Cell Cycle-Dependent Localization and Properties of a Second Mitochondrial DNA Ligase in Crithidia fasciculata

Abstract

The mitochondrial DNA in kinetoplastid protozoa is contained in a single highly condensed structure consisting of thousands of minicircles and approximately 25 maxicircles. The disk-shaped structure is termed kinetoplast DNA (kDNA) and is located in the mitochondrial matrix near the basal body. We have previously identified a mitochondrial DNA ligase (LIG kβ) in the trypanosomatid Crithidia fasciculata that localizes to antipodal sites flanking the kDNA disk where several other replication proteins are localized. We describe here a second mitochondrial DNA ligase (LIG kα). LIG kα localizes to the kinetoplast primarily in cells that have completed mitosis and contain either a dividing kinetoplast or two newly divided kinetoplasts. Essentially all dividing or newly divided kinetoplasts show localization of LIG kα. The ligase is present on both faces of the kDNA disk and at a high level in the kinetoflagellar zone of the mitochondrial matrix. Cells containing a single nucleus show localization of the LIG kα to the kDNA but at a much lower frequency. The mRNA level of LIG kα varies during the cell cycle out of phase with that of LIG kβ. LIG kα transcript levels are maximal during the phase when cells contain two nuclei, whereas LIG kβ transcript levels are maximal during S phase. The LIG kα protein decays with a half-life of 100 min in the absence of protein synthesis. The periodic expression of the LIG kα transcript and the instability of the LIG kα protein suggest a possible role of the ligase in regulating minicircle replication.

The mitochondrial DNA of trypanosomatid protozoa, termed kinetoplast DNA (kDNA), is a network of catenated circular DNAs consisting of interlocked minicircles and maxicircles (19, 40). The kDNA of Crithidia fasciculata contains approximately 5,000 minicircles and 25 maxicircles. The minicircles are 2.5 kb in size and encode small guide RNAs, which are involved in editing of maxicircle transcripts (44). Maxicircles range in size from 23 kb for Trypanosoma brucei to approximately 37 kb for C. fasciculata and encode rRNAs and mitochondrial proteins involved in oxidative metabolism (43). In vivo, C. fasciculata kDNA is condensed into a disk-shaped structure, 1 μm in diameter and ≈0.4 μm thick located at the base of the flagellum, with the flagellum perpendicular to the face of the disk (13, 26). Minicircles in the disk are relaxed, unlike most other circular DNAs in nature, which are negatively supercoiled (36). Electron microscopic studies of sections through the kDNA disk suggest that minicircles are stretched taut and aligned parallel to the axis of the disk, whereas maxicircles appear to be present as a catenated network within the overall network (26, 38, 40).

Replication of kDNA takes place in approximate synchrony with nuclear DNA during S phase (41, 49) unlike in higher eukaryotes, where mitochondrial DNA replicates throughout the cell cycle (6, 14, 48). Each minicircle replicates only once in every cycle, leading to a doubling of the size of the kDNA network, which then segregates into two daughter networks with each cell receiving one progeny network. The unusual kDNA structure and its replication mechanism have been the focus of several recent reviews (19, 20, 25, 40). These studies have shown that different steps of kDNA replication are carried out at discrete sites in and around the kDNA by protein complexes localized at those sites (19). Minicircles are released from the kDNA vectorially as covalently closed circles prior to initiation of replication in a specialized zone called the kinetoflagellar zone (KFZ) at the flagellar face of the kDNA disk (9). Universal minicircle sequence-binding protein, which binds to the two minicircle origins (1) and localizes to two neighboring sites on the flagellar face of the kDNA disk (2), likely plays a role in the initiation of replication. Multiple mitochondrial DNA polymerases have been identified in T. brucei, a related kinetoplastid, and RNA interference studies have identified some as being involved in kDNA replication (21). Minicircle replication is RNA primed (5, 31), and the RNA primers are likely synthesized by a DNA primase that localizes to the two faces of the kDNA disk in C. fasciculata (24). Minicircles replicate by a unidirectional theta-type mechanism in C. fasciculata (5, 11, 17, 18, 39). Initiation of minicircle replication occurs from one of two origins of replication located 180° apart on the circular DNA. Synthesis of the leading strand (L strand) is continuous, and results in a daughter minicircle with a nick or a gap of several nucleotides at the origin of the L strand. A few ribonucleotides remain at the 5′ terminus of the newly synthesized L strands (5, 31). The H strand is synthesized discontinuously after being initiated at a site just downstream of the l-strand origin (4, 5, 17). Minicircle replication appears to initiate in the KFZ (9) and may continue as minicircles associate with two antipodal sites flanking the kDNA disk, where they are ultimately rejoined to the kDNA network. Replication proteins present at the antipodal sites include topoisomerase II (Topo II) (29), DNA polymerase β (DNA Pol β) (13), structure-specific endonuclease (SSE1) (10), and the mitochondrial DNA ligase, LIG kβ (42). SSE1 has RNase H activity and has been implicated in the removal of RNA primers (15), whereas DNA Pol β, a nonprocessive DNA polymerase, is proposed to fill discontinuities between Okazaki fragments (45). The coimmunoprecipitation of LIG kβ and DNA Pol β suggests that they function together to repair gaps between Okazaki fragments (42). Minicircle replication intermediates are only partially repaired prior to their reattachment to the kDNA network by Topo II (47), with nicks and gaps remaining at the replication origins, which are not closed until all minicircles have replicated. The final sealing of all discontinuities occurs only after all minicircles have been replicated and precedes kinetoplast division (13).

