The cell biology of malaria infection of mosquito: advances and opportunities

Abstract

Recent reviews (Feachem et al.; Alonso et al.) have concluded that in order to have a sustainable impact on the global burden of malaria, it is essential that we knowingly reduce the global incidence of infected persons. To achieve this we must reduce the basic reproductive rate of the parasites to < 1 in diverse epidemiological settings. This can be achieved by impacting combinations of the following parameters: the number of mosquitoes relative to the number of persons, the mosquito/human biting rate, the proportion of mosquitoes carrying infectious sporozoites, the daily survival rate of the infectious mosquito and the ability of malaria-infected persons to infect mosquito vectors.

This paper focuses on our understanding of parasite biology underpinning the last of these terms: infection of the mosquito. The article attempts to highlight central issues that require further study to assist in the discovery of useful transmission-blocking measures.

Introduction

Recent reviews (Feachem et al., ²⁰¹⁰; Alonso et al., ²⁰¹¹) have concluded that in order to have a sustainable impact on the global burden of malaria, it is essential that we knowingly reduce the global incidence of infected persons. To achieve this we must reduce the basic reproductive rate of the parasites to < 1 in diverse epidemiological settings. This can be achieved by impacting combinations of the following parameters: the number of mosquitoes relative to the number of persons, the mosquito/human biting rate, the proportion of mosquitoes carrying infectious sporozoites, the daily survival rate of the infectious mosquito and the ability of malaria-infected persons to infect mosquito vectors.

This paper focuses on our understanding of parasite biology underpinning the last of these terms: infection of the mosquito (Fig. 1). The article attempts to highlight central issues that require further study to assist in the discovery of useful transmission-blocking measures.

Figure 1.

Some of the earlier and significant steps in the analysis of the cell biology of sexual development of Plasmodium spp.

	I	II	III	IV	V	Gamete formation	Fertilization	Ookinete	Oocyst
Light microscopy1881-	(Hawking et al., 1971; Carter and Miller, 1979)					(Laveran, 1881)	(McCallum, 1897)		(Ross, 1897)
Induction mechanisms1926-	(Sinton et al., 1926; Kafsack et al., 2014; Sinha et al., 2014)					(Billker et al., 1998)
Electron microscopy1965-	(Sinden et al., 1978)					(Garnham et al., 1967)	(Sinden et al., 1976)	(Mehlhorn et al., 1980)	(Garnham, 1965)
Nuclear organization1973-					(Janse et al., 1986b)	(Sinden et al., 1976)		(Sinden and Hartley, 1985)	(Canning and Sinden, 1973)
Transcriptome2000-	(Lanfrancotti et al., 2007)				(Hayward et al., 2000)			(Hall et al., 2005)
Translation repression1990-					(Paton et al., 1993; Guerreiro et al., 2014)		(Mair et al., 2010)
Proteins first identified1974-	(Silvestrini et al., 2005)				(Vermeulen et al., 1985)	(Kumar and Carter, 1984)	(Liu et al., 2008)	Vermeulen et al., 1985)	(Krotoski et al., 1974)
Proteome2005-					(Khan et al., 2005)	(Talman et al., 2014)		(Hall et al., 2005)
Metabolome2014-	(Lamour et al., 2014)

From vertebrate host to insect vector

Noting the apparent differences in the sexual biology of malaria parasites in the subgenera Laverania and Plasmodium, where comment or data specifically refers to just one subgenus it will be stated explicitly, and where it is felt by the author to apply to both subgenera it will not be stated.

Transmission from host to vector is the exclusive role of the intra-erythrocytic gametocytes. The development of which can be divided into induction, maturation and gamete formation. The often extended periods of gametocyte induction raises the question whether any generalization can be made as to a physical site where gametocytogenesis is triggered. Rarely it is appreciated that there is evidence that merozoites released directly from the liver schizont can produce gametocytes (Suhrbier et al., ¹⁹⁸⁷; Sinden et al., ¹⁹⁹⁰), and that in the ‘ancestral’ haemoproteids all the merozoites released from the tissue schizonts make gametocytes, i.e. induction is not necessarily a consequence of prior blood infection. Past research (Smalley et al., ¹⁹⁸¹) showed the young falciparum gametocytes were found in the bone marrow. Trager and Gill (1992) recognized the eminent logic of the slow-growing gametocytes of this species (subgenus Laverania), initiating their protracted development in hematopoietic tissues. These observations have been elegantly extended using both molecular techniques (Aguilar et al., ²⁰¹⁴; Joice et al., ²⁰¹⁴) and in vitro systems (Farfour et al., ²⁰¹²; Tiburcio et al., ²⁰¹²; Peatey et al., ²⁰¹³). The latter illustrate how both the rigidity of the red blood cell (RBC) infected by the elongated immature P. falciparum (Pf) gametocyte, and surfin expression on the infected RBC (iRBC), might correlate with retention of the immature gametocyte in the complex extravascular environment of the bone marrow. The former attribute is unlikely to apply to the spherical parasites of the subgenus Plasmodium that lack the elongate sub-pellicular cytoskeleton of Laverania. When considering induction of gametocytogenesis in the peripheral blood, current in vitro evidence suggests that in P. falciparum induction occurs in the trophozoite of the asexual generation preceding gametocyte formation, and that the merozoites from any one committed schizont are predetermined to be either male or female (i.e. multiple fate ‘decisions’ have already been made prior to merozoite formation) (Bruce et al., ¹⁹⁹⁴). A marker of this commitment has recently been described as Pf10_0164 (ETRAMP 10.3) (Brancucci, unpubl.). In the rodent malaria parasite P. berghei gametocyte determination possibly occurs following invasion by the merozoite (Mons et al., ¹⁹⁸⁵). Myriad potential external inducers of gametocytogenesis have been described, all of which may be broadly classified as ‘stress related’, but molecular characterization of any consensus induction pathway still eludes us (Sinton, 1926; Bhasin and Trager, ¹⁹⁸⁴; Dyer and Day, ²⁰⁰⁰; Carter et al., ²⁰¹³). Future experiments to understand sexual/asexual differentiation should at least entertain the hypothesis that sexual development is the ancestral – and therefore default – pathway.

