A toolbox of engineered mosquito lines to study salivary gland biology and malaria transmission

Mosquito saliva is a vehicle for the transmission of vector borne pathogens such as Plasmodium parasites and different arboviruses. Despite the key role of the salivary glands in the process of disease transmission, knowledge of host-pathogen interactions taking place within this organ is very limited. To improve the experimental tractability of the salivary glands, we have generated fluorescent reporter lines in the African malaria mosquito Anopheles coluzzii using the salivary gland-specific promoters of the anopheline antiplatelet protein (AAPP), the triple functional domain protein (TRIO) and saglin (SAG) coding genes. Promoter activity was specifically observed in the distal-lateral lobes or in the median lobe of the salivary glands. Besides a comparison of the expression patterns of the selected promoters, the fluorescent probes allowed us to evaluate the inducibility of the selected promoters upon blood feeding and to measure intracellular redox changes. We also combined the aapp-DsRed fluorescent reporter line with a pigmentation-deficient yellow(-) mosquito mutant to assess the feasibility of in vivo microscopy of parasitized salivary glands. This combination allowed locating the salivary gland through the cuticle and imaging of individual sporozoites in vivo, which facilitates live imaging studies of salivary gland colonization by Plasmodium sporozoites.


Author summary
Over the course of evolution, a number of pathogens have developed the ability to replicate within and to be transmitted by mosquitoes. An important adaptation of pathogens to be successfully transmitted is their ability to colonize mosquito slivary glands, as the timing and efficiency with which pathogens enter the glands directly affects their transmission. Despite the importance of salivary glands in disease transmission, host-pathogen interactions in this organ are poorly understood. To facilitate biological studies, we developed fluorescent reporter lines in the malaria mosquito Anopheles coluzzii by placing fluorescent probes under the control of the salivary gland-specific promoters of the anopheline antiplatelet protein (AAPP), triple functional domain protein (TRIO), and saglin (SAG) genes. We used these lines to characterize inducibility by blood meals and Introduction Vector borne diseases have been a major health burden in the history of mankind. An efficient way to counteract diseases transmitted by blood sucking arthropods is to prevent transmission by reducing the number of vectors in the environment or by preventing vectors from biting humans in risk areas. As a consequence of malaria control interventions, approximately 663 million clinical malaria cases could be averted between 2000 and 2015. The most effective measure was the use of insecticide treated bet nets, which prevented an estimated 68% of cases [1]. Still, preventive measures are counteracted by the very effective transmission of vector borne pathogens, which hampers eradication approaches. The key organ for transmission of vectorborne diseases is the salivary gland, as pathogens must colonize this organ to subsequently enter the saliva and be deposited in the skin during a blood meal. Host-pathogen interactions that take place in this organ are crucial for the spread of pathogens, as the concentration of pathogens in the mosquito saliva, as well as the composition and quantity of the saliva itself, are critical for successful transmission. To successfully colonize salivary glands, different pathogens have developed different strategies. While vector-borne viruses such as DENV cause systemic infection of the mosquito and replicate in the salivary glands [2], malaria parasites of the genus Plasmodium proliferate exclusively in the mosquito midgut. After infection of the mosquito midgut through the uptake of a blood meal from an infected host, Plasmodium parasites form cysts, called oocysts, below the basal lamina of the midgut epithelium where they undergo multiple rounds of DNA replication. Rapid cell division, called sporogony, then follows, leading to the generation of several thousands of sporozoites per oocyst. These cells are released from oocysts and travel from the midgut to the salivary glands that they specifically recognize and invade. After being deposited along with mosquito saliva in the skin during a blood meal, sporozoites traverse the dermis, enter blood capillaries and reach the human liver where they pursue development. On the path of sporozoites from the mosquito midgut to the liver of its intermediate host, the salivary glands represent a special niche in which they mature before being transferred to the host with the mosquito saliva.
Although the general life cycle of Plasmodium is well known, we lack detailed knowledge of the temporal relationships and host-parasite interactions that occur in the various tissues and cells which the parasite, particularly the sporozoite, traverses and invades. While it is believed that the invasion process of the salivary glands by sporozoites that have egressed from oocysts is rapid (~1-10 minutes), in vivo data supporting this assumption is lacking [3]. Still, the time a sporozoite needs to reach and successfully invade the glands is an important parameter, because at this timepoint, a mosquito is considered infective to humans. In addition, microscopic examinations of infected glands revealed that sporozoites form large aggregations in organized bundles in the cavities of the acinar cells [4]. The degree of order in these structures raises the question of whether the sporozoites recognize each other and possibly enter a quiescent phase to bridge the time until the vector next blood meal, or whether these bundles form by mechanical constraints alone. Based on the accumulation of sporozoites in specific sub-regions of salivary glands, it has also been hypothesized that sporozoites may preferentially invade the gland at specific sites [5]. However, since no receptor-ligand interaction between sporozoites and acinar cells has yet been clearly identified, the detection of such entry sites is difficult to prove. The study of gland-sporozoite or sporozoite-sporozoite interactions in the gland has generally been limited to the observation of infected tissues ex vivo because the highly pigmented cuticle of the mosquito makes microscopy of parasite stages in living mosquitoes difficult and, at the same time, an in vitro culture system for the insect stages of Plasmodium is not available. While ex vivo studies significantly contributed to the knowledge about Plasmodium we have today, samples obtained in this way are less suitable for imaging dynamic processes because dissected tissues change and decompose rapidly. In addition, the study of processes taking place in salivary glands has been complicated by the lack of molecular tools for mosquitoes like fluorescent reporter lines with tissue-specific gene expression, which have been slow to develop due to difficulties in transgenesis. The implementation of CRISPR/Cas9 technology in mosquito research has been a decisive step forward, which will give access to the analysis of the molecular mechanisms underlying mosquito-pathogen interactions at an unprecedented level of details. Females of Aedes spp. and Anopheles spp. mosquitoes possess two salivary glands each consisting of two lateral lobes and one median lobe. In addition, the lateral lobes are further subdivided into a distal-lateral and a proximal-lateral part which differ visually in cell size and in the size of formed cavities in which cells secrete saliva components [6]. Transcriptomic studies have shown that salivary glands display a tissue-specific expression profile in which many proteins thought to have a function in blood uptake are indeed absent in male salivary glands [7]. Consistent with the reduction in expressed proteins, male salivary glands are smaller and differ in their morphology, consisting of only one lobe per gland. Detailed knowledge about the expression of proteins in the salivary glands could be obtained by applying ex vivo approaches. In situ hybridization of gland-specific transcripts on dissected Aedes aegypti salivary glands has shown that genes are spatially expressed in the median, distal-lateral, and proximal-lateral lobes or combinations thereof [8], suggesting that different sections of the salivary gland may provide different components of mosquito saliva. In addition, Plasmodium sporozoites have been shown to invade primarily the distal-lateral lobes [5], which has led to speculation about a specific protein interface used by sporozoites to recognize the distal-lateral sections.
Recombinant salivary proteins considered to have therapeutic potential have been produced and studied in vitro for their vasodilator, anticoagulant and anti-inflammatory properties. Examples include aegyptin [9], anopheline [10], and anopheline anti-platelet protein (AAPP) [11], all of which act as anticoagulants. From this selection, AAPP is one of the few proteins that have also been studied in vivo. In A. stephensi, the generation of one of the first transgenic mosquito lines showed that the aapp promoter drives DsRed expression exclusively in the distal-lateral lobes [12]. In addition to their primary physiological function, salivary gland proteins suspected to play a role in the spread of pathogens such as the malaria parasite Plasmodium have been the focus of several studies. Protein-protein interaction-based approaches using sporozoite surface proteins and salivary gland proteins identified the proteins saglin and CSPBP as potential determinants for sporozoite invasion [13,14]. Knockdown experiments of both proteins have shown that sporozoite numbers in the salivary glands are reduced compared to controls, indicating a lower invasion rate. However, overexpression of saglin in the distal-lateral lobes of A. stephensi did not result in an increase in the number of sporozoites colonizing the salivary glands, raising doubt about the importance of saglin in this process [15]. Another protein that has been shown to affect sporozoite transmission is AgTRIO (triple functional domain protein), that, like saglin, belongs to the SG1-protein family [7,16]. Active and passive immunization of mice with recombinant AgTRIO was shown to diminish transmission of P. berghei sporozoites [17,18]. Interestingly, TRIO expression increases in the presence of P. berghei sporozoites, suggesting that sporozoites in salivary glands may influence protein expression of their host cells to potentially increase transmission efficiency [17]. Here we generated and characterised three fluorescent salivary gland reporter lines using the upstream regions of the aapp, trio and saglin genes in the African malaria vector Anopheles coluzzii, and we used one of them in combination with a pigment-deficient line to perform in vivo live imaging of sporozoites invading salivary glands. We expect the lines described here will be of use to investigate salivary gland and sporozoite biology.

