The EMBO Journal Vol.19 No.7 pp.1731–1742, 2000
`
`DNA polymerase mu (Pol μ), homologous to TdT,
`could act as a DNA mutator in eukaryotic cells
`
`Orlando Domı´nguez, Jose´ F.Ruiz,
`Teresa Laı´n de Lera1, Miguel Garcı´a-Dı´az,
`Manuel A.Gonza´ lez1, Tomas Kirchhoff,
`Carlos Martı´nez-A1, Antonio Bernad1 and
`Luis Blanco2
`
`Centro de Biologı´a Molecular Severo Ochoa (CSIC-UAM) and
`1Departamento de Inmunologı´a y Oncologı´a, Centro Nacional de
`Biotecnologı´a (CSIC), Universidad Auto´noma, 28049 Madrid, Spain
`2Corresponding author
`e-mail: lblanco@cbm.uam.es
`
`O.Domı´nguez and J.F.Ruiz contributed equally to this work
`
`This work is dedicated to the memory of Professor Eladio Vin˜uela
`
`A novel DNA polymerase has been identified in human
`cells. Human DNA polymerase mu (Pol μ), consisting
`of 494 amino acids, has 41% identity to terminal
`deoxynucleotidyltransferase (TdT). Human Pol μ, over-
`produced in Escherichia coli in a soluble form and
`purified to homogeneity, displays intrinsic terminal
`deoxynucleotidyltransferase activity and a strong pref-
`erence for activating Mn2⫹ ions. Interestingly, unlike
`TdT, the catalytic efficiency of polymerization carried
`out by Pol μ was enhanced by the presence of a
`template strand. Using activating Mg2⍣ ions, template-
`enhanced polymerization was also template-directed,
`leading to the preferred insertion of complementary
`nucleotides, although with low discrimination values.
`In the presence of Mn2⍣ ions, template-enhanced poly-
`merization produced a random insertion of nucleotides.
`Northern-blotting and in situ analysis showed a pref-
`erential expression of Pol μ mRNA in peripheral
`lymphoid tissues. Moreover, a large proportion of the
`human expressed sequence tags corresponding to Pol μ,
`present in the databases, derived from germinal center
`B cells. Therefore, Pol μ is a good candidate to be the
`mutator polymerase responsible for somatic hyper-
`mutation of immunoglobulin genes.
`Keywords: human DNA polymerase/mutator/somatic
`hypermutation/terminal deoxynucleotidyltransferase
`
`Introduction
`The maintenance and stability of genetic information in
`DNA depends largely on the high fidelity of DNA synthesis
`displayed by replicative polymerases, which in most cases
`are capable of proofreading insertion errors (reviewed by
`Bebenek and Kunkel, 1995). A further improvement in
`DNA stability relies on different systems of DNA repair,
`able to detect and repair most kinds of DNA damage that,
`if left unrepaired, could lead to cell transformation and
`death. Despite all this DNA maintenance enzymology,
`capable of excising bases and nucleotides, repairing breaks
`
`and correcting mismatches, there is some ‘necessary risk’
`of accumulating DNA mutations as the driving force
`allowing evolution. Such background mutability could be
`due to misfunction of replication and the repair machinery,
`but also to the participation of a specific enzymology,
`contributed by mutator (error-prone) DNA polymerases,
`which should work in opposition to DNA repair.
`DNA polymerase beta (Pol β) is one of the known
`cellular DNA polymerases for which a specific role in
`DNA repair has been proposed (Wilson, 1998; Dianov
`et al., 1999). However, a working hypothesis is that some
`altered versions of Pol β could represent mutator DNA
`polymerases acting as dominant error-prone enzymes
`capable of altering the normal
`levels of DNA repair
`(Bhattacharyya and Banerjee, 1997; Clairmont and
`Sweasy, 1998; Clairmont et al., 1999). Recently, three
`novel non-essential cellular DNA polymerases, zeta (ζ;
`Lawrence and Hinkle, 1996), eta (η; Johnson et al., 1999a)
`and theta (θ; Sharief et al., 1999) have been reported.
