Strict Pairing of var Promoters and Introns Is Required for var Gene Silencing in the Malaria Parasite Plasmodium falciparum*

The human malaria parasite, Plasmodium falciparum, maintains a persistent infection altering the proteins expressed on the surface of the infected red blood cells, thus avoiding the host immune response. The primary surface antigen, a protein called PfEMP1, is encoded by a multicopy gene family called var. Each individual parasite only expresses a single var gene at a time, maintaining all other members of the family in a transcriptionally silent state. Previous work using reporter genes in transiently transfected plasmid constructs implicated a conserved intron found in all var genes in the silencing process. Here we have utilized episomal recombination within stably transformed parasites to generate different var promoter and intron arrangements and show that loss of the intron results in var promoter activation. Further, in multicopy plasmid concatamers, each intron could only silence a single promoter, suggesting a one-to-one pairing requirement for silencing. Transcriptionally active, “unpaired” promoters remained active after integration into a chromosome; however, they were not recognized by the pathway that maintains mutually exclusive var gene expression. The data indicate that intron/promoter pairing is responsible for silencing each individual var gene and that disruption of silencing of one gene does not affect the transcriptional activity of neighboring var promoters. This suggests that silencing is regulated at the level of individual genes rather than by assembly of silent chromatin throughout a chromosomal region, thus providing a possible explanation of how a var gene can be maintained in a silent state while the immediately adjacent var gene is transcriptionally active.

The human malaria parasite, Plasmodium falciparum, maintains a persistent infection altering the proteins expressed on the surface of the infected red blood cells, thus avoiding the host immune response. The primary surface antigen, a protein called PfEMP1, is encoded by a multicopy gene family called var. Each individual parasite only expresses a single var gene at a time, maintaining all other members of the family in a transcriptionally silent state. Previous work using reporter genes in transiently transfected plasmid constructs implicated a conserved intron found in all var genes in the silencing process. Here we have utilized episomal recombination within stably transformed parasites to generate different var promoter and intron arrangements and show that loss of the intron results in var promoter activation. Further, in multicopy plasmid concatamers, each intron could only silence a single promoter, suggesting a one-to-one pairing requirement for silencing. Transcriptionally active, "unpaired" promoters remained active after integration into a chromosome; however, they were not recognized by the pathway that maintains mutually exclusive var gene expression. The data indicate that intron/promoter pairing is responsible for silencing each individual var gene and that disruption of silencing of one gene does not affect the transcriptional activity of neighboring var promoters. This suggests that silencing is regulated at the level of individual genes rather than by assembly of silent chromatin throughout a chromosomal region, thus providing a possible explanation of how a var gene can be maintained in a silent state while the immediately adjacent var gene is transcriptionally active.
Malaria caused by the protozoan parasite Plasmodium falciparum continues to place a heavy health and economic burden on the developing world, particularly in sub-Saharan Africa (1). P. falciparum parasites infect the circulating red blood cells of their human hosts, modifying the red blood cell membrane and placing the parasite-encoded protein PfEMP1 on the cell surface. This makes infected cells cytoadherent, resulting in the sequestration of the parasites within the deep vascular beds and resulting in disruption of circulation within such organs as the brain and placenta. The syndromes of cerebral malaria and placental malaria are thought to be a direct consequence of this process (2). PfEMP1 is encoded by the multicopy var gene family (3)(4)(5). Each haploid parasite possesses ϳ60 var genes within its genome (6) but expresses only a single gene at a time, maintaining all other var gene copies in a transcriptionally silent state (7,8). By switching which gene is expressed, parasites alter both the cytoadherent properties of the infected cells and their antigenic phenotype, thus avoiding the antibody response of the infected individual and maintaining a persistent infection. This process of var gene expression switching and antigenic variation is dependent on strict control of var gene transcription, such that only a single gene is active at a time. Silencing of the remaining members of the family is therefore imperative for parasite survival.
Recent work has shed some light on the process of var gene regulation. Changes in var gene expression are not accompanied by alterations in the sequence of the genes or in their position in the genome (8), and activation or silencing of specific genes is unlikely to result from changes in the presence or absence of specific transcription factors (9). Changes in the transcriptional status of individual var genes have, however, been linked to alterations in chromatin structure and subnuclear localization, implicating an epigenetic mechanism for var gene regulation (10,11).
Transfection experiments using episomes containing var promoters have demonstrated a role for the conserved var intron in promoter silencing (12)(13)(14). This silencing was shown to be S-phase-dependent, a characteristic typical of silencing that is the result of alterations in chromatin structure (15,16). In the context of a transiently transfected episome, the ability of the intron to function as a silencer requires its own independent promoter activity, giving rise to sterile transcripts that initiate within the central region of the intron (13,14).
