The human homolog of Escherichia coli Orn degrades small single-stranded RNA and DNA oligomers.

We report here the identification of human homologues to the essential Escherichia coli Orn protein and the related yeast mitochondrial DNA-escape pathway regulatory factor Ynt20. The human proteins appear to arise from alternatively spliced transcripts, and are thus identical, except the human Ynt20 equivalent contains an NH(2)-terminal extension that possesses a predicted mitochondrial protease cleavage signal. In vitro analysis revealed that the smaller human protein exhibits a 3' to 5' exonuclease activity for small (primarily </=5 nucleotides in length) single-stranded RNA and DNA oligomers. We have named this human protein Sfn for small fragment nuclease to reflect its broad substrate range, and have termed the longer protein hSfnalpha. Sfn prefers Mn(2+) as a metal cofactor and displays a temperature-resistant (to 50 degrees C) nuclease activity. Kinetic analysis indicates that Sfn exhibits a similar affinity for small RNAs and DNAs (K(m) of approximately 1.5 micrometer), but degrades small RNAs approximately 4-fold more efficiently than DNA. Mutation of a conserved aspartate (Asp(136)) to alanine abolishes both nuclease activities of Sfn. Northern blot analysis revealed that a 1-kilobase transcript corresponding to SFN and/or SFNalpha (these mRNAs differ by only two nucleotides) is expressed at varying levels in all fetal and adult human tissues examined. Expressed tag sequence clone analysis found that the two splice variants, SFN to SFNalpha, are present at a ratio of roughly 4 to 1, respectively. The results presented within suggest a role for human Sfn in cellular nucleotide recycling.

Nucleases are critical components of DNA and RNA metabolism, carrying out functions in DNA repair, replication and recombination, and RNA processing and degradation. Several candidate nucleases were recently classified as members of the 3Ј to 5Ј exonuclease superfamily (1). This superfamily includes RNases (such as RNase T and D), the proofreading domains of pol I family DNA polymerases, and DNases that exist as independent proteins (such as Escherichia coli Exo1) or as domains within larger polypeptides (such as a region within the Werner's syndrome helicase) (1)(2)(3). Homology within the superfamily is centered around three conserved exonuclease motifs (Exo I, II, and III) (2). The crystal structure of the Klenow fragment of E. coli DNA polymerase I and complementing site-directed mutagenesis studies indicate that the Exo I, II, and III motifs are clustered around the active site and contain four conserved negatively charged residues that are critical for coordinating the two metal ions involved in phosphodiester bond cleavage (reviewed in Ref. 4).
Based on conservation of the Exo motifs, the protein encoded by the E. coli open reading frame YjeR was placed into the 3Ј to 5Ј exonuclease superfamily (5). This protein was subsequently shown in vitro to be an exonuclease specific for RNA molecules shorter than about 5 nucleotides (nt) 1 in length (6). Based on this ribonuclease activity, YjeR was renamed Orn for oligoribonuclease (6). Data base mining revealed a human EST homolog to YjeR, indicating an evolutionary conserved role for the encoded protein (5).
Orn is one of eight distinct 3Ј to 5Ј exoribonucleases present in E. coli (7). It is a processive enzyme that initiates attack at a free 3Ј hydroxyl group on single-stranded RNA, releasing 5Ј-mononucleotides in a sequential manner (6). Notably, unlike previously characterized E. coli ribonucleases, ORN is an essential gene. Experiments using a temperature-sensitive ORN construct and pulse-chase experiments with radiolabeled RNA revealed an accumulation of small RNA oligonucleotides at the non-permissive temperature (8). These observations suggested a role for Orn in the final steps of mRNA degradation, since mRNA-degrading enzymes such as RNase II and polynucleotide phosphorylase generate small oligonucleotide fragments that require further processing to mononucleotides (8).
An ORN homologue is also present in yeast Saccharomyces cerevisiae (5,9). This YNT20 gene encodes a larger protein that has been shown to localize to the mitochondria (9). YNT20 plays a role in a S. cerevisiae mitochondrial DNA escape pathway, a process that involves the transfer of genetic material from the mitochondria to the nucleus (9).
