African trypanosomes have a remarkable mitochondrial DNA termed kDNA (kinetoplast DNA) that contains several thousands of topologically interlocked DNA rings. Because of its highly unusual structure, kDNA has a complex replication mechanism. Our approach to understanding this mechanism is to identify the proteins involved and to characterize their function. So far approx. 30 candidate proteins have been discovered and we predict that there are over 100. To identify genes for more kDNA replication proteins, we are using an RNA interference library, which is the first forward genetic approach used for these parasites.
Trypanosoma brucei is the protozoan parasite that causes sleeping sickness. Being among the earliest branching eukaryotes, trypanosomes have unusual biological properties and one of their most remarkable features is a mitochondrial DNA known as kDNA (kinetoplast DNA) (see [1,2] for recent reviews). kDNA is a giant network of topologically interlocked DNA rings and its topology resembles that of chain mail in medieval armour (see Figure 1 for electron micrograph of an isolated kDNA network). Within the cell, the network is condensed into a disc-shaped structure located within the mitochondrial matrix adjacent to the flagellar basal body. The T. brucei network contains several thousands of minicircles (each 1 kb) and a few dozen maxicircles (each 23 kb). Maxicircles, like mitochondrial DNAs of higher organisms, encode rRNAs and subunits of respiratory complexes. Maxicircle transcripts are cryptic and must be edited to form functional mRNAs. Editing is an amazing process involving addition or deletion of uridylate residues at specific sites within the transcript to form an open reading frame. The templates for editing are small guide RNAs that are encoded by minicircles. Many guide RNAs are needed and therefore the kDNA network contains a large repertoire of different minicircle species, each encoding different guide RNAs. For a recent review on editing, see .
Electron micrograph of a segment of a kDNA network from
kDNA network replication
During kDNA synthesis each minicircle and maxicircle replicates once, forming a network that is double in size. This structure then divides into two unit-sized progeny networks that segregate into the two daughter cells during cytokinesis. A critical feature of kDNA replication, which could account for much of its complexity, must be a mechanism ensuring that each daughter network receives a full minicircle repertoire. Otherwise, essential guide RNAs will be lost and the cells will die. For reviews on kDNA replication, see [1,2].
Figure 2 shows the current model for kDNA organization and minicircle replication. The network is condensed into a disc with its covalently closed minicircles stretched parallel to the disc's vertical axis. Remarkably, the kDNA is surrounded by its own replication machinery and a variety of replication proteins are assembled in distinct positions within and around the kDNA disc. The first step in kDNA replication is the topoisomerase-mediated vectorial release of minicircles into the KFZ (kinetoflagellar zone), the region between the kDNA disc and the mitochondrial membrane nearest to the flagellar basal body . These free minicircles, that are covalently closed, then bind proteins localized within the KFZ and initiate replication as theta structures. The proteins include UMSBP (universal minicircle sequence-binding protein) (the replication origin binding protein) , primase , two DNA polymerases  and other proteins (such as single-strand-binding protein and a helicase) that are not yet characterized. The sister minicircles probably segregate in the KFZ. They then migrate to the antipodal sites, two protein assemblies that flank the kDNA disc 180° apart, where later steps in minicircle replication occur. These include removal of a primer by a structure-specific endonuclease-I  and repair of many (but not all) of the minicircle gaps by a DNA polymerase β  and a DNA ligase . These three enzymes, as well as a topoisomerase II , are specifically positioned in the antipodal sites. The function of this topoisomerase is to reattach the newly replicated minicircles, still containing at least one gap, back on to the network periphery .
The kDNA replication model showing the location of replication proteins and minicircle replication intermediates
For every minicircle released from the network for the purpose of replication, two progeny are reattached and thus the network grows in size. As replication proceeds, there is a shrinking of the network's central zone (consisting of covalently closed minicircles that have not yet replicated) and an enlargement of the two peripheral zones adjacent to the antipodal sites (containing newly replicated minicircles that are gapped). Finally, when all minicircles have replicated, the double-size network (which by now contains only gapped minicircles) splits in two. Then all the minicircle gaps are repaired, presumably by a DNA polymerase β and a DNA ligase [9,10]. These enzymes, both localized within the kDNA disc, are distinct from their counterparts in the antipodal sites. The progeny networks, each identical to the parent, then segregate into the daughter cells during cytokinesis.
As mentioned above, a key requirement of the replication mechanism is the maintenance of the minicircle repertoire. There are two features of this model that may be involved in this process. One is that minicircles are covalently closed before replication and gapped afterwards. Thus the gaps distinguish minicircles that have already replicated and provide a mechanism for ensuring that each minicircle replicates once and only once per generation. Furthermore, if the sister minicircles segregate in the KFZ and migrate to different antipodal sites, then they will almost certainly segregate into different progeny networks. Therefore the full complement of minicircles will be preserved.