Generation of transgenic mosquito lines with salivary gland-specific reporter expression and low pigmentation
To generate salivary gland reporter lines, the 5' upstream sequences of saglin (AGAP000610), aapp (AGAP009974) and trio (AGAP001374) were used to drive the expression of DsRed and hGrx1-roGFP2 from transgenic constructs integrated in the X1 docking line (Figs 1A and S1A and S1 Appendix) with an attP site on chromosome 2L [19]. The fluorescence reporter hGrx1-roGFP2 was chosen because it bears the advantages of EGFP and enables the measurement of the oxidation level of the intracellular glutathione pool by ratiometric quantification [20]. We used the aapp promoter sequence from Anopheles coluzzii orthologous to the one used to drive salivary gland-specific transgene expression in Anopheles stephensi [12]. For both saglin and trio, we defined putative promoter regions according to the genomic context. In addition, we also used CRISPR/Cas9 to integrate a fluorescence cassette encoding EGFP into the endogenous saglin locus, replacing the coding sequence, to study saglin promoter activity in its native genomic context. Indeed, activity of the cloned promoter selected for ectopic expression was uncertain due to the shortness of the apparent saglin promoter region (220bp, S2 Fig). For this, site directed knock-in mutagenesis was performed by injecting a plasmid encoding three guide RNAs targeting saglin and carrying a repair template marked with 3xP3-EGFP into mosquito eggs from a line with germline-specific expression of Cas9 (vasa-Cas9), generating a sag(-)KI line (Figs 1B and S3A). To place the EGFP gene directly under control of the saglin promoter, the 3xP3 promoter used to select transgenic mosquito larvae was flanked by lox sites to enable its excision by Cre recombination (S3A Fig, Material & Methods). Similarly, we generated a knockin (KI) into the yellow gene that encodes a protein required for the synthesis of black pigments in insects [21,22]. For this, we used a plasmid with guide RNAs specific for yellow and corresponding homology arms flanking a fluorescence cassette expressing EGFP surrounded by lox sites (S3B Fig). Through crossing of salivary gland reporter lines with low pigmented yellow(-)KI mosquitoes, we aimed to improve the fluorescence signal of the expressed reporter and extend the range of applications, e.g. for in vivo imaging. Genotyping by PCR confirmed the successful generation of the salivary gland reporter lines aapp-DsRed, aapp-hGrx1-roGFP2, trio-DsRed and sag-DsRed (S1B Fig) as well as the mosquito lines sag(-)KI and yellow(-)KI (S3C and S3D Fig). Loss of the 3xP3 promoter in sag(-)EX mosquitoes (mosquitoes lacking the 3xP3 promoter after Cre mediated excision) was confirmed by PCR in the progeny of a cross between the sag(-)KI line and a mosquito line expressing Cre recombinase in the germline (S3E Fig) [19].

Fig 1. Lobe-specific reporter expression observed in female aapp-DsRed, aapp-hGrx1-roGFP2, trio-DsRed and sag(-)EX mosquitoes. A)
FC31 integrase-mediated transgene insertion. The aapp, trio and saglin promoter sequences were fused to DsRed or hGrx1-roGFP2 within the pDSAP vector which enables chemical selection of mosquito larvae by puromycin. Transgenes were integrated by FC31 integrase-mediated recombination into the X1 line harbouring an attP site on chromosome 2L. B) Cas9-mediated knockin of a fluorescence cassette in the saglin gene (AGAP000610). Upstream and downstream sequences of saglin were cloned into the pENTR vector flanking a fluorescence cassette expressing EGFP under the control of the 3xP3 promoter. Three guides targeting saglin controlled by U6 promoters were encoded on the same pENTR vector. Transgenesis was performed in mosquitoes with germline-specific expression of Cas9 (vasa-Cas9). C) Fluorescent reporter expression in mosquitoes homozygous for trio-DsRed, aapp-DsRed, aapp-hGrx1-roGFP2 and sag(-)EX (3xP3 promoter excised) as well as in females heterozygous for trio-DsRed and aapp-DsRed, and trio-DsRed and aapp-hGrx1-roGFP2. Salivary glands of wild type (G3) are shown as controls. Illustrations on the left of each panel indicate the genotype of the gland as well as the observed fluorescence pattern. Images of salivary glands are shown in brightfield beside an image of the same gland in the relevant fluorescence channel. Red arrows in brightfield images highlight the pink parts of aapp-DsRed and trio-DsRed salivary glands in the absence of fluorescence excitation indicating the high expression of DsRed. Median (M) and distal-lateral lobes (DL) are indicated on the images of wild type glands. Salivary glands were dissected from 2-10 day old females that had not previously received blood feeding. Scale bars: 250 μm. https://doi.org/10.1371/journal.ppat.1010881.g001