Here we report the generation of stably transformed P. falciparum parasites carrying reporter genes driven by either active or silent var promoters. These lines were generated using constructs that contain only the transcriptional activities found in var promoters and introns, thus maintaining as closely as possible the natural architecture of an intact var gene. Further, by taking advantage of intramolecular recombination events within episomally replicating multicopy concatamers, we were able to analyze how different arrangements of var promoters and introns influence expression. Recombination events that deleted an intron from the concatamer led to high levels of expression from the resulting "free" var promoter, thus confirming the role of the intron in var promoter silencing. In large concatamers, an equal number of var promoters and introns always resulted in complete silencing; however, concatamers in which var promoters outnumbered introns always displayed high levels of promoter activity, regardless of whether the concatamer was episomal or integrated into the chromosome. These assays indicate that there is a strict one-to-one pairing requirement between var promoters and introns for var promoter silencing to occur and that each intron can silence only a single var promoter. Parasites containing a chromosomally integrated active "unpaired" var promoter were also assayed for expression of the rest of the var gene family. Expression of this var promoter appears to have no effect on expression of the endogenous var genes within the genome, indicating that it is not "counted" within the allelic exclusion pathway that maintains mutually exclusive var gene expression.

EXPERIMENTAL PROCEDURES
Reporter Constructs-The plasmid pVLH/IDH was described previously ( Fig. 1) (13). This reporter construct carries two reporter cassettes, the first utilizing a var promoter driving expression of a firefly luciferase gene and the second using the promoter activity of a var intron to drive expression of the human dihydrofolate reductase (hdhfr)-selectable marker. Both cassettes use the hrp2 3Ј-UTR 2 to terminate the expressed mRNA and ensure efficient protein expression.
Parasite Culture and Transfection-P. falciparum parasites were cultivated at 5% hematocrit in RPMI 1640 medium, 0.5% albumax II (Invitrogen), 0.25% sodium bicarbonate, and 0.1 mg/ml gentamicin. Parasites were incubated at 37°C in an atmosphere of 5% oxygen, 5% carbon dioxide, and 90% nitrogen. Reporter constructs were transfected into the NF54 parasite line by using "DNA-loaded" red blood cells as described previously (17). Briefly, 0.2-cm electroporation cuvettes were loaded with 0.175 ml of erythrocytes and 50 g of plasmid DNA in incomplete cytomix solution. For stable transfection, NF54 parasites were cultured in media containing 40 ng/ml pyrimethamine. Plasmid rescue experiments were performed by transforming Escherichia colicompetent cells with 100 ng of purified P. falciparum genomic DNA.
After obtaining stable transformants of the NF54 P. falciparum strain clonal cultures were generated by limiting dilution using 96-well microtiter plates (18). As a control for endogenous var gene switching, we also generated clonal cultures of wild type NF54 parasites. Media changes were performed at 7, 14, 21, 25, and 30 days. Plates were supplemented with 200 l of red blood cells each on day 14. Individual plates were screened for parasites during media changes on days 21, 25, and 30. This gave rise to the clones D10, E4, G6, and E5. To select for chromosomal integration, the clonal parasite lines were cycled on and off drug pressure and recloned by limiting dilution resulting in the parasite lines D10/integrated, E4/integrated, E5/integrated, and G6/integrated. Plasmid rescue experiments were employed before and after drug cycling to quantitate episomal load of each parasite line. Individual parasite cultures were grown up to 20-ml cultures, synchronized, and used for determination of luciferase activity as well as for DNA and RNA extraction. The cell line D10/retransformed was obtained by retransfecting a rescued episome from the D10 parasite line into the wild type NF54 parasites.
Luciferase Assays-Parasites were synchronized by the Percoll-sorbitol method (13,19). Schizonts were isolated from 20-ml cultures using a 70%/40% Percoll-sorbitol gradient and subsequently inoculated into a 20-ml culture at 5% hematocrit. Parasites were allowed to reinvade overnight, and luciferase activity was measured at ϳ12-14 h post-invasion. Synchronization was confirmed by light microscopy. In our hands this procedure yielded almost 100% pure ring stage parasites. Parasitemias were counted for 1000 red blood cells. Parasites were obtained from 200 l of culture by centrifugation and subsequent lysis in 100 l of Glo Lysis Buffer. 100 l of Bright-Glo luciferase reagent system was added to the lysate. Luciferase activity was measured immediately in a TD-20/20 luminometer. Luciferase activity was expressed per 1% ring stage parasitemia. The luciferase activity of each clonal cell line was determined in at least three independent experiments. Southern Blots and Diagnostic PCR-Analysis of the rearrangements of both episomal and integrated constructs was performed using Southern blots and diagnostic PCR. Southern blots were performed according to established protocols (20). Briefly, genomic DNA isolated from recombinant parasites was digested to completion by restriction enzymes and subjected to gel electrophoresis using 1% agarose in Tris/ boric acid/EDTA buffer. The DNA was transferred to high-bond nitrocellulose membrane by capillary action after alkaline denaturation. DNA detection was performed using the Amersham Biosciences nonradioactive detection kit according to the manufacturer's protocols.
