Babesia divergens egress from host cells is orchestrated by essential and druggable kinases and proteases

Apicomplexan egress from host cells is fundamental to the spread of infection and is poorly characterized in Babesia spp., parasites of veterinary importance and emerging zoonoses. Through the use of video microscopy, transcriptomics and chemical genetics, we have implicated signaling, proteases and gliding motility as key drivers of egress by Babesia divergens. We developed reverse genetics to perform a knockdown screen of putative mediators of egress, identifying kinases and proteases involved in distinct steps of egress (ASP3, PKG and CDPK4) and invasion (ASP2, ASP3 and PKG). Inhibition of egress leads to continued intracellular replication, indicating exit from the replication cycle is uncoupled from egress. Chemical genetics validated PKG, ASP2 and ASP3 as druggable targets in Babesia spp. All taken together, egress in B. divergens more closely resembles T. gondii than the more evolutionarily-related Plasmodium spp. We have established a molecular framework for biological and translational studies of B. divergens egress.

continue replicating in the absence of either an egress signal or the downstream host lytic effector molecules. Moreover, chemical genetic approaches demonstrate that these conserved essential molecules present validated druggable targets in Babesia spp.

An induced egress assay to study B. divergens host cell egress and invasion
To make the study of egress accessible in B. divergens, we developed a method to induce egress, which otherwise occurs asynchronously in in vitro culture. A ow cytometry-based assay was used to screen for egress induction in B. divergens using known T. gondii egress-inducing compounds (Figs. 1A, S1B and S1D). 8-Br-cGMP, BIPPO (phosphodiesterase (PDE) inhibitor) and to a lesser extent, H89 (PKA inhibitor), induced egress and inhibited parasite replication (Figs. 1A, S1D-H) 12,15 . A summary of the putative B. divergens egress signaling pathway can be found in S1C.
To elucidate the cellular feature of B. divergens egress and invasion, we used video microscopy to follow 8-Br-cGMP mediated induced egress. As has been previously observed 16 , B. divergens egress is frequently initiated when the intracellular parasite contacts and deforms the host cell (observed in 76% (16/21) of egress events, Fig. 1B -red arrow). Soon after the initial deformation, the RBC 'rounds up' (mean ± S.D. = 2.6 s ± 3 s, n = 17) (Fig. 1B, yellow arrow). 'Rounding' is observed in P. falciparum immediately preceding PVM rupture and may be associated with increased PVM permeability, however, the mechanisms remains unclear [17][18][19] . In B. divergens, which lack a PVM, rounding is likely due to permeabilization of the RBC. B. divergens parasites then become motile and escape the permeabilized RBC to invade new RBCs (mean ± S.D. = 1 s ± 2.4 s, n = 28, to escape after host cell permeabilization). Upon contact with a new RBC, the parasite strongly deforms the host cell around itself (Fig. 1C, green to purple arrows; mean ± S.D. = 6 s ± 2.7 s, n = 14), a process known as pre-invasion in P. falciparum, followed by the internalization phase when the parasite enters the RBC with relatively little deformation of the RBC (Fig. 1C, purple to blue arrows, mean ± S.D. = 4.5 s ± 1.1 s, n = 10; Video S1) 20 . In contrast to P. falciparum egress that takes ~ 5-10 minutes (from initial PVM permeabilization to merozoite release) 18,21 , B. divergens egress is rapid (mean ± S.D. = 10.8 s ± 3.0 s, n = 10).
To test the ability of B. divergens to egress throughout replication, parasites were synchronized to a 20-minute window and egress was induced every 1-2 hrs. The increased sensitivity to 8-Br-cGMP and increased number of parasites that egress as the parasites mature suggest that while B. divergens can egress throughout the replication cycle, it is more strongly primed to egress when parasites are fully mature (Fig. 1D). This feature of B. divergens egress is similar to T. gondii, which can be induced to egress throughout the replication cycle, albeit at reduced e ciency during S and M/C phase 22 , whereas P. falciparum egress is restricted to a narrow window at the end of the lytic cycle 15,23 .

Loss of host cell integrity induces B. divergens motility and egress
To test if B. divergens motility is induced by exposure to extracellular conditions, as has been demonstrated in T. gondii, we observed parasites after the RBC was lysed using a low concentration of saponin in buffers mimicking either intracellular (IC, 140 mM K + , 5 mM Na + ) or extracellular (EC, 5 mM K + , 140 Na + ) concentrations of potassium and sodium, plus or minus calcium (2 mM). Upon saponininduced RBC lysis in EC-Ca 2+ or IC-Ca 2+ buffer, parasites became motile and were able to escape from the permeabilized RBC (Fig. 1E, Video S3). In contrast, parasites in either IC or EC buffer without Ca 2+ did not become motile (Fig. 1E, Video S4). PKG activation, through the addition of 8-Br-cGMP, bypassed the requirement for extracellular calcium in either buffer to initiate gliding motility (Fig. 1E).
