). Northern analysis (Fig. 2C) showed a single band of ≈1.5 kb in both embryo and adult D. melanogaster, implying that the full-length cDNA had been cloned.

Figure 1.

Sequence comparison of human, murine, D. melanogaster, and C. elegans Nit1 and Fhit proteins. Identities are shown in black boxes, similarities are shown in shaded boxes. For human and mouse FHIT GenBank accession nos. are U46922 and AF047699, respectively.

Figure 2.

Expression of Nit1 and Fhit mRNAs in murine and human tissues. (A) Mouse multiple tissues Northern blot. Lanes 1–8: heart, brain, spleen, lung, liver, skeletal muscle, kidney, and testis. (Top) Fhit probe; (Middle) Nit1 probe; (Bottom) actin probe. (B) Human blot, NIT1 probe. Lanes 1–8: heart, brain, placenta, lung, liver, skeletal muscle, kidney, and pancreas. (C) Lanes 1 and 2: D. melanogaster adult, D. melanogaster embryo; D. melanogaster NitFhit probe. Lane 3: C. elegans adult; C. elegans NitFhit probe.

The 460-aa predicted protein sequence was used in a blastp search. Of the top 50 scoring alignments, 22 aligned with the 145-residue C-terminal segment (Fhit-related sequences) and 28 aligned with the 315-residue N-terminal segment. The 28 sequences aligning with the N terminus were led by an uncharacterized gene from chromosome X of Saccharomyces cerevisiae (P-value of 1.4 × 10⁻⁴⁵), followed by uncharacterized ORFs of many bacterial genomes and a series of enzymes from plants and bacteria that have been characterized as nitrilases and amidases (11). Thus, the 460-aa predicted protein contains a N-terminal nitrilase domain and a C-terminal Fhit domain and was designated NitFhit.

The D. melanogaster NitFhit cDNA probe was used to screen a D. melanogaster lambda genomic library. Sequencing of positive clones revealed that the gene is intronless and, interestingly, the 1.5-kb NitFhit gene is localized within the 1.6-kb intron 1 of the D. melanogaster homolog of the murine glycerol kinase (Gyk) gene (18). The direction of transcription of the NitFhit gene is opposite to that of the Gyk gene (Fig. 3A). It is not known whether such localization affects transcriptional regulation of these two genes.

Figure 3.

Genomic organization of human and murine Nit1 genes and D. melanogaster and C. elegans NitFhit genes. (A) Exon-intron structure of the genes. (B) Alternative processing of human NIT1 gene.

The cytological position of the NitFhit gene was determined by in situ hybridization to salivary gland polytene chromosomes. These experiments showed that there is only one copy of the sequence, which is localized to region 61A, at the tip of the left arm of chromosome 3 (not shown). Digoxigenin-labeled RNA probes were hybridized to whole-mount embryos to determine the pattern of expression during development. NitFhit RNA was found to be uniformly expressed throughout the embryo (not shown), suggesting that NitFhit protein could be important for most of the embryonic cells.

Because human Fhit protein and the D. melanogaster Fhit domain of NitFhit were only 40% identical, to show that we cloned the authentic D. melanogaster Fhit homolog, we tested its enzymatic activity. Fig. 4 shows that recombinant D. melanogaster NitFhit is capable of cleaving ApppA to AMP and ADP and therefore possesses ApppA hydrolase activity.

Figure 4.

Cleavage of ApppA by D. melanogaster NitFhit. At indicated times of incubation, samples were spotted on TLC plates with appropriate nucleotide standards.

C.

elegans. Fhit genomic sequences were identified from the Sanger database (contig Y56A3) by using blast searches. 5′ and 3′ RACE with C. elegans Fhit specific primers yielded a 1.4-kb cDNA [including polyadenylation signal and poly(A) tail] coding for a 440-aa protein (Fig. 1). Northern analysis (Fig. 2C) showed a single band of a similar size in adult worms. Similarly to D. melanogaster, the C. elegans protein containes an N-terminal nitrilase domain and a C-terminal Fhit domain (Fig. 1) with 50% identity and 57% similarity to human Fhit. Comparison between C. elegans NitFhit cDNA and genomic sequences from the Sanger database revealed that the C. elegans NitFhit gene comprises eight exons and is more than 6.5 kb in size (Fig. 3A); the nitrilase domain is encoded by exons 1–6, and the Fhit domain is encoded by exons 6–8. D. melanogaster and C. elegans NitFhit proteins are 50% identical and 59% similar and exhibit several conserved domains (Fig. 1).

Cloning and Characterization of Human and Murine NIT1 cDNAs and Genes.

Because Fhit and nitrilase domains are part of the same polypeptides in D. melanogaster and C. elegans, it is reasonable to suggest that they may be involved in the same biochemical or cellular pathway(s) in these organisms. Because we found that nitrilase homologs are conserved in animals, the mammalian nitrilase homologs were cloned as candidate Fhit-interacting proteins.

