THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1985 by The American Society of Biological Chemists, Inc.

Vol. 260, No. 19, Issue of September 5, pp. 10495-10502,1985 Printed in U. S. A.

Expression of Human Terminal Deoxynucleotidyl Transferase in Escherichia coli*

(Received for publication, April 22, 1985)

Ronald C. Peterson, Ling C. Cheung, Robert J. MattalianoS, Stephen T. White, L. M. S. Chang, and F. J. Bollum From the Department of Biochemistry, Uniformed Services University of the Health Sciences, Bethesdq Maryland 20814

A cloned DNA fragment related to pT17 containing a partial cDNA sequence of human terminal deoxynu- cleotidyl transferase was used as a probe to screen for the full length cDNA sequence of the enzyme in a Xgtll library constructed from human lymphoblastoid KM-3 cDNA. A recombinant containing a 2068-base pair insert was isolated and recloned into the EcoRI site of the sequencing plasmic pUC-8 as two subclones, pT711 and pT106. DNA sequencing and hybridization studies showed that pT711 contains the pT17 sequence and an additional 172 upstream nucleotides. pT711 repre- sents the coding sequence for the carboxyl half of the terminal transferase protein. pT106, containing a 965-base pair insert, hybridizes to the same mRNA as pT711 on Northern blots and contains an open reading frame that is in phase with the reading frame of the insert in pT711. Amino acid sequencing of the 58-kDa peptide of the calf thymus terminal transferase failed, indicating that the N terminus is blocked. N-Terminal sequencing of a 56-kDa form of the protein produced 24 amino acids corresponding to the translated human cDNA coding sequence starting at residue 398 of the insert in pT106 with 83% homology between bovine and human sequence. The initiation codon is assigned to an ATG sequence at nucleotide 329 of the insert in pT106. Comparison of the translated human terminal transferase sequence with peptides from the calf thy- mus enzyme showed that the homology between the human and bovine enzyme is better than 90% among 263 amino acids determined.

The coding sequences in pT106 and pT711 were recloned into an expression plasmid pUC-19 down- stream from the lac promoter and in phase with the coding sequence of the lac Z gene. Lysates of bacteria carrying the reconstructed coding sequence of human terminal transferase contain a fused protein of 60 kDa that reacts with rabbit antibody to terminal transfer- ase on immunoblots and exhibits enzyme activity. Iso- lation of this fused protein from bacterial lysates with mouse monoclonal antibody to human terminal trans- ferase produces the expected protein of 60 kDa.

Terminal deoxynucleotidyl transferase is a 58-kDa protein

* This investigation was supported by United States Public Health Service Grants GM31393 and CA23262. 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.

$ Present address: Biogen Research Corporation, 14 Cambridge Center, Cambridge, MA 02142. The N-terminal analysis of the 58- and 56-kDa calf thymus terminal transferase peptides was carried out with the facilities at Biogen.

expressed in pre-B and pre-T lymphocytes during early dif- ferentiation (1). Cytological studies demonstrated a nuclear localization in primitive bone marrow lymphocytes (2) and nuclear and cytoplasmic localization in cortical thymocytes (3). Biochemical analysis showed that the enzyme can be phosphorylated with protein kinases suggesting hormonal regulation of the enzyme activity and/or turnover (4,5). The precise biological function of the terminal transferase is not known although recent data suggest that the enzyme is in- volved in the generation of the non-germline elements (N regions) in immunoglobulins (6). Because of the restricted localization of the enzyme in the lymphoid differentiation scheme, a direct approach to determination of the biological function of the enzyme has not been feasible. We are exploring the possibility of using the cloned terminal transferase DNA to probe the function of the enzyme.

The availability of reliable polyclonal (7) and monoclonal antibodies (8) to terminal transferase and development of the expression vector Xgtll (9) allowed us and others to generate recombinant phages that carry and express cDNA sequence of human (10) and murine terminal transferase (11). Using the cloned human terminal transferase cDNA as a probe, we have now isolated near full length terminal transferase cDNA containing the complete coding sequence for the enzyme. Insertion of the coding sequence of the enzyme into a pUC- 19 expression plasmid (12) allows the expression of human terminal transferase in Escherichia coli.

EXPERIMENTAL PROCEDURES

Materials

Poly(A)+ mRNA from human lymphoblast cell line KM-3 (lo), Xgtll vector (9), pT711 (pUC-9 containing the 1100-bp’ human TdT cDNA sequence) (lo), rabbit antibody to calf terminal transferase (7), monoclonal antibody to human terminal transferase covalently linked to Protein A Sepharose (13), d(pA), (14), and high molecular weight form of terminal transferase from KM-3 cells and calf thymus glands (131, goat antibody (IgG) to rabbit IgG (15) were as previously described. Endonucleases (EcoRI and BamHI), X phage packaging system, E. coli DNA polymerase I, large fragment of E. colipolymerase I, EcoRI methylase, exonuclease 111, and the 15-base pair M13 phage- sequencing primer were purchased from Bethesda Research Labora- tories. Rabbit antibody to horseradish peroxidase-peroxidase complex was purchased from Accurate Chemicals, avian myeloblastosis virus reverse transcriptase from Life Sciences, BAL 31 nuclease from New England Biolabs, pancreatic DNase I and RNase from Worthington, d(pC)lp.lB from P-L Biochemicals, proteinase K and calf intestinal alkaline phosphatase from Boehringer Mannheim, bacteriophage T4 ligase, and EcoRI linker from Collaborative Research, standard pro-

’ Abbreviations used for nucleotides and polynucleotides are those of IUPAC-IUB Commission on Biochemical Nomenclature (CBM) J. Biol. Chem. 245, 5171-5176. Other abbreviations used are: bp, base pair; EDTA, ethylenediaminetetraacetic acid; kDa, kilodalton.