Together these results suggest that B. divergens motility is induced by exposure to extracellular concentrations of Ca 2+ , likely acting upstream or in parallel of PKG, but does not respond to Na + or K + concentrations. Similarly, T. gondii can sense exposure to the extracellular environment through serum albumin, a drop in potassium and an increase in extracellular calcium, which induces microneme secretion and motility 22,24−28 . P. falciparum is also able to sense lipids and potassium which alter microneme secretion, egress or invasion, although their roles in the parasite are less clearly de ned [29][30][31] . Extracellular calcium is required for e cient invasion, but not egress, of P. falciparum 30,32,33 . Chemical inhibition of PKG, proteases and calcium release all impair egress To identify the molecular processes required for B. divergens egress, we used the ow cytometry-based egress assay to screen for compounds that inhibit 8-Br-cGMP-induced egress. The compounds were selected based on known egress inhibition in P. falciparum or T. gondii. BAPTA-AM, which chelates intracellular calcium, and Compound 1 (C1), an inhibitor of apicomplexan PKG, both inhibited 8-Br-cGMP induced egress (Fig. 1F). Inhibitors of serine proteases, including TPCK, TLCK and PMSF, reduced induced egress at similar concentrations to those that inhibit P. falciparum gametocyte egress, whereas inhibitors of other protease classes exhibit modest inhibition (< 20%) ( Figure S1H) [34][35][36] . The short incubation time (15 min) may not identify inhibitors of certain proteases such as PfPMX/TgASP3 which act several hours prior to egress. Inhibitors of the lipid signaling pathway that inhibit egress in T. gondii and/or P. falciparum, including U73122 and propranolol which target PI-PLC and phosphatidic acid phosphatase, respectively, did not in uence egress by B. divergens, whereas the diacylglycerol kinase inhibitor R59022 instead enhanced induced egress ( Figure S1H) 27,29,37,38 . In line with our ow cytometry assay, E64d treated parasites showed no obvious defect in egress, motility or invasion by video microscopy ( Figure S1J). Taken together, these data demonstrate a requirement for cGMP signaling, calcium signaling (both intracellular and extracellular), serine proteases and gliding motility for e cient egress by B. divergens. Unlike in T. gondii, release of calcium with the ionophore A23187 does not induce egress or motility in B. divergens, suggesting calcium release is required, but not su cient for egress. We note that a previous study found that A23187 could induce egress in B. bovis which could be due to differences between species, host cells, or technical such as the extracellular concentration of calcium 39 .
Motility is required for B. divergens merozoites to egress from permeabilized RBCs but not for permeabilization The initial deformation of the RBC in egress appears to be caused by direct contact with the parasite indicating the parasites actinomyosin motor may be used to physically disrupt the host membrane ( Fig. 1B) 16 . To test if motility is required during egress, egress was induced with 8-Br-cGMP in the presence of phalloidin-Alexa Fluor 488, which selectively stains the cytoskeletons of permeabilized RBCs, and cytochalasin D, which inhibits actin polymerization and gliding motility. Cytochalasin D-treated parasites can be induced to lyse the host cell, but do not escape the permeabilized cell ( Fig. 1G-H, Video S2). Immediately preceding egress, the host cell becomes ru ed, but no localized deformation is observed, followed by the RBC rapidly showing a reduced diameter, becoming round and being in ltrated by phalloidin, demonstrating the RBC membrane has been permeabilized (Fig. 1H, Video S2). T. gondii secreted lytic factors (e.g. PLP1) and host calpains damage the host cell, allowing the motile parasite to escape the permeabilized cell 27,40,41 . The mechanisms of RBC lysis in asexual Plasmodium spp. remain unclear 42,43 . B. bovis PLP1 is not strictly required for egress, however, its absence does lead to continued intracellular replication (4 parasites per cell instead of 2 which is typical of B. bovis), suggestive of a partial egress defect 44 . In contrast to P. falciparum which fractures the RBC cytoskeleton during egress ( Figure   S1I) 45 , the permeabilized RBC remains relatively intact throughout egress in B. divergens (Figs. 1B-C, 1G-H) 16 . P. falciparum secretes proteases (SERA6) which degrade the RBC cytoskeleton, leading to the to rupture of the RBC to release free merozoites, thus bypassing the requirement for gliding motility in egress 17,46,47 . Consistent with this difference, Babesia spp. do not have an ortholog of any SERA protease nor an ortholog of the phospholipase TgLCAT/PbPL. Taken together, these data argue for a mechanism of host cell lysis in B. divergens relying centrally on secretion of lytic factors (e.g PLPs). While, we do not observe a requirement for parasite motility for host cell lysis, we cannot rule out that motility contributes towards cell lysis as has been observed in T. gondii 40 . Identi cation of putative egress, motility and invasion genes through transcriptomic analyses The lytic cycle of B. divergens has been morphologically de ned but remains poorly characterized at the molecular level in any Babesia spp. 5 There are several Babesia spp. transcriptomes at different stages of the life cycle, however, there are no synchronous intraerythrocytic stage time course transcriptomes 48-52 . We generated a synchronous bulk transcriptome of one replication cycle (0-12 hrs) for B. divergens. As a parallel strategy, we utilized a recent asynchronous single-cell transcriptome where expression pro les of individual genes were generated using a pseudo-time analysis with the start of gene expression identi ed by cross-correlation analysis between the bulk and single-cell expression curves 53 . Genes that showed expression changes over time display a transcriptional cascade associated with "just-in-time" gene expression that has been observed in Plasmodium spp. and T. gondii (Figs. 2A and S2A) 54,55 . The number of genes detected and that display signi cant changes over time for each method can be found in Fig. 2B. The timing of peak expression between the bulk and single-cell transcriptomes was well correlated for the majority of genes, supporting the use of the pseudo-time data for downstream analysis (Fig. 2D). 