`Pol ζ and Pol η are capable of altering the outcome of
`the DNA repair process, since these enzymes are able to
`use damaged (unrepaired) DNA efficiently as a template
`(see Friedberg and Gerlach, 1999). In yeast and fungi,
`most spontaneous and damage-induced mutations are
`introduced by Pol ζ responsible for trans-lesion DNA
`synthesis (Han et al., 1998). A model for somatic hyper-
`mutation of Ig genes has been proposed whereby Pol ζ is
`recruited to the Ig locus (Diaz and Flajnik, 1998). Pol η
`is required for error-free replication of UV lesions, and
`its absence produces the variant (V) form of xeroderma
`pigmentosum, an autosomal recessive disease character-
`ized by a high incidence of skin cancers (Johnson et al.,
`1999b; Masutani et al., 1999). More recently, a novel
`DNA polymerase homologous to Pol β, named Pol lambda
`(λ), has been described and predicted to be involved
`in DNA repair synthesis during meiotic recombination
`(M.Garcı´a-Dı´az, O.Domı´nguez, L.A.Lo´pez-Ferna´ndez,
`T.Laı´n de Lera, M.L.Sanı´ger,
`J.F.Ruiz, M.Pa´rraga,
`M.J.Garcı´a, T.Kirchhoff,
`J.del Mazo, A.Bernad and
`L.Blanco, submitted).
`A clear example of the existence of a particular enzymol-
`ogy for the generation of diversity is the enzyme terminal
`deoxynucleotidyltransferase (TdT), although its action
`appears to be restricted to specific genes such as those
`coding for antigen receptors (reviewed in Bentolila et al.,
`1995). TdT is a DNA-independent DNA polymerase, i.e.
`it has the ability to add nucleotides to DNA without any
`template information. This unusual ability is exploited at
`the broken ends of the V(D)J recombination intermediates
`of rearranging antigen receptor genes. The specificity of
`TdT action on these intermediates depends not only on
`its expression pattern, restricted to primary lymphoid
`organs, but also on specific interactions with the Ku
`protein, the DNA-binding component of DNA-dependent
`
`© European Molecular Biology Organization
`
`1731
`
`Illumina Ex. 1115
`IPR Petition - USP 10,435,742
`
`

`

`O.Domı´nguez et al.
`
`protein kinase (DNA-PK), which functions in DNA repair,
`V(D)J recombination and isotype switching (Mahajan
`et al., 1999). Interestingly, TdT and Pol β are evolutionarily
`related, belonging to the family X of DNA polymerases
`(see Ito and Braithwaite, 1991).
`We describe here the identification and preliminary
`biochemical characterization of a novel human DNA
`polymerase, named DNA polymerase mu (Pol μ), with a
`high similarity to TdT, and whose mRNA is highly
`expressed in secondary lymphoid tissues. The purified
`enzyme exhibited both terminal deoxynucleotidyltransfer-
`ase activity and unprecedented error-proneness on primer–
`template structures. We propose that Pol μ could act as a
`DNA mutator polymerase responsible for the somatic
`hypermutation of immunoglobulin genes.
`
`Results
`Identification of a novel terminal transferase in
`human cells
`The identification and cloning of the complete cDNA
`sequence of the novel human DNA polymerase Pol μ
`belonging to family X was carried out as described in
`Materials and methods. Pol μ is closely related to TdT, a
`member of the Pol X family whose template-independent
`polymerization capacity contributes to generation of
`diversity in antigen receptor genes (reviewed in Bentolila
`et al., 1995). Figure 1 shows a multiple alignment of human
`Pol μ and TdTs from different origins, demonstrating
`an overall amino acid identity of 41%. This value is
`significantly higher than that relating Pol μ and Pol β
`(23%) or Pol β and TdT (22%). A nuclear localization
`signal (NLS) of the most common type (SV40 large
`T antigen) is predicted at
`the sequence ‘PKRRRAR’
`(residues 3–9) of Pol μ. A similar sequence in TdTs was
`proposed to act as an NLS (Bentolila et al., 1995).
`Interestingly, residues 22–120 of Pol μ are predicted (see
`Materials and methods) to form a BRCT domain (Bork
`et al., 1997; Callebaut and Mornon, 1997). This domain,
`whose name derives from its initial identification at the
`C-terminal domain of the BRCA1 protein, is proposed to
`mediate protein–protein interactions in a variety of proteins
`involved in DNA repair and cell cycle checkpoint regula-
`tion upon DNA damage (Bork et al., 1997). As indicated
`in Figure 1, Pol μ residues 141–494 form a conserved
`Pol β core, whose three-dimensional structure in the
`presence of DNA and ddCTP has been solved at high
`resolution (Pelletier et al., 1994). Pol μ shares 139 of the
`209 amino acid residues (66%) that are invariant among
`TdTs from very different origins, implying evolutionarily
`conserved structural and perhaps functional relationships.