10495

10496 Terminal Transferase in E. coli tein markers from Sigma, and [cY-~*P]~ATP from New England Nuclear. Other chemicals were reagent grade.

Methods

Construction of the u t 1 1 Library and Screening for Terminal Transferase Positive Recombinant Clones-A new Xgtll library was constructed with double-stranded cDNA from KM-3 cells essentially as previously described (10) except that an oligo-dG tail was added to the single-stranded cDNA, and second strand synthesis was initi- ated with d(pC)12-18 and extended using the large fragment of E. coli polymerase I. The double-stranded cDNA was then methylated using EcoRI methylase and ligated to EcoRI linkers. After packaging, recombinant phages were plated on E. coli Y1088 and screened by hybridization with the nick-translated insert from pT711. The ter- minal transferase positive recombinants were plaque purified and amplified. Amplified recombinant phages were treated with 100 pg/ ml pancreatic RNase for 30 min at room temperature followed by incubation with 100 pg/ml of proteinase K in 0.5% sodium dodecyl sulfate at 70 "C for 1 h, extracted with phenol and chloroform, precipitated with ethanol, and finally redissolved in 1 mM EDTA in 10 mM Tris-HC1 at pH 7.6. Partial digestion with EcoRI nuclease was carried out on DNA from each recombinant isolated. DNA digests were separated on 1% agarose gel, transferred onto nitrocellulose filters, and hybridized with nick-translated insert of pT711. A recom- binant showing the presence of a 2100-bp insert was selected for further investigation.

Recloning of Terminal Transferase cDNA Fragments in pUC-8 and DNA Sequencing-The purified recombinant phage DNA was par- tially digested with EcoRI nuclease to produce three nonphage DNA fragments with 2100,1100, and 1000 bp. Each fragment was recloned in EcoRf-cleaved phosphatase-treated pUC-8 to produce pT711 (1100-bp insert) and pT106 (about 1000-bp insert). Recloning the 2100-bp fragment in pUC-8 resulted in recombinants containing inserts of less than 1200 bp. DNA sequence analysis for the inserts in pT711 and pT106 was carried out by the dideoxy (16) and Maxam- Gilbert (17) procedures on BAL 31 derivatives of pT711 and pT106 essentially as previously described (10).

Blot Hybridization Analysis-Poly(A)+ mRNA from KM-3 cells was denatured and separated on a 1.5% agarose gel, transferred onto a nitrocellulose sheet, and hybridized with nick-translated pT711 and pT106 essentially as previously described (10).

Sequencing of Calf Thymus Peptides-Sequencing of tryptic pep- tides of succinylated calf thymus terminal transferase was carried out as described previously (10). The 58- and 56-kDa peptides were isolated by separation of the peptides present in immunoaffinity- purified preparations by electrophoresis in 12.5% polyacrylamide gel in the presence of sodium dodecyl sulfate (13) followed by electroe- lution of the peptides of interest. 1.3 nmol of 56-kDa peptide and 1.7 nmol of 58-kDa peptide were subjected to automated Edman degra- dation in an Applied Biosystems gas phase sequenator.

Detection of Terminal Transferase Peptides on Polyacrylamide Gels-Protein solutions were separated on 12.5% polyacrylamide gel in the presence of sodium dodecyl sulfate as described by Laemmli and Favre (18) and electrophoretically transferred onto nitrocellulose sheets. After transfer the nitrocellulose sheet was blocked with 20% fetal calf serum in TBS (0.1 M NaCl in 50 mM Tris-HC1 at pH 8.01, reacted 1 h with rabbit antibody to terminal transferase, washed with TBS, reacted 1 h with goat anti-rabbit IgG, washed with TBS, reacted 1 h with rabbit anti-horseradish peroxidase-peroxidase complex, washed with TBS, and developed for 30 min at 0.5 mg/ml 4-chloro- 1-naphthol, 0.001 % H2O2 in TBS.

ity was assayed with 5 p~ d(pA)&, 1 mM [8-3H]dGTP at 62 cpm/ Terminal Transferase Enzyme Assay-Terminal transferase activ-

prnol, 8 r n M MgC12, 0.2 M potassium cacodylate buffer at pH 7.2, and 1 mM 2-mercaptoethanol as previously described (19). Incubation was at 35 "C. Products of the reaction were scored as acid-insoluble radioactivity on glass fiber filters. One unit of enzyme is defined as 1 nmol of dGMP polymerized in 1 h.

Terminal Transferase Activity in Bacterial Extracts-Crude ex- tracts of midlog bacterial cells were prepared by 1-min sonication of suspensions of 1.2 X 10' cells in 1 ml of 0.1 M potassium cacodylate buffer at pH 7.2. The lysates were clarified by centrifugation for 20 min at 13,000 rpm in a Sorvall SS-34 rotor. Terminal transferase activity was measured as described above, and protein concentration was estimated using biuret reagent (20).