Not knowing the molecular properties of the inducers, it is unsurprising that we do not know the receptors initiating the signalling pathways for gametocytogenesis. Things we do know however are that the down-regulation of the histone deacetylase gene pfHda2, the heterochromatin protein PfHP1 and the ABC transporter-encoding gene gabcg2 increase the proportion of gametocytes in bloodstage infections (Alano, 2015; Brancucci et al., ²⁰¹⁴; Coleman et al., ²⁰¹⁴; Tran et al., ²⁰¹⁴). Perhaps the closest we have come to understanding these pathways is through the production and characterization of mutations resulting in the loss of gametocyte formation. The genetic regulation of gametocytogenesis was first appreciated in studies correlating the loss of chromosomal integrity (notably P. falciparum chromosome 9) with the loss of sexual potential (Birago et al., ¹⁹⁸²; Janse et al., ¹⁹⁸⁹; Day et al., ¹⁹⁹³; Ikadai et al., ²⁰¹³), but it was transposon-mediated mutagenesis (Ikadai et al., ²⁰¹³) that identified 16 loci, nine of whose disruption led to loss of any identifiable (stage I) gametocytes [this group includes a member of the ap2 family (PF13_0097)] and seven loci essential to formation of stage II gametocytes. Of these 16 ‘mutagenized’ loci only five were successfully complemented. Perhaps, unsurprisingly, the former nine loci were classifiable as genes encoding regulators/signalling moieties, and the latter seven included genes with potential roles in the formation of the extensive cytoskeletal structures defining Pf gametocyte morphology. These conclusions are consistent with parallel observations (Young et al., ²⁰⁰⁵) that noted up-regulation of 246 gametocyte-specific gene classes in the mature gametocytes – shown to include kinase/phosphatase enzymes (Guttery et al., ^2012b; Brochet et al., ²⁰¹⁴). Genes that have been considered of particular note include pfgdv1 (Eksi et al., ²⁰¹²), pfgig, pfgly and the plasma membrane transporter NPT1 (Baker, 2010; Boisson et al., ²⁰¹¹). Elegant recent studies have confirmed a pivotal role for ap2-g in gametocytogenesis (Kafsack et al., ²⁰¹⁴; Sinha et al., ²⁰¹⁴); transcription of ap2-g is regulated by PfHP1-H3K9me3 histone modification (Kafsack et al., ²⁰¹⁴). Pathways regulating gametocytogenesis have been reviewed (Morahan and Garcia-Bustos, 2014); these authors suggest ap2g may regulate pfgdv1, thence pfnek4 and nek2 – genes that are additionally regulated by extracellular factors via cyclic adenosine monophosphate-mediated pathways. Interestingly, the protein phosphatase PPM2 (see Table 1) has been shown to have a positive role in the determination of sex ratio (female bias) (Guttery et al., ²⁰¹⁴). Have we identified all the key regulator(s) of sexual differentiation, certainly not, but we are now developing the tools for doing so.

Table 1.

Suggested stages of Plasmodium development in the mosquito vector at which identified protein phosphatases and protein kinases are active

	Female sex allocation	Male gamete exflagellation	Zygote/ookinete differentiation	Ookinete movement	Ookinete invasion	Oocyst development	Sporogony
Protein phosphatases	PPM2	PPM1	PPKL PPM2	PPKL	SHLP1	PPM5	PTPLA
Protein kinases	–	SRPK CDPK4 MAP2	NEK2 NEK4 GAK PK7	CDPK3	CDPK3	GAK PK7	CDLK

Data from Tewari et al. (2010) and Guttery et al. (2014).

CDLK, cyclin-dependent-like kinase; CDPK3,4, calcium-dependent protein kinase; GAK, cyclin g-associated kinase; MAP2, mitogen-activated protein kinase; NEK2, NEK4, NimA-related kinase; PK7, protein kinase; PPKL, protein phosphatase containing an N-terminal β-propeller formed by kelch-like motifs; PPM1, PPM2, PPM5, metallo-dependent protein phosphatases; PTPLA, protein tyrosine phosphatase-like A; SHLP1, Shewanella-like protein phosphatase; SRPK, SR protein kinase.