`Pol μ also conserves 21 of the 23 amino acid residues
`that are conserved among all members of the hetero-
`geneous Pol X family, including all those residues acting
`as metal, dNTP and DNA ligands, or triggering conforma-
`tional changes upon ternary complex formation (see legend
`to Figure 1; for a review, see Oliveros et al., 1997).
`Therefore, as will be described later, characterization of
`the polymerization activity associated with Pol μ was
`necessary in order to determine whether this TdT homo-
`logue has terminal deoxynucleotidyltransferase activity.
`
`1732
`
`Chromosomal mapping of human Pol μ
`The human gene (POLM) coding for Pol μ was mapped
`initially to chromosome 7 by using a panel of human–
`rodent somatic cell hybrids (see Materials and methods).
`By radiation hybrid analysis (see Materials and methods),
`the SHGC marker that best linked with the POLM gene
`was SHGC-6115, with a lod score of 8.2. Based on the
`correspondence of this marker with the GCK gene, the
`POLM gene has been mapped within band 7p13. This
`region constitutes one of the four known fragile sites in
`lymphocytes, with a high incidence of molecular altera-
`tions such as deletions, inversions and translocations.
`DNA polymerase activity associated with Pol μ
`Human Pol μ was overproduced in Escherichia coli and
`purified to homogeneity as described in Materials and
`methods. The 55 kDa recombinant protein was obtained
`in soluble form in a high yield (see Figure 2A). Taking
`into account that family X DNA polymerases are low
`processive enzymes with no proofreading 3⬘–5⬘ exo-
`nuclease, assay conditions were selected to favour detec-
`tion of a TdT-related enzyme such as Pol μ versus
`endogenous E.coli DNA polymerases. Thus, labelling of
`activated (gapped) heteropolymeric DNA was assayed in
`the presence of a low concentration of dATP as a single
`nucleotide, and activating Mn2⫹ ions. These conditions
`would favour incorporation of complementary and non-
`complementary nucleotide by terminal deoxynucleotidyl-
`transferases (DNA independent), and by low-fidelity DNA-
`dependent DNA polymerases without proofreading activ-
`ity. As shown in Figure 2B, under these conditions, DNA
`labelling with commercial TdT was ~10-fold more efficient
`using Mn2⫹ rather than Mg2⫹ activating ions; on the other
`hand, the opposite metal preference was obtained when
`labelling with the Klenow fragment of E.coli Pol I. As
`shown in Figure 2C, DNA labelling activity was detected
`in the 50% ammonium sulfate precipitate corresponding
`to the Pol μ-induced extracts, but it was very low in
`the corresponding uninduced fraction. Interestingly, the
`catalytic efficiency of the induced DNA polymerase was
`20-fold higher in the presence of Mn2⫹ versus Mg2⫹ ions.
`This induced DNA polymerase activity was co-purified
`with the overproduced 55 kDa polypeptide throughout the
`purification procedure. As a further demonstration that the
`induced DNA polymerase activity was intrinsic to Pol μ,
`the heparin–Sepharose fraction (HS) was sedimented on
`a glycerol gradient. As shown in Figure 3, a DNA
`polymerase activity, preferentially activated by Mn2⫹ ions,
`co-sedimented perfectly at a molecular weight correspond-
`ing to the monomeric form of the Pol μ polypeptide. No
`3⬘–5⬘ exo- or endonucleolytic activities were associated
`with the Pol μ peak (data not shown) and, therefore, the
`glycerol gradient fractions 9 and 10 (pooled) were used
`as the enzyme source for further activity assays.
`Pol μdisplays terminal deoxynucleotidyltransferase
`activity but requires a template–primer structure
`for optimal activity
`In agreement with the structural similarity of Pol μ
`and TdT, a terminal deoxynucleotidyltransferase activity
`associated with Pol μ was demonstrated by using different
`oligonucleotides as single-stranded primer substrates,
`again in the presence of Mn2⫹ as the preferred cation.