It was suggested, based on sequence comparisons, that the S. cerevisiae Ynt20 protein is potentially capable of hydrolyzing both DNA and RNA (9). Although it has been reported that the E. coli Orn protein exhibits no detectable DNase activity on double stranded T7 chromosomal DNA (10), its ability to degrade short deoxyribonucleotides apparently has not been examined. We set out to examine the substrate specificity of the human Orn equivalent, with a particular interest in its activity for DNA molecules, and demonstrate here that this protein cleaves not only short RNAs but also short DNAs in a 3Ј to 5Ј direction. Thus, we suggest that the human enzyme be named Sfn for small fragment nuclease. We discuss our results as they may relate to the cellular role of this protein in nucleic acid metabolism.

EXPERIMENTAL PROCEDURES
Identification of SFN and SFN␣ cDNAs-The EST data base was searched using the TBLASTN program to identify human cDNA clones that encode a protein homologous to the E. coli Orn protein (GenBank accession number P39287). IMAGE clone 663460 was chosen, sequenced (11), and found to contain an open reading frame encoding a 205-amino acid protein. This protein is identical to CGI-114 protein (accession number is AF151872). The SFN␣ cDNA was found in the GenBank data base (accession number AL110239).
Chromosomal Localization of the Human SFN Gene-The sequence at the 5Ј end (accession number AA224296) or the 3Ј end (accession number AA224194) of the SFN gene, when compared with the "Sequenced Tagged Site" data base at National Cancer Biology Institute, identified STS-H98169 as the genetic locus of SFN. This sequencetagged site is positioned at chromosome 11 in a region between markers D11S1347 and D11S939 (110.3-117.9 centimorgan), which corresponds to the region 11q23.1-11q23.2.
Buffers and Reagents-All reagents were purchased from Sigma unless otherwise indicated. Restriction enzymes were purchased from New England Biolabs. Labeled nucleotides were from Amersham Pharmacia Biotech. Spectrophotometric grade glycerol was obtained from Fisher. Olio(dT) 4 -22 Ladder was purchased from Life Technologies, Inc. and consists of single-stranded DNA oligos from 4 to 22 nt in length, increasing by 1-nucleotide increments. DNA oligos were obtained from Operon Biotechnologies (Alameda, CA). Synthetic RNA was obtained from Dharmacon Research (Boulder, CO). L buffer (Lysis buffer) con- The stars indicate the catalytically important tyrosine and the four conserved negatively charged residues that are involved metal ion binding (1,4). The nuclear localization signal (NLS), KKRK, is boxed (18). q is used to mark the putative mitochondrial targeting signal, ARGVR, for hSFN␣ (17). B, sequences of the putative splice site in hSFN (accession number AF151872) and hSFN␣ (AL110239) mRNAs. The first nucleotide of the respective protein coding sequence is designated as ϩ1. The first codon of hSFN is boxed. The differences in the two sequences are underlined.
Purification of Recombinant Sfn-His Protein-The SFN coding region was PCR amplified using primers NCO5ЈHYJER (5Ј-GCATGCCT-GGCGGCAGGGGAGAGCAT) and HIND3ЈYJER (3Ј-GCAGTAAGCT-TACTCACGGTCTTCTCAT), and subcloned after digestion into the NcoI and HindIII sites of pET28d (Novagen, Madison, WI) to generate phyjeR-His. This construct allows for expression of six histidine residues on the carboxyl terminus under the control of a T7 RNA polymerase promoter. phyjeR-His plasmid was sequenced and no PCR errors were found.
The phyjeR-His plasmid was transformed into BL21(DE3)/pLysS E. coli strain (Novagen). An overnight culture of 100 ml was grown at 37°C in LB (1% bactotryptone, 0.5% bactoyeast extract, and 1% NaCl) with 50 g/ml kanamycin and 25 g/ml chloramphenicol. The overnight culture was used to inoculate 1 liter of the same medium. This culture was grown at 37°C with vigorous aeration until the A 550 nm was 0.8. Isopropyl-1-thio-␤-galactopyranoside (final concentration of 1 mM) was then added to induce Sfn-His protein expression. At 4 h after induction, cells were harvested. The cell paste was resuspended in 30 ml of L buffer. The cell suspension was sonicated using a Misonix XL sonicator by three 1-min bursts with a microtip at the maximum setting. The cell lysate was centrifuged at 27,000 ϫ g for 30 min at 4°C. To the supernatant, polyethyleneimine was added slowly with constant stirring to a final concentration of 0.25% to remove nucleic acids (12). The suspension was centrifuged and the supernatant was loaded at 1.5 ml/min onto a 2-ml S2 cation exchange column equilibrated with L buffer using the BioLogic Workstation FPLC system (Bio-Rad). Fractionation over the cation exchange column removed the contaminating proteins found to bind nonspecifically to the subsequent Ni affinity column. The S2 flowthrough was collected, and NaCl and imidazole were added to final concentrations of 500 and 5 mM, respectively. The S2 flow-through was then incubated with 2 ml of Ni-NTA resin (Qiagen, Santa Clarita, CA) for 1 h at 4°C with gentle rocking. This suspension was centrifuged, and the resin was washed 2 times each with 20 ml of W5 buffer, followed by 4 washes each with 10 ml of W40 buffer. Sfn-His proteins were then eluted from the Ni-NTA resin 4 times, each with 2 ml of E250 buffer, dialyzed overnight against L buffer containing 0.1 mM dithiothreitol, and stored at Ϫ70°C. Concentrations of protein solutions were determined by measuring the absorbance at 280 nm and using the theoretical molar extinction coefficient E 280 nm of 25,940 M Ϫ1 cm Ϫ1 calculated for Sfn-His (13).