PLOS PATHOGENS
A toolbox of engineered mosquito lines to study salivary gland biology glands (Fig 1C). While no fluorescence was observed for DsRed driven by the cloned upstream sequence of the saglin gene in dissected salivary glands (S2A Fig), promoter activity was detected in the line where EGFP replaced the saglin gene (Fig 1C), indicating that essential transcriptional elements localize outside of the intergenic sequence between AGAP000609 and AGAP000610 (saglin) (S2B Fig). Interestingly, fluorescence signals in glands of female aapp-DsRed, aapp-hGrx1-roGFP2 mosquitoes were restricted to the distal-lateral lobes, while female trio-DsRed and sag(-)EX mosquitoes displayed fluorescence exclusively in the median lobe (Fig 1C). This difference became especially evident when crossing aapp-hGrx1-roGFP2 mosquitoes with trio-DsRed mosquitoes giving rise to a trans-heterozygous offspring displaying hGrx1-roGFP2 expression in distal-lateral lobes combined with DsRed expression in the median lobe (Fig 1C). Similarly, when crossing aapp and trio DsRed reporter lines, the entire salivary gland was red fluorescent in the F1 progeny (Fig 1C). Expression of trio-DsRed and aapp-DsRed was so strong that the respective parts of the salivary gland appeared pink colored even in brightfield microscopy (Fig 1C). Of note, fluorescent reporter expression in the salivary glands of aapp-DsRed, aapp-hGrx1-roGFP2 and trio-DsRed mosquito colonies was restricted to females while no fluorescence signal could be observed in this organ in males (S4A Fig). Imaging of aapp-DsRed, aapp-hGrx1-roGFP2 and trio-DsRed pupae during development revealed no distinct fluorescence signal in the head part where the salivary glands are located (S4B Fig). Still, DsRed expression was visible in the salivary glands of aapp-DsRed females that had been dissected directly after emergence (Fig 2A), indicating that the aapp promoter is already active in pharate adults. In contrast, no DsRed expression was observed in the glands of freshly hatched trio-DsRed females (Fig 2A). To evaluate the expression profile over time we performed a time-course experiment by dissecting and imaging salivary glands of trio-DsRed and aapp-DsRed females every day. DsRed expression controlled by aapp and trio promoters increases steadily until day 5 after hatching (Fig 2B). The increase in DsRed expression over time in the salivary glands of aapp-DsRed and trio-DsRed females could also be followed in transmitted light based on the pink coloration of the distal-lateral lobes (S4C Fig). In Anopheles stephensi, aapp expression was shown by qRT-PCR to be upregulated upon blood feeding [12]. We followed DsRed fluorescence levels in the salivary glands of aapp-DsRed and trio-DsRed sibling mosquitoes that were fed on blood or kept on sugar (Fig 2C and  2D). Our results confirm that the aapp promoter is induced upon ingestion of a blood meal while there was no difference between blood-fed and sugar-fed in females expressing the trio-DsRed transgene.

Spatial expression pattern of the saglin and trio promoters in the median lobe
While expression of the fluorescent probes DsRed and hGrx1-roGFP2 in the mosquito lines aapp-DsRed, trio-DsRed and aapp-DsRed is driven by the cloned aapp and trio promoter sequences, the selection of transgenic sag(-)KI larvae required the use of the synthetic 3xP3 promoter to allow screening for expression of EGFP in neonate larvae. To assess the activity of the native saglin promoter, the 3xP3 promoter was flanked by lox sites that allowed its excision using Cre to generate the sag(-)EX line (S3E Fig). Imaging of salivary glands (dissected tissues or through the cuticle in live mosquitoes) from sag(-)KI and sag(-)EX females revealed that EGFP expression is much stronger in specimens expressing the sag(-)KI transgene, indicating that the synthetic 3xP3 promoter enhances the native transcriptional activity of the saglin promoter (Fig 3A). As for the aapp promoter, reporter expression in Sag(-)EX was detected in salivary glands of mosquitoes that had just emerged (Fig 2A), and increased after blood feeding (Fig 2C and 2D). Of note, EGFP signal was 1.8 and 2.6 fold stronger at the base than at the

PLOS PATHOGENS
A toolbox of engineered mosquito lines to study salivary gland biology apex of the median lobe of the salivary gland in sag(-)KI and sag(-)EX females, respectively (Fig 3A and 3B). The mean fluorescence of sag(-)KI was 5.5-fold and 3.8-fold higher than that of sag(-)EX at the apex and base, respectively (Fig 3B). Next, we crossed sag(-)KI and trio-DsRed mosquitoes. Confocal imaging of salivary glands dissected from heterozygous individuals expressing both transgenes revealed a striking difference in the expression pattern of the trio and saglin promoters. DsRed expression driven by trio was observed strongest in the apex of the median lobe with low to no expression in cells of the base (Fig 3C and 3D). In contrast EGFP expression driven by the saglin promoter was detectable along the median lobe in an opposite gradient (Fig 3B and 3D). Imaging of the base of a gland expressing trio-DsRed and sag(-)KI showed a remarkably sharp expression boundary between cells with high and low saglin promoter activity. Only few high producer cells close to the boundary were observed to Images of salivary glands dissected directly after mosquitoes emerged. Fluorescence signals were observed in salivary glands expressing aapp-DsRed and sag(-)EX (marked by red arrowheads) but not trio-DsRed. Scale bar: 250 μm. B) Mean fluorescence intensity measurements of dissected aapp-DsRed and trio-DsRed salivary glands from day one to day five after hatching. Data were normalized for the highest exposure time. Each dot represents a single gland (2-5 glands / timepoint) and bars the median. � p<0.01, one-way-ANOVA (Dunn's Multiple Comparison Test). C) The blood meal inducibility of the trio, aapp and saglin promoters was assessed by providing sibling females either with a blood meal or sugar solution. Approximately 24h later, salivary glands were dissected, imaged and the mean DsRed or EGFP fluorescence intensities were quantified. Data pooled from three (aapp-DsRed and trio-DsRed) or four (sag(-)EX) independent experiments. Horizontal black bars represent the median. �� p<0.001 and ��� p<0.0001, Mann Whitney test. ns: not significant. D) Images of salivary glands dissected 24h after blood feeding in comparison to sugar fed controls. Scale bar: 250 μm. https://doi.org/10.1371/journal.ppat.1010881.g002

PLOS PATHOGENS
A toolbox of engineered mosquito lines to study salivary gland biology

PLOS PATHOGENS
A toolbox of engineered mosquito lines to study salivary gland biology co-express trio-DsRed (Fig 3E). Using z-stacks of confocal images from dissected sag(-)KI salivary glands stained with Hoechst 33342, we further determined that the median lobe consists of 144 cells on average, of which 49 cells display high and 95 low saglin promoter activity. Considering the observed differences in the mean fluorescence between apex and base and assuming that the expression of EGFP represents the expression of saglin, cells at the base of the median lobe contribute~88% to Saglin expression, while the cells in the apex contribute onlỹ 12%, although the apex area contains nearly twice as many cells as the base.

Subcellular localization of fluorescent reporters
During first microscopic examinations using low magnification, different intracellular localizations of the fluorescent proteins DsRed, EGFP and hGrx1-roGFP2 were observed (Fig 1C), which was confirmed by confocal microscopy. DsRed expressed in the salivary glands of aapp-DsRed and trio-DsRed mosquitoes appeared spotty and displayed a vesicle-like pattern within the cytoplasm of acinar cells, while hGrx1-roGFP2 and EGFP expressed in the glands of aapp-hGrx1-roGFP2 and sag(-)EX mosquitoes displayed a more cytoplasmic localization (Fig 4).
Due to the anatomy of the secretory cells, which form an inward cavity into which saliva is secreted, the vesicle-like DsRed accumulations localized mainly at the outer rim of the salivary gland. In contrast, localization of hGrx1-roGFP2 and EGFP was homogenous in the cytoplasm but appeared to be enriched around nuclei (S5 Fig).