Despite long-established protocols for the culture of P. falciparum gametocytes en masse (Ponnudurai et al., ¹⁹⁸³; Sinden et al., ¹⁹⁸⁴) and the detailed descriptions of the complex morphological changes occurring in the skeletal, cytoplasmic and nuclear organization of the parasites (Sinden et al., ¹⁹⁷⁸; Dixon et al., ²⁰¹²; Hliscs et al., ²⁰¹⁵) and in the organization of the mitochondrion and apicoplast (Okamoto et al., ²⁰⁰⁸), we remain disturbingly ignorant of the molecular profile of the maturing gametocyte. Transcriptomic and proteomic studies (Hall et al., ²⁰⁰⁵; Khan et al., ²⁰⁰⁵; Tao et al., ²⁰¹⁴) have revealed, for example, 174 male- and 258 female-enriched proteins in mature (dimorphic/separable) P. falciparum gametocyte preparations. The gametocyte metabolomes are incompletely described (Lamour et al., ²⁰¹⁴); indirect observations, e.g. drug sensitivity, currently suggest the immature gametocytes (stages I–III) are more readily killed by inhibitors of RNA and protein synthesis (Sinden and Smalley, 1979) and schizonticides (e.g. chloroquine) than the mature (stage IV–V) gametocytes (Smalley, 1977; Sinden, ¹⁹⁸²; Targett et al., ²⁰⁰¹), an observation previously linked to the substantial reduction in protein synthesis/ribosome population seen in the more mature forms (Sinden et al., 1976; 1978^,). To date one of the most interesting of the unexplored aspects of gametocyte biology lies in protein translation for two reasons: first, the mature female gametocyte (which retains a ‘normal’ ribosome population) accumulates a reported 169–731 species of translationally repressed gene transcripts (Mair et al., ²⁰⁰⁶; Guerreiro et al., ²⁰¹⁴), which may be localized in cytoplasmic ‘granules’ (Thompson and Sinden, 1994; Vervenne et al., ¹⁹⁹⁴) where they are presumably complexed with the translational regulators DOZI and CITH (Mair et al., 2006; 2010^,; Guerreiro et al., ²⁰¹⁴); but what is often overlooked is the evidence that the structure of the ribosome population itself changes as the parasite moves from vertebrate to insect host – the population transitions from A to S-form ribosomes in which different alleles encoding the SsuRNAs 30 and 31 are expressed (Waters et al., ¹⁹⁸⁹). Why these changes occur and whether they have any relevance to the significant role of translational regulation in gametogenesis has yet to be tested.

Structural studies have shown that the mitochondrion enlarges significantly throughout gametocyte maturation (Aikawa et al., ¹⁹⁶⁹; Sinden et al., ¹⁹⁷⁶; Jensen, ¹⁹⁷⁹; Okamoto et al., ²⁰⁰⁸). The seminal early electron microscopy studies of Aikawa et al. (1969) demonstrated beyond question that the mitochondrion (at the time considered to be numerous organelles) of the gametocytes was morphologically distinguished from those of the asexual bloodstages by the presence of prominent tubular cristae (Aikawa et al., ¹⁹⁶⁹); subsequent studies revealed that in the ookinete and oocyst stages the cristae were even more prominently developed. Jim Jensen (1979) was the first to suggest that the multiple profiles of mitochondria seen in the gametocytes were in fact derived from a single network-like organelle. By contrast the apicoplast remains comparatively small and closely associated with the nucleus. Recent application of elegant whole-cell confocal studies combined with the use of organelle-specific fluorochrome-tagged markers has emphasized how the mitochondrion and apicoplast subsequently proliferate in the oocyst (Stanway et al., ²⁰⁰⁹). Recent metabolomic studies (Lamour et al., ²⁰¹⁴) suggest that oxidative – and lipid – metabolism may differ significantly between the asexual and sexual bloodstage parasites. Recognizing the female gametocyte provides the majority of the biomass of the zygote/ookinete, this expansion of energy provision is understandable, but the male gametocyte forms eight microgametes devoid of both plastids, suggesting in the male cell that it performs a vital function during gametocytogenesis or during the dramatic events of gamete formation in the mosquito? It has been shown that the apicoplast provides isoprenoids essential for gametocytogenesis (Wiley et al., ²⁰¹³), and up-regulation of glyoxalase provides defence against oxidative stress (Okamoto et al., ²⁰⁰⁸) that has recently been shown to be a component of the mosquitoes' responses to malarial infection (Shrinet et al., ²⁰¹⁴). Up-regulation of genes encoding type II fatty acid, and 15 of 16 TCA enzymes in the mature gametocyte are also consistent with the cytological observations. In marked contrast it is the enzymes of the glycolytic pathway that are among the most abundant proteins in the free swimming microgamete (Wass et al., ²⁰¹²; Talman et al., ²⁰¹⁴). The male gamete lacks both plastids (Sinden et al., ¹⁹⁷⁶; Okamoto et al., ²⁰⁰⁸); as a consequence their inheritance is maternal (Creasey et al., ¹⁹⁹³), and the male gametes rely entirely on hexose import and glycolysis for energy to drive their vigorous swimming (Slavic et al., ²⁰¹¹; Talman et al., ²⁰¹⁴). Reflecting on this body of data, it is interesting to note that some of the antimalarial drugs, e.g. atovaquone with known mitochondrial targets such as the cytochrome B1 complex and adenosine triphosphate (ATP) binding cassette transporter (Fry and Pudney, 1992; Rijpma et al., ²⁰¹⁴), are exquisitely active against transmission of parasites from vertebrate to mosquito hosts, and the ookinete in particular (Fowler et al., ¹⁹⁹⁵; Delves et al., ²⁰¹²).