To obtain human and murine Nit1 sequences we used the D. melanogaster nitrilase domain sequence in blast searches of the GenBank EST database. We found numerous partially sequenced human and murine Nit1 homologous ESTs. All mouse Nit1 ESTs were identical as were all human NIT1 ESTs, suggesting the presence of a single NIT1 gene in mouse and human. To obtain the full-length human and mouse cDNAs we completely sequenced several human and mouse ESTs and human 5′ and 3′ RACE products. This resulted in the isolation of a ≈1.4-kb full-length human sequence encoding 327 amino acids and a ≈1.4-kb mouse full-length sequence coding for 323 amino acids (Fig. 1), although several alternatively spliced products were detected in both cases (see below and Fig. 3B). Both cDNAs are polyadenylated, but lack polyadenylation signals, although AT-rich regions are present at the very 3′ end of each cDNA. Mouse and human Nit1 amino acid sequences are 90% identical; the human Nit amino acid sequence is 58% similar and 50% identical to the C. elegans nitrilase domain and 63% similar and 53% identical to the D. melanogaster nitrilase domain (Fig. 1).

Murine lambda and human bacterial artificial chromosome genomic libraries were screened with the corresponding NIT1 cDNA probes, yielding one mouse lambda clone and one human BAC clone containing the NIT1 genes. The human and murine NIT1 genomic regions were sequenced and compared with the corresponding cDNA sequences. The genomic structure of human and mouse NIT1 genes is shown in Fig. 3A. Both genes are small: the human gene is ≈3.2 kb in size and contains seven exons; the murine gene is ≈3.6 kb in size and contains eight exons. Southern analysis confirmed that both human and mouse genomes harbor a single NIT1 gene (not shown).

A radiation hybrid mapping panel (GeneBridge 4) was used to determine the chromosomal localization of the human NIT1 gene. By analysis of PCR data at the Whitehead/Massachusetts Institute of Technology database (http://www-genome.wi.mit.edu), the NIT1 gene was localized to 6.94 cR from the marker CHLC.GATA43A04, which is located at 1q21–1q22.

A full-length murine Nit1 cDNA probe was used to determine the chromosomal location of the murine gene by linkage analysis. Interspecific backcross analysis of 180 N₂ mice demonstrated that the Nit1 locus cosegregated with several previously mapped loci on distal mouse chromosome 1. The region to which Nit1 maps was further defined by PCR of genomic DNA from 92 N₂ mice using the markers D1Mit34, D1Mit35, and D1Mit209 (Research Genetics). The following order of the genes typed in the cross and the ratio of recombinants to N₂ mice was obtained: centromere - D1Mit34 - 7/78 - D1Mit35 - 8/90 - Nit1 - 11/91 - D1Mit209 - telomere. The genetic distances given in centiMorgans (±SE) are as follows: centromere - D1Mit34 - 9.0 ± 3.2 - D1Mit35 - 8.9 ± 3.0 - Nit1 - 12.1 ± 3.4 - D1Mit209 - telomere. This region of mouse chromosome 1 (1q21–1q23) is syntenic to human chromosome 1q and is therefore consistent with the localization of the human ortholog of Nit1.

Expression and Alternative Splicing of Human and Murine NIT1 Genes.

For the human gene, Northern analysis revealed two major transcripts of ≈1.4 kb and ≈2.4 kb in all adult tissues and tumor cell lines tested. A third band of ≈1.2 kb was observed in adult muscle and heart (Fig. 2B). Our longest cDNA (≈1.4 kb) corresponds to the ≈1.4-kb transcript observed on Northern blots. The 1.2-kb band corresponds to transcript 1 on Fig. 3B (see below). It is not known whether the ≈2.4-kb RNA represents an additional transcript or an incompletely processed mRNA. No significant variation in human NIT1 mRNA levels was observed in different tissues (Fig. 2B). On the contrary, different mouse tissues showed different levels of expression of Nit1 mRNA (Fig. 2A). The highest levels of Nit1 mRNA were observed in mouse liver and kidney (Fig. 2A Middle, lanes 5 and 7). Interestingly, the pattern of Nit1 expression was similar to the pattern of expression of Fhit (Fig. 2A Top and Middle), supporting the hypothesis that the proteins may act in concert or participate in the same pathway.

Analysis of mouse Nit1 ESTs revealed that some transcripts lack exon 2 and encode a 323-aa protein. An alternative transcript containing exon 2 encodes a shorter, 290-aa protein starting with the methionine 34 (Fig. 1).