Variation of DsRed expression in trio-DsRed mosquitoes
During the generation of the salivary gland reporter lines and their subsequent outcrossing we observed that most L1 mosquito larvae carrying the trio-DsRed transgene displayed fluorescence in the eyes (Fig 5A), which was absent in the two other reporter lines aapp-DsRed and aapp-hGrx1-roGFP2. Interestingly, a few L4 stage larvae of the trio-DsRed line even displayed strong ubiquitous DsRed expression (Fig 5A). At the pupal stage, DsRed expression was detected in the ocellus, an eye spot close to the main compound eye, the antennae and/or throughout the body. The observed expression patterns were highly variable and ranged from completely non-fluorescent pupae to pupae expressing DsRed in all three tissues (S6A Fig). In adult mosquitoes, DsRed signal was less distinct compared to pupae although complete body fluorescence was occasionally observed in both males and females, with the strongest expression of DsRed in the salivary glands of females and the palps and antennae of males (Fig 5B).
Dissections of adult mosquitoes displaying whole body fluorescence at the pupal stage revealed strong and sex-independent DsRed signals in Malphigian tubules, an insect organ that performs functions similar to the vertebrate kidney (Fig 5C). Of note, fluorescence in tubules strictly correlated to the high pupal fluorescence phenotype and was never observed in mosquitoes hatched from low fluorescent pupae (Fig 5D). We next investigated sex-specificity of each fluorescence pattern at the pupal stage and inheritance of the high and low fluorescent trio-DsRed transgene. Non-fluorescent pupae were always females, while fluorescent antennae number of measured salivary glands is indicated above each column. Data from four (sag (

PLOS PATHOGENS
A toolbox of engineered mosquito lines to study salivary gland biology were observed exclusively in male pupae (S6B Fig). In contrast, the presence of a fluorescent ocellus or body fluorescence was sex-independent. Quantification of the observed expression pattern revealed that a fluorescent ocellus, either alone or in combination with body fluorescence, was most frequently observed, and that the ratio of pupae that displayed body fluorescence to those that did not was close to 50% (S6C

Glutathione redox state and morphology of acinar cells expressing hGrx1-roGFP2
To assess the redox sensitivity of hGrx1-roGFP2 expressed in the aapp-hGrx1-roGFP2 line, salivary glands were dissected and treated either with diamide or dithiothreitol to oxidize or reduce the fluorescent probe, respectively. In addition, salivary glands were treated with Nethyl-maleimide to preserve the native redox state of the dissected tissue. Samples were imaged at 405 and 488 nm and the ratio of both wavelengths was plotted (Fig 6A), revealing a dynamic range of 4.2 between fully oxidized and fully reduced probes (Fig 6B). Interestingly, the ratio of native salivary glands was very similar to measurements performed on samples treated with dithiothreitol (Red), suggesting that the glutathione pool is largely reduced in acinar cells ( Fig  6B). Infection with P. berghei sporozoites did not affect the redox state of salivary glands ( Fig  6C). Of note, initial imaging experiments pointed towards a thinning of the tips of the distallateral lobes in salivary glands expressing aapp-hGrx1-roGFP2, while glands obtained from aapp-DsRed, trio-DsRed and sag(-)KI or sag(-)EX females appeared similar to wild type (Figs 1  and 6D). Measurement of the diameter at the "neck-region" of the distal-lateral lobes confirmed its thinning in salivary glands expressing aapp-hGrx1-roGFP2 compared to wild type and aapp-DsRed (Fig 6E). This difference is likely due to a flattening of the cell body of the acinar cells within the tips, which was observed by confocal microscopy (Fig 4).

Salivary gland reporter lines as a tool to investigate sporozoite interactions with salivary glands
We next investigated whether our highly fluorescent salivary gland reporter lines could facilitate in vivo imaging of Plasmodium development through the cuticle of a living mosquito using a GFP expressing parasite strain (Δp230p-GFP) of the rodent malaria parasite P. berghei (Fig 7A) [23]. This strain is bright enough to make parasites visible through the mosquito cuticle, allowing pre-sorting of infected mosquitoes based on fluorescence in the mosquito midgut, the salivary glands and the wing joint, a localization where hemolymph sporozoites tend to accumulate (Fig 7B). To test if the expression of a fluorescent reporter in the salivary glands alters parasite development in the mosquito, we infected the three reporter lines trio-DsRed, aapp-DsRed and aapp-hGrx1-roGFP2, as well as the parental line G3. Parasite counts observed in reporter lines varied from 118 to 142 oocysts and were not significantly different from controls (121 oocysts An average of 10.000 salivary gland sporozoites per mosquito is usually observed in our laboratory, with numbers fluctuating from 5.000 to more than 20.000 sporozoites per mosquito in non-transgenic lines, indicating no significant difference between the G3 control and reporter lines. To normalize the counts of salivary gland sporozoites relative to differences in oocyst numbers, we calculated the number of sporozoites that had successfully invaded the salivary glands per oocyst. Again, no significant difference was detected between reporter lines (86-146 sporozoites per oocyst) and G3 control (133) mosquitoes (S8C Fig).
While the expression of a fluorescence reporter makes it simple to detect the location of the salivary glands through the cuticle, light absorption by pigments and light scattering by chitin generate signal losses reducing image quality and resolution. To decrease light absorption, we crossed aapp-DsRed mosquitoes, displaying highly fluorescent salivary glands without altered

PLOS PATHOGENS
A toolbox of engineered mosquito lines to study salivary gland biology morphology, with yellow(-)KI mosquitoes presenting a pale and yellowish cuticle (Fig 7A and  7B). To assess if the 3xP3 driven expression of EGFP in the low pigmented yellow(-)KI line would interfere with the imaging of GFP positive P. berghei parasites, adult yellow(-)KI and WT females were imaged in the GFP channel to compare emitted fluorescence signals (S8D

PLOS PATHOGENS
A toolbox of engineered mosquito lines to study salivary gland biology EGFP expression in the ovaries is likely due to the endogenous promoter activity of the yellow promoter. However, no significant increase in green fluorescence was observed at the throat area where the salivary glands localize compared with WT, suggesting that imaging of GFPpositive parasites would be feasible.

PLOS PATHOGENS
A toolbox of engineered mosquito lines to study salivary gland biology As expected, yellow(-)-aapp-DsRed mosquitoes displayed a low pigmented cuticle in combination with highly fluorescent salivary glands (Fig 7B). The decrease in pigmentation greatly increased the observed GFP signal visible through the cuticle, notably in strongly pigmented areas that appear almost black in WT mosquitoes (Fig 7C). We also investigated whether GFP-or mCherry-positive sporozoites can be imaged in fluorescent salivary glands. Imaging of dissected infected glands showed that fluorescent sporozoites can be clearly distinguished from the fluorescent signal of salivary glands at low magnification while lobe-specific fluorescence expression allowed for simple differentiation between median and distal-lateral lobes (Fig 7D). In addition, imaging showed that all salivary gland reporter lines may have a high sporozoite load, suggesting that fluorescence expression in salivary glands does not affect individual salivary gland sporozoite colonization.
We next imaged mosquitoes infected with P. berghei Δp230p-GFP between day 17 and day 20 after infection, when the majority of sporozoites have invaded the salivary glands. Mosquitoes were pre-sorted for GFP fluorescence in the salivary glands (S9A Fig) as well as in the wing joint (Fig 7B) to confirm the presence of sporozoites. In a subset of mosquitoes (less than 1 out of 5), we observed that the salivary glands localized very close to the cuticle (S9B and S9C Fig). To facilitate in vivo imaging, we further selected mosquitoes positive for sporozoites and with cuticular localization of at least one salivary gland and prepared them for imaging (S9D Fig). Confocal imaging enabled to visualize precisely single sporozoites through the cuticle close to the salivary gland as well as subcellular structures, like the cavities of acinar cells of the gland itself (Fig 8A). However, we noted that a significant portion of the DsRed signal was also seen in the GFP channel, especially at lower magnification, suggesting a possible spillover of the DsRed signal. This could possibly occur due to the concentration difference between DsRed in the acinar cells and EGFP in the sporozoites (S9E Fig). Nevertheless, spillover effects did not affect detection of sporozoites at higher magnification. Quantification of the recorded movies revealed that 26% of persistent sporozoites in the thorax area (at least 5 minutes in the field of view) perform pattern of active movement while~19% are attached but not moving and~55% are floating ( Fig  8A). The observed movement patterns resembled in most cases the patch gliding previously described in vitro. During this type of movement, the sporozoite adheres in the center and performs jerky back-and-forth movements along its adhesive patch (Fig 8B and S1 and S2 Movies) [24,25]. Due to the brevity of the recorded movies (5 minutes), many actively moving sporozoites appeared to remain on the site without making large terrain gains. However, in some cases, the position of the adhesive patch was observed to change slightly over time, such that sporozoites made small terrain gains in one direction while continuously sliding back and forth. Importantly, we were also able to observe sporozoites inside salivary glands (Fig 8C).