93 B. divergens genes were identi ed as orthologs or family members of known egress, motility or invasion genes we have collated for apicomplexan parasites. 91 of these genes are expressed in in vitro culture and 67 showed expression changes over time (Table S1). Genes that are expected to co-localize within the same subcellular compartment (e.g. micronemes) based on orthology display similar expression pro les as has been observed in Plasmodium spp. and T. gondii (Figs. 2C and S2B) 54,56,57 . Secondary messenger-responsive proteins, including PKG, PKAc1, PKAr, and CDPK4 (PfCDPK4/TgCDPK3) peak ~ 8-10 hours post invasion (hpi) and remain high until 12 hpi when parasites naturally egress (Figs. 2C). Other putative members of the egress signaling pathway did not display stage-speci c expression indicative of a role in egress or invasion (Table S1). The cysteine and aspartyl proteases, DPAP1 (no direct ortholog) and ASP3 (ortholog of PfPMX/TgASP3), respectively, display expression pro les matching that of microneme proteins (Figs. 2C). The aspartyl protease ASP2 (closely related to BdASP3) displays an expression pro le matching rhoptry proteins (Figs. 2C). All nine B. divergens PLP genes are expressed (Data not shown). We found that only PLP1 (TgPLP1/PfPLP3) and PLP4 (no direct ortholog) display an expression pro le matching other microneme and rhoptry proteins, respectively, suggestive of a role in egress and invasion or post-invasion PV breakdown, respectively (Figs. 2C).
Multiple invasion ligands were identi ed by orthology to known apicomplexan ligands and display expression pro les matching that of rhoptry or microneme proteins (Figs. 2E). These included TRAP2/P18, RAP1 and AMA1, which all have a demonstrated role in Babesia spp. invasion (Fig. 2C) 58-60 . The B. divergens orthologs of the Plasmodium spp. or T. gondii rhoptry protein, SRA, and microneme proteins, CLAMP, GAMA, MAEBL and CelTOS represent novel vaccine targets that have not been investigated in Babesia spp. 61-65 Inner membrane complex (IMC) proteins, which are involved in cell structure and motility, are transcribed in three groups ( Figure S2B), suggesting that IMC assembly with distinct pro les warrants future study.
A comparative transcriptomic approach using data from P. falciparum and T. gondii was used to identify novel proteins putatively involved in egress, motility or invasion (methods outlined in Figure S2C). 104 genes were identi ed, 31 of which were identi ed by multiple methods and whose orthologs have no known function (Figs. 2E; Table S1). Of the 8 genes found in all three species, 5 have a growth phenotype in the T. gondii CRISPR screen and the P. falciparum piggyBac screen (Table S1).
A genetic screen of high priority candidates reveals the essentiality of PKG, CDPK4, ASP2 and ASP3 for parasite proliferation We focused on 11 high priority candidates based on transcriptomics as well as orthology to apicomplexan egress genes, including the kinases, PKG, PKAc1, PKAc2, CDPK4, CDPK5 and CDPK7, the perforin-like proteins, PLP1 and PLP4, and the proteases, ASP2, ASP3 and DPAP1. PKG is conserved between B. divergens, P. falciparum and T. gondii and has a well-characterized role in egress and invasion in these latter species 66,67 . PKAc1 suppresses egress in T. gondii but is instead required for invasion by P. falciparum 10,12,68−71 . The CDPK family of proteins are also required for P. falciparum and T. gondii egress and invasion, although the direct orthologs (de ned by reciprocal blast) do not always have the same function between species (Reviewed in 72,73 ). PLPs are involved in host cell permeabilization and egress in T. gondii, Plasmodium spp. sexual stages and B. bovis 40,44,74−76 . PfDPAP family members and BdASP2/ASP3 orthologs (PfPMX, PfPMIX and TgASP3) are required for egress and/or invasion in P. falciparum or T. gondii [77][78][79][80][81][82] .
To develop a stable transfection system for B. divergens, multiple transfection methods and selection drugs with established resistance markers were tested for stable transfection of a GFP reporter plasmid ( Figure S3A and C) (summarized in Table S2). Nucleofection of isolated merozoites and blasticidin-S selection was the most e cient method and was used for all further transfections. A CRISPR/Cas9 system was generated to introduce a HA tag, as well as the glmS riboswitch with and without a destabilization domain (DD) inducible knockdown system in series to the 3' end of each gene (Figs. 3A and S3B) 83,84 . Parasites containing the correct integration reached > 1% parasitemia 12-16 days after transfection for all constructs, except PKAc1, PLP1 and PLP4 which we were unable to tag ( Figure S3F and H). PKG-HA-DD-glmS parasites were initially used to determine the effectiveness of the knockdown systems. Induction of the DD system, which induces protein degradation, generated a more rapid (~ 6 hrs) and stronger reduction of protein levels than the glmS system (Fig. 3B). The glmS system, which destabilizes mRNA, resulted in slower knockdown that was rst observable by 24 hrs, but generated strong knockdown by 48 hrs (Fig. 3B). Combining both systems resulted in a stronger knockdown than either alone (Fig. 3B). This pattern was re ected in the effect of these knockdown systems on parasite growth when targeting PKG, CDPK4, ASP2 or ASP3, with the double knockdown producing the strongest defect in all lines, whereas the DD or glmS alone were not su cient for all genes (Fig. 3D). DPAP1, PKAc2, CDPK5 and CDPK7 knockdown did not affect proliferation (Fig. 3D, data not shown).
