Genomic and cDNA sequence data were accessed via PlasmoDB 10. The pfnek-4 gene PlasmoDB identifier is PF3D7_0719200, MAL7P1.100 in earlier versions. The Pfnek-4-GFP plasmid (pCHD-Pfnek-4) was generated by using the pHGB and pCHD-1/2 transfection vectors based on Gateway^TM recombinational cloning and described in Tonkin et al.11. The 997-bp 5’-flanking region of Pfnek-4 was amplified from 3D7 genomic DNA using the forward OL-306 (GGGTCGACGAACTCATCATTCATA) and reverse OL-308 (CCAGATCTTGAATGGTTATAAGATATAC) oligonucleotides containing SalI and BglII sites, respectively. The 933-bp Pfnek-4 open reading frame was amplified from a gametocyte cDNA library using the oligonucleotides forward OL-598 (CCCAGATCTATGAATAAATATGAAAAGATTAGAG) and reverse OL-599 (CCCCCTAGGAGTATCAACAACATCCAG) containing BglII and AvrII sites, respectively. The digested products, ~1-kb 5’-flanking region and open reading frame of Pfnek-4, were sequentially ligated into the plasmid pHGB to produce the pHGB-Pfnek-4-GFP entry clone. This plasmid was used in a recombination reaction with the pCHD-1/2 destination vector containing the cassette responsible for expression of hDHFR conferring resistance to WR99210 treatment, to produce the final transfection vector pCHD-Pfnek-4-GFP.

The pCC1 and pCC4 vectors constructed for negative selection of single crossover recombinants in the generation of knock-out parasites have been described 12 and formed the basis for the generation of plasmids pScCDUP_Pfnek-4 and phDHFR_Pfnek-4 described here. The pScCDUP_Pfnek-4 plasmid was constructed by replacing the 857-bp SacII-XhoI hsp86 5’ sequence of pCC4 by the 997-bp 5’-flanking region of pfnek-4 amplified from genomic DNA with forward OL-1082 (GGCCGCGGGAACTCATCATTCATA) and reverse OL-1083 (GGCTCGAGTGAATGGTTATAAGATATAC) containing SacII and XhoI sites, respectively, generating an expression vector in which the yeast cytosine deaminase gene is placed under the control of the pfnek-4 promoter. The phDHFR_Pfnek-4 plasmid was constructed by replacing the ~1.1–kb XhoI-XmaI yeast cytosine deaminase coding sequence from the pScCDUP_Pfnek-4 plasmid by ~0.6–kb of the hDHFR sequence amplified from plasmid pCC1 with forward OL-1084 (GGCTCGAGATGCATGGTTCGCTAAACTGC) and reverse OL-1085 (GGCCCGGGTTAATCATTCTTCTCATATAC) containing XhoI and XmaI sites, respectively, generating an expression vector in which the hDHFR gene is placed under the control of the pfnek-4 promoter.

A pfnek-4 gene disruption plasmid was produced in the plasmid pCAM-BSD that contains the gene conferring resistance to blasticidin. The oligonucleotide pair forward OL-46 (GGGGGGATCCAATTATGGAAATACAATACT) and reverse OL-47(GGGGCGCCGGCGTGGACTTAAATAATAAGG) containing BamHI and NotI sites was used to amplify a 1,017 bp fragment for insertion to pCAM-BSD. Ring-stage parasites were electroporated with 50–100 μg plasmid DNA, as previously described. Blasticidin (Calbiochem) was added to a final concentration of 2.5 μg/ml 48 hours after transfection to select for transformed parasites. Resistant parasites appeared after three to four weeks and were maintained under selection. After verification by PCR that pfnek-4^- parasites were present, the population was cloned by limiting dilution in 96-well plates (0.25/0.5/1.0 parasite per well). Genotypic analysis enabled selection of independent pfnek4^- clones for further phenotypic analysis. All constructs were sent to the Dundee Sequencing Service, University of Dundee, UK, for sequence verification before being used.

The 3D7 clone of P. falciparum, its F12 subclone and the 3D7 and F12 transfectants were grown in human erythrocytes using 0.5% Albumax II (Invitrogen) and synchronized using sorbitol as described previously 13. For transfections, synchronized ring-stage parasites (3D7 and F12) were electroporated with 50–100 μg of plasmid DNA using standard procedures. Transformed parasites were selected in presence of 5 nM WR22910 (Jacobus Pharmaceutical Co Inc, Princeton, NJ, USA) or 2.5 μg ml^-1 blasticidin (Calbiochem). Parasitaemia was monitored by Giemsa-stained thin blood smears. Induction of gametocytogenesis was performed according to the protocol of Carter et al.14, culturing asexual blood stage parasites for four to five days in 6%-haematocrit blood cultures to high 8-10% parasitaemia.