Discussion
Mosquito salivary glands have long interested entomologists, parasitologists, virologists and microbiologists because they are a key organ in the transmission of vector borne diseases. Still, the host-pathogen interactions that allow the specific recognition and colonization of this organ by different pathogens has been understudied, probably in part because of the relative difficulty to access and visualize these processes in comparison to midgut invasion. The reporter lines we generated in this study represent valuable tools to circumvent some of these issues: they enable [1] easy visualization of the glands, even in brightfield or DIC microscopy, [2] distinguishing distal-lateral and median lobes of the salivary glands, and [3] imaging of single sporozoites in proximity to and inside salivary glands in vivo.
We used the promoters of three salivary gland genes, aapp, trio and saglin, to drive fluorescent reporter expression in A. coluzzii mosquitoes. The different fluorescence patterns

PLOS PATHOGENS
A toolbox of engineered mosquito lines to study salivary gland biology obtained in the salivary glands of these lines show that the proteomes of median and distal-lateral lobes are diverse and most likely contribute different saliva components. In addition, the variable fluorescence intensity in the distal-lateral lobes of both aapp lines (stronger towards the apex) point towards an even more complex expression pattern along the lobe. This hierarchical expression was even more evident for the trio and saglin promoters, both of which are active in the median lobe but with opposite expression patterns. Saglin is strongly active in

PLOS PATHOGENS
A toolbox of engineered mosquito lines to study salivary gland biology cells at the base and less active in the apex, in contrast, trio expression is absent in cells with high saglin activity, but very strong in cells with low saglin activity. While the expression pattern of trio has not been investigated before, the observed activity of the saglin promoter is coherent with immunofluorescence studies performed in A. stephensi [15]. Expression of saglin in the median lobe is puzzling since the protein has been shown to affect colonization of the salivary glands by Plasmodium sporozoites, [13], whereas parasites prefer to invade the distal-lateral lobes [5] that lack activity of the saglin promoter. A possible explanation is that its site of expression is not the site where it interacts with parasites.
The absence of reporter expression in mosquitoes expressing DsRed controlled by the saglin 5' upstream sequence is likely due to the shortness of the cloned sequence. The saglin gene localizes in the middle of an array of four genes believed to be salivary gland-specific [16] and separated by short intergenic regions (S2 Fig). Regulatory sequences controlling transcription are likely located upstream and/or downstream of this array and might be shared by all four genes.
Notably, the reporter lines display different subcellular localizations of hGrx1-roGFP2, EGFP and DsRed. While all expressed fluorescent proteins lack any targeting signal which could explain a different localization, only hGrx1-roGFP2 (aapp) and EGFP (saglin) localized within the cytoplasm. In contrast, DsRed displayed a vesicular pattern in both the median (trio) and the distal-lateral lobes (aapp), while a DsRed2 variant driven by the aapp promoter in A. stephensi mosquitoes was shown to be cytoplasmic [12]. Although AAPP was shown previously to localize in a vesicle-like fashion [6], we believe that the observed localization difference is likely artefactual due to fortuitous trafficking signals present in the reporter proteins. First, vesicular localization of DsRed was observed with both the trio and aapp promoters, and second, only DsRed appeared vesicular, whereas hGrx1-roGFP2 was cytoplasmically localized, although both proteins were expressed from the same aapp sequence. Comparison of the DsRed sequence used here with the previously described DsRed2 sequence expressed in A. stephensi revealed mutations mainly in the N-terminal region and the addition of 11 unrelated amino acids at the C-terminus of DsRed2 [12]. It is likely that these sequence differences between the two DsRed versions lead to the formation of a cryptic trafficking signal recognized by the acinar cells. Interestingly, although we did not investigate their nature, we also observed the presence of vesicles in the salivary glands of wild type mosquitoes with similar distribution and morphology and displaying weak red autofluorescence (S2 Fig). DsRed could potentially be trafficked to these or similar compartments explaining its unique localization. Differences in salivary gland morphology between aapp-DsRed and aapp-hGrx1-roGFP2 expressing mosquitoes are likely due to the expression of human Glutaredoxin 1 (hGrx1) and not of roGFP2, a protein derived from EGFP with 2 amino acid changes. Indeed, no change in morphology was detected in sag(-)EX and sag(-)KI mosquitoes expressing EGFP alone in the median lobe. Still, EGFP appeared less expressed in sag(-)EX compared to hGrx1-roGFP2 in aapp-hGrx1-roGFP2 (Fig 1) and we cannot exclude that the acinar cells of the median and distal-lateral lobes react differently to the expression of the respective reporters, or that the cells show different responses to roGFP2 and EGFP. Despite its influence on cell morphology, the salivary glands of hGrx1-roGFP2 expressing mosquitoes retained their integrity even in old mosquitoes and apoptosis of cells expressing the fluorescent reporter was not observed.
We also report a surprisingly large panel of fluorescence patterns in mosquitoes expressing the trio-DsRed transgene. The trio promoter activated transcription of DsRed in the salivary glands of all females (but not in males), but also in ocelli, antennae and Malpighian tubules as well as whole body in some but not all mosquitoes, independently of their sex but for antennae that were strictly restricted to males, and with a strict correlation between strong DsRed expression in pupal body and adult Malpighian tubules. The fact that trio expression was

