Mutant/mutation
The mutant expresses a C-terminal 3xHA tagged version of ROVER and expresses the reporter GFP under control of the constitutive eef1a promoter. 

Protein (function)
ROVER was identified in a genetic screen with barcode sequencing that allowed identifying gametocyte expressed genes required for mosquito-infection. ROVER encodes a 367 amino acid protein with no predicted domains.

Phenotype
The ROVER::3xHA protein was detected only in mature ookinetes as 2 bands: the first at the expected size of ~43 kDa, and the second, more predominant band at ~25 kDa. This may indicate proteolytic cleavage of the protein. In immunofluorescence assays, STONES::3xHA was specifically detected at a distinctive membrane region located on the convex side of the mature ookinete, posterior to the apical structure. This region is critical for ookinete motility and has been termed Ookinete Extrados Site (OES). In non-triton X-100 treated mature ookinetes, no signal at the OES could be detected, suggesting that the N-terminal HA-tagged part of STONES is intracellular, which is consistent with its topology predictions.

Additional information
ROVER was identified in a genetic screen with barcode sequencing that allowed identifying gametocyte expressed genes required for mosquito-infection.

From the paper:
We selected for further analysis, three genes with strong phenotypes, STONES, CRYSP and CRONE, as well as two genes for which the screen revealed no phenotype, ROVER and SPM1; an independent preliminary study had indicated that ROVER may be involved in infection, while SPM1 was used as a control. Integration of the disruption cassettes and gene deletion in the clonal Δsto (RMgm-5345), Δcry (RMgm-5346), Δcro (RMgm-5347), Δrov (RMgm-5348) and Δspm1 (RMgm-5349) parasite lines was confirmed by PCR

We used PlasmoGEM disruption vectors for STONES and CRYSP and conventional disruption vectors for CRONE, ROVER and SPM1 to generate clonal P. berghei mutants in the c507 GFP-expressing wild type (wt) line. 
For all mutant parasites exflagellation events per the number of male gametocytes, was comparable to that of the wt control and ookinete conversion rates, i.e., the ratio of ookinetes to female gametocytes counts were also not significantly different from the control for all the mutants. 
Δcry, Δcro and Δspm1 mutants produced oocysts that were not significantly different in number from wt. A 99% decrease of mean oocyst numbers was observed for Δsto and Δrov mutants, with Δrov showing a maximum of only one oocyst in some midguts. Whilst the STONES phenotype was consistent with that obtained from the screen, the ROVER phenotype was unexpected and could only be justified by re-expression of the gene in the ookinetes and male wt allele rescue of the phenotype in the screen.
We investigated this by crossing Δrov to either the female-donor Δhap2 or the male-donor Δnek4 followed by oocyst counting in A. coluzzii 8 days post infection on coinfected mice. The Δc57 line that harbors a disruption of PIMMS57 was also included in these assays, as the screen also failed to detect this gene that has been previously shown to be important for ookinete-to-oocyst transition. The results confirmed that the oocyst-deficient phenotypes of both genes can be rescued by both the male and female  wt alleles, consistent with the expected limitation of the screen to reveal recessive phenotypes in diploid cells, after the wt allele introduced into the zygote by the microgamete is transcribed. PIMMS57 is known to be specifically expressed in ookinetes, likely by both parental alleles, and the results suggest that the gametocyte-enriched ROVER gene is also expressed in ookinetes and that this expression is important for its function.
These results were also reflected in the salivary gland sporozoite counts for Δsto, Δcro  and Δrov,which ranged from very few to none. Importantly, and consistent with the results of the screen, none of the thousands of Δcry oocyst sporozoites were capable of infecting the salivary glands, again corroborating the results of the screen. A statistically significant 57% reduction in sporozoite counts was detected for Δspm1 suggesting that the effect seen in midgut sporozoites may also be true. 
The ability of mutant parasites to transmit to the mouse host and infect RBCs was assessed using mosquito-to-mouse (C57/BL6 strain) bite-back infections 21 days post infection. As expected, no transmission and development of mouse parasitemia was detected for any of the Δsto, Δcro, Δcry and Δrov mutants, leading us to conclude that loss-of-function of the respective proteins leads to malaria transmission blockade. However, the reduction seen in Δspm1 salivary gland sporozoites did not bear any impact on the capacity of mutant sporozoites to infect the mouse host, suggesting that  SPM1 is dispensable for sporozoite development and transmission.