Site-directed Mutagenesis of the SFN cDNA-Mutagenesis was performed using the overlapped PCR method as described (14). The sitespecific mutant oligo used was D136A, 5Ј-TATAGAATAATTGCTGT-GAGCACTGTTAAG-3Ј, with the mutated codon underlined. We chose to mutate Asp 136 because a structurally equivalent mutation in E. coli DNA pol I (D501A) caused a 13,000-fold decrease in 3Ј-5Ј exonuclease activity (15). The final SFN mutant PCR product was digested with NcoI and HindIII and subcloned into the same restriction sites of pET28d. This construct was sequenced as described (11).
Kinetics of Sfn Exonuclease Activity on Single-Stranded DNA and RNA Substrates-Nuclease assays were performed under standard buffer conditions as described above at 37°C, with Sfn at a final concentration of 10 nM. The reaction was incubated for 5 min with RNA5 or 40 min with DNA5, and was within the linear range of enzymatic activity (i.e. Յ15% of the substrate was converted to product). The substrate concentration range was 0.1-3.2 M. The apparent Michaelis-Menten constant (K m ) and maximal velocity (V max ) were obtained from a double-reciprocal Lineweaver-Burk plot. A linear plot of 1/[S] versus 1/V (where [S] is the substrate concentration in M and V is the velocity in M min Ϫ1 units) produces a slope of K m /V max and a y intercept of 1/V max . k cat was calculated from the equation (k cat )(E t ) ϭ V max , where E t is the total concentration of enzyme in the assay. Linear regression analysis was performed using CricketGraph software (Cricket Software, Philadelphia, PA).
Northern Blot Analysis-Northern blots were prehybridized for 3 h at 65°C in ExpressHyb Solution (CLONTECH, Palo Alto, CA). The cDNA probe (SFN PCR product described above) was labeled using the Megaprime DNA Labeling System (Amersham Pharmacia Biotech) and [␣-32 P]dCTP (Amersham Pharmacia Biotech), and hybridized in Ex-pressHyb Solution at 65°C for 2-3 h. Blots were washed once at room temperature for 30 min and twice at 50°C for 30 min with 50 mM NaPO 4 , pH 7.4, 0.5% SDS, and 1 mM EDTA. Images were exposed on Kodak Bio-Max MS film for 16 h and developed. Blots were normalized with ␤-actin transcripts.

RESULTS
The SFN cDNA, Genomic Location, and mRNA Expression Pattern-A cDNA containing the SFN coding region was obtained from the EST data base (see "Experimental Procedures"). By blasting the sequence of the 5Ј end (accession number AA224296) or the 3Ј end (accession number AA224194) of the SFN cDNA against the NCBI UniGene data base, we found that the human SFN gene mapped to chromosome 11 at position 11q23.1-11q23.2. Four additional genes also map within this vicinity: apolipoprotein A-I (apoA-I), human serotonin receptor 3 (HTR3), human nicotinamide N-methyltransferase (NNMT), and zinc finger protein ZNF259 (PLZF).