PLOS PATHOGENS
A toolbox of engineered mosquito lines to study salivary gland biology reported to be salivary gland-specific [17] suggests that the region we selected upstream of trio does not contain all regulatory elements that control its natural expression, or that the genomic context of the X1 locus may influence its expression. To date, the number of available promoters able to drive ubiquitous expression of transgenes in Anopheles mosquitoes is limited [26], thus replacing DsRed in a selected highly fluorescent line could provide a new means to ubiquitously express transgenes. By sorting and crossing according to the observed pupal fluorescence phenotype, it was possible to separate alleles conferring high and low pupal body fluorescence, which enabled the production of a colony enriched with the high pupal fluorescence phenotype and a colony showing only low pupal body fluorescence. This observation pointed towards a genetically encoded difference which is inherited in a Mendelian manner rather than a transient (epi)genetic modification. Genotyping confirmed the presence of a single copy of the transgene in the X1 locus in both weakly and highly fluorescent mosquitoes. Since fluorescent ocelli, antennae and whole-body expression were only observed in the trio-DsRed population, we propose that the different DsRed expression patterns are locus-specific variegation effects caused by interactions between the selected trio promoter sequence and polymorphic regulatory elements that could be linked or unlinked to the transgene. Although unlikely, we cannot rule out the possibility that some mosquitoes carried an extra copy of the transgene that caused the highly fluorescent phenotype. Still, DsRed positive ocelli and antennae were observed in both high and low fluorescence populations, which is most likely explained by a variable DsRed expression from the transgene inserted at X1. While the observed DsRed expression patterns in larvae and pupae are helpful in the laboratory to select transgenic individuals, it also urges caution in interpreting results as it clearly shows that transcription is contextual.
Intravital imaging has revealed the dynamics of parasite development inside its vertebrate host [27,28]. While optical windows in mice is a well implemented method [29], in vivo imaging through the cuticle of insects is rarely performed, mostly because it is difficult to achieve good resolution by imaging through the insect cuticle that scatters and absorbs light. Recent studies have explored the possibility to circumvent this issue through the application of clearing protocols on infected mosquitoes to improve resolution and enable imaging through whole mosquitoes [30,31]. However, while these improvements are remarkable, imaging of cleared specimens only represents snapshots whereas mapping of dynamic processes are desirable. Imaging of tissues ex vivo in combination with the use of hemolymph-like media allows the observation of dynamic processes [32]. Still, imaging ex vivo samples can only be performed for a limited amount of time and artifacts due to the artificial tissue environment cannot be excluded. A step towards intravital imaging was made recently by developing a protocol that combines feeding of dyes relevant for imaging with blood feeding prior to sample preparation [33]. Due to the inflation of the mosquito abdomen directly after feeding, cuticle plates open up and create temporary optical windows that can be exploited for microscopy. Still, in this case, mosquitoes were decapitated and tightly flattened between microscopy slide and cover slip, thus enabling optimal imaging but maintaining in vivo conditions only for a limited time. To demonstrate the use of the salivary gland reporter lines for imaging purposes, we crossed aapp-DsRed mosquitoes with the weakly pigmented mosquito line yellow(-)KI deficient for black pigment synthesis. In a subset of mosquitoes, salivary glands localized very closely to the cuticle allowing the recognition of subcellular structures of the gland in living mosquitoes. In addition, the more transparent cuticle enabled better visibility of individual sporozoites through the cuticle. It is generally believed that sporozoites, after they have been released from the oocyst, are passively transported to the salivary glands and engage in active motility only upon contact with acinar cells [3]. Our in vivo observations show that~26% of persistent thorax sporozoites perform active motility mostly in a patch gliding manner (S1 and

PLOS PATHOGENS
A toolbox of engineered mosquito lines to study salivary gland biology S2 Movies) [24,25]. This could be an indication that the motility of the sporozoites is already crucial to reach the salivary glands and ensure efficient colonization. Testing mutants that produce immobile sporozoites could provide more sophisticated information on the extent to which sporozoite motility promotes invasion of salivary glands.
Here we provide a proof of concept of in vivo salivary gland imaging, and we hope that the method can be further improved to gain a better insight in the invasion process of salivary glands by sporozoites. Potential spillover effects of the DsRed signal in salivary glands of aapp-DsRed females could possibly be avoided by using a P. berghei reporter line with stronger GFP expression [34]. Furthermore, the injection of fluorescent beads or dyes could be used as vitality measure to monitor the active movement of the hemolymph while, at the same time, highlighting contact areas with hemolymph and salivary glands. The localization of most sporozoites, separated from the objective by several cell layers, was a limiting factor in our settings, which could be improved by using two-photon microscopy instead of confocal microscopy, thus providing a better resolution when imaging in deep tissues. Recently, this method has been used for long-term monitoring of the ventral nerve cord in live adult Drosophila, although the cuticle was previously partially removed for this approach [35]. Sample selection and preparation, like partial embedding of mosquitoes as well as the choice of the immersion medium, could also lead to significant improvements. Furthermore, we did not exhaust the repertoire of possible genetic modifications to decrease pigmentation of the cuticle. Deletion of the genes tan and ebony, two enzymes which regulate the homeostasis of the pigment precursors dopamine and N-ß-alanyldopamine, in combination with the deletion of yellow could potentially further decrease cuticle pigmentation [21,36]. Another possibility could be the overexpression of arylalkylamine-N-acetyltransferase, an enzyme converting dopamine into the sclerotizing precursor molecule N-acetyldopamine, which was shown to affect pigmentation in different insect species by enhancing the production of colorless components of the cuticle [37]. A key improvement for in vivo imaging would be an increase in the number of mosquitoes displaying ideally positioned salivary glands. Preliminary monitoring experiments showed that the position of the gland does not significantly change over time in adult mosquitoes indicating that positioning occurs during metamorphosis. Should salivary gland positioning be genetically determined, selective breeding may help to improve the ratio of mosquitoes with cuticular salivary glands. Taken together, we hope that the transgenic lines developed in this study can contribute to improve in vivo imaging and salivary gland dissections in Anopheles mosquitoes, and enable real-time imaging of Plasmodium and other vector borne pathogens.

Ethics statement
Experiments were carried out in conformity with the 2010/63/EU directive of the European Parliament on the protection of animals used for scientific purposes. Our animal care facility received agreement #I-67-482-2 from the veterinary services of the département du Bas Rhin (Direction Départementale de la Protection des Populations). The use of animals for this project was authorized by the French ministry of higher education, research and innovation under the number APAFIS#20562-2019050313288887 v3. The generation and use of transgenic lines (bacteria, mosquitoes, parasite) was authorized by the French ministry of higher education, research and innovation under the number 3243.

Animals
Transgenesis was performed in the Anopheles coluzzii strain X1 [19] or in a strain expressing Cas9 from the X1 locus under control of the vasa promoter (transgenesis plasmid: Addgene #

PLOS PATHOGENS
A toolbox of engineered mosquito lines to study salivary gland biology 173670) and introgressed into the Ngousso genetic background. Wild type Anopheles coluzzii strains Ngousso and G3 were used as controls. CD1 mice purchased from Charles River or from our own colonies were used for both mosquito maintenance through blood meals, and for parasite maintenance and mosquito infections. For infections with rodent malaria parasites, we made use of the Plasmodium berghei lines Δp230p-GFP and Δp230p-mCherry derived from the wild type background P. berghei ANKA [41].

Breeding of mosquitoes
Anopheles coluzzii mosquitoes were kept in standard conditions (27˚C, 75-80% humidity, 12-hr/12-hr light/dark cycle). Larvae were hatched in pans of osmosed water and fed with pulverized fish food (TetraMin, Melle, Germany). Adult mosquitoes were maintained on a 10% sucrose solution. To propagate colonies, four to seven-day old mosquitoes were blood fed for 10-15 minutes on anesthetized mice and allowed to lay eggs on wet filter paper 2-3 days later. For mosquito blood feeding, female CD-1 mice (> 35g) were anesthetized with a mixture of Zoletil (42.5 mg/kg) and Rompun (8.5 mg/kg) in 0.9% NaCl solution.