All PCR reactions were carried out on a PTC-2000 Peltier thermal cycler using Taq polymerase (Invitrogen) under the following conditions: 95°C for 5 min followed by 35-38 cycles of 94°C for 30 s, 52°C for 40 s, 68°C for 3 min, and a final extension step of 68°C for 10 min. Reaction products were analyzed by gel electrophoresis and automated sequencing.
Genomic DNA Extraction-Infected red blood cells were pelleted by centrifugation at 6000 rpm. After the supernatant was discarded, the pellet was divided into two microcentrifuge tubes followed by resuspension in 500 l of phosphate-buffered saline and 20 l 10% saponin. Parasites were pelleted by centrifugation and washed with 1000 l of phosphate-buffered saline. This process was repeated twice. The parasite pellet was then taken up in 200 l of Tris/sodium chloride/EDTA buffer to which 40 l of 10% SDS and 20 l of 6 M NaClO 4 were added. This suspension was placed on a rocker overnight and was phenol/ chloroform-extracted the next morning. The final aqueous phase was ethanol-precipitated and resuspended in 10 l of sterile distilled H 2 O. Final DNA concentration was determined by absorbance at 260 nm. Yields ranged from 100 ng/l to 1 g/l depending on parasitemia and stage of parasites.
RNA Extraction and Real-time RT-PCR for Assaying Expression of the var Gene Family-RNA was extracted from synchronized ring stage parasites 12-14 h post-invasion. RNA extraction was performed with the TRIzol LS Reagent as described previously (21). RNA to be used for cDNA synthesis was treated with Deoxyribonuclease I (Invitrogen) as described by the manufacturer. A total of 1600 ng of RNA was digested in a 20-l reaction. Samples were incubated at room temperature for 30 min followed by 10 min of heat inactivation at 65°C. cDNA synthesis was performed with Superscript II RNase H reverse transcriptase (Invitrogen) with random primers (Invitrogen) as described by the manufacturer. cDNA was synthesized from 800 ng of RNA in a 40-l reaction. For each cDNA synthesis reaction, a control reaction without reverse transcriptase was performed with identical amounts of template. Transcription of the var gene family was quantified using the primer set published by Salanti et al. (22). With this primer set 150-bp fragments of exon I of each individual var gene present in the 3D7/NF54 P. falciparum strain are generated. To compare expression levels of the endogenous genes with the luciferase reporter gene, the primer pair luciferase forward (LF) (see above) and luciferase reverse (LR) (5Ј-GT-GTTCGTCTTCGTCCCAGT-3Ј) were designed using primer 3 software. Amplification efficiency was verified by testing all primer pairs using a 3-log range of concentrations of genomic DNA obtained from transformed parasites. var gene transcription was expressed relative to the expression of a housekeeping gene (⌬CT method). All reactions included the three control genes published by Salanti et al.: seryl-tRNA synthetase (PF07_0073), fructose-bisphosphate aldolase (PF14_0425), and actin (PFL2215w). However, an additional two control genes were also included: arginyl-tRNA synthetase (PFL0900c), using (5Ј-AAGAGATGCATGTTGGTC-3Ј) and (5Ј-GTACCCCAATCACCTACA-3Ј); and glutaminyl-tRNA synthetase (PF13_0170), using (5Ј-GGCACTTCAAGGGTACCT-3Ј) and (5Ј-TAATATAGCCTCACAAGC-3Ј).
Reactions were performed at a final primer concentration of 0.5 M using Bio-Rad ITAQ Sybr Supermix in 20-l reactions on an ABI Prism 7900HT real-time PCR machine. The ⌬CT for each individual primer pair was determined by subtracting the CT value from the CT value of the control the seryl-tRNA synthetase (Applied Biosystems, User Bulletin 2). ⌬CTs were then converted to relative copy numbers with the formula 2 ⌬Ct .
Quantitation of Intron/Promoter Copies by Real-time PCR-To quantify the relative amounts of introns and promoters in the different cell lines, primer pairs specific to the promoter and intron of the pVLH/ IDH plasmid were designed.
