Comparison of the Two Murine Deoxynucleotidyltransferase Terminal Isoforms A 20-AMINO ACID INSERTION IN THE HIGHLY CONSERVED CARBOXYL-TERMINAL REGION MODIFIES THE THERMOSENSITIVITY BUT NOT THE CATALYTIC ACTIVITY*

Terminal deoxynucleotidyltransferase (TdT) cata-lyzes the addition of nucleotides to 3 * -hydroxyl ends of DNA strands in a template-independent manner and has been shown to add N-regions to gene segment junctions during V(D)J recombination. TdT is highly conserved in all vertebrate species, with a second isoform, character-ized by a 20-amino acid insertion near the COOH-termi-nal end, described only in the mouse. The two murine isoforms differ in their subcellular localization, and the long isoform (TdTL) has previously been found to be unable to add N-regions. Using purified protein pro-duced in a high level expression system in Escherichia coli , we were able to carry out detailed catalytic com-parisons of these two TdT isoforms. We discovered that TdTL exhibits terminal transferase activity with kinetic parameters similar to those of the conserved TdT isoform (TdTS). We observed, however, that TdTL is inac-tivated at physiologic temperature but stable at lower temperatures. This thermal sensitivity of TdTL polymerase activity is not correlated with a significant change in the circular dichroism spectrum of the protein. Thus, the 20-amino acid insertion in TdTL does not affect the catalytic activity but modifies the thermosensitivity. Terminal deoxynucleotidyltransferase (TdT, 1 EC 2.7.7.31) with microsep 10K centrifugal concentrator (Pall Filtron, Northborough, MA) and stored at 2 80 °C after addition of 50% glycerol. Purified proteins were analyzed by SDS-polyacrylamide gel electrophoresis, using an 8% acrylamide gel. Gels were stained with Coomas- sie Blue or probed with polyclonal rabbit anti-bovine TdT serum (a generous gift from F. J. Bollum). No degradation products were detected on the Western blot. Densitometric analysis with a MasterScan I In-terpretive densitometer (Scananalytics) of the Coomassie-stained gel Thermostability— To measure the thermostability of TdT isoforms, proteins at 3 m M in the storage buffer (25 m M Hepes, pH 7.0, 25 m M MgOAc, 25 m M (NH 4 ) 2 SO 4 , and 100 m M NaCl, 50% glycerol) were preincubated at various temperatures ranging from 20 to 60 °C for 10 min or at 35 °C for various times and chilled on ice. The residual activity was measured at 35 °C in the standard kinetic assay buffer with 100 m M (dA) 10 , 1 m M [ a - 32 P]dATP at 70 cpm/pmol, and 200 n M TdTS or TdTL. Dichroism (CD) Spectrum— Secondary structures of recom- binant proteins measured using circular dichroism on a Jobin-Yvon CD6 spectrometer France).

Terminal deoxynucleotidyltransferase (TdT) catalyzes the addition of nucleotides to 3-hydroxyl ends of DNA strands in a template-independent manner and has been shown to add N-regions to gene segment junctions during V(D)J recombination. TdT is highly conserved in all vertebrate species, with a second isoform, characterized by a 20-amino acid insertion near the COOH-terminal end, described only in the mouse. The two murine isoforms differ in their subcellular localization, and the long isoform (TdTL) has previously been found to be unable to add N-regions. Using purified protein produced in a high level expression system in Escherichia coli, we were able to carry out detailed catalytic comparisons of these two TdT isoforms. We discovered that TdTL exhibits terminal transferase activity with kinetic parameters similar to those of the conserved TdT isoform (TdTS). We observed, however, that TdTL is inactivated at physiologic temperature but stable at lower temperatures. This thermal sensitivity of TdTL polymerase activity is not correlated with a significant change in the circular dichroism spectrum of the protein. Thus, the 20-amino acid insertion in TdTL does not affect the catalytic activity but modifies the thermosensitivity.