An outstanding dilemma in the understanding of gametocyte biology is whether the pattern of metabolism revealed by the studies on slow maturing/long-lived gametocytes of the subgenus Laverania (Smalley and Sinden, 1977; Eichner et al., ²⁰⁰¹; Bousema and Drakeley, ²⁰¹¹) is representative of the subgenus Plasmodium. Evidence suggesting this may be the case stems from the observations that purified populations of viable mature infectious gametocytes of the rodent malaria parasites, like those of P. falciparum, can be prepared by treating mixed bloodstage infections with schizonticides (Beetsma et al., ¹⁹⁹⁸; Rodrıguez et al., ²⁰⁰²).

The impact of gametocytes on the biology of host/parasite–vector interactions has for decades provoked intriguing discussion and experimentation. ‘Is the infectious host less responsive to mosquito bites?’ (Rossignol et al., ¹⁹⁸⁵); ‘Is it more attractive to mosquitoes?’ (Batista et al., ²⁰¹⁴); ‘Is the sporozoite-infected mosquito more likely to probe?’ (Anderson et al., ¹⁹⁹⁹). At the cellular level the oft-raised question whether mature (infectious) gametocytes are preferentially retained in the capillaries of the skin where they would be accessible to the mosquito remains an interesting but still unresolved question. The potential now to use late gametocyte promoters to regulate luciferase expression in the mature gametocyte, combined with IVIS technology, might provide the quickest method of resolving this question. Should gametocytes exhibit this tropism, it would seriously impact the methodologies required to describe the infectious reservoir. However, recognizing the now widely acknowledged fact that ∼ 80% of persons infected with P. falciparum will also be gametocyte positive (Muirhead-Thomson and Mercier, 1952a,b^,; Muirhead-Thomson, ¹⁹⁵⁷; Bousema and Drakeley, ²⁰¹¹) does raise the question as to how any further knowledge, beyond correlating peripheral gametocytaemia with the probability of infecting mosquitoes (Churcher et al., ²⁰¹³), is going to change any control strategy (Bousema et al., ²⁰¹²) – which must surely treat every infected individual as if they were infectious to the vector.

Sexual development in mosquito

For this author, the events of malaria microgametogenesis (exflagellation) in the mosquito vector remain one of the most beautifully orchestrated and dramatic developmental transformations in eukaryote biology.

Induction of exflagellation

Our understanding of the induction of gametogenesis at the molecular level has advanced little since its discovery in 1998 (Billker et al., ¹⁹⁹⁸). We know two conditions must prevail: a fall in temperature of > 5°C (Roller and Desser, 1973; Sinden and Croll, ¹⁹⁷⁵), suggesting an as yet undescribed role for heat shock proteins in protein remodelling, and the presence of elevated levels of the mosquito waste product xanthurenic acid (XA) (Billker et al., ¹⁹⁹⁸). The fall in temperature alone may trigger the secretion of the osmiophilic bodies leading to RBC breakdown (see below).

In marked contrast the downstream pathways regulating the component events of gamete formation, namely – escape from the RBC and molecular remodelling of the gamete surfaces and exclusively in the male cell – genome replication, genome separation and axoneme assembly, are now extensively although still incompletely characterized. Current models of the signalling cascade regulating gametogenesis have been outlined (Baker, 2010; Morahan and Garcia-Bustos, ²⁰¹⁴; Sinha et al., ²⁰¹⁴). Although prior data suggested raised extracellular pH (∼pH 8.0) is a useful laboratory ‘inducer’ (Nijhout and Carter, 1978), the overall pH of the bloodmeal only rises by some 0.2 pH units in the 24 h following bloodmeal ingestion (Billker et al., ¹⁹⁹⁷), suggesting that a transient rise in intracellular pH may form part of the signalling cascade. Studies on the role of gametocyte ion pumps (Kawamoto et al., 1992; 1993^,; Kawamoto, ¹⁹⁹³) nonetheless suggested that the cytoplasmic pH of the gametes might change both rapidly and significantly.

Activation (directly or indirectly) of guanylyl cyclase by XA produces cGMP that is in turn converted to guanidine 5 monophosphate by phosphodiesterase (knockout of pdeδ ablates exflagellation). cGMP activates PKG (McRobert et al., ²⁰⁰⁸) that in turn maintains elevated levels of cytosolic Ca²⁺ (see below) (Brochet et al., ²⁰¹⁴), which regulates rounding up (i.e. increase in volume = water uptake?) of the cells and the expression of at least some translationally repressed surface proteins. Deletion of pbcax, however, prevents the onward development of the rounded female cell into an ookinete (Guttery et al., ²⁰¹³).

XA additionally stimulates PIPLC to produce PIP2 and IP3; these in turn give rise to elevated cytoplasmic calcium just 10 s after activation (Billker et al., ²⁰⁰⁴; Sebastian et al., ²⁰¹²) which, through CDPK4, results in DNA replication and axoneme assembly in the male cell, and translation release in the female. Downstream of CDPK4, and mediated by NEK1 and NEK3, MAP2 and SPRK regulate the motility of the axonemes and cytokinesis of the microgametocyte (Tewari et al., ²⁰¹⁰). The writer is amused by the observation that a viagra-like molecule (Zaprinast) can obviate the need for XA in male gamete release (McRobert et al., ²⁰⁰⁸).