`
`

`

`Characterization of human Pol μ
`
`Fig. 1. Pol μ, a novel eukaryotic DNA polymerase homologous to TdT. Multiple alignment of human Pol μ (this study) with TdTs from human (Hs; sp
`P04053), bovine (Bt; sp P06526), murine (Mm; sp P09838), Monodelphis domestica (Md; sp Q02789), chicken (Gd; sp P36195) and Xenopus laevis (Xl;
`sp P42118). Numbers between slashes indicate the amino acid position relative to the N-terminus of each DNA polymerase. A putative nuclear
`localization signal (NLS) at residues 3–9 of human Pol μ is boxed. Amino acid residues 22–118 of Pol μ (boxed) are predicted to form a BRCT domain
`(Bork et al., 1997). Amino acid residues 141–494 of Pol μ (boxed) form a conserved Pol β core (see text for details). Invariant residues between Pol μ and
`TdTs are indicated with white letters (on a black background). Identical residues among TdTs are in bold and boxed (grey). Other relevant similarities
`between Pol μ and TdTs are in bold. Conservative substitutions were considered as follows: K, H and R; D, E, Q and N; W, F, Y, I, L, V, M and A; G, S, T,
`C and P. The 23 residues that are invariant among DNA polymerase X members (Oliveros et al., 1997) are indicated with an asterisk. Dots at the bottom
`of the alignment indicate putative homologues to Pol β residues (Pelletier et al., 1994) shown to act either as DNA ligands (Gly64, Gly66, Gly105,
`Gly109, Lys234, Arg254, Arg283 and Tyr296; grey), or as dNTP and metal ligands (Phe272, Gly274, Arg183; Asp190, Asp192 and Asp256; black).
`Squares at the bottom of the alignment indicate putative homologues to Pol β residues involved in interactions between the ‘palm’ and ‘thumb’
`subdomains (Gly179/Phe272; Arg182/Glu316). The total length, in number of amino acid residues, is indicated in parentheses.
`
`1733
`
`

`

`O.Domı´nguez et al.
`
`Fig. 3. Co-sedimentation of a DNA polymerase activity with the Pol μ
`polypeptide. The heparin–Sepharose fraction (HS) shown in Figure 2A
`was sedimented on a glycerol gradient (15–30%) and fractionated as
`described in Materials and methods. The inset shows an SDS–PAGE
`analysis followed by Coomassie Blue staining of some selected
`fractions. Fractions are numbered from the bottom (1) to the top (22).
`Arrows indicate the sedimentation position of several molecular mass
`markers centrifuged under identical conditions. Quantitation of the
`Pol μ band corresponding to each fraction is expressed in arbitrary
`units of optical density (a.u.; right ordinates). DNA polymerase
`activity ([α-32P]dATP labelling of activated DNA) of each fraction,
`assayed for 15 min at 37°C in the presence of 1 mM MnCl2 (see
`Materials and methods), is expressed as dAMP incorporation (left
`ordinates).
`
`Interestingly, the level of Pol μ-catalysed dTMP incorp-
`oration obtained on single-stranded substrates such as
`oligo(dT) or poly(dA), assayed independently, increased
`up to 370-fold when these were pre-hybridized to form a
`template–primer structure (Figure 4B). On the contrary,
`TdT catalysed a similar incorporation on both single-
`stranded homopolymers and a poly(dA)/oligo(dT) sub-
`strate, in agreement with its template independence (data
`not shown). Therefore, and in spite of its intrinsic terminal
`deoxynucleotidyltransferase activity, Pol μ may be defined
`as a DNA-dependent DNA polymerase, since it requires
`a template–primer for optimal activity.
`Pol μis an error-prone DNA-dependent DNA
`polymerase
`When the polymerization assay on activated DNA (used
`to monitor Pol μ activity during purification) was carried
`out in the presence of all four deoxynucleotides, incorpor-
`ation of the labelled dATP substrate by Pol μ was strongly
`inhibited (Figure 5A). In fact, under the standard conditions
`used to assay most DNA polymerases (⬎100-fold unla-
`belled versus labelled nucleotide precursors), Pol μ activity
`would not be detectable. A similar inhibition was obtained
`with TdT whereas, in the case of the Klenow enzyme, an
`increase (11-fold) in dAMP incorporation was obtained
`by addition of all four nucleotide substrates, as expected.