Terminal deoxynucleotidyltransferase (TdT, 1 EC 2.7.7.31) is the only known enzyme to catalyze the nontemplated addition of nucleotides to 3Ј-hydroxyl ends of DNA strands (1). This polymerase function has an important role in the diversification of the immunoglobulin and T cell receptor repertoires in vertebrates, enabling TdT to add random nucleotides, called N-regions, to the junctions of gene segments during V(D)J recombination (2,3). TdT is highly conserved across the vertebrate phylum, from bony fish to man (4 -8), yet the mouse is the only species in which two TdT isoforms have been discovered. Analysis of a mouse genetic clone containing the 3Ј-region of the TdT gene revealed the presence of an additional exon (exon X-bis). We identified two alternatively spliced TdT mRNAs in mouse bone marrow and thymus. The longer TdT mRNA has an insertion of 60 nucleotides encoded by the addi-tional exon near the carboxyl-terminal end of the gene and codes for a protein of 529 amino acids (TdTL). The shorter mRNA codes for a protein of 509 amino acids (TdTS) (9). Catalytic and functional differences between the short and the long murine TdT isoforms have been described but not explained. We have previously shown, in transfection experiments using either fibroblasts or COS cells, that TdTS and TdTL differ in their subcellular localization and their ability to add N-regions to V(D)J junctions during recombination of an episomic substrate. In this system, TdTS is strictly nuclear and adds Nnucleotides at V(D)J junctions, whereas TdTL is essentially cytoplasmic and does not add N-nucleotides. Comparison of terminal transferase activity in extracts of TdTS and TdTL transfected cells led to the hypothesis that the inability of TdTL to add N-regions could be due to a defective polymerase activity (10).
The lack of an adequate, unproteolysed protein source has been a constraint on research on TdT. We have recently developed a method of production and purification of TdT in Escherichia coli (11), allowing us to carry out extensive kinetics and structural analyses of the protein. Here we report for the first time that TdTL displays the same terminal transferase activity as TdTS but that the 20-amino acid insertion conveys heat sensitivity.

EXPERIMENTAL PROCEDURES
Purification of the Two Murine TdT Isoforms-The protocol used for expression and purification of recombinant TdTS has been described (11). It was modified for the expression and purification of TdTL, since the protein was recovered in inclusion bodies when the cells were grown at 15°C and was poorly retained on the nickel affinity column at high stringency. Cells transformed with the TdTL expression vector were grown at 25°C until OD 600 reached 0.8. After addition of 1 mM IPTG, cultures were incubated at 10°C during 24 -36 h. Cells were harvested by centrifugation at 2600 g for 15 min, resuspended in 0.05 volume of buffer A L (10 mM Tris-HCl, pH 8.0, 0.25 M NaCl), and lysed in a French press at 16,000 p.s.i. The lysate was cleared by centrifugation for 30 min at 20,000 ϫ g. The supernatant was then sonicated, filtered through a 0.45-m filter, and loaded at a rate of 1 ml/min on a 5-ml Hitrap affinity column (Amersham Pharmacia Biotech, Uppsala, Sweden) charged with Ni 2ϩ ions and equilibrated with buffer A L . The column was washed with 3 volumes of buffer A L , 100 mM imidazole, and TdTL was eluted with a gradient of 100 -300 mM imidazole on a FPLC system (Amersham Pharmacia Biotech) at a rate of 2 ml/min. The purest fractions were pooled and dialyzed against buffer B (50 mM Hepes, pH 7.0, 50 mM MgOAc, 50 mM (NH4) 2 SO 4 , 200 mM NaCl). Dialyzed fractions were concentrated with microsep 10K centrifugal concentrator (Pall Filtron, Northborough, MA) and stored at Ϫ80°C after addition of 50% glycerol. Purified proteins were analyzed by SDS-polyacrylamide gel electrophoresis, using an 8% acrylamide gel. Gels were stained with Coomassie Blue or probed with polyclonal rabbit anti-bovine TdT serum (a generous gift from F. J. Bollum). No degradation products were detected on the Western blot. Densitometric analysis with a MasterScan I Interpretive densitometer (Scananalytics) of the Coomassie-stained gel * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
C. P. dedicates her contribution to this work to the memory of Dr. Philippe Jérôme Lecomte deceased in Paris on January 18, 2000.
‡ Supported by a fellowship from the Ministère de l'Education Nationale de la Recherche et de la Technologie.
led to an estimation of the purity of the two proteins superior to 80%. Protein concentration was calculated using a theoretical extinction coefficient of 54,870 M Ϫ1 cm Ϫ1 (12).