Purification of terminal transferase was carried out on a 0.6-ml mouse monoclonal antibody column containing 1.2 mg of mouse

monoclonal antibody to human terminaI transferase covalently cou- pled to 0.6 ml of Protein A Sepharose (21). Bacterial cells (4 g), grown in the presence of 1 mM isopropyl 0-D-thiogalactopyranoside, were washed and resuspended in 25% sucrose in 50 mM Tris-HC1 buffer at pH 8.0. After digestion with 2.5 mg/ml egg white lysozyme for 1 h at 4 "C and treatment with 50 mM EDTA for 5 min at 4 "C, the cells were lysed by dilution with an equal volume of a buffer containing 0.5% Nonidet P-40,O.S M NaCl, 2 mM phenymethylsulfonyl fluoride, 1% dimethyl sulfoxide in 50 mM Tris-HC1 at pH 8.0. The lysate was clarified by centrifugation for 1 h at 42,000 rpm in a Ti-60 rotor in a Beckman ultracentrifuge. The clarified extract (150 ml) was loaded directly onto the 0.6-ml mouse monoclonal antibody column equili- brated with 0.25% Nonidet P-40,0.5 M NaCl, 1 mM phenymethylsul- fonyl fluoride, 1% dimethyl sulfoxide in 50 mM Tris-HC1 at pH 8.0. The column was then washed with 25 ml of column equilibration buffer, 5 ml of 0.5 M NaCl in 50 mM Tris-HC1 at pH 8.0, 2 ml of 5 mg/ml bovine serum albumin in 0.5 M NaCl in 50 mM Tris-HC1 at pH 8.0, 50 ml of 0.5 M NaCl in 50 mM Tris-HCI at pH 8.0, and protein was eluted with 3.2 M MgCL in 50 mM Tris-HC1 at pH 8.0. The eluted protein was dialyzed for a total of 24 h with 4 changes of 0.5 M NaC1, 50 mM Tris-HC1 at pH 8.0 in 10% glycerol, 16 h against 50 mM potassium phosphate at pH 7.4, 1 mM 2-mercaptoethanol in 50% glycerol, and stored at -20 "C.

Reconstruction of the Coding Sequence of Human Terminal Trans- ferase in an Expression Vector-Expression vector pUC-19 was di- gested with EcoRI nuclease and dephosphorylated with calf intestinal alkaline phosphatase. The cleaved vector was deproteinized and re- digested with BamHI. The insert from pT711 was obtained by EcoRI digestion. A 634-bp fragment was obtained from pT106 by cleaving with EcoRI nuclease, dephosphorylation with calf intestine alkaline phosphatase, and digestion with BamHI nuclease. The 634-bp frag- ment from pT106 and the 1103-bp fragment from pT711 were sepa- rated from the vector fragment by electrophoresis on a l% agarose gel and extracted from the gel.

Reconstruction of the coding sequence of terminal transferase in pUC-19 was accomplished by ligation of the cleaved vector with the 634- and 1103-bp fragments using bacteriophage T4 ligase. The ligated DNA was used to transform E. coli HB101, and the resultant transformants were analyzed for size of DNA insert and production of material reacting with rabbit antibody to terminal transferase (22). Three nonidentical recombinants in pUC-19 were obtained; pT201 contains a 634-bp insert which does not produce protein reacting with anti-terminal transferase; pT223 contains a 1737-bp insert that does produce immunoreactive material; and pT226 contains a 1737-bp insert that does not produce immunoreactive material probably be- cause the pT711 segment is in the wrong orientation. Restriction analysis proved that pT223 contains the terminal transferase se- quences from pT106 and pT711 in proper orientation to the lac Z promoter and sequence.

RESULTS

Screening for Full Length Terminal Transferase cDNA Clones-We previously reported (10) the cloning of a fragment of human terminal transferase using the expression vector Xgtll and rabbit antibody to terminal transferase to detect fused P-galactosidase and transferase protein (9). The trans- ferase cDNA sequence (931 bp), coding for the carboxyl end of the protein, was subsequently cloned in pBR 322 and pUC- 8 and was designated pT17. Using the insert in pT17 as a nick-translated probe to rescreen the original Xgtll library, we were able to isolate several recombinant phages containing about 1100 bp. Recloning the 1100-bp insert in pUC-8 gen- erated pT711. Blot hybridization studies on KM-3 poly(A)+ mRNA and KM-3 double-stranded cDNA showed that the mRNA for human terminal transferase as well as full length double-stranded cDNA is about 2100 bp (10). The DNA sequence of pT711 showed that the EcoRI sites in this frag- ment do not arise from the addition of EcoRI linker but are present as EcoRI sites in the sequence of transferase cDNA. Since the original KM-3 cDNA library in Xgtll was con- structed without protection of EcoRI sites on the double- stranded cDNA, the presence of EcoRI sites in the coding sequence of terminal transferase cDNA resulted in recombi-

Terminal Transferase in E. coli 10497

nants containing parts of the complete terminal transferase cDNA sequence.

In order to obtain recombinant clones containing longer cDNA sequences, a new Xgtll library was constructed with cDNA from KM-3 cells in which the double-stranded cDNA was methylated with EcoRI methylase to protect internal EcoRI sites. Using the nick-translated insert of pT711 as probe to screen this second library, we were able to find 35 recombinants containing terminal transferase sequence. Par- tial EcoRI digests of the recombinant phage DNAs produced only one recombinant that contained the desired insert of about 2100 bp. The partial digests contained three DNA fragments derived from the insert, 2100,1100, and about 1000 bp. Both the 2100- and 1100-bp fragments hybridized to nick- translated pT711 insert. Attempts to reclone the 2100-bp fragment in pUC-8 and pBR 322 were unsuccessful because the resulting recombinants always contained fragments of about 1000 or 1100 bp. The reason for our failure to transfer the entire 2100-bp fragment into pUC-8 or pBR 322 is not known. It is possible that there are sequences present on the 2100-bp fragment that promote recombination. Recloning of the 1000- and 1100-bp fragments from the recombinant phage into pUC-8 resulted in pT106 and pT711. Preliminary DNA sequence analysis showed that the newly generated pT711 sequence is identical to that isolated from the original KM-3 cDNA library. A summary of the relationships between var- ious recombinants of terminal transferase cDNA is presented in Fig. 1.