PKG and CDPK4 are essential for egress and their depletion results in continued intracellular replication We analyzed knockdowns of PKG and CDPK4 following synchronized invasion to understand when the block in proliferation occurs.
PKG or CDPK4 knockdown parasites continued to replicate within a single iRBC, to form up to 16 parasites per RBC by 40 hpi (Figs. 3C-E and S3I). The fraction of parasites that replicate to > 4 per iRBC is dependent on the degree of knockdown (Fig. 3E). Knockdown of PKG prevented induced egress by 8-Br-cGMP, further supporting that 8-Br-cGMP induces egress through activation of PKG ( Figure S3D). The reduced susceptibility to 8-Br-cGMP mediated induced egress in PKG-HA-DD-glmS parasites without inducing knockdown compared to WT parasites is likely due to a partial knockdown in the absence of induction, due to tagging as observed in other parasite recombinants ( Figure S3D) 85,86 . A model of the molecular pathways involved in B. divergens egress can be found in Fig. 3F.

The proteases ASP2 and ASP3 are required for egress and invasion
The orthologs of B. divergens ASP2 and ASP3 in P. falciparum (PfPMIX/PMX) and T. gondii (ASP3) are the most upstream proteases known to date in the protease cascade that matures many proteins in the micronemes and rhoptries that are required for egress and invasion 80,81 . ASP2 and ASP3 knockdown resulted in an increased number of free merozoites in culture (Fig. 3D). In the ASP3 knockdown culture, clusters of parasites were also observed in lightly stained RBCs (Fig. 3C, ASP3 left panel). To further de ne these phenotypes, induced egress in knockdown parasites was observed by video microscopy. The time to egress and rate of escape from the lysed cells was the same between ASP2 knockdown and WT parasites (Fig. 4A, B and E). After egress, ASP2 parasites typically bound to a single RBC and were able to strongly deform it (pre-invasion) similar to WT, however, were unable to complete invasion and eventually detached from the cell (Figs. 4C-E; Video S5). The time from initial deformation of the RBC by the intracellular ASP3 knockdown parasite to the time of lysis of the RBC was signi cantly longer than in WT parasites and fewer parasites were able to escape the lysed RBC (Figs. 4A-B, F-G). Fewer ASP3 knockdown parasites were able to escape the lysed cell, and parasites that successfully egressed were able to bind to and deform the RBC, however, unlike WT or ASP2 knockdown parasites that typically bind strongly to a single host cell, ASP3 parasites maintained gliding motility over the surface of the cell, often contacting multiple host cells, but were rarely able to complete invasion ( Fig. 4B-D, 4F-G; Video S6). Together, these results are consistent with the transcriptomic data suggesting ASP2 functions in the rhoptries and knockdown parasites are able to reorient but are not able to undergo the nal step of invasion which requires rhoptry proteins 87-89 . The ASP3 phenotype suggests that it is required for maturation of microneme proteins that are required for RBC lysis and reorientation and/or anchoring of the apical end of the parasite to the RBC prior to rhoptry release 21,23 . In comparison to T. gondii, which requires only TgASP3 to process proteins in the rhoptries and micronemes, our data suggest that BdASP2 and BdASP3 are functionally orthologous to PfPMIX and PfPMX, respectively, and likely responsible for processing proteins required for invasion in the rhoptries, and egress and invasion in the micronemes, respectively. PKG, ASP2 and ASP3 are druggable targets required for egress and/or invasion Signi cant effort has been invested to develop speci c inhibitors of apicomplexan PKG, CDPKs and aspartyl proteases (PfPMIX, PfPMX and TgASP3) (reviewed in [90][91][92][93][94]. To demonstrate that PKG is a druggable target in B. divergens we used the apicomplexan PKG inhibitors compound 1 (C1), and ML10 a potent and highly speci c P. falciparum PKG inhibitor with in vivo activity 95,96 . To determine the speci city of C1 and ML10 for BdPKG, CRISPR/Cas9 was used to introduce a putative resistance mutation (T651Q) equivalent to the gatekeeper mutations that confer resistance in P. falciparum and T. gondii (Fig. 5A) 67,97 . A 500 bp region within a plasmid was used as a repair template, achieving ~ 50% editing, with decreasing e ciency further from the cut site ( Figure S3E and F). We utilized the ability to perform sequential transfection with the same resistance marker to generate double mutants, containing the T651Q mutation and the glmS knockdown system. Treatment of WT parasites with C1 or ML10 resulted in the same continued intracellular replication as was observed with PKG knockdown (Fig. 5B). The potency of C1 and ML10 against B. divergens is ~ 25-fold and 281-fold less than against P. falciparum, and 4.4-and 8.4-fold less than against B. bovis 66,96,98,99 . A 3-and 10-fold for C1, and 15-and 8-fold for ML10, reduction in the IC 50 was observed between WT compared to PKG-glmS parasites with or without knockdown, respectively (Figs. 5C-D).
