TbKAP6, a Mitochondrial HMG Box-Containing Protein in Trypanosoma brucei, Is the First Trypanosomatid Kinetoplast-Associated Protein Essential for Kinetoplast DNA Replication and Maintenance

Abstract

Kinetoplast DNA (kDNA), the mitochondrial genome of trypanosomatids, is a giant planar network of catenated minicircles and maxicircles. In vivo kDNA is organized as a highly condensed nucleoprotein disk. So far, in Trypanosoma brucei, proteins involved in the maintenance of the kDNA condensed structure remain poorly characterized. In Crithidia fasciculata, some small basic histone H1-like kinetoplast-associated proteins (CfKAP) have been shown to condense isolated kDNA networks in vitro. High-mobility group (HMG) box-containing proteins, such as mitochondrial transcription factor A (TFAM) in mammalian cells and Abf2 in the budding yeast, have been shown essential for the packaging of mitochondrial DNA (mtDNA) into mitochondrial nucleoids, remodeling of mitochondrial nucleoids, gene expression, and maintenance of mtDNA. Here, we report that TbKAP6, a mitochondrial HMG box-containing protein, is essential for parasite cell viability and involved in kDNA replication and maintenance. The RNA interference (RNAi) depletion of TbKAP6 stopped cell growth. Replication of both minicircles and maxicircles was inhibited. RNAi or overexpression of TbKAP6 resulted in the disorganization, shrinkage, and loss of kDNA. Minicircle release, the first step in kDNA replication, was inhibited immediately after induction of RNAi, but it quickly increased 3-fold upon overexpression of TbKAP6. Since the release of covalently closed minicircles is mediated by a type II topoisomerase (topo II), we examined the potential interactions between TbKAP6 and topo II. Recombinant TbKAP6 (rTbKAP6) promotes the topo II-mediated decatenation of kDNA. rTbKAP6 can condense isolated kDNA networks in vitro. These results indicate that TbKAP6 is involved in the replication and maintenance of kDNA.

INTRODUCTION

Trypanosoma brucei is a unicellular eukaryotic parasite that causes human African trypanosomiasis (HAT) and nagana in livestock in sub-Saharan Africa. The high toxicity of most current chemotherapies and the emergence of drug-resistant parasites are stimulating efforts to identify promising molecular targets and develop next-generation therapeutics (1, 2). These efforts require better understanding of trypanosome basic biology, which differs markedly from that of its host. For instance, trypanosome mitochondrial DNA is organized as a massive chain mail-like network, known as kinetoplast DNA (kDNA). kDNA consists of several thousand minicircles (1 kb) catenated with a few dozen maxicircles (23 kb). This giant planar network is condensed into a disk within the mitochondrial matrix and connected to the flagellar basal body by a transmembrane filament system named the tripartite attachment complex (see the review in reference 3). The lack of a similar DNA network in mammalian cells suggests that kDNA and proteins involved in its metabolism could be appealing therapeutic targets. Indeed, kDNA replication is the primary therapeutic target for ethidium bromide, a drug still used to treat nagana in livestock (4).

Trypanosome kDNA replication is unusual in comparison with those of the mitochondrial genome in other eukaryotes (see reviews in references 3, 5, and 6). Instead of the asynchronous replication of the mitochondrial genome, as in other eukaryotes, kDNA replication occurs nearly simultaneously with nuclear replication during the S phase of the cell cycle. Replication starts with the release of covalently closed minicircles into the kinetoflagellar zone (KFZ), a region between the proximal face (facing the flagellum) of the disk and the mitochondrial membrane. In the KFZ, each free minicircle is thought to undergo unidirectional theta-replication, producing two minicircle progeny. The progeny then migrate to the antipodal sites (AS), two nucleoprotein complexes flanking the kDNA disk ∼180° apart and containing a mitochondrial topoisomerase II (mtTopo II) and other enzymes. Here, some of later stages of replication occur, such as the removal of RNA primers from Okazaki fragments and the filling of gaps, prior to reattachment to the network by mtTopo II. Maxicircles also replicate unidirectionally as theta structures but, unlike minicircles, they replicate while still attached to the network.

Prior to replication, the kDNA network is a highly condensed disk-shaped structure, in which there must be proteins or protein complexes that stabilize the disk architecture. Within the disk, the minicircles are stretched out and stand side by side, interlocked with their neighbors. This arrangement explains why the disk thickness is half of the circumference of a minicircle. It has been shown that small basic proteins such as histone H1-like, kinetoplast-associated proteins (KAP) in Crithidia fasciculata not only condense the isolated kDNA networks in vitro but colocalize with the kDNA disk in vivo (7, 8). These proteins are likely involved in stabilizing the kDNA disk structure in vivo. For example, a CfKAP1-null mutant strain has an altered kDNA structure in vivo with DNA fibers packed into much thicker strands separated by the electron-lucent zones, which are not present in the wild-type cells. Moreover, the abnormal kDNA structure can be rescued to a nearly normal phenotype by the ectopic expression of CfKAP1 in the null mutant strain (9). Interestingly, maxicircle-encoded mRNA levels increase 2- to 4-fold in CfKAP2 and CfKAP3 double-knockout cells, which indicates CfKAP2 and CfKAP3 may also play a role in regulating kDNA gene expression (10). Trypanosoma cruzi KAP4 and KAP6 localize on kDNA, but their localizations change during the parasite differentiation process, which led to the speculation that TcKAP4 and TcKAP6 might be involved in the kDNA architectural rearrangement (11).

As mentioned above, kDNA minicircles replicate outside kDNA disks. This requires covalently closed minicircle replication precursors to be released from the highly condensed kinetoplast DNA disk. The major protein to be involved in minicircle release must be a type II topoisomerase (12, 13). More recently, in vitro studies have demonstrated that Crithidia fasciculata universal minicircle sequence-binding protein (CfUMSBP) can decondense CfkDNA networks that had been condensed by CfKAP3 or CfKAP4 (14). CfUMSBP is well known to bind the origin sequence (universal minicircle sequence [UMS]), but CfUMSBP-mediated decondensation depends upon interactions between two proteins and not the DNA. This decondensation rendered the kDNA network accessible to human topo II, yielding free kDNA minicircle monomers. These in vitro results led to the proposal that CfUMSBP-mediated remodeling of condensed kDNA networks may function in vivo in promoting the accessibility of the network to topoisomerase II, which results in the release of minicircles from the network, enabling the initiation of minicircle replication.