Construct cloning for site-directed mutagenesis into yellow and saglin genes
For the design of guide RNAs (gRNAs), the coding sequence of saglin as well as the first exon of the yellow gene were manually searched for PAMs (NGG), with N representing any nucleotide. Candidate gRNA sequences were investigated for potential off-target effects using http:// www.rgenome.net/cas-designer/ (Cas-OFFinder) [43]. We selected gRNAs starting with a G considering the transcription start from the U6 promoter and differing by �3 nucleotides from putative off-target sequences in the A. coluzzii genome. For each knockin construct, linkers encoding the protospacer motif of three gRNAs ( Table 1) were generated by annealing two primers ordered from IDT DNA and cloned using BbsI into pKSB-sgRNA (Addgene #173671-3) vectors containing the U6 snRNA polymerase III promoter (AGAP013557), the invariable sequence of a standard crRNA-tracerRNA fusion [44], and the RNA polIII TTTTT terminator. The three complete gRNAs blocks were then assembled into a single pKSB-sgRNA destination plasmid by GoldenGate cloning using different but matching BsaI overhangs. For the construction of repair templates, regions of homology on the 5' side (saglin: primers P29 and P30; yellow: primers P19 and P20) and on the 3' side (saglin: primers P31 and P32; yellow: primers P21 and P22) of the region targeted by the gRNAs were amplified from Ngousso genomic DNA and subcloned in the pJET vector, generating plasmids pJET-5'saglin, pJET-5'yellow,

Salivary gland reporter lines sag-DsRed, trio-DsRed, aapp-DsRed and aapp-hGrx1-roGFP2
Following egg micro-injection, surviving pupae were sorted according to sex, hatched in separate cages (G0 generation) and crossed with~100 mosquitoes of the other sex from a mosquito line expressing CFP from the X1 docking site that was used as a balancer of the new, non-fluorescent but puromycin resistant transgenic insertions at the same locus. Neonate larvae from this cross were treated with puromycin (0.05 mg/mL). G1 pupae were subsequently sex sorted and crossed with CFP positive mosquitoes to expand the population. Adult puromycin resistant G2 females were sorted according to the observed fluorescence phenotype (DsRed or roGFP2) in their salivary glands, reflecting the expression of the trio-DsRed, aapp-DsRed or aapp-hGrx1-roGFP2 transgenes and crossed to CFP balancer males. For sag-DsRed females, sorting was not necessary, as transgenesis was performed separately by injection of the pDSAP-saglin-DsRed plasmid alone. Puromycin-resistant G3 larvae were raised, intercrossed and the G4 offspring was sorted by COPAS (Marois et al., 2012) for the absence of CFP to select lines homozygous for sag-DsRed, trio-DsRed, aapp-DsRed and aapp-hGrx1-roGFP2. Lines were then raised without puromycin treatment. Single females of all generated colonies were genotyped by PCR to monitor correct integration of the transgene into the docking site (S1A and S1B Fig). Successfully genotyped mosquito colonies were expanded and used to perform experiments.

yellow(-)KI and sag(-)KI lines
Pupae obtained from plasmid injected eggs were sorted according to sex and crossed with 100 Ngousso wild type partners. G1 neonate larvae were screened for the expression of EGFP using a Nikon SMZ18 fluorescence Stereomicroscope. Selected transgenic larvae were sex sorted at the pupal stage and crossed with an excess of wild type partners. The G2 progenies (from a single positive G1 female for sag(-)KI and from 10 positive individuals for yellow(-)KI) were COPAS sorted to enrich for the integrated transgene and to remove the vasa-Cas9 transgene (marked with DsRed). The generated sag(-)KI colony was kept in equilibrium with the wild type allele, while the yellow(-)KI colony was completely homozygotized for the transgene by selecting for mosquitoes with a yellow body color. Homozygous sag(-)KI mosquitoes were crossed with the C2S line expressing Cre in the germline [19] for Cre recombinase-mediated excision of the lox cassette to generate sag(-)EX mosquitoes.

PLOS PATHOGENS
A toolbox of engineered mosquito lines to study salivary gland biology polymerase (ThermoFisher Scientific) or the Phire Tissue Direct PCR Master Mix (Thermo-Fisher Scientific) (S2 and S7 Figs) according to manufacturer recommendations. Primers used for genotyping are described in Table 1.

Mosquito infection and evaluation of parasite numbers
Parasites were maintained as frozen stocks at -80˚C or by passage between mice. For this, blood was taken by heart puncture from a donor mouse with a parasitaemia of 3-5%, diluted to 2.10 7 parasitized erythrocytes in 100 μL 0.9% NaCl that were injected intravenously into a naïve mouse. Parasitemia was monitored via flow cytometry (Accuri C6, BD biosciences) three days later. Mosquitoes were fed for 10-15 min on anesthetized mice with 2.5-3.5% parasitemia. Blood fed females were sorted and kept on 10% sucrose at 21˚C and 50-70% humidity. Midguts and salivary glands were dissected in PBS under a SZM-2 Zoom Trinocular stereomicroscope (Optika). Images were acquired with the Nikon SMZ18 Stereomicroscope (salivary glands were imaged without and midguts with coverslip) or with a Zeiss LSM 780 confocal microscope (sealed coverslip). To count oocysts, midguts were dissected 7-8 days post infection. Fluorescence images of infected midguts were processed using the watershed segmentation plugin [47] and oocysts were subsequently counted using the "analyze particles"function implemented in Fiji [39]. To count salivary gland sporozoites, salivary glands were collected in PBS 17-18 days post infection and ground with a plastic pestle for one minute. The number of sporozoites was measured using a Neubauer hemocytometer under a light microscope (Zeiss).

Fluorescence imaging, fluorescence measurements and image analysis
Samples imaged using the Nikon SMZ18 Stereomicroscope were usually acquired using the respective fluorescence filter to visualize expression of DsRed or roGFP2/EGFP combined with brightfield to visualize the whole specimen. Images were taken using a 63-fold magnification objective (NA 1.4) on a Zeiss Axio Observer Z1 confocal LSM 780 microscope equipped with an Hamamatsu Orca Flash 4.0 V1 and a Yokogawa CSU X-1 camera (e.g. All other images as well as images used for fluorescence measurements were acquired with a Nikon SMZ18 Stereomicroscope. For fluorescence quantifications, images were acquired in the DsRed channel and settings were kept the same for all experiments. For each salivary gland, only a single lobe was measured by analyzing a square of 51×51 pixels covering the central part of the fluorescent area to determine the mean fluorescence intensity. The image processing for all experiments was performed with Fiji [39]. To count cells in median lobes, salivary glands were dissected in PBS supplemented with Hoechst 33342 and incubated for approximately 10 min at room temperature before imaging using a Zeiss LSM 780 confocal microscope. Median lobes free of external tissues and with entire nuclear staining were imaged for GFP and Hoechst 33342 fluorescence. To ensure complete imaging of all nuclei, imaging was started at the bottom of the median lobe and continued in z-direction to the top. The step size was set manually in this process in order not to miss weaker stained nuclei. To determine cell numbers, all images acquired for a given median lobe were opened simultaneously in Fiji and stained nuclei were counted using the "Multipoint Tool". Cells were visually divided into low and high production cells by identifying and defining a transcriptional boundary between the two populations (see Fig 3C, 3D and 3E).