We report here the identification of a second mitochondrial DNA ligase from C. fasciculata. We have named it LIG kα based on its kinetoplast localization and its location upstream of LIG kβ. Localization of LIG kα specifically to dividing kinetoplasts during kinetoplast division and segregation suggests that LIG kα may play a role in the final sealing of discontinuities at the replication origins.

MATERIALS AND METHODS

Cloning of the LIG kα gene of C. fasciculata.

A search of Leishmania major and Trypanosoma brucei genome databases revealed the presence of two putative DNA ligase genes. We have termed these genes LIG kα and LIG kβ (8). A phage isolate obtained by screening a C. fasciculata genomic DNA library in λGEM-11 (Promega) to identify the C. fasciculata LIG kβ gene was subsequently used for identification of the upstream LIG kα gene. A Southern blot of the phage DNA probed with a MluI-SalI 540-bp restriction fragment containing the 5′-flanking region of the LIG kβ gene identified an approximately 7.0-kb SalI fragment, which was cloned into similarly digested pGEM-5Zf (+) (Promega) to create the plasmid pLigA.2. Sequence analysis of pLigA.2 showed that it contained both the LIG kβ and the LIG kα coding sequences (accession number AY380335).

Episomal expression of epitope-tagged LIG kα.

pX.2-KO (32) was used to construct an expression vector. pLigA.2 was used as a template to amplify a 1,561-bp fragment containing 900 bp of the 5′-flanking region of LIG kα, along with 661 bp of coding sequence by using primers M36 (GGTACCGATATCTGTGCAGGACGA) and M11 (TCACGGCGACAGCACCAG), and then subcloned into pCRII-TOPO (Invitrogen), yielding plasmid pLigA.9. pLigA.2 was also engineered to remove one Bsp120I site and introduce an in-frame Bsp120I site just before the termination codon of LIG kα. Three copies of the influenza virus hemagglutinin (HA) epitope tag were introduced at this Bsp120I site, yielding pLigA.8. A KpnI-to-NcoI digest of pLigA.9 released a 1,143-bp fragment providing upstream portions of LIG kα, whereas a NcoI-to-SalI digest of LigA.8 released the remainder of the gene with three copies of the HA epitope tag on a 2,914-bp fragment. These fragments were joined with KpnI-SalI-digested pX.2-KO to produce the expression plasmid pLigA.10. Wild-type C. fasciculata cells were transfected with pLigA.10 and selected on agar plates containing Difco brain heart infusion medium and 50 μg of G418/ml as described previously (32).

Immunolocalization of LIG kα.