Escape from the RBC

Both male and female cells increase in volume (Sinden and Croll, 1975); thus, stressing the host cell, the males additionally release very motile gametes that undoubtedly assist the disruption of a weakened host cell (Sologub et al., ²⁰¹¹; Deligianni et al., ²⁰¹³; Wirth et al., ²⁰¹⁴). Escape is mediated by the secretion of the osmiophilic bodies (more abundant in the sessile female; Sinden et al., ¹⁹⁷⁶). The biogenesis and lytic function of the osmiophilic bodies is dependent on the following moieties Pfg377: exclusively in females (Alano et al., ¹⁹⁹⁵; Olivieri et al., ²⁰¹⁴), MDV-1/PEG3 (Lanfrancotti et al., ²⁰⁰⁷; Lal et al., ²⁰⁰⁹; Ponzi et al., ²⁰⁰⁹), pbGEST (Talman et al., ²⁰¹¹) cysteine and aspartic protease, and a perforin-like protein 2 (Wirth et al., ²⁰¹⁴).

Genome replication in male gametocyte

The ploidy and replication of the gametocyte genome has been a controversial topic, resolved in large part by the studies of Cornelissen and Janse in the 1980s, who, using cytochemical techniques, showed that mature gametocytes like the merozoite are haploid, and that the male cell replicated its DNA three times in brief period (∼10 min) between induction and microgamete release – exflagellation (Janse et al., 1986a,b^,; Cornelissen, ¹⁹⁸⁸). This replication is cdpk4 dependent (Guttery et al., ^2012a). In cdc20 and map2 knockout lines subsequent mitotic separation of the replicated genomes in the male cells is blocked after metaphase (Guttery et al., ^2012a). The incredible speed of replication suggested that thousands of replication forks are distributed across the 14 chromosomes. DNA polymerase-α mediates replication, a process sensitive to aphidicolin but not mitomycin C (Janse et al., ^1986a). Electron microscopy revealed three conventional mitotic divisions within the single large male nucleus at ∼ 3, 8 and 15 min following activation (conventional with the no exception that at no point do the chromosomes condense! – a fact that might correlate with the very rapid replication of the genomes). Although this observation is entirely consistent with the absence of histone 1 from the genome, it also raises fascinating questions as to how the initial 14, and final 112, filamentous chromosomes (whose telomeres are likely embedded in the nuclear envelope) can be moved by the kinetochores on the spindle microtubules with such speed and precision without breaking. The female gamete like the originating immature female gametocyte is haploid, but the intervening nuclear events are unclear; data suggested that the DNA content may rise above 1°C (Cornelissen, 1988) and descriptions of spindle-like structures in the developing (putative) macrogametocyte are either erroneous or suggest the female genome is somehow modified (Sinden et al., ¹⁹⁷⁸).

Axoneme assembly in male gametocyte

The cytoplasm of the mature microgametocyte contains large quantities of tubulin, which upon activation rapidly polymerize into microtubules. Polymerization begins with the de novo assembly of eight basal bodies (apparently as two conventionally orientated orthogonal tetrads). Basal body structure/patterning and function requires SAS6 expression (Marques et al., ²⁰¹⁵). Axoneme polymerization, being intracytoplasmic, is independent of intra-flagellar transport and consequentially very rapid; all eight axonemes reaching 14 μm in length in just 10 min, when formed, lay coiled around the persistent envelope of the single nucleus. Axonemes are largely of conventional 9 + 2 design (Sinden et al., ²⁰¹⁰). The formation of the central pair is severely, although incompletely, disrupted when the armadillo repeat protein PF16 is deleted (Straschil et al., ²⁰¹⁰). At the onset of exflagellation, the axonemes become motile and swim basal body first out of the parental cell. The prior attachment of each basal body to a mitotic spindle pole ‘ensures’ each basal body drags a haploid genome into the gamete; the gamete then swims through the dense mosquito bloodmeal using alternating periods of fast and slow rotary ‘ambidextrous’ sinusoidal waves (Sinden and Croll, 1975; Wilson et al., ²⁰¹³). Microgamete motility is entirely dependent on the import of hexose from the bloodmeal (Slavic et al., ²⁰¹¹).

Protein remodelling of the gamete surfaces

The surface proteins of the male and female gametes differ both from each other and from that of the parental gametocytes. Sexual differences between gametes, e.g. expression of P48/45, P230 and HAP2 in male and P47 and the LAP/CCCP proteins in female (van Schaijk et al., ²⁰⁰⁶; Raine et al., ²⁰⁰⁷; Liu et al., ²⁰⁰⁸), may reflect functions in fertilization, and differences between gametocyte and gametes may indicate functions related to gamete survival in the challenging (both immunologically and enzymatically) environment of the bloodmeal. Expression of some 169–731 proteins in the female gamete, including the surface proteins P25 and P28 on the fertilized zygote, is dependent on the release of their mRNA from translation repression (Mair et al., 2006; 2010^,; Guerreiro et al., ²⁰¹⁴), a process in part regulated by CDPK1 (reviewed by Morahan and Garcia-Bustos, ²⁰¹⁴). Early defence of the gametes/zygote by expression of a complement regulator factor H-like protein (Simon et al., ²⁰¹³) inactivates complement ingested in the bloodmeal (Grotendorst and Carter, 1987; Margos et al., ²⁰⁰¹; Simon et al., ²⁰¹³); however, the gametes have no defence against antibody attack targeted against proteins secreted onto the gamete surface (e.g. P45/48; P230), making these molecules excellent targets for ‘contraceptive’ vaccines (Carter et al., ²⁰⁰⁰; Sauerwein, ²⁰⁰⁷). P230 is reportedly responsible for the attachment of the male gametes to RBCs in the bloodmeal (Templeton et al., ¹⁹⁹⁸), although what possible advantage this is to the parasite escapes this writer.