The ookinete to oocyst defective phenotypes of the Δsto and Δrov parasites were further investigated in midgut invasion assays using infections of A. coluzzii silenced for CTL4. CTL4 is a key hemolymph regulator of melanization, and its silencing leads to readily melanized P. berghei ookinetes that have succeeded in invading the midgut epithelium and reached the sub-epithelial space. The results showed that both Δsto and Δrov ookinetes displayed a great defect in midgut invasion as the number of melanized ookinetes were significantly reduced compared to wt controls. Defective midgut invasion can be due to the inability of ookinetes to move, and we assessed this by measuring the forward speed of ookinetes on matrigel. The results confirmed that both Δsto and Δrov mutants exhibit strong motility defects which likely cause their decreased ability to traverse the midgut epithelium and form oocysts and sporozoites. To further examine this, Δsto and Δrov ookinetes were injected directly into the hemocoel to assess if the oocyst and sporozoite defective phenotypes could be rescued. Indeed, it has been previously shown that mosquito transmission of P. berghei mutants with ookinete motility defects can be rescued if midgut invasion is bypassed. The result confirmed that this was the case for both Δsto and Δrov, as both the salivary gland sporozoite numbers and the ability of mutants for mouse transmission through bite-back were restored.

The endogenous STONES, ROVER, CRYSP and CRONE genes were tagged with C-terminal 3xHA tag via double crossover homologous recombination in the c507 line, and the resulting transgenic lines were designated stones::3xha (RMgm-5350), rover::3xha (RMgm-5351), crysp::3xha (RMgm-5352) and crone::3xha (RMgm-5353), respectively We first analyzed STONES and ROVER, of which the disruption leads to defective ookinete phenotypes. The STONES::3xHA protein could not be detected at the predicted size of ~125 kDa in Triton X-100 soluble extracts of blood stages, gametocytes or mature ookinetes. Instead, a band size of ~16 kDa was detected predominantly in mature ookinetes and less in blood stages and gametocytes. However, in Triton X-100 insoluble extracts, a band of ~80 kDa was specifically detected  in mature ookinetes, with traces of it also in gametocytes. As the full length protein was never detected in any of the extracts, these results suggest that STONES undergoes proteolytic processing, and that its C-terminal ~80kDa fragment is embedded within the membrane owing to the multiple transmembrane domains. The ROVER::3xHA protein was detected only in mature ookinetes as 2 bands: the first at the expected size of ~43 kDa, and the second, more predominant band at ~25 kDa. This may indicate proteolytic cleavage of the protein. In immunofluorescence assays, STONES::3xHA was specifically detected at a distinctive membrane region located on the convex side of the mature ookinete, posterior to the apical structure. This region is critical for ookinete motility and has been termed Ookinete Extrados Site (OES). In non- riton X-100 treated mature ookinetes, no signal at the OES could be detected, suggesting that the N-terminal HA-tagged part of STONES is intracellular, which is consistent with its topology predictions. ROVER::3xHA was localized in discrete cytoplasmic spots of mature ookinetes, resembling exocytic vesicles, commonly but not always positioned toward the apical end and in proximity to the cell membrane.

Western blot analyses using an anti-HA antibody detected a ~33 kDa CRYSP::3XHA protein in extracts from purified in vitro cultured crysp::3xha ookinetes and, to a much lower levels, from gametocytes, both prior to and after induction of gametogenesis. Similarly, the ~33 kDa CRONE::3xHA protein was detected in ookinete and, less so, gametocyte extracts of the crone::3xha line. We examined the cellular localization of the two proteins in immunofluorescence assays of gametocytes and ookinetes. In both cases, a clear and distinct spot pattern that always colocalized with the hemozoin (visible in bright field) was detected in the ookinete. This pattern is the hallmark of crystalloid localization in P. berghei. Multiple ookinete observations revealed that the number of spots varied from 1 to 3, which were always in association with the hemozoin containing vesicles. The two proteins were henceforth named CRONE for “crystalloid oocyst not evolving” and CRYSP for “crystalloid needed for sporozoites”. In the crone::3xha line, a vesicle-like albeit less prominent staining pattern was also detected in the female gametocytes, consistent with the high CRONE protein abundance in gametocyte extracts. Crystalloids are organelles known to be specific to ookinetes and young oocysts, thought to form soon after fertilization through active assembly of endoplasmic reticulum-derived vesicles. Some of the known crystalloid proteins are also synthesized in the gametocytes. Therefore, one can speculate that the CRONE::3xHA-stained gametocyte vesicles are crystalloid precursor subunits. Whilst this may be true, the expression of CRONE in gametocytes could be due to the CRONE::3xHA expression design that used the P. berghei dihydrofolate reductase (DHFR) 3′ untranslated region (UTR). Cis-acting elements in the 5′ UTR or 3′ UTR of DOZI- regulated genes have been shown to be important for translational repression. Indeed, a previous study that expressed a GFP-tagged version of CRONE using the 3′ UTR of P28 that is also translationally repressed by DOZI found that GFP is restricted to the ookinete crystalloid. To examine this, we raised rabbit polyclonal antibodies against a codon-optimized fragment of CRONE (amino acids 24-235) expressed in Escherichia coli cells. Using these antibodies in immunofluorescence assays, we detected a clear ookinete crystalloid signal, but this signal was absent from gametocytes. This indicated that the gametocyte signal detected in the crone::3xha line is likely due to leaky DOZI post-transcriptional repression

Other mutants