`Moreover, Pol μ-catalysed dAMP incorporation on
`poly(dT)/oligo(dA) was also inhibited strongly by rela-
`
`Fig. 2. Expression of human Pol μ in Escherichia coli. (A) Coomassie
`Blue staining after SDS–PAGE separation of control non-induced (NI)
`and IPTG-induced (I) extracts of E.coli BL21(DE3) cells transformed
`with the recombinant plasmid pRSET-hPolμ, and further purification
`steps of the latter extracts. The mobility of the induced protein Pol μ
`was compatible with its deduced molecular mass (55 kDa/494 amino
`acids). After PEI precipitation of the DNA, Pol μ was precipitated
`with 50% ammonium sulfate (AS), and purified further by
`phosphocellulose (PC) and heparin–Sepharose (HS) chromatography,
`as described in Materials and methods. The electrophoretic migration
`of a collection of molecular mass markers (MW) is shown at the left.
`(B) Relative activation by Mg2⫹ versus Mn2⫹ of TdT and Klenow
`enzymes during DNA polymerization ([α-32P]dATP labelling) on
`activated DNA. TdT (5 U) and Klenow (1 U) were assayed for 30 min
`at 37°C, in the presence of either 10 mM MgCl2 or 1 mM MnCl2 as a
`source of activating metal ions. DNA polymerase activity, expressed as
`dAMP incorporation, was quantitated as described in Materials and
`methods. (C) DNA polymerization activity associated with Pol μ
`expression. The 50% AS fraction corresponding to either non-induced
`(N.I.) or induced extracts was assayed and quantitated as described in
`(B).
`
`As shown in Figure 4A, Pol μ was able to catalyse
`polymerization of any of the four dNTPs to a single-
`stranded DNA primer in the absence of a template. The
`catalytic efficiency of
`the terminal deoxynucleotidyl-
`transferase activity of Pol μ varied as a function of the
`nucleotide used, dTTP and dCTP (both pyrimidines) being
`inserted the most efficiently, and dATP the least efficiently.
`A different nucleotide preference was observed when
`using TdT, dGTP and dCTP being the preferred nucleotide
`substrates under these conditions. The terminal deoxynu-
`cleotidyltransferase associated with Pol μ was also active,
`although less efficient, on double-stranded DNA substrates
`where the primer terminus was paired with a 5⬘-terminal
`complementary nucleotide (blunt-ended), a behaviour
`already described for TdT (results not shown).
`
`1734
`
`

`

`Characterization of human Pol μ
`
`tively low concentrations of any of the other three (non-
`complementary) nucleotides (Figure 5B). An identical
`behaviour was obtained by using TdT in a parallel assay
`(results not shown). On the contrary, non-complementary
`nucleotides did not inhibit polymerization by the Klenow
`enzyme (see Figure 5B). These results suggest that dAMP
`incorporation by Pol μ is being competed by the other
`nucleotides, as would be expected either for a template-
`independent terminal transferase such as TdT, or for a
`DNA-dependent DNA polymerase with a poor template-
`directed nucleotide discrimination.
`The ability of Pol μ to select among the four
`deoxynucleotides (base discrimination) to catalyse faithful
`template-directed DNA synthesis was evaluated initially
`on the four template–primer structures depicted in Figure 6,
`obtained as described in Materials and methods. The
`four dNTPs, at varying concentrations, were assayed
`individually as a substrate for each of the four template–
`primer structures,
`thus representing the 16 possible
`template–substrate nucleotide pairs (four matched ⫹ 12
`mismatched). The same primer molecule (without any
`template) was assayed in parallel to estimate the residual
`terminal transferase activity of Pol μ with each of the four
`dNTPs. As shown in Figure 6, under the conditions used
`in this experiment, only dTMP was incorporated, with
`either Mg2⫹ or Mn2⫹,
`into the single-stranded DNA
`primer. On the contrary, on the four template–primer
`structures, preferential insertion of the nucleotide comple-
`mentary to the first template base was observed in the
`presence of Mg2⫹ ions, indicating that the catalytic effici-
`ency (Kcat/Km) was improved greatly by template selection
`of the incoming nucleotide (note that the complementary
`nucleotide is provided at a 10-fold lower concentration
`than that used in the control without template). However,
`template instructions appear not to be very rigorous, since
`the enzyme is able to add non-complementary nucleotides
`at 100 μM (dT and dA in Figure 6), and non-complement-
`ary dCTP at higher concentrations (data not shown). The
`probability of G:A misincorporation was estimated to be
`only 10- to 50-fold lower than that of a correct G:C pair
`(data not shown). Considering the efficiency of dTMP
`incorporation on single-stranded DNA, it cannot be ruled
`out
`that
`the observed insertion of dTMP using non-
`complementary templates might be due to residual amounts
`of non-hybridized primer.