In higher eukaryotes, nuclear chromosomes are packaged into condensed chromatin structures by histones. The chromatin structures and functions are modulated by two major groups of abundant chromosomal proteins, the histone H1 family and the nonhistone high-mobility group (HMG) superfamily, such as HMG box-containing proteins (15). The HMG box is a DNA-binding domain (∼75 amino acids [aa]) which binds into the DNA minor groove without sequence specificity, slightly intercalates base pairs, and induces a strong bend in the DNA helix axis. Similar to its nuclear counterpart, mitochondrial DNA (mtDNA) is also compacted into a higher-order structure called the mitochondrial nucleoid (mt-nucleoid) predominantly by nonhistone HMG box-containing proteins (16). For instance, Saccharomyces cerevisiae, an HMG box-containing protein, Abf2, compacts mitochondrial DNA into mt-nucleoids and plays an important role in the maintenance and metabolism of mtDNA (17, 18). In mammalian cells, mitochondrial transcription factor A (TFAM) is a central player in the mt-nucleoid packaging, gene expression, and mitochondrial genome copy number (16, 19). Since s classical histones exist in trypanosome mitochondria, we thought that HMG box-containing proteins could be involved in the kDNA condensation and maintenance throughout the cell cycle. We then examined proteins annotated as KAPs in the T. brucei genome (http://tritrypdb.org/tritrypdb/) or unannotated T. brucei KAP homologues (11) and found four of them are HMG box-containing proteins with mitochondrial targeting signals. Here, we report studies on TbKAP6, a mitochondrial HMG box-containing protein in T. brucei, which is homologous to CfKAP4 and TcKAP6 (see Fig. S1 in the supplemental material). This protein localizes in the kDNA disk and, in vitro, it can condense isolated kDNA networks into DNA-protein aggregates with a size similar to that in vivo. RNA interference (RNAi) knockdown of TbKAP6 stops cell growth and inhibits release of covalently closed minicircle. Overexpression of TbKAP6 quickly promotes the abnormally high minicircle release. Both RNAi and overexpression alters the kDNA organization and results in kDNA loss. Overexpression also causes defects in kDNA segregation. Therefore, we conclude that TbKAP6 is essential for cell viability and is involved in kDNA replication and maintenance.

MATERIALS AND METHODS

Trypanosome strains and growth.

T. brucei strain TREU 927 procyclic form was grown at 27°C in SDM-79 medium supplemented with 10% fetal bovine serum (Sigma) and used for the localization of TbKAP6. Procyclic strain 29-13 (from G. Cross, Rockefeller University) was cultured at 27°C in SDM-79 medium supplemented with 10% fetal bovine serum (Sigma), 15 μg/ml G418, and 50 μg/ml hygromycin B. Cell cultures were routinely diluted with fresh medium when cell density reached late-log phase. Crithidia fasciculata (UC strain from Larry Simpson but cultured in our laboratory for many years) was grown in 100 ml 3.7% brain heart infusion medium (supplemented with 10 μg/ml hemin) at room temperature for the isolation of a large amount of kDNA.

TbKAP6-Myc tagging and immunofluorescence microscopy.

The C-terminal 3×Myc tagging of one allele of TbKAP6 was conducted as described previously (20). The fixation, permeabilization, and immunostaining of TbKAP6-Myc cells were performed as described previously (21), using rabbit anti-Myc polyclonal antibody (1:200) (Santa Cruz) and Alexa Fluor 594-conjugated goat anti-rabbit IgG (1:200) (Molecular Probes).

RNAi and overexpression of TbKAP6.

For RNAi experiments, we used the RNAit program (http://trypanofan.path.cam.ac.uk/software/RNAit.html) for the specific target selection and identification of primers (22). The selected target (500 bp) in TbKAP6 coding sequence was then amplified with primers in Table S1 in the supplemental material and Tb29-13 genomic DNA as the PCR template. A stem-loop construct, SLTbKAP6, was assembled as described previously (23). For the overexpression of TbKAP6, the full-length coding sequence was PCR amplified with primers in Table S1 in the supplemental material and cloned into the overexpression vector pLew100V5BSR (a gift from Ruslan Aphasizhev, University of California, Irvine) to obtain the iOETbKAP6 construct (24). The final SLTbKAP6 and iOETbKAP6 constructs were linearized and transfected into 29-13 cells as described previously (23). Stable transfectants were selected with 2.5 μg/ml phleomycin and 1 μg/ml blasticidin and cloned by limiting dilution in SDM-79 containing 15% fetal bovine serum (FBS). The clonal RNAi and overexpression cell lines, named SLTbKAP6F2 and iOETbKAP6A10, were used for further analyses. The induction of RNAi or overexpression was carried out by adding 1 μg/ml tetracycline.

Expression of recombinant TbKAP6 in Escherichia coli and protein purification.

The TbKAP6 coding sequence minus the first 48 bp (encoding the mitochondrial targeting signal [MTS], predicted by Mitoprot II [http://ihg.gsf.de/ihg/mitoprot.html]) was amplified with the primers listed in Table S1 in the supplemental material and cloned into pET-28a(+) (Novagen) at NheI/XhoI sites. The resulting pET28a-TbKAP6 plasmid contains a six-His sequence in frame with the 3 terminus of TbKAP6. The plasmid was then transformed into E. coli Rosetta [BL21(DE3)] pLysS competent cells (Novagen). Transformants were cultured in LB medium containing 34 μg/ml chloramphenicol and 30 μg/ml kanamycin at 37°C to an optical density at 600 nm (OD₆₀₀) of 0.6. Protein expression was induced by adding 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG) to the medium and further incubating for 3 h before harvesting cells. Purification of recombinant TbKAP6 (rTbKAP6) was performed as previously described (25). The BCA method was then used to determine the concentration of rTbKAP6 by following the manufacturer's instructions (Thermo Scientific Pierce BCA protein assay kit).

C. fasciculata kDNA decatenation assays.

CfkDNA networks (200 ng) were incubated with 0 to 1 unit of human topoisomerase IIα (USB) in 10 mM Tris-HCl (pH 7.9), 50 mM NaCl, 50 mM KCl, 5 mM MgCl₂, 1 mM ATP, and 15 μg/ml bovine serum albumin (BSA) in the absence or presence of recombinant TbKAP6 (1.6 μM) in a final volume of 20 μl at 37°C for 30 min. The reactions were terminated by adding 4 μl DNA 6× sample loading buffer (Thermo Scientific). The decatenated products were analyzed on 1% agarose-EtBr gels.

Plasmid relaxation assay.

A total of 300 ng of supercoiled plasmids pJN6 (26) and pLew111 (http://tryps.rockefeller.edu/) were incubated with human topoisomerase IIα (0 to 0.4 units) in the presence or absence of rTbKAP6 (0 to 0.6 μM) at 37°C for 40 min in 10 mM Tris-HCl (pH 7.9), 50 mM NaCl, 50 mM KCl, 5 mM MgCl₂, 0.1 mM EDTA, 15 μg/ml BSA, and 1 mM ATP with a final volume of 20 μl. The reactions were terminated by adding 4 μl DNA 6× sample loading buffer (Thermo Scientific). The relaxation products were loaded onto 1.5% agarose gel in 1 Tris-acetate-EDTA (TAE) buffer. The DNA topoisomers were visualized by UV-illumination of an ethidium bromide-stained gel (1 μg/ml).

Condensation assays with recombinant TbKAP6.

kDNA networks were isolated from C. fasciculate cells as described previously (27). Networks (300 ng; ∼1 μM minicircles) were incubated with 0.04 to 4 μM recombinant TbKAP6, or 10 mM spermidine, in 25 mM Tris-HCl (pH 7.5), 15 mM KCl, 2 mM MgCl₂, and 1 mM dithiothreitol (DTT) with a final volume of 20 μl at room temperature for 5 min. The reaction mixtures were allowed to adhere to 8-well slides and stained with 5 μg/ml 4′,6-diamidino-2-phenylindole (DAPI) for 10 min before imaging by fluorescence microscopy.