Because the new library was screened with a nick-translated insert of pT711, the sequence present in pT106 could not be detected. To demonstrate that pT106 represents terminal transferase sequence, nick-translated pT711 and pT106 were used to probe nitrocellulose blots of KM-3 poly(A)+ mRNA, and the results are presented in Fig. 2. Inserts from both pT106 and pT711 hybridize to an mRNA of about 2100 nucleotides.

Complete DNA Sequences of pT711 and pT106 and Com- parison of Translated Human Terminal Transferase Sequence with Determined Amino Acid Sequences of Calf Thymus Ter- minal Transferase-DNA sequences for the inserts of pT711 and pT106 were determined by dideoxy (16) and Maxam- Gilbert (17) procedures on BAL 31 derivatives. The combined insert size was found to be 2068 bp. We have shown previously by DNA sequencing and amino acid sequence analyses that pT17 contained an open reading frame spanning 238 codons (10) and represents the cDNA coding for the carboxyl half of human terminal transferase. pT711 contains the entire DNA sequence of pT17 and an additional 172 upstream nucleotides. Translation of the DNA sequence of pT711 showed 57 addi- tional upstream codons with an open reading frame in phase with pT17. The DNA sequence of the pT106 insert is 965 bp long with an ATG codon starting at nucleotide 329 and an open reading frame in phase with pT711 spanning 212 codons.

I +pTl06 pT7111 I

i pT2231 pT171

I I 1 I I 00 400 800 1 2 0 0 1 6 0 0 2 0 0 0 b . p .

FIG. 1. Schematic diagram of the cDNA sequence coding for human terminal transferase. Abbreviations used for restriction enzymes are: A, ApaI; B, BarnHI; Bg, BglI; E, EcoRI; P, PuuII; and s, SstI.

A B

-3.25 kb

”1.11 kb

FIG. 2. Blot hybridization of KM-3 RNA with cloned ter- minal transferase cDNA fragments. Poly(A)+ RNA from KM-3 cells was separated by electrophoresis on a 1.5% agarose gel contain- ing 2.2 M formamide in 0.2 M morpholinopropanesulfonate at pH 7.0, 0.05 M sodium acetate, and 5 mM EDTA, and transferred to a nitrocellulose sheet. Lane A shows the autoradiogram of mRNA hybridizing to nick-translated pT711, and Lane B shows hybridiza- tion to nick-translated pT106. Size markers used were restriction enzyme (BglI and HzncII) fragments of pBR 322 DNA.

The entire coding region for the human terminal transferase cDNA (Fig. 3) starts with nucleotide 329 and terminates at nucleotide 1852, translating to a 508-amino acid residue pro- tein with a calculated molecular weight of 58,308. The conti- nuity of pT106 sequence with pT711 sequence was confirmed by finding an overlap peptide in our calf thymus terminal transferase sequence (Fig. 3, Peptide 5).

Although the recombinant phage with the 2100-bp insert contains the entire coding sequence for human terminal trans- ferase, this does not represent the full length cDNA sequence of terminal transferase. Sequence analysis showed no poly(A) tail on pT711 although two poly(A) attachment sites (AAA- TAAAT) (23) are present within 174 nucleotides of the end of the pT711 insert (10). The reason for our failure to obtain the poly(A) sequence in the recombinants may be explained by the presence of an EcoRI site near the 5’-end of the pT711 insert.

Comparison of the translated human terminal transferase sequence with amino acid sequences determined for calf thy- mus terminal transferase peptides showed greater than 90% homology between the amino acid sequences of the bovine and human enzyme (Fig. 3). Of the 263 amino acids deter- mined for the calf enzyme, only 24 mismatches with the translated human sequence are found. It would be of great interest to determine the N-terminal amino acid sequence of the 58-kDa bovine protein, but Edman degradation of this peptide indicates that the N-terminal amino acid is blocked. On the other hand, sequencing of a 56-kDa peptide from calf enzyme produced a sequence of 24 amino acids beginning 22

10498 Terminal Transferase in E. coli

Met Asp P ro Pro A r g A l a Ser H i s Leu Ser Pro A r g L y s L y s A r g Pro A r g G l n Thr G l y 20 ATG GAT CCA CCA CGA GCG TCC CAC TTG AGC CCT CGG AAG AAG AGA CCC CGG CAG ACG GGT

Ala Leu Met Ala S e r Ser Pro G l n Asp I l e Lys Phe G l n Asp Leu V a l V a l Phe I l e Leu 40 GCC TTG ATG GCC TCC TCT CCT CAA GAC ATC AAA TTT CAA GAT TTG GTC GTC TTC ATT TTG

756 - - PRO H I S - ""- ASN LEU " "-

G l u Lys Lys Met G l y Thr Thr A r g A r g A l a Phe Leu Met G l u Leu A l a A r g A r g L y s G l y 6 0 GAG AAG AAA ATG GGA ACC ACC CGC AGA GCG TTC CTC ATG GAG CTG GCC CGC AGG AAA GGG """

113 GLN """ 720 - -

Phe A r g V a l G l u Asn G l u Leu Ser Asp Ser V a l Thr H i s I l e V a l Ala G l u Asn Asn Ser 80 TTC AGG GTT GAA AAT GAG CTC AGT GAT TCT GTC ACC CAC ATT GTA GCA GAG AAC AAC TCG - 710""- LEU TYR "- TW"-

G l y Ser Asp V a l Leu G l u Trp Leu G l n A l a G l n Lys V a l G l n V a l Ser Ser G l n Pro G l u 100 GGT TCG GAT GTT CTG GAG TGG CTT CAA GCA CAG AAA GTA CAA GTC AGC TCA CAA CCA GAG