Fertilization

Fertilization of malaria gametes was first recognized in 1897 (McCallum, 1897) and described at the ultrastructural level some 40 years ago (Sinden et al., ¹⁹⁷⁶; Aikawa et al., ¹⁹⁸⁴). Gamete recognition and binding is mediated through GPI-anchored P47 on female interacting with P230, complexed to GPI-anchored P48/45 on the male cell. The ‘ambidextrous’ ability of P230 to bind both male- and female-specific moieties may explain the formation of intimately bound same-sex and mixed-sex clusters of gametes/gametocytes in vitro (Janse et al., ¹⁹⁸⁵; Sinden and Hartley, ¹⁹⁸⁵; Sinden et al., ¹⁹⁸⁵). The fusion of the gamete membranes is now known to be mediated by male-specific HAP2/CSC1 (Liu et al., ²⁰⁰⁸; Mori et al., ²⁰¹⁰) and can be severely disrupted by anti-HAP2 antibodies (Blagborough and Sinden, 2009; Miura et al., ²⁰¹³). Fusion of the gamete nuclei occurs within 1 h of plasma membrane fusion (Aikawa et al., ¹⁹⁸⁴), and is preceded by the de-condensation of the chromosomes in the male nucleus. There is substantial transcriptional up-regulation of the zygote genome via AP2O (Yuda et al., ²⁰⁰⁹; Akinosoglou et al., ²⁰¹⁵). We do not yet have definitive evidence as to when expression of male-inherited genes begins, but zygote development reportedly can progress in the absence of RNA polymerase II mediated transcription (Guerreiro et al., ²⁰¹⁴).

Within the mosquito the parasite is from the time of zygote formation until budding of daughter sporozoites from the oocyst a polyploid cell, but cytological study shows that the genetic organization of the parasite, following fertilization, is immediately returned to a haploid state, within a single nucleus, by a conventional two-step meiotic division producing four haploid genomes in the ookinete (Janse et al., ^1986b; Sinden, ¹⁹⁹¹). Thus, any one oocyst has the potential to contain four discrete and potentially recombinant genotypes; this has been confirmed experimentally (Ranfordcartwright et al., ¹⁹⁹¹).

It must be remembered that the process of gamete production and fertilization occurs in an environment composed essentially of unaltered host blood to which the standing components of the mosquito gut have been added, among which the trypsin-like proteases are significant factor (Muller et al., ¹⁹⁹³). The introduction of proteolytic enzymes to the bloodmeal degrades vertebrate complement in the first 3 h (Margos et al., ²⁰⁰¹), a change that is intriguingly mirrored by the short-lived initial defence of the parasite against complement (see above) – a truly fascinating interplay between host, vector and parasite.

Host antibodies persist, at potentially lethal concentrations, in the bloodmeal for at least 24 h, and can escape through the midgut epithelium into the haemocoele (possibly enhanced by the rupture of the epithelium by the invading ookinetes; Han and Barillas-Mury, ²⁰⁰²). Convincing evidence has been provided that cytokines, which can be present at significant levels in the hosts' blood especially following synchronous schizogony (Motard et al., ¹⁹⁹⁰), can, when present in the bloodmeal, significantly suppress both gametogenesis and oocyst infection (Dearsly et al., ¹⁹⁹⁰; Naotunne et al., ¹⁹⁹³). The duration for which active host cytokines persist in the bloodmeal is however not known. Interesting data suggesting a dialogue between vertebrate insulin and mosquito receptors (Pakpour et al., ²⁰¹²) raise the fascinating possibility that the antiparasitic activity of mosquito may be both directly and indirectly mediated.

The mosquito places both physical and immune barriers in the path of the parasite. The physical barrier to infection/invasion offered by the peritrophic matrix (PTM) was revealed when inhibitors of parasite chitinase (e.g. allosamidin) were shown to decrease oocyst numbers (Shahabuddin et al., 1993; 1995^,; Shahabuddin and Kaslow, 1994a,b^,; Zieler et al., 1999; 2000^,). Humoral insect defence mechanisms however remain potent inhibitors when the ookinete crosses the epithelial layer (James, 1928; Dimopoulos, ²⁰⁰³; Osta et al., ²⁰⁰⁴; Christophides, ²⁰⁰⁵; Povelones et al., ²⁰¹¹; Shrinet et al., ²⁰¹⁴).