Total RNA samples were extracted from parasite pellets using TRIzol lysis solution (Invitrogen). DNase treatment of RNA samples prior to RT-PCR was performed by incubation at 37°C for 30 min using the RQ1 RNase-free DNase I purchased from Promega. The DNase was inactivated by incubation at 65°C for 10 min. RT-PCRs were performed with 500 ng of total RNA/reaction using the ImPromII reverse transcription system purchased from Promega. The RT reactions were incubated at 42°C for 1 h. For the first round of PCR (30 cycles at 94°C for 45 sec, 55°C for 45 sec, and 68°C for 2 min), Pfnek-4-specific primers were forward OL-587 (GAGAGGGATCCATGAATAAATATGAAAAGA) and reverse OL-586 (GAGAGGTCGACTTAAGTATCAACAACATCC). For the second round of PCR (25 cycles at 94°C for 45 sec, 55°C for 45 sec, and 68°C for 2 min), 1 μl (1/25) of each PCR product was reamplified using the Pfnek-4-specific primers forward OL-46 (GGGGGGATCCAATTATGGAAATACAATACT) and reverse OL-47 (GGGGCGCCGGCGTGGACTTAAATAATAAGG).

Disruption of the pfnek-4 gene was analysed by diagnostic PCR using three primer pairs. The primer pair forward OL-761(CACGACATTACATAATAAAAGC) and reverse OL-763 (ATCCCTTTTATGAATTTACTG) produced a 1288-bp fragment corresponding to the undisrupted Pfnek-4 locus from wild-type 3D7. Primer pairs forward OL-761 and reverse OL-168 (CAATTAACCCTCACTAAAG), and forward OL-167 (TATTCCTAATCATGTAAATCTTAAA) and reverse OL-764 (TCGAAGAGGTCATTATATATC) amplified across the 5’ and 3’ ends of the integration site, giving rise to 1,179- and 1,277-bp products only in the disrupted locus, respectively.

Western blot analysis was performed on cell-free extracts prepared by resuspending parasite pellets in M-PER Mammalian Protein Extraction Reagent (Pierce) supplemented with 1 mM phenylmethylsulphonyl fluoride and Complex^TM mixture protease inhibitor tablet from Roche Applied Science. Immunoblotting was performed as described 13, using monoclonal anti-GFP antibodies (Roche) and horseradish peroxidase-conjugated sheep anti-mouse IgG antiserum (Sigma). Lambda protein phosphatase (New England BioLabs) was used to dephosphorylate protein extracts (20 μg) 30 min at 30°C following the supplier recommendations prior to western blot analysis.

Live imaging of parasite transfectants was performed on a Delta vision deconvolution fluorescence microscope (100x/1.4 oil immersion objective Olympus IX-70). Images were processed using IMARIS version 7.0. For live imaging, parasite nuclei were stained by incubation 10 min at 37°C in complete medium containing 1 μg ml^-1 Hoechst 3342 (Invitrogen).

Flow cytometer analysis of parasite subpopulations was performed on a FACScan flow cytometer (Becton Dickinson Biosciences). Cells were either fixed in 0.025% glutaraldehyde and propidium iodide (PI)-stained or directly stained with 1 μg ml^-1 Hoechst 33258 (Invitrogen) prior to analysis. The threshold was set to a value eliminating uninfected red cells. For sorting Pfnek-4-GFP-expressing transfectants, late trophozoite-stage parasites were first enriched by using the VarioMACS separator and CS MACS columns (MiltenyiBiotec), then returned to culture conditions for three to four hours prior to cell sorting using a FACS ARIA II SORP Instrument (Becton Dickinson Biosciences). In two independent experiments, Pfnek-4-GFP^- (4 10⁶ and 12 10⁶) and Pfnek-4-GFP⁺ (0.8 10⁶ and 1.6 10⁶) parasites, respectively, were collected over a four-hour cell sorting period and returned to culture conditions in flat-bottom 96-well plates with fresh red blood cells to 2% -haematocrit and 1% -parasitaemia. Gating of GFP^- and GFP⁺ parasite populations has been set in a restrictive way such that the sorted cells exhibited high levels of purity in both experiments (purity of GFP^- sorted cells, 100%; GFP⁺ sorted cells, 92.8 and 97.2%) when re-analysed by FACS analysis.