`In the presence of Mn2⫹ as metal activator, the pattern
`and efficiency of nucleotide incorporation changed drastic-
`ally (see Figure 6). In all cases, dNTP incorporation was
`driven by the presence of a template DNA (note that the
`nucleotide concentration was 1000-fold lower than that
`used for Mg2⫹ activation), but with a poor or null base
`selectivity. As an example, when dC is the first template
`base, the four dNTPs appear to have similar probabilities
`of being inserted. Exceptionally, dGTP incorporation
`occurred mainly in front of its complementary nucleotide.
`Moreover,
`inserted errors are elongated efficiently,
`favoured not only by complementarity but also as reitera-
`tive misinsertions, particularly when using dTTP and
`dATP substrates. In the same assay, a Pol β-like enzyme
`of only 20 kDa (ASFV Pol X) was shown to extend the
`four template–primer structures by adding only the correct
`(complementary) deoxynucleotide, but not by adding an
`excess (400 μM) of each of the three incorrect (non-
`
`Fig. 4. Pol μ has terminal transferase activity, but requires a template–
`primer structure for optimal efficiency. (A) Terminal transferase
`activity associated with human Pol μ. The assay was carried out as
`described in Materials and methods, using 3.2 nM 5⬘-labelled single-
`stranded 19mer (P19) as substrate, 1 mM MnCl2 as a source of
`activating metal ions, 80 μM each individual deoxynucleotide, and
`either TdT (2.5 U/41 ng) or Pol μ (20 ng). A control reaction in the
`absence of enzyme (C) was also carried out. After incubation for
`30 min at 30°C, extension of the 5⬘-labelled oligonucleotide was
`analysed by 8 M urea–20% PAGE and autoradiography. (B) Template-
`dependent polymerization catalysed by Pol μ. Polymerization
`efficiency was assayed comparatively on either poly(dA) (s),
`oligo(dT) (u) or a poly(dA)/oligo(dT) hybrid (d) to provide a
`homopolymeric template (dA)n. The assay was carried out in the
`presence of 1 mM MnCl2, 13 nM [α-32P]dTTP, Pol μ (20 ng) and
`0.5 μM each DNA substrate. After incubation for the indicated times
`at 37°C, dTMP incorporation was quantitated as described in Materials
`and methods.
`
`

`

`O.Domı´nguez et al.
`
`complementary) deoxynucleotides (Oliveros et al., 1997).
`Similar results were obtained when Mn2⫹ was used instead
`of Mg2⫹ as metal activator (results not shown).
`All these results demonstrate that Pol μ is an error-
`prone DNA-dependent DNA polymerase. Interestingly, in
`the presence of its preferred activator (Mn2⫹), Pol μ
`
`1736
`
`behaves as a strong mutator, lacking base discrimination
`during nucleotide insertion on a DNA template–primer
`structure. Exceptionally, Pol μ preferentially inserts a dG
`in front of its complementary dC template base even in
`the presence of Mn2⫹ ions.
`Pol μmRNA is expressed predominantly in
`peripheral lymphoid tissues
`Quantitative analyses of Pol μ transcription levels in
`different human tissues were carried out by Northern
`blotting using commercial membranes containing normal-
`ized amounts of poly(A)⫹ RNA from different human
`tissues (see Materials and methods). As shown in Figure 7,
`a major transcript migrating at ~2.6 kb, in agreement with
`the size of the cDNA isolated (2589 nucleotides), was
`accumulated at the highest level in lymph nodes, followed
`by spleen, thymus, pancreas and peripheral blood lympho-
`cytes. Lower levels of this transcript were present in the
`other tissues examined, being undetectable only in lung.