A trending, but not statistically signi cant shift in the IC 50 was observed between WT and the PKG-T651Q parasites (10.5 vs 14.1 µM for C1, and 591 nM vs 1209 nM for ML10), whereas the same mutation in knockdown parasites restores susceptibility to near WT levels (Figs. 5C-D). Together these results support a model in which C1 and ML10 are able to target PKG, however, at concentrations required for killing B. divergens there are secondary targets contributing to growth inhibition, which is likely to be CDPK4 that is the only other CDPK to share the same small gatekeeper residue as PKG ( Figure S3G). The ability to rapidly and speci cally inhibit PKG with C1 when combined with partial knockdown was used to investigate the role of PKG during RBC invasion. PKG knockdown increased the sensitivity to C1 invasion inhibition by 20-fold (Fig. 5E), suggesting PKG signaling is also required for host cell invasion.
To determine the druggability of ASP2 and ASP3, we screened 22 aspartyl protease inhibitors, including compounds known to be active against P. falciparum PfPMIX and PfPMX as well as clinical inhibitors of human beta-secretase, against WT and knockdown parasite lines ( Figure S4) 81,100−103 . 14 compounds displayed an IC 50 < 25 µM, with the most potent compound (TCMDC-134675) having an IC 50 of 891 nM against WT parasites ( Fig. 5G and Table 1) 81,103 . We observe a 2.7-fold and 3.6-fold increase in sensitivity to TCMDC-134675 in ASP2 and ASP3 knockdown parasites, respectively. CWHM-0000166 was less active against WT parasites (8.8 µM), but displayed the largest increase in sensitivity of 7.7-fold and 26-fold against ASP2 and ASP3 knockdown parasites, respectively (Fig. 5F). In P.
falciparum, incomplete PMX inhibition results in parasites lysing the PVM, whereas complete inhibition blocked PVM rupture 80, 104 . To determine if the phenotype observed in ASP3 knockdown parasites was incomplete, we combined knockdown with CWHM-0000166 treatment. Under these conditions we observed that parasites continued to replicate intracellularly, as was observed with inhibition of PKG and CDPK4, further supporting a role of ASP3 in egress (Fig. 5B). We also observed clusters of parasites within a lysed RBC, in line with the live microscopy results that parasites that do lyse the host cell escape at a reduced rate. The similar phenotype seen following inhibition of ASP3 to the protein kinases CDPK4 and PKG, suggests that failure to egress by multiple mechanisms results in a default for continuation of the intracellular lytic cycle. No synergy was observed between ASP2 or ASP3 knockdown and ML10, suggesting the kinase and protease egress pathways act independently. Together these data demonstrate that PKG, ASP2 and ASP3 are essential and druggable proteins in B. divergens that are required for egress and/or invasion.

Discussion
Here, we establish B. divergens as a genetically tractable in vitro model to study Babesia spp. cell biology. The synchronous B. divergens bulk transcriptome and single-cell transcriptome will serve as a resource for the study of Babesia spp. biology as well as for further comparative studies of apicomplexan biology. The majority of molecular research in apicomplexan parasites has been limited to P. falciparum, P. berghei and T. gondii. Transient or stable transfection systems exist for Babesia spp., Theileria spp., Cryptosporidum spp., Eimeria spp., Sarcocystis spp. and Neospora spp. (reviewed in Suarez, et al. 105 ). However, the number of studies and the range of available tools and resources remains comparatively limited in these organisms. Additional studies of a wider range of apicomplexan parasites will expand our knowledge of conserved and unique biological mechanisms of apicomplexan parasitism, that are often pathogens of metazoans.
Egress is a unique aspect of apicomplexan parasitism compared to their hosts that requires specialized processes and offers novel druggable targets. The overall process of egress is similar between Plasmodium spp. and T. gondii, however, there are notable differences in their cell biology (e.g. differences in division mechanisms) and in the host cell niche that place unique pressures on the parasite (T. gondii resides in nucleated cells, while the asexual stages of Babesia spp. and Plasmodium spp. reside in enucleated RBCs). While the signaling and molecular mechanisms differ in some aspects, we have found that egress of B. divergens at the cellular level closely resembles that of T. gondii and has several notable differences to Plasmodium spp. egress despite being evolutionarily closer and sharing the same RBC niche. B. divergens egress begins with localized deformation of the RBC by the intracellular parasite. The deformation requires the parasites actinomyosin motor but is not necessary for disruption of the RBC membrane, although it may still contribute. Once the parasite has permeabilized the RBC membrane, it utilizes its actinomyosin motor to pass through the RBC cytoskeleton which remains relatively intact throughout egress. This is similar to T. gondii but in sharp contrast to Plasmodium spp. that fracture the RBC cytoskeleton to release parasites without requiring motility.
T. gondii is able to sense both intrinsic signals of parasite density and extrinsic signals of host cell damage to induce egress 22,[24][25][26][27]37,106 . Similarly, P. falciparum can sense lipids and potassium, although their role in egress and invasion are less clearly de ned 29,30 .