Immunofluorescent localization of HA epitope-tagged LIG kα was performed essentially as described previously (26). In brief, 2 × 10⁷ cells expressing HA-tagged LIG kα were harvested from either an exponential culture at 2 × 10⁷ cells per ml or from a synchronous culture at 120 min after release from a hydroxyurea block (32), resuspended in phosphate-buffered saline (PBS), and adhered to poly-l-lysine-coated glass slides. Approximately 30% of the cells from the synchronous culture contained two nuclei and either a dividing kinetoplast or two newly divided kinetoplasts. The cells were allowed to adhere for 15 min and then fixed in 4% paraformaldehyde in PBS for 5 min. Fixation was stopped by two 5-min washes in 0.1 M glycine in PBS, followed by incubation for 10 min in 0.025% Triton X-100 in PBS. The slides were kept in methanol at −20°C overnight and rehydrated by three washes in PBS for 5 min each time. The cells were then blocked in 20% goat serum in PBST (PBS-0.05% Tween 20) at 23°C for 60 min, followed by a 60-min incubation with either mouse monoclonal 12CA5 antibody at a 1:500 dilution (Babco, Richmond, CA) or with HA.11 monoclonal antibodies conjugated to Alexa Fluor 488 (Molecular Probes) at a 1:500 dilution to detect HA-tagged LIG kα. Cells were washed three times in PBST for 5 min each, and cells incubated with 12CA5 antibodies were incubated further with goat anti-mouse immunoglobulin G (IgG) Alexa Fluor 594 (1 μg/ml; Molecular Probes) for 60 min at 23°C. The slides were washed again three times for 5 min each time in PBST and were mounted by using SlowFade antifade containing 10 μg per ml of DAPI (4′,6′-diamidino-2-phenylindole; Molecular Probes), which stains both the nucleus and the kinetoplast. Some slides were also immunostained with polyclonal antibodies to the nuclear protein RPA1 and then with goat anti-rabbit IgG conjugated with Alexa Fluor 488. All of the incubations were performed in a humid chamber. For fluorescence microscopy, cells were imaged by using a Zeiss Axioskop II compound microscope with a 63× Plan-neofluor oil-immersion objective lens, and images were captured by using a Zeiss Axiocam digital camera and Zeiss Axiovision 3.0 software. Red, blue, green, and phase-contrast images of the same field were captured independently and were merged by using Adobe Photoshop (v. 7.0). Confocal images were taken with a 100×/1.4 Planapo lens on a Leica TCS-SP MP Confocal and Multiphoton Inverted Microscope (Heidelberg, Germany) equipped with argon (488-nm blue excitation) and 561-nm (green) diode lasers and a two-photon laser setup consisting of a Spectra-Physics Millenia X 532-nm green diode pump laser and a Tsunami Ti-Sapphire picosecond pulsed infrared laser tuned at 768 nm for UV excitation.

Western blot analysis of cycloheximide-treated cells.

C. fasciculata cells expressing HA epitope-tagged LIG kα were treated with 100 μg of cycloheximide/ml for 4 h. Cell aliquots were collected at 1-h intervals starting immediately after the addition of cycloheximide. Total cell lysates were prepared from the cell samples, and equal amounts of protein from each aliquot were fractionated by polyacrylamide gel electrophoresis. The fractionated protein was immunoblotted as described previously (33) and probed with a 1:5,000 dilution of 12CA5 monoclonal antibodies (Babco) to the HA epitope and anti-CSBPA polyclonal antibodies to CSBPA protein (28).

Northern blots.

Wild-type C. fasciculata cells were synchronized by hydroxyurea treatment as described previously (32). Cell aliquots of 1.5 ml each were collected every 30 min after release from a hydroxyurea block (6 h with 200 μg of hydroxyurea per ml). RNA was prepared from the cell aliquots by using an RNeasy kit (QIAGEN). Then, 10 μg of RNA isolated at each time point were electrophoresed on a 1.2% formaldehyde-agarose gel for 17 h at 25 V with continuous circulation of buffer. RNA was transferred onto Hybond-XL membrane (Amersham Biosciences), UV cross-linked, and subsequently probed with ³²P-labeled probes for LIG kα, LIG kβ, and DHFR-TS coding-sequence probes. Northern blots were quantitated by using a Molecular Dynamics PhosphorImager.

Expression and purification of recombinant LIG kα.