`By searching the expressed sequence tag (EST) database
`using Pol μ cDNA, we identified a collection of 40 ESTs
`corresponding to human Pol μ; 37% of these ESTs derive
`from different tumours. It is worth noting that 36% of the
`non-tumoural ESTs of Pol μ derive from human tonsillar
`cells enriched for germinal centre B cells (CD20⫹, IgD–)
`by flow sorting (library NCI-CGAP-GCB1). The signifi-
`cance of this finding is higher considering the fact that
`only 7% of the available ESTs corresponding to Pol β, a
`housekeeping DNA repair polymerase, derive from germ-
`inal centre B cells. All these data suggested that Pol μ
`mRNA could be expressed preferentially in human B
`cells, particularly in populations associated with the germ-
`inal centre structures present
`in secondary lymphoid
`organs. To confirm this suggestion, we used in situ
`hybridization to analyse the localization, at the cellular
`level, of the Pol μ mRNA present in different human
`tissues. Using a specific antisense probe corresponding to
`the first 1200 nucleotides of Pol μ cDNA (see Materials
`and methods), expression of Pol μ mRNA was observed
`in tissue sections corresponding to human secondary
`lymphoid organs. Thus, as shown in Figure 8, and in
`agreement with the data obtained by Northen blotting, the
`stronger signal was found in peripheral
`lymph node
`
`Fig. 5. Inhibition of DNA-directed synthesis by non-complementary
`dNTPs. (A) Inhibition of [α-32P]dATP labelling of activated (gapped)
`DNA by addition of different concentrations of a mixture of dC, dG
`and dTTP, in the presence of 1 mM MnCl2 (a scheme is depicted).
`Under the standard conditions described in Materials and methods,
`only dATP (13 nM) is used as substrate for this assay. After
`incubation for 15 min at 37°C in the presence of either TdT
`(2.5 U/41 ng), Klenow (1 U) or Pol μ (20 ng), and the concentration
`indicated of dNTPs, dAMP incorporation on activated DNA was
`expressed as a percentage of that obtained under standard assay
`conditions: 100% represents either 73 (TdT), 13 (Klenow) or 8 (Pol μ)
`fmol of incorporated dAMP. (B) A similar analysis was carried out,
`but using a poly(dT)/oligo(dA) hybrid to provide a homopolymeric
`template (dT)n. The assay was carried out in the presence of 1 mM
`MnCl2, 13 nM [α-32P]dATP as the correct nucleotide, either 20 ng of
`Pol μ (circles) or 1 U of Klenow (squares), and the concentration
`indicated (on the abscissa) of individual non-complementary dNTPs.
`After 5 min at 37°C, dAMP incorporation on poly(dT)/oligo(dA) was
`expressed as a percentage of that obtained when non-complementary
`nucleotides were added: 100% represents either 23 (Pol μ) or 127
`(Klenow) fmol of incorporated dAMP.
`
`

`

`Characterization of human Pol μ
`
`other human organs such as muscle, lung or even bone
`marrow (the latter shown in Figure 8), a myelo-lymphoid
`tissue, were negative or faintly positive in comparison
`with the corresponding negative control. As shown in
`Figure 8, the level of expression of the Pol μ mRNA in
`lymph nodes seems to be high, in comparison with other
`markers for centroblastic populations such as A-myb (data
`not shown). The in situ hybridization pattern obtained is
`compatible with a preferential expression in the follicular
`lymphoid region, with a variable expression in different
`areas and not restricted to a particular cell subpopulation.
`The expression level of Pol μ mRNA in spleen is lower
`in comparison with that observed in lymph nodes, and
`more restricted to particular structures, such as that shown
`in Figure 8, which resemble the typical organization of a
`germinal centre in a secondary follicular area. Also in this
`case, expression of Pol μ mRNA does not seem to be
`associated preferentially with a discrete cell subpopulation.
`In summary, it seems that Pol μ mRNA is expressed
`preferentially in the follicular areas of human secondary
`lymphoid organs. Therefore, although immunostaining of
`Pol μ in similar tissue sections is still
`lacking,
`it
`is
`tempting to speculate that Pol μ expression could be
`associated with the generation, in these organs, of germinal
`centres, specialized structures that are critical for the
`maturation of the immune humoral response.