Miscellaneous methods.

DNA and RNA preparation, Southern, Northern, and Western blotting (28), kDNA isolation and DAPI staining (27), (29), electron microscopy of isolated kDNA (30) and thin sections (20), and TdT labeling (20) were performed as described.

RESULTS

Characteristics of TbKAP6.

The coding sequence of TbKAP6 (GenBank accession number XP_823113) is 630 bp in length and encodes a 209-residue polypeptide with a predicted pI of 11.3. The first 16 residues at the N terminus are a potential mitochondrial targeting signal (MTS) with a probability of 91% predicted by Mitoprot II (http://ihg.gsf.de/ihg/mitoprot.html). TbKAP6 shares 46.48% identity with TcKAP6, 32.67% with LiKAP6, and 31.25% with CfKAP4 (see Fig. S1 in the supplemental material). We performed the domain annotation and three-dimensional (3D) modeling analyses at the SWISS-MODEL workspace (http://swissmodel.expasy.org/), which is a fully automated protein structure homology-modeling server. Our analyses showed that TbKAP6 contains two tandem HMG boxes, a degenerate short box A and a conserved box B (see Fig. S2A and B in the supplemental material), but lacks the acidic C-tail (the hatched box in Fig. S2A). The comparisons of 3D models of TbKAP6's two HMG boxes with the nuclear magnetic resonance (NMR) solution structure of rat HMGB1's HMG box B suggested that both HMG boxes of TbKAP6 form common helical structures (see Fig. S2C, left). However, TbKAP6's HMG box A lacks helix III, whereas its HMG box B has all three of the helixes found in HMGB1 (see Fig. S2C, right).

TbKAP6 localizes on the kinetoplast disk.

Since TbKAP6 has a predicted N-terminal mitochondrial targeting sequence (see above), we examined the localization of TbKAP6-Myc by C-terminal tagging of one endogenous allele of TbKAP6. Fluorescence microscopy revealed that TbKAP6-Myc is localized in the kinetoplast disk (Fig. 1) and appears throughout the cell cycle: 1N1k, 1N2k, and 2N2k cells (N and k stand for nucleus and kinetoplast, respectively). TbKAP6-Myc seemed to have stronger fluorescence when viewed on top of the disk (stars) than when viewed on the side (arrowheads), which was very similar to TcKAP6 (11). Inspection of some kinetoplasts suggested that since the DAPI signal (red) merged with Myc signal (green) only at the edge, TbKAP6-Myc might not be uniformly distributed throughout the kDNA disk but appears to be distributed on one face of the disk, at least in some stages of the cell cycle. Regardless, this conclusion awaits further study.

FIG 1.

TbKAP6-Myc localized on kDNA disk. Stars show kinetoplasts viewed from the top and arrowheads show kinetoplasts viewed from the edge. N, nucleus; k, kinetoplast DNA. In merged panels, DAPI is in red and anti-Myc is in green. Bar, 5 μm.

RNAi of TbKAP6 causes kDNA loss and arrests cell growth.

We conducted RNAi by inserting two opposing copies of a 500-bp fragment of TbKAP6 coding sequence into a stem-loop vector that produces double-stranded RNA (dsRNA) upon induction with tetracycline. This 500-bp coding sequence was checked by RNAit software and shows no homology with other T. brucei sequences that could cause RNAi depletion of other proteins. This construct (SLTbKAP6) was then linearized and transfected into 29-13 cells that constitutively express both T7 polymerase and tetracycline (Tet) repressor. We recovered three clonal cell lines (A3, F2, E12), and they displayed similar properties, such as kDNA loss in DAPI-stained cells. We used one of these three clonal cell lines (SLTbKAP6F2) for all the following analyses except for the analyses of free minicircle replication intermediates where we conducted parallel experiments on all three. The induction of TbKAP6 RNAi by addition of tetracycline caused >90% of TbKAP6 mRNA to be degraded after 24 h, and TbKAP6 transcript did not return during 5 days of RNAi due to recovery from RNAi (inset panel in Fig. 2A). The RNAi started to affect cell growth around day 4 and stopped it around day (Fig. 2A). DAPI staining of cells revealed that TbKAP6 RNAi caused kDNA shrinkage and loss in some cells (Fig. 2B). To further evaluate the significance of depletion of TbKAP6 on kDNA metabolism, we evaluated the shrinkage and loss of kinetoplast in more than 500 randomly chosen DAPI-stained cells at each time point (Fig. 2C). The kinetics of kDNA loss in Fig. 2C showed that the percentage of cells with normal-sized kDNA (“normal k”) dropped dramatically after 2 days of TbKAP6 RNAi. In the meantime, the percentage of cells with smaller-than-normal kDNA (“small k”) went up to 44% after 4 days of RNAi, and the percentage of cells without kDNA (“no k”) dramatically rose to 82% before day 6. By day 6, there were less than 5% of cells containing normal kDNA.

FIG 2.

TbKAP6 RNAi arrested cell growth and caused kDNA loss in some cells. (A) Effects of TbKAP6 RNAi on cell growth. The cumulative cell count on the y axis is the measured value times the dilution factor. Inset, Northern blot of TbKAP6 mRNA (630 bp) isolated from uninduced cells or cells induced for RNAi for 5 days. Tubulin mRNA is probed as the loading control. (B) Examples of kDNA morphology changes following TbKAP6 RNAi, shown by DAPI staining of uninduced cells (day 0) and cells induced for RNAi for 4 days (day 4). N, nucleus; k, kinetoplast DNA. (C) Kinetics of kDNA loss during a 6-day course of TbKAP6 RNAi. At each time point, at least 500 randomly chosen DAPI-stained cells were evaluated.

Although most cells have lost their kDNA by day 6 of RNAi, we note that cell division does not cease until after day 7. As we have reported earlier (4), this lag is likely due to the time needed to turn over preexisting kDNA-encoded proteins.

TbKAP6 RNAi causes kDNA disorganization and shrinkage.

We isolated kDNA networks from uninduced cells or cells undergoing TbKAP6 RNAi for 2, 4, and 6 days. DAPI staining of isolated networks (Fig. 3A) revealed the disorganization and shrinkage after TbKAP6 depletion. For kDNA networks isolated from cells uninduced for RNAi (day 0 in Fig. 3A), most were of unit size, round (stars), with elongating networks of intermediate size that were undergoing replication (arrowhead); networks that presumably had finished replication (arrow) are double in size and dumbbell-shaped. However, for those networks isolated from TbKAP6 RNAi cells (days 2, 4, and 6 in Fig. 3A), they became increasingly heterogeneous in size and fluorescence intensity and displayed various shapes. Meanwhile, compared to day 0, from day 2 there were fewer dumbbell-shaped networks, an indicator of replicating kDNA. Very tiny networks appeared from day 4 (maybe earlier than day 4, since we did not examine networks at day 3). At day 4, there were a variety of kDNA sizes and shapes. The surface area quantitation of DAPI-stained well-spread kDNA networks (≥250 networks at each time point) (Fig. 3B; see also Fig. S3A in the supplemental material) further confirmed the shrinkage of kDNA. Since spreading of kDNA networks on slides is not always ideal, we chose flat, well-dispersed networks to best measure surface area. In Fig. 3B, the relative value of average surface area decreased almost 50% by day 4. The distribution of percentages of kDNA networks within different surface area ranges (see Fig. S3A) exhibited a gradual shift from higher to lower surface area ranges and further confirmed that the networks shrank.