Analysis of human ESTs and 5′ RACE products from HeLa and testis also suggested alternative processing. To investigate this possibility, a series of RT-PCR experiments was carried out. Fig. 5 shows the results obtained from HeLa RNA (similar results were obtained by using RNAs from the MDA-MB-436 breast cancer cell line and adult liver). The alternatively spliced transcripts are shown on Fig. 3B. Transcript 1, lacking exon 2, was represented by several ESTs in the GenBank EST database. This transcript probably corresponds to the ≈1.2-kb transcript observed on Northern blots in adult muscle and heart. Transcript 2 encoding the 327-aa Nit1 protein (Fig. 1) is a major transcript of human NIT1 at least in the cell lines tested. This transcript lacks exons 1a and 1b. Transcript 3 has exons 1a and 1b; transcript 4 has exon 1a but lacks exon 1b (Fig. 3B). It is not known whether transcript 5 (lacking exon 2) starts from exon 1 or 1c.

Figure 5.

Analysis of alternative transcripts of human NIT1 by RT-PCR. RT-PCR of HeLa RNA was performed with primers in different exons. Lanes 1–6: exons 1 and 3 (transcript 2); exons 1C and 3 (transcript 5); exons 1A and 3 (transcripts 3, upper band and 4, lower band); exons 2 and 3 (transcripts 2–4); exons 1 and 1C (transcript 5); and exons 1 and 2 (transcript 2).

The alternative initiating methionines of different transcripts are shown on Fig. 3B. Future immunoblot experiments will determine which of these methionines are used in vivo. Our preliminary data suggest that at least in COS-7 cells transfected with a construct containing transcript 2, the methionine in exon 3 (shown in transcripts 1 and 3, Fig. 3B) initiates more efficiently than the methionine in exon 2 (Fig. 3B, transcript 2).

DISCUSSION

Although the frequent loss of Fhit expression in several common human cancers is well documented (19–21), and results supporting its tumor suppressor activity have been reported (7), the role of Fhit in normal and tumor cell biology and the mechanism of its action in vivo are unknown. The ApppA hydrolytic activity of Fhit (9) seems not to be required for its tumor suppressor function (7), and it has been suggested that the enzyme-substrate complex is the active form of Fhit (10). To facilitate an investigation of Fhit function, we initiated an analysis of the gene in model organisms by cloning and characterizing D. melanogaster and C. elegans Fhit genes.

Surprisingly, in flies and worms, Fhit is expressed as a fusion protein with the Fhit domain fused to a “Nit” domain showing homology to plant and bacterial nitrilases (11). We further isolated human and murine NIT1 genes. Nit and Fhit are expressed as separate proteins in mammals but, at the mRNA level, are coordinately expressed in mouse tissues.

In several eukaryotic biosynthetic pathways multiple steps are catalyzed by multifunctional proteins containing two or more enzymatic activities. The same steps in prokaryotes frequently are carried out by monoenzymatic proteins that are homologs of each domain of the corresponding eukaryotic protein (22). For example, Gars, Gart, and Airs are domains of the same protein in D. melanogaster and mammals. These domains catalyze different steps in de novo synthesis of purines. In yeast, Gart homolog (Ade8) is a separate protein, and Gars and Airs homologs (Ade5 and Ade7) are domains of a bienzymatic protein; in bacteria, all three homologs (PurM, PurN, and PurD) are separate proteins (22). De novo pyrimidine biosynthesis illustrates a similar case (23). Recently, a fusion protein of a lipoxygenase and catalase, both participating in the metabolism of fatty acids, has been identified in corals (24). In all of these examples, if domains of a multienzymatic protein in some organisms are expressed as individual proteins in other organisms, the individual proteins participate in the same pathways. This observation and the fact that Fhit and Nit1 exhibit similar expression patterns in murine tissues suggest that Fhit and Nit1 participate in the same cellular pathway in mammalian cells.

The critical questions are, what is this pathway and how is it altered in human cancers? Nothing is known about the function of nitrilase homologs in mammalian cells, and there is no obvious connection between the ability to bind and hydrolyze ApppA and the hydrolysis of nitrile-containing compounds. Enzymatic activity has yet to be demonstrated for animal nitrilase homologs. At this point we can only hypothesize about possible connections between the two enzymes. The best known nitrilases are plant enzymes responsible for synthesis of the plant growth factor auxin (25). One possibility is that an active form of Fhit stimulates or inhibits a nitrilase in production of growth-regulating compounds. Some of the plant nitrilases are tightly associated with plasma membranes (25). If this is true for the mammalian nitrilases and they can interact (directly or indirectly) with Fhit, they could be a factor that places Fhit in the proximity of the plasma membrane, similarly to the recruitment of Raf-1 by Ras (26). Genetic and biochemical analysis of the Nit/Fhit genes should help to reveal their roles in tumor progression.

ABBREVIATIONS

RT-PCR

reverse transcription–PCR

HIT

histidine triad

ApppA

diadenosine triphosphate

EST

expressed sequence tag

RACE

rapid amplification of cDNA ends