Since Plasmodium spp. egress is strictly required at the end of each replication cycle, the initial egress signal is likely to be linked to the cell cycle, however, this remains to be demonstrated. The only egress signal identi ed in B. divergens is extracellular calcium, but not potassium or sodium, which acts as an extrinsic signal of host cell damage (Fig. 1E, Video S2). Whether B. divergens responds to its environment to determine if parasites will egress after one or two replication cycles, or whether this is regulated stochastically remains unclear.
In T. gondii and P. falciparum, the initial signals converge to either activate guanylate cyclase (GC) or inhibit PDE and thus raise cGMP levels, which in turn activates PKG. In T. gondii, PKAc1 suppresses cGMP levels to prevent premature egress, putatively by activating a PDE, whereas in P. falciparum PKAc1 is required for invasion 10,12,68−71 . Chemical inhibition or activation of PKA suggests that PKA suppresses B. divergens egress and is also required for invasion (Fig. 1F, S1D and S1G). It is possible these functions are performed separately by the two PKAc orthologs, although only PKAc1 displays stage speci c expression (Figs. 2C). The PKAc2 knockdown line did not display an obvious growth defect and we were unable to generate a PKAc1 knockdown line. The inability to generate several of the attempted knockdown lines could be due to the DD tag interfering with the proteins function, or the proteins being sensitive to partial knockdown induced by the DD or glmS even in the absence of induction. Alternative genetic systems will need to be developed for B. divergens to determine the function of these proteins.
In P. falciparum and T. gondii, PKG activates PI-PLC, which generates lipid second messengers. These eventually lead to the release of calcium from intracellular stores that are required for microneme/exoneme secretion. We have demonstrated that PKG is a central component of the B. divergens egress signaling pathway that is both necessary and su cient to induce egress (Figs. 1A, 1F and S3D). Small molecules targeting the lipid signaling pathway, including propranolol and U73122, had little to no effect on B. divergens egress and will require further genetic studies to determine the role of lipid signaling. The egress signaling pathway leads to the release of the micronemes/exonemes. These organelles contain lytic factors, motility proteins and invasion ligands, many of which are proteolytically matured. In P. falciparum, proteolytic maturation in the rhoptries and micronemes is done in part by PMIX and PMX, respectively 81 . In contrast, T. gondii only contains one ortholog, TgASP3, which is localized in a post-Golgi compartment and matures a subset of microneme and rhoptry proteins 80,82 . The putative localization of BdASP2 and BdASP3 based on transcriptomics and their knockdown phenotypes suggest they are functionally orthologous to PfPMIX and PfPMX, respectively. While further work will be required to determine the substrates of these proteases and overlap between apicomplexans, we hypothesize that the continued replication phenotype observed with ASP3 inhibition is due to a block in maturation of lytic factors and therefore host cell lysis. Inhibition of either the signaling pathway or the downstream lytic factors produces a similar continued intracellular replication phenotype in T. gondii 40,67 . This implies that T. gondii and Babesia spp. lack a checkpoint to exit the cell cycle and that the default pathway is to continue replication. Inhibition of egress signaling in P. falciparum blocks replication, however, the transcriptional pro le of stalled parasites continues to progress, suggesting that they may also lack a checkpoint at the transcriptional level 109 .
Small molecules have been developed that target the kinases and proteases required for egress of Plasmodium spp. and T. gondii. Identi cation of shared parasite targets of these compounds in Babesia spp. and other apicomplexan parasites will help leverage drug development efforts for malaria that have signi cantly more resources. Here, we have developed transfection, alongside CRISPR/Cas9 and inducible knockdown systems to modify the B. divergens genome. Through knockdown and the introduction of resistance mutations, we identi ed C1 and ML10 as inhibitors of PKG, however, at concentrations required for killing it is likely that other parasite molecules are also inhibited, with CDPK4 being the most likely target as it also contains a small gatekeeper residue (Fig. 5C-E and S3G). We also identi ed several compounds that dually target ASP2 and ASP3, including compounds that have known activity against PfPMIX and PfPMX lines 81,100,101 . The dual activity against these proteins may reduce the ability of the parasite to develop resistance to small molecule inhibitors. The same methods could be applied to these and other proteins to aid future drug development for Babesia spp. Notably, multiple new classes of potent inhibitors have been developed against PfPMX and PfPKG with drug-like properties and in vivo activity 104,110,111 .
Here, for the rst time we have established a molecular framework for the events surrounding Babesia spp. egress. Several major questions remain about Babesia spp. egress, such as how they selectively lyse the PVM after invasion, the role of lipid signaling, the initial signal to induce egress and how they regulate egress to occur after one or more replication cycles. With the cellular, transcriptomic and genetic tools developed here, future studies will be able to answer these questions and reveal the unique biology of Babesia spp. as well as conserved processes throughout Apicomplexa, providing a rational basis for the development of therapeutic interventions.

Declarations Acknowledgments
We would like to thank Dr.

RESOURCE AVAILABILITY
Lead contact Further information and requests for resources and reagents should be directed to and will be ful lled by the lead contact, Manoj Duraisingh (mduraisi@hsph.harvard.edu).