FIG 3.

TbKAP6 RNAi caused kDNA disorganization and shrinkage. (A) DAPI-stained kDNAs isolated from uninduced cells (day 0) and cells induced for RNAi for 2, 4, and 6 days. Star, kDNA of unit size; arrowhead, kDNA of intermediate size; arrow, kDNA of double size. Bar, 5 μm. (B) Kinetics of average surface area of DAPI-stained kDNA isolated from cells without or with RNAi. Relative value is the ratio of average surface area at each time point to that of day 0. At each time point, at least 250 well-spread kDNA networks were evaluated. (C) TdT labeling of kDNA isolated from uninduced cells (day 0) and cells induced for RNAi for 2, 4, and 6 days (days 2, 4, and 6). DAPI is in red and TdT labeling is in green. Bar, 5 μm. (D) Electron micrographs of kDNA isolated from uninduced cells (day 0) and cells induced for RNAi for 4 days (day 4). Bar, 500 nm. (E) Electron micrographs of thin sections of resin-embedded cells without (day 0) or with RNAi (day 4). Arrow, kinetoplast DNA; arrowhead, additional DNA fibers close to the kinetoplast; FP, flagellar pocket; FL, flagellum; BB, basal body. Bar, 500 nm.

The structural changes of the kDNA network are assessed by TdT labeling of isolated kDNA (Fig. 3C). As outlined in the introduction, during kDNA replication the progeny minicircles (nicked/gapped [N/G]) are attached to the network poles, and the last gap is not repaired until the network has finished replication and is undergoing segregation. TdT labeling can be used to visualize the distribution of these N/G minicircles on networks. For networks isolated from uninduced cells, antipodal labeling of intermediate-sized networks (star and arrowhead in Fig. 3C, day 0) and uniform labeling of double-sized dumbbell-shaped networks (arrow in Fig. 3C, day 0) are two characteristic TdT labeling patterns. While there is no significant change in the percentage of TdT labeling-positive kDNA networks (see Fig. S3B), the shift of TdT labeling patterns indicates RNAi-mediated structural changes on kDNA. As shown at day 2 (Fig. 3C), besides the conventional antipodal labeling pattern, a ring-labeling pattern (star) appeared, but it was not common in the total population. By day 4 or later, irregular labeling (arrows) of kDNA networks of different sizes increased to >70% of TdT-positive networks with uniform labeling of small networks (star) to more than 15% (days 4 and 6 in Fig. 3C), which indicated dramatic structural changes on kDNA networks after the depletion of TbKAP6.

Next we used electron microscopy (EM) to examine changes in the organization of kDNA networks isolated from TbKAP6-depleted cells. As shown in the upper panel of Fig. 3D, kDNA networks from uninduced cells (day 0) appeared as unit-sized and planar-shaped (left panel) or double-sized and dumbbell-shaped (right panel). The maxicircle loops (black arrows) were distributed around the planar network or were concentrated in the center of the dumbbell-shaped network. For the unit-sized kDNA networks isolated from uninduced cells, there is a thick line (white arrow) with much higher electron density at the periphery of the network. This line is usually less prominent in T. brucei networks, but it is common in networks of C. fasciculata. The lower panel of Fig. 3D presented the networks isolated from cells that had undergone RNAi for 4 days. These networks were smaller, were heterogeneous in size, and may have lost the planar shape. The electron-dense material is randomly distributed within the network.

We further examined these changes in the kinetoplast in situ by thin-section EM of uninduced cells and TbKAP6-depleted cells (RNAi for 4 days) (Fig. 3E). In uninduced cells (day 0, Fig. 3E), cross sections of the kinetoplast disk exhibited a characteristic striated, densely stained disk structure. As expected, the thickness of the disk is about half the circumference of a minicircle. They assumed a unit-sized disk shape (cell in left subpanel of day 0, Fig. 3E), an elongating disk shape (cell in the middle subpanel of in day 0, Fig. 3E), or a shallow V-configuration (cell in right subpanel of day 0, Fig. 3E). In contrast, after TbKAP6 RNAi for 4 days, most kinetoplasts lost the normal disk shape and became smaller (cells 1 to 4, day 4, Fig. 3E). The kinetoplast shrinkage seemed extreme in cells 3 and 4 with two basal bodies, which would in wild-type cells normally be found with double-sized networks. A few kinetoplasts had additional densely stained DNA fibers along with a unit-sized or elongated kDNA disk (cells 5 and 6). We will address these results in Discussion.

TbKAP6 RNAi decreases total minicircle, total maxicircle, and free minicircle replication intermediates.

To assess kDNA loss at the molecular level, we conducted Southern blotting of HindIII/XbaI-digested total DNA (these enzymes cleave all minicircles and maxicircles) isolated from uninduced cells or from three independent clonal TbKAP6-depleted cell lines as mentioned above. Upon RNAi of TbKAP6, the levels of total minicircle and total maxicircle underwent a similar decrease in three independent clonal cell lines, and Fig. 4A showed results from one of three clonal cell lines named SLTbKAP6F2. This gel was quantitated on a phosphorimager (Fuji Film BAS-2500 photo image scanner). Phosphorimaging values for each band were corrected by subtracting a background (in the nearby apparently blank area, the same size and shape as the band itself) and normalized for load. Backgrounds have little effect on strong bands but have a large effect on weak bands. Thus, the values for bands 6, 7, and 8 are most likely unreliable. Nonetheless, we measured total minicircle and total maxicircle levels in a representative clonal cell line and plotted results as shown in Fig. 4B. From Fig. 4B, during first 4 days of RNAi, total minicircle abundance declined ∼50% and total maxicircle abundance dropped ∼20%.

FIG 4.

TbKAP6 RNAi inhibited kDNA replication. (A) Effects on kDNA abundance. Total DNA (10⁶ cell equivalents/lane) isolated from cells without or with RNAi was digested with HindIII/XbaI, Southern blotted, and probed for minicircles (1.0-kb linearized minicircles), maxicircles (a 1.4-kb fragment), and nucleus-encoded trypanosome hexose transporter (THT) genes as the loading control (Load). (B) Changes of total minicircle and maxicircle abundance in a representative clonal cell line during the course of TbKAP6 RNAi. DNA abundance was measured by phosphorimaging of the Southern blot shown in panel A. Relative value is the ratio of total minicircle or maxicircle abundance to that of the loading control (THT). (C) Effects on free minicircle intermediates. Total DNA (10⁶ cell equivalents/lane) isolated from cells without or with RNAi was fractionated by agarose gel electrophoresis, Southern blotted, and probed for minicircles and THT for loading control. N/G, nicked/gapped minicircle; CC, covalently closed minicircle. (D) The abundances of N/G and CC minicircles decreased upon the depletion of TbKAP6. DNA abundances from Southern blots shown in panel C, which is a representative example of three independent clonal cell lines, and the other two blots (data not shown) were averaged to obtain the error bars and plotted. (E) Effects on maxicircle replication intermediates. Total DNA (10⁶ cell equivalents/lane) isolated from cells without or with RNAi was decatenated by topo IV, fractionated by agarose gel electrophoresis, Southern blotted, and probed for maxicircles and THT for loading control. N/G, nicked/gapped maxicircle; CC, covalently closed maxicircle; L., linear maxicircle.