1 4 ALA LEU - Leu Leu Asp V a l Ser Trp Leu I l e G l u C y s I l e G l y Ala G l y Lys Pro V a l G l u Met Thr 120 CTC CTC GAT GTC TCC TGG CTG ATC GAA TGC ATA GGA GCA GGG AAA CCG GTG GAA ATG ACA """""

MET """- I L E - G l y Lys H i s G l n Leu V a l V a l A r g A r g Asp T y r Ser Asp Ser T h r Asn Pro G l y Pro P ro 140 GGA AAA CAC CAG CTT GTT GTG AGA AGA GAC TAT TCA GAT AGC ACC AAC CCA GGC CCC CCG

Lys T h r Pro Pro I l e Ala V a l G l n Lys I l e Ser G l n Tyr A l a C y s G l n A r g Arg Thr Thr 160 AAG ACT CCA CCA ATT GCT GTA CAA AAG ATC TCC CAG TAT GCG TGT CAG AGA AGA ACC ACT

Leu Asn Asn C y s Asn G l n I l e Phe Thr Asp A l a Phe Asp I l e Leu Ala G l u Asn C y s G l u 180 TTA AAC AAC TGT AAC CAG ATA TTC ACG GAT GCC TTT GAT ATA CTG GCT GAA AAC TGT GAG

Phe A r g G l u Asn G l u Asp Ser C y s V a l Thr Phe Met A r g Ala A l a Ser V a l Leu Lys Ser 200 TTT AGA GAA AAT GAA GAC TCC TGT GTG ACA TTT ATG AGA GCA GCT TCT GTA TTG AAA TCT

75 """- pT7 11--->

Leu Pro Phe Thr I le I l e Ser Met Lys Asp Thr G l u G l y I l e Pro C y s Leu G l y Ser Lys 220 CTG CCA TTC ACA ATC ATC AGT ATG AAG GAC ACA GAA GGA ATT CCC TGC CTG GGG TCC AAG """"""""""

V a l Lys G l y I l e I l e G l u G l u I l e I l e G l u Asp G l y G l u S e t Ser G l u V a l Lys Ala V a l 240 GTG AAG GGT ATC ATA GAG GAG ATT AT" GAA GAT GGA GAA AGT TCT GAA GTT AAA GCT GTG """""

Leu Asn Asp G l u A r g T y r G l n Ser Phe Lys Leu Phe Thr S e r V a l Phe G l y V a l G l y Leu 260 TTA AAT GAT GAA CGA TAT CAA TCC TTC AAA CTC TTT ACT TCT GTA TTT GGA GTG GGG CTG

78 ALA " """"""

pT17---> Lys Thr Ser G l u L y s Trp Phe A r g Met G l y Phe A r g Thr Leu Ser Lys V a l A r g Se r Asp 280 AAG ACT TCT GAG AAG TGG TTC AGG ATG GGT TTC AGA ACT CTG AGT AAA GTA AGG TCG GAC " LEU - - - - 719 - - - Lys Ser Leu L y s Phe Thr A r g Met G l n Lys Ala G l y Phe Leu T y r T y r G l u Asp L e u V a l 300 AAA AGC CTG AAA TTT ACA CGA ATG CAG AAA GCA GGA TTT CTG TAT TAT GAA GAC CTT GTC

Ser C y s V a l Thr A r g Ala G l u A l a G l u Ala V a l S e r V a l L e u V a l L y s G l u A l a V a l T r p 320 AGC TGT GTG ACC AGG GCA GAA GCA GAG GCC GTC AGT GTG CTG GTT AAA GAG GCT GTC TGG

71 """ GLY """"

residue 329-1852 of the human cDNA, and the cDNA sequence are shown. Peptides from calf thymus terminal FIG. 3. Complete coding sequence of human terminal transferase. Amino acid residues translated from

transferase (lln) are shown with lines for identity and mismatches spelled out. The N-terminal sequence of the 56- kDa peptide ((56) and the 5'-ends of the pT711 and pT17 sequences are also shown.

Terminal Transferase in E. coli

Ala Phe Leu Pro Asp Ala Phe Val Thr Met Thr Gly Gly Phe Arg Arg Gly Lys Lys Met GCA TTT CTT CCG GAT GCT TTC GTC ACC ATG ACA GGA GGG TTC CGG AGG GGT AAG AAG ATG """ 12 ""

Gly H i s Asp Val Asp Phe Leu I le Thr Ser Pro Gly Ser Thr Glu Asp Glu Glu Gln Leu GGG CAT GAT GTA GAT TTT TTA ATT ACC AGC CCA GGA TCA ACA GAG GAT GAA GAG CAA CTT """"""- ALA "- Leu Gln Lys Val Met Asn Leu Trp Glu Lys Lys Gly Leu Leu Leu T y r Tyr Asp Leu Val TTA CAG AAA GTG ATG AAC TTA TGG GAA AAG AAG GGA TTA CTT TTA TAT TAT GAC CTT GTG

Glu Ser Thr Phe Glu Lys Leu Arg Leu Pro Ser Arg Lys Val Asp Ala Leu Asp H i s Phe GAG TCA ACA TTT GAA AAG CTC AGG TTG CCT AGC AGG AAG GTT GAT GCT TTG GAT CAT TTT

Gln Lys Cys Phe Leu Ile Phe Lys Leu Pro Arg Gln Arg Val Asp Ser Asp Gln Ser Ser CAA AAG TGC TTT CTG ATT TTC AAA TTG CCT CGT CAA AGA GTG GAC AGT GAC CAG TCC AGC

Trp Gln Glu Gly L y s Thr Trp Lys Ala I l e Arg Val Asp Leu Val Leu C y s Pro T y r Glu TGG CAG GAA GGA AAG ACC TGG AAG GCC ATC CGT GTG GAT TTA GTT CTG TGC CCC TAG GAG