The coding region of LIG kα minus its first 37 amino acid residues was cloned into pET22b (Novagen) at the NdeI/XhoI sites. The resulting gene construct contained a His₆ tag in frame with the 3′ end of the cloned fragment. The plasmid construct was transformed into Escherichia coli BL21(DE3) cells and protein expression was induced with 1 mM IPTG (isopropyl-β-d-thiogalactopyranoside) for 45 min at 37°C. The bacterial cells were harvested and resuspended in 20 mM Tris-HCl (pH 8.0) and protease inhibitor cocktail (Novagen). Lysozyme was added to 1 mg/ml and incubated at 30°C for 15 min. Triton X-100 and NaCl were added to 1% and 0.5 M, respectively. The lysate was passed through an 18-gauge needle five times and centrifuged at 15,000 × g for 15 min at 4°C. The supernatant was filtered through a 0.45-μm-pore-size filter and purified by using His-Bind resin (Novagen). The protein was eluted in 20 mM Tris-HCl (pH 8.0)-0.5 M NaCl with 0.3 M imidazole. A 100-μl aliquot of the peak fraction was loaded onto a 1-ml G25 column in 20 mM Tris-HCl (pH 8.0)-100 mM NaCl-2 mM dithiothreitol-0.2 mM EDTA, and fractions of 100 μl were collected. An equal volume of glycerol was added to the eluted fractions, and the protein was stored at −20°C.

Adenylation and deadenylation of recombinant LIG kα.

Adenylation reactions were performed by using recombinant LIG kα protein in reactions containing 10 μl of 20 mM Tris-HCl (pH 8.0), 10 mM MgCl₂, 5 mM dithiothreitol, 8% glycerol, 0.02% Triton X-100, 0.2 μCi of [α-³²P]ATP, and 20 ng of the recombinant protein. Reactions were incubated at 25°C for 15 min. For deadenylation of the ligase-AMP, the reactions were further incubated with 2 μl of DNase I-treated calf thymus DNA (2.5 μg/μl), 10 μl of poly(dA) · oligo(dT)₂₀ (250 ng/μl), or 10 μl of poly(rA) · oligo(dT)₂₀ (250 ng/μl) at 25°C for 15 min. Both the adenylation and deadenylation reactions were stopped by boiling with sodium dodecyl sulfate sample buffer for 5 min, and the products were separated by polyacrylamide gel electrophoresis. The gel was fixed in 30% methanol and 10% acetic acid for 30 min, dried, and autoradiographed (42).

DNA ligase activity of recombinant LIG kα.

The DNA joining activity of the recombinant LIG kα was assayed by using 5′ end-labeled oligo(dT)₂₀ annealed to poly(dA). The substrate was incubated with various amounts of recombinant LIG kα at 27°C for 3 h in ligase buffer (Gibco-BRL). The ligation reaction was stopped by heating it at 85°C for 3 min, and the ligation products were analyzed on a 12% polyacrylamide gel containing 6 M urea.

RESULTS

Identification of a second kinetoplast DNA ligase gene in C. fasciculata.

We have previously described the identification and characterization of a kinetoplast DNA ligase (LIG kβ) in C. fasciculata. This identification was based on the presence of a putative mitochondrial DNA ligase gene in the L. major genome database (42). Subsequent examination of both the T. brucei and L. major genome databases revealed the presence of a second putative mitochondrial DNA ligase gene (LIG kα) upstream of LIG kβ. We have cloned the C. fasciculata LIG kα gene and describe here the characterization of the encoded ligase. The open reading frame of the C. fasciculata LIG kα consists of 663 amino acid residues and contains six colinear motifs characteristic of ATP-dependent DNA ligases (46), including the conserved active site motif-K-DG- (amino acid residues 283 to 286), which is essential for ligase activity. The predicted LIG kα protein does not contain a mitochondrial leader sequence at its N terminus similar to the 9-amino-acid cleavable presequences present in several C. fasciculata kinetoplast proteins (45, 50). However, the MitoProt II computational method for predicting mitochondrially imported proteins predicts mitochondrial import with a probability score of 0.9895. Like LIG kβ, LIG kα does not have significant amino acid sequence homology outside of the conserved motifs with any other known ligases, including human mitochondrial DNA ligase, LIG III (22). The C. fasciculata LIG kα has 37% identity with the T. brucei LIG kα and 58% identity with the L. major LIG kα. The C. fasciculata LIG kα and LIG kβ are only 24% identical (8).

Localization of LIG kα is cell cycle dependent.