Next we analyzed the changes in free minicircle replication intermediates by Southern blotting of total DNA isolated from uninduced cells and TbKAP6-depleted cells from three independent clonal cell lines as mentioned above. As described in the introduction, during kDNA replication, covalently closed (CC) minicircles and nicked/gapped (N/G) minicircles coexisted in the KFZ, and their abundances can be detected by Southern blotting of total undigested DNA, which contained all free minicircle replication intermediates. Figure 4C showed the immediate decrease of both CC and N/G minicircle abundances upon depletion of TbKAP6 in SLTbKAP6F2 cells. In the other two independent clonal cell lines (SLTbKAP6A3 and E12), we had similar results (data not shown). We then measured the abundances of CC and N/G minicircles in these three independent clonal cell lines and plotted the results in Fig. 4D. From Fig. 4D, the abundance of CC minicircles quickly declined nearly 3 times faster than that of N/G minicircles between day 0 and day 1 (at day 1, the abundance of CC and N/G minicircles declined 38% and 13%, respectively). Their abundances continued to decrease at similar rates, with that of N/G minicircles declining slightly faster between day 2 and day 5.

We further studied the effects of TbKAP6 RNAi on the abundance of CC maxicircles and N/G maxicircles. In wild-type cells, CC maxicircles are replication precursors and N/G maxicircles are replication products. We assumed the same is true in TbKAP6 RNAi cells. First, we isolated total DNA (including kDNA) from uninduced cells and TbKAP6-depleted cells. Then we used topoisomerase IV to treat total DNA so that kDNA networks can be completely decatenated and covalently closed or nicked/gapped maxicircles can both be resolved on an agarose-EtBr gel and detected by Southern blotting. Figure 4E demonstrated that TbKAP6 RNAi also affected maxicircle replication. As shown in Fig. 4F, upon TbKAP6 RNAi, N/G maxicircle abundance dramatically declined from day 5 and nearly disappeared at day 8.

TbKAP6 overexpression causes cell death and kDNA segregation defects.

To overexpress TbKAP6, we inserted the entire coding sequence of TbKAP6 into pLEW100V5BSR vector (24). The final construct iOETbKAP6 was linearized and transfected into 29-13 cells. We used one of three clonal cell lines (TbKAP6OE-A10) with similar properties for all the following experiments. Upon the overexpression of TbKAP6, the TbKAP6 mRNA (630 bp) level increased at least 6-fold (inset panel in Fig. 5A). TbKAP6 overexpression slowed cell growth between day 2 and day 3 and nearly stopped it around day 4 (Fig. 5A). DAPI-stained cells revealed that TbKAP6 overexpression caused kDNA loss in some cells (Fig. 5B). To further assess the significance of TbKAP6 overexpression on kDNA loss, we assessed ∼400 to 600 randomly chosen DAPI-stained cells at each time point, and the results were plotted in Fig. 5C. In Fig. 5C, the percentage of “normal k” cells declined from the very beginning of TbKAP6 overexpression. In contrast, the percentages of “small k” and “no k” cells increased to 24% and 31% at day 6, respectively.

FIG 5.

Overexpression of TbKAP6 arrested cell growth and caused kDNA loss and segregation defects in some cells. (A) Effects on cell growth. The cumulative cell count on the y axis is the measured value times the dilution factor. Inset, Northern blot of TbKAP6 mRNA isolated from uninduced cells or cells induced for overexpression for 1 day and 2 days. Tubulin mRNA is probed as the loading control. (B) Effects on kDNA morphology changes following overexpression, shown by DAPI staining of uninduced cells (day 0) and cells induced for overexpression for 2, 4, and 6 days. N, nucleus; k, kinetoplast DNA. Star, kinetoplasts that underwent asymmetric segregation. Bar, 5 μm. (C) Kinetics of kDNA loss during a 6-day course of overexpression. At each time point, 400 to 600 randomly chosen DAPI-stained cells were evaluated. (D) Effects on kDNA segregation, shown by DAPI staining of uninduced cells (day 0) and cells induced for overexpression for 2 and 4 days. Arrows, the larger daughter kinetoplast from asymmetric segregation; arrowheads, the smaller daughter kinetoplast from asymmetric segregation. Bar, 5 μm. (E) Bar graphs showing increasing kDNA segregation defects during a 6-day course of overexpression. At different time points, 40 to 100 2K cells were analyzed. White bar, cells with normal symmetric segregation of sister kDNAs; black bar, cells with kDNA segregation defects.

The examination of DAPI-stained cells revealed TbKAP6 overexpression caused kDNA segregation defects (Fig. 5B and D). In uninduced cells, kDNA replication and segregation coordinate with the nuclear DNA replication, mitosis, and cytokinesis in a spatial and temporal manner (day 0, Fig. 5D). However, in TbKAP6-overexpressed cells (days 2 and 4, Fig. 5D), various kDNA segregation defects appear as early as day 2, when the cell growth was affected. The segregation defects included asymmetric division (left subpanels at days 2 and 4, Fig. 5D), two nascent kinetoplasts with a thread (middle subpanels at days 2 and 4, Fig. 5D), and a large kDNA, which cannot undergo segregation after replication (right subpanels at days 2 and 4, Fig. 5D). Figure 5E showed that the percentage of cells with kDNA segregation defects quickly jumped to 61% at day 2 and reached 70% by day 6.

TbKAP6 overexpression causes kDNA shrinkage and disorganization.

We examined the effects of TbKAP6 overexpression on kDNA networks in vitro and in vivo (Fig. 6). Compared to kDNA networks isolated from uninduced cells (day 0, Fig. 6A), networks isolated from TbKAP6-overexpressed cells (days 2, 4, and 6, Fig. 6A) became heterogeneous in shape, size, and fluorescence intensity. As early as day 2, there were small networks which may result from the asymmetric division. Meanwhile, almost 50% of networks (arrowheads, at days 2, 4, and 6, Fig. 6A) or parts of networks have lower fluorescence intensity, which meant that there was less DNA in the entire disk or in some regions. Very interestingly, some networks with higher fluorescence intensity also appeared from day 2. These large-size networks were either the larger daughter networks of asymmetric division or those giant kDNA networks which failed to divide in vivo (right subpanels at days 2 and 4, Fig. 5D). The surface area quantitation of DAPI-stained well-spread kDNA networks (≥250 networks at each time point) also confirmed the shrinkage of kDNA (Fig. 6B; see also Fig. S4A in the supplemental material). The relative value of average surface area decreased by 40% at day 2 and 50% at day 4 as shown in Fig. 6B. We plotted the distribution of percentages of kDNA networks within different surface area ranges at different time points (see Fig. S4A). The dramatic shift from higher to lower surface area ranges further confirmed the shrinkage of kDNA networks following TbKAP6 overexpression. Even at day 2, the percentage of networks with surface areas in the range of 0.5 to 1 μm² went up to 56% from 6% on day 0 and networks with surface areas less than 0.5 μm² appeared and reached 9.6%. By day 4, the percentage of networks with surface area less than 0.5 μm² dramatically rose to ∼30%; that of networks with surface areas between 1 and 1.5 μm² declined to 15% from 55% on day 0.