The primer pair promoter forward (PF) (5Ј-ATTTTGAGTGTAAC-CAAGCGTAT-3Ј) and promoter reverse (PR) (5Ј-TCCACA-CAAAAACATAAACCTCA-3Ј) and the primer pair intron forward (IF) (5Ј-GGGGGATCCATTGCTTTTT-3Ј) and intron reverse (IR) (5Ј-AAATCACATACATATACACAAACACTT-3Ј) were used to amplify 150-bp fragments of the promoter and intron, respectively. The first 5 G residues in primer IF belong to the multicloning site of the original plasmid and ensure that the primer pair will only amplify the original plasmid intron. The promoter contained in pVLH/IDH was originally obtained from the Dd2 parasite line (9) and is not present in the NF54 genome. Both primer pairs showed identical amplification profiles across 3 logs of DNA template dilutions. Real-time PCR conditions were the same as used for the expression analysis of the var gene family. Relative DNA copy numbers were determined using the ⌬CT method. The upstream promoter is silenced by the intron in an S-phase-dependent manner. In the pVLH/IDH plasmid the var upstream promoter drives luciferase expression but is silenced by the intron promoter. The intron promoter drives human dhfr expression. The hrp2 3Ј-UTR region follows both the luciferase and hdhfr genes to ensure efficient expression of these marker genes. B, active and silent parasite lines obtained after transfection with the pVLH/IDH plasmid. Luciferase activity was measured in synchronized parasites in mid-ring stage (ϳ14 h post-invasion) and is expressed as luminescence units per percent parasitemia. Luciferase activity in the active parasites D10 and E4 was at least 100-fold higher than in the silent parasite lines E5 and G6. The parasite line D10/retransformed (D10/re) was generated by retransfecting a rescued plasmid from the D10 parasite line into wild type parasites.

Stable Transformation of Parasites with the Plasmid pVLH/IDH
Results in Either an Active or a Silent Phenotype-To investigate var promoter regulation on stably replicating rather than transiently transfected plasmids, it is necessary that the constructs also contain a selectable marker cassette. Typically this involves the placement of a constitutively active promoter driving a drug resistance gene onto the plasmid, thus allowing drug selection of transformed parasites (23)(24)(25). However, previously it was shown that the ability of a var intron to silence an associated var promoter is dependent on the independent promoter activity of the intron itself (13), indicating that some sort of interaction, possibly promoter competition or interference, is responsi-ble for the silencing effect. To avoid the potentially confounding effects of placing an additional transcriptionally active promoter next to the var promoter, we instead used the promoter activity of the intron itself to drive expression of the human dhfr gene. In the construct pVLH/IDH (Fig. 1), a var promoter is used to drive expression of the firefly luciferase reporter gene. However, the presence of a var intron on the plasmid effectively silences the var promoter, and thus parasites transiently transfected with this construct are silent with respect to luciferase expression (13).
Cultured parasites were transfected with this construct and placed under selection with the antimalarial drug pyrimethamine. After approximately 3 weeks of drug pressure, stably transformed parasites During episomal replication multimers are formed by rolling circle replication, and the size of the concatamer is maintained through intramolecular homologous recombination. The original pVLH/IDH plasmid carries a duplicated hrp2 3Ј-UTR sequence. Homologous recombination between these two sequences can result in two new plasmid species. In one species the intron is deleted from the VLH/IDH expression cassette, whereas in the other the cassette now has an additional (double) intron. PCR with the primers luciferase forward (LF) and vector reverse (VR) amplify a 1330-bp fragment from the cassette that deleted its intron or a 3281-bp fragment from the original VLH/IDH cassette. The primers DHFR forward (DF) and DHFR reverse (DR) amplify a 1826-bp fragment only from the cassette that carries the double intron. were detectable in the culture, and plasmid rescue experiments demonstrated that they were indeed carrying episomes. Parasites were then cloned by limiting dilution to assess whether clonal parasite cultures exhibited the silent phenotype observed in transient transfection experiments. Surprisingly, about half of these new clones exhibited an active phenotype with very high levels of luciferase expression, whereas the remainder of the clones exhibited the expected silent phenotype. This indicated that ϳ50% of the stably transformed parasites had activated the var promoter in the construct and were expressing luciferase. Fig. 1B shows the levels of luciferase expression from four representative clones demonstrating the active and silent phenotypes.
Uneven Recombination within Episomal Concatamers Results in Deletion of the Intron from Some Expression Cassettes-Transformation with the silent pVLH/IDH plasmid generated both active and silent parasite clones in approximately equal proportions. Previous reports have suggested that switches in var gene expression occur at a rate of 0.1-2% per generation (26,27). If var promoter activation in our transformants had occurred at this rate, only a small fraction of the parasite population should have exhibited the active phenotype. We therefore hypothesized that var promoter activation in the clonal parasite lines was unlikely to be due to "switching" but most likely was secondary to a recombination event that resulted in var promoter activation.