Immunostaining of an asynchronous C. fasciculata culture expressing HA-tagged LIG kα from a plasmid construct showed that localization of LIG kα was not detected in all cells but was primarily observed in cells undergoing division and having a dividing kinetoplast or recently divided kinetoplast and two nuclei (Fig. 1A, a to c; B, a to h; and D, a to b; and Table 1). Some cells containing a single nucleus that also show LIG kα localization possibly represent newly divided cells in which the localization is still observable (Fig. 1A, d to f, and C, a and b). In cells from a synchronized culture harvested at a time when ca. 30% of the cells contain either two nuclei and a dividing kinetoplast or two nuclei and two newly divided kinetoplasts, essentially all of the dividing cells show kinetoplast localization of LIG kα. The lack of detection of localization of LIG kα in a few of the dividing cells may be a consequence of the signal being below the level of detection in those cells.

FIG. 1.

Immunolocalization of HA-tagged LIG kα. (A) A dividing cell stained with mouse HA.11 antibodies conjugated with Alexa Fluor 488 (a) and DAPI (b) and a phase-contrast image (c) are shown. A cell with a single nucleus stained overnight with mouse HA.11 antibodies conjugated with Alexa Fluor 488 (d) and with DAPI (e) and a phase-contrast image (f) are also shown. (B) Dividing cells stained with mouse 12CA5 antibodies and goat anti-mouse IgG conjugated with Alexa Fluor 594 and with rabbit anti-RPA1 and goat anti-rabbit IgG conjugated with Alexa Fluor 488 to stain the nucleus (a and e) and DAPI (b), a DAPI and AlexaFluor488 merged image (f), phase images (c and g), and merged images (d and h) are shown. (C) Cells with a single nucleus stained with mouse 12CA5 antibodies and goat anti-mouse IgG conjugated with AlexaFluor 594 and with DAPI (a) and a phase image (b). (D) Confocal image of a dividing cell stained with mouse HA.11 antibodies conjugated with Alexa Fluor 488 and with DAPI shown in false color (red) (a); a phase image is also shown (b).

TABLE 1.

Localization of LIG kα in 1N and 2N cells

Culture	Total no. of cells	No. of ligases localized	% Localized
Exponential culturea
1N	375	41	11
2N	11	8	73
Synchronous cultureb
1N	114	18	16
2N	42	40	95

^aExponential culture at 2 × 10⁷ cells per ml.

^bSynchronized culture harvested 120 min after hydroxyurea release.

In Fig. 1A (a and d) and B (a) localization of LIG kα appears to localize to the two faces of kDNA disk in dividing cells and in a cell containing a single nucleus and a nondividing kinetoplast. An unstained central region is evident in these kinetoplasts, indicating an absence of the ligase in the interior of the disk. Although the epitope-tagged protein appears to localize on both faces of the kinetoplast disk, the localization observed in Fig. 1B (d and h) and D (a) is more pronounced on the flagellar face of the disk in the KFZ, which is proposed to be the site for initiation of minicircle replication (9). Figure 1C shows a pair of cells each having a single nucleus but with the upper cell having greater localization of LIG kα in the KFZ, while the lower cell has greater localization of LIG kα on the opposite face of the kDNA disk. The significance of this latter localization is unknown. LIG kβ colocalizes with DNA Pol β, Topo II, and SSE1 at the two antipodal sites (42) but to a lesser degree also to the two faces of the kDNA disk similar to that of DNA primase (24).

Adenylation and DNA ligase activity of the recombinant LIG kα.

To demonstrate the enzymatic activity of the ligase protein, a His-tagged form of the recombinant LIG kα was expressed and purified from E. coli (Fig. 2). Although a high level of expression of the recombinant protein was obtained (Fig. 2A, lane 2), the protein was mostly insoluble and was removed by centrifugation. However, a sufficient amount of soluble protein remained in the supernatant (Fig. 2A, lane 3) for further purification by metal chelate affinity chromatography and gel filtration (Fig. 2B). The purified recombinant LIG kα was assayed for its adenylation, deadenylation, and DNA ligase activities. ATP-dependent DNA ligases form a ligase-AMP complex as an intermediate in the ligation reaction. Recombinant LIG kα was adenylated in the presence of [α-³²P]ATP as shown in Fig. 3A, lane 1. The ³²P-labeled ligase-AMP intermediate was subsequently discharged upon incubation with DNA ligase substrates (lanes 3 to 5). Nicked calf thymus DNA and oligo(dT)₂₀ annealed to poly(dA) or to poly(rA) were all effective in deadenylation of the ligase-AMP. DNA joining activity of the recombinant protein was also demonstrated in reactions using 5′-end-labeled oligo(dT)₂₀ annealed to poly(dA) as a substrate. Analysis of the reaction products on a denaturing polyacrylamide gel showed that the dT oligonucleotides were ligated into higher oligomers (Fig. 3B), confirming the identification of the protein as a DNA ligase.