FIG 6.

Overexpression of TbKAP6 caused kDNA disorganization and shrinkage. (A) Examples of DAPI-stained kDNAs isolated from uninduced cells (day 0) and cells induced for overexpression for 2, 4, and 6 days. Star, kDNA of unit size; arrowhead in the panel for day 0, kDNA of intermediate size; arrow, kDNA of double size. Arrowheads in panels for days 2, 4, and 6, kDNA with lower fluorescent density. (B) Kinetics of average surface area of DAPI-stained kDNA isolated from cells without or with overexpression. Relative value is the ratio of average surface area at each time point to that of day 0. At each time point, 400 to 600 well-spread kDNA networks were quantified. (C) TdT labeling of kDNA isolated from uninduced cells (day 0) and cells induced for overexpression for 2, 4, and 6 days. DAPI is in red and TdT labeling is in green. (D to H) Electron micrographs of thin sections of resin-embedded cells without (day 0; D) or with overexpression (day 4; E to H). FP, flagellar pocket; FL, flagellum; BB, basal body; white arrows, kDNA. For all electron micrographs (D to H), bar = 500 nm.

TdT labeling of isolated kDNA from TbKAP6-overexpressed cells revealed the disorganization of networks (Fig. 6C) and a slight increase (∼9% by day 4) in the percentage of TdT-labeling-positive kDNA networks (see Fig. S4B in the supplemental material). As mentioned above, the antipodal labeling and dumbbell-shaped labeling are two typical TdT labeling patterns in wild-type/uninduced cells or isolated networks (day 0, Fig. 6C). Upon the overexpression of TbKAP6, two typical patterns were replaced by irregular labeling patterns, such as ring labeling (star at day 4, Fig. 6C), uniform labeling of unit-sized or smaller networks (arrowheads at days 4 and 6, Fig. 6C), or patterns difficult to classify (arrows at day 4, Fig. 6C). In fact, some networks (arrows at day 4, Fig. 6C) have very low DAPI fluorescence (data not shown) but have similar or higher fluorescence intensity than that of day 0, which indicated much fewer DNA fibers, and these DNA fibers were nicked/gapped.

We also evaluated the effects of TbKAP6 overexpression on the kinetoplast in vivo by electron microscopy of thin sections of uninduced cells and TbKAP6-overexpressed cells (Fig. 6D to G). Figure 6D showed uninduced cells with a unit-sized kinetoplasts in disk shape (left subpanel in Fig. 6D), an elongating kinetoplast with a V configuration (middle subpanel in Fig. 6D), and two nascent segregated kinetoplasts of unit size (right panel in Fig. 6D). After TbKAP6 overexpression for 4 days, some kinetoplasts shrunk to tiny spots but with an electron density similar to that of wild-type cells (Fig. 6E). Surprisingly, some cells (cells 13 and 14, Fig. 6E) revealed more than two densely stained spots. Figure 6F showed some small kinetoplasts lost their disk shape (cells 15 to 17, Fig. 6F) and some had heterogeneous electron densities in two nascent kDNA disks (cell 15, Fig. 6F). Kinetoplasts in Fig. 6G are roundish (cells 18 and 19, 22 to 24) or disk-like (cells 20 and 21) but have higher electron density. In cells 22 to 24 (Fig. 6G), kinetoplasts seemed to have undergone asymmetric division. Note two nascent kinetoplasts in cell 22 (Fig. 6G) have different electron density. In Fig. 6H, elongating or replicated kinetoplasts were distorted and could not undergo segregation.

TbKAP6 overexpression promotes the release of covalently closed minicircles.

Since TbKAP6 RNAi inhibited the release of minicircle and eventually led to kDNA loss, we evaluated whether TbKAP6 overexpression may have opposite effects. We analyzed the effects of TbKAP6 overexpression on free minicircle replication intermediates by Southern blotting of total DNA isolated from cells with or without TbKAP6 overexpression (Fig. 7). As shown in Fig. 7A and B, the levels of both CC minicircle and N/G minicircle dramatically increased within 24 h of overexpression. By day 1, their abundances have risen about ∼2.5-fold and by day 5 their abundances have increased 3-fold. After day 5, their abundances gradually declined but were still higher than those of uninduced cells.

FIG 7.

Effects of TbKAP6 overexpression on free minicircle intermediates. (A) Effects on free minicircle intermediates. Total DNA (10⁶ cell equivalents/lane) isolated from cells without or with overexpression was fractionated by agarose gel electrophoresis, Southern blotted, and probed for minicircles and THT for loading control. N/G, nicked/gapped minicircle; CC, covalently closed minicircle. (B) Changes of the abundances of N/G and CC minicircles during the course of TbKAP6 overexpression. DNA abundances were quantified by phosphorimaging of the Southern blot shown in panel A. Relative value is the ratio of N/G or CC minicircle abundance to that of the loading control (THT).

rTbKAP6 can promote topoisomerase II-mediated decatenation of the kDNA network.

As mentioned in the introduction, at the beginning of kDNA replication, CC minicircles are released from the networks by a type II topoisomerase (topo II). The Southern blots in Fig. 4D and Fig. 7 suggested that TbKAP6 plays a role in the release of CC minicircle in vivo. Moreover, it has been shown that HMGB1 can stimulate the topo IIα-mediated decatenation of CfkDNA in vitro (31). These findings prompted us to determine whether TbKAP6 could also promote the decatenation activity of topo II in vitro. Unfortunately, T. brucei mitochondrial topo II is not available, so we used commercially available human topo IIα. To investigate the effect of recombinant TbKAP6 (rTbKAP6; see Fig. S5A in the supplemental material) on topo II-mediated kDNA decatenation, we added increasing amounts of topo II to CfkDNA solutions in the absence or presence of rTbKAP6 (1.6 μM), and the reaction products were analyzed by agarose-EtBr gel electrophoresis (Fig. 8A). As shown in Fig. 8A, in the absence of rTbKAP6, decatenation products, including CC and N/G minicircles, were not visible until 0.2 unit topo II was used. Without rTbKAP6, kDNA (300 ng) can be fully decatenated (no detectable kDNA left in the gel slot) until the amount of topo II increased to one unit. In contrast, in the presence of rTbKAP6, 0.05 unit of topo II can lead to the detectable amount of CC and N/G minicircles migrating into the gel. Moreover, the same amount of kDNA (300 ng) was completely decatenated with only 0.2 unit topo II.

FIG 8.