`
`Discussion
`The amino acid sequence derived from the Pol μ cDNA
`(a total of 494 amino acids) showed 41% identity with
`TdTs from different origins, suggesting that Pol μ could
`be involved in specific processes such as V(D)J recombina-
`tion of Ig genes, in which TdT contributes to variability
`by adding non-template nucleotides. Interestingly, Pol μ
`contained 21 of the 23 residues that are invariant in all
`DNA polymerases from family X, including all those
`involved in DNA and nucleotide binding, catalysis of
`polymerization and conformational changes involved in
`the polymerization cycle, as defined for Pol β, the para-
`digmal enzyme for DNA polymerase family X. In order
`to delineate the function of Pol μ, we carried out a
`preliminary biochemical characterization of the human
`Pol μ protein overproduced in E.coli, together with expres-
`sion analysis in different tissues at the RNA level.
`Pol μ, a mutator DNA polymerase
`The results presented here demonstrate that Pol μ shares
`enzymatic properties with different members of the Pol X
`family. Like TdT, its closest homologue, Pol μ has an
`intrinsic terminal transferase activity that preferentially
`inserts pyrimidine nucleotides. Thus, its relative nucleotide
`usage, different
`from that observed for TdT,
`is:
`dT⬎⬎dC⬎dG⬎dA. As for Pol β and other DNA-depend-
`ent DNA polymerases, but unlike TdT,
`the catalytic
`efficiency of nucleotide incorporation by Pol μ was greatly
`enhanced by the presence of a template strand. Both
`the terminal
`transferase and DNA-dependent DNA
`polymerization activities of Pol μ are strongly activated
`in vitro by manganese ions. Interestingly, in the presence
`of its preferred activator (Mn2⫹), Pol μ behaves as a
`strong mutator, lacking any base discrimination during
`nucleotide insertion.
`
`1737
`
`Fig. 6. Pol μ-catalysed misinsertion at the four template bases. The
`four template–primer structures used, which differ only in the first
`template base (outlined), are indicated on the left. The single-stranded
`oligonucleotide corresponding to the primer strand was assayed in
`parallel as a control of DNA-independent nucleotide insertion. Mg2⫹-
`activated nucleotide insertion on each 5⬘-labelled DNA substrate
`(3.2 nM) was analysed in the presence of either the complementary
`nucleotide (10 μM) or each of the three incorrect dNTPs (100 μM),
`as described in Materials and methods. Mn2⫹-activated nucleotide
`insertion was assayed with each of the four dNTPs (0.1 μM). After
`incubation for 15 min at 30°C in the presence of 20 ng of human
`Pol μ, extension of the 5⬘-labelled (*) strand was analysed by
`electrophoresis in an 8 M urea–20% polyacrylamide gel and
`autoradiography.
`
`Fig. 7. Pol μ mRNA is expressed preferentially in secondary lymphoid
`organs. Northern blotting analysis of TdT-2 mRNA was carried out as
`indicated in Materials and methods, using commercial blots (MTN and
`MTN-II blots, Clontech) containing poly(A)⫹ RNA from the human
`tissues indicated. The membrane was hybridized with a specific
`32P-labelled DNA probe containing 1141 nucleotides of the Pol μ
`cDNA 3⬘-terminal sequence. The hybridized probe, revealing a major
`transcript (2.6 kb), was detected by autoradiography.
`
`sections; a strong signal was also found in sections from
`spleen. Hybridization specificity was assessed by using a
`sense riboprobe under the same experimental conditions
`and in a close parallel tissue section, not producing a
`comparable signal (see Figure 8). In similar experiments,
`
`

`

`O.Domı´nguez et al.
`
`Fig. 8. In situ hybridization of Pol μ mRNA in different human tissues. DIG-labelled sense and antisense riboprobes, corresponding to the first 1200
`nucleotides of Pol μ cDNA, were obtained and hybridized to human tissue sections (Human Tissue Set I and Human Hematal and Immune Tissue
`Set, Novagen) under the conditions described in Materials and methods. After hybridization, detection of the RNA probes in tissue sections was
`carried out by incubation with anti-DIG–alkaline phosphatase antibody.The dark blue staining, observed in lymph nodes and spleen with the
`antisense riboprobe, outlined regions largely expressing Pol μ mRNA. No comparable signal was obtained by using a sense riboprobe under the
`same experimental conditions and in a close parallel tissue section.
`
`Structural studies demonstrated that two metal ions,
`coordinated by a triad of carboxylate residues, are required
`at