TdT Enzymatic Assays and Determination of Kinetic Parameters-Two substrates are involved in the terminal transferase nucleotidyl transfer reaction: a single stranded DNA (initiator) and a deoxynucleoside triphosphate. Michaelis-Menten kinetic parameters (K m dATP , K m (dA)10 , and k cat ) were obtained by titrating each substrate in the presence of a saturating concentration of a second substrate, as described previously (11). (dA) 10 and (dA) 50 oligonucleotide primers were purchased from Genset (Paris, France). Ultrapure deoxyribonucleotide solutions were purchased from Amersham Pharmacia Biotech (Uppsala, Sweden).
Terminal transferase activity was detected by incorporation of dATP into single-stranded DNA using the following standard assay: recombinant proteins were incubated at 35°C in 200 mM potassium cacodylate, 25 mM Tris-HCl, pH 6.6, 0.25 mg/ml bovine serum albumin, 4 mM MgCl 2 , 4 M ZnSO 4 , 1 mM [␣-32 P]dATP (70 cpm/pmol), and 100 M oligonucleotide primer unless otherwise specified. Reactions were stopped at different times with 15 mM EDTA. Aliquots were spotted onto Whatman DE81 discs. Unpolymerized nucleotides were washed away by immersing the discs three times for 5 min in NH 4 COOH/ Na 4 P 2 O 7 ⅐10H 2 O (300 mM/10 mM) and then in H 2 O and in EtOH. Dried discs were counted in a scintillation fluid (Econofluor from Packard, Göningen, The Netherlands). Kinetic curves show the percentage of total dATP present in the reaction incorporated over time.
Owing to the distributive mechanism of TdT polymerization, analysis of chain length distribution was done at low oligonucleotide primer concentration. 5 M 5Ј-32 P-labeled (dA) 10 primer was used as the radioactive substrate. Aliquots were withdrawn at different times, supplemented with a formamide dye mix (10 mM NaOH, 95% formamide, 0.05% bromphenol blue, 0.05% xylene cyanole), and electrophoresed, after heat denaturation, on a 16% acrylamide denaturing gel. TdT products were visualized after exposure of the wet gel under a Kodak film (Biomax MR) at Ϫ70°C. Thermostability-To measure the thermostability of TdT isoforms, proteins at 3 M in the storage buffer (25 mM Hepes, pH 7.0, 25 mM MgOAc, 25 mM (NH 4 ) 2 SO 4 , and 100 mM NaCl, 50% glycerol) were preincubated at various temperatures ranging from 20 to 60°C for 10 min or at 35°C for various times and chilled on ice. The residual activity was measured at 35°C in the standard kinetic assay buffer with 100 M (dA) 10  Circular Dichroism (CD) Spectrum-Secondary structures of recombinant proteins were measured using circular dichroism on a Jobin-Yvon CD6 spectrometer (Longjumeau, France). Measurements were done at wavelengths from 195 to 260 nm, at two temperatures. First measurement was done at 25°C. The temperature was then raised progressively to 35°C and proteins incubated further for 20 min, before the second measurement was done. For each temperature three independent measurements were performed. Concentrations of protein solutions were 1 mg/ml in 2ϫ storage buffer without glycerol.