Recombinant TbKAP6 (rTbKAP6) promotes decatenation of C. fasciculata kinetoplast DNA by human topoisomerase II (topo II) and relaxation of supercoiled plasmid DNAs. (A) The decatenation of CfkDNA by topo II is stimulated by rTbKAP6. CfkDNA (200 ng) was incubated with increasing amounts of topo II (0.05 to 1 unit) in the presence (lanes 1 to 5) and absence (lanes 6 to 10) of rTbKAP6. The reaction mixtures were then examined by agarose-EtBr gel electrophoresis. N/G, nicked/gapped minicircle; CC, covalently closed minicircle. M, 1 kb plus DNA ladder (Invitrogen). (B) The relaxation of supercoiled plasmid DNAs by topo II is stimulated by rTbKAP6. Supercoiled plasmids (350 ng), including a Trypanosoma equiperdum minicircle-containing plasmid pJN6 (lanes 1 to 12) and a nonrelated plasmid pLew111 (lanes 13 to 19), were incubated with 0.1 to 0.4 unit of topo II in the presence of 0.3 to 0.6 μM TbKAP6 or absence of rTbKAP6 (lanes 4, 7, 10) at 37°C for 45 min. As controls, the plasmids incubated only with rTbKAP6 in the same topo II relaxation reaction buffer (lanes 2, 3, 14) as described above or without incubation (lanes 1 and 13) were loaded for comparison.

rTbKAP6 can promote topoisomerase II-mediated relaxation of supercoiled plasmids.

We then assayed the effect of rTbKAP6 on topo II relaxation of supercoiled plasmids. As shown in Fig. 8B, the incubation of rTbKAP6 alone (lanes 2, 3, 14) with supercoiled plasmid DNA has no effect on their relaxation. When increasing amounts of topo II alone were incubated (lanes 4, 7, 10, 15) with negatively supercoiled plasmid DNAs, DNA was relaxed proportionally to the amount of topo II. Then, when rTbKAP6 was present (lanes 5, 6, 8, 9, 11, 12, 16 to 19), the relaxation by topo II was more efficient, even leading to the complete relaxation of the same amount of both plasmids (lanes 6, 8, 9, 11, 12, 16 to 19). These results showed that rTbKAP6 could stimulate the relaxation activity of topo II.

rTbKAP6 can nonspecifically compact DNA in vitro.

Some CfKAPs can condense kDNA in vitro (8). We evaluated rTbKAP6 (see Fig. S5A in the supplemental material) condensation effects on isolated kDNA networks (see Fig. S5). The upper panel of Fig. S5B represented intact kDNA networks isolated from C. fasciculata cells. Most DAPI-stained networks exhibited uniform fluorescence intensity. The nonreplicative kDNAs are of unit size (arrow), with replicating kDNAs of intermediate size (star) and double size (arrowhead). We first incubated C. fasciculata networks with 2 mM spermidine (lower panel in Fig. S5B). DAPI staining showed that those networks became condensed and their average surface areas declined to one-third to one-fifth of those of intact networks. The condensation of networks by spermidine led to a higher DAPI fluorescence intensity than that of normal networks.

We then used the purified rTbKAP6 to assess its packaging effects on C. fasciculata networks (see Fig. S5C in the supplemental material). Using the same reaction condition for spermidine, increasing amounts of rTbKAP6 progressively compacted C. fasciculata networks. The incubation of 40 nM rTbKAP6 with networks (see panel 1 in Fig. S5C) had small effects on some networks. Some were slightly smaller and had higher fluorescence intensity (arrowheads); some had one or two bright spots within the network which could be nucleation points for condensation (stars). These effects became much stronger when rTbKAP6 increased to 400 nM (see panel 2 in Fig. S5C). The number of bright spots within one network (star) increased. More interestingly, once the concentration of rTbKAP6 rose to 1 μM (see panel 3 in Fig. S5C), those bright spots became bigger and brighter (star) following the dramatic decrease of fluorescence intensity of other regions of the network. Upon incubation with 2 μM rTbKAP6 (see panel 4 in Fig. S5C), most networks were compacted into bright roundish spots, much smaller than those condensed by spermidine (see the lower panel in Fig. S5B). Panel 5 in Fig. S5C showed that almost all networks were condensed as tiny bright dots in the presence of 4 μM rTbKAP6.

DISCUSSION

It has been estimated that more than 100 proteins are required for trypanosomatid kDNA replication and maintenance (6). Small basic histone H1-like proteins from C. fasciculata, CfKAP2 to CfKAP4, can condense kDNA in vitro and rescue an E. coli HU mutant with a defect in chromosome condensation and segregation (8). These results suggested that CfKAPs play a role in kDNA organization and maintenance. The BLASTp search using CfKAP protein sequences as the query within the genome databases of 5 trypanosomatid species (T. cruzi, T. brucei, and three Leishmania species) led to the identification of 35 proteins related to CfKAPs (11), including five T. brucei KAPs: TbKAP3, TbKAP4.1 and TbKAP4.2 (different by only one amino acid), TbKAP6, and TbKAP7. Our domain/motif analyses of these five TbKAPs showed that TbKAP4.1, TbKAP4.2, TbKAP6, and TbKAP7 are HMG box-containing proteins.

Even though CfKAP1 to CfKAP4 (8, 9), TcKAP3, TcKAP4, and TcKAP6 (11, 32) were localized to kinetoplast in vivo, none of them has been shown to be essential for cell viability, and knockout studies of CfKAP2, CfKAP3, and TcKAP3 barely resulted in changes in kDNA morphology. In contrast, TbKAP6 in this report is the first essential trypanosomatid KAP protein of which the depletion by RNAi arrested cell growth and caused the disorganization, shrinkage, and loss of kDNA.

The functional exploration of TbKAP6 in vivo led to the most interesting finding in this report: TbKAP6 RNAi primarily and immediately inhibited the release of CC minicircles. Within 1 day after induction of TbKAP6 RNAi, the abundance of N/G minicircles dropped 13%, but the abundance of CC free minicircles declined almost 40% (Fig. 4C and D). This could imply that TbKAP6 plays a role in minicircle release. Since N/G minicircles are the replication progeny from precursor CC minicircles, once the CC minicircle abundance declined, the N/G minicircle abundance also decreased, which eventually eliminates minicircle replication and contributes to kDNA shrinkage and loss. Consistent with these results, overexpression of TbKAP6 elevates the release of CC minicircles, increasing almost 2.5-fold within 24 h of overexpression. This result is also consistent with TbKAP6 being required for minicircle release; it must control the rate-limiting step in minicircle release. The overrelease of CC minicircle gradually led to kDNA shrinkage or loss shown by DAPI staining of cells (Fig. 5B) and isolated kDNA networks (Fig. 6A) and EM of thin sections (Fig. 6D to G). The lower DAPI fluorescence intensity of more than 50% isolated networks indicates a reduction of DNA fibers in networks. The TdT labeling (Fig. 6C) further confirmed that the residual minicircles within these low-DAPI-fluorescence-intensity networks are largely nicked/gapped minicircles since they have similar or even stronger TdT fluorescence intensity than those of kDNAs isolated from uninduced cells. The RNAi and overexpression studies together suggest that TbKAP6 may control minicircle release. Low levels of TbKAP6 following RNAi cause very low release of CC minicircles, wild-type levels maintain normal release, and overexpression causes 2.5- to 3-fold-higher release than normal.

A less likely alternative interpretation of the data in the previous paragraph is that the major effects are on minicircle reattachment. RNAi, instead of blocking minicircle release, could accelerate minicircle reattachment. However, the immediate deficiency is in CC minicircles, but it is N/G minicircles that are reattached. Similarly, overexpression, instead of stimulating minicircle release, could slow down minicircle reattachment. This model is inconsistent with simultaneous increase in both CC and N/G minicircles upon overexpression.