The SFN cDNA codes for a 205-amino acid protein of 23,754 daltons, with a theoretical pI of 5.6. SFN belongs to the YjeR/ ORN family, which is a distinct subgroup of the 3Ј to 5Ј exonuclease superfamily (1,5). Fig. 1 shows a comparison of the amino acid sequence of the Sfn-like proteins from bacteria, yeast, plant, worm, mouse, and human. The human Sfn protein is ϳ50% identical to its E. coli counterpart, the Orn protein (5). Sfn and E. coli Orn possess the three characteristic sequence motifs, termed Exo I, II, and III (2), of the 3Ј to 5Ј exonuclease superfamily (1) (Fig. 1A). Four conserved negatively charged residues within these three Exo motifs (shown in Fig. 1A) are involved in positioning of the two divalent cations required for catalysis and phosphodiester bond cleavage (reviewed in Ref. 4).
Upon further examination of the GenBank data base, an identical protein to hSfn, but with an extended NH 2 -terminal domain, was identified (Fig. 1A). This protein, which we propose to call Sfn␣, appears to be the human equivalent to the yeast Ynt20 protein and to have arisen from an alternative RNA splicing event where two additional nucleotides were introduced (Fig. 1B). Using the PSORT II computer search program, a putative consensus cleavage site motif for mitochondrial processing proteases (ARGVR) (17) was identified within the unique NH 2 -terminal portion of Sfn␣. Both Sfn and Sfn␣ contain a consensus nuclear localization signal (KKRK) (18) (Fig. 1A). It seems logical that Sfn would be targeted to the nucleus and that Sfn␣ would translocate predominantly to the mitochondria.
Northern blotting revealed that the SFN cDNA probe detects a single transcript of ϳ1 kilobase in all fetal and adult tissues examined (Fig. 2). Comparatively, relatively low mRNA levels are observed in adult lymph nodes, brain, lung, liver, spleen, and thymus, with highest levels observed in heart. Since the human SFN and SFN␣ cDNAs appear to differ by only two nucleotides, it was not possible to distinguish the two transcripts here. Of the 10 human ESTs found in the data base, 8 contained a sequence identical to the SFN cDNA and two maintained a sequence identical to SFN␣.
Purification of Overproduced SFN-His Fusion Protein-We have characterized here Sfn, and expect that Sfn␣ will exhibit similar substrate specificity as it maintains the same core nuclease domain (Fig. 1A). The SFN gene was subcloned into the pET28d plasmid to produce a COOH-terminal-tagged Sfn-His fusion protein, and the recombinant protein was purified as described under "Experimental Procedures." From 1 liter of induced bacterial culture (about 5 g of wet cell pellet), we obtained ϳ4 mg of Sfn-His fusion protein of Ͼ95% purity (Fig.  3A). As shown in Fig. 3B, the Sfn-His fusion protein migrates as a single symmetrical peak on an analytical gel filtration column that corresponds to a globular protein with a molecular mass of ϳ90 kDa. Since the calculated molecular weight of Sfn is 24,000, Sfn-His fusion protein appears to migrate as a tetramer. However, since the elution position on a gel filtration column is dependent on size and shape of the protein, it is possible that the Sfn-His fusion protein exists in solution as a rod-shaped protein of lower oligomeric state, since rod-shaped proteins are known to migrate slower than expected for their molecular weight. 2 Unlike the human Sfn-His protein, the E.
coli Orn protein was reported to be a dimer based on its gel filtration chromatographic profile, although it should be noted that the experimental conditions differed slightly from those used here (6).
Exonuclease Activity of Sfn-His Protein-Short fragments of Յ5 ribonucleotides were shown to be the optimal substrate for E. coli Orn protein (6, 10). Using a 5-nucleotide RNA (RNA5), labeled at the 5Ј end, as the substrate, we determined the divalent metal cation preference of the human Sfn. At metal concentrations of 0.01, 0.05, 0.1, 0.5, 1, 5, and 10 mM, only Mg 2ϩ and Mn 2ϩ had any stimulatory effect on Sfn-His exonuclease activity, with Mg 2ϩ being ϳ10-fold less effective than Mn 2ϩ at 10 mM (Fig. 4A and data not shown)  (and a preference for) Mn 2ϩ (Fig. 4A; Ref. 10). The RNase activity of Sfn-His also has a temperature optimum around 50°C (Fig. 4B), similar to that of E. coli Orn protein (10). Thus both proteins are quite thermostable. The RNA-specific degradation activity of Sfn at 37°C is ϳ2-fold less than that at 50°C (Fig. 4B). Since we observed degraded 4-mer products prior to detecting mononucleotide products (Fig. 4C), Sfn-His appears, as expected, to be a 3Ј to 5Ј exonuclease.