P. falciparum 3D7 parasites were transfected with a plasmid containing the Pfnek-4 coding sequence fused C-terminally to GFP under the control of its cognate (Pfnek-4) promoter, using ~1 kb of genomic sequence upstream of the Pfnek-4 translation initiation codon. Episomal propagation of the construct was maintained in the presence of antifolate drug pressure. Stage II transgenic gametocytes displayed a single punctuate Pfnek-4-GFP fluorescent structure, which coincides with or is located nearby one tip of the parasite and which is often closely associated with the nucleus (Figure 1A-B). The Pfnek-4-GFP protein appeared to accumulate in the cytosol of stage III gametocytes (Figure 1C) and remained present at high levels throughout the formation of mature stage V gametocytes (Additional file 1, panel A). Interestingly, while the bulk of examined schizonts did not show any detectable signal, as expected for a protein thought to be gametocyte-specific, fluorescence microscopy of asexual stage cultures consistently identified a small subpopulation (a few percents, see below) of schizonts with dots (single or doublets) of concentrated Pfnek-4-GFP protein associated with nuclei (Figure 1D). Fluorescence was never detected in ring and trophozoite stages. Noteworthy, all nuclei within a single schizont appeared to be associated with punctuate Pfnek-4-GFP fluorescence from early developing schizont to multinucleated schizonts (Additional file 1, panel B).

Western blot analysis of parasite extracts from gametocyte and asexual blood stage parasite cultures, using anti-GFP antibodies, detected a specific protein of ~62 kDa, consistent with the predicted mass of the Pfnek-4-GFP fusion protein (Figure 1E). This is in contrast to the previously reported lack of detectable Pfnek-4 protein in cultures of asexual 3D7 parasites, using chicken anti-Pfnek-4 IgY antibodies. Possible explanations for this discrepancy include: (i) the use of synchronized schizont stage parasites grown under conditions inducing gametocyte formation in the present series of experiments (in contrast to mixed asexual not under gametocytogenesis-inducing conditions in the earlier study 8), and (ii) higher sensitivity of the anti-GFP monoclonal antibodies versus the anti-Pfnek-4 IgY. Pfnek-4-GFP transfectants produced in the F12 clone background, a 3D7 subclone having lost the ability to produce gametocytes, allow to exclude that the Pfnek-4 signal from 3D7 asexual cultures is solely caused by contaminating gametocytes (Figure 1E). Anti-GFP reactive protein species of slightly higher molecular mass likely corresponds to Pfnek-4 phosphorylated forms as indicated by their susceptibility to λ protein phosphatase treatment (Figure 1E).

Nested RT-PCR using RNA isolated from asexual stages of the 3D7 parental clone and the gametocyte-less F12 clone yielded amplicons of the expected size for Pfnek-4 cDNA (Figure 2), the identity of which was verified by cloning and sequencing, further demonstrating that Pfnek-4 is not strictly gametocyte-specific, and that low mRNA levels are also present in untransformed asexual parasites.

Pfnek-4-GFP transgenic 3D7 and F12 parasites, either maintained under standard culture conditions or grown under conditions inducing commitment to sexual differentiation were subjected to FACS analysis. Figure 3 shows dot plots of samples of glutaraldehyde-fixed, propidium iodide (PI)-stained 3D7 and F12 transfectants grown under standard culture conditions, the GFP-positive (GFP⁺) schizonts consisting of 13.7% and 2.1% of the total schizont population, respectively; the high DNA content as measured by PI fluorescence allow to exclude that these are gametocytes. Under culture conditions stimulating sexual differentiation, 3D7 and F12 GFP⁺ schizonts significantly increased to 26.5% and 9.6% of the total schizont population, respectively. FACS analysis of live, 3D7 Hoechst-stained Pfnek-4-GFP transfectants confirmed the ~2-fold increase (paired t-test; p<0.0001) in the asexual multinucleated GFP⁺ parasite population in culture conditions inducing sexual differentiation (Table 1). It should be stressed that the GFP⁺ asexual parasites generated at high parasitaemia display a punctuate pattern and intensity of Pfnek-4-GFP fluorescence similar to the GFP⁺ asexual parasites observed in standard culture conditions, arguing against any major deregulation of the episomal promoter due to stress conditions (data not shown).