Within parasites, plasmids are thought to multiply into head-to-tail concatamers of 9 -15 copies, with the size of the concatemer maintained through an ill-defined recombination-dependent mechanism (28,29). The pVLH/IDH plasmid carries a duplicated 3Ј-hrp2 terminator sequence following both the luciferase and hdhfr coding sequences. We hypothesized that homologous recombination between these two sequences generated two new episome species (Fig. 2), one that deleted the intron from the VLH/IDH cassette and the other one carrying a cassette with a double intron. Because the intron was previously shown to be required for silencing, this recombination event could have resulted in var promoter activation. PCR reactions using diagnostic primer pairs designed to specifically detect these two new species of plasmids confirmed their existence within genomic DNA extracted from the transformed parasites (Fig. 3B). In accordance with our hypothesis, all parasites that exhibited the active phenotype had deleted an intron from the original VLH/IDH cassette and contained a free var promoter. Interestingly, the parasite line E5 contained both products of the recombination event, and in these parasites the luciferase gene was silent. This suggested that in this parasite line the arrangement of promoters and introns had changed, whereas their numbers remained equal.
Analysis of Rescued Plasmids and Retransformation of Parasites-To further study the structure and expression phenotype of the recombined episomes, we attempted to recover them by "plasmid rescue" in E. coli. Episomes were readily rescued, and restriction enzyme and Southern blot analysis of the recovered plasmids clearly identified the intron deletion in the episomes rescued from the parasites expressing high levels of luciferase (Figs. 3C and 4A). The plasmid containing the double intron recombination product was also easily detected by colony screen PCR of E. coli transformed with DNA from silent parasites (data not shown); however, we were unable to obtain these episomes intact from E. coli by miniprep. Because we clearly detected the double intron recombination product in Southern blots, and by PCR using genomic DNA from the silent parasites we suspected that our difficulty in rescuing this episome was the result of the high AT content of the tandemly repeated double intron, leading to instability in E. coli. Some of the plasmids recovered from the active clones contained a VLH/IDH cassette with an intron deletion as well as an intact VLH/IDH cassette (Fig.  3C). This suggested that transcription from the var promoter in the cassette without an intron was not affected by the additional promoter/ intron pair on the same episome.
As final confirmation that activation of luciferase expression was the result of the deletion of the intron from the VLH/IDH cassette, we retransformed an episome rescued from the active clone D10 into wild type parasites. This episome carried the original VLH/IDH cassette in addition to a cassette that had deleted its intron (Fig. 4). Luciferase activity was measured as soon as drug-resistant parasites were detected in culture, when the transformed cells carried almost exclusively episomal copies. As expected, the parasites displayed the active phenotype (Fig. 1B). Plasmids were again rescued from the transformed parasites and analyzed using diagnostic restriction enzyme digestion. All recovered episomes carried the cassette that had deleted the intron in addition to the original VLH/IDH cassette (Fig. 4B), confirming the original observation.

In Chromosmally Integrated Concatamers Luciferase Expression Levels Reflect a Strict Pairing Requirement between var Promoters and
Introns for Silencing-To investigate the effect of integration of the constructs into the chromosome on the active or silent phenotype, we cycled transformed parasites on and off drug pressure and recloned individual high and low expressing parasite lines. Each cloned line was analyzed for site of integration, concatamer arrangement, and luciferase FIGURE 4. Analysis of rescued and retransfected episomes. A, Southern blot of an active episome used for retransfection of wild type parasites. The episome was rescued from the D10 parasite line. Restriction enzyme analysis shows the 1.6-kb EcoRI/BstXI fragment that results from the intron deletion as well as the three fragments of the original pVLH/IDH plasmid. The 1.6-kb EcoRI/BstXI fragment as well as the original 2.4-kb EcoRI/NotI fragment hybridize strongly with the luciferase probe. B, the rescued episome was retransfected into wild type parasites, resulting in the active parasite line D10/retransformed (see Fig. 1B). DNA from this parasite line was again rescued in E. coli and analyzed by restriction enzyme digest. All rescued episomes were identical to the original D10 episome. Lane C is a control digest of the original pVLH/IDH, which does not contain the 1.6-kb fragment representing the free var promoter.