Materials availability
All unique reagents in this study are available from the lead contact. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Data and code availability
Single-cell RNAseq and bulk RNAseq data have been deposited at NCBI Sequence Read Archive (SRA) and are publicly available as of the date of publication. Accession numbers are listed in the key resources table. Microscopy data reported in this paper will be shared by the lead contact upon request. All code has been deposited at GitHub. DOIs are listed in the key resources

Live microscopy
All videos were taken on a Zeiss Axio Observer using a 60x oil immersion lens inside a chamber heated to 37 o C. Prior to imaging, parasites were allowed to settle on the bottom of a glass bottom slide (Ibidi, Cat#80827/81817) at 37 o C in a 5% CO 2 incubator for 10-15 min. Small molecule inhibitors were included in this incubation as necessary. Immediately prior to imaging, the media was removed and replaced with the RPMI/IC/EC with 500 µM 8-Br-cGMP, and any small molecules inhibitors being tested. For IC/EC experiments, the nal buffer contained 0.0075% (w/v) saponin (Calbiochem, Cat#558255) to lyse the RBC. Alexa Fluor™ 488 Phalloidin (Invitrogen, Cat#A12379) was added to RPMI at a concentration of 1/150 where stated. All images were taken within 20 minutes of removal from the 5% CO 2 incubator. Plasmodium falciparum mature schizonts were isolated by magnetic a nity puri cation (MACs LS column, Miltenyi) and imaged as per the B. divergens protocol. Images were processed in Zen 2 (Zeiss) and ImageJ/Fiji. Isolation of free merozoites for invasion assays, synchronization and transfectioñ 1-2 ml of packed iRBCs at 20-30% parasitemia was used to isolate free merozoites using a modi ed protocol from 5 . Brie y, the iRBC was resuspended to 10% HCT in RPMI and passed through two 1.2 µM lters. The isolated merozoites and RBC debris were pelleted at 3000 x g for 3 min and the supernatant was removed. For invasion assays, the merozoites were resuspended in RPMI and added to a 96 well u-bottom plate containing 1.33x nal drug concentration (30 µl total) and incubated at 37 o C for 10 min. Fresh RBCs were then added to a nal of 2% HCT and 1x drug concentration (40 µl total). The plate was then incubated at 37 o C shaking at 600 rpm for 20 min. The assay was stopped by washing the parasites three times with 200 µL PBS, 400 x g for 2 min or by adding 200 µl 4% paraformaldehyde. Parasitemia was determined by ow cytometry as per induced egress section above. For synchronization, isolated merozoites were resuspended to a nal volume of 1 ml of 20% HCT RBCs in RPMI and allowed to invade for 20 min shaking at 600 rpm and 37 o C. Parasites were then washed 3x with 10 ml RPMI at 400 x g to remove free merozoites and cell debris. 100 µg/ml heparin was added to prevent re-invasion throughout the time course when stated in the gure legend.

Plasmid construction
The sequences of all primers and synthesis products used in this study are found in table S3. The bi-directional promoter between the EF1alpha (Bdiv_030590) and LON peptidase (Bdiv_030580c) was ampli ed using primers BE-8 and BE-9, and cloned into XhoI/BamHI sites of pPfEF-GFP-BSD (EF1alpha side drives GFP/Cas9 expression). The DHFR (Bdiv_030660, primers BE-21/22) and HSP90 (Bdiv_037120c, primers BE-32/33) 3'UTRs were cloned into EcoRI/HindIII and SpeI/NotI sites, respectively, for the selection marker or GFP/Cas9, respectively. Cas9 was ampli ed from pDC2-Cas9 using primers BE-124/125 and cloned into the XhoI/SpeI sites in place of GFP. The U6 promoter, bbs1 sites (for the guide), guide tracer/scaffold, U6 terminator and PKG-T651Q repair template were synthesized by IDT (Synthesis 1) and cloned into the EcoRI site by gibson assembly to make the pEF-Cas9-PKG-T651Q plasmid. For the sgRNA, oligos were either phosphorylated, annealed and cloned into the BbsI sites, or the sgRNA was ampli ed using the corresponding 'guide PCR' primer (e.g. BE-551) and BE-550 (universal for all). For all inducible knockdown lines, the HA-glmS or HA-glmS-DD tags were ampli ed using primers BE-536/537. The 5'HR (HR1, 3'end of the gene) and 3'HR (HR2, 3'UTR of gene) were ampli ed using the corresponding HR1 and HR2 primers from Table S3 for each gene (e.g. BE_512-515 for CDPK4). The sgRNA, HR1, DD-glmS and HR2 PCR fragments were cloned into the Bbs1/PacI sites of the pEF-Cas9-PKG-T651Q plasmid in a single Gibson reaction. Correct integration of plasmids was con rmed with primers from Table S4 labelled "test integration" (e.g. BE_610/611 for CDPK4).