*Set as the parasite population with Hoechst fluorescence intensity > 2–3 fold higher than that of the ring-stage (1n) parasite population; values are mean ± SD calculated from three separate cultures obtained in two independent experiments at day-5 of culture.

**Paired t-test: two-tailed p value < 0.0001.

A cell-sorter approach was used to isolate live GFP-negative (GFP^-) and GFP-positive (GFP⁺) parasites from the same schizont population and examine the production of gametocytes when returned to in vitro tissue culture. Two independent experiments revealed that at day-4 after the FACS-sorted parasites were re-cultivated, the number of gametocytes observed by Giemsa-staining was about 10 times higher in GFP ⁺ −sorted cells (2.2 ± 1.2%) than in the GFP^--sorted cells (0.2 ± 0.1%) (Figure 4A). During the same four-day assay -period, asexual parasites increased to 5.0 ± 3.0% parasitaemia for the GFP⁺-sorted cells, and 6.5 ± 2.5% for GFP^--sorted cells, respectively. At day 3, 5 and 7 of culture, gametocytes mainly displayed the D shape, elongated shape and pointed ends that characterize stage II, III and IV, respectively (Figure 4B), consistent with development from sorted asexual schizont-stage parasites, rather than commitment following the sorting procedure or the sorting of gametocytes present in the asexual stage population. FACS analysis of Hoechst-stained parasite cultures undergoing sorting in experiments I and II, indicated that early stage gametocytes, i.e. GFP⁺ parasites with low DNA content, might account for approximately 5.3 and 12.9% of the sorted GFP⁺ cells, respectively.

Expression of the gametocyte-specific pfnek-4 gene in a subpopulation of asexual stage parasites was further investigated using a selection system containing an expression cassette for the bifunctional yeast protein cytosine deaminase/uridyl phosphoribosyl transferase (ScCDUP), providing efficient negative selection of Plasmodium upon treatment with the prodrug 5-fluorocytosine (5-FC) 1215. Transformed 3D7 parasites expressing the ScCDUP gene under control of the constitutive Hsp86 promoter (pScCDUP_Hsp86) were rapidly lost in the presence of 5-FC, as expected (Figure 5A). In contrast, transformed 3D7 parasites expressing ScCDUP driven by ~1-kb Pfnek-4 5’ flanking region (pScCDUP_Pfnek-4) showed a marginal reduction in growth rate in presence of 5-FC, consistent with the Pfnek-4 being active in only a small sub-population of asexual parasites. In parallel, to achieve a positive selection of parasites expressing Pfnek-4, the expression cassette was modified such that the human dihydrofolate reductase (hDHFR) was placed under control of ~1-kb 5’ flanking region from Pfnek-4 (phDHFR_Pfnek-4). Figure 5B shows that positive selection of transformed 3D7 parasites expressing hDHFR driven by the Pfnek-4 promoter with the antifolate drug WR99210 caused a marked reduction in parasitaemia growth rate, but was not sufficient to completely abolish proliferation. In contrast parental 3D7 parasites were fully susceptible, as expected.

The function of the nek-4 gene has been previously investigated in the rodent malaria parasite Plasmodium berghei; it was shown that disruption of Pbnek-4 does not affect asexual replication, gametocytogenesis and gametogenesis, but impairs differentiation of zygotes into ookinetes 8. To investigate a possible phenotype caused by the absence of a functional Nek-4 enzyme in the commitment of P. falciparum parasite to sexual differentiation, the pfnek-4 gene of the 3D7 parasite line was disrupted by single-crossover homologous recombination (Figure 6A). Clonal lines derived from two independent transfection experiments were established by limiting dilution, and their genotypes were analysed by PCR. The amplicon corresponding to the wild-type locus was not detected in clones cl3 and cl4, but was observed in untransfected 3D7 parasites. In contrast, amplicons diagnostic for the 5’ and 3’ boundaries of plasmid integration were detectable in the transgenic parasites (Figure 6B). The generation of pfnek-4^- mutant parasite clones demonstrates that the gene is not essential for replication in erythrocytes. Moreover, no effect on parasite growth rate of the pfnek-4^- mutant parasites was observed (data not shown). At day 12 to 14 of culture the number of gametocytes (~0.5-0.7%) appears undistinguishable from those of 3D7 wild-type parasite cultures and the majority of gametocytes were at stage IV and V as assessed by Giemsa-stained smears (Figure 6C). Most stage V gametocytes displayed the morphological characteristics of mature female gametocytes 16, indicating that pfnek-4 disruption does not affect the formation of mature female gametocytes.