fragments. The cassette with the double intron carries a new 1.9-kb NotI/NotI fragment in addition to the fragments in the original cassette. The primer pairs detecting the corresponding PCR fragments are also indicated. B, PCR reactions using genomic DNA extracted from active and silent parasite lines as template. The left panel shows PCR reaction products after amplification with the primer pairs LF and VR. The 1330-bp PCR fragment was amplified from the active parasite lines D10 and E4, showing that both lines carry a VLH/IDH cassette that has deleted its intron. The silent parasite G6 carries the original intact cassette. The silent parasite E5 also carries a cassette with an intron deletion. The right panel shows PCR reactions products amplified with the primers DF and DR. The double intron 1826-bp PCR fragment is only amplified from the silent parasite lines G6 and E5. The silent parasite line E5 therefore carries both recombination products: a cassette with an extra intron as well as a cassette with a deleted intron. C, restriction enzyme analysis of different episomes rescued in E. coli. All episomes were digested with BstXI, EcoRI, and NotI. Lane 2 represents the digest of the original pVLH/IDH plasmid (C, control). Lane 3 shows an identical episome recovered from a silent parasite line. Lanes 4 and 5 show recombined episomes from active parasites. The episome in lane 4 carries the 1.6-kb fragment representing the VLH/IDH cassette that has lost its intron. The episome in lane 5 shows a multimer that consists of the original pVLH/IDH plus a cassette that has deleted the intron.
var Promoter/Intron Pairing APRIL 14, 2006 • VOLUME 281 • NUMBER 15 FIGURE 5. Analysis of parasites carrying chromosomally integrated concatamers. A, luciferase activity of transgenic parasites with chromosomally integrated concatamers. The active parasite lines D10/integrated (inte) and E4/integrated parasite lines exhibit at least a 15-50-fold higher luciferase activity than the silent parasite lines E5/integrated and G6/integrated. After integration, parasites were kept in continuous culture for up to 10 months and maintained stable luciferase expression levels. B, schematic representation of the single crossover integration events at the endogenous var intron and the hrp2 3Ј-UTR. The active concatamers of the parasite lines D10 and E4 integrated into the 3Ј-UTR region of the hrp2 gene. The silent concatamers of the parasite lines G6 and E5 integrated at the intron of the var gene PFB1055c. Because integration of the different concatamers at each locus occurred in the same way, only one type of concatamer is shown. C, Southern blot analysis of active and silent parasites carrying integrated concatamers. Only one lane is shown for expression. After selection for integration and recloning, all parasite lines derived from the original silent episomes maintained their silent phenotype. Similarly, all integrated clones derived from the original active parasites continued to express high levels of luciferase (Fig. 5A), with the levels of expression varying ϳ3-fold. Selection for integration and the loss of episomes coincided with a relative decrease of luciferase activity compared with the exclusively episomal active cell lines. However, all chromosomally integrated cell lines were kept in continuous culture for at least 10 months and maintained stable luciferase expression throughout this time. This suggests that once the concatamers had integrated into the genome their expression and arrangements remained stable.
PCR amplification and sequencing were employed to determine the exact site of integration (Fig. 5B). The silent episomes G6 and E5 integrated by a single crossover event at the intron of var gene PFB1055c on chromosome 2. The active concatamers in the cell lines D10 and E4 integrated by single crossover in the 3Ј-UTR of the hrp2 locus located in the subtelomeric region of chromosome 8 (30,31). Both of these loci are located within the subtelomeric heterochromatin that has recently been described to extend up to 140 kb into the chromosome (11).
Southern blots (Fig. 5C) and real-time PCR were used to distinguish the arrangement of var promoters and introns in the different integrated clones. The high expressing clones E4/integrated and D10/integrated contained both an intact cassette and a cassette that had deleted the intron. No double intron-containing cassettes were found in these clones, and both cell lines exhibited an active phenotype. The clone G6/integrated contained both the original VLH/IDH cassette and the double intron recombination product. It did not however contain a cassette without an intron, and luciferase assays showed that the var promoters in this arrangement were essentially silent. The clone E5/integrated contained a complex tandem array of an original VLH/IDH intact cassette, a cassette that had deleted its intron, and a cassette containing a double intron. This confirmed the observation that the episomes in the original E5 clone contained both recombination products (Fig. 3B). Because luciferase expression is silent in parasites that contain this arrangement, we hypothesize that in this concatamer the extra intron in the double intron cassette is able to interact with the free var promoter to silence its expression. This again suggested that for complete silencing to occur, every var promoter needs to be "paired" with an intron.
To verify that strict "one to one" pairing between var promoters and introns is required for silencing, quantitative real-time PCR was performed using primer pairs specific for the var promoter and intron included in the construct. As controls the original silent pVLH/IDH plasmid and a rescued active episome carrying the original VLH/IDH cassette plus a cassette that had deleted the intron were analyzed. In the original pVLH/IDH plasmid the number of var promoter and intron copies was identical (Table 1), whereas the active episome carried 2-fold more var promoters than introns, as expected. The same relationship was observed for total genomic DNA extracted from the active and silent chromosomally integrated cell lines. All silent parasites carried integrated concatemers containing equal numbers of var promoters and introns, whereas all active parasite lines carried more var promoters than var introns. These data again suggest that silencing requires strict var promoter and intron pairing.