'19 ""-"" Arg Arg Ala Phe Ala Leu Leu Gly Trp Thr Gly Ser Arg Phe Glu Arg Asp Leu Arg Arg CGT CGT GCC TTT GCC CTG TTG GGA TGG ACT GGC TCC CGG TTT GAG AGA GAC CTC CGG CGC

123

T y r Ala Thr H i s Glu Arg Lys Met I l e Leu Asp Asn H i s Ala Leu Tyr Asp Lys Thr Lys TAT GCC ACA CAT GAG CGG AAG ATG ATT CTG GAT AAC CAT GCT TTA TAT GAC AAG ACC AAG

112 """" -

""- q17 - - MET """"-" Arg I l e Phe Leu Lys Ala Glu Ser Glu Glu Glu I l e Phe Ala H i s Leu Gly Leu Asp Tyt AGG ATA TTC CTC AAA GCA GAA AGT GAA GAA GAA ATT TTT GCG CAT CTG GGA TTG GAT TAT '13 VAL

I l e Glu Pro Trp Glu Arg Asn Ala - ATT GAA CCG TGG GAA AGA AAT GCC TAG

"""""""""

""- FIG. 3-continued.

10499

340

360

380

400

420

440

460

480

500

so8

amino acids from N-terminal methionine with 4 mismatches with the translated human sequence. A summary of the com- parison of translated amino acid sequence for the human enzyme and the determined amino acid sequences for the calf enzyme is presented in Fig. 3.

In examining the amino acid sequence of the human ter- minal transferase several interesting points are immediately obvious. At the N terminus, residues 11 to 17 are -Pro-Arg- Lys-Lys-Arg-Pro-Arg, representing a perfect nuclear localiza- tion sequence for the protein (24). Terminal transferase in human lymphoblastoid cells indeed has a nuclear localization (2). On the other hand, terminal transferase isolated from the calf thymus by immunoaffinity purification contains a mix- ture of peptides, including 58, 56, 44, and 42 kDa (13). The nuclear and cytoplasmic localization of the enzyme in thymus (3) may be the result of the removal of residues 11-17 (25) during conversion of 58-kDa protein to 56-kDa protein.

The second interesting feature concerns our previous ob- servation that the 58-kDa protein can be phosphorylated with a CAMP-dependent protein kinase while the 56-kDa protein does not have a phosphorylation site (4). The translated human terminal transferase sequence has the sequence -Arg- Ala-Ser-His-Leu-Ser- at residues 5-10 where the Ser-7 could be phosphorylated and the sequence -Pro-Arg-Gln-Thr-Gly- Ala-Leu- at residues 16-22 where Thr could be a potential

phospho~lation site. Analysis of phosphoamino acids in ter- minal transferase showing 85% phosphoserine and 15% phos- phothreonine (5 ) is consistent with our postulated phospho- rylation sites. It is possible that hormone or phorbol ester- induced phosphorylation of the terminal transferase alters the binding of enzyme in the nucleus resulting in degradation of the protein and/or change of cellular localization.

Finally, it is possible to deduce a rough localization of the a- and ,@-peptides of the 32-kDa form of terminal transferase by analysis of the distribution of amino acids in the human cDNA sequence. This analysis assumes that the published amino acid composition of a- and @-peptides for calf thymus terminal transferase (26) provides an appropriate basis for comparison. We believe this assumption is correct because of the high degree of protein sequence homology obtained so far.

Searching the human sequence for cysteine residues we find that 5 cysteines are rather closely grouped between amino acid residues 165 and 402. The bovine @-peptide contains 5 cysteines. There are 2 cysteines remaining in residues 403 to 508 and 2 cysteines in the bovine a-peptide. Table I shows the c o m p ~ i ~ n of amino acid residues present in these frag- ments from bovine and translated human sequences. The localization of composition within the fragments shows good agreement for several additional amino acids. These data suggest that the amino-terminal 158 amino acids are unnec-

10500 Terminal Transferase in E. coli

essary for the catalytic activity of terminal transferase since the 32-kDa calf thymus terminal transferase is fully active.

Expression of Human Terminal Transferase in E. coli-To direct the expression of human terminal transferase in E. coli, it was necessary to reconstruct a continuous coding sequence for the terminal transferase cDNA under the control of a bacterial promoter. pT711 and pT106 were joined through an EcoRI site in the original 2100-bp insert with untranslated regions at both the 5' and 3' ends. A BamHI site is present at nucleotide 331 of the insert in pT106 and corresponds to the second amino acid of the predicted terminal transferase sequence. When the 634-bp BamHI-EcoRI fragment from pT106 is joined to the BamHI site of pUC-19, the terminal transferase-coding region is in the same reading frame as the N-terminal codons of the lac Z gene carried by this plasmid. The protein produced should be under the control of the lac promoter with 15 amino acids derived from the plasmid DNA followed by the amino acids coded by the insert of pT106 minus methionine. The remainder of the terminal transferase coding sequence on the EcoRI fragment of pT711 can be inserted in either orientation in relation to the plasmid DNA. In the correct orientation (pT223) a fused protein would be produced containing 15 nonterminal transferase amino acids at the N terminus followed by the entire human terminal transferase sequence minus the methionine (residue 1). The plasmid with the opposite orientation (pT226) would not be expected to produce active terminal transferase.

When the transformants were screened for production of

TABLE I Amino acid composition comparison

&Peptide Amino acid

a-Peptide

Human cDNA Calf" Human cDNA Calf"

Ala Arg Asp + Asn cy; Glu + Gln GlY His Ile Leu LYS Met Phe Pro Ser Thr Trp TY r Val Total residues

- 13 14 12 10

-

22 21 5 5

2G 25

- - -

- 14 14 2 4

-

10 12 25 21 21 19

7 3 18 10 5 8

19 15 16 13 3 2 5 6

13 13 244 215

8 12 9 2

13 4 3 6

12 7 1 6 3 5 4 4 4 3

106

-

-

-

6 5 8 2 li 6 2 3 7 7 1 3 2 6 4 1 2 4

80

-

-

-

-

Values taken from Ref. 14.