Transfection
For transfection of iRBCs using the Biorad Genepulser II, 100 µg DNA in 30 µl combined with 370 ul of cytomix (120 mM KCl, 0.15 M CaCl2, 2mM EGTA, 5mM MgCl2, 10mM K2HPO4/KH2PO4, 25mM HEPES, pH 7.6) and 200 µl iRBCs (~ 15% parasitemia). Transfection was carried out with Biorad genepulser II set to 0.31 kV, 950µF. For Amaxa nucleofection of free merozoites, merozoites were isolated from ~ 3x10 9 iRBCs and all supernatant was removed. The pellet, containing free merozoites and cell debris, was resuspended in 100 µl of P3 solution (Lonza) plus 10 µl of water containing 2-10 µg of DNA. For Amaxa nucleofection of intact iRBCs, 10 µl iRBCs at 10% parasitemia was resuspended in 110 µl P3 + DNA solution, as per above. Transfection of either free merozoites or iRBCs was carried out in a 4D-Nucleofector System (Lonza), using the FP158 program. After electroporation of free merozoites, the parasite and buffer mixture was immediately transferred to 1 ml of RPMI containing 200 µl packed RBCs and pre-heated to 37 o C. Parasites were allowed to invade at 37 o C shaking at 600 rpm for 20 min before being washed with 10 ml RPMI to remove the P3 solution and returned to culture. Electroporated intact iRBCs were washed 1x with 10 ml RPMI and returned to culture. Cultures were selected with 15-20 µg of Blasticidin-S.

RNAseq analysis
RNA was isolated from parasites using a hybrid protocol of organic extraction combined with column puri cation. Brie y, parasite pellets were resuspended in 500 µl TRIzol and then extracted with chloroform. The aqueous layer was then puri ed using Qiagen RNeasy Mini spin-columns following manufacturers protocols. RNA was quanti ed and normalized into individual wells, and libraries were prepared following the Smart-seq2 protocol 117 . Libraries were sequenced on the illumina platform. Bulk synchronous RNAseq analysis work ow: The quality of reads was assessed using FastQC (Version 0.10.1). The reads were trimmed with Cutadapt from Trim Galore package (Version 0.3.7) (http://www.bioinformatics.babraham.ac.uk/projects/trim_galore/). The trimmed reads were mapped against the Bdivergens1802A reference genome (PiroplasmaDB release 46) and assembled with HISAT2 (Version 2.0.5 released) ( https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4655817/). SAM les obtained from alignment results were processed using SAMtools (Version 1.4.1) and the relative abundance of transcripts were estimated using featureCounts (https://academic.oup.com/bioinformatics/article/30/7/923/232889). Normalization & Noise removal: Counts per million (cpm) values per gene was calculated using cpm() from edgeR R package (Version 3.24.3) (https://www.bioconductor.org/packages/release/bioc/vignettes/edgeR/inst/doc/edgeRUsersGuide.pdf). Genes with cpm value > 2 in at least 3 samples were maintained for further analysis. Gene counts were normalized and scaled to logarithmic form using edgeR's TMM method (trimmed mean of M values) with DGEList(), calcNormFactors() and cpm() functions. The cpm() parameters were as following: y = DGEList.obj, log = TRUE, prior.count = 3, normalized.lib.sizes = TRUE. Batch effected samples were Identi ed through the analysis of hierarchical clustering and dissimilarities method using the R function hclust ( gondii data was manipulated to change the timepoints to match the other datasets (i.e. it starts at 6h from paper which is straight after invasion). Pearson correlation was calculated in R using the smoother curves for AMA1 and RON2. Genes were considered to be highly correlated if they displayed ≥ 2-fold change over time and had a Pearson correlation value of ≥ 0.9.
scRNA-Seq work ow scRNA-Seq data was processed using the 10x cell-ranger pipeline and aligned to the Babesia divergens 1802A genome. Counts were subsequently normalized and processed using the R Seurat package. A total of 9450 cells and 3620 genes were retained after removing cells and features with low counts (Seurat parameters: min.cells = 10, min.features = 100, nFeature_RNA > 200 nFeature_RNA < 1200).
Dimensionality reduction and clustering analysis were performed using PCA and graph-based KNN as implemented in the Seurat Package. A total of 4 clusters were identi ed by the KNN algorithm. To make the size of the data manageable, each cluster was downsampled to include 800 cells. Global differential expression performed on each cluster (log(FC) > 1 and adjusted p-value < 0.01) identi ed 544 differentially expressed genes. Pseudo-time analysis: Pseudo-time analysis was performed on the rst two PCA components by tting a principal curve to the data and orthogonally projecting the cells on the curve. Gene expression curves were then constructed using the pseudo-time and the start of gene expression was identi ed by cross correlation analysis between the bulk and single cell expression curves. Further details can be found in 118 .

QUANTIFICATION AND STATISTICAL ANALYSIS
All statistical analysis was performed in Graphpad PRISM 9. IC 50 values were calculated using the Non-linear regression function (variable slope -four parameters, least squares regression). Statistical signi cance of IC 50 changes were determined using ANOVA. All image analysis was performed in Zen 2 (Zeiss) and ImageJ/Fiji. Details of each analysis can be found in the corresponding gure legend.    increases cAMP levels. The increased cAMP activates PKAc1, which is required for invasion. PKAc1 activity also reduces egress, potentially through inhibition of the PKG signaling pathway. Solid lines represent pathways with a high con dence based on available data for B. divergens. Dashed lines and proteins marked with a "?" indicate processes based primarily on the RNAseq data from this paper and orthology to P. falciparum or T. gondii.