The 20-Amino Acid Insertion in the Long Isoform of Murine TdT Is Localized in the Highly Conserved Carboxyl-terminal
Region-A multiple alignment of all known TdT sequences was drawn using the Pileup program (Genetics Computer Group version 9.0), and a consensus sequence was deduced (Fig. 1). 153 amino acids are shared by all the sequences analyzed, corresponding to 30% identity, and 71 amino acids are functionally equivalent, leading to an overall similarity of 44%. TdT protein can be divided into three regions: the NH 2 -terminal domain, which is dispensable for TdT activity (13), 2 and the so-called ␤ and ␣ peptides (14). Positions are given for the  10 . The percentage of total dATP incorporated was measured at 0, 5, 7.5, 10, 15, 30, and 60 min (A). Each experiment was repeated several times, and results were averaged. Error bars represent the S.D. Reaction conditions for analysis of chain length distribution were as described above, except that (dA) 10 (15) and later in many other proteins, including some involved in DNA damage repair and recombination (16,17). It was recently demonstrated that, in vitro, this domain in TdT interacts with the Ku70/86 heterodimer that binds DNA and is involved in double-strand break repair and V(D)J recombination (18). Mutagenesis experiments of human TdT showed that two amino acids analogous to Asp 343 and Asp 345 in murine TdT are essential for catalysis (19). We have shown that Asp 434 is also required for catalysis. 2 The catalytic center of the protein thus contains amino acids from both ␤ peptide (aa 132-421, 41% similarity) and ␣ peptide (aa 422-510, 67% similarity). The 20-amino acid insertion is located at position 482 in the ␣ peptide, which is the most conserved region in all the analyzed sequences (46% identity). TdTL Is a True Terminal Deoxynucleotidyltransferase-Catalytic activities of purified TdTS and TdTL were measured using a standard assay (described under "Experimental Procedures") with dATP and (dA) 10 as substrates. Initial velocities (V i ) were determined using linear regression on appropriate time courses (every 30 s for 3.5 min) and were normalized to the concentration of TdT. For both TdTS and TdTL, we observed a hyperbolic dependence of initial rates on the concentration of the varying substrates (data not shown). Michaelis-Menten kinetic parameters were obtained using the Kaleidagraph program (Synergy Software, FCF Inc., Reading PA) and are presented in Table I. There is only a 2-fold reduction in TdTL specific activity as compared with TdTS. TdTL has the same apparent affinity for dATP, a higher apparent affinity for (dA) 10 (over 2-fold), and a somewhat lower polymerization rate (about 2-3-fold difference in k cat for dATP and (dA) 10 ). Thus, TdTL has true terminal transferase activity with kinetic parameters similar to those of TdTS.
Thermosensitivity of TdTL Catalytic Activity-When we compared the behavior of the two proteins over a long time course in our standard assay we noticed that TdTL quickly reached a plateau. Less than 10% (about 5.0 nmol) of the initial dATP could be incorporated, whereas TdTS was capable of consuming, in an hour, 80% (40 nmol) of the total amount of dATP present in the reaction (Fig. 2A). The chain length distributions at low (dA) 10 concentration (5 M) also reflected the differences between the two isoforms. TdTS synthesized much longer products than TdTL. The size of the TdTS products grew over time, whereas the size distribution of the TdTL products remained uniform (Fig. 2B). The same differences in the TdTS and TdTL chain length distribution patterns were observed in the presence of the four deoxyribonucleotides (Fig. 2C). One hypothesis for the distinctive behavior of TdTL could be the inability of the protein to extend longer DNA chains. We thus measured the catalytic activity of TdTL on a (dA) 50 oligonucleotide primer. (dA) 50 was elongated efficiently, and a plateau of incorporation was again observed after 20 min of incubation (data not shown). Alternatively, the plateau could reflect a lack of stability and, thus, an inactivation of the protein in the reaction conditions. The residual activity of the two TdT isoforms was measured at 35°C after preincubation for 10 min at varying temperatures from 20 to 60°C. The reactions were started upon the addition of the enzyme and initial velocities were measured. The temperature inactivation curves are shown in Fig. 3A. The inflection point occurred at 37°C for TdTL and at 45°C for TdTS. This result suggested that TdTL was indeed unstable around the temperature (35°C) commonly used. This was further confirmed by measuring the enzymes residual activities after preincubation at 35°over varying periods of time from 15 to 60 min. As can be seen in Fig. 3B, the residual activity was 60% for TdTS but only 20% for TdTL after 60 min of preincubation at 35°C. We verified, using SDSpolyacrylamide gel electrophoresis analysis, that there was no apparent protein degradation even after an hour of incubation at 35°C. Furthermore, addition to the reaction buffer of bacterial protease inhibitors did not prevent the inactivation of TdTL (data not shown). These observations led us to assess the retention of secondary structures of the two proteins by comparing their CD spectra. The analyses were first done at 25°C, where both isoforms are stable. At this temperature, TdTS and TdTL CD spectra were similar with a mean molar ellipticity at 210 nm around Ϫ15,000 degree⅐cm 2 dmol Ϫ1 . This low value reflects the abundance of ␣-helices and ␤-sheets. The proteins were then slowly brought to 35°C and further incubated for 20 min at this temperature. Both isoforms retained a low molar ellipticity (Fig. 4). The thermosensitivity of TdTL catalytic activity is thus not correlated with a loss of secondary structure.