To exclude the possibility that luciferase activity differences were simply a result of differences in luciferase copy number in the respective concatamers, we performed quantitative real-time PCR for luciferase and a single-copy housekeeping gene on genomic DNA extracted from active and silent parasite lines. These experiments revealed that the silent cell line E5/integrated carried more luciferase copies than the active cell line E4/integrated, thus demonstrating that the number of luciferase copies per concatamer did not correlate with the active or silent phenotype (Table 2). However, within the active cell lines, the concatamer in the D10/integrated line was larger than in the E4/integrated line. This suggested the possibility that differences in luciferase expression levels between different active cell lines could be because of differences in the number of integrated cassettes that had unpaired, active var promoters.
Unpaired var Promoters Are Not Counted as Part of the Allelic Exclusion Pathway-Within any given parasite, only a single var gene is thought to be expressed, indicating that a mechanism of allelic exclusion ensures that only one var promoter is actively transcribed at a time. In the constructs described here, recombinational deletion of an intron renders a var promoter constitutively active, regardless of whether it is located on an episome or integrated into the chromosome. In the active clone D10/integrated, the level of luciferase expression was constant for more than 10 months of continuous culture, supporting the conclusion that an unsilenced promoter is constitutively active and does not switch

Relative copy number of var promoter and intron per concatamer
Quantitative real-time PCR was performed on the DNA of silent and active concatamers. Experiments were performed across 3 logs of DNA template concentrations without affecting the relative amounts of intron and promoter signal. Data from a single dilution are shown. A ⌬CT value of 0 reflects an equal number of intron and promoter copies. A value of 1 implies 2-fold more promoter than intron copies. The negative value for G6/inte reflects the presence of the double intron in this concatamer, resulting in more intron than promoter copies.

DNA
⌬CT  var Promoter/Intron Pairing between the on and off state. If this active promoter is recognized by the mechanism that ensures that only a single var gene is expressed at a time, the remainder of the var gene family should be turned off in these parasites.
To investigate how active expression of this unpaired var promoter affects transcription of the rest of the var gene family, we employed real-time reverse transcription-PCR and the method of Salanti et al. (22) to quantitatively measure mRNA transcribed from every var gene in the parasite genome. This method uses unique oligonucleotide primer pairs specific to every var gene in the genome. Using cDNA from mid-ring stage parasites, expression levels of each individual var gene were measured. We recloned the D10/integrated parasite line to ensure that there  (22). The copy number is relative to the housekeeping gene t-RNA synthetase. Genes were sorted according to their level of expression. The NF 54 population is dominated by two major var transcripts, but multiple other transcripts are detectable at a low copy number (B). This is consistent with a population that is switching away from a predominant var transcript. The D10/integrated parasite population expresses luciferase at a higher level than the housekeeping gene, suggesting that all parasites in this population express luciferase from the unsilenced promoters within the integrated plasmid construct (A). However the expression pattern of the endogenous var genes is similar to the NF54 wild type population. Because luciferase expression was stable over a period of 10 months, the data suggest that the promoter expressing luciferase is not counted as part of the pathway regulating monoallelic var expression and does not affect endogenous var gene switching.
var Promoter/Intron Pairing would be a relatively small number of dominant var transcripts within the population of parasites (27). As a control we cloned wild type NF54 parasites. RNA transcriptional analysis was done on both lines at ϳ7 weeks after cloning. The transcriptional analysis showed that the D10/ integrated cell line showed a similar switching pattern as the wild type parasite despite high expression of luciferase from the unsilenced promoter (Fig. 6). This suggests that the var promoter that is driving luciferase expression is not recognized by the pathway that regulates mutually exclusive var gene expression.

DISCUSSION
The experiments described here use the reporter construct pVLH/ IDH to study the requirements for var promoter silencing. Because firefly luciferase is a neutral marker, it likely reflects the natural propensity of a var promoter to assume an active or silent state, and the use of the intron promoter activity to drive expression of the selectable marker avoids the potentially confounding influence that additional exogenous promoters might have on var promoter activity. In addition, the intramolecular recombination event described here provided us with a unique opportunity to study stably transfected parasites in which a var promoter has been separated from the influence of the intron. We observed that var promoters that were paired with an intron were always silent, whereas those that became unpaired as a result of recombination were always transcriptionally active, regardless of whether they were located on an episome or after integration into the subtelomeric region of a P. falciparum chromosome. Importantly, although active promoters appeared to be unaffected by neighboring intron/promoter pairs, they were affectively silenced by an "extra" intron in the same concatamer (illustrated in Fig. 7). Thus, although successful pairing and silencing did not require that the intron be in its normal position directly downstream of the var promoter, it did however appear to be strictly one to one, requiring an equal or greater number of introns to silence all of the var promoters within an integrated concatameric construct. These results therefore suggest that each var promoter/intron pair acts as a separate unit such that each intron silences only a single var promoter.