TABLE I1 Terminal transferase activity in E. coli extracts

Soluble extracts from sonicated cells were prepared and terminal transferase activitv was determined as described under "Methods."

E. coli cells Terminal transferase activitv

unitslmg protein HB 101 2.3" HB 101 with pUC-19 2.0 HB 101 with pT201 1.7 HB 101 with pT223 207 HB 101 with ~ T 2 2 6 0.5

_____

Terminal transferase is not present in prokaryotic cells. The low level of activity detected in the assay is probably caused by nonter- minal transferase-polymerizing activities present in the extract.

material reacting with rabbit anti-terminal transferase and the sizes of inserts in the plasmids, three groups were identi- fied. One recombinant of each group was characterized fur- ther. Among pT201, pT223, and pT226, only pT223 produced immunoreactive material. The insert in pT201 was found to be 634 bp, and the inserts in pT223 and pT226 were 1737 bp. When the extracts of cells carrying these plasmids were prepared and analyzed for terminal transferase activity, the results shown in Table I1 were obtained. Significant enzyme activity was detected only when the extract was prepared from cells carrying pT223. Inclusion of isopropyl P-D-thiogalacto- pyranoside in the growth media of HBlOl cells carryingpT223 only increased terminal transferase activity about 10% (data not shown). This is not unexpected for a high copy number plasmid since the number of plasmid molecules/cell exceeds the number of lac repressors. Analysis of crude extracts from cells carrying pT223 on polyacrylamide gels by staining with Coomassie Blue show no additional protein bands (data not shown), indicating that only modest amounts of terminal transferase protein are present. When the separated proteins were subjected to immunoblot analysis, a peptide reacting with rabbit anti-terminal transferase was readily detected in the extract from cells carrying pT223, but not in extracts from any of the controls. This peptide is about 60 kDa, about 2 kDa larger than the KM-3 terminal transferase (data not shown).

Terminal transferase in HBlOl carrying pT223 can be isolated from the bacterial extract by the use of an immu- noaffinity column made with mouse monoclonal antibody to human terminal transferase covalently linked to Protein A Sepharose. Using a one-step procedure, terminal transferase present in the bacterial extract was purified to a specific activity of 25,000 units/mg, about 125-fold purification. Total enzyme activity recovered from 4 g of E. coli cells was about 14,000 units, about 7-fold greater than an equivalent amount of KM-3 cells. Analysis of peptides in the immunoaffinity- purified bacterial terminal transferase by polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate showed three major stained peptides, 60,58, and 54 kDa (Fig. 4-4, lane a). Immunoblot analysis showed that the 60- and the 58-kDa peptides react with rabbit antibody to terminal trans- ferase (Fig. 4B, lane a) demonstrating that the 60-kDa peptide is the recombinant terminal transferase. The 54-kDa peptide represents a contaminating peptide. Some degradation of the recombinant terminal transferase peptide is observed, and the 58-kDa peptide was found to be the major degraded species (Fig. 4B, lane a). These results are an unequivocal demonstra- tion that the cDNA sequence isolated represents human ter- minal transferase sequence.

DISCUSSION

The results obtained in this investigation provide the proof of structure for the coding sequence of human terminal de- oxynucleotidyl transferase, and by homology the basic struc- ture for this highly conserved family of proteins. It is impor- tant to review the validity of our information before it is used for more penetrating analysis of the structure and function of this protein.

The original cloned partial sequence of human terminal transferase (insert in pT17) was detected by screening Xgtll recombinants with polyclonal antibody. pT17 was then used for selection of an mRNA that could be translated in a reticulocyte lysate system to a 58-kDa protein that immuno- precipitated with polyclonal and monoclonal antibody (10). The insert in pT17 was sequenced, and the translated human sequence shows greater than 90% homology with the peptide

Terminal Transferase in E. coli 10501

A m a

”

. . L

B a

W

b c

FIG. 4. Analysis of recombinant terminal transferase puri- fied from E. coli extract. Panel A shows the Coomassie Blue- stained gel of purified E. coli terminal transferase, and Panel B shows the immunohlot of the same set of samples. Lane a represents 3 pg of immunoaffinity-purified E. coli terminal transferase, lune b represents 2 pg of immunoaffinity-purified KM-3 terminal transferase, and lane 3 represents 5 pg of immunoaffinity-purified low molecular weight calf thymus terminal transferase. Molecular weight markers (lune m ) used were bovine serum albumin, ovalbumin, glyceraldehyde-3-phos- phate dehydrogenase, carbonic anhydrase, trypsinogen, trypsin inhib- itor, and a-lactalbumin.

sequences we determined for calf thymus terminal transferase. Thus, two independent determinations demonstrated that pT17 contained partial cDNA sequence for terminal transfer- ase.