roGFP2 measurements
Dissected salivary glands were treated as in [48]. In brief, tissues were alkylated in 20 mM Nethyl-maleimide (NEM) for 15 min to preserve the redox state of roGFP2 and then fixed in

PLOS PATHOGENS
A toolbox of engineered mosquito lines to study salivary gland biology paraformaldehyde (4% PFA) for 20 min. In parallel, full oxidation and reduction controls were prepared by incubating some samples in 2 mM diamide or 20 mM dithiothreitol (DTT), respectively, prior to alkylation with NEM and fixation as described before. All compounds were prepared in phosphate-buffered saline. Salivary glands were mounted with VectaShield (Vector Laboratories) and samples were stored at 4˚C in the dark until analysis. Samples were imaged with a Zeiss Axio Observer Z1 confocal LSM780 microscope with GaAsP detector by using a Plan-Apochromat 63x/1.4 oil objective. The fluorescent probe hGrx1-roGFP2 was excited sequentially at 405 and 488 nm, and its emission signal was recorded at 500-530 nm. Images were processed in ImageJ following the protocol described in [49]. In brief, background was subtracted by using the rolling ball procedure set to 50 pixels and images were converted to 8-bit. Then, images were smoothed and stacked. The intensity of the 488 nm image was thresholded with the default settings in black and white; values below threshold were set to "not a number". The ratio image was generated by dividing the 405 nm picture by the 488 nm picture using the "Ratio Plus" plugin, and colored using the lookup table "fire". Subsequently, intensities were normalized to the DTT control.

In vivo imaging
For in vivo imaging experiments, infected yellow(-)-aapp-DsRed mosquitoes were pre-screened between days 17-21 post infection for the presence sporozoites (GFP fluorescence) at the hinge of the wings and in salivary glands, as well as for cuticular positioning of the salivary glands (DsRed signal) (S9A and S9B Fig). After removing their legs, mosquitoes were glued on a microscopy slide using small amounts of UV-light reactive glue (Bondic Pocket) and covered with a coverslip prepared with plasticine feet at each corner to prevent a squeezing of the sample. Corners of the cover slip were coated with nail polish to prevent sliding. Imaging was performed on a Zeiss Axio Observer Z1 confocal LSM780 microscope with GaAsP detector using a Plan-Apochromat 63x/1.4 oil objective. Movies of sporozoites and salivary glands were acquired for a duration of 2-5 minutes and 2-3 seconds between frames using the Metamorph acquisition software.

Statistical analysis
Statistical analysis was performed using GraphPad Prism 5.0 (GraphPad, San Diego, CA, USA). Data sets were either tested with a one-way ANOVA or a Mann Whitney test. A value of p<0.05 was considered significant.  4 -Supplement 1D) and of a female mosquito heterozygous for the trio-DsRed high fluorescence transgene (trio-DsRed high fluo) and the wild type allele (WT) obtained from family 1 (Fig 4 -Supplement 1E). An illustration of the modified X1 locus containing the trio-DsRed transgene in comparison to the unmodified wild type allele is shown on top. Genotyping primers and expected products are shown as black arrows and lines, respectively. The length of the expected PCR products is indicated below the gel images. Note that for the heterozygous female carrying the trio-DsRed high fluorescence transgene, the smaller PCR fragment amplifying the unmodified locus dominates while the long PCR fragment representing the transgene containing allele is visible as a faint band at the upper edge (marked by a red arrow). The OpiE2-Puro PCR to verify the presence of the transgene gave several unspecific bands. PCRs were performed on genomic DNA obtained from single mosquitoes. (TIF)

S8 Fig. Evaluation of P. berghei parasite infection in salivary gland reporter lines and background fluorescence in yellow(-)KI mosquitoes.
Oocyst (A) and salivary gland sporozoite (spz) (B) counts in infected aapp-DsRed, aapp-hGrx1-roGFP2 and trio-DsRed females in comparison to wild type (G3) females. Data pooled from 3-4 experiments generated with three different mosquito generations. The total number of dissected mosquitoes is given above each genotype. Data points represent parasites counted per midgut (A) and mean number of sporozoites per mosquito in independent experiments (B). C) Number of salivary gland sporozoites (spz) per oocyst for all three reporter lines in comparison to wild type (WT: G3). The invasion rate was calculated using the data shown in (A) and (B). Each dot represents the mean of an independent experiment, the median is indicated by a bar. All data were tested for significance using a Kruskal-Wallis test. ns: not significant (p>0.05). D) GFP background fluorescence of yellow(-)KI in comparison to a wild-type (Ngousso) female. Combinatorial transcriptional

PLOS PATHOGENS
A toolbox of engineered mosquito lines to study salivary gland biology activity guided by the 3xP3 and the endogenous yellow promoter drive EGFP expression in the eye and the ovaries of yellow(-)KI females (indicated by red asterisks). No EGFP-like fluorescence was observed in the throat region where the salivary glands localize (indicated by dashed red circle). Scale bar: 1 mm. The image on the right shows EGFP fluorescence from a dissected ovary of a yellow(-)KI female at higher magnification. (TIF) ). The number of evaluated mosquitoes is indicated above each column. D) Sample preparation before imaging. Legless mosquitoes were glued on a microscopy slide and covered with a cover slip underlaid with orange plasticine to avoid squeezing. Corner areas were subsequently coated with teal-colored nail polish to prevent the cover slip from moving. E) DsRed, GFP and merging the two signals of the image shown in Fig 8A. Note that a significant portion of the DsRed signal is also visible in the GFP channel, whereas the sporozoites are exclusively visible in the GFP channel, potentially indicating spillover of the red signal due to the high DsRed concentration. (TIF) S1 Movie. Sporozoites display active motility inside the mosquito. Movie showing three sporozoites imaged in close proximity to the salivary gland. Sporozoites appear to be attached to a larger vessel and observed movements might be a combination of active motility and passive hemolymph flow. Imaging performed with a Hamamatsu Orca Flash 4.0 V1 camera using a 63x (NA 1.4) objective. GFP signal of a 3 min movie with 3 seconds between frames. (AVI) S2 Movie. Sporozoites form accumulations close to the salivary gland. Movie showing several sporozoites directly below the cuticle. Three sporozoites display different patterns of active movement. Imaging performed with a Hamamatsu Orca Flash 4.0 V1 camera using a 63x (NA 1.4) objective. GFP signal of a 2 min and 30 s movie with 2 seconds between frames. (AVI) S1 Appendix. Promoter sequences of the trio, saglin and aapp genes used to generate the mosquito lines aapp-DsRed, aapp-hGrx1-roGFP2, trio-DsRed and sag-DsRed. Nucleotides highlighted in red are part of the primer sequence used to amplify the promoter sequence. Sequences with a grey background indicate the 5'UTRs of the trio and aapp transcript, respectively. (DOCX) assistance during microscopy and the whole mosquito immune responses (MIR) team for discussions and assistance with mosquito breeding. We also thank the CNRS, Inserm and the University of Strasbourg for providing the infrastructure, salaries and for their support.

PLOS PATHOGENS
A toolbox of engineered mosquito lines to study salivary gland biology