Since TbKAP6 seems to stimulate minicircle release, we wondered how this occurs. The trypanosome genome encodes two type II topoisomerases, one nuclear enzyme (33) and one mitochondrial enzyme (12, 13). It is likely that the mitochondrial topo II (mtTopo II) is responsible for minicircle release. However, RNAi studies have always been competent for minicircle release but deficient in minicircle reattachment and repair of holes in networks caused by minicircle release (34). They are also deficient in decatenating multiply interlocked minicircles (25). Since the genome encodes only one mitochondrial topo II, it is possible that residual topo II following RNAi is adequate for minicircle release, or it could be a noncanonical enzyme that does the job.

Since our data suggested that TbKAP6 may stimulate minicircle release, we wanted to test whether in vitro TbKAP6 can stimulate topo II decatenation activity. There is a precedent in that human HMGB1 can stimulate the topo II-mediated decatenation of CfkDNA in vitro (15, 31). Since T. brucei mtTopo II is not available, we used human topo II. We found that in in vitro assays, rTbKAP6 can stimulate human topo II-mediated decatenation of the kDNA network about 5-fold (Fig. 8A). It seems likely that either TbKAP6 directly stimulated human topo II decatenation activity, or it facilitated the kDNA network accessibility for human topo II, or possibly both. However, it still remains unknown whether in vivo TbKAP6 can stimulate T. brucei mtTopo II decatenation activity and how TbKAP6 can stimulate topo II decatenation activity or facilitate kDNA accessibility for mtTopo II. We will further discuss this below.

Recent in vitro studies showed that CfKAP3 alone condensed the CfkDNA network and inhibited its decatenation by topo II. However, protein-protein interactions between CfKAP3 and CfUMSBP can decondense these networks and restore topo II-mediated decatenation (14). This indicates that trypanosomatid KAP proteins could have different roles in different species, especially when they have limited homology, such as between CfKAP3 and TbKAP6. In T. brucei, in vivo TbKAP6 promotes minicircle release from networks and in vitro TbKAP6 stimulates topo II-mediated decatenation. However, CfKAP3 inhibits decatenation in vitro.

Our in vitro studies also showed that rTbKAP6 can condense isolated kDNA into a size comparable to that in vivo, similar to CfKAP2 to CfKAP4. This suggested the architectural role of TbKAP6 in kDNA organization and condensation, which can also be implied from TbKAP6-Myc kDNA disk localization (Fig. 1). Upon the depletion or overexpression of TbKAP6, isolated kDNA networks from induced cells become heterogeneous in shape and size, shown by DAPI staining. These kDNAs also lost typical antipodal or dumbbell-shaped TdT labeling, which means their structures are disrupted. These architectural changes are further confirmed by EM thin sections of TbKAP6-depleted or -overexpressed cells, in which the kinetoplasts either lose their normal disk shape or become heterogeneous in electron density. Interestingly, there are additional DNA fibers along unit-sized or double-sized kinetoplasts from RNAi cells (cells 5 and 6 in Fig. 3E), which indicated that TbKAP6 also played structural roles in kDNA packaging/condensation.

As stated above, as a structural component, TbKAP6 appears to be always associated with kDNA. Meanwhile, it is required for covalently closed minicircle release during kDNA replication. This raises a question: if TbKAP6 is always present at the kDNA disk, why is KAP6 not constantly stimulating the minicircle release, which only happens during kDNA S phase? One explanation is that the minicircle release needs the decatenation activity of mtTopo II, and mtTopo II expression is regulated and periodically expressed at the boundary of G₁/S phase (35). Therefore, TbKAP6 could stimulate minicircle release only when mtTopo II is available. Meanwhile, the distribution of TbKAP6 within the kinetoplast might also be dynamically regulated during the cell cycle, similar to that of TcKAP4 and TcKAP6 (11). The ultrastructural immunolocalization analyses showed that both TcKAP4 and TcKAP6 redistributed corresponding to the kDNA rearrangement during the T. cruzi differentiation process. They dispersed through the disk-shaped kDNA network and then redistributed mainly to the kDNA periphery in intermediate and rounded kinetoplasts. Even though the ultrastructural localization of TbKAP6 needs be further characterized, it is possible that TbKAP6 stimulates minicircle release during kDNA S phase through direct binding to minicircles, but then it is insulated from the kDNA disk and situated at the periphery through the replacement by other TbKAPs once kDNA replication ends.

As mitochondrial tandem HMG box-containing proteins, it is interesting that TbKAP6 seems to be an evolutionary intermediate between yeast Abf2 and mammalian TFAM in terms of their functions. Abf2 plays an essential role in yeast mitochondrial nucleoid DNA packaging and has no role in mtDNA replication or transcription. The abundance of Abf2 in nucleoids determines the degree of compaction and the overall structure of nucleoids (36, 37). Here, we report that TbKAP6 is essential for both kDNA replication and kDNA packaging/maintenance. Compared to Abf2 and TbKAP6, mammalian TFAM has more essential roles in mitochondrial nucleoid: mtDNA nucleoid packaging, mtDNA replication or genome copy number, and transcription initiation (19, 38). It has been shown that a charged C-terminal tail of TFAM (39), which is absent from Abf2 or TbKAP6, is critical for activation of promoter-specific mtDNA transcription (40). More recently, X-ray crystallography studies have shown that TFAM imposes a U-turn on mitochondrial promoter DNA, and the strong bending of DNA depends on the cooperative binding of both HMG boxes and the intervening linker (38), which serves as a molecular mechanism for its role in mitochondrial nucleoid packaging and mtDNA transcription. Besides binding to DNA, the HMG box is also the binding region for proteins and involved in a variety of protein-protein interactions (15, 41). Through HMG box-mediated protein-protein interactions, HMGB proteins have been shown to modulate chromatin structural changes by interacting with chromatin remodeling factors (multiprotein complexes which can couple ATP hydrolysis to alter chromatin structure and displacement of histone H1), which resulted in the unlocking of nucleosome (nucleosome sliding), promoting DNA bending, and stimulation of the activity and nucleosome binding of chromatin remodeling complexes (42, 43). Since TbKAP6 also has two HMG boxes and its box B is structurally conserved to that of mammalian HMGB1, it is conceivable that TbKAP6 or its HMG boxes might play a role in modulating the accessibility of kinetoplast in vivo in a similar mechanism: through HMG box-mediated DNA binding and protein-protein interactions, TbKAP6 (and its unknown binding partners) binds to kinetoplast and displaces other kinetoplast packaging proteins to transiently loosen the condensed disk and promote the accessibility of the kDNA disk for mtTopo II.

Taken together, we have demonstrated that TbKAP6 is the first trypanosomatid KAP protein shown to be essential, while it shares classic features with CfKAPs and TcKAPs, such as low molecular weight, basic nature, association with kinetoplast, the ability to compact kDNA networks in vitro, and a role in kDNA maintenance. We also showed that TbKAP6 is involved in kDNA replication, possibly through its interactions with topo II, to control the minicircle release in a spatial and temporal manner which needs further characterization. Furthermore, it remains unknown what are the roles of its two HMG boxes.