To determine whether short single-stranded DNA can serve as substrate for Sfn-His, we used a single-stranded oligo(dT) ladder from 4 to 22 bases in length, and single-stranded DNA oligos of 5 and 8 nt. As shown in Fig. 5, Sfn-His is capable of degrading short single-stranded DNA, although with less efficiency than 5-mer RNAs. Furthermore, its DNase activity is inversely related to the length of the DNA oligo, as seen for the RNase activity of E. coli Orn (19). The data above is also consistent with the lack of DNase activity reported for Orn on double stranded T7 chromosomal (large) DNA (10). The DNase and RNase Activities Are Intrinsic to Sfn-His-To confirm that the observed DNase and RNase activities are intrinsic to the Sfn-His protein, a catalytically inactive Sfn-His mutant was constructed. Amino acid residue Asp 136 (Fig. 1, in Exo III motif) was selected because an aspartate to alanine mutation at the structurally equivalent residue (5) in E. coli DNA pol I causes a 13,000-fold decrease in 3Ј to 5Ј exonuclease activity (14). The D136A Sfn-His mutant was purified using the same procedure as described for wild type Sfn-His protein (Fig.  6A). When compared with wild type Sfn-His, D136A Sfn-His mutant displayed a Ն50-fold reduced exonuclease activity for both RNA5 and DNA5 substrates (Fig. 6B). These data are consistent with both nuclease activities being intrinsic to the Sfn-His protein, and not being the result of protein contaminants. The results also demonstrate the importance of this acidic residue in the enzymatic function of the 3Ј to 5Ј exonuclease family members (1).
We subsequently examined Sfn nuclease activity on DNA structures known to be products/substrates of DNA repair processes. As might be expected from the results above, in the presence of Mn 2ϩ or Mg 2ϩ , we did not detect any protein-dependent degradation of 42-base pair double-stranded DNAs of different configurations, including gapped, nicked, 4-nucleotide recessed or overhanged 3Ј ends, or flap structures (data not shown; substrates described in Ref. 20). FIG. 4. Metal cofactor requirement, temperature optimum, and time course of Sfn-His RNase activity. A, metal preference of Sfn-His. 100 fmol of labeled RNA5 was incubated with 1 pmol of Sfn-His at 37°C for 30 min in the presence of 10 mM of the different divalent metal cations as indicated. The y axis is relative exonuclease activity. Minimal detection for this assay was Յ1 fmol of product which is equivalent to ϳ2.5% of the activity detected with Mn 2ϩ as shown for Cu 2ϩ , Ca 2ϩ , Ni 2ϩ , and Zn 2ϩ . B, temperature optimum of Sfn-His. 100 fmol of labeled RNA5 was incubated with 500 fmol of Sfn-His for 3 or 10 min at the temperature indicated. Lane 1 is no protein control. Lanes 2 and 3 represent reactions performed at 21°C; lanes 4 and 5, 37°C; lanes 6 and 7, 50°C; lanes 8 and 9, 65°C; and lanes 10 and 11, 80°C. %P is the percentage of labeled substrate that has been converted to the final mononucleotide product. S is the labeled RNA5 substrate. P is the final product. Shown is a representative of three experiments. C, a time course of Sfn-His RNase activity with 5 mM MnCl 2 at 37°C. Lane 1 is the no protein control. Kinetic Parameters of the Sfn Nuclease Activities-To determine the reason(s) for the different nuclease efficiencies of Sfn-His, we determined the Michaelis-Menten constant (K m ), maximal velocity (V max ), and the apparent rate of catalysis (k cat ) for comparable RNA and DNA substrates (Fig. 7). The K m of RNA5 (K m ϭ 1.56 M) is essentially identical to that of DNA5 (K m ϭ 1.51 M). In contrast, the V max and k cat values were ϳ4-fold higher for RNA5 (V max ϭ 0.015 M min Ϫ1 , k cat ϭ 1.5 min Ϫ1 ) than for DNA5 (V max ϭ 0.004 M min Ϫ1 , k cat ϭ 0.39 min Ϫ1 ). Our K m value for RNA5 is consistent with the micromolar range reported for E. coli Orn on p(A) 5 single-stranded RNA substrates (19). DISCUSSION The human SFN gene belongs to the YjeR/ORN subfamily of the 3Ј to 5Ј exonuclease superfamily previously described (1,5). Some of the members of this superfamily, most notably the Werner syndrome gene product and the polymyositis-scleroderma overlap syndrome 100-kDa autoantigen (PM-Scl 100), are associated with human disease (21,22). While a connection of SFN to a specific human disease is not obvious, the ubiquitous expression of its transcript may suggest a general and essential role for the encoded protein in mammalian cells. Notably, the human SFN gene maps to chromosome position 11q23.1-11q23.2, a region that undergoes translocation events in several leukemias (23,24), although none of these translocation breakpoints have been finely mapped to SFN.