FIG. 2.

Purification of recombinant LIG kα. (A) Metal chelate chromatography of His-tagged LIG kα. Lane 1, molecular mass markers (kDa); lane 2, total cell extract; lane 3, column load; lane 4, flow through; lane 5, wash with loading buffer; lane 6, wash with 60 mM imidazole in loading buffer; lanes 7 to 13, elution with 0.3 M imidazole in column buffer. (B) G25 Sephadex desalting column of fraction 9 from panel A. Lane 1, molecular mass markers (in kilodaltons); lane 2, column load; lanes 3 to 13, eluted fractions.

FIG. 3.

Adenylation and ligation by the recombinant LIG kα. (A) Lane 1, adenylation of 20 ng of the recombinant protein in the presence of [α-³²P]ATP. Deadenylation of the adenylated protein upon incubation with 5 μg of DNase I-treated calf thymus DNA (lane 3), 2.5 μg of poly(dA) · oligo(dT) (lane 4), or 2.5 μg of poly(rA) · oligo(dT) (lane 5). Lane 2 shows a mock deadenylation reaction. Reaction products were analyzed by sodium dodecyl sulfate-gel electrophoresis and autoradiography of the dried gel. (B) DNA ligation by recombinant LIG kα. 5′ End-labeled oligo(dT)₂₀ annealed to poly(dA) was incubated with 0, 5, or 10 ng of the recombinant protein (lanes 1 to 3, respectively) in the presence of 1 mM ATP. The reaction products were analyzed on a urea-12% polyacrylamide gel and imaged by autoradiography of the dried gel.

Transcripts of LIG kα and LIG kβ vary out of phase with each other during the cell cycle.

In light of the cell cycle-dependent localization of LIG kα protein, we have investigated the variation in transcript levels of the LIG kα and LIG kβ genes during the cell cycle. A Northern blot analysis of the transcript levels of LIG kα, LIG kβ and DHFR-TS in a synchronized wild-type C. fasciculata cell culture showed that the variation in the transcript levels of both LIG kα and LIG kβ during the cell cycle are out of phase with one another (Fig. 4). The variation in the transcript level of LIG kβ is identical to that of DHFR-TS, which was shown previously to be maximal during S phase and minimal during cell division (32). The transcript levels of both LIG kβ and DHFR-TS peak immediately after the release from a hydroxyurea block when the cells are in S phase and then increase again during 180 to 210 min after hydroxyurea release during the S phase of the next cycle. In contrast, the transcript levels of LIG kα are maximal during 60 to 120 and 210 to 270 min after hydroxyurea release. These periods correspond to the passage of the cells through mitosis and subsequent cell division.

FIG. 4.

Transcript levels of LIG kα and LIG kβ genes during the cell cycle. (A) Transcript levels of LIG kα, LIG kβ, and DHFR-TS genes in a synchronized C. fasciculata culture were analyzed by Northern blot after release from a hydroxyurea block. Numbers at the top indicate the time at which cell aliquots were collected for RNA isolation after hydroxyurea release. (B) PhosphorImager quantitation of the transcript levels in panel A. The relative transcript levels are shown as a function of time after release from the hydroxyurea block.

Cycling sequences flanking the LIG kα and LIG kβ genes.

Previous studies identified an octameric consensus sequence CAUAGAA(A/G) present in the 5′ and/or 3′ untranslated regions of mRNAs encoding several DNA replication proteins in C. fasciculata (3, 7, 27, 30, 32, 34). The central hexamer AUAGAA is highly conserved in these transcripts and has been used, along with additional constraints, to screen the genome database of Leishmania major, a closely related parasite (51). This screen was highly successful in identifying genes expressed during S phase in L. major. We have examined the DNA sequences flanking the C. fasciculata LIG kα and LIG kβ genes for the presence of the conserved hexamer sequence. Although the LIG kβ transcript cycles similarly to those of TOP2, KAP3, RPA1, and DHFR-TS, the hexamer sequence is not present within sequences flanking the LIG kβ gene (Fig. 5). However, there are five copies of the sequence ATAGA in 5′ and 3′ flanking sequences and one copy of the sequence CATAGAGG. DNA LIG kα, on the other hand, has two copies of the conserved hexamer ATAGAA, one of which is within the sequence CATAGAA (Fig. 5). Transcripts of LIG kα would have been predicted to cycle in the same manner as those of TOP2, KAP3, RPA1, and DHFR-TS. Instead, the LIG kα mRNA cycles out of phase with the other cycling transcripts.