The partial sequence in pT17 was used to obtain longer DNA sequences by hybridization screening of our KM-3 cDNA libraries. We obtained two contiguous recombinants, pT106 and pT711, that span a major part of terminal trans- ferase mRNA and detect an identical mRNA on Northern blots. In situ and chromosomal DNA blot hybridization with pT106, pT711, and pT223 provide a self-consistent set local- izing the terminal transferase sequence to the long arm of chromosome 10.’ These recombinant DNAs have been se- quenced and are found to share a continuous open reading frame that translates to amino acid sequence homologous to sequences we determined from calf thymus terminal transfer- ase peptides. Within the 2068-bp cDNA sequence translation fron nucleotide 329 to a stop codon a t nucleotide 1855 pro- duces a protein sequence of 508 residues with a molecular weight of 58,308. Our earlier size estimates had suggested 58

* M. Isohe, K. Huebner, and C. M. Croce, personal communication.

kDa (28) for human terminal transferase. Several regions of the translated protein sequence are of

special interest. From the distribution of amino acids, partic- ularly cysteine, we can deduce that the @-peptide (24 kDa) of the low molecular weight form of terminal transferase (26) must reside within residues 159-402. The a-peptide must lie within residues 403-508. Similar conclusions have been drawn by comparison of amino acid sequences from a and @ peptides derived from bovine terminal transferase (27) with pT17 DNA sequence (10). Residues 1-158 are, therefore, not required for catalytic activity and may contain structure related to other functions of the enzyme. We note that residues 1-23 form a hydrophilic N-terminal “tail” that contains a phosphorylation site and a nuclear localization site. It will be of interest to see if manipulation of these sites will affect the biological prop- erties of the enzyme.

The final proof of terminal transferase structure is provided by demonstration of enzyme activity in a fused protein syn- thesized by E. coli. We constructed an expression plasmid containing coding sequences for 15 residues derived from plasmid DNA fused to 507 codons for human terminal trans- ferase (pT223). Bacterial lysates of cells containing pT223 exhibit terminal transferase activity by direct enzyme assay whereas appropriate controls do not. The enzyme activity present in E. coli lysates containing pT223 can be isolated on monoclonal anti-human terminal transferase adsorbents and is shown to contain enzyme activity on a protein of the expected molecular weight. We believe that we are now in a position to attempt variations of this experiment that will provide further insight into the biological role and function attributes of terminal transferase.

REFERENCES 1. Bollum, F. J. (1979) Blood 54, 1203-1215 2. Gregoire, K. E., Goldschneider, I., Barton, R. W., and Bollum, F.

3. Goldschneider, I., Gregoire, K. E., Barton, R. W., and Bollum, F.

4. Chang, L. M. S., and Bollum, F. J. (1982) J . Biol. Chem. 257,

5. Elias, L., Longmire, J., Wood, A., and Ratliff, R. (1982) Biochem. Biophys. Res. Commun. 106,458-465

6. Desiderio, S. V., Yancopoulos, G. D., Paskind, M., Thomas, E., Boss, M. A., Landau, N. R., Alt, F. W., and Baltimore, D. (1984) Nature 3 11, 752-755

7. Bollum, F. J. (1975) Proc. Natl. Acad. Sci. U. S. A. 72, 4119- 4122

8. Bollum, F. J., Augl, C., and Chang, L. M. S. (1984) J. Biol. Chem.

9. Young. R. A., and Davis, R. W. (1983) Proc. Natl. Acud. Sci. U.

J. (1977) Proc. Natl. Acad. Sei. U. S. A. 74,3993-3996

J . (1977) Proc. Natl. Acad. Sci. U. S. A. 74, 734-738

9588-9592

259,5848-5850

1

1

s. A: so, 1194-1198 0. Peterson, R. C., Cheung, L. C., Mattaliano, R. J., Chang, L. M.

S., and Bollum, F. J. (1984) Proc. Natl. Acad. Sci. U. S. A. 81,

1. Landau, N. R., St. John, T. P., Weissman, I. L., Wolf, S. C., Silverstone, A. E., and Baltimore, D. (1984) Proc. Natl. Acad. Sei. U. S. A. 81,5836-5840

2. Norvander, J., Kempe, T., and Messing, J. (1983) Gene 26,101- 106

3. Augl, C., Lee, S., Breviario, D., Chang, L. M. S., and Bollum, F. J . (1983) Fed. Proc. 42. 2147

4363-4367

14. Chang, L. M. S., and Bolium, F. J . (1971) Biochemistry 10,536- 542

15. Bollum, F. J., and Chang, L. M. S. (1981) J. Biol. Chem. 256, 8767-8770

16. Sanger, F., Nicklen, S., and Coulson, A. R. (1977) Proc. Natl.

17. Maxam, A., and Gilbert, W. (1980) Methods Enzymol. 65,499-

18. Laemmli, U. K., and Favre, M. (1973) J. Mol. Biol. 80, 575-599

Acud. Sei. U. S. A. 74, 5463-5467

560

10502 Terminal Transferase in E. coli 19. Chang, L. M. S. (1971) Biochem. Biophys. Res. Commun. 44, 214

20. Gornall, A. G., Bardawifl, C. J., and David, M. M. (1949) J. Biol. E. (1984) Ce&39,499-509

21. Chang, L. M. S., Rafter, E., Augl, C., and Bollum, F. J. (1984) J, Chem. 257,5700-5706

22. Helfman, D. M., Feramisco, J. R., Fiddes, J. C., Thomas, G. P., 909-916

124-131 24. Kalderon, D., Roberts, B. L., Richardson, W. D., and Smith, A.

Chem. 177,751-766 25. Chang, L. M. S., PIevani, P., and Bollum, F. J. (1982) J. Biol.

Biol. Chem. 259, 14679-14687 26. Chang, L. M. S., and Bollum, F. J. (1971) J. BWZ. Chem. 246,

and Hughes, S. H. (1983) Proc. Natl. Acad. Sci. U. S. A. 80, 27. Beach, C. M., Chan, S. K., Vanaman, T. C., and Coleman, M. S. 31-35 (1985) Fed. Proc. 44,1057

23. Proudfoot, N. J., and Brownlee, G. G. (1976) Nature 2 6 3 , 211- 28. Bollum, F. J., and Brown, M. (1979) Nature 278 , 191-192