The unique NH 2 terminus of hSfn␣ contains a consensus cleavage site pattern for mitochondrial processing proteases (17), whereas both Sfn␣ and Sfn possess a nuclear targeting signal in their COOH-terminal domains. This observation suggests that an alternative splicing event has evolved to give rise to a mitochondrial (Sfn␣) and a nuclear (Sfn) version of the Ynt20/Orn equivalents in human. However, at present, the genomic DNA sequence of the entire hSFN gene is not available, and thus the alternatively spliced products cannot be confirmed. Notably, in the mouse EST data base, there are also multiple transcripts, suggesting that alternatively spliced murine SFN transcripts exist as well. Future studies will need to address whether there is tissue-specific expression of the mRNA splice variants found in mammals. Interestingly, while S. cerevisiae maintains a homologue to hSFN␣ (called YNT20 or REX2), we were unable to find a homologue to SFN in the NCBI data base using the Blastp search program.
A knockout of the E. coli SFN homologue, the ORN gene, results in cellular lethality (7). An E. coli temperature-sensitive mutant is not only lethal at the nonpermissive temperature (i.e. where Orn is inactive), but accumulates small oligoribonucleotides, indicating that Orn maintains an essential activity to degrade RNA (8). Three possibilities were provided as to why ORN deletion mutants are inviable (8): 1) accumulation of oligoribonucleotides results in a depletion of cellular mononucleotides; 2) accumulated oligoribonucleotides inhibit certain enzymes and interfere with essential metabolic processes; or 3) Orn has an additional unknown function that is responsible for the growth cessation. Since Sfn is capable of degrading both RNA and DNA, we propose that the human protein operates to remove not only short RNAs (likely resulting from RNA degradation processes) (25,26), but also short single-stranded DNAs that might arise as products of DNA repair and recombination. Thus, by extension (8), Sfn would function globally to recycle nucleotides or to remove nucleic acids that may interfere with essential cellular processes. How q, DNA5 substrate. The data presented here is the average of five independent experiments for RNA5 and of four independent experiments for DNA5 in the presence of its preferred metal cation Mn 2ϩ . Error bars represent the standard deviation. Linear regression analysis showed a correlation coefficient of 0.997 for RNA5 and 0.986 for DNA5. the nuclease activities of the Orn/Ynt20 proteins would function in mitochondrial DNA escape, a process that involves translocation of mitochondrial DNA to the nucleus (9) and is possibly linked to cellular aging or senescence (27), is presently unclear. Furthermore, although unlikely, whether the NH 2terminal differences between these proteins affect substrate specificity will need to be determined.
Not surprisingly, Sfn-His, which shares ϳ50% identity in amino acid sequence to its bacterial counterpart, has very similar biochemical properties to Orn (5). In particular, both proteins degrade short RNAs in a 3Ј to 5Ј direction, prefer Mn 2ϩ as its metal cofactor, and have an enzymatic activity temperature optimum of ϳ50°C. However, gel filtration chromatography suggests that the human Sfn-His fusion protein is a tetramer, in contrast to E. coli Orn, which exists as a homodimer (6), suggesting that these proteins may maintain differing activities as well.
The nuclease activities of both Sfn and Orn (19) are inversely proportional to the length of the single-stranded substrate. Furthermore, previous kinetic data of E. coli Orn has shown that the K m values for short and long single-stranded RNAs are similar (19). We observed that Sfn has a similar K m for short singlestranded RNAs and DNAs, but that such short RNAs are degraded ϳ4-fold more efficiently than DNAs. Such experimental observations raise questions of how the substrate length or the nucleic acid chemistry influences Sfn/Orn enzymatic activity. To answer these questions will require determining which step(s) is influenced by nucleotide length or nucleic acid composition. High resolution structural data of Sfn, alone and in complex with RNA and DNA, would also shed light on the recognition and catalytic mechanisms of these proteins. Lastly, our studies emphasize that future experiments should pay particular attention to the poten-tial range of substrate diversity recognized by the other 3Ј to 5Ј exonuclease superfamily members.