FIG. 5.

Sequences related to mRNA cycling sequence elements flanking kinetoplast DNA ligase genes. The highly conserved central hexamer ATAGAA of the consensus octamer cycling sequence (7, 34) and the related sequence ATAGA are shown flanking the coding sequences of LIG kα and LIG kβ.

LIG kα is unstable.

Since ligation of remaining discontinuities at the replication origins of newly synthesized minicircles appears to be inhibited until all minicircles have been replicated, periodic expression of LIG kα mRNA would be particularly relevant if the LIG kα protein had a short half-life. We have therefore investigated the possibility that LIG kα might be an unstable protein. We examined the relative levels of LIG kα by Western blotting of cell extracts from cells in which protein synthesis was inhibited by cycloheximide (Fig. 6A). LIG kα was observed to decay with a half-life of approximately 100 min under conditions where the cell doubling time was 3 to 3.5 h, whereas the C. fasciculata CSBPA protein was stable. No turnover of LIG kβ was observed under identical conditions (Fig. 6B). The striking difference in the signals of CSBPA in Fig. 6A and B is due to the much longer exposure of the blot in Fig. 6A required to detect the LIG kα signal and reflects a lower abundance of LIG kα compared to that of LIG kβ.

FIG. 6.

Turnover of LIG kα in the absence of protein synthesis. C. fasciculata cultures expressing HA epitope-tagged LIG kα (A) and HA epitope-tagged LIG kβ (B) were treated with cycloheximide at 100 μg/ml. Cell aliquots were collected at 1-h intervals and analyzed by Western blotting of total cell extracts by probing with 12CA5 monoclonal antibodies and rabbit polyclonal antibodies to the CSBPA protein used as loading control.

DISCUSSION

The repair of discontinuities in newly replicated minicircles is a highly regulated process and may play an important role in regulating kDNA replication (25, 40). Many, but not all, discontinuities in the newly synthesized minicircle strands are filled and sealed before their attachment to the kDNA network. Pol β and the SSE1 nuclease, present at the antipodal sites, have been shown to be capable of primer removal and gap filling (15). LIG kβ, also present at the antipodal sites is proposed to be involved in joining Okazaki fragments at these sites. However, discontinuities at the replication origins consist of a nick or gap of a few nucleotides and are particularly resistant to closure (5, 18). These remaining discontinuities have been suggested to distinguish the replicated from unreplicated minicircles (12). Once all of the minicircles are replicated there is a final sealing of the discontinuities before the segregation of progeny kDNA networks to the daughter cells (35).

We have sought to identify the proteins involved in the late steps of kDNA replication. The LIG kα identified here is an ATP-dependent DNA ligase and is only distantly related to other mitochondrial DNA ligases and bacterial DNA ligases. Unlike the E. coli DNA ligase that uses NAD⁺ for activation of the enzyme, LIG kα utilizes ATP and forms an enzyme-AMP intermediate in the ligase reaction like most other eukaryotic DNA ligases (23).

LIG kα showed a cell cycle-dependent localization to the kDNA disk and was also concentrated in the KFZ. Localization of LIG kα was observed primarily in cells that had undergone mitosis and contained a dividing kinetoplast or newly divided kinetoplast. In some of these dividing cells LIG kα was present on both faces of the kDNA disk in addition to being present in the KFZ. Cells containing only a single nucleus rarely showed kinetoplast localization of LIG kα and possibly represent newly divided cells with residual LIG kα associated with the kDNA. In these cells LIG kα was often present on the opposite faces of the kDNA disk and was absent from the interior of the disk.