To test the possibility that the two isoforms could interact in vitro and exert an influence on each other, we compared the activity of a mixture of the two enzymes to the activity of each enzyme tested separately. As can be seen in Fig. 5, with a standard kinetic assay for incorporation of ␣-32 P-labeled dATP on a (dA) 10 primer, the activities of TdTS and TdTL were additive. The same result was obtained when the experiment was carried out with different ratios of the two enzymes (data not shown). DISCUSSION TdT has attracted great interest because of its unique catalytic capability, role in V(D)J recombination, unexplained evo- lutionary lineage, and numerous applications in molecular biology. We have previously demonstrated the presence of two TdT isoforms in the mouse, both expressed in the thymus and bone marrow (9). Our initial attempts to compare the activities of the two murine TdT isoforms were carried out with extracts of TdTS or TdTL transfected cells, where the proteins were poorly expressed and labile (10). To overcome these problems, we recently developed a method of production of murine TdT in E. coli by which large amounts of purified, unproteolysed TdTS are easily obtained (11). Since the presence of the 20-amino acid insertion modifies the biochemical properties of the protein, certain minor modifications in the expression and purification steps were required to optimize the recovery of TdTL.
Both recombinant short and long TdT isoforms exhibit terminal transferase activity, with similar kinetic parameters. They are both capable of polymerizing long chains of singlestranded DNA, but a plateau of incorporation of deoxynucleotides is always observed with TdTL after a short period of incubation at standard temperature. Our analysis of heat-induced denaturation of the two TdT isoforms argues for a correlation between the plateau and the thermal instability of TdTL. The TdT active site, as inferred from sequence alignments, cross-linking (20), and mutagenesis experiments (19), 2 contains a constellation of amino acids from both ␣ and ␤ peptides. Interestingly, the 20-amino acid insertion in the long murine TdT isoform is located near the COOH-terminal end of the protein, in the region that is the most conserved in all the TdT sequences analyzed. It is conceivable that the 20-amino acid insertion destabilizes the tertiary structure of the protein and affects the geometry of the active site at higher temperatures, precluding the interactions with substrates, nucleotides, and/or primer. These changes must be subtle, since the thermosensitivity of TdTL polymerase activity is not correlated with a loss of secondary structures, as inferred from analyses of the CD spectrum. The possibility of an interaction between TdTS and TdTL in vitro was considered. Kinetic analysis of a mixture of the two TdT isoforms indicates that TdTL does not inhibit the activity of TdTS. Furthermore, if the rapid loss of activity were due to a contaminant in the TdTL preparation, such as a protease, we would have expected to observe an effect of the mixing on TdTS stability as well. Such was not the case, thus arguing instead for an intrinsic instability of the long isoform at physiologic temperature.
Temperature sensitivity acquired by spontaneous mutations has been described for several proteins. A defect in the cellular trafficking of thermosensitive mutants of human tyrosinase and cystic fibrosis transmembrane conductance regulator has been correlated with a loss of activity in vivo (21,22). It remains to be determined whether cultivating transfected cells at lower temperatures can modify the subcellular localization of TdTL and rescue its putative cellular function. The possibility that TdTL in vivo plays a role distinct from N-nucleotide addition will need to be explored.
TdT has only been found in vertebrates and no homolog to murine TdTL has been reported. Genomic data are yet to be collected to allow the search for an exon homologous to the murine TdT exon X-bis. Murine TdTL could represent an ancestral, perhaps vestigial, form of the enzyme (23) or may result from an evolutionary happenstance, such as the late capture of an additional exon. The evolutionary origin of TdT is a matter of interest and debate. Based upon some sequence similarity with polymerase ␤, TdT has been classified in the family X polymerases (24), a subclass of an ancient nucleotidyltransferase superfamily whose members share a common signature in the active site and catalyze the same chemical reaction but have diverse biological roles (25,26). Confirmation of this filiation awaits the elucidation of TdT tertiary structure.
FIG. 5. Activity of TdTS and TdTL mixture. Terminal transferase activity was measured by incorporation of ␣-32 P-labeled dATP on a (dA) 10 primer in our standard enzymatic assay. Reactions were carried out with 290 nM TdTS (black circles), 290 nM TdTL (black squares), or 290 nM amounts of each enzyme (black triangles). The percentage of total dATP incorporated was measured between 0 and 40 min. The curve corresponding to the addition of the dATP incorporated by TdTS and TdTL tested in separate assays is also displayed (open triangles). Each experiment was repeated several times, and results were averaged. Error bars represent the S.D.