AgESP is constitutively expressed in the mosquito midgut and is localized in the submicrovillar region of epithelial cells, an area that ookinetes interact with during invasion. This protease is critical for Plasmodium
midgut invasion, as silencing AgESP greatly reduces P. berghei
and P. falciparum
infection in An. gambiae
. Transcriptional analysis of the responses of An. gambiae
midgut to P. berghei
revealed that the most noticeable class of genes induced in response to ookinete invasion were those that modulate the architecture of the actin cytoskeleton. Silencing either gelsolin 
or AgESP () results in a similar phenotype in which the intensity and prevalence of Plasmodium
infection is greatly reduced. These observations suggest that the critical role of AgESP during ookinete invasion may be through induction of gelsolin expression, as AgESP silencing greatly reduces gelsolin expression () without affecting the expression of other genes thought to be involved in midgut cytoskeletal modifications in response parasite invasion, such as CP, Ciboulot or WASP.
The kinetics of AgESP induction in response to Plasmodium infection is different between P. berghei and P. falciparum-infected midguts. In P. berghei AgESP expression is induced at 24 h PF. This transcriptional response requires midgut invasion and SRPN6 expression, a sensitive marker of cell invasion, is also induced at this time. However, the induction of AgESP and SRPN6 expression is observed much earlier (12 h PF) in P. falciparum-infected midguts, suggesting that at least a few P. falciparum ookinetes may begin to invade the midgut earlier than previously thought. It is also possible that there is a long pre-invasion interaction of P. falciparum parasites with the midgut surface that triggers expression of these markers.
Apoptotic responses also trigger major rearrangements in the actin cytoskeleton 
. For example, in vertebrates, gelsolin is cleaved by caspase-3, and microinjection of the cleaved gelsolin fragment that contains the actin-severing activity triggers rapid depolymerization of the actin cytoskeleton 
. AgESP protein expression is low 24 h PF in the midgut of mosquitoes fed on a healthy mouse (), and expression is highly induced in response to Plasmodium
infection (). This induction requires cell invasion, as it is not triggered by CTRP–
parasites, which form ookinetes that do no invade the midgut (). Furthermore, induction of AgESP expression in the midgut of P. berghei
-infected females peaks at 32 h PF (), a time when many invaded cells are undergoing apoptosis and AgESP protein is highly expressed in the cytoplasm of cells undergoing apoptosis (). This suggests that, besides its involvement in ookinete midgut invasion, AgESP probably has a second function and may also mediate the apoptotic response triggered by parasite invasion. Interestingly, the induction of AgESP expression is more transient in P. falciparum
-infected midguts, as it is no longer observed between 24 and 48 h PF, a time when invaded midgut cells are known to be undergoing apoptosis 
. This suggests that there may also be functional differences in the apoptotic responses of midgut cells to invasion by theses two parasite species.
SRPN6 is a well-studied marker of midgut 
and salivary gland 
parasite invasion. Silencing SRPN6 does not affect AgESP expression (data not shown), indicating that AgESP acts upstream of SRPN6. The fact that silencing AgESP significantly reduces SRPN6 expression in the midgut as well as the salivary gland, indicates that AgESP is required for SRPN6 to be induced. Although it is possible that SRPN6 could act as a direct inhibitor of AgESP, we believe this unlikely because silencing SRPN6 does not enhance P. berghei
infection in A. gambiae
G3 females 
. The effect of AgESP on SRPN6 expression could be indirect, as SRPN6 is not expressed constitutively in the midgut or salivary gland, and the lack of invasion when AgESP is absent may prevent activation of the signals that mediates the induction of SRPN6 by the parasite.
Immunofluorescence staining revealed that AgESP protein is abundant on the luminal side of the mosquito midgut (), while in the SGs, AgESP is present on the basal surface of the distal lobes (). AgESP is constitutively expressed in high abundance in the distal lobes, unlike SRPN6 
, which is detected only at sites of sporozoite invasion in the lateral lobes. Sporozoite SG invasion is asynchronous, taking place between 13 and 19 days post infection, and is facilitated by specific receptor-ligand interactions 
. AgESP silencing in SGs decreased the number of sporozoites, indicating that this protease is also involved in sporozoite SG invasion. Unlike midgut cells, each SG cell is invaded by numerous sporozoites through the formation of parasitophorous vacuoles 
. An increase in the number of microtubules in the basal region of infected cells has been documented in transmission electron microscopy studies, but no signs of apoptosis have been observed 
, probably due to the “gentler” mechanism of parasite invasion in the SG. The fact that AgESP silencing does not affect gelsolin expression in the SG suggests that this protease promotes sporozoite invasion through a mechanism different from that in the midgut, probably not involving changes in the cytoskeletal architecture mediated by the induction of gelsolin expression.
Together, our findings indicate that AgESP is expressed on the surface of the two epithelial barriers that Plasmodium
parasites need to cross to complete their life cycle in the mosquito. Plasmodium
parasites activate a cascade of events in mosquito epithelial cells to traverse these organs, and AgESP is a major player in the regulation of the invasion process. The recent development of a synthetic homing endonuclease gene drive system in An. gambiae
opens the possibility of spreading the disruption of genes such as AgESP in natural populations. This strategy could result in mosquitoes with reduced ability to transmit malaria because Plasmodium
parasites would no longer be able to invade the midgut or the salivary glands efficiently.