Although Figure 2 presents a coherent replication mechanism for kDNA minicircles, there is still much to be learned. For example, we do not even know all the proteins functioning at the replication fork of a free minicircle and we know almost nothing about proteins involved in maxicircle replication. We have no candidate for the topoisomerase II that releases minicircles into the KFZ. We know little about factors controlling the timing of kDNA replication in the cell cycle (nearly concurrent with the nuclear S phase) . We know nothing about the mechanisms governing the relative movement between the kDNA disc and the antipodal sites . We do not know how the progeny minicircles move from the KFZ to the antipodal sites; the distance seems too large to depend on free diffusion. We do not know what controls the accurate scission of the double-size network into two unit-sized progeny and the details of how the progeny distribute into the two daughter cells .
How do we answer these questions?
The best approach is to discover new proteins involved in network replication, localize them around the kDNA disc and determine their biochemical function. The model in Figure 2 lists ten proteins and we know approx. 20 more that are in various stages of characterization in our laboratory and in other laboratories. Based on the complexity of the replication mechanism there may well be another 100.
How does one discover new proteins? The first kDNA replication proteins were identified by enzyme purification. More recently, a genomic approach has proven fruitful in identifying trypanosome proteins that are similar to replication proteins in other systems. Both of these approaches require a clear picture of what you are looking for. Unfortunately, many steps in kDNA replication have no counterpart in other systems, and therefore it is impossible to predict what proteins may be involved. To overcome this obstacle, we developed a forward genetic approach, an RNAi (RNA interference) library.
In 1998, came the first report of RNAi in T. brucei . Subsequently our laboratory developed a convenient RNAi vector, pZJM, which in most cases mediates specific knockdown of expression of T. brucei genes (see map in Figure 3) . In this vector, a fragment of the gene of interest (shown by the hatched region) is inserted between two head-to-head T7 promoters, each regulated by a tetracycline operator. The vector is linearized by NotI and then transfected into transgenic trypanosomes that constitutively express T7 RNA polymerase and the tetracycline repressor; it integrates by homologous recombination into the non-transcribed rDNA spacer . Synthesis of double-stranded RNA and subsequent RNAi is induced by the addition of tetracycline to the culture medium. pZJM has been used effectively in functional analysis of kDNA replication proteins (see e.g. [10,12]).
The pZJM RNAi vector
Identification of kDNA replication proteins using an RNAi library
In our first uses of the library, we introduced random genomic fragments (averaging 0.7 kb) into pZJM and then transfected the recombinant vectors into T. brucei . We prepared enough transfectants to cover the genome several times. Our plan was to induce RNAi with tetracycline and then clone cells with the desired phenotype; we identified the responsible gene by PCR of the pZJM insert and by sequencing. In one case, we screened for cells whose surface glycoproteins had lost the ability to bind the lectin concanavalin A . In another study, we selected cells resistant to tubercidin, a toxic adenosine analogue .
More recently, we have used the library to discover new genes for kDNA replication proteins. Since we had previously found that RNAi of several replication enzymes resulted in shrinking and loss of the network (see e.g. ), we decided to screen for RNAi-induced loss of kDNA. The problem is that kDNA loss results in cell death, making it impossible to clone and study these cells. We overcame this obstacle by cloning cells from the library before screening. Our current strategy is to clone parasites from the library by limiting dilution in 96-well plates, induce RNAi in each clone and then after 7 days assay for kDNA loss by microscopy of cells stained with a Hoechst dye. Clones with small or absent kDNA are candidates for further study.
In one case, we found that the pZJM vector had integrated not in the expected rDNA spacer, but instead into the gene corresponding to its insert . This misintegration positioned one of the vector's T7 promoters so that there was a high level of transcription of nearly 10 downstream genes. Subsequently, we found that overexpression of two of these genes, individually, caused kDNA loss. The function of these genes in network replication is under study (S.A. Motyka, unpublished work).
We have so far used this strategy to conduct small screens of 300–1400 cloned cell lines and we have identified six novel proteins whose knockdown (or in two cases, overexpression) causes kDNA loss. One of these proteins (a helicase), we already knew from other studies, is an obvious candidate for involvement in replication and another (a subunit of an aminoacyl-tRNA synthetase) has so far no clear direct connection with kDNA synthesis although a related protein is a processivity factor for mitochondrial DNA polymerase in higher eukaryotes . The other proteins have sequences that provide no clues to their function, and they are strong candidates for a role in some of the unique steps in the kDNA replication pathway. It will be challenging and exciting to identify their functions.
Large-Scale Screening: A Focus Topic at BioScience2005, held at SECC Glasgow, U.K., 17–21 July 2005. Edited by B. Baum (Ludwig Institute, London, U.K.), K. Brindle (Cambridge, U.K.), S. Eaton (Institute of Child Health, London, U.K.) and I. Johnstone (Glasgow, U.K.).
We thank the National Institutes of Health (grants AI58613 and AI21334) for support of our research. We thank Dr David Pérez-Morga for the micrograph.