Cell cycle-dependent localization of kinetoplast DNA replication proteins is a common feature of several kinetoplast replication proteins and has been observed for DNA Pol β, Topo II (16), and SSE1 (10) in C. fasciculata cells. These proteins are localized to antipodal sites at the edges of the kDNA disk. Localization of the C. fasciculata universal minicircle sequence-binding protein was also observed to vary during the cell cycle (2). However, unlike these examples, the localization of LIG kα appears to be specific for dividing kinetoplasts or newly divided kinetoplasts. Periodic assembly of kinetoplast replication proteins at distinct sites relative to the kDNA disk may dictate the timing and order of individual steps in kDNA replication as cells progress through the cell cycle.

The variation in the transcript level of LIG kα differs from those of TOP2, KAP3, RPA1, and DHFR-TS (32), transcripts that are expressed at their highest levels during S phase. The transcript level of LIG kα is low after the release from a hydroxyurea block when the cells are in S phase and then increases 90 to 120 min after the release when the cells are going through mitosis and cell division. This also corresponds to a phase of the cell cycle when nuclear S phase has been completed but kinetoplast duplication is not yet completed, as evidenced by the presence of cells containing two nuclei but still have a single elongated kDNA that has not yet divided. The possible links between the expression of the LIG kα transcript, the cell cycle-dependent kDNA localization of the LIG kα protein, and the presence of networks having all of the minicircles nicked/gapped at the replication origins remain to be investigated.

There is still much to be learned concerning the mechanisms of cell cycle regulation of mRNA levels in trypanosomatids and analysis of the cycling of the LIG kα and LIG kβ transcripts should provide new opportunities for further defining the factors regulating transcript levels. Even though sequences flanking LIG kβ do not contain the consensus octamer sequence CATAGAA(A/G) that has been shown to be necessary for cycling of the TOP2, KAP3, RPA1, and DHFR-TS transcripts, the LIG kβ transcripts nonetheless cycle in the same manner. There are, however, five copies of the sequence ATAGA and a single copy of the sequence CATAGAGG near the 5′ and 3′ termini of the LIG kβ coding sequence. Earlier studies showed that multiple copies of closely related sequences including a single copy of the sequence CATAGACC were sufficient to confer cycling on a truncated form of the RPA1 gene (7). In addition, binding to the octamer consensus sequence by the cycling sequence binding protein CSBP in C. fasciculata extracts is reduced but not eliminated by an A-to-C substitution at nucleotide 7 in the octamer sequence (nucleotide 6 in the conserved hexamer core) (27). In contrast, single nucleotide substitutions in any of the first five nucleotides of the hexamer essentially abolished binding by CSBP. Taken together, these results implicate the sequence ATAGA as the most important element in the cycling of these transcripts.

It is much less clear why the LIG kα transcripts do not cycle in the same manner but are 180 degrees out of phase with transcripts of these other genes. The LIG kα gene has two copies of the conserved hexamer sequence 5′ of the LIG kα coding sequence, one of which is in the sequence CATAGAA. It also contains the sequence ATAGAG just 3′ of the LIG kα coding sequence. This result suggests that whereas these conserved sequence elements may be required for transcript cycling in the manner of TOP2, KAP3, RPA1, and DHFR-TS, they are not sufficient. Further examination of the factors involved in the cycling of the LIG kα transcripts may reveal new aspects of the cycling mechanism.

There are previous instances where multiples of replication proteins have been found in kinetoplastid parasites. T. brucei has four mitochondrial Pol I-like DNA polymerases (21) and two DNA Pol β-like polymerases (37). Distinct localizations of these polymerases may be indicative of distinct roles in kDNA replication. The different localization and expression patterns of LIG kα and LIG kβ may also reflect involvement in different stages of kDNA replication. LIG kβ is suggested to be primarily involved in the repair of Okazaki fragments at the two antipodal sites (42), whereas LIG kα may be involved in the final sealing of the discontinuities at the origin of the newly replicated minicircles. LIG kα may have additional roles as well. The higher level of LIG kα in the KFZ where minicircle replication initiates suggests that LIG kα might also participate in joining Okazaki fragments in nascent minicircles prior to their attachment to the antipodal sites. Regulation of LIG kα at the levels of periodic mRNA expression, protein stability, and kinetoplast localization suggests that LIG kα may play a key role in the timing of the final closure of the remaining discontinuities in the minicircle population prior to scission of the double-size networks.