The ookinete

Notwithstanding the critical genetic events of meiosis (Sinden, 1985; Sinden and Hartley, ¹⁹⁸⁵) – recently shown to be regulated by NIMA kinases, NEK2 and 4 (Reininger et al., 2005; 2009; 2012^,,) and reviewed in Morahan and Garcia-Bustos (2014) – the key role of differentiation of the zygote into the ookinete is to produce a polarized motile cell that can escape the increasingly potent digestive enzymes of the bloodmeal to which the ookinete is susceptible (Gass, 1977; Gass and Yeates, ¹⁹⁷⁹). This differentiation is both fertilization dependent and reliant upon PK7, PPKL (protein phosphatase with Kelch-like domains) and GA kinase activities (Guttery et al., ^2012b; Philip et al., ²⁰¹²; Morahan and Garcia-Bustos, ²⁰¹⁴). Although formation of the single ookinete from each zygote in very large part mirrors the assembly of both the merozoite and the sporozoite from their respective ‘schizonts’, we know less of the cellular events. Although the assembly and cytoskeletal polarity of apicomplexan ‘zoites’ is clearly determined by the position of a single nuclear spindle plaque/microtubule organizing centre (Dubremetz, 1975), it is challenging to understand how/whether the ookinete nucleus, which contains four recombinant haploid genomes/spindle plaques, similarly dictates the polar assembly of the single ookinete cytoskeleton (unless polarity was determined by the early polarization of the zygote nucleus and formation of the perinuclear array of cytoplasmic microtubules; Aikawa et al., ¹⁹⁸⁴). The functions of the elaborate, and ookinete-specific, anterior apical collar, beyond providing apparent rigidity to the ookinete penetrative apparatus, have yet to be determined. The functions of those components driving motility are better but still incompletely understood. Conventional thinking suggests motility is dependent on CDPK3, Ca⁺⁺ and cGMP signalling pathways (Siden-Kiamos and Louis, 2008; Moon et al., ²⁰⁰⁹; Guttery et al., ^2012b), controlling the activity of the extensively described actin-myosin glideosome motor (Raibaud et al., ²⁰⁰¹; Opitz and Soldati, ²⁰⁰²; Kan et al., ²⁰¹⁴) that is linked to transmembrane surface adhesins such as CTRP, a large protein composed of seven thrombospondin and six von Willebrand factor A domains, of which only the proximal A domains appear to be essential to locomotion (Ramakrishnan et al., ²⁰¹¹). The validity of the conventional glideosome model has recently been challenged by controversial data from Toxoplasma suggesting neither actin nor myosin is essential for locomotion (Andenmatten et al., ²⁰¹²). The spiral architecture of the microtubular ookinete skeleton dictates both the overall shape of the ookinete and the direction of movement (essentially a left-handed corkscrew; Vlachou et al., ²⁰⁰⁴; Kan et al., ²⁰¹⁴), which may be optimal to drive the parasite through the particulate and increasingly ‘viscous’ bloodmeal.

Whether the movement of the ookinete is in any way directed to the midgut epithelium is not known, but interaction with the complex layers of the midgut epithelium is certainly not understood. The epithelium is composed of three layers: the chitinous PTM, which is secreted at the time of bloodmeal ingestion (Shao et al., ²⁰⁰¹); the tubular network (Zieler et al., ²⁰⁰⁰); and the cell epithelium monolayer. Traversal of the PTM (if formed) is assisted by the secretion of one or more chitinase enzymes, one secreted as a pro-enzyme that is reportedly activated by the mosquito trypsins (Shahabuddin et al., ¹⁹⁹³). Addition of the chitinase inhibitor allosamidin to bloodmeals containing P. gallinaceum or P. falciparum reduced oocyst numbers (Shahabuddin et al., ¹⁹⁹³), and in P. berghei knockout of chitinase 1 reduces infectivity significantly (Dessens et al., ²⁰⁰¹).

Microscopic evidence has been interpreted as suggesting that ookinetes preferentially invade/lyse epithelial cells at the posterior of the midgut, and close to intercellular junctions. Early work suggested that the ‘preferential’ oocyst distribution at the posterior midgut was simply due to the sedimentation of the iRBC in the bloodmeal during the prolonged period when the bloodfed female mosquitoes rest ‘head-up’ after the feed. The validity of this early observation has, however, been questioned (Kan et al., ²⁰¹⁴). Studies in the 1990s showing that ookinetes in the midgut wall frequently co-located with v-ATPase activity led workers to suggest that ookinetes invaded a specific subclass of epithelial cell, which they termed the ‘Ross Cell’ (Shahabuddin, 2002). The study, however, failed to recognize that v-ATPase might be up-regulated following, and not before, invasion. To some, the most compelling data suggesting there is a specific epithelial ligand bound by the ookinete comes from antibody blockade studies where anti-P25, -P28 (Gozar et al., ¹⁹⁹⁸; Miura et al., ²⁰¹³), enolase (through interaction with host plasminogen; Ghosh et al., ²⁰¹¹) and AgAPN1 antibodies (Armistead et al., ²⁰¹³) can inhibit invasion. However, it must be recognized that coating abundant surface moieties with antibodies does not necessarily identify specific ligands. Other molecules reducing ookinete invasion by unknown mechanisms when added to the bloodmeal include PLA2 (Rodrigues et al., ²⁰⁰⁸), bee venom (Moreira et al., ²⁰⁰²) and the peptides SHIVA (Boisbouvier et al., ¹⁹⁹⁸), and SM1 (Fang et al., ²⁰¹¹). Early data (Rosales-Ronquillo and Silverman, 1974) strongly suggested that ookinetes can recognize and lyse RBC – exactly the same mechanisms now believed to mediate midgut invasion, it is thus tempting to speculate that the host molecules inducing secretion of the lytic contents of the abundant micronemes are shared between the RBC and mosquito epithelial cells (and are perhaps somewhat less specific than commonly hypothesized). In our efforts to develop ever improved transmission-blocking vaccines, we must recognize we know too little of the molecular architecture and immunogenic potential of the ookinete surface.