Many var genes are located in clusters along the chromosome in very close proximity to one another. Repositioning into distinct perinuclear sites has been associated with var gene activation; however, only a single var gene becomes activated, whereas all adjacent var genes remain transcriptionally silent, indicating that locus repositioning does not uniformly activate all var genes (32). Similarly, if alterations in chromatin structure play a significant role in regulating var gene expression, as has been proposed, the changes must be localized within the confines of a single gene rather than across a large chromosomal segment encompassing multiple var genes. This suggests that each var gene represents an independent transcription unit and that the silencing activity of var introns is different from some other previously described transcriptional silencers. For instance, the sir2-dependent silencers that flank the mating type locus in Saccharomyces cerevisiae work cooperatively to form a silent chromatin structure over an entire region of the chromosome, including all promoters that exist within this region (33). The var introns, however, each appear to silence exactly one var promoter, even if several var promoters are in close proximity. This suggests that rather than simply mediating the assembly of silent chromatin throughout a locus, each var intron instead specifically interacts with an individual var promoter to silence its transcription, thus maintaining one var gene in a transcriptionally silent state while an adjacent var gene can be transcriptionally active. The exact nature of this interaction remains ill defined; however, it appears to rely on the promoter activity of the intron itself. Similarly, is seems likely that the allelic exclusion pathway recognizes promoter/intron pairs or other regulatory sequences at the var locus. This would explain why in our experiments a transcriptionally active var promoter that was separated from an intron and integrated at the hrp2 locus was not recognized as part of the allelic exclusion pathway and had no effect on expression of other members of the gene family. These conclusions are supported by several other recently published observations. Disruption of the var2CSA gene through insertion within exon I of a selectable expression cassette renders this promoter constitutively active (34). This may be because the insertion has disrupted pairing between the intron and promoter. This transcriptionally active, unpaired promoter is also not recognized as part of the var gene family and has no effect on expression of the other var genes within the FIGURE 7. Model for var promoter silencing based on strict intron/promoter pairing. The top line reflects the intron/promoter arrangement in a parasite that actively expresses luciferase from the integrated construct. A VLH/IDH cassette without an intron carries an unpaired active promoter that results in high levels of luciferase expression. Transcription from this promoter is not affected by the neighboring intron/promoter pair. The middle line represents the situation in the silent G6 clone. Each promoter is paired with its original intron and all promoters are therefore silenced. The extra intron in the double intron does not participate in the silencing process. The bottom line represents the conformation of the concatamer in the silent E5 parasite. A double intron is located next to a promoter that has lost its intron. The extra intron is now "pairing" with the promoter and silencing luciferase expression.
var Promoter/Intron Pairing genome. The endogenous var1CSA gene (also called varcommon) possesses an intron that carries a substantial deletion in a region previously shown to be important for silencing (13). This gene is constitutively transcribed in many parasite isolates, indicating that it cannot be silenced (35,36). However expression of this gene does not preclude expression of other members of the gene family, indicating that it is not recognized by the allelic exclusion mechanism. Finally, Gannoun-Zaki et al. (14) reported that in stably inherited episomal constructs, a var promoter that is paired with an disabled intron is transcriptionally active and does not affect expression of the rest of the var gene family.
Subtelomeric heterochromatin has been shown to be important for gene silencing in yeast and recently has been implicated in silencing at least a subset of subtelomerically located var genes in P. falciparum (10,11). In contrast, we observed var promoter silencing on episomes as well as after chromosomal integration, indicating that silencing based on promoter/intron pairing is independent of chromosomal position. However we also observed a decrease in activity of unpaired var promoters upon integration into the subtelomeric region of a chromosome, indicating that subtelomeric heterochromatin may indeed affect var promoter expression. Interestingly, the combination of the two layers of regulation may support a model in which different var genes have different propensities to turn "on" or "off" as proposed by Horrocks et al. (27). The chromatin structure found in subtelomeric regions could influence var gene switching rates compared with var genes located in central regions of a chromosome, thus resulting in different on and off rates for individual var genes based on their position on the chromosome.
Understanding the transcriptional regulation of individual var gene promoters and the mechanism of allelic exclusion within this large multicopy gene family remain difficult challenges. Interactions between promoters, sterile RNA molecules, and alterations in subnuclear localization are all likely to be part of a complex and tightly regulated process that ultimately results in the elegant pattern of antigenic variation that malaria parasites display. Additional complexities may involve the different types of var upstream regions (37,38) and the possibility that different var genes are explicitly expressed in different hosts or different tissues (22,39). The regulation of this gene family, and also potentially the other multicopy gene families within the parasite genome, is therefore likely to be composed of multiple "layers" of regulation, each contributing to the expression patterns observed during natural infections.