Twelve to 36 h following ookinete–epithelium interaction, the secretions from micronemes lyse the host cell and the ookinete migrates into the resulting ‘cadaver’. Should many ookinetes invade in the same locality epithelial destruction is considerable (Ecker et al., ²⁰⁰⁷), with the consequent opportunity of transfer of bloodmeal contents to the haemocoele, significant parasite density-dependent mosquito death may occur at this time. The writer is fascinated by the question of how the ookinete escapes, in a directed manner, the sac of the lysed midgut cell, particularly noting the very different molecular profiles presented by the cytoplasmic membrane surfaces potentially encountered. The ‘time bomb’ theory of midgut invasion (Han and Barillas-Mury, 2002) suggests the lysed cells will be ejected from the epithelium by a draw-string repair mechanism, involving substantial reorganization of the epithelial–cell actin network (Shiao et al., ²⁰⁰⁶); thus, escape from the lysed cell is essential for ‘infection’ to be a success. We do not yet understand what, on reaching the luminal/midgut face of the basal lamina, induces the emergent ookinete to become immotile and begin differentiation into an oocyst, but it is tempting to suggest that the molecular recognition of the collagen-rich basal lamina is a regulatory factor. Note, however, that the ookinete is very motile in Matrigel® (Corning Life Sciences), an artificial intercellular matrix rich in (mouse) laminin, enactin and collagen (Moon et al., ²⁰⁰⁹).

Irrespective of the extent of the damage caused by the ookinete to the epithelial cells, the mosquito mounts a significant response to the ookinete and developing oocyst within the epithelial layer. Current evidence in a rapidly changing area of study suggests that, in a nitration-dependent event (Molina-Cruz et al., ²⁰⁰⁸), the mosquito thioester (complement-like) molecule TEP-1 forms a complex with LRIM1 and APL1 and the ookinete surface, causing the lethal deposition of melanin on the parasites' membrane systems (Sinden and Garnham, 1973; Fraiture et al., ²⁰⁰⁹; Povelones et al., ²⁰¹¹). The presence of the protein Pfs47 on the parasite surface reportedly reduces this nitration-dependent attack mechanism (Molina-Cruz et al., ²⁰¹³). Other PKC-mediated antimalarial immune response pathways have been identified (Pakpour et al., ²⁰¹³).

Although infection of the mosquito is theoretically ‘completed’ following the induction of sporogony in the ookinete under the basal lamina, the parasite nonetheless undergoes further unique and poorly understood developmental steps that intrigue the writer, topmost being the restructuring of its ribosome population. In the early 1990s pioneering work in the McCutchan laboratory recognized that the sSUrRNA from sporozoites (S) and asexual blood stages (A) differed (Zhu et al., ¹⁹⁹⁰; Waters, ¹⁹⁹⁴). Subsequently, it was demonstrated that shortly after fertilization a third form (O) of the 18S RNA is expressed to be replaced in the late oocyst by the S form (Li et al., ¹⁹⁹⁷). It is interesting to note that the small ribosomal subunit protein S1 has the highest transcript abundance in the gametocyte and S2 in the ookinete; whether this is related to the skeletal 18S rRNA changes is unclear. What biological critical advantage is gained from these fundamental and fascinating changes in ribosome design are still unresolved.

The future?

Over the past 50 years the advances in our understanding of malaria cell biology have been truly amazing, what of the next 50 years? Ongoing metabolomic studies are opening a broader understanding of parasite function and might be expected to play a central role in future drug development programmes. Image resolution does not need to improve on that of the electron microscope (at the cellular level) and X-ray crystallography (at the molecular level), but without doubt the new abilities to determine in live or un-fixed specimens, the exact position of identified molecules in complex molecular/cellular machines will be utterly fascinating (Wong et al., ²⁰¹⁴). This combined with in vivo single-cell imaging has the potential to transform our understanding of these amazing cells.

Recognizing the paucity of technologies available to cell biologists in the 1970s, it is somewhat bizarre to believe that it is almost to be taken for granted that we will have at our disposal fluorescent-tagged constructs, and knockouts of every parasite gene of interest, but to be provocative, if we are to make significant steps in our understanding of the parasites molecular machines perhaps we should already consider the construction of appropriately tagged protein libraries for the ultrastructural localization of molecules in situ by cryo-tomography. As ever we will be dependent on emerging technologies, such as single-cell ‘omics’, but perhaps we should think more of the difficult areas of lipid and carbohydrate biology and the roles of small nucleic acids in parasite biology.

It is now clear that in our efforts to reduce the impact of these potent pathogens, the sexual and early sporogonic development of malarial parasites offer unique opportunities to attack vulnerable bottleneck populations with both drugs (Ruecker et al., ²⁰¹⁴) and ‘simple’ antibody-based vaccines (Miura et al., ²⁰¹³). It should therefore be a collective priority to understand: first, the contrasting metabolic organizations of the quiescent (and comparatively schizonticide insensitive) mature gametocyte and the diverse pathways mediating gamete, zygote and ookinete development; and second, how the molecular composition and architecture of the exposed surfaces of the gametes, zygotes and ookinetes define the immunological vulnerability of both fertilization and zygote development. We live in exciting times with ever new and powerful technologies at our disposal; we must however accept our responsibility to those infected by these formidable pathogens that perhaps the most important task in front of us is to ensure the considerable knowledge we have already gathered and will continue to gather is translated into useful intervention measures, a process which of itself provides academic challenges and rewards